Animal Communication Networks

http://www.cambridge.org/9780521823616

P1: JZZ/JZZ P2: JZZ0521823617agg.xml CU1917B/McGregor 0 521 582361 7 April 12, 2005 13:42

ii

P1: IYP0521809592agg.xml CB777B/Carrington 0 521 80959 2 April 9, 2005 15:59

This page intentionally left blank


Animal Communication Networks

Most animal communication has evolved and now takes place in the context of a

communication network: several signallers and receivers within communication

range of each other. This idea follows naturally from the observation that many

signals travel further than the average spacing between animals. This is

self-evidently true for long-range signals, but at a high density the same is true

for short-range signals (e.g. begging calls of nestling birds). This book provides a

current summary of research on communication networks and appraises future

prospects. It combines information from studies of several taxonomic groups

(insects to people via fiddler crabs, fish, frogs, birds and mammals) and several

signalling modalities (visual, acoustic and chemical signals). It also specifically

addresses the many areas of interface between communication networks and

other disciplines (from the evolution of human charitable behaviour to the

psychophysics of signal perception, via social behaviour, physiology and

mathematical models).

P. K. McGregor was Head of the Department of Animal Behaviour at

Copenhagen University; he is now Reader in Applied Zoology at Cornwall

College, Newquay, UK. He is editor of the journal Bioacoustics and on the editorial

board of several other academic journals.

i


ii


AnimalCommunicationNetworks

Edited by

P. K. McGregorUniversity of Copenhagen andCornwall College, Newquay, UK

iii

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University PressThe Edinburgh Building, Cambridge , UK

First published in print format

- ----

- ----

© Cambridge University Press 2005

2005

Information on this title: www.cambridg e.org /9780521823616

This book is in copyright. Subject to statutory exception and to the provision ofrelevant collective licensing agreements, no reproduction of any part may take placewithout the written permission of Cambridge University Press.

- ---

- ---

Cambridge University Press has no responsibility for the persistence or accuracy ofs for external or third-party internet websites referred to in this book, and does notguarantee that any content on such websites is, or will remain, accurate or appropriate.

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org

hardback

eBook (EBL)

eBook (EBL)

hardback

http://www.cambridge.org

http://www.cambridge.org/9780521823616


Contents

List of contributors viii

Preface xiii

1 Introduction 1Peter K. McGregor

Part I Behaviours specific to communication networks

Introduction 9

2 Eavesdropping in communication networks 13Tom M. Peake

3 Public, private or anonymous? Facilitating and countering

eavesdropping 38Torben Dabelsteen

4 Performing in front of an audience: signallers and the social

environment 63Ricardo J. Matos & Ingo Schlupp

5 Fighting, mating and networking: pillars of poeciliid

sociality 84Ryan L. Earley & Lee Alan Dugatkin

6 The occurrence and function of victory displays within

communication networks 114John L. Bower

Part II The effects of particular contexts

Introduction 129

v


vi Contents

7 Enlightened decisions: female assessment and

communication networks 133Ken A. Otter & Laurene Ratcliffe

8 Predation and noise in communication networks of

neotropical katydids 152Alexander B. Lang, Ingeborg Teppner, Manfred Hartbauer &Heiner Römer

9 Nestling begging as a communication network 170Andrew G. Horn & Marty L. Leonard

10 Redirection of aggression: multiparty signalling within a

network? 191Anahita J. N. Kazem & Filippo Aureli

11 Scent marking and social communication 219Jane L. Hurst

Part III Communication networks in different taxa

Introduction 247

12 Waving in a crowd: fiddler crabs signal in networks 252Denise S. Pope

13 Anuran choruses as communication networks 277T. Ulmar Grafe

14 Singing interactions in songbirds: implications for social

relations and territorial settlement 300Marc Naguib

15 Dawn chorus as an interactive communication network

320John M. Burt & Sandra L. Vehrencamp

16 Eavesdropping and scent over-marking 344Robert E. Johnston

17 Vocal communication networks in large terrestrial

mammals 372Karen McComb & David Reby

18 Underwater acoustic communication networks in marine

mammals 390Vincent M. Janik

19 Looking for, looking at: social control, honest signals and

intimate experience in human evolution and history 416John L. Locke


Contents vii

Part IV Interfaces with other disciplines

Introduction 445

20 Perception and acoustic communication networks 451Ulrike Langemann & Georg M. Klump

21 Hormones, social context and animal communication 481Rui F. Oliveira

22 Cooperation in communication networks: indirect

reciprocity in interactions between cleaner fish and client

reef fish 521Redouan Bshary & Arun D’Souza

23 Fish semiochemicals and the evolution of communication

networks 540Brian D. Wisenden & Norman E. Stacey

24 Cognitive aspects of networks and avian capacities 568Irene M. Pepperberg

25 Social complexity and the information acquired during

eavesdropping by primates and other animals 583Dorothy L. Cheney & Robert M. Seyfarth

26 Communication networks in a virtual world 604Andrew M. R. Terry & Robert Lachlan

Index 628


Contributors

Filippo AureliSchool of Biological and Earth Sciences, Liverpool John Moores University, Byrom St,Liverpool L3 3AF, UK

John L. BowerFairhaven Office 348, Fairhaven College, Western Washington University,Bellingham, Washington 98225-9118, USA

Redouan BsharyDepartment of Zoology, University of Cambridge, Downing St, Cambridge CB2 3EJ,UK. Present address: Evolutionary Psychology and Behavioural Ecology ResearchGroup, School of Biological Sciences, Crown St, University of Liverpool, LiverpoolL69 7ZB, UK

John M. BurtCornell Laboratory of Ornithology, 159 Sapsucker Woods Rd, Ithaca, NY 14850, USA.Present address: Department of Psychology, Box 351525, University of Washington,Seattle, WA 98195-1525, USA

Dorothy L. CheneyDepartment of Biology, University of Pennsylvania, Philadelphia, PA 19104, USA

Torben DabelsteenDepartment of Animal Behaviour, Copenhagen University Zoological Institute,Tagensvej 16, DK-2200 Copenhagen N, Denmark

Arun D’SouzaDepartment of Animal Ecology and Tropical Biology, University of Wurzburg, 97074Wurzburg, Germany

Lee Alan DugatkinDepartment of Biology, University of Louisville, Louisville, KY 40292, USA

viii


List of contributors ix

Ryan L. EarleyDepartment of Biology, Georgia State University, 20 Peachtree Center Ave NE, 402Kell Hall, Atlanta GA 30303, USA

T. Ulmar GrafeDepartment of Animal Ecology and Tropical Biology, University of Wurzburg, 97074Wurzburg, Germany

Manfred HartbauerInstitute of Zoology, Karl-Franzens-University, Universitatsplatz 2, A-8010 Graz,Austria

Andrew G. HornDepartment of Biology, Life Science Centre, Dalhousie University, 1355 Oxford St,Halifax, Nova Scotia, Canada B3H 4J1

Jane L. HurstFaculty of Veterinary Science, University of Liverpool, Leahurst Veterinary FieldStation, Neston, South Wirral L64 7TE, UK

Vincent M. JanikCentre for Social Learning and Cognitive Evolution and the Sea Mammal ResearchUnit, Gatty Marine Laboratory, University of St Andrews, Fife KY16 8LB, UK

Robert E. JohnstonDepartment of Psychology, Uris Hall, Cornell University, Ithaca, NY 14853, USA

Anahita J. N. KazemSchool of Biological Sciences, University of Wales Bangor, Brambell Building, DeiniolRd, Bangor LL57 2UW, UK

Georg M. KlumpCarl von Ossietzky Universitat Oldenburg, AG Zoophysiologie and Verhalten, FB 7,26111 Oldenburg, Germany

Robert LachlanDepartment of Biology, Coker Hall, University of North Carolina at Chapel Hill, NorthCarolina 27599, USA

Alexander B. LangInstitute of Zoology, Karl-Franzens-University, Universitatsplatz 2, A-8010 Graz,Austria

Ulrike LangemannCarl von Ossietzky Universitat Oldenburg, AG Zoophysiologie and Verhalten, FB 7,26111 Oldenburg, Germany

Marty L. LeonardDepartment of Biology, Life Science Centre, Dalhousie University, 1355 Oxford St,Halifax, Nova Scotia, Canada B3H 4J1


x List of contributors

John L. LockeNew York University, 719 Broadway (Suite 200), New York, NY 10003, USA. Presentaddress: Department of Speech-Language-Hearing Sciences, Lehman College, CityUniversity of New York, 250 Bedford Park Boulevard West, Bronx, New York 10468,USA

Ricardo J. MatosDepartment of Animal Behaviour, Copenhagen University Zoological Institute,Tagensvej 16, DK-2200 Copenhagen N, Denmark

Karen E. McCombExperimental Psychology, School of Biological Sciences, University of Sussex, Falmer,Brighton BN1 9QG, UK

Peter K. McGregorDepartment of Animal Behaviour, Copenhagen University Zoological Institute,Tagensvej 16, DK-2200 Copenhagen N, Denmark. Present address: Centre for AppliedZoology, Cornwall College Newquay, Trenance Gardens, Newquay Cornwall TR7 2LZ,UK

Marc NaguibDepartment of Animal Behavior, University Bielefeld, PO Box 10 01 31, 33501Bielefeld, Germany

Rui F. OliveiraUnidade de Investigacao em Eco-Etologia, Instituto Superior de Psicologia Aplicada,Rua Jardim do Tabaco 34, 1149–041 Lisbon, Portugal

Ken A. OtterEcosystem Science and Management Program, University of Northern BritishColumbia, 3333 University Way, Prince George, British Columbia, Canada V2N 4Z9

Tom M. PeakeDepartment of Animal Behaviour, Copenhagen University Zoological Institute,Tagensvej 16, DK-2200 Copenhagen N, Denmark

Irene M. PepperbergMIT School of Architecture and Planning, Brandeis University and Department ofPsychology, Waltham, MA 02454, USA

Denise S. PopeDepartment of Animal Behaviour, Copenhagen University Zoological Institute,Tagensvej 16, DK-2200 Copenhagen N, Denmark. Present address: Department ofBiology, Trinity University, 1 Trinity Place, San Antonio, TX 78212-7200, USA

Laurene RatcliffeDepartment of Biology, Queen’s University, Kingston, Ontario, Canada K7L 3N6

David RebyExperimental Psychology, School of Biological Sciences, University of Sussex, Falmer,Brighton BN1 9QG, UK


List of contributors xi

Heiner RomerInstitute for Zoology, Karl-Franzens-University, Universitatsplatz 2, A-8010 Graz,Austria

Ingo SchluppZoologisches Institut, Universitat Zurich, Winterthurerstrasse 190, CH-8057 Zurich,Switzerland and Section of Integrative Biology C0930, University of Texas, Austin,TX 78712, USA. Present address: Biozentrum Grindel, Universitat Hamburg,Martin-Luther-King Pl. 3, D-20146 Hamburg, Germany

Robert M. SeyfarthDepartment of Psychology, University of Pennsylvania, Philadelphia, PA 19104-6196,USA

Norman E. StaceyDepartment of Biological Sciences, University of Alberta, Edmonton, Alberta, CanadaT6G 2E9

Ingeborg TeppnerInstitute of Zoology, Karl-Franzens-University, Universitatsplatz 2, A-8010 Graz,Austria

Andrew M. R. TerryDepartment of Animal Behaviour, Copenhagen University Zoological Institute,Tagensvej 16, DK-2200 Copenhagen N, Denmark. Present address: IUCN – The WorldConservation Union, Regional Office for Europe, Rue Vergote 15, 1030 Bruxelles,Belgium

Sandra L. VehrencampCornell Laboratory of Ornithology, 159 Sapsucker Woods Rd, Ithaca, NY 14850, USA

Brian D. WisendenDepartment of Biology, Minnesota State University Moorhead, 1104 7th St S,Moorhead, MN 56563, USA


xii


Preface

This book attempts to reflect the state of current research on communica-

tion networks: groupings of several individuals that constitute the social context

in which communication takes place. In my view, a structured collection of chap-

ters by active researchers best conveys the excitement of the research findings as

well as the underlying expertise of the authors, especially when a wide range of

taxa and signalling modalities are addressed.

The motivation to edit such a book came from the interest in the topic that

was evident after seminars and conference presentations. However, it was the

symposium on communication networks at the XXVIIth International Ethological

Conference held in Tubingen that converted motivation into action. The symposium

showed (at least to my satisfaction) how well the topic integrated research on

different taxa and signalling modalities. It was also the opportunity to meet Shana

Coates of Cambridge University Press and to appreciate her enthusiasm for a ‘book

of the symposium’.

The book has turned out to be much more than a collection of symposium

papers. First, it covers considerably more ground in its 26 chapters than was pos-

sible in a symposium of nine spoken papers. Second, some of the stimulating

informal discussions that characterize a good conference have contributed to the

section introductions. However, the main ‘added value’ comes from the willing-

ness of the authors to comment on the chapters of others, to incorporate comments

and cross-references into their own chapters and, above all, to look at communi-

cation from a network perspective. In many instances, this has led to insights that

are likely to have a major effect on the direction of research on animal communi-

cation – real Eureka moments. It has been a privilege to share in these moments.

Many people deserve my thanks for the role they have played in the cre-

ation of this book. Marc Naguib suggested that we submit the symposium topic

xiii


xiv Preface

‘communication networks’ to the IEC scientific committee. This committee and

the conference main organizer, Raimund Appfelbach, were kind enough to accept

the topic and in doing so set the ball rolling. Shana Coates of Cambridge University

Press gently, but firmly, ensured that my timescale for editing a book on communi-

cation networks was advanced from ‘some time in the future’to ‘in the next couple

of months’. Considerable credit is also due to Tom Peake and Andrew Terry, who

applied their own particular brand of pressure (accompanied by several cappucci-

nos) to stimulate me to draft the book proposal on the flight back from Tubingen.

Shana gave excellent advice in the early stages of the book, since when Tracey

Sanderson and Martin Griffiths have overseen production. Of course, there would

be no book without the authors. I am very grateful to all of them for finding time

in overcrowded schedules to write their own chapters and to comment on those

of others. Denmark’s Statens Naturvidenskabelige Forskningsrad has supported

my research for the last 5 years. København Universitet supported me during the

initial stages of the book, but the bulk of the work was done with support from

the EU and Cornwall College via a Marie Curie Category 40 Fellowship. Last, but by

no means least, I thank Leonie and Tom McGregor – for sustaining me throughout

the project with their love, good humour and flexibility at all times, particularly

when the going got tough.

P1: IKB/... P2: KOD0521823617c01.xml CU1917B/McGregor 0 521 582361 7 April 10, 2005 15:52

1

Introduction

p e t e r k . m c g r e g o r

University of Copenhagen, Denmarkand Cornwall College, Newquay, UK

Some of the most conspicuous behaviours performed by an animal are

related to communication – communication that mediates reproduction and sur-

vival. As explained below, a knowledge of animal communication is important in

more respects than simply its role in understanding such fundamental aspects of

animals’lives. This book is about a perspective that can increase our understanding

of animal communication.

One way in which animal communication is important is that it interfaces

with and links several fields of study. In the field of behaviour, for example, com-

munication is often used to illustrate Niko Tinbergen’s four types of question

(function, mechanism, development and evolution) and how the answers comple-

ment each other (e.g. Krebs & Davies, 1993). Communication has interfaces with

many other areas of biology including evolution, ecology, population genetics,

neurobiology and physiology. For example, it can be a window into the cognitive

worlds of animals (e.g. Ch. 24). Links with other sciences are shown by the use

of ideas and techniques from psychology to understand how communication is

perceived (Ch. 20), and using information from physics and chemistry to explain

how communication is achieved (e.g. Bradbury & Vehrencamp, 1998).

Communication cannot occur in isolation; it is an inherently social behaviour.

This makes it even more surprising that the wider social context in which commu-

nication takes place is rarely considered explicitly. As explained in the next para-

graph, it is likely that communication commonly occurs in the context of a net-

work of several animals. This chapter is both a brief introduction to this context –

animal communication networks – and an explanation of this book’s structure.

Animal Communication Networks, ed. Peter K. McGregor. Published by Cambridge University Press.c© Cambridge University Press 2005.

1


2 P. K. McGregor

About communication networks

A communication network is a group of several animals within signalling

and receiving range of each other. If signals travel further than the average spacing

between individuals, then there is potential for a communication network to exist.

This is as true for the ocean-spanning songs of whales as it is for the begging calls

of songbird nestlings crammed into a nest cavity, and it is why networks can be

considered to be the commonest context in which communication occurs (e.g.

McGregor & Peake, 2000).

This would seem to be stating the obvious, especially to those new to the field of

animal communication. Indeed, those studying chorusing animals, particularly

insects and anuran amphibians, have long adopted a network perspective and rec-

ognized the importance of doing so (e.g. Otte, 1974). However, it is only relatively

recently that other types of communication have been considered explicitly in a

network context. Communication was, and still is in many instances, treated as oc-

curring between two individuals – the signaller–receiver dyad – perhaps because

this is the simplest relationship possible between the three basic components

found in communication (the signaller, the signal and the receiver). In this sense,

a dyadic view of communication follows from the stricture of Occam’s razor (also

known as the law of parsimony) to employ the simplest explanation consistent

with the facts. While agreeing wholeheartedly with this standard scientific prac-

tice, it is clear that a dyadic view of communication is often not consistent with

the facts. One example is the high signal level used in close-range aggressive en-

counters – human antagonists nose to nose, yet shouting at each other – surely

high signal levels are not needed to achieve signalling at such close range? In a

network context, such high levels make more sense, because there may be more

distant intended receivers (the gathering crowd in the human example) in addi-

tion to the opponent (Zahavi, 1979). Many further examples of communication

that are best considered in the context of a communication network are found

throughout this book.

Another reason for explicitly considering communication in a network context

is that it identifies communication behaviours that cannot occur in a dyad. A good

example is eavesdropping, particularly social eavesdropping (Ch. 2) in which the

eavesdropper extracts information from a signalling interaction between others.

Social eavesdropping requires a minimum of three individuals (one eavesdrop-

ping, two more interacting) and, therefore, falls outside a dyadic view of com-

munication. The evidence for eavesdropping and its wider implications (e.g. for

comparative cognition) is presented in many of the chapters of this book. Eaves-

dropping and similar network behaviours discussed in this book are considered

by many to be a compelling reason to adopt a network perspective.


Introduction 3

Communication networks and eavesdropping

It is perhaps worth emphasizing that, while eavesdropping is a good ex-

ample of communication network behaviour, it is not the only one, and the value

of the communication network perspective does not depend on a demonstration

of eavesdropping. The reason for its current prominence is that it was considered

first and, therefore, at the moment it is more prevalent in the literature. There is

no merit in shoe-horning a natural example into a definition of eavesdropping,

nor in judging the value of any natural communication behaviour by how well it

fits this (or any other) definition. As several chapters demonstrate (e.g. Chs. 9 and

23), such examples from the real world can probe and challenge our definitions

(e.g. of interactions and of communication more generally) and the thinking that

follows from them. The result can be considerable insight and lead to progress for

the whole field of communication.

A note on definitions

Clear and workable definitions are the essential basis for meaningful

discussion. I have tried to ensure that terms are used clearly and consistently

within a chapter, but there may be good reasons why chapters differ in the detail

of their definitions (e.g. for reasons discussed in the previous paragraph). There

are no instances in this book where the same term is used in a markedly different

way in different chapters, but readers should bear in mind that the detail of the

definition may be important to the topics discussed by the chapter.

There are two nice illustrations of the problems that definitions can create. The

first concerns eavesdropping. Alan Grafen pointed out a problem with the term

eavesdropping after I had used it when presenting ideas on communication net-

works at the Royal Society Meeting on Signalling in 1992 (McGregor, 1993). The prob-

lem he foresaw was that in everyday use the term means secret information gath-

ering, and it was clear to him that there may be advantages to the signallers in pro-

viding information (i.e. promoting eavesdropping), especially if the signaller had

won the agonistic contest (see also Zahavi, 1979). The everyday meaning of eaves-

dropping and its implicit association with acoustic signals have been at the root

of several misunderstandings that could perhaps have been avoided if a more neu-

tral term had been used (at the time Grafen suggested type II receivers). Tom Peake

has sorted out this and other problems to do with definitions of eavesdropping

with admirable clarity in Ch. 2. Nevertheless, information gathered without the

source’s knowledge may have particular value, as John Locke discusses in Ch. 19. I

think this demonstrates that identifying the secrecy or otherwise of information

gathering is the route to progress, rather than rigidly applying a definition.


4 P. K. McGregor

The second example concerns the relationship between information and com-

munication. In my view, the terms are clearly not synonymous; rather signals

are a subset of information because they are specialized to transmit information

(more details in McGregor & Peake (2000)). This could have created a problem with

semiochemicals: if they are not signals (i.e. they contain information but are not

specialized to transmit it) then the behaviour involving them is not communica-

tion and the concept of communication networks would not apply. Fortunately

for the book, Brian Wisenden and Norm Stacey thought carefully about the is-

sue and realised that there were many important similarities that gave them an

opportunity to discuss the functional and evolutionary relationships between in-

formation, signals and networks (Ch. 23). So a problem arising from definitions

has given real insight, rather than the acrimonious defence of definitions that is

all too common in the literature.

About this book

Coverage

There are several types of book on animal communication. Some are syn-

optic treatments of the whole topic (e.g. Hauser, 1996; Bradbury & Vehrencamp,

1998) whereas others concentrate on particular types of signal such as pheromones

(Wyatt, 2003) or on a group of animals such as arthropods (Greenfield, 2002).

Many books do both, for example dealing with acoustic communication in insects

(Gerhardt & Huber, 2002) or birds (Kroodsma & Miller, 1996). This book is rather

different in that it looks at a specific topic in communication and covers several

modalities and taxonomic groups.

Organization

Each chapter has been written so that it can be read alone, since this

is a common way for edited volumes to be read. Inevitably, this has led to some

similarity between chapters in their opening remarks, but I think this is more than

offset by each chapter having its own reference section. The many cross-references

to other chapters in the book also show the extent to which authors have taken

account of material in other chapters and made links between them.

A second way in which the book has been given overall coherence is to group the

chapters into four parts that reflect major aspects of communication networks.

Each of these parts is prefaced by a short overview that identifies chapter themes

and highlights some of the issues that remain to be tackled. The fact that many

chapters could have been put into any of the four parts further demonstrates

the extent of overall coherence of the book and the wide-ranging nature of the


Introduction 5

chapters. Within each part, there is no particular order of chapters, although in

Part III the order is loosely phylogenetic.

The chapters grouped into Part I deal with communication behaviours, such

as eavesdropping and audience effects, that involve three or more individuals (i.e.

a communication network) and as such fall outside the ‘classical’ or traditional

dyadic (one signaller and one receiver) approach to communication.

Part II groups particular contexts that are fruitful to consider from a communi-

cation network perspective: mate choice, predation, begging, aggression and scent

marking.

The reason for grouping chapters in Part III is taxonomic: from fiddler crabs

to humans via most groups of vertebrate. While communication networks may

be more or less ubiquitous, features of different taxa (e.g. main senses, social

organization) can have a major effect on the details of communication networks

and provide insight into the topic as a whole.

The final part contains chapters that, to a greater or lesser degree, link com-

munication and other disciplines in biology and more widely in science. From the

evidence of these chapters, a network perspective seems to be particularly valuable

at such subject interfaces.

Summary

There are several reasons for considering that the natural context in which

communication occurs (and in which it has evolved) is a network of several animals

in signalling and receiving range of each other. However, this context has not been

considered explicitly in many studies of animal communication. The chapters in

this book apply a communication network perspective to a variety of taxa using a

number of signal modalities in several circumstances. The results are illuminating.

To modify a marketing phrase used for mobile phones: the future is bright; the

future is a network view of communication.

References

Bradbury, J. W. & Vehrencamp, S. L. 1998. The Principles of Animal Communication.

Sunderland, MA: Sinauer.

Gerhardt, H. C. & Huber, F. 2002. Acoustic Communication in Insects and Anurans: Common

Problems and Diverse Solutions. Chicago, IL: Chicago University Press.

Greenfield, M. D. 2002. Signalers and Receivers: Mechanisms and Evolution of Arthropod

Communication. Oxford: Oxford University Press.

Hauser, M. D. 1996. The Evolution of Communication. Cambridge, MA: MIT Press.

Krebs, J. R. & Davies, N. B. 1993. An Introduction to Behavioural Ecology, 3rd edn. Oxford:

Blackwell Scientific.


6 P. K. McGregor

Kroodsma, D. E. & Miller, E. H. 1996. Ecology and Evolution of Acoustic Communication in

Birds. Ithaca, NY: Cornell University Press.

McGregor, P. K. 1993. Signalling in territorial systems: a context for individual

identification, ranging and eavesdropping. Philosophical Transactions of the Royal

Society of London, Series B, 340, 237–244.

McGregor, P. K. & Peake, T. M. 2000. Communication networks: social environments

for receiving and signalling behaviour. Acta Ethologica, 2, 71–81.

Otte, D. 1974. Effects and functions in the evolution of signaling systems. Annual

Review of Ecology and Systematics, 5, 385–417.

Wyatt, T. D. 2003. Pheromones and Animal Behaviour: Communication by Smell and Taste.

Cambridge, UK: Cambridge University Press.

Zahavi, A. 1979. Why shouting? American Nauralist, 113, 155–156.

P1: JZZ/IKB P2: JZZ/...0521840716agg.xml CU1917B/McGregor 0 521 582361 7 April 10, 2005 16:0

Part I B E H A V I O U R S S P E C I F I C T O

C O M M U N I C A T I O N N E T W O R K S

7


8


Introduction

The reason for grouping together the chapters that appear in this part

of the book is that each of them concerns communication behaviours that are

best viewed from a communication network perspective, rather than from the

more common dyadic (one signaller to one receiver) standpoint. It is a fact that,

with the exception of choruses, most studies to date have implicitly or explicitly

considered communication between a dyad. Although the communication net-

work perspective of several signalling and receiving individuals seems to follow

logically from what we know of natural communication, the dyadic viewpoint has

historical precedence and considerable inertia. A network perspective will become

more commonly adopted only if it is clearly better able to explain communication

behaviours than a dyadic approach. It is for this reason that a network perspective

has long been adopted in studies of choruses; the effect on an individual’s signal

timing of the signals of nearby conspecifics can be striking patterns, such as signal

synchrony in the chorus (e.g. Greenfield, 2002; Ch. 13). Such patterns cannot be

explained by considering communication as a dyad. All of the chapters in this

book demonstrate the value of adopting a network perspective; however, it gives

this demonstration more emphasis to begin with a section covering communi-

cation behaviours that are particularly suited to, or associated with, a network

perspective.

Eavesdropping

In Ch. 1, eavesdropping is identified as a receiving behaviour that has been

particularly identified with, and is only possible in, a communication network.

The first two chapters of Part I look at eavesdropping in more detail.


9


10 Part I

In Ch. 2, Tom Peake summarizes the evidence for eavesdropping in different

contexts and also discusses the diverse use of the term in the literature. His di-

vision of eavesdropping into interceptive eavesdropping (e.g. predators locating

prey from prey signals) and social eavesdropping (extracting information from a

signalling interaction) is an important clarification. However, as Tom points out,

clarifying definitions is more important as means of moving arguments on from

the question of whether a given behaviour can be called eavesdropping or not and

towards a more fruitful and general approach based on the nature of information

transfer.

Torben Dabelsteen deals mainly with social eavesdropping on the acoustic sig-

nals of birds in Ch. 3. He identifies the potential costs and benefits of eavesdrop-

ping and uses information from studies of how bird song transmits through the

habitat to explore how eavesdropping is best achieved. The overall balance of costs

and benefits of being eavesdropped upon will determine whether signallers pro-

mote eavesdropping on their signals or whether they try to avoid it. One intriguing

possibility that Torben discusses is whether the costs of being eavesdropped upon

could be avoided if signallers made their signals anonymous by removing infor-

mation on signaller identity.

Audience effects

In communication networks, several receivers are likely to be present

during signalling interactions between others; these receivers do not take part in

the interaction and have been referred to as an audience. The effects they can have

on signalling behaviour are the subject of Ch. 4, in which Ricardo Matos and Ingo

Schlupp draw the distinction between an apparent audience and an evolutionary

audience. The distinction is important because selection pressures imposed by

the presence of audiences in the evolutionary past of the animals may result in

features of the signalling interactions despite the absence of an audience during

any particular interaction. Also, whether an audience is apparent to the signallers

involved in interactions may depend on signal modality: individuals have to be

in the line of sight of visual signals to receive them and, therefore, an audience is

likely to be apparent; however, the same is not true of widely broadcast acoustic

signals.

Bystanders

Being a bystander (i.e. present, but not directly involved) during an agonis-

tic or mating interaction can affect subsequent aggressive and mating behaviour

and is explored in Ch. 5. Ryan Earley and Lee Dugatkin focus on social eaves-

dropping (a subset of bystanding) by two species of poeciliid fishes that are likely


Behaviours specific to communication networks 11

to be familiar to many – green swordtails Xiphophorus helleri and guppies Poecilia

reticulata – and that communicate largely with visual signals. Their chapter shows

how a network view can encompass and organize diverse aspects of fighting and

mating behaviour (including mate copying); it also identifies the many conditions

that favour eavesdroppers and how the effects of eavesdropping are manifested.

Victory displays

In the final chapter of this section, John Bower examines victory displays:

signals produced by the winner (but not the loser) after an aggressive interaction.

There has been surprisingly little work specifically on this topic, despite the wealth

of studies of signalling before and during aggressive displays, and such informa-

tion is widely scattered. Chapter 6 collates the information on victory displays

and then interprets its functional significance, first from a dyadic perspective and

then from a network perspective. It may have been premature to include victory

displays in this section, because on current evidence it is not clear that victory dis-

plays always function in a network context rather than in the winner–loser dyad.

However, even if their main function is dyadic, their conspicuous nature makes

it likely that other individuals could gain useful information by paying attention

to victory displays.

Future directions

The authors dealing with eavesdropping make several suggestions for the

directions future research should take: incorporating eavesdropping into theoret-

ical models to derive testable predictions that can contribute to understanding

signal evolution (Ch. 2); finding evidence of eavesdropping in non-experimental

natural contexts (perhaps by using a combination of tracking and acoustic location

technologies to follow the individuals in a network), and continuing such studies

long enough to identify differences in reproductive success (Ch. 3); unravelling the

complex interrelationships between features of individuals, their social and wider

environment and the role of bystanders in order to understand communication

fully (Ch. 5).

The authors dealing with eavesdropping clearly consider that the phenomenon

is now well characterized. In contrast, victory displays clearly need more detailed

study in order to establish the phenomenon and to elucidate its function and

whether it is network phenomenon. It is likely that controlled laboratory experi-

ments are the best way to investigate what effect, if any, victory displays have on

other members of the communication network (Ch. 6).

Progress in understanding audience effects seems likely to come from a differ-

ent type of approach. In addition to modelling and controlled experiments, there is


12 Part I

the potential to integrate information on audience effects with underlying mech-

anisms. Suitable candidate mechanisms exist in the literature (e.g. the hormonal

basis of priming effects) and deserve to be investigated more fully (Ch. 4).

References



P1: JZZ/... P2: JZZ/...0521823617c02.xml CU1917B/McGregor 0 521 582361 7 April 10, 2005 16:16

2

Eavesdropping in communicationnetworks

t o m m . p e a k e

University of Copenhagen, Denmark

Introduction

All communication occurs in a network environment with the exception

of a subset of systems that unequivocally meet both of the following criteria: (a) a

signal can never be received by more than one receiver; (b) a receiver can never

receive more than one signal simultaneously. In other words, all communication

networks have at least one of two defining properties: (a) signals can be, at least

potentially, received by several receivers; and (b) receivers can, at least potentially,

receive signals from several signallers at any one time. Consequently, in moving

from a dyadic consideration of communication to a network view, signallers and

receivers both take on a range of costs and benefits, which are the theme of this

book. In this chapter, I will consider the implications of a particular type of re-

ceiving behaviour that becomes possible in a network, namely eavesdropping. I

will begin by reviewing different definitions of eavesdropping that are found in

the literature and the evidence for different types of eavesdropping, distinguish-

ing between eavesdropping on signals and eavesdropping on signal interactions. I

will then examine the costs, benefits and implications of eavesdropping on inter-

actions, as recognition of this phenomenon emerged from considerations of

qualitative differences between dyadic and network views of communication

(McGregor, 1993; McGregor & Dabelsteen, 1996).

Defining eavesdropping

The verb eavesdrop is defined by the New Oxford Shorter English Dictionary as

‘Listen secretly to (a person, private conversation), orig, by standing beneath the


13


14 T. M. Peake

eaves of a house. Formerly also, stand beneath the eaves of (a building) in order to

overhear conversation within.’ This word, at least to native English speakers, has

an evocative quality that makes it appealing to authors in a variety of often quite

different contexts. The use of the word in everyday language also has connotations

that it may be useful to discard at this stage. First, in everyday use the term applies

only to the acoustic modality; as a technical term in animal communication there

is no good reason why this should be, although undoubtedly this has contributed

to the term not being used by authors working in some modalities. Second, the

idea of secrecy contained in the above definition need not necessarily be carried

over to its use in animal communication.

In the context of animal communication, the term has been used in a number of

different ways that can be summed up by a general definition: the use of information

in signals by individuals other than the primary target. This definition excludes the

use of eavesdropping to describe behaviours such as detecting prey by cues that

are not designed to enhance information transfer (e.g. extraneous noise caused

by movement); in this sense the definition differs from that given by Bradbury &

Vehrencamp (1998, p. 3). Eavesdroppers have been called ‘illegitimate’ (Otte, 1974),

‘unintended’ (Wiley, 1994) or ‘third party’ (Zahavi, 1979) receivers or ‘bystanders’

(e.g. Dugatkin, 2001) according to the context in which they were defined. In the

general definition above, I use the phrase ‘individuals other than the primary

target’ on the grounds that, as I shall outline below, eavesdropping individuals do

not under all circumstances impose a cost on signallers as is implied by some of the

alternative terms above. In situations where the presence of eavesdroppers benefits

signallers, there may be selection pressure to allow information transmission to

eavesdroppers, while the major selective force remains the more apparent (or

primary) receiver.

Within this general definition, there are two classes of use of the term eaves-

dropping that are sufficiently different, yet sufficiently commonly used, to war-

rant discussion and clarification. Here I call these classes interceptive and social

eavesdropping. Interceptive eavesdroppers benefit by intercepting signals in-

tended (in an evolutionary sense) for another individual, usually to the cost of

the signaller. Social eavesdroppers gather information on other individuals by

attending to their signalling interactions with conspecifics. At first glance, these

two types of behaviour may seem very similar; however, as I shall argue in the

remainder of the chapter, the nature of information transfer and resulting selec-

tion pressures differ markedly between them. These terms were chosen carefully

to indicate the source from which eavesdroppers gather information (i.e. inter-

cepting signals versus attending to social interactions), without undue prejudice

towards certain aspects more commonly (but rarely exclusively) associated with

either behaviour. The aim of this distinction is to move arguments or points of

confusion away from the question of whether a given behaviour can be called


Eavesdropping in communication networks 15

Table 2.1. Generalizations concerning the two types of eavesdropping behaviour

defined in the text

Type of eavesdropping Interceptive Social

Source of information Signals Signal interactions

Type of signal Usually broadcast Always directed

Eavesdropper–signaller

relationship

Usually heterospecific Usually conspecific

Payoff to signaller Usually negative or zero Positive, negative or zero

Information gathered Absolute Relative information also available

eavesdropping or not and towards a more fruitful approach based on the nature

of information transfer. While these definitions of eavesdropping require that sig-

nals be transmitted to more than one receiver (network property (a) above), social

eavesdropping further requires that receivers can detect more than one signal at

the same time (network property (b)). One fact is unavoidable: eavesdropping is,

by definition, a behaviour that can only occur in a network as it requires at least

three individuals: a signaller, a target receiver and an eavesdropper.

A number of generalizations may be made that show the distinctions between

the two types of eavesdropping (summarized in Table 2.1). (a) Interceptive eaves-

dropping usually involves the reception of broadcast signals (i.e. those that have

a class of targets such as females of the signaller’s species rather than a spe-

cific target); social eavesdropping, by definition, involves the exchange of sig-

nals directed towards specific receiving individuals. (b) Interceptive eavesdrop-

ping is most commonly identified in situations where eavesdroppers are a differ-

ent species from the signaller; social eavesdropping is usually identified within

a species. (c) Interceptive eavesdropping usually has a negative or zero effect on

signallers; the payoff to a signaller resulting from social eavesdropping is much

less clear, as will be discussed below. (d) Interceptive eavesdropping focuses on

the absolute signalling behaviour of the signaller (in many cases this may be sim-

ple presence/absence of information); social eavesdropping additionally allows

information to be gathered on the relative performance of interacting signallers

(allowing both direct comparison of interactants and assessment of relationships

between them).

Interceptive eavesdropping

Wiley (1983) defined eavesdropping as the behaviour by which ‘signals

intended for one receiver are intercepted by another’; this definition is explic-

itly given as an example of a receiver ‘obtaining information about the signaller


16 T. M. Peake

against its own best interests’. Bradbury & Vehrencamp (1998) adapted this defi-

nition to include as eavesdropping situations where the signaller obtains a zero

benefit, terming as ‘exploitation’ cases in which eavesdroppers are detrimental to

signallers. In defining interceptive eavesdropping above, I make no assumptions

about the nature of the signaller payoff. The definition is intended to be simply

descriptive and one could imagine many subdivisions that could be made. One

clear distinction is in the taxonomic relationship between signaller and eaves-

dropper. Where signaller and eavesdropper are of different species, as is the case

in most examples of signal interception found in the literature, the payoff to the

signaller is almost certainly negative. Information obtained by eavesdropping in

this case may be something as simple as the location of a suitable prey item. The

effect of eavesdropping within a species is likely to be more difficult to determine

and the kinds of information gathered may well be more related to features of the

signaller.

As a final point, it is suggested by some authors (e.g. Bradbury & Vehrencamp,

1998; Greenfield, 2002) that signals intercepted by another species, particularly

in such cases as predators locating prey, should be considered cues as they are not

designed to enhance information transfer to those receivers. This is certainly true

if one considers predator and prey in isolation; however, when considering the

structure of the source of information and factors associated with production it

is important that the wider context is included.

Interspecific examples

The most commonly cited examples of eavesdropping, and those that have

been best studied, are those that occur between trophic levels, i.e. predators and

parasites detecting the signals of prey or hosts or prey detecting predator presence

by their signals. Selection pressures imposed by these kinds of eavesdropper are un-

derstandably high and have been shown to lead to a range of counter-adaptations

that aim to ameliorate or avoid such pressures (e.g. Greenfield, 1994, 2002; Heller,

1995; Stoddard, 1999; Gerhardt & Huber, 2002; Ch. 8). Examples are particu-

larly prevalent in acoustic (e.g. Cade, 1975; Ryan et al., 1982; Sakaluk & Belwood,

1984; Belwood & Morris, 1987) and chemical (Aldrich, 1995; Roberts et al., 2001)

signalling and there are good reviews available (e.g. Stowe et al., 1995; Zuk &

Kolluru, 1998). While many visual signals are conspicuous and may be used by

predators to find prey (e.g. Lloyd & Wing, 1983), it is rare to find such examples

called eavesdropping (see Bruce et al. (2001) for such an example).

Such interceptive eavesdropping reflects communication networks working on

a community level. The selection pressures on communication between trophic

levels are widely acknowledged in most considerations of the evolution of commu-

nication. Much less widely studied and appreciated is the importance of networks



operating within a species, and there is evidence that interceptive eavesdropping

occurs at this level.

Intraspecific examples

Within a species, individuals may eavesdrop on the signalling behaviour

of others for a variety of reasons. In some cases, animals of one sex eavesdrop

on signals intended for the opposite sex (Ch. 12). For example, Kiflawi & Gray

(2000) looked at eavesdropping by male crickets Acheta domesticus on competing

males’ mating calls. Smaller males showed a phonotactic response towards speak-

ers broadcasting calls preferred by females in a two-speaker choice design, while

larger males varied in their phonotaxis. The suggestion here is that males with

unattractive calls can potentially intercept females as they move towards attrac-

tive males.

A recent example concerns the use of female signals, apparently intended for

mates, as a means of detecting fertile females for extra-pair copulations. Female

robins Erithacus rubecula produce ‘seep’ calls to obtain provisioning from their

mate (East, 1981). Mate removal experiments show that females may attract other

males, which provide courtship feeding that may result in copulation (Tobias &

Seddon, 2000). Tobias & Seddon (2002) found that neighbouring males approached

‘seep’ calls when they were played back at a high rate near the territory boundary,

on occasion bearing provisions. They suggest that, if the female call is a hunger

signal directed towards the mate, neighbouring males might be eavesdropping

(in the interceptive sense). In this case, females may derive a benefit from the

presence of eavesdroppers while the primary benefit comes from the response of

the mate. Tobias & Seddon (2002) also acknowledged the possibility that the ‘seep’

call is directed towards extra-pair males as a means of ‘blackmailing’the mate into

providing food. In either case, the results highlight the influence of operating in

a social network.

Individuals may eavesdrop on signals designed to warn others of the presence of

predators. For example, Shennan et al. (1994) describe the behaviour of group mem-

bers paying attention to the vigilance activities of others in order to avoid preda-

tors. Convict cichlids Cichlasoma nigrofasciatum fin-flick in order to warn young; fish

not guarding young do not fin-flick. Parents were shown to fin-flick in response to

fin-flicking models, suggesting that they are capable of monitoring the vigilance

activities of others in order to warn their own young sooner. Here, the primary tar-

gets of the signal are likely to be relatives of the signaller, while non-relatives may

also benefit by paying attention to the signals at no obvious cost to the signaller.

This early warning feature of the signalling behaviour of others has been sug-

gested as a possible function of territoriality (e.g. Eason & Stamps, 1993). Obser-

vations of red-capped cardinals Paroaria gularis showed that territorial males had


18 T. M. Peake

a high chance of detecting intrusions where the intruder had been recently re-

pelled from a neighbour’s territory as a consequence of the conspicuousness of

behaviours involved in eviction. Intruders that had not been detected by neigh-

bours were unlikely to be detected by territorial subjects.

In all of these examples, individuals use signals produced by conspecifics for

their own benefit, as fits the definition of interceptive eavesdropping. Less clear is

the effect of such eavesdropping on the signaller. In some cases (e.g. Kiflawi & Gray,

2000), the signaller may suffer a cost because of the presence of eavesdroppers,

while in some there is no obvious benefit or cost to the signaller (e.g. Shennan et al.,

1994). In some cases (e.g. Tobias & Seddon, 2002), the presence of eavesdroppers

may actually benefit the signaller. These examples show how definitions of eaves-

dropping and communication based on costs to the signaller (e.g. Wiley, 1983;

Bradbury & Vehrencamp, 1998) may not apply to all circumstances. It is for this

reason that I prefer the descriptive definition of eavesdropping in general as the

use of signals by receivers other than the primary target (see above).

Autocommunication and eavesdropping

Eavesdropping has also been used to describe the interception of informa-

tion contained in sounds produced by animals in order to investigate their environ-

ment. Although not strictly within the general definition of eavesdropping given

above, because these sounds are not designed to transmit information to others,

the examples are interesting enough to be worth mentioning. Little brown bats

Myotis lucifugus gather information by paying attention to the echolocation calls

of foraging conspecifics (Barclay, 1982). Bats approached speakers broadcasting

echolocation calls of conspecifics and a heterospecific Eptesicus fuscus, suggesting

that eavesdropping could substantially increase potential prey detection distance.

Balcombe & Fenton (1988) found similar results in M. lucifugus, a congener Myotis

yumanensis and another species (Lasiurus borealis), in which individuals apparently

used others’ echolocation calls to identify ‘vulnerable’ prey in order to steal them

from the eavesdropped bat. Similarly, Xitco & Roitblat (1996) use the term eaves-

dropping to describe the ability of a bottlenose dolphin Tursiops truncatus to identify

objects inspected by another dolphin, via the reception of the inspector’s echolo-

cation clicks.

Social eavesdropping

In one of the first explicit considerations of territorial systems as a commu-

nication network, McGregor (1993) suggested that the term eavesdropping could

be applied within species, particularly in the context of paying attention to inter-

actions between neighbours and rivals. While initially McGregor (1993) considered



this analogous to interceptive eavesdropping, McGregor & Dabelsteen (1996) later

made the distinction, refining their definition of eavesdropping to ‘extracting in-

formation from an [signalling] interaction between other individuals’. They con-

sidered it ‘a prerequisite of eavesdropping that a third party (the eavesdropper)

gains information from an interaction that could not be gained from a signal

alone’. This, they suggested, was a different level of information transfer than

simply locating a prey item by its signals. McGregor & Peake (2000) attempted to

clarify the hierarchical relationship between information, signals and signalling

interactions, concluding that interactions may be under additional selection pres-

sures to those acting on the signals themselves. The most obvious source of these

selection pressures is those with an interest in the outcome: rivals and mates.

Experimental evidence for social eavesdropping has recently increased dra-

matically as clear experimental paradigms have emerged. Studies that explicitly

address eavesdropping in this context have thus far been carried out exclusively

on acoustic interactions in territorial songbirds and visual interactions in teleost

fish. However, evidence from other experiments not designed to test eavesdrop-

ping per se are strongly supportive of the existence of eavesdropping as a means

of gaining information on the qualities of and/or social relationships between

conspecifics.

Acoustic interactions in songbirds

McGregor et al. (1997) addressed the issue in songbirds using interactive

playback (Dabelsteen et al., 1996) to simulate intrusion upon the neighbours of

subject male great tits Parus major (Fig. 2.1ai). Neighbours were presented with

one of two types of intruder. One type indicated its willingness to escalate by

beginning each song immediately following the onset of neighbour song (over-

lapping: Hulsch & Todt, 1982; Dabelsteen et al., 1996, 1997) and increasing song

length. The other type of intruder playback signalled a lower level of willingness

to escalate by beginning songs only after the neighbour songs had been completed

(alternating) and reducing song length. After a short amount of time, an intru-

sion by the same intruder was simulated in the subject male’s territory (Fig. 2.1aii)

singing an alternating pattern with matched song length. Subjects responded to

previously aggressive intruders by keeping their distance and overlapping song,

while less-aggressive intruders were approached quickly.

A similar experiment carried out on great tits looked at the behaviour of fe-

males in response to intruders interacting with their mates and neighbours (Otter

et al., 1999). In this case, experiments were carried out on dyads of neighbouring

territories each defended by a mated pair. Playback was used to intrude on each

territory on successive days (Fig. 2.1bi, bii) such that the same intruder showed a

high willingness to escalate to one male and a low willingness to escalate to the


20 T. M. Peake

(a) (b)

(c)

(d)

(e)

KEY

(i)

(i) (i)

(i)

(i)

(ii) (ii)

(ii)

(ii)

(ii) (iii)

(iii)

Loudspeaker

Loudspeakerbroadcastingplayback

Interaction between two loudspeakers

Interaction between male and loudspeaker

Fig. 2.1. Schematic representations of experiments described in the text investigating

social eavesdropping on acoustic interactions in songbirds. (a) Representation of

design used by McGregor et al. (1997) showing (i) interaction between loudspeaker and

neighbouring male and (ii) subsequent playback intrusion in subject’s territory.

(b) Design used by Otter et al. (1999) and Mennill et al. (2002) showing (i) interaction

between one male and a loudspeaker, (ii) subsequent interaction between

neighbouring male and loudspeaker and (iii) observation of female behaviour.

(c) Design used by Naguib & Todt (1997) and Naguib et al. (1999) showing (i) interaction

between two loudspeakers inside the subject’s territory and (ii) subsequent playback

from loudspeaker not approached initially by the subject. (d) Design used by



neighbouring male. Otter et al. (1999) then followed the females for some time

subsequent to the treatments (Fig. 2.1biii) and showed that females whose mates

had suffered from an intruder that a neighbour dealt with easily were more likely

to trespass onto neighbouring territories, particularly that of the neighbour who

had performed well during playback. These female forays were not converted into

offspring, however (Otter et al., 2001), suggesting that the short-term nature of

the information was not enough to convince females of the poor quality of their

mates. Mennill et al. (2002), however, did find such an effect in female black-capped

chickadees Poecile atricapillus using a similar experimental paradigm. In this case,

Mennill et al. (2002) had information on males’ dominance ranks during winter

feeding flocks so that dyads of neighbouring territories each consisted of one

mated pair in which the male was high ranking and the other in which the male

was low ranking. Playback was carried out to these dyads in a similar way to that

used by Otter et al. (1999) and was followed by microsatellite paternity analysis

in order to assess female reproductive decisions. The results showed that high-

ranking males that had lost to playback showed a much greater incidence of lost

paternity with extra-pair young in 12 of 23 nests, compared with 2 of 20 in control

nests. Low-ranking males that did well against intruders lost paternity to the same

extent as controls.

In all of the above experiments, the conclusion is that the responses of subjects

are a result of information gained by paying attention to the interactions between

intruders and known males. This interpretation is, however, somewhat limited by

the fact that, in each case, the response of the subject may be affected by the

response of the known male: because the interaction of interest was between

a male simulated by playback and a live male, the subsequent response of the

subject may be affected by changes in behaviour of the live male. For example,

in McGregor et al. (1997), the response of the subject may result from changed

behaviour of the neighbour following different levels of intrusion; similarly, in

the latter two studies, females may have changed their behaviour in response to

changed behaviour of their mates. Mennill et al. (2002) did, in fact, examine the

Fig. 2.1 (cont.) Peake et al. (2004) showing (i) interaction between two loudspeakers

outside the territory boundary and (ii) subsequent intrusion by one of the

loudspeakers. (e) Design used by Peake et al. (2002) showing (i) intrusion by

loudspeaker, (ii) subsequent interaction between that and another loudspeaker

outside the territory boundary and (iii) intrusion by the second loudspeaker. Rounded

rectangles represent territory boundaries; male and female symbols represent

approximate positions of resident males and females; arrows represent movements by

loudspeakers; arrows with curved lines represent monitoring of female movements.

See text and cited references for more details of each experiment.


22 T. M. Peake

behaviour of males following each playback treatment and could find no effect

on subsequent behaviour.

The problem of lack of control over the signalling behaviour of interactants

and subsequent changes in behaviour can be avoided in songbirds by replacing

males with playback; dyadic encounters can then be simulated using two loud-

speakers. The first study to use this approach was carried out on nightingales

Luscinia megarhynchos by Naguib & Todt (1997), who examined the effect of asym-

metric interactions on the responses of territorial males. The asymmetry in this

case was achieved by having one loudspeaker producing songs that overlapped

the other. Loudspeakers were placed inside the territory boundary of the subject

and interactions lasted for two minutes (Fig. 2.1ci). Males responded by spending

more time near, spending more time singing near, and singing more songs near

the overlapping speaker. Ten minutes after the ‘interaction’ had finished, play-

back for one minute was broadcast from the speaker that was not approached

first during the interaction (Fig. 2.1cii). Males sang more at the location of the

formerly overlapping speaker regardless of whether that speaker was producing

song. Naguib et al. (1999) repeated this experiment with a different kind of inter-

action in which songs did not overlap but were still asymmetrical as one speaker

(the follower) always directly followed the output of another (the leader). In this

case, males showed a stronger response to the speaker that ‘led’; once again the

subjects responded differently to the two types of apparent opponent (Naguib &

Todt, 1997).

In these two experiments, the design meant that subjects could associate roles

during an interaction with the location of a singing intruder. Peake et al. (2001)

looked at whether similar associations could be made between roles and song

features using a similar experimental paradigm in great tits. In this case inter-

actions were carried out between two loudspeakers situated outside the territory

boundary and thus in an area that subjects would be less willing to approach di-

rectly (Fig. 2.1di). Information extracted by subjects was then assayed by means of

a third speaker placed well inside the subject’s territory, which broadcast songs

of one of the interactants 15 minutes after the interaction (Fig. 2.1dii). Three

types of interaction were used, again based on song timing between the speakers:

overlapping, alternating and random. Intruders were then one of four types: over-

lappers, alternators, random (i.e. no consistent role) or males that had been over-

lapped by the opponent. In response to the assay intrusion, subjects responded

with equally high song output to overlappers, alternators and random interactants

but showed a twofold reduction in song towards males that had been overlapped.

In all three of these experiments (Naguib & Todt, 1997; Naguib et al., 1999;

Peake et al., 2001), the only difference between intruders was in relative song

timing during the interaction, i.e. there was no absolute information available



upon which subjects could base their responses. The results then clearly showed

that subjects had eavesdropped on the interaction as a whole and associated the

roles of interactants with either the location of the singer (Naguib & Todt 1997;

Naguib et al., 1999) or features of his song (Peake et al., 2001).

These three experiments also share the feature that, in each case, the subjects

had no prior experience of either interactant; therefore, decisions must have been

made purely on the basis of the interaction. In reality, territorial songbirds are

likely to have knowledge of the relative strengths of neighbouring males as a result

of direct interactions during territory establishment and maintenance and indi-

rectly from hearing them interact with others. Therefore, individuals may be able

to use these known individuals as ‘yardsticks’against which to measure previously

unencountered individuals. The first three studies mentioned in this section at-

tempted to address this issue by looking at eavesdropping on encounters between

intruders and neighbouring males (McGregor et al., 1997) and/or mates (Otter et al.,

1999; Mennill et al., 2002). In these studies, however, there was little control over

eavesdroppers’prior experience with these yardsticks and the possibility that they

may have themselves contributed to the responses shown (see above).

Peake et al. (2002) attempted to address these problems by carefully control-

ling prior experience with an individual. This experiment was similar to the two-

speaker experiment mentioned above (Peake et al., 2001). The difference was that

one of the interactants (A) was introduced to the subject prior to the interaction by

means of a territorial intrusion simulated by interactive playback (Fig. 2.1ei). The

initial intruder either played an aggressive role, overlapping the subject’s song, or

a much less-aggressive role, beginning a song one second after the subject had fin-

ished each song, allowing the subject to overlap playback. Following this intrusion,

an interaction was simulated outside the territory between the recent intruder

and a male (B) unknown to the subject (Fig. 2.1eii); here either A or B played the

aggressive role by overlapping the song of the other. By combining the outcomes

of the two interactions, four treatment types were carried out that provided in-

formation on the status of B relative to the subject. In two cases the information

available did not clearly show the status of B relative to the subject: either A was

aggressive to both B and the subject or A received aggression from both. In the

other two cases, the information available was clear: A showed low aggression to

the subject but high aggression to B, indicating B to be of low status, or A was

highly aggressive to the subject but received aggression from B, indicating that

B was of high status relative to the subject. In response to subsequent intrusion

by B (Fig. 2.1eiii), males showed a threefold reduction in song towards males that

were of low relative status (as indicated by the information available from the

treatment type), compared with the response to high-status males or males about

which information did not reliably determine status. This result shows that males


24 T. M. Peake

combined information from the two interactions, one they took part in and one

they heard, in deciding how to respond to subsequent intrusion.

In all of these experiments, great care was taken to ensure that the informa-

tion available to eavesdroppers was purely relative. This is important in order to

demonstrate social eavesdropping, i.e. that individuals pay attention to the inter-

action rather than simply the absolute outputs of either male. However, during

real interactions it is likely that both absolute and relative information is avail-

able to, and indeed used by, social eavesdroppers. A recent experiment on great tits

(Peake et al., 2004) attempted to address this issue. Male great tits have a repertoire

of one to six song types, many of which are shared by neighbouring individuals

and used during song interactions (matched counter-singing; Krebs et al., 1981;

Falls et al., 1982). Peake et al. (2004) used the two-loudspeaker design of Peake et al.

(2001; Fig. 2.1d) to simulate interactions in which interactants differed in their

use of song types. In each interaction, one speaker (A) produced the same song

type throughout the interaction. The other speaker (B) began producing a differ-

ent song type from A and then switched song types halfway through. On half of

the occasions, B switched to the same song type as A (matching); on the other

half of occasions B switched to a song type that was different from A. Thus, there

were four possible intruders during the assay intrusion: males that switched to

match (matchers), males whose opponent switched to match (matched), males

who switched but did not match their opponent (switchers) and males whose op-

ponents switched but did not match (switched). In both types of interaction there

is a clear absolute difference between males in signalling behaviour, i.e. singing

one song type versus two song types. Between the two types of interaction there is

also relative information, available only in the interaction as a whole, in whether

the switching individual matched his opponent. Subjects responded to simulated

intruders by singing much shorter songs to those individuals that used two song

types compared with those singing one song type. In addition, subjects did not

approach or spend time near switched intruders, compared with no difference

in approach response to the other types of intruder. Therefore, it seems that in

this case the response of males to simulated intruders used both relative informa-

tion in the interaction (switching and matching) and absolute information in the

signalling behaviour of the individual interactants (one or two song types).

Visual interactions in fish

Eavesdropping on visual displays given by male Siamese fighting fish Betta

splendens during male–male interactions has been shown by both males (Oliveira

et al., 1998) and females (Doutrelant & McGregor, 2000). In these experiments, sub-

jects (who could see other males without themselves being seen) were allowed

to witness interactions between males displaying across a transparent barrier



(Fig. 2.2ai). At the same time, two other males were taking part in a similar in-

teraction that could not be seen by the subject, allowing a control for changes

in the male opponents providing information on the outcome rather than (or as

well as) information from the dynamics of the interaction. Male subjects (Oliveira

et al., 1998) were then introduced to each of the four interactants (two seen and

two unseen) in turn (Fig. 2.2aii) and the response measured. Males responded

to individuals that they had seen lose by approaching and displaying sooner

than with males that they had seen win. No such differences were seen in re-

sponse to the winners and losers of displays that had not been witnessed. Features

of the subjects’ behaviour during the interaction strongly suggested that the

information used by subjects in responding was gathered by eavesdropping (see

discussion in McGregor & Peake, 2000). In experiments with females (Doutrelant

& McGregor, 2000), the seen and unseen interactions were temporally separated

rather than concurrent (Fig. 2.2bi, bii). Following interactions, female subjects

were allowed to move freely so as to exhibit a proximity preference for either

male (Fig. 2.2biii). Females visited seen winners first, more often and spent more

time near and displaying to seen winners than seen losers. Unseen losers were

visited first more often than unseen winners, with no difference in the time spent

near or displaying towards either male.

Similar results have been found in green swordtails Xiphophorus helleri by

Earley & Dugatkin (2002). Males allowed to view contests between other males

without themselves being seen (Fig. 2.2ci) responded more cautiously to perceived

winners, being less willing to initiate contests. Males that had witnessed contests

were much less willing to escalate contests than males that had not seen contests.

Males allowed to interact with contesting males during the contest (Fig. 2.2cii),

and hence assess contestants more directly, were as likely to win contests as those

that had not seen contests, suggesting that individual differences between the fish

settled those contests. Males that had not witnessed contests (Fig. 2.2ciii) tended

to escalate, whereas males that had interacted with contestants were unlikely to

escalate, suggesting that these individual differences were assessed previously.

These results also suggested that, where information on an opponent’s fighting

ability was available from direct interaction, a presumably more reliable source,

males tended to ignore the less-direct information gathered by eavesdropping.

The difficulties of providing visual stimuli in the absence of live animals makes

it much less straightforward to achieve the level of control over interactions

afforded to songbird studies by acoustic playback. The use of models (e.g. Shennan

et al., 1994) or video playback (Oliveira et al., 2000) potentially allows control over

stimuli at a comparable level to acoustic playback, but neither has yet been used to

simulate interactions. In the absence of such an approach, one way to delve deeper

into the relationship between interactions and eavesdroppers is to decouple the


26 T. M. Peake

(a)

(d)

(c)

(b)

KEY

(i)

(ii)

(ii)

(i)

(iii)

(i) (ii) (iii)

(i) (ii)

Opaque partition

Transparent partition

One-way glass (arrow shows visible direction)

Fig. 2.2. Schematic representations of experiments described in the text investigating

social eavesdropping on visual interactions in fish. All fish were physically isolated by

partitions; opaque, solid line; transparent, dotted line; one-way glass, dashed line

with arrow showing direction in which visual contact was possible. (a) Design used by

Oliveira et al. (1998) showing (i) two visual signalling interactions across transparent

partitions, one witnessed by the central male, the other not; and (ii) subsequent

presentation (indicated by arrow) of each male to the subject. (b) Design used by

Doutrelant & McGregor (2000) showing (i) interaction between two males witnessed by

a female, (ii) not witnessed and (iii) subsequent observation of female movements

(indicated by arrow). (c) Treatments used by Earley & Dugatkin (2002) in which

(i) interacting males were witnessed by a male that could not be seen by the



experience of interactants and eavesdroppers, i.e. both parties view the interaction

differently. McGregor et al. (2001) attempted to do this using male Siamese fighting

fish. Subject males were allowed to view two conspecific males apparently inter-

acting across a small gap into which the eavesdropper could not see. The gap was

in reality filled by a small aquarium that either was empty (allowing males to inter-

act across the gap: the ‘real’ interaction; Fig. 2.2di) or contained two fish separated

by an opaque partition such that each of the males viewed by the eavesdropper

was, in fact, interacting with a fish that could not be seen (the ‘apparent’ interac-

tion; Fig. 2.2dii). In the apparent interaction, the eavesdropper’s interpretation of

the aggressive signals given by each visible male was decoupled from that male’s

actual experience in his own interaction with a hidden fish. The results showed

that eavesdroppers responded more aggressively to individuals that had displayed

more during apparent interactions, with no such differentiation between males

involved in real interactions. McGregor et al. (2001) suggested that the proximity

of interactants to (hidden) opponents during the apparent interaction (compared

with the relatively large distance between males (about one fish length) in the

real interaction) resulted in an increase in aggressive displays (tail beating) and

behaviour (attempted biting) in these interactions. Therefore, males either paid

more attention to the information in these particularly aggressive encounters or

viewed the winner of a highly aggressive encounter as a different level of threat

to the winner of an encounter of lower general aggression.

The evidence presented above demonstrates that fish pay attention to informa-

tion available in interactions between conspecifics; however, the source of that

information has not been conclusively shown to be the signals exchanged during

interactions. While the results are consistent with the idea that fish eavesdrop,

the difficulties in presenting subjects with fully controlled visual signalling inter-

actions (cf. acoustic playback) means that using the term ‘social eavesdropping’

for this behaviour may be premature.

Other evidence for social eavesdropping

There is, in addition to the examples given above, a variety of evidence

supporting the importance of social eavesdropping as a means of information

gathering. In some cases, there may be clear physiological effects; adult male

Fig. 2.2 (cont.) interactants, (ii) were witnessed by a male that could be seen (and

interacted with) or (iii) were not seen. (d) Treatments used by McGregor et al. (2001) in

which subjects witnessed two males (i) interacting across an empty divide into which

the subject could not see (the real interaction) or (ii) apparently interacting but in fact

interacting with males hidden from the subject (the apparent interaction). See text

and cited references for more details of each experiment.


28 T. M. Peake

cichlid fish Oreochromis mossambicus that witnessed fights between conspecifics

showed elevated androgen levels (testosterone and 11-ketotestosterone) compared

with controls (Oliveira et al., 2001).

Female domestic fowl Gallus gallus that witness a known dominant being de-

feated by a stranger readily submit to that stranger in subsequent interactions

(Hogue et al., 1996). When the stranger lost to a known dominant, the witnessing

individual was able to dominate this stranger subsequently on 50% of occasions. A

similar situation was found in juvenile rainbow trout Onchorhynchus mykiss, when

subjects were allowed to interact with fish that had been seen to be dominant in a

previous encounter or who had been dominant but had not been seen (Johnsson &

Akerman, 1998). Individuals that lost to either dominant (seen or unseen) reduced

aggression more rapidly to seen dominants. Individuals that won over these domi-

nants increased aggression more rapidly to seen dominants. These results suggest

that, while the final outcome may be a result of individual differences, informa-

tion obtained before direct interactions occurred enabled individuals to make

decisions about how to respond to these individuals more quickly. In these stud-

ies, the extent to which social eavesdropping occurs is difficult to assess, as it

is not clear whether information extracted by observers is contained in signal

interactions between participants or in other aggressive behaviours.

Knowledge of the social rank relationships of others through observation of

social interactions is also an important part of forming strategic alliances in some

primate species (Seyfarth & Cheney, 2002; Ch. 25). The required amount of knowl-

edge of this kind quickly becomes enormous as group sizes increase and has been

suggested as one selection pressure driving large brain sizes in primates (Seyfarth &

Cheney, 2002).

Identifying types of eavesdropping

So far I have considered examples that clearly fall into one of the two

classes of eavesdropping; interception of signals or of signal interactions. Situ-

ations could be imagined that are not so clearly placed in one category or the

other. Some of these problems may be caused by the difficulties of defining signal

interactions (Chs. 9 and 14). All the examples considered so far have dealt with in-

teractions in which the signallers use the same modality. Many behaviour patterns

that can be viewed as signalling interactions may occur in different modalities or

switch between modalities as the interaction proceeds. If an acoustic signal given

by a male is responded to by a visual signal from the female, then we can clearly

say there has been an interaction involving signals, but does this constitute a sig-

nalling interaction? There is no good reason why signalling interactions, at least in

terms of information transfer, have to occur in the same modality for each party.



However, there may be logistic difficulties in demonstrating social eavesdropping

in such cases; one would have to show that both signals are involved in producing

a response by the eavesdropper, i.e. that neither signal alone produced the same

response.

An example in which the level of information transfer is currently unclear con-

cerns the use of courtship signals by neighbouring males to detect reproductive

attempts in the whitethroat Sylvia communis (Balsby & Dabelsteen, 2003). In this

case, the response of neighbouring males was recorded during experiments that

simulated a territorial male interacting with an intruding male (via playback) or a

receptive female (via a remotely controlled dummy and playback of female calls).

The simulation of a receptive female resulted in greater song flight activity by

neighbours than simulation of an intruding male, and intrusions by the neigh-

bour (and subsequent evictions) were only seen during simulated courtship events.

Song output by the experimental male could not explain the incidence of intru-

sion during courtship interactions by neighbours. As Balsby & Dabelsteen (2003)

acknowledged, their experiment does not rule out the possibility that either sig-

nal alone provides sufficient information to explain the pattern of intrusion, thus

interceptive eavesdropping may be an appropriate description of this behaviour.

However, the detection of a courtship attempt in which the female is receptive

is much facilitated by the presence of both male and female signals; therefore,

it may be the interaction that is important, making this an example of social

eavesdropping.

Information gathering and implications of eavesdropping

So far I have discussed the different ways in which animals may be con-

sidered to be eavesdropping and some of the contexts in which this behaviour has

been shown to occur. As yet I have considered the kinds of information that eaves-

droppers may gain in only the broadest terms. As discussed by McGregor & Peake

(2000), the information available in signalling interactions that is not available in

signals alone is the important distinction between social and interceptive eaves-

dropping. Just as signals are a subset of the information available in an animal’s

environment, signal interactions are a further subset of signalled information.

The way in which animals use signals during interactions may in many cases be

more revealing than the underlying content of the signals themselves, and in any

case the use and content of signals need not necessarily be directly related.

A number of features of signalling interactions may be particularly important

in this respect; many of these features are particularly obvious in agonistic sig-

nalling encounters. Signal exchanges during agonistic encounters are generally

thought to function so as to reduce the likelihood of direct physical aggression


30 T. M. Peake

(and its ensuing costs) by allowing assessment of the likely outcome of a direct

fight (e.g. Enquist, 1985). Reliability of signals used in this context may be high

because of the threat of having one’s bluff called by one’s opponent. To this end,

and given the immediacy of the potential punishment for cheating, signals given

during these kinds of exchange may be particularly reliable. It is likely that the

most accurate picture of an opponent’s fighting ability (short of actually fighting

them) is gained by becoming involved in an aggressive signal exchange. The relia-

bility of signals given in this context may also make social eavesdropping a good

alternative in terms of obtaining accurate information while avoiding the risk of

escalation. Similarly, while information on the underlying quality of an individual

may be available in signals, immediate quality (e.g. condition, motivation) may be

assessed more accurately when an opponent calls those factors directly into ques-

tion. Thus signal interactions may provide reliable and up-to-date information on

the current quality of participants.

Second, interactions enable a direct comparison to be made between partici-

pants on a relative scale. Simply knowing the outcome of an interaction provides

the information that A is stronger than B. By paying attention to an interaction it

may be possible to extract information on relative quality, e.g. A is much stronger

than B. If selection favours individuals that ‘just do enough’ to win an interaction,

the available information will underestimate the relative difference in quality

between the opponents. However, the nature of the interaction may provide such

information. For example, one might expect interactions involving highly asym-

metrical opponents to be shorter and less intense than those involving closely

matched opponents. Of course, this assumes that the relative quality of oppo-

nents is the only influence on the information contained in the interaction; the

presence of eavesdroppers may well affect the dynamics of interactions (Ch. 4).

In most situations, the presence of eavesdroppers imposes selection pressures

on signallers. In the case of interceptive eavesdropping, the selection pressures

may be severe, especially in the case of predatory eavesdropping. In this case, the

signaller must accept or avoid the costs of signalling; such avoidance mechanisms

are particularly well understood in insects and anurans (e.g. Gerhardt & Huber,

2002; Ch. 8). In the case of selection pressures imposed by social eavesdroppers, the

situation may be much less clear. In some situations, it is apparent that eavesdrop-

pers have imposed strong selection pressures on signalling interactions by the na-

ture of those interactions; for example, intense song duels in birds often involve

switching to ‘quiet’ song when they approach physical aggression (Dabelsteen

et al., 1998). In some cases, however, the opposite seems to be true: signalling

interactions seem to be much more conspicuous than necessary to transmit infor-

mation between participants (Zahavi, 1979). In these cases, it may be reasonable

to assume that signals used here are, at least partly, ‘intended’ for eavesdroppers,



i.e. that communication is no longer restricted to the interaction. This view is

supported by work on the effects of audiences on the dynamics of interactions

(Ch. 4) and recent work on altruism (see below).

In cases where asymmetries become clear to interactants, there may be different

pressures acting on each party: for ‘winning’ individuals to advertise the fact and

‘losing’ individuals to hide it. Here eavesdroppers, as part of the selection regime

in which communication systems have evolved, provide individuals with different

payoffs depending on the current social context. In this case, one would expect

a variety of adaptations in signalling behaviour to allow individuals dynamically

to advertise or privatize information in interactions; many such adaptations have

been suggested (e.g. Chs. 3 and 10).

While considerations of social eavesdropping have focused on aggressive in-

teractions, there are a number of similarities between the considerations of in-

formation gathering in this context and in the context of acts of apparent altru-

ism (Johnstone, 2001). Suggestions that altruists may benefit by being perceived

as such assume that observers are able to associate those acts with the individ-

ual performing them and subsequently use that information (Nowak & Sigmund,

1998). Studies of eavesdropping provide clear evidence that this level of association

occurs, albeit in a different context. Many other aspects of cooperative and non-

cooperative behaviour occurring between individuals may similarly be explained

by the passage of information outside the apparent dyad (Ch. 22).

Costs of eavesdropping

Social eavesdropping has been suggested as a relatively cost-free means

of gathering reliable information on rivals or potential mates. However, the costs

have not been explicitly discussed in the literature aside from a sentence by Mc-

Gregor & Dabelsteen (1996) that ‘listening at a distance only involves forgoing

other behaviours such as feeding’. While gathering information in this way is

undoubtedly less costly than becoming involved in aggressive, possibly physical,

interactions, the costs of listening may be greater than commonly assumed.

Evidence from studies of prey detection strongly suggests that animals have lim-

its to the attention they can give to competing tasks (Dukas, 1998; Dukas & Kamil,

2001). Dukas (1998) suggested that the brain has finite capacity to process informa-

tion such that animals can only process a limited amount of information at any

one time. While studies have so far concentrated on visual attention, presumably

a more demanding process than listening, it may be that similar reasoning ap-

plies to the acoustic sense. In this case, eavesdropping may limit the attention

available for other important tasks such as predator vigilance, particularly if so-

cial eavesdropping is a more cognitively demanding task than simply attending


32 T. M. Peake

to signals. A study on humans provided support for this view (Pendry, 1998). Hu-

man subjects were asked to form an impression of a target person based upon

information provided to them one item at a time on a computer screen. Without

prior warning and with no specific instruction, participants were simultaneously

played a tape recording of a conversation that was either relevant to them or

not (the contents of the conversations were identical; relevance differences were

achieved by changing the object of the conversation). Subsequent tests showed

that participants extracted more information than expected by chance from the

relevant conversation but not from the irrelevant one. Participants hearing the

relevant conversation were much more likely to obtain stereotypical impressions

of the target and recalled many fewer items of information related to the target

than those hearing a conversation of little relevance.

These sorts of cost may still be comparatively low, particularly if eavesdrop-

ping, or otherwise monitoring the social environment, is relatively rare. However,

social monitoring may represent a large proportion of some animals’ lives. A study

of brown capuchin monkeys Cebus apella suggested that monitoring the social en-

vironment was the main function of vigilance behaviour in this species, and that

the amount of time spent in such activity was highly correlated with the number

of neighbouring individuals (Hirsch, 2002). Individuals spent, on average, 12.7%

of their time vigilant, of which nearly 30% could be directly attributed to social

monitoring and less than 10% to predator vigilance (see also Chs. 19 and 25).

Summary and future possibilities

A defining property of communication networks is that more than one

receiver may detect signals. In many cases, at least some of the receivers are not the

primary target of the signal and in some the majority of potential receivers may

fall into this category. Cases where individuals other than the primary target use

information obtained in signals have been termed eavesdropping by a number of

authors. In this chapter, I have distinguished between examples of eavesdropping

that involve the interception of signals (interceptive eavesdropping) and a more

recently suggested phenomenon of gathering information from signalling inter-

actions between conspecifics (social eavesdropping). Interceptive eavesdropping

can have significant effects on signal design and signalling behaviour when eaves-

droppers impose large costs on signallers (e.g. where such eavesdropping occurs

between trophic levels). The effects of interceptive eavesdroppers within species

and social eavesdroppers in general are less well understood.

The focus of studies on social eavesdropping has so far been territorial species,

reflecting the biases of researchers active in this area. However, many different

social situations would seem ideally suited to promoting eavesdropping as a social



behaviour. Group and/or colonial living species presumably have an even greater

opportunity to extract information from interactions because of the close proxim-

ity of individuals to one another. Similarly, the focus of most studies of eavesdrop-

ping so far has been on male–male aggressive interactions. Interactions between

males and females (Ch. 7), between group members (Chs. 10 and 25) and between

parents and offspring (Ch. 9) are just some of the areas that potentially offer an

important source of information to eavesdroppers and would benefit from fur-

ther research. Equally as interesting would be the possibility of taking studies of

eavesdropping beyond the trio of two interacting signallers and an eavesdropper

and in so doing place the effect of eavesdroppers in a more extensive and natural

network environment.

The study of social eavesdropping is still in its infancy, yet the number of stud-

ies showing that such eavesdropping occurs has increased dramatically since the

idea was mooted by McGregor & Dabelsteen in 1996. As the prevalence of studies

demonstrating eavesdropping in communication networks increases, the impor-

tance of eavesdropping as a selective force on signalling and social structure will be

better understood. We currently lack a clear theoretical framework within which

to place the importance of social eavesdroppers in the evolution of signalling

systems. As empirical studies continue to provide evidence that animals clearly

have these capabilities, we eagerly await the emergence of models including eaves-

droppers, such as those of Johnstone (2001) and Terry & Lachlan (Ch. 26); models

that make clear predictions about where, how and when eavesdropping should

occur.

Acknowledgements

I would like to thank the following people for providing discussion and comments

that greatly improved both this chapter and my thoughts on communication networks: Pete

McGregor, Thorsten Balsby, Torben Dabelsteen, Giuliano Matessi, Ricardo Matos, Denise Pope

and Andy Terry. While writing this chapter, I was funded by the Zoological Institute, University

of Copenhagen.

References

Aldrich, J. R. 1995. Chemical communication in true bugs and exploitation by

parasitoids and commensals. In Chemical Ecology of Insects II, ed. R. T. Carde & W. J.

Bell. New York: Chapman & Hall, pp. 318–363.

Balcombe, J. P. & Fenton, M. B. 1988. Eavesdropping by bats: the influence of

echolocation call design and foraging strategy. Ethology, 79, 158–166.


34 T. M. Peake

Balsby, T. J. S. & Dabelsteen, T. 2003. Male calling between courtship sequences in

whitethroats: a way to counter intrusions from neighbouring rivals. Behavioural

Processes, 63, 149–157.

Barclay, R. M. R. 1982. Interindividual use of echolocation calls: eavesdropping by bats.

Behavioral Ecology and Sociobiology, 10, 271–275.

Belwood, J. J. & Morris, G. K. 1987. Bat predation and its influence on calling behavior

in neotropical katydids. Science, 238, 64–67.

Bradbury, J. W. & Vehrencamp, S. L. 1998. Principles of Animal Communication.

Sunderland MA: Sinauer.

Bruce, M. J., Herberstein, M. E. & Elgar, M. A. 2001. Signalling conflict between prey

and predator attraction. Journal of Evolutionary Biology, 14, 786–794.

Cade, W. 1975. Acoustically orienting parasitoids: fly phonotaxis to cricket song.

Science, 190, 1312–1313.

Dabelsteen, T., McGregor, P. K., Shepherd, M., Whittaker, X. & Pedersen, S. B. 1996. Is

the signal value of overlapping singing different from that of alternating singing

during matching in great tits? Journal of Avian Biology, 27, 189–194.

Dabelsteen, T., McGregor, P. K., Holland, J., Tobias, J. A. & Pedersen, S. B. 1997. The

signal function of overlapping singing in male robins (Erithacus rubecula). Animal

Behaviour, 53, 249–256.

Dabelsteen, T., McGregor, P. K., Lampe, H. M., Langmore, N. E. & Holland, J. 1998. Quiet

song in songbirds: an overlooked phenomenon. Bioacoustics, 9, 89–106.

Doutrelant, C. & McGregor, P. K. 2000. Eavesdropping and mate choice in female

fighting fish. Behaviour, 137, 1655–1659.

Dugatkin, L. A. 2001. Bystander effects and the structure of dominance hierarchies.

Behavioral Ecology, 12, 348–352.

Dukas, R. 1998. Constraints on information processing and their effects on behavior.

In Cognitive Ecology, ed. R. Dukas. Chicago, IL: University of Chicago Press,

pp. 89–127.

Dukas, R. and Kamil, A. 2001. Limited attention: the constraint underlining search

image. Behavioral Ecology, 12, 192–199.

Earley, R. L. & Dugatkin, L. A. 2002. Eavesdropping on visual cues in green swordtail

(Xiphophorus helleri) fights: a case for networking. Proceedings of the Royal Society of

London, Series B, 269, 943–952.

Eason, P. K. & Stamps, J. A. 1993. An early warning system for detecting intruders in a

territorial animal. Animal Behaviour, 46, 1105–1109.

East, M. 1981. Aspects of courtship and parental care of the European robin Erithacus

rubecula. Ornis Scandinavica, 12, 230–239.

Enquist, M. 1985. Communication during aggressive interactions with particular

reference to variation in choice of behaviour. Animal Behaviour, 33, 1152–1611.

Falls, J. B., Krebs, J. R. & McGregor, P. K. 1982. Song matching in the great tit (Parus

major): the effect of similarity and familiarity. Animal Behaviour, 30, 977–1009.

Gerhardt, H. C. & Huber, F. 2002. Acoustic Communication in Insects and Anurans. Chicago,

IL: University of Chicago Press.

Greenfield, M. D. 1994. Cooperation and conflict in the evolution of signal

interactions. Annual Review of Ecology and Systematics, 25, 97–126



2002. Signalers and Receivers. New York: Oxford University Press.

Heller, K. G. 1995. Acoustic signaling in palaeotropical bushcrickets (Orthoptera:

Tettigonioidea: Pseudophyllidae): does predation pressure by eavesdropping

enemies differ in the Palaeo- and Neotropics? Journal of Zoology, 237, 469–485.

Hirsch, B. T. 2002. Social monitoring and vigilance behavior in brown capuchin

monkeys (Cebus apella). Behavioral Ecology and Sociobiology, 52, 458–464.

Hogue, M. E., Beaugrand, J. P. & Lague, P. C. 1996. Coherent use of information by hens

observing their former dominant defeating or being defeated by a stranger.

Behavioral Processes, 38, 241–252.

Hulsch, H. & Todt, D. 1982. Temporal performance roles during vocal interactions in

nightingales (Luscinia megarhynchos B.). Behavioral Ecology and Sociobiology, 11,

253–260.

Johnsson, J. I. & Akerman, A. 1998. Watch and learn: preview of the fighting ability of

opponents alters contest behaviour in the rainbow trout. Animal Behaviour, 56,

771–776.

Johnstone, R. A. 2001. Eavesdropping and animal conflict. Proceedings of the National

Academy of Sciences, USA, 98, 9177–9180.

Kiflawi, M. & Gray, D. A. 2000. Size-dependent response to conspecifics mating calls by

male crickets. Proceedings of the Royal Society of London, Series B, 267, 2157–2161.

Krebs, J. R., Ashcroft, R. & van Orsdol, K. 1981. Song matching in the great tit, Parus

major L. Animal Behaviour, 29, 918–923.

Lloyd, J. E. & Wing, S. R. 1983. Nocturnal aerial predation of fireflies by light-seeking

fireflies. Science, 222, 634–635.




McGregor, P. K. & Dabelsteen, T. 1996. Communication networks. In Ecology and

Evolution of Acoustic Communication in Birds, ed. D. E. Kroodsma & E. H. Miller.

Ithaca, NY: Cornell University Press, pp. 409–425.



McGregor, P. K., Dabelsteen, T. & Holland, J. 1997. Eavesdropping in a territorial

songbird communication network: preliminary results. Bioacoustics, 8, 253–254.

McGregor, P. K., Peake, T. M. & Lampe, H. M. 2001. Fighting fish Betta splendens extract

relative information from apparent interactions: what happens when what you

see is not what you get. Animal Behaviour, 62, 1059–1065.

Mennill, D. J., Ratcliffe, L. M. & Boag, P. T. 2002. Female eavesdropping on male song

contests in songbirds. Science, 296, 873.

Naguib, M. & Todt, D. 1997. Effects of dyadic interactions on other conspecifics

receivers in nightingales. Animal Behaviour, 54, 1535–1543.

Naguib, M., Fichtel, C. & Todt, D. 1999. Nightingales respond more strongly to vocal

leaders of simulated dyadic interactions. Proceedings of the Royal Society of London,

Series B, 266, 537–542.

Nowak, M. A. & Sigmund, K. 1998. Evolution of indirect reciprocity by image scoring.

Nature, 393, 573–577.


36 T. M. Peake

Oliveira, R. F., McGregor, P. K. & Latruffe, C. 1998. Know thine enemy: fighting fish

gather information from observing conspecifics interactions. Proceedings of the

Royal Society of London, Series B, 265, 1045–1049.

Oliveira, R. F., Rosenthal, G. G., Schlupp, I. et al. 2000. Considerations on the use of

video playbacks as visual stimuli: the Lisbon workshop consensus. Acta Ethologica,

3, 61–65.

Oliveira, R. F., Lopes, M., Carneiro, L. A. & Canario, A. V. M. 2001. Watching fights raises

fish hormone levels. Nature, 409, 475.

Otte, D. 1974. Effects and functions in the evolution of signaling systems. Annual


Otter, K. A., McGregor, P. K., Terry, A. M. R. et al. 1999. Do female great tits Parus major

assess extra-pair males by eavesdropping? A field study using interactive song

playback. Proceedings of the Royal Society of London, Series B, 266, 1305–1309.

Otter, K., Stewart, I. K., Terry, A. M., McGregor, P. K. & Burke, T. 2001. Extra-pair

paternity in great tits in relation to manipulation of male signals. Journal of Avian

Biology, 32, 338–344.

Peake, T. M., Terry, A. M. R., McGregor, P. K. & Dabelsteen, T. 2001. Male great tits

eavesdrop on simulated male-to-male vocal interactions. Proceedings of the Royal


Peake, T. M., Terry, A. M. R., Dabelsteen, T. & McGregor, P. K. 2002. Do great tits assess

rivals by combining direct experience with information gathered by

eavesdropping? Proceedings of the Royal Society of London, Series B, 269, 1925–1929.

Peake, T. M., Matessi, G., Dabelsteen, T. & McGregor, P. K. 2004. Song type matching,

song type switching and eavesdropping in male great tits. Animal Behaviour,

in press.

Pendry, L. 1998. When the mind is otherwise engaged: resource depletion and social

stereotyping. European Journal of Social Psychology, 28, 293–299.

Roberts, S. C., Gosling, L. M., Thornton, E. A. & McClung, J. 2001. Scent-marking by

male mice under the risk of predation. Behavioral Ecology, 12, 698–705.

Ryan, M. J., Tuttle, M. D. & Rand, A. S. 1982. Bat predation and sexual advertisment in a

neotropical anuran. American Naturalist, 119, 136–139.

Sakaluk, S. K. & Belwood, J. J. 1984. Gecko phonotaxis to cricket calling song: a case of

satellite predation. Animal Behaviour, 32, 659–662.

Seyfarth, R. M. & Cheney, D. L. 2002. What are big brains for? Proceedings of the National


Shennan, M. G. C., Waas, J. R. & Lavery, R. J. 1994. The warning signals of parental

convict cichlids are socially facilitated. Animal Behaviour, 47, 974–976.

Stoddard, P. K. 1999. Predation enhances complexity in the evolution of electric fish

signals. Nature, 400, 254–256.

Stowe, M. K., Turlings, T. C. J., Loughrin, J. H., Lewis, W. J. & Tumlinson, J. H. 1995. The

chemistry of eavesdropping, alarm, and deceit. Proceedings of the National Academy

of Sciences, USA, 92, 23–28.

Tobias, J. A. & Seddon, N. 2000. Territoriality as a paternity guard in the European

robin Erithacus rubecula. Animal Behaviour, 60, 165–173.



2002. Female begging in European robins: do neighbours eavesdrop for extra-pair

copulations? Behavioral Ecology, 5, 637–642.

Wiley, R. H. 1983. The evolution of communication. In Animal Behaviour, Vol. 2:

Communication, ed. P. J. B. Slater & T. R. Halliday. Oxford: Blackwell, pp. 156–189.

1994. Errors, exaggeration and deception in animal communication. In Behavioral

Mechanisms in Evolutionary Ecology, ed. L. Real. Chicago, IL: University of Chicago

Press, pp. 157–189.

Xitco, M. J., Jr & Roitblat, H. L. 1996. Object recognition through eavesdropping: passive

echolocation in bottlenose dolphins. Animal Learning and Behavior, 24, 355–365.

Zahavi, A. 1979. Why shouting? American Nauralist, 113, 155–156.

Zuk, M. & Kolluru, G. R. 1998. Exploitation of sexual signals by predators and

parasitoids. Quarterly Review of Biology, 73, 415–438.

P1: GFZ/... P2: JZZ/...0521823617c03.xml CU1917B/McGregor 0 521 582361 7 April 10, 2005 17:1

3

Public, private or anonymous?Facilitating and counteringeavesdropping

t o r b e n da b e l s t e e n


Introduction

Animals often live in environments where several conspecifics are within

signalling range of each other simultaneously. This is obvious for lekking species,

but it also applies to territorial species. In theory, this allows complicated patterns

of information flow between the individuals, which can be considered members

of a communication network (Dabelsteen, 1992; McGregor, 1993). A special case of

a network is when dyads of signalling individuals are temporarily or, in the case

of a sparse population, permanently isolated from other signallers or receivers.

Within a network, an individual may gather information about another individ-

ual from simple reception of its signals, but it may also be in a position that allows

simultaneous reception of the signals from two individuals engaged in a dyadic

signalling interaction. Such a position provides a special option for gathering rel-

ative information about the two interacting individuals, for example about their

state, strength or quality (McGregor & Dabelsteen, 1996). The relative information

results from how the two individuals use their signals in the interaction; the in-

formation is relative in the sense that it expresses the relative state or ‘value’ of

the two individuals without necessarily giving information about their absolute

states or values. The extraction of information from a signalling interaction is,

therefore, fundamentally different from simple receiving and so deserves its own

term, eavesdropping (McGregor & Dabelsteen, 1996) or, more specifically, social

eavesdropping (Ch. 2). In theory, relative information about different individu-

als could also be obtained by receiving the signals from each individual in turn,


38


Facilitating and countering eavesdropping 39

followed by a process in which the absolute information about different individ-

uals is compared.

Do animals eavesdrop?

There are observations of how natural signalling interactions between

two animals may cause conspecifics to approach the interactants and/or signal in

immediate or delayed response to the interaction or directly to interfere with the

interaction (e.g. McGregor & Dabelsteen, 1996). Such observations, whether anec-

dotal (e.g. Snow, 1958) or the result of planned studies (e.g. Bower, 2000), suggest

that animals eavesdrop. There is also experimental evidence that a conspecific

may interfere with a dyadic interaction by signalling predominantly towards the

apparent superior individual of the interaction, suggesting that eavesdropping

took place before interference (Naguib & Todt, 1997; Naguib et al., 1999; Ch. 14).

The ability to gather relative information by eavesdropping and utilize this in

later dyadic encounters with one of the previously interacting individuals has been

demonstrated experimentally in field studies with male birds (e.g. McGregor et al.,

1997), female birds (e.g. Otter et al., 1999; Mennill et al., 2002), captive male fish

(e.g. Oliveira et al., 1998; McGregor et al., 2001; Ch. 5) and captive female fish (e.g.

Doutrelant & McGregor, 2000). Some of the most convincing evidence comes from

field experiments with territorial male great tits, Parus major, that were allowed

to eavesdrop on male–male interactions simulated by means of playback from

two different loudspeakers. The experiments with simulated interactions, which

allowed the best possible control over the relative information made available to

the test subjects, demonstrated that males have the ability to extract relative in-

formation about rivals engaged in a song dual and utilize this in later encounters

with them (Peake et al., 2001). They also demonstrated the ability to combine such

information with the eavesdropper’s own previous direct experience with one of

interacting males (Peake et al., 2002). Given the potential advantages to eavesdrop-

pers of gaining such relative information (see below), this ability is likely to be

used also in non-experimental natural contexts, but we still lack firm evidence

that this really happens.

Potential gains and costs of eavesdropping

Gaining at least relative information about the state, quality or strength

of rivals or potential mates when absolute information is not available, or

when the capacity to compare such information is lacking, must constitute the

aim of any assessment process preceding decision making (e.g. in mate choice).

Eavesdropping was, therefore, predicted to be a widespread phenomenon in both

sexes and because it may constitute a low-cost and low-risk alternative to gathering


40 T. Dabelsteen

the same relative information through direct interactions with the individuals

eavesdropped upon (McGregor & Dabelsteen, 1996). For instance, a male that is

eavesdropping on male–male interactions probably uses less energy and runs a

lower risk of injury than it would through a direct interaction, which might es-

calate to actual fighting. Escalated interactions, whether hostile or collaborative

(as in courtship), may also increase predation risk because vigilance is reduced

(e.g. Dabelsteen & Pedersen, 1990; Jakobsson et al., 1995). As eavesdroppers must

divide their attention between interactants, they may have to reduce their vig-

ilance more than simple receivers, which can focus their attention on a single

signaller (e.g. Dukas & Kamil, 2001). It is difficult to identify other potential costs

that are specific to eavesdropping rather than costs that are common to any sort

of information gathering. Overall, the advantages of eavesdropping seem obvious

and predict the evolution of eavesdropping strategies that increase the possibility

of gaining relative information about the participants of an interaction.

Whereas the advantages of eavesdropping are clear, it is not necessarily advan-

tageous to give away information to eavesdroppers. For instance, an individual

that ends up losing a hostile interaction is unlikely to benefit from having its

loss advertised, whereas a winner would have a clear interest in advertising its

superiority (see also discussions in Chs. 2, 4 and 10). An individual should only

start an agonistic interaction if it has no prior knowledge that it will lose to its op-

ponent (e.g. Dabelsteen, 1985). Therefore, participants in an agonistic interaction

should not attempt to withhold information at the start of the interaction, but

do so later should the outcome become uncertain and the interaction escalate.

When the outcome is uncertain, both of the interactants would have an interest

in keeping the interaction private until the interaction has been settled. I know

of no experiments in which subjects were presented with an interaction that had

no difference in relative information, but I predict that a possible response of sub-

jects to such playback would be territorial intrusions and extra-pair behaviour in

relation to the interactants.

Courtship interactions may also be sensitive to eavesdropping. During early

stages of courtship, the two sexes may have conflicting interests. Males have an

interest in preventing rivals from discovering, and perhaps interfering with, their

courtship (Balsby & Dabelsteen, 2003a,b). Females may wish to attract more males

(e.g. Wiley & Poston, 1996), for instance to provoke an interaction between males

upon which they could eavesdrop. When the female has chosen a mate, both the fe-

male and the chosen male have an interest in preventing rivals from eavesdropping

on the intensive courtship interaction that often precedes copulation, because

information that copulation is imminent could lead to attempts to prevent or

even interrupt copulation by rivals. Intrusion by neighbours and subsequent



interference with courtship (including interruption of copulation) have been ob-

served in a number of different species, for example the robin Erithacus rubecula

(Lack, 1940), the dunnock Prunella modularis (Davies, 1992) and the blackbird Turdus

merula (Snow, 1958). As Snow (1958, p. 86) wrote:

The sight of a pair copulating or about to copulate has an immediate

and powerful effect on neighbouring males. In nearly every case that I

observed, copulations were interfered with by the sudden arrival of one

or two males, who either knocked the copulating male off the female or

prevented him from mounting. And these attacks have been directed

against a territory-holder in the middle of his own territory, where the

neighbours normally never go or, if they do, only with every sign of

nervousness.

Snow (1958) also noticed that courtship of an impassive blackbird female did not

result in such interference. Only when the female responded with copulation

solicitation behaviour (i.e. when there was a real courtship interaction) did the

rival males intrude.

In the whitethroat Sylvia communis, the presence of a female in a territory leads

to significantly more intrusions from neighbouring males than when no female

is present, and territory owners always respond by chasing intruders out of their

territories even when they have to interrupt their courtship of the female (Balsby &

Dabelsteen, 2003a,b). A recent experiment suggested that it is the courtship inter-

actions that make neighbours intrude. The experiment compared the intrusion

rate of male subjects that could eavesdrop on either their neighbours interacting

with a loudspeaker playing normal full whitethroat song (song duel treatment)

or courtship interactions between their neighbours and Jumping Sylvia (Balsby &

Dabelsteen, 2002), a remotely controlled stuffed female that could jump and vocal-

ize (courtship treatment). Whereas the song duel treatment never elicited intru-

sions into the neighbour’s territory by the male subjects, the courtship treatment

did so in 56% of trials, and 44% of intrusions led to direct interference with the

neighbour’s courtship of, or copulation with, Jumping Sylvia (Balsby & Dabelsteen,

2004).

The different potential gains and costs of eavesdropping to interactants in

agonistic and courtship contexts predict the evolution of strategies that counter,

or reduce, the negative consequences and strategies that ignore, or even facilitate,

it. In the rest of this chapter, I discuss how communication behaviour can be made

public, private or anonymous and how eavesdroppers should behave. The focus is

on vocal interactions of territorial songbirds.


42 T. Dabelsteen

How best to eavesdrop

An eavesdropper should attempt to achieve the best conditions for receiv-

ing the signals of both interactants simultaneously and at the same time reduce

the potential costs of missing more important interactions or being detected by

predators. Since eavesdropping always involves increased risk of predation be-

cause of divided attention, the duration of an eavesdropping session should be

restricted to the time necessary for the ‘intended’ gathering of information. In

addition, a male eavesdropper should attempt to stay undetected by the interac-

tants because at least one (e.g. the loser) will always have an interest in concealing

the interaction irrespective of its nature, i.e. whether it is agonistic or sexual. A

detected female eavesdropper, by comparison, may benefit from the intensified

interaction between male interactants. Such intensification of males’displays and

fights is often observed in lekking species when females pass or arrive on a lek

(e.g. Lack, 1939; Hovi et al., 1995).

Predictions on how best to eavesdrop

Simply approaching the interactants or moving to a position where the

signals from both individuals can be received simultaneously and equally well will,

of course, enhance eavesdropping. When the information gathered from an inter-

action depends on the timing of airborne sound signals from two interactants,

eavesdropping may be complicated by the relatively slow speed of sound trans-

mission in air. In such cases, eavesdroppers should approach to positions where

the distances to the two interactants are equal (e.g. Dabelsteen, 1992). Like simple

receivers, eavesdroppers may also improve the conditions for receiving sound sig-

nals by ascending to a high perch. Depending on the nature of the surrounding

vegetation, this may improve the possibilities for observing visual displays. The

evidence for the improvement of sound reception comes from sound transmission

experiments that quantified the degradation of natural sound signals transmit-

ted over natural communication distances using natural signaller and receiver

positions in the appropriate habitat for the study species (Box 3.1).

Sound transmission experiments indicate that simply leaving the ground and

flying a few metres up to the undergrowth of a forest will not necessarily im-

prove the receiving conditions if the signaller is already located higher up in the

vegetation. This is perhaps not surprising given the ‘ground’effect (attenuation, es-

pecially of low-frequency sounds, when sounds are transmitted along the ground)

when both the signaller and the receiver are close to the ground (e.g. Wempen,

1986; Embleton, 1996; Nemeth et al., 2001). However, ascent by the receiver to

perches above the undergrowth may improve receiving conditions considerably

depending on the type of habitat. For instance, a whitethroat that moves from



4 to 9 m above ground level in an open whitethroat habitat will not improve

sound receiving conditions further (Balsby & Dabelsteen, 2003c). In whitethroats,

high perches mainly seem to help visual surveying of the surroundings. There is

a very different result in a closed forest habitat before leaf burst. Experiments

with songs of three different species, the blackbird (Dabelsteen et al., 1993), the

wren Troglodytes troglodytes (Holland et al., 1998, 2001) and the blackcap Sylvia atri-

capilla (Schmitz et al., 2000; Mathevon et al., 2004), show that receivers may obtain a

considerable improvement by ascending to high perches. For instance, a blackcap

receiver that moves from 4 to 9 m above ground level obtains improvements that

would correspond to a horizontal approach towards the singer of up to 23 m, i.e.

almost half an average territory diameter (Fig. 3.1 and Box 3.1) (Schmitz et al., 2000;

Mathevon et al., 2004).

Box 3.1

Transmission-caused sound degradation has at least four aspects. (a) Sound

signals are attenuated because of spherical spreading (6 dB per doubling of

distance) and excess attenuation (EA) caused by absorption and multiple

scattering (e.g. Michelsen, 1978). (b) This attenuation will, together with the

addition of background noise, reduce the signal-to-noise ratio (SNR).

(c) Selective frequency filtering, atmospheric turbulence and reverberation

will result in a distortion or blurring within the sounds of their frequency

and amplitude patterns over time (e.g. Wiley & Richards, 1982), which can be

quantified by a blur ratio (BR; Dabelsteen et al., 1993). (d) Reverberation will

cause an elongation of the sounds with tails of echoes, which can be

quantified by a tail-to-signal ratio (TSR; Holland et al., 2001). Sound

transmission experiments indicate that all these aspects of sound

degradation change with distance: EA, BR and TSR increase and SNR

decreases with distance. The experiments also indicate that sound

degradation decreases with increasing height above ground level of signaller

(loudspeaker) and receiver (microphone). This means that both signaller and

receiver may improve communication by sound signals by moving up to

high perches. Exactly how large these improvements are is perhaps best

understood when the regressions of the values of each of the four

degradation measures against the logarithm of the distance are used to

translate the improvements obtained by moving upwards into virtual

horizontal distances that a signaller or receiver would have to approach a

receiver or signaller, respectively, to obtain the same improvements.

Figure 3.1 shows an example.


44 T. Dabelsteen

ReceiverTSR EABRSNR

Sender

BR TSREASNR

Fig. 3.1. Virtual horizontal distances (see text) that a blackcap receiver or signaller

would have to approach a signaller or a receiver, respectively, to obtain

improvements similar to those obtained by moving in the vegetation from 4 to 9 m

above ground level. In this example, the distance between signaller and receiver is

50 m and the calculations are based on average degradation values for 10 different

blackcap song elements. For a signaller, the average improvements for each of four

degradation measures (indicated by silhouettes of flying birds for SNR, BR, EA and

TSR; defined in Box 3.1 text) correspond to virtual approaches only slightly longer

than the 5 m ascent. For a receiver, the virtual approaches are considerably longer,

corresponding to almost half the average territory diameter of 25 m, indicated by

the horizontal line along which the bird silhouettes are flying. Movements and

vegetation are to scale, whereas bird silhouettes are enlarged. (Modified from

Schmitz et al., 2000.)

These sound transmission experiments predicted that eavesdroppers on vocal

interactions should ascend to high perches in a forest habitat before leaf burst. By

ascending to high perches instead of approaching interactants, an eavesdropper

would save energy and also avoid moving too far away from locations where future

interactions of potential interest might take place. By staying somewhere in the

middle of its territory during periods with high-singing activity, for instance at

dawn or during most of the morning, a perching bird would also potentially be

able to monitor a number of simultaneously occurring interactions and switch its

attention between different interactions depending on where the most interest-

ing developments happen. Staying at some distance from the interactants would



also help to conceal eavesdropping activity. Exactly where and how high above

ground level an eavesdropper should position itself will, of course, depend on the

local vegetation and the availability of cover to avoid detection by predators or

interactants.

Female eavesdroppers have the same interest as males with respect to predators,

but not necessarily with respect to the interactants (see above). If the predation risk

constrains them to stay hidden in the vegetation, they could, in theory, announce

their presence vocally.

For how long should individuals eavesdrop? Eavesdropping should be as brief

as possible to reduce the risk of predation caused by divided attention, but suf-

ficiently long to allow the extraction of relative information about the interac-

tants. What exactly ‘sufficient’ means seems to depend heavily on the context.

For instance, intensive courtship interactions will almost immediately inform an

eavesdropper that both members of a pair are ready to copulate, whereas agonistic

interactions between males may sometimes progress very slowly and the signalling

may only reveal the superior male after some time has passed. In such cases, we

cannot predict the duration of eavesdropping because this will be controlled by

the interactants.

In some species, song repertoire size is correlated with morphometric measures

of males, suggesting that repertoire size is capable of providing absolute informa-

tion about male quality. In such species, relative information about the quality of

different individuals could be deduced by receiving the songs from individuals in

turn. However, in some of these species, studies have failed to show that repertoire

size has signal value. In experimental studies where the design has only allowed

simple receiving, the negative results could reflect the fact that the species do not

have the capacity for determining absolute repertoire size. It would be interesting

to investigate if such species eavesdrop on song duels to gather relative informa-

tion about repertoire sizes. If so, the minimum time needed to obtain the relative

information by eavesdropping could be predicted from cumulative plots of the

number of new songs sung as a function of the total number of songs sung by

each of the interactants. The time taken to get a reliable indication of the relative

repertoire sizes of two interactants would, of course, depend on who is interacting

with whom since the cumulative plots of some pairs of individuals become dif-

ferent sooner than other pairs. Male whitethroats, for instance, apparently do not

vary their responses to playback of different repertoire sizes (Balsby & Dabelsteen,

2001): maybe because they do not consider repertoire size in agonistic contexts;

maybe because they do not have the capacity for determination of absolute reper-

toire size. If they could perceive relative repertoire sizes, such males would have

to eavesdrop for between two minutes (on an interaction between males A and B


0

100

200

300

400

500

600

700

040

00

8000

1200

016

000

Tot

al n

umbe

r of

ele

men

ts

A

B

C

Number of different elements

Fig.

3.2.

Cu

mu

lati

vep

lots

ofn

ewso

ng

elem

ents

fou

nd

asa

fun

ctio

nof

son

gel

emen

tssu

ng

by18

dif

fere

nt

wh

itet

hro

atm

ales

.In

10m

inu

tes,

app

roxi

mat

ely

1000

elem

ents

wer

esu

ng.

The

lett

ers

refe

rto

dif

fere

nt

mal

esd

iscu

ssed

inth

ete

xt.(

Mod

ifie

dfr

om

Bal

sby,

2000

.)

46



in Fig. 3.2) and more than 20 minutes (when males B and C are interacting) to

detect which male has the largest repertoire (Balsby, 2000).

Do eavesdroppers fit the predictions?

The experiments that sought to establish eavesdropping behaviour in

songbirds did not monitor the locations of subjects in a way that allows the above

predictions to be tested. However, there is evidence (personal observation) that

male great tits may approach and/or be perched during experimentally induced op-

portunities for eavesdropping. There are also observations of neighbouring males

approaching during interactive playback to territorial blackbirds, and male black-

birds may interrupt ongoing behaviour and fly to high perches and stay silent in

response to other males starting to sing (personal observation). Although anecdo-

tal, such observations support the idea that eavesdroppers may attempt to achieve

the best conditions for receiving the signals of interactants. It is also possible that

some of the variation in the results obtained in the field experiments with great

tits (Otter et al., 1999; Peake et al., 2001, 2002) was due to the ‘uncontrollable’

eavesdroppers (i.e. subjects) having been in positions that varied in how well the

interactants could be heard.

Evidence suggesting that birds probably engaging in eavesdropping behave in

a way that would best receive signals of interactants comes from a radio-tracking

study of 11 unmated female great reed warblers Acrocephalus arundinaceus (Bensch

& Hasselquist, 1992). Their routes were taken as evidence for female assessment

preceding mate choice (Bensch & Hasselquist, 1992; but see Ch. 7). Figure 3.3 shows

the route followed by one such female. During the eight hours the female was fol-

lowed, it had four relatively long stays within a limited area, indicated by clusters

of black dots in Fig. 3.3. The first cluster indicates a stay of 40 minutes at the

location of the female early in the morning before it starts to move. The other

three clusters, each corresponding to stays of about 60 minutes’ duration, are at

locations with almost exactly equal, relatively short, distances to two males at a

time: perfect positions for eavesdropping on vocal interactions between males.

One cluster would allow eavesdropping on interactions between males VII and III,

the next on males III and II, and finally on males V and VI. These results suggest

that about 60 minutes was needed to gather information from two males at a time,

including relative information from the occasional singing interactions between

the males.

It has been hypothesized that females of some bird species use loud fertility

advertisement calls to incite male–male interactions (e.g. Montgomerie & Thorn-

hill, 1989; Hoi, 1997). Although the actual fertility advertisement function of the

loud female calls is doubtful, it is possible that such calls may initiate or even

intensify ongoing male–male interactions and hence act as an aid in mate choice


48 T. Dabelsteen

Fig. 3.3. Map showing the movements of a female great reed warbler in an area with

nine male territories, the positions of which are marked with bird silhouettes. The

female, which was equipped with a radio transmitter and followed for eight hours,

had its position monitored every 10 minutes, as indicated by black dots. Four clusters

each consisting of four to six dots have been encircled. The one to the left marks the

position of the female when she was first tracked at 04:50 hours. The three clusters to

the right mark positions from which the female may have been eavesdropping on

singing interactions between dyads of males.

(e.g. Sæther, 2002; Ch. 7). A female that utters such calls to intensify an ongoing

interaction would be fitting one of the above predictions for how best to eavesdrop.

Public signalling: facilitating eavesdropping

The term advertising is normally used for signalling that makes the sig-

naller advertise itself to a wide audience with respect to some quality or capacity.

Advertising signals transmit over relatively long distances and are used in solo



signalling as well as in signalling interactions. A good example of such an ad-

vertising signal is the full song of songbirds, which often has a dual function:

to attract potential mates and repel rivals. Interactants using advertising signals

will, of course, expose themselves to eavesdropping; however, if such eavesdrop-

ping has no immediate or subsequent adverse consequences, then it should not

change the signalling behaviour of interactants. Yet there could also be situations

where one of the interactants might benefit from the presence of eavesdroppers,

for instance by making its superiority relative to the opponent widely known.

Here the superior individual should continue signalling or even attempt to make

it more effective, whereas the inferior individual should stop using advertising sig-

nals and ultimately stop interacting. It could be argued that an individual should

always make advertising signalling as effective as possible. However, when effec-

tiveness depends on energy used or some other cost, animals may wish to limit

the costs during solo signalling more than during the usually shorter signalling

interactions, perhaps to avoid exhaustion.

Predictions for public signalling

Signallers could allow eavesdropping simply by choosing signals that

transmit effectively in their physical environment and by signalling from posi-

tions and at times of the day that would maximize signal transmission. The ev-

idence for such choices in use of sound signals comes from sound-transmission

experiments. All of the experiments with bird song mentioned above show that

some sound types transmit better than others, even among functionally equivalent

types. For instance, low-frequency, narrowband and unmodulated sounds seem to

transmit best in a forest habitat. In the full song of the blackbird (Fig. 3.4a), the

introductory low-frequency whistle or motif sounds transmit much better than the

terminating highly modulated and broadband twitter sounds; among the whistle

sounds, the relatively unmodulated and constant frequency ‘CF-sounds’ transmit

better than sounds that are frequency modulated (‘FM-sounds’)and/or have energy

rich overtones (‘MIX-sounds’) (Dabelsteen et al., 1993). Also in great tit song, rela-

tively ‘pure-toned’sound elements seem to transmit better than highly modulated

‘buzz’ elements (Blumenrath et al., 2004). In blackcap song the terminating rela-

tively low frequency, pure tone and narrowband motif sounds are less attenuated

than the introductory highly modulated broadband twitter sounds (Dabelsteen &

Mathevon, 2002). In wren song, low frequency sounds are least attenuated

(Holland et al., 1998), and in antbird (Thamnophilidae) song, the narrowband and

low-frequency sounds transmit best (Nemeth et al., 2001).

Low-frequency sounds also have another property that make them suited for

advertising: they usually radiate from the vocalizing individual more or less

equally well in all directions (i.e. they are omnidirectional). This has been shown


50 T. Dabelsteen

0.5 1.0 1.5 2.0 2.5 3.0

2

4

6

8

10

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

2

4

6

8

10kHz

s

s

(a)

(b)

kHz

Fig. 3.4. Sound spectrograms of blackbird song. (a) A male full song recorded in the

field; the introductory whistle part is underlined. (b) A male full song, which is

framed, has the highest frequencies of the terminating twitter part overlapped by a

female copulation trill, the duration of which is indicated by a horizontal line. This

was recorded in an anechoic chamber and shows the part of a playback trial where

playback of full song to an oestradiol-treated female elicits a copulation solicitation

posture and a copulation trill (see Dabelsteen, 1988). The movements associated with

the female posturing are indicated by broadband white noise, especially over the

whistle part of the male’s full song but also over the last third of the copulation trill.

Spectrograms were produced in Avisoft (FFT 512 points, flat top, overlap 75%,

frequency resolution 43 Hz, time resolution 5.805 ms).

in laboratory studies with blackbirds (Larsen & Dabelsteen, 1990). Species using

sounds with a more directional radiation pattern may facilitate advertising by

moving their head from side to side during singing (Brumm & Todt, 2003).

Sound transmission is normally believed to be most effective from high perches.

The sound transmission experiments with bird song also indicate that a signaller

should at least ascend to a few metres above ground level to make sound transmis-

sion effective (e.g. Nemeth et al., 2001); transmission may sometimes be improved

further by ascending to high perches, but only slightly. In a forest, the improve-

ment that a signaller obtains by moving to higher perches sometimes seems very

small relative to that a receiver might obtain by the same movement (e.g. Fig. 3.1);

consequently, high perches should perhaps be called listening posts rather than

song posts in these cases (Schmitz et al., 2000; Mathevon et al., 2004). Sound trans-

mission is also believed to be most effective at dawn (e.g. Henwood & Fabric, 1979).



However, very few studies have investigated this with sound transmission experi-

ments using natural sound signals. A recent study with blackcap song, at the peak

season of their singing immediately after their return to the breeding sites in

the spring, failed to show that their songs propagate most effectively at dawn

(Dabelsteen & Mathevon, 2002).

Overall, the sound transmission experiments suggest that the most widely prop-

agating sound signals would be relatively loud, low frequency, narrowband and

unmodulated; they would be emitted from at least a few metres above the ground.

The effect of higher song perches is small and the optimal time of day is uncer-

tain and probably depends strongly on the weather. Windy and rainy conditions

will constrain advertising for different reasons (e.g. Lengagne & Slater, 2002), al-

though birds, in theory, might compensate for a high level of background noise

by increasing the output level of songs (Lombard effect; Lombard, 1911). For in-

stance, nightingales Luscinia megarhynchos may use a higher song output level in

noisy environments than at less-noisy locations (Brumm, 2004).

Do interactants use public signals to facilitate eavesdropping?

Observations suggest that birds sometimes increase the loudness of their

singing when they shift from solo singing to counter-singing (i.e. interacting).

Unfortunately, such observations have rarely been verified with sound pressure

level (SPL) measurements. An exception is a study by Brumm & Todt (2002), which

showed that male nightingales singing full songs increase the SPL by more than

5 dB when they shifted from solo singing to playback-induced counter-singing.

However, it is a question of whether the increase in SPL has evolved to facilitate

eavesdropping or whether it represents an increased arousal of the singers, which

as a side effect inevitably facilitates eavesdropping on the vocal interaction. Play-

back experiments support the arousal hypothesis because a higher SPL elicits a

stronger response (e.g. Dabelsteen & Pedersen, 1992).

There is, as yet, no evidence that interactants facilitate eavesdropping dur-

ing interactions by using signals from their repertoire of functionally equivalent

advertising signals that transmit more effectively than those used during solo

signalling. For song types, this would imply that blackbirds engaged in a song

duel across the border between their territories should use more CF-sounds and

fewer FM- and MIX-sounds than during solo singing. However, the opposite seems

to happen, perhaps because the sound types are not functionally equivalent, with

the FM-sounds expressing the highest degree of arousal (Dabelsteen & Pedersen,

1992). Neither is there any evidence that higher perches are used during interac-

tive singing than during solo singing. In both contexts, the use of relatively high

perches is probably mainly to improve sound reception, either of the songs of the

opponent or of vocal responses to the solo songs.


52 T. Dabelsteen

When singing interactions using advertising signals escalate, the interactants

often shift to special types of close range, private singing (see below). However,

after agonistic encounters, one or both of them may shift back to loud advertising

songs. When both individuals do this and continue to interact, the shift back to

advertising songs can be said to facilitate eavesdropping on their interaction. The

advertising singing of the eventual winner of the escalated interaction is often

different from that of the loser and, therefore, is referred to as an ‘acoustic victory

display’ (e.g. Bradbury & Vehrencamp, 1998; Ch. 6). When only one of the birds

shifts back to advertising songs, it is usually the winner of the interaction. In this

case, the winner’s singing cannot be said to facilitate eavesdropping since there is

only one signaller left.

Private signalling: countering eavesdropping

An interactant should use private signals in an interaction whenever pub-

lic signals would incur potential risks. This is true when the risks are immediate,

for example from predators because of reduced vigilance during interactions com-

pared with solo signalling, or from male eavesdroppers, which may take advan-

tage of an interactant’s involvement with a rival to pay its mate a visit during the

interaction. It is also true when the risks are less immediate, for example eaves-

droppers that can extract information about an interactant and utilize this in

future encounters with it. Whatever the risks, they seem likely to increase with

the duration of the interaction. An important step to make an interaction private

would, therefore, be to make it as brief as possible, because this would reduce the

risk of detecting the signalling interaction and of obtaining useful information.

However, as explained in the next section, there are other options.

Predictions for private signalling

Anything that limits signal transmission and reception may, of course,

help to make signal interactions private, in effect the opposite of advertising.

Private signalling should, therefore, employ signals, signaller positions and sig-

nalling times that are the opposite of those used for advertising. Vocally interact-

ing birds that wish to signal privately should use sound signals that are relatively

high frequency, broadband and highly modulated, and emit them with a low SPL.

This would reduce the number of potential receivers for two reasons: such sounds

attenuate and degrade relatively fast with distance (e.g. Dabelsteen et al., 1993;

Holland et al., 1998) and are directional in the sense of being beamed away from

the sender in one direction (e.g. Larsen & Dabelsteen, 1990). Sound transmission

experiments also show that privatizing would be most effective if the interactions

take place close to the ground or in the undergrowth of a forest.



Background noise, constant as well as transient, may mask communication

sounds and hence contribute to private signalling. This may be especially impor-

tant when eavesdropping involves information gathered from the timing of songs.

For instance, the degree to which songbirds delay their songs relative to each other

or overlap each other’ssongs seem to be important indicators of social dominance

(e.g. Dabelsteen et al., 1997, 1998; Langemann et al., 2000) that are utilized by

eavesdroppers (e.g. McGregor et al., 1997; Peake et al., 2001). Receivers in general

may have internal representations of songs which would help them to reconstruct

songs which are partially masked by transient background noise, for instance the

vocalizations of other birds. However, such representations would probably not

be of much help to eavesdroppers in reconstructing how interactants delay their

songs relative to each other or overlap each other’s songs (e.g. Poesel et al., 2001).

Therefore, private signalling should take place at times of the day when the level

of background noise is high.

Do interactants use private signals to counter eavesdropping?

So-called quiet song in songbirds, sometimes referred to as soft or whis-

per song, seems to be a good candidate for a private signal (e.g. Dabelsteen et al.,

1998). Quiet singing is still relatively unexplored, but wherever it has been discov-

ered it has been accompanied by an active behaviour rather than perching, for

instance posturing and/or escorting or chasing another individual, and it is often

sung more or less continuously without the intersong pauses characteristic of full

singing. Quiet singing usually occurs at close range during escalated interactions,

collaborative as well as competitive, i.e. in contexts of eavesdropping (e.g. Titus,

1998; Balsby, 2000; Balsby & Dabelsteen, 2003a,b).

Good examples of quiet singing occur in the blackbird, the redwing Turdus ili-

acus, the robin, the dunnock and the alpine accentor Prunella collaris (Dabelsteen

et al., 1998). The blackbird has at least three types of quiet singing. Counter-singing

males switch from loud full songs with whistle sounds to quiet twittering without

whistles when the interaction escalates and the interactants approach each other

to within about 10 m. A male also twitters quietly during intense courtship of a

soliciting female immediately before copulation, and the soliciting female sings

a very quiet copulation trill (Figs. 3.4b and 3.5). Until recently, the aggressive and

sexual twitter were believed to be identical because that is how they sound to the

human observer (e.g. Dabelsteen et al., 1998). However, recent spectrographic anal-

yses have suggested that sexual twitter may also include specific sounds (Fleron,

2003) and hence have the potential to communicate the sexual arousal of the

singer in the same way as the accompanying posturing of the male (e.g. Snow,

1958).


54 T. Dabelsteen

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

2

4

6

8

10kHz

5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5

2

4

6

8

10kHz

s

Fig. 3.5. Sound spectrogram of the quiet vocalizations uttered during 10 seconds of a

natural blackbird courtship interaction early in the morning soon after the dawn

chorus. The courtship vocalizations dominate among the sounds that are visible on

the spectrogram above 4 kHz, most of them being male sexual twitter. A horizontal

line indicates a trill that is likely to be the female copulation trill. Two full songs of a

robin that is singing close by are indicated by dashed horizontal lines: the first has

visible energy up to about 5 kHz; the second starts with a high-frequency part

(approximately 8 kHz) and ends with a more low-frequency part (approximately

4 kHz). Most of the sounds between 1.8 and 3 kHz are whistle parts of more distant

singing blackbird males. Spectrograms were produced as in Fig. 3.4.

Quiet singing in the blackbird seems to fit nearly all of the predictions for a

private signal, with respect to sound type, variability and SPL, and with respect to

where, when and for how long it is used. Relative to the whistle sounds of full song,

all three types of quiet song consist of relatively high-frequency, broadband and

modulated sounds (Figs. 3.4, and 3.5). Twitters are at least 10 dB(A) quieter than

whistles (Dabelsteen, 1984) and sound transmission experiments have shown that

the twitters degrade and attenuate much faster than whistles (Dabelsteen et al.,

1993). Whereas whistles seem capable of transmitting over at least two to four

territory diameters, twitters usually transmit less than one diameter (Dabelsteen

et al., 1993). Twitters are also more directional than whistles (Larsen & Dabelsteen,

1990). Female copulation trills sound even quieter than male twitters (personal

observations) and probably have a radiation pattern and transmission capability



resembling that of twitters. Both aggressive twitter interactions and courtship

interactions are most frequently performed under cover close to the ground or in

the undergrowth (personal observation) and early in the morning after the males’

dawn chorus at a time where the level of background noise from the vocalizations

of other species is very high (Fig. 3.5) (e.g. Messmer & Messmer, 1956). Finally,

the duration of aggressive twitter interactions can vary quite a lot, but courtship

interactions are always very brief (e.g. Snow, 1958).

The term quiet singing should not be taken too literally since some of the bird

vocalizations referred to as calls seem to be used in the same contexts and to

have the same acoustic structure as quiet singing. Good examples are the dscharp-

and ze-calls of the female whitethroat, which are used in courtship interactions

where the male performs a special diving song display (Balsby & Dabelsteen, 2002,

2003a,b). The two calls fit the requirements for private signals: they have sound

structures and low SPL (measured for dscharp-call, too low to be measured for

ze-call), which make them short-range signals, and they are uttered from low

positions in the vegetation (Balsby & Dabelsteen, 2003a,b).

The eavesdropping contexts with quiet vocalizing are characterized by risks to

the interactants from predation and eavesdropping. This probably applies to all

eavesdropping contexts, and both these risks seem strong enough to cause the

evolution of private signals. It will, therefore, be very difficulty to disentangle the

relative influence of the two risks in the evolution of quiet singing and calling.

Quiet vocalizing might also have evolved simply to save energy during close-range

communication. However, this seems unlikely given that quiet vocalizations are

always accompanied by movements and sometimes even by posturing. This, to-

gether with the continuous nature of quiet vocalizing, suggests that birds do

not necessarily save energy by switching from loud advertising singing to quiet

singing. If energy saving was the main purpose, the bird could simply lower the

output level of singing as in the non-social subsong (e.g. Thorpe & Pilcher, 1959).

They do not have to switch to another song type and otherwise behave in a way

that would counter eavesdropping. However, there are contexts where predation

risk may be the main factor responsible for private signals, but these contexts

do not, as far as I know, involve a signalling interaction and, therefore, do not

constitute eavesdropping in the sense used in this chapter. A good example is

the so-called nest-relief song or calling-out song produced by males of many song-

bird species (e.g. Gompertz, 1961; Stork, 1971; Ficken et al., 1978; Lind et al., 1996).

Structurally, these songs seem identical to the full songs used during advertis-

ing singing, but they are usually sung with a very low SPL (although higher than

in subsong; personal observations) and they are also relatively short or few in

number.


56 T. Dabelsteen

Anonymity: another way to counter eavesdropping

Private signals counter eavesdropping but do not necessarily prevent it.

However, if interactants can remain anonymous, eavesdroppers cannot attribute

information gained from the interaction to any particular individual and, there-

fore, this cost of eavesdropping to interactants is removed. For instance, eaves-

droppers on an interaction between unknown individuals could not make use of

information in subsequent encounters with them.

Predictions for anonymity

Anything that makes individual identification of a signaller based on its

signalling activity more difficult would help anonymity. One way to constrain

individual recognition of vocalizing birds might be to increase the variability of

their vocal output. For instance, experiments suggest that song repertoires of fewer

than about 25 song types do not interfere with song-based neighbour recognition,

whereas repertoires of more than about 100 do (e.g. Stoddard, 1996; Molles &

Vehrencamp, 2001). Birds that want to make their singing anonymous should,

therefore, change the way they use their repertoire or somehow increase it during

interactions, for instance by switching to another type of singing. If a male can

be identified by the repertoire being sung from particular song posts, switching

to unusual posts combined with frequent post shifts might also help to achieve

anonymity.

Do animals make themselves anonymous in eavesdropping contexts?

Quiet singing in songbirds may aid anonymity because, unlike full song,

it is used almost everywhere and not from preferred posts, and most importantly it

seems a lot more variable than full song (e.g. Dabelsteen et al., 1998). For instance,

when a rival intrudes, a male blackbird uses the aggressive twitter everywhere in

its territory rather than from its usual song posts. Also its repertoire of different

aggressive twitter motifs far exceeds the threshold value (> 100; Fleron, 2003)

that constrains individual identification (e.g. Stoddard, 1996). The repertoire size

of full-song whistles averages 44 (Rasmussen & Dabelsteen, 2001).

The idea of anonymity in signals is novel and, therefore, not really studied yet,

but it seems clear that, unlike private signals, anonymity is difficult to explain

as a response to predation risk. Unlike the aspects of private signals mentioned

above, which can reduce the risk of detection by predators, increased variation

per se does not seem to have such an effect. Relative to full song, quiet song

endowed with a variation similar to that of full song would still be very difficult

for a predator to detect. The evolution of the relatively larger variation in quiet

singing could have resulted from sexual selection, which is believed to play a role



in the evolution of avian repertoire sizes (e.g. Searcy & Yasukawa, 1996). However,

this would require sexual selection to have acted more strongly on quiet song

than on full song, even though both types of song seem to play important roles in

mate choice and deterrence of rivals. Finally, the relatively large variation of quiet

song could be a side effect of a sound production mechanism, coupling low SPL

or high frequencies with large variation. However, this also seems unlikely given

that full song is sometimes uttered in non-social contexts as very quiet so-called

subsong (Thorpe & Pilcher, 1959) and that high-frequency quiet singing contains

the same type of fixed combinations of sound elements as full song (e.g. Rasmussen

& Dabelsteen, 2001; Fleron, 2003). At the moment, it seems likely that the large

variation of quiet singing relative to full singing in some species reflects the need

for singer anonymity to counter negative consequences of eavesdropping.

Summary

There is now experimental evidence that animals have the ability to gather

relative information about interactants by eavesdropping and utilize this informa-

tion in subsequent decision making about how to behave towards the interactants.

We still lack good observational evidence that this happens in non-experimental

natural contexts, but this seems likely given the obvious advantages to the eaves-

dropper. It is also likely that eavesdroppers behave in ways that enhance their

ability to eavesdrop. The potential gains of being eavesdropped upon are more dif-

ficult to identify but may exist in special situations and, therefore, have led to the

facilitation of eavesdropping, including enhancing advertising signals. The poten-

tial costs of being eavesdropped upon are much more obvious and set the scene for

an evolutionary arms race between eavesdroppers and interactants, with private

signals and anonymity reducing the costs of being subjected to eavesdropping.

In this chapter, I have concentrated on vocal interactions between songbirds

and made predictions for what eavesdroppers and interactants should do in terms

of their relative positioning and the type of sound signals that interactants should

use. Some of the predictions may seem trivial or speculative, while others are

more substantial because they are derived from the results of sound-transmission

experiments. At the moment, there is anecdotal evidence for all of the predictions

except aspects of public signals, where there is stronger evidence. There is also

strong evidence for private signals. Quiet singing in songbirds fulfils most of the

predictions for private signalling and seems to do so for anonymity as well. Other

selection pressures are also important; for instance predation risk is likely to have

played an important role in the evolution of private signals.

An interesting challenge for future research on these matters will be to investi-

gate eavesdropping in natural non-experimental contexts. In territorial songbirds,


58 T. Dabelsteen

this could be done by quantifying how territorial subject males behave during vo-

cal interactions between other males and in subsequent vocal interactions with

them. Such a study would need recordings of the vocal activity in a local network

of males using acoustic location systems, as done by, for instance, Bower (2000)

and Burt & Vehrencamp (Ch. 15), combined with monitoring the movements of

silent individuals by radio-tracking. If the study located birds in three dimensions,

it might also test the predictions on positioning and signal use discussed above. If

the study was long term and included collection of data on predation, survival and

reproductive success, it might help to disentangle the influence of the different

selection pressures on the evolution of communicating in a network.

Acknowledgements

Thorsten Balsby kindly commented on the manuscript. Sandra Blumenrath drew Fig. 3.1

and Thorsten Balsby assisted with the production of the remaining figures. Henrik Brumm gave

me access to unpublished manuscripts. The main part of my own research on communication

network activities forming the basis for this chapter has been funded by the Danish National

Science Foundation.

References

Balsby, T. J. S. 2000. Song activity and variability in relation to male quality and female

choice in whitethroats Sylvia communis. Journal of Avian Biology, 31, 56–62.

Balsby, T. J. S. & Dabelsteen, T. 2001. The meaning of song repertoire size and song

length to male whitethroats Sylvia communis. Behavioural Processes, 56, 75–84.

2002. Female behaviour affects male courtship in whitethroats, Sylvia communis. An

interactive experiment using visual and acoustic cues. Animal Behaviour, 63,

251–257.

2003a. Male singing behaviour and female presence in the territory in whitethroats.

Acta Ethologica, 5, 81–88.

2003b. Male calling between courtship sequences in whitethroats: a way to counter

intrusions from neighbouring rivals. Behavioural Processes, 63, 149–157.

2003c. Degradation of whitethroat Sylvia communis song: implications for song flight

and communication network activities. Behaviour, 140, 695–719.

2004. Simulated courtship interactions elicit neighbour intrusions in the

whitethroat Sylvia communis. Animal Behaviour, in press.

Bensch, S. & Hasselquist, D. 1992. Evidence for active female choice in a polygynous

warbler. Animal Behaviour, 44, 301–311.

Blumenrath, S. H., Dabelsteen, T. & Pedersen, S. B. 2004. Being inside nest boxes: does

it complicate the receiving conditions for great tit Parus major females?

Bioacoustics, in press.



Bower, J. L. 2000. Acoustic interactions during naturally occurring territorial conflict

in a song sparrow neighborhood. Ph.D. Thesis, Cornell University, Ithaca,

New York.

Brumm, H. 2004. The impact of environmental noise on song amplitude in a

territorial bird. Journal of Animal Ecology, 73, 434–440.

Brumm, H. & Todt, D. 2002. Noise-dependent song amplitude regulation in a

territorial songbird. Animal Behaviour, 63, 891–897.

2003. Facing the rival: directional singing behaviour in nightingales. Behaviour, 140,

43–53.

Bradbury, J. W. & Vehrencamp, S. L. 1998. The Principles of Animal Communication.


Dabelsteen, T. 1984. An analysis of the full song of the blackbird Turdus merula with

respect to message coding and adaptations for acoustic communication. Ornis

Scandinavica, 15, 227–239.

1985. Messages and meanings of bird song with special reference to the blackbird

(Turdus merula) and some methodology problems. Det Kongelige Danske

Videnskabernes Selskab Biologiske Skrifter, 25, 173–208.

1988. The meaning of the full song of the blackbird, Turdus merula, to untreated and

oestradiol-treated females. Ornis Scandinavica, 19, 7–16.

1992. Interactive playback: a finely tuned response. In: Playback and Studies of Animal

Communication, ed. P. K. McGregor. New York: Plenum Press, pp. 92–109.

Dabelsteen, T. & Mathevon, N. 2002. Why do songbirds sing intensively at dawn? A test

of the acoustic transmission hypothesis. Acta Ethologica, 4, 65–72.

Dabelsteen, T. & Pedersen, S. B. 1990. Song and information about aggressive

responses of blackbirds Turdus merula: evidence from interactive playback

experiments with territory owners. Animal Behaviour, 40, 1158–1168.

1992. Song features essential for species discrimination and behaviour assessment

by male blackbirds (Turdus merula). Behaviour, 121, 259–287.

Dabelsteen, T., Larsen, O. N. & Pedersen, S. B. 1993. Habitat-induced degradation of

sound signals: quantifying the effects of communication sounds and bird

location on blur ratio, excess attenuation, and signal-to-noise ratio in blackbird

song. Journal of the Acoustical Society of America, 93, 2206–2220.

Dabelsteen, T., McGregor, P. K., Holland, J., Tobias, J. & Pedersen, S. B. 1997. The signal

value of overlapping singing in male robins (Erithacus rubecula). Animal Behaviour,

53, 249–256.

Dabelsteen, T., McGregor, P. K., Lampe, H. M., Langmore, N. E. & Holland, J. 1998.

Quiet singing in song birds: an overlooked phenomenon. Bioacoustics, 9,

89–105.

Davies, N. B. 1992. Dunnock Behaviour and Social Evolution. Oxford: Oxford University

Press.



Dukas, R. & Kamil, A. C. 2001. Limited attention: the constraint underlying search

image. Behavioral Ecology, 12, 192–199.


60 T. Dabelsteen

Embleton, T. F. W. 1996. Tutorial on sound propagation outdoors. Journal of the

Acoustical Society of America, 100, 31–48.

Ficken, M. S., Ficken, R. W. & Witkin, S. R. 1978. Vocal repertoire of the black-capped

chickadee. The Auk, 95, 34–48.

Fleron, G. M. W. 2003. Stille sang hos solsorten, Turdus merula. M.Sc. Thesis, University

of Copenhagen.

Gompertz, T. 1961. The vocabulary of the great tit. British Birds, 54, 369–394,

409–418.

Henwood, K. & Fabric, A. 1979. A quantitative analysis of the dawn chorus: temporal

selection for communicatory optimisation. American Naturalist, 114,

260–274.

Hoi, H. 1997. Assessment of the quality of copulation partners in the monogamous

bearded tit. Animal Behaviour, 53, 277–286.

Holland, J., Dabelsteen, T., Pedersen, S. B. & Larsen, O. N. 1998. Degradation of wren

Troglodytes troglodytes song: implications for information transfer and ranging.

Journal of the Acoustical Society of America, 103, 2154–2166.

Holland, J., Dabelsteen, T., Pedersen, S. B. & Paris, A. L. 2001. Potential ranging cues

contained within the energetic pauses of transmitted wren song. Bioacoustics, 12,

3–20.

Hovi, M., Alatalo, R. V. & Siikamaki, P. 1995. Black grouse leks on ice: female mate

sampling by incitation of male competition? Behavioral Ecology and Sociobiology, 37,

283–288.

Jakobsson, S., Brick, O. & Kullberg, C. 1995. Escalated fighting behaviour incurs

increased predation risk. Animal Behaviour, 49, 137–142.

Lack, D. 1939. The display of the blackcock. British Birds, 32, 290–303.

1940. Observations on captive robins. British Birds, 33, 262–270.

Langemann, U., Tavares, J. P., Peake, T. M. & McGregor, P. K. 2000. Responses of great

tits to escalating patterns of playback. Behaviour, 137, 451–471.

Larsen, O. N. & Dabelsteen, T. 1990. Directionality of blackbird vocalization.

Implications for vocal communication and its further study. Ornis Scandinavica,

21, 37–45.

Lengagne, T. & Slater, P. J. B. 2002. The effects of rain on acoustic communication:

tawny owls have good reason for calling less in wet weather. Proceedings of the


Lind, H., Dabelsteen, T. & McGregor, P. K. 1996. Female great tits can identify mates by

song. Animal Behaviour, 25, 667–671.

Lombard, E. 1911. Le signe de l’elevation de la voix. Annales des Maladies de l’Oreille et du

Larynx, 37, 101–119.

Mathevon, N., Dabelsteen, T. & Blumenrath, S. H. 2004. Are high perches in the

blackcap Sylvia atricapilla song or listening posts? A sound transmission study.

Journal of the Acoustical Society of America, in press.


identification, ranging and eavesdropping. Philosophical Transactions of the. Royal




McGregor, P. K. & Dabelsteen, T. 1996. Communication networks. In: Ecology and







see isn’t what you get? Animal Behaviour, 62, 1059–1065.



Messmer, E. & Messmer, I. 1956. Die Entwicklung der Lautausserungen und einiger

Verhaltensweisen der Amsel (Turdus m. merula L.) unter naturlichen Bedingungen

und nach Einzelaufzucht in schalldichten Raumen. Zeitschrift fur Tierpsychologie,

13, 341–441.

Michelsen, A. 1978. Sound reception in different environments. In: Sensory Ecology, ed.

M. A. Ali. New York: Plenum Press, pp. 345–373.

Molles, L. E. & Vehrencamp, S. L. 2001. Neighbor recognition by resident males in the

banded wren, Thryothorus pleurostictus, a tropical songbird with high song type

sharing. Animal Behaviour, 61, 119–127.

Montgomerie, R. & Thornhill, R. 1989. Fertility advertisement in birds: a means of

inciting male–male competition? Ethology, 81, 209–220.

Naguib, M. & Todt, D. 1997. Effects of dyadic vocal interactions on other conspecific




Series B, 266, 537–542.

Nemeth, E., Winkler, H. & Dabelsteen, T. 2001. Differential degradation of antbird

songs in a Neotropical rainforest: adaptation to perch height? Journal of the






assess males by eavesdropping? A field study using interactive song playback.

Proceedings of the Royal Society of London, Series B, 266, 1305–1309




2002. Do great tits assess rivals by combining direct experience with information

gathered by eavesdropping? Proceedings of the Royal Society of London, Series B, 269,

1925–1929.

Poesel, A., Dabelsteen, T. & Pedersen, S. B. 2001. Making yourself heard: a study of

masking effects on blue tit Parus caeruleus singing interactions. In Proceedings of the

XVIII International BioAcoustics Conference, p. 12.


62 T. Dabelsteen

Rasmussen, R. & Dabelsteen, T. 2001. Song repertoires and repertoire sharing in a local

group of blackbirds. Bioacoustics, 13, 63–76.

Sæther, S. A. 2002. Female calls in lek-mating birds: indirect mate choice, female

competition for mates, or direct mate choice? Behavioral Ecology, 13, 344–352.

Schmitz, S., Dabelsteen, T. & Mathevon, N. 2000. High perches in the blackcap Sylvia

atricapilla: song posts or listening posts? In Proceedings of the 8th International

Behavioral Ecology Congress, p. 175.

Searcy, W. A. & Yasukawa, K. 1996. Song and female choice. In: Ecology and Evolution of

Acoustic Communication in Birds, ed. D. E. Kroodsma and E. H. Miller. Ithaca, NY:

Cornell University Press, pp. 454–473.

Snow, D. W. 1958. A Study of Blackbirds. London: Allen and Unwin.

Stoddard, P. K. 1996. Vocal recognition of neighbours by territorial passerines. In:

Ecology and Evolution of Acoustic Communication in Birds, ed. D. E. Kroodsma and E. H.

Miller. Ithaca, NY: Cornell University Press, pp. 356–374.

Stork, H.-J. 1971. Zur socialen Funktion des Gesanges der Amsel Turdus merula L.

Zeitschrift fur Tierpsychologie, 28, 54–58.

Thorpe, W. H. & Pilcher, P. M. 1959. The nature and characteristics of sub-song. British

Birds, 51, 509–514.

Titus, R. C. 1998. Short-range and long-range songs: use of two acoustically distinct

song classes by dark-eyed juncos. The Auk, 115, 386–393.

Wempen, J. 1986. Outdoor sound propagation close to the ground. In: Sound

Propagation in Forested Areas and Shelterbeds, ed. M. J. M. Martens. Nijmegen: Faculty

of Sciences, Catholic University, Nijmegen, pp. 83–106.

Wiley, R. H. & Poston, J. 1996. Indirect mate choice, competition for mates, and

coevolution of the sexes. Evolution, 50, 1371–1380.

Wiley, R. H. & Richards, D. G. 1982. Adaptations for acoustic communication in birds:

transmission and signal detection. In: Acoustic Communication in Birds, Vol. 1, ed.

D. E. Kroodsma & E. H. Miller. New York: Academic Press, pp. 131–181.


4

Performing in front of an audience:signallers and the social environment

r i c a r d o j . m a t o s 1 & i ng o s c h l u p p 2

1University of Copenhagen, Denmark2University of Zurich, Switzerland and University of Texas, Austin, USA

Introduction

Several signallers and receivers sharing the same active signalling space

constitute a communication network. This type of environment imposes addi-

tional selection pressures on both signallers and receivers other than those clas-

sically considered in signaller–receiver dyads. In this chapter, we shall discuss

how communication networks influence the behaviour of a signaller and, more

specifically, the effect of an audience (defined below) on signalling behaviour.

An individual signaller has to cope with two main issues when signalling in a

network: (a) it has to compete or cooperate with other signallers, and (b) it has to

deal with the presence of several receivers. Signalling at the same time as other

individuals poses a problem for the signaller: how does it ensure that its specific

signal is detected by a receiver when other conspecifics are signalling? Signallers

solve or minimize this problem by either cooperating or competing for the signal

broadcast space. For example, in frog and insect choruses, individuals time their

signals to avoid acoustic interference (e.g. alternating their calls) or compete for

call order in the chorus (Gerhardt & Huber, 2002; Ch. 13). At the community level,

different species with similar signals may broadcast their signals at different times

of the day (Endler, 1992).

The presence of several receivers presents two additional problems for the sig-

naller. The first is how to direct the signal to a specific receiver. For example, bird

song often has a range that encompasses several neighbouring territories. When a

bird sings, the song could potentially reach all the neighbours in surrounding ter-

ritories. During interactions with neighbours, individuals may need to direct the


63


64 R. J. Matos & I. Schlupp

signal to a specific individual, for example because that neighbour starts to sing

close to the territory boundary. McGregor & Peake (2000) discussed several ways

in which songbirds can direct the signal to a specific rival neighbour or intruder.

For example, matched counter-singing (Stoddard et al., 1992; Beecher et al., 1996)

is a good candidate for directing the signal to a specific individual bird.

The second issue that arises from the presence of several receivers, and one that

this chapter covers in more detail, is how signallers communicate in the presence

of additional receivers other than the primary target receiver. We will concentrate

on conspecific receivers because heterospecific receivers, especially predators and

parasites, have received considerable attention and are known to be important

in shaping signals and signalling interactions (Bradbury & Vehrencamp, 1998;

Chs. 2 and 8). The term audience has been used to describe conspecific receivers

in the context of a communication network (McGregor & Peake, 2000; Doutrelant

et al., 2001). In this chapter, we shall begin by discussing this term and its use in

the context of communication networks. We shall then discuss how the presence

of several receivers may affect signalling behaviour and the choice and evolution

of signalling strategies.

Definitions of audience and audience effects

Audiences

We define audiences as individuals that are present during, but do not

take part in, signalling interactions between others. We distinguish two types of

audience: evolutionary audiences and apparent audiences.

Evolutionary audiences

By evolutionary audiences we mean individuals that were historically

common in the environment of the signaller and that may have generated selec-

tion on the form and content of signalling behaviour. For example, it is widely

accepted that bird song has a dual function, both as a signal to attract females

and as a signal used in male–male competition (Berglund et al., 1996; Searcy &

Nowicki, 2000). The evolution of this dual function has been widely discussed in

the literature (e.g. Searcy & Nowicki, 2000). One hypothesis suggests that song first

evolved as a male–female signal with males acting as eavesdroppers (see Ch. 12 for

similar discussion on fiddler crabs). This eavesdropping pressure caused by male

audiences may have induced new selective forces on the form and content of the

signal, resulting in the appearance of a dual function signal. If this hypothesis is

correct, then males have acted as an evolutionary audience in bird song evolu-

tion. An evolutionary audience does not need to be present or apparent to affect


Signallers and the social environment 65

signalling behaviour at any instant in time, because selection has acted in the past

(and presumably continues to act) on the signal (e.g. introducing or emphasizing

features in the design of songs that males use in male–male competition). For

more information on the effects and importance of evolutionary audiences, we

refer the reader to Chs. 2 and 14.

Apparent audiences

Apparent audiences are individuals that affect the behaviour of the sig-

naller only when they are present and detected. For example, in the presence of

females, interacting male Siamese fighting fishes Betta splendens decrease highly

aggressive behaviours (attempted bites) and increase the intensity of conspicuous

displays (tail beats and gill cover display) (Doutrelant et al., 2001). Unlike evolution-

ary audiences, the effects produced by this type of audience are triggered when

the audience is present; males show no such effects on the different displays when

the female is absent.

In this chapter, we are mainly concerned with the study of apparent audiences,

as their effects can be studied experimentally and, unlike studies of evolutionary

audiences, they do not rely on historical inference.

Audience effects

We define an audience effect as changes in the signalling behaviour during

an interaction between individuals caused by the mere presence of an audience.

Matos & McGregor (2002) found that male fighting fish engaged in visual signalling

interactions changed their signalling (i.e. the visual displays directed towards the

rival male) when a male audience was present. It is important to emphasize that

the change in signalling behaviour occurred between the two individuals involved

in the signalling interaction and not directly towards the audience. Whether the

information content of signalling changes will depend on the balance of cost and

benefit to the signallers (see below). This type of effect is specific to a commu-

nication network as it can only occur in situations where a minimum of three

individuals is present: two individuals engaged in a signalling interaction and

one individual making up the audience.

Why audience?

Different authors have used different terms to designate extra potential

receivers in a communication network, such as bystanders (Dugatkin, 2001), unin-

tended receivers (Endler, 1993) or illegitimate receivers (Otte, 1974). Most of these

terms, however, have been used in an interspecific context (with the exception

of bystander) to describe predator detection of prey signalling behaviour (Otte,

1974; Endler, 1993; Bradbury & Vehrencamp, 1998; Ch. 2). Because we restrict our



definition of an audience to conspecifics, we exclude predators or parasites re-

sponding to the signal (Ch. 2). We chose the term audience because it is more

descriptive of the role of the individual during the signalling interaction in two

ways. First, it implies that individuals are present but do not take part in the

interaction, although they are clearly able to. Second, it implies that the individ-

uals may pay attention to the signalling interaction and thus potentially extract

information.

We think that it is important to link the term audience to other network be-

haviours such eavesdropping in this way because the presence of eavesdroppers

can impose costs and benefits on signallers and to link these costs–benefits to the

information content of the interaction. For example, the finding that eavesdrop-

pers behave more aggressively to individuals that behave as losers in an aggres-

sive signalling interaction (e.g. Chs. 2 and 14) identifies an immediate cost of an

audience on the losers. It is worth noting though that we do not have to show

that individuals are able to extract information to cause an audience effect. For

example, audiences may be costly just because there is a high risk of the audi-

ence disrupting the signalling interaction (e.g. intervention behaviour of semi-

captive zebras Equus quagga; Schilder, 1990). By comparison, non-apparent eaves-

droppers do not promote an audience effect because signallers are unaware of their

presence.

Other uses of audience and audience effect in the literature

The terms audience and audience effects have been used in the communi-

cation network literature to describe the effects on signalling interactions of the

presence of additional potential receivers that do not take part in the interaction

(Doutrelant et al., 2001; Matos & McGregor, 2002). However, these terms have also

been used in other studies in animal communication. In the following paragraphs

we shall talk about these studies and underline the differences between the two

uses of the term audience.

The first studies to use the terms audience and audience effects looked at the

effect of the presence of a conspecific on the incidence of alarm and food calls

in birds (Gyger et al., 1986; Marler et al., 1986; Gyger, 1990; Evans & Marler, 1994).

These authors were interested in whether these calls were elicited by and directed

to a specific class of individuals or audiences, namely conspecifics (e.g. conspecific

versus predator; male versus female). In these studies, an audience is defined as

any individual that is present in the same location as the subject (an apparent

audience), and the audience effect is the change in signalling behaviour (e.g. in-

crease in food call rate: Marler et al., 1986) caused by the presence of the audience.

In both cases, the signal was assumed to be directed towards the audience; for

example, Gyger et al. (1986) performed two experiments to investigate whether



male cockerels Gallus domesticus modulated their alarm calls in the presence of

an audience when a model of a predator was presented. The protocol of both

experiments was the same; the birds were placed in a cage above which a model

of a predator was ‘flown’. The audience was housed in another cage next to the

male’s cage; both individuals could see the predator. In the first experiment, the

audience was either their own mate or a female that was mated to another male,

with an empty cage as a control. The second experiment was similar to the first

one with the difference that instead of another male’s female the authors used an

unfamiliar male. The authors found that in both experiments males increased the

rate of alarm calls when a conspecific was present compared with when alone. No

significant difference was found between the presence of the male’s mate com-

pared with another male’s mate, or between the male’s mate compared with an

unfamiliar male. The authors concluded from these results that the presence of

a conspecific audience has an effect on alarm calling and that these calls may be

primarily directed towards conspecifics and not towards the predator. Because

there was no significant effect of the type of conspecific (own mate, other’s mate,

unfamiliar male), one can rule out the hypothesis that the observed increase in

call rate is a result of sharing the risk with the other prey (Gyger et al., 1986).

There are two main differences between the use of the terms audience and

audience effects in these studies and our own use. First, we restrict audience effects

to the signalling interaction between the two individuals; the audience is not the

primary receiver of the signals but acts as a potential non-targeted receiver. In the

predator/food call studies, the distinction between the audience and a primary

receiver of the signal is blurred as the target receiver is the audience (Fig. 4.1). The

second difference is that our definition is specific to communication networks. In

the predator/food call studies, this was not necessarily true; only two conspecifics

were necessary to produce the audience effect: the signaller and the audience. For

example, in a similar study to the one described above, Marler et al. (1986) showed

that male cockerels increased their food calls in response to the presence of one

hen; such a situation is a signaller–receiver dyad.

We point out these differences in use of the terms to ensure that different

phenomena are not confused by the use of a common term and suggest that the

terms should be clearly defined when used.

Audience effects

Relatively few studies have addressed directly the question of whether

audience effects occur. In this section, we summarize these studies and discuss

other systems in which audience effects appear to have an important influence

on signalling behaviour.



Fig. 4.1. The audience effect in different types of study. (a) In predator/food call

studies (Gyger, et al., 1986), the change in behaviour (dotted arrows) is triggered by the

presence of the audience (the hen) and directed towards the audience. (b) In the

audience effects described in this chapter, the change in behaviour (dotted arrows) is

triggered by the presence of the audience (non-target receiver, the hen) and directed

towards the target receiver (the other male).

Male–male aggressive signalling interactions

Individuals often use signals to compete for resources such as territories,

food or mates. These displays are used to assess the opponents’ fighting ability

and motivation (Huntingford & Turner, 1987; Bradbury & Vehrencamp, 1998). In

a communication network, this information is available not only to the opponent

but also to other individuals that are within signal range. This audience of non-

targeted receivers may introduce extra costs or benefits to signallers; as explained

above, some studies show that eavesdropping fish are more likely to initiate ag-

gressive interactions with a loser than with a winner (Oliveira et al., 1998; Earley &

Dugatkin, 2002; Chs. 2 and 5). If an audience has high costs or benefits to signallers,

then signallers should adjust their behaviour towards the opponent in order to

conceal or enhance information, respectively (McGregor & Peake, 2000).

Siamese fighting fish

Siamese fighting fish often use signals to mediate competition over re-

sources such as territories, food or mates, and such visual displays have been



(a) (b)

Fig. 4.2. Representation of the experimental design used in Matos (2002) to study the

effect of a male audience on male–male interactions in Betta splendens. (a) In the first

10 minutes, both males were allowed to interact in the absence of an audience. (b) In

the second 10 minute period, either an audience or an empty tank was revealed

(removal of the opaque partition) to the males. Ma and Mb are the interacting males;

A is the audience tank; o.p. is an opaque partition; arrows represent the direction in

which visual contact was possible.

used as a model system to address different questions related to communication

networks (e.g. eavesdropping: Oliveira et al., 1998; McGregor et al., 2001). One of

the first experiments to address specifically whether male Siamese fighting fish

were affected by the presence of an audience during an aggressive interaction was

performed by Matos (2002). Two males were allowed to interact through a clear

partition (tank walls), and a third male (the audience) was placed at a small dis-

tance from these males (Fig. 4.2). This small distance prevented the audience from

taking part in the interaction yet, at the same time, allowed both males to see the

audience. Each trial of the experiment was divided into two 10 minute periods: in

the first period the two individuals were allowed to interact without the audience

being present; the second period started when an opaque partition that separated

the audience from the two males was removed, allowing the males to see the au-

dience while interacting. Previous studies have shown that one can predict the

winner of a fight between two male fighting fish from display difference at the

beginning of the interaction (Simpson, 1968). In this experiment, the winner of

the signalling interaction was defined as the individual that displayed most during

the first 10 minutes of the interaction (the other male was the loser). It is important

to note that the barriers between males prevented actual fighting and none of the

interactions reached an outcome (e.g. displaying submissive colouration). No dis-

plays directed towards the audience were observed. Matos (2002) found that ‘win-

ners’ did not change their signalling behaviour in the presence of an audience. In



contrast, when an audience was present ‘losers’ reduced the time they spent in

gill cover display (a purely visual display) and the time spent near the opponent

compared with when there was no audience. However, there was no significant

change in the more aggressive displays that had both tactile and visual compo-

nents (i.e. attempted bites and tail beats). This change in behaviour may be viewed

as an attempt by the loser to restrict the information available to the eavesdropper

while at the same time providing adequate information for assessment by the op-

ponent. Another hypothesis is that by reducing the less-aggressive displays whilst

maintaining the more aggressive forms, ‘losers’ may seem more aggressive to the

audience. Thus even though the audience may have seen that individual lose, it

would be more reluctant to interact with it because of its aggressiveness (‘good

loser’ hypothesis: Peake & McGregor, 2004).

This study (Matos, 2002) suggests that there is an audience effect when a male

audience is present during male–male interactions and that the presence of the

audience can be more costly for the individual that is losing the interaction than

for the winner. The finding that the audience effects in this situation involved

a change of signalling behaviour by the loser fits both observations that losers

are more rapidly approached by males that saw them lose (Siamese fighting fish:

Oliveira et al., 1998; McGregor et al., 2001; swordtail fish Xiphophorus helleri: Earley &

Dugatkin, 2002) and that this effect disappears in combats where both individuals

escalated (Earley & Dugatkin, 2002).

In an earlier study, Doutrelant et al. (2001) also found that female audiences

affected male–male B. splendens aggressive displays. In this experiment, a female

audience was presented to a pair of males that interacted through a clear partition.

The effect of the presence of an audience was then compared with a treatment

where males were allowed to interact with no audience present. Males increased

the amount of conspicuous displays (e.g. tail beats and time with gill cover erect)

and decreased the more aggressive displays (e.g. attempted bites) towards oppo-

nents when a female was present. The authors interpreted this result as males try-

ing to compromise between having to interact with an opponent and at the same

time provide information to the audience by using more conspicuous displays,

which are more often used in both aggressive and courtship contexts. Doutrelant

et al. (2001) also performed a second experiment to examine whether male audi-

ences affected signalling interactions but did not find an audience effect (except

for a tendency for males to spend less time near the opponent). However, the re-

sult of these two experiments cannot be compared directly because of differences

in the experimental design and procedure (i.e. the audience was closer to the

males and the males were pre-exposed to the audience in the female experiment,

while in the male experiment the audiences were further away and there was no



(a) (b)

Fig. 4.3. Schematic representation of the experimental design used in both Matos &

McGregor (2002) and Matos et al. (2003). (a) In a five minute pre-exposure period, both

males could see the audience tank. (b) In the 10 minute interaction period, the opaque

partition was removed and both males were allowed to interact with each other in

front of or in the absence of the audience. Ma and Mb are the interacting males; A is

the audience tank; o.p. is an opaque partition; arrows represent the direction in

which visual contact was possible.

pre-exposure period). Both distance and pre-exposure to another individual have

been shown to have a strong effect on male aggressive display (Bronstein, 1989;

Halperin et al., 1998; also see below).

In a more recent experiment, Matos & McGregor (2002) looked directly at the

effect of the sex of the audience. Three different types of audience were used:

male, female B. splendens and female Xiphophorus spp. (to control for responses not

specific to conspecifics). A control with no audience present was also used. The

design and procedure of the experiment was similar to that in Matos (2002), ex-

cept that the males were first pre-exposed to the audience and then were allowed

to see and interact with the opponent (Fig. 4.3). The audience was visible for the

entire trial. No distinction was made between winners and losers as data were only

collected from one of the individuals involved in the interaction. No differences

were observed between the female Xiphophorus spp. treatment and no audience;

therefore the Xiphophorus spp. treatment was used as the control. Males behaved

more aggressively (i.e. attempted more bites and spent less time near the oppo-

nent) when a male audience was present than with a female audience (Matos &

McGregor, 2002). To explain this difference, the authors suggested that the pres-

ence of a female might confront the males with a trade-off between expelling

their male opponent and not driving away a potential mate. Males of this species

often bite when courting a female and highly aggressive males may cause females

to flee because of the high risk of injury (Bronstein, 1984). The results of these



experiments suggest that the sex of the audience is important in determining

how males should behave during aggressive signalling interactions.

Field crickets

Tachon et al. (1999) studied male–male competition for resources in the

field cricket Gryllus bimaculatus. They tested whether the presence of a female

influences the aggressive behaviour between males. In each test, a group of five

males in an arena under three different treatments was observed. Besides the two

obvious treatments, presence and absence of females, they used a third condition

where a paper impregnated with female scent was introduced into the arena.

Previous studies had shown that this scent elicited behavioural responses from

males of this species (Otte & Cade, 1976; Hardy & Shaw, 1983).

Tachon et al. (1999) found that males increased their level of aggressive displays

(e.g. aggressive stridulation and mandible flaring) towards other males in the

treatment where the females were present. Interestingly, there was no evidence

that the female scent produced the same effect as the actual presence of a female.

Female scent alone in this system may be a poor predictor of female presence and

the cost of escalating increases when there is a high probability that the female

is not present. However, in this example, it is not clear what effect direct female–

male interactions had on male–male competition, as opposed to the effect of the

mere presence of the female. Further studies are needed to attempt to distinguish

these effects and thus to confirm whether this is an example of an audience effect.

Parental behaviour

Male parental care is common in many species. If there is a direct link

between the care provided to the young and their survival until reproductive age,

it might be of advantage for the females to choose a good father as a potential

mate. One way of assessing paternal care is to observe male interactions with

young (e.g. affiliate signalling behaviour). If females do choose a good father for

their future mate, then it should be to the advantage of the male to try to perform

as a better ‘parent’ when a female is present.

Vervet monkeys

Vervet monkeys Cercopithecus aethiops have a complex social system where

individuals influence their own or other group members’ dominance rank by

socializing with individuals of different rank. In such a system, female mate choice

or preference to associate with a male can influence the male’s future position in

the hierarchy (Ch. 25). Interactions between males and infant are quite common

and males often form strong protective relationships with the females and their

young. These relationships may reduce the harassment that females and infants



receive from other group members. Therefore, females may prefer to associate

with males that perform more affiliatively towards their infant.

Hector et al. (1989) investigated whether male vervet monkeys changed their

interaction with an infant in the presence versus ‘absence’ of the mother. In this

experiment, the females were placed (a) behind a one-way mirror, where they could

see both male and infant but not vice versa; (b) behind a Plexiglas partition, where

male, female and infant could see each other; and (c) behind a metal partition,

where the female could not see the dyad and the male and infant could not see

the female. The results of this experiment showed that males are sensitive to

the presence of the mother and engaged in more affiliative and less-antagonistic

behaviour toward the infant when the male was able to see the mother. However,

it is not clear whether the effect is simply caused by the presence of the female or

occurs because the females could still potentially signal to the dyad through the

Plexiglas, affecting the behaviour of both infant and male. A further treatment

would be needed to address this question, where the female is placed behind a

one-way mirror and the dyad can see the female but not vice versa.

The authors further studied if females varied their behaviour towards males

that they saw performing more affiliative behaviours towards their infants and

found that females tolerated the males more and also performed more affiliative

behaviours towards them. In spite of the lack of an appropriate control, this study

showed that potentially individuals may adjust their behaviour when an audience

is present and that there are direct consequences to the individual.

Budgerigars

Female birds may assess male parental care behaviour by the male’sextra-

pair behaviour during the period prior to egg laying. In species with obligate

biparental care, males that provide more care to the young should be preferred as

a mate, as less-committed males increase the female’scosts of feeding and spending

more time with the young. Extra-pair activity by the male (e.g. displaying to another

female) may provide information to the female on the male’s attentiveness to-

wards the female and the nest.

Budgerigars Melopsittacus undulatus are socially monogamous birds where both

members of the pair provide parental care. The males of this species provide most

of the food to the nest, both at the start of the nesting period and through brooding.

As a consequence, male commitment to the female and brood is very important to

the female and survival of the brood, and females may use cues of male commit-

ment when they are choosing a potential mate. Baltz & Clark (1994) investigated

whether male budgerigars were less likely to court another female when their own

mate was present. In other words, they tested whether there is an effect of an audi-

ence (their mate) on the male’s extra-pair behaviour. The study was conducted on



a captive population housed in an outdoor aviary. Nestboxes were provided, sim-

ulating the nests in natural cavities observed in the wild. The authors assumed

that the females lost visual contact with the male when inside the nestbox. The

behaviour of each male and its mate was recorded in the periods where the female

was inside (no audience) and outside (audience) the nest. Males significantly in-

creased extra-pair courtship behaviour when out of view of the female (i.e. when

the female was inside the nestbox) relative to when the female was in view. How-

ever, the results of this experiment can also be explained by an alternative hypoth-

esis. Males may reduce the time courting other females because with their mate

outside the nest they are more vulnerable to extra-pair courtship and copulations

by other males in the flock. Therefore, the reduction of courtship may be a re-

sult of mate guarding (Baltz & Clark, 1994). In another study, Baltz & Clark (1997)

showed that the necessity for mate guarding did not change the males’ response

to the extra-pair female. The authors used the same experimental design as before

but this time the mate was separated from the rest of the flock in both treat-

ments. This procedure prevented other males from interacting with the female

(subject’s mate) and thus reduced the necessity for mate guarding. Once again,

males reduced courtship behaviour towards extra-pair females when their mate

was visible. Although this study suggests that there maybe an audience effect, we

consider it poor evidence for audience effects as we define them in this chapter.

The main problem with the experimental design of both studies is that the audi-

ence effect is not caused by the mere presence of the audience, the male–female

pair are only separated visually by an opaque partition, and, as the authors state,

both individuals could still contact each other through calls even when they could

not see each other. We suggest that further studies would be required to confirm

the presence of an audience effect in such system.

Human behaviour

Social psychologists have long recognized that audiences have an im-

portant effect on human behaviour (e.g. Zajonc, 1965; Blumstein, 1973; Felson,

1982; Ch. 19). These effects extend from a change in the performance of sim-

ple motor tasks, when compared with apparently ‘non-social’ contexts (Zajonc,

1965), to changes in more complex forms of social behaviour such as interper-

sonal strategies used during social interactions (Blumstein, 1973). One interesting

area of study with regard to audience effects in humans is impression manage-

ment theory. This theory focuses on the principle that a person is aware of being

characterized or typified by others when performing a behaviour and responds

by trying to make these characterizations favourable. As a consequence, most

human behaviour is designed to obtain ‘favourable’ reactions from an audience

(Felson, 1978, 1982). For example, Felson (1982) found an effect of third-party



presence on aggressive interactions between humans. The study was based on in-

terviews with patients with previous mental health problems, with ex-criminals

and with a sample of the general population. All groups answered a questionnaire

asking them to describe in detail four aggressive incidents. The replies showed

that the outcome of an interaction between individuals of the same sex was more

severe when an audience was present (when allowing for third-party instigation

or mediation of the fight). There was a higher probability that individuals would

escalate from verbal insults to actual physical contact. However, the authors also

found that the same was not true in conflicts between the sexes; the cause of

such a difference may be that the audience is more likely to disapprove of severe

aggression in between-sex conflicts (Felson, 1982).

The general idea that individuals may try to manipulate their characterization

by others has recently been used to explain altruistic behaviour in humans and

non-human animals (Zahavi & Zahavi, 1997; Wedekind & Milinski, 2000; Milinski

et al., 2001; Bshary, 2002). This idea is discussed by Bshary & D’Souza in Ch. 22.

Priming: a mechanism of audience effects or a functional alternative?

In the experiments discussed above showing that male Siamese fighting

fish behaved more aggressively towards an opponent when a male audience was

present (Matos & McGregor, 2002), the trial procedure allowed males to see the

audience before they started interacting. This procedure was used to ensure that

the males were aware of the presence of the audience during the interaction. In

a further series of experiments, Matos et al. (2003) found that the presence of an

audience before an interaction affected how male B. splendens behaved during the

interaction. Using a similar design to that described by Matos & McGregor (2002),

the authors divided each trial into two continuous periods: a pre-exposure period

(when males could either see an empty tank or a tank containing an audience) and

an interaction period (when both males where allowed to interact with each other

in the presence or absence of an audience). In the first experiment, four different

treatments where used in which the audience was (1) present in the pre-exposure

period, (2) present during the interaction period, (3) present in both periods or (4)

absent in both periods. The authors then separated the behaviours overt aggression

(i.e. attempted bites and latency to first bite) and a display score (combined measure

of the other displays, i.e. time spent flaring the gill cover, number of tail beats and

time spent near the opponent); for details on the method see Matos et al. (2003).

Overall, males behaved more aggressively (i.e. shorter latency to attempt to bite

the opponent) during the interaction in the treatments where the males were pre-

exposed to the audience (treatments 1 and 3). This effect is similar to aggressive

priming. The presence of the audience before the interaction may have increased



(a) (b)

(c) (d)

Fig. 4.4. Schematic representation of the second experimental design used in Matos

et al. (2003). (a–c) The five minute pre-exposure period when both males were

pre-exposed to an empty tank (no pre-exposure) (a); both males were pre-exposed to an

audience (b); and only one of the males (Ma) was pre-exposed to the audience (c).

(d) The 10 minute interaction period following all treatments, where both males were

allowed to interact in front of an audience. Ma and Mb are the interacting males; A is

the audience; thick lines between the tanks represent opaque partitions; arrows

represent the direction in which visual contact was possible.

the motivation to behave aggressively. As a result, individuals escalated more

rapidly into more aggressive forms of behaviour when they interacted with the

opponent.

The authors also found that priming effects overrode any effect of presenting

the audience only during the interaction. The levels of aggression between the two

treatments where males were pre-exposed (treatments 1 and 3) were similar, in-

dependent of audience presence during the interaction period, while much lower

levels of aggression were seen in treatments 2 and 4. In fact, there was no signifi-

cant difference between the treatments with the audience absent in both periods

(treatment 4) and with the audience present during the interaction (treatment 2).

These results may suggest that audiences do not affect male–male fighting fish

interactions, as the audience affected only treatments with pre-exposure. In this

respect the results matched those of Doutrelant et al. (2001), in which male au-

diences did not have an effect on male–male signalling interactions (see above).

However, we should also note that in both studies the authors did not look at

losers and winners separately.

Matos et al. (2003) performed a second experiment to look at the interac-

tion between audience effects and pre-exposure to the audiences; the design al-

lowed independent pre-expose of the two opponent males (Fig. 4.4). As in the first



experiment, the trials were divided in two periods: five minutes of pre-exposure

and a 10 minute period in which the two opponents were allowed to interact.

There were three treatments in the pre-exposure period: both males pre-exposed

to an empty tank (no pre-exposure; Fig. 4.4a), both males pre-exposed to the audi-

ence (symmetric pre-exposure; Fig. 4.4b), and one of the opponents pre-exposed to

the audience while the other male was pre-exposed to an empty tank (asymmetric

pre-exposure; Fig. 4.4c). The audience was always present in the interaction period

(Fig. 4.4d). The results confirmed that pre-exposed males tend to behave more ag-

gressively (higher display scores and overt aggression); both the no pre-exposure

and the symmetric treatments showed the same tendencies. In the asymmetric

treatment, pre-exposed males also tended to display more than the ones not pre-

exposed with one exception: non-pre-exposed males matched the number of at-

tempted bites of the pre-exposed males. A possible explanation is that it may be

costly for individuals not to retaliate when its opponent escalates, because of the

high risk of injury, especially in a confined space such as the experimental tanks

(Maan et al., 2001). However, this cost may be enhanced by the presence of the

audience. By matching the opponent in more aggressive behaviour, males may be

either decreasing the ability of an audience to discriminate loser from winner or

manipulating the information to seem more aggressive. These results support the

previously discussed idea that males, particularly losers, may gain by performing

more aggressively during an interaction in the presence of an audience, as it may

decrease the chances of future harassment by that individual (Earley & Dugatkin,

2002; Matos et al., 2003).

These two studies taken together support the idea that previous studies on

audience effects (i.e. Doutrelant et al., 2001; Matos & McGregor, 2002) have under-

estimated the effect of pre-exposure on male aggression. Nevertheless, they also

suggest that the social environment (i.e. audiences) is important in determining

the dynamics of signalling interaction.

Previous studies have shown that priming is an important mechanism mediat-

ing aggressive interactions (e.g. Potegal & Popken, 1984; Bronstein, 1989; Halperin

et al., 1992) as it affects the individual’s aggressive motivation. For example, prim-

ing may decrease the time to initiate aggression or increase the attack behaviour

of individuals (e.g. Potegal & ten Brink, 1984; Halperin et al., 1998). However, the

effect on the outcome of interactions is not always clear. It seems that priming

may have a more pronounced effect during the initial stage of the fight, either

causing the individual to display more actively at the beginning of the inter-

action or to escalate and initiate aggression more quickly (Potegal & Popken,

1984; Bronstein, 1989; Halperin et al., 1998). In several species, individuals that

display more intensively and escalate earlier during an interaction usually gain a



competitive advantage over their opponents (Huntingford & Turner, 1987). In such

a case, priming may produce a positive effect as it increases the probability that

the individual will win the fight. In some cases, however, priming can have a

negative effect, male B. splendens that have been isolated and then primed with

a conspecific image behaved more aggressively towards their opponents but lost

most of the interactions (Halperin et al., 1998). These individuals could have been

manipulated by priming into aggressive levels that they were not able to sustain

during the entire fight, causing them to tire faster than the opponents and sub-

sequently lose the interaction. We conclude that priming may have an important

impact on the outcome of the interaction, but whether this impact is positive or

negative may depend on whether the initial stages of the interaction determine

the outcome and on the length of the interaction.

One potential mechanism behind such aggressive priming is the production

of hormones caused by the presentation of a social stimulus. In a recent study,

Oliveira et al. (2000) showed that watching a fight raises the androgen levels of

adult male cichlid fish Oreochromis mossambicus. Priming may involve a similar

mechanism, and the facilitation of aggressive behaviour through pre-exposure

may be caused by an increase in androgen levels initiated by the pre-exposure to

the audience. Oliveira et al. (2000) suggested that these hormones mediate changes

in the perceptual abilities and readiness to interact of males, which, in turn, would

enhance their success in social interactions.

Further studies are needed to comprehend fully the relation between the adap-

tive value of priming and the presence of an audience. Advances in the under-

standing of the effects of the social environment on the neuroendocrinological

system may be an important contribution in this area (Ch. 21).

Summary and future directions

One important question in the general context of communication net-

works is how narrowly or broadly we wish to define the social context of sig-

nalling. Recent studies have shown that mate preferences can be altered by view-

ing sexual interactions (Westneat et al., 2000). In several species, seeing a male

mate enhances this male’s attractiveness to females (Dugatkin, 1992; Ch. 5). Such

choosing females would be eavesdropping on the signalling–mating interaction

of two other individuals and responding accordingly. The same reasoning might

apply for other interactions as well. It appears that most dyadic interactions are

actually embedded in a social context or network. This raises the question of

how common the well-studied dyadic interactions actually are, as these studies

have only considered them in a social void. This situation might be more of an

exception than the rule. In this context, more knowledge on sensory ecology and



especially the role of private channels would be very helpful. Communication via

private channels uses sensory channels not available to the audience. This has

been documented for swordtails (Xiphophorus spp.): males signal in the ultraviolet,

a part of the spectrum that cannot be detected by a predator, the Mexican tetra

Astyanax mexicanus (Cummings et al., 2003). True dyadic interactions may be brief

and limited to signals transmitted in close contact. A potential example might be

nipping in poeciliid fishes; here, males nibble a female’s genital region and chem-

ical signals are transmitted (Parzefall, 1973). Such signals are not available to any

other individual, although the male’s response to the signal might be (Parzefall,

1973).

Another aspect to consider is that many social interactions relevant to several

aspects of an individual’s life may happen simultaneously and influence each

other. Any given individual will have to include this into its signalling decisions.

For instance, a singing bird may simultaneously be faced with the problems of at-

tracting a female, discouraging a neighbour from entering its territory and avoid-

ing predators. This leads to a more complicated network of social interactions,

the components of which may influence each other to shape a ‘social interac-

tion network’. Our singing bird example also illustrates that each context alone

would select for a different signal or signalling strategy. Signals have to be effective

enough to transmit accurate information to target receivers but private enough to

prevent this information from being detected by ‘unwanted’untargeted receivers.

Any signal that is under such conflicting demands will be a compromise, depend-

ing on the associated costs and benefits. Only recently has formal modelling been

used to address this problem (Johnstone, 2001; Ch. 26).

We have attempted to show that the presence and type of audience can have

important effects on the signalling strategies of individuals. The nature of infor-

mation and the extent to which it is broadcast may depend on the type of audience

and on the role of each signaller during an interaction. Audiences may also in-

fluence the evolution of new types of signal. During signal evolution, different

pressures may arise in signal design depending on whether it is specialized to ad-

vertize or privatize information (e.g. ‘normal’ song versus quiet song: Dabelsteen

et al., 1998; Ch. 3). Audience effects may be closely linked with mechanisms such as

priming effects, which may influence motivation of signallers and consequently

their signalling strategy. In natural systems, the social environment affects how

animals make behavioural decisions. Individuals can use signalling interactions

between others as a source of information; this can in many ways have important

consequences for the fitness of individuals. In order to improve our understanding

of the evolution of signals and signalling strategies, we must take into account

the individuals’ social environment and the costs and benefits associated with the

presence of audiences and eavesdroppers.



Acknowledgements

We would like to express our sincere thanks to Peter McGregor for providing us with the

opportunity to write this chapter and to Tom Peake, Ryan Earley, Denise Pope, Giuliano Matessi

and Andrew Terry for their valuable comments and discussion of the manuscript. We thank the

Fundacao para a Ciencia e Tecnologia (Portugal) for funding R. M., whose Ph.D. provided data and

ideas included in this chapter. I. S. was supported by a Heisenberg Fellowship of the Deutsche

Forschungogemeinschaft.

References

Baltz, A. P. & Clark, A. B. 1994. Limited evidence for an audience effect in budgerigars,

Melopsittacus undulatus. Animal Behaviour, 47, 460–462.

Baltz, A. P. & Clark, A. B. 1997. Extra-pair courtship behaviour of male budgerigars and

the effect of an audience. Animal Behaviour, 53, 1017–1024.

Beecher, M. D., Stoddard, P. K., Campbell, S. E. & Horning, C. L. 1996. Repertoire

matching between neighboring song sparrows. Animal Behaviour, 51, 917–923.

Berglund, A., Bisazza, A. & Pilastro, A. 1996. Armaments and ornaments: an

evolutionary explanation of traits of dual utility. Biological Journal of the Linnean

Society, 58, 385–399.

Blumstein, P. W. 1973. Audience, Machiavellianism, and tactics of identity bargaining.

Sociometry, 36, 346–365.



Bronstein, P. M. 1984. Agonistic and reproductive interactions in Betta splendens.

Journal of Comparative Psychology, 98, 421–431.

Bronstein, P. M. 1989. The priming and retention of agonistic motivation in male

Siamese fighting fish, Betta splendens. Animal Behaviour, 37, 165–166.

Bshary, R. 2002. Biting cleaner fish use altruism to deceive image-scoring client reef

fish. Proceedings of the Royal Society of London, Series B, 269, 2087–2093.

Cummings, M. E, Rosenthal, G. G. & Ryan, M. J. 2003. A private ultraviolet channel in

visual communication. Proceedings of the Royal Society of London, Series B, 270,

897–904.

Dabelsteen, T., McGregor, P. K., Lampe, H. M., Langmore N. E. & Holland J. 1998.

Quiet song in song birds: an overlooked phenomenon. Bioacoustics, 9,

89–105.

Doutrelant, C., McGregor, P. K. & Oliveira, R. F. 2001. The effect of an audience on

intra-sexual communication in male Siamese fighting fish, Betta splendens.


Dugatkin, L. A. 1992. Sexual selection and imitation females copy the mate choice of

others. American Naturalist, 139, 1384–1389.








Endler, J. A. 1992. Signals, signal conditions, and the direction of evolution. American

Naturalist, 139, S125–S153.

Endler, J. A. 1993. Some general comments on the evolution and design of animal

communication systems. Philosophical Transactions of the Royal Society of London,

Series B, 350, 215–225.

Evans, C. S. & Marler, P. 1994. Food calling and audience effects in male chickens,

Gallus gallus: their relationship to food availability, courtship and social

facilitation. Animal Behaviour, 47, 1159–1170.

Felson, R. B. 1978. Aggression as impression management. Social Psychology, 41,

94–105.

Felson, R. B. 1982. Impression management and the escalation of aggression and

violence. Social Psychology Quarterly, 45, 245–254.


Problems and Diverse Solutions. Chicago, IL: University of Chicago Press.

Gyger, M. 1990. Audience effects on alarm calling. Ethology Ecology and Evolution, 2,

227–232.

Gyger, M., Karakashian, S. J. & Marler, P. 1986. Avian alarm calling: is there an

audience effect? Animal Behaviour, 34, 1570–1572.

Halperin, J. R. P., Dunham, D. W. & Ye, S. 1992. Social isolation increases social display

after priming in Betta splendens but decreases aggressive readiness. Behavioural

Processes, 28, 13–32.

Halperin, J. R. P., Giri, T., Elliott, J. & Dunham, D. W. 1998. Consequences of

hyper-aggressiveness in Siamese fighting fish: cheaters seldom prospered. Animal

Behaviour, 55, 87–96.

Hardy, T. N. & Shaw, K. C. 1983. The role of chemo-reception in sex recognition by

male crickets: Acheta domesticus and Teleogryllus oceanicus. Physiological Entomology, 8,

151–166.

Hector, A. C. K., Seyfarth, R. M. & Raleigh, M. J. 1989. Male parental care, female

choice and the effect of an audience in vervet monkeys. Animal Behaviour, 38,

262–271.

Huntingford, F. A. Turner, A. K. 1987. Animal Conflict. London: Chapman & Hall.

Johnstone, A. R. 2001. Eavesdropping and animal conflict. Proceedings of the National


Maan, M. E., Groothuis T. G. G. & Wittenberg, J. 2001. Escalated fighting despite

predictors of conflict outcome: solving the paradox in a South American cichlid

fish. Animal Behaviour, 62, 623–634.

Marler, P., Dufty, A. & Pickert, R. 1986. Vocal communication in the domestic chicken:

II. Is a sender sensitive to the presence and nature of a receiver? Animal Behaviour,

34, 194–198.

Matos, R. J. 2002. Social influences on signalling: audience effects and communication

networks. Ph.D. Thesis, University of Copenhagen.



Matos, R. J. & McGregor, P. K. 2002. The effect of the sex of an audience on

male–male displays of Siamese fighting fish (Betta splendens). Behaviour, 139,

1211–1221.

Matos, R. J., Peake, T. M. & McGregor, P. K. 2003. Timing of presentation of an

audience: aggressive priming and audience effects in male Siamese fighting fish

(Betta splendens). Behavioural Processes, 63, 53–61.





see isn’t what you get. Animal Behaviour, 62, 1059–1065.

Milinski, M., Semmann, D., Bakker, T. C. M. & Krambeck, H.-J. 2001. Cooperation

through indirect reciprocity: image scoring or standing strategy? Proceedings of the



gather information from observing conspecific interactions. Proceedings of the



fish hormone levels. Nature, 409, 475–475.

Otte, D. 1974. Effects and functions in the evolution of signalling systems. Annual


Otte, D. & Cade, W. 1976. On the role of olfaction in sexual and interspecies

recognition in crickets (Acheta and Gryllus). Animal Behaviour, 24, 1–6.

Parzefall, J. 1973. Attraction and sexual cycle of poeciliids. In: Genetics and Mutagenesis

of Fish, ed. J. H. Schroder. Berlin: Springer, pp. 177–183.

Peake, T. M. & McGregor, P. K. 2004. Information and aggression in fishes. Learning and

Behavior, in press.

Potegal, M. & ten Brink, L. 1984. The behavior of attack-primed and attack satiated

female golden hamsters, Mesocricetus auratus. Journal of Comparative Psychology, 98,

66–75.

Potegal, M. & Popken, J. 1984. The time course of attack priming effects in female

hamsters. Behavioural Processes, 11, 199–208.

Schilder, M. B. J. 1990. Interventions in a herd of semi-captive plains zebras. Behaviour,

112, 53–83.

Searcy, W. S. & Nowicki, S. 2000. Male–male competition and female choice in the

evolution of vocal signalling In: Animal Signals: Signalling and Signal Design in

Animal Communication, ed. Y. Espmark, T. Amundsen & G. Roseqvist. Trondheim:

Tapir Academic Press, pp. 301–315.

Simpson, M. J. A. 1968. The display of the Siamese fighting fish Betta splendens. Animal

Behaviour Monographs, 1, 1–73.

Stoddard, P. K., Beecher, M. D., Campbell, S. E. & Horning, C. L. 1992. Song-type

matching in the song sparrow. Canadian Journal of Zoology, 70, 1440–1444.

Tachon, G., Murray, A.-M., Gray, D. A. & Cade, W. H. 1999. Agonistic displays and the

benefits of fighting in the field cricket, Gryllus bimaculatus. Journal of Insect

Behaviour, 12, 533–543.



Wedekind, C. & Milinski, M. 2000. Cooperation through image scoring in humans.

Science, 288, 850–852.

Westneat, D. F., Walters, A., McCarthy, T. M., Hatch, M. I. & Hein, W. K. 2000.

Alternative mechanisms of nonindependent mate choice. Animal Behaviour, 59,

467–476.

Zahavi, A. & Zahavi, A. 1997. The Handicap Principle. Oxford: Oxford University Press.

Zajonc, R. B. 1965. Social facilitation. Science, 149, 269–274.


5

Fighting, mating and networking:pillars of poeciliid sociality

r y a n l . e a r l e y 1 & l e e a l a n d u g a t k i n 2

1Georgia State University, Atlanta, USA2University of Louisville, USA

We are both spectators and actors in this great drama of existence

Niels Bohr

Introduction

Poeciliid fishes such as green swordtails Xiphophorus helleri and guppies

Poecilia reticulata aggregate in social groups called shoals. In addition to reducing

predation risk and increasing foraging efficiency (e.g. Magurran & Pitcher, 1987;

Ranta & Juvonen, 1993), fish shoals promote the transfer of social information

within the group. For instance, information about foraging routes is transmit-

ted from trained individuals to naive fish in guppy shoals (Laland & Williams,

1997; Swaney et al., 2001; Brown & Laland, 2002). The type of information transfer

demonstrated in the social learning and foraging literature involves the trans-

mission of signals from one or more individuals to the remaining group mem-

bers. Investigations of social foraging and anti-predator behaviour have demon-

strated that poeciliids attend to a variety of cues emitted by both conspecifics and

heterospecifics (e.g. predators: Brown & Godin, 1999; Mirza et al., 2001; Brosnan

et al., 2003). Although social learning and anti-predator responses constitute

important aspects of group living in poeciliids, this chapter focuses more on how

individuals gain information from observing interactions that occur in their social

environment. Indeed, the concept of communication networks was founded on

the premise that the information exchanged during social interactions (e.g.

agonistic or courtship displays) may be available not only to the participants but

also to bystanders within signal detection range (McGregor, 1993; McGregor et al.,

2000).


84


Poeciliid sociality: fighting, mating and networking 85

Throughout this chapter, we make a clear distinction between bystanders and

eavesdroppers, though the two have been used synonymously in the past (Ch. 2).

Bystanders are any individuals within detection range of signalling interactions

while eavesdroppers represent a subset of bystanders that extract information

from these interactions. The primary aim of this chapter is to examine how

social eavesdropping – extracting information from signalling interactions be-

tween others (Ch. 2) – influences aggressive contest behaviour and female mate

choice in male X. helleri and female P. reticulata, respectively. Therefore, we focus

on how eavesdropping affects the subsequent behaviour of poeciliid bystanders

rather than on how the behaviour of participants in an interaction is modified

in the presence of an audience (Ch. 4). Swordtails and guppies are well suited to

investigations of networking phenomena because they are highly social, exhibit

stereotypical agonistic and courtship displays and are especially responsive to a

host of stimuli (e.g. visual, chemical) in their social surroundings.

Social eavesdropping and contest behaviour

Although sociality confers fitness-related benefits to individuals within

the group, competition for social status and limiting resources (e.g. food, mates)

often increases with group size (Pulliam & Caraco, 1984). In many animals, overt

aggressive interactions are most common during hierarchy or territory establish-

ment. Among fishes, rank-order fights involve a series of gradually escalating dis-

plays that convey information about strength, size or willingness to persist in the

encounter. The intensity and/or duration of aggressive contests depend largely on

differences in fighting ability between adversaries. When substantial differences

in fighting ability exist, the interaction may terminate following a bout of non-

contact displays. Interactions between well-matched opponents, however, may

intensify to physical combat, where behavioural tactics such as mouthwrestling

are used to settle the dispute (Enquist et al., 1990). Escalated contests often yield

unambiguous dominant–subordinate relationships among closely matched com-

petitors but the costs associated with such interactions can also be quite high. For

instance, Neat et al. (1998) revealed that prolonged fights result in the accumula-

tion of anaerobic metabolites and depletion of sugar reserves in the muscle tissue

of cichlid fish Tilapia zillii. Other potential contest costs include physical injury,

increased susceptibility to predation, lost mating or foraging opportunities and

increased stress hormone levels, which may impede future reproductive activity

(Haller, 1995; Jakobsson et al., 1995; Schuett, 1997; Halperin et al., 1998; Neat et al.,

1998).

Most theoretical and empirical work on aggressive contest behaviour assumes

that information about fighting ability is available only via direct interactions


86 R. L. Earley & L. A. Dugatkin

(Enquist & Leimar, 1983; Payne & Pagel, 1997; Mesterton-Gibbons & Adams, 1998;

Payne, 1998). Although this may be the case for solitary species, the social environ-

ment of group-living animals is ripe with opportunities for indirect assessment

of fighting ability via eavesdropping. Since observing fights may provide infor-

mation about fighting ability without the associated costs of physical combat,

eavesdropping should be an advantageous assessment strategy when both contest

costs and the opportunities for watching interactions are high. In this section, we

review a series of experiments that elucidate how eavesdropping modulates the

agonistic behaviour of male X. helleri. Specifically, we focus on how a bystander’s

behaviour changes after observing fights and the levels at which these behavioural

modifications are manifest.

Spectators in swordtail networks: empirical work

Cast of characters

Green swordtail fish are an excellent system in which to examine visually

based network effects. Although there are few studies on the costs of combat in X.

helleri, corticosteroid hormone levels are elevated above control levels for at least

six hours after contest settlement, particularly in subordinate fish (Hannes et al.,

1984). This finding, together with the data from a number of other studies on fish

(e.g. Neat et al., 1998), indicates that fighting is likely to be costly for male green

swordtails. Moreover, in both laboratory and field settings, male X. helleri establish

social hierarchies where rank-order fights and/or attack–retreat sequences are

common (Beaugrand et al., 1984; Franck et al., 1998). Thus, bystanders probably

have ample opportunity to observe aggressive interactions that occur within their

social environment and, given the potential costs of fighting, may benefit from

doing so. Figure 5.1 depicts a simplified version of the swordtail social network as

being composed of fighters engaged in aggressive signalling interactions, solitary

individuals not involved in dominance interactions, and bystanders. Furthermore,

the solitary individuals and fighters can either be observed or not observed by a

bystander within range to extract information relevant to fighting ability (e.g.

signals exchanged by the fighters or subtle behavioural/morphological cues of

solitary individuals). Of course, each individual within the swordtail groups can

assume the solitary, fighter or bystander position depending on the social circum-

stances they are exposed to at any given moment.

Basic paradigm

Within the swordtail network, bystanders may extract information from

fighters or solitary fish and integrate this information for use in future encounters

with the observed individual(s). Furthermore, eavesdropping may elicit changes


(a)

(b)

+−

+−

+−

+−

Fig.

5.1.

The

ind

ivid

ual

sco

mp

risi

ng

swor

dta

iln

etw

orks

and

the

hyp

oth

etic

alef

fect

sth

atw

atch

ing

figh

tsco

uld

hav

eon

aby

stan

der

’sre

spon

seto

war

dob

serv

edw

inn

ers,

obse

rved

lose

rs,o

rso

lita

ryco

nsp

ecif

ics

that

had

been

wat

ched

orn

otw

atch

ed.A

fter

obse

rvin

gfi

ghts

orso

lita

ryin

div

idu

als

(a;d

ash

edli

nes

),th

eby

stan

der

sm

ay

resp

ond

ina

vari

ety

ofw

ays

toot

her

sin

its

soci

aln

etw

ork

(b;s

olid

lin

es).

Effe

cton

byst

and

erbe

hav

iou

r:0,

no

effe

ct;−

,avo

idan

cere

spon

se;+

,in

itia

tion

ofag

gres

sion

ores

cala

tion

and

less

like

lyto

con

ced

e.

87



in the motivational state of the bystander that could influence its interactions

with individuals that had not been observed (Fig. 5.1). To address these issues in

the laboratory, we established a protocol where a bystander was visually isolated

from, or able to observe, fighters or solitary conspecifics. This was accomplished

by placing either an opaque plastic partition (opaque treatment) or a one-way

mirror (mirror treatment) between the bystander and either a pair of interacting

individuals or a solitary individual in laboratory aquaria (Earley & Dugatkin, 2002;

Earley et al., 2003, 2004). In all of the experiments, the bystander and the fighters

(or solitary conspecific) were matched for lateral surface area, a composite mea-

surement of body length, body depth and sword length that corresponds better

with fighting ability than standard length alone (Beaugrand et al., 1996). Following

opportunity or no opportunity to observe, the bystander was pitted against one

of the following individuals: the observed solitary individual, the winner of the

observed fight, the loser of the observed fight, or a solitary fish that had not been

observed (solitary-naive).

In the context of aggressive interactions, eavesdropping could influence a

bystander’s behaviour toward the observed individual(s) in many hypothetical

ways. First, observation could have no effect (0) on the behaviour of the bystander.

In this case, the behaviour of bystanders that had observed should be similar to

the behaviour of naive bystanders. Second, the bystander may exhibit an avoid-

ance response (−) if, through eavesdropping, it assessed the fighting ability of the

observed individual(s) to exceed its own. Avoidance responses include refraining

from initiating aggression or escalation and withdrawing from the contest. Third,

if the bystander assessed the fighting ability of the observed individual(s) to be less

than its own, it may be more inclined to initiate aggression or escalate and less

likely to concede to its opponent (+). It is also possible that eavesdropping could

affect the aggressive behaviour of the bystander outside the context of interacting

with the individuals that had been observed, for example when interacting with

solitary conspecifics that were not observed (Fig. 5.1).

Eavesdropping on fights: confronting winners and losers

Our work on communication networks in X. helleri began with a rela-

tively simple question: does watching a fight influence the agonistic response of

a bystander toward the observed contestants? To determine precisely how eaves-

dropping influences a bystander’s interactions with the observed contestants, one

must also recognize that the fighters enter the interaction with previous domi-

nance or subordination experience. Previous winning or losing experiences are

known to influence a host of contest characteristics in many fish species (e.g.

probability of initiating aggression or winning against future opponents: Bakker



& Sevenster, 1983; Bakker et al., 1989; Chase et al., 1994; Hsu & Wolf, 1999, 2001).

Prior winning experiences tend to increase the probability of future contest suc-

cess, while prior losing experiences decrease the likelihood of winning in the

future; these experiential effects have been dubbed the ‘winner effect’ and ‘loser

effect’, respectively. The opaque treatment can be seen as a control for the effects

of previous fighting experience. Because the bystander is not allowed to observe

the interaction, the dynamics of bystander versus winner or bystander versus loser

contests are influenced primarily by the fighters’previous experience. In contrast,

both eavesdropping and prior experience effects can mediate bystander contest

dynamics in the mirror treatment. Thus, when the dynamics of contests involv-

ing the bystander are compared between the opaque and mirror treatments, the

contribution of eavesdropping can be determined explicitly.

When the effects of the fighters’ prior experience are controlled, we found

that watching fights had a considerable influence on bystander behaviour, partic-

ularly when confronted with the winner of the observed interaction (Earley &

Dugatkin, 2002). Bystanders that observed the contest exhibited a more

pronounced avoidance response toward winners than bystanders that had not

observed the interaction (Fig. 5.2; Winner). However, the intensity of the observed

contest had no bearing on the eavesdroppers’response toward winners (see caption

to Fig. 5.3). Thus, eavesdroppers avoided observed winners regardless of whether

they defeated their opponent by escalated or non-escalated means. A radically dif-

ferent scenario emerged in the bystander versus loser contests. Here, bystanders

responded in a similar way to all losers, regardless of whether their defeat was

witnessed (Fig. 5.2; Loser). Nevertheless, our data revealed that eavesdroppers were

less likely to initiate aggressive behaviour and win against losers that persisted

versus losers that retreated immediately in the observed contest (Fig. 5.3).

These results demonstrate that swordtails not only make the dichotomous

assessment of ‘winner versus loser’ but also calculate the fighting ability of

each contestant, particularly the losers, independent of final outcome. The ca-

pacity to tease apart fighting ability from final status (i.e. winner or loser) may

be of particular benefit during the initial establishment of dominance hierar-

chies, where the rank of each group member relative to the others, including

the bystander, remains unclear. For instance, if eavesdroppers base their future

agonistic decisions on status assessment alone (winner versus loser), challenging

a relatively strong loser may prove costly. However, eavesdroppers that assess the

fighting ability of the observed contestants based on, for example, persistence

or willingness to escalate may be better equipped to adjust their agonistic be-

haviour in a manner consistent with the actual fighting ability of others in the

network.


Bys

tand

er in

itiat

es a

ggre

ssio

n (p

ropo

rtio

n of

con

test

s)

observed naive

(a)

Prop

ortio

n of

esc

alat

ed c

onte

sts

(b)

observed naive

Fig. 5.2. The effect of watching fights on a swordtail bystander’s propensity to (a)

initiate aggression, (b) participate in escalated contests and (c) win contests against

observed winners, observed losers, solitary conspecifics that had been observed, or

previously unknown individuals (solitary-naive). Grey bars represent the mirror

treatment, where a bystander could see either the fighters or the solitary individual

without being seen; white bars represent the opaque treatment, where a bystander

remained naive to the presence of the fighters or the solitary individual prior to

confrontation. An asterisk indicates significant differences (p < 0.05) between the

mirror and opaque treatments.

90



Bys

tand

er w

ins

(pro

port

ion

of c

onte

sts)

(c)

observed naive

Fig. 5.2. (cont.)

Eavesdropping on fights: observing solitary individuals

Given that male swordtails modify their agonistic behaviour based on

what appears to be an independent assessment of each contestant, bystanders

may be getting more information from the individual than from the interaction

itself. To address this possibility, we allowed some bystanders to observe a solitary

fish through a one-way mirror; the remaining bystanders were visually isolated

from the solitary individual using an opaque plastic partition. After the obser-

vation period (or lack thereof), the bystander was confronted with its opponent.

In this experiment, we allowed small variation in body length, body depth and

sword length between the two contestants. Asymmetries in any of the three size

measurements did not lead to substantial mismatches in lateral surface area.

When asymmetries in body size between the bystander and the watched individ-

ual were not considered, observation did not appear to elicit modifications in

the bystanders’ behaviour (Earley et al., 2003; Fig. 5.2, Solitary-observed). However,

when small differences in body size were examined as potentially informative

cues, an interesting result emerged. The observers’ propensity to initiate attack

increased as a function of relative body length. Somewhat surprisingly, observers

were more prone to initiate against larger opponents, a trend evident only in the

mirror treatment. Rapid escalation was a key predictor of contest success, with

initiators of attack winning 81% of the interactions (Earley et al., 2003). Therefore,



Prob

abili

ty o

f in

itiat

ing

aggr

essi

on a

gain

st lo

ser

Prob

abili

ty th

at b

ysta

nder

win

s ag

ains

t los

er

(a) (b)

Fig. 5.3. The probability that bystanders will a) initiate aggression or b) win against

losers from the observed contest as a function of the degree to which the observed

contest in the mirror treatment escalated. The number of reciprocal acts refers to the

frequency with which an aggressive act from the eventual winner of the initial contest

was countered with an aggressive act from the eventual loser. Solid lines indicate the

probability of the event; dashed lines indicate the upper and lower confidence limits.

Results of the logistic analyses for initiation (Wald χ12 = 4.4; p = 0.036) and winning

(Wald χ12 = 4.5; p = 0.034) against persistent losers were statistically significant.

Similar analyses on bystander versus winner contests yielded insignificant results

(initiation: χ12 = 0.4; p = 0.54; winning: χ1

2 = 0.003; p = 0.96) and are not shown here.

when a potential fighting disadvantage is perceived, observers adopt tactics that

enhance the probability of contest success against slightly larger opponents.

Because relative body size had no influence on the behaviour of individuals that

did not preview their opponent, prior observation is the likely trigger for modi-

fications in the bystanders’ attack behaviour. These findings demonstrate that

swordtails are capable of detecting small disparities in body length and that they

adjust their agonistic behaviour in response to perceived size asymmetries. The

fact that behavioural modifications elicited by watching solitary individuals were

distinct, even opposite, from changes generated by observing fights suggests that

different information is being integrated in each case. Information that accu-

rately reflects superior fighting ability, such as outcome or persistence, may be

more effective at deterring eavesdroppers than information about relative body



size alone. In systems where prior experience effects, or any other social factor,

have considerable influences on contest behaviour, information about size alone

may not be sufficient to deter eavesdroppers, particularly when asymmetries are

small.

An interesting point that was not addressed by the above experiment is whether

observing individuals exhibiting contest-type behaviour, without actually witness-

ing the interaction itself, would modify a bystander’s agonistic response. The

importance of this question lies in partitioning how the fighting tactics of each

contestant versus the dynamics of the actual interaction influence bystander

behaviour. McGregor et al. (2001) allowed a bystander to observe either ‘real’ or

‘apparent’interactions between two male Betta splendens. The ‘real’ interaction pro-

vided the bystander with information about two contestants that were actually

fighting with one another. In the ‘apparent’interaction, the bystander was exposed

to two males that appeared to be interacting with each other but were actually

fighting against different opponents (Fig. 5.3 and Fig. 2.2d, p. 26). In the ‘apparent’

interactions, bystanders responded to winners (i.e. the individual of a pair that

displayed longest) more strongly than to losers. It is important to note that, in

the ‘apparent’ interactions, an individual that was perceived to have won/lost by

the bystander may have actually obtained a different experience. Therefore, these

results demonstrate that bystanders utilize information about individual contest

behaviour (e.g. display duration), in addition to interaction dynamics, to gauge

their future agonistic decisions. Another way to test this idea would be to allow a

bystander to observe an individual exhibiting aggressive behaviour toward a stim-

ulus that is out of view of the bystander (e.g. mirror image; conspecific opponent).

Following observation, the bystander could be exposed to the individual it had

observed; as potential controls, the bystander could be pitted against individu-

als that were observed not interacting with the stimulus and/or individuals that

were not seen interacting with the stimulus. Provided that the watched individuals

show substantial variation in aggressive behaviour, this type of experiment could

elucidate whether watching individuals exhibiting contest-typical behaviour is

sufficient to elicit modifications in bystander behaviour.

Eavesdropping on fights: confronting naive conspecifics

From the experiments described above, it is clear that observing aggres-

sive interactions prompts an agonistic response in swordtail bystanders. Initially,

we interpreted this as evidence that bystanders extract information about the

fighting ability of each contestant and respond accordingly when confronted

with the observed individuals. Although our results are consistent with the

existence of eavesdropping in swordtails, it is possible that observing fights elicits

general changes in a bystander’s aggressive motivation that could affect future



contest behaviour. This alternative hypothesis does not require that bystanders

extract information from signalling interactions between others. Studies on

Mozambique tilapia Oreochromis mossambicus (Oliveira et al., 2001) and Siamese

fighting fish B. splendens (Clotfelter & Paolino, 2003) have shown that observing

fights increases urinary androgen levels and aggressive behaviour/contest success,

respectively. This type of response to social stimuli is best labelled as ‘priming’

(Hollis et al., 1995). However, the motivational changes experienced by swordtail

bystanders, if any, may be quite different from priming. Recall that swordtail spec-

tators responded with increased avoidance behaviour toward winners and losers

that had persisted in the observed contest. Based on these data, any changes in the

motivational state of swordtail bystanders should be manifest as decreases, rather

than increases, in aggressive behaviour, possibly as a consequence of elevated

corticosteroid stress hormones.

To address the ‘motivational’ hypothesis from a behavioural perspective, some

bystanders were exposed to aggressive interactions while others were visually iso-

lated from a pair of fighters. Following the observation period, or the lack thereof,

the bystander was confronted with an inexperienced fish that was not seen.

Bystanders that observed conflict were equally likely to initiate aggression, es-

calate and win against the inexperienced fish as bystanders that were not exposed

to the fight (Earley et al., 2004; Fig. 5.2: Solitary-naive). Therefore, observing fights

does not appear to precipitate general increases or decreases in the aggressive

motivation of swordtail bystanders. Given the lack of support for the ‘motiva-

tional’ hypothesis, it is reasonably clear that swordtail behaviour is modulated

by more sophisticated mechanisms than observation-induced priming or stress.

Namely, the agonistic response of swordtail eavesdroppers is influenced by the

acquisition, integration and retention of information that accurately reflects the

fighting ability of others in the network.

Communication networks and fighting: future considerations

Eavesdropping in aggressive contexts: influence of individual differences

Most investigations of networking phenomena in fishes employ a ‘sym-

metrical’design; that is, all participants (e.g. fighters and bystanders) are matched

for attributes such as size or previous social experience. This type of design mini-

mizes the effect of extraneous variables on the dynamics of contests involving the

bystander and has helped to pinpoint how eavesdropping modulates bystander

behaviour in X. helleri (Earley & Dugatkin, 2002) and B. splendens (Oliveira et al.,

1998). However, the symmetrical design is probably not representative of com-

munication networks in nature where substantial individual variation in size,

previous fighting experience, physiological condition and social status are likely



to exist. Individual variation in characteristics related to physical prowess may

have important consequences for how the benefits and costs of fighting and/or

observing contests are perceived and, thus, how the effects of eavesdropping are

manifest. For instance, dominant and subordinate members of a social hierarchy

may respond differently to the winners and losers of observed interactions. Main-

tenance of dominance status is likely a priority for dominant individuals while

increasing status may benefit subordinates. Therefore, dominants should respond

more vigorously to individuals that pose the greatest threat of rank usurpation

(i.e. winners), while subordinates should respond more aggressively to losers so as

to take advantage of opportunities to increase rank. Similarly, territorial defence

is essential for resident individuals while non-territorial individuals may be most

interested in seizing a territory of their own. In this case, residents may respond

most aggressively toward upstart winners (Naguib & Todt, 1997; Naguib et al., 1999;

Peake et al., 2001) while intruders should exploit recent losers.

In addition to mediating the ‘direction’ of a bystander’s agonistic response,

differences in individual perceptions of the costs and benefits of fighting may

influence the degree to which eavesdropping is utilized as an assessment strategy

(Johnstone, 2001). The relationship between the benefits and costs of fighting may

be perceived as high for consistent winners, intermediate for inexperienced ani-

mals and low for consistent losers. Given that eavesdropping is most advantageous

under circumstances where combat bears a relatively high cost (low benefit to cost

ratio), inexperienced animals and consistent losers may benefit most by observ-

ing fights. However, as an individual’s perception of combat costs increases past

a certain threshold, it may refrain from aggressive interactions and eavesdrop-

ping altogether (e.g. playing ‘dove’; Johnstone, 2001). An individual’s perception

of the costs and benefits of fighting can be influenced by a host of additional fac-

tors including size, status, ownership, physiological or immunological condition

(e.g. hungry versus satiated; healthy versus weak), reproductive state and resource

value. Whether animals eavesdrop or how they respond to observing fights may

be integrally related to each of these factors. Though the state dependency of

eavesdropping effects has not yet gained empirical attention, this is likely to be

an important avenue of future research in the area of animal communication

networks.

Eavesdropping in aggressive contexts: environmental influences

Just as factors associated with physical prowess can affect eavesdropping,

so too can environmental or population-based variables. The presence of preda-

tors, or cues indicative of predator presence, influences the frequency of aggres-

sive interactions in fishes (e.g. Martel & Dill, 1993; Wisenden & Sargent, 1997;

Ch. 23). Individuals engaging in aggressive contests become more conspicuous to



predators; that is, the costs of fighting are increased considerably when predators

are around. Predation risk may also have a negative impact on the frequency of

eavesdropping via at least three potential mechanisms: (a) if fights are less com-

mon in the presence of a predator, (b) if observing fights makes the bystander more

conspicuous to predators (e.g. by association with the fighters), or (c) if the capacity

to dedicate simultaneous attention to fights and predators is limited (see Dukas

(2002) for review on selective attention). Although the frequency of eavesdropping

may decline with predation risk, the efficacy of eavesdropping as an assessment

strategy would not necessarily be compromised. In fact, when aggressive contests

are rare and information flow through the network is reduced, bystanders may

take every opportunity to extract information about the fighting abilities of oth-

ers. In this sense, predation risk decreases the frequency and intensity of aggres-

sive encounters and, by necessity, the opportunities to eavesdrop and the absolute

amount of information available. However, the net benefit of eavesdropping under

these circumstances may remain unaffected (or may even be increased).

To test whether the frequency of eavesdropping is modified under different pre-

dation pressures would entail an analysis of how the proportion of eavesdroppers

in a population changes across predation regimes or how predation risk affects

an individual’s propensity to eavesdrop. Mathematical models may be best suited

to address population-level questions (e.g. how the proportion of eavesdroppers

changes with predation risk). Questions more amenable to empirical study include

whether predation risk affects the extent to which bystanders gather information

or whether bystanders compromise information acquisition in order to remain

hidden from predators. For instance, one could compare whether bystanders

derived from high-predation and low-predation sites differ in their response to

the watched contestants. In addition, one could vary the quality or availability of

refuges, the degree of habitat heterogeneity or the distance of refuges from the

focal fight and subsequently quantify how the bystander’sresponse to the observed

contestants changes with environmental condition.

Besides predation risk, the social and/or mating system of the species in ques-

tion may have a significant impact on how, and to what extent, exposure to fights

alters a bystander’s agonistic decisions. The challenge hypothesis postulated that

individuals should respond to social instability with increased testosterone levels

and, presumably aggression levels, to deter rival males and secure reproductive

opportunity (Wingfield et al., 1990; Ch. 21). In the broad sense, ‘social instability’

could include instances where individuals are being challenged directly (e.g. in

territorial disputes) or where individuals are exposed to but not directly engag-

ing in aggressive interactions. Mozambique tilapia and Siamese fighting fish fit

nicely within this broad scope, as males of these species respond to observing fights

with increased 11-ketotestosterone levels or increased aggression levels (Oliveira



et al., 2001; Clotfelter & Paolino, 2003). However, we failed to uncover evidence

that general changes in agonistic motivation accompany eavesdropping in green

swordtail fish (Earley et al., 2004). We argued that differences in the mating system

(e.g. breeding seasonality; monogamous versus polygamous) and/or social system

(e.g. territorial versus hierarchical; stable versus unstable) could explain appar-

ent species-specific differences in the response to observing contests (Earley et al.,

2004; Ch. 21 has a more in-depth treatment).

A piece of the hierarchical puzzle?

Eavesdropping is inherently a social phenomenon because it requires at

least two individuals actively engaging with one another and a third party that

extracts information from the signalling interchange. Nevertheless, there have

been no controlled studies in fishes that examine aspects of networking above

and beyond its effects at the dyadic level (for birds see Peake et al., 2002; Ch. 15).

For example, in our work with X. helleri, we exposed a previously inexperienced

bystander to an aggressive interaction and then assessed its response toward one of

the contestants by staging dyadic contests (bystander versus winner or bystander

versus loser). However, is it possible that the effects of eavesdropping are manifest

differently when bystanders interact with a previously observed individual in the

presence of other network members (e.g. observed winners or losers, unknown

conspecifics, previous opponents that defeated or were defeated by the bystander;

see Ch. 4)? Also, since social groups provide the opportunity for a wide range of

interactions, each individual may have a different blend of prior social experience

(e.g. several winning, losing or eavesdropping experiences or any combination

thereof). Could the ways in which several social experiences are integrated over

time, and the mere presence or absence of winner, loser and eavesdropping effects,

have implications for the structure of animal social systems? From an empirical

standpoint, these questions remain unanswered. However, the role of bystanders

in the establishment of linear dominance hierarchies has been a question of con-

ceptual interest for quite some time and, more recently, has attracted theoretical

attention.

Chase (1980, 1982) developed a conceptual model of linear hierarchy formation,

the ‘jigsaw model’,that involved two interacting individuals and a bystander. Once

a dominance relationship was established between the initial contestants, four

possible interaction sequences could follow: initial dominant defeats bystander

(double dominance), initial subordinate submits to the bystander (double subor-

dination), bystander defeats initial dominant, or bystander submits to the initial

subordinate. Double dominance and double subordination most often led to the

establishment of linear hierarchies while the remaining two interaction sequences

generated intransitive dominance orders.



Chase (1980, 1982) did not address explicitly the behavioural mechanisms

responsible for the double dominance and double subordination sequences and,

thus, the genesis of linear hierarchies. There are at least two mechanisms that

could give rise to such sequences. On the one hand, the initial contestants may

update their estimation of their own fighting ability after having won or lost, i.e.

winners increase and losers decrease their perceived fighting ability (Hsu & Wolf,

2001). As a consequence, previous winners should be more likely to defeat a

bystander with average fighting ability and losers should be more likely to defer to

the same bystander. In this instance, winner and loser effects are the behavioural

mechanisms responsible for the double dominance and double subordination

sequences and, in turn, the formation of linear hierarchies. On the other hand,

the bystander may update its perception of the fighting ability of the initial con-

testants after observing the fight. For instance, the bystander may increase its

estimate of the winner’s fighting ability and decrease its perception of the loser’s

prowess. As a result, the bystander may be liable to attack the loser and submit

to the winner. In this case, eavesdropping is the behavioural mechanism that

generates the double dominance and double subordination sequences and linear

hierarchies. These two scenarios need not be mutually exclusive. That is, winner,

loser and eavesdropping effects may act in concert to promote linear hierarchy

formation in animal groups.

Dugatkin (2001) developed a simulation model to illustrate the potential inter-

actions between eavesdropping and prior-experience effects in shaping dominance

hierarchies. In this model, individuals could increase or decrease their own fight-

ing ability and that of others in the network. The model made two assumptions:

the first was that winner, loser and eavesdropping effects change individual per-

ceptions with equal magnitude; the second was that all bystanders were privy to

every interaction that occurred in their social environment. When eavesdropping

and prior experience were considered separately, only winner effects produced

a linear hierarchy. Conversely, when eavesdropping and prior experience effects

operated simultaneously, a linear hierarchy always emerged. Therefore, linear

hierarchies are most likely to occur when some combination of winner, loser and

eavesdropping effects operate.

In order to improve understanding of how these effects operate in natural sys-

tems, some of the assumptions employed by Dugatkin (2001) need to be relaxed. For

at least two reasons, all bystanders are probably not capable of observing every

interaction that occurs in their social network: first, time spent observing one

interaction interferes with a bystander’sability to observe additional contests and,

second, bystanders are unlikely to be within signal detection range of all aggres-

sive encounters. Moreover, eavesdroppers probably have less information about

the fighting ability of each observed contestant than the contestants themselves.



Therefore, updates of others’ fighting ability may be of lesser magnitude than up-

dates of one’s own fighting ability. R. L. Earley, S. Brosnan & J. Bragg (unpublished

data) developed a simulation model of hierarchy formation in animals, in part to

examine the consequences of relaxing these assumptions (Ch. 26).

Several signalling modalities: seeing is not everything

Our work on eavesdropping and aggression examined the influence of

visual signals on a bystander’s future behaviour. Nevertheless, visual signals may

convey only part of the story in piscine duels. Acoustic (Lugli, 1997; Ladich, 1998;

Amorim & Hawkins, 2000; Thorson & Fine, 2002), chemical (Waas & Colgan, 1992;

Giaquinto & Volpato, 1997) and electrical (McGregor & Westby, 1992) stimuli have

all been implicated as potential modes of communication in aggressive contests

in fishes. Whether these signals elicit similar changes in bystander behaviour as

visual signals has yet to be tested. Thorson & Fine (2002) demonstrated that male

gulf toadfish Opsanus beta emit acoustic signals during the calls of neighbouring

males, a phenomenon they called ‘acoustic tagging’ and interpreted as an aggres-

sive display. If overlapping versus non-overlapping acoustic signals in toadfish pro-

vide information about willingness to escalate (or putative status), then playback

experiments such as those used in territorial bird systems (e.g. Peake et al., 2001;

Ch. 2) may be worth conducting, provided an anechoic aquatic chamber can be

developed. Insights into multimodal signalling, the transmission of these signals

within the network and the availability of such signals to bystanders will surely

weave a more comprehensive story of how communication networks operate in

nature.

Social eavesdropping and female mate choice

As a general rule, females invest more time and energy in the reproduc-

tive process than males (e.g. production of viable eggs, gestation, maternal care,

etc.) and, therefore, should be the choosier of the two sexes. Over the past several

decades, an abundance of conceptual, theoretical and empirical work has focused

on the factors that mediate female mate choice or male success in attracting

females (Ryan, 1997). Most of the female mate-choice models have investigated

how exaggerated male secondary sex characters and female preferences for these

characters evolve, through either direct or indirect selection. For instance, Fisher’s

runaway selection hypothesis postulated that, over evolutionary time, the alleles

responsible for the male trait and the female’spreference for the male trait become

genetically correlated (Fisher, 1958). This genetic linkage initiates a positive feed-

back loop whereby male traits can become more exaggerated as the preference for



such traits strengthens and, in turn, further exaggeration of male traits intensifies

female preference.

Fisher’s genetic model, together with alternative models that address direct

(e.g. sensory bias) and indirect (e.g. good genes) selection on female preferences

(Ryan, 1997), have helped to elucidate how exaggerated male traits exist when they

appear to have a negative impact on survival and how strong female preferences

for these traits arise and persist. However, almost all sexual selection models

assume that females choose mates independent of the choices made by other

females (an exception is Kirkpatrick & Dugatkin, 1994; also see below). Is it possible

that female choice depends not only on intrinsic preferences but also on the

preferences of other females in the network? Two lines of evidence suggest that

the decisions a female makes with respect to choosing mates is influenced in large

part by observing interactions. First, monitoring apparent male–male interactions

alters female mating decisions, for example initial mate choice (Doutrelant &

McGregor, 2000) or loyalty to partner (Otter et al., 1999; Mennill et al., 2002; Ch. 7).

Second, observing male–female courtship and/or mating interactions influences

the subsequent mate choice decisions of a female peripheral to the interaction

(e.g. Dugatkin, 1992, 1996a; Dugatkin & Godin, 1992; Grant & Green, 1996; Witte &

Ryan, 1998; Witte & Noltemeier, 2002).

In this section, we focus on the latter aspect of networking in poeciliid fishes:

namely, how intersexual courtship rituals mediate the mating decisions of female

P. reticulata that are not directly involved in the interaction. The principal con-

cept linking communication networks to courtship interactions and female mate

choice is mate copying. Mate copying occurs when “the conditional probability

of choice of a given male by a female is either greater or less than the absolute

probability of choice depending on whether that male mated previously or was

avoided, respectively” (Pruett-Jones, 1992, p. 1001). Furthermore, the female must

obtain information about a male’s mating history (or some part of it) by observa-

tion (Dugatkin, 1996b). In other words, the information gained by eavesdropping

on mating interactions may sway a female’sdecision toward or away from mating

with the observed male. We confine our discussion to unambiguous cases of mate

copying, that is, where a shift in the mating decisions of females is based solely

on observing interactions between males and females other than oneself.

Spectators in guppy networks: empirical work

Cast of characters

Guppies are an ideal species for examining networking phenomena such

as mate copying for a number of reasons. First and foremost, guppies live in mixed-

sex shoals, within which females likely have opportunities to view (and potentially



copy) the mate choice of nearby conspecifics. In addition, ample evidence suggests

that social information is utilized by guppies in the context of mate choice (see

below), foraging (Laland & Williams, 1997; Laland & Reader, 1999) and in the dy-

namics of shoal motion (Lachlan et al., 1998). Mate choice has been studied exten-

sively in this species (for reviews see Kodric-Brown, 1990; Endler & Houde, 1995;

Houde, 1997) and guppies exhibit normal courtship behaviour when placed in

small aquaria; therefore, they are ideal for manipulative laboratory experiments.

Typical courtship interactions involve the male directing sigmoid displays at the

female; receptive females respond to male displays with a ‘gliding’ motion (for

a more complete description, see Liley (1966)). Lastly, female guppies from the

Paria river (the population that was used in the experiments described below) ex-

hibit heritable preferences for certain male traits (e.g. orange colour: Houde, 1987,

1988; Endler & Houde, 1995). Experiments on such populations provide a unique

opportunity to examine the interaction between genetic-based preferences and

those resulting from interactions within the guppy social network. The cast of

characters in the guppy network resembles that described for the swordtails. The

principal difference is that the network is partitioned into (a) males that are either

quite similar with respect to exaggerated colour patterns or that differ by varying

degrees; (b) at least one female being courted by, or exhibiting a preference for,

a focal male; and (c) a female bystander within range to detect courtship and/or

mating signals (Fig. 5.4). In theory, any of the females within the guppy network

can assume a bystander role or a courtship role at any given time; males however,

are restricted to a courtship role.

Basic paradigm

In the guppy network, we are concerned primarily with how a female

bystander responds to a male that was recently preferred by another female (here,

the focal female). To address this question, one of the authors (Dugatkin) devel-

oped a protocol where a female was either exposed or not exposed to a courtship

interaction between a male and a focal female. Following the observation period,

the female bystander was given the opportunity to make a choice between the

male that was preferred by the focal female and the male that was not preferred

(Dugatkin, 1992). The preference of the focal female was ‘staged’ because she was

restricted to the side of the aquarium occupied by one of the two males; therefore,

the bystander female observed an apparent choice by the focal female. Eavesdrop-

ping on the mate choice of others could have three potential effects on the future

behaviour of the female bystander (Fig. 5.4). If observing courtship interactions

has no effect (0) on the female bystander then she should choose both males with

equal frequency. If females increase their assessment of a male’s quality (+) after

observing him successfully court, then the bystander should choose the male that



Brightly coloured male Relatively drab male

Brightly coloured male Model female Relatively drab male

Bystander

+ / 0 /− + / 0 /−

Fig. 5.4. The individuals comprising guppy networks and the hypothetical effects that

watching mating interactions could have on a female bystander’s response toward the

observed, apparently successful male. After witnessing a courtship interaction

(dashed lines), the female bystander can respond in a variety of ways toward males in

her social network (solid lines). Effect on bystander behaviour: 0, no effect; −,

decreased assessment of male quality; +, increased assessment of male quality.

was preferred by the focal female a significant majority of the time, i.e. mate copy-

ing. It is also possible that females decrease their assessment of a male’squality (−)

after observing a courtship interaction, for instance if a recent mating depletes

the male’s sperm supply (Nakatsuru & Kramer, 1982). In this case, the bystander

female should avoid the male that was recently preferred by the focal female. In

the following sections, we provide a general overview of the empirical work on

mate copying that has been conducted in P. reticulata, with special emphasis an

how eavesdropping mediates a female’s subsequent mating decisions.

Female mate copying: does it occur?

Dugatkin’s (1992) research on guppies provided the first controlled study

of female mate copying. In this study, the bystander female chose the male that



was preferred by the focal female in a significant proportion of the trials. In a

series of five additional experiments, Dugatkin (1992) ruled out a host of alterna-

tive explanations, including female biases toward one side of the aquarium over

the other, female preference for areas of the tank recently occupied by the largest

group of fish (‘schooling preference’) and female choice for males that courted

most recently and thus were more active. These findings provided substantial evi-

dence that females do copy the mate choice of others and that female bystanders

assess the quality of recently preferred males to be higher than those that were

not preferred. Moreover, Dugatkin (1992) demonstrated that courtship inter-

actions between the focal female and a male are crucial for eliciting mate copying

behaviour. These results support the notion that mate choice decisions in P. reticu-

lata are, in part, socially modulated. Two other studies (Brooks, 1996; Lafleur et al.,

1997) found no evidence of mate copying in guppies. It is critical to note, how-

ever, that neither of these studies used guppies from natural streams in Trinidad.

Nonetheless, support for mate copying has also been reported in other poeciliid

species (Poecilia latipinna; Witte & Ryan, 1998) and the Japanese medaka Oryzias

laticeps (Grant & Green, 1996). Given the pivotal role of male–female interactions

in mediating a female bystander’s future mating decisions, the social system of

guppies and other species in which there is unambiguous evidence for mate copy-

ing are clearly amenable to interpretation from the perspective of communication

networks.

Socially modulated versus intrinsic preferences

In the absence of mate copying opportunities, female guppies distinguish

between males based on a number of phenotypic traits, for example tail size

(Bischoff et al., 1985) or colouration patterns (Houde, 1988; Houde & Endler, 1990;

Endler & Houde, 1995). Furthermore, female preferences for male traits such as

orange colouration have a significant heritable component (Houde, 1988; Houde

& Endler, 1990). Because female preferences in guppies are shaped by both ge-

netic and social factors, an interesting question is whether and how these factors

interact. Dugatkin & Godin (1992) conducted an experiment where a female was

initially allowed to choose between two males that differed only in their colour pat-

tern. Following this choice, the female was either exposed or not exposed to a focal

female placed beside the male not chosen in the initial preference test. After the

observation period, the female was again allowed to choose between the same two

males. Interestingly, the female reversed her choice in 75% of the trials, suggesting

that social factors (i.e. eavesdropping on the apparent preference of another female

for the less preferred male) are capable of overriding intrinsic preferences.

Dugatkin (1996a) employed a similar protocol but systematically altered the

asymmetries in orange colouration between the two males provided to the female



Prop

ortio

n of

tim

es f

emal

e pr

efer

red

mor

e-or

ange

mal

e

Mean difference in total orange body colour between males

Fig. 5.5. The proportion of times female bystanders chose the more orange of the two

males in the presence (solid squares) or absence (open circles) of a focal female.

(Adapted from Dugatkin, 1996a.)

bystander. When there was no opportunity for mate copying, females exhibited a

more pronounced preference for males with more orange as the asymmetries in

colouration increased (Fig. 5.5). However, when a focal female was placed beside

the male with less orange, so as to simulate an apparent preference for drab

males, the female bystander chose the less-orange male significantly more often

in all cases except when the asymmetry was most drastic (40% difference in total

orange body colouration). Even when substantial asymmetries in male colouration

existed, a female bystander could be coerced into choosing the less-orange male by

increasing his perceived attractiveness via social manipulation (Dugatkin, 1998).

This was accomplished by increasing the number of focal females that exhibited

an apparent preference for the drab male or by increasing the amount of time

a single focal female spent near the less-orange male. Witte & Noltemeier (2002)

obtained strikingly similar results in female sailfin mollies P. latipinna, a related

poeciliid species where females presumably exhibit a genetically based preference

for large males (Marler & Ryan, 1997). Witte & Noltemeier (2002) also demonstrated

that reversals in mate choice incited by a copying bout persisted for long periods



of time (five weeks), even when a female was allowed to choose between previously

unknown large and small males. This marks one potential direction for the guppy

work: to determine whether observing courtship bouts influences female choice

even in the absence of the observed, previously chosen male.

In all, these results provide compelling evidence that observing courtship inter-

actions is sufficient to overturn female poeciliid’s genetically predisposed choice.

Furthermore, as the amount of social information available to the female by-

stander about a male’s potential quality increases, the more apt she is to rely on

social signals in lieu of the preference algorithm engrained in her genes. It is

important to note that in all of the studies on mate choice copying the observed

female was placed near a male of lesser quality (e.g. smaller or with less orange).

The low-quality male was then considered to be a suitable mate by virtue of his

being chosen earlier. This type of design is necessary to decouple mate choice copy-

ing from established, genetically based preferences. In nature, however, female

bystanders likely observe other females choosing relatively high-quality mates.

Therefore, mate choice copying is likely to reinforce rather than contradict pre-

existing preferences (Brooks, 1996).

Mate copying: a theoretical perspective

Relative to eavesdropping in an aggressive context, mate choice copying

has received an abundance of theoretical attention. Mathematical treatments of

mate copying have addressed two principal evolutionary questions: first, how es-

tablished mate copying strategies influence the evolution of female preferences

and male secondary sex characteristics or the variance in male mating success

(Wade & Pruett-Jones, 1990; Kirkpatrick & Dugatkin, 1994; Laland, 1994; Agrawal,

2001) and, second, the emergence and persistence of copying strategies (Losey

et al., 1986; Pruett-Jones, 1992; Dugatkin & Hoglund, 1995; Servedio & Kirkpatrick,

1996; Stohr, 1998). An exhaustive comparison of these models is beyond the scope

of this chapter. However, mate copying theory has revealed that network phe-

nomena, in particular the behavioural modifications that result from observing

social interactions, can have both short-term effects on individual decision mak-

ing and substantial evolutionary consequences. For instance, established mate

copying strategies can increase the variance in male mating success and, thus,

the opportunity for selection on male phenotypic traits that correlate well with

attractiveness (Wade & Pruett-Jones, 1990). When bystanders most often observe

interactions between females and males that exhibit the most common pheno-

type and when the effects of observing courtship interactions are independent

of male phenotype, mate copying impedes the spread of rare (or novel) male

traits (Kirkpatrick & Dugatkin, 1994; Laland, 1994; Agrawal, 2001). Interestingly,

when eavesdropping has a graded effect on bystander behaviour (e.g. depending



on the phenotype of the observed male), a number of additional scenarios

emerge, including the potential for the spread of rare male phenotypes (Agrawal,

2001).

Given that mate copying can have dramatic effects on the distribution of male

traits in a population, it is important to understand how mate copying strategies

could emerge in the first place. A recurrent theme in the literature in this area,

particularly in models that assume female choice is under direct selection (but see

Servedio & Kirkpatrick, 1996), is that copying strategies will thrive when female

choice is costly (e.g. sampling costs, search time, predation risk: Pruett-Jones 1992)

or when substantial differences exist in females’ability to discriminate low- versus

high-quality males (Stohr, 1998). In the next section, we elaborate on how the costs

and benefits of female mate choice (or copying) potentially influence the degree

to which copying is used as a mate-assessment strategy.

Communication networks and mate choice: future directions

Eavesdropping and mate choice: influence of individual differences

In most studies on mate choice copying, the bystander and focal female

are matched for characteristics such as size, age and previous mating experience

(but see Dugatkin & Godin, 1993; Witte & Ryan, 1998). Under natural circum-

stances, females involved in the interaction are likely to differ in some respect.

Individual differences may influence how observation of courtship interactions

modulates bystander behaviour by adjusting the expected costs and benefits of

mate copying or mate choice itself. Dugatkin & Godin (1993) demonstrated that

small (young) females copy the mate choice of large (old) females, but not vice

versa. If older females have more mate choice experience and if experience de-

creases the rate at which errors in mate choice are committed (e.g. choosing a

poor-quality mate), then older females are liable to be better at discriminating

low- from high-quality mates than younger females. Therefore, the fitness benefit

of copying older females, namely having a higher probability of mating with supe-

rior males, may be quite high for young females. Conversely, the costs of copying a

younger, error-prone female may be sufficiently high to discourage mate copying

in older females. A host of other variables, including gravidity, previous mating

experience independent of age and physiological condition may affect how the

benefits and costs of mate choice and/or mate copying are perceived. For instance,

if female mate choice entails substantial sampling or search costs, individuals

who are under significant time constraints (e.g. hungry individuals whose time

would better be spent foraging) should be more likely to rely on the choices of

others. As with eavesdropping in an aggressive context, the state dependency of



mate copying represents an understudied and potentially important aspect of this

field.

Eavesdropping and mate choice: environmental influences

Environmental factors may also modify an individual’spropensity to copy

the choice of other females. Predation risk is known to influence female choosiness

and this is likely because females encounter a trade-off between the benefits of

remaining vigilant toward predators and the costs of spending time searching for,

or assessing, potential mates (Magnhagen, 1991; Pocklington & Dill, 1995). As the

costs of female choice intensify, mate copying should become increasingly bene-

ficial; therefore, under high predation risk, we might expect females to increase

their propensity to copy other females. However, this also assumes that bystanders

are less conspicuous to predators than individuals involved in active mate choice.

Briggs et al. (1996) found little support for this hypothesis in guppies derived from

streams with relatively high predation risk; apparent predation risk did not influ-

ence the proportion of females reversing their choice in the presence of a focal

female. However, as Briggs et al. (1996) acknowledged, females from populations

derived from streams with a high predation risk may not exhibit differential mate

copying responses under different predation regimes because, given the high cost

to female mate choice under natural circumstances, they may already exhibit the

maximal propensity to copy. Although Briggs et al. (1996) discarded this thesis,

they did not test whether female guppies derived from low-predation streams ex-

hibit divergent responses in the presence or absence of a predator. Nevertheless,

their work marks one of the first attempts to broach a largely unexplored area of

communication networks: namely how predation risk can modulate individual

tendencies to eavesdrop on mating interactions. The probability or frequency for

females to copy the choices of others may also be influenced by temporal factors

(e.g. early vs. late in the breeding season: Dugatkin & Hoglund, 1995), the quality

of one’s own mate, the availability of refuge or the opportunity to observe mating

interactions. Identifying influential environmental factors may aid in formulat-

ing a comparative communication network concept and should certainly trigger

empirical work dedicated to distinguishing the relative effects of individual- and

environment-based factors on decision-making processes in animals.

Summary

The aim of this chapter was to illustrate the importance of extending re-

search on mating and aggressive behaviour beyond the dyad and into a broader

social milieu. We have demonstrated that poeciliid fishes are capable of extracting



information from various types of social interaction and integrating this infor-

mation for use in future encounters with the observed individuals. A recurrent

theme in this chapter is how the actual (or perceived) costs and benefits of fighting,

mating or eavesdropping can influence the short-term effects of observing social

interactions and the evolutionary viability of such strategies. This is where mate

copying theory and current, albeit sparse, theory on eavesdropping in an aggres-

sive context intersect. Irrespective of a bystander’s focus, when social interactions

bear a high cost, it pays to be attentive to signalling exchanges between others.

Whether the same selection pressures act on eavesdropping regardless of the con-

text in which it is used remains to be explored theoretically. Nevertheless, we have

attempted to highlight a wealth of factors that could affect the circumstances un-

der which eavesdropping is favoured and how the effects of eavesdropping are

manifest behaviourally. Daunting as the list of candidate influences may be, a

comprehensive understanding of communication networks relies, in part, on our

ability to partition the relative effects of each of these factors using comparative,

theoretical or empirical approaches. Unveiling the complex interactions between

individual (e.g. size, status, age), social (e.g. mating systems) and ecological (e.g.

predation risk, seasonality) variation and bystander behaviour marks a compelling

future direction for the field of communication networks.

Acknowledgements

We express our sincere thanks to Matthew Druen, Trish Sevene-Adams, Meredith McGee,

Michael Boles, Megan Tinsley and Blair Gilliland for their assistance in the laboratory. We are

also grateful to Peter McGregor for extending the invitation to contribute to this book and to

Matthew Grober, Cathleen Drilling and Ed Rodgers for insightful discussion on earlier versions

of this chapter. The work described in this chapter was funded in part by the National Science

Foundation, Sigma Xi, Kentucky Academy of Science, Animal Behavior Society and the American

Livebearers Association.

References

Agrawal, A. 2001. The evolutionary consequences of mate copying on male traits.


Amorim, M. & Hawkins, A. 2000. Growling for food: acoustic emissions during

competitive feeding of the streaked gurnard. Journal of Fish Biology, 57, 895–907.

Bakker, T. C. M. & Sevenster, P. 1983. Determinants of dominance in male sticklebacks

(Gasterosteus aculeatus L.). Behaviour, 86, 55–71.



Bakker, T. C. M., Bruijn, E. F. D. & Sevenster, P. 1989. Asymmetrical effects of prior

winning and losing on dominance in sticklebacks (Gasterosteus aculeatus). Ethology,

82, 224–229.

Beaugrand, J. P., Caron, J. & Comeau, L. 1984. Social organization of small

heterosexual groups of green swordtails (Xiphophorus helleri, Pisces, Poeciliidae)

under conditions of captivity. Behaviour, 91, 24–60.

Beaugrand, J. P., Payette, D. & Goulet, C. 1996. Conflict outcome in male green

swordtail fish dyads (Xiphophorus helleri): interaction of body size, prior

dominance/subordination experience and prior residency. Behaviour, 133,

303–319.

Bischoff, R., Gould, J. & Rubenstein, D. 1985. Tail size and female choice in the guppy

(Poecilia reticulata). Behavioral Ecology and Sociobiology, 17, 252–255.

Briggs, S., Godin, J. G. J., Dugatkin, L. A. 1996. Mate-choice copying under predation

risk in the Trinidadian guppy (Poecilia reticulata). Behavioral Ecology, 7, 151–157.

Brooks, R. 1996. Copying and the repeatability of mate choice. Behavioral Ecology and

Sociobiology, 39, 323–329.

Brosnan, S. F., Earley, R. L., Dugatkin, L. A. 2003. Observational learning and predator

inspection in guppies (Poecilia reticulata). Ethology, 109, 823–833.

Brown, C. & Laland, K. N. 2002. Social learning of a novel avoidance task in the guppy:

conformity and social release. Animal Behaviour, 64, 41–47.

Brown, G. & Godin, J. G. J. 1999. Chemical alarm signals in wild Trinidadian guppies

(Poecilia reticulata). Canadian Journal of Zoology, 77, 562–570.

Chase, I. D. 1980. Social process and hierarchy formation in small groups: a

comparative perspective. American Sociological Review, 45, 905–924.

1982. Dynamics of hierarchy formation: the sequential development of dominance

relationships. Behaviour, 80, 218–239.

Chase, I. D., Bartolomeo, C. & Dugatkin, L. A. 1994. Aggressive interactions and

inter-contest interval: how long do winners keep winning? Animal Behaviour, 48,

393–400.

Clotfelter, E. & Paolino, A. 2003. Winner effect is induced in bystanders to aggressive

contests between male fighting fish. Animal Behaviour, 66, 343–347.



Dukas, R. 2002. Behavioural and ecological consequences of limited attention.

Philosophical Transactions of the Royal Society of London, Series B, 357, 1539–1547.

Dugatkin, L. A. 1992. Sexual selection and imitation: females copy the mate choice of

others. American Naturalist, 139, 1384–1389.

1996a. Interface between culturally based preferences and genetic preferences:

female mate choice in Poecilia reticulata. Proceedings of the National Academy of

Sciences, USA, 93, 2770–2773.

1996b. Copying and mate choice. In: Social Learning in Animals: the Roots of Culture ed.

C. Heyes & G. Galef, New York: Academic Press, pp. 85–105.

1998. Genes, copying, and female mate choice: shifting thresholds. Behavioral

Ecology, 9, 323–327.



2001. Bystander effects and the structure of dominance hierarchies. Behavioral

Ecology, 12, 348–352.

Dugatkin, L. A. & Godin, J. G. J. 1992. Reversal of female mate choice by copying in the

guppy (Poecilia reticulata). Proceedings of the Royal Society of London, Series B, 249,

179–184.

1993. Female mate copying in the guppy (Poecilia reticulata): age-dependent effects.


Dugatkin, L. A. & Hoglund, J. 1995. Delayed breeding and the evolution of mate

copying in lekking species. Journal of Theoretical Biology, 174, 261–267.




Earley, R. L., Tinsley, M. & Dugatkin, L. A. 2003. To see or not to see: does previewing a

future opponent affect the contest behavior of green swordtail males (Xiphophorus

helleri)? Naturwissenschaften, 90, 226–230.

Earley, R. L., Druen, M. & Dugatkin, L. A. 2004. Watching fights does not alter a

bystander’s response towards naıve conspecifics in male green swordtail fish

(Xiphophorus helleri). Animal Behaviour, in press.

Endler, J. A. & Houde, A. E. 1995. Geographic variation in female preferences for male

traits in Poecilia reticulata. Evolution, 49, 456–468.

Enquist, M. & Leimar, O. 1983. Evolution of fighting behaviour: decision rules and

assessment of relative strength. Journal of Theoretical Biology, 102, 387–410.

Enquist, M., Leimar, O., Ljungberg, T., Mallner, Y. & Segerdahl, N. 1990. A test of the

sequential assessment game: fighting in the cichlid fish, Nannacara anomala.

Animal Behaviour, 40, 1–14.

Fisher, R. A. 1958. The Genetical Theory of Natural Selection, 2nd edn. New York: Dover.

Franck, D., Klamroth, B., Taebel-Hellwig, A. & Schartl, M. 1998. Home ranges and

satellite tactics of male green swordtails (Xiphophorus helleri) in nature. Behavioural

Processes, 43, 115–123.

Giaquinto, P. C. & Volpato, G. L. 1997. Chemical communication, aggression, and

conspecific recognition in the fish Nile tilapia. Physiology and Behavior, 62,

1333–1338.

Grant, J. & Green, L. 1996. Mate copying versus preference for actively courting males

by female Japanese medaka (Oryzias latipes). Behavioral Ecology, 7, 165–167.

Haller, J. 1995. Biochemical background for an analysis of cost–benefit interrelations

in aggression. Neuroscience and Biobehavioral Reviews, 19, 599–604.

Halperin, J., Giri, T., Elliot, I. & Dunham, D. 1998. Consequences of hyper-

aggressiveness in Siamese fighting fish: cheaters seldom prosper. Animal


Hannes, R. P., Franck, D. & Leimann, F. 1984. Effects of rank-order fights on whole-body

and blood concentrations of androgens and corticosteroids in the male swordtail

(Xiphophorus helleri). Ethology, 65, 53–65.

Hollis, K. L., Dumas, M. J., Singh, P. & Fackelman, P. 1995. Pavlovian conditioning of

aggressive behavior in blue gourami fish (Trichogaster trichopterus): winners become

winners and losers stay losers. Journal of Comparative Psychology, 109, 125–133.



Houde, A. E. 1987. Mate choice based on naturally occurring colour-pattern variation

in a guppy population. Evolution, 41, 1–10.

1988. Genetic differences in female choice between two guppy populations. Animal

Behaviour, 36, 510–516.

1997. Sex, Colour and Mate Choice in Guppies. Princeton, CT: Princeton University Press.

Houde, A. E. & Endler, J. A. 1990. Correlated evolution of female mating preference

and male colour pattern in the guppy, Poecilia reticulata. Science, 248, 1405–1408.

Hsu, Y. & Wolf, L. L. 1999. The winner and loser effect: integrating multiple

experiences. Animal Behaviour, 57, 903–910.

2001. The winner and loser effect: what fighting behaviours are influenced? Animal

Behaviour, 61, 777–786.

Jakobsson, S., Brick, O. & Kullberg, C. 1995. Escalated fighting behaviour incurs

increased predation risk. Animal Behaviour, 49, 235–239.

Johnstone, R. 2001. Eavesdropping and animal conflict. Proceedings of the National


Kirkpatrick, M. & Dugatkin, L. A. 1994. Sexual selection and the evolutionary effects of

copying mate choice. Behavioral Ecology and Sociobiology, 34, 443–449.

Kodric-Brown, A. 1990. Mechanisms of sexual selection: insights from fishes. Annales

Zoologici Fennici, 27, 87–100.

Lachlan, R. F., Crooks, L. & Laland, K. N. 1998. Who follows whom? Shoaling

preferences and social learning of foraging information in guppies. Animal

Behaviour, 56, 181–190.

Ladich, F. 1998. Sound characteristics and outcome of contests in male croaking

gouramis (Teleostei). Ethology, 104, 517–529.

Lafleur, D. L, Lozano, G. A. & Sclafani, M. 1997. Female mate-choice copying in guppies,

Poecilia reticulata: a re-evaluation. Animal Behaviour, 54, 579–586.

Laland, K. N. 1994. Sexual selection with a culturally transmitted mating preference.

Theoretical Population Biology, 45, 1–15.

Laland, K. N. & Reader, S. M. 1999. Foraging innovation in the guppy. Animal Behaviour,

57, 331–340.

Laland, K. N. & Williams, K. 1997. Shoaling generates social learning of foraging

information in guppies. Animal Behaviour, 53, 1161–1169.

Liley, N. R. 1966. Ethological isolating mechanisms in four sympatric species of

poeciliid fishes. Behaviour (Supplement), 31, 1–197.

Losey, G., Jr Stanton, F., Telecky, T., Tyler III, W. and the Zoology 691 Graduate

Seminar Class 1986. Copying others, an evolutionary stable strategy for mate

choice: a model. American Naturalist, 128, 653–664.

Lugli, M. 1997. Response of male goby, Padogobius martensii, to aggressive sound

playback following pre-experimental visual stimulation. Behaviour, 134,

1175–1188.

Magnhagen, C. 1991. Predation risk as a cost of reproduction. Trends in Ecology and

Evolution, 6, 183–186.

Magurran, A. E. & Pitcher, T. J. 1987. Provenance, shoal size, and the socio-biology of

predator evasion behaviour in minnow shoals. Proceedings of the Royal Society of




Marler, C. A. & Ryan, M. J. 1997. Origin and maintenance of a female mating

preference. Evolution, 51, 1244–1248.

Martel, G. & Dill, L. M. 1993. Feeding and aggressive behaviours in juvenile coho

salmon (Oncorhynchus kisutch) under chemically mediated risk of predation.





McGregor, P. K. & Westby, G. W. M. 1992. Discrimination of individually characteristic

electric organ discharges by a weakly electric fish. Animal Behaviour, 43, 977–986.

McGregor, P. K., Otter, K. & Peake, T. M. 2000. Communication networks: receiver and

signaller perspectives. In: Animal Signals: Signalling and Signal Design in Animal

Communication ed. Y. Espmark, T. Amundsen & G. Rosenqvist. Trondheim: Tapir

Academic Press, pp. 405–416.



see isn’t what you get. Animal Behaviour, 62, 1059–1065.

Mennill, D., Ratcliffe, L. & Boag, P. 2002. Female eavesdropping on male song contests

in songbirds. Science, 296, 873.

Mesterton-Gibbons, M. & Adams, E. 1998. Animal contests as evolutionary games.

American Scientist, 86, 334–341.

Mirza, R., Scott, J. & Chivers, D. 2001. Differential responses of male and female red

swordtails to chemical alarm cues. Journal of Fish Biology, 59, 716–728.





Series B, 266, 537–542.

Nakatsuru, K. & Kramer, D. L. 1982. Is sperm cheap? Limited male fertility and female

choice in the lemon tetra (Pisces, Characidae). Science, 216, 753–755.

Neat, F., Taylor, A. & Huntingford, F. 1998. Proximate costs of fighting in male cichlid

fish: the role of injuries and energy metabolism. Animal Behaviour, 55, 875–882.





fish hormone levels. Nature, 409, 475.

Otter, K., McGregor, P. K., Terry, A. M. R. et al. 1999. Do female great tits (Parus major)


Proceedings of the Royal Society of London, Series B, 266, 1305–1309.

Payne, R. 1998. Gradually escalating fights and displays: the cumulative assessment

model. Animal Behaviour, 56, 651–662.

Payne, R. & Pagel, M. 1997. Why do animals repeat displays? Animal Behaviour, 54,

109–119.







gathered by eavesdropping? Proceedings of the Royal Society of London, Series B, 269,

1925–1929.

Pocklington, R. & Dill, L. M. 1995. Predation on females or males: who pays for bright

male traits? Animal Behaviour, 49, 1122–1124.

Pruett-Jones, S. 1992. Independent versus nonindependent mate choice: do females

copy each other. American Naturalist, 140, 1000–1009.

Pulliam, H. R. & Caraco, T. 1984. Living in groups: is there an optimal group size? In:

Behavioural Ecology. An Evolutionary Approach, 2nd edn. ed. J. R. Krebs & N. B. Davies.

pp. Oxford: Blackwell Scientific, pp. 122–147.

Ranta, E. & Juvonen, S. K. 1993. Interference affects food-finding rate in schooling fish.

Journal of Fish Biology, 43, 531–535.

Ryan, M. J. 1997. Sexual selection and mate choice. In: Behavioural Ecology. An

Evolutionary Approach, 4th edn. ed. J. R. Krebs & N. B. Davies. Oxford: Blackwell

Scientific, pp. 179–202.

Schuett, G. 1997. Body size and agonistic experience affect dominance and mating

success in male copperheads. Animal Behaviour, 54, 213–224.

Servedio, M. & Kirkpatrick, M. 1996. The evolution of mate choice copying by indirect

selection. American Naturalist, 148, 848–867.

Stohr, S. 1998. Evolution of mate-choice copying: a dynamic model. Animal Behaviour,

55, 893–903.

Swaney, W., Kendal, J., Capon, H., Brown, C. & Laland, K. N. 2001. Familiarity facilitates

social learning of foraging behaviour in the guppy. Animal Behaviour, 62, 591–598.

Thorson, R. & Fine, M. 2002. Acoustic competition in the gulf toadfish Opsanus beta:

acoustic tagging. Journal of the Acoustical Society of America, 111, 2302–2307.

Waas, J. R. & Colgan, P. W. 1992. Chemical cues associated with the visually elaborate

aggressive displays of threespine sticklebacks. Journal of Chemical Ecology, 18,

2277–2284.

Wade, M. J. & Pruett-Jones, S. G. 1990. Female copying increases the variance in male

mating success. Proceedings of the National Academy of Sciences, USA, 87, 5749–5753.

Wingfield, J. C., Hegner, R. E., Dufty, A. M. & Ball, G. F. 1990. The challenge hypothesis:

theoretical implications for patterns of testosterone secretion, mating systems,

and breeding strategies. American Naturalist, 136, 829–846.

Wisenden, B. D. & Sargent, R. C. 1997. Anti-predator behaviour and suppressed

aggression by convict cichlids in response to injury-released chemical cues of

conspecifics but not to those of an allopatric heterospecific. Ethology, 103, 283–291.

Witte, K. & Noltemeier, B. 2002. The role of information in mate-choice copying in

female sailfin mollies (Poecilia latipinna). Behavioral Ecology and Sociobiology, 52,

194–202.

Witte, K. & Ryan, M. J. 1998. Male body length influences mate-choice copying in the

sailfin molly Poecilia latipinna. Behavioral Ecology, 9, 534–539.


6

The occurrence and functionof victory displays withincommunication networks

j o h n l . b ow e r

Fairhaven College, Western Washington University, Washington, USA

Introduction

Much recent research has focused on communication that occurs prior

to and during agonistic interactions in animals, leading to theoretical and em-

pirical advances in our understanding of the evolution of signalling before and

during agonistic contests (Maynard Smith, 1982; Huntingford & Turner, 1987;

Bradbury & Vehrencamp, 1998; Johnstone, 2001). However, very little research

has focused on the signalling that occurs at the conclusion of agonistic con-

tests (but see Ch. 10) despite the fact that such signalling may have important

consequences for animals involved in such contests and nearby conspecifics.

Post-contest signals may be given by the winner or loser of a contest or may

occur when there is no clear winner or loser. Such signalling by a winner or

loser may be directed towards the other combatant or may be directed to oth-

ers, such as potential rivals and mates. One type of post-contest signal has been

called a ‘victory display’ (e.g. Bradbury & Vehrencamp, 1998). Here, I define a vic-

tory display as a display performed by the winner of a contest but not by the

loser. In this chapter, I review the known occurrence of victory displays and then

use those examples to explore the functional significance of victory displays. I

first consider functions within the combatant dyad and then expand the view

to consider functions within a communication network. Along the way, I illus-

trate some difficulties in studying victory displays and suggest areas for further

research.


114


The occurrence and function of victory displays 115

Occurrence of victory displays

It is not known how common victory displays are because few researchers

have studied them or have looked for them, and I am unaware of any published

review of these behaviours. Here, I present examples of displays that are plausi-

bly victory displays. These examples come from a literature survey, enquiries to

researchers and requests for information on victory displays posted to listservers

that focus on behavioural ecology or taxonomic groups. Most potential victory

displays discovered were either side notes within papers devoted to other aspects

of behaviour or were anecdotes from researchers. I will then explore the displays

described here to begin the discussion of how one identifies a victory display.

Because of the inefficiency of the methods I was forced to use, I expect that my list

of potential victory displays does not represent all the occurrences in the litera-

ture, and most certainly the list does not reflect the frequency of its occurrence in

the natural world. In the following paragraphs, I begin with the earliest described

victory display (in waterfowl), then consider other bird examples before travelling

a conventional phylogenetic route from arthropods to humans.

Waterfowl

The earliest described victory display is the ceremonie der triumphe in the

greylag goose Anser anser (Heinroth, 1910). This display occurs in several contexts,

but its use as a victory display occurs when males return to their mates and/or fam-

ilies after ritualized or actual contests. Returning males cackle loudly with their

necks extended and wings half-raised. Their mates sometimes join the display,

creating what Lorenz (1965a) considered to be ‘the most impressive vocal display

of the greylag goose’. Similar displays appear to be common in swans and geese.

Examples from the literature include Canada geese Branta canadensis (Radesater,

1974), barnacle geese Branta leucopsis (Bigot et al., 1995) and black swans Cygnus

atratus (Kraaijeveld & Mulder, 2002). In each of these four examples, the displays

are characterized by acoustic (raucous ‘cackling’ vocalizations) and visual (wing

flapping and water splashing) signals that carry far beyond the area in which the

contest took place. In particular, exaggerated rolling of the neck and cackling

occur simultaneously after a male has won an agonistic contest in the greylag

goose (Lorenz, 1965b). In Canada geese, male ‘high intensity cackling’,occurs after

contests (Radesater, 1974). Raveling (1967) found that triumph ceremonies were

performed by both males following some contests, but that only victors performed

ceremonies following severe attacks. In addition, victors gave more prolonged dis-

plays after severe attacks than they did after less-severe attacks. Thus, while graded

expressions of the triumph ceremony may occur in a variety of behavioural con-

texts, the exaggerated and simultaneous head rolling and cackling may be the


116 J. L. Bower

identifying characteristics of a victory display. Furthermore, it may be only af-

ter particularly serious contests that the ceremony can be considered a victory

display.

Other bird groups

In other birds, there are several examples of victory displays. These include

parrots of the genus Trichoglossus, which show post-contest behaviours analogous

to the triumph display in geese (Serpell, 1981) and the ‘bow flipper spread’ given

by winners of contests in the little blue penguin Eudyptula minor (Waas, 1990).

In little blue penguins, aggression commonly occurs between unpaired males

gathered on non-breeding sites within caves that contain a nesting colony. About

10% of agonistic interactions between males escalate to physical fights, which

sometimes caused serious flesh wounds. At the conclusion of such a contest, the

winner typically bows forward with his flippers spread and vocalizes while the

loser remains stationary or retreats.

Some examples involve mainly acoustic signals. For example, a victory display

has recently been described in duetting tropical boubou Laniarius aethiopicus (Grafe

& Bitz, 2004). In playback experiments involving 26 pairs of boubous, pairs sang

one of their 12 shared song types much more often than any other song type

following the cessation of playback. This song type was only rarely sung prior to

playback or during playback, suggesting that the song type functioned as a post-

conflict display. Because only presumptive winners sang the duet and not losers,

the song appears to be a victory display. This song type had unique signal design

features (longer song, higher frequencies, more overlap of male and female notes:

Grafe & Bitz, 2004) and was sung from higher perches and carried farther than

other song types (T. U. Grafe, unpublished data). This is a particularly striking

example of a bird species using a specific song type in a specific behavioural

context.

A second acoustic example of a victory display occurs in song sparrows Melospiza

melodia, in which the winner of a naturally occurring territorial contest (defined

as the bird who remains in the contest area after the contest) increases his song

rate to match the highest song rates sung prior to territorial contests (Table 6.1;

Bower, 2000). During the minute following the conclusion of a contest, the win-

ning bird’ssong rate almost always exceeds those of the other dozen or so males in

the song sparrow neighbourhood (Table 6.2). Like the tropical boubous, winning

song sparrow males typically sang from higher perches after contests ( J. Bower,

unpublished data), making them highly conspicuous to their neighbours and sug-

gesting that neighbours may be intended receivers of the victory display.

A third example of a victory display with a striking acoustic component is the

yodel call of black-throated divers Gavia immer. In a low-density Scottish popula-

tion, yodels were produced by males that had just successfully defended their loch



Table 6.1. Song rates of winners, losers, neighbours and non-neighbours during the

minute before and the minute after the end of territorial contests

Songs/min (mean ± SE)

Category Before After No.a p valueb

Winner 3.08 ± 0.99 6.08 ± 0.45 12 0.02∗

Loser 1.17 ± 0.66 2.25 ± 0.63 12 0.26

Unpaired neighbour 4.06 ± 0.59 3.25 ± 2.14 8 0.18

Paired neighbour 1.07 ± 0.41 1.86 ± 0.83 7 0.40

Unpaired non-neighbour 2.32 ± 0.44 2.36 ± 0.40 12 0.86

Paired non-neighbour 0.78 ± 0.25 1.05 ± 0.39 9 0.40

SE, standard erroraSample sizes vary between tests because not all contests included birds of every category,

but males do not appear more than once in the data.bDifferences were compared with Wilcoxon matched-pairs signed-ranks test. Statistically

significant results are marked with an asterisk.

Table 6.2. Song rates of winners, losers, neighbours and non-neighbours during the

minute following the end of territorial contests

Category Songs/min Category Songs/min No.a p valueb

Winner 6.1 ± 0.5 Loser 2.3 ± 0.6 12 0.002∗

Winner 6.3 ± 0.7 Unpaired neighbour 3.3 ± 0.8 8 0.03∗

Winner 6.0 ± 0.4 Paired neighbour 1.9 ± 0.8 7 0.03∗

Winner 6.1 ± 0.5 Unpaired non-neighbour 2.4 ± 0.4 12 0.02∗

Winner 5.7 ± 0.4 Paired non-neighbour 1.1 ± 0.4 9 0.01∗

aSample sizes vary between tests because not all contests included birds of every category.bDifferences were compared with Wilcoxon matched-pairs signed-ranks test. Statistically

significant results are marked with an asterisk.

against another male, and playback of yodels elicits searching rather than yodels

(Gilbert, 1993).

Arthropods to humans

Victory displays have been described in several arthropod species. Reichert

(1978) described one such display in female funnel-web spiders Agelenopsis aptera,

in which the winners and losers of contests over web ownership showed striking

differences in stereotyped behaviours after the contest. Females who take over

a web, in particular, display behaviours such as biting the web manipulating

prey, circling the web or laying new silk. The performance of these behaviours is


118 J. L. Bower

exaggerated during post-contest periods beyond their non-contest performance.

For instance, the spiders exaggerate their abdomen movements when laying silk

and walk particularly slowly when circling the web. In crickets and their allies,

post-contest stridulation by winners of male–male contest appears to be common

in several species, including Gryllus bimaculatus (Alexander, 1961; Simmons, 1986;

Adamo & Hoy, 1995), Teleogryllus oceanicus (Burk, 1983) and Acheta domesticus (Hack,

1997). In three species of Australian tree wetas (Hemideina crassidens, H. femorata and

H. ricta), winners of male–male contests stridulate after winning contests while

losers do not (Field & Rind, 1992; Field, 2001).

One species of reptile and one amphibian show what appear to be victory dis-

plays. In some lizards, the end of a contest is marked by ritualized positions, which

may communicate the status of winners and losers to others. For instance, in the

pygmy Mulga monitor Varanus gilleni, individuals who win contests end up atop

the loser and attempt to ride the loser until either a new contest occurs or the

two separate and the loser leaves the contest area (Carpenter, 1976). In green frogs

Rana clamitans, territorial males engage in splashing displays after expelling an

intruding male from a territory (Wells, 1978).

In mammals, potential victory displays have been recorded for a number of

species. Natoli & de Vito (1991) and Natoli et al. (2000) reported that some feral

domestic cats Felis catus roll on their back on the ground, exposing their under-

sides, in front of the contest loser. Males that engaged in this behaviour were

highly ranked within the dominance hierarchy of the feral cat groups studied. In

wolves Canis lupus and coyotes Canis latrans, winners of contests often run about

with their tails held high in the air after winning contests ( J. Way, personal com-

munication). In observations of eastern coyotes at the Stoneham Zoo (Stoneham,

MA), Way reported that winners of pinning contests emerge from the contests

with a high and bouncing gait while losers stay low to the ground in a submissive

pose. Antelope territorial males sometimes engage in a scent-marking behaviour

after expelling an intruder from their territory. For instance, hartebeest Alcelaphus

buselaphus sometimes add to a boundary dung pile after contests (L. M. Gosling,

personal communication).

In marine mammals, the term ‘victory squeal’ is used for the vocalization given

within a second of a fish being seized by a bottlenose dolphin Tursiops truncatus

(S. H. Ridgeway, personal communication) or by a white whale Delphinapterus leucas

(Ridgeway & Carder, 1998). Whether this is a victory display by my definition

depends on whether the fish was caught in a competitive situation: analogous to

goal scoring in human examples.

Lastly, it is worth noting that various human behaviours may be considered vic-

tory displays. For instance, in ritualized sporting contests, winners often engage in

conspicuous displays after goals are scored or victories occur: ice hockey players



raise their sticks high after a goal has been scored and football (soccer) players

sometimes take off their shirts and run around the field after scoring a goal. Many

American football players engage in stereotyped displays from ‘spiking’ the ball

against the ground to displays involving dances that are unique to the individual

performing it. At the conclusion of many important team sports events, winning

teams climb on top of each other in ways that would seemingly leave them vulner-

able to aggression from the other team. On a more serious note, victory parades

following armed conflict can be considered a group victory display. It would be

fascinating to know of other examples of victory displays following individual or

group conflict in either adult or juvenile humans.

Functional significance of victory displays

Distinctive design features of victory displays

Before addressing the functional significance of victory displays, I must

address the problem of determining if a victory display has a unique function and

meaning that can be separated from that of signals occurring prior to and dur-

ing contests. One difficulty in answering this question is that many post-contest

signals have similar design features to signals used in other behavioural contexts.

For example, wetas and crickets stridulate and birds sing prior to and during con-

tests as well as after contests, leading one to question whether the meaning of a

post-contest signal is to communicate the end of a contest or whether it is simply

performed in anticipation of the continuation of aggression.

One solution to this problem is to examine whether the context for the post-

contest signalling differs markedly from contexts in which signals with similar

design features are produced. If the sender and potential receivers can determine

that the signal is given in the context of the conclusion of a contest, then one can

surmise that the signal’s meaning in the post-contest context may be specific to

that context. For instance, in song sparrows, the victory display of singing at a

high song rate is similar to the high song rate typical of aggressors before they

initiate contests (Bower, 2000). However, the context in which this post-contest

signal occurs differs from the pre-contest singing in important ways. A major

difference is that at the termination of song sparrow contests, the loser leaves the

area, ending the almost continual chasing and physical contact that marks a song

sparrow territorial contest. Following contests, losers typically remain hidden

and quiet for 30 minutes or more after losing (J. Bower, unpublished data). Thus,

the contest winner and surrounding individuals are likely to be aware very soon

after the contest ends that he has entered a post-contest period in which further

aggression is unlikely to occur in the near future. During most pre-contest periods,

both males sit on perches separated by only a few metres and sing. Consequently,


120 J. L. Bower

the context for the victory display is very different from the context for singing

in other stages of interactions. Further research on victory displays may further

our understanding of how such signalling differs from signals with similar design

features given in other contexts.

A second solution to this problem is to test whether the design features of post-

signal displays differ, even subtly, from similar signals given in other contexts. For

instance, one difficulty in considering the triumph ceremony in geese as a victory

display is that the triumph display is composed of a variable and complex mix

of discrete displays, which occur in a variety of social contexts. For instance, the

rolling of the neck and some degree of cackling occurs even when male greylag

geese only feign attacks on other geese and do not actually engage in a contest.

However, both the rolling and cackling are most exaggerated and occur simul-

taneously only after a male has won a real contest (Lorenz, 1965b). Similarly, in

Canada geese, male ‘high intensity cackling’ is the part of the multifaceted tri-

umph display that occurs most commonly after territorial aggression (Radesater,

1974). Therefore, while graded expressions of the triumph ceremony may occur in

a variety of behavioural contexts, the exaggerated and simultaneous head rolling

and cackling may be specific design features of a victory signal.

In other species, behaviours or signal design features may also differ from sig-

nals used in different contexts. For instance, winners in song sparrow contests

often rise quickly above the thick shrubby vegetation to sing from higher perches

than they typically sing from, increasing the active space of their song and the

number of potential receivers ( J. Bower, unpublished data). High perches may

also facilitate listening (Ch. 3). In playback experiments with the duetting tropical

boubou, pairs typically sang one of their 12 shared song-types following the cessa-

tion of playback (Grafe & Bitz, 2004). This song type is rarely sung outside of the

post-contest context, suggesting that it functions as a victory display. Design fea-

tures of this song type differ significantly from other song types in several ways.

Male and female notes overlap more; male notes reach higher frequencies and

the songs are longer in duration than other song types. So, while victory displays

may at first appear to be very similar to other signals, with closer inspection one

may find that they have specific and unique design features that identify them as

victory displays.

Functions of victory displays

Victory displays could function within the winner–loser dyad or more

widely in the communication network. For example, victory displays could make

the victory more memorable (in the sense of Guilford & Dawkins (1991)) to the

loser of the dyad, to other receivers in a network or both. It is also possible that

the displays have no function and are consequences of mechanisms driving



contests, such as the hormone changes underlying such interactions (e.g. Ch. 21).

While recognizing that function and mechanism (ultimate and proximate) levels

of explanation are complementary rather than alternatives, the detailed features

of victory displays discussed below make it unlikely that they are functionless

by-products of recent aggression.

Functions within the dyad

If victory displays function within the winner–loser dyad, the winner

would be directing the signal to the loser, most likely in an attempt to decrease

the probability that the loser will initiate a new contest. By discouraging the loser

from starting a new contest, the winner might achieve a lasting victory, thus

avoiding the costs of further contests and creating the opportunity to resume

other activities, such as searching for a mate, maintaining a pair bond, vigilance

against other intruders and predators, and feeding and other maintenance tasks.

The examples described above seem to fit into one of two categories.

First, there are victory displays that seem to invite an extension of the contest.

For instance, in the bow flipper spread display in little blue penguins the winner

bows low the ground and spreads his flippers apart. This appears to put the winner

in a position where he is vulnerable to attack from the loser. Likewise, feral cats

that roll over on their back and expose their undersides would seem to be choosing

a position that is vulnerable to further attacks. It is possible that by providing the

loser with a stimulus for attacking the winner just after the loser has retreated

from the contest, the winner helps to crystallize dominance over the loser. At

a mechanistic level, it is possible that future displays by the winner similar to

the victory display used after a contest victory may result in a change of mental

state in the loser. Such associative learning (e.g. Staddon, 1983), in which the

loser associates losing the contest with the signal used during the winner’svictory

display, may function to discourage a contest loser from initiating future contests

with the winner.

Second, there are victory displays in which the winner may give a display that is

energetically or otherwise costly. For instance, in my song sparrow study (Bower,

2000), winners almost always sang at very high rates in the minutes following a

contest. Since singing is a moderately energetically costly behaviour (Oberweger

& Goller, 2001), the ability to sing at high rates immediately following a contest

may advertise the winner’s vigour or quality. This should be especially true since

song sparrow contests are energetically costly, often characterized by almost con-

tinuous chasing and occasional physical fighting for an hour or more, with no

breaks for feeding or resting (Bower, 2000). By singing at high rates just after a

contest ends, a male may be sending an honest signal to the loser that the winner

has sufficient endurance to defend the area he has just secured from subsequent


122 J. L. Bower

challenges, despite the energy costs of the recent contest. Other examples of pos-

sible victory displays that may be metabolically or otherwise costly include post-

victory stridulation in crickets and wetas, circling the web and laying new silk in

funnel-web spiders, splashing displays in frogs and the bouncing gait of eastern

coyotes. As with song sparrows, while victory displays may be relatively short in

duration, and thus not likely to be more than moderately energetically costly, their

occurrence at the end of often long and energetically costly contests may make

them difficult to perform. Consequently, they may function as honest indicators

of aspects of the winner’s quality (for further discussion of handicaps and honest

signalling, see Dawkins (1995)). The ability to perform such a display following a

contest may reduce the chances that the contest loser will re-engage the winner

in a later contest.

Functions within the network

Theory suggests that animals are likely to gain fitness benefits by assess-

ing potential mates and rivals in contest situations and altering their behaviour

according to their assessments (e.g. Cox & Le Boeuf, 1977; McGregor & Dabelsteen,

1996; McGregor & Peake, 2000). If this were so, then one would expect selection

pressure for winners to advertise their victory to other members of a communica-

tion network. Victory displays may inform social eavesdroppers (Ch. 2) and other

members of the network that did not pay attention to the interaction that the

winner has just won a contest. As discussed above, the displays may also provide

further information about the winner’s vigour and/or other measures of quality

by displaying after energetically costly contests.

This possible function of victory displays is distinct from audience effects (Ch. 4)

and considerations of the nature of signals and signalling during interactions (e.g.

private signals (Ch. 3) and the ‘good loser’ hypothesis (Peake & McGregor, 2004)).

However, no studies have been attempted that test whether performing a victory

display modifies the behaviour of network members. At present, investigating the

network function of victory displays has to rely on indirect evidence. One obvious

criterion, of course, is whether conspecifics other than (or as well as) the loser

are able to receive the signal (for similar discussion on the intended receivers for

post-copulatory displays in ducks, see Johnson et al. (2000)). This depends both on

the signal features and the spatial arrangement of the conspecifics. For instance,

of the examples given above, it is least likely that funnel-web spider post-victory

behaviour has a communication network function, since the widely spaced webs

(Reichert, 1978) and reliance on vibration for communication means that rivals

are unlikely to be aware of contests on other webs. Likewise, since varanid lizard

territories tend to be large (R. Earley, personal communication) and the possible

victory display is a visual signal, it is unlikely that rivals in adjoining territories



will witness a contest between two lizards. Therefore, the pygmy Mulga monitor’s

‘back riding’post-contest behaviour is more likely directed to the loser rather than

for another potential rival. However, it is not known how often females might

witness such contests, nor how often other males encroaching on the territory

witness the contest. Therefore, even in territorial species in which territories are

large and signals have relatively short ranges, there may be a network function if

other animals are present on territories.

In contrast to these two examples, the other victory displays described above

occur in species in which potential mates and rivals are likely to witness the

display. Territorial songbirds, for instance, often reside in neighbourhoods, with

several territories constituting a neighbourhood. During the song sparrow territo-

rial contests I studied, I noticed that neighbourhood males and females often flew

up to perches from which they could see the contest ( J. Bower, unpublished data).

Such behaviour seemed to indicate that members of the communication network

were paying close attention to contests in their neighbourhood. Bird song is a

moderately long-range signal, typically carrying into and beyond neighbouring

territories (e.g. Brenowitz, 1982; Ch. 20). Therefore, the post-contest vocalizations

of a winning song sparrow, tropical boubou pair or goose would likely be heard

by other conspecifics in the bird’s neighbourhood. Similarly, stridulating crickets

and wetas, as well as splashing green frogs, all produce acoustic signals that are

likely to be accessible to rivals and potential mates. Visual signals made within

open habitats or by species with close spacing may also be candidates for victory

displays directed at the network. Thus, post-contest scent marking by hartebeest,

the movements of geese and the bouncing gait of eastern coyotes all are accessible

to conspecific receivers in the communication network.

Summary

Victory displays are post-conflict signals given by the winner (but not the

loser) of an agonistic contest. There has been little work specifically addressing

such displays and most of the evidence for their existence comes from incidental

descriptions or asides and footnotes. Much remains to be done to characterize

victory displays and to identify their function, including whether they are network

phenomena or are directed at the loser of the interaction. The difficult but exciting

work that lies ahead is to demonstrate that such displays, in conjunction with or

independent from the social eavesdropping that may occur during a contest, alter

the behaviour of other members of the communication network. It is likely that

such a test could be developed more easily in a laboratory setting than in situ;

one approach would be to prevent observers from watching the contest but allow

them to observe the following victory display.


124 J. L. Bower

Acknowledgements

I thank the many researchers who responded to my call for potential examples of

victory displays, Ulmar Grafe for sharing unpublished data, and three anonymous referees and

Peter McGregor for comments on previous drafts of this chapter.

References

Adamo, S. & Hoy, R. 1995. Agonistic behaviour in male and female field crickets,

Gryllus bimaculatus, and how behavioural context influences its expression. Animal

Behaviour, 49, 1491–1501.

Alexander, R. 1961. Aggressiveness, territoriality, and sexual behaviour in field

crickets (Orthoptera gryllidae). Behaviour, 17, 130–223.

Bigot, E., Hausberger, M. & Black, J. 1995. Exuberant youth: The example of triumph

ceremonies in barnacle geese (Branta leucopsis). Ethology, Ecology and Evolution, 7,

79–85.


in a song sparrow (Melospiza melodia) neighborhood. Ph.D. Thesis, Cornell

University, Ithaca, New York.



Brenowitz, E. A. 1982. Long-range communication of species identity by song in the

red-winged blackbird. Behavioral Ecology and Sociobiology, 10, 29–38.

Burk, T. 1983. Male aggression and female choice in a field cricket (Teleogryllus

oceanicus): the importance of courtship song. In: Orthopteran Mating Systems: Sexual

Competition in a Diverse Group of Insects, ed. D. T. Gwynne & G. K. Morris. Boulder,

CO: Westview Press, pp. 97–119.

Carpenter, C. C. 1976. A further analysis of the combat ritual of the pygmy mulga

monitor, Varanus gilleni (Reptilia: Varanidae). Herpetologica, 32, 35–40.

Cox, C. R. & Le Boeuf, B. J. 1977. Female incitation of male competition: a mechanism

for mate selection. American Naturalist, 111, 317–335.

Dawkins, M. S. 1995. Unravelling Animal Behaviour, 2nd edn. Harlow, UK: Longman.

Field, L. H. 2001. Aggression behaviour in New Zealand tree wetas. In: The Biology of

Wetas, King Crickets, and Their Allies, ed. L. H. Field & T. H. Jarman. Oxford: CAB

International, pp. 333–350.

Field, L. H. & Rind, F. 1992. Stridulatory behaviour in a New Zealand weta, Hemideina

crassidens. Journal of Zoology, 228, 371–394.

Gilbert, G. 1993. Vocal individuality as a census and monitoring tool: practical

considerations illustrated by a study of the bittern Botaurus stellaris and the

black-throated Diver Gavia arctica. Ph.D. Thesis, University of Nottingham, UK.

Grafe, T. U. & Bitz, J. H. 2004. An acoustic victory display in the duetting tropical

boubou (Laniarius aethiopicus): a signal of victory. BMC Ecology, 4, 1.

http://www.biomedcentral.com/bmcecol/.

Guilford, T. & Dawkins, M. S. 1991. Receiver psychology and the evolution of animal

signals. Animal Behaviour, 42, 1–14.



Hack, M. A. 1997. The energetic costs of fighting in the house cricket, Acheta domesticus

L. Behavioral Ecology, 8, 28–36.

Heinroth, O. 1910. Beitrage zur Biologie, namentlich Ethologie und Psychologie der

Anatiden. In: Verhandlungen der V Internationalis Ornithologen Kongressus in Berlin,

pp. 589–702.

Huntingford, F. & Turner, A. K. 1987. Animal Conflict. Cambridge, UK: Cambridge

University Press.

Johnson, K. P., McKinney, F., Wilson, R. & Sorenson, M. D. 2000. The evolution of

postcopulatory displays in dabbling ducks (Anatini): a phylogenetic perspective.


Johnstone, R. A. 2001 Eavesdropping and animal conflict. Proceedings of the National


Kraaijeveld, K. & Mulder, R. 2002. The function of triumph ceremonies in the black

swan. Behaviour, 139, 45–54.

Lorenz, K. 1965a. Here I Am – Where Are You?: The Behavior of the Greyleg Goose. New York:

Harcourt Brace.

1965b. The triumph ceremony of the greylag goose. Philosophical Transactions of the


Maynard Smith, J. 1982. Evolution and the Theory of Games. Cambridge, UK: Cambridge

University Press.






Natoli, E. & de Vito, E. 1991. Agonistic behaviour, dominance rank and copulatory

success in a large multi-male feral cat, Felis catus L., colony in central Rome.


Natoli, E., De Vito, E. & Pontier, D. 2000. Mate Choice in the domestic cat (Felis silvestris

catus L.). Aggressive Behavior, 26, 455–465.

Oberweger, K. & Goller, F. 2001. The metabolic cost of bird song production. Journal of

Experimental Biology, 204, 3379–3388.


Behavior, 32, 114–121.

Radesater, T. 1974. Form and sequential associations between the triumph ceremony

and other behaviour patterns in the Canada Goose Branta canadensis L. Ornis

Scandinavica, 5, 87–101.

Raveling, D. G. 1967. Sociobiology and ecology of Canada Geese in winter. Ph.D. Thesis,

Southern Illinois University, Carbondale, USA.

Reichert, S. E. 1978. Games spiders play: behavioral variability in territorial disputes.


Ridgeway, S. H. & Carder, D. A. 1998. Net-aided foraging by two white whales. Marine

Mammal Science, 14, 332–334.

Serpell, J. 1981. Duets, greetings and triumph ceremonies: analogous displays in the

parrot genus Trichoglossus. Zeitschrift fur Tierpsychologie, 55, 268–283.


126 J. L. Bower

Simmons, L. 1986. Inter-male competition and mating success in the field cricket,

Gryllus bimaculatus (De Geer). Animal Behaviour, 34, 567–579.

Staddon, J. E. R. 1983. Adaptive Behavior and Learning. Cambridge, UK: Cambridge

University Press.

Waas, J. R. 1990. An analysis of communication during the aggressive interactions of

little blue penguins (Eudyptula minor). In: Penguin Biology, ed. LS Davis & JT Darby.

San Diego, CA: Academic Press, pp. 345–376.

Wells, K. D. 1978. Territoriality in the green frog (Rana clamitans): vocalizations and

agonistic behaviour. Animal Behaviour, 26, 1051–1063.

P1: JZZ/... P2: JZZ/...0521823617agg.xml CU1917B/McGregor 0 521 582361 7 April 11, 2005 10:51

Part II T H E E F F E C T S O F P A R T I C U L A R

C O N T E X T S

127


128


Introduction

The rationale behind the grouping of chapters into this section is to

facilitate comparisons between communication networks found in very different

contexts: mate choice, predation, nestling begging, redirection and scent mark-

ing. One of the attractions of communication networks is that the idea applies to

any context in which the signals used travel far enough to encompass several other

individuals. However, each context will have distinctive features affecting the na-

ture of the information transmitted, the signals used and their travelling power;

therefore, the nature of the communication network may differ. Comparison of

networks found in different contexts could, therefore, advance our understanding

of the topic.

Mate choice

It is probably a fair generalization to say that in recent years the most

widely considered, modelled and experimented upon context for communication

has been the simplest mate choice situation, i.e. that involving a male signaller and

a female receiver. However, Ken Otter and Laurene Ratcliffe point out in Ch. 7 that

a communication network is a more likely context because females have access

to the widely broadcast mate attraction signals of several males. This chapter

discusses which traits females in a communication network use when choosing

between males: both as pair mates and as extra-pair partners. It also discusses the

way choice is achieved (e.g. simultaneous versus sequential assessment) and how

sampling by females can be inferred from the pattern of movement through a

network of signalling males.


129


130 Part II

Predation

Communication networks are not restricted to individuals of the same

species; predators have long been recognized to be the unwanted guests at com-

munication feasts, with long-range signals advertising the location of potential

prey. The commonest response of choruses of insects and anurans to the detec-

tion of a predator is to cease calling (Gerhardt & Huber, 2002). This may be an

efficient way to avoid predators, but it is an inefficient way to communicate with

the intended receivers of the signals, such as potential mates. In Ch. 8, Alexander

Lang, Ingeborg Teppner, Manfred Hartbauer & Heiner Romer explain how pseu-

dophylline katydids can to some extent overcome this problem by signalling with

vegetation-borne vibrations (tremulations) rather than airborne sounds, because

tremulations cannot be detected by passive listening bat predators. While such bat

predators are an important selection pressure on katydids, they are not the only

ones. This chapter uses neurophysiological preparations in the field and decision-

tree learning algorithms to investigate how katydids communicate in a noisy rain-

forest environment.

Nestling begging

In birds, the begging of nestlings has come to rival mate choice as a model

system for the study of the evolution of biological signalling (Wright & Leonard,

2002). However, in parallel with most of the work on mate choice, begging is

considered as a dyad, with a nestling (or the brood collectively) as the signaller and

one parent as the receiver. This seems odd, perverse even, given the close proximity

of nestlings (both to nestmates and to their parents) and the conspicuous vocal and

visual signals that nestlings produce. In Ch. 9, Andy Horn and Marty Leonard show

how considering begging as a communication network can yield new insights into

begging behaviour. Furthermore, issues for communication networks, in general,

are raised by the close proximity of individuals and the possibility of direct physical

action in a crowded nest. Such issues will be particularly relevant to many highly

social species such as social hymenoptera.

Redirection of aggression

The nature of communication in aggressive encounters has a long history

of study, from Darwin’s antithesis principle (Darwin, 1872) and Lorenz’s classics

King Solomon’s Ring and On Aggression (Lorenz, 1952, 1966) onwards. Most attention

has focused on signal exchanges before and during aggressive encounters. How-

ever, redirection is a puzzling behaviour performed by losers after an aggres-

sive contest. As the target of redirection is an individual other than the winner


The effects of particular contexts 131

(commonly a lower-ranking individual in social groups), it extends even dyadic

interactions into a network. Ani Kazem and Filippo Aureli discuss explanations

for redirection in Ch. 10, focusing on primates where it has been best described.

They conclude that redirection is best interpreted from a network viewpoint, in

terms of how it can influence the behaviour of bystanders.

Scent marking

Scent marking is a rather different context from the others considered in

this section. Whereas the other contexts could be loosely considered to be aspects

of social behaviour, scent marking is a particularly distinct aspect of chemical

communication and can be involved in several social contexts. Scent marking is

also rather different from most other signals because scent marks persist, often for

considerable periods, in the absence of the signaller. Any conspecific visiting the

scent mark can obtain information from it; in this respect, such marks could be

considered analogous to public noticeboards. In Ch. 11, Jane Hurst considers the

selection pressures that result from such undirected and long-lasting signals. Most

of her examples come from studies of mice, where many studies have examined

the behavioural and biochemical basis of scent communication.

Future directions

A common theme of the chapters in this section (and indeed throughout

this book) is that communication needs to be considered in a more complex way in

order to make progress. Far from being a standard recourse to complexities of the

real world when simple explanations fail (and being even further from a counsel

of despair), these chapters demonstrate how adopting a network perspective can

explain troublesome aspects of communication and also indicate directions for

future research.

One future challenge in mate choice is to characterize female assessment be-

haviour, because in many instances potential mates can be assessed at long range,

with close approach possibly representing the outcome of choice rather than

assessment in action. Scent marks seem to offer an opportunity for such char-

acterization because close approach and perhaps contact are required to gather

information from scent marks. Video tracking individuals in a naturalistic enclo-

sure may relatively easily provide data on patterns of visits to scent marks and, by

extension, on information gathering.

Many communication behaviours including mate choice are likely to be con-

strained by the presence of predators, as the chapter on katydids demonstrates. In

such circumstances, information gathering could be a costly exercise if proximity

to signallers increases the risk of being preyed upon. Similarly, being a bystander


132 Part II

at an aggressive interaction could be costly if the bystander becomes the target of

redirected aggression. These examples indicate that animals may have to trade-off

several somewhat disparate costs and benefits when gathering information in a

communication network. As the costs and benefits are likely to change diurnally

and seasonally as well as on much shorter timescales, these trade-offs may be best

explored by modelling.

Many of the chapters have suggested or implied that more detailed empirical

studies, both experimental and observational, are needed to further our under-

standing. In some sense, this is always going to be true because of the enormous

variation in biological systems; however, one value of a communication network

perspective is that it suggests what types of information (e.g. mate choice assess-

ment patterns, variation in the form of begging calls) would allow the field to

develop.

The chapters in this part demonstrate the advantages of a broader view of

communication over and above the advantages of a network view: understand-

ing of communication in any particular context can come from contexts other

than that under immediate consideration. For example, nestling begging would

seem to have little to offer contexts such as mate choice, aggression and resource

defence, because the detailed circumstances of the contexts are so different. How-

ever, begging behaviour focuses attention on issues that are fundamental to all

three contexts and indeed to communication in general, such as the distinction

between signals and physical action. Similarly, several explanations for redirection

behaviour operate over a timescale encompassing a sequence of several contests,

suggesting that some aspects of audience effects and eavesdropping would benefit

from consideration over such longer timescales. Difficult as it may be with an ever-

expanding communication literature, it would seem a good idea to keep an eye on

developments in several communication contexts. Because communication net-

works can be seen to apply to several contexts, adopting a network perspective

helps to promote such a breadth of interest.

References

Darwin, C. 1872. The Expression of the Emotions in Man and Animals. London: John Murray.



Lorenz, K. Z. 1952. King Solomon’s Ring; New Light on Animal Ways. New York: Crowell.

1966. On Aggression. New York: Harcourt, Brace and World.

Wright, J. & Leonard, M. L. 2002. The Evolution of Begging: Competition, Cooperation, and

Communication. Dordrecht: Kluwer.

P1: IYP/... P2: JZZ/...0521823617c07.xml CU1917B/McGregor 0 521 582361 7 April 11, 2005 11:49

7

Enlightened decisions: femaleassessment and communicationnetworks

k e n a . o t t e r 1 & l au r e n e r a t c l i f f e 2

1 University of Northern British Columbia, Prince George, Canada2 Queen’s University, Kingston, Canada

Introduction

Asymmetry in parental investment often predicts that females should

be choosier about prospective mates than males. It is commonly assumed that

females assess male characteristics during mate choice, but which traits are

assessed, and how they influence female decision making, is not well understood.

Current models of mate choice suggest females may sequentially sample a pool

of males, memorizing levels of trait expression among comparison males, or else

accept the first male that exceeds some minimum threshold value of mate qual-

ity. Recent tests of communication network theory suggest that these models may

have to be revised because females can tap into advertising signals broadcast in

a network fashion. Such behaviour could reduce costs of mate searching, as sig-

nals are perceived simultaneously, allowing instantaneous relative comparisons.

In this chapter, we explore the potential of females to extract comparative infor-

mation on the relative quality of males for use in reproductive decision making.

We focus primarily on primary mate choice decisions (i.e. initial selection of a

mating partner) and secondary mate choice decisions (i.e. mating decisions that

arise after social pairing, which may include extra-pair copulations or ‘divorce’ of

the current mate to pair with another male) based on acoustic signals in territorial

passerines; however, the ideas that we present should be applicable to other taxa

and other sensory modalities. Finally, we discuss the potential impacts of habitat

alteration on females’ abilities to use network assessments for mate choice.


133


134 K. A. Otter & L. Ratcliffe

In the early 1990s, it was realised that bird song and other animal commu-

nication took place in a more extensive (network) context than male–female

or male–male dyads; several receivers may perceive each signaller simultane-

ously and, conversely, signallers may direct their signal towards several receivers

(McGregor, 1993; McGregor & Dabelsteen, 1996). The idea that females seeking

mates may be able to assess several males simultaneously, not only on leks but

also in situations where males defend more dispersed territories, complicates the

study of female mate choice. For example, we need to rethink how females may

be sampling males, especially during initial territorial settlement and any subse-

quent pursuit of extra-pair copulations.

Communication network theory challenges us to reconsider traditional models

of mate assessment. Constraints on signal transmission and reception may limit

the spread of information to receiving parties. Such constraints should bias assess-

ment to particular kinds of signal and may also help to explain why some signals

important in initial mate choice need not necessarily correlate with secondary

mate choice (e.g. red plumage in house finches, Carpodacus mexicanus, is selected

by females in initial mate choice but does not seem to affect extra-pair success:

Hill, 2002). We must also consider the kinds of signal that are received by females;

how habitat alteration might influence communication in a network, and how,

in turn, that might influence reproductive success.

Deciphering which attributes of males are of most interest to females has

proved to be a challenging task; a common approach has been to ask whether

females pick superior males as partners, where superiority is defined by the expres-

sion of condition-dependent signals. In birds, considerable evidence from dyadic

mate choice studies suggests that colour and vocal cues believed to be energeti-

cally expensive are correlated with mate selection (reviewed by Gil & Gahr, 2002).

Whether such dyadic studies accurately reflect female assessment in natural cir-

cumstances is still unclear. Considerations of communication in a network con-

text have stimulated new experimental approaches that seek to determine how

males evaluate potential rivals (e.g. Naguib et al., 1999; Todt & Naguib, 2000; Ch. 2)

and similar types of study may help researchers to decipher how females evaluate

potential mates.

This chapter reviews data on which traits appear to be important in female

choice in songbirds and then suggests ways to model and test female choice using

communication network theory. Although work in this field is still limited, we

review published and in-progress studies that discuss how signals might spread in

networks and how females might assess such information. Finally, we discuss how

habitat alteration can affect the propagation of signals, and how this influences

the ability of females to assess males in communication networks.


Enlightened decisions: female assessment 135

How do females assess males?

Traditional models of primary female mate choice assume that females

assess males sequentially ( Janetos, 1980; Wittenberger, 1983). In choosing social

mates, females move from one male to the next and select mates via either sequen-

tial comparison of the current male versus the last male visited, or via a best-of-n

model, where females sample all males and then return to the individual of highest

quality. Although less-formally modelled, most studies on secondary mate choice

assume similar kinds of decision-making strategy; females select a social partner

then assess this male relative to other available males to determine whether to en-

gage in extra-pair copulations or divorce (e.g. the ‘better options’ model of divorce

(Ens et al., 1993), in which birds select a social mate then assess opportunities to

desert and pair with a male of better quality, or similar strategies in the ‘genetic

benefits’ models of extra-pair copulation (Kempenaers & Dhondt, 1993)). Recent

models of primary mate choice are more realistic because they incorporate costs

of mate searching (Real, 1990; Wiegmann et al., 1996) but still assume that males

are assessed sequentially (e.g. Fig. 7.1a). However, the signals used by females to

assess males may propagate sufficient distance in some circumstances for males

to be assessed simultaneously (Gibson & Langen, 1996).

In communication networks, female receivers can simultaneously detect the

signals of several territorial males. For example, a female moving from one male

to the next may still be able to detect the signals of males she has visited previ-

ously. Thus, females may be able to assess the relative expression of several males’

signals simultaneously, without having to rely on memory of absolute expression.

Females could continue to search in such a fashion until no new male exceeds

a preceding male. However, female searching may not be even this constrained.

If females use a best-of-n strategy, they need not rely solely on memory of trait

expression of each male. All they need do is remember the territory locations of

males of perceived high quality and position themselves in a manner that allows

simultaneous comparison. If signal transmission is sufficiently long range, fe-

males may be able to assess and eliminate a number of males without even closely

approaching them, as described in anuran mating aggregations (e.g. Murphy &

Gerhardt, 2002; see below). Females positioning themselves strategically within

networks and making choices on relative trait expression could decrease search

time, maintain safe distances from territory owners and reduce the chance of

mistaken decisions (Fig. 7.1b).

Field studies typically use female movement patterns to assess the number

of males that are sampled by females. Some studies have used radio telemetry

to track female movement and have assumed that close approach is evidence


AB

C

DE

i

ii

iii

iv

(d)

(a)

(c)

(b)

AB

C

DE

AB

C

DE

AB

C

D

E

Fig. 7.1. Sampling of males by females is often assumed when a female closely

approaches a male; however, if females assess males by long-range signals the

relationship between sampling and close approach may be rather different:

compare (a) with (b), and (c) with (d). Males are represented by capital letters (A–E)

and the female’s path is shown as a dashed line. In (a) and (c), dark lines represent

their territorial boundaries. In (b) and (d) these lines are shown in grey so as not

to obscure the dotted lines representing the range of effective signal

transmission. (a) The female travels along path i–iv approaching males E, C, D

and then C, with whom the female finally settles. In this scenario, the female

may be considered to be using a best-of-n sampling method, having sampled

males E, C and D but not A and B. (b) If we consider female sampling in relation to

the signal transmission range, we see that the female’s movement may, in fact,

also allow her to sample males A and B (i.e. her path lies within their effective

signal range) without ever approaching them directly. During the period that she

is in the territory of male C, the female could hypothetically sample the signals of

males B, A, E and D, and she would be within transmission range of two to three

males at any given time along her whole route. (c) In this scenario, the female

appears to avoid close contact with all males until approaching male A. This

could be interpreted as the female not sampling any of the males prior to making

a mating decision, but considering the situation in relation to signal range (d)

reveals that the female would be able to assess males E, C and B by their signals

en route to male A. Moreover, the female’s movements place her within the

signal range of at least two males at any point along her path, allowing her

potentially to compare males simultaneously, as well as comparing each male

newly encountered with the last male along her pathway. If male A exceeds the

traits of the other males, this female may be interpreted as adopting either a

threshold-style model or a best-of-n model in mate sampling.

136



of assessment of a signalling male (Bensch & Hasselquist, 1992; Neudorf et al.,

1997). The pied flycatcher Ficedula hypoleuca offers an extreme example of such

movement, as the nestbox is an assessed resource (Slagsvold et al., 1988; Dale

et al., 1990) and assessment requires close inspection, in much the same way

that a short-range signal would. However, movement patterns can be difficult

to interpret. For example, a female moving along the boundary between terri-

torial males, apparently undetected by them, may still be sampling these males

(Fig. 7.1c). If the males are producing long-range signals, then the female could

sample males based on this signalling network (Fig. 7.1d). Her movement pattern

allows at least two males to be assessed at almost all points along her route. By

such surreptitious sampling (i.e. moving silently and apparently remaining unde-

tected by males along the route: Neudorf et al., 1997), females may avoid some of

the costs associated with close approach to males. These costs could be harassment

from males or aggression from mated females (e.g. Dale & Slagsvold, 1995). After

surreptitious sampling, the female may then closely approach the male she has

selected. In such a scenario, close approach indicates choice rather than sampling

and the amount of sampling is underestimated (in Fig. 7.1d four males have been

sampled rather than one). Murphy & Gerhardt (2002) described such a scenario

in female barking treefrogs Hyla gratiosa in the field; females approach only a

single male in a chorus, suggesting no sampling has occurred and that the first

male encountered is selected. However, further anecdotal evidence suggested that

females may be assessing several males at a distance and then approaching only

the selected male (Murphy & Gerhardt, 2002).

Therefore, we urge caution when patterns of female movement are used to

infer sampling behaviour. We suggest that it may be more appropriate to deter-

mine female movement in relation to the transmission distance of signals used

in mate choice. Such considerations will allow us to determine whether females

are strategically placing themselves in areas that maximize the number of males

that can be simultaneously assessed while concurrently minimizing search costs.

What are females looking for?

Research on mate choice since the early 1990s has focused on females

selecting males based on perceived quality. But how are such distinctions made?

Females are presumably unable to assess male genetic quality directly but can

infer this through assessment of traits that tightly correlate to resource-holding

potential of the male (Grafen, 1990).

Visual signals from plumage are known to be associated with male condition

and ability to acquire resources (Hill, 2002). Dominance status and aggressive

behaviour also reflect relative male condition, as they predict access to limited



resources (Ekman & Askenmo, 1984; Hogstad, 1988; Desrochers, 1989; Ficken

et al., 1990; Ekman & Lilliendahl, 1992). Singing behaviour is also indicative of

male condition. Repertoire size in several species is related to age (Yasukawa et al.,

1980; Lampe & Espmark, 1994; Birkhead et al., 1997; Eens, 1997), providing infor-

mation on individual survival. Singing behaviour may also reflect a male’s ability

to secure access to limited resources (Reid, 1987; Alatalo et al., 1990; Thomas,

1999a,b; Thomas & Cuthill, 2002). Song output is known to be associated with a

male’s dominance rank (Otter et al., 1997), level of parasite infestation (Møller,

1991) and immune response (Saino et al., 1997a). Even the fine structure of song

may give cues to the survivorship (Forstmeier et al., 2002) or rank (Christie et al.,

2004) of males. Many studies have shown that these behavioural and morpho-

logical characteristics are important in female choice, both for primary mates

(Radesater et al., 1987; Alatalo et al., 1990; Andersson, 1994; Hoi-Leitner et al.,

1995; Buchanan & Catchpole, 1997) and for extra-pair paternity (Smith, 1988;

Morton et al., 1990; Houtman, 1992; Wetton et al., 1995; Hasselquist et al., 1996;

Kempenaers et al., 1997; Saino et al., 1997b; Møller et al., 1998; Otter et al., 1998;

Forstmeier et al., 2002).

Which trait is the best indicator of quality and what is meant by a good indicator

of quality? Strong correlations between male quality and expression of the trait

are assumed for the trait to be ‘reliable’ (Grafen, 1990) and several condition-

dependent traits will often be intercorrelated. To determine which of these signals

are likely to be assessed by females, however, we should focus on the perception

of signals by females rather than the production of signals by males. It does not

necessarily follow that a female will be able to discriminate amongst males even

if male quality is correlated with the expression of the trait. The traits may all

potentially be reliable, but the important question for females in networks is

at what distance are they detectable and discriminable (Ch. 20)? The debate on the

evolution of multiple signals (Møller & Pomiankowski, 1993; Pomiankowski &

Iwasa, 1993; Iwasa & Pomiankowski, 1994; Johnstone, 1996) has focused largely on

whether multiple signals are of use to the female in an additive way, or whether

their apparent redundancy is used by females to confirm their assessment. The

debate assumes that females are able to assess all traits simultaneously. It seems

more likely that, in a natural network context, signals that target different sensory

modalities may be assessed sequentially in relation to their transmission distance.

Distance can have a profound effect on both detection and discrimination

(Wiley & Richards, 1982). Many morphological and behavioural traits that require

visual inspection can only be discriminated at close range, particularly in habi-

tats with dense vegetation. Tactile signals may be similarly restricted. In contrast,

olfactory or auditory signals, particularly song, have evolved to transmit at least

the average interterritory spacing within a species (Brenowitz, 1982; Calder, 1990);



thus, detection and discrimination are possible at greater range. Long-range sig-

nals may form the basis of initial assessment, which can then be confirmed on

closer inspection by assessment of short-range signals. The apparent redundancy

of intercorrelated signals may thus reflect the hierarchical way in which they are

assessed (Bradbury & Vehrencamp, 1998).

Imagine a situation where a female is assessing a prospective male, either as

a mate or as an extra-pair partner. In many cases, female movement may be con-

strained by such factors as mate guarding, aggression of nearby mated females

(Slagsvold & Lifjeld, 1994; Dale & Slagsvold, 1995) or risk of predation. Under these

circumstances, any mechanism that enables females to narrow down the pool of

acceptable males from a position of relative safety would be favoured (Gowaty,

1996). In songbirds, male advertising song provides an ideal signal for assessment

at a distance (Fig. 7.2a–c), because of its long transmission range. Initial decisions

about males can be made via this single cue; females can then directly approach

subsets of males deemed to be of the best quality among the available pool. Further

discrimination may then occur by shifting assessment to short-range signals, such

as plumage, the expression of which we would expect to correlate with quality

indicated by long-range signals. This sequential assessment of signals may increase

the certainty of assessment.

Females may employ information derived from networks not only to narrow

the pool of potential mates; females may also instigate network communication

to evaluate male quality during close approach. For example, during intrusions

across territorial boundaries, the attraction of neighbouring males may incite

competitive interactions between a female’smate and his neighbours (Fig. 7.2d–f).

Montgomerie & Thornhill (1989) suggested that females might incite interactions

between males as a mechanism for sperm competition, but it is also possible

that such behaviour provides females with more information about the general

quality of available males (Sæther, 2002). This possibility is also supported by

demonstrations that females use information from male–male singing interac-

tions (i.e. they eavesdrop; Ch. 2) in extra-pair behaviour decisions (Otter et al., 1999;

Mennill et al., 2002). Future work on this topic should target species where females

readily engage in secondary mate choice, for example the pursuit of extra-pair

copulations.

Recently, Sæther (2002) demonstrated that such incitement occurs in the great

snipe Gallinago media, a species in which females call from the edges of males’

territories within leks. Using playbacks, he showed that female calls from such

boundaries increase the competitive interactions between neighbouring males,

providing a potential source of information to prospecting females. Similarly,

in territorial songbirds, females may exhibit behaviour that draws their mates

and neighbours into interactions on territorial boundaries. Ramsay et al. (1999)



(a) (c)(b)

(d) (e) (f)

Fig. 7.2. Secondary mate choice, such as the decision to engage in extra-pair

copulations or divorce a current mate for another male, may be constrained in some

territorial species. Unlike primary (i.e. initial) mate choice, females may not be able to

move freely through the territories of various males because of the presence of

resident females (a). However, long-range signals of males, such as song, transmit

beyond the boundaries of the territories and effectively signal presence to

neighbouring territories (b). In (a, b, d and e), dark lines represent territorial

boundaries with males and females shown as symbols. Dotted lines represent the

range of effective signal transmission and some territory boundaries are show in grey

in (c) and (f ) to prevent obscuring signal ranges. (c) A female may be able to assess all

neighbouring males as well as her mate without having to leave the territory. (d) A

female may incite interactions among males by moving (arrow) towards a boundary.

(e) Her movement may draw neighbouring males to that boundary (arrows) and the

resulting interactions may allow her to assess other signals, including short-range

types of display (plumage, direct dominance interactions or fights). (f ) Thus the

female could assess a subset of the original males using multiple signals in

succession, possibly leading to increased certainty of her assessment.

found that female black-capped chickadees Poecile atricapillus place their nests close

to territorial edges, despite evidence that these areas offer no better resources

or nesting opportunities than central nest locations. One explanation for this

pattern is that females can more easily monitor neighbours in relation to their

mates and capitalize on opportunities for secondary mate choice. One observed

outcome of this nesting pattern by females is increased numbers of territorial

disputes between the resident male and his neighbours (Ramsay et al., 1999), which



may also make it easier for females to make these relative distinctions. Female

hooded warblers Wilsonia citrina at the peak of their fertility give characteristic

chip calls, which result in increased intrusions by neighbouring males (Neudorf,

1996); female European robins Erithacus rubecula show similar patterns in their

rates of seep calls (Tobias & Seddon, 2002). While such ‘fertility announcements’

are often interpreted as mechanisms that increase female choice through potential

sperm competition, they may also function more directly in active assessment by

females if they increase interactions between mates and other males.

Evidence for female assessment of network information

Few studies have asked whether female mate choice incorporates informa-

tion derived from signals in a communication network, although the potential for

females to use such information seems considerable (e.g. Otter et al., 1999; Mennill

et al., 2002). K. A. Otter, T. M. Peake, A. M. R. Terry and P. K. McGregor (unpublished

data) found that dawn chorus singing of neighbouring male great tits Parus major is

clearly recorded by microphones placed within the nestboxes of roosting females;

therefore, it is likely that females could assess a network of singing males without

leaving the nestbox. Relative song output among males during the dawn chorus is

known to correlate with male condition in a number of species (Reid, 1987; Alatalo

et al., 1990; Otter et al., 1997; Thomas, 1999a,b; Thomas & Cuthill, 2002), and it is,

therefore, a useful cue of quality (Hutchinson et al., 1993). However, surprisingly

little work has investigated whether females attend to variation in male dawn

song. Otter & Ratcliffe (1993, 1996) suggested that changes in dawn singing of

males who have lost their mates might function as useful cues for neighbouring

females seeking better mates, and anecdotal evidence in black-capped chickadees

suggests that divorces occur soon after the dawn chorus ends. Further studies

should be conducted to determine whether the generally higher song output at

dawn is used in assessment by females. This could be done by elevating male song

output by supplementary feeding (e.g. Reid, 1987; Alatalo et al., 1990; Thomas,

1999a,b) and seeing whether radio-tracked females appear attentive to increased

song output of neighbours.

To date, the few studies that have investigated female assessment in commu-

nication networks have focused on eavesdropping upon dyadic male aggressive

singing interactions (Otter et al., 1999; Mennill et al., 2002). These eavesdrop-

ping experiments have used interactive playback (Dabelsteen & McGregor, 1996)

to manipulate the outcome of aggressive singing interactions between males.

Otter et al. (1999) showed that female great tits appear to be aware of the relative

ease with which males interact with a ‘strange intruder’ (the interactive play-

back). The mates of males who lost interactions were more likely to intrude into



neighbouring territories in following days than were those whose mates won

against the same intruder. As the only difference in singing between the two male

treatments was in their song rates relative to the playback, females appear to base

their movement patterns on the perceived interaction. Moreover, females visited

more frequently the neighbouring male heard to win against the same intruder to

which their mate had lost. Our results, however, found no evidence that females

produced more young with these males; there was no pattern of extra-pair cop-

ulation associated with the playback treatment, although males who had been

cuckolded were of lower genetic heterozygosity than males who were not cuck-

olded (Otter et al., 2001). As genetic heterozygosity appears to be associated with

survival and fecundity (Coulson et al., 1998; Hansson et al., 2001), this result might

indicate that females drawn to males via song may have found other signals (e.g.

colour patterns, which were not assessed during the study) that contradicted the

assessment via song.

By contrast, Mennill et al. (2002) recently showed that female black-capped

chickadees exposed to similar eavesdropping opportunities did modify extra-pair

behaviour, although there was no apparent effect on observed intrusions. Males

of high and low social rank were exposed to challenges simulated by interactive

playback. The challenges either reinforced their rank disparity (e.g. de-escalating

playback to a dominant male, escalating to the subordinate), or countered rank

disparity (escalate to dominant male, de-escalate to subordinate). While females

mated to low-ranking males showed no influence of playback on their decisions

to engage in extra-pair copulation, females mated to high-ranking males that had

lost against the playback ‘intruder’ were more likely to have extra-pair young in

their broods. As females in this species mated to high-ranking males usually forego

extra-pair copulation (Otter et al., 1998), this result suggests that the protocols used

by Mennill et al. (2002) had a profound impact on female decisions. Normally, if

females mated to high-ranking males do engage in extra-pair copulation, they se-

lect males of similar or higher rank than their mate (Otter et al., 1998). Yet, D. J.

Mennill and colleagues (unpublished data) found that extra-pair males selected by

these high-ranking females were nearly random with respect to the relative rank

of their mate; further evidence that assessment by eavesdropping in a network

can have dramatic influences on behavioural decisions.

There is also experimental evidence that female Siamese fighting fish Betta

splendens eavesdrop on male–male aggressive visual displays and are more will-

ing to mate with males that they have seen win such interactions (Doutrelant &

McGregor, 2000; see also Ch. 2) These initial studies provide impetus for future

work. However, a number of fundamental questions still need to be addressed.

For example, the nature and accuracy of the information on relative male quality

available to females in interactions remains to be determined. Another major is-

sue is the extent to which features of signals used for individual identification are



affected by transmission over long distances. Many ideas related to communica-

tion networks assume that individual signallers can be identified; for example, in

the context of mate choice in birds, females are assumed to be able to distinguish

different males by voice. This has been shown in a number of species (e.g. Davis,

1986; Weary & Krebs, 1992; Lind et al., 1996); however, these studies have not re-

quired females to make these discriminations at long distance. Although song is a

long-distance signal, all acoustic signals are subject to degradation over distance

(Bradbury & Vehrencamp, 1998), which could negatively affect features females

use in discrimination and assessment.

Implications for networks of environmental alteration

Song transmission is affected not only by distance but also by the medium

through which it must travel. Reverberation, differential attenuation and other

effects on sound are imposed by habitat characteristics and may shape the songs of

species inhabiting different areas (Catchpole & Slater, 1996). Habitat alteration can

result in a change in the characteristics of signals to maintain maximum transmis-

sion range within new habitats, for example in rufous-collared sparrows Zonotrichia

capensis inhabiting forested versus grassland habitats (Tubaro et al., 1993; Tubaro &

Segura, 1994). However, it is unknown how long it takes for changes in song to oc-

cur in response to habitat change and how females respond to such changes. Most

habitat alterations occur over very short timeframes, and unless reproductive iso-

lation occurs between undisturbed and disturbed habitats, selection in response

to the altered landscape may be slow (e.g. Dhondt et al., 1992; Dias & Blondel, 1996).

Therefore, changes in the structure of male song may not keep pace with changes

in the habitats, leading to song structure that is mismatched for transmission in

the present environment.

In many species, habitat alteration may change sound transmission conditions

and also decrease resource availability or breeding success (Blondel, 1985; Blondel

et al., 1993; Fort & Otter, 2004). If habitat alteration simultaneously reduces signal

transmission and enlarges territory size in response to lowered resources, the

extent of communication networks and the information to be gained from them

may be seriously reduced. For example, the size of male song networks in black-

capped chickadees appears to be constrained by habitat change. While recording

focal males during the dawn chorus, we conducted standard avian point counts

at three-minute intervals to determine the number and direction of other males

audible at the location of the focal male. The result was that fewer males were

audible to male chickadees that occupy early successional forests (characterized

by a low canopy and dense understorey) than to those occupying nearby mature,

mixed forests (I.-J. Hansen, K. A. Otter & H. van Oort, unpublished data). This is likely

a consequence of decreased transmission of song (I.-J. Hansen, K. A. Otter & H. van



Oort, unpublished data) and increased territory size (Fort & Otter, 2004) of males

occupying this disturbed habitat.

Such changes to the transmission of signals following habitat alteration can

potentially diminish assessment ability of females. Not only could the networks

decline, reducing the number of males a female could assess for secondary mate

choice, but, in addition, females in such circumstances could also fail to locate

primary mates, or become polygynous. Alternatively, females may mate monog-

amously with males of lesser quality during primary mate selection, because as-

sessment of neighbouring males is constrained (M. Kasumovic, L. M. Ratcliffe &

P. T. Boag, unpublished data). If such assessment is critical in female mating tac-

tics (e.g. Wagner, 1991), females may fail to settle in such altered habitats, even if

the resources would support a breeding effort.

Another impact of habitat alteration that could affect female assessment in

networks is the close relationship between resource access and the ability of males

to produce condition-dependent traits. If habitat quality is poor, the absolute

expression of traits may be diminished (Hill, 1995); moreover Qvarnstrom and

Forsgren (1998) also predict that dominant males may suffer disproportionately

in poor habitat. The costs of achieving dominance status are normally countered by

the benefit of access to rich resources, but if the habitat is unable to produce these

benefits, the high costs paid by dominants may put them in a net metabolic deficit.

In support of this, H. van Oort, K. A. Otter, F. Fort & C. I. Holschuh (unpublished

data) found that song output in the dawn chorus of black-capped chickadees varies

across habitats. As predicted by Otter et al. (1997), birds occupying mature forests in

northern British Columbia, Canada had song output that reflected their relative

rank: high-ranked birds tended to have higher song output than lower-ranked

birds. By comparison, birds settling in neighbouring young, regenerating forests

did not show this same trend. Overall, the birds in the disturbed forests had lower

song output than birds in the undisturbed forests, as predicted by Hill (1995),

but this relationship was driven by abnormally low song output by high-ranking

males in the disturbed woods, as predicted by Qvarnstrom and Forsgren (1998).

Males of either high or low rank in the disturbed forest could not be differentiated

based on song output. Therefore, the transmission of signals may not be the only

impact of landscape alterations; the reliability of signals may also be influenced

by habitat context and may diminish the ability of females to use long-range

signalling networks in assessment.


The role of communication networks in female mate choice is ripe for

study using the techniques that simulate signalling interactions. In territorial



(a) (b)

(c)

Fig. 7.3. A proposed experiment to test females’ use of information in communication

networks in male assessment. The figure shows an aviary in plan view with release

points at the corners (solid squares). (a) The aviary contains four cells (squares) each

containing a choice stimulus (represented by a loudspeaker symbol here, but they

could contain live males. (b) Signals are designed so that, when broadcast, only a

position in the middle of the aviary would allow the female to assess all males

simultaneously (i.e. the point of overlap of the effective signal ranges, shown as dotted

lines). (c) The pattern of female movement observed (dashed lines), for example

consistently moving towards the central areas prior to entering a cell, would indicate

that several males were being compared.

songbirds, studies of eavesdropping (Ch. 2) may help us to understand female

secondary mating tactics. By manipulating the relative signals emanating from

neighbouring males, and using radio-tracking and genetics to measure female

preferences, we should be able to obtain a clearer idea of how (and perhaps why)

socially monogamous females choose secondary partners.



Considering communication in a network context should also aid our under-

standing of primary mate choice, an area of particular interest. It is particularly

important to determine at what range females can discriminate between signals.

As we argued earlier, measures of close approach to males may be insufficient

to determine whether males have been ‘sampled’ by females. This is particularly

the case if females are assessing males at long distance, approaching only chosen

males. Such cryptic choice is potentially difficult to monitor. One way of address-

ing this problem is to plot female movements in relation to known signal trans-

mission range. Speaker replacement studies, such as those used with flycatchers

(Eriksson & Wallin, 1986) or starlings Sturnus vulgaris (Mountjoy & Lemon, 1991),

could simulate clusters of signalling males to determine the effects of altered

singing patterns on female assessment routes and strategies. Aviary studies may

also be informative in this regard. Rather than the traditional dyadic choices pre-

sented to females, aviaries with several males could be presented (Fig. 7.3). The

idea that females position themselves to assess males in a network could be inves-

tigated by manipulating the transmission range of auditory or visual signals and

observing female movements (e.g. Fig. 7.3).

Acknowledgements

We thank Peter McGregor, Bart Kempenaers, Dan Mennill, Harry van Oort, Carmen

Holschuh, Tania Tripp and David Nordstrom for discussion on the topics in this chapter. Ingebjørg-

Jean Hansen, Kevin Fort, Harry van Oort, Peter Christie, Dan Mennill, Mike Kasumovic, Peter

McGregor, Tom Peake and Andrew Terry kindly allowed us to cite the results of unpublished,

co-authored data. Peter McGregor, Dan Mennill, Marc Naguib, Harry van Oort and an anonymous

reviewer also provided useful suggestions on early drafts of the manuscript. Both authors were

funded by NSERC (Canada) research grants during the preparation of this work.

References

Alatalo, R. V., Glynn, C. & Lundberg, A. 1990. Singing rate and female attraction in the

pied flycatcher: an experiment. Animal Behaviour, 39, 601–602.

Andersson, M. 1994. Sexual Selection. Princeton, NJ: Princeton University Press.

Bensch, S. & Hasselquist, D. 1992. Evidence for active female choice in a polygynous

warbler. Animal Behaviour, 44, 301–311.

Birkhead, T. R., Buchanon, K. L., de Voogd, T. J. et al. 1997. Song, sperm quality and

testes asymmetry in the sedge warbler. Animal Behaviour, 53, 965–971.

Blondel, J. 1985. Breeding strategies of the blue tit and coal tit (Parus) in mainland and

island Mediterranean habitats: a comparison. Journal of Animal Ecology, 54,

531–556.



Blondel, J., Dias, P. C., Maistre, M. & Perret, P. 1993. Habitat heterogeneity and

life-history variation of Mediterranean blue tits (Parus caeruleus). The Auk, 110,

511–520.



Brenowitz, E. A. 1982. The active space of red-winged blackbird song. Journal of

Comparative Physiology, A, 147, 511–22.

Buchanan, K. L. & Catchpole, C. K. 1997. Female choice in the sedge warbler,

Acrocephalus schoenobaenus: multiple cues from song and territory quality.


Calder, W. A. 1990. The scaling of sound output and territory size: are they matched?

Ecology, 71, 1810–1816.

Catchpole, C. K. & Slater, P. J. B. 1996. Bird Song. Cambridge, UK: Cambridge University

Press.

Christie, P. J., Mennill, D. J. & Ratcliffe, L. M. 2004. Pitch shifts and song structure

indicate male quality in the dawn chorus of black-capped chickadees. Behavioral

Ecology and Sociobiology, 55, 341–348.

Coulson, T. N., Pemberton, J. M., Albon, S. D. et al. 1998. Microsatellites reveal

heterosis in red deer. Proceedings of the Royal Society of London, Series B, 256, 489–495.

Dabelsteen, T. & McGregor, P. K. 1996. Dynamic acoustic communication and

interactive playback. In: Ecology and Evolution of Acoustic Communication in Birds, ed.

D. E. Kroodsma & E. H. Miller. Ithaca, NY: Cornell University Press, pp. 398–408.

Dale, S. & Slagsvold, T. 1995. Female contests for nest sites and mates in the pied

flycatcher, Ficedula hypoleuca. Ethology, 99, 209–222.

Dale, S., Amundsen, T, Lifjeld, J. T. & Slagsvold, T. 1990. Mate sampling behaviour of

female pied flycatchers: evidence for active mate choice. Animal Behaviour, 27,

87–91.

Davis, W. J. 1986. Acoustic recognition in the belted kingfisher: cardiac response to

playback vocalizations. The Condor, 88, 505–512.

Desrochers, A. 1989. Sex, dominance, and microhabitat use in wintering black-capped

chickadees: a field experiment. Ecology, 70, 636–645.

Dhondt, A. A., Kempenaers, B. & Adriaensen, F. 1992. Density-dependent clutch size

caused by habitat heterogeneity. Journal of Animal Ecology, 61, 643–648.

Dias, P. C. & Blondel, J. 1996. Local specialization and maladaptation in the

Mediterranean blue tit (Parus caeruleus). Oecologia, 107, 79–86.



Eens, M. 1997. Understanding the complex song of the European starling: an

integrated ethological approach. Advances in the Study of Behaviour, 26, 355–434.

Ekman, J. & Askenmo, C. 1984. Social rank and habitat use in willow tit groups. Animal

Behaviour, 32, 508–514.

Ekman, J. B. & Lilliendahl, K. 1992. Using priority to food access: fattening strategies in

dominance-structured willow tit (Parus montanus) flocks. Behavioral Ecology, 4,

232–2.



Ens, B. J., Safriel, U. N. & Harris, M. P. 1993. Divorce in the long-lived and monogamous

oystercatcher, Haematophagus ostralegus: incompatibility or choosing a better

option? Animal Behaviour, 45, 1199–1217.

Eriksson, D. & Wallin, L. 1986. Male bird song attracts females: a field experiment.


Ficken, M. S., Weise, C. M. & Popp, J. W. 1990. Dominance rank and resource access in

winter flocks of black-capped chickadees. Wilson Bulletin, 102, 623–633.

Forstmeier, W., Kempenaers, B., Meyer, A. & Leisler, B. 2002. A novel song parameter

correlates with extra-pair paternity and reflects male longevity. Proceedings of the


Fort, K. & Otter, K. A. 2004. The impact of habitat disturbance on reproduction of

chickadees. The Auk, 121, in press.

Gibson, R. M. & Langen, T. A. 1996. How do animals choose their mates? Trends in

Ecology and Evolution, 11, 468–470.

Gil, D. & Gahr, M. 2002. The honesty of bird song: multiple constraints for multiple

traits. Trends in Ecology and Evolution, 17, 133–141.

Gowaty, P. A. 1996. Multiple mating by females selects for males that stay: another

hypothesis for social monogamy in passerine birds. Animal Behaviour, 51,

482–484.

Grafen, A. 1990. Biological signals as handicaps. Journal of Theoretical Biology, 144,

517–546.

Hansson, B., Bensch, S., Hasselquist, D., & Åkesson, M. 2001. Microsatellite diversity

predicts recruitment of sibling great reed warblers. Proceedings of the Royal Society

of London, Series B, 268, 1287–1291.

Hasselquist, D., Bensch, S. & von Schantz, T. 1996. Correlation between male song

repertoire, extra-pair paternity and offspring survival in the great reed warbler.

Nature, 381, 229–232.

Hill, G. E. 1995. Ornamental traits as indicators of environmental health. BioScience, 45,

25–31.

2002. A Red Bird in a Brown Bag: The Function and Evolution of Colourful Plumage in the

House Finch. Oxford: Oxford University Press.

Hogstad, O. 1988. Rank-related resource access in winter flocks of willow tit Parus

montanus. Ornis Scandinavica, 19, 169–174.

Hoi-Leitner, M., Nechtelberger, H. & Hoi, H. 1995. Song rate as a signal for nest site

quality in blackcaps (Sylvia atricapilla). Behavioral Ecology and Sociobiology, 37,

399–405.

Houtman, A. M. 1992. Female zebra finches choose extra-pair copulation with

genetically attractive males. Proceedings of the Royal Society of London, Series B, 249,

3–6.

Hutchinson, J. M. C., McNamara J. M. & Cuthill, I. C. 1993. Song, sexual selection,

starvation, and strategic handicaps. Animal Behaviour, 45, 1153–1177.

Iwasa, Y. & Pomiankowski, A. 1994. The evolution of mate preferences for multiple

sexual ornaments. Evolution, 48, 853–867.

Janetos, A. C. 1980. Strategies of female mate choice: a theoretical analysis. Behavioral




Johnstone, R. A. 1996. Multiple displays in animal communication: ‘backup signals’

and ‘multiple messages’. Philosophical Transactions of the Royal Society of London, Series

B, 351, 329–338.

Kempenaers, B. & Dhondt, A. A. 1993. Why do females engage in extra-pair

copulations? A review of the hypotheses and their predictions. Belgian Journal of

Zoology, 123, 93–103

Kempenaers, B., Verheyen, G. R. & Dhondt, A. A. 1997. Extrapair paternity in the blue

tit (Parus caeruleus): female choice, male characteristics and offspring quality.


Lampe, H. M. & Espmark, Y. O. 1994. Song structure reflects male quality in pied

flycatchers, Ficedula hypoleuca. Animal Behaviour, 47, 869–876.

Lind, H., Dabelsteen, T. & McGregor, P. K. 1996. Female great tits can identify mates by

song. Animal Behaviour, 52, 667–671.


identification, ranging, and eavesdropping. Philosophical Transactions of the Royal







Møller, A. P. 1991. Parasite load reduces song output in a passerine bird. Animal

Behaviour, 41, 723–730.

Møller, A. P. & Pomiankowski, A. 1993. Why have birds got multiple sexual

ornaments? Behavioral Ecology and Sociobiology, 32, 167–176.

Møller, A. P., Saino, N., Taramino, G., Galeotti, P. & Ferrario, S. 1998. Paternity and

multiple signalling: effects of a secondary sexual character and song on paternity

in the barn swallow. American Naturalist, 151, 236–242.

Montgomerie, R. & Thornhill, R. 1989. Fertility advertisment in birds: a means of

inciting male–male competition? Ethology, 81, 209–220.

Morton, E. S., Forman, L. & Braun, M. 1990. Extra-pair fertilisations and the evolution

of colonial breeding in purple martins. The Auk, 107, 275–283.

Mountjoy, D. J. & Lemon, R. E. 1991. Song as an attractant for male and female

European starlings, and the influence of song complexity on their response.


Murphy, C. G. & Gerhardt, H. C. 2002. Mate sampling by female barking treefrogs (Hyla

gratiosa). Behavioral Ecology, 13, 472–480.

Naguib, M., Fitchel, C. & Todt, D. 1999. Nightingales respond more strongly to vocal


Series B, 266, 537–542.

Neudorf, D. L. 1996. A dual system of female control of extra-pair copulations in the

hooded warbler (Wilsonia citrina). Ph.D.; Thesis. York University, North York,

Ontario.

Neudorf, D. L., Stutchbury, B. J. M. & Piper, W. H. 1997. Covert extraterritorial behavior

of female hooded warblers. Behavioral Ecology, 8, 595–600.



Otter, K. & Ratcliffe, L. 1993. Changes in the singing behaviour of male black-capped

chickadees (Parus atricapillus) following mate removal. Behavioral Ecology and


1996. Female initiated divorce in a monogamous songbird: abandoning mates for

males of higher quality. Proceedings of the Royal Society of London, Series B, 263,

351–354.

Otter, K., Chruszcz, B. & Ratcliffe, L. 1997. Honest advertisement and singing during

the dawn chorus of black-capped chickadees, Parus atricapillus. Behavioral Ecology,

8, 167–173.

Otter, K., Ratcliffe, L., Michaud, D. & Boag, P. T. 1998. Do female black-capped

chickadees prefer high ranking males as extra-pair partners? Behavioral Ecology and


Otter, K., McGregor, P. K., Terry, A. M. R. et al. 1999. Do female great tits (Parus major)



Otter, K., Stewart, I. K., Terry, A. M. R., McGregor, P. K. & Burke, T. 2001. Extra-pair

paternity in great tits in relation to manipulation of male signals. Journal of Avian

Biology, 32, 338–344.

Pomiankowski, A & Iwasa, Y. 1993. Evolution of multiple sexual preferences by Fisher’s

runaway process of sexual selection. Proceedings of the Royal Society of London, Series

B, 253, 173–181.

Qvarnstrom, A. & Forsgren, E. 1998. Should females prefer dominant males? Trends in


Radesater, T., Jakobsson, S., Andbjer, N., Bylin, A. & Nystrom, K. 1987. Song rate and

pair formation in the willow warbler, Phylloscopus trichilus. Animal Behaviour, 35,

1645–1651.

Ramsay, S. M, Otter, K. & Ratcliffe, L. M. 1999. Nest-site selection by female

black-capped chickadees: settlement based on conspecific attraction? The Auk,

116, 604–617.

Real, L. 1990. Search theory and mate choice I: models of single-sex discrimination.

American Naturalist, 136, 376–405.

Reid, M. L. 1987. Costliness and reliability in the singing behaviour of Ipswich

sparrows. Animal Behaviour, 35, 1735–1743.

Sæther, S. A. 2002. Female calls in lek-mating birds: indirect mate choice, female

competition for mates or direct mate choice? Behavioral Ecology, 13, 344–352.

Saino, N., Galeotti, P., Sacchi, R. & Møller, A. P. 1997a. Song and immunological

condition in male swallows (Hirundo rustica). Behavioral Ecology, 8, 364–371.

Saino, N., Primmer, C. R., Ellegren, H. & Møller, A. P. 1997b. An experimental study of

paternity and tail ornamentation in the barn swallow (Hirundo rustica). Evolution,

51, 562–570.

Slagsvold, T. & Lifjeld, J. T. 1994. Polygyny in birds: the role of competition between

females for male parental care. American Naturalist, 143, 59–94.

Slagsvold, T., Lifjeld, J. T., Stenmark, G. & Breiehagen, T. 1988. On the cost of searching

for a mate in female pied flycatchers Ficedula hypoleuca. Animal Behaviour, 36,

433–442.



Smith, S. M. 1988. Extra-pair copulations in black-capped chickadees: the role of the

female. Behaviour, 107, 15–23.

Thomas, R. J. 1999a. Two tests of a stochastic dynamic programming model of daily

singing routines in birds. Animal Behaviour, 57, 277–284.

1999b. The effect of variability in the food supply on the daily singing routines of

European robins: a test of a stochastic dynamic programming model. Animal

Behaviour, 57, 365–369.

Thomas, R. J. & Cuthill, I. C. 2002. Body mass regulation and the daily singing routines

of European robins. Animal Behaviour, 63, 285–292.

Tobias, J. A. & Seddon, N. 2002. Female begging in European robins: do neighbors

eavesdrop for extrapair copulations? Behavioral Ecology, 13, 637–642.

Todt, D. & Naguib, M. 2000. Vocal interactions in birds: the use of song as a model in

communication. Advances in the Study of Behaviour, 29, 247–295.

Tubaro, P. L. & Segura, E. T. 1994. Dialect differences in the song of Zonotrichia capensis

in the Southern Pampas: a test of the acoustic adaptation hypothesis. The Condor,

96, 1084–1088.

Tubaro, P. L., Segura, E. T & Handford, P. 1993. Geographic variation in the song of the

rufous-collared sparrow in eastern Argentina. The Condor, 95, 588–595.

Wagner, R. H. 1991. Evidence that female razorbills control extra-pair copulations.

Behaviour, 118, 157–169.

Weary, D. M. & Krebs, J. R. 1992. Great tits classify songs by individual voice

characteristics. Animal Behaviour, 43, 283–287.

Wetton, J. H., Burke, T., Parkin, D. T. & Cairns, E. 1995. Single-locus DNA fingerprinting

reveals that male reproductive success increases with age through extra-pair

paternity in the house sparrow (Passer domesticus). Proceedings of the Royal Society of


Wiegmann, D. D., Real, L. A., Capone, T. A. & Ellner, S. 1996. Some distinguishing

features of models of search behavior and mate choice. American Naturalist, 147,

188–204.

Wiley, R. H. & Richards, D. G. 1982. Adaptations for acoustic communication in birds:

sound transmission and signal detection. In: Acoustic Communication in Birds, Vol. 1,

ed. D. E. Kroodsma & E. H. Miller. New York: Academic Press, pp. 131–181.

Wittenberger, J. F. 1983. Tactics of mate choice. In: Mate Choice, ed. P. Bateson.

Cambridge, UK: Cambridge University Press, pp. 435–447.

Yasukawa, K., Blank, J. L. & Patterson, C. B. 1980. Song repertoires and sexual selection

in the red-winged blackbird. Behavioral Ecology and Sociobiology, 7, 233–238.

P1: IRK/KAA P2: IYP-KOD0521823617c08.xml CU1917B/McGregor 0 521 582361 7 April 11, 2005 13:38

8

Predation and noise in communicationnetworks of neotropical katydids

a l e x a n d e r b . l a ng , i ng e b o rg t e p p n e r , m a n f r e dh a r t b au e r & h e i n e r r Om e r

Karl-Franzen University, Graz, Austria

Introduction

Intraspecific acoustic communication in grasshoppers or katydids ap-

pears to be a very simple and straight forward behaviour: one sex – usually the

male – produces an acoustic signal, and the female, once perceiving and recogniz-

ing the signal as species specific, shows some kind of response, either an acoustic

reply or a phonotactic movement to the male. However, the system is far from

being that simple and involves more than just a sender and receiver. First, com-

munication usually takes place in a physically complex environment, where sound

signals are subject to attenuation and degradation, depending on the carrier fre-

quencies, which are often in the high-sonic or ultrasonic range because of the

small size of the sound radiating structures (Wiley & Richards, 1978; Michelsen,

1992). In addition, the physical conditions of the transmission channel for the

sound may vary strongly during day or night and with weather conditions; con-

sequently, the ability to detect and localize a signal undergoes strong variations.

Second, insects often aggregate and communicate in areas rich in resources or

at periods of the day or night favouring mate attraction. As a result of many sig-

nallers calling in close proximity, masking interference will take place at the site

of receivers, depending on the spacing, as well as the kind and extent of signal tim-

ing. Since such favourable areas and times for signalling are similar for different

species, heterospecific choruses may be formed with impressive sound pressure

levels of biological background noise, which further complicates the detection

of a signal (reviewed for katydids by Schatral (1990)). Third, a female might gain

fitness benefits (directly or indirectly) by choosing a male based on variation of


152


Neotropical katydids: predation and noise in networks 153

particular properties of the calling song. Such a female preference may drive the

evolution of the signal in a particular direction. For example, in insects, patterns of

preferences based on call rate or the duration of long calls are usually highly direc-

tional, whereas those based on pulse rate or frequency are stabilizing (reviewed by

Gerhardt & Huber, 2002). Female preferences can also result from a sensory bias in

the sensory or nervous system of receivers (Ryan, 1990; Ryan & Keddy-Hector, 1992;

Ryan & Rand, 1993; Endler & Basolo, 1998). Thus sexual selection by female choice

can result in signal traits that enhance the mating success of males. Fourth, traits

preferred by females may also decrease male survivorship by increasing exposure

to predators. Acoustically orienting predators or parasitoids can use the same sig-

nals produced for mate attraction to identify, localize and home in on the signaller

(Cade, 1975; Belwood & Morris, 1987; Lehmann & Heller, 1998; Zuk & Kolluru, 1998;

Allen, 2000). Male fitness can also be decreased by increased competition resulting

from the signal attracting conspecific rivals. Both are cases of interceptive eaves-

dropping in the sense of Peake (Ch. 2). Fifth, signalling at the long duration and

high rate preferred by females may be limited by energetic constraints, as sound

production for small animals is rather inefficient and probably costly (Bailey et al.,

1993; Wagner & Hoback, 1999). Finally, as ectothermic animals, the motor output

of insects depends on the ambient temperature, and in consequence, the temporal

properties of calling songs are influenced by environmental temperature. Vertical

temperature gradients in a grasshopper’s habitat can be 10 ◦C in 30 cm (Romer,

2001); therefore, senders and receivers can differ strongly in body temperature.

As a consequence, the signaller’s temporal patterning of song may not match the

preference function of a receiver.

From this short summary, it is clear that some of the factors contributing to

the evolution of acoustic communication systems could interact in a complex

way. For example, if high predation risk forces a species to communicate acous-

tically at a time of day or night when acoustic competition with other species is

high, the consequence is a high degree of song interference and masking, and the

calling activity of one species can inhibit that of other species (Greenfield, 1988;

Romer et al., 1989). In this chapter we emphasize the importance of an ecological

(integrated) approach to communication networks. By focusing on the intraspe-

cific communication of a subfamily of neotropical katydids, we demonstrate the

complex dependency of predation and signalling, nocturnal ambient light levels,

masking noise levels and alternative signalling strategies.

Predation and antipredator defences in rainforest katydids

A key paper by Belwood & Morris (1987) (see also Belwood, 1990; Morris

et al., 1994) suggested that the evolution of specific anti-predator defences in a


154 A. B. Lang, I. Teppner, M. Hartbauer & H. Romer

family of neotropical katydids (Pseudophyllinae) is strongly influenced by an-

tipredator defences. Members of this subfamily differ strongly in appearance, be-

haviour and hearing from those of the other larger taxon, the Phaneropterinae.

The latter live in the canopy, have in general a green, leaf-like appearance and

are good flyers. Their ears are about 15 dB more sensitive than the ears of pseudo-

phyllines (unpublished results). By contrast, pseudophyllines live in the rainforest

understorey, they have a long, slender, fusiform body and are bad flyers.

Katydids are a major source of protein for diurnal predators such as birds (e.g.

Formicariidae, Furnariidae and others), rodents and small primates (Nickle &

Heymann, 1996; Martins & Setz, 2000). Some of these birds feed almost exclu-

sively on arthropods by searching curled dead leaves that hang from vegetation

in the lower understorey (Gradwohl & Greenberg, 1980, 1982, 1984; Remsen &

Parker, 1984), thereby counteracting one of the katydids’ primary defence strate-

gies, namely crypsis by general appearance and behaviour (Nickle & Castner, 1995).

During the night, foliage-gleaning bats (Micronycteris hirsuta, Lophostoma silvicolum)

eat large numbers of Pseudophylline katydids (Belwood, 1988). These bats are at-

tracted by calling songs or other sounds involved in phonotactic activities of their

prey. Forest-living katydids exhibit a range of behaviours and signal characters

that appear to be adaptations to avoid predation by these bats: a reduction in call

redundancy (duty cycles of 3% and less), high carrier frequencies over 20 kHz and

the partial (or in one species, complete) replacement of airborne sound signals by

substrate-borne vibrations (tremulation) (Belwood & Morris, 1987).

In this chapter, we present data about the antipredator behaviour of a neotrop-

ical katydid, and its consequences for signal detection in noisy rainforest.

Predation pressure and roost site selection

The study was conducted on Barro Colorado Island (BCI), Panama and

on nearby peninsulas and small islands. The 1500 ha island is located in central

Panama (09◦10′N, 79◦51′W) in Gatun Lake, part of the Panama Canal. BCI is almost

totally covered with secondary and primary semideciduous lowland tropical forest

(Foster & Brokaw, 1982). The study took place in February/March (dry season) and

June/July 2002 (beginning of the rainy season). We studied Docidocercus gigliotosi,

a Pseudophylline katydid with a medium-sized, long and slender brown body.

Its natural history is only poorly known, although it is one of the most common

katydids on the island (Belwood, 1988). M. hirsuta and Micronycteris megalotis are two

insectivorous bats that glean highly cluttered spaces (Kalko et al., 1996) and feed on

D. gigliotosi (established by identification of remains at bat roosts). This katydid con-

stitutes about 20% of the diet of M. hirsuta (Belwood, 1988; personal observations).



(a) (b)

(c)

20m

Fig. 8.1. Roosting of Docidocercus gigliotosi (Pseudophyllinae) in Aechmea magdalenae.

(a) The A. magdalenae plant; (b) the 2.5 cm leaf edge spines; (c) the location of A.

magdalenae plants in part of field ‘Zetek 15’ on Barro Colorado Island (Panama),

mapped with a geographical information system. Each plant is marked with a circle;

those occupied with one or more D. gigliotosi are shown with a filled circle.

We regularly found D. gigliotosi roosting during the day in Aechmea magdalenae,

a terrestrial bromeliad of the pineapple family that can grow to a height of 2.5 m

(Fig. 8.1a). A striking characteristic of these plants are numerous, inch-long spines

along the leaf edges (Fig. 8.1b). It is abundant throughout BCI, forming dense

stands of sometimes more than 1000 plants. The part of the field ‘Zetek 15’ shown

in Fig. 8.1 comprises about 480 plants covering an area of 2600 m2. The leaves of

the plant form a long tube in the centre and that is where most katydids were

found roosting.

We observed several individuals shortly before sunset and during the night

using infrared video cameras. Sunset occurred around 18:30 h and katydids usually

became active (climbing and cleaning themselves within the plant) between 19:00

and 19:30 h. Three males were observed tremulating for several seconds. Between

20:00 and 21:00 h they used nearby lianas or trees to climb up into the lower

canopy, where they could no longer be observed. We presume that they are active



in the canopy throughout the night, because they returned to their host plants

around 04:00 h and climbed back into the central tube by 04:30 h, which is about

one hour before sunrise. With a mark-and-recapture study, we confirmed this

general scheme of daytime inactivity within the roost plant, nocturnal activity in

the canopy approximately between 21.00 h and 04.00 h and the return to the plant.

We recaptured 35 (out of 65) adults within 17 days; only three recaptures were

more distant than 2 m from the marking site. The maximum recapture distance

was 10 m and we never found marked individuals in different A. magdalenae fields.

Some individuals were found in the same plant for a period of more than two

weeks and 66% of katydids were recaptured in the same plant.

D. gigliotosi were not randomly distributed among A. magdalenae plants but

were found to roost in taller plants in above average condition with leaf-litter-free

central tubes close to canopy access ‘walkways’ (A. B. Lang & H. Romer, unpublished

data). Figure 8.1c shows a field of such plants (Zetek 15), in which plants occupied

by one or more individuals are marked with filled dots. D. gigliotosi roosted in

plants that were significantly taller (mean height (± standard deviation) 1.68 ±0.3 m (n = 32)) than unoccupied plants (1.36 ± 0.36 m (n = 320); (two-tailed Mann-

Whitney U test, p < 0.0001). Similar results were obtained for two study periods in

February/March and May/June 2002. We attempted to quantify the quality of roost

plants by ranking the condition of each plant on a subjective scale from bad (0) to

excellent (3). This ranking included the state of desiccation, number of damaged

leaves, and number of fresh, fleshy leaves, in particular those in the centre.

A survey performed in July 2002 found that the average condition of A. mag-

dalenae plants occupied by katydids was significantly better than that of unoccu-

pied plants (occupied plants, mean rank (± STD) 2.55 ± 0.69 (n = 36); unoccupied

plants, mean rank (± STD) 1.72 ± 0.96 (n = 248); two-tailed Mann–Whitney U test

p< 0.0001). Similar results were found for a survey carried out in March 2002. Most

(81%) of the plants in which adult D. gigliotosi roosted had direct contact with, or

grew within 1 m of a tree or liana reaching at least to the lower canopy.

These data indicate that the life history of the Pseudophylline katydid D.

gigliotosi is strongly influenced by predators. Two pieces of evidence are consis-

tent with D. gigliotosi attempting to avoid predation. First, by roosting during the

day in the spiny bromeliad A. magdalenae they are protected from predatory birds

and mammals. Insectivorous birds can have a pronounced effect on populations

of their arthropod prey (e.g. Lepidoptera larvae: Holmes et al., 1979) and on BCI

Myrmotherula fulviventris (Formicariidae) spends 98% of its foraging time search-

ing aerial leaf litter for arthropods and about 20% of its prey items are crickets

and katydids (Gradwohl & Greenberg, 1982). Therefore, the usual pseudophyllines

habit of roosting in curled leaves is potentially risky. Heavy predation by birds and



other visually hunting predators selects for various patterns of crypsis in their

katydid prey (Belwood, 1990; Castner & Nickle, 1995) and also for their choice of

backgrounds on which to hide, and on other life-history traits. Second, D. gigliotosi

is not active when gleaning bats show most flight activity (which is likely to be

related to foraging activity). I. silvicolum has a peak in flight activity about one hour

after sunset, when it flies to its hanging perch (C. D. Weise, E. K. V. Kalko, personal

communication), and one hour before sunrise, when it flies back. Radio telemetry

showed a similar pattern of activity in M. hirsuta (S. Spehn, personal communica-

tion). This correlates with the finding that D. gigliotosi, one of the bat’s common

prey species, does not exhibit night-time activity until the major period of bat

flight activity is over, and they also return to their bromeliad roost before the bats

return to their roosts. Although one may argue that flight activity of bats does not

necessarily reflect the time of highest predation pressure for their prey, it is reason-

able to assume that bats would home in on katydid song and the noises caused by

prey flight or landing activity during this time.

The costs of nocturnal communication: masking interference

Many species of insect and anuran communicate acoustically at night

and the resulting multispecies choruses have high sound pressure levels (SPL)

and complex spectral properties. Figure 8.2a shows measurements of the SPL in

the rainforest on BCI over a period of 24 hours. During the day, the SPL was

rather low, measuring 40–50 dB. It rapidly increased shortly after sunset by some

20 dB as a result of calling activity of insects and frogs. SPL declined throughout

the night (depending on the moon cycle; see below), until it reached daytime

levels after sunrise. In such chorus noise, different species occupy different fre-

quency bands. In our recording, the most prominent frequencies were below 8 kHz

(calling songs of crickets) (Fig. 8.2b), but frequencies well above 20 kHz were also

obvious.

We used a ‘biological microphone’ to ‘listen’ through the ears of the biological

receiver to analyse the challenging problem of signal detection after sunset for

D. gigliotosi at the position of potential receivers. The biological microphone is a

small, portable outdoor neurophysiological set-up that records the action poten-

tial activity of a single, identified auditory interneuron of a katydid (Rheinlaender

& Romer, 1986; Romer & Lewald, 1992). Rather than analysing the properties of

signals and noise at the position of potential receivers with conventional micro-

phones, such a method allows one to listen through the ears of the biological

receivers and to draw conclusions from the analysis of afferent nervous activity

under these natural conditions.



Time of day (h)

Time (s)

Bac

kgro

und

nois

e le

vel (

dB S

PL)

Fre

quen

cy (

kHz)

(a)

(b)

70

60

50

40

06:00 12:00 18:00 24:00 06:00

Full moon

New moonLast quarter

Sunset Sunrise

30

20

10

15 20 25

Fig. 8.2. Measurement of the multispecies chorus on Barro Colorado Island (Panama).

(a) Sound pressure level (SPL) in the rainforest over a period of 24 hours, at three moon

phases. (b) Sonagram of a 16 second sound recording at the same site after sunset at

19.00 h. Note the different frequency bands between 3 and 7.5 kHz produced

predominantly by crickets, and those in the high-frequency and ultrasonic range

produced by katydids. The short duration and low redundancy calling song of

Docidocercus gigliotosi (frequency range 22–25 kHz) is marked by arrows.

A typical result is shown in Fig. 8.3a. The receiver was placed within the rain-

forest at 17.00 h, 10 m from a speaker broadcasting a conspecific calling song.

Since a female has no a priori knowledge about the presence of a signal, her only

information about the presence or absence of a signal is encoded in afferent ner-

vous activity such as shown in Fig. 8.3a. This task is apparently easy before sunset

(Fig. 8.3a, upper trace), because each burst of action potential activity (increase

in spike frequency) was the result of a conspecific stimulus. A detection criterion

based on bursts of action potentials or the corresponding increase in spike rate



(a)

(b)Spike frequency

17.00 h

18.30 h

150

100

50

5 s

Fig. 8.3. Outdoor action potential recording of an identified nerve cell of the afferent

auditory pathway of Docidocercus gigliotosi (see text). (a) Recording obtained within the

rainforest at 17.00 h (upper trace), and at 18.30 h (lower trace). Before sunset, masking

noise level is low and each conspecific stimulus elicits bursts of action potentials (hits

shown by open dots). After sunset, many bursts of action potentials are elicited by

noise (false alarms shown by stars). (b) Longer recording of action potentials after

sunset (lower trace) and the corresponding spike rate (upper trace). Note that a signal

presented at five second intervals (arrows on upper trace, points between traces)

elicits an increase in the spike rate, and noise may result in a stronger increase in the

spike rate than the signal.

would give ‘hits’ in terms of signal detection (Green & Swets, 1966). Indeed, in all

cases when there was an acoustic signal during the experiment at 17.00 h, there

was bursting activity in the nerve cell and there was no such activity when a signal

was absent; therefore, there were no ‘misses’ or ‘false alarms’, respectively.

However, this ideal situation for signal detection changed completely after sun-

set, when most katydids and other insects started to communicate acoustically.

The same preparation at exactly the same position in the rainforest now exhibited

high action potential activity (Fig. 8.3a, lower trace) and only an a priori knowledge

of the time of signalling allows correct detection of the stimuli. Using the same

detection criterion as in the situation before sunset would result in many false

alarms (i.e. identifying background noise as signals: stars in Fig. 8.3a, lower trace).

We measured a false alarm rate of more than 1400 in only five minutes, thus ex-

ceeding the hit rate dramatically, rendering communication between conspecifics



Male signal

Representation of signalwithin CNS of receiver

Fig. 8.4. The amplitude-modulated calling song of a male Thamnobates subfalcata

katydid (lower trace) and the corresponding representation of this song in the spike

discharge of a nerve cell (omega-neuron) in the afferent auditory pathway of the

central nervous system (CNS) in a female Docidocercus gigliotosi receiver.

rather ineffective. Figure 8.3b illustrates this point over a longer time scale and

also shows that a signal presented at regular intervals of five seconds elicited a

corresponding increase in the spiking rate, but that noise pulses may result in a

stronger increase in the spike rate than the signal.

Signal detection would be improved by increasing either the duration or the

rate of signalling, and indeed we found such an effect with our preparation. Sim-

ilar experiments to those shown in Fig. 8.3 clearly indicate that the rate of hits

increases and the rate of false alarms decreases with increased signal redundancy

and duration. However, as pointed out by Belwood & Morris (1987), eavesdropping

(interceptive eavesdropping in the sense of Peake (Ch. 2)) by gleaning bats excludes

such a solution and illustrates the opposing selection pressures of avoiding pre-

dation and signalling effectively.

How do the insects solve the problem? A closer look at the signals used by males

offers a possible solution. Figure 8.4 shows a typical, short amplitude-modulated

signal of a male katydid Thamnobates subfalcata and the corresponding represen-

tation of this signal in the spike discharge of a nerve cell (omega-neuron) in the

afferent pathway of a receiver. If the parameters of the spike discharge in response

to a species-specific call differ from those in bursts elicited by background noise,

this difference could be used by the nervous system to discriminate signals from

noise.

To investigate this further we used the biological microphone to record action

potentials in the rainforest at about 21.00 h, when the level of the background

noise was still high (Fig. 8.2). A male signal was broadcast every two seconds

at 10 m from the preparation. The SPL of the signal was adjusted to a value

of approximately 10 dB above the neuron’s masked threshold. We then used a

self-developed Delphi-application (Delphi 6, Borland Software Corporation, Scotts

Valley, CA 95066-3249, USA) to extract bursts from afferent spike recordings. We



compared several features of bursts (spike variables: the number of spikes in the

sequence; the mean, maximum, minimum; variance; average deviation; standard

deviation; skewness; kurtosis of the interval between two spikes) between hits (i.e.

bursts induced by the playback call) and false alarms (i.e. bursts elicited by back-

ground noise). One half of the recording (30 minutes) was used as the training set to

compute the decision tree: that is, the program learned the specific spike variables

that characterize hits. This tree was then evaluated for the remaining 30 minutes of

the recording, the validation set, to see whether it could detect such responses in

a noisy background. Machine learning on the basis of decision-tree learning is

one of the most widely used and practical data-mining methods to classify very

large amounts of data. It is a method for approximating discrete-valued functions,

in which the learned function is represented by a decision tree that is robust to

noisy data and capable of learning disjunctive expressions (Mitchell, 1997). We

used the algorithms J48 (pruned, unpruned) (Quinlan, 1993) and PART (Frank &

Witten, 1998) for classification of bursts within spike trains, calculating the re-

sults with the Java application WEKA (Trigg et al., 1999). Methods of decision-tree

learning such as J48 and PART search a completely expressive hypothesis space

and thus avoid the difficulties of restricted hypothesis spaces (Mitchell, 1997).

To avoid so-called ‘over fitting’ (see also Mitchell, 1997), we conducted a tenfold

cross-validation.

The unpruned J48 decision tree classified 95.4% of the bursts correctly, meaning

that it was able to distinguish between the double pulse signal and background

noise with an error of only 4.6% (Table 8.1). The effect of varying the duration

of the playback signal (7, 70 and 700 milliseconds) at a constant signal rate of

0.5/second is shown in Table 8.1. The most obvious effect is the lower detectability

of the 7 millisecond signal by all three algorithms; however, there is little increase

in detectability between the 70 and 700 millisecond signals. A similar result was

found for the grasshopper Chorthippus biguttulus, where signal detection improved

with increasing signal duration up to 450 milliseconds but did not improve further

with longer signals (Ronacher et al., 2000).

Although such results do not tell us that the insect’s nervous system makes

use of this information, it does show how bursts of action potentials in response

to conspecific song can, in principle, be discriminated from bursts produced by

other sound sources. The most important spike variables used by the decision trees

in discriminating signals from noise were the kurtosis, number of spikes, mean

and variance. The minimum and maximum spike interval was often used in final

decisions of the trees. Because some of these parameters (e.g. number of action

potentials or the minimum or maximum spike intervals) are also relevant in real

nervous systems for processing and discriminating sensory information, one can

assume that the insect nervous system can also solve this discrimination task.



Table 8.1. The effects of duration (7, 70, 700 milliseconds) of the

playback signal (presented at 0.5/second) on signal detection in

afferent spike trains of a receiver based on decision tree learning

with three different algorithms: J48, unpruned J48, and PARTa

Algorithm Correctly classified bursts (%)

Stimulus 7 ms (n = 609)

J48 80.13

J48 (unpruned) 79.64

PART 85.22


J48 94.77


PART 94.77


J48 95.37


PART 95.37

aSee text for further explanation.

Using machine-learning procedures to evaluate afferent spike patterns in sen-

sory systems may also enable us to look more closely at the strategic design of

signals (Guilford & Dawkins, 1991). For example, given the advantage of short-

duration and low-redundancy signalling in the presence of interceptive eaves-

dropping predators, what degree of amplitude modulation in a signal is necessary

to make its representation in afferent channels reliably different from heterospe-

cific signals? This is part of ongoing research on a variety of katydid species on BCI,

some of which pose a real challenge for signal detection by using signals of only a

few milliseconds duration. Additional behavioural experiments under noisy con-

ditions with the same species are urgently needed in order to show whether the

insects’ nervous systems can solve the task.

Variation in ambient light and noise levels, and the use

of a conditional communication strategy

Two features of neotropical rainforest at night vary with the lunar cycle:

insect abundance and noise level. We shall argue that they are related to predation

pressure on communication. We quantified the effect of lunar cycle on katydid

abundance on BCI by collecting at mercury vapour lights in December 1999 and

April 2001 at 21:00 and 24:00 h.



Day in moonphase

Col

lect

ed k

atyd

ids

per

nigh

t

New

moo

n

Ful

l moo

n

40

30

20

10

0

1 4 7 10 13 16 19 22 25 28

Fig. 8.5. The number of katydids collected at mercury vapour lights on Barro Colorado

Island (Panama) in relation to moon phase. Note that the higher variation at new

moon may be because of rainfall during these nights, when many katydids hide under

leaves and do not fly.

There was a significant relationship between moon phase and the number

of katydids collected (Fig. 8.5): at full moon the number of katydids collected

approached zero and it reached a maximum at new moon. As these data are similar

to results obtained with other insects (e.g. Hardwick, 1972) and are also similar

to collections made with suction traps, these cycles in abundance reflect natural

activity patterns. Light intensity may vary by three to four orders of magnitude

between full moon and new moon (Erkert, 1974). Full-moon ambient light levels

are high enough for humans to orient easily in the forest understorey with their

dark-adapted eyes. It is, therefore, likely that these light conditions allow a variety

of predators to hunt visually and, consequently, their potential prey must adopt

a cryptic lifestyle.

We quantified the effect of lunar cycle on background noise level on BCI with a

continuous recording system. The system consisted of a sound level meter (CEL 414

plus attached CEL-296 digital filter with settings A- weighting and slow time con-

stant) with a condenser microphone (LD 2540, type 4133, range 4–40 kHz). The

set-up was protected from humidity and rainfall and heated to 2 ◦C above am-

bient temperature with an infrared bulb to prevent fogging of the microphone

membrane. The DC output of the sound level meter was monitored at intervals of

five seconds with a Maclab/Powerlab 4e data acquisition system (AD Instruments

Pty Ltd) connected to a portable computer (Sony PCG-F707). Recordings were made

from the end of October to early December 2001, as well as in February, May and

June 2002.

Figure 8.2a shows representative examples of noise measurements over

24 hours at full moon, new moon and the last quarter of a lunar cycle. The increase



Tremulation

Stridulation 2 s

Fig. 8.6. The two types of mating signal produced by a male Docidocercus gigliotosi.

Tremulations (vibratory signals, upper trace) were recorded without mechanical

contact by laser doppler vibrometry, simultaneously with airborne sound signals

produced by elytral stridulation (lower trace). The photograph shows a male on a large

leaf.

in noise at sunset and the decrease at sunrise are common to all recordings. How-

ever, at full moon, noise levels decrease after sunset and for the rest of the night

the noise level is only 10 dB above the daytime level. As a result, the masking noise

level between 09:00 and approximately 05:00 h varies cyclically with the moon

phase; the amplitude of variation is about 10 dB.

Given the fact that the masking noise is the result of acoustic signalling, pre-

dominantly by insects, the drop in noise level at full moon is best explained by

species and/or individuals reducing or abandoning signalling with airborne sound.

We have argued above that the nocturnal lifestyle of many insects avoids predation

from visually hunting predators (e.g. rodents) and that light intensity at full moon

may allow increased visual predation. Therefore, a cryptic lifestyle may include

cryptic signalling.

Direct evidence for this hypothesis comes from the signalling behaviour of

D. gigliotosi. Males produce airborne sound with the usual elytral stridulation and

also tremulations, when the insect rapidly shakes its body up and down or drums

with the abdomen on the substratum (Morris et al., 1994). Tremulations are trans-

mitted through the substratum and females respond with tremulations of their

own. Although the active space of a tremulation signal is limited to the plant

where the tremulation is produced, it may travel 2–3 m (Michelsen et al., 1982;

Markl, 1983). Tremulation is an effective way to communicate in the presence of

acoustic interceptive eavesdroppers such as gleaning bats because only receivers

equipped with sensitive vibration receptors (e.g. spiders) can intercept the signal.

Figure 8.6 demonstrates that the duration and redundancy of the tremulation



New moon

Time of day

Num

ber

of e

vent

s/5

min

Num

ber

of e

vent

s/5

min

Full moon

60

50

40

30

20

10

0

60

50

40

30

20

10

0

18:0

0

19:1

2

21:3

6

00:0

0

01:1

2

02:2

4

03:3

6

22:4

8

20:2

4

(a)

(b)

Fig. 8.7. The pattern of production of the two types of mating signal (tremulation,

open squares; airborne sound, filled squares and grey area) produced by a male

Docidocercus gigliotosi at new moon (a) and full moon (b). Males were separated from

females a day before data were collected, caged and placed in the rainforest

understorey. An infrared camera recorded both signals over a period of about

nine hours, from sunset to 03.00 h.

signal of D. gigliotosi is higher than the stridulatory airborne sound signal by orders

of magnitude.

As expected if male D. gigliotosi vary the relative proportion of these two modes

of signalling in relation to the chance of predation from gleaning bats, at new

moon, or in the laboratory in complete darkness, males become active about half

an hour after sunset and begin signalling with a period of high-rate tremulation,

followed by a prolonged period of airborne sound production, often for many

hours (Fig. 8.7a). At full moon the onset of signalling after sunset is often delayed,

airborne sound signalling is strongly reduced and tremulation is more common.

It should also be noted that D. gigliotosi reduces acoustic signalling at full moon

despite the fact that the lower noise levels would allow better detection of conspe-

cific signals. The observations are consistent with D. gigliotosi having a conditional

strategy of signalling, where fairly cryptic (i.e. short duration, low redundancy)

airborne sound production is replaced by the even more private mode of commu-

nication with tremulations.



Summary

In this chapter we have emphasized the importance of predation on the

ecology and acoustic communication of a subfamily of neotropical katydids. Pseu-

dophylline katydids switch from airborne sound signalling to tremulation be-

cause of high predation by passive listening bats. This results in either decreased

active space of their signals or decreased detectability of airborne sound signals

in high levels of background nocturnal noise. Neurophysiological experiments

and decision-tree learning algorithms nevertheless indicate that information for

discrimination between signals and noise is still preserved within afferent spike

discharges. Daytime and night-time predation also appears to restrict individuals

to roost sites within bromelid plants. D. gigliotosi exhibits a strong site dependency

for a particular plant in a field over several weeks; consequently, there is very

little horizontal movement of males and females in a population. In conjunction

with a reduced active space of acoustic and/or vibrational signalling, and reduced

flight capability, this situation may strongly reduce the chances of matings with

individuals of neighbouring populations. Current population genetic analysis will

show whether genetic exchange between populations is suppressed, despite the

lack of a geographical barrier between populations.

Acknowledgements

Inspiring discussion with Elisabeth K. V. Kalko, Christa D. Weise, Dina K. N. Dechmann

and Sabine Spehn played a major role in maturing the ideas presented here. We acknowledge the

logistic support of the Smithsonian Tropical Research Institute, Panama. Barbara Bliem, Franz

Kainz, Birgit Roehnfeld and Iris Strauss assisted in the field and in the laboratory. We also thank

Peter McGregor and two anonymous reviewers for many suggestions to improve the manuscript.

Research was supported by the Austrian Science Fund (FWF P14257-BIO to H. R.) and a Ph.D.

scholarship (Austrian Academy of Sciences to A. B. L.).

References

Allen, G. R. 2000. Call structure variability and field survival among bushcrickets

exposed to phonotactic parasitoids. Ethology, 106, 409–423.

Bailey, W. J., Withers, P. C., Endersby, M. & Gaull, K. 1993. The energetic costs of calling

in the bushcricket Requena verticalis (Orthoptera: Tettigoniidae: Listroscelidinae).

Journal of Experimental Biology, 178, 21–37.

Belwood, J. J. 1988. The influence of bat predation on calling behaviour in Neotropical

forest katydids (Insecta: Orthoptera: Tettigoniidae). Ph.D. Thesis, University of

Florida, Gainesville, FL.


Neotropical katydids: predation and noise in communication networks 167

1990. Anti-predator defences and ecology of Neotropical forest katydids,

especially the Pseudophyllinae. In: The Tettigoniidae: Behaviour, Systematics,

Evolution, ed. W. J. Bailey & D. C. F. Renz. Bathurst, Australia: Crawford House

Press, pp. 6–26.

Belwood, J. J. & Morris, G. K. 1987. Bat predation and its influence on calling behaviour

in Neotropical katydids. Science, 238, 64–67.

Cade, W. 1975. Acoustically orienting parasitoids: fly phonotaxis to cricket song.

Science, 190, 1312–1313.

Castner, J. L. & Nickle, D. A. 1995. Observations on the behaviour and biology of

leaf-mimicking katydids (Orthoptera: Tettigoniidae: Pseudophyllinae:

Pterochrozini). Journal of Orthopteran Research, 4, 93–98.

Endler, J. A. & Basolo, A. L. 1998. Sensory ecology, receiver biases and sexual selection.

Trends in Ecology and Evolution, 13, 415–420.

Erkert, H. G. 1974. Der Einfluß des Mondlichtes auf die Aktivitatsperiodik

nachtaktiver Saugetiere. Oecologia, 14, 269–287.

Foster, R. B. & Brokaw, N. V. L. 1982. Structure and history of the vegetation of Barro

Colorado Island. In: The Ecology of a Tropical Forest, ed. E. G. Leigh, A. S. Rand & D. M.

Windsor. Washington, DC: Smithsonian Institute Press, pp. 151–172.

Frank, E. & Witten, I. H. 1998. Generating accurate rule sets without global

optimization. In: Proceedings of the 15th International Conference on Machine

Learning, Madison, WI, ed. J. Shavlik. San Francisco, CA: Morgan Kaufmann,

pp. 144–151.



Gradwohl, J. A. & Greenberg, R. 1980. The formation of antwren flocks on Barro

Colorado Island, Panama. The Auk, 97, 395.

1982. The effect of a single species of avian predators on the arthropods of aerial

leaf litter. Ecology, 63, 581–583.

1984. Search behaviour of the checker-throated antwren foraging in aerial leaf

litter. Behavioral Ecology and Sociobiology, 15, 281–285.

Green, D. M. & Swets, J. A. 1966. Signal Detection Theory and Psychophysics. New York:

Wiley.

Greenfield, M. D. 1988. Interspecific acoustic interactions among katydids

Neoconocephalus: inhibition-induced shifts in diel activity. Animal Behaviour, 36,

684–695.

Guilford, T. & Dawkins, M. S. 1991. Receiver psychology and the evolution of animal

signals. Animal Behaviour, 42, 1–14.

Hardwick, D. F. 1972. The influence of temperature and moon phase on the activity of

noctuid moths. Canadian Entomologist, 104, 1767–1770.

Holmes, R. T., Schultz, J. C. & Nothnagle, P. 1979. Bird predation on forest insects: an

exclosure experiment. Science, 206, 462–463.

Kalko, E. K. V., Handley, C. O., Jr & Handley, D. 1996. Organization, diversity, and

long-term dynamics of a Neotropical bat community. In: Long-term Studies of



Vertebrate Communities, ed. M. L. Cody & J. A. Smallwood. New York: Academic

Press, pp. 503–553.

Lehmann, G. U. C. & Heller, K. G. 1998. Bushcricket song structure and predation by

the acoustically orienting parasitoid fly Therobia leonidei (Diptera: Tachinidae:

Ormiini). Behavioral Ecology and Sociobiology, 43, 239–245.

Markl, H. 1983. Vibrational communication. In: Neuroethology and Behavioural

Physiology, ed. F. Huber & H. Markl. Berlin: Springer, pp. 332–353.

Martins, M. M. & Setz, E. F. Z. 2000. Diet of buffy tufted-eared marmosets (Callithrix

aurita) in a forest fragment in southeastern Brazil. International Journal of

Primatology, 21, 467–476.

Michelsen, A. 1992. Hearing and sound communication in small animals:

evolutionary adaptations to the law of physics. In: The Evolutionary Biology of

Hearing, ed. E. D. Webster, R. R. Fay & A. N. Popper. New York: Springer, pp. 61–77.

Michelsen, A., Fink, F., Gogala, M. & Traue, D. 1982. Plants as transmission channels

for insect vibrational songs. Behavioral Ecology and Sociobiology, 11, 269–281.

Mitchell, T. M. 1997. Machine Learning. New York: McGraw-Hill.

Morris, G. K., Mason, A. C., Wall, P. & Belwood, J. J. 1994. High ultrasonic and

tremulation signals in Neotropical katydids (Orthoptera: Tettigoniidae). Journal of

Zoology (London), 233, 129–163.

Nickle, D. A. & Castner, J. L. 1995. Strategies utilized by katydids (Orthoptera:

Tettigoniidae) against diurnal predators in rainforests of northeastern Peru.

Journal of Orthopteran Research, 4, 93–98.

Nickle, D. A. & Heymann, E. W. 1996. Predation on orthoptera and related orders of

insects by tamarin monkey, Saguinus mystax and Saquinus fuscicollis (Primates:

Callitrichidae), in northeastern Peru. Journal of Zoology, (London), 239, 799–819.

Quinlan, J. R. 1993. C.4.5: Programs for Machine Learning. San Francisco, CA: Morgan

Kaufmann.

Remsen, J. V. & Parker, T. A. 1984. Arboreal dead-leaf-searching birds of the neotropics.

The Condor, 86, 36–41.

Rheinlaender, J. & Romer, H. 1986. Insect hearing in the field: the use of identified

nerve cells as biological microphones. Journal of Comparative Physiology, A, 158,

647–651.

Ronacher, B., Krahe, R. & Hennig, R. M. 2000. Effects of signal duration on the

recognition of masked communication signals by the grasshopper Chorthippus

biguttulus. Journal of Comparative Physiology, A, 186, 1065–1072.

Romer H. 2001. Ecological constraints for sound communication: from grasshoppers

to elephants. In: Ecology of Sensing, ed. F. G. Barth & A. Schmid. Berlin: Springer,

pp. 59–77.

Romer, H. & Lewald, J. 1992. High-frequency sound transmission in natural habitats:

implications for the evolution of insect acoustic communication. Behavioral


Romer, H., Bailey, W. & Dadour, I. 1989. Insect hearing in the field. III. Masking by

noise. Journal of Comparative Physiology, A, 164, 609–620.


Neotropical katydids: predation and noise in communication networks 169

Ryan, M. J. 1990. Sensory systems, sexual selection, and sensory exploitation. Oxford

Surveys in Evolutionary Biology, 7, 157–195.

Ryan, M. J. & Keddy-Hector, A. 1992. Directional patterns of female mate choice and

the role of sensory biases. American Naturalist, 139, S4–S35.

Ryan, M. J. & Rand, S. A. 1993. Sexual selection and signal evolution: the ghost of biases

past. Philosophical Transactions of the Royal Society of London, Series B, 340, 187–195.

Schatral, A. 1990. Interspecific acoustic behaviour among bushcrickets. In: The

Tettigoniidae: Behaviour, Systematics, Evolution, ed. W. J. Bailey & D. C. F. Renz.

Bathurst, Australia: Crawford House Press, pp. 150–165.

Trigg, L., Hall, M. & Frank, E. 1999. WEKA (Java collection of machine learning

algorithms). Hamilton, New Zealand: University of Waikato;

http://www.cs.waikato.ac.nz/ml/weka/.

Wagner, W. E., Jr & Hoback, W. W. 1999. Nutritional effects on male calling behaviour

in the variable field cricket. Animal Behaviour, 57, 89–95.

Wiley, R. H. & Richards, D. G. 1978. Physical constraints on acoustic communication in

the atmosphere: implications for the evolution of animal vocalizations. Behavioral

Ecology and Sociobiology 3, 69–94.

Zuk, M. & Kolluru, G. R. 1998. Exploitation of sexual signals by predators and

parasitoids. Quarterly Review of Biology, 73, 415–438.


9

Nestling begging as acommunication network

a n d r e w g . h o r n & m a r t y l . l e o n a r d

Dalhousie University, Halifax, Canada

Introduction

In many bird species, young beg for care from their parents. A parent

arriving at the nest with food is met by begging nestlings, which are waving their

wings, calling and stretching to expose brightly coloured gapes, all within the

confines of a nest that may contain several other begging nestlings. This mode

of parent–offspring communication has become a model for the study of the

evolution of biological signalling.

Hungrier nestlings beg more intensely, so the parent can use the display to

decide which nestling to feed and to decide how soon it should return to the nest

with food (reviewed by Budden & Wright, 2001). The fact that the parent can extract

information on nestling hunger from such a confusing burst of signalling raises

numerous questions. How does each nestling ensure that its own signal of need

is received above the din of its nestmates’ displays? How do parents differentiate

among these displays to choose which nestling to feed? How much do the displays,

as opposed to the physical jostling toward the parent that also goes on in the nest,

determine which nestlings are fed?

To answer such questions we need to understand how the begging behaviours

of whole broods function together. Concepts derived from the new field of com-

munication networks seem well suited to this task but have not yet been explicitly

applied to begging. As currently defined (McGregor & Dabelsteen, 1996; McGregor

& Peake, 2000), a communication network forms whenever several individuals

communicate within transmission range of each other’s signals. Nestlings noisily

begging within the confines of a nest clearly fit this definition, since most or all of

the nestlings within a brood are within transmission range of each other’ssignals.


170


Nestling begging as a communication network 171

In this chapter, we hope to show that considering begging as a communication

network yields new insights, not only into begging behaviour but also into commu-

nication networks in general. We begin by briefly summarizing previous research

on begging, most of which has treated the display as dyadic communication: that

is, as signalling from one individual, the nestling (or the brood considered as send-

ing one joint signal), to one receiver, the parent. We then apply ideas from studies

of communication networks to nestling begging, identifying several conceptual

issues that we think studies of begging can help to clarify. We also discuss aspects

of the design of begging and parental behaviour that may have evolved in re-

sponse to the network environment and finally we make some suggestions for

future work.

Begging as dyadic communication

In this section, we summarize the theoretical and empirical work on beg-

ging to date, most of which has treated parent–offspring communication as a

dyadic communication system. This summary provides background information

for the discussion of communication networks that follows, while also illustrat-

ing some of the strengths and weaknesses of the dyadic approach to begging

behaviour.

Theoretical work

Begging has attracted considerable attention from evolutionary biologists

largely because of its apparently needless conspicuousness. Because parents are

only a few centimetres away from their young, it is not clear why offspring signal

for food with such an elaborate display. Perhaps the best-known explanation for

this apparent extravagance stems from parent–offspring conflict theory (reviewed

by Godfray, 1995; Parker et al., 2002). Natural selection favours parents that dis-

tribute resources optimally amongst both their current and future offspring. Each

of these offspring, however, is selected to solicit resources so as to benefit its own

fitness, rather than the inclusive fitness of its siblings. Therefore, offspring might

signal for resources that parents would do better giving to siblings or reserving

for future broods. To overcome parental reluctance, offspring may have to send

exaggerated signals of need (Trivers, 1974; Godfray, 1995).

This basic explanation has been revised or extended in various ways, making

the parent–offspring dyad one of the most thoroughly modelled animal communi-

cation systems. Some of the most influential models, both for begging and for ani-

mal signals in general, have asked how reliable signalling can evolve in the face of

conflict between signallers and receivers (reviewed by Godfray & Johnstone, 2000;

Johnstone & Godfray, 2002). Specifically, if young are prone to exaggerate, then


172 A. G. Horn & M. L. Leonard

why would parents respond at all to begging signals? The answer is that, whereas

parents might easily be able to assess some aspects of their nestlings, like their

size, parents might not be able to assess important aspects of their nestlings’needs,

for example their immediate need for food. If begging provides information on

these aspects of nestling need, then parents should provision nestlings accord-

ing to variation in the begging signal. This situation can be evolutionarily stable,

however, only if the signal is costly for nestlings to produce. Therefore, in effect,

nestlings might have to put on a costly begging display to prove that they really

are hungry (Godfray, 1991).

These results have been largely responsible for the general acceptance of the

idea that reliable signals must be costly if they are to evolve. Some of the com-

plexities of this story are less widely known, however. For instance, recent models

have suggested that, in some situations, nestlings might signal their needs ac-

curately without large costs, for example if exaggeration draws so much care

away from siblings that the cost to the signaller’s inclusive fitness outweighs

the direct benefits of the extra signalling (Maynard Smith, 1994; Bergstrom &

Lachmann, 1998; Johnstone, 1999; Price et al., 2002).

For the purposes of this chapter, two features of theoretical work on begging

particularly stand out. First, these models have focused on fundamental issues

in dyadic communication, such as how signalling can evolve despite conflicts of

interest between signallers and receivers. Thus they are relevant to our under-

standing of a wide range of communication systems. Second, the emphasis these

models have placed on particular aspects of signalling, such as its honesty and

costliness, has led empirical studies to focus on these aspects of begging to the

neglect of others (see below). One of these neglected aspects is the communication

network in which begging occurs; although recent attempts to model the effects

of signalling on nestmates (reviewed by Royle et al., 2002; Johnstone & Godfray,

2002), which we discuss further below, are steps in that direction.

Empirical work

The theoretical possibility that begging might be exaggerated led many

researchers to test whether begging is indeed a reliable signal of need. Studies in

a wide range of species confirm that the intensity of both the visual and vocal

aspects of the display increase with food deprivation (Budden & Wright, 2001). In

turn, parents use the begging signal in two ways to make provisioning decisions.

First, the more intense the begging of the brood as whole, the more often parents

return to the nest with food. This level of response has been shown most clearly

in experiments in which playback of nestling begging calls stimulates higher

provisioning rates (Budden & Wright, 2001). Second, once parents arrive at the

nest, nestlings that beg more intensely than their nestmates are more likely to be



fed. Experiments again provide the clearest demonstrations of this effect: parents

are more likely to direct feedings to nestlings with brighter gapes (Gotmark &

Ahlstrom, 1997; Kilner, 1997; Saino et al., 2000; Saino and Møller, 2002) or to

nestlings placed next to speakers playing higher call rates (Leonard & Horn, 2001a;

Kilner, 2002a; but see Glassey & Forbes, 2002a). Therefore, begging appears to

communicate to parents the requirements both of the brood as a whole and of

individual nestlings.

Begging is more than a simple cry for food, however, for two reasons. First, food

deprivation is not the only aspect of nestling need that the begging display adver-

tises. For example, in some species begging may signal long-term nutritional need

as opposed to the short-term hunger described above, with nestlings in poorer

condition (e.g. having lower mass than nestmates) begging more than their nest-

mates (Price et al., 2002). Additionally, some aspects of begging, especially begging

calls, can change when nestlings lose heat, thus signalling the need for brooding

(Evans, 1994; Leonard & Horn, 2001b; Clotfelter et al., 2003; B. Glassey, personal

communication). Finally, gape colour in some species may advertise a nestling’s

immunocompetence (Saino & Møller, 2002). Clearly, the message that begging is

sending may be more complex than just short-term hunger.

A second complicating factor is the effect of siblings on nestling begging. Beg-

ging intensity, whether measured by the intensity of the postural display or overall

call rate, increases with brood size in many species (Budden & Wright, 2001) and

may also increase when nestmates beg (e.g. Leonard & Horn, 1998). Also, nestlings

compete physically for access to parents (see below) and their display and its effect

on parents may vary according to the nature and intensity of this physical com-

petition (e.g. Price et al., 1996; Cotton et al., 1999). Interest in the effects of both

signalling interactions and physical competition among nestmates has mainly fo-

cused on how they complicate honest signalling of need (e.g. Rodrıguez-Girones et

al., 2001; Price et al., 2002). We will be discussing them further below because they

are clearly central to any discussion of begging as a communication network.

Summary

This brief review shows that the main emphasis of work on begging has

been on how it functions as a signal of need from nestlings to parents. Begging has

been treated mainly as a dyadic signalling system: that is involving one signaller

(the nestling or the brood considered as sending one joint signal) and one re-

ceiver (the parent). Siblings have been included in the picture, but mainly because

they might affect the dyadic signalling of need. Only recently have researchers

started to consider the effects of competing signalling by nestmates in any de-

tail, an important step toward treating the begging brood as a communication

network.



Begging as a communication network

If we are to broaden our view of begging to include the communication

network in which it occurs, we must first characterize that communication net-

work. By definition, a brood of begging young is a communication network because

nestmates are all within range of each other’s signals (McGregor & Dabelsteen,

1996; McGregor & Peake, 2000). Going beyond this definition, however, to charac-

terize the network and explore its implications, raises more conceptual challenges

than this simple definition might suggest.

In this section, we discuss three of these issues. First, to apply the definition of

communication networks at all, we must distinguish signalling from other acts.

This can be especially problematic in the case of begging, in which signalling and

direct physical competition are tightly linked. Second, to examine some of the

more interesting implications of the network, we must carefully consider the na-

ture of signals and signalling interactions – again, a challenging distinction when

applied to begging. Third, there are factors, such as the genetic relatedness of

nestlings, which are at least as important for characterizing this communication

network as the overlapping transmission ranges of signals that define it. While all

three of these areas present challenges for studies of begging networks, they also

provide opportunities for testing some key concepts in the study of communica-

tion networks.

Physical competition versus signalling

Nestlings form a communication network because they are within sig-

nalling range of each other. Unlike members of many other communication net-

works, however, nestlings are also in direct physical contact with each other. This

tight proximity highlights difficulties that can arise when we try to distinguish

between signalling and other acts, in this case physical competition. Since a com-

munication network, by definition, consists of signalling (i.e. of behaviours spe-

cialized to communicate information (McGregor & Peake, 2000)), this distinction

is fundamental for understanding any communication network.

Nestlings jostle with one another for access to parental feeding locations

within the nest and their success at reaching the parent strongly affects which

nestlings are fed (Budden & Wright, 2001). Nestlings can physically compete in

several ways, for example by usurping positions close to where parents arrive

at the nest, by blocking parents’ access to other nestlings or, particularly in non-

passerine species, by directly pushing or pecking one another (Mock & Parker, 1997;

Budden & Wright, 2001; Drummond, 2002).

Much of this physical competition is hard to distinguish from signalling.

Jostling for position and direct aggression seem to be non-signalling acts by which



nestlings get better access to parents. Parents may nonetheless get information

about nestling need and quality from these physical interactions, which they then

use to choose which nestling to feed (Rodrıguez-Girones, 1996; Lotem et al., 1999).

This informativeness alone does not make them signals. If, however, the interac-

tions are designed to affect that choice, rather than merely to thrust a nestling

forward to rob the parent of its choice, then they are signals, by the above defini-

tion, despite their outward appearance. Conversely, some features of begging that

appear to have been designed partly to convey information and thus are signals by

definition (McGregor & Peake, 2000), such as posturing (Kilner, 2002b), seem just as

clearly designed for effective jostling toward the parent. Even the design features

of begging displays that are adaptations for overcoming interference from nest-

mates (reviewed below) may be seen either as ways to signal information on need

more effectively to parents (Horn & Leonard, 2002) or as scrambles for parental

attention (Rodrıguez-Girones et al., 2001; Royle et al., 2002). In the latter case, their

ultimate function would differ little from that of physical competition, since by

dominating the parents’ visual and acoustic fields they too would not inform

parents so much as reduce the parents’ opportunity to choose which nestling to

feed.

Therefore, a nest full of begging nestlings is part communication network,

part scrum toward the parent. Which view of begging is more accurate depends

largely on how parents interpret begging signals and physical competition, a topic

we discuss further below. Given that display behaviours ultimately evolve from

non-signalling acts, however, we can at least conclude that begging offers an

interesting system for studying how social behaviours besides signalling affect

communication networks.

Signalling interactions versus just signalling

One of the aspects of communication that has become more prominent

as a result of the communication network approach is the information content of

signalling interactions: the give and take of signals among members of the network

(McGregor & Peake, 2000). It is from the interactions between signallers, rather

than the signals themselves, that some particularly interesting consequences of

communication networks arise, such as signalling to avoid interference (Ch. 13)

and eavesdropping (Peake et al., 2002; Ch. 2).

Distinguishing signals (directed at the parent) from signal interactions (di-

rected at nestmates) in the case of nestling begging is difficult, however. On the

one hand, several lines of evidence show that nestmates’ signals influence how

a nestling signals. In many studies, nestlings beg more intensely when in bigger

broods or when with nestmates than when alone (Budden & Wright, 2001; but see

Cotton et al., 1996). More direct evidence comes from studies in which nestlings



increase their postural display when their nestmates do (e.g. Leonard & Horn, 1998)

or call more when they hear nestmates calling (e.g. Leonard & Horn, 2001c).

On the other hand, it is not clear that these changes in signalling constitute

signalling interactions in the sense implied by current discussions of commu-

nication networks, especially work on social eavesdropping (McGregor & Peake,

2000; Ch. 2). According to this work, a signalling interaction consists of a sender

directing a signal at a receiver, which then responds. To the degree that begging

is directed at the parent, then competitive interactions among nestlings to catch

the parents’ attention are not signalling interactions in this sense (Royle et al.,

2002). By extension, parents that choose to feed nestlings that beg more than their

nestmates (Budden & Wright, 2001), like the predators that are attracted to nests

whose calling is increased by competition (Haskell, 2002), are interceptive rather

than social eavesdroppers, because social eavesdroppers must base their response

on signalling interactions not just on signals (Ch. 2).

This conclusion may partly reflect our still sketchy understanding of nestling

interactions. For example, Roulin (2002) has recently suggested that at least some

signalling by nestlings may be directed at nestmates. Nestling barn owls Tyto alba,

for example, appear to have calling contests between parental visits, in which

nestlings negotiate which of them will receive a feeding when the parent next

returns (Roulin, 2002). If nestlings do direct signals to each other in this way, then

parents that extract information from these interactions would fit the definition

of social eavesdroppers (Ch. 2).

In the particular case of barn owls, nestling negotiations occur when the par-

ent is absent and so cannot be overheard by parents. In principle, however, there

is no reason why similar interactions between nestlings could not also occur in

the parent’s presence, especially in species in which parents spend enough time

transferring food to their young that the young have time to interact (e.g. par-

rots (Psittaciformes); Krebs, 2002). Certainly, if nestlings do direct their signals to

each other, the importance of considering nestling begging as a communication

network is considerably strengthened.

Functional relationships among nestlings and network structure

Communication networks were first defined in the context of communi-

cation among territorial songbirds, which are widely separated on different terri-

tories but are interconnected by the overlapping transmission ranges of their songs

(McGregor & Dabelsteen, 1996). Song is, thus, the main way in which these birds

interact; consequently, characterizing interacting songbirds as a communication

network captures much of how they affect each other’s signalling behaviour.

Nestlings packed together within a nest, however, are interconnected in many

ways besides the overlapping ranges of their signals. We have already discussed



how they interact through physical competition and how that may have strong

effects on their signalling behaviour. In this section, we briefly list three other

interconnections that are integral to any explanation of how nestmates affect

each other’s signalling behaviour.

Unlike physical competition, these effects do not present difficulties for defin-

ing signals and hence applying the definition of communication networks to

begging. They do, however, illustrate that, in some communication networks, sig-

nallers are so mutually dependent on one another that the overlapping transmis-

sion ranges of their signals are only one way in which their signalling behaviours

are interconnected.

We will list three such relationships among nestlings: genetic relatedness,

shared fate and heat transfer. For each category, we touch briefly on their pos-

sible implications for signalling. We then discuss perhaps their most interesting

implication, which is how all these relationships might combine to give a structure

to the communication network within the brood.

Genetic relatedness

Genetic relatedness is perhaps the most important of the relationships

among nestlings, because it so heavily influences the fitness consequences of all

the other types of relationship. Since nestlings tend to be highly related to one

another, relatedness probably affects signalling in this communication network

more than in most of the other networks described in this volume. Indeed, for

most theoretical models of begging, the main route of sibling effects on begging is

through a nestling’s inclusive fitness. In general, theory predicts less-exaggerated

or less-costly begging the higher the relatedness among nestmates ( Johnstone &

Godfray, 2002; Price et al., 2002). Consistent with such predictions, interspecific

brood parasites, whose relatedness with their host nestmates is zero, such as Euro-

pean cuckoos Cuculus canorus, great spotted cuckoos Clamator glandarius and brown-

headed cowbirds Molothrus ater, call more loudly and more frequently than their

nestmates (Dearborn & Lichtenstein, 2002; Redondo & Zuniga, 2002).

Evidence for non-parasitic species, however, is scant. In one comparison across

species for which data on genetic parentage were available, begging calls were

louder in species with more frequent mixed parentage (Briskie et al., 1994). This

result suggests that a species’ average level of relatedness within broods might set

its average level of begging. A more relevant result for communication networks,

however, would be if nestlings within a species could assess their relatedness to

broodmates and adjust their levels of competitive signalling accordingly. Nestlings

are generally thought to lack the cues by which their nestmates could assess

their relatedness (e.g. Whittingham & Dunn, 2001); indeed there may be selection

against such cues ( Johnstone, 1997). As for kin recognition in birds in general



(Komdeur & Hatchwell, 1999), addressing this issue directly will require more

sophisticated experiments than have been applied to date.

Shared fate

Along with relatedness, a fundamental feature underlying nestling inter-

actions is that, like the proverbial eggs in one basket, nestmates often share the

same fate. For better or worse, they have the same adults feeding them, share the

same local environmental conditions around the nest and, therefore share their

chances of survival to a greater degree than participants in most other types of

communication network.

This shared fate has inevitable consequences for signalling behaviour; if one

nestling begs more loudly, for example, the parents might return more often to

feed all the nestlings or a predator might be more likely to find the nest and eat

all the nestlings. Thus, both the benefits and the costs of begging by any given

nestling are at least partially visited on the whole brood. Indeed, Wilson & Clark

(2002) went still further and suggested that broods are subject to a form of group

selection which may lead nestlings to signal cooperatively. Aspects of begging that

are usually presented as competitive, such as signal characteristics that ostensibly

serve to circumvent interference (see below), might instead function cooperatively

to coordinate nestmates’ signals (Wilson & Clark, 2002). How individual signals

fit together to form aggregate brood signals has not been studied yet, but we can

safely expect that the shared fate of nestlings will make signalling interactions

within their networks differ in interesting ways from those of signallers with more

independent fates, such as chorusing frogs.

Heat transfer

Nestling birds cannot thermoregulate until partway through the nesting

period. Before that point, they rely not only on brooding by parents but also on

heat from their nestmates. Nests where young hatch asynchronously may consist

of older, heat-producing nestlings and younger, heat-consuming nestlings (e.g.

Hill & Beaver, 1982). Such thermal relationships among nestlings may increase the

variety of their signals and signalling interactions. Specifically, in several species,

some aspects of begging, especially begging calls, change when nestlings lose heat

and may signal their need for brooding (see above). Nestlings might, therefore, have

to compete for attention from nestmates that are sometimes signalling for food

and sometimes for warmth, and they might adopt different signalling strategies

for each situation. Thermal relationships might also affect signalling through

more direct effects on individual signallers. For example, some evidence suggests

that house sparrow Passer domesticus nestlings lose heat when the stretching and

gaping of begging increases their surface area (Ovadia et al., 2002). They might,



therefore, be able to beg more when next to larger nestmates, since any thermal

loss during begging would be reduced. Thermal relationships among nestmates

are still poorly understood, but, like physical competition and signalling, they

are probably readily perceived by nestlings and thus may have immediate and

dynamic effects on patterns of signalling within the nest.

Network structure

The net result of all the relationships listed above, including the physical

competition also discussed, is that they may lend structure to the communica-

tion network within the nest. By ‘structure,’ we mean a pattern in which not all

nestlings have the same sorts of relationship with one another. Most obviously,

physical competition can lead to dominance hierarchies, with larger or stronger

nestlings suppressing the begging signals of smaller nestlings or displacing them

from positions near the parent where their begging signals would attract the

parent’s attention more effectively (Mock & Parker, 1997).

Hierarchies, however, are only one of a variety of network architectures

that might arise. Speaking more generally, Glassey & Forbes (2002b) noted that

nestlings can often be divided into ‘core’ and ‘marginal’ nestlings (Mock & Forbes,

1995). Survival of core nestlings is usually predictable, whereas marginal nestlings,

which may be smaller, in poorer condition, younger, subordinate and/or less able

to thermoregulate, survive only if ecological conditions are favourable. This ‘struc-

tured sibship’ (Glassey & Forbes, 2002b) may yield three different sorts of nestling

relationships within the brood: core to core, marginal to marginal, and core to

marginal (Glassey & Forbes, 2002b).

Variation among species in this underlying structure will affect physical com-

petition and signalling interactions within the nest. For example, one core and

one marginal nestling might yield a simple dominance hierarchy, whereas three

nestlings in each category might yield two ‘cliques’ of nestlings, between which

there is a dominance hierarchy but within which signalling behaviours are simi-

lar and physical competition is equitable. In any case, the underlying structure of

relationships within the brood, even though they do not consist of signalling rela-

tionships, nevertheless may strongly affect the structure of the overlying commu-

nication network – no doubt a recurring theme for most communication networks

(e.g. Chs. 10 and 25).

Summary

We have raised three complexities in applying the concept of communi-

cation networks to nestling begging. First, characterizing the communication net-

work entails a difficult distinction between signalling and physical interactions.

Second, demonstrating some of the more interesting effects of communication



networks entails another difficult distinction: between signalling to the parent

and signalling interactions with nestmates. Third, any realistic description of the

communication network must include interrelationships among nestmates that

do not involve signalling but nevertheless may shape the structure of the network.

These particular issues, of course, have less of an impact on communication in

some other kinds of network. Territorial birds singing from their song posts, for ex-

ample, are far beyond the range of physical interaction, are clearly directing their

signals at each other (but see Ch. 14), and are generally unrelated to one another.

Nonetheless, the issues we have raised are not unique to begging nestlings. Even

territorial birds, for example, can engage in close-range interactions that combine

signals with direct aggression, sing in ways that can be seen either as signalling

interactions or as attempts to overcome interfering signals, and have dominance

relationships that structure their communication network. If communication net-

works are indeed ‘the commonest social environment in which communication

occurs’(McGregor & Peake, 2000), then network concepts will inevitably be applied

to other systems that do present some of the complications we have discussed to

varying degrees. If we are to understand how these networks function, we need

to clarify these issues and begging should prove to be a particularly useful system

for doing so.

Consequences of the network for begging

We now turn from attempting to characterize the communication net-

work within the nest to exploring how it might affect communication, from both

signallers’ and receivers’ perspectives. Most discussions of communication net-

works have emphasized two consequences in particular (e.g. McGregor & Peake,

2000; McGregor et al., 2000; see also other chapters in this volume) and we begin

with these. First, from the signaller’s point of view, signals must be designed to

catch the receiver’s attention in the face of interference from other signals in the

network. Second, receivers, for their part, can more readily compare signallers in

a network because they are in transmission range of several signallers at once. A

third possible consequence has received less attention: communication networks

might reduce error in the information that signals convey. Specifically, as we

explain below, nestlings are particularly error prone in deciding when and how

intensely to beg. When nestlings partly base these decisions on the behaviour of

other nestlings, as they can when signalling within a network, these errors might

have less effect on their signals of need.

Design to catch receiver attention

McGregor and Peake (2000) suggested that the main effect of networks on

signal design arises through competition for receiver attention, as each signaller



attempts to circumvent the interference caused by competing signals in the net-

work. Perhaps no other communication system is more obviously a competition

for receiver attention than a brood of noisy nestlings. Given the interest in the

exaggeration of this signal and its role in nestmate competition, however, there

are surprisingly few studies that specifically address how begging signals are de-

signed to overcome interference from nestmates. Our understanding of begging

and nestling competition might be considerably enhanced by thinking of begging

nestlings as a communication network.

In particular, we suspect that many of the most striking characteristics of the

begging display may be designed for overcoming interference. If so, then the con-

spicuousness and complexity of the display, which seems unnecessarily extrava-

gant for such a short-range signal, may, in fact, be a proportionate response to

signal interference (Dawkins & Guilford, 1997; Horn & Leonard, 2002). Here we

briefly discuss how selection for overcoming interference might account for a few

of the more obvious features of begging (see also Horn & Leonard, 2002).

High output

The most straightforward way to overcome any background noise is to

increase the amplitude or duty cycle of one’s signal. There is ample evidence that

nestlings respond in this way to signalling by nestmates (Budden & Wright, 2001;

but see Cotton et al., 1996). For example, nestlings in some species beg more in-

tensely when placed near a begging nestmate (Leonard & Horn, 1998) and call at

higher rates when they can hear a nestmate calling (Leonard & Horn, 2001c).

Locatable signals

Surprisingly small apparent angular separation between stimuli can sig-

nificantly enhance a receiver’s ability to tell them apart (Ch. 20). Thus design

features that enhance the locatability of nestlings are likely to enhance how well

they stand out from competing signals and so focus parental attention on an in-

dividual nestling. The visual components of begging, brightly coloured gapes in

particular, seem designed to be readily locatable targets for parental attention.

These gapes have particularly bright outlines in species that nest in cavities, most

likely so that the location of each nestling’s gape is distinct despite the darkness

(Kilner & Davies, 1998; Heeb et al., 2003).

Begging calls, in contrast, do not seem as obviously suited for locating nestlings

because they are broadcast noisily throughout the nest. Also, there is little evidence

so far that their structures are individually distinct in ways that would make

them easy for parents to distinguish (Leonard & Horn, 2001c; but see Popp &

Ficken, 1991). Indeed, some theoretical models suggest that they should not be

individually distinct because that would risk rejection by the parent (Beecher,

1991; Johnstone, 1997).



Fig. 9.1. Spectrograms of three nestling begging calls: tree swallow Tachycineta bicolor,

hairy woodpecker Picoides pubescens and white-browed scrubwren Sericornis frontalis.

Vertical bar is 10 kHz, horizontal bar is 500 milliseconds and filter bandwidth is

700 Hz.

Nonetheless, many calls do display features thought to enhance locatability,

including abrupt onsets and offsets, broad frequency ranges and use of frequencies

to which parents are most acutely tuned (Horn & Leonard, 2002; Fig. 9.1). Whether

these features really do enhance locatability within the confines of a nest has

not been tested directly. Comparative evidence, however, suggests that begging

calls do display some of these features, except when subject to counteracting

selective pressure from predators that use locatable calls to find and depredate

nests (Haskell, 2002; Horn & Leonard, 2002).

Multiple components

Which features of signals stand out from the noise of competing signals

will depend on the situation, and the multiple components of the begging display

may allow nestlings to signal effectively in each of these different situations. For

example, a nestling competing with a nestmate in the front of a cavity nest might

gain more from gaping wider and posturing more intensely than a nestling stuck

in the back of the nest, because the nestling in the front is in plain sight of the

parent. In contrast, a nestling in the back of the box cannot be clearly seen by the

parent and would probably gain more from large increases in call rate than from

any changes in the visual signal (Leonard et al., 2003). Therefore, in addition to

the numerous other psychological advantages of multimodal signal components

(Rowe, 1999), they may provide nestlings with a toolbox of ways to make their

signal stand out despite changing conditions.



Precedence

Precedence effects, the tendency of receivers to take more notice of signals

that occur first, may favour signallers that signal before their competitors do

(McGregor et al., 2000). Note that such effects, as shown in insect and frog choruses,

for example, may (Greenfield, 2002) or may not (Gerhardt & Huber, 2002) be the

result of certain psychological effects also known as precedence (for which Dent &

Dooling (2003a,b) provide an avian example). Begging may provide a particularly

good example of this effect on signalling. Parents in a wide range of species are

more likely to feed nestlings that beg before their nestmates (Budden & Wright,

2001) and nestlings appear to have been selected for hair-trigger responses to

the first sign of the parent’s arrival (Leonard & Horn, 2001d). The importance of

precedence effects may vary considerably among species, providing interesting

opportunities for comparative tests of their effects on signalling. For example,

they may be less important in species in which parents spend more time assessing

begging signals at each visit (e.g. Krebs, 2002) or in which hasty responses by

nestlings might waste energy or attract predators (Leonard & Horn, 2001d).

Signal suppression

All the aspects of signal design we have outlined so far can overcome sig-

nal competition by enhancing the signaller’s own signal. Signallers might also,

however, overcome competition by suppressing the signals of competitors. For ex-

ample, nestling whydahs Vidua spp. spread their wings to block their parents’view

of nestmate signals (B. Mines, personal communication) and dominant nestlings

of many non-passerine species aggressively punish subordinate nestmates that

beg in their presence (Drummond, 2002; Roulin, 2002). Subtler versions of such

direct approaches to signal competition may be widespread and should be looked

for in other species.

Comparison among signals

A second consequence of communication networks is that they allow re-

ceivers to compare information from several signallers at once. Social eavesdrop-

ping, extracting information from a signalling interaction (Ch. 2), is a particularly

interesting special case of such comparisons. However, receivers might also ben-

efit from the network simply by being able to compare signals simultaneously

rather than having to assess each signaller in succession (Chs. 7 and 14).

Surprisingly, how or even whether parents compare begging signals to decide

which nestling to feed is still poorly understood. Many studies, using various mea-

sures of begging intensity, have shown that more intensely begging nestlings are

more likely to be chosen, but such evidence is only correlational. Only a few re-

cent studies have experimented on parental choice in sufficient detail to separate



the roles of non-signalling and signalling components of begging, or to demon-

strate preferences based on individual components of the begging display (Horn &

Leonard, 2002; Kilner, 2002a,b). Demonstrating whether parents use information

from signalling interactions among nestlings will require still more refined ex-

periments (see above).

Interestingly, recent models suggest that parents must assess interactions

among nestlings if begging is to evolve as a signal at all (Rodrıguez-Girones et al.,

2001; Royle et al., 2002). Specifically, if parents simply select the most obvious

signal, then the information content of begging becomes irrelevant and begging

consists merely of a scramble for the parents’ attention. If, however, parents can

calibrate the information in the signals to correct for competitive differences

among nestlings, whether those are expressed via signalling (e.g. Roulin, 2002)

or physical competition, then begging can indeed convey information on need

(Rodrıguez-Girones et al., 2001; Royle et al., 2002). Under this scenario, a network

environment may have been of central importance in the evolution of begging.

Error reduction

The last possible consequence of communication networks that we will

discuss has received little attention, although it seems simple in principle and

broad in implications. Specifically, because information in a network is transferred

via not just one but several signals, the impact of error from any given signal might

be reduced. To explain this possibility, we first outline some possible sources of

error in begging displays and then discuss how the communication network may

reduce this error.

Begging by individual nestlings may be considerably error prone for at least

two reasons (Clark, 2002; Horn & Leonard, 2002). First, nestlings may be poor at

assessing their own needs, especially since doing so requires integrating their cur-

rent condition with their future requirements and their likely returns from beg-

ging, both of which are partly under control of their parents, their nestmates and

the vagaries of the environment (Clark, 2002). Second, nestlings are often poor

at distinguishing the parent’s arrival at the nest from other sights and sounds

and, therefore, often beg in response to irrelevant stimuli. In older tree swallow

Tachycineta bicolor nestlings, for example, while nestlings often simply start beg-

ging after their nestmates do, many of the initial begging responses are to events

other than the parent, like the wind blowing through the trees or the bump of

another bird species landing by the nest (Leonard & Horn, 2001d; Horn & Leonard,

2002). Conversely, nestlings apparently hold back on begging when they are un-

sure whether the parent actually has arrived and so may miss the parent’s arrival

or may send an inappropriately weak signal (Clark, 2002). From the nestling’spoint

of view, these are errors in how they deliver the begging signal. From the parent’s



point of view, however, such errors corrupt any information that the parent might

obtain from the begging signal.

Begging in a network may buffer such errors, because each nestling bases its

decision of when and how intensely to beg partly on the begging of its nestmates

(e.g. Leonard & Horn, 1998, 2001c). This influence of nestmates should reduce the

impact of the errors that each nestling would make if it were begging on its own;

from the parent’spoint of view, it would provide a more reliable signal of offspring

need (Clark, 2002).

This argument could, of course, be reversed. Specifically, one might argue that

the more links in the information chain from nestlings to parents, the less accurate

and reliable information will be (Royle et al., 2002). Determining whether networks

reduce or increase error requires modelling of information flow through the net-

work. A nestling’s decision of when to beg, to take the first step in the chain, may

be seen as a game of signal detection, in which the nestling can either try to be the

first to detect the parent’s arrival, at the risk of more false alarms (as shown above

for tree swallows), or can free-ride by eavesdropping on the responses of nestmates,

at the risk of begging later than its nestmates (Erev et al., 1995). Notwithstanding

the promise of such models, probably the most pressing need for understand-

ing information flow through networks, indeed for all the possible consequences

of the begging network surveyed above, is for more empirical research on how

parents assess begging signals.


In this chapter we have tried to show that begging by nestling birds is a

promising system for clarifying fundamental aspects of communication networks,

particularly the grey but conceptually fruitful areas between physical acts and

signals, between signalling competitively and interacting, and between commu-

nication and other functional relationships among signallers. Theoretical work

on the evolution of begging has already started exploring each of these areas,

but it has been inspired more by field workers’ insistence that begging behaviour

is complex than by any attempt to treat begging as a communication network.

In the future, theoretical work would likely benefit from a more explicit appli-

cation of network concepts, much as studies of economics and cooperation in

humans have benefited from models of social networks (e.g. Slikker & van den

Nouweland, 2001). Conversely, those studying other communication networks

will likely benefit from staying abreast of theoretical developments in the study of

begging.

Perhaps the greatest opportunities for future work, however, are in empirical

studies that focus on signalling and nestmate interactions in more detail. Despite



enormous variation in the form of begging calls within and across species, for

example (Popp & Ficken, 1991), only a handful of studies have addressed the func-

tion of this variation in any detail (Horn & Leonard, 2002; see also Kilner (2002b) for

the display as a whole). Similarly, despite a long history of interest in intrabrood

competition in birds (Mock & Parker, 1997), few studies have tried to identify the

specific functions of the various behaviours that nestlings use in competition,

especially what information they might convey to both parents and nestmates

(Clark, 2002; Roulin, 2002). Perhaps most importantly, how, or even whether, par-

ents choose which nestling to feed remains largely unknown because the requisite

experiments have not been done (Royle et al., 2002). Hopefully, greater appreciation

that nestlings communicate within a network of signallers, with all its attendant

challenges and opportunities, will inspire more research on all of these funda-

mental questions.

Acknowledgements

We thank Pete McGregor for the chance to contribute to this volume and for his pa-

tience and constructive advice during the preparation of this chapter. We also thank John Bower

and an anonymous reviewer for their helpful comments on an earlier draft. Conversations with

participants at the Gregynog 2000 Begging Workshop and with authors of the resultant book have

been invaluable in developing our ideas about begging, as has our collaboration with Rob Magrath

and his students at the Australian National University and with the many students and assistants

who have worked on tree swallows with us. We also thank the Coldwell, Hines and Minor families

for allowing us to work on their land, and NSERC for financial support.

References

Beecher, M. D. 1991. Successes and failures of kin recognition in animals. In: Kin

Recognition, ed. P. G. Hepper. Cambridge, UK: Cambridge University Press,

pp. 94–124.

Bergstrom, C. T. & Lachmann, M. 1998. Signaling among relatives. III. Talk is cheap.

Proceedings of the National Academy of Sciences, USA, 95, 5100–5105.

Briskie, J. V., Naugler, C. T. & Leech, S. M. 1994. Begging intensity of nestling birds

varies with sibling relatedness. Proceedings of the Royal Society of London, Series B, 258,

73–78.

Budden, A. E. & Wright, J. 2001. Begging in nestling birds. Current Ornithology, 16,

83–118.

Clark, A. B. 2002. Appetite and the subjectivity of hunger. In: The Evolution of Begging:

Competition, Cooperation, and Communication, ed. J. Wright & M. L. Leonard.

Dordrecht: Kluwer, pp. 173–198.



Clotfelter, E. D., Shubert, K. A., Nolana, V. Jr & Ketterson, E. D. 2003. Mouth color

signals thermal state of nestling dark-eyed juncos ( Junco hyemalis). Ethology, 109,

171–182.

Cotton, P. A., Kacelnik, A. & Wright, J. 1996. Chick begging as a signal: are nestlings

honest? Behavioral Ecology, 7, 178–182.

Cotton, P. A., Wright, J. & Kacelnik, A. 1999. Chick begging strategies in relation to

brood hierarchies and hatching asynchrony. American Naturalist, 153, 412–420.

Dawkins, M. S. & Guilford, T. 1997. Conspicuousness and diversity in animal signals.

Perspectives in Ethology, 12, 55–72.

Dearborn, D. C. & Lichtenstein, G. 2002. Begging behaviour and host exploitation in

parasitic cowbirds. In: The Evolution of Begging: Competition, Cooperation, and

Communication, ed. J. Wright & M. L. Leonard. Dordrecht: Kluwer, pp. 361–

387.

Dent, M. K. & Dooling, R. J. 2003a. Investigations of the precedence effect in

budgerigars: effects of stimulus type, intensity, duration, and location. Journal of

the Acoustical Society of America, 113, 2146–2158.

2003b. Investigations of the precedence effect in budgerigars: the perceived location

of auditory images. Journal of the Acoustical Society of America, 113, 2159–2169.

Drummond, H. 2002. Begging versus aggression in avian broodmate competition. In:

The Evolution of Begging: Competition, Cooperation, and Communication, ed. J. Wright &

M. L. Leonard. Dordrecht: Kluwer, pp. 337–360.

Erev, I., Gopher, D., Itkin, R. & Greenshpan, Y. 1995. Toward a generalization of signal

detection theory to n-person games: the example of two person safety problem.

Journal of Mathematical Psychology, 39, 360–375.

Evans, R. M. 1994. Cold-induced calling and shivering in young American white

pelicans: honest signalling of offspring need for warmth in a functionally

integrated thermoregulatory system. Behaviour, 129, 14–34.



Glassey, B. & Forbes, S. 2002a. Muting individual nestlings reduces parental feeding

for the brood. Animal Behaviour, 63, 779–786.

2002b. Begging and asymmetric nestling competition. In: The Evolution of Begging:



Godfray, H. C. J. 1991. The signalling of need by offspring to their parents. Nature, 353,

328–330.

1995. Evolutionary theory of parent–offspring conflict. Nature, 376, 133–138.

Godfray, H. C. J. & Johnstone, R. A. 2000. Begging and bleating: the evolution of

parent–offspring conflict and sibling rivalry. Philosophical Transactions of the Royal


Gotmark, F. & Ahlstrom, M. 1997. Parental preferences for red mouth of chicks in a

songbird. Proceedings of the Royal Society of London, Series B, 264, 959–962.





Haskell, D. G. 2002. Begging behaviour and nest predation. In: The Evolution of Begging:



Heeb, P., Schwander, T. & Faoro, S. 2003. Nestling detectability affects parental feeding

preferences in a cavity-nesting bird. Animal Behaviour, 66, 637–642.

Hill, R. W. & Beaver, D. L. 1982. Inertial thermostability and thermoregulation in

broods of redwing blackbirds. Physiological Zoology, 55, 250–260.

Horn, A. G. & Leonard, M. L. 2002. Efficacy and the design of begging signals. In: The

Evolution of Begging: Competition, Cooperation, and Communication, ed. J. Wright &


Johnstone, R. A. 1997. Recognition and the evolution of distinctive signatures: when

does it pay to reveal identity? Proceedings of the Royal Society of London, Series B, 264,

1547–1553.

1999. Signaling of need, sibling competition, and the cost of honesty. Proceedings of

the National Academy of Sciences, USA, 96, 12644–12649.

Johnstone, R. A. & Godfray, H. C. J. 2002. Models of begging as a signal of need. In: The

Evolution of Begging: Competition, Cooperation, and Communication, ed. J. Wright &


Kilner, R. M. 1997. Mouth colour is a reliable signal of need in begging canary

nestlings. Proceedings of the Royal Society of London, Series B, 264, 963–968.

2002a. The evolution of complex begging displays. In: The Evolution of Begging:



2002b. Sex differences in canary (Serinus canaria) provisioning rules. Behavioral


Kilner, R. M. & Davies, N. B. 1998. Nestling mouth colour: ecological correlates of a

begging signal. Animal Behaviour, 56, 705–712.

Komdeur, J. & Hatchwell, B. J. 1999. Kin recognition: function and mechanism in avian

societies. Trends in Ecology and Evolution, 14, 237–241.

Krebs, E. A. 2002. Sibling competition and parental control: patterns of begging in

parrots. In: The Evolution of Begging: Competition, Cooperation, and Communication, ed.

J. Wright & M. L. Leonard. Dordrecht: Kluwer, pp. 319–336.

Leonard, M. L. & Horn, A. G. 1998. Need and nestmates affect begging in tree swallows.


2001a. Begging calls and parental feeding decisions in tree swallows (Tachycineta

bicolor). Behavioral Ecology and Sociobiology, 49, 170–175.

2001b. Acoustic signalling of hunger and thermal state by nestling tree swallows.


2001c. Dynamics of calling by tree swallow (Tachycineta bicolor) nestmates. Behavioral


2001d. Begging in the absence of parents by nestling tree swallows. Behavioral

Ecology, 12, 501–505.

Leonard, M. L., Horn, A. G. & Parks, E. 2003. The role of calling and posturing in the

begging behavior of nestling birds. Behavioral Ecology and Sociobiology, 54, 188–

193.



Lotem, A., Wagner, R. H. & Balshine-Earn, S. 1999. The overlooked signaling

component of nonsignaling behavior. Behavioral Ecology, 10, 209–212.

Maynard Smith, J. 1994. Must reliable signals always be costly? Animal Behaviour, 42,

1034–1035.






McGregor, P. K., Otter, K. & Peake, T. M. 2000. Communication networks: receiver and

signaller perspectives. In: Signalling and Signal Design in Animal Communication, ed.

Y. Espmark, T. Amundsen & G. Rosenqvist. Trondheim: Tapir Academic Press,

pp. 329–340.

Mock, D. W. & Forbes, L. S. 1995. The evolution of parental optimism. Trends in Ecology

and Evolution, 10, 130–134.

Mock, D. W. & Parker, G. A. 1997. The Evolution of Sibling Rivalry. Oxford: Oxford

University Press.

Ovadia, O., Pinshow, B. & Lotem, A. 2002. Thermal imaging of house sparrow nestlings:

the effect of begging behavior and nestling rank. The Condor, 104, 837–842.

Parker, G. A., Royle, N. J. & Hartley, I. R. 2002. Intrafamilial conflict and parental

investment: a synthesis. Philosophical Transactions of the Royal Society of London, Series

B, 357, 295–307.

Peake, T. M., Terry, A. M. R., McGregor, P. K. & Dabelsteen, T. 2002. Do great tits assess



Popp, J. & Ficken, M. S. 1991. Comparative analysis of acoustic structure of passerine

and woodpecker nestling calls. Bioacoustics, 3, 255–274.

Price, K., Harvey, H. & Ydenberg, R. 1996. Begging tactics of nestling yellow-headed

blackbirds, Xanthocephalus xanthocephalus, in relation to need. Animal Behaviour, 51,

421–435.

Price, K., Ydenberg, R. & Daust, D. 2002. State-dependent begging with asymmetries

and costs: a genetic algorithm approach. In: The Evolution of Begging: Competition,

Cooperation, and Communication, ed. J. Wright & M. L. Leonard. Dordrecht: Kluwer,

pp. 21–42.

Redondo, T. & Zuniga, J. M. 2002. Dishonest begging and host manipulation by

Clamator cuckoos. In: The Evolution of Begging: Competition, Cooperation, and

Communication, ed. J. Wright & M. L. Leonard. Dordrecht: Kluwer, pp. 389–412.

Rodrıguez-Girones, M. A. 1996. Siblicide: the evolutionary blackmail. American

Naturalist, 148, 101–122.

Rodrıguez-Girones, M. A., Enquist, M. & Lachmann, M. 2001. Role of begging and

sibling competition in foraging strategies of nestlings. Animal Behaviour, 61,

733–745.

Roulin, A. 2002. The sibling negotiation hypothesis. In: The Evolution of Begging:





Rowe, C. 1999. Receiver psychology and the evolution of multicomponent signals.


Royle, N. J., Hartley, I. R. & Parker, G. A. 2002. Begging for control: when are offspring

solicitation behaviours honest? Trends in Ecology and Evolution, 17, 434–440.

Saino, N. & Møller, A. P. 2002. Immunity and begging. In: The Evolution of Begging:



Saino, N., Ninni, P., Calza, S. et al. 2000. Better red than dead: carotenoid based mouth

coloration reveals infection in barn swallow nestlings. Proceedings of the Royal


Slikker, M. & van den Nouweland, A. 2001. Social and Economic Networks in Cooperative

Game Theory. Dordrecht: Kluwer.

Trivers, R. L. 1974. Parent–offspring conflict. American Zoologist, 14, 249–264.

Whittingham, L. A. & Dunn, P. 2001. Male parental care and paternity. Current

Ornithology, 16, 257–298.

Wilson, D. S. & Clark, A. B. 2002. Begging and cooperation: an exploratory flight. In:

The Evolution of Begging: Competition, Cooperation, and Communication, ed. J. Wright &



10

Redirection of aggression: multipartysignalling within a network?

a na h i t a j . n . k a z e m 1 & f i l i p p o au r e l i 2

1University of Wales, Bangor, UK2Liverpool John Moores University, UK

Introduction

In many species, an individual that finds itself in a losing position may

interrupt a contest to harass a bystander (i.e. an apparently uninvolved third party)

aggressively or may do so immediately after hostilities with the original oppo-

nent have ceased. Such ‘redirection’ of aggression (Bastock et al., 1953) is often

interspecific; for example, rollers and chasseur-type kingfishers (Coraciiformes)

are reported to dash away frequently during disputes to attack small passerines,

doves and plovers (Moynihan, 1998). The scapegoats are typically not ecological

competitors but do tend to be smaller and inoffensive individuals, both literally

and figuratively, and thus relatively safe targets. In socially living taxa, however,

redirection is most commonly directed towards a lower-ranking group member

(where available) and, therefore, is usually intraspecific. Both aspects of redirec-

tion are conveyed by the description of tensions between spotted hyaenas Crocuta

crocuta at a kill, producing a cascade of aggression, in which ‘A chases B, B chases

C, C chases D, and D chases vultures’ (Zabel et al., 1992, p. 129).

Redirection of aggression has traditionally been explained as a means of reduc-

ing the physiological arousal associated with participation in a conflict. The neuro-

endocrine responses underlying the preparation for ‘fight or flight’, whilst

essential in the immediate context, can be detrimental if they remain activated

over prolonged periods. Chronically elevated secretion of glucocorticoids, for

example, is associated with a range of cardiovascular pathology, depressed im-

mune function and compromised digestion, growth and reproduction (reviewed

by Sapolsky, 1998). Any action that prompts the endocrine stress response to


191


192 A. J. N. Kazem & F. Aureli

attenuate to baseline levels more rapidly can, therefore, reduce the physiological

costs of being victimized. When an animal is unable to retaliate directly against

an aggressor (for example because it is a higher-ranking or otherwise superior

competitor), redirecting offers an outlet for ‘frustration’ and allows the actor to

perceive a measure of ‘control’over the social situation. In experimental studies on

rodents and primates, both psychological effects modulate the elevation in heart

rate and glucocorticoid levels that repeated exposure to an unavoidable physical

stressor usually produces (reviewed by Levine et al., 1989; Sapolsky, 1998). Further-

more, rats Rattus norvegicus given the opportunity to attack a conspecific when

subjected to a mild electric shock subsequently developed fewer gastric lesions

than yoked controls that were not provided with this outlet (Weiss et al., 1976).

Amongst wild olive baboons Papio anubis, males that readily initiate aggression in

appropriate contexts and that frequently respond to losing fights by redirecting

aggression against others have significantly lower basal glucocorticoid levels (and

a better response to acute challenge) than similarly ranked individuals that do

not exhibit these behaviour patterns as frequently (Sapolsky & Ray, 1989; Virgin &

Sapolsky, 1997). However, the extent to which this can be attributed to beneficial

consequences of redirection per se is not clear, as it was only one of a suite of

traits concerning temperamental style in handling male–male competition that

characterized these individuals. Nevertheless, recent experimental work has con-

firmed that redirection does have an inhibitory effect on neuroendocrine stress

responses in rainbow trout Oncorhynchus mykiss (Øverli et al., 2004).

Here we summarize evidence suggesting there may, in some species, be more

to redirection than this. The behaviour is particularly prevalent and has been

best demonstrated in primates, notably several members of the genus Macaca

(Table 10.1). In these species, harassing a conspecific would seem a rather costly

way of gaining stress relief. Aggression carries the risk of injury, attracting preda-

tors and, despite a careful choice of target, may still provoke retaliation from its

allies. Why harass bystanders rather than quietly chew on wood (which has bene-

ficial effects in rats (Weiss et al., 1976))? Naturally, redirection may confer benefits

in terms of resource acquisition or reinforcing one’s status over the target; yet

this would not explain why individuals are particularly likely to instigate such

aggression immediately after losing a conflict: a time when their energy reserves

may be depleted and the former opponent likely to join in coalition against them.

Instead, the answer may lie in how redirection influences the behaviour of by-

standers, rather than the target. A number of additional hypotheses have been put

forward to explain redirection behaviour in macaques: for example that it diverts

the aggressor’s attention (Itani, 1963), provides an opportunity for the two oppo-

nents to resolve their differences by joining forces against a common foe (de Waal &

Yoshihara, 1983) or encourages the former aggressor to participate in a conciliatory


Redirection of aggression: multiparty signalling 193

Table 10.1. Non-human primate studies of intraspecific redirected aggression within

intact social groups

Species Holding conditions Redirection Evidencea Source

Barbary macaque, Captive Yes 1a Aureli et al., 1994

Macaca sylvanus

Japanese macaque, Captive Yes 1c Aureli et al., 1992b

Macaca fuscata Captive Yes 1a Aureli et al., 1993

Captive Yes 2 Eaton, 1984; Scucchi et al.,

1988

Longtailed

macaque,

Wild and captive Yes 1a Aureli & van Schaik, 1991a;

Aureli, 1992

Macaca fascicularis

Pigtail macaque, Captive Yes 1b Judge, 1982b

Macaca nemestrina

Rhesus macaque,

Macaca mulatta

Free ranging and

captive

Yes 1a de Waal & Yoshihara,

1983;c Kazem, 1999

Captive Yes 2 Gore, 1994

Stumptail macaque, Captive Yes 2 Walker Leonard, 1979

Macaca arctoides

Hamadryas baboon,

Papio hamadryas

Captive Yes 1a F. Zaragoza &

F. Colmenares,

unpublished data

Captive (females) No 2 Gore, 1994

Olive baboon, Wild (females) No 1a Castles & Whiten, 1998a

Papio anubis Wild (males) Yes 2 Sapolsky & Ray, 1989;

Virgin & Sapolsky, 1997

Sooty mangabey, Captive Yes 1a Gust & Gordon, 1993

Cercocebus

torquatus atys

Vervet monkey, Wild Yes 1b,c Cheney & Seyfarth, 1986,

Cercopithecus 1989b,c

aethiops

Spectacled langur, Captive No 1a Arnold & Barton, 2001

Trachypithecus

obscurus

Mountain gorilla,

Gorilla gorilla

Wild (males and

immatures)

Yes 1a Watts, 1995bd

beringei Wild (females) No 1a Watts, 1995b

(cont.)



Table 10.1. (cont.)

Species Holding conditions Redirection Evidencea Source

Common Captive Yes 1a Fuentes et al., 2002e

chimpanzee

Pan troglodytes

White-faced Wild Yes 2 S. E. Perry,

capuchin, unpublished data

Cebus capucinus

Black lemur, Semi free ranging No 1a Roeder et al., 2002

Eulemur macaco

Brown lemur,

Eulemur fulvus

Captive No 1a Roeder et al., 2002

aThe nature of quantitative evidence varies: 1, comparison of post-conflict period with matched-

controls (1a), with baseline focal observations or pre-conflict period (1b) or with other data (1c); 2,

inference from quantitative but uncontrolled data. Type 2 data do not necessarily demonstrate

that a significant post-conflict increase in aggression against bystanders exists. For Japanese,

longtailed and rhesus macaques, additional anecdotal citations can be found in several other

studies.bAnalysis of kin-oriented redirection, not overall incidence against all targets.cSubject’s role in previous conflict not distinguished; therefore, analyses potentially include for-

mer aggressors.dResult for adult males was trend only, owing to sample size (n = 4).eConspecific and human targets were not distinguished in analyses.

reunion with the actor (Aureli & van Schaik, 1991a). Although not explicitly

couched in signalling terms, all rely on the principle that the former aggressor

perceives the act and responds in ways that indirectly reduce the likelihood of

further aggression against the redirecting individual. We develop this notion by

proposing that redirection functions as a signal aimed at both the former oppo-

nent and other bystanders, which conveys information about the perpetrator’s

competitive ability and current state and thus directly reduces challenges from

these receivers. Such pre-emptive strikes offer a novel interpretation of the maxim

that ‘offence is the best defence’.

The macaque system

The results we describe are primarily drawn from three studies examin-

ing post-conflict behaviour in both captive and wild longtailed macaques Macaca

fascicularis (Aureli & van Schaik, 1991a,b; Aureli, 1992), and in juveniles from two

social groups within a free-ranging colony of rhesus macaques Macaca mulatta



(Kazem, 1999, A. J. N. Kazem, unpublished data). Both species form permanent

multimale–multifemale groups and within the genus are regarded as possessing

a relatively ‘despotic’ dominance style (de Waal & Luttrell, 1989; Thierry, 2000).

Clear-cut dominance relationships are apparent in both sexes, and aggressive dis-

putes are frequent, often injurious and overwhelmingly directed down the social

hierarchy. Although individuals will defend relatives and close associates against

attack, the most common pattern of intervention is support for the aggressor, with

dominant group members receiving assistance from both kin and non-kin against

subordinates that attempt to contravene the established hierarchy (Chapais, 1995).

Unsurprisingly, direct retaliation by targets against aggressors is uncommon (e.g.

less than 9% of conflicts involve counter-aggression in rhesus macaques (de Waal &

Luttrell, 1989)).

Anecdotal reports of redirected aggression abound in the literature on non-

human primates, but the phenomenon has been explored and statistically con-

firmed in rather fewer species (Table 10.1). Demonstrating that a defeat influences

the likelihood of initiating aggression against bystanders requires a comparison

with the victim’s behaviour during an equivalent period not preceded by a con-

test. In naturalistic studies within intact social groups, this is typically achieved

by comparing the immediate post-conflict period with a ‘matched-control’ obser-

vation collected on the next possible day (PC–MC method (de Waal & Yoshihara,

1983)) or, less frequently, with periods selected from a distribution of baseline

focal observations on the same individual. Current best-practice protocols entail

matching the conditions at the start of these paired samples with respect to fac-

tors likely to influence rates of aggression or other social interactions between

relevant parties. These include the time of day, prevailing climatic conditions,

predominant activity of both the subject and the wider group (if different) and,

in analyses of interactions between former adversaries, inter-opponent distance

(e.g. Aureli, 1992; Kazem, 1999). Studies differ in whether the demonstration of

redirection emphasizes the occurrence of a single critical early post-conflict event

(as in the PC–MC method), or simply compares overall rates of aggression initiated

within a defined timeframe. Discussion of the relative merits of methods that can

be used to demonstrate a significant post-conflict increase and/or operationally

identify particular bouts of aggression as ‘redirection’ events, and the statistical

issues involved, can be found in Veenema (2000), Das et al. (1997) and Kazem (1999).

Characteristic post-conflict aggressive phenomena

Together our three studies provide a database of more than 2550 post-

conflict samples from 137 individuals, in which the subject had been the victim of

unidirectional aggression with a clear outcome (i.e. the initial recipient was always

the loser). Despite differences in species, holding conditions, group composition,



age of subjects and methods of analysis, the patterns documented were strikingly

similar. Subjects threatened or attacked third parties significantly earlier and

more frequently in the aftermath of a defeat than under control conditions. In

both species, the increase was most pronounced within the first two minutes,

quickly declining to baseline levels thereafter (Aureli & van Schaik, 1991a; Kazem,

1999). On average, rhesus victims harassed at least one bystander within the first

minute after 22% of contests, corresponding to a 10-fold increase over control lev-

els (an underestimate, as the methodology used excludes bouts initiated during

the original conflict and individuals may redirect multiple times). The scapegoats

were almost always lower ranking, but lack of a suitable conspecific did not neces-

sarily stop individuals. For example, a young female rhesus victim, finding herself

surrounded by members of the alpha matriline, proceeded to energetically and

noisily pursue several lizards and a rat before peering intently into and repeatedly

threatening a small bush (completely devoid of vertebrates), whilst continually at-

tempting to solicit assistance from her former foe situated some distance away (A.

J. N. Kazem, personal observation). Such scenes are not uncommon but were not

operationally classified as redirection.

By comparison, the post-conflict behaviour of former aggressors has received

little attention. In rhesus macaques, victors also exhibit a post-conflict elevation

in attacks against bystanders (and disproportionately target relatives of their for-

mer victim (Kazem, 1999)), while in longtailed macaques they do not (Das, 1998).

Nevertheless, rhesus subjects were significantly more likely to harass bystanders

after losing a conflict than after winning one, identifying redirection by former

victims as the more distinctive and pervasive phenomenon.

Another notable feature is that, having lost one contest, individuals are also

liable to receive further aggression. In the aftermath of conflicts where the victim

had neither reconciled with its opponent (i.e. engaged in a post-conflict affiliative

reunion (Aureli et al., 2002)) nor redirected against a third party, defeated individu-

als were subjected to significantly elevated rates of threats and attacks in all three

studies. The levels received were particularly high for the first three to four min-

utes (and at least 10 minutes in longtailed macaques), gradually waning toward

baseline incidence over time. Interestingly, the renewal of hostilities by the former

adversary was not the only cause; in many cases, the aggressor was a previously

uninvolved bystander. Such increased receipt of aggression is a common find-

ing, being the predominant sequence of events in triadic interactions between

Japanese macaques Macaca fuscata (‘mobbing’; Eaton, 1984) and documented in

controlled post-conflict studies of both macaques (e.g. de Waal & Yoshihara, 1983;

Cords, 1992; Kutsukake & Castles, 2001) and other cercopithecines (e.g. mountain

gorillas Gorilla gorilla beringei (Watts, 1995a) and olive baboons (Castles & Whiten,

1998a)). The effect is specific to former victims; neither the victor nor participants



in bidirectional contests suffer an enhanced risk (Castles & Whiten, 1998a; Das,

1998; Kazem, 1999). In the rhesus study, the number of aggressive incidents be-

tween other group members recorded within a 10 m radius of the focal animal did

not differ consistently between post-conflict and control samples (A. J. N. Kazem

unpublished data). This rules out the possibility that the increases observed in

both initiation and receipt of aggression by victims could have been caused by

the post-conflict samples being obtained during periods of generally heightened

group aggressivity, and it confirms that control samples effectively matched post-

conflict conditions even in the more variable free-ranging situation.

A consequence of loser effects and eavesdropping?

The most likely explanation for the temporal patterning of challenges

from bystanders is that victims undergo some form of ‘loser experience’(cf. Scott &

Fredericson, 1951), which renders them more easily beaten than under other

circumstances. Research in a broad range of vertebrates has documented a ten-

dency for animals who have suffered a defeat to lose in subsequent interactions

against randomly selected and otherwise equally matched individuals (reviewed by

Chase et al., 1994; Hsu & Wolf, 1999). In some cases, prior losers are even at a

disadvantage against considerably smaller opponents – ones that they would nor-

mally be expected to defeat easily in any other encounter. To our knowledge, the

standard protocols used to demonstrate this behavioural pattern (staging succes-

sive dyadic contests between unfamiliar individuals, with all factors other than

the competitors’ prior social experiences held constant) have not been applied

in a primate. However, in experiments in which novel triads (trios) of unfamil-

iar rhesus macaques were convened, the sequences of agonistic interactions

observed were consistent with the operation of a loser (and indeed a winner) effect

(Mendoza, 1993). Triad members had been matched for size, age, sex and activity

level, making it unlikely that the predominance of consecutive losses (or wins)

against different opponents was a result of pre-existing differences in intrinsic

attributes.

These patterns may arise from physiological changes precipitated by an indi-

vidual’sexperiences in a prior contest. During the initial minutes of an encounter,

both contestants typically exhibit rapid, and often qualitatively similar, changes

in central neurotransmitter activity (serotonergic, dopaminergic and noradren-

ergic) as well as increased secretion of adrenal axis hormones such as gluco-

corticoids and testosterone (e.g. van Erp & Miczek, 2000; Summers et al., 2003).

However, as a contest progresses and the outcome becomes perceived, the neu-

roendocrine profiles of winner and loser diverge. Notably, in many vertebrates,

central serotonin activity and peripheral glucocorticoid concentrations return to

baseline levels relatively rapidly in victors, while greater initial increases and more



prolonged elevation are characteristic of defeated individuals (Schuurman, 1980;

Hannes et al., 1984; Øverli et al., 1999; Summers et al., 2003). These differences

are compounded if the protagonists remain confined together in the longterm,

with subordinate individuals often exhibiting chronic elevation of these para-

meters (Blanchard et al., 1993; Gust et al., 1993; Winberg & Lepage, 1998). Winners

are also reported to undergo increases in levels of circulating androgens such as

testosterone, while those in losers appear temporarily suppressed (Bernstein et al.,

1974; Rose et al., 1975; Hannes et al., 1984; Booth et al., 1989). In keeping with this,

dominant animals generally possess higher basal testosterone levels than do sub-

ordinates (see Ch. 21), although neither the causal direction nor the cumulative

influence of successive contests can be distinguished in such data.

High levels of circulating testosterone are known to sharpen concentration

and enhance social attention and memory processes (e.g. Andrews, 1991; Cynx &

Nottebohm, 1992), and they are associated with both greater risk taking (Kavaliers

et al., 2001) and expression of offensive aggression (e.g. Delville et al., 1996; Higley

et al., 1996; Ch. 21). In contrast, serotonin generally exerts an inhibitory effect on

aggressive behaviour. Experimentally enhancing central serotonin levels reduces

an animal’s readiness to initiate aggressive acts (Olivier et al., 1995; Ferris et al.,

1997; Perreault et al., 2003), while primates with chronically low serotoninergic

functioning exhibit greater impulsivity, perseverance and use of severe unre-

strained aggression (i.e. engage in aggression without regard for its consequences

(Mehlman et al., 1994; Higley et al., 1996; Fairbanks et al., 2001)). It is, therefore,

possible that the physiological changes typical of winners and losers produce tran-

sient alterations in factors that affect actual fighting ability. Alternatively, recent

evidence in fish suggests that it may simply be an individual’s perception of its

own relative ability that is modified (Hsu & Wolf, 2001), hence affecting subse-

quent decisions to initiate, escalate or withdraw. Either way, bystanders can take

advantage of these changes to gain a temporary competitive edge over recent

losers, offering an opportunity to reverse or reinforce an existing dominance rela-

tionship at relatively low cost. However, the behavioural consequences of a single

prior loss are short lived (and species specific). They persist for only a matter of min-

utes or hours in many taxa, even after the prolonged and intense fights often char-

acteristic of experimentally staged encounters between unfamiliar competitors

(Chase et al., 1994; Hsu & Wolf, 1999). Timing may assume additional importance

in species where coalitions are common, because a challenger’scosts will be further

reduced if, by choosing this moment to attack, its actions are also likely to receive

support from the victim’sformer opponent (an event that may produce additional

dividends by strengthening the assailant’s bonds with the latter individual).

One means of being alerted to a possible loser effect is by attending to the out-

come of conflicts between other group members. In some cases, simple cues might



betray a recent defeat (e.g. the dorsal darkening and flank patterns that appear

in newly subordinate individuals in many fish), but in primates the only possibil-

ity would seem to be specific postural or behavioural changes only apparent after

particularly severe losses. Instead, experimental evidence from a growing number

of species suggests that third parties ‘eavesdrop’ (McGregor, 1993), extracting in-

formation on the relative fighting abilities of conspecifics from interactions they

have witnessed (Ch. 2). Eavesdroppers subsequently treat perceived winners and

losers differently, in ways consistent with having identified those individuals as

relatively strong or weak competitors, respectively (e.g. Hogue et al., 1996; Oliveira

et al., 1998; McGregor et al., 2001; Peake et al., 2001; Earley & Dugatkin, 2002). A mul-

titude of evidence suggests that cercopithecines are similarly aware of the nature

and outcome of contests in which they are not themselves involved. Macaques and

baboons commonly exhibit apparent knowledge of the dominance relations be-

tween their groupmates (e.g. Silk, 1999; Ch. 25). In these species, the relative rank

of third parties typically cannot be deduced directly from cues such as relative

body size (especially in females) and is most likely derived by scrutinizing ago-

nistic interactions between the individuals concerned. As an illustration, female

chacma baboons Papio ursinus pay greater attention when presented with manipu-

lated sequences of calls in which the affiliative grunt of a subordinate individual

is closely followed by a scream from a higher-ranking female – a situation incon-

sistent with the existing dominance relationship between those particular group

members – than they do to a control (and causally consistent) sequence (Cheney

et al., 1995). Eavesdropping may also have contributed, in part, to the pattern of

double-wins and double-losses observed in the rhesus experiment described above

(Mendoza, 1993), given that the interactions occurred within a triad setting (see

Chase et al. (2002) for similar logic).

Eavesdropping is often regarded as a means of gathering information on an un-

known competitor’s abilities without incurring the costs of directly confronting

the party concerned. For example, a bystander might integrate information on

how animal A fares against B with prior knowledge (gained via direct interaction)

of its own standing relative to A, to extrapolate how it too might fare against

the unknown B. Alternatively, even if informed only about the relative prowess

of two strangers, this may still provide a probabilistic indication of how the de-

feated individual might rate relative to oneself and thus be useful in guiding

behaviour. We suggest that another function of eavesdropping is simply to detect

that an individual (known or unknown) has suffered a loss and, therefore, may

be undergoing a loser experience. After all, in the cercopithecine systems being

described, individuals are generally well aware of who outranks whom within the

group; therefore, when contest outcome is in the expected direction, it may be

the information about temporary changes in an individual’s current state that



is more important in the decision to challenge. Note that while eavesdropping

can account for why known losers are more readily challenged than their victorious

counterparts (as in the experiments cited above), it does not in itself explain why an

individual should be challenged more frequently following a defeat than at other

times (as in our data). Acquiring information on the relative abilities of potential

competitors does not necessitate that it be put to use immediately; in contrast,

the transient nature of a putative loser effect would require immediate use of the

information.

Redirection influences the behaviour of bystanders

In both longtailed and rhesus macaques, the amount of aggression a vic-

tim received in the minutes following a redirection event was significantly lower

than that during comparable periods on occasions when it had lost but had not

redirected (Aureli & van Schaik, 1991b; Kazem, 1999; A. J. N. Kazem, unpublished

data). In other words, targeting third parties after a defeat appears to confer a

protective effect. It is always possible that some feature of the preceding conflict

influenced both the likelihood of redirection and of receiving subsequent aggres-

sion. Two results counsel against this view. First, in the rhesus study the likelihood

of receiving post-conflict harassment in the period before redirection took place

did not differ systematically between contests in which the victim did, or did not,

go on to redirect (the individual’s mean latency to redirect was used to define the

relevant timeframe in the latter). Second, rhesus youngsters redirected aggression

more frequently after low-level contests (threats or minor lunges) than after more

intense confrontations (involving prolonged pursuits or physical contact) – a point

to which we shall return. However, less-intense disputes were not in themselves

associated with receipt of low levels of subsequent harassment; victims incurred

virtually identical rates of aggression following mild or severe incidents. There-

fore, the reduction in harassment documented appears to be associated with the

act of redirecting itself.

It is not yet known whether the beneficial effect is achieved primarily via an

alteration in the behaviour of the former opponent or in that of opportunistic

bystanders. A change in the disposition of the former aggressor, at least, seems

likely because redirection apparently influences its behaviour in other respects.

Among longtailed macaques, redirecting is associated with an increased likelihood

that the former aggressor will later participate in an affiliative reunion with the

perpetrator (Aureli & van Schaik, 1991a), although a causal connection remains

to be demonstrated. Reconciliation between former adversaries is known to have

positive consequences in terms of restoring tolerance and reducing subsequent

aggression between the protagonists (e.g. Cords, 1992); although note that this

effect cannot have been indirectly responsible for the results we report above,



because post-conflict periods in which reconciliation had occurred were excluded

from analyses.

An honest indicator of post-conflict condition and motivation?

The evidence that redirection events may influence the disposition and

behaviour of individuals other than the target of aggression suggests there is a sig-

nalling advantage to be gained by behaving antagonistically in post-conflict con-

texts. In essence, the primary benefit of the behaviour might derive from it being

witnessed by third (or often, fourth) parties. It is often claimed that redirection

serves its purpose by focusing attention upon an alternative target, thus cutting

short the original contest and/or persuading potential challengers to look else-

where (Itani, 1963; Gust & Gordon, 1993). Data supporting such an outcome are

rarely, if ever, presented. This tactic may be successful in spotted hyaenas, social

carnivores that exhibit a high degree of within-group coordination in activities

and are liable to join in against any animal that is currently losing – to the point

where supporters of an aggressor will often switch sides simply because their target

counter-attacked its original opponent (Zabel et al., 1992). It seems less plausible

in macaques, where there is little compelling reason why the former aggressor

(at least) should so readily divert to a different target. Nor would it account for

why aggressors seem more willing to reconcile with victims that have redirected

against others.

We propose that redirecting does more than merely draw attention to a

new stimulus; the act itself may provide bystanders with useful information. As

outlined above, the physiological and psychological consequences of a defeat gen-

erally reduce the likelihood that an individual will initiate aggression in the

ensuing minutes, and they may render it less likely to persist and be less effective

in combat when challenged. However, a recent victim that is nevertheless suffi-

ciently confident and capable of rapidly redirecting against a third party thereby

demonstrates (to the scapegoat, and more importantly to others) that it has not

been unduly compromised by the preceding experience. The act may serve as an

unfakeable marker of the perpetrator’spost-conflict state, indicating that it would

be ready and/or able to defend itself, thus dissuading renewed or opportunistic

challenges from bystanders. It might also make individuals that redirect appear

more formidable rivals within the group, both to the former aggressors and their

kin (see below; Aureli & van Schaik, 1991a), which could explain why former aggres-

sors become more willing to reconcile. This hypothesis assumes that the neuroen-

docrine response to a perceived loss is not an all-or-nothing affair: that features of

the prior contest (primarily its intensity and duration, and perhaps the opponent’s

identity) can modulate the magnitude and/or type of changes undergone and the

time course of recovery. The fact that rhesus macaques were much more likely to



redirect following receipt of low-intensity as opposed to high-intensity harassment

(see above) is consistent with this argument; comparable results are also available

in rainbow trout (Øverli et al., 2004). Furthermore, in many taxa, individuals differ

consistently not only in their baseline levels of physiological parameters but also

in the magnitude and nature of neuroendocrine response (‘reactivity’) produced

by stressors such as received aggression (reviewed by Koolhaas et al., 1999). Conse-

quently, the degree to which a victim has been compromised following any specific

defeat is not straightforward for bystanders to surmise. Variation in post-conflict

state provides both the impetus for conveying this information to potential chal-

lengers and the means (motivation or ability) to do so.

Ensuring others notice the event

Redirection events possess several features likely to draw these acts to

the attention of bystanders. For example, reports often emphasize that victims

redirect ‘in front of ’ their former assailant (e.g. de Waal & Yoshihara, 1983; Aureli

et al., 1992; but see Watts, 1995b). Bouts of redirection by young rhesus macaques

in a free-ranging situation were also more likely to take place within 5 m of (and

hence within view and earshot of ) their former opponents than were equivalent

bouts instigated under control conditions, even when inter-opponent distance at

the start of paired observation periods was statistically controlled (Kazem, 1999).

As in other studies, it was not uncommon to observe subjects glancing back at

their previous adversary both prior to and while threatening the target, imply-

ing that the aggressor’s presence (and perhaps even line of gaze: Emery et al.,

1997; Tomasello et al., 1998) was actively taken into account. Experiments have

demonstrated that vervet monkeys Cercopithecus aethiops and macaques often take

the presence and composition of bystanders into account before behaving an-

tagonistically toward others (Keddy Hector et al., 1989; Cheney & Seyfarth, 1990).

Equally, a rule of thumb simply prompting victims to act quickly could ensure

that their former adversary was likely to have remained nearby (the mean latency

to redirect was 12 and 28 seconds in longtailed and rhesus macaques, respec-

tively, excluding those bouts which occurred within the original conflict (Aureli

& van Schaik, 1991b; Kazem, 1999)). In some cases, victims even approached and

attempted to enlist their former opponent’s support in the venture with con-

spicuous head-flagging (a recruitment gesture; see also Cords (1988) for simi-

lar behaviour in young longtailed macaques). It has been speculated that such

solicitation may be aimed at using partnership in a coalition to achieve some

form of ‘reconciliation’ with the former adversary (de Waal & Yoshihara, 1983;

Aureli & van Schaik, 1991b), as well as more directly reducing the likelihood

of renewed aggression from that quarter. Again, aggression against the target



appears an almost incidental by-product of communication with one’s former

opponent.

Redirection is also a particularly noisy affair. These interactions disproportion-

ately often incorporate vocal threats in comparison with equivalent bouts initiated

during control observations (Kazem, 1999). Increased incidence of vocal forms of

aggression might simply reflect greater arousal in animals that have themselves

recently been subjected to attacks, but whatever the proximate mechanism the

resulting events will be more effective at alerting bystanders. This is reminiscent

of Zahavi’s (1979) observation that many aggressive signals are far louder than

actually required for effective information transfer between the two parties di-

rectly concerned. He interpreted the probable interception of these signals by

several more distant receivers as imposing greater costs upon the signaller, there-

fore ensuring the reliability of the degree of threat conveyed to the opponent. We

suggest that one ‘shouts’ precisely in order to advertise the signal to those more

distant receivers, because one gains a benefit by doing so. Furthermore, evidence

for individual discrimination by voice exists in macaques, although the extent of

individual signatures varies according to call type (rhesus: Gouzoules et al., 1986;

Rendall et al., 1996, 1998). Perception of identity in threat calls does not appear to

have been tested but, if present, would allow the signaller’sactions to be identified

even if the redirection event was not observed.

Kin-oriented redirection: a special case?

An intriguing variant is that victims appear specifically to target ma-

ternal relatives of their former assailant. Examples have been reported in some

cercopithecines: vervet monkeys (Cheney & Seyfarth (1989), although they did not

specify whether the actor was victim or aggressor in the original conflict), juvenile

longtailed macaques (Aureli & van Schaik, 1991a), Japanese macaques (Aureli et al.,

1992) and pigtail macaques Macaca nemestrina ( Judge, 1982). As maternal relatives

share the (typically higher) familial rank of the original aggressor, the strategy

is not without its risks. Japanese macaques circumvent this issue by selecting

younger – and therefore often lower-ranking and more vulnerable – relatives of

their former opponent and take advantage of ‘safe’ opportunities to join ongoing

coalitions against the target, making it difficult for the target (or the former ag-

gressor) to retaliate (Aureli et al., 1992). Unsurprisingly, kin-oriented threats and

attacks typically account for only a small fraction of redirection events and may

take place over a longer timescale (minutes or even hours) because of the need

to encounter appropriate conditions. In Japanese macaques the majority (74%) of

these incidents occurred within view of the former aggressor, leading Aureli et al.

(1992) to propose that inflicting indirect fitness costs might serve as a form of so-

cial leverage, to deter further aggression from the same individual over the longer



term (assuming the latter associates the two events). The suggestion, therefore,

relies on a signalling argument, in this case restricted to a particular third-party

receiver. It has recently been demonstrated that apparently ‘spiteful’ acts (such as

these) can be evolutionarily stable in systems where observers accord each other

status on the basis of aggression witnessed, at least under certain simplified social

conditions ( Johnstone & Bshary, 2004).

Kin-oriented redirection need not be restricted to primates. Appropriate con-

ditions are provided in any system where individuals are often constrained from

retaliating directly against aggressors, reside in groups composed of a mixture of

related and unrelated conspecifics, and close kin are preferred associates or coali-

tion partners. Macaques and baboons can discriminate kin relationships between

third parties (e.g. Dasser, 1988; Cheney & Seyfarth, 1999), possibly by observing the

association patterns of other group members, and are often assumed to act on

this basis. However, use of simpler proximity-based rules may suffice. If relatives

of protagonists tend to cluster at the scene of conflicts (as is often the case in taxa

where individuals preferentially support their kin in coalitions), they will be over-

represented among the bystanders present. Therefore, a tendency to strike at any

vulnerable individual nearby (other than one’sown close associates) could have the

effect of disproportionately targeting the opponent’s kin under post-conflict con-

ditions. Spotted hyaenas and greylag geese Anser anser, both of which are reported

to redirect aggression (Table 10.2), might be promising candidates. Spotted hyae-

nas form large multimale–multifemale clans with high variance in within-group

relatedness and a matrilineal structure similar to macaques (Frank, 1986; Mills,

1990). Many geese form cohesive family units (pairs and their immature offspring)

that aggregate in large feeding flocks during winter. Both are highly competitive

societies with pronounced dominance relationships and a high propensity to in-

tervene aggressively on behalf of relatives in disputes; consequently, rank is highly

dependent on social support (hyaenas: Zabel et al., 1992; Engh et al., 2000; geese:

Lamprecht, 1986; Black & Owen, 1987; K. Kotrschal, personal communication). Kin

are valuable allies and often in spatial proximity, affording competitors the op-

portunity to learn (or otherwise locate) the habitual associates of others, as well

as ‘safe’ opportunities to redirect in coalition.

Intraspecific redirection in other taxa

A number of predictions can be made regarding systems in which redi-

rection might operate as a signal. First and foremost, conditions should facilitate

eavesdropping: the communication modality, social structure and typical habitat

should be such that agonistic signals often transmit further than the average spac-

ing between conspecifics and can, therefore, be detected by several receivers. The



extent of the communication network will often constrain whether bystanders are

likely to be aware of a contestant’s defeat, providing both the impetus to redirect

and the means by which its occurrence is detected.

Where individuals reside in permanent and cohesive social groups with de-

fined dominance relationships between group members (e.g. many primates, so-

cial carnivores, some ungulates), the majority of interactions witnessed are likely

to involve animals whose capabilities are already known to the individual and

aggression in the expected direction within a dyad (the exceptions being con-

tests involving immigrants and rank reversals occurring as individuals mature

or decline in ability). As we have argued for macaques, much of the utility of

eavesdropping may then lie in detecting transient loser effects (determining the

optimum timing of challenges), and redirection may primarily convey updated

information on a victim’spost-conflict state (as well as reaffirming the ‘status quo’

to both target and bystanders). If so, redirection should occur in conjunction with

a physiological/behavioural loser effect (at least in the wake of intense contests)

in systems where victims often receive several attacks from different individuals

in quick succession (coyotes Canis latrans may be an example; Table 10.2).

However, the redirection principle may be more broadly applicable. In many

taxa, individuals have knowledge of the abilities of a consistent subset of con-

specifics yet frequently encounter others whose status is not yet known. Examples

include species where kin units coalesce into larger but spatially structured aggre-

gations (e.g. winter feeding flocks in geese and some corvids), or individuals form

dominance relationships within a temporary display aggregation (e.g. male ducks).

Other possibilities include species where individuals defend nest sites within a

breeding colony or form closely spaced territories during the breeding season (as

in many birds and fish), thus forming relationships with their immediate neigh-

bours. Where eavesdropping is used mainly to estimate the fighting ability of un-

familiar competitors, redirecting could limit the negative impression conveyed

by a loss. Observers should be wary of challenging those losers that have neverthe-

less demonstrated the motivation and ability still to dominate some opponents

( just as eavesdroppers in some species respond more cautiously toward losers

that have exhibited persistent counter-aggression during their previous contest

(Earley & Dugatkin, 2002)). In this case, loser effects need not be present for redi-

rection to be worthwhile (although where they are, the incentive for redirecting

may be further enhanced).

There are also many systems where a signalling aspect to redirection seems

unlikely. For example, both eavesdropping and redirection would seem to be of less

use when individuals reside in large and continually shifting aggregations (either

year-round or during certain seasons), in which they possess little information on

conspecifics’ identities. In such cases, ‘badges of status’ may be used to mediate


Tabl

e10

.2.

Repo

rts

ofin

tras

peci

fic

redi

rect

ion

ofag

gres

sion

inot

her

spec

ies

Spec

ies

Con

text

inw

hic

hSy

stem

aC

oun

ter-

Lose

rV

icti

ms

Mai

nso

urc

ee

red

irec

tion

obse

rved

aggr

essi

onb

effe

ctc

chal

len

ged

d

Rai

nbo

wtr

out,

Onc

orhy

nchu

s

myk

iss

Juve

nil

etr

iad

s(u

nse

xed

),m

eeti

ng

insu

cces

sive

pai

rwis

e

exp

erim

enta

lco

nte

sts

(cap

tive

)

BV

aria

ble

1Ye

si–

Øve

rli

etal

.,20

04f

Atl

anti

csa

lmon

,Sal

mo

sala

rPa

rrh

eld

inm

ixed

-sex

quin

tets

and

sext

ets

infl

um

e(s

emin

atu

ral

stre

am)j

BYe

sH

.C.S

ute

r,p

erso

nal

com

mu

nic

atio

ng

Blu

egil

lsu

nfi

sh,L

epom

is

mac

roch

irus

Nes

tin

gm

ales

wit

hin

bree

din

g

colo

ny

(wil

d)

AYe

s?k

M.R

.Gro

ss,p

erso

nal

com

mu

nic

atio

ng

Am

aril

lofi

sh,G

irar

dini

chth

ys

mul

tira

diat

us

Mal

esin

mix

ed-s

exex

per

imen

tal

grou

ps

ofse

ven

(cap

tive

)

AV

aria

ble

1C

.Mac

ıas

Gar

cia,

per

son

al

com

mu

nic

atio

ng

Gre

ylag

goos

e,A

nser

anse

rG

and

ers

inw

inte

rfe

edin

gfl

ocks

(fre

era

ngi

ng)

CLo

wYe

sK

.Kot

rsch

alet

al.,

un

pu

blis

hed

dat

ah

Mal

lard

,Ana

spl

atyr

hync

hos

Dra

kes

wit

hin

pre

-pai

rin

gd

isp

lay

aggr

egat

ion

s(w

ild

and

outd

oor

pen

s)

CLo

wYe

sE.

J.A

.Cu

nn

ingh

am,

per

son

alco

mm

un

icat

ion

g

Gre

atti

t,Pa

rus

maj

orA

du

lts,

typ

ical

lym

ales

,in

smal

l

floc

ks(a

viar

ygr

oup

san

dw

inte

r

fora

gin

gfl

ocks

inw

ild

)

DV

aria

ble

2Ye

slYe

slP.

J.D

ren

t,p

erso

nal

com

mu

nic

atio

n;D

ren

t,

1983

g

Labo

rato

ryra

t,Ra

ttus

norv

egic

us

Dom

inan

tm

ales

inm

ixed

-sex

colo

nie

s(t

riad

s),i

nst

aged

resi

den

t–in

tru

der

test

s(c

apti

ve)n

BYe

smS.

Pell

is,p

erso

nal

com

mu

nic

atio

ng

Spot

ted

hya

ena,

Croc

uta

croc

uta

Ad

ult

san

dju

ven

iles

ofbo

thse

xes,

ofte

nfe

mal

es(w

ild

clan

san

d

cap

tive

coh

orts

)

ELo

wYe

soZ

abel

etal

.,19

92;M

.L.E

ast

&

H.H

ofer

,per

son

al

com

mu

nic

atio

ng

206


Coy

ote,

Cani

sla

tran

sB

oth

sexe

s,cu

bsw

ith

insi

blin

g

sext

ets

(wil

dan

dca

pti

veli

tter

s)

ELo

wYe

spYe

sM

.Bek

off,

per

son

al

com

mu

nic

atio

ng

Hu

man

,Hom

osa

pien

sB

oth

sexe

s,ad

ult

ssu

bjec

ted

to

verb

alp

rovo

cati

on

(exp

erim

enta

l)

E−

Mar

cus-

New

hal

let

al.,

2000

f

aSo

cial

syst

emty

pic

alof

this

con

text

inth

ew

ild

:A–D

,in

div

idu

als

fam

ilia

rw

ith

only

asu

bset

ofco

nsp

ecif

ics

regu

larl

yen

cou

nte

red

;A

,def

ence

ofn

est

or

cou

rtin

gsi

tein

seas

onal

aggr

egat

ion

byon

ese

x;B

,def

ence

ofte

rrit

ory

orfe

edin

gsi

te,d

omin

ance

rela

tion

ship

sbe

twee

nin

div

idu

als

resi

den

tin

loca

lare

a;C

,

def

ence

ofsh

ifti

ng

feed

ing

zon

ear

oun

din

div

idu

alor

fam

ily,

dom

inan

cere

lati

onsh

ips

betw

een

subs

ets

ofin

div

idu

als

ofon

eor

both

sexe

s;D

,mem

bers

hip

of

non

-ter

rito

rial

floc

kor

terr

itor

ialp

air,

dom

inan

cere

lati

onsh

ips

betw

een

subs

ets

ofco

nsp

ecif

ics;

E,m

embe

rsh

ipof

coh

esiv

egr

oup

,dom

inan

cere

lati

onsh

ips

betw

een

all

grou

pm

embe

rs.

bLo

w,b

etw

een

fam

ilia

rin

div

idu

als

occu

rsin

≤10

%of

dya

dic

con

test

s;va

riab

le1,

typ

ical

lylo

wbe

twee

nfa

mil

iar

ind

ivid

ual

s,bu

th

igh

erin

this

inst

ance

,

e.g.

beca

use

con

test

ants

un

fam

ilia

rat

star

tof

pro

toco

l;va

riab

le2,

typ

ical

lylo

wbe

twee

nfa

mil

iar

ind

ivid

ual

s;−

,pre

ven

ted

byex

per

imen

talp

roto

col.

The

like

lyfr

equ

ency

ofco

un

ter-

aggr

essi

ond

ecre

ases

from

low

tova

riab

le2

tova

riab

le1

toex

per

imen

tall

yp

reve

nte

d.

c Exp

erim

enta

lly

dem

onst

rate

du

sin

gsu

cces

sive

pai

rwis

eco

nte

sts,

orin

ferr

edfr

omse

quen

ces

ofin

tera

ctio

nin

inta

ctgr

oup

s.dR

epor

tsth

atvi

ctim

sre

ceiv

esu

bseq

uen

th

aras

smen

tfr

omby

stan

der

sin

the

min

ute

sfo

llow

ing

anin

itia

ld

efea

t.e Ev

iden

cefo

rre

dir

ecti

ond

emon

stra

ted

exp

erim

enta

lly.

f inco

mp

aris

onw

ith

con

trol

dat

a;gba

sed

onan

ced

otal

rep

orts

from

auth

ors

coll

ecti

ng

quan

tita

tive

dat

aon

aggr

essi

vein

tera

ctio

ns;

hor

infe

rred

from

quan

tita

tive

but

un

con

trol

led

dat

a.i A

bbot

tet

al.,

1985

.j Se

min

atu

ral

grou

ps

hel

dat

rela

tive

lyh

igh

den

sity

;red

irec

tion

not

obse

rved

inw

ild

.k B

ehav

iou

ral

lose

ref

fect

has

been

doc

um

ente

din

Lepo

mis

gibb

osus

(Bea

cham

&N

ewm

an,1

987;

Ch

ase

etal

.,19

94)a

nd

L.cy

anel

lus

(McD

onal

det

al.,

1968

).l V

erbe

ek,1

998;

Ver

beek

etal

.,19

99.

mSe

war

d,1

946;

van

der

Poll

etal

.,19

82.

nN

otkn

own

wh

eth

erth

esu

bjec

tw

aslo

ser

orw

inn

erof

the

init

ial

con

test

.o O

nly

occa

sion

ally

rece

ived

from

byst

and

ers,

but

Wah

ajet

al.,

2001

dem

onst

rate

rece

ipt

from

form

erop

pon

ent.

pB

ekof

f&

Du

gatk

in,2

000.

207



access to resources such as food, where individual items are of relatively low

value (e.g. certain finches: Rohwer, 1975; Johnstone & Norris, 1993). Displaying a

badge means an individual’s relative aggressive propensity and/or ability should

be immediately apparent to those it meets and to others who witness its social

interactions, reducing the utility of eavesdropping. However, if dominance is not

mediated solely via plumage badges (e.g. individuals may still form relationships

with a subset of familiar conspecifics), and as the magnitude of any loser effect

may still vary, a signalling role for redirection is not ruled out.

Finally, a low probability of counter-aggression during contests is thought to

promote redirection (e.g. Thierry, 1985), because this factor influences both the

necessity and the costs of the behaviour. As mentioned above, redirection offers an

outlet for arousal and frustration in situations where animals cannot retaliate ag-

gressively against an initiator, either because they are physically prevented from

doing so (in experiments) or because the adversary is perceived as too powerful.

A signalling interpretation can also predict a negative association between the

two forms of response; they might be alternative methods of ‘proving oneself ’ to

pre-empt strikes from conspecifics that had witnessed the defeat (although indi-

viduals might sometimes do both to reinforce the message; cf. a multicomponent

signal). Second, if there is a high risk of reprisal from the target, redirection may

become an excessively costly option. Both arguments have been used to explain

the absence of operationally defined redirection in adult female olive baboons

and mountain gorillas (Watts, 1995b; Castles & Whiten, 1998b); bidirectional ag-

gression is common between adult females in these species. Potential retaliation

from targets might also explain why the majority of ‘redirection’ in a small cap-

tive group of chimpanzees Pan troglodytes was directed toward human caretakers,

rather than conspecifics (Malone et al., 2000). Suitable targets are expected to be

those likely to capitulate immediately. Selection may be based on familiarity:

targeting a known subordinate makes sense, especially in species with a ‘strict’

dominance style. However, certain classes of unfamiliar conspecific can also be

identified as unlikely to dispute the outcome, for example targeting first-year in-

dividuals in birds (identifiable by their size and plumage) or smaller floater and

satellite males in fish.

Reports of intraspecific redirection

When placed in situations where they are unable to retaliate directly to

verbal provocation, humans will readily redirect hostility by verbally or physically

punishing substitute targets, both animate and inanimate (reviewed by Marcus-

Newhall et al., 2000). Experimentally induced ‘displaced aggression’ has also been

observed in rodents. Although females are not ordinarily subjected to aggression

by males, mates may become the target of ‘redirected’ attacks in the seconds



following staged contests between resident males and same-sex intruders (mice

Peromyscus spp.: Eisenberg, 1962; Simmel & Walker, 1970; montane and prairie

voles Microtus montanus and Microtus ochrogaster: S. Pellis, personal communication),

especially if continued access to the opponent(s) is suddenly barred. Unfortunately,

in rodent studies it is often unclear whether the redirecting individual was, or

would eventually have become, the loser in the original contest. A rather artificial

situation is also imposed by the research foci of these experiments; the subject

may not have the option of retaliating against the original aggressor because

s/he is no longer present (often the case in human studies) or may be physically

prevented from pursuing the original contest to its conclusion (e.g. because of

partitions separating opponents in work on rodents). Consequently the subject is

generally presented with a limited range of options and stimuli against which to

direct any response. However, apparent redirection of aggression by victims has

been reported under more naturalistic social contexts in a wide range of species

(a selection, by no means a comprehensive survey, is presented in Table 10.2).

In some species, the behaviour may be performed purely for the physiologi-

cal effects: the benefits of rapidly reducing arousal and perhaps of experiencing

victory. However, the occurrence of redirection is likely to be used as a cue by

bystanders, because it inevitably carries information about an unknown competi-

tor’s relative ability within a population and, where winner or loser effects are

in operation, its current state. If this means bystander behaviour is influenced to

the redirecting individual’sadvantage, one would expect redirection to have been

selected as a signal that is performed more than required for strictly physiological

reasons (or even in the absence of any physiological benefits) and adjusted so as

to be effectively publicized. The challenge is to identify where particular taxa fall

within this spectrum of possibilities.

Testing the occurrence and function of redirection

An essential preliminary is verifying whether an identifiable phe-

nomenon exists – i.e. a pronounced increase in initiation of aggression after

being victimized in comparison with the incidence under appropriate control

conditions. As the primate data illustrate (Table 10.1), redirection may fail to be

demonstrated operationally in species (or age–sex classes) where, on the basis of

uncontrolled data, it had been assumed to be present. It would also be useful to clar-

ify whether counter-aggression and redirection are dissociated, and what factors

favour expression of one tactic over the other. In observational work to date, there

is often insufficient data per individual to allow analyses that compare the inci-

dence of redirection after unidirectional versus bidirectional contests, whilst also

controlling for systematic differences in aggression intensity, duration and num-

ber of participants in the contest. Crucially, the existence of a causal relationship



between redirecting and a reduction in the number of challenges subsequently

received by the actor requires experimental verification. This could be addressed

by experimentally manipulating the access of the former aggressor and other

bystanders to visual/acoustic information concerning the initial conflict and its

aftermath – how do counter-aggression and redirection compare in terms of their

influence upon the behaviour of witnesses? Furthermore, demonstrating that vic-

tims are sensitive to the perceived presence and composition of bystanders when

redirecting (and not just the availability of a suitable target) would clearly sup-

port a signalling interpretation. Given the likely costs of engaging in aggression,

such facultative use of redirection is predicted. Although absence of an ‘audience

effect’ (i.e. a change in behaviour when in the perceived presence of conspecifics;

see Ch. 4) does not preclude a signalling function, one would expect that use of an

unconditional rule could evolve only if, under natural circumstances, a relevant

audience is almost invariably present. Finally, in taxa that exhibit a loser effect,

it should be possible to confirm whether conflicts concluded after differing dura-

tions and intensity of aggression produce different physiological changes in the

loser. This would provide a basis for exploring how neuroendocrine profiles in the

immediate post-conflict period, and the time course of recovery, might differ in

cases where the animal did or did not redirect.

Summary

We propose that redirection of aggression may function, in part, as a sig-

nal, used to pre-empt subsequent challenges from the former aggressor and/or

other bystanders that witness the act. In systems where behavioural loser effects

exist, such as macaques, this might be achieved because redirection serves as an

honest indicator of post-conflict condition and motivation, demonstrating that

the perpetrator has not been unduly compromised by its preceding defeat. Al-

ternatively, redirection might simply limit the extent of the negative impression

usually conveyed by a loss, reaffirming one’s position within a status hierarchy to

bystanders or, where the bystander is unfamiliar with the protagonist, conveying

an ability to defeat at least some individuals within the population.

This suggestion has a number of broader implications. Being observed to lose a

contest incurs social penalties, affecting not only the likelihood of being chal-

lenged by others but also the prospects of acquiring or retaining mates (e.g.

Doutrelant & McGregor, 2000; Mennill et al., 2002). So far, attention has focused

exclusively on how contestants counter these pressures by modifying their dis-

plays during the original agonistic exchange if in the perceived presence of an

audience (e.g. Doutrelant et al., 2001; Matos & McGregor, 2002; A. J. N. Kazem,

R. J. Motos & P. K. McGregor, unpublished data). In reality, there may be several



points at which the eventual loser can influence others’ perceptions of its ability:

in the decision to counter-attack rather than defer, to fight harder (or differ-

ently) in the current exchange, or even by redirecting after the contest has con-

cluded. First, this suggests it would be profitable to consider audience effects and

eavesdropping over a longer timescale, encompassing sequences of contests. Sec-

ond, individuals who approach conflicts between others risk becoming targets

of redirection (this may explain why, in our study species, low-ranking individ-

uals often flee the scene of incipient or ongoing aggression). Therefore, depend-

ing on the signal modality used, the costs of eavesdropping may not be as low

as previously assumed – particularly for low-ranking or otherwise poor competi-

tors, which stand to gain the most from acquiring information via this route.

Finally, redirection may constitute a rare example of a signal aimed primar-

ily at ‘secondary’ receivers. The communication networks perspective has high-

lighted how many dyadic displays evolve within the context of, and may even

be designed to advertise to, several receivers. However, redirection may be an in-

stance of a signal performed almost entirely for its effect on third parties rather

than upon the apparent receiver. There has been substantial theoretical inter-

est in the possibility that apparently altruistic behaviour may be maintained via

this route (e.g. Zahavi, 1977; Nowak & Sigmund, 1998; Roberts, 1998; Leimar &

Hammerstein, 2001), but few empirical examples have been discovered outside

humans. Redirection has the potential to prove a taxonomically widely distributed

case, but experimental verification and analyses to explore the evolutionary sta-

bility of such a signalling system in aggression are still required.

Acknowledgements

We are very grateful to the many correspondents who provided access to unpublished

data and detailed observations on their study taxa. We would also like to thank Peter McGregor

for inviting us to contribute to this volume, and his encouragement and support of the first

author’s ongoing work on redirection. The latter’s research on rhesus macaques was financially

supported by awards from the H. F. Guggenheim Foundation, L. S. B. Leakey Foundation, Wenner

Gren Foundation for Anthropological Research, Zunz Foundation and a University of Durham

Research Studentship; the generosity of these organizations is greatly appreciated.

References

Abbott, J. C., Dunbrack, R. L. & Orr, C. D. 1985. The interaction of size and experience

in dominance relationships of juvenile steelhead trout (Salmo gairdneri). Behaviour,

92, 241–253.

Andrews, R. J. 1991. Testosterone, attention and memory. In: The Development and

Integration of Behaviour, ed. P. P. G. Bateson. Cambridge, UK: Cambridge University

Press, pp. 171–190.



Arnold, K. & Barton, R. A. 2001. Postconflict behavior of spectacled leaf monkeys

(Trachypithecus obscurus). II. Contact with third parties. International Journal of


Aureli, F. 1992. Post-conflict behaviour among wild long-tailed macaques (Macaca

fascicularis). Behavioral Ecology and Sociobiology, 31, 329–337.

Aureli, F. & van Schaik, C. P. 1991a. Post-conflict behaviour in long-tailed macaques

(Macaca fascicularis): I. The social events. Ethology, 89, 89–100.

1991b. Post-conflict behaviour in long-tailed macaques (Macaca fascicularis): II.

Coping with the uncertainty. Ethology, 89, 101–114.

Aureli, F., Cozzolino, R., Cordischi, C. & Scucchi, S. 1992. Kin-oriented redirection

among Japanese macaques: an expression of a revenge system? Animal Behaviour,

44, 283–291.

Aureli, F., Veenema, H. C., van Panthaleon van Eck, C. J. & van Hooff, J. A. R. A. M. 1993.

Reconciliation, consolation, and redirection in Japanese macaques (Macaca

fuscata). Behaviour, 124, 1–21.

Aureli, F., Das, M., Verleur, D. & van Hooff, J. A. R. A. M. 1994. Postconflict social

interactions among Barbary macaques (Macaca sylvanus). International Journal of


Aureli, F., Cords, M. & van Schaik, C. P. 2002. Conflict resolution following aggression

in gregarious animals: a predictive framework. Animal Behaviour, 64, 325–343.

Bastock, M., Morris, D. & Moynihan, M. 1953. Some comments on conflict and

thwarting in animals. Behaviour, 6, 66–84.

Beacham, J. L. & Newman, J. A. 1987. Social experience and the formation of

dominance relationships in the pumpkinseed sunfish, Lepomis gibbosus. Animal

Behaviour, 35, 1560–1563.

Bekoff, M. & Dugatkin, L. A. 2000. Winner and loser effects and the development of

dominance relationships in young coyotes: an integration of data and theory.

Evolutionary Ecology Research, 2, 871–883.

Bernstein, I. S., Rose, R. M. & Gordon, T. P. 1974. Behavioral and environmental

events influencing primate testosterone levels. Journal of Human Evolution, 3,

517–525.

Black, J. M. & Owen, M. 1987. Determinants of social rank in goose flocks: acquisition

of social rank in young geese. Behaviour, 102, 129–146.

Blanchard, D. C., Sakai, R. R., McEwen, B. & Blanchard, R. J. 1993. Subordination stress:

behavioral, brain and neuroendocrine correlates. Behavioural Brain Research, 58,

113–121.

Booth, A., Shelley, G., Mazur, A., Tharp, G. & Kittok, R. 1989. Testosterone, and winning

and losing in human competition. Hormones and Behavior, 23, 556–571.

Castles, D. L. & Whiten, A. 1998a. Post-conflict behaviour of wild olive baboons. II.

Stress and self-directed behaviour. Ethology, 104, 148–160.

1998b. Post-conflict behaviour of wild olive baboons. I. Reconciliation, redirection

and consolation. Ethology, 104, 126–147.

Chapais, B. 1995. Alliances as a means of competition in primates: evolutionary,

developmental and cognitive aspects. Yearbook of Physical Anthropology, 38, 115–

136.



Chase, I. D., Bartolomeo, C. & Dugatkin, L. 1994. Aggressive interactions and


393–400.

Chase, I. D., Tovey, C., Spangler-Martin, D. & Manfredonia, M. 2002. Individual

differences versus social dynamics in the formation of animal dominance

hierarchies. Proceedings of the National Academy of Sciences, USA, 99, 5744–5749.

Cheney, D. L. & Seyfarth, R. M. 1986. The recognition of social alliances by vervet

monkeys. Animal Behaviour, 34, 1722–1731.

1989. Redirected aggression and reconciliation among vervet monkeys, Cercopithecus

aethiops. Behaviour, 110, 258–275.

1990. Attending to behaviour versus attending to knowledge: examining monkeys’

attribution of mental states. Animal Behaviour, 40, 742–753.

1999. Recognition of other individuals’ social relationships by female baboons.


Cheney, D. L., Seyfarth, R. M. & Silk, J. B. 1995. The responses of female baboons (Papio

cynocephalus ursinus) to anomalous social interactions: evidence for causal

reasoning? Journal of Comparative Psychology, 109, 134–141.

Cords, M. 1988. Resolution of aggressive conflicts by immature long-tailed macaques,

Macaca fascicularis. Animal Behaviour, 36, 1124–1135.

1992. Post-conflict reunions and reconciliation in long-tailed macaques. Animal


Cynx, J. & Nottebohm, F. 1992. Testosterone facilitates some conspecific song

discriminations in castrated zebra finches Taeniopygia guttata. Proceedings of the

National Academy of Sciences, USA, 89, 1376–1378.

Das, M. 1998. Conflict management and social stress in long-tailed macaques. Ph.D.

Thesis, University of Utrecht, the Netherlands.

Das, M., Penke, Z. & van Hooff, J. A. R. A. M. 1997. Affiliation between aggressors and

third parties following conflicts in long-tailed macaques (Macaca fascicularis).

International Journal of Primatology, 18, 157–179.

Dasser, V. 1988. A social concept in Java monkeys. Animal Behaviour, 36, 225–230.

Delville, Y., Mansour, K. M., Ferris, C. F. 1996. Testosterone facilitates aggression by

modulating vasopressin receptors in the hypothalamus. Physiology and Behavior,

60, 25–29.

de Waal, F. B. M. & Luttrell, L. M. 1989. Toward a comparative socioecology of the

genus Macaca: different dominance styles in rhesus and stumptail macaques.

American Journal of Primatology, 19, 83–109.

de Waal, F. B. M. & Yoshihara, D. 1983. Reconciliation and redirected affection in

rhesus monkeys. Behaviour, 85, 224–241.






Drent, P. J. 1983. The functional ethology of territoriality in the great tit (Parus major

L.). Ph.D. Thesis, University of Groningen, the Netherlands.






Eaton, G. G. 1984. Aggression in adult male primates: a comparison of confined

Japanese macaques and free-ranging olive baboons. International Journal of


Eisenberg, J. F. 1962. Studies on the behavior of Peromyscus maniculatus gambelli and

Peromyscus californicus parasiticus. Behaviour, 19, 177–207.

Emery, N. J., Lorincz, E. N., Perret, D. I., Oram, M. W. & Backer, C. I. 1997. Gaze

following and joint attention in rhesus monkeys (Macaca mulatta). Journal of

Comparative Psychology, 111, 286–293.

Engh, A. L., Esch, K., Smale, L. & Holekamp, K. E. 2000. Mechanisms of maternal rank

‘inheritance’ in the spotted hyaena, Crocuta crocuta. Animal Behaviour, 60,

323–332.

Fairbanks, L. A., Melega, W. P., Jorgensen, M. J., Kaplan, J. R. & McGuire, M. T. 2001.

Social impulsivity inversely associated with CSF 5-HIAA and fluoxetine exposure

in vervet monkeys. Neuropsychopharmacology, 24, 370–378.

Ferris, C. F., Melloni, R. H., Jr, Koppel, G. et al. 1997. Vasopressin/serotonin interactions

in the anterior hypothalamus control aggressive behavior in golden hamsters.

Journal of Neuroscience, 17, 4331–4340.

Frank, L. G. 1986. Social organization of the spotted hyena (Crocuta crocuta). 2.

Dominance and reproduction. Animal Behaviour, 34, 1510–1527.

Fuentes, A., Malone, N., Sanz, C., Matheson, M. & Vaughan, L. 2002. Conflict and

post-conflict behavior in a small group of chimpanzees. Primates, 43, 223–235.

Gore, M. A. 1994. Dyadic and triadic aggression and assertiveness in adult female

rhesus monkeys, Macaca mulatta, and hamadryas baboons, Papio hamadryas. Animal

Behaviour, 48, 385–392.

Gouzoules, H., Gouzoules, S. & Marler, P. 1986. Vocal communication: a vehicle for the

study of social relationships. In: The Cayo Santiago Macaques: History, Behavior and

Biology, ed. R. G. Rawlins & M. J. Kessler. Albany, NY: State University of New York

Press, pp. 111–129.

Gust, D. A. & Gordon, T. P. 1993. Conflict resolution in sooty mangabeys. Animal

Behaviour, 46, 685–694.

Gust, D. A., Gordon, T. P., Hambright, M. K. & Wilson, M. E. 1993. Relationship between

social factors and pituitary-adrenocortical activity in female rhesus monkeys

(Macaca mulatta). Hormones and Behavior, 27, 318–331.

Hannes, R. P., Franck, D. & Liemann, F. 1984. Effects of rank order fights on body and

blood concentrations of androgens and corticosteroids in the male swordtail

(Xiphophorus helleri). Zeitschrift fur Tierpsychologie, 65, 53–65.

Higley, J. D., Mehlman, P. T., Poland, R. E. et al. 1996. CSF testosterone and 5-HIAA

correlate with different types of aggressive behaviors. Biological Psychiatry, 40,

1067–1082.

Hogue, M.-E., Beaugrand, J. P. & Lague, P. C. 1996. Coherent use of information by hens


Behavioural Processes, 38, 241–252.






Behaviour, 61, 777–786.

Itani, J. 1963. Vocal communication of the wild Japanese monkeys. Primates, 4, 11–66.

Johnstone, R. A. & Bshary, R. 2004. Evolution of spite through indirect reciprocity.


Johnstone, R. A. & Norris, K. 1993. Badges of status and the cost of aggression.


Judge, P. G. 1982. Redirection of aggression based on kinship in a captive group of

pigtail macaques. International Journal of Primatology, 3, 301.

Kavaliers, M., Choleris, E. & Colwell, D. D. 2001. Brief exposure to female odors

‘emboldens’ male mice by reducing predator-induced behavioral and hormonal

responses. Hormones and Behavior, 40, 497–509.

Kazem, A. J. N. 1999. Conflict management in free-ranging immature rhesus

macaques, Macaca mulatta. Ph.D. Thesis, University of Durham, UK.

Keddy Hector, A., Seyfarth, R. M. & Raleigh, M. J. 1989. Male parental care, female

choice, and the effect of an audience in vervet monkeys. Animal Behaviour, 38,

262–271.

Koolhaas, J. M., Korte, S. M., de Boer, S. F. et al. 1999. Coping styles in animals: current

status in behavior and stress-physiology. Neuroscience and Biobehavioral Reviews, 23,

925–935.

Kutsukake, N. & Castles, D. L. 2001. Reconciliation and variation in post-conflict stress

in Japanese macaques (Macaca fuscata fuscata): testing the integrated hypothesis.

Animal Cognition, 4, 259–268.

Lamprecht, J. 1986. Structure and causation of the dominance hierarchy in a flock of

bar-headed geese (Anser indicus). Behaviour, 96, 28–48.

Leimar, O. & Hammerstein, P. 2001. Evolution of cooperation through indirect

reciprocity. Proceedings of the Royal Society of London, Series B, 268, 745–753.

Levine, S., Coe, C. & Wiener, S. G. 1989. Psychoneuroendocrinology of stress: a

psychobiological perspective. In: Psychoendocrinology, ed. S. Levine & R. Brush.

London: Academic Press, pp. 341–377.

Malone, N., Vaughan, L. & Fuentes, A. 2000. The role of human caregivers in the

post-conflict interactions of captive chimpanzees (Pan troglodytes). Laboratory

Primate Newsletter, 39, 1–3.

Marcus-Newhall, A., Pedersen, W. C., Carlson, M. & Miller, N. 2000. Displaced

aggression is alive and well: a meta-analytic review. Journal of Personality and Social

Psychology, 78, 670–689.

Matos, R. J. & McGregor, P. K. 2002. The effect of the sex of an audience on male–male

displays of Siamese fighting fish (Betta splendens). Behaviour, 139, 1211–1221.

McDonald, A. L., Heimstra, N. W. & Damkot, D. K. 1968. Social modification of

agonistic behaviour in fish. Animal Behaviour, 16, 437–441.






McGregor, P. K., Peake, T. M. & Lampe, H. M. 2001. Fighting fish, Betta splendens, extract


see is not what you get. Animal Behaviour, 62, 1059–1065.

Mehlman, P. T., Higley, J. D., Faucher, I. et al. 1994. Low CSF 5-HIAA concentrations and

severe aggression and impaired impulse control in nonhuman primates. American

Journal of Psychiatry, 151, 1485–1491.

Mendoza, S. P. 1993. Social conflict on first encounters. In: Primate Social Conflict, ed.

W. A. Mason & S. P. Mendoza. Albany, NY: State University of New York Press,

pp. 85–110.



Mills, M. G. L. 1990. Kalahari hyaenas: Comparative Behavioural Ecology of Two Species.

London: Unwin Hyman.

Moynihan, M. H. 1998. The social regulation of competition and aggression in animals.

Washington, DC: Smithsonian Institution Press.


Nature, 393, 573–577.




Olivier, B., Mos, J., van Oorschot, R. & Hen, R. 1995. Serotonin receptors and animal

models of aggressive behavior. Pharmacopsychiatry, 28, 80–90.

Øverli, Ø., Harris, C. A. & Winberg, S. 1999. Short-term effects of fights for social

dominance and the establishment of dominant-subordinate relationship on brain

monoamines and cortisol in rainbow trout. Brain Behavior and Evolution, 54,

263–275.

Øverli, Ø., Korzan, W. J., Larson, E. T. et al. 2004. Behavioural and neuroendocrine

correlates of displaced aggression in trout: cycle of violence is evolutionarily

conserved. Hormones and Behavior, 45, 3224–3349.




Perreault, H. A. N., Semsar, K. & Godwin, J. 2003. Fluoxetine treatment decreases

territorial aggression in a coral reef fish. Physiology and Behavior, 79,

719–724.

Rendall, D., Rodman, P. S. & Emond, R. E. 1996. Vocal recognition of individuals and

kin in free-ranging rhesus monkeys. Animal Behaviour, 51, 1007–1015.

Rendall, D., Owren, M. J. & Rodman, P. S. 1998. The role of vocal tract filtering in

identity cueing in rhesus monkey (Macaca mulatta) vocalizations. Journal of the


Roberts, G. 1998. Competitive altruism: from reciprocity to the handicap principle.


Roeder, J.-J., Fornasieri, I. & Gosset, D. 2002. Conflict and postconflict behaviour in two

lemur species with different social organizations (Eulemur fulvus and Eulemur

macaco): a study on captive groups. Aggressive Behavior, 28, 62–74.



Rohwer, S. 1975. The social significance of avian plumage variability. Evolution, 29,

593–610.

Rose, R. M., Bernstein, I. S. & Gordon, T. 1975. Consequences of social conflict on

plasma testosterone levels in rhesus monkeys. Psychosomatic Medicine, 37, 50–61.

Sapolsky, R. M. 1998. Why Zebras Don’t get Ulcers: An Updated Guide to Stress, Stress-related

Diseases, and Coping, 2nd edn. New York: Freeman.

Sapolsky, R. M. & Ray, J. C. 1989. Styles of dominance and their endocrine correlates

among wild olive baboons (Papio anubis). American Journal of Primatology, 18, 1–13.

Schuurman, T. 1980. Hormonal correlates of agonistic behaviour in adult male rats.

Progress in Brain Research, 53, 415–520.

Scott, J. P. & Fredericson, E. 1951. The causes of fighting in mice and rats. Physiological

Zoology, 24, 273–309.

Scucchi, S., Cordischi, C., Aureli, F. & Cozzolino, R. 1988. The use of redirection in a

captive group of Japanese monkeys. Primates, 29, 229–236.

Seward, J. P. 1946. Aggressive behavior in the rat. IV. Submission as determined by

conditioning, extinction and disuse. Journal of Comparative Psychology, 39,

51–57.

Silk, J. B. 1999. Male bonnet macaques use information about third-party rank

relationships to recruit allies. Animal Behaviour, 58, 45–51.

Simmel, E. C. & Walker, D. A. 1970. Social priming for agonistic behavior in a ‘docile’

mouse strain. American Zoologist, 10, 486–487.

Summers, C. H., Summers, T. R., Moore, M. C. et al. 2003. Temporal patterns of limbic

monoamine and plasma corticosterone response during social stress. Neuroscience,

116, 553–563.

Thierry, B. 1985. Patterns of agonistic interactions in three species of macaque (Macaca

mulatta, M. fascicularis, M. tonkeana). Aggressive Behavior, 11, 223–233.

2000. Covariation of conflict management patterns in macaque societies. In: Natural

Conflict Resolution, ed. F. Aureli & F. B. M. de Waal. Berkeley, CA: University of

California Press, pp. 106–128.

Tomasello, M., Call, J. & Hare, B. 1998. Five primate species follow the visual gaze of

conspecifics. Animal Behaviour, 55, 1063–1069.

van de Poll, N. E., de Jonge, F., van Oyen, H. G. & van Pett, J. 1982. Aggressive behaviour

in rats: effects of winning and losing on subsequent aggressive interactions.


van Erp, A. M. M. & Miczek, K. A. 2000. Aggressive behavior, increased accumbal

dopamine and decreased cortical serotonin in rats. Journal of Neuroscience, 15,

9320–9325.

Veenema, H. C. 2000. Methodological progress in post-conflict research. In: Natural

Conflict Resolution, ed. F. Aureli & F. B. M. de Waal. Berkeley, CA: University of

California Press, pp. 21–23.

Verbeek, M. 1998. Bold or cautious: behavioural characteristics and dominance in

great tits. Ph.D. Thesis, University of Wageningen, the Netherlands.

Verbeek, M. E. M., de Goede, P., Drent, P. J. & Wiepkema, P. R. 1999. Individual

behavioural characteristics and dominance in aviary groups of great tits.




Virgin, C. E., Jr & Sapolsky, R. M. 1997. Styles of male social behavior and their

endocrine correlates among low-ranking baboons. American Journal of Primatology,

42, 25–39.

Wahaj, S. A., Guse, K. R. & Holekamp, K. E. 2001. Reconciliation in the spotted hyena

(Crocuta crocuta). Ethology, 107, 1057–1074.

Walker Leonard, J. 1979. A strategy approach to the study of primate dominance

behaviour. Behavioural Processes, 4, 155–172.

Watts, D. P. 1995a. Post-conflict social events in wild mountain gorillas (Mammalia,

Hominoidea). I. Social interactions between opponents. Ethology, 100, 139–157.

1995b. Post-conflict social events in wild mountain gorillas. II. Redirection, side

direction, and consolation. Ethology, 100, 158–174.

Weiss, J. M., Pohorecky, L. A., Salman, S. & Gruenthal, M. 1976. Attenuation of gastric

lesions by psychological aspects of aggression in rats. Journal of Comparative and

Physiological Psychology, 90, 252–259.

Winberg, S. & Lepage, O. 1998. Elevation of brain 5-HT activity, POMC expression, and

plasma cortisol in socially subordinate rainbow trout. American Journal of

Physiology, 274, R645–R654.

Zabel, C. J., Glickman, S. E., Frank, L. G., Woodmansee, K. B. & Keppel, G. 1992.

Coalition formation in a colony of prepubertal spotted hyenas. In: Coalitions and

Alliances in Humans and Other Animals, ed. A. H. Harcourt & F. B. M. de Waal. Oxford:

Oxford University Press, pp. 113–135.

Zahavi, A. 1977. Reliability in communication systems and the evolution of altruism.

In: Evolutionary Ecology, ed. B. Stonehouse & C. M. Perrins. London: Macmillan,

pp. 253–259.

1979. Why shouting? American Naturalist, 113, 155–156.


11

Scent marking and socialcommunication

j a n e l . h u r s t

University of Liverpool, Liverpool, UK

Introduction

The use of chemical scents for communication between individuals is

widespread among both vertebrate and invertebrate animals. Scent signals ema-

nating from an animal’s body can be used for intimate and immediate commu-

nication when two or more individuals interact at close quarters, but scents can

also be deposited in the environment in the form of scent marks. Unlike most

visual or acoustic signals used by animals, scent marks persist in the absence of

the signaller, often over extended periods. The prolonged duration of signals in

deposited scent marks makes them particularly suited for broadcasting informa-

tion to all conspecifics that visit a scent-marked site. Further, scent marks might be

deposited to signal to certain individuals, such as when animals are attempting to

attract potential mates or to indicate a territorial boundary to neighbours. Once

deposited in the environment, however, the scent is not physically directed to-

wards specific recipients and the information will be available to any other ani-

mals in the locality. This ready availability of scent marks to third parties is likely

to provide strong selection pressure to ensure that the information deposited

in scent marks is appropriate for communication to any individual likely to en-

counter the scent. Consequently, scent marks are likely to have evolved to be used

for network communication rather than as signals between specific individuals.

While volatile components of the scent may be detected at some distance from

a scent mark, alerting animals to the presence and location of scent signals, non-

volatile components can be detected only by close contact investigation. Typically,

animals approach and sniff scent marks very closely; many species will also lick the

scent. Scent marks may thus communicate information via both volatile odorants


219


220 J. L. Hurst

and non-volatile components or those of low volatility, and perception may involve

more than one chemosensory system. For example, mammals have both a main

olfactory system linked to the olfactory epithelium (reviewed by Nibu, 2002) and an

accessory olfactory system linked to the vomeronasal organ (reviewed by Brennan,

2001; Takami, 2002; Zufall et al., 2002).

Scent marking is particularly common among terrestrial, arboreal and subter-

ranean mammals and plays an important role in a number of social contexts,

including recognition of group members and kin, the advertisement of territory

ownership and social dominance, assessment of the quality of competitors and

potential mates, and the advertisement and control of reproductive status. In

each of these contexts, communication characteristically involves all individuals

that deposit and detect scent marks within a particular area rather than private

communication between two individuals. To illustrate this, I will review the use

of scent marks for communication among house mice Mus musculus domesticus,

in which both the behavioural and biochemical mechanisms underlying scent

communication have been well studied, both in the laboratory and among wild

mice in more naturalistic enclosure experiments. In the concluding part of the

chapter, I will discuss the advantages of using scent marks as broadcast signals to

communicate to a network of conspecifics rather than directing signals to specific

individuals.

Scent marking among house mice

Urine is the primary source of social odours among house mice and con-

tains ‘fixed’ (genomic) information about the species, sex, individual identity, ma-

jor histocompatibility complex (MHC) type and other genetic background of the

owner, as well as ‘variable’ (metabolic) information on the owner’s current social,

reproductive and health status, and its food resources (reviewed by Brown, 1985a,

1995; Hurst et al., 2001a; Malone et al., 2001). Once old enough to leave the nest,

mice of both sexes scent mark by depositing urine in small spots and streaks on

the substrate in a deliberate pattern of deposition as they move around their home

area (Fig. 11.1). All surfaces are rapidly covered with this background scent mark-

ing and scent mark density patterns generally correspond with spatial patterns of

activity, most marks being deposited around physical edges such as walls, feeding

sites and near nest sites (Fig. 11.2). Mice do not appear to urine mark within their

nest site. Once mice have scent marked a patch of substrate thoroughly, the rate of

further background marking decreases but it persists at a low level so the animals

are continually refreshing their scent. Because mice generally live in territorial so-

cial groups consisting of one dominant male together with one or more breeding

females and their offspring, and variable numbers of non-breeding adult females


Scent marking and social communication 221

(c) Urine scent posts

(a) Water stimulus (b) Intruder urine stimulus

Fig. 11.1. Scent marks of wild house mice. Urine scent marks deposited by a resident

male territory owner on a patch of substrate (15 cm × 15 cm of absorbent paper)

introduced into the territory for 30 min. The patch was first treated with 10µl of

either water (a) or urine from an intruder male (b). The white box indicates location of

the stimulus). Urinary scent marks (white) were visualized under ultraviolet light

using a FluorS scanner (Bio-Rad Laboratories Ltd, Hemel Hempsted, UK). (c) Urine scent

posts (approx 2 cm tall, see white box) on a wooden batten where mice frequently

rested in an infested poultry house.

and subordinate males (reviewed by Barnard et al., 1991), the substrate becomes

smothered in urinary scent marks from all mice that share the same home area

(Hurst, 1989). Not all mice contribute equally to this background scent marking;

the dominant male territory owner, responsible for most defence of the territory,

marks at a much higher rate than other individuals. In situations where males


222 J. L. Hurst

1.2 m

Male mouse

Male mouse

(a)

(b)

Fig. 11.2. Correspondence between activity and scent deposition when an adult male

mouse explored a clean enclosure neighbouring his territory for 15 min. Arrow

indicates the hole through which the male entered from his own territory.

(a) Location sampled three times per second. (b) Urine marks were sampled by

swabbing each 15 cm × 15 cm floor section. The darkness of spots shows the amount

of urine recovered, visualized by immunoassay of the major urinary proteins. The

nestbox (circle) and food station (rectangle) were not swabbed. (Data collected by

Karen Sanders.)

have to defend their territory from other males, dominant male territory owners

deposit large streaks as well as smaller spots of urine wherever they go (Desjardins

et al., 1973; Hurst, 1990), aided by hairs on the end of the adult male prepuce

(Maruniak et al., 1975).

Adult mice excrete a number of species- and sex-specific volatiles in their urine

that are under hormonal control, in addition to a large number of non-specific

volatile metabolites (Novotny et al., 1984, 1990; Schwende et al., 1986; Harvey

et al., 1989). Male-specific signalling volatiles include two sesquiterpenes, (E , E )

α-farnesene and (E ) β -farnesene, which are secreted into the urine by the preputial

glands, and 2-sec-butyl-4,5-dihydrothiazole and 2,3-dehydro-exo-brevicomin, which



are present in the bladder before any additions from accessory glands (Harvey et al.,

1989; Novotny et al., 1990). A large number of volatiles in female urine vary over the

oestrous cycle (Andreolini et al., 1987) and during pregnancy and lactation ( Jemiolo

et al., 1987), although it is not yet known which (if any) of these are chemosignals or

metabolic by-products. One urinary constituent, 2,5-dimethylpyrazine, produced

under adrenal control by group-housed females is known to inhibit reproduction

in other females (Novotny et al., 1986; Jemiolo & Novotny, 1993, 1994).

In addition to these sex-specific volatiles, mouse urine is characterized by the

presence of a high concentration of protein, over 99% of which is contributed

by a group of 18–20 kDa lipocalins known as the major urinary proteins (MUPs).

The lipocalins are synthesized in the liver and secreted into the plasma; they

subsequently pass through the glomerular filter into the urine (Beynon et al., 2001).

Although urine of both sexes contains a substantial quantity of these proteins,

adult male urine typically contains 20–30 mg/ml protein, approximately three

times as much as female urine (Beynon & Hurst, 2003). This sex difference occurs

at puberty when there is an increase in excretion among males (Payne et al., 2001).

These lipocalin proteins have a central cavity that binds lipophilic molecules.

In males, MUPs bind a number of ligands but principally the two male-specific

signalling volatiles 2-sec-butyl-4,5-dihydrothiazole and 2,3-dehydro-exo-brevicomin

(Bacchini et al., 1992; Robertson et al., 1993; Novotny et al., 1999a). Once urine is

deposited as a scent mark, the binding of signalling volatiles to MUPs greatly slows

down their evaporation from the scent mark (Hurst et al., 1998; Robertson et al.,

2001), extending the duration over which volatiles can be detected (Hurst et al.,

1998; Humphries et al., 1999). No MUP ligands have yet been identified in female

urine.

Although volatiles may be detected from a distance of several centimetres or

more, depending on the amount of scent deposited, the response of mice is almost

always to approach and contact the scent mark, investigating it at very close

quarters. Consequently, it is likely that mice gain information from both the

volatile and non-volatile components in scents (Humphries et al., 1999). Recent

evidence suggests that the vomeronasal system only detects scents when these

are pumped to the vomeronasal organ after contact with a stimulus (Luo et al.,

2003).

Advertising territory ownership and competitive ability

The network of scent marks deposited and investigated by all individuals

within the local population provides a mechanism by which animals can advertise

their competitive ability to both competitors and to potential mates in a manner

that makes it difficult, if not impossible, to cheat. This reliable mechanism involves


224 J. L. Hurst

both the spatial and the temporal pattern of competitive scent marking as well as

the quality of an individual’s scent.

Spatial and temporal distribution of scent

Like many other mammals, dominant male house mice advertise their

territory ownership by scent marking throughout their defended area, marking

at a much higher rate than other individuals (Ralls, 1971; Desjardins et al., 1973;

Hurst, 1990). Territory owners continually refresh these scents at a high rate so

that, in frequently used sites, small posts of dried urine can build up like small

stalagmites (Fig. 11.1c). Because only animals that dominate a territory can ensure

that their marks predominate in that area, the spatial pattern and density of scent

marks bearing the owner’s individual identity signature provide physical proof of

an individual’s territory ownership (Gosling, 1982; Hurst, 1993). Further, the spa-

tial and temporal pattern of scent marks from other males indicates the success

with which an owner dominates its scent-marked territory. Only males that de-

fend their territory effectively can ensure that no other males deposit competing

signals that might attract mates (Hurst, 1993; Hurst & Rich, 1999). Conversely,

the presence of any competing signals that are as fresh or fresher than those of

the owner will indicate that the owner is not stopping competitors from deposit-

ing competing marks and, therefore, is not being very successful in dominating

the area, even if the area is suffused with the owner’s scent. Accordingly, domi-

nant male mice rapidly counter-mark if they encounter competing scent signals

from other males in their territory, as well as attacking and excluding from the

territory any competitors that deposit such competing scent marks (Ralls, 1971;

Hurst & Rich, 1999). In house mice, counter-marking consists of a rapid increase

in the rate of urine scent marking in the vicinity of, but not specifically on top

of, a competitor male’s scent (Fig. 11.1b; Hurst, 1989; Humphries et al., 1999). In

other species, counter-marking may take the form of over-marking the competi-

tor’sscent to prove which scent was deposited most recently ( Johnston et al., 1997;

Johnston, 1999; Ch. 16). House mice do not attempt to over-mark the scent marks

of a competitor, perhaps because their urinary scent marks are scattered so widely

in numerous spots and streaks. Instead, mice assess the difference in age of nearby

scent marks from competing males to determine which male’sscent was deposited

most recently (Rich & Hurst, 1999). Because scent marks remain in the environ-

ment and are long lived (see p. 232), the spatial and temporal pattern of scents

from different individuals provides a continuous record of any challenges for dom-

inance over the area and, crucially, the outcome of those challenges. This record is

available for investigation by any other animal in the area. Animals do not need to

witness, or to eavesdrop on, individual challenges for dominance while these are

occurring, because the outcome will be readily apparent from scent marks for as



long as the marks retain the individual signatures of the depositors (but see Ch. 2

for a discussion of extra information available from interactions). Scent marks

provide the ‘minutes’ of a meeting that are made public to all interested parties,

although they may not include a transcript of all of the information involved in

the detailed arguments expressed during the meeting.

Experimental manipulation of the scent marks within male mouse territories

has shown that this information is used by third parties to assess the competitive

ability of territory owners: (a) by females when selecting high-quality mates, and

(b) by other males when deciding whether to avoid a territory owner or to challenge

the owner for dominance themselves. These are discussed separately below.

Assessment by females

Although female house mice generally nest and raise their offspring

within one male’s territory, they often visit or range over several neighbouring

male territories (Hurst, 1987) and extraterritorial matings occur frequently (e.g.

43% of all matings in large captive populations occurred when a female travelled

to, and mated with, a male owning a nearby territory (Potts et al., 1991)). Inter-

estingly, among wild house mice, resident male territory owners appear to show

little or no discrimination against offspring sired by other males that are reared

within their territory (Hurst & Barnard, 1992). Therefore, females usually have a

choice between several high-quality male territory owners as potential mates in

the local population, regardless of where they choose to nest, because there is

little risk to their offspring.

Most simply, females can compare the scent marks left by two competing males

to assess which male’s scent was deposited most recently. Since only a male that

is successfully preventing other males from depositing competing scents can en-

sure that his marks are the freshest in that location, this male must have been the

winner of the conflict. Subsequently given a choice between the two signalling

males, females generally prefer the owner of the most recently deposited scent

(e.g. Johnston et al., 1997; Fig.11.3a). However, females can also use the presence

of scent marks from any males in an owner’s territory to make a much more

general assessment of each territory owner’s competitive ability when choosing

between territory owners. By manipulating the scent marks in equivalent male

territories, Rich & Hurst (1998) showed that female house mice prefer the owners

of exclusively scent-marked territories (those containing no scents from competi-

tor males) over neighbouring males whose territories contain some competing

counter-marks from an intruder male (Fig. 11.3b). After exploring the males’ ter-

ritories, females spent more time sniffing and chewing a barrier to gain access

to the owner of the exclusively scent-marked territory and were more affiliative

and invited attempted mounts from this territory owner when they were allowed


226 J. L. Hurst

Fig. 11.3. Effects of counter-marking between males on female preferences in

different types of test. (a) Female encounters a single scent mark from male A that

is deposited on top of scent from male B (e.g. Johnston et al., 1997). (b) Females

explore neighbouring male scent-marked territories, one of which contains

patches of intruder (i) scent counter-marking the owners scent (Rich & Hurst, 1998).

(c) Both neighbouring male territories contain patches of intruder (i) scent (Rich &

Hurst, 1999).

to interact. When given a choice between two males whose territories both con-

tained some intruder scent marks, but one owner had counter-marked the intruder

scents while some of the other owner’s scent had been counter-marked by the in-

truder, females preferred the owner that had counter-marked the intruder’s scent

(Fig. 11.3c; Rich & Hurst, 1999). In both cases, intruder scent marks came from

unfamiliar males; therefore, females were not making a simple choice between

two interacting males. Neither were females simply responding to the freshest

scents encountered because both territories contained fresh scent marks from the



territory owners. Female preferences resulted from their assessment of the pattern

of male scent marks and counter-marks as an indicator of each male’scompetitive

ability. Scent marks were manipulated in such a way that only the females were

exposed to the manipulation, while the territory-owning males were temporar-

ily removed from their territories, so responses were not a result of any changes

in male behaviour in response to the intruder scent marks introduced into each

male’s territory.

Female mice appear to distinguish the most recent scent (i.e. the counter-mark)

by the age difference between the male’s scents (Rich & Hurst, 1999). When both

territory owners’ and intruders’ scents were of very similar age, females failed to

show a preference in favour of an owner that had counter-marked intruder scents.

This could be because females were unable to discriminate between scent marks

and counter-marks without a substantial age difference in the scents (24 hours

in these experiments), or because the similar age of the scents indicated that the

competition between the males had yet to be resolved.

Males can thus gain a reproductive advantage from scent marking their territo-

ries and from counter-marking the scents of any competitors to ensure that their

own scent marks are those most recently deposited. Although it is often assumed

that the very high rates of marking at borders between neighbouring territory

owners are signals to warn neighbours to keep out, the main function of frequent

scent marking at shared borders may be the need of both territory owners to en-

sure that their own scent is as fresh as their neighbour’s wherever their scents are

in close proximity. This would require both animals continually to refresh their

scent at a shared border as an advertisement to females in the locality (including

females resident in the male’s territory and those living in neighbouring areas

(Hurst & Rich, 1999)). Male territory owners can thus compete with each other

simply through their scent-marking behaviour because competitive scent marks

are used by females when selecting a mate.

Assessment by males

Third-party competitors also appear to use the record of competitive scent-

mark signals to identify and avoid challenging owners that are defending their

territory effectively against other males. Unfamiliar intruders use the scent marks

deposited around a territory to identify the territory owner and are much less

likely to challenge a male whose individual scent signature matches the local

scent marks than a male whose scent does not match (Gosling & McKay, 1990).

Adding a small drop of fresh urine from the territory owner onto one of the own-

ers’ scent marking posts increases the frequency with which intruders and resi-

dent subordinates spontaneously flee when they encounter the owner, without

any attack or pursuit. In contrast, addition of urine from a neighbouring territory

owner reduces their evasion and increases challenges against the territory owner,


228 J. L. Hurst

regardless of their own previous experience of the high competitive ability of

the territory owner during direct interactions (Hurst, 1993; Hurst & Rich, 1999).

Thus, territory owners also appear to gain a strong advantage in competitive in-

teractions with other males through broadcasting their scent around the territory

and through advertising their ability to overcome challenges from other males,

reducing their need to invest in direct aggression. Other third-party males gain

by avoiding challenging a male that is successfully defending his territory against

other competitors but will also rapidly detect when the owner is struggling to

maintain dominance (for example, if his competitive ability is reduced by ageing,

injury or disease). This may be particularly important to resident subordinates,

which are likely to be highly familiar with the greater competitive ability of the

dominant male and reluctant to challenge if this might result in their exclusion

from the territory.

Deposition patterns and scent age

In contrast to isolated or subordinate (non-competitive) mice, males that

are advertising territory ownership or competing to establish a territory change

their pattern of scent marking by scattering their urine in a much larger number

of streaks and small spots. While this helps to ensure that their scent is distributed

throughout the territory, numerous scent marks are deposited close together in

the same local area (see Fig. 11.1b). When counter-marking another male’s scent,

mice do not attempt to deposit a bigger scent mark than that of the competitor,

which would contain a greater intensity of volatile signalling molecules. Instead,

they deposit many small marks in the vicinity, returning repeatedly to add more

scent marks usually over a period of several hours (Humphries et al., 1999). By drib-

bling out their urine rather than depositing it all in one go, they are maximizing

the freshness of their scent marks by increasing the rate of replenishment (Hurst

et al., 2001a). Thus, each time they deposit a new scent mark, they are increas-

ing the age difference between their own scent and that of the competitor, while

volatiles in their own fresh scent attract the attention of others to the aged scents

of a competitor. Notably, males counter-mark both fresh and aged scents from

competitors but deposit most marks near to the aged competitor scent where the

contrast will be greatest (Humphries et al., 1999). Most male signalling volatiles

are lost from scents within a few hours (Hurst et al., 1998; Humphries et al., 1999;

Robertson et al., 2001), but non-volatile components of scent marks continue to be

detected for at least seven days if males are aware of the presence of scent marks

in the area (Humphries et al., 1999, 2001).

Both volatile and non-volatile components of a scent mark are likely to be in-

volved in providing a reliable signal of scent-mark age. While volatile components



will be lost as a scent mark ages, the intensity of a volatile signal at any point

in time will depend on the amount initially deposited as well as the time since

deposition. Receivers will not be able to assess the age of the mark from a volatile

signal alone without knowing the amount that was deposited (Hurst et al., 2001b).

In contrast, non-volatile components are not lost through time and may provide a

measure of the amount of scent deposited. As yet, we do not know the molecular

mechanism used to assess scent-mark age, but MUPs that bind and slowly release

volatile ligands in mouse urine have the capacity to provide a very reliable indi-

cator of scent-mark age. Because each protein molecule can only bind one ligand

molecule, and ligands are slowly released and evaporate from the scent mark,

the proportion of protein molecules that contain ligands will decrease with time

since deposition. Making one component the ligand of the other defines implic-

itly the relationship between them. By contrast, the ratio between two unrelated

volatile components that have different rates of evaporation requires that the re-

ceiving animal knows the ratio between them at the time of deposition. Although

MUPs are not odorants, these non-volatile proteins appear to be detected through

the vomeronasal system (Brennan et al., 1999; Krieger et al., 1999). Direct contact

with the scent source appears to be essential for activation of this system, sug-

gesting that non-volatile components like MUPs are necessary to deliver volatile

pheromones to the vomeronasal organ (Luo et al., 2003). The release of volatiles

from MUPs also plays an important function in alerting mice to the presence of a

scent mark (Hurst et al., 1998; Humphries et al., 1999).

Advertising subordinate status

Females are attracted by sex-specific volatiles in the urine of adult

male mice ( Jemiolo et al., 1985, 1991), which also act as reproductive priming

pheromones to stimulate female oestrus cycling (see below). These same male sig-

nalling volatiles elicit aggression from other competitive males (Novotny et al.,

1985), or avoidance by subordinates ( Jones & Nowell, 1989; Novotny et al., 1990;

Gosling et al., 1996; Mucignat-Caretta et al., 1998). However, subordinate males

that live within the territory of another male and are defeated repeatedly by the

dominant territory owner reduce their production of these male-specific volatiles

( Jones & Nowell, 1989; Harvey et al., 1989); their preputial glands are smaller than

those of dominant males (Hucklebridge et al., 1972; Bronson & Marsden, 1973) and

they are much less likely to initiate competitive interactions (Crowcroft & Rowe,

1963; Hurst, 1987). As a consequence of these changes in scent quality, subordinate

male urine is no longer attractive to females (Bronson & Caroom, 1971; Jones &

Nowell, 1974; Jemiolo et al., 1991) and females will not mate with subordinate

males (Wolff, 1980; Hurst, 1987; Potts et al., 1991). In compensation, subordinate


230 J. L. Hurst

male odours elicit less aggression from other males (Mugford & Nowell, 1970; Jones

& Nowell, 1973, 1975; Novotny et al., 1985). Subordinate male mice also show a dra-

matic and immediate reduction in scent-marking behaviour, although this is not

completely suppressed and they continue to deposit scent marks around their

home area in larger spots and pools (Desjardins et al., 1973; Sandnabba, 1986).

Because their urinary scent differs in quality from that of dominant males, these

scent marks advertise their subordinate status within the territory to all other

animals in the area, including females.

Why should subordinates advertise their low quality in such a public manner?

Experimental manipulations of these substrate scents indicate that they are criti-

cal in determining tolerance of the subordinate by other resident males. Male mice

generally attempt to exclude other adult males from their scent-marked territo-

ries (relatives or non-relatives) and they are highly aggressive towards unfamiliar

mice or familiar neighbours that intrude into the territory even if intruders are

of subordinate status (Barnard et al., 1991). However, complete exclusion can be

extremely difficult to achieve in complex habitats where persistent males can hide

(Crowcroft, 1966; Poole & Morgan, 1976). Familiar males living in the same terri-

tory establish a social structure in which one male becomes dominant and main-

tains dominance over familiar subordinates through brief attacks and aggressive

postures, rather than attempting to evict the subordinates from the territory. If

these familiar males are housed in separate cages but their soiled cage substrate is

regularly mixed to maintain their contact with group scent cues, males continue

to be relatively tolerant of each other. However, if a familiar subordinate male

is suddenly prevented from contributing fresh scent to the mixed group-marked

substrate, although the subordinate itself continues to encounter group substrate

scents as if it were still a group member, within 24–48 hours both the resident

dominant and other subordinate males in the group start to investigate and to

attack and chase this male as if he was no longer a tolerated group member (Hurst

et al., 1993). In contrast, scent marks deposited to compete with the signals of the

dominant territory owner, for example on territorial scent-marking posts, induce

attack against the familiar subordinate (Hurst, 1993).

Competitive pressure from dominant male territory owners thus appears to

force subordinate males both to change the quality of their urinary scent and to

deposit urinary scent marks that advertise their subordinate status to any animals

using or visiting the area. Because the scent marks remain in the environment once

deposited, and carry the subordinate’s individual identity signature (see below), a

subordinate would be unable to cheat by altering the quality of his scent during

direct interactions with females. In order to be tolerated and allowed to remain

within another male’s territory, subordinates appear to be forced to broadcast

honest signals of their low competitive ability at the cost of their reproductive



success, because their scent marks will be encountered by females. The outcome

of this is to reduce considerably the risk that subordinate males might compete

for females through sneaky matings (mating with subordinate males is very rarely

observed in seminaturalistic studies of mouse populations), making it safe for the

territory owner to tolerate subordinates within the territory that contribute to

group substrate scents. Notably, experiments have revealed that animals of low

competitive ability only suppress the production of competitive male scents when

they live in very close proximity to a dominant individual; they do not suppress

scent signals simply in response to defeat by another male. Jones and Nowell (1989)

confirmed that, if males are repeatedly defeated by a higher quality competitor

and are kept in continuous olfactory contact with the scent of their victor (as if

they lived within his territory), the defeated subordinate male’s scent loses the

aversive effect on other males caused by male signalling volatiles. However, if

males experience the same frequency of defeat but are housed in separate cages

from their victor (as if they could escape to another territory), the defeated male’s

scent retains the high levels of male signalling volatiles that are aversive to other

males. Not surprisingly, males do not advertise their low quality unless they are

forced to do so by the constant threat of attack and displacement from higher

quality competitors.

Female reproductive priming

Female reproductive physiology is strongly influenced by a number of

reproductive priming pheromones in the urine of male or female mice (reviewed

by Brown, 1985b; Novotny et al., 1999a). Volatile priming pheromones in the urine

of adult male mice have stimulatory effects on female physiology, accelerating

puberty in young females (Vandenberg, 1969; Novotny et al., 1999b) and stimulat-

ing oestrous cycling (Whitten, 1956; Jemiolo et al., 1986), while urinary odours

from pregnant or lactating females have similar though not identical effects

(Drickamer & Hoover, 1979; Hoover & Drickamer, 1979). The stimulatory effects of

urine from pregnant or lactating females may reflect the preference of house mice

for communal nesting, because females raise more offspring when cooperating

with another female than they can when breeding alone (Konig, 1994a). In contrast,

nestling survival is greatly reduced in overcrowded nest sites, particularly among

females of low social status (Southwick, 1955; Hurst, 1987). Accordingly, when

non-breeding females live in groups with several other non-breeding females, or

have frequent contact with the urine of other non-breeding females, they produce

a priming pheromone in their own urine that inhibits oestrous cycling in adults

(Champlin, 1971) and delays puberty in prepubertal females (Drickamer, 1977;

Jemiolo & Novotny, 1994). However, females of high social status are not affected


232 J. L. Hurst

and continue to breed (Lloyd & Christian, 1969). Lastly, if a female encounters the

scent of a novel male within four days of mating, implantation fails and she will

abort if she is not protected by continued exposure to urine from the familiar

stud male, a phenomenon known as the Bruce effect (Bruce, 1959; Brennan, 1999).

This is likely to provide females with the opportunity to mate with a new territory

owner if their previous mating partner is displaced. Since male territory owners

do not appear to discriminate against the offspring of other males born within

their territory (Hurst & Barnard, 1992), this may be a tactic to increase offspring

quality.

Although priming pheromones are volatile, these are detected through the

vomeronasal system and animals must make contact with the scent source to

allow chemical stimuli to be pumped to the vomeronasal organ (Brown, 1985b;

Luo et al., 2003). Under natural conditions, females are surrounded by urinary scent

marks from all individuals using the same sites, exposing them to both stimulatory

and inhibitory priming pheromones. Because these urinary scents are very widely

distributed and females do not appear deliberately to control their exposure to

these cues (Hurst & Nevison, 1994), this network of scent signals from all animals

using the same area appears to provide a mechanism for females to adjust their

own reproductive physiology appropriately, according to the current local social

conditions and to the individual’s own age and social status.

Individual scent signatures

Scent marks are deposited in the environment to provide information in

the absence of the signaller (unlike other types of signal), so it is essential that they

provide stable and persistent information about the donor’s individual identity.

Ideally, individuality scents should be ‘hard-coded’ in the individual’s genome,

exhibit a high degree of individual polymorphism to uniquely identify the donor

and be expressed by all individuals regardless of social status or sex (Beynon et

al., 2001). Attention has focused largely on the volatile components of scents as

sources of individuality signatures, particularly those associated with the highly

polymorphic major histocompatibility complex (MHC) odortypes, although many

other genetic loci also influence individual differences in urinary volatile profiles

(Boyse et al., 1987; Beauchamp et al., 1990; Eggert et al., 1996). The MHC encodes

for glycoproteins involved in individual (self versus non-self) recognition at the

cellular level but has also been shown to affect the volatile scent signals produced

by animals such as mice, rats, fish and humans (reviewed by Jordan & Bruford,

1998; Singh, 2001; see also Olsen et al., 1998; Reusch et al., 2001; Jacob et al., 2002).

Laboratory studies using MHC congenic strains of mice and rats have confirmed

that rodents are able to discriminate differences in urinary odours from donors



that differ genetically at alleles within the MHC region, even when donors differ

only at a single MHC locus (Yamazaki et al., 1999; Singh, 2001; Carroll et al., 2002).

Although the molecular basis of MHC-associated odours is not known, it appears to

involve a complex mixture of volatile metabolites bound and released by urinary

proteins (Singer et al., 1993, 1997). One hypothesis is that soluble fragments of MHC

class I and class II molecules in urine differentially bind volatile metabolites in

the antigen-binding groove once the peptide normally bound in this groove is lost

(Singh, 2001). Alternatively, MHC haplotype may affect the volatile metabolites

that are released into the urine (Singer et al., 1997; Yamazaki et al., 1999). These

MHC-dependent volatiles might then be bound and released by the MUPs, which

are present at protein concentrations up to a million times higher than MHC

class I molecule fragments and possess a large flexible binding pocket for small

lipophilic (and thus potentially volatile) molecules (Beynon et al., 2001).

Although MHC polymorphism results in differences in scents that are discrim-

inable by mice, these volatile signals appear to be easily disrupted by environ-

mental factors that affect an individual’s metabolite profile, such as changes in

food type, bacterial gut flora or social status (reviewed by Brown, 1995; Nevison

et al., 2000). This presents a problem for an individual recognition signature, sug-

gesting that MHC-associated odours may not provide sufficient stability or per-

sistence to act as individuality signals in scent marks. Indeed, mice do not use

MHC-associated odours to discriminate their own scent marks from those of other

males, despite their clear ability to detect differences in their own MHC type and

those of other individuals ( J. L. Hurst, unpublished data). Instead, they use the

different patterns of MUPs that individual mice express in their urine (Hurst et al.,

2001b).

MUPs are coded by a multigene family on chromosome 4 and are expressed

at high concentration by adult house mice of both sexes, although males invest

more than females in both scent marking and MUP production (Beynon et al.,

2001; Payne et al., 2001; Beynon & Hurst, 2003). MUPs exhibit a very high level

of genetic polymorphism and individual mice express a combination of MUPs

(typically at least 7–12) such that the combinatorial diversity of individual MUP

profiles among wild mice may be as great as for MHC (Robertson et al., 1997;

Beynon et al., 2001, 2002; Payne et al., 2001). These individual-specific patterns of

urinary MUPs in the scent marks of wild mice appear to be essential in allowing

outbred wild mice to distinguish another individual’s scent marks from their

own, regardless of differences at many other genetic loci such as MHC (Hurst et al.,

2001b). If a male territory owner encounters a competing scent mark from another

male in his territory, he normally investigates closely and then counter-marks

and spends time in the vicinity of the scent mark. However, males only recognize

an intruder’s scent mark if it carries a different MUP pattern from their own.


234 J. L. Hurst

Scent marks with their own MUP pattern draw initial investigation but no further

response (Hurst et al., 2001b; J. L. Hurst, unpublished data). This is not because

males avoid competing with a relative (i.e. kin discrimination) as they respond

strongly to urinary scents of different MUP type whether from a close relative

or an unrelated male. In contrast, if their own urinary MUP profile is altered by

adding a recombinant MUP to their urine, males counter-mark as if the scent was

from an intruder (Hurst et al., 2001b). There is a high degree of individual variability

in MUP patterns expressed by males captured from the same population (Payne

et al., 2001) and MUPs show little or no degradation in scent marks (Hurst et al.,

2001a). These patterns, therefore, provide a stable and persistent individual scent

signature that is hard-coded in the individual’s genome and remains constant

throughout the individual’s lifetime.

Kin and group member recongnition

Inherited scent signatures are also important in allowing animals to rec-

ognize whether others are likely to be close relatives. Many genetic differences

appear to contribute to inherited scents used for kin recognition, including MHC-

associated odours. For example, female mice prefer to rear their offspring com-

munally with close relatives rather than with unrelated females, and offspring

survival is greater when cooperating with a familiar sister (Konig, 1994b). When

unfamiliar female mice are mixed together in seminatural populations, they ap-

pear to recognize the similar scents of other females of the same MHC type as

themselves and are more likely to share nests with these females (Manning et al.,

1992).

Recognition of close relatives is particularly important to avoid inbreeding.

There is considerable evidence that MHC type affects female preference between

male scents, with females generally preferring the scents of males of different MHC

type to themselves or their parents (e.g. Egid & Brown, 1989; Penn & Potts, 1998).

By crossing wild mice with laboratory strains to create wild-type mice of known

homozygous MHC type, Potts et al. (1991) confirmed that females in seminatural

populations showed a significant preference for MHC-disassortative mating. Inter-

estingly, females showed no MHC bias when mating with the owner of the territory

in which they nested (suggesting that MHC type did not influence territory prefer-

ence despite the genetic information in the owner’s scent marks). However, when

females went outside their territory to mate, they preferred owners of neighbour-

ing territories that had a different MHC type to their own familiar MHC-associated

odours.

In mice, the mechanism of kin recognition appears to be largely through im-

printing on scents experienced during early life rather than on cues inherited by



the female herself (D’Udine & Alleva, 1983). Mating preferences can be reversed

by cross-fostering. Females fostered onto a different strain will later avoid mating

with males related to their foster parents rather than with those related to them-

selves (e.g. Penn & Potts, 1998; though see Eklund, 1997). Because relatives will

carry a much greater range of alleles than the female herself, imprinting on the

scents of relatives experienced in their early environment may be more effective

than phenotype matching to self for recognizing potential relatives or those from

a similar genetic background.

In addition to inherited scents, mice acquire scents on their bodies from other

group members that influence recognition when mice interact (Aldhous, 1989).

All group members are likely to become tainted with the scents of the resident

territory owner from the sticky scent marks deposited throughout the area. This

acquired group scent may make an important contribution to the ability of mice

to recognize their own group members regardless of their inherited scents (e.g.

Hurst & Barnard, 1992, 1995).

Scent marking as broadcast signals

Animals spontaneously scent mark their territories in the absence of in-

teraction with others, although scent marking is usually significantly enhanced

by competitive and sexual interaction with others or with their scent marks. In

addition to the chemical information in an individual’s scent, the spatial and

temporal patterning of scent marks from all individuals in a locality provides

information about their social and genetic relationships. Scent marks are partic-

ularly suitable for network communication between many individuals. By their

very nature, they are long lasting and readily available for inspection by any indi-

viduals that visit a scent-marked site. Conversely, scent marks are not appropriate

for private communication (see Ch. 3) unless individual access to the scent marked

site is restricted.

There are two ways in which scent marking might be used in a signalling net-

work. First, scent marks may be deposited as broadcast signals, designed to com-

municate information to all other animals in the area. Alternatively, scents may

be deposited as signals to specific individuals, with third parties making use of

information in scent marks and counter-marks by eavesdropping on the commu-

nication between others. Eavesdropping has been defined as ‘extracting informa-

tion from signalling interactions between others’ (McGregor & Dabelsteen, 1996;

McGregor & Peake, 2000; Ch. 2). The implication here is that signals are designed

to provide information to one or more interacting individuals (e.g. during the in-

teraction between two competitors), not to provide information to eavesdroppers.

However, eavesdroppers make use of this information to their own advantage. It


236 J. L. Hurst

is, therefore, important to establish who gains an advantage from the use of these

signals.

With respect to competitive counter-marking by territory owners in response

to aggressive challenges and scent marking from other males, the main fitness ad-

vantage to signallers appears to be the response of females to these scent marks,

although females may be viewed as ‘third parties’ responding to signalling inter-

actions between males (see Ch. 7 for parallels with bird song). Evidence from our

scent-manipulation studies in house mice indicates that these scent signals have

highly significant effects on female preferences, in accordance with our hypothe-

sis that this is a very reliable way to assess the competitive ability of different po-

tential mates. Similarly, territorial scent marking and the counter-marking of any

intruder scents by the owner increases avoidance responses and decreases compet-

itive challenges from third-party males (i.e. males that are not the owners of either

the scent marks or counter-marks). As this affects the responses of all males in the

vicinity, this is likely to have a big impact on the ease with which males defend

their territories. Therefore, a successful territory owner gains clear advantages

from advertising his territory ownership and competitive ability widely, includ-

ing his ability to overcome the challenges of competitors. While increased scent

marking to counter-mark a competitor’s scent might, at first sight, appear to be a

signalling interaction between two competitors, the main selective advantage to

successful competitors comes from broadcasting this information to all animals in

the vicinity. There is no evidence that counter-marking the scent of an aggressive

competitor reduces challenges from the competitor itself without direct aggres-

sion, although once defeated, competitors will show a generalized avoidance of

competitor scents (e.g. Jones & Nowell, 1989; Hurst et al., 1997). McGregor & Peake

(2000) pointed out that there seem to be few demonstrated advantages to signallers

of communicating in the social environment of a network. However, competitive

scent marking provides an excellent example of the advantages that successful

territory owners can gain from depositing competitive scent marks within the

network of signals from other males, because this provides a mechanism for the

reliable advertisement of their own competitive ability. These scent marks are

clearly broadcast signals, designed to communicate information to all other ani-

mals in the area. As such, the concept of eavesdropping on signals aimed at others

does not seem to be appropriate.

The deposition of scent signals that advertise an individual’s subordinate sta-

tus appears to be enforced by the resident territory owner and other resident

subordinate males, and thus subordinate scent marking might be viewed as a

signal aimed principally at local competitors to reduce aggression against the

subordinate. However, the selective advantage to local competitors of reducing

their aggression is that such scent marks broadcast the subordinate status and



identity of a competitor to females. Subordinates that signal their subordinate

status through their body scent but not through substrate scent marks are not

tolerated. From an evolutionary viewpoint, it again seems more appropriate to

view scent marks deposited to advertise subordinate status as broadcast signals,

aimed at publicising this information to others despite the immediate reduction

in reproductive opportunities to the signaller from doing so. This at least allows

unsuccessful males to remain in a suitable habitat with the potential to become

a successful territory owner, and gain reproductive success, in the future.

Summary

In conclusion, since scent marks persist in the environment and cannot be

directed towards specific recipients (unlike most visual and acoustic signals), scent

marks are only likely to be used as broadcast signals and are used in social contexts

where the signaller can gain an advantage from communicating information to

a public audience.

Acknowledgements

I am grateful to Rob Beynon for providing Fig. 11.2, for invaluable discussion and for

comments on the manuscript. Research carried out in my laboratory was supported by grants

from the Biotechnology and Biological Sciences Research Council.

References

Aldhous, P. 1989. The effects of individual cross-fostering on the development of

intrasexual kin discrimination in male laboratory mice. Animal Behaviour, 37,

741–750.

Andreolini, F., Jemiolo, B. & Novotny, M. 1987. Dynamics of excretion of urinary

chemosignals in the house mouse (Mus musculus) during the natural estrous cycle.

Experientia, 43, 998–1002.

Bacchini, A., Gaetani, E. & Cavaggioni, A. 1992. Pheromone binding proteins of the

mouse, Mus musculus. Experientia, 48, 419–421.

Barnard, C. J., Hurst, J. L. & Aldhous, P. 1991. Of mice and kin: the functional

significance of kin bias in social behaviour. Biological Reviews of the Cambridge

Philosophical Society, 66, 379–430.

Beauchamp, G. K., Yamazaki, K., Duncan, H., Bard, J. & Boyse, E. A. 1990. Genetic

determination of individual mouse odour. In: Chemical Signals in Vertebrates 5, ed.

D. W. Macdonald, S. Natyncuk & D. Muller-Schwarze. Oxford: Oxford University

Press, pp. 244–255.

Beynon, R. J. & Hurst, J. L. 2003. Multiple roles of major urinary proteins in the house

mouse, Mus domesticus. Biochemical Society Transactions, 31, 142–146.


238 J. L. Hurst

Beynon, R. J., Hurst, J. L., Gaskell, S. J. et al. 2001. Mice, MUPs and myths:

structure-function relationships of the major urinary proteins. In: Chemical Signals

in Vertebrates, ed. A. Marchelewska-Koj, D. Muller-Schwarze & J. Lepri. New York:

Plenum Press, pp. 149–156.

Beynon, R. J., Veggerby, C., Payne, C. E. et al. 2002. Polymorphism in major urinary

proteins: molecular heterogeneity in a wild mouse population. Journal of Chemical

Ecology, 28, 1429–1446.

Boyse, E. A., Beauchamp, G. K. & Yamazaki, K. 1987. The genetics of body scent. Trends

in Genetics, 3, 97–102.

Brennan, P. A. 1999. Bruce effect. In: Encyclopedia of Reproduction, ed. E. Knobil & J. Neill.

San Diego, CA: Academic Press, pp. 433–438.

2001. The vomeronasal system. Cellular and Molecular Life Science, 58, 546–555.

Brennan, P. A., Schellinck, H. M. & Keverne, E. B. 1999. Patterns of expression of the

immediate-early gene egr-1 in the accessory olfactory bulb of female mice exposed

to pheromonal constituents of male urine. Neuroscience, 90, 1463–1470.

Bronson, F. H. & Caroom, D. 1971. Preputial gland of the male mouse; attractant

function. Journal of Reproductive Fertility, 25, 279–282.

Bronson, F. H. & Marsden, H. M. 1973. The preputial gland as an indicator of social

dominance in male mice. Behavioral Biology, 9.

Brown, R. E. 1985a. The rodents II: suborder Myomorpha. In: Social Odours in Mammals,

ed. R. E. Brown & D. W. Macdonald. Oxford: Clarendon Press, pp. 345–457.

1985b. The rodents I: effects of odours on reproductive physiology (primer effects).

In: Social Odours in Mammals, ed. R. E. Brown & D. W. Macdonald. Oxford:

Clarendon Press, pp. 245–344.

1995. What is the role of the immune system in determining individually distinct

body odours? International Journal of Immunopharmacology, 17, 655–661.

Bruce, H. M. 1959. An exteroceptive block to pregnancy in the mouse. Nature, 184, 105.

Carroll, L. S., Penn, D. J. & Potts, W. K. 2002. Discrimination of MHC-derived odors by

untrained mice is consistent with divergence in peptide-binding region residues.


Champlin, A. K. 1971. Suppression of oestrus in grouped mice: the effects of various

densities and the possible nature of the stimulus. Journal of Reproductive Fertility,

27, 233–241.

Crowcroft, P. 1966. Mice All Over. London: Foulis.

Crowcroft, P. & Rowe, F. P. 1963. Social organization and territorial behaviour in the

wild house mouse (Mus musculus L.). Proceedings of the Zoological Society of London,

140, 517–531.

Desjardins, C., Maruniak, J. A. & Bronson, F. H. 1973. Social rank in the house mouse:

differentiation revealed by ultraviolet visualisation of urinary marking patterns.

Science, 182, 939–941.

Drickamer, L. C. 1977. Delay of sexual maturation in female house mice by exposure

to grouped females or urine from grouped females. Journal of Reproductive Fertility,

51, 77–81.

Drickamer, L. C. & Hoover, J. E. 1979. Effects of urine from pregnant and lactating

female house mice on sexual maturation of juvenile females. Developmental

Psychobiology, 12, 545–551.



D’Udine, B. & Alleva, E. 1983. Early experience and sexual preferences in rodents. In:

Mate Choice, ed. P. Bateson. Cambridge, UK: Cambridge University Press,

pp. 311–327.

Eggert, F., Holler, C., Luszyk, D., Muller-Ruchholtz, W. & Ferstl, R. 1996.

MHC-associated and MHC-independent urinary chemosignals in mice. Physiology

and Behavior, 59, 57–62.

Egid, K. & Brown, J. L. 1989. The major histocompatibility complex and female mating

preferences in mice. Animal Behaviour, 38, 548–550.

Eklund, A. 1997. The major histocompatibility complex and mating preferences in

wild house mice. Behavioral Ecology, 8, 630–634.

Gosling, L. M. 1982. A reassessment of the function of scent marking in territories.


Gosling, L. M. & McKay, H. V. 1990. Competitor assessment by scent matching: an

experimental test. Behavioral Ecology and Sociobiology, 26, 415–420.

Gosling, L. M., Atkinson, N. W., Dunn, S. & Collins, S. A. 1996. The response of

subordinate male mice to scent marks varies in relation to their own competitive

ability. Animal Behaviour, 52, 1185–1191.

Harvey, S., Jemiolo, B. & Novotny, M. 1989. Pattern of volatile compounds in dominant

and subordinate male-mouse urine. Journal of Chemical Ecology, 15, 2061–2072.

Hoover, J. E. & Drickamer, L. C. 1979. Effects of urine from pregnant and lactating

female house mice on oestrous cycles of adult females. Journal of Reproductive

Fertility, 55, 297–301.

Hucklebridge, F. H., Nowell, N. W. & Wouters, A. 1972. A relationship between social

experience and preputial gland function in the albino mouse. Journal of

Endocrinology, 55, 449–450.

Humphries, R. E., Robertson, D. H. L., Beynon, R. J. & Hurst, J. L. 1999. Unravelling the

chemical basis of competitive scent marking in house mice. Animal Behaviour, 58,

1177–1190.

Humphries, R. E., Robertson, D. H. L., Nevison, C. M., Beynon, R. J. & Hurst, J. L. 2001.

The role of urinary proteins and volatiles in competitive scent marking among

male house mice. In: Chemical Signals in Vertebrates, ed. A. Marchelewska-Koj, D.

Muller-Schwarze & J. Lepri. New York: Plenum Press, pp. 353–360.

Hurst, J. L. 1987. Behavioural variation in wild house mice Mus domesticus Rutty: a

quantitative assessment of female social organization. Animal Behaviour, 35,

1846–1857.

1989. The complex network of olfactory communication in populations of wild

house mice Mus domesticus Rutty: urine marking and investigation within family

groups. Animal Behaviour, 37, 705–725.

1990. Urine marking in populations of wild house mice Mus domesticus Rutty. 1.

Communication Between Males. Animal Behaviour, 40, 209–222.

1993. The priming effects of urine substrate marks on interactions between male

house mice, Mus musculus domesticus Schwarz and Schwarz. Animal Behaviour, 45,

55–81.

Hurst, J. L. & Barnard, C. J. 1992. Kinship and social behavior in wild house mice:

effects of social group membership and relatedness on the responses of dominant

males toward juveniles. Behavioral Ecology, 3, 196–206.


240 J. L. Hurst

1995. Kinship and social tolerance among female and juvenile wild house mice: kin

bias but not kin discrimination. Behavioral Ecology and Sociobiology, 36, 333–342.

Hurst, J. L. & Nevison, C. M. 1994. Do female house mice, Mus domesticus, regulate their

exposure to reproductive priming pheromones. Animal Behaviour, 48, 945–959.

Hurst, J. L. & Rich, T. J. 1999. Scent marks as competitive signals of mate quality. In:

Advances in Chemical Communication in Vertebrates, ed. R. E. Johnson, D.

Muller-Schwarze & P. W. Sorensen. New York: Plenum Press, pp. 209–226.

Hurst, J. L., Fang, J. M. & Barnard, C. J. 1993. The role of substrate odours in

maintaining social tolerance between male house mice, Mus musculus domesticus.


Hurst, J. L., Gray, S. J., Davey, P. et al. 1997. Social interaction alters attraction to

competitor’s odour in the mouse Mus spretus Lataste. Animal Behaviour, 54, 941–953.

Hurst, J. L., Robertson, D. H. L., Tolladay, U. & Beynon, R. J. 1998. Proteins in urine scent

marks of male house mice extend the longevity of olfactory signals. Animal

Behaviour, 55, 1289–1297.

Hurst, J. L., Beynon, R. J., Humphries, R. E. et al. 2001a. Information in scent signals of

competitive social status: the interface between behaviour and chemistry. In:

Chemical Signals in Vertebrates, ed. A. Marchelewska-Koj, D. Muller-Schwarze & J.

Lepri. New York: Plenum Press, pp. 43–52.

Hurst, J. L., Payne, C. E., Nevison, C. M. et al. 2001b. Individual recognition in mice

mediated by major urinary proteins. Nature, 414, 631–634.

Jacob, S., McClintock, M. K., Zelano, B. & Ober, C. 2002. Paternally inherited HLA alleles

are associated with women’s choice of male odour. Nature Genetics, 30, 175–179.

Jemiolo, B. & Novotny, M. 1993. Long-term effect of a urinary chemosignal on

reproductive fitness in female mice. Biology of Reproduction, 48, 926–

929.

1994. Inhibition of sexual maturation in juvenile female and male mice by a

chemosignal of female origin. Physiology and Behavior, 55, 519–522.

Jemiolo, D., Alberts, J., Sochinski-Wiggins, S., Harvey, S. & Novotny, M. 1985.

Behavioural and endocrine responses of female mice to synthetic analogs of

volatile compounds in male urine. Animal Behaviour, 33, 1114–1118.

Jemiolo, B., Harvey, S. & Novotny, M. 1986. Promotion of the Whitten effect in female

mice by synthetic analogs of male urinary constituents. Proceedings of the National


Jemiolo, B., Andreolini, F., Wiesler, D. & Novotny, M. 1987. Variations in mouse (Mus

musculus) urinary volatiles during different periods of pregnancy and lactation.

Journal of Chemical Ecology, 13, 1941–1956.

Jemiolo, B., Xie, T. M. & Novotny, M. 1991. Socio-sexual olfactory preference in female

mice: attractiveness of synthetic chemosignals. Physiology and Behavior, 50,

1119–1122.

Johnston, R. E. 1999. Scent over-marking. How do hamsters know whose scent is on

top and why should it matter? In: Advances in Chemical Signals in Vertebrates, ed.

R. E. Johnson, D. Muller-Schwarze & P. W. Sorensen. New York: Plenum Press,

pp. 227–238.



Johnston, R. E., Sorokin, E. S. & Ferkin, M. H. 1997. Female voles discriminate males’

over-marks and prefer top-scent males. Animal Behaviour, 54, 679–690.

Jones, R. B. & Nowell, N. W. 1973. Aversive and aggression-promoting properties of

urine from dominant and subordinate male mice. Animal Learning and Behavior, 1,

207–210.

1974. A comparison of the aversive and female attractant properties of urine from

dominant and subordinate male mice. Animal Learning and Behavior, 2,

141–144.

1975. Effects of clean and soiled sawdust substrates and of different urine types

upon aggressive behaviour in male mice. Aggressive Behavior, 1, 111–121.

1989. Aversive potency of urine from dominant and subordinate male laboratory

mice (Mus musculus): resolution of a conflict. Aggressive Behavior, 15, 291–296.

Jordan, W. C. & Bruford, M. W. 1998. New perspectives on mate choice and the MHC.

Heredity, 81, 239–245.

Konig, B. 1994a. Components of lifetime reproductive success in communally and

solitarily nursing house mice: a laboratory study. Behavioral Ecology and


1994b. Fitness effects of communal rearing in house mice: the role of relatedness

versus familiarity. Animal Behaviour, 48, 1449–1457.

Krieger, J., Schmitt, A., Lobell, D. et al. 1999. Selective activation of G protein subtypes

in the vomeronasal organ upon stimulation with urine-derived compounds.

Journal of Biological Chemistry, 274, 4655–4662.

Lloyd, J. A. & Christian, J. J. 1969. Reproductive activity of individual females in three

experimental freely growing populations of house mice (Mus musculus L.). Journal

of Mammalogy, 50, 49–59.

Luo, M., Fee, M. S. & Katz, L. C. 2003. Encoding pheromonal signals in the accessory

olfactory bulb of behaving mice. Science, 299, 1196–1201.

Malone, N., Payne, C. E., Beynon, R. J. & Hurst, J. L. 2001. Social status, odour

communication and mate choice in wild house mice. In: Chemical Signals in

Vertebrates, ed. A. Marchelewska-Koj, D. Muller-Schwarze & J. Lepri. New York:


Manning, C. J., Wakeland, E. K. & Potts, W. K. 1992. Communal nesting patterns in

mice implicate MHC genes in kin recognition. Nature, 360, 581–583.

Maruniak, J. A., Desjardins, C. & Bronson, F. H. 1975. Adpatations for urinary marking

in rodents: prepuce length and morphology. Journal of Reproductive Fertility, 44,

567–570.






Mucignat-Caretta, C., Caretta, A. & Baldini, E. 1998. Protein-bound male urinary

pheromones: differential responses according to age and gender. Chemical Senses,

23, 67–70.


242 J. L. Hurst

Mugford, R. A. & Nowell, N. W. 1970. Pheromones and their effect on aggression in

mice. Nature, 226, 967–968.

Nevison, C. M., Barnard, C. J., Beynon, R. J. & Hurst, J. L. 2000. The consequences of

inbreeding for recognising competitors. Proceedings of the Royal Society of London,

Series B, 267, 687–694.

Nibu, K. 2002. Introduction to olfactory neuroepithelium. Microscopy Research and

Technique, 58, 133–134.

Novotny, M., Schwende, F. J., Wiesler, D. Jorgenson, J. W. & Carmack, M. 1984.

Identification of a testosterone-dependent unique volatile constituent of male

mouse urine: 7-exo-ethyl-5-methyl-6,8- dioxabicyclo(3.2.1)-3-octene. Experientia, 40,

217–219.

Novotny, M., Harvey, S., Jemiolo, B. & Alberts, J. 1985. Synthetic pheromones that

promote inter-male aggression in mice. Proceedings of the National Academy of

Sciences, USA, 82, 2059–61.

Novotny, M., Jemiolo, B., Harvey, S., Wiesler, D. & Marchlewska-Koj, A. 1986.

Adrenal-mediated endogenous metabolites inhibit puberty in female mice.

Science, 231, 722–725.

Novotny, M., Harvey, S. & Jemiolo, B. 1990. Chemistry of male dominance in the house

mouse, Mus domesticus. Experientia, 46, 109–113.

Novotny, M. V., Ma, W., Zidek, L. & Daev, E. 1999a. Recent biochemical insights into

puberty acceleration, estrus induction and puberty delay in the house mouse. In:

Advances in Chemical Communication in Vertebrates, ed. R. E. Johnson, D.

Muller-Schwarze & P. W. Sorensen. New York: Plenum Press, pp. 99–116.

Novotny, M. V., Ma, W., Wiesler, D. & Zidek, L. 1999b. Positive identification of the

puberty-accelerating pheromone of the house mouse: the volatile ligands

associating with the major urinary protein. Proceedings of the Royal Society of London,

Series B, 266, 2017–2022.

Olsen, K. H., Grahn, M., Lohm, J. & Langefors, A. 1998. MHC and kin discrimination in

juvenile Arctic charr, Salvelinus alpinus (L.). Animal Behaviour, 56, 319–327.

Payne, C. E., Malone, N., Humphries, R. E. et al. 2001. Heterogeneity of major urinary

proteins in house mice: population and sex differences. In: Chemical Signals in

Vertebrates, ed. A. Marchelewska-Koj, D. Muller-Schwarze & J. Lepria. New York:


Penn, D. & Potts, W. 1998. MHC-disassortative mating preferences reversed by

cross-fostering. Proceedings of the Royal Society of London, Series B, 265, 1299–

1306.

Poole, T. B. & Morgan, H. D. R. 1976. Social and territorial behaviour in mice (Mus

musculus L.) in small complex areas. Animal Behaviour, 24, 476–480.

Potts, W. K., Manning, C. J. & Wakeland, E. K. 1991. Mating patterns in seminatural

populations of mice influenced by MHC genotype. Nature, 352, 619–621.

Ralls, K. 1971. Mammalian scent marking. Science, 171, 443–449.

Reusch, T. B., Haberli, M. A., Aeschlimann, P. B. & Milinski, M. 2001. Female

sticklebacks count alleles in a strategy of sexual selection explaining MHC

polymorphism. Nature, 414, 300–302.



Rich, T. J. & Hurst, J. L. 1998. Scent marks as reliable signals of the competitive ability

of mates. Animal Behaviour, 56, 727–735.

1999. The competing counter-marks hypothesis: reliable assessment of competitive

ability by potential mates. Animal Behaviour, 58, 1027–1037.

Robertson, D. H. L., Beynon, R. J. & Evershed, R. P. 1993. Extraction, characterization

and binding analysis of two pheromonally active ligands associated with major

urinary protein of house mouse (Mus musculus). Journal of Chemical Ecology, 19,

1405–1416.

Robertson, D. H., Hurst, J. L., Bolgar, M. S., Gaskell, S. J. & Beynon, R. J. 1997. Molecular

heterogeneity of urinary proteins in wild house mouse populations. Rapid

Communications in Mass Spectrometry, 11, 786–790.

Robertson, D. H. L., Marie, A. D., Veggerby, C., Hurst, J. L. & Beynon, R. J. 2001.

Characteristics of ligand binding and release by major urinary proteins. In:

Chemical Signals in Vertebrates, ed. A. Marchelewska-Koj, D. Muller-Schwarze & J.

Lepri. New York: Plenum Press, pp. 169–176.

Sandnabba, N. K. 1986. Changes in male odors and urinary marking patterns due to

inhibition of aggression in male mice. Behavioral Processes, 12, 349–361.

Schwende, F. J., Wiesler, D., Jorgenson, J. W., Carmack, M. & Novotny, M. 1986. Urinary

volatile constituents of the house mouse, Mus musculus, and their endocrine

dependency. Journal of Chemical Ecology, 12, 277–296.

Singer, A. G., Tsuchiya, H., Wellington, J. L., Beauchamp, G. K. & Yamazaki, K. 1993.

Chemistry of odortypes in mice: fractionation and bioassay. Journal of Chemical

Ecology, 19, 569–579.

Singer, A. G., Beauchamp, G. K. & Yamazaki, K. 1997. Volatile signals of the major

histocompatibility complex in male mouse urine. Proceedings of the National


Singh, P. B. 2001. Chemosensation and genetic individuality. Reproduction, 121,

529–539.

Southwick, C. H. 1955. Regulatory mechanisms of house mouse populations: social

behavior affecting litter survival. Ecology, 36, 627–634.

Takami, S. 2002. Recent progress in the neurobiology of the vomeronasal organ.

Microscopy Research and Technique, 58, 228–250.

Vandenberg, J. G. 1969. Male odor accelerates female sexual maturation in mice.


Whitten, W. K. 1956. Modification of the oestrous cycle of the mouse by external

stimuli associated with the male. Journal of Endocrinology, 13, 399–404.

Wolff, R. J. 1980. Mating behaviour and female choice: their relation to social

structure in wild caught house mice (Mus musculus) housed in a semi-natural

environment. Journal of Zoology (London), 207, 43–51.

Yamazaki, K., Singer, A. & Beauchamp, G. K. 1999. Origin, functions and chemistry of

H-2 regulated odorants. Genetica, 104, 235–240.

Zufall, F., Kelliher, K. R. & Leinders-Zufall, T. 2002. Pheromone detection by

mammalian vomeronasal neurons. Microscopy Research and Techniques, 58, 251–

260.


244


Part III C O M M U N I C A T I O N N E T W O R K S

I N D I F F E R E N T T A X A

245


246


Introduction

Communication networks can be found in any taxonomic group of an-

imals, all that is required is that their signals travel further than the average

distance between individuals. This potential for taxonomically widespread oc-

currence is one of the reasons that communication networks are likely to be an

important concept for the understanding of communication in general. However,

taxa vary considerably in several aspects that could affect communication net-

works, including the senses used by receivers (signal modality), processing power

and social organization. The potential insights gained from such taxon-related

differences are the reason for grouping chapters into this section.

Not all taxa are covered in Part III: for example, fish do not appear, but they do

in Parts I and IV (Chs. 4, 5, 21, 22 and 23). Also some taxa are underrepresented:

there is a preponderance of endothermic vertebrate groups, which is recognized

to be a general feature of the literature (Bonnet et al., 2002), and invertebrates

have many fewer chapters than their species richness would seem to require. The

invertebrate balance is redressed slightly by the fact that insects are the focus of

a chapter elsewhere in this book (Ch. 8) and by recent books on insect communi-

cation that deal extensively with chorus behaviour (e.g. Gerhardt & Huber, 2002;

Greenfield, 2002). Nevertheless, this part does have chapters ranging from fiddler

crabs to humans and that is a sufficiently broad taxonomic coverage to demon-

strate common themes and illuminating differences.

Fiddler crabs

To the casual observer, fiddler crabs on a mudflat would seem to be a

clear example of a communication network because of the density of crabs and


247


248 Part III

the male behaviour of conspicuously waving an enlarged claw. However, as Denise

Pope points out in Ch. 12, that is a human perception; we need to know whether

the crabs consider themselves part of a network. After carefully weighing up the

evidence, particularly on their visual abilities, she concludes that it is likely that

most fiddler crab populations do function as networks. Careful review also in-

dicates that the claw-waving display functions predominantly in male–female

contexts, focusing attention on the intriguing receiver behaviour of conspecific

interceptive eavesdropping and types of competitive signalling interaction.

Anuran amphibians

The frogs and toads (anurans) were one of the first taxonomic groups to

be considered from a communication network perspective. This early interest was

a result of their habit of communicating in striking choruses: groups of calling

individuals that can number several thousand. The calls of male anurans incor-

porate adaptations to enhance the effectiveness of mate attraction in the noisy

environment of a breeding pool. However, calling males also have to repel male

competitors and avoid the unwanted attentions of predators and parasites. In

Ch. 13, Ulmar Grafe considers how such compromises affect the design of acoustic

signals. He also points out that in natural circumstances the precise timing of calls

(e.g. whether calls are synchronized or alternated with the calls of neighbouring

males) may be as important in determining a male’s reproductive success as the

acoustic properties of the calls themselves.

Songbirds

Bird song has long excited the interest and admiration of humans, a fact

reflected in the large literature on songbirds. Song is a long-range advertising

vocalization, so although males defending territories may be widely separated (in

contrast to birds on leks, in flocks or at roosts) they can function as a network

because of song. In many parts of the world, there is also a distinct dawn chorus,

when most individuals of most species are singing at much the same time. Such

characteristics explain why this section has two chapters on songbirds and may

explain why studies of songbirds from a communication network perspective are

becoming more common.

Marc Naguib concentrates on vocal interactions between territorial (usually

male) songbirds in Ch. 14, integrating such information with aspects of territo-

rial behaviour such as settlement patterns. The spatial and social relationships

that define territorial neighbours are mediated by vocal interactions and have

implications for general spacing behaviour.


Communication networks in different taxa 249

John Burt and Sandy Vehrencamp (Ch. 15) tackle the dawn chorus of songbirds.

It is a striking acoustic phenomenon when song rate, singing diversity and song

complexity reach their peak. However, the function of the dawn chorus is not

readily explained by a single hypothesis. In Ch. 15, the dawn chorus is considered

from a network perspective, which seems more likely to reveal its function.

Terrestrial mammals

Terrestrial mammals are the second taxonomic grouping represented by

two chapters in this section. This reflects the diversity of terrestrial mammals;

for example, they span a size range from shrews to elephants, with obvious con-

sequences for the feasibility of laboratory studies, and employ signal modalities

such as scent and sound.

Bob Johnston deals with scent communication in small terrestrial mammals

in Ch. 16. Scent marks may contain information for several weeks and during that

time the original scent mark may be over-marked by several individuals. As some

species can determine the order of over-marking, this will create a sort of scent

bulletin board.

Scent marking cannot be detected unaided by human observers; this is in con-

trast to the loud calls of many large terrestrial mammals but similar to the infra-

sonic signals of very large mammals. Karen McComb and David Reby, in Ch. 17,

consider the loud calls of large mammals that can and cannot be heard by humans.

They also point out the implications of social organization for communication net-

works, particularly how the fluid fission–fusion nature of many large mammal

groups is likely to increase opportunities for contact.

Marine mammals

Far-carrying acoustic signals and fluid social systems are also characteris-

tics of marine mammals. In Ch. 18, Vincent Janik summarizes the effect of these

factors on communication networks of pinnipeds and cetaceans and explains how

the nature of sound transmission in water means that sounds can potentially travel

much further than in air. He also discusses whether the communication networks

of marine mammals have been reduced in size in recent decades as oceans have

become noisier (Andrew et al., 2002).

Humans

Many of the terms used to discuss communication network behaviours

(e.g. eavesdropping, audience effects) have their origin in our everyday human


250 Part III

experiences. It is, therefore, a surprise to find that studies of human language

generally consider dyads, in close parallel to other animal communication. John

Locke in Ch. 19 argues that a network perspective is more realistic and that the laws

forbidding eavesdropping found in some of the earliest known legal codes show

that such behaviour has always been common. He also argues that information

gained surreptitiously by eavesdropping may be particularly reliable and hence

valuable, repaying the considerable effort and ingenuity often expended by human

eavesdroppers.

Future directions

There is consensus between the chapters in this section that a communica-

tion network perspective is an advance in understanding, but most of the chapters

also point out that more information is necessary in order to evaluate fully the

utility of this perspective by establishing the costs and benefits of communicating

in a network. For this reason, it is premature to contemplate a formal comparative

study of features of communication networks at the level of taxa represented by

chapters in this section. However, it should be possible to attempt comparative

analysis considerably sooner in species-rich groups with diverse features likely to

affect networks (e.g. density and habitat) such as fiddler crabs and anurans.

A key feature of any communication network addressed by most of the chapters

in Part III is the extent of the network: how many individuals are encompassed by a

signal? Theoretical estimates of maximum signal transmission distance combined

with average separation distances between individuals are a very useful first ap-

proximation (for the role of perceptual abilities, see Ch. 20). However, it should be

remembered that such maximum estimates of network size might be considerably

larger than the actual size to which individuals respond. For example, in many

anuran choruses, males adjust their call timing only to immediate neighbours.

Similarly, it would be interesting to know the effect of fission–fusion societies on

network size. Resolving such issues requires detailed study of actual networks,

sometimes involving relatively new techniques (such as passive acoustic location)

or features of signalling modalities that have been recognised as important rela-

tively recently (e.g. scent over-marking).

Several chapters raise the issue of the cognitive requirements for operating

in a communication networks; for example, the extent to which an ability to

identify individuals constrains social eavesdropping. Similarly, the nature of the

information contained in signals will help to determine information flow through

the network: from our human viewpoint, it would seem obvious that human

language is a far richer source of information than the long-range advertising

signals of other animals. Cognitive aspects of communication networks are also


Communication networks in different taxa 251

dealt with by several chapters in Part IV, indicating that such questions are of

interest to many.

References

Andrew, R. K., Howe, B. M., Mercer, J. A. & Dzieciuch, M. A. 2002. Ocean ambient sound:

comparing the 1960s with the 1990s for a receiver off the California coast. Acoustic

Research Letters Online, 3, 65–70.

Bonnet, X., Shine, R. & Lordais, O. 2002. Taxonomic chauvinism. Trends in Ecology and






P1: IYP/KAB P2: KPB-KOD0521823617c12.xml CU1917B/McGregor 0 521 582361 7 April 11, 2005 17:36

12

Waving in a crowd: fiddler crabs signalin networks

d e n i s e s . p o p e


Introduction

A communication network is formed when more than one receiver can in-

tercept the signal produced by a signaller, and when more than one signal reaches

a receiver at the same time (McGregor & Dabelsteen, 1996). Communication net-

work theory broadens the consideration of selection pressures on signallers and

receivers to include selection on signallers by receivers other than the primary or

target receiver, and selection on receivers when they receive more than one signal

simultaneously or intercept a signal that was not targeted at them (McGregor &

Dabelsteen, 1996; McGregor & Peake, 2000).

A mudflat full of male fiddler crabs (genus Uca, family Ocypodidae), all rhyth-

mically waving their enlarged claw, seems a perfect example of a communication

network: there are several signallers and receivers in close proximity, and many

signals are being produced simultaneously. However, this human perception of a

coordinated network may be partly a product of our excellent visual ability and

large size in relation to these small crabs. What about the crabs themselves: how

many receivers does a signal reach, and how many signals can individuals receive

simultaneously? Is our impression that they form signalling networks simply an il-

lusion caused by our extreme size and high visual acuity? Most importantly, what

can we learn about the communication system of fiddler crabs by considering

networks of signallers and receivers rather than simple sender–receiver dyads?

In this chapter, I will first introduce the biology of fiddler crabs, then review

the evidence that groups of displaying males of these species form communica-

tion networks; finally, I will examine the implications that such networks may

have for our understanding of the fiddler crab communication system and suggest


252


Fiddler crabs signal in networks 253

possible routes for future investigation. Although I am focusing on a single taxon,

my hope is that these ideas may stimulate similar lines of investigation into com-

parable signalling systems in other taxa. In particular, much of the attention in

communication network theory has, to date, focused on the phenomena of social

eavesdropping and audience effects (see Chs. 2 and 4). Such social eavesdropping

is unlikely in fiddler crabs, for reasons explained below, but in this review I aim to

illustrate the utility of the network approach for identifying other consequences

for signallers and receivers that are not predicted from the dyadic view of animal

communication. In addition, much of the previous work on networks has focused

on agonistic interactions between males, while in this chapter I focus on signals

used by males to attract and court females.

The biology of fiddler crabs

Fiddler crabs are small, deposit-feeding semiterrestrial crabs that inhabit

protected shores worldwide in tropical and some warm temperate regions (Crane,

1975). There are 97 recognized species and subspecies in the genus (Rosenberg,

2001). Their intertidal and semiterrestrial existence governs their life in many

ways: their activities are confined to low-tide periods and their lives are cen-

tred around their individual burrows, which they defend against conspecifics

and which serve as shelters during tidal inundation and as refuges from heat,

desiccation, and predation during low tide periods. Their reproductive lives are

also constrained by an obligate pelagic larval stage. This larval stage often results

in lunar or semilunar cycles of reproductive activity (Christy, 1978; Zucker, 1978;

Greenspan, 1982; Yamaguchi, 2001a), as egg hatching and larval release are timed

to coincide with optimal times for larval transport (Morgan & Christy, 1995). The

timing of peak mating activity is, therefore, set by the timing of larval release and

the duration of egg incubation (approximately two weeks depending on temper-

ature and species; reviewed in Yamaguchi, 2001b).

Among behavioural biologists, fiddler crabs are perhaps best known for their

striking sexual dimorphism: males have highly asymmetrical claws, with the ma-

jor claw greatly enlarged (up to five times in length) relative to both the male’s

own minor claw and the female’s two symmetrical small claws (Rosenberg, 2002).

Since the minor claw is used for scooping up the sediment for deposit feeding,

females have the advantage of two feeding appendages while males have only one

and hence have lower intake rates (Weissburg, 1992). The male’smajor claw is used

primarily in two types of activity: fighting and signalling. Both males and females

defend their burrows and the surrounding area from intruders, and the major

claw is a very effective weapon in these disputes. Male–male aggression generally

progresses through a series of stereotyped threat postures with the major claw to


254 D. S. Pope

eventual pushing and grappling with interlocked claws, which sometimes results

in one opponent being thrown (Crane, 1975; Hyatt & Salmon, 1977; Jennions &

Backwell, 1996). When males are displaced from burrows they ‘wander’, either

searching for an empty burrow or attempting to take over a burrow from another

male (Crane, 1975; Jennions & Backwell, 1996; Backwell et al., 2000). Much of the

variation in major claw morphology between species may be related to differences

in fighting techniques (Crane, 1975).

Fights between males for burrow ownership may be common because male bur-

rows serve as a breeding resource in many species of fiddler crab (Christy, 1982;

Backwell & Passmore, 1996). In what has sometimes been regarded as the ‘typical’

fiddler crab mating system, females leave their own burrows to ‘wander’ and sam-

ple courting males and their burrows, eventually choosing to stay in one burrow

to mate with the male and lay her eggs; she then usually remains there for the

duration of egg incubation until the larvae are released (Christy, 1983; Backwell

& Passmore, 1996). A second mating system involves copulation on the surface

close to the female’s own burrow (Crane, 1975; Salmon, 1984; Christy & Salmon,

1984); in this case, a male may defend his burrow as a base from which to court

neighbouring females (Salmon, 1984). Recently, it has been recognized that many

species exhibit both of these modes of mating in the same population; hence, they

might best be thought of as alternative mating tactics (Koga et al., 1998; de Rivera

& Vehrencamp, 2001; de Rivera et al., 2003). Several factors probably contribute

to the opportunity for species to engage in surface mating in addition to, or in-

stead of, burrow mating: small clutch size (Christy & Salmon, 1984; de Rivera &

Vehrencamp, 2001), anatomical receptivity of females (if gonopore opercula are

decalcified to allow copulation throughout the lunar cycle (Salmon, 1984)), preda-

tion level (Koga et al., 1998), density (de Rivera et al., 2003) and the spatial overlap

between feeding areas where females burrow and breeding areas where males

court (Christy, 1982, 1983), because in species where males and females gener-

ally do not inhabit adjacent burrows, the opportunity for surface mating is very

limited.

In addition to fighting, males use their major claw in a variety of movement

signals and signalling postures, most conspicuously in the claw-waving display.

This display is a species specific (Crane, 1975) and relatively stereotyped (Hyatt,

1977; Doherty, 1982) pattern of claw elevation (and in some species, unflexing),

sometimes accompanied by movements of the minor claw, legs and body, and

occasionally combined with a stereotyped pattern of locomotion (Crane, 1975).

The display is performed only during the breeding season (Salmon, 1965; Crane,

1975; Wolfrath, 1993) and generally only by territorial, burrow-holding males.

This context of the display led to suggestions that it may function to attract

receptive females to the burrow, repel rival males from it or serve the dual



function of signalling to both types of receiver (reviewed by Crane, 1975; Moriito &

Wada, 2000).

Recent experimental work has attempted to distinguish between these pro-

posed functions of claw waving by identifying the target receiver of claw wav-

ing in several species of fiddler crabs (U. pugilator (Pope, 2000); U. beebei and U.

terpsichores (formerly U. musica; Rosenberg, 2001; D. S. Pope, unpublished data);

U. annulipes (P. R. Y. Backwell, unpublished data); U. tangeri (D. S. Pope & P. K.

McGregor, unpublished data)) and one other species of ocypodid crab, Scopimera

globosa (Moriito & Wada, 2000). The details of the experimental design differed

between studies, but the general experimental approach involved isolating males

either in cages or with temporary fences to control their visual environment,

and then exposing them to different categories of potential receiver: neighbour-

ing and/or introduced males, and neighbouring and/or introduced females. The

results of these studies are best understood by first explaining that waving can

be classified into at least two categories based on the intensity of the display,

as has also been pointed out by other authors (von Hagen, 1962; Salmon, 1965;

Crane, 1975; Doherty, 1982). High-intensity waving can be differentiated from low-

intensity waving both by an increased rate and, in some species, by the addition

or deletion of display components. In all six species studied, high-intensity wav-

ing was evoked only by the introduction of females, simulating the presence of a

mate-searching female in the male’svicinity, strongly implying that high-intensity

waving is directed exclusively to wandering females in these species.

High-intensity waving is, therefore, clearly part of the courtship sequence in

burrow-mating species. During times of peak mating, males in good condition

generally wave at the low intensity, or background level, more or less continuously.

von Hagen (1962) suggested that low-intensity waving functions to orient mate-

searching females towards the male from relatively long distances. When a male

detects a wandering female near his burrow, he switches to high-intensity waving

by increasing the wave rate and adding or subtracting display components. When

females approach closely, males often switch to yet another courtship signal (e.g.

rapping the major claw against the substrate in U. pugilator (Salmon, 1965) and the

raised carpus display in U. beebei (Christy, 1988a)). The rate of waving also varies

with temperature (Hyatt, 1977; Doherty, 1982) and male size (Hyatt, 1977; Jennions

& Backwell, 1998).

While high-intensity waving seems clearly directed to a particular receiver, low-

intensity waving may function more as a broadcast signal, in the sense that it is

not directed to any particular individual; however, as von Hagen (1962) suggested,

it may be targeted at a general class of receiver: the mate-searching female. This

function may imply that the presence of other individuals should have no effect

on the ‘background’ level of low-intensity waving. The six species studied so far


256 D. S. Pope

differed in the effect of neighbours’ presence on low-intensity waving: in other

words, whether the rate of waving differed when males were completely visually

isolated and when they could see their near neighbours. S. globosa, U. pugilator and

U. annulipes showed no background, or low-intensity, waving in the experiments

(Moriito & Wada, 2000; Pope, 2000; P. R. Y. Backwell, unpublished data); U.

terpsichores waved at the same rate when alone as when surrounded by neighbours

(D. S. Pope, unpublished data), and in both U. beebei and U. tangeri the presence of

neighbours increased the wave rate above the background level when visually iso-

lated (D. S. Pope, unpublished data; D. S. Pope & P. K. McGregor, unpublished data).

The lack of background waving in the S. globosa and U. pugilator studies may have

been experimental artefacts as the males in these cases were translocated to caged

areas, so the natural level of background waving in those species remains to be

clarified. The U. annulipes study also involved caged males, although in this case the

males were not displaced; the fact that males waved only when females approached

accords with observations of natural interactions in this species (Backwell et al.,

1998). The reasons for lack of low-intensity or background waving in this species

deserve further investigation.

The different effects of the presence of neighbours on the waving rate in the

remaining three species may be attributable to differences among the species in

the spatial overlap between the sexes. In U. terpsichores, males and females are

generally spatially segregated while in U. beebei and U. tangeri, males and females

intermingle in the same microhabitat (personal observation); therefore, U. terpsi-

chores males have only male neighbours while the neighbours of the other two

species would include both males and females. This potential correspondence be-

tween the presence of female neighbours and an increased rate of low-intensity

waving may indicate either that low-intensity waving is simply stimulated by the

presence of females in the vicinity, whether they are neighbours or wandering

females, or that female neighbours themselves are part of the target receivers of

the low-intensity display. Another difference between U. terpsichores and the other

two species is that the other species engage in surface mating in addition to bur-

row mating. In these species, female neighbours are potential mates, and if these

females assess the quality of neighbouring males, low-intensity claw waving may

be targeted at them (see discussion on p. 268). However, a comparative study of

four Panamanian species that exhibit different combinations of mating tactics,

including U. terpsichores and U. beebei (D. S. Pope, unpublished data) found no in-

dication that males of any species faced female neighbours while waving, and in

every species, males waved most often not facing any individual in the vicinity,

suggesting that they may actively avoid facing neighbours while waving. If female

neighbours do prove to be part of the target receivers of low-intensity waving, it

is likely still more accurate to think of this level of waving as a broadcast display,



directed not at specific individuals but rather targeted at females in general as po-

tential receivers. Despite suggestions based on behavioural observation that claw

waving appears to be directed sometimes to males as a territorial or threat display

(reviewed by Moriito & Wada, 2000), the experimental evidence indicates that this

is not the case in the six species studied so far.

From current knowledge of fiddler crab mating systems, it appears that the

majority of species engage in burrow mating, either alone or in combination with

surface mating and other less-common tactics (Pope, 1998). All of the experimen-

tal investigations into the targeted receivers of the display involved species with

burrow mating only or mixed tactics; therefore, in these cases, the importance

of attracting females for burrow mating is clear. No studies have investigated

the function of waving in species that copulate only on the surface, but several

authors have noted that surface mating is not preceded immediately by waving

(Salmon, 1984; Yamaguchi, 2001c), implying that it does not serve as a courtship

signal in this context in the same sense that it does in burrow mating. The fact

that the waving display has been retained in these species implies that it continues

to serve some function as a communication signal, although phylogenetic com-

parative evidence suggests that the complexity of the display may be reduced in

these species (Pope, 1998; de Rivera & Vehrencamp, 2001). Further investigations

are warranted into the possibilities that neighbouring females assess males for

surface mating by their waving display (see discussion below) or that waving is

used more in male–male interactions in these species.

In addition to claw waving, fiddler crabs have a rich repertoire of other signals,

not all of which involve the major claw. Using the major claw held outstretched,

males produce threat signals to other males (similar to the threat displays of

many other brachyuran crabs (Wright, 1968)). By rapping the claw against the

substrate or through stridulation of body parts, males also produce vibration sig-

nals (Salmon & Horch, 1972; Popper et al., 2001). These vibration signals are most

commonly produced at night by species that are nocturnally active, although they

may also play a role in the final sequence of diurnal courtship, as described above.

The other major class of signals in fiddler crabs are structures constructed from

the sediment (sand or mud), generally in close proximity to the signaller’sburrow.

Courting males of 16 species build structures that range from the small semidomes

of U. pugilator to the elaborate pillars of U. beebei and hoods of U. terpsichores (re-

viewed by Christy, 1988b; Christy et al., 2002). Hoods and pillars increase the attrac-

tiveness of the males that build them to mate-searching females (Christy, 1988a;

Christy et al., 2002). While I will focus on claw waving in this review, some of

the conclusions may also be applicable to the other potentially long-range signals

of fiddler crabs, specifically vibration signals and structures such as pillars and

hoods.


258 D. S. Pope

Much of the preceding descriptions of the crabs’ biology, as well as many of

the suggestions to come in this chapter, may also apply to other members of the

family Ocypodidae. There are five subfamilies (Kitaura et al., 1998) and fiddler crabs

belong to the subfamily Ocypodinae, along with their closest relatives, the ghost

crabs (genus Ocypode). This close relationship with ghost crabs can be misleading

as ghost crabs are unusual within the family (many ghost crabs are specialized

predators; many are nocturnal; they often inhabit exposed shorelines; and they

are, on average, larger than other members of the family), and many of the other

species more closely resemble fiddler crabs in ecology and behaviour than do ghost

crabs. Crabs in the genera Macrophthalmus, Ilyoplax and Scopimera, in particular,

resemble fiddler crabs in that they are small, deposit-feeding crabs inhabiting

protected shores, which show a rich repertoire of signalling behaviour including

claw waving (e.g. Wada, 1991; Kosuge et al., 1994; Moriito & Wada, 1997) and

structure building (e.g. Wada, 1994, Kitaura et al., 1998). They also show both

surface- and burrow-mating tactics, with many species exhibiting both tactics

(e.g. Wada, 1984; Henmi et al., 1993).

Do fiddler crabs signal in networks?

All signals produced by fiddler crabs have four classes of potential con-

specific adult receiver: wandering females, burrow-holding females, wandering

males and burrow-holding males. At least in species with substantial overlap in

microhabitat use by males and females, all four classes of individual are potentially

within receiving range of the display at any given time, creating the conditions

necessary for the formation of communication networks: that is, the active space

of a signal exceeds the average spacing between individuals (McGregor, 1993;

McGregor & Dabelsteen, 1996). What is the evidence that these conditions are met

within fiddler crabs?

Interindividual spacing

Fiddler crab colonies are often described as dense, but this is of course

from the human observer’s point of view. In fact, average densities vary both be-

tween and within species. Larger species are in general less densely distributed

than smaller species (de Rivera & Vehrencamp, 2001). The density of fiddler crab

colonies can be estimated by counting surface-active individuals, by counting the

number of open burrows or by excavating the sediment and counting all crabs un-

covered in a given area; the last gives the best estimate of true population density

(Macia et al., 2001). Counts of surfacing individuals underestimate the true popula-

tion density in U. annulipes while burrow counts tend to overestimate the number

of excavated crabs because some burrows are empty (Macia et al., 2001). However,



when considering the social environment available to waving male crabs, the

density of individuals on the surface at any given time is the best estimate of

the potential density of interactants in the putative communication network. We

can get an estimate of the range of interindividual distances by considering two

well-studied species at opposite ends of the size spectrum: U. beebei and U. tangeri

(with average adult male body sizes, measured as carapace width or the distance

between the outer edge of the eye sockets, of approximately 0.9 cm and 3.0 cm,

respectively; D. S. Pope, unpublished data). Density measures can be transformed

into estimates of interindividual distances by taking the reciprocal of the density

to yield the average space per individual; if it is then assumed that the area is circu-

lar, the radius of that circle can be calculated from the area. Multiplying the radius

by two gives the average distance between two individuals. U. beebei is found at high

densities of 49 active individuals per m2 on average (D. S. Pope, unpublished data),

which translates into 16 cm between individuals. The larger species U. tangeri is

more widely spaced, at an average of 4.6 active individuals per m2, giving an aver-

age interindividual distance of 53 cm (D. S. Pope, unpublished data). Each of these

densities was estimated from areas of high crab activity, but crab distribution is

patchy within the colonies of both species (U. beebei (de Rivera et al., 2003), U. tangeri

(D. S. Pope, unpublished data)), so there will be areas with larger spacing between

individuals. These estimates may, therefore, be regarded as the optimal conditions

for communication networks in these species. In addition, other species are found

at lower densities. For example, U. terpsichores is in the same size class as U. beebei

but is often found in sandier areas with densities of 17.5 individuals per m2, or

27 cm between neighbours. U. stylifera is a larger crab (approximately 2.4 cm body

size (Crane, 1975)) and is found at 2.6 individuals per m2 (70 cm spacing) at high

local densities (D. S. Pope, unpublished data).

These estimated interindividual distances represent the average spacing of

burrow-holding individuals, but the distance of a wandering individual from con-

specifics will be approximately half this distance, assuming that a wanderer main-

tains an equal distance between conspecific burrows. Therefore, wandering males

or wandering mate-searching females of U. beebei would be, on average, 8 cm from

the closest burrow-holding conspecifics and U. tangeri wanderers would be, on

average, 26.5 cm from the closest burrow-holding conspecifics.

Detection distance of conspecifics

Given that we know the average spacing of individuals, we now need to

know at what distance fiddler crabs are likely to perceive a waving male conspeci-

fic. On an absolute level, the vision of fiddler crabs is constrained by the size

of the ommatidia in the crabs’ eyes, as its size determines the resolution of the

eye. The ability of the eye to resolve an object depends on both the angular size


260 D. S. Pope

of the object and the object’s contrast with the background. Therefore, a con-

servative estimate of the limit of detectability is that an object with an angular

size less than the size of a single ommatidium is unlikely to be resolved from

the background unless it contrasts strongly with it (Land & Nilsson, 2001; Zeil &

Hofmann, 2001). However, ommatidial size varies throughout the eye. Fiddler

crabs have eyes well adapted to their primarily flat visual environment: like all

‘flat world’ crabs, they have a band of high vertical resolution around the hori-

zon of the eye, which they align with the visual horizon (Zeil et al., 1986; Land

& Layne, 1995a; Zeil & Al-Mutairi, 1996). Since fiddler crabs carry their eyes on

long stalks, objects below the height of their eyes, including the bodies of most

conspecifics, will be seen below the horizon line. Several authors have argued that

this visual horizon can allow crabs easily to categorize stimuli into either ‘con-

specific’ or ‘predator’, as predators, being larger than the crab, would be imaged

above the horizon line (Land & Layne, 1995a; Layne, 1998). In this high-resolution

zone close to the horizon, the theoretical resolution threshold is approximately

0.5–1◦. Given this resolution and assuming an eye height of 2.5 cm for crabs of

1.0 cm average body size, a 1 cm conspecific should be easily resolvable at 57 cm;

this distance doubles to 114 cm for a 2 cm conspecific ( J. Zeil, personal communi-

cation). Larger species, such as U. tangeri, should theoretically be able to detect con-

specifics at greater distances, both because of the larger stimulus size and because

of the increased height of the eyes above the ground (estimated to be 4 cm in U.

tangeri (D. S. Pope, unpublished data)), which expands the range over which dis-

tance judgements can be made based on retinal elevation (J. Zeil, personal com-

munication). Therefore, the likely detection distance for conspecifics exceeds the

average interindividual spacing in both large and small species of fiddler crab.

Unfortunately, we are technically constrained in our measurements of what a

crab can see by what its behaviour tells us. In other words, a field measurement of

the distance at which a crab reacts to a specific stimulus is a combined measure

of both the crab’s ability to detect it and the relevance of the stimulus. A given

stimulus may well be detectable at much greater distances but it only elicits a re-

sponse once it enters a specific zone around the individual’s burrow. Given these

caveats, what reaction distances have been reported for fiddler crabs? Reaction

distances to conspecifics and conspecific-sized stimuli have been tested in only a

few species. Land & Layne (1995a) reported 30 cm as the distance at which court-

ing male U. pugilator (large male carapace width 1.6 cm (de Rivera & Vehrencamp,

2001)) seemed to notice a wandering conspecific, based on an increase in wave rate.

They reported 10–15 cm as the distance at which males apparently discriminated

the sex of an approaching individual (presumably by the presence or absence of a

major claw), based on the approach distance at which males switched from wav-

ing to threat behaviour to males. The 30 cm distance corresponds to an angular



size of two to three interommatidial angles, which is two to three times larger

than the theoretical detection threshold. These distances should be interpreted

as minimum estimates of detection distance, however, because they are probably

modulated by the size of a male’s territory, as illustrated by the fact that these

authors also measured responses to predator-like stimuli above the horizon (to

which the motivation to respond is presumably higher) at much smaller angular

sizes. Hemmi & Zeil (2003) performed experiments on burrow surveillance by male

U. vomeris by moving crab-sized dummies (2.25 cm wide and 1.2 cm high) across the

surface towards the male’s burrow and measuring the distance at which the crabs

responded by returning to their burrow to defend it. Male response was better

predicted by the distance of the dummy to the burrow than by the distance of the

dummy to the crab: they consistently responded when the dummy was an average

of 23.8 cm from the burrow, suggesting that this distance might represent the

radius of their defended territory. However, the distance between the dummy and

the crab when the crab reacted varied greatly, since this depended on the male’s

distance to his burrow and the orientation of approach of the dummy towards the

burrow. Crabs responded to dummies at distances of up to 80 cm (J. Zeil, personal

communication), indicating that these conspecific-sized objects were clearly re-

solvable at that distance. Finally, in U. tangeri (3 cm, as above), response distances

to wandering females have been estimated as 150–200 cm (von Hagen, 1962). In

a preliminary experiment on the same species, I attempted to determine a max-

imum reaction distance by controlling a male’s visual environment with fences

and then providing him with a stimulus to which he should be highly motivated

to respond, a tethered but realistically moving female. In these preliminary mea-

surements, I estimated a reaction threshold distance of 100–150 cm. More field

experiments and observations, in addition to neurobiological work on crab vi-

sion, will help us to understand better how fiddler crabs process and respond to

their visual environment.

All of these measured reaction distances were reactions to non-waving con-

specifics, or conspecific-sized objects. There are several reasons to suspect that the

claw-waving display makes a male more detectable to conspecifics than a motion-

less male. First, at adult male body sizes, the claw length exceeds the carapace

width of the male (Crane, 1975), increasing the size of the visual stimulus. Second,

the claw itself is probably the most detectable part of the male: in displaying males

it is often bleached white to contrast strongly with the body (Crane, 1975) and it

is the part of the body that contrasts most strongly with the substrate in terms of

spectral reflectance and polarization (Zeil & Hofmann, 2001). There is, as yet, no

definitive evidence that fiddler crabs have colour vision, but recent findings are

suggestive of a dual-pigment system (Horch et al., 2002), and the smooth wet cuti-

cle of fiddler crabs generate ample specular and ultraviolet reflectance contrasts


262 D. S. Pope

with the mudflat background (Zeil & Hofmann, 2001). The movement of the claw

provides motion signatures that should also make the crab more detectable (Zeil &

Zanker, 1997). In addition, during claw waving, the claw itself is elevated above the

level of the crabs’ eyestalks and crosses the visual horizon of viewing conspecifics,

hence entering the zone above the horizon that is used to detect predators (Land

& Layne, 1995a; Zeil & Zanker, 1997). Some authors have suggested that claw wav-

ing thus taps into the predator-escape responses of conspecifics (e.g. Land & Layne,

1995a). Clearly, even if the waving display did initially exploit the female’sreceiver

biases for detection of moving objects in the zone above the horizon, selection has

since modified the response from a generalized negative or inhibitory response

towards threatening stimuli to a more specialized positive or attractive response

towards preferred male conspecifics (see discussion of exploitation of antipredator

receiver biases in Greenfield (2002)). In summary, it is likely that several aspects of

signal design work together to increase the detectability of displaying male fiddler

crabs to conspecifics, although we do not yet know the absolute ‘signal space’ of a

male’s claw-waving display.

Inferences and assumptions

The evidence reviewed above suggests that both the distances at which

crabs can theoretically detect conspecific-sized objects and the distances at which

they have been shown to react to conspecifics exceed the average spacing between

individuals, especially the distances between a wandering individual and the clos-

est burrow owner, thus setting the stage for potential communication networks to

exist, at least in species or populations with relatively high densities of individuals.

Clearly more research is needed on the perceptual ability of crabs, including poten-

tial differences between species resulting from factors such as phylogeny (fiddler

crabs are traditionally divided into broad-front and narrow-front species, based on

the space between their eyes and hence a difference in relative eyestalk length be-

tween species – what effect does this have on their vision and behaviour?), crab size

(does ommatidial size and number scale with body size?) and visual environment

(do mangrove-dwelling fiddler crabs have any special adaptations to deal with

their more complex visual environment?). All of these considerations are likely

to mean that the extent of communication networks will vary both between and

within species of fiddler crabs, and this variation could be a fruitful avenue for

further research.

Implications of network signalling for fiddler crab communication

I will now examine the consequences of assuming that fiddler crabs signal

in networks. What are the possible behavioural effects this will have on how fiddler



crabs produce and respond to signals, as distinct from what would be predicted

from a traditional ‘dyadic’ signalling scenario?

Production of signals by fiddler crabs in a network: strategies for signal competition

In a communication network, signallers are faced with a more complex

problem than in a simple sender–receiver dyad in the sense that they must neces-

sarily compete with other signallers for the attention of the targeted receiver. This

is self-evident in many cases of sexual signalling, where males are competing to

attract females. In most situations in fiddler crabs, as in many chorusing species of

insects and anurans, females will be in a position to receive more than one signal

simultaneously, so the environment is by default a network (Ch. 2). Such signal

competition probably has effects on at least two timescales: in terms of gross sig-

nal timing (whether to signal or not), and in terms of the fine-scale patterns of

signal timing among neighbouring males. I will term these two timescales bout

timing and signal timing, respectively (Gerhardt & Huber, 2002; Greenfield, 2002).

Bout timing

At the level of bout timing, males undoubtedly use other males as cues for

when to start signalling. Presumably because of energetic constraints, males do not

constantly wave their claws, so they should time their signalling to coincide with

the maximum likelihood of attracting receptive females, assuming that, at least

in burrow-mating species, the display functions primarily in attracting females

to the male’s burrow and persuading them to mate. As noted above, male fiddler

crabs track the timing of receptive females on the scale of the breeding season

(Salmon, 1965; Crane, 1975; Wolfrath, 1993), the lunar cycle (Christy, 1978; Zucker,

1978) and also the daily cycle (Christy et al., 2001). In addition, males are probably

selected to signal whenever their neighbours are signalling lest they miss mating

opportunities; hence males are probably likely to begin signalling if another male

does so. There is some evidence that males do respond this way in U. pugilator,

both to waving neighbours and acoustically signalling neighbours (Salmon, 1965;

Pope, 1998). This effect may also be inferred in U. beebei and U. tangeri, as males of

these species show an increase in waving rate with an increasing number of male

neighbours (D. S. Pope, unpublished data). There is good evidence that males use

other males as cues to begin signalling bouts in acoustic insects (Greenfield, 2002)

and so this should be a fruitful avenue for research in fiddler crabs as well.

Signal timing

Choruses of acoustically signalling anurans and insects often exhibit

group coordination of signals such as synchronous or alternating calling, which

must be accomplished through fine-scale timing adjustments of individual males


264 D. S. Pope

to the calls of other calling males (Gerhardt & Huber, 2002; Greenfield, 2002;

Ch. 13). These chorusing interactions are thought to be epiphenomena result-

ing from the preference of females for leading signals (Greenfield, 1994, 2002).

Recent experiments using acoustic playback have revealed in many anuran and

insect species that if two signals overlap in time but are otherwise equal, females

will prefer the leading of the two signals (reviewed by Gerhardt & Huber, 2002;

Greenfield, 2002). Such a preference is thought to result from psychoacoustic con-

straints such as masking (Gerhardt & Huber, 2002) or the precedence effect (a

phenomenon by which two signals, overlapping in time, are perceived as a single

acoustic object by the receiver (Greenfield et al., 1997; Greenfield, 2002)). Males

have thus been selected to avoid being the following male and hence have evolved

timing mechanisms by which they delay their calling if a rival male calls within a

certain critical interval following the male’s own call (Greenfield et al., 1997). As a

consequence, males subtly adjust their timing in an effort always to be the first to

call of a pair of calling males. Such mutual adjustment leads to either synchrony

or call alternation, depending on the call timing of the species. There is no evi-

dence that either synchronous or alternating patterns of calling are cooperative

in the sense that there is any benefit owing to greater attraction of females per

capita by grouped versus solitary signallers (Gerhardt & Huber, 2002; Greenfield,

2002); hence signal competition is the most parsimonious explanation for these

chorusing phenomena.

In a few species of fiddler crabs and a related ocypodid, there is some very good

evidence that males adjust the timing of their signals in relation to their neigh-

bours, resulting in the production of synchronous signals: U. annulipes (Gordon,

1958; Backwell et al., 1998, 1999), U. perplexa and U. saltitanta (P. R. Y. Backwell, M. D.

Jennions, K. Wada, M. Murai & J. H. Christy, unpublished data), and Ilyoplax pusilla

(Aizawa, 1998). The phenomenon of synchronous waving has been most thor-

oughly documented in U. annulipes (Backwell et al., 1998, 1999), an Indo-West Pacific

broad-fronted species. U. annulipes is unusual in that males apparently do not pro-

duce low-intensity, or background, waves but only wave in the presence of a female

(Backwell et al., 1998). When a mate-searching female approaches, males cluster

around her, and males within this cluster synchronize their waves with each other.

As in most synchronously calling insects and anurans, Backwell et al. (1998) also

found a female preference for the leading male of a group of synchronous wa-

vers. These males also signal at a faster rate than their neighbours (Backwell et al.,

1999), producing some waves that are not overlapped by other males. It is not clear,

therefore, whether the female preference for the leading male is the result of a per-

ceptual constraint as it appears to be in insects and anurans, or whether it is simply

a consequence of females preferring males that signal at the fastest rate, which

is condition-dependent in this species (Jennions & Backwell, 1998). The fact that



synchronous waving results, however, would imply that perhaps the overriding

preference is for the leader (Backwell et al., 1999), and the fastest waving male

is simply the one who least needs to adjust his signal to those of his neighbours

(Greenfield, 2002). In both U. perplexa and U. saltitanta, males produce synchronous

waves in both the presence and absence of mate-searching females and increase

their wave rate in the presence of females (P. R. Y. Backwell, M. D. Jennions, K. Wada,

M. Murai & J. H. Christy, unpublished data), as expected if high intensity waving

serves to court females and persuade them to mate, as discussed above. The waves

of U. perplexa become less synchronous in the presence of females, perhaps because

the increased rate of high-intensity waving means that some males are not able to

produce waves at such a high rate and fall out of synchrony with their neighbours

(P. R. Y. Backwell, M. D. Jennions, K. Wada, M. Murai & J. H. Christy, unpublished

data). In these two species, females also prefer the male waving at the fastest rate,

but only in U. perplexa is that male also most often the leading male. This would

suggest that the overwhelming preference is for high-quality males, waving at the

fastest rate. The fact that U. perplexa and U. saltitanta continue to wave in synchrony

even when mate-searching females are not nearby suggests other aspects of signal

competition: perhaps it results from a female preference for males with the high-

est wave rate even at a distance, a preference for groups of synchronous wavers, or

because such signal competition allows males to compete better with their male

neighbours in other contexts such as territory acquisition and maintenance (P. R. Y.

Backwell, M. D. Jennions, K. Wada, M. Murai & J. H. Christy, unpublished data). The

results of these studies clearly demonstrate that the network phenomenon of sig-

nal competition for receivers’ responses, resulting in signal synchrony, occurs in

these species. If a high rate of waving indicates male quality in other species of

fiddler crabs, females may be expected to prefer males that wave at the highest

rate in these species as well, which leads to the question of why signal competition

resulting in synchronous signalling is not more widespread in fiddler crabs. Fu-

ture research should be directed at investigating what factors may have promoted

the evolution synchronous waving in some species of Uca and not others.

While synchronous waving has been observed in a few other Uca species (P. R. Y.

Backwell, personal communication), bouts of synchrony do not appear to be as

sustained or as tightly timed as they are in U. annulipes, U. perplexa and U. saltitanta.

In the vast majority of species (i.e. those that do not signal synchronously), it is not

clear whether males adjust the timing of their signals in relation to each other,

other than at the gross level of initiating bouts when neighbours do. However,

results from the ocypodid I. pusilla suggest that males may adjust their signal

timing in more subtle ways. Aizawa (1998) found that males delayed the timing of

their waves so that they overlapped with both live and videotaped male neighbours

in a laboratory setting, agreeing with the observations that neighbouring males


266 D. S. Pope

of this species signal synchronously. More surprisingly, field measurements also

demonstrated that male I. pusilla adjusted the timing of their signals to match that

of the much larger fiddler crab U. lactea, whereas there was no evidence that U.

lactea adjusted their timing reciprocally (Aizawa, 2000). These results suggest that

males of other species of fiddler crab (and other waving ocypodids) may make fine-

scale adjustments to their signal timing in ways that are not immediately obvious

to the naked eye and will only be uncovered by careful analysis. Burford et al. (1998)

found some suggestion of an adjustment of wave rate by large males of U. tangeri

when signalling in the presence of smaller neighbours, but the comparisons were

made only at the level of wave rate, and not the males’ timing of signal initiation

in relation to each other’ssignals. The considerations above lead to the conclusion

that males in many fiddler crab species are easily able to perceive the signals of

at least their immediate neighbours, implying that such timing adjustments are

likely.

Reception of signals by fiddler crabs in a network:strategies for information gathering

Some of the most intriguing possible consequences of fiddler crab commu-

nication networks relate to how receivers may use signals to gather information in

ways not traditionally considered in a dyadic framework. This section is somewhat

speculative as, to date, there is little evidence for these effects. However, given the

clear existence of network effects on the production of signals by male fiddler

crabs (reviewed above), the network environment is likely to have consequences

for receivers as well.

As outlined in the introduction, the experimental evidence from the five fiddler

crab species so far studied points to wandering, receptive females as the primary

receivers to which claw waving is directed. In this section, I will work under the

assumption that wandering females are the target receivers. As all of these five

species engage in burrow mating, and since less research to date has been focused

on species that only mate on the surface, the discussion is biased towards species

that mate in burrows. Our understanding of communication networks in fiddler

crabs will benefit from more in-depth study into the function and use of the claw-

waving display and other signals in surface-mating species, and such study will

provide a useful test of the generality of the following inferences about fiddler

crab networks.

While I assume here that wandering females are the primary receivers of the

claw-waving display, I reiterate that by ‘receivers’ here I mean not only the ‘in-

tended’ receivers, or primary targets, of the signal but also other individuals that

may be paying attention to the signal. As described above, in most species there

are at least two other classes of adult conspecifics within receiving range of the



display: neighbouring males and wandering males. In addition, in species with ex-

tensive spatial overlap between males and females, neighbouring females will also

be within receiving range. Such potential receivers may be considered ‘conspecific

interceptive eavesdroppers’,that is, conspecifics that benefit from intercepting sig-

nals targeted at another individual (see Ch. 2). In addition, other receivers such

as bird predators, juvenile conspecifics and heterospecific crabs may also bene-

fit from intercepting the signals of males (by better locating or avoiding waving

males, depending on the context), but I will not consider these additional receivers

further. I will consider each of the four categories of adult conspecific receivers in

separate sections.

Wandering females: the target receivers

Females may be attracted to groups of signalling males because the group

as a whole is more detectable or because it provides the female with an enhanced

opportunity to compare mates. Such opportunities might benefit a female because

they reduce search time (presumably also reducing search risk) and perhaps result

in a higher overall mate quality if signalling in some way allows her to assess male

quality. There is mixed evidence from acoustically signalling insects and anurans

that females are differentially attracted to groups of signalling males. Studies test-

ing whether choruses themselves attract females better than lone signallers have

found primarily negative results (reviewed by Gerhardt & Huber, 2002; Greenfield,

2002). Yet, there is fairly good evidence that females may be attracted to some

groups of males over others within choruses, based on the size or density of the

group (reviewed by Gerhardt & Huber, 2002). As resources in fiddler crabs are not

as clumped as in these acoustic species, the situation in fiddler crabs may be more

analogous to the second case, of females moving within choruses; as such, the

potential for similar effects exist: females may be more attracted to areas with a

higher density of signalling males. In the eastern Pacific species U. beebei, higher

density increased the likelihood of mate searching by females in experimentally

manipulated areas, and wandering females more frequently entered areas of nat-

urally higher density than low-density areas (de Rivera et al., 2003). If females are

directly attracted to higher densities of waving males, this can be tested by exper-

imentally offering females a choice of patches of males that differ in density of

signallers.

Factors affecting the attractiveness of claw waving to females have been

best studied in synchronously waving species (Backwell et al., 1998, 1999; P. R. Y.

Backwell, M. D. Jennions, K. Wada, M. Murai & J. H. Christy, unpublished data).

Further work is needed on other species to determine what aspects of waving are

most attractive to females and how these factors correlate with male quality and re-

source (burrow) quality. More information on what attracts females to claw waving


268 D. S. Pope

may provide insight into the potential exploitation of receiver biases and how

males may use waving to enhance their detectability and attractiveness to fe-

males, as Christy and colleagues have done for male courtship structures (Christy

et al., 2001, 2002).

Neighbouring females

As reviewed above, there is as yet no evidence that claw waving functions

as a direct prelude to surface copulation. However, in species that have both mixed-

sex colonies and either mixed mating tactics (both surface and burrow mating)

or surface mating only, the possibility exists that females may assess and com-

pare their male neighbours (who are their most likely surface copulation partners

(Yamaguchi, 2001c)) for potential future copulations. In an unpublished study of

four Uca species in Panama, I found no evidence that males directed their displays

to female neighbours, but this does not preclude the possibility that females may

still use the displays to gain information. A recent study by Murai et al. (2002) found

that in U. paradussimieri, a species with a unique mating tactic in which males enter

female’s burrows for mating, territorial males showed evidence of being able to

assess the reproductive state of their female neighbours and directed non-waving

courtship at ones that were close to being receptive, mating with them up to three

days later. Therefore, if males are able to assess and integrate such information

over such time periods, females may in a parallel fashion be able to assess and com-

pare male neighbours and integrate such information into their decisions about

surface-mating partners. The potential for longer-term assessment by neighbour-

ing females in species that surface mate only should be investigated in concert

with experimental manipulations to assess the target receivers of waving in these

species; this would improve our understanding of how waving functions in these

species. Such knowledge will also help us to assess how waving has evolved across

the genus and family as a whole.

Wandering males

The traditional view of claw waving as a dual function territorial signal

holds that claw waving would repel a wandering male from approaching the bur-

row of a waving male, because the display would in this case signal the male’s

ownership of the burrow and his willingness to defend it. Given that there is no

evidence that males address their waving to wandering males, and the fact that

they switch to distinct threat displays when another male intrudes (Salmon &

Stout, 1962; Land & Layne, 1995a; D. S. Pope & P. K. McGregor, unpublished data),

claw waving may not have this function in fiddler crabs. Conversely, the network

view suggests that the display might have the opposite effect: that of attracting a

wandering male to the vicinity. If claw waving is directed at females, wandering

males might use it to gather information about potential areas to establish a new



burrow or attempt a take over of an occupied one. If the goal of burrow ownership

is to attract females for mating (over and above the necessity of a burrow as a

shelter and refuge), then areas of high waving activity might indicate to the male

an area of high likelihood of mate attraction. This effect may be magnified in the

case of high-intensity waving, as it indicates the actual presence of a female. A

wandering male’s response to a waving male may be modulated by the wanderers’

size relative to the resident, since larger crabs usually win fights, although there

is also a resident advantage ( Jennions & Backwell, 1996). Thus, waving may actu-

ally repel smaller males, whereas equally sized or larger males may be attracted

by it.

Male attraction to rival males’ advertisement signals has been suggested in

a few taxa. Stamps (1988) found that juvenile Anolis aeneus lizards showed ‘con-

specific attraction’ when settling in new habitat and suggested that they might

use the territorial advertisement displays of head-bobbing to assess conspecific

density. Alatalo et al. (1982) demonstrated that broadcasting the song of the pied

flycatcher Ficedula hypoleuca attracted settlement by conspecifics in nearby nest

boxes. Playback of male advertisement calls in laboratory phonotaxis experiments

attracted conspecific males in the spadefoot toad Spea multiplicata (Pfennig et al.,

2000) and the house cricket Acheta domesticus (Kiflawi & Gray, 2000). In both of

these cases, smaller males in particular differentially approached calls that were

most attractive to females. Such behaviour can result in male aggregations that

are independent of resource distribution, and it may also be a prerequisite for

satellite male-calling behaviour (reviewed by Gerhardt & Huber, 2002). Conspe-

cific interceptive eavesdropping of this type, involving male attraction to signals

of their rivals, may be much more common than is widely recognized, and the

dearth of examples may simply result from an absence of studies. Observational

studies of the movements of wandering males and experimental studies testing

whether males respond aversively or positively to other males’ signals would help

to clarify the existence of this phenomenon in fiddler crabs.

Neighbouring males

One final possible consequence of fiddler crab communication networks

is another form of interceptive eavesdropping, in this case by neighbouring terri-

torial males. The fact that males are more likely to switch to high-intensity waving

when wandering females approach (D. S. Pope & P. K. McGregor, unpublished data)

suggests the possibility that other males could use their neighbours as ‘female

detectors’. By monitoring the signalling of neighbouring males, in particular not-

ing when they switch to high-intensity waving, they might effectively expand the

distance at which they can detect a wandering female. This monitoring would

presumably need to be upregulated by the actual detection of the female her-

self, so that males do not end up wasting energy unnecessarily. The number of


270 D. S. Pope

neighbours that are high-intensity waving may also have a synergistic effect, such

that a male is more likely to pay attention if more than one neighbour switches to

high intensity, because this is a more reliable indicator of female presence. Recent

experiments on U. tangeri suggest that this effect may indeed be occurring (D. S.

Pope & P. K. McGregor, unpublished data). The high-intensity waving display of

this species is easily differentiated from the low-intensity display by the addition

of an introductory curtsey, and males reliably switch to high-intensity waving

when a female approaches (D. S. Pope & P. K. McGregor, unpublished data). The vi-

sual environment of neighbouring pairs of males was controlled with an opaque

fence so that the males could view each other but females could be introduced

on one side of the fence in such a way that one male and not the other could see

them. Males that could view the introduced females significantly increased their

rate of high-intensity waving, as expected. Males that could see their neighbour,

but not the introduced female, significantly increased their rate of low-intensity

waving; the rate of high intensity waving also increased, but not significantly.

Thus these ‘interceptor’ males showed an intermediate level of waving between

the lower background level and the higher level in the presence of a female. These

results suggest that males do attend to the waves of their neighbours and use

the information to compete to attract females, but they are more responsive to

the actual detection of the female nearby. A similar result was found in the syn-

chronously waving species U. annulipes (M. D. Jennions, unpublished data): males

began to wave in synchrony with a neighbouring male even when their view of the

female was blocked with a fence. This ‘female detector’effect might be common in

fiddler crab species but may not normally be noted by a human observer because

it is often not clear whether a male is responding to seeing the female herself or

to his neighbour.

The possibility of males using rival males’ signals as ‘female detectors’ may be

extended to other taxa as well. Such interceptive eavesdropping by rival males

is a type of socially acquired information (Giraldeau et al., 2002) in which ani-

mals collect information from other conspecifics on resource location and quality.

When the probability of acting on socially acquired information goes up when the

same information is obtained from more than one individual, information cas-

cades can result in which many individuals make the same behavioural decision

without obtaining direct information themselves (Giraldeau et al., 2002; Watts,

2002). Such information cascades can result in information being transmitted

faster than the rate of direct information acquisition, for example when escape

responses are transmitted throughout a group faster than the approach of a model

predator (Treherne & Foster, 1981), which the authors termed the Trafalgar effect.

These information cascades can sometimes lead to suboptimal behaviour if the

behavioural decisions of the initiators of the cascades were erroneous (e.g. escape

response to a sudden movement of vegetation rather than a predator (Giraldeau



et al., 2002)). The experiments described above on U. tangeri and U. annulipes involved

only exposure to the behavioural decision (in this case, high-intensity waving) of a

single conspecific; it remains to be tested if the effect is stronger if males witness

the waving of more than one individual, as would be predicted if the situation is

a true information cascade (Giraldeau et al., 2002). The potential also exists that

males may track the movement of a female through a group of males by monitor-

ing the signals of other males, as suggested by McGregor & Dabelsteen (1996) for

territorial intrusions in songbirds.


The ecology of fiddler crabs (typically high densities and open habitat)

as well as their sensory physiology (good visual resolution) argue that most fid-

dler crab colonies operate as communication networks. The fact that their most

common and conspicuous signal, the claw-waving display, does not appear to be

used in male–male interactions makes it unlikely that the network effect of so-

cial eavesdropping (i.e. eavesdropping on signalling interactions; Ch. 2) occurs in

these crabs. However, many other consequences of communication networks sug-

gest themselves, particularly involving competitive signalling interactions and

novel information-gathering strategies, including forms of conspecific intercep-

tive eavesdropping. Such possibilities should provide fruitful avenues for future

investigation.

The best-studied cases of interceptive eavesdropping, which involve intercep-

tion by predators or parasites, clearly harm the signallers, but the situation is

less clear for conspecific interceptive eavesdropping (reviewed in Ch. 2). In fiddler

crabs, males are likely to benefit from any interception of their signals by females,

whether or not a given individual female was the target of the signal. Even in cases

where a male’s signal compares unfavourably with that of a neighbour because it

is given at a lower rate or is not a leading signal in a synchronous species, males are

still likely to benefit more from waving than from not signalling at all. However,

males may be more likely to suffer costs, in terms of missed mating opportuni-

ties, from the interception of their signals by other males. Are there any strategies

that males can use to minimize these potential costs? The broadcast nature of the

claw-waving display (at least at low-intensity, or background, levels) make it un-

likely that males can eliminate the potential for eavesdropping (cf. in songbirds;

Ch. 3); however, they may be able to target specific individuals when signalling

by orienting either their front or back sides towards the approaching individual

(female). This would present the largest visual stimulus to the female while reduc-

ing the view for neighbouring males situated off-axis to the signalling male. There

is some suggestion that males orient in this way towards approaching females in

U. pugilator (Land & Layne, 1995b), although such orientation was not in evidence


272 D. S. Pope

in U. beebei (Christy, 1988a). The effectiveness of such a strategy for minimizing the

information available to neighbours is unknown and more information is needed

on how widespread this orientation strategy is in fiddler crabs. Future research

should be directed towards investigating not only the existence of the potential

conspecific interceptive eavesdropping described here but also whether males suf-

fer substantial costs from such eavesdropping, and if so, what counter-strategies

they might employ for minimizing these costs. Only further work can illuminate

the costs and benefits to fiddler crab signallers and receivers of operating in a

network environment and uncover behavioural strategies each party might use

to exploit the situation to their best advantage. The diversity among fiddler crab

species, in terms of habitats (mudflat versus mangrove), mating tactics and dis-

play form and function will create variation in the extent to which the network

perspective is applicable to fiddler crabs and could serve as useful tests of the

predicted consequences to these crabs of signalling in a network environment.

Acknowledgements

I would like to thank Peter McGregor and the ‘communication crew’at the University of

Copenhagen for welcoming me into their network and stimulating and clarifying my thinking on

the communication networks of fiddler crabs. Michael Jennions, Giuliano Matessi, Ricardo Matos,

Peter McGregor, Tom Peake, Andrew Terry and an anonymous reviewer all provided very helpful

feedback on earlier versions of this chapter. I would like to thank Jochen Zeil in particular for

his comments and insight into the visual world of fiddler crabs. Patricia Backwell and Catherine

de Rivera generously shared their results with me before they were published. My own research

described here was funded by the US National Science Foundation and the Danish Natural Science

Research Council, which also supported me during the writing of this chapter.

References

Aizawa, N. 1998. Synchronous waving in an ocypodid crab, Ilyoplax pusilla: analyses of

response patterns to video and real crabs. Marine Biology, 131, 523–532.

2000. Heterospecific timing of waving display in ocypodid crabs Ilyoplax pusilla and

Uca lactea. Journal of the Marine Biological Association of the UK, 80, 947–948.

Alatalo, R. V., Lundberg, A. & Bjorkland, M. 1982. Can the songs of male birds attract

other males? An experiment with the Pied Flycatcher Ficedula hypoleuca. Bird

Behavior, 4, 42–45.

Backwell, P. R. Y. & Passmore, N. I. 1996. Time constraints and multiple choice criteria

in the sampling behavior and mate choice of the fiddler crab, Uca annulipes.


Backwell, P., Jennions, M., Passmore, N. & Christy, J. 1998. Synchronized courtship in

fiddler crabs. Nature, 391, 31–32.



Backwell, P. R. Y., Jennions, M. D., Christy, J. H. & Passmore, N. 1999. Female choice in

the synchronously waving fiddler crab, Uca annulipes. Ethology, 105, 415–421.

Backwell, P. R. Y., Christy, J. H., Telford, S. R., Jennions, M. D. & Passmore, N. I. 2000.

Dishonest signalling in a fiddler crab. Proceedings of the Royal Society of London, Series

B, 267, 719–724.

Burford, F. R. L., McGregor, P. K. & Oliveira, R. F. 1998. Chorusing by male European

fiddler crabs, Uca tangeri: a study of visual communication networks. Acta

Ethologica, 1, 33–41.

Christy, J. H. 1978. Adaptive significance of reproductive cycles in the fiddler crab Uca

pugilator: a hypothesis. Science, 199, 453–455.

1982. Burrow structure and use in the sand fiddler crab, Uca pugilator (Bosc). Animal

Behaviour, 30, 687–694.

1983. Female choice in the resource-defense mating system of the sand fiddler crab,

Uca pugilator. Behavioral Ecology and Sociobiology, 12, 169–180.

1988a. Pillar function in the fiddler crab Uca beebei (II): competitive courtship

signaling. Ethology, 78, 113–128.

1988b. Pillar function in the fiddler crab Uca beebei (I): effects on male spacing and

aggression. Ethology, 78, 53–71.

Christy, J. H. & Salmon, M. 1984. Ecology and evolution of mating systems of fiddler

crabs (genus Uca). Biological Reviews 59, 483–509.

Christy, J. H., Backwell, P. R. Y. & Goshima, S. 2001. The design and production of a

sexual signal: hoods and hood building by male fiddler crabs Uca musica.

Behaviour, 138, 1065–1083.

Christy, J. H., Backwell, P. R. Y., Goshima, S. & Kreuter, T. 2002. Sexual selection for

structure building by courting male fiddler crabs: an experimental study of

behavioral mechanisms. Behavioral Ecology, 13, 366–374.

Crane, J. 1975. Fiddler Crabs of the World (Ocypodidae: Genus Uca). Princeton, CT: Princeton

University Press.

de Rivera, C. E. & Vehrencamp, S. L. 2001. Male versus female mate searching in fiddler

crabs: a comparative analysis. Behavioral Ecology 12, 182–191.

de Rivera, C. E., Backwell, P. R. Y., Christy, J. H. & Vehrencamp, S. L. 2003. Density

affects female and male mate searching in the fiddler crab, Uca beebei. Behavioral


Doherty, J. A. 1982. Stereotypy and the effects of temperature on some spatio-temporal

subcomponents of the ‘courtship wave’ in the fiddler crabs Uca minax (Le Conte)

and Uca pugnax (Smith) (Brachyura, Ocypodidae). Animal Behaviour, 30, 352–363.


Problems and Diverse Solutions. Chicago, IL: University of Chicago Press.

Giraldeau, L.-A., Valone, T. J. & Templeton, J. J. 2002. Potential disadvantages of using

socially acquired information. Philosophical Transactions of the Royal Society of London,

Series B, 357, 1559–1566.

Gordon, H. R. S. 1958. Synchronous claw-waving of fiddler crabs. Animal Behaviour, 6,

238–241.

Greenfield, M. D. 1994. Cooperation and conflict in the evolution of signal

interactions. Annual Review of Ecology and Systematics, 25, 97–126.


274 D. S. Pope

2002. Signalers and Receivers: Mechanisms and Evolution of Arthropod Communication.

Oxford: Oxford University Press.

Greenfield, M. D., Tourtellot, M. K. & Snedden, W. A. 1997. Precedence effects and the

evolution of chorusing. Proceedings of the Royal Society of London, Series B, 264,

1355–1361.

Greenspan, B. N. 1982. Semi-monthly reproductive cycles in male and female fiddler

crabs, Uca pugnax. Animal Behaviour, 30, 1084–1092.

Hemmi, J. M. & Zeil, J. 2003. Robust judgement of inter-object distance by an

arthropod. Nature, 421, 160–163.

Henmi, Y., Koga, T. & Murai, M. 1993. Mating behavior of the sand bubbler crab

Scopimera globosa. Journal of Crustacean Biology, 13, 736–744.

Horch, K., Salmon, M. & Forward, R. 2002. Evidence for a two pigment visual system in

the fiddler crab, Uca thayeri. Journal of Comparative Physiology A, 188, 493–499.

Hyatt, G. W. 1977. Quantitative analysis of size-dependent variation in the fiddler crab

wave display (Uca pugilator, Brachyura, Ocypodidae). Marine Behavior and Physiology,

5, 19–36.

Hyatt, G. W. & Salmon, M. 1977. Combat in the fiddler crabs Uca pugilator and U.

pugnax: a quantitative analysis. Behaviour, 65, 182–211.

Jennions, M. D. & Backwell, P. R. Y. 1996. Residency and size affect fight duration and

outcome in the fiddler crab Uca annulipes. Biological Journal of the Linnean Society, 57,

293–306.

1998. Variation in courtship rate in the fiddler crab Uca annulipes: is it related to

male attractiveness? Behavioral Ecology, 9, 605–611.

Kiflawi, M. & Gray, D. A. 2000. Size-dependent response to conspecific mating calls by

male crickets. Proceedings of the Royal Society of London, Series B, 267, 2157–2161.

Kitaura, J., Wada, K. & Nishida, M. 1998. Molecular phylogeny and evolution of unique

mud-using territorial behavior in ocypodid crabs (Crustacea: Brachyura:

Ocypodidae). Molecular Biology and Evolution, 15, 626–637.

Koga, T., Backwell, P. R. Y., Jennions, M. D. & Christy, J. H. 1998. Elevated predation risk

changes mating behaviour and courtship in a fiddler crab. Proceedings of the Royal


Kosuge, T., Poovachiranon, S. & Murai, M. 1994. Male courtship cycles in three species

of tropical Ilyoplax crabs (Decapoda, Brachyura, Ocypodidae). Hydrobiolgia 285,

93–100.

Land, M. & Layne, J. 1995a. The visual control of behavior in fiddler crabs: I.

Resolution, thresholds and the role of the horizon. Journal of Comparative Physiology

A, 177, 81–90.

1995b. The visual control of behavior in fiddler crabs: I1. Tracking control systems

in courtship and defence. Journal of Comparative Physiology A, 177, 91–103.

Land, M. F. & Nilsson, D.-E. 2001. Animal Eyes. Oxford: Oxford University Press.

Layne, J. E. 1998. Retinal location is the key to identifying predators in fiddler crabs

(Uca pugilator). Journal of Experimental Biology, 201, 2253–2261.

Macia, A., Quincardete, I. & Paula, J. 2001. A comparison of alternative methods for

estimating population density of the fiddler crab Uca annulipes at Saco Mangrove,

Inhaca Island (Mozambique). Hydrobiologia, 449, 213–219.





Society, Series B, 340, 237–244.






Morgan, S. G. & Christy, J. H. 1995. Adaptive significance of the timing of larval release

by crabs. American Naturalist 145, 457–479.

Moriito, M. & Wada, K. 1997. When is waving performed in the ocypodid crab

Scopimera globosa? Crustacean Research, 26, 47–55.

2000. The presence of neighbours affects waving display frequency by Scopimera

globosa (Decapoda, Ocypodidae). Journal of Ethology, 18, 43–45.

Murai, M., Koga, T. & Yong, H.-S. 2002. The assessment of female reproductive state

during courtship and scramble competition in the fiddler crab, Uca

paradussumieri. Behavioral Ecology and Sociobiology, 52, 137–142.

Pfennig, K. S., Rapa, K. & McNatt, R. 2000. Evolution of male mating behavior: male

spadefoot toads preferentially associate with conspecific males. Behavioral Ecology

and Sociobiology 48, 69–74.

Pope, D. S. 1998. The fiddler crab claw waving display: function and evolution of a

sexually selected signal. Ph.D. thesis, Duke University, USA.

2000. Testing function of fiddler crab claw waving by manipulating social context.


Popper, A. N., Salmon, M. & Horch, K. W. 2001. Acoustic detection and communication

by decapod crustaceans. Journal of Comparative Physiology A, 187, 83–89.

Rosenberg, M. S. 2001. The systematics and taxonomy of fiddler crabs: a phylogeny of

the genus Uca. Journal of Crustacean Biology, 21, 839–869.

2002. Fiddler crab claw shape variation: a geometric morphometric analysis across

the genus Uca (Crustacea: Brachyura: Ocypodidae). Biological Journal of the Linnean

Society, 75, 147–162.

Salmon, M. 1965. Waving display and sound production in the courtship behavior of

Uca pugilator, with comparisons to U. minax and U. pugnax. Zoologica, 50,

123–150.

1984. The courtship, aggression and mating system of a ‘primitive’ fiddler crab (Uca

vocans: Ocypodidae). Transactions of the Zoological Society of London, 37, 1–50.

Salmon, M. & Horch, K. 1972. Acoustic signaling and detection by semi-terrestrial

crabs of the family Ocypodidae. In: Behavior of Marine Animals, vol. 1, ed. H. E. Winn

& B. L. Olla, New York: Plenum Press, pp. 60–96.

Salmon, M. & Stout, J. F. 1962. Sexual discrimination and sound production in Uca

pugilator Bosc. Zoologica 47: 15–19.

Stamps, J. A. 1988. Conspecific attraction and aggregation in territorial species.

American Naturalist 131, 329–347.

Treherne, J. E. & Foster, W. A. 1981. Group transmission of predator avoidance

behaviour in a marine insect: the Trafalgar effect. Animal Behaviour 29, 911–917.


276 D. S. Pope

von Hagen, H.-O. 1962. Frielandstudien zur sexual- und forpflanzungs-biologie von Uca

tangeri in Andalusien. Zietschrift fur Morphologie und Okologie der Tiere, 51, 611–725.

Wada, K. 1984. Pair formation in the two forms of Macrophthalmus japonicus De Haan

(Crustacea: Brachyura) at a co-occurring area. Journal of Ethology, 2, 7–10.

1991. Biogeographic patterns in waving display, and body size and proportions of

Macrophthalmus japonicus species complex (Crustacea: Brachyura: Ocypodidae).

Zoological Science, 8, 135–146.

1994. Earthen structures built by Ilyoplax dentimerosa (Crustacea, Brachyura,

Ocypodidae). Ethology, 96, 270–282.

Watts, D. J. 2002. A simple model of global cascades on random networks. Proceedings of

the National Academy of Sciences, USA, 99, 5766–5771.

Weissburg, M. 1992. Functional analysis of fiddler crab foraging: sex-specific

mechanics and constraints in Uca pugnax (Smith). Journal of Experimental Marine

Biology and Ecology, 156, 105–124.

Wolfrath, B. 1993. Observations on the behaviour of the European fiddler crab Uca

tangeri. Marine Ecology Progress Series, 100, 111–118.

Wright, H. O. 1968. Visual displays in brachyuran crabs: field and laboratory studies.

American Zoologist 8, 655–665.

Yamaguchi, T. 2001a. The breeding period of the fiddler crab, Uca lactea (Decapoda,

Brachyura, Ocypodidae) in Japan. Crustaceana 74, 285–293.

2001b. Incubation of eggs and embryonic development of the fiddler crab, Uca

lactea (Decapoda, Brachyura, Ocypodidae). Crustaceana 74, 449–458.

2001c. The mating system of the fiddler crab, Uca lactea (Decapoda, Brachyura,

Ocypodidae). Crustaceana 74, 389–399.

Zeil, J. & Al-Mutairi, M. M. 1996. Variation of resolution and of ommatidial dimensions

in the compound eyes of the fiddler crab Uca lactea annulipes (Ocypodidae,

Brachyura, Decapoda). Journal of Experimental Biology, 199, 1569–1577.

Zeil, J. & Hofmann, M. 2001. Signals from ‘crabworld’: cuticular reflections in a fiddler

crab colony. Journal of Experimental Biology, 204, 2561–2569.

Zeil, J. & Zanker, J. M. 1997. A glimpse into crabworld. Vision Research, 37, 3417–3426.

Zeil, J., Nalbach, G. & Nalbach, H.-O. 1986. Eyes, eye stalks and the visual world of

semi-terrestrial crabs. Journal of Comparative Physiology A, 159, 801–811.

Zucker, N. 1978. Monthly reproductive cycles in three sympatric hood-building

tropical fiddler crabs (genus Uca). Biological Bulletin, 155, 410–424.


13

Anuran choruses as communicationnetworks

t. u l m a r g r a f e

University of Wurzburg, Germany

Introduction

It is becoming more and more apparent that communication often takes

place in a network of several signallers and receivers (as shown by most of the chap-

ters in this volume and reviewed by McGregor & Peake (2000)). The network view

of communication stresses that signallers and receivers have additional costs and

benefits to those usually found in dyadic interactions. For example, in communica-

tion networks signallers often face the problem of intense intra- and interspecific

competition whereas receivers must discriminate information from individuals

under conditions of high background noise.

In many frogs and toads, males aggregate in large choruses to advertise for

females. The signals they use are conspicuous and long range; therefore, choruses

constitute a classic example of a communication network. The challenge of com-

municating in such large choruses is to balance the costs and benefits of attracting

a mate, repelling rivals and avoiding predators and/or parasites. Consequently, ad-

vertising in choruses will have far-reaching effects on vocal behaviour. If we want

to understand signal design and signalling behaviour in such aggregations, we

need to look at communication in the network context in which these different

selective pressures operate.

In this chapter, I will review why it is important to investigate communication

in chorusing anurans within the network environment. I will focus on the be-

haviour of both signallers and receivers. First, I will discuss patterns of male–male

vocal competition that can best be understood within the network environment.

In particular, I will discuss how the timing of signals within the chorus determines

mating success. The variation in signal timing between anurans suggests that they


277


278 T. U. Grafe

are fine-tuned to both the level of competition and the receiver biases. Further-

more, high chorus density necessitates that males interact with only a subset of

males. Determining the degree of selective attention (i.e. the number of individu-

als males interact with) is, therefore, an important parameter of connectivity in

the network. Aggregations of chorusing males allow females to monitor male–

male interactions prior to mate choice. What evidence from anurans is there that

females eavesdrop? Similarly, both calling and satellite males may benefit from

monitoring male–male interactions and position themselves within the chorus to

increase their active space, associate with attractive callers or gain information

about the location of gravid females.

Signaller behaviour

Patterns of male–male vocal competition

Many anurans form large and conspicuous aggregations in which high lev-

els of background noise reduce their effectiveness in attracting females (Gerhardt

& Klump, 1988; Narins & Zelick, 1988; Gerhardt & Schwartz, 1995). Similar effects

are found in insects (e.g. Romer et al., 1989) and birds (Klump, 1996). Chorusing

intensifies competition between males over females or over resources of interest

to females. Temporal segregation of calling activity or the partitioning of calling

frequencies are solutions available to some species in some situations. However,

in many cases, males cannot avoid calling in dense aggregations and are subject to

high levels of intra- and interspecific acoustic interference. This is especially evi-

dent in species that breed in temporary ponds where reproductive success depends

on laying eggs as early as possible to ensure that larval development is completed

before the breeding sites dry out.

Males that vocalize in dense aggregations deal with high levels of acoustic

competition in several ways (reviewed by Wells, 1988; Gerhardt & Schwartz, 1995;

Gerhardt & Huber, 2002). They may increase their call repetition rate, increase the

complexity of their calls or defend calling sites and acoustic space against other

males. Many species show all of these adaptations. The classic example is that of the

tungara frog Physalaemus pustulosus where males add chuck notes to their whines

to form complex advertisement calls and increase their call rate depending on

the social milieu (Ryan, 1985). Some frogs even increase call intensity in response

to playbacks of conspecific advertisement calls (Lopez et al., 1988) or lower the

dominant frequency of their advertisement calls during aggressive interactions

with other males (reviewed by Bee & Bowling, 2002). In addition, chorusing males

alternate or synchronize their calls with neighbouring males (e.g. Zelick & Narins,

1983; Forester & Harrison, 1987; Klump & Gerhardt, 1992; Greenfield & Roizen,

1993; Grafe, 1999; Bosch & Marquez, 2002). The precise timing of calls may, in


Anuran choruses as communication networks 279

fact, be as important in determining a male’s mating success as the acoustic call

properties that, traditionally, are investigated. I discuss this in more detail in

following sections.

Repertoire sizes can be large in some anurans, reflecting the potential to mod-

ulate social interactions on a fine scale. Repertoires appear especially impressive

in Boophis madagascariensis, with 28 distinct calls reported (Narins et al., 2000) and

Amolops tormotus, with an amazing variety of calls that have defied categorization

(Feng et al., 2002). Such call variety may reflect the need to signal to males and

females simultaneously (see below).

Signals used during aggressive interactions

Aggressive calls play an important role in mediating the spacing between

male anurans within a chorus and there is tremendous variation in their use

(Schwartz, 2001; Gerhardt & Huber, 2002). In fact, many species never produce

distinct aggressive calls. In the genus Kassina, for example, aggressive calls have

only rarely been reported (Fleischack & Small, 1978), despite extensive playback

experiments in most species (T. U. Grafe, unpublished data). In some species, the

advertisement call grades into the aggressive call (e.g. Wells, 1989; Grafe, 1995a);

in others the advertisement call is distinct from the aggressive call, with the ag-

gressive call being graded (e.g. Schwartz, 1989). Some species have several kinds

of aggressive call (e.g. Given, 1987). Much discussion revolves around the func-

tional significance of graded versus discrete signalling systems and how cheat-

ing can be prevented. Handicap models predict that aggressive signals should be

graded to convey best information about the probability that the signaller will

attack (Grafen, 1990). In contrast, conventional signalling (Enquist et al., 1998) and

discrete handicap models (Johnstone, 1994) both predict that signals should be

discrete. This discussion has not considered the necessity for chorusing males si-

multaneously to repel rivals and attract females. Here, signals need to reach two

different classes of receiver (males and females) and may need to be designed dif-

ferently. In general in anurans, aggressive calls are less attractive to females than

advertisement calls (e.g. Brenowitz & Rose, 1999). Therefore, signals that grade be-

tween advertisement and aggressive calls or discrete aggressive call variants may

allow males to increase the aggressive content of a signal gradually or discretely,

while only partially reducing a male’s attractiveness to females. This may be a

general solution to signalling when trying to reach different classes of receiver

and a resolution to a signalling conflict that is typical of a network environment.

The variation in the use of aggressive calls in anurans may depend on the degree

of flexibility needed in dealing with the conflicting demands of signalling to

males and females simultaneously (Brenowitz et al., 2000; Marshall et al., 2003). An

elegant solution to this problem is the use of two-part advertisement calls, in which


280 T. U. Grafe

one part is directed to females and the other to males. In Geocrinia victoriana, for

example, males produce an introductory note that is directed towards males and

a series of shorter repeated notes that are directed towards females (Littlejohn &

Harrison, 1985). A similar case can be made for the diphasic call of Eleutherodactylus

coqui (Narins & Capranica, 1978). Viewed from an energetic perspective, such a

solution appears wasteful because only every second note is directed to females.

This may explain why such a solution is not very common.

Fine-scale patterns of signal timing

Signal timing can be an important feature influencing mating success

in chorusing anurans and, therefore, males need to monitor the call timing of

competitors (reviewed by Greenfield, 1994a; Gerhardt & Huber, 2002). The fine-

scale patterns of signal timing are often described by taking the calling period

of one male as a reference point and relating the timing of the second individ-

ual to this reference (reviewed by Klump & Gerhardt, 1992; Greenfield, 1994a). At

one end of the continuum of call-timing patterns, individuals signal in perfect

synchrony (relative phase of 0◦); at the other end signals are spaced with equal

intervals between them resulting in perfect alternation (relative phase of 180◦).

Most anurans do not show either of these extreme patterns on a regular basis.

Synchrony can, therefore, refer to signalling patterns in which signals overlap,

whereas alternation characterizes patterns in which signals regularly are more or

less evenly spaced in time (Gerhardt & Huber, 2002). Examples of different call-

timing patterns are shown in Fig. 13.1. I use the term entrainment to refer to

call-timing patterns with relative phase angles below 45◦ but not overlapping

(a similar classification is used in insects (Greenfield, 1994a)). This represents

an operational definition that helps to classify a continuously varying param-

eter. Alternating would then refer to call-timing patterns with relative phase

angles above 45◦. In dyadic interactions of entrained calling, the calls of one

male will lead and the calls of the other male will follow. Leading and lagging

roles often switch between individuals (Fig. 13.1; Gerhardt & Huber, 2002; Grafe,

2003) are less discernible in alternating species (Klump & Gerhurdt 1992; Bosch &

Marquez, 2002).

Levels of overlap rise dramatically in aggregations of three and four males (e.g.

Schneider et al., 1988; Schwartz, 1993; Grafe, 1996a). However, overlap remains

lower than expected if frogs were calling at random. This suggests that males

are interacting in ways that prevent high levels of overlap. In most cases, males

will attempt to place their calls in relatively silent gaps, thus avoiding masking

interference by other males. Zelick & Narins (1985), in their pioneering work with

the Puerto Rican treefrog E. coqui, documented experimentally that males were

able to place their calls in unpredictable gaps of silence. Males were triggered to



Fig. 13.1. Spectrograms of call-timing patterns (defined in text) in five different

species of running frogs (Kassina). K. cassinoides shows alternation, K. schioetzi

entrainment, K. senegalensis entrainment with occasional overlap, K. fusca synchronous

calling and K. kuvangensis synchronous call groups with alternating calls. The letters

(A, B) designate individual males. Note the switching of leader and follower roles

between males in the top four spectrograms. These species call syntopically (i.e. in the

same pond and at the same time) and were recorded in the Guinea savannah region of

the Comoe National Park, Ivory Coast. K. kuvangensis was recorded at Hillwood Farm,

northwestern Zambia.


282 T. U. Grafe

call by the rapid offset of background sound. Similar studies on other anurans

have supported these basic results and the underlying mechanisms of call ini-

tiation (Schwartz, 1993; Grafe, 1996a). Interestingly, constant high noise levels

do not prevent calling, suggesting that call inhibition subsides with time. Like

E. coqui, Broadley’s painted reed frogs Hyperolius marmoratus broadleyi initiate their

calls in response to drops in background noise levels, showing an ‘off-response’

(Grafe, 1996a). The modal response latencies in the reaction of males were be-

tween 40 and 80 milliseconds (in some bushcrickets it is less than 20 milliseconds

(Robinson, 1990)), suggesting that higher auditory centres such as the thalamus

are not involved in processing such a fast response (Walkowiak, 1992). Although

modal response latencies are very short, most calls are given with much longer

latencies. Males avoid call overlap by calling within windows of low noise levels

(‘silence’) by selectively attending to near and thus loud neighbours.

The flexibility with which males can adjust their calls even on a note-by-note

basis is remarkable. Schwartz (1993), studying Hyla microcephala, found that males

increased the spacing between their calls when interrupted, thus avoiding further

overlap of subsequent notes in their call. I found a similar response in the Central

African frog Kassina kuvangensis (Grafe, 2003). Calling in this species is characterized

by synchronizing call groups while at the same time alternating advertisement

calls with those of neighbouring males. As in H. microcephala, males readjusted their

inter-call intervals within milliseconds in response to the playback stimulus.

In contrast to alternating or entrained calling, synchronous calling is unusual

in anurans. It has been reported for only a handful of species: the neotropical hylids

Smilisca sila (Ryan, 1986), Hyla ebraccata (Wells & Schwartz, 1984) and Centrolenella

granulosa (Ibanez, 1993) as well as the African running frogs Kassina fusca and

Kassina senegalensis (Grafe, 1999; T. U. Grafe & H. Lussow, unpublished data). In

pairwise interactions of the savannah running frog K. fusca, 81.5% (overall median)

of calls overlapped (relative phase of 8.6 ± 4.4◦) with a median degree of overlap

of 20.8%. In the Senegal running frog K. senegalensis, only 21.6% of calls between

neighbouring males overlapped on average (T. U. Grafe & H. Lussow, unpublished

data).

Synchrony in anurans, in contrast to many insects (reviewed by Greenfield,

1994a), is achieved by a rapid acoustic response to the onset of a concurrent signal

produced by a neighbour. This ‘on-response’ mechanism is thus fundamentally

different from the ‘off-response’ found in the alternating and entrained response

type. It should be stressed, however, that at least one anuran, K. fusca, can vary

its response type. When presented with playbacks of conspecific advertisement

calls, males showed a synchronous response whereas they entrained calls to het-

erospecific calls or white noise stimuli, thus showing both on- and off-responses

(Grafe, 1999).



There is considerable variation in the patterns of call timing between species. In

insects and anurans, call alternation is found predominantly in species that show

low call rates, whereas synchrony is more common in species that call at high rates

(Narins, 1982; Greenfield, 1994a). An exception to this rule is the genus Kassina.

Males show synchronous or entrained calling at both low and high call rates inde-

pendent of density, suggesting that call rate alone is not the prime determinant of

signal timing (Grafe, 1999). Moreover, some species show both synchrony and al-

ternation (e.g. Moore et al., 1989). In K. kuvangensis, males synchronize call groups

while at the same time alternating calls within call groups. Such synchronized

interdigitated calling may serve to reduce predation while maintaining species-

specific temporal information important to females (Grafe, 2003). This may be a

solution to reducing the costs associated with synchrony and alternation.

Adaptive significance of call timing

Both cooperative and competitive hypotheses can explain the evolution

and maintenance of call timing in anurans and insects. Cooperative explanations

for synchrony include (a) confusion of predators by decreasing the locatability of

signals; (b) enhancement of detection by females by increasing the peak ampli-

tude of signals; and (c) improving the detection of female acoustic responses. Sev-

eral authors have noted that there is little support for these cooperative explana-

tions (Greenfield, 1994a; McGregor & Peake, 2000). Only in one species, S. sila, may

synchronous calling provide protection against frog-eating bats Trachops cirrhosus,

since bats were shown to be more attracted to alternating than to synchronous

playback of calls (Tuttle & Ryan, 1982). There is no evidence for enhanced detection

of overlapping calls by females. Peak amplitude of synchronous calls or choruses of

males is not much higher than that of individual signallers (Bradbury, 1981), thus

providing only a marginal increase in the active space of males. Although females

are attracted to larger choruses in a number of species, the mating success per

male often declines as lek size increases (Deutsch, 1994; Widemo & Owens, 1995).

Similarly, playback experiments with the grasshopper Ligurotettix coquilletti and

the treefrog H. microcephala showed that females were attracted to larger arrays of

speakers but the attractiveness per speaker was not higher for larger arrays (Shelly

& Greenfield, 1991; Schwartz, 1993). Finally, acoustic responses to male advertise-

ment calls are given in only a few species, most notably in midwife toads Alytes spp.

(Bosch, 2001). Since calling effort (duty cycle) is generally not very high in midwife

toads, there is little gained by synchronizing calls to improve the detection of

female acoustic responses.

There is strong evidence that competition to produce leading calls often ex-

plains call timing in anurans (and insects) because there is a strong preference

by females for leading signals (e.g. Grafe, 1996a; Greenfield et al., 1997; Bosch &


284 T. U. Grafe

Marquez, 2002). This preference for leading calls has been termed the precedence

effect (reviewed by Gerhardt & Huber, 2002). In species in which females prefer

leading calls, males that produce trailing calls will benefit by delaying the onset

of their calls to avoid overlap or by speeding up calls to produce leading calls.

Selective attention

In the previous section, I have outlined why chorusing anurans should

generally avoid interference with other signalling males. The problem in a large

aggregation is that if a male is to avoid interference with all males in the chorus,

he would have to stop calling. The solution is to attend only to a subset of nearby

males. Brush & Narins (1989) were the first to investigate systematically the ques-

tion of to how many chorus members an individual male attends. They found that

E. coqui males typically avoided overlap with just two neighbours, only rarely with

three individuals. Monitoring small choruses of four to six male H. microcephala,

Schwartz (1993) found that males generally responded to their loudest rivals. In

addition, more centrally located males typically attended to more males (one to

four) than those at the periphery of the chorus (one). Greenfield & Rand (2000)

have further demonstrated the plasticity involved in selective attention. Male P.

pustulosus responded to two to three neighbouring males depending on the chorus

structure and intensity of those males’ calls. These results suggest that selective

attention is a dynamic process that will vary as males enter the chorus and move

within it.

Selective attention to nearest neighbours may occur even in anurans that do not

avoid call overlap. In K. fusca, call overlap itself is a measure of the attention paid

to other males. Preliminary work (T. U. Grafe, unpublished data) suggests that the

spatial distribution of males in the chorus is an important factor in determining

the number of males that are paid attention.

Another interesting case is the gray treefrog Hyla versicolor (Schwartz, 2001;

Schwartz et al., 2001). Although females discriminate against overlapping calls,

neighbouring males in small choruses do not avoid overlap. Avoidance of overlap

appears to be overridden by female preference for longer calls and a step-like

decrease in attractiveness of short calls even if they are unmasked. Given the

preferences of females, the best strategy for male gray treefrogs appears to be the

production of long calls that partly overlap with those of neighbours. The plasticity

in response to acoustic competition and the differences in auditory perception

of receivers suggest the absence of a unifying general rule governing selective

attention.

Energetics of calling

Important determinants of the interactions of calling male anurans are

the energetic constraints of calling. Calling is the most energetically expensive



behaviour of ectothermic vertebrates (reviewed by Wells, 2001). In the European

treefrog Hyla arborea, for example, instantaneous rates of oxygen consumption

can reach 41 times the resting rate (Grafe & Thein, 2001). Such high levels of

expenditure cannot be maintained for long and they set an upper limit to the

rate and complexity of calling. Female anurans prefer to mate with males calling

at high rates in all species tested (Ryan & Keddy-Hector, 1992; Gerhardt & Huber,

2002). Consequently, males need to adjust the rate and complexity of calls to the

levels of competition or risk having to drop out of the chorus prematurely because

they have run out of energy. In species that vary the duration of calls, such as

H. versicolor, the best predictor of energy expenditure is the product of call rate

and call duration (calling effort or duty cycle; reviewed by Wells, 2001).

In most anurans, calling effort increases with chorus density (e.g. Taigen et al.,

1985; Grafe, 1996b). In H. versicolor, however, males reduce the rate of calling while

increasing call duration as chorus density increases, thus maintaining a constant

level of energy expenditure (Wells & Taigen, 1986; Grafe, 1997a). Females prefer

long calls at low rates to short calls at high rates (Klump & Gerhardt, 1987). These

studies show that the energetic constraints of calling require males to monitor

the behaviour of others to maintain their attractiveness towards females and that

the way males partition their energy depends on female preferences.

Receiver behaviour

Mechanisms of female preferences

Important selection pressures on signal design and signalling behaviour

are the sensory and neuronal abilities of females. In anurans, acoustic commu-

nication plays a central role in mate choice. The wide range in the threshold of

auditory neurons and the sensitivity for narrow frequency bands in the peripheral

auditory system are important in allowing females to choose between conspecific

males in the presence of background noise (reviewed by Narins & Zelick, 1988).

Numerous studies have shown that females prefer males that produce loud and

conspicuous signals with large active space (Ryan & Keddy-Hector, 1992). Conse-

quently, receivers generally exert strong selection for loud and ritualized signals.

In recent years, it has become clear that the fine-scale patterns of signal timing

have a large influence on female choice. Females of many taxonomic groups show

a preference for leading, but not necessarily overlapping, signals in the olfactory

(voles: Johnston et al., 1997), visual (fiddler crabs: Backwell et al., 1998; fireflies:

Vencl & Carlson, 1998) and auditory modalities (field crickets: Wyttenbach &

Hoy, 1993; katydids: Greenfield, 1994b; Greenfield et al., 1997; frogs: Gerhardt &

Huber, 2002; rats: Kelly, 1974; cats: Cranford & Oberholzer, 1976). This preference

for leading signals has entered the literature under the term precedence effect. It

was originally described by Wallach et al. (1949) in humans and describes the


286 T. U. Grafe

observation that when two spatially separated sounds are presented with a brief

delay in onset the leading sound dominates localization. However, apart from

the work on mammals, the experimental designs of the studies listed above do

not distinguish between masking of the trailing call and the inability to locate

the trailing call even though it is easily heard (i.e. it is not masked). To distin-

guish between these alternatives, psychoacoustic studies will be necessary. From

an evolutionary perspective, however, the selection pressures on males to produce

leading calls will be strong irrespective of the underlying mechanisms.

I was able to demonstrate a strong preference for leading calls in Broadley’s

painted reed frog H. m. broadleyi (Grafe, 1996a) and the savannah running frog

K. fusca (Grafe, 1999). In reed frogs, the preference by females for the leading call

was largely independent of sound pressure, underscoring the robustness of this

preference (see also Dyson & Passmore, 1988a). In the synchronously calling run-

ning frog, preference for both leading and trailing calls was observed depending

on the degree of overlap (Grafe, 1999). Females preferred leading calls when calls

overlapped by 75% and 90% but switched their preference to trailing calls at 10%

and 25% of overlap. Thus, males are selected to overlap the calls of neighbours;

however, they should not do so with high degrees of overlap. Interestingly, play-

back experiments also showed that males were able to initiate their calls sooner

than they actually do, suggesting that special mechanisms are involved that in-

hibit males from calling with high degrees of overlap. For the savannah running

frog, the adaptive significance of synchronous calling is explained, at least in part,

by the auditory preferences of females.

Whereas the preference for leading signals appears to be a basic design feature

of nervous systems and thus a constraining feature of receivers that males need

to attend to, the preference of females for trailing signals is likely to be a more

fine-tuned adaptation by receivers to specific signalling environments. It remains

unclear why female savannah running frogs prefer trailing calls at low degrees of

overlap.

A comparative analysis within the genus Kassina may provide some answers as

to how a species’environment, in particular the communication network in which

a population finds itself, influences signal design and signalling behaviour. Pre-

liminary female choice experiments suggest that the call-timing pattern of males

is also tuned to the respective preference functions of females in K. senegalensis

(T. U. Grafe & H. Lussow, unpublished data). However, call-timing patterns do not

correlate with habitat characteristics such as degree of cover or calling site. Fur-

ther comparative work needs to be done to elucidate the environmental correlates

of call timing in anurans.

It should be noted that the physical characteristics of the communication chan-

nel, the transmission properties of the environment and the network structure



(i.e. distance between network members) might select for different signals (Wiley

& Richards, 1978; Ryan et al., 1990; Staaden & Romer, 1997). Signal attenuation and

degradation through reverberations or irregular amplitude fluctuations are fac-

tors that limit the active space of signals (Wiley & Richards, 1978; see also Ch. 20).

These effects vary between habitats, with distance between sender and receivers

and with height above ground. Spectral components of the call are degraded least,

for example, if either sender or receivers are elevated (Wiley & Richards, 1978).

Ryan & Wilczynski (1991) demonstrated that differences in habitat characteris-

tics explained a large part of the variance in the frequency of advertisement calls

between populations of the chorus frog Acris crepitans (see Wiley (1991) for bird

examples). In several recent studies with anurans, however, none of the predicted

differences in call features was found between species from different habitats

(Penna & Solis, 1998; Kime et al., 2000). The evidence for the influence of transmis-

sion properties of the environment and network structure on signal design and

signalling behaviour in anurans is equivocal at best.

Comparing female choice in two-choice trials and in natural choruses

To identify which acoustic parameters are important in determining male

mating success, researchers traditionally use two-choice trials in which female

anurans are given the choice between two acoustic stimuli. Gravid female frogs

and toads readily phonotactically approach one of the speakers and will search for

the male on or in the speaker. Typically population-wide preferences of females

are then noted and inferences drawn about the importance of male acoustic traits

(Gerhardt, 1994). Regarded from a network perspective, such experiments should

be viewed with caution because females are being tested in very simplified envi-

ronments in which background noise is reduced to a minimum (see also Sullivan

et al., 1995). To illustrate this point, I will review three examples of how prefer-

ences demonstrated in simple arena trials may not translate into sexual selection

in natural populations.

The first example is from the detailed studies on the South African painted reed

frog Hyperolius marmoratus marmoratus by Neville Passmore and his colleagues. They

showed that females preferred lower frequency calls, suggesting that larger males

should have a mating advantage. This, however, was only the case when comparing

the mating success and body size of males in small choruses (Telford et al., 1989). In

large choruses, large males no longer had an advantage. Instead of preferring large

males, females tested in the field preferred males calling at high rates (Passmore

et al., 1992). Therefore, under noisy field conditions, female preference for calls of

lower frequency were overridden by preferences for call rate.

I made a similar observation when studying mate choice in Broadley’s painted

reed frog H. m. broadleyi. To determine female preferences in the small choruses


288 T. U. Grafe

in the field, I monitored male calling behaviour using an array of microphones

set around the periphery of small natural ponds (Grafe, 1997b). This technique of

passive sound localization (McGregor & Dabelsteen, 1996) enabled me to match

precisely each call on the recording to an individual male in the chorus. I moni-

tored both the spectral and the temporal components of each male’s calls while a

female, released from the edge of the pond, was choosing among males. The analy-

sis showed that call rate and proximity to the female release site were the best

predictors of male mating success. These preferences were corroborated in tradi-

tional two-choice trials. Interestingly, females also preferred medium-frequency

calls to high or low frequencies in two-choice trials, a preference not mani-

fested under noisy field conditions. Furthermore, call parameters of interest to

females are often intensity dependent: that is, preference for a call trait can be

reversed by increasing the intensity of an alternative stimulus (e.g. Arak, 1988;

Gerhardt, 1988).

The third example showing the importance of testing females under natural

conditions is the work on H. versicolor by Schwartz (2001). As mentioned above, fe-

males tested in arena choice trials prefer males producing longer calls even if they

call at a lower rate as long as calling effort remains the same. The preference for

long calls is non-linear, with strong discrimination against very short calls. Under

quiet conditions, the discrimination was remarkable. On average, females discrim-

inated in favour of calls on the basis of 1.5 pulses (out of 20). Background noise

played to females over an additional speaker reduced the ability to discriminate

to 2.3 pulses. Field experiments supported the importance of call duration in mat-

ing success; however, it was limited. An array of eight speakers was placed along

the perimeter of a pond and calls of varying duration and call rate were broad-

cast over many nights. Naturally arriving females were trapped at the speaker of

their choice. The extent of the preference for call duration was quite restricted,

with only the shortest call being discriminated against. Overall, the preference

for long calls explained less than 10% of the variation in male mating success in

the field.

These examples do not argue against the utility of two-choice trials but suggest

that they should not stand alone. In chorusing anurans, females have to make

decisions under acoustically unfavourable environments, often under the risk of

predation, and must, therefore, limit their choosiness. The utility of two-choice

trials comes into play when testing hypotheses generated by field observations or

experiments, as recently demonstrated by Murphy & Gerhardt (2002). They com-

bined field observations of mate sampling by female barking treefrogs Hyla gratiosa

with two- and three-speaker choice trials to determine the influence of distance

of calling males on female choice. As in Broadley’s painted reed frog, most female

barking treefrogs approached the closest male and mated with him. Evidence sug-

gests that these species simultaneously sample males. Such sampling is especially



vulnerable to background noise and will be limited by the perceptual abilities of

females (Murphy & Gerhardt, 2002). It should be noted, however, that females of

other anurans probably use sequential sampling techniques and, therefore, are

less influenced by background noise because they closely approach several males

(reviewed by Murphy & Gerhardt, 2002).

Eavesdropping by females

Do females extract information from male–male interactions that influ-

ences their mating decisions? Rephrasing this question in communication net-

work terminology: are females eavesdropping on male–male interactions (i.e. in-

dulging in social eavesdropping; Ch. 2)? The evidence that they are comes from

observing female choice for the relative timing of male advertisement calls. Like

bird song, the advertisement call of anurans is directed to both males and females

(i.e. it is not just a mating call). Females can potentially extract information from

how males interact. Chorusing male anurans adjust the timing of their advertise-

ment calls to that of neighbouring males in a competitive way (Klump & Gerhardt,

1992; Schwartz, 2001). As outlined above, in most species, females prefer the calls

of leading males, thereby often overriding their preferences for other call param-

eters (e.g. Dyson & Passmore, 1988b; Grafe, 1996a). In a few cases, females prefer

follower calls (Schwartz & Wells, 1984; Grafe, 1999).

Evidence that females eavesdrop on male–male interactions requires simulta-

neous monitoring of males, i.e. that they show simultaneous mate choice. Good

evidence for simultaneous mate choice comes from species that approach males

only after spending some time, often several minutes, at the edge of breeding

ponds and from the relative preferences of females tested in two-choice trials

(Grafe, 1997b; Murphy & Gerhardt, 2002). However, a convincing study of social

eavesdropping would need to show that females are not just approaching the first

male they can distinguish from the background noise, in most cases this would be

the nearest male. In H. m. broadleyi, females based their choice not only on nearby

males but also on male call rate (Grafe, 1997b). Likewise, female Hyla gratiosa did

not just approach the first male they could distinguish from the background noise

(Murphy & Gerhardt, 2002). Such observations and experimental evidence suggests

that females monitor the calling behaviour of more than one male. Two-choice

trials have shown that females prefer males that jam the calls of other males.

For social eavesdropping to occur, one need not assume high cognitive abilities.

In fact, the proximate mechanisms underlying female preference for leading or

follower calls may not even require the involvement of higher auditory centres.

Eavesdropping by males

Potentially, anuran choruses provide ample opportunities for males to

eavesdrop on the interactions between other males or between males and females.


290 T. U. Grafe

0

5

10

15

20

25

30

0 2.5 5 7.5 10 12.5 15 17.5 20

Cal

l rat

e (c

alls

/0.5

min

)

Time (min)

Fig. 13.2. Call rates of seven males (open and closed symbols) in a small chorus over

20 minutes. Note that the ‘opportunistic male’ (closed symbol) is not calling until

1.25 minutes before he went into amplexus (at 19.8 minutes). Since call rates were

high at other times during the recording, other cues, such as seeing the female, in

addition to high call rates are likely to have directed the attention of the

‘opportunistic male’ to the female.

Males of many species increase their call rate substantially when approached by

a female and this can lead to local interactions with heightened activity between

males. Other silent or satellite males in the vicinity may make use of such infor-

mation by approaching this chorus area and attempting to intercept the female

or attract females by vigorously starting to call themselves. I recorded three cases

of such ‘opportunistic’ calling in Broadley’s painted reed frog using a microphone

array (Grafe, 1995b). These males were silent throughout most of the recording

period (3, 19, 20 minutes) and started calling vigorously only after a female had

been introduced into the chorus and was moving towards other calling males. In

all three cases, males approached the area of heightened activity and were success-

ful in attracting the female’s attention and mating with her. One case is shown in

Fig. 13.2. Documenting the behaviour of non-signalling individuals is particularly

difficult and may be one of the reasons why little is known about ‘silent’ (eaves-

dropping) strategies. This underscores the utility of using microphone arrays to

record chorusing activity because the absence of calling can be documented using

this technique.

Interceptive eavesdropping (Ch. 2) is known to occur in some anurans. In spade-

foot toads and green frogs, for example, satellite males associate with speak-

ers that produce attractive advertisement calls (Pfennig et al., 2000; Gerhardt &

Huber, 2002), suggesting that these males are monitoring the activity of other



males and making adaptive decisions. Such behaviour is probably more common

than reported.

In anurans, in which females give acoustic responses to male advertisement

calls (e.g. midwife toads, see above), it would be interesting to see if either sex

uses these male–female interactions to interfere with the courting pair. For exam-

ple, female Australian bushcrickets Elephantodeta nobilis give acoustic responses to

male advertisement calls (Bailey & Field, 2000). Males that are probably satellites

are attracted by these duets and produce advertisement calls, thereby occasionally

attracting these females themselves. Eavesdropping on male–female vocal interac-

tions may occur in Alytes spp. as well. Two studies report eavesdropping in captive

Alytes obstreticans in which females competed for male parental care by approach-

ing vocalizing pairs, attempting to block the path of other females and displacing

amplectant males (Verrell & Brown, 1993; Grafe et al., 1999).

Indirect evidence for eavesdropping comes from observing the signal type and

signalling intensity used during communication in a network environment. Priva-

tizing an interaction is a likely consequence of eavesdropping (Ch. 3). Many male

anurans have distinct courtship calls that are quieter than their advertisement

calls (reviewed by Wells, 1988). In E. coqui, for example, males use these quiet

courtship calls to lead females to oviposition sites on the forest floor (Townsend &

Stewart, 1986). Since females are not being mate guarded (i.e. are not in amplexus),

it is important for the male to prevent interference by other males.

Another point of interest is that anuran advertisement signals are generally

omnidirectional (e.g. Passmore, 1981), probably because males cannot predict from

where females approach. Spherical spreading, however, facilitates eavesdropping.

One would predict that courtship calls should be more directional; however, sound

fields of courtship calls have not been measured in anurans.


Aggregations of calling frogs and toads are characterized by high levels of

background noise. The common problems of communicating in a noisy environ-

ment, with its variety of conflicting selection pressures that act on both signallers

and receives, have led to a diversity of solutions. Masking interference, for exam-

ple, is reduced in most species by adjusting the timing of signals to avoid overlap.

In some species, however, overlap increases as males compete to become more

attractive to females.

Many features of communicating in networks, such as signal timing interac-

tions, selective attention and simultaneous mate choice, have been relatively well

studied in anurans. Other specific signalling behaviours, predicted from commu-

nication network theory, such as social eavesdropping (Ch. 14), audience effects


292 T. U. Grafe

(Ch. 4) and victory displays (Ch. 6), have generally not been considered or investi-

gated explicitly. Such behaviours would be predicted to occur in species that have

several encounters with the same individuals, such as in many territorial ranid

frogs that show individual recognition (e.g. Bee & Gerhardt, 2002) or in dendro-

batid frogs with year-round territorial behaviour and complex patterns of parental

care (e.g. Summers, 1992).

Social eavesdropping is likely to be a general phenomenon of anuran choruses.

Identifying additional cases of social eavesdropping in anurans, as outlined above,

would provide a fruitful avenue for future studies and would further highlight that

anuran choruses are complex communication networks. An open question in this

context is the functional significance of female preferences for leading (or lagging)

advertisement calls. Determining any indirect and direct benefits females may

obtain from their choice of leading males would be highly desirable. In addition,

studies that demonstrate social eavesdropping through aggressive interactions

would be of interest.

The relative ease with which phonotaxis can be induced in females in the lab-

oratory and the availability of sound synthesis software for the production of

synthetic signals that can be constructed with signal parameters varying indepen-

dently of each other have diverted attention from testing female preferences in

the chorus. New techniques, such as multiple channel recordings as well as mi-

crophone and speaker arrays, will undoubtedly contribute to our understanding

of patterns of male vocal competition and female choice. More observational data

on female sampling behaviour would also contribute to revealing how receivers

deal with complex signalling networks (e.g. Murphy & Gerhardt, 2002).

The perceptual basis of communication in noisy environments remains largely

unexplored (Ch. 20). It seems likely that receivers group sounds into auditory

streams in order to improve recognition and to assign them to individual signallers

(i.e. auditory scene analysis: Feng & Ratnam 2000; Hulse 2002). A recent study

by Farris et al. (2002) demonstrated auditory grouping in female tungara frogs

in which the whine and chuck are grouped together even when presented from

widely different directions. Understanding how auditory systems group incoming

signals or signal components will help to explain how animals communicate in

noisy networks and how they achieve selective attention; it also has the potential to

provide a mechanistic basis for understanding the evolution of multicomponent

signals.

Anurans offer unique opportunities to study communication networks. Males

of many species aggregate in choruses of varying size and complexity. Advertise-

ment calls of anurans are long-range signals used to attract females and repel rival

males. Investigations are facilitated by the species-specific and highly stereotyped

signals as well as by the generally small signal repertoires. The consequences of



strategic decisions can readily be observed because fertilization is generally

external, making paternity analysis unnecessary. Consequently, the evolutionary

consequences of communicating in network environments can be assessed with

relative ease. In addition, anurans occupy a wide variety of habitats and use a

variety of reproductive strategies. This diversity provides numerous opportunities

for comparative analyses.

Acknowledgements

I thank Peter McGregor and two anonymous reviewers for their helpful comments on

a previous version of this chapter. My work on running frogs in West Africa was supported by the

Deutsche Forschungsgemeinschaft (Gr 1584).

References

Arak, A. 1988. Female mate selection in the natterjack toad: active choice or passive

attraction? Behavioral Ecology and Sociobiology, 22, 317–327.

Backwell, P., Jennions, M., Passmore, N. & Christy, J. 1998. Synchronized courtship in

fiddler crabs. Nature, 391, 31–32.

Bailey, W. J. & Field, G. 2000. Acoustic satellite behaviour in the Australian bushcricket

Elephantodeta nobilis (Phaneropterinae, Tettigonidae, Orthoptera). Animal

Behaviour, 59, 361–369.

Bee, M. A. & Bowling, A. C. 2002. Socially mediated pitch alteration by territorial male

bullfrogs, Rana catesbeiana. Journal of Herpetology, 36, 140–143.

Bee, M A. & Gerhardt, H. C. 2002. Individual voice recognition in a territorial frog

(Rana catesbeiana). Proceedings of the Royal Society of London, Series B, 269, 1443–1448.

Bosch, J. 2001. Female reciprocal calling in the Iberian midwife toad (Alytes cisternasii)

varies with male call rate and dominant frequency: implications for sexual

selection. Naturwissenschaften, 88, 434–437.

Bosch, J. & Marquez, R. 2002. Female preference function related to precedence effect

in an amphibian anuran (Alytes cisternasii): tests with non-overlapping calls.


Bradbury, J. W. 1981. The evolution of leks. In: Natural Selection and Social Behavior, ed. R.

D. Alexander & D. W. Tinkle. New York: Chiron Press, pp. 138–169.

Brenowitz, E. A. & Rose, G. J. 1999. Female choice and plasticity of male calling

behaviour in the Pacific treefrog. Animal Behaviour, 57, 1337–1342.

Brenowitz, E. A., Rose, G. J. & Alder, T. 2000. The neuroethology of acoustic

communication in Pacific treefrogs. In: Anuran Communication, ed. M. J. Ryan.

Washington, DC: Smithsonian Institute Press, pp. 145–155.

Brush, J. S. & Narins, P. M. 1989. Chorus dynamics of a Neotropical amphibian

assemblage: comparison of computer simulation and natural behaviour. Animal



294 T. U. Grafe

Cranford, J. & Oberholzer, M. 1976. Role of neocortex in binaural hearing in the cat. II:

The ‘precedence effect’ in sound localization. Brain Research, 111, 225–239.

Deutsch, J. C. 1994. Uganda kob mating success does not increase on larger leks.


Dyson, M. L. & Passmore, N. I. 1988a. Two-choice phonotaxis in Hyperolius marmoratus

(Anura: Hyperoliidae): the effect of temporal variation in presented stimuli.


1988b. The combined effect of intensity and the temporal relationship of stimuli on

phonotaxis in female painted reed frogs Hyperolius marmoratus. Animal Behaviour,

36, 1555–1556.

Enquist, M., Ghirlanda, S. & Hurd, P. L. 1998. Discrete conventional signalling of a

continuous variable. Animal Behaviour, 56, 749–754.

Farris, H. E., Rand, A. S. & Ryan, M. J. 2002. The effects of spatially separated call

components on phonotaxis in tungara frogs: evidence for auditory grouping.

Brain, Behavior and Evolution, 60, 181–188.

Feng, A. S. & Ratnam, R. 2000. Neural basis of hearing in real-world situations. Annual

Review of Psychology, 51, 699–725.

Feng, A. S., Narins, P. M. & Xu, C.-H. 2002. Vocal acrobatics in a Chinese frog, Amolops

tormotus. Naturwissenschaften, 89, 352–356.

Fleischack, P. C. & Small, C. P. 1978. The vocalizations and breeding behaviour of

Kassina senegalensis (Anura, Rhacophoridae) in summer breeding aggregations.

Koedoe, 21, 91–99.

Forester, D. C. & Harrison, W. K. 1987. The significance of antiphonal vocalisation by

the spring peeper, Pseudacris crucifer (Amphibia, Anura). Behaviour, 103, 1–15.

Gerhardt, H. C. 1988. Acoustic properties used in call recognition by frogs and toads.

In: The Evolution of the Amphibian Auditory System, ed. B. Fritzsch, M. J. Ryan, W.

Wilczynski, T. E. Hetherington & W. Walkowiak. New York: John Wiley,

pp. 455–483.

1994. The evolution of vocalization in frogs and toads. Annual Review of Ecology and

Systematics, 25, 293–324.



Gerhardt, H. C. & Klump, G. M. 1988. Masking of acoustic signals by the chorus

background noise in the green treefrog: a limitation on mate choice. Animal

Behaviour, 36, 1247–1249.

Gerhardt, H. C. & Schwartz, J. J. 1995. Interspecific interactions in anuran courtship.

In: Amphibian Biology: Social Behaviour, ed. H. Heatwole & B. K. Sullivan. Chipping

Norton, UK: Surrey Beatty, pp. 601–632.

Given, M. F. 1987. Vocalizations and acoustic interactions of the carpenter frog, Rana

virgatipes. Herpetologica, 43, 467–481.

Grafe, T. U. 1995a. Graded aggressive calls in the African painted reed frog Hyperolius

marmoratus (Hyperoliidae). Ethology, 101, 67–81.

1995b. Sexual selection and communication in the African painted reed frog

(Hyperolius marmoratus). Ph.D. Thesis, Cornell University, Ithaca, NY.



1996a. The function of call alternation in the African reed frog Hyperolius

marmoratus: precise call timing prevents auditory masking. Behavioral Ecology and


1996b. Energetics of vocalization in the African reed frog (Hyperolius marmoratus).

Comparative Biochemistry and Physiology A, 114, 235–243.

1997a. Use of metabolic substrates in the gray treefrog, Hyla versicolor: implications

for calling behavior. Copeia, 1997, 356–362.

1997b. Costs and benefits of mate choice in the lek-breeding reed frog, Hyperolius

marmoratus. Animal Behaviour, 53, 1103–1117.

1999. A function of synchronous calling and a novel female preference shift in an

anuran. Proceedings of the Royal Society of London, Series B, 266, 2331–2336.

2003. Synchronised interdigitated calling in the Kuvangu running frog (Kassina

kuvangensis). Animal Behaviour, 66, 127–136.

Grafe, T. U. & Thein, J. 2001. Energetics of calling and metabolic substrate use during

prolonged exercise in the European treefrog Hyla arborea. Journal of Comparative

Physiology B, 171, 69–76.

Grafe, T. U., Spieler, M. & Konig, B. 1999. Soziobiologische Erklarungsansatze

des Verhaltens von Amphibien und Reptilien. Zeitschrift fur Feldherpetologie, 6,

1–42.

Grafen, A. 1990. Biological signals as handicaps. Journal of Theoretical Biology, 144,

517–546.

Greenfield, M. D. 1994a. Cooperation and conflict in the evolution of signal

interactions. Annual Review of Ecology and Systematics, 25, 97–126.

1994b. Synchronous and alternating choruses in insects and anurans: common

mechanisms and diverse functions. American Zoologist, 34, 605–615.

Greenfield, M. D. & Rand, A. S. 2000. Frogs have rules: selective attention algorithms

regulate chorusing in Physalaemus pustulosus (Leptodactylidae). Ethology, 106,

331–347.

Greenfield, M. D. & Roizen, I. 1993. Katydid synchronous chorusing is an evolutionary

stable outcome of female choice. Nature, 364, 618–620.



1355–1361.

Hulse, S. 2002. Auditory scene analysis in animal communication. Advances in the Study

of Behavior, 31, 163–200.

Ibanez, D. R. 1993. Female phonotaxis and call overlap in the Neotropical glassfrog

Centrolenella granulosa. Copeia, 1993, 846–850.

Johnston, R. E., Sorokin, E. S. & Ferkin, M. H. 1997. Female voles discriminate males’


Johnstone, R. A. 1994. Honest signalling, perceptual error and the evolution of

‘all-or-nothing’ displays. Proceedings of the Royal Society of London, Series B, 256,

169–175.

Kelly, J. B. 1974. Localization of paired sound sources in the rat: small time differences.



296 T. U. Grafe

Kime, N. M., Turner, W. R. & Ryan, M. J. 2000. The transmission of advertisement calls

in Central American frogs. Behavioral Ecology, 11, 71–83.

Klump, G. M. 1996. Bird communication in the noisy world. In: Ecology and Evolution of

Acoustic Communication in Birds, ed. D. E. Kroodsma & E. H. Miller. Ithaca, NY:


Klump, G. M. & Gerhardt, H. C. 1987. Use of non-arbitrary acoustic criteria in mate

choice by female gray tree frogs. Nature, 326, 286–288.

1992. Mechanisms and function of call-timing in male–male interactions in frogs.

In: Playback and Studies of Animal Communication, ed. P. K. McGregor. New York:


Littlejohn, M. J. & Harrison, P. A. 1985. The functional significance of the diphasic

advertisement call of Geocrinia victoriana (Anura: Leptodactylidae). Behavioral


Lopez, P. T., Narins, P. M., Lewis, E. R. & Moore, S. W. 1988. Acoustically induced call

modification in the white-lipped frog, Leptodactylus albilabris. Animal Behaviour, 36,

1295–1308.

Marshall, V. T., Humfeld, S. C. & Bee, M. A. 2003. Plasticity of aggressive signalling in

male spring peepers (Pseudacris crucifer). Animal Behaviour, 65, 1223–1234.






Moore, S. W., Lewis, E. R., Narins, P. M. & Lopez, P. T. 1989. The call-timing algorithm of

the white-lipped frog, Leptodactylus albilabris. Journal of Comparative Physiology A,

164, 309–319.

Murphy, C. G. & Gerhardt, H. C. 2002. Mate sampling by female barking treefrogs (Hyla

gratiosa). Behavioral Ecology, 13, 472–480.

Narins, P. M. 1982. Behavioral refractory period in neotropical treefrogs. Journal of

Comparative Physiology A, 148, 337–344.

Narins, P. M. & Capranica, R. R. 1978. Communicative significance of the two-note call

of the treefrog Eleutherodactylus coqui. Journal of Comparative Physiology A, 127, 1–9.

Narins, P. M. & Zelick, R. 1988. The effects of noise on auditory processing and

behavior in amphibians. In: The Evolution of the Amphibian Auditory System, ed. B.

Fritzsch, T. Hetherington, M. Ryan, W. Wilczynski & W. Walkowiak. New York:

John Wiley, pp. 455–483.

Narins, P. M., Lewis, E. R. & McClelland, B. E. 2000. Hyperextended call note repertoire

of the endemic Madagascar treefrog Boophis madagascariensis (Rhacophoridae).

Journal of Zoology, 250, 283–298.

Passmore, N. I. 1981. Sound levels of mating calls of some African frogs. Herpetologica,

37, 166–171.

Passmore, N. I., Bishop, P. J. & Caithness, N. 1992. Calling behaviour influences mating

success in male painted reed frogs, Hyperolius marmoratus. Ethology, 92, 227–241.



Penna, M. & Solis, R. 1998. Frog call intensities and sound propagation in the South

American temperate forest region. Behavioral Ecology and Sociobiology, 42, 371–381.

Pfennig, K. S., Rapa, K. & McNatt, R. 2000. Evolution of male mating behavior: male

spadefoot toads preferentially associate with conspecific males. Behavioral Ecology

and Sociobiology, 48, 69–74.

Robinson, D. 1990. Acoustic communication between the sexes in bushcrickets. In: The

Tettigoniidae: Behavior, Systematics, Evolution, ed. W. J. Bailey & D. C. F. Rentz.

Bathurst, Australia: Crawford House Press, pp. 110–129.

Romer, H., Bailey, W. J. & Dadour, I. 1989. Insect hearing in the field. III. Masking by

noise. Journal of Comparative Physiology A, 164, 609–620.

Ryan, M. J. 1985. The Tungara Frog. Chicago, IL: University of Chicago Press.

1986. Synchronized calling in a treefrog (Smilisca sila). Brain, Behavior and Evolution,

29, 196–206.

Ryan, M. J. & Keddy-Hector, A. 1992. Directional patterns of female mate choice and

the role of sensory biases. American Naturalist, 139, S4–S35.

Ryan, M. J. & Wilczynski, W. 1991. Evolution of intraspecific variation in the

advertisement call of a cricket frog (Acris crepitans). Biological Journal of the Linnean

Society, 44, 249–271.

Ryan, M. J., Cocroft, R. B. & Wilczynski, W. 1990. The role of environmental selection in

intraspecific divergence of mate recognition signals in the cricket frog, Acris

crepitans. Evolution, 44, 1869–1872.

Schneider, H., Hoermann, G. & Hodl, W. 1988. Calling and antiphonal calling in four

Neotropical anuran species of the family Leptodactylidae. Zoologische Jahrbucher

Physiologie, 92, 77–103.

Schwartz, J. J. 1989. Graded aggressive calls of the spring peeper, Pseudacris crucifer.

Herpetologica, 45, 172–181.

1993. Male calling behavior, female discrimination and acoustic interference in the

Neotropical treefrog Hyla microcephala under realistic acoustic conditions.


2001. Call monitoring and interactive playback systems in the study of acoustic

interactions among male anurans. In: Anuran Communication, ed. M. J. Ryan.


Schwartz, J. J. & Wells, K. D. 1984. Interspecific acoustic interactions of the neotropical

treefrog Hyla ebraccata. Behavioral Ecology and Sociobiology, 14, 211–224.

Schwartz, J. J., Buchanan, B. W. & Gerhardt, H. C. 2001. Female mate choice in the gray

treefrog (Hyla versicolor) in three experimental environments. Behavioral Ecology


Shelly, T. E. & Greenfield, M. D. 1991. Dominions and desert clickers (Orthoptera:

Acrididae): influences of resources and male signaling on female settlement

patterns. Behavioral Ecology and Sociobiology, 28, 133–140.

Staaden, M. J. v. & Romer, H. 1997. Sexual signalling in bladder grasshoppers: tactical

design for maximizing calling range. Journal of Experimental Biology, 200,

2597–2608.


298 T. U. Grafe

Sullivan, B. K., Ryan, M. J. & Verrell, P. A. 1995. Female choice and mating system

structure. In: Amphibian Biology: Social Behaviour, ed. H. Heatwole & B. K. Sullivan.

Chipping Norton, UK: Surrey Beatty, pp. 469–517.

Summers, K. 1992. Dart-poison frogs and the control of sexual selection. Ethology, 91,

89–107.

Taigen, T. L., Wells, K. D. & Marsh, R. L. 1985. The enzymatic basis of high metabolic

rates in calling frogs. Physiological Zoology, 58, 719–726.

Telford, S. R., Dyson, M. L. & Passmore, N. I. 1989. Mate choice occurs only in small

choruses of painted reed frogs (Hyperolius marmoratus). Bioacoustics, 2, 47–53.

Townsend, D. S. & Stewart, M. M. 1986. Courtship and mating behavior of a Puerto

Rican frog, Eleutherodactylus coqui. Herpetologica, 42, 165–170.

Tuttle, M. D. & Ryan, M. J. 1982. The role of synchronized calling, ambient light, and

ambient noise, in anti-bat-predator behavior of a treefrog. Behavioral Ecology and


Vencl, F. V. & Carlson, A. D. 1998. Proximate mechanisms of sexual selection in the

firefly Photinus pyralis (Coleoptera: Lampyridae). Journal of Insect Behavior, 11,

191–207.

Verrell, P. A. & Brown, L. E. 1993. Competition among females for mates in a species

with male parental care, the widwife toad Alytes obstetricans. Ethology, 93, 247–257.

Walkowiak, W. 1992. Acoustic communication in the fire-bellied toad: an integrative

neurobiological approach. Ethology, Ecology and Evolution, 4, 63–74.

Wallach, H., Newman, E. B. & Rosenzweig, M. R. 1949. The precedence effect in sound

localization. American Journal of Psychology, 62, 315–336.

Wells, K. D. 1988. The effect of social interactions on anuran vocal behavior. In: The

Evolution of the Amphibian Auditory System, ed. B. Fritzsch, M. J. Ryan, W. Wilczynski,

T. E. Hetherington & W. Walkowiak. New York: John Wiley,

pp. 433–454.

1989. Vocal communication in a Neotropical treefrog, Hyla ebraccata: responses of

males to graded aggressive calls. Copeia, 1989, 461–466.

2001. The energetics of calling in frogs. In: Anuran Communication, ed. M. J. Ryan.


Wells, K. D. & Schwartz, J. J. 1984. Vocal communication in a Neotropical treefrog, Hyla

ebraccata: advertisement calls. Animal Behaviour, 32, 405–420.

Wells, K. D. & Taigen, T. L. 1986. The effect of social interactions on calling energetics

in the gray treefrog (Hyla versicolor). Behavioral Ecology and Sociobiology, 19, 9–18.

Widemo, F. & Owens, I. P. F. 1995. Lek size, male mating skew and the evolution of

lekking. Nature, 373, 148–151.

Wiley, R. H. 1991. Associations of song properties with habitat for territorial oscine

birds of eastern North America. American Naturalist, 138, 973–993.




Wyttenbach, R. A. & Hoy, R. R. 1993. Demonstration of the precedence effect in an

insect. Journal of the Acoustical Society of America, 94, 777–784.



Zelick, R. D. & Narins, P. M. 1983. Intensity discrimination and the precision of call

timing in two species of neotropical treefrogs. Journal of Comparative Physiology A,

153, 403–412.

1985. Characterization of the advertisement call oscillator in the frog

Eleutherodactylus coqui. Journal of Comparative Physiology A, 156, 223–229.

P1: IRK/KFO P2: IRK-KOD0521823617c14.xml CU1917B/McGregor 0 521 582361 7 April 11, 2005 17:1

14

Singing interactions in songbirds:implications for social relationsand territorial settlement

m a rc n ag u i b

University of Bielefeld, Germany

Introduction

Interactions between individuals make up a significant part of life in so-

cial animals. They form a crucial behavioural mechanism establishing and main-

taining particular spacing patterns among individuals and groups of individuals

and are inherent in the regulation of social relations. Animals interact with each

other in a broad range of contexts, such as during intersexual competition, mate

choice, or parent–offspring communication, but still many of the underlying prin-

ciples share common ground (Hauser, 1996; Bradbury & Vehrencamp, 1998). It is

well documented that the performance of individuals in interactions has profound

implications for the resolution of conflicts over resources, such as mates, food or

space. Interactions may consist of complex behavioural displays or may be based

exclusively on signals in either one or several signalling modalities. Vocal interac-

tions are among the most conspicuous forms of interactions and have been well

studied in several taxonomic groups, such as insects, anurans and birds (Bradbury

& Vehrencamp, 1998). In birds, vocal interactions are most evident in parent–

offspring communication (Kilner & Johnstone, 1997; Ch. 9), calling and singing in

group-living species (Farabaugh & Dooling, 1996; Zann, 1996), duetting in tropi-

cal songbirds (von Helversen, 1980; Farabaugh, 1982) and in singing interactions

between male territorial songbirds (Todt & Naguib, 2000).

In this review, I will focus on singing interactions in male territorial song-

birds. Their vocal interactions are among the most striking examples of bird

vocal communication and are an established model for studies on territoriality

and communication networks (McGregor, 1993; McGregor & Dabelsteen, 1996;


300


Singing interactions in songbirds 301

Todt & Naguib, 2000). Male songbirds commonly hold adjacent territories forming

neighbourhoods; so usually several males sing within signalling range of each

other. The typical pattern of settlement and spacing of individuals of the same

species thus sets the framework for the evolution of communication behaviour,

including the evolution of complex patterns of vocal interactions and the strate-

gies for gathering information on conspecifics.

Singing interactions between males take place in a variety of different contexts

and the information exchanged may strongly depend on the singers’ social and

spatial relations. They take place during immediate competition over resources

such as space or mates in addition to being a conspicuous component in the social

relations between established territorial neighbours. In general, vocal interactions

differ from the classical song traits such as singing activity (Kempenaers et al.,

1997; Gil et al., 1999; Amrhein et al., 2002, 2004), singing versatility (Hasselquist

et al., 1996; Searcy & Yasukawa, 1996) or other performance-related traits (Podos,

1996; Forstmeier et al., 2002). These classical traits can be regarded as ‘individual’

traits that are present regardless of the social context in which a male is singing

(Fig. 14.1). Vocal interactions, in contrast, have an additional interactive dimension

as the message conveyed depends strongly on the pattern of song interchange

between the interacting singers (Todt & Naguib, 2000). This interactive dimension

has resulted in considerable current research interest in vocal interactions, as the

performance of singers during an interaction provides immediate information on

relative differences between them: information that also is used by eavesdropping

individuals in a communication network (McGregor & Dabelsteen, 1996).

My principal goal in this chapter is to integrate current knowledge on strategies

of vocal interactions in territorial songbirds with concepts of territorial behaviour

and territorial settlement (Waser & Wiley, 1980; Stamps, 1994; Stamps & Krishnan,

2001) and to explore how recent advances in studies on vocal interactions con-

tribute to our understanding of the social relations among neighbouring territo-

rial songbirds. The social and spatial relationships among neighbouring males can

be mediated by their vocal interactions; consequently, vocal interactions can have

profound implications for the evolution of strategies for territorial settlement

and spacing behaviour in general. I will also evaluate how principles of vocal in-

teractions contribute to our understanding of the evolution of singing strategies

and the evolution of receivers’adaptations to gather information from conspecific

signalling behaviour. A key trait of vocal interactions responsible for much of the

interest is that they are commonly asymmetric in the sense that each of the singers

involved uses its songs differently in relation to those of its counterpart. These

asymmetries can reflect differences in motivation or quality among singers and

consequently provide information about the relationship between them (Todt &

Naguib, 2000). Therefore, this chapter will consider the internal aspects of


302 M. Naguib

Pathways of information gathering

Social traits

Vocal interactions (e.g. asymmetries in vocal matching, or in timing of songs)

Song structure

Diurnal singing activity

Singing rate

Individual song traits

Singing versatility

Extracting information from individual signals Extracting information from interactions and their asymmetries (eavesdropping)

Fig. 14.1. Singing traits and pathways of information gathering. Receivers can gather

information from individual traits during and in the absence of interactions.

Information gathering by extracting information from signalling interactions

(eavesdropping) provides additional information that cannot be extracted from

individual song traits. In interactions, additional social factors come into play such as

the relative relation of the songs, which can provide immediate information on

differences between singers.

interactions (i.e. their function in the interaction between the singers) as well as

their external implications (i.e. their wider importance as a source of information

for other listening (eavesdropping) individuals; see also Ch. 2).

After reviewing some recent studies of vocal interactions and by drawing several

examples from our own studies on nightingales Luscinia megarhynchos (reviewed

in more detail by Todt & Naguib, 2000), I will evaluate more closely the general

association of social and spatial relationships between males and their strategies in

vocal interactions, as well as their strategies in gathering information by attending

to others’ vocal interactions.

Nature of interactions

Interactions in communication can be defined as the exchange of in-

formation through signals by at least two individuals where the signals of both

signallers have some direct relation to each other. In acoustic communication,

interactions can be best determined when there are two individuals signalling;

such interactions are referred to commonly as dyadic interactions. When several

individuals are signalling, they may form a communication network with highly

complex modes of interactions (see Chs. 13 and 15) but this need not be the case



(Shackleton & Ratcliffe, 1994). Although interactions pragmatically can be defined

broadly as instances when two or more singers are singing at the same time, true

interactions are best identified by demonstrating that the singers influence each

other in their choice and timing of song types or in other aspects of their singing

strategy. Such a definition is analogous to our understanding of dialogues in hu-

man speech and allows us to extract and define specific strategies and to study

their causations and evolutionary implications.

Song matching

Most songbirds have song repertoires, allowing them flexibility in deci-

sions on the next song to be sung (Kroodsma, 1982). The most conspicuous way of

using specific song types during an interaction is song matching, a situation in

which a singer replies with a song of the same type as the preceding song sung

by the opponent. Such matching of signals is also found in other taxa (Ch. 18) and

can give insights into the functions of signalling strategies as well as addressing

mechanistic questions such as how animals perceive and categorize signals (Falls

et al., 1982, 1988; Weary et al., 1990; Naguib et al., 2002). Song matching is known to

be used to address a particular rival and often it appears to function as an aggres-

sively directed signal to increase the level of threat towards a specific rival (Krebs

et al., 1981; McGregor et al., 1992; Nielsen & Vehrencamp, 1995; Vehrencamp, 2001).

However, several studies have shown that song matching is not always a strong

aggressive signal and may be used as a graded signal of intent (Searcy et al., 2000;

Burt et al., 2002; Naguib et al., 2002). Moreover, males may sometimes match some,

but not all, features of songs. Males may match only parts of a song or certain song

parameters such as the frequency of the full song (Otter et al., 2002) or specific song

components (Burt et al., 2002; Naguib et al., 2002), duration of songs (Weary et al.,

1990) or categories of song (Wiley et al., 1994; Naguib et al., 2002). Neighbouring

song sparrows Melospiza melodia have been shown to match repertoires by reply-

ing with non-matching songs that are shared with the singing opponent (Beecher

et al., 1996). The meaning of matching also may vary with the distance between

singers, the general context, the song type or even the specific song component

that is matched. In nightingales, we discovered patterns of song matching that

clearly differed from the most widespread principle that song matching increases

with the level of perceived threat (Naguib et al., 2002). In playbacks conducted

on nightingales’ nocturnal song, males increased the precision of matching the

pitch of whistles in so-called whistle songs (Fig. 14.2) with increasing distance

to the simulated unfamiliar opponent but not the overall rate of matching the

song category. Matching whistle songs and specifically the pitch of the whistles

may have a different biological significance than matching ‘normal’ songs. The

narrow spectral bandwidth of the whistles implies that they transmit with less


304 M. Naguib

2 4 6 s

2

4

6

8

10kHz

2 4 s

2

4

6

8

10kHz

1 12

21

(a)

(b)

Fig. 14.2. Spectrographic examples of whistle song matching and overlapping in

nightingales. Hatched arrows indicate beginning of songs, numbers indicate different

males. (a) The whistle song in the centre from male 2 is a non-overlapping, full song

type match of the first song of male 1, i.e. the whistle part and terminal sections are

the same in both songs and there is no noticeable overlap of songs. Male 1 replied

immediately to match and clearly overlap his opponent. (b) The two whistle songs

match but also differ in frequency so that the whistle parts do not mask each

other.

spectral degradation over long distances and thus may function particularly in

long-range signalling (Wiley & Richards, 1978; Slabbekoorn et al., 2002). At first

glance, more matching at long distance is puzzling as the social importance and,

therefore, the urgency of addressing a rival is assumed to decrease with interindi-

vidual spacing. However, even distant neighbours that do not share a territorial

boundary are part of the same neighbourhood, in which males have to establish

and maintain social and spatial relations that are likely to be regulated through

long-range vocal interactions. Radio-tracking data have shown that males make

substantial excursions into territories of direct neighbours and even more dis-

tant ones within auditory range from the own territory (Hanski, 1992; Chandler

et al., 1997; Pitcher & Stutchbury, 2000; Naguib et al., 2001), indicating that the

social and spatial relations between males go beyond their immediate neighbours

with whom territorial boundaries are shared. Long-distance matching, therefore,

may be a mechanism involved directly in establishing and maintaining spatial

relations between males.



In summary, the extensive research on song matching (reviewed more fully

by Todt & Naguib, 2000) shows that matching can do more than simply address a

rival; it can have several functions, depending on the social context, the territorial

relationships of the interacting males and the song type or its specific parameters

that are matched. It will be interesting to see how the refinements of playback

design and playback technologies (e.g. Dabelsteen, 1992) will address more subtle

questions on the function of song matching in different contexts and thus provide

more detailed insights into the kinds of message conveyed.

Timing of songs

In a singing interaction, the relative timing of song production by the

interactants can signal specific information on the singer’s state and intention,

such as its readiness to escalate the contest (Todt & Naguib, 2000). The relative

timing of songs during an interaction differs from song matching in two ways.

First, song sharing or detailed knowledge of a rival’s repertoire is not required.

Second, relative song timing can vary continuously whereas song matching is more

categorical (i.e. matching occurs or does not occur) even though recent research

has emphasized that song matching can be subtle with graded components and

that it is not restricted to matching full song types, as discussed above.

Despite the continuous nature of relative song timing, two categories of timing

of songs have been shown to occur to date (Hultsch & Todt, 1982) and to have func-

tionally different signal value: song alternating and song overlapping (Brindley,

1991; Dabelsteen et al., 1996; Naguib et al., 1999; Langemann et al., 2000). Song al-

ternating is a common strategy where males take turns in delivering their songs.

This strategy can also be observed in concurrent interspecific singing where males

avoid acoustic competition (Ficken et al., 1974). During song overlapping, in con-

trast, males begin to sing a song before the opponent has ended its song. Song

overlapping has been shown in several species to function as a directed agonistic

signal (Brindley, 1991; McGregor et al., 1992; Dabelsteen et al., 1996, 1997; Naguib,

1999) whereas song alternating is the seemingly predominating singing strategy

during less-intense contexts. Interestingly, song overlapping is not only treated

as an agonistic signal by the singer whose songs are overlapped but also is used

by eavesdropping males and females to assess differences in the relative quality

or motivation in two interacting males (Naguib & Todt, 1997; Naguib et al., 1999;

Otter et al., 1999; Peake et al., 2001, 2002; Mennill et al., 2002), as discussed below

(see also Ch. 2). Song overlapping can result in considerable masking of part of the

rival’s song, but this effect is not inherent in song overlapping. Masking effects

will depend on the distance between singers (i.e. their relative difference in am-

plitude), the amount of song overlapped in terms of duration and the similarity

in phonology between the overlapping parts of the two songs (Fig. 14.2; see also


306 M. Naguib

Ch. 20). In addition, the extent of perceived masking will vary with whose percep-

tion is considered: the singer whose songs are overlapped, the overlapper or other

individuals, which may be at any relative location to the two singers (see below). As

a result, benefits through masking the opponent’ssongs may be limited to certain

conditions and locations and, despite the attractiveness of the argument, it is un-

likely to be the primary consequence that led to the evolution of song overlapping

as a singing strategy in long-distance interactions. Although overlap will affect

detection and recognition of subtle sound features (Wiley, 1983, 1994) by other lis-

tening conspecifics, spatial release from masking (Klump, 1996) is a compensatory

mechanism that can help in coping with problems resulting from masking. Todt

and Naguib (2000) further suggested that a male that is overlapping the songs

of its opponent might benefit by shifting the attention of eavesdroppers to the

overlapper. Finally, despite such a range of basic effects of song overlap on sig-

nal perception, song overlapping may also be a conventional signal of dominance

that is maintained by retaliation costs if overlapping increases the probability to

escalate a contest.

Song overlapping and song alternating have been shown to be of biological

significance, but the issue of how much overlap needs to be achieved in order to

accomplish a certain function remains to be studied. Similarly, the precise timing

of songs during alternating singing and also during song overlapping in natural

conditions may have specific signal value (McGregor et al., 1992), another issue

that deserves to be explored in more detail in future studies. For instance, some

studies suggest that song alternating is not a homogeneous strategy but that the

exact timing during alternating is also of functional significance. Specifically,

leader–follower relationships, in which the follower sings soon after the leader,

have been interpreted as the leader representing the more dominant singer (Smith

& Norman, 1979; Popp, 1989; Naguib et al., 1999). Gathering information on very

fine temporal differences in timing of songs will require knowledge of the dis-

tance to the opponent (Naguib & Wiley, 2001) and eavesdroppers will require to

know the distance to each interactant because of the different time delay of the

songs originating from two sources at unequal distance. However, confusion over

whether or not a song is overlapping can occur only in long-range interactions in

two extremes: when both singers start their song at about the same time (so that

each singer will perceive the opponent’s song as overlapping) or when an overlap-

ping song sets on late during the song that is overlapped (so that the opponent

may receive it as non-overlapping). It remains to be studied if overlapping events

that fall into this ‘confusion range’are interpreted differently from unambiguous

events. In nightingales, overlapping songs most commonly fall outside these con-

fusion ranges (Hultsch & Todt, 1982; Naguib, 1999); consequently, in most cases

overlapping is an unambiguous event.



Matching and timing

In the preceding sections, we have seen that both song matching and

song timing are characteristic of bird vocal interactions and in many cases are

related to a specific meaning of singing. However, the possibility that both may

be dependent on each other (Todt, 1981; Wolffgramm & Todt, 1982) has been little

explored from a functional perspective. There are several possible combinations

of these two aspects. For example, males may match and overlap a rival’s song, as

nightingales frequently do when matching whistle songs (Fig. 14.2), or they may

match a song with varying delays during alternating singing.

Matching songs during boundary disputes often occurs in interactions with

high song rates and thus short delays in responses. Matching with short delays

may be of specific value in signalling the willingness to escalate a contest. When

males interact where no immediate dispute is apparent, such as during the dawn

chorus or with low song rates in long-range interactions between neighbours that

sing at the same time of day or night, matching may be timed differently and,

therefore, may have a different meaning and consequence for the social relations

between singers. One possibility is that frequent matching between established

territory holders acts to repel non-territorial males seeking to establish a territory

(Amrhein et al., 2004) (thus benefiting both territorial males) by signalling long-

term territory tenure and an established spatial and social relationship. There is

evidence that males sharing song types have longer territory tenure (Beecher et al.,

2000) and neighbouring males often share more songs than non-neighbouring

males (Kroodsma, 1974; Hultsch & Todt, 1981; McGregor & Krebs, 1982; Schroeder

& Wiley, 1983; Beecher, 1996; Payne, 1996; Beecher et al., 2000; Griessmann &

Naguib, 2002): both features that make song matching more likely. An increase

in song sharing over the season, as shown for thrush nightingales Luscinia luscinia

(Sorjonen, 1987), may increase the probability of matching during vocal interac-

tions as the season progresses. If so, matching during long-range vocal interactions

such as the thrush nightingales’ nocturnal song may function to strengthen ter-

ritorial residency in neighbouring males rather than being an agonistic signal.

Vocal interactions and social relationships among singers

Songbirds interact with song in at least five different social contexts that

need to be considered when singing strategies and their evolutionary implications

are studied (Fig. 14.3). Territorial males interact by song with; (a) neighbours

over long distances when there is no immediate dispute noticeable (Fig. 14.3a);

(b) neighbours in immediate disputes over territorial boundaries or possibly over

access to females (Fig. 14.3b); (c) neighbours that have crossed the shared territo-

rial boundary (Fig. 14.3c); (d) unfamiliar rivals that have intruded into the territory


308 M. Naguib

Fig. 14.3. Five contexts of vocal interactions between male songbirds. Territories are

represented by elliptical shapes. (a) Long-range interaction between established

neighbours (filled and open circles); (b) boundary interaction between established

neighbours with both males singing in their own territory; (c) interaction in which

the resident male of the right-hand territory has intruded; (d) interaction between a

territory holder and a non-territorial stranger (hatched circle) that has intruded into

the territory; (e) interaction between a territory holder and a distant stranger

(hatched circle) that may attempt to establish a territory nearby.



and start vocally claiming part of it (Fig. 14.3d); or (e) unfamiliar rivals that attempt

to establish a territory nearby without directly threatening the resident’sterritory

(Fig. 14.3e). Singing strategies in these different contexts are under different selec-

tion pressures as the social context differs; accordingly, the implications of specific

singing strategies will vary in these situations. In encounters between a resident

male and an intruder, asymmetries in site-specific dominance are inherent, as

the payoff for each male differs because of the prior investment of the resident

male in establishing and maintaining a territory before the contest (Maynard

Smith & Parker, 1976; Waser & Wiley, 1980). Interactions with strangers are often

single and time-limited events and, therefore, males clearly have to signal their

strength and should signal a higher readiness to escalate the contest. Territorial

residents are more likely to win a contest than intruders, so it is adaptive for

residents to invest more in the interaction (Pusey & Packer, 1997). Interactions

among neighbours, in contrast, are repeated and it may pay males to use a differ-

ent singing strategy. Moreover, the asymmetries in site-specific dominance that

is evident in all encounters between residents and intruders does not apply to

neighbour–neighbour interactions, provided that both are singing from within

their territories at locations that are not under direct dispute. The exact ways

males interact with their neighbours will depend, therefore, on their locations

and on previous experience of the dyad (Wiley & Wiley, 1980). Interactions be-

tween neighbours, as between residents and intruders, are still characterized by

asymmetries, presumably as a result of inherent differences among males, such as

differences in age, duration of prior residency or mating status. Remaining asym-

metries in status between territorial neighbours can then well be reflected in the

way they interact with each other vocally. Moreover, males may develop specific

expectations when interacting with specific neighbours because of the specific

ontogenetic trajectory of their relationship; consequently a given singing strat-

egy may have a different functional significance when used with neighbours than

when it is used with strangers. Biologically, the interactions among neighbours

are particularly interesting as they presumably reflect the social relationships be-

tween them and, therefore, provide deeper insights into the territorial and social

system of songbirds.

Experiments that focused explicitly on vocal interactions have simulated a

stranger’s or neighbour’s intrusion into a territory (or appearance near the ter-

ritory), simulating the contexts illustrated in Fig. 14.3d,e (Todt & Naguib, 2000).

Vocal interactions in such high-intensity contexts have served as an important

experimental model to unravel the functions of specific singing strategies during

an interaction. In general, such situations are characterized by high song rates,

song overlap, song matching and song switching (in species in which males usu-

ally repeat the same song type several times before switching to a different one).


310 M. Naguib

Depending on locations of simulated intrusions, territorial males distinguish be-

tween neighbours and strangers in their responses (Falls, 1982; Stoddard, 1996).

Such individual recognition is the prerequisite for different response strategies

with specific neighbours and new rivals and raises issues about how interactions

vary in dynamics and functions with the familiarity of the singers and their spatial

and social relationship.

The information on functions and principles of interactions obtained from

these playback experiments simulating intrusions are likely to be of general value

and applicability also to long-range interactions among established neighbours

that are interacting when singing on their own territory. However, it is impor-

tant to consider that neighbours can be expected to exchange much more subtle

information in vocal interactions in the absence of an immediate dispute. Neigh-

bours having prior experience with each other may be better at using nuances

in variation of singing patterns, such as song rate, quality of sound production

or use of specific song variants or song types. Communication among established

males may thus reach a much higher level of complexity with higher cognitive de-

mands than communication among unfamiliar males, where disputes are driven

by more immediate contests in specific contexts over specific resources. More de-

scriptive and experimental studies on long-range vocal interactions in the absence

of immediate disputes will be needed to test these ideas further.

Functions of vocal interactions in territorial defence against intruders

The functions of vocal interactions among residents and intruders are ob-

vious as there is an immediate conflict over space. Vocal interactions here make

up a significant fraction of the behaviour during such conflicts, underlying their

importance in spacing behaviour. Asymmetries in these interactions may be an im-

portant predictor for subsequent behaviour over the spatial conflict and, therefore,

may set the stage for the occurrence of subsequent and intermittent movements

and the probability of physical encounters.

Vocal interactions commonly escalate in intensity in immediate disputes over

territorial boundaries. Consequently, most intense interactions can be observed

early in the season when territories are established, whenever boundaries are vi-

olated by neighbours, and when males attempt to establish a new territory in an

area with males that have been resident for some time. In these situations, terri-

tory holders are highly aroused and attempt to drive the rival from the disputed

area by intense singing and an interactive strategy that signals high readiness

for escalation. Commonly, males then sing at a high song rate and interactions

are characterized by frequent song overlapping and high rates of song matching

or song switching, as discussed above. Most experimental research on the func-

tion of singing during vocal contests has used playback simulating an unfamiliar



intruder and demonstrated that song matching, song switching (not reviewed

here) and song overlapping are strategies used and perceived as an agonistic sig-

nalling behaviour, as discussed above. Therefore, these singing variables are likely

to determine the outcome of a contest over space. Future playback experiments

that focus on consequences of song overlap and song matching in terms of choice

of song posts by the opponent will be important to address this issue in more

detail. Males may avoid singing at posts where they provoke intense responses,

for example being overlapped (e.g. Todt, 1981) and challenged by high song rates,

and retreat earlier when their singing evokes such responses by resident males.

Functions of vocal interactions among neighbouring males

Songbirds sing extensively in the phase when territories are established

but continue to interact vocally when conflicts over space become less intense,

that is when territories appear to be established. The functions of these contin-

uing vocal interactions must have a different evolutionary significance, as their

outcome is less likely to have drastic effects in conflicts over space. Territorial

neighbours exhibit site-specific dominance; therefore, their vocal interactions are

not associated with spatial asymmetries unless intrusions take place. In these sit-

uations, when vocal interactions are unlikely to function to resolve immediate

conflicts over space, such interactions are more likely to function in maintaining

a spacing pattern and keeping remaining asymmetries of the territorial neigh-

bours at an equilibrium that avoids conflicts in which no clear winner is likely to

emerge. Information on conspecifics will be imperfect; consequently, neighbour-

ing males may need to continuously update their information on neighbours to

refine their assessment. Therefore, after the spatial arrangement in a territorial

system becomes established, vocal interactions occur most frequently between fa-

miliar neighbouring males singing from their own territories in the absence of im-

mediate boundary disputes. However, long-range interactions among neighbours

are much less well studied than interactions in high-intensity contexts (Kramer &

Lemon, 1983; Kramer et al., 1985).

Factors that are important to consider when assessing the function of specific

singing strategies are that basic principles of singing are likely to depend on (a) the

specific prior relationship of the singers, (b) the current singers’ relationship and,

(c) the expected future relationship. Many of the basic principles of singing, such as

song matching and strategies of song timing, may have the same function regard-

less of the singers’ specific relationship as long as they reflect the singers’ internal

states. However, the interpretation of a singing strategy may differ depending on

whether or not the singers are familiar with each other: whether or not they are ter-

ritorial neighbours. Neighbouring males often already have substantial previous

experience with each other and have to expect a long-term relationship. Singing


312 M. Naguib

strategies during their interactions may be different as the strategy is an integral

component of the previous relationship and that expected in the future. Thus,

males may be able to code information more subtly in choice of song patterns or

in the timing of songs. Males who have established dominance over rivals through

previous interactions may not escalate substantially during subsequent interac-

tions. Rather a few occurrences of song overlap (for instance) may suffice to signal

alertness or the readiness to escalate a contest. Therefore, the function and the

consequences of singing strategies during vocal interactions among neighbours

may depend not only on the current singing strategy but also on how it relates to

the strategy in previous interactions. By changing song posts during vocal inter-

actions with neighbours, males may probe each other. If males avoid song posts at

which they encounter repeated vocal aggression, not only territory boundaries but

also the choice of song posts may be determined by neighbours’ singing behaviour

during vocal interactions. Overall, more descriptive studies on the nature of vocal

interactions between established males in relation to their spatial relationships

will be needed to answer questions on how vocal interactions reflect and affect

the social relations between males in long-term spatial relationships.

Vocal interactions and territorial settlement

The role of vocal interactions in territorial settlement is particularly inter-

esting when singing strategies during vocal interactions reflect the males’qualities

or their motivation to defend a particular space. Although the function of song

in territorial defence is well established once a male has an established territory

(Krebs et al., 1978; Nowicki et al., 1998; Naguib et al., 2001), there has been little

discussion of how song and vocal interactions determine spatial relations of males

during territorial settlement or when territories shift in the course of the breeding

season. Interpretations that particular singing strategies reflect a winner (Peake

et al., 2001, 2002) can apply to single interactions, but single interactions may

not necessarily reflect the overall relationship between the singers. For example,

the singing behaviour of individuals establishing territories may differ from that

when they are defending an established territory. The terms winner and loser in

competition over resources usually imply that one individual gets all (or has first

access) and the other gets nothing (retreats). Stamps and Krishnan (1997, 1999,

2001) pointed out that during territorial settlement the consequences of winning

and losing contests are much more complex. During territorial establishment, in-

dividuals are dividing up space and such division of space usually involves repeated

interactions from different locations. Therefore, the outcome of vocal interactions

may be determined by the net outcome of repeated interactions at different lo-

cations rather than a single interaction. This situation presumably predominates



during establishment of territories in areas that are not yet saturated and where

both interacting males can expect to succeed in establishing a territory in a par-

ticular area. In some instances, males may even divide up space without a definite

winner. By shifting song posts, space may be divided up passively when males

avoid song posts in which they elicit high-intensity interactions with their rival.

In this way, losing an interaction in territorial conflicts does not necessarily mean

that losers fully retreat but rather that they shift song posts and they may be the

winner at a different location. The final spatial arrangement of singing territo-

ries may be determined by the pattern of repeated vocal interactions with males

singing at different song posts. Vocal interactions between established neighbours

may reveal information on remaining differences among them that does not re-

sult from asymmetries in site-dependent dominance, where each male may be the

winner at a certain location. In nightingales, males may systematically ‘win’ re-

peated interactions with a particular neighbour, but this pattern is not true for all

neighbouring males (M. Naguib, unpublished data). If recurring asymmetries exist

in the interactions between particular males, this may reflect a stable dominance

relationship that is maintained after space is divided up.

Vocal interactions in communication networks

Vocal interactions have received particular attention in recent years as

their asymmetries have been shown to be used by other individuals as a source

of information. Vocal interactions in songbirds are a clearly defined signalling

context and so have become one of the main models in studies of communica-

tion networks (McGregor & Dabelsteen, 1996; McGregor & Peake, 2000; Whitfield,

2002). To date, several studies have shown that male songbirds eavesdrop on rival

vocal interactions and have expanded the understanding of information gather-

ing when several individuals are within signalling rage of each other (Naguib &

Todt, 1997; Naguib et al., 1999; Otter et al., 1999; Peake et al., 2001, 2002; Mennill

et al., 2002). In our own studies, we showed in two-loudspeaker experiments that

territorial nightingales discriminated between asymmetries in vocal interactions:

subjects responded significantly more strongly to a simulated rival that was over-

lapping the songs of the opponent, i.e. was the more aggressive intruder (Naguib &

Todt, 1997). When songs in the interaction simulated by the loudspeakers did not

overlap each other but were played in an alternating order with songs of one loud-

speaker leading (closely followed by the songs of the other loudspeaker), subjects

responded more strongly to the loudspeakers playing the leading songs (Naguib

et al., 1999). These combined experiments indicate that different proximate cues

were used depending on the kind of asymmetry simulated. The use of oppos-

ing proximate cues by subjects depending on the kind of asymmetry perceived


314 M. Naguib

provides evidence that subjects use an adaptive strategy that is independent of

a putative general proximate shift of attention to a first or last heard stimuli.

Peake and coworkers (2001) elegantly expanded this playback design and showed

that great tits Parus major attended to such asymmetries in interactions played

to them from outside their territory and then responded differently to subse-

quent intrusions, depending on which previous singer was simulated as intruder.

They further showed that males varied their song output in responses to intrud-

ers depending on the kind of experience they had with the intruder prior to an

interaction with another male (Peake et al., 2002). This suggests more complex

ways of gathering information than shown in any of the previous experiments

on eavesdropping. Studies by Otter et al. (1999) and Mennill et al. (2002) indicated

that females also use asymmetries in male–male interactions as sources of infor-

mation in their responses and seemingly even in reproductive decisions (see also

Ch. 7). Therefore, it is well documented that songbirds not only attend to vocal in-

teractions between males but also extract information coded in the asymmetries

in singing strategies and use the information adaptively. An interesting question

to be answered in order to understand the further implications of eavesdropping

in communication networks with widely spaced individuals is whether birds are

able accurately to extract meaningful asymmetries in an interaction occurring at

a distance, or whether they are only able to do so when they are close to, or at

equal distance from, the singers. We already have clear evidence that vocal inter-

actions have much wider implications than the exchange of information between

interacting males. Strategies of singing during vocal interactions then are likely

not only to evolve through responses of the opponent but also through effects on

other listening individuals (Ch. 2). Viewing communication from the perspective

of communication networks considerably broadens the view of social implica-

tions of the prevalence of vocal interactions and the functional implications of

certain singing strategies during vocal interactions. Most experimental studies of

the functional significance of different singing strategies during an interaction

have used different playback protocols and there is a need to complement these

studies by more descriptive studies on natural interactions.


Vocal interactions in songbirds have many facets that need to be inte-

grated into models of information gathering in communication. Because of the

accessibility of song, vocal interactions are a suitable model to address wider con-

cepts in communication, to address questions in cognitive ecology (Ch. 24) and

to obtain new insights into the social relationships between territorial neigh-

bours. Singing strategies during vocal interactions and strategies for information

gathering from vocal interactions by participants and by eavesdroppers have



evolved under ecological constraints, such as the spacing patterns of conspecifics

or more specifically the abundance and distribution of signallers in space and

time. As we continue to gain more information on strategies of vocal interactions

in natural settings among established neighbours and during establishment of

territories, we are likely to obtain new insights into how social relations are me-

diated by song and how strategies of singing interactions are related to spatial

ecology.

Acknowledgements

I thank Peter McGregor, Ken Otter and an anonymous referee for helpful comments on

a previous version of the manuscript

References

Amrhein, V., Korner, P. & Naguib, M. 2002. Nocturnal and diurnal singing activity in

the nightingale: correlations with mating status and breeding cycle. Animal

Behaviour, 64, 939–944.

Amrhein, V., Kunc, H. P. & Naguib, M. 2004. Non-territorial nightingales prospect

territories during the dawn chorus. Proceedings of the Royal Society of London, Series B,

(Supplement), 271, S167–S169.

Beecher, M. D. 1996. Bird song learning in the laboratory and field. In: Ecology and

Evolution in Acoustic Communication in Birds, ed. D. E. Kroodsma & E. H. Miller.



matching between neighbouring song sparrows. Animal Behaviour, 51, 917–923.

Beecher, M. D., Campbell, S. E. & Nordby, J. C. 2000. Territory tenure in song sparrows

is related to song sharing with neighbours, but not to repertoire size. Animal




Brindley, E. L. 1991. Response of European robins to playback of song: neighbour

recognition and overlapping. Animal Behaviour, 41, 503–512.

Burt, J. M., Bard, S. C., Campbell, S. E. & Beecher, M. D. 2002. Alternative forms of song

matching in song sparrows. Animal Behaviour, 63, 1143–1151.

Chandler, C. R., Ketterson, E. D. & Nolan, V. 1997. Effects of testosterone on use of

space by male dark-eyed juncos when their mates are fertile. Animal Behaviour, 54,

543–549.

Dabelsteen, T. 1992. Interactive playback: a finely tuned response. In: Playback and

Studies of Animal Communication, ed. P. K. McGregor. New York: Plenum Press,

pp. 97–109.

Dabelsteen, T., McGregor, P. K., Shepherd, M., Whittaker, X. & Pedersen, S. B. 1996. Is

the signal value of overlapping different from that of alternating during matched

singing in great tits? Journal of Avian Biology, 27, 189–194.


316 M. Naguib

Dabelsteen, T., McGregor, P. K., Holland, J., Tobias, J. A. & Pedersen, S. B. 1997. The

signal function of overlapping singing in male robins. Animal Behaviour, 53,

249–256.

Falls, J. B. 1982. Individual recognition by sounds. In: Acoustic Communication in Birds,

Vol. 2, ed. D. E. Kroodsma & E. H. Miller. New York: Academic Press, pp. 237–278.

Falls, J. B., Krebs, J. R. & McGregor, P. K. 1982. Song matching in the great tit (Parus

major): the effect of similarity and familiarity. Animal Behaviour, 30, 997–1009.

Falls, J. B., Horn, A. G. & Dickinson, T. E. 1988. How western meadowlarks classify their

songs: evidence from song matching. Animal Behaviour, 36, 579–585.

Farabaugh, S. M. 1982. The ecological and social significance of duetting. In: Acoustic

Communication in Birds, Vol. 2, ed. D. E. Kroodsma & E. H. Miller. New York:


Farabaugh, S. M. & Dooling, R. J. 1996. Acoustic communication in parrots: laboratory

and field studies of budgerigars, Melopsittacus undulatus. In: Ecology and Evolution of



Ficken, R. W., Ficken, M. S. & Hailman, J. P. 1974. Temporal pattern shifts to avoid

acoustic interference in singing birds. Science, 183, 762–763.

Forstmeier, W., Kempenaers, B., Meyer, A. & Leisler, B. 2002. A novel song parameter

correlates with extra-pair paternity and reflects male longevity. Proceedings of the


Gil, D., Graves, J. A. & Slater, P. J. B. 1999. Seasonal patterns of singing in the willow

warbler: evidence against the fertility announcement hypothesis. Animal

Behaviour, 58, 995–1000.

Griessmann, B. & Naguib, M. 2002. Song sharing in neighboring and non neighboring

thrush nightingales (Luscinia luscinia) and its implications for communication.

Ethology, 108, 377–387.

Hanski, I. K. 1992. Territorial behaviour and mixed reproductive strategy in the

chaffinch. Ornis Scandinavica, 23, 475–482.

Hasselquist, D., Bensch, S. & von Schantz, T. 1996. Correlation between male song

repertoire, extra-pair paternity and offspring survival in the great reed warbler.

Nature, 381, 229–232.


Hultsch, H. & Todt, D. 1981. Repertoire sharing and song-post distance in nightingales

(Luscinia megarhynchos B.). Behavioral Ecology and Sociobiology, 8, 183–188.

1982. Temporal performance roles during vocal interactions in nightingales

(Luscinia megarhynchos). Behavioral Ecology and Sociobiology, 11, 253–260.

Kempenaers, B., Verheyren, G. R. & Dhondt, A. A. 1997. Extrapair paternity in the blue

tit (Parus caeruleus): female choice, male characteristics, and offspring quality.


Kilner, R. & Johnstone, R. A. 1997. Begging the question: are offspring solicitation

behaviours signals of need? Trends in Ecology and Evolution, 12, 11–15.

Klump, G. 1996. Bird communication in the noisy world. In: Ecology and Evolution of





Kramer, H. G. & Lemon, R. E. 1983. Dynamics of territorial singing between

neighboring song sparrows (Melospiza melodia). Behaviour, 85, 198–223.

Kramer, H. G., Lemon, R. E. & Morris, M. J. 1985. Song switching and agonistic

stimulation in the song sparrow (Melospiza melodia). Animal Behaviour, 33, 135–149.

Krebs, J. R., Ashcroft, R. & Webber, M. 1978. Song repertoires and territory defense in

the great tit Parus major L. Nature, 271, 539–542.

Krebs, J. R., Ashcroft, R. & van Orsdol, K. 1981. Song matching in the great tit Parus

major L. Animal Behaviour, 29, 918–921.

Kroodsma, D. E. 1974. Song learning, dialects and dispersal in the Bewick’s wren.


1982. Song repertoires: problems in their definitions and use. In: Acoustic

Communication in Birds, Vol. 1, ed. D. E. Kroodsma & E. H. Miller. New York:


Langemann, U., Tavares, J. P., Peake, T. M. & McGregor, P. K. 2000. Response of great

tits to escalating patterns of playback. Behaviour, 137, 451–471.

Maynard Smith, J. & Parker, G. A. 1976. The logic of asymmetric contests. Animal

Behaviour, 24, 159–175.







McGregor, P. K. & Krebs, J. R. 1982. Mating and song types in the great tit. Nature, 297,

60–61.



McGregor, P. K., Dabelsteen, T., Shepherd, M. & Pedersen, S. B. 1992. The signal value of

matched singing in great tits: evidence from interactive playback experiments.



contests in songbirds. Science, 296, 873–873.

Naguib, M. 1999. Effects of song overlapping and alternating on nocturnally singing

nightingales. Animal Behaviour, 58, 1061–1067.



Naguib, M. & Wiley, R. H. 2001. Estimating the distance to a source of sound:

mechanisms and adaptations for long-range communication. Animal Behaviour,

62, 825–837.


leaders in simulated dyadic interactions. Proceedings of the Royal Society of London,

Series B, 265, 537–542.

Naguib, M., Altenkamp, R. & Griessmann, B. 2001. Nightingales in space: song and

extra-territorial forays of radio tagged song birds. Journal fur Ornithologie, 142,

306–312.


318 M. Naguib

Naguib, M., Mundry, R., Hultsch, H. & Todt, D. 2002. Responses to playback of whistle

songs and normal songs in male nightingales: effects of song category, whistle

pitch, and distance. Behavioral Ecology and Sociobiology, 52, 216–223.

Nielsen, B. M. B. & Vehrencamp, S. L. 1995. Responses of songs sparrows to song-type

matching via interactive playback. Behavioral Ecology and Sociobiology, 37, 109–117.

Nowicki, S., Searcy, W. A. & Hughes, M. 1998. The territory defense function of song in

song sparrows: a test with the speaker occupation design. Behaviour, 135, 615–628.

Otter, K., McGregor, P. K., Terry, A. M. R. et al., 1999. Do female great tits (Parus major)



Otter, K. A., Ratcliffe, L., Njegovan, M. & Fotheringham, J. 2002. Importance of

frequency and temporal song matching in black-capped chickadees: evidence

from interactive playback. Ethology, 108, 181–191.

Payne, R. B. 1996. Song traditions in indigo buntings: origin, improvisation, dispersal,

and extinction in cultural evolution. In: Ecology and Evolution of Acoustic

Communication in Birds, ed. D. E. Kroodsma & E. H. Miller. Ithaca, NY: Cornell

University Press, pp. 198–220.




Peake, T. M., Terry, A. M. R., McGregor, P. K. & Dabelsteen, T. 2002. Do great tits assess



Pitcher, T. E. & Stutchbury, B. J. M. 2000. Extraterritorial forays and male parental care

in hooded warblers. Animal Behaviour, 59, 1261–1269.

Podos, J. 1996. Motor constraints on vocal development in a songbird. Animal

Behaviour, 51, 1061–1070.

Popp, J. W. 1989. Temporal aspects of singing interactions among territorial ovenbirds

(Seiurus aurocapillus). Ethology, 82, 127–133.

Pusey, A. E. & Packer, C. 1997. The ecology of relationships. In: Behavioural Ecology, an

Evolutionary Approach, ed. J. R. Krebs & N. B. Davies. Oxford: Blackwell Scientific,

pp. 254–283.

Schroeder, D. J. & Wiley, R. H. 1983. Communication with shared song themes in

tufted titmice. The Auk, 100, 414–424.




Searcy, W. A., Nowicki, S. & Hogan, C. 2000. Song type variants and aggressive context.


Shackleton, S. A. & Ratcliffe, L. 1994. Matched counter-singing signals escalation of

aggression in black-capped chickadees (Parus atricapillus). Ethology, 97, 310–316.

Slabbekoorn, H., Ellers, J. & Smith, T. B. 2002. Bird song and sound transmission: the

benefits of reverberations. The Condor, 104, 564–573.

Smith, D. G. & Norman, D. O. 1979. ‘Leader–follower’ singing in red-winged blackbirds.

The Condor, 81, 83–84.



Sorjonen, J. 1987. Temporal and spatial differences in traditions and repertoires in the

song of the thrush nightingale (Luscinia luscinia) birds. Behaviour, 102, 196–212.

Stamps, J. 1994. Territorial behavior: testing the assumptions. Advances in the Study of

Behavior, 23, 173–232.

Stamps, J. A. & Krishnan, V. V. 1997. Functions of fights in territory establishment.


1999. A learning-based model of territory establishment. Quarterly Review of Biology,

74, 291–318.

2001. How territorial animals compete for divisible space: a learning-based model

with unequal competitors. American Naturalist, 157, 154–169.

Stoddard, P. K. 1996. Vocal recognition of neighbors by territorial passerines. In:

Ecology and Evolution of Acoustic Communication in Birds, ed. D. E. Kroodsma & E. H.


Todt, D. 1981. On functions of vocal matching: effects of counter-replies on song post

choice and singing. Zeitschrift fur Tierpsychologie, 57, 73–93.

Todt, D. & Naguib, M. 2000. Vocal interactions in birds: the use of song as a model in

communication. Advances in the Study of Behaviour, 29, 247–296.

Vehrencamp, S. L. 2001. Is song-type matching a conventional signal of aggressive

intentions? Proceedings of the Royal Society of London, Series B, 268, 1637–1642.

von Helversen, D. 1980. Structure and function of antiphonal duets. Acta XVIIIth

International Ornithological Congress, Berlin, 682–688.

Waser, P. M. & Wiley, R. H. 1980. Mechanisms and evolution of spacing in animals. In:

Handbook of Behavioral Neurobiology: Social Behavior and Communication, ed. P. Marler

& J. G. Vandenbergh. New York: Plenum Press, pp. 159–233.

Weary, D., Falls, J. B. & McGregor, P. K. 1990. Song matching and perception of song

types in great tits, Parus major. Behavioral Ecology, 1, 43–47.

Wiley, R. H. 1983. The evolution of communication: information and manipulation.

In: Animal Behaviour : Communication, Vol. 2, ed. T. R. Halliday & P. J. B. Slater.

Oxford: Blackwell Scientific, pp. 156–189.

1994. Errors, exaggeration and deception in animal communication. In: Behavioral

Mechanisms in Ecology, ed. L. Real. Chicago, IL: University of Chicago Press,

pp. 157–189.




Wiley, R. H. & Wiley, M. S. 1980. Territorial behavior of a blackbird: mechanisms of site

dependent dominance. Behaviour, 73, 130–154.

Wiley, R. H., Godard, R. & Thompson, A. D. 1994. Use of two singing modes by hooded

warblers as adaptations for signaling. Behaviour, 129, 243–278.

Whitfield, J. 2002. Nosy neighbours. Nature, 419, 242–243.

Wolffgramm, J. & Todt, D. 1982. Pattern and time specificity in vocal responses of

blackbirds Turdus merula. Behaviour, 81, 264–286.

Zann, R. A. 1996. The Zebra Finch: a Synthesis of Field and Laboratory Studies. Oxford:

Oxford University Press.


15

Dawn chorus as an interactivecommunication network

j o h n m . b u r t & s a n d r a l . v e h r e nc a m p

Cornell Laboratory of Ornithology, Ithaca, USA

Introduction

Dawn chorus singing is a striking behaviour pattern, performed by some

temperate-zone and tropical songbird species, as well as a few non-passerine and

non-avian species. In a typical chorusing songbird species, all territorial males in

a neighbourhood synchronously start singing 30 to 90 minutes before sunrise.

During the ensuing chorus period, song rate, singing diversity and song complex-

ity reach maximal levels, and often birds do not seem to be interacting with any

one particular neighbour (Hultsch & Todt, 1982). Then, as the light level increases

around sunrise, this mode of singing usually abruptly ends. Soon after dawn cho-

rus is over, birds begin to forage and patrol their borders, and they switch to

courtship singing or dyadic (i.e. paired) counter-singing with nearby neighbours.

Post-chorus singing is typically more sporadic and overall song rates tend to be

lower and much more variable than they are at dawn chorus (for a review of dawn

chorus behaviour, see Staicer et al. (1996)).

Numerous hypotheses have been proposed to explain dawn chorus singing.

In an insightful review, Staicer et al. (1996) outlined 12 non-exclusive hypotheses

and compared their predictions against the existing empirical evidence across

many songbird and non-songbird species. The 12 hypotheses were grouped into

three categories: intrinsic, environmental and social. Intrinsic explanations such

as circadian cycles of testosterone and self-stimulation are likely proximate mecha-

nisms for dawn singing (Wingfield & Farner, 1993; Goodson, 1998). Environmental

explanations, such as low predation risk, good acoustic transmission, inefficient

foraging with low ambient light and unpredictable night-time conditions lead-

ing to excess energy reserves on most mornings, provide reasons why singing at


320


Dawn chorus as an interactive network 321

dawn might be less costly than singing at other times (Henwood & Fabrick, 1979;

Kacelnik & Krebs, 1982; Mace, 1987; McNamara et al., 1987; Hutchinson, 2002;

Dabelsteen & Mathevon, 2002). However, none of these intrinsic or environmental

hypotheses provide a functional explanation of the selective advantage for vigor-

ous, continuous, complex vocal displays prior to sunrise, nor do they explain why

only some species exhibit this phenomenon.

The social hypotheses outlined by Staicer et al. (1996) that do attempt to provide

functional explanations for dawn chorus singing include mate attraction and/or

stimulation, territory defence and resolution of social dynamics. Predictions for

these alternatives have been used to support or reject certain hypotheses in several

species. For example, mate attraction and mate stimulation can be ruled out as

primary explanations in species where dawn chorus singing is uniformly high

across the breeding season, rather than being concentrated during periods of mate

attraction and mate fertility as these hypotheses predict (Kroodsma et al., 1989;

Part, 1991; Slagsvold et al., 1994). Moreover, males have generally been observed to

stop dawn chorus singing as soon as their mates emerge from their night roost,

and males of some species drop out of the dawn chorus altogether on days of

peak mate fertility, findings that do not support mate stimulation hypotheses

for those species (Mace, 1986; Cuthill & Macdonald, 1990; Part, 1991; Otter &

Ratcliffe, 1993). The territory defence hypothesis, suggesting that dawn song is an

extra vigorous keep-out signal, is contradicted in some sedentary species because

males cease dawn chorus singing during the non-breeding season even though

they continue to occupy and defend a territory (Staicer et al., 1996). Staicer et al.

concluded that the social hypothesis that best fits the existing evidence is the social

dynamics hypothesis, which proposes that the function of dawn chorus is the

interactive communication and adjustment of social relationships among males.

Their conclusion is based partly on findings in numerous species that dawn chorus

singers use signals and modes of signalling that are specifically associated with

male–male interaction and that dawn chorus singers of some species appear to be

listening to and directing their songs towards particular neighbours (Kroodsma

et al., 1989; Staicer, 1989; Nelson & Croner, 1991; Spector, 1991; Dabelsteen, 1992;

Staicer et al., 1996).

Females of at least some species may acquire information about mate qual-

ity from dawn singing. In several species, individual differences in dawn song

output are correlated with male age and/or dominance and with female laying

date and/or fecundity (Welling et al., 1995; Otter et al., 1997; Poesel et al., 2001;

Ballentine et al., 2003). Females could possibly assess the quality of singing males

at dawn chorus either by attending to song traits that are directly associated with

indicators of fitness, such as stamina, age or dominance, or by eavesdropping to


322 J. M. Burt & S. L. Vehrencamp

acquire information about relative quality if males are interacting (Otter et al.,

2001; Mennill et al., 2002; Ch. 7).

Taken as a whole, the pattern of dawn chorus behaviour across species does not

seem to fit any single currently hypothesized function. Furthermore, the social

hypotheses are not mutually exclusive, suggesting that the dawn chorus might

have multiple functions that may differ in relative importance, depending on

species. If that is the case, then it may be necessary to examine more closely (or

re-examine) the singing behaviour of each species that has a dawn chorus, pay-

ing attention to such factors as the intended receiver(s) and whether interactions

are occurring among singers within the chorus. Staicer et al. (1996) provided a set

of predictions for such evidence that could be helpful in distinguishing which

hypotheses might apply to a particular species (cf. Table 24.1 in Staicer et al.,

1996).

Investigation of these target and interaction issues would benefit from an ap-

proach that considers the neighbourhood of singing males as a communication

network. Network communication is broadly defined as the involvement of at least

three individuals, one or more of them signalling and all receiving (McGregor &

Dabelsteen, 1996; see other chapters in this volume). The dawn chorus, with many

simultaneous signallers (possibly interacting) and many potential receivers, cer-

tainly fits the broad definition of a communication network. However, to date,

no study has examined dawn chorus from a network perspective. We suggest

that a study of the characteristics of the communication network that occurs

at dawn chorus for a given species could provide further information about its

function. In this chapter, we discuss what kinds of communication network the

different functional hypotheses might predict for dawn chorus and how one might

go about testing for them. We then test some of these ideas using a dawn chorus

recording of a neighbourhood of banded wrens Thryothorus pleurostictus as an

example.

Communication network structures at dawn chorus

The detailed structure and complexity of communication networks can

vary and depend largely on the degree to which the signalling ‘links’between com-

municating individuals are one way or are interactive (i.e. signals flow both ways

between individuals). Using three hypothetical individuals (Fig. 15.1), it is possible

to define three basic network structures: broadcast networks, in which one sender

broadcasts a one-way omnidirectional signal to two receivers (Fig. 15.1a); eaves-

dropping networks, where two senders interact and a third receiver eavesdrops

on the interaction to obtain information about the interactants (Fig. 15.1b); and in-

teractive networks, with three senders all interacting with each other (Fig. 15.1c).



(a)

(b)

(c)

Fig. 15.1. Three basic network components may occur within a communication

network, either singly or in combination: (a) broadcast networks, in which at least one

sender produces undirected one-way signals that are received by potentially many

receivers; (b) eavesdropping networks, in which two signallers interact and

eavesdroppers obtain relative information about each interactant; (c) interactive

networks containing three or more individuals signalling interactively to one another

and eavesdropping on nearby interactions.

Any real-world communication network is likely to include many more than just

three individuals, and networks can theoretically consist of one or any combina-

tion of these basic components, adding more potential variety and complexity

to network structure. For example, it is likely that any bird-song network with

an interactive component probably also has an eavesdropping component in the

form of non-interacting listeners such as females and floater males, as well as

interacting males who are eavesdropping on their neighbours’ interactions.



Investigating the structure of a dawn chorus communication network might

provide useful clues about its function, with a key variable being the degree to

which signallers within the network are interactive. For example, omnidirectional

non-interactive signalling at dawn would indicate a broadcast network. A broad-

cast network at dawn chorus would support male keep-out, female attraction or

direct male quality-assessment functions, since these hypotheses do not strictly

require male–male interaction. Conversely, evidence of two-way or multi-way sig-

nalling between singers at dawn would indicate an interactive network, support-

ing the notion that the dawn chorus serves an inter-male communication role

(the social dynamics hypothesis). Evidence of eavesdropping would also support a

relative male quality-assessment function for dawn chorus.

Searching for interactions at dawn

The presence or absence of an interactive network may be a key diagnostic

for the function of dawn chorus, but on a practical level how do we go about

looking for interactions at dawn? One possible first step would be to examine

what we know about how males in a given dawn-chorusing species use their songs

to communicate during the daytime. Indeed, much effort has been devoted to

analysing daytime counter-singing interactions between pairs of adjacent males.

Observational studies of daytime dyadic counter-singing between focal birds and

a neighbour or intruder are relatively easy to conduct, and playback experiments

have been used to test hypotheses for the function of male song interactions. In

territorial species, song has generally been found to function as a keep-out signal to

other males (Krebs et al., 1978; Yasukawa et al., 1982; Nowicki et al., 1998). Studies

of daytime singing have found that song can also be used in complex ways to

mediate aggression between neighbours. For example, in populations with high

levels of song-type sharing between neighbours, birds can match their neighbour’s

song with their own version of that type. Matching is particularly useful as a

directed signal, since by replying with the same type a bird can unambiguously

address a rival. Post-chorus, birds have been shown to use song matching as a

directed threat to indicate subsequent aggressive intentions (Krebs et al., 1981;

McGregor et al., 1992; Burt et al., 2001; Vehrencamp, 2001). The rate of switching

between song types, temporal overlapping of songs, duration matching and pitch

matching are additional potential directed signalling strategies that vary with

the intensity of agonistic interaction and serve to mediate aggression between

neighbours (reviewed by Vehrencamp, 2000).

The directional properties of signals such as song matching make them poten-

tially useful for detecting interactions at dawn since an audio recording of both

singers can determine the singer and target. However, conventional one- or two-

channel recording methods will not usually be adequate for this task for three



reasons. First, a dawn chorus interactive network would span a large area of ad-

jacent territories and birds could interact with any neighbour, thereby reducing

the effectiveness of focal recording. Second, identification and location of singers

by visual means is very difficult in the twilight before dawn; third, the sheer quan-

tity of vocalizations that occur at dawn in many locations tends to mask easy

identification of specific interacting participants. In fact, these difficulties may be

the reason why other researchers have not detected interactive networks at dawn

chorus, even in species where males clearly interact with their neighbours during

daylight.

New methods for studying dawn chorus

One solution to the problem of detecting interactions at dawn is the use of

distributed microphone arrays, which have been proposed as an ideal method for

monitoring communication networks in territorial neighbourhoods (McGregor &

Dabelsteen, 1996). Such systems are ideal for studying dawn chorus because they

can simultaneously record the songs of many singers in a large area. Another

advantage to using microphone arrays is the ability to determine the location

(and, therefore, in many cases the identity) of each singer using sound arrival

time differences (Watkins & Schevill, 1972; Speisberger & Fristrup, 1990). The

specific details of how acoustic location systems work is reviewed more thor-

oughly in McGregor et al. (1997). Array recording is a particularly good method

for detecting interactive networks in species that are known to use some form

of directed signal, such as immediate matching or overlapping during vocal ex-

changes. Changes in singing behaviour associated with movement to different

parts of the territory may also indicate interaction, most likely of a dyadic na-

ture. Additionally, microphone array recordings can be used to document changes

in the singing behaviour of non-interacting individuals before, during and af-

ter an intense interaction between two other individuals in a neighbourhood

(Eason & Stamps, 1993; Bower, 2000). Evidence that non-vocalizing receivers act

on information gained by eavesdropping (i.e. eavesdropping networks) is best ac-

quired with carefully designed playback experiments (e.g. Naguib & Todt, 1997;

Oliveira et al., 1998; Naguib et al., 1999; McGregor et al., 2001; Peake et al., 2001;

see also Ch. 2).

Banded wren song behaviour

In this chapter, we describe one of the first attempts to study a dawn cho-

rus communication network using a microphone array system (see also Bower,

2000). Our study species is the banded wren, in which males possess repertoires

of discrete, distinctive song types and usually switch to a different type after

each consecutive song. Neighbouring males share many of the same song types



and prior studies have documented the use of type matching and other singing

patterns during aggressive encounters (Molles & Vehrencamp, 1999, 2001). This

species is also a vigorous dawn chorus singer. Here we present an initial analysis

of the singing behaviour and interactions among four neighbouring male wrens

recorded during a single morning encompassing dawn chorus and the subsequent

hour of post-dawn chorus. We look for evidence of network communication in-

teractions involving two, three or more birds by searching for the presence of

directed signals such as matching and overlapping.

The banded wren is a common and vocally active species that inhabits the trop-

ical dry deciduous forest of the Pacific slope of Central America. It breeds only

during the first half of the rainy season (May–August) but remains resident and

paired on the same territory during the rest of the year. The mating system is so-

cially monogamous and each pair defends an all-purpose territory approximately

0.4 ha in area. Although not a true duet, a female occasionally sings short male-

like songs following or overlapping her mate’s songs. Males possess a repertoire

of 15 to 30 discrete song types, which may be delivered with a high rate of switch-

ing between song types (immediate variety mode) or in a more repetitive fashion

(eventual variety mode). Young males tend to copy whole song types from nearby

males and generally do not disperse very far from their natal territory, so estab-

lished adjacent neighbours share between 50 and 90% of their song-type repertoire

(Molles & Vehrencamp, 1999).

In the course of our research, we have identified a variety of song-delivery

patterns that banded wrens use to communicate with their neighbours in the

daytime during bouts of counter-singing. Song matches appear to be a threat sig-

nal; repertoire matches (singing a song shared with but not currently sung by the

neighbour) are used as a low-threat directed signal maintaining the interaction,

and switches to non-shared song types indicate a desire to deescalate (Molles &

Vehrencamp, 2001). Finally, banded wrens also appear to use song overlapping

during escalated interactions. Although the function of overlapping is not well

understood, it is often associated with (and often simultaneously combined with)

high rates of song matching, suggesting it also has a threat function. Overlapping

is apparently avoided during low-intensity counter-singing interactions between

distant males (Molles & Vehrencamp, 1999).

Banded wrens have a pronounced dawn chorus during the breeding season

months and are relatively silent at dawn during the rest of the year, despite remain-

ing on their territories. The chorus starts at twilight (approximately 05:00 h) and

lasts about 30 minutes. During this period, males sing vigorously and loudly. They

initially perch high in an emergent tree (10–20 m) and sing without pause from

one location for several minutes while constantly changing their body orientation.

As light levels increase, dawn chorus singers usually shift to other high perches



in other parts of their territories. Prior to the array study, we had anecdotally ob-

served numerous type matching events at dawn, but, given the conditions, it was

unknown to what extent, and with whom, the birds were matching. These prelim-

inary observations of song matching suggested to us that dawn chorus might be

an interactive phenomenon.

A male usually abruptly ceases dawn chorusing behaviour when his mate ap-

proaches and interacts with him. She may join him in a brief, uncoordinated duet.

Occasionally a male forgoes the dawn chorus completely to interact with his (pre-

sumably fertile) mate. Unmated males continue to sing at a high rate for another

30 to 60 minutes. After the dawn chorus, males begin foraging, interspersed with

bouts of singing, and patrol the borders of their territories more actively. At this

time, males seem to shift to more focused counter-singing with nearby neighbours.

Each territorial male has three to four adjacent neighbours with whom he

regularly interacts. Males construct bulky covered nests and are constantly ini-

tiating new ones because of high nest-predation rates. Females appear to select

the nest location and sometimes choose a site near a territorial border, which

forces the male to renegotiate that boundary with his neighbour through close-

range counter-singing and fights. The dynamic nature of territory boundaries in

this species could be one source of changes in social status hypothetically being

signalled during the dawn chorus.

Recording methods and subjects

As part of our research project studying the function of banded wren

song, we developed a microphone array recording system as a tool for acousti-

cally monitoring several vocalizing individuals. The technique involves placing

an array of many microphones at strategic locations within and around a small

neighbourhood of banded wren territories and simultaneously recording all song

interactions picked up by the microphones on a central multiple-channel receiving

unit. With these array recordings, we can quantify neighbourhood-wide singing

patterns and also focus on individual birds to gain a more complete picture of

their interactions with all of their neighbours.

The array data presented in this chapter were taken from an analysis of a

recording made on 20 June, 2001 at one of our study sites in Santa Rosa Park,

Costa Rica. This recording was selected from our library of daily recordings made

during the 2000, 2001 and 2002 May–July field seasons as a representative example

of dawn chorus singing in our study population. The chapter dataset runs from

05:02 h (Central American time zone) to 06:42 h, a total of 100 minutes. On that

day, civil twilight occurred at 04:58 h and sunrise at 05:21 h. Behaviourally, the

recording covered the first song sung in the neighbourhood at 05:04 h, the entire

dawn chorus and about one hour of post-chorus singing when dyadic counter-



450

400

LE

BO

N

N

N

N

OH

PU

YO

WBUB

WE

350

300

250

200

150

100

50

50 100 150 200

Distance (m)

Dis

tanc

e (m

)

250 300 350 400

*

*

* *

*

***

**

**

*

Fig. 15.2. Map of banded wren territorial neighbourhood recorded by the array. Solid

lines show the boundaries of the focal neighbours, while dotted lines are the

boundaries of adjacent neighbours. An asterisk indicates a microphone position and

N indicates current nest sites for each focal male. The shaded region in bird OH’s

territory indicates the area recently annexed from bird YO by OH.

singing predominated. The array consisted of 13 microphones, situated among

and surrounding four focal neighbours (males OH, YO, BO and WB; Fig. 15.2 shows

array configuration and territories). Since the focal birds were usually within

the array, nearly all of the songs sung by these four birds could be identified

and located. Songs of four other neighbours adjacent to the central four often

could be detected and identified if the bird was close to the array, but usually

not located (birds WE, PU, UB and LE; Fig. 15.2). Songs of these outlying birds

were included in calculations of overlapping and song matching to create more

accurate song statistics for the central focal birds. During the recording, four

observers were posted near each focal bird to take behavioural notes so that later

we could reconstruct patterns of interaction.



The recording on 20 June 2001 was typical for a mid-breeding season dawn

chorus and post-dawn chorus day of singing. The focal birds varied in age and

stage of nesting. OH had recently built a nest near his border with YO and had

extended his territory into that of YO (shaded area on OH’s territory in Fig. 15.2).

This boundary shift was a source of ongoing aggression between OH and YO, as well

as with BO, another of OH’s neighbours who was also affected by the shift. OH was

a long-term resident of the field site (at least seven years old). YO was banded as an

adult two years earlier (so was three or more years of age) and on the recording day

had just started building a new nest, which was located approximately 10–20 m

from OH’s border and approximately 50 m from OH’s nest. WB was an offspring of

the previous owner of PU’s territory, hatched in 1999 (two years old) and had an

active nest with nestlings. BO was a newly banded bird (probably first year), whose

nest was predated the day before by capuchin monkeys. During the recording, BO

was observed to be building a new nest.

Temporal patterns of singing behaviour

In this section, we provide a quantitative description of changes in several

key singing behaviours over time, based on the four focal birds in our recording. To

help the reader to visualize these dynamic patterns, we have adopted a presenta-

tion format that plots a running mean and standard error of the mean (SEM) of the

four focal birds’scores, starting at the time of the first song of the morning (05:04 h)

over a series of overlapping five minute intervals that move forward in time in

one minute increments (Fig. 15.3). Since the numbers of matches and overlaps are

highly dependent on the number of songs delivered in each interval, percentages

are given for these measures. Table 15.1 shows the correlations among these mea-

sures. For matching and overlapping analyses, all focal birds were considered to be

‘adjacent’neighbours (see Fig. 15.1), because all focal birds could easily hear all the

other focals and we had previously observed matched counter-singing between all

combinations of focal neighbours during the daytime.

Bout structure

Intersong interval (ISI), measured as the time from the end of a song

to the beginning of the next song, was used as an index of bout structure. This

variable is particularly sensitive to shifts between continuous singing and bout

singing (periods of relatively high song rates interspersed with pauses in singing)

and can be used to indicate the point in time that dawn chorus ends and post-

chorus bout singing begins. When all birds sing continuously at high rates, as in

the classic dawn chorus, the mean and variability of ISI will be small. When birds

shift to singing in asynchronous bouts, both the mean and the variability of ISI

will increase. Figure 15.3a shows the running mean and SEM of ISI over the course



200twilight

(a)

(b)

(c)

(d)

(e)

150

100

50

0

20

15

10

5

0

100

50

0

100

50

0

100

50

0

05:00 05:15 05:30 05:45 06:00 06:15 06:30 06:45

05:00 05:15 05:30 05:45 06:00 06:15 06:30 06:45

05:00 05:15 05:30 05:45 06:00 06:15 06:35 06:45

05:00 05:15 05:30 05:45 06:00 06:15 06:30 06:45

ISI (

s)S

ongs

/bird

Mov

emen

t (m

)M

atch

es (

%)

Ove

rlaps

(%

)

Time of day

05:00 05:15 05:30 05:45 06:00 06:15 06:30 06:45

Fig. 15.3. Changes in song behaviour and movement patterns over the course of the

recording, averaged across the four focal birds OH, YO, BO and WB. Values are

measures calculated over a series of overlapping five minute intervals that move

forward in time in one minute increments. Values are plotted at the centre time for

each segment. Means are shown as a solid line and the grey region represents ± SEM.

(a) Intersong interval (ISI); (b) number of songs per bird; (c) distance moved from each

bird’s average position at the previous to the current time segment; (d) percentage of

songs within a segment that were matches; (e) percentage of songs within a segment

that were overlaps.



Table 15.1. Pearson correlation coefficients among the key singing behaviours above the

diagonal, p values below the diagonala

Intersong

Song rate interval Movement Matches (%) Overlaps (%)

Song rate 0.046 −0.222 0.102 0.378

Intersong interval 0.657 0.280 −0.0128 −0.062

Movement 0.029 0.005 −0.149 −0.163

Matches (%) 0.322 0.211 0.147 0.251

Overlaps (%) 0.0001 0.549 0.111 0.013

a Significant correlation coefficients are shown in bold. The time series data were separated into

one minute bins. Each variable was corrected for autocorrelation by regressing it against its

lagged values and the residuals were used for the correlations (n = 97).

of the recording. Song bout structure for these banded wrens clearly differed

between dawn chorus and later. From the start of dawn chorus and for 20 minutes

into the recording, the mean ISI was very brief (around 10 seconds) and SEM was

very low, indicating that all birds were singing more or less continuously. At

around 05:21 h, the ISI measure exhibits a noticeable break from the previous

trend and both the mean and SEM increase and become slightly more variable

over time. At 05:45 h, a major break in ISI occurred, reflecting the fact that birds

had begun to sing in asynchronous bouts with variable interbout pauses. From

that point on, each bird stopped singing at least once for five minutes or longer.

These pauses are shown as spikes in mean ISI in Figure 15.3a, which occurred at a

different time for each bird (WB: 05:44–05:50 and 06:16–06:23 h; BO: 06:06–06:12 h;

YO: 06:23–06:29 h; OH: 06:25–06:32 h).

Song rates

In this recording, birds sang at uniformly high rates throughout dawn cho-

rus (Fig. 15.3b). OH began singing two minutes before the other birds, then YO and

BO began to sing, and two minutes later WB finally joined the chorus. At 05:22 h,

WB began to sing at a lower rate, while OH, BO and YO continued to sing at high but

more variable rates (shown as an increase in SEM in Fig. 15.3b). At about 05:40 h, all

four birds began a bout of intense counter-singing, and OH continued to sing at an

exceptionally high rate. After this synchronized bout, song rates declined again

but were punctuated by four more peaks, which reflect brief bouts of counter-

singing between different sets of neighbours. Individually, the four birds clearly

differed in their overall song output, with WB in particular being consistently

lower in song rate, starting later and quitting earlier compared with the others.



Movement patterns

Patterns of movement were analysed using passive acoustic location to

calculate the position of each bird whenever he sang. Figure 15.3c plots distances

between mean locations for successive time segments. This measure has low values

when birds are singing continuously from one position and increases when birds

move to new positions. The focal birds showed an initial spike of movement at the

beginning of dawn chorus (05:04–05:07 h) and then remained relatively stationary

until about 05:22 h. The initial movement spike was caused by several birds (BO,

OH and WB), who appear to have sung for a brief period from positions near their

sleeping nests and then moved to a more centrally located high song post for the

bulk of their dawn chorus singing. Between 05:27 and 05:43 h, birds made two

or three short distance movements to different parts of their territories while

still singing fairly vigorously and continuously. After 05:43 h, movements became

larger and more frequent. At this time, birds moved relatively quickly between

song perches and tended to stay at each location for several minutes before moving

again (these movements are seen as a pattern of brief spikes in Fig. 15.3c).

Song matching

Figure 15.3d plots the mean percentage across birds of songs sung that

were matches. A song was judged to be a match if it was the same type as an

adjacent neighbour’s recent song (either a song the neighbour had just sung,

or the one previous) and occurred less than 30 seconds after the matched song.

Matching was initially lower because OH had sung alone for two minutes, but

as soon as the other birds began to sing, matching rates quickly increased. Song

matching peaked at 70% at 05:14 h (12 minutes into dawn chorus). After the first

peak, matching declined slightly (although it remained quite high at around 40%)

and peaked again at 05:30 h (52%). Matching declined thereafter, but peaked again

at 05:56 (43%), 06:07 (30%) and 06:26 h (25%).

The four focal birds sang an average of 22 different song types during the record-

ing (range, 20–24) and shared 82% of their song types with any given neighbour

(range, 77.3–85.1). With high singing rates, frequent song type switches and high

levels of song type sharing, the possibility of birds matching by chance will be

higher at dawn chorus. We created a model to test whether the observed rates

of matching were higher than that expected through chance. To estimate chance

matching rates, we generated new datasets using the observed singing data for

all males with randomly shuffled song-type assignments within each bird’s reper-

toire. By averaging the matching rates obtained over multiple permutations of

shuffled song types, we could estimate the probability of chance matching if the

birds had chosen their songs without regard to other singers. Figure 15.4 compares

observed matching versus expected chance matching calculated on the basis of



Mea

n pe

rcen

tage

mat

ches

per

5 m

in

Fig. 15.4. Observed (± SEM) and expected percentage matching rate per five minute

interval. Expected values were calculated by averaging matching outcomes of 100

random permutations of song types within focal birds. Observed matching rates that

were significantly higher than expected are marked with an open circle (two-tailed

binomial tests, Holm corrected for multiple comparisons; criterion p < 0.05).

100 permutations of random song-type shuffling. Observed rates were significantly

higher than expected throughout the first 15 minutes of dawn chorus and during

the four subsequent peaks later on. The peak at 05:28 h was caused by an intense

interaction between the four focal birds, with a three-way matching interaction

between OH, YO and BO, and a separate matching interaction between WB and

BO (conclusions drawn from analysing individual bird data not shown in the fig-

ure). The three later peaks are attributable to further intense bouts of matched

counter-singing between pairs and trios of neighbours (OH, YO and PU at 05:56 h;

WB, UB and WE at 06:07 h; and WB, BO and LE at 06:27 h).

If song matching is an indicator of conflict, knowing who is matching with

whom can provide useful information about what is going on in a neighbourhood.

Bird OH was involved in many of the interactions that morning. We think that his

singing behaviour and his neighbours’ responses to him were related to his recent

annexation of space at the corner of his, YO’sand BO’sboundaries to defend a newly

active nest in that area (Fig. 15.2). Figure 15.5 shows the patterns of matching by



Fig. 15.5. Patterns of matching by and to bird OH throughout the recording period.

Grey squares indicate OH’s song matches to one of his four neighbours. Black squares

indicate that OH sang a non-matching song. Open squares indicate a match by a

particular neighbour to OH.

and to bird OH, giving us a finer picture of the interactions between him and the

other birds that morning. OH’s bout of solo singing two minutes before his other

neighbours accounts for the lower mean song matching at the very beginning of

dawn chorus. During the initial peak of song at 05:12 h, OH was matched by and

was matching the neighbours at the disputed corner (BO and YO). At this time, the

majority of matches were initiated by the neighbours (each matched a different

one of OH’s songs) and OH replied with his own matches to those neighbours (i.e.



Table 15.2. Percentage of matches by and to each focal bird

Target bird Matching bird

BO OH YO WB Other neighbours

BO – 25 38 20 17

OH 32 – 43 16 8

YO 22 49 – 11 17

WB 21 38 24 – 17

Other neighbours 14 8 16 25 –

in Fig. 15.5 the open squares indicating a neighbour match usually preceded OH’s

own match to that neighbour, so the neighbour matched first). OH’s border shift

had particularly affected YO, and YO and OH matched most often throughout the

morning (43% of YO’s matches were to OH, while 49% of OH’s matches were to

YO; Table 15.2), as would be predicted by an agonistic function for dawn chorus

singing. Another trend is visible in Fig. 15.5, primarily during the dawn chorus

(05:07–05:32 h) and briefly later (05:52–05:57 h): OH, as well as the other birds not

shown in this figure, frequently alternately matched multiple neighbours within

a short period of time. Often these multiple matches involved using a different

song to match each neighbour. For example, at the start of dawn chorus, OH had

been switching between two song types. The first song sung by YO was a match

to one of OH’s types, while BO first sang a match to the other type. Matching was

not merely isolated between the OH/YO/BO trio – all of these birds were often

matching other birds at the same time. Based on our observations that morning,

we know that WB spent most of his time interacting with two other neighbours

(WE and UB) but, nevertheless, at various times did match and was matched by

OH, BO and YO.

Overlapping

In prior work, we had noticed many occurrences of what appeared to be

deliberate overlapping during close-range counter-singing interactions between

neighbouring males. Furthermore, overlapping often occurred in conjunction

with a song match, timed to cover the majority of the other bird’s song as if

the matcher were trying to ‘jam’ the other singer. The context of the ‘overlapping

match’ phenomenon suggested that overlapping is used as an aggressive signal.

Figure 15.3e plots the mean percentage across birds of songs that overlap an-

other song. A song was considered to be an overlap if it ‘covered’ an adjacent

neighbour’ssong by 50% or more of its duration. Songs are sufficiently long in this

species (mean duration is 3.4 seconds) and the territories relatively small (centres

120 m apart) that errors in perceived overlapping caused by the slow speed of



sound are small for this high overlap criterion (Dabelsteen, 1992). Overlapping

was strongly correlated with mean song rate (Table 15.1). Our overlapping mea-

sure cannot distinguish between accidental and deliberate overlapping and it is

likely that much of the trend in overlapping resulted from an increase in the prob-

ability of accidental overlapping during times of increased song rates. Overlapping

may still be used deliberately, as we have observed, but much less frequently or

under very specific contexts. We have some evidence for this: overlapping rate was

significantly correlated with matching rate, even after controlling for the effects

of song rate (partial r = 0.215; p = 0.024), a pattern that would occur if birds oc-

casionally deliberately combined overlapping with song matching. There were 27

occurrences of overlapping matches out of 924 songs delivered by the focal birds

during the recording.

Overall patterns of banded wren singing at dawn chorus

The structure of the banded wren dawn chorus follows the pattern com-

monly described for other species: males begin singing at twilight and sing con-

tinuously at high rates. Then, coincident with increasing light levels and female

emergence, males change to a daytime pattern of interacting individually with

nearby neighbours. Banded wrens also show high rates of song matching to neigh-

bouring singers at dawn chorus. Although dawn chorus matching has been noted

in several other species (Todt, 1970; Spector, 1991), it had not been quantified be-

fore our study, making it difficult to know whether high-rate matching at dawn

occurs in many other species. Therefore, to the degree that dawn chorus has been

characterized in other birds, banded wrens appear to behave similarly to other

chorusing species.

We found clear differences among the four focal males recorded on this single

morning in their rate and duration of singing, which could be caused by vari-

ation in male quality, condition, dominance status or territory quality, as de-

scribed for several species (Cuthill & Macdonald, 1990; Otter et al., 1997; Poesel

et al., 2001). However, there was evidence that these differences could be caused

by short-term variation in motivation arising from differences in nesting stage or

territory boundary disputes. The lowest-rate singer in this recording, WB, was a

highly successful male who was feeding nestlings, whereas the other three males

had recently lost their nests and were engaged in boundary disputes resulting

from OH’s incursion into YO’s and BO’s territories. Analysis of additional record-

ings would clearly be needed to determine whether individual differences in song

output are consistent over time or vary with breeding conditions and/or short-term

motivation.



Evidence for an interactive network

Our strongest evidence for an interactive network during the banded wren

dawn chorus is the high level of song matching that takes place during the first

half hour of singing. Matching at rates significantly higher than chance indicates

that song types delivered by one bird are affecting song types delivered by other

birds and that implies, by definition, that they are interacting. The dawn matching

was likely of a competitive nature too, given that bouts of matching also occur

in the daytime during synchronized counter-singing between dyads or trios of

neighbouring males and when males approach each others’ boundaries for closer

interaction. Our simultaneous recordings showed that males alternately matched

different neighbours in rapid succession with different song types, the key type of

evidence for a fully multi-way interactive communication network. Furthermore,

matching was strongly directed toward one male, OH, by his neighbours YO and

BO at a time when OH was expanding his territory into mostly YO’s territory to

accommodate a new nest site. We thus see particularly clear evidence of multi-way

competitive interactions among these three males.

An observational study conducted on the same population in 2000 indirectly

corroborates our claim of male–male interaction at dawn. In that study, we found

that banded wren males use their most vigorous song type renditions during the

most intense period of singing at dawn, for example longer song-type variations,

compound songs (two or more types sung together) and song types with longer

trills, wider bandwidth and rattle and buzz elements, all of which are associated

with intense male–male interactions at other times (S. L. Vehrencamp & A. Trillo,

unpublished data). Similar patterns have been described for the European black-

bird Turdus merula and yellow warbler Dendroica petechia, which use louder, longer

songs of higher intensity during the dawn chorus and when counter-singing from

a distance with other males (Dabelsteen, 1992; Lowther et al., 1999).

Comparing song patterns with aggression

Patterns of matching and switching in this recording could also reveal

short-term changes in motivation and the outcomes of recent interactions. For

example, OH’s prior aggressive behaviour (annexing portions of two of the focal

neighbours’ territories) appears to have had a strong effect on the behaviour ob-

served in this recording. From the start of dawn chorus, the pattern of matching

toward individual birds was strongly asymmetrical, with most of the matching

directed toward OH (Table 15.2). OH’s boundary shift had probably involved pro-

longed bouts of escalated counter-singing and physical fighting with the affected

neighbours (YO and BO), making OH a particularly threatening neighbour. OH’s



early start to dawn chorus, two minutes before the others, may have been intended

to reinforce his tentative ownership of the disputed property. The matching re-

sponse directed toward OH by the affected neighbours may, in turn, have been

retaliatory threats in response to OH’s announcement of continued occupation.

Frequent type switching during a matching interaction may enable a bird to as-

sess more easily which neighbour is feeling most threatened or to challenge each

neighbour with a distinctive signal.

Dawn chorus singing as an indicator of male quality

In addition to revealing information about short-term changes in moti-

vation, dawn singing could also give eavesdropping receivers information about

longer-term or intrinsic differences among males related to dominance, condi-

tion and age. Montgomerie (1985) suggested that energy reserves should be lowest

at dawn, imposing a handicap such that the vigour and amount of singing hon-

estly reflects a male’s condition or territory quality. Food supplementation was

shown to increase the amount of song in blackbirds (Cuthill & Macdonald, 1990).

Peak song rate during the dawn chorus was correlated with winter dominance at

feeders in black-capped chickadees Parus atricapillus (Otter et al., 1997) and with

earlier female laying date in the blue tit Parus caeruleus (Poesel et al., 2001). Banded

wrens attain peak daily song rates during the dawn chorus. In addition, their

songs seem to be especially loud at this time, although this impression could be

caused by the high song perches (Dabelsteen & Mathevon, 2002). One drawback

to array recording is that individuals who do not vocalize are ‘invisible’ to the

analysis and so we have not been able to show any evidence for eavesdropping

with this dataset. However, with more recordings and more detailed analysis,

it may be possible to show some direct effects of eavesdropping on vocalizing

interactants.

The value of matching at dawn chorus

A defining feature of the dawn chorus is a continuous high rate of singing

by all territorial neighbours. To a listener in a forest at dawn chorus, there is a

confusingly high density of song coming from many directions. This unique ‘song

environment’ poses a challenge to singers, who may be trying to direct signals

to specific neighbours and simultaneously listen for signals directed at them

from all of their neighbours. For this reason, highly directional signals may in-

crease in value and birds would be predicted to shift their singing strategies to

favour more directional signals and avoid using signals that might create ambi-

guity as to the singer’s intended target. In particular, dawn chorusing birds may

avoid using signals that rely on song rate and timing such as song overlapping,



synchronized song rates and synchronized song-type switching, because these

signals are more likely to be masked when many individuals are singing nearby at a

high rate.

Song matching is one of the few signals that retains its usefulness as a direc-

tional signal at dawn chorus, because it is based on song-type selection, rather

than song timing or rate. In addition, the higher rates of singing and song-type

switching often seen at dawn chorus provide birds with more opportunities for

directed song matching than at other times. For these reasons, it is possible that

song matching accompanied by rapid switching rates will be a common occur-

rence in dawn chorusing species that engage in neighbour–neighbour interactive

networks.

Summary

The single array recording presented here gives the reader a glimpse into

the behaviour of banded wrens at dawn chorus. Our observations provide evi-

dence of a highly interactive communication network, which is most consistent

with the social dynamics hypothesis for the function of dawn chorus, as presented

in Staicer et al. (1996). We are currently analysing a number of similar recordings,

made between 2000 and 2002, on the same focal birds, as well as on different

sets of focal neighbours. With more recordings and more birds, we intend to test

more rigorously the trends we saw in the recording presented in this chapter. In

particular, evidence that the bird who is the focus of matching changes on dif-

ferent days, in relation to current patterns of boundary and nest-site movements,

would greatly strengthen the argument for dawn chorus mediating changes in

social status. Repeated observations on the same birds will also be necessary to

determine whether any of the dawn chorus behaviours are related to male age,

repertoire size, sharing level or quality, which would indicate an additional male

quality-assessment role for the banded wren dawn chorus, and the existence of

eavesdropping.

Based on our findings, we think that the microphone array recording tech-

nique is an ideal method for studying the details of dawn chorus song behaviour.

In particular, multimicrophone recording in some form is possibly the only fea-

sible way to detect and study multi-way interactive communication networks,

such as we found in the banded wren dawn chorus. It is our hope that other

species can be studied using techniques similar to ours so that cross-species com-

parisons can be made of dawn chorus communication networks. Until that time,

it will not be known whether the highly interactive dawn chorus network we

have documented in banded wrens is unique or common among dawn chorusing

species.



Acknowledgements

Logistical support was provided by the staff of the Area de Conservacion Guanacaste.

We thank Alex Trillo, Liz Campbell, Carlos Botero, Richard Mills, Dan Pendleton and Harold Mills

for helping us set up the array and make observations during recording sessions, and Cary Leung

for analysing much of the array data. This research was funded by NIH grant R01-MH60461.

References

Ballentine, B., Badyaev, A. & Hill, G. 2003. Changes in song complexity correspond to

periods of female fertility in Blue Grosbeaks (Guiraca caerulea). Ethology, 109,

55–66.


in a song sparrow neighbourhood. Ph.D. Thesis, Cornell University, New York.

Burt, J. M., Campbell, S. E. & Beecher, M. D. 2001. Song type matching as threat: a test

using interactive playback. Animal Behaviour, 62, 1163–1170.

Cuthill, I. C. & MacDonald, W. A. 1990. Experimental manipulation of the dawn and

dusk chorus in the blackbird Turdus merula. Behavioral Ecology and Sociobiology, 26,

209–216.

Dabelsteen, T. 1992. Interactive playback: a finely tuned response. In: Playback and

Studies of Animal Communication, ed. P. K. McGregor. New York: Plenum Press,

pp. 97–109.

Dabelsteen, T. & Mathevon, N. 2002. Why do songbirds sing intensively at dawn? A test

of the acoustic transmission hypothesis. Acta Ethologica, 4, 65–72.

Eason, P. K. & Stamps, J. A. 1993. An early-warning system for detecting intruders in a

territorial animal. Animal Behaviour, 46, 1105–1109.

Goodson, J. L. 1998. Territorial aggression and dawn song are modulated by vasotocin

and vasoactive intestinal polypeptide in male field sparrows (Spizella pusilla).

Hormones and Behavior, 34, 66–77.

Henwood, K. & Fabrick, A. 1979. A quantitative analysis of the dawn chorus: temporal

selection for communicatory optimization. American Naturalist, 114, 260–274.

Hultsch, H. & Todt, D. 1982. Temporal performance roles during vocal interactions in

nightingales (Luscinia megarhynchos). Behavioral Ecology and Sociobiology, 11, 253–260.

Hutchinson, J. M. C. 2002. Two explanations of the dawn chorus compared: how

monotonically changing light levels favour a short break from singing. Animal

Behaviour, 64, 527–539.

Kacelnik, A. & Krebs, J. R. 1982. The dawn chorus in the great tit (Parus major):

proximate and ultimate causes. Behaviour, 83, 287–309.

Krebs, J. R., Ashcroft, R. & Webber, M. 1978. Song repertoires and territory defence in

the great tit. Nature, 271, 539–542.

Krebs, J. R., Ashcroft, R. & van Orsdol, K. 1981. Song matching in the great tit Parus

major. Animal Behaviour, 29, 918–923.



Kroodsma, D. E., Bereson, R. C., Byers, B. E. & Minear, E. 1989. Use of song types by the

chestnut-sided warbler: evidence for both intra- and inter-sexual functions.

Canadian Journal of Zoology, 67, 447–456.

Lowther, P. E., Celada, C., Klain, N. K., Rimmer, C. C. & Spector, D. A. 1999. Yellow

warbler. In: The Birds of North America, no. 454, ed. A. Poole & F. Gill. Philadelphia:

The Birds of North America, Inc.

Mace, R. H. 1986. The importance of female behaviour in the dawn chorus. Animal

Behaviour, 34, 621–622.

1987. Why do birds sing at dawn? Ardea, 75, 123–132.







McGregor, P. K., Dabelsteen, T., Clark, C. W. et al. 1997. Accuracy of a passive acoustic

location system: empirical studies in terrestrial habitats. Ethology, Ecology and




see is not what you get? Animal Behaviour, 62, 1059–1065.

McNamara, J. M., Mace, R. H. & Houston, A. I. 1987. Optimal daily routines of singing

and foraging in a bird singing to attract a mate. Behavioral Ecology and Sociobiology,

20, 399–405.



Molles, L. E. & Vehrencamp, S. L. 1999. Repertoire size, repertoire overlap, and

singing modes in the banded wren, Thryothorus pleurostictus. The Auk, 116,

677–689.

2001. Songbird cheaters pay a retaliation cost: evidence for auditory conventional

signals. Proceedings of the Royal Society of London, Series B, 268, 2013–2019.

Montgomerie, R. D. 1985. Why do birds sing at dawn? In: Proceedings of the XIX

International Ethology Congress, p. 242.





Series B, 266, 537–542.

Nelson, D. A. & Croner, L. J. 1991. Song categories and their functions in the field

sparrow (Spizella pusilla). The Auk, 108, 42–52.

Nowicki, S., Peters, S. & Hughes, M. 1998. The territory defense function of song in

song sparrows: a test with the speaker occupation design. Behaviour, 135,

615–628.



Oliveira, R. F., McGregor, P. K. & Latuffe, C. 1998. Know thine enemy: fighting fish



Otter, K. & Ratcliffe, L. 1993. Changes in singing behaviour of male black-capped

chickadees (Parus atricapillus) following mate removal. Behavioral Ecology and


Otter, K. A., Chruszcz, B. & Ratcliffe, L. 1997. Honest advertisement and song output

during the dawn chorus of black-capped chickadees. Behavioral Ecology, 8, 167–173.

Otter, K. A., Stewart, I. R. K., McGregor, P. K. et al. 2001. Extra-pair paternity among

great tits Parus major following manipulation of male signals. Journal of Avian

Biology, 32, 338–344.

Part, T. 1991. Is dawn singing related to paternity insurance? The case of the collared

flycatcher. Animal Behaviour, 41, 451–456.




Poesel, A., Foerster, K. & Kempenaers, B. 2001. The dawn song of the blue tit Parus

caeruleus and its role in sexual selection. Ethology, 107, 521–531.

Slagsvold, T., Dale, S. & Saetre, G.-P. 1994. Dawn singing in the great tit (Parus major):

mate attracting, mate guarding, or territorial defence? Behaviour, 131, 115–138.

Spector, D. A. 1991. The singing behaviour of yellow warblers. Behaviour, 117, 29–52.

Speisberger, J. L. & Fristrup, K. M. 1990. Passive location of calling animals and sensing

of their acoustic environment using acoustic tomography. American Naturalist,

135, 107–135.

Staicer, C. A. 1989. Characteristics, use, and significance of two singing behaviours in

Grace’s Warbler (Dendroica graciae). The Auk, 106, 49–63.

Staicer, C. A., Spector, D. A. & Horn, A. G. 1996. The dawn chorus and other diel

patterns in acoustic signaling. In: Ecology and Evolution of Acoustic Communication in

Birds, ed. D. E. Kroodsma & E. H. Miller. Ithaca, NY: Cornell University Press,

pp. 426–453.

Todt, D. 1970. Gesang und gesangliche Korrespondenz der Amsel. Naturwissenschaften,

57, 61–66.

Vehrencamp, S. L. 2000. Handicap, index, and conventional signal elements of bird

song. In: Signalling and Signal Design in Animal Communication, ed. Y. Espmark,

T. Amundsen & G. Rosenqvist. Trondheim: Tapir Academic Press,

pp. 277–300.

2001. Is song-type matching a conventional signal of aggressive intentions?


Watkins, W. A. & Schevill, W. E. 1972. Sound source location by arrival times on a

non-rigid three-dimensional hydrophone array. Deep-Sea Research, 19,

691–706.

Welling, P., Koivula, K. & Lahti, K. 1995. The dawn chorus is linked with female

fertility in the willow tit Parus montanus. Journal of Avian Biology, 26, 241–246.



Wingfield, J. C. & Farner, D. S. 1993. Endocrinology of reproduction in wild species. In:

Avian Biology, Vol. 9, ed. D. S. Farner, J. R. King & K. C. Parkes. London: Academic

Press, pp. 163–327.

Yasukawa, K., Bick, E. I., Wagman, D. W. & Marler, P. 1982. Playback and

speaker-replacement experiments on song-based neighbour, stranger, and self

discrimination in male red-winged blackbirds. Behavioral Ecology and Sociobiology,

10, 211–215.


16

Eavesdropping and scent over-marking

ro b e r t e . j o h n s t o n

Cornell University, Ithaca, USA

Introduction

Compared with communication in other sensory domains and with scents

that are released into the air, scent marking is unusual because the signal remains

long after the signalling behaviour; for example, the flank gland marks of male

golden hamsters Mesocricetus auratus deposited on glass in the laboratory are de-

tected by other hamsters 40 days later and vaginal secretion marks are detected

at least 100 days after deposition (Johnston & Schmidt, 1979). In the field, the

paste scent marks deposited by brown hyaenas Crocuta crocuta can be detected

by humans for at least 30 days (Gorman, 1990) and klipspringers Oreotragus oreo-

tragus respond to preorbital gland marks that have been exposed to direct sun

for at least seven days by an increase in scent marking (Roberts, 1998). In many

species, especially those that live solitarily, there is often no receiver present

when the marks are deposited. Consequently, scent marks are necessarily gen-

eral broadcast signals that usually have several functions, depending on the age,

sex, reproductive status, social status and individual identities of both senders and

receivers.

One type of marking, scent counter-marking, is directed at the scent marks

of other individuals, but again these individuals are often not present to ob-

serve the signalling behaviour. I consider the term scent counter-marking to

include two different types of behaviour: (a) over-marking, in which the sec-

ond individual’s scent at least partially overlaps that of the first individual; and

(b) adjacent marking, in which the second individual’s scent is close to that of

the first individual but does not overlap it. In the species with which I am most

familiar (golden hamsters and meadow voles Microtus pennsylvanicus), both types of


344


Eavesdropping and scent over-marking 345

counter-marking are usually intermixed, but it is possible that some species would

engage in one type of counter-marking but not the other, or would vary the type of

counter-marking or the proportions of different types of counter-mark in different

contexts.

Based on the variety of situations in which counter-marking is observed, it no

doubt has a variety of functions (Brown & Macdonald, 1985). Counter-marking of-

ten occurs only between adult males, suggesting a sexually selected trait related

to competition for females (Hurst & Rich, 1999; Johnston, 1999), but in some

species adult males mark over the marks of their mates, suggesting mate guard-

ing (Mertl, 1977; Moore & Byers, 1989; Kappeler, 1998; Roberts, 2000; Powzyk,

2002). Females may counter-mark other females (e.g. golden hamsters (Johnston,

1977), house mice Mus musculus (Hurst, 1990)) or mark over the marks of males

(e.g. moustached tamarin Saguinus mystax (Heymann, 1998)). Among species that

live in social groups, most or all of the members of a group may mark in the same

place sequentially, producing a group counter-mark (e.g. Rasa, 1973; Mills et al.,

1980; Gorman & Mills, 1984); sometimes dominant individuals in the group mark

most often (Peters & Mech, 1975), but sometimes subordinate, subadult individ-

uals mark most (Lazaro-Perea et al., 1999). Groups may also counter-mark during

territorial encounters (Jolly, 1966). There is a great need for more observational

and experimental field studies focused on the functions of over-marking.

Several years ago, I proposed that the functions of over-marking could be ap-

proached from the question of what kinds of information third parties could

obtain about the marking individuals from scents in over-marks (Johnston et al.,

1994). I suggested that there were three different types of effect that might occur:

(a) masking, in which the most recently deposited scent covers previous scent

marks and thereby eliminates access to the information they contain (e.g. individ-

ual identity, sex, reproductive state); (b) mixing, in which the scents of different

individuals become a chemical mixture, thus producing a new odour quality (e.g.

a group odour) and thereby eliminating information about particular individu-

als; and (c) posting, in which each scent mark to some extent remains separate or

distinguishable, thus producing a bulletin board at which information about all

individuals that marked there can be obtained. Subsequent research suggests that

an important fourth possibility is that scent over-marks can also provide informa-

tion about the relationships between the odours of different individuals, perhaps

including the relative freshness, amount of scent, number of marks, top or bottom

position, or geometric layout (Wilcox & Johnston, 1995; Johnston et al., 1997a;

Johnston & Bhorade, 1998; Hurst & Rich, 1999). Although these early experiments

addressed the perception of over-marks and the subsequent memory for individ-

ual odours, they also can be viewed as indicating the kinds of information that

are available to eavesdroppers.


346 R. E. Johnston

Scent marking, scent over-marking and eavesdropping

Regardless of the different functions that counter-marking might have,

eavesdropping can occur when a third individual investigates a place marked by

two or more other individuals. I use eavesdropping in the broad sense of an animal

witnessing some type of interaction or an exchange of information between two

other individuals (similar to ‘social eavesdropping’ sensu Peake, Ch. 2). This does

not imply conscious intent or secrecy on the part of the animal doing the observ-

ing nor any awareness or attempt to conceal the interaction by the animals whose

interaction is observed. Eavesdropping on scent counter-marking may be particu-

larly common because the marks are so long lasting, allowing many individuals

to investigate them, not just the individuals present when the marking behaviour

was performed. Furthermore, when eavesdropping on scent counter-marks, indi-

viduals can more easily avoid a potential cost of eavesdropping on vocalizations or

visual displays, namely being detected and threatened, chased or attacked. Among

vertebrates, neither scent marking nor other aspects of chemical communication

have been explicitly analysed using the framework of eavesdropping. Among inver-

tebrates, however, numerous cases have been described using these concepts, but

authors studying insects seem to have adopted a different meaning for the term.

All of the examples I have found involve predators, parasites or parasitoids using

the chemical signals of the host or prey as a means of locating them (e.g. Stowe

et al., 1995). This is a different set of phenomena (‘interceptive eavesdropping’; see

Chs. 2 and 23) and will not be considered here.

What evidence is there for eavesdropping based on scent counter-marking

among mammals? I will discuss three different types of evidence, primarily from

our own work: (a) apparent sensory and perceptual specializations for the evalua-

tion of scent over-marks; (b) evidence that the information in over-marks leads to

differential responses by the perceiver towards individuals whose marks are in dif-

ferent positions in over-marks (top versus bottom); and (c) specialized mechanisms

for the production of over-marks.

Scent marking in golden hamsters

There are two types of scent marking in golden hamsters. Flank marking

is carried out by both males and females and scent is deposited from the flank

gland, which is a region of specialized, pigmented sebaceous glands on the poste-

rior flank. This glandular field is larger in males than in females and is testosterone

dependent (Vandenbergh, 1973). Flank marking is performed as a part of general

maintenance activities (e.g. shortly after animals wake up and groom themselves)

and is especially prevalent in potentially agonistic situations (e.g. in the presence

of odours of other individuals). Hamsters do not usually flank mark during actual



interactions (Johnston, 1975a,b,c, 1977, 1985). In seminatural laboratory environ-

ments, hamsters mark just inside the tunnel to their burrow, in the vicinity of

the burrow entrance and in other locations (Johnston, 1975c). Subordinate males

in these environments have been observed marking in their nest when a dom-

inant male is attempting to enter (Johnston, 1975c). Therefore, it seems likely

that flank marking is involved in defending the burrow and food hoard by both

males and females (Johnston, 1975c). The second type of scent marking is a type

of anogenital marking, called vaginal marking, that deposits vaginal secretions.

Hamsters have a specialized pouch surrounding the distal vagina that produces

and collects this secretion. The frequency of vaginal marking is related to the fe-

male’sreproductive state, peaking during the night 12–24 hours before receptivity

(Johnston, 1977, 1985). This secretion is highly attractive to males; it stimulates

copulatory behaviour, reduces aggressive behaviour and causes increases in cir-

culating luteinizing hormone and testosterone (Johnston, 1985, 1990). In a study

in seminatural enclosures in the laboratory, females attracted males the night

before receptivity, slept with them during the day, mated early the next day and

then drove the male away (Lisk et al., 1983). Therefore, one primary function of this

type of marking is to advertise sexual receptivity and to attract males. In addition,

females may over-mark the vaginal marks of other females (Fischer & McQuiston,

1991), perhaps as a means of competing for the attention of males or as a secondary

aspect of defence of the burrow and food hoard against other females.

Specialized mechanisms for evaluation of scent over-marks

The first evidence suggesting special mechanisms for evaluation of scent

over-marks came from experiments aimed at understanding the information

obtained about individuals from the scents in an over-mark, as described above.

In particular, these experiments were designed to discover what golden hamsters

would remember after investigating an over-mark consisting of the scents of two

individuals (Johnston et al., 1994). Corresponding to the idea that an over-mark by

one individual might mask, mix or remain distinguishable from the underlying

individual’s scent, would hamsters remember just the top scent, neither scent, or

both scents? We used an habituation technique in which male subjects were first

exposed to a newly deposited, flank-gland over-mark from two donor males (male

B always on top of male A) on four or five successive trials and then were tested on

a final trial with the scent of one of these donors and the scent of a novel donor.

In the first experiment, we wanted to be sure that the scent of male B covered the

scent of male A, so we simulated natural scent marks by picking up the animals

and rubbing their flank gland region against the substrate, a glass plate. We placed

a paper cardstock template over the glass so that scent was deposited only in a


348 R. E. Johnston

limited area on the plate (the exact size of the area corresponded to the usual size

of flank or vaginal scent marks (Johnston et al., 1994)). Across repeated habituation

trials, the investigation of the scent over-mark decreased, as it would to a single

individual’s scent. On the test trial, subjects should investigate the familiar scent

significantly less than the novel scent, indicating memory for the familiar scent

(e.g. Johnston et al., 1993). The results were quite interesting: subjects treated the

flank gland odour of the top-scent individual as familiar (investigated it less than

that of a novel individual) but investigated the flank odour of the bottom scent

male the same amount as that of the novel individual (Johnston et al., 1994). We

obtained similar results when males were habituated to vaginal secretion over-

marks of females and then tested with each scent individually compared with a

novel vaginal secretion. Because the scent of the second donor was placed on top

of that of the first donor, our interpretation was that the top scent masked the

bottom scent and, therefore, the bottom scent actually was novel to the subjects

(Johnston et al., 1994).

This first experiment was, however, somewhat unrealistic in that the scent of

the second individual completely covered the scent of the first individual; when

hamsters deposit their own scent they are usually not so thorough or precise. In

a second experiment, we placed the top and bottom scents at right angles to one

another such that they formed a cross; that is, there was a region of overlap of

the two individuals’ scent marks, but also regions where each individual’s scent

was by itself. The same results were obtained as in the first experiment: male

subjects treated the vaginal scent of donor B as familiar but treated the scent of the

bottom-scent individual (donor A) the same as that of a novel individual (Fig. 16.1a).

Results using flank-gland secretions showed the same pattern (Johnston et al., 1995;

Johnston, 1995). These results indicate that the subjects had a preferential memory

for the top scent compared with the bottom scent, despite being able to investigate

both scents during the habituation trials. This preferential memory suggests that

the subjects have a mechanism for evaluating over-marks and either preferentially

remember the top scent, selectively forgetting the bottom scent, or tag the memory

of the top scent so that it is more salient than that of the bottom scent. In further

experiments of this type, we found evidence for preferential memory for the top

scent even if, during the habituation trials, there was an additional scent mark of

the bottom-scent donor that was not marked over (Fig. 16.1b; Wilcox & Johnston,

1995). These experiments suggest that hamsters extracted information about the

relative position (top or bottom) or the relative freshness of the two individuals’

scent marks and either selectively remembered just one of them or had placed

a different value on the memory of the top and bottom odours. The existence

of either of these mechanisms suggests that the relative position of an

individual’s scent in an over-mark is important to the perceiver and that this



0.005

Mea

n in

vest

igat

ing

tim

e (s

)

0.0025

Fig. 16.1. The time that male hamsters spent investigating the vaginal scents of

females on the test trial after habituation to a pattern of vaginal scent-marks as

shown above the graphs. In both (a) and (b), males investigated the scent from the

top-scent female significantly less than the novel female’s scent (n = 9 in (a); n = 10

in (b)) whereas there was no significant difference in time spent investigating the

scent of the bottom-scent male and the novel scent (n = 10 in both groups), thus

indicating a preferential memory for the top scent of the over-mark. The bar indicates

the standard error; p values derived from t-tests. (From Wilcox & Johnston, 1995.)

information may influence subsequent social interactions between the perceiver

and the individuals that deposited the scent marks.

Did the subjects in these experiments actually forget the bottom scent in an

over-mark or did they just attach less value or importance to it? The latter seems

more likely. First, it is difficult to believe that hamsters would not remember one

of two individually distinctive odours after investigating them four or five times,

since a single scent is remembered at least 10 days after such exposures (Johnston,

1993). Second, later experiments showed that hamsters would remember two ad-

jacent scents (see below). Third, in another experiment, we obtained evidence

that males did have some memory of the bottom scent from an over-mark. Male

hamsters were first exposed to experimenter-produced over-marks of male flank

glands in the pattern of a cross during four habituation trials with 15 minutes

between trials. Investigation of the crossed scents decreased significantly in all

three groups (10 per group) across the habituation trials: the results for trial 1 and

trial 4, respectively were 16.8 and 3.8 seconds in group 1 (t = 10.216; p < 0.001),

21.1 and 3.2 seconds in group 2 (t = 14.802; p < 0.001), and 19.3 and 3.7 seconds

in group 3 (t = 7.156; p < 0.001). These groups were then tested 15 minutes after


350 R. E. Johnston

0.0007

0.01M

ean

inve

stig

atin

g ti

me

(s)

Fig. 16.2. The time male hamsters spent investigating the flank scent of one male in

the test trial after habituation to a scent over-mark in the pattern of a cross (shown

above) in a series of trials. The top scent from the over-mark was investigated least,

indicating habituation to and memory for this odour compared with the novel odour.

The bottom scent from the scent mark was investigated an intermediate amount,

suggesting some memory for this odour but one that was significantly different than

that for the top scent. The bar indicates the standard error; p values derived from

t-tests; 10 animals in each group. (R. E. Johnston & M. Schiller, unpublished data.)

the last habituation trial with the flank scent from just one male on the test

trial: the flank scent of the top-scent male for group 1, the bottom-scent male for

group 2 and a novel male for group 3. The novel flank scent was investigated most

and the top scent was investigated significantly less, as in the previous tests in

which there were two stimuli present in the test trial (Fig. 16.2; R. E. Johnston &

M. Schiller, unpublished data). The bottom scent, however, was investigated an

intermediate amount, significantly more than the top scent but significantly less

than the novel scent (Fig. 16.2). This experiment suggests that hamsters do re-

member the scent of the bottom-scent male but the memory is not as strong or as

salient, or that the behaviour based on the memory is different from that for the

top-scent male. We do not know exactly in what way it is different, but we suspect

that the bottom scent is less important to the subjects (see p. 359–360). Experiments

on meadow voles support the notion that the odour of the bottom-scent male

is devalued relative to top-scent males or novel males (Woodward et al., 2000).

Before describing our experiments on how hamsters and voles determine which

scent is on top, it is useful to review what we know about the mechanisms under-

lying discrimination between odours from different individuals. As in any other



sensory domain, individual recognition is accomplished by pattern recognition

mechanisms: perceptual mechanisms by which animals discriminate between sim-

ilar, complex stimuli (e.g. faces, voices, odours). In such processes, specific features

are generally much less important than the relationships between features. In the

case of odours, the pattern is generated by a large number of individual chemical

compounds that occur in differing proportions across individuals (Gorman, 1976;

Bagneres et al., 1991; Gamboa et al., 1996; Singer et al., 1997; Smith et al., 2001). It is

these differences in proportions of chemical compounds that give each individual

its distinctive odour quality. The particular chemicals that differ in proportion

vary across pairs of individuals; that is, there does not seem to be a particular set

of chemical compounds that are used for this purpose (Smith et al., 2001). Conse-

quently, one might expect that a mixture of two scents would be created when

one individual over-marks another’s scent that was different from either of the

two original mixtures and that this new mixture would produce a new odour

quality. The results of the experiments reported in the preceding paragraphs, in

which subjects were habituated to over-marks, argue against this hypothesis be-

cause in the test trial hamsters showed memory for one individual but not the

other (Johnston et al., 1994, 1995; Wilcox & Johnston, 1995).

We have undertaken a series of experiments to try to characterize the mecha-

nisms used to distinguish top and bottom scents in an over-mark. The strategy in

these experiments was to determine if a particular kind of information in over-

marks was sufficient, by itself, to promote differential responses to the odour of

one donor versus the other donor.

One possible cue is the relative freshness of the two scents, because the top scent

of an over-mark is necessarily fresher than the bottom scent. In the experiments

discussed above, the difference in the age of the two scent marks in over-marks

was about 30 seconds (not more than 60 seconds); however, in nature, scent marks

are likely to differ more than this in freshness (e.g. by at least tens of minutes

and often by hours or days). Therefore, a series of experiments were carried out to

see if differences in freshness alone would cause the differential memory effect.

In addition, we wanted to be sure that animals could detect and remember the

scents of two individuals that were in close proximity. In the first experiment, sub-

jects were habituated to vaginal secretion scents from two individuals (A and B)

that were placed adjacent to one another in an open cross pattern, as shown

in Fig. 16.3a, and then were tested for their responses to scent A or B versus a

novel scent. Males investigated both scents (A and B) less than the novel scent,

indicating that they could remember the scents of two individuals from habitu-

ation trials (for flank gland scents, see Cohen et al. (2001)). Then we tested males

to determine if differences in freshness between the two scents would result in

differential memory for the fresher scent. The scents were again presented in the


Mean investigating time (s)

(a)

Hab

ituat

ion

stim

uli

Scen

t 1

Scen

t 2

0.00

10.

005

0.00

250.

025

0.02

5

Scen

t 1gr

oup

Scen

t 2gr

oup

scen

tnove

l

scen

t

Fig.

16.3

.Th

eti

me

spen

tby

mal

eh

amst

ers

inve

stig

atin

gva

gin

alse

cret

ion

scen

tsfo

llow

ing

hab

itu

atio

nto

vagi

nal

secr

etio

nm

arks

inth

ep

atte

rnsh

own

abov

eth

egr

aph

.(a)

Aft

erh

abit

uat

ion

totw

osc

ent

mar

ksin

the

pat

tern

ofan

open

cros

s,bo

thsc

ents

wer

ein

vest

igat

edsi

gnif

ican

tly

less

than

an

ovel

scen

t,in

dic

atin

gm

emor

yfo

rbo

th.(

b)A

fter

hab

itu

atio

nto

two

scen

tsof

dif

feri

ng

deg

rees

offr

esh

nes

s,bo

thsc

ents

are

inve

stig

ated

sign

ific

antl

yle

ssth

anth

en

ovel

scen

ts,i

nd

icat

ing

that

ad

iffe

ren

cein

fres

hn

ess

doe

sn

otle

adto

dif

fere

nti

alm

emor

yfo

ron

eof

them

.(c)

Aft

er

hab

itu

atio

nto

scen

tsin

an‘a

pp

aren

tcr

oss’

(on

eco

nti

nu

ous

scen

tan

don

ein

terr

up

ted

scen

t),t

he

con

tin

uou

ssc

ent

was

inve

stig

ated

sign

ific

antl

yle

ssth

anth

en

ovel

scen

tbu

tth

ere

was

no

sign

ific

ant

dif

fere

nce

inti

me

spen

tin

vest

igat

ing

the

inte

rru

pte

dsc

ent

and

the

nov

elsc

ent,

sugg

esti

ng

poo

rer

mem

ory

for

the

inte

rru

pte

dsc

ent

than

for

the

con

tin

uou

s

scen

t.Th

eba

rin

dic

ates

the

stan

dar

der

ror;

pva

lues

der

ived

from

t-te

sts;

10an

imal

sin

each

grou

p.(

From

Joh

nst

on&

Bh

orad

e,19

98.)

352



open cross pattern during habituation trials, but one vaginal secretion scent was

30 seconds old and the other was 24 hours old. On the test trial, male hamsters

investigated novel scents more than familiar scents of the same age, indicating

memory of both of the familiar scents. The investigation of the familiar, 30 second

scent did not differ from that of the familiar 24 hour scent, indicating no preferen-

tial memory for the fresher scent (Fig. 16.3b; Johnston & Bhorade, 1998; for similar

data on male flank gland scent, see Cohen et al. (2001)). Similar experiments with

meadow voles exposed to adjacent anogenital marks differing in age by 60 minutes

yielded no significant difference in response to the donor of the fresh scent versus

the donor of the 60-minute-old scent (Ferkin et al., 1999). Therefore, freshness by

itself did not lead to preferential treatment of the scent of one individual. It is

worth noting that four different types of odour were used in these experiments:

vaginal secretions and flank glands in hamsters (Johnson & Bhorade, 1998), anogen-

ital area scent in voles (Ferkin et al., 1999) and urine in voles (M. H. Ferkin,

J. Dunsavage & R. E. Johnston, unpublished data). This suggests that the lack of

an effect of freshness is not caused by the chemistry of one particular type of

scent (such as the sebaceous scent from flank glands) that might change little over

24 hours. We do not know how long individually specific information lasts, but

hamster flank and vaginal scents deposited on glass in the laboratory are detected

and investigated after 40 and 100 days, respectively (Johnston & Schmidt, 1979).

A second possible cue that might be used as a guide for preferential responses to

one individual over another is the relative amount of scent or number of marks de-

posited by two individuals. Among many species, dominant or high-ranking indi-

viduals mark more often than subordinate or low-ranking individuals (Ralls, 1971;

Eisenberg & Kleiman, 1972; Johnson, 1973; Johnston, 1975a,c; Bronson, 1976; Hurst

& Rich, 1999) and the relative amount of over-marking is probably correlated with

the frequency of marking. Therefore, the number of marks or over-marks could

be an indication of dominance status. Even when there was a single over-mark, it

might be the case that subjects would perceive that the top-scent individual de-

posited more scent if the top scent masked some of the bottom-scent individual’s

mark. The ‘amount of scent’ and ‘number of marks’ hypotheses were, however,

ruled out as explanations of the differential effects in hamsters and meadow

voles. That is, hamsters still remembered the top-scent individual and meadow

voles investigated the top-scent individual more even when, during the expo-

sure to over-marks, there were more marks or more area covered by the bottom-

scent individual (e.g. Fig. 16.1b; Wilcox & Johnston, 1995; Johnston et al., 1997a,b;

Johnston & Bhorade, 1998; Ferkin et al., 1999). In one experiment with meadow

voles, for example, female subjects investigated the home cage of a male that

had been briefly investigated and marked by an ‘intruder’ male. When tested in

the Y-maze, females spent more time investigating the whole-body odour of the


354 R. E. Johnston

Mea

n in

vest

igat

ing

time

(s)

0.01

Fig. 16.4. The time that female meadow voles spent investigating whole-body odours

from cotton bedding in a Y-maze after exposure to the home cage of one male that

had been briefly investigated and scent marked by another male, as represented by

the drawing on the left. The bar indicates the standard error; p values derived from

Wilcoxon matched-pairs signed-rank test; n = 12. (From Johnston et al., 1997b.)

intruder male than that of the home-cage male, even though the home-cage male

must have had many more marks and covered more area with his marks than

the intruder did (Fig. 16.4; Johnston et al., 1997b). Therefore, neither hamsters nor

voles seem to use the relative amount of scent or number of marks, by themselves,

as a means of selective responses to other individuals.

Other experiments suggest that it is some type of information from the area of

overlap, or the area of overlap compared with adjacent areas with non-overlapped

scents, that leads to preferential responses to the top-scent individual. One line

of evidence for this conclusion is that among all of the experiments that we have

done with hamsters and meadow voles, using a variety of testing methods, we

have found differential responses in the test phase only when there were scent

marks that overlapped during the exposure phase (with one exception, see below)

(Johnston et al., 1994, 1995, 1997a,b; Johnston & Bhorade, 1998; Wilcox & Johnston,

1995; Ferkin et al., 1999; Cohen et al., 2001). A second line of evidence for the

importance of scent overlap comes from experiments specifically designed to test

whether overlap of two individuals’ scent marks was necessary for differential

responses to the two animals or to their odours. For example, we first habituated



Fig. 16.5. The time that female meadow voles spent investigating whole-body odours

from cotton bedding in a Y-maze after an exposure to scent marks in the patterns

shown above the graph. (a) After exposure to a large area of scent from one male with

a small spot of scent from another male on top, females (n = 12) spent significantly

more time investigating the top-scent male. (b) When the small spot was placed in a

clean area of the slide, so that there is no overlap, females (n = 10) show no significant

difference in investigation time. Bar indicates the standard error; p values derived

from Wilcoxon test. (From Johnston et al., 1997a.)

male golden hamsters either to two scent marks in the pattern of a cross or to

two scent marks in a pattern of an ‘open cross’ – that is, there was no overlap in

the middle, just an unscented space. On the test trial, males exposed to this latter

pattern showed an equivalent response to the scent of the two donors, treating

them both as familiar (e.g. Fig. 16.3a; Cohen et al., 2001), but males exposed to

the crossed scents showed memory for the top scent but not for the bottom scent

(Fig. 16.1a; Johnston et al., 1995; Johnston & Bhorade, 1998; Cohen et al., 2001). In

another example, one group of female meadow voles were exposed for 15 minutes

to a microscope slide largely covered by the anogenital scent of one male but with

a small spot of scent from a second male placed on top. During the test trial, these

females spent significantly more time investigating the whole-body odours of the

small-spot, top-scent male (Fig. 16.5a). If, however, the second male’s scent was

placed in a clean ‘hole’ surrounded by scent from the first male, female meadow

voles showed no significant difference in response to the odours of the two males

(Fig. 16.5b; Johnston et al., 1997a). In both hamsters and voles, an area of overlap

was necessary to produce a differential response to their odours. We do not know


356 R. E. Johnston

exactly how hamsters or voles use the area of overlap to determine which scent is

on top, but one hypothesis is that they compare the odour quality of the overlap

region with that in the adjacent areas. Assuming that there is partial mixing in the

area of overlap, the odour in this region should smell more like the top-scent male

than the bottom-scent male. In the hamster experiment, animals could compare

the area of overlap with the two adjacent, non-overlapped scents and determine

which was the top-scent by which scent was closest in odour quality to the region

of overlap. This mechanism would not work for the vole experiment, however,

because in this situation there is just an area of overlap and an area of scent from

the bottom-scent male.

Finally, there is some evidence that hamsters may be able to use the geomet-

ric relationships between the two scent marks (interrupted versus uninterrupted

streaks) to determine which is the top scent. We reasoned that, by analogy with

depth perception in vision, if one scent occludes another, it must be on top. If,

when investigating scent over-marks, hamsters or other animals develop a repre-

sentation of the geometrical layout of the marks, they might be able to determine

which of two scents was on top by determining which one occluded the other. To

test this possibility while at the same time eliminating cues from a region of over-

lap, we investigated how male hamsters would respond after being habituated

to a pattern of scent marks in which it might appear that one individual’s scent

occluded the other but in fact there was no region of overlap; rather, there was one

continuous scent and, at right angles to it, two scent marks that approached this

scent closely but did not touch it (Fig. 16.3c). Male hamsters showed a preferential

memory for the continuous scent compared with the interrupted scent for vaginal

scent marks (Fig 16.3c; Johnston & Bhorade, 1998) and male flank marks (Cohen

et al., 2001). This is the only case in which a region of overlap was not necessary to

obtain a differential response to one animal or its odours after exposure to scent

marks of two individuals. In contrast, meadow voles show no evidence of using

the same kind of spatial information (Ferkin et al., 1999). Perhaps the primary rea-

son for this species difference is that, whereas hamsters deposit marks in linear

streaks, voles’ marks are more often small spots or larger irregularly shaped areas

(especially urine marks) and it is not obvious how spots and blobs could provide

spatial cues about which scent was on top.

The results summarized above indicate that at least two species of rodents

have evolved specialized mechanisms for the perception and analysis of scent over-

marks. These abilities are quite striking – indeed, amazing – and we do not yet fully

understand them. Many other species have probably evolved similar mechanisms,

but, because of the diversity of functions served by scent marking, it is not likely

that all species have evolved such mechanisms. The existence of these abilities,



however, should stimulate us to think about other ways in which animals might

extract information from arrays of scent marks.

Specialized mechanisms for production of over-marks

If over-marking has functions that are distinct from the functions of reg-

ular scent marking, specialized mechanisms should have evolved to promote ac-

curate placement of scent on top of that of another individual: that is, mecha-

nisms to target the scent marks of others. Numerous observations of a variety of

mammalian species in nature indicate that such targeted over-marking does oc-

cur (Ralls, 1971; Eisenberg & Kleiman, 1972; Johnson, 1973; Brown & Macdonald,

1985). Several recent field studies have provided quantitative data on the extent of

over-marking in natural environments (Kappeler, 1998; Lazaro-Perea et al., 1999).

In perhaps the most dramatic case, male diademed sifakas Propithecus diadema over-

mark 94% of the scent marks deposited by their mates (Powzyk, 2002). Observations

of over-marking in nature are convincing evidence for a targeting mechanism be-

cause the probability of even one over-mark occurring by chance is extremely low,

given that there are many possible places to mark. In addition, most animals that

over-mark also engage in other related activities when encountering scent marks

from another individual, such as careful investigation, pawing or scratching the

ground, biting the bark of the tree on which the marks are deposited or becoming

visibly aroused or agitated, indicating that they are reacting to this odour and are

focused on it.

Virtually all experimental studies of the mechanisms underlying scent mark-

ing and over-marking, however, have been carried out in captivity or in laboratory

settings, where space is limited. In such circumstances, it is much more difficult

to determine if over-marks occur by targeting mechanisms or occur by chance.

For example, a common (but not universal) observation is that a particular odour

(e.g. urine, flank gland) will increase the frequency of scent marking with the

same scent (Ralls, 1971; Johnston, 1975a, 1977; Hurst, 1990). Some of these scent

marks will be over-marks. In the limited spaces used in the laboratory, however,

it is difficult to know whether the increase in the number of over-marks occurs

just because the rate of marking increased and, by chance, some were deposited

in places that had been marked previously, or because a specific targeting mecha-

nism exists. Similarly, other mechanisms that increase marking frequency, such as

changes in hormone levels with reproductive state, can increase the frequency of

over-marking. Furthermore, many species have preferred types of site for marking,

such as particular topological features (visually prominent rocks or vegetation, in-

tersections of trails, particular types of plant or tree, water holes or other rare


358 R. E. Johnston

resources). Hamsters in laboratory environments, for example, prefer to mark in

confined spaces, such as the corners of a rectangular arena and just outside of their

burrow entrance (Johnston, 1975c). An increase in the number and the percentage

of over-marks could occur if the rate of marking is increased by stimulation and

there are a limited number of preferred places in which to mark.

We designed a simple method to determine if individuals have a specific mecha-

nism for targeting another individual’sscent mark to produce an over-mark, which

we have applied to both flank marking and vaginal marking by hamsters (R. E.

Johnston, S. Szmuilowicz, D. J. Mayeaux, S. K. Barot, & N. S. Schwarz, unpublished

data). This involves comparing the number of marks that are deposited over scent

marks to the number of marks deposited over mirror-symmetric locations in the

same arena that have no scent (imaginary marks). Since the types of locations are

identical and are in the same arena, this method controls for both the problem of

preferred locations and the problem of odour-stimulated increase in overall mark-

ing frequency. In one recent experiment of this type (S. K. Barot, N. S. Schwarz &

R. E. Johnston, unpublished data), we found that the mean number of flank marks

by 12 male hamsters that overlapped another male’s flank marks was 6.0, whereas

the number of flank marks that overlapped imaginary marks (symmetric locations

but clean) was 2.8, (t = 3.171; p < 0.01). Females, however, did not selectively flank

mark over the flank scents of other females or males. We are currently replicating

and refining these experiments, but these initial experiments suggest the exis-

tence of a specific mechanism in adult, male golden hamsters that targets the

flank marks of other adult males but not the marks of juvenile males or females.

These results suggest that flank over-marking by males is a sexually selected char-

acteristic related to male–male competition, but that flank marking by females is

a more generalized kind of broadcast signalling.

Functions of over-marking and eavesdropping on over-marks

What is the function of scent over-marking, and why have hamsters and

meadow voles evolved specialized mechanisms for evaluating over-marks? We

have hypothesized that in meadow voles and golden hamsters, both of which live

solitarily, scent over-marking by males may be a type of intrasex competition in

which each male targets its male neighbours (Johnston et al., 1997a,b; Johnston

& Bhorade, 1998; Johnston, 1999; Cohen et al., 2001). This mutual over-marking

presumably reflects a struggle for dominance between males. If each male is, at

least during the breeding season, continuously trying to keep its scent on top of

the scents of its male neighbours, over-marking should be an energetically costly

activity. I am not aware of any data on the energetic costs of over-marking in

natural settings, but the cost of marking for defence of territory is important for



some species (Mills et al., 1980; Gorman & Mills, 1984; Gorman, 1990; Gosling et al.,

2000; Gosling & Roberts, 2001). The degree to which an individual is successful at

keeping its marks on top of those of its neighbours should be an honest indicator

of phenotypic vigour and quality and, to the extent that these characteristics are

dependent on genotype, genetic quality. Information gathered by third par-

ties from over-marks about ‘whose scent is on top’ should, therefore, be valu-

able information for mate-choice decisions by females and it may influence in-

teractions between like-sex rivals as well. Is there evidence to support these

hypotheses?

The evidence suggests that analysis of scent over-marks does affect preferences

for opposite-sexed individuals and that such preferences are likely to be impor-

tant in mate-choice decisions. Most of the tests we carried out with meadow voles

described above were preference tests rather than habituation tests: female voles

were first exposed to anogenital or urine over-marks from two males for 15 min-

utes; 10 minutes later they were tested for their preferences for the whole-body

odours (cotton bedding material, that contained additional body odours and pos-

sibly urine odours) of donor males in a Y-maze. Females spent more time close

to and investigating the whole-body odour of the top-scent male than that of

the bottom-scent male in a variety of experiments (Figs. 16.4, 16.5a; Johnston et al.,

1997a,b). Males and females were housed in long-day light cycles and were in re-

productive condition, so this preferential behaviour suggests that females would

prefer the top-scent males as mates (Johnston et al., 1997b). In addition, we also

exposed female meadow voles to naturally deposited over-marks of males and

found that females again preferred the whole-body odours (cotton bedding) from

top-scent males over that of bottom-scent males (Fig. 16.4; Johnston et al., 1997b).

It is worth noting that in several of these experiments there was far more of

the bottom-scent male’s odour present during the exposure phase than of the

top-scent male’s odour, but females still preferred the top-scent male (Figs. 16.4

and 16.5; Johnston et al., 1997a,b). We interpret these results as a preference for

the top-scent male because the test stimulus (whole-body odours in bedding ma-

terial) incorporated whole set of body odours and was a different stimulus from

the one odour collected by us and presented during the exposure phase. Further-

more, the test was carried out in a different environment from the one used for

the exposure to the over-marks (Y-maze rather than subject’shome cage). These re-

sults cannot be explained as merely the result of habituation for two reasons. First,

females were exposed to the odours of both stimulus animals during the presenta-

tion of the over-marks. Second, our studies using a habituation paradigm showed

that after habitation to an over-mark subjects spent less time investigating the

scent from the top-scent male than that from the bottom-scent male, which is the

opposite of what we found in our Y-maze experiments.


360 R. E. Johnston

Pilot experiments with hamsters suggest that after females eavesdrop on males’

over-marks they show preferences for the top-scent over the bottom-scent male.

Female hamsters living in a seminatural enclosure in the laboratory were allowed

to explore regions of the environment that had been explored and marked by two

males, but one male was always first (male A) and the other was always second

(male B); consequently, when the female explored the arena, the marks of male

B should have been on top. (Although we did observe males marking, we did not

have a method of determining the location of the marks with sufficient accuracy

to be certain for all cases that they overlapped or not.) Females were tested in

the seminatural environment approximately 18 hours prior to receptivity (when

they are soliciting males) and also when they were receptive. The stimulus males

(A and B) were confined in small, wire-mesh enclosures. Females spent more time

investigating male B (that had explored and marked the arena second) than male A

(there first): combined investigation time for day before oestrous was 238.1 ± 39.8

seconds and for day of oestrous was 159.1 ± 24.6 seconds (degrees of freedom (df ) =9; t = 2.39; p = 0.04). Females also vaginal marked more in the vicinity of male B (7.4

± 1.4 seconds for male B and 4.9 ± 1.0 seconds for male A; df = 9; t = 2; p = 0.057

(S. K. Barot, N. S. Schwarz & R. E. Johnston, unpublished data)). In other experiments

with hamsters, males explored an arena in which there were six vaginal secretion

marks of female B overlapping those of female A and two marks of female A

by themselves. After 45–60 minutes, males were tested for their preference in a

simultaneous choice apparatus (Steel, 1984). Males spent more time sniffing the

top-scent female B (108 ± 4.8 seconds) than female A (88.7 ± 5.4 seconds; df =15; t = 2.956; p = 0.01 (R. E. Johnston & C. Lee, unpublished data)). These results

suggest that males may prefer as mates females that successfully over-mark other

females with vaginal secretions; such over-marking by females may reflect their

vigour and ability to defend their burrow and food hoard from other females.

Maintaining a safe burrow and a food hoard could lead to greater pup survival and

if so it would be advantageous for males to mate preferentially with such females.

Experiments with other species also indicate that information gained by inves-

tigation of scent marks influences mate preferences. After female house mice ex-

plored areas in which the marks of only one male were present (exclusively marked

territory) and another area in which a second male had also marked (and over-

marked), females preferred the males that had the exclusively marked areas (Rich &

Hurst, 1998). Females also preferred a male that had counter-marked another

male to one that had been counter-marked (Rich & Hurst, 1999; see also Ch. 11).

Similar supporting evidence comes from studies with a primate, the pygmy loris

Nycticebus pygmaeus (Fisher et al., 2003). Male lorises over-mark the urine marks of

other males. Females were exposed to naturally produced male over-marks over

a period of 14–20 weeks, until the females came into oestrous. Several measures



of female behaviour (proximity and orientation to males, investigation of odours

and socio-sexual behaviour) all indicated a preference for the top-scent male by

oestrous females.

The evidence so far suggests that in several species eavesdropping on scent

over-marks influences mate preferences, perhaps because over-marks usually re-

flect the relative phenotypic and possibly genotypic quality of individuals. Much

more research needs to be done to determine how widespread this phenomenon

is, the degree to which eavesdropping on over-marks influences mate choice in

natural settings, and the factors leading to the evolution of over-marking as a

sexually selected trait that provides useful information about potential mates. It

would also be valuable to determine the degree to which over-marking correlates

with other measures of behaviour or physiology that are related to the geno-

typic or phenotypic quality of individuals within a population. Although I have

stressed the usefulness of over-marks as a means by which third parties might

assess opposite-sexed individuals, analysis of over-marks may also influence com-

petitive interactions between third parties of the same sex as those that deposited

the over-marks (e.g. in competition for territory, food or water resources, nesting

sites or mates) could be seen as more likely to dominate in confrontations. Also yet

to be investigated are the effects of over-marks on the individuals that are engaged

in over-marking contests. No doubt there are many interesting phenomena yet to

be discovered.

Field tests of specific hypotheses about over-marking

Both field and laboratory studies are needed to test specific hypotheses

about the functions of over-marking. If over-marking is a type of advertising con-

test between like-sex individuals for mates for example, one would expect that

over-marking the scent of potential rivals (adults of the same sex in reproductive

condition) would be especially prevalent. There are many studies that have shown

that scent marking in general is stimulated by like-sex adults, but relatively few

studies that have measured over-marking and even fewer that have compared

over-marking towards rivals and non-rivals. Likewise, if over-marking is a means

of mate-guarding, one would expect it to be predominantly done by the sex that is

most actively competing for mates (usually males) and that males would mark over

the scent of their mate, as has been observed in the pronghorn antelope Antilocapra

americana (Byers & Bekoff, 1986). Particularly valuable should be field experiments

in which scent marks are manipulated in ways similar to song playbacks (Sliwa &

Richardson, 1998), for example, by experimentally over-marking some residents

but not others, or by using the scent of one resident to over-mark the scents marks

of rivals.


362 R. E. Johnston

Audience effect on scent marking?

Audience effects refer to an alteration in behaviour because of the pres-

ence of specific individuals or classes of individuals: the audience (see Ch. 4).

Generally, the presence of other animals is defined as the subject being able to

see or hear these other animals. Scent provides interesting possibilities that have

not been systematically explored. For example, might the behaviour of one indi-

vidual toward a second individual be altered by the presence of scent from some

individuals but not by scent from others? If so, would fresh scent be more effective

than older scent? Although fresh scent is not exactly the same as being observed

or heard by another individual, it could indicate the recent presence of an individ-

ual, and thus it could indicate a high probability of being observed or discovered

by the animal that deposited the scent. One can imagine that very fresh scent of

a dominant individual, for example, could inhibit some types of aggressive be-

haviour of a subordinate or that fresh scent of a male paired with that of a female

might inhibit another male from courting her.

In a more traditional sense, do animals alter their scent marking behaviours

based on the social environment? Both of the species that we have studied in

detail (golden hamsters and meadow voles) live solitarily and in nature one would

predict that they usually mark when alone. In laboratory settings, golden hamsters

do not usually flank mark in the presence of another hamster but often mark

vigorously just after social encounters (Johnston, 1975a). This could be considered

a type of audience effect, but not one that depends on the presence or absence of a

specific audience. The one exception that I have observed to this pattern of marking

when alone is that subordinate males do sometimes mark just inside their burrow

entrance when they defend their burrows from a dominant male (Johnston, 1975c).

This behaviour, however, seems to be readily explained by flank marking as an

agonistic behaviour involved in the defence of the burrow or food hoard: the

notion of an audience effect does not aid our understanding. Female hamsters

are stimulated to vaginal mark in the presence of a male or a male’s odours but

not in the presence of a female or her odours: they mark most frequently in the

period 12–24 hours before sexual receptivity (Johnston, 1977, 1979). Once again,

this pattern seems to be primarily related to the function of sexual advertisement

and is not a modulation of behaviour based on specific relationships between

individuals.

Perhaps the most likely situations in which audience effects might be observed

would be in gregarious species. For example, in the ring-tailed lemur, Lemur catta,

individuals are attentive to the marking behaviour of others and 62% of all marks

are investigated within 30 seconds of deposition; in 89% of these cases the original

mark is over-marked (Kappeler, 1998). In this study, however, no evidence was



found for an influence of social context on the likelihood of scent marking. In

species that live in groups that have a clear dominance hierarchy, such as wolves

Canis lupus or free-ranging packs of domestic dogs, dominant individuals mark

much more than subordinates (Peters & Mech, 1975; Bekoff, 1979). One could

imagine that subordinates might alter their marking based on the presence or

absence of a more dominant individual, but I am not aware of any data that

explicitly supports this speculation. In some group-living species, such as dwarf

mongooses Helogale undulata, all members of the group may mark in the same

place, especially when first emerging from the burrow in the morning (Rasa, 1973).

Marking by some individuals could be stimulated by observing others mark (social

facilitation) or by the mere presence of others in the group at that time of day (a

possible audience effect). Another situation in which an audience effect might be

observed is in cases in which the scent-marking behaviour serves as a visual signal

as well as a means of depositing a chemical signal. For example, males among

all species of gazelles engage in stereotyped visual displays when marking their

territories. This type of marking has been called ‘demonstrative marking’ because

of its obvious nature and probable value as a simultaneous visual and olfactory

signal (Estes, 1967). Again, one could hypothesize that the occurrence or vigour

of such marking displays might vary depending on the relationships between the

marker and specific males that were present or the presence, or absence, of females.

I am not aware of any data that have been analysed in this context, however. The

alpha male and female dogs of a pack are more likely to mark with a raised leg

urination after observing another dog do so than are lower-ranking individuals

(Bekoff, 1979), but again it is not clear if this is a competitive reaction to observing

the mark or is dependent on the presence of the audience of the other members

of the pack.

Networks, cognition and individual recognition: speculations on

species differences in underlying mechanisms

The notion of a communication network is important because it empha-

sizes that individuals are a part of a community of interacting individuals (for an

early version of this view, see Estes (1969)). This is true even in species in which in-

dividuals spend most of their time by themselves. The concepts of eavesdropping

and audience effects draw attention to two specific ways that the social context

can modulate the behaviour of individuals. Although evidence demonstrating

eavesdropping or audience effects does not imply anything specific about the

mechanisms underlying these effects, most researchers working in this area seem

to assume that the animals they work with treat one another as unique individuals

with unique sets of distinctive characteristics. This is an inference about cognitive


364 R. E. Johnston

and neural processes that I will refer to as true individual recognition or having

a concept of other individuals. One important aspect of having such an attribute

is that animals should have memories consisting of several types of information

about other individuals and these memories should be stored as organized units

or representations of individuals (Johnston & Jernigan, 1994; Johnston & Bullock,

2001). This type of representation contrasts with other, simpler mechanisms that

are nonetheless sufficient to explain many of the findings that are referred to

as individual recognition, neighbour recognition, etc. Although the mechanisms

involved in recognition and memory are not directly observable, it is possible

to characterize these mechanisms by appropriate behavioural and physiological

measurements. The fields of cognitive psychology and cognitive neuroscience, for

example, depend on this kind of analysis. With regard to recognition of individ-

uals by non-human animals, for example, it is possible to discover the kinds of

information that one animal knows about another (Johnston & Bullock, 2001). Fur-

ther, this information can help to explain why individuals behave the way they

do. Indeed, the complexity of an individual’s knowledge about other individuals

is likely to have a profound influence on how that individual behaves in social

interactions with them (e.g. Chs. 24 and 25).

I have previously argued that many demonstrations of individual recognition

in the field and in the laboratory do not allow us to discriminate between merely

recognizing familiar versus unfamiliar cues or combinations of cues and recogniz-

ing individuals as unique entities (Johnston & Jernigan, 1994; Johnston & Bullock,

2001). For example, in the literature on neighbour recognition by song in birds,

the majority of studies merely demonstrate that a territorial male responds more

strongly to a novel song than a familiar song, or a novel song–direction combi-

nation than to a familiar song–direction combination. They do not provide proof

that the birds recognize their neighbours as individuals. A slightly more complex

mechanism might be categorization of information into heterogeneous categories

(e.g. a frequently heard song versus a rarely heard song) but again, not a catego-

rization based on individuals as the unit of analysis (Barrows et al.,1975; Caldwell,

1985). Cases in which birds engage in repertoire matching, in contrast, suggest the

existence of true individual recognition. That is, when a song sparrow Melospiza

melodia hears a neighbour sing one song type and then responds with a song

type that he shares with this neighbour but that is different from the one that the

neighbour just sang (Beecher et al., 1996), this suggests that the male knows several

characteristics of his neighbour and that they differ for different neighbours. This

type of observation provides evidence that song sparrows respond to neighbours

as unique individuals. Another example that demonstrates complex, multicom-

ponent representations of individuals comes from work with golden hamsters, in

which we showed that after males had interacted with several females they had



memories of these females that incorporated at least three different odours. Using

an habituation task, we showed that habituation to one of these odours resulted

in habituation to the other odours as well. That is, when becoming habituated to a

particular stimulus the males also became habituated to the individual and thus

other features of this individual (Johnston & Jernigan, 1994; Johnston & Bullock,

2001). These effects do not occur if the males have not interacted with the stimulus

females and, therefore, have not had an opportunity to learn about the features

of these females; that is, the effects are not a result of inherent similarities across

odours and consequent generalization across odours.

Similarly, an experimental demonstration of eavesdropping may result from

true individual recognition or it could result from a simpler mechanism. For

example, if one male fighting fish observes a fight between two other males of

the species, he could remember that fish A with the purple fringe on this dorsal

fin and red stripes on his tail and a distinctive wiggle in his display is much

more aggressive than fish B, characterized by all red fins and tail but a purple

spot in the middle of the tail fin. The observer fish could have memories of these

two individuals, and each memory would consist of an integrated memory of

that male’s physical and behavioural characteristics. A simpler type of memory

would, however, also be sufficient to explain the effects. The observer fish might

associate ‘purple edge above and red stripes in back’ with fear and ‘all red’ with

lack of fear. He could learn an association between a few specific cues and fear

or danger but not have memories of individuals as such. Both types of memory

would result in differential responses to the two individuals, but the mechanisms

underlying the responses would be different and indicate different levels of neural

and cognitive complexity. Additional experiments could provide evidence for or

against the existence of true individual recognition. For example, does the subject

react differently to two individuals that were observed to have similar experiences

(e.g. lost a fight) but some aspects of the interaction were different (e.g. how quickly

the animal lost)? Does the subject react differently to the two individuals in a

non-aggressive context? Does the subject cross-habituate to different features (e.g.

odours, sounds, other visual information) from the same individuals (Johnston &

Jernigan, 1994; Johnston & Bullock, 2001)?

There is probably a continuum of complexity in the types of representation that

animals have of other individuals, and these differences in complexity should cor-

relate roughly with the complexity of the social behaviour of different species

and taxonomic groups. At higher levels of complexity are species that not only

remember individuals as such but also remember something about the relation-

ships between different individuals (Cheney & Seyfarth, 1990a, b; Chs. 24 and 25).

At the most complex end of the continuum would be species, such as humans and

perhaps some other highly social animals, that have partially or well-developed


366 R. E. Johnston

abilities to understand that other individuals have different knowledge or inten-

tions than themselves; that is, they have a so-called ‘theory of mind’ (Premack &

Woodruff, 1978; Wimmer & Perner, 1983; Cheney & Seyfarth, 1990a,b; Ch. 25). One

crucial task is to identify the simplest level of representation that can explain a

particular phenomenon because any phenomena could be explained by a more

complex mechanism. Identifying the simplest mechanism that could be used can

provide a starting place for a taxonomy of the kinds of representation that indi-

viduals have of other individuals, species differences in these representations and

hypotheses about the neural mechanisms underlying them. Ultimately, it may

help us to understand the evolution of social behaviour.

The complexity of representations that individuals have of other individuals is

important in the context of communication networks because these representa-

tions will influence the way in which individuals interact with others, the kinds

of information they extract from observing interactions between others and the

effects that such information has on their own behaviour. This, in turn, will influ-

ence the nature of the networks that develop and the ways that individuals interact

within those networks (Chs. 24 and 25). I suggest that animals with complex, in-

tegrated representations of individuals will have a number of advantages over

those with simpler mechanisms. For example, complex representations contain

more information and should reduce errors in recognition, especially recognition

over long intervals during which changes in relevant cues may have occurred

through age, injury, nutrition or hormonal status. More complex representations

may also facilitate the evaluation of relationships between two or more individuals

obtained via observation of interactions.


The observations described in this chapter suggest that individuals in

some species obtain information from over-marks about the relative quality of

the individuals that deposited these marks and that this information influences

subsequent responses to those individuals, thus providing evidence for eavesdrop-

ping (but see Ch. 11). In addition, some species appear to have specialized sensory

mechanisms for the evaluation of scent over-marks and specialized mechanisms

that promote accurate placement of a scent mark over the scent of another in-

dividual (targeting). At present there is little evidence for audience effects on

scent marking, but this may be because few observers have looked for such ef-

fects. More studies are desperately needed on the functions of scent over-marking,

eavesdropping on over-marks and the effects of such eavesdropping, especially

observations and experiments in natural settings. Also important are compara-

tive studies on a set of related species with a rich, diverse repertoire of scent-

marking behaviours. Little is understood about the variety of functions that scent



over-marking has or about the kinds of information that may be obtained from

such over-marks.

Studies of eavesdropping on singing interactions in birds suggest some inter-

esting possibilities for further experiments with scent over-marking. For exam-

ple, use of information gathered by eavesdropping to direct extra-pair behaviour

(Mennill et al., 2002; Chs. 2 and 7). Among mammals, similar behaviour may be

found in monogamous pairs. In species in which females usually mate with sev-

eral partners, individual females might reduce their interactions with additional

males if the first mating partner was known to be highly successful in over-mark

competitions; in contrast, females that live near males that are less successful in

over-marking competitions might be more vigorous in advertising for, or in seek-

ing out, other males. That is, a female might be influenced by knowledge of the

history of over-marking interactions between numerous males in her vicinity.

Acknowledgements

Thanks to E. Regan, Peter McGregor, Jane Hurst and two anonymous reviewers for

comments on the manuscript. Thanks to Joan Johnston for help with graphics and other technical

help.

References

Bagneres, A.-G., Killian, A., Clement, J.-L. & Lange, C. 1991. Interspecific recogntion

among termites of the genus Reticulitermes: evidence for a role for the cuticular

hydrocarbons. Journal of Chemical Ecology, 17, 2397–2420.

Barrows, E. M., Bell, W. J. & Michner, C. D. 1975. Individual odor differences and their

social functions in insects. Proceedings of the National Academy of Sciences, USA, 72,

2824–2828.


matching between neighboring song sparrows. Animal Behaviour, 51, 917–923.

Bekoff, M. 1979. Scent marking by free-ranging domestic dogs: olfactory and visual

components. Biology of Behavior, 4, 123–139.

Bronson, F. H. 1976. Urine marking in mice: causes and effects. In: Mammalian

Olfaction, Reproductive Processes and Behavior, ed. R. L. Doty. New York: Academic

Press, pp. 119–141.

Brown, R. E. & Macdonald, D. W. 1985. Social Odours in Mammals, Vols. 1 & 2. Oxford:


Byers, J. A. & Bekoff, M. 1986. What does ‘kin recognition’ mean? Ethology, 72, 342–345.

Caldwell, R. L. 1985. A test of individual recognition in the stomatopod, Gonodactylus

festae. Animal Behaviour, 33, 101–106.

Cheney, D. L. & Seyfarth, R. M. 1990a. How Monkeys See the World. Chicago, IL: University

of Chicago Press.

1990b. The representation of social relations by monkeys. Cognition, 37, 167–196.


368 R. E. Johnston

Cohen, A. B., Johnston, R. E. & Kwon, A. 2001. How golden hamsters (Mesocricetus

auratus) discriminate top from bottom flank scents in over-marks. Journal of


Eisenberg, J. F. & Kleiman, D. 1972. Olfactory communication in mammals. Annual


Estes, R. D. 1967. The comparative behavior of Grant’s and Thomson’s gazelles. Journal

of Mammalogy, 48, 189–209.

1969. Territorial behavior of the wildebeest (Connochaetes taurinus Bruchell, 1823).


Ferkin, M. H., Dunsavage, J. & Johnston, R. E. 1999. What kind of information do

meadow voles, Microtus pennsylvanicus, use to distinguish between the top and

bottom scent of an overmark? Journal of Comparative Psychology, 113, 43–51.

Fischer, R. B. & McQuiston, J. 1991. A possible role for Syrian hamster, Mesocricetus

auratus, vaginal secretion in inter-female competition. Animal Behaviour, 42,

949–954.

Fisher, H. S., Swaisgood, R. R. & Fitch-Snyder, H. 2003. Countermarking by male pygmy

lorises (Nycticebus pygmaeus): do females use odor cues to select mates with high

competitive ability? Behavioral Ecology and Sociobiology, 53, 123–130.

Gamboa, G. J., Grudzien, T. A., Espelie, K. E. & Bura, E. A. 1996. Kin recognition

pheromones in social wasps: combining chemical and behavioural evidence.


Gorman, M. L. 1976. A mechanism for individual recognition by odour in Herpestes

auropunctatus (Carnivora: Viverridae). Animal Behaviour, 24, 141–145.

1990. Scent marking strategies in mammals. Revue suisse de Zoologie, 97, 3–29.

Gorman, M. L. & Mills, M. G. L. 1984. Scent marking strategies in hyaenas (Mammalia).

Journal of Zoology, 202, 535–547.

Gosling, L. M. & Roberts, S. C. 2001. Scent-marking by male mammals: cheat-proof

signals to competitors and mates. Advances in the Study of Behavior, 30, 169–217.

Gosling, L. M., Roberts, S. C., Thorton, E. A. & Andrew, M. J. 2000. Life history costs of

olfactory status signalling in mice. Behavioral Ecology and Sociobiology, 48,

328–332.

Heymann, E. W. 1998. Sex differences in olfactory communication in a primate, the

moustached tamarin, Saguinus mystax (Callitrichinae). Behavioral Ecology and


Hurst, J. L. 1990. Urine marking in populations of wild house mice Mus domesticus

Rutty. II. Communication between females. Animal Behaviour, 40, 223–232.

Hurst, J. L. & Rich, T. J. 1999. Scent marks as competitive signals of mate quality. In:

Advances in Chemical Signals in Vertebrates, ed. R. E. Johnston, D. Muller-Schwarze &

P. W. Sorensen. New York: Kluwer Academic/Plenum, pp. 209–225.

Johnson, R. P. 1973. Scent marking in mammals. Animal Behaviour, 21, 521–535.

Johnston, R. E. 1975a. Scent marking by male hamsters: I. Effects of odors and social

encounters. Zeitschrift fur Tierpsychologie, 37, 25–98.

1975b. Scent marking by male hamsters II. The role of flank gland odor in the

causation of marking. Zeitschrift fur Tierpsychologie, 37, 138–144.



1975c. Scent marking by male hamsters III. Behavior in a semi-natural environment.


1977. The causation of two scent marking behaviour patterns in female golden

hamsters. Animal Behaviour, 25, 317–327.

1979. Olfactory preferences, scent marking and ‘proceptivity’ in female hamsters.


1985. Communication. In: The Hamster: Reproduction and Behavior, ed. H. I. Siegel. New

York: Plenum Press, pp. 121–124.

1990. Chemical communication golden hamsters: from behavior to molecules and

neural mechanisms. In: Contemporary Issues in Comparative Psychology, ed. D. A.

Dewsbury. Sunderland, MA: Sinauer, pp. 381–409.

1993. Memory for individual scent in hamsters (Mesocricetus auratus) as assessed by

habituation methods. Journal of Comparative Psychology, 107, 201–207.

1995. Responses to individual signatures in scent counter-marks: a striking new

phenomenon. In: Chemical Signals in Vertebrates 7, ed. R. Apfelbach, D.

Muller-Schwarze, K. Reutter & E. Weiler. Oxford: Elsevier, pp. 361–369.

1999. Scent over-marking: how do hamsters know whose scent is on top and why

should it matter? In: Advances in Chemical Communication in Vertebrates, ed. R. E.

Johnston, D. Muller-Schwarze & P. W. Sorensen. New York: Kluwer

Academic/Plenum, pp. 227–238.

Johnston, R. E. & Bhorade, A. 1998. Perception of scent over-marks: novel mechanisms

for determining which individual’s mark is on top. Journal of Comparative


Johnston, R. E. & Bullock, T. A. 2001. Individual recognition by use of odours in golden

hamsters: the nature of individual representations. Animal Behaviour, 61,

545–557.

Johnston, R. E. & Jernigan, P. 1994. Golden hamsters recognize individuals, not just

individual scents. Animal Behaviour, 48, 129–136.

Johnston, R. E. & Schmidt, T. 1979. Responses of hamsters to scent marks of different

ages. Behavioral and Neural Biology, 26, 64–75.

Johnston, R. E., Derzie, A., Chiang, G., Jernigan, P. & Lee, H.-C. 1993. Individual scent

signatures in golden hamsters: evidence for specialization of function. Animal

Behaviour, 45, 1061–1070.

Johnston, R. E., Chiang, G. & Tung, C. 1994. The information in scent over-marks of

golden hamsters. Animal Behaviour, 48, 323–330.

Johnston, R. E., Munver, R. & Tung, C. 1995. Scent counter marks: selective memory for

the top scent by golden hamsters. Animal Behaviour, 49, 1435–1442.

Johnston, R. E., Sorokin, E. S. & Ferkin, M. H. 1997a. Female voles discriminate males’


1997b. Scent counter-marking by male meadow voles: females prefer the top-scent

male. Ethology, 103, 443–453.

Jolly, A. 1966. Lemur Behavior. Chicago, IL: University of Chicago Press.

Kappeler, P. M. 1998. To whom it may concern: the transmission and function of

chemical signals in Lemur catta. Behavioral Ecology and Sociobiology, 42, 411–421.


370 R. E. Johnston

Lazaro-Perea, C., Snowdon, C. T. & Arruda, M. F. 1999. Scent-marking behavior in wild

groups of common marmosets (Callithrix jacchus). Behavioral Ecology and


Lisk, R. D., Ciaccio, L. A., & Catanzaro, C. 1983. Mating behaviour of the golden

hamster under seminatural conditions. Animal Behaviour, 31, 659–666.



Mertl, A. S. 1977. Habituation to territorial scent marks in the field by Lemur catta.

Behavioral Biology, 21, 500–507.

Mills, M. G. L., Gorman, M. L. & Mills, M. E. J. 1980. The scent marking behaviour of the

brown hyena Hyaena brunnea. South African Journal of Zoology, 15, 240–248.

Moore, J. D. & Byers, J. A. 1989. The function of scent marking by males on female

urine in pronghorns. Journal of Mammalogy, 70, 812–814.

Peters, R. & Mech, L. D. 1975. Scent marking in wolves. American Scientist, 63,

628–637.

Powzyk, J. A. 2002. Scent wars. Natural History, 111, 82–85.

Premack, D. & Woodruff, G. 1978. Does the chimpanzee have a theory of mind?

Behavioral and Brain Sciences, 4, 515–526.

Ralls, K. 1971. Mammalian scent marking. Science, 171, 443–449.

Rasa, O. A. E. 1973. Marking behaviour and its social significance in the African dwarf

mongoose, Helogale undulata rufula. Zeitschrift fur Tierpsychologie, 32, 293–318.

Rich, T. J. & Hurst, J. L. 1998. Scent marks as reliable signals of the competitive ability

of mates. Animal Behaviour, 56, 727–735.

1999. The competing countermarks hypothesis: reliable assessment of competitive

ability by potential mates. Animal Behaviour, 58, 1027–1037.

Roberts, S. C. 1998. Behavioural responses to scent marks of increasing age in

klipspringer Oreotragus oreotragus. Ethology, 104, 585–592.

2000. Female territoriality and the function of scent-marking in a monogamous

antelope (Oreotragus oreotragus). Behavioral Ecology and Sociobiology, 47, 417–423.

Singer, A. G., Beauchamp, G. K. & Yamazaki, K. 1997. Volatile signals of the major

histocompatibility complex in male mouse urine. Proceedings of the Naional


Sliwa, A. & Richardson, P. R. K. 1998. Responses of aardwolves, Proteles cristatus,

Sparrman 1783, to translocated scent marks. Animal Behaviour, 86, 137–146.

Smith, T. E., Tomlinson, A. J., Mlotkiewicz, J. A. & Abbott, D. H. 2001. Female marmoset

monkeys (Callithrix jaccus) can be identified from the chemical composition of

their scent marks. Chemical Senses, 26, 449–458.

Steel, E. 1984. Effect of the odour of vaginal secretion on non-copulatory behaviour of

male hamsters (Mesocricetus auratus). Animal Behaviour, 32, 597–608.

Stowe, M. K., Turlings, T. C. J., Loughrin, J. H., Lewis, W. J. & Tumlinson, J. H. 1995. The

chemistry of eavesdropping, alarm and deceit. Proceedings of the National Academy of

Sciences, USA, 92, 23–28.

Vandenbergh, J. G. 1973. Effects of gonadal hormones on the flank gland of the golden

hamster. Hormone Research, 4, 28–33.



Wilcox, R. M. & Johnston, R. E. 1995. Scent counter-marks: specialized mechanisms of

perception and response to individual odors in golden hamsters, Mesocricetus

auratus. Journal of Comparative Psychology, 109, 349–356.

Wimmer, H. & Perner, J. 1983. Beliefs about beliefs: representation and constraining

function of wrong beliefs in young childrens’s understanding of deception.

Cognition, 13, 103–128.

Woodward, R. L. J., Bartos, K. & Ferkin, M. H. 2000. Meadow voles (Microtus

pennsylvanicus) and prairie voles (M. ochrogaster) differ in their responses to

over-marks from opposite- and same-sex conspecifics. Ethology, 106, 979–992.


17

Vocal communication networks in largeterrestrial mammals

k a r e n m c c o m b & dav i d r e b y

University of Sussex, Falmer, Brighton, UK

Introduction

Many mammals give long-range calls that can be received over wide areas,

often containing large numbers of receivers. In the case of mammals with fluid so-

cial systems, opportunities for exposure to the calls of others are further enhanced

by the movement of individuals with respect to one another. In our chapter, we

discuss the relevance of eavesdropping and communication networks in a range

of mammal species, first considering how these concepts apply in cases where

loud calls are used to exchange social information in static territorial and fluid

fission–fusion societies, and then exploring their potential importance where

mammals use loud sexual calls to broadcast information about resource-holding

potential. We also outline the mechanisms by which information in mammalian

calls is encoded, broadcast and acquired, and we consider the possible fitness con-

sequences that attending to calling interactions can confer. Finally, we evaluate

how the vocal communication networks described for non-human mammals differ

from human communication networks and discuss possible explanations for these

differences.

When mammals give loud calls, the area over which the signal can be re-

ceived is potentially extensive. Such calls are typically emitted at high sound-

pressure levels (greater than 100 dB at 1 m) and while spherical spreading and

excess attenuation from the environment eventually result in the signal being

engulfed in background noise, it often remains intelligible over distances of

several kilometres from the source: for example the calls of lions Panthera leo

(Ogutu & Dublin, 1998; Funston, 1999; K. McComb, unpublished data), hyaenas

Crocuta crocuta (Ogutu & Dublin, 1998) and elephants Loxodonta africana (McComb


372


Vocal networks in large terrestrial mammals 373

et al., 2003). Many mammals occur at relatively high densities; therefore, the ac-

tive space may contain a large number of potential receivers. In addition, some

mammals live in fluid social systems where there are unusual opportunities for

exposure to vocal signals from others in the population. In such social systems,

where the identity of immediate neighbours constantly changes because of the

movement of individuals and groups in relation to one another, an individual may

broadcast to and receive signals from a larger section of the overall population

than is normally possible in territorial systems. These two important characteris-

tics of mammal signalling systems have the potential to generate a much wider

audience for the signaller.

On the basis of the characteristics outlined above, it seems appropriate to con-

ceptualize the production and perception of mammal loud calls in the context

of an array of several receivers (after McGregor & Dabelsteen, 1996). Indeed we

will argue, based on the examples presented in our review, that this is the best

way to view communication involving loud calls. There is direct evidence that

mammals attend to vocal signals that are not explicitly directed at them. For ex-

ample, elephant contact calls, although directed at family and bond group mem-

bers, are attended to by others in the population, who exhibit knowledge of these

calls and adjust their social behaviour on the basis of them (McComb et al., 2000).

Diana monkeys Cercopithecus diana attend to the alarm calls of another primate

(Campbell’s monkey Cercopithecus campbelli) and appear to obtain functionally rel-

evant information from the detailed combination of different vocalizations used

(Zuberbuhler, 2002, see also below). However, in considering the extent to which

this form of audience effect (Ch. 4) in loud-calling mammals involves ‘eavesdrop-

ping’ or constitutes a ‘communication network’, three important issues need to

be considered.

First, eavesdropping in the context of animal signalling has been defined as

‘extracting information from an interaction between other individuals’(McGregor

& Dabelsteen, 1996; see also Ch. 2). This is a technical definition of a term that

in colloquial usage implies more specifically that receivers ‘secretly listen to a

conversation’(Concise Oxford English Dictionary). While behavioural acts constituting

secrecy or deception are notoriously difficult to identify in mammals (Semple &

McComb, 1996), it seems important to distinguish between cases of eavesdropping

in which transmission of information to receivers other than the main recipient

would be selected for and those where it would not. This can be achieved by

conducting cost–benefit analyses of particular signalling interactions (e.g. Ch. 3).

Second, an important aspect of the McGregor & Dabelsteen (1996) definition

of eavesdropping is that it involves extracting information from ‘an interaction’

rather than simply attending to the call itself. At this stage, evidence that mammals

attend to the signalling interaction (rather than just the signal) is very sparse. An


374 K. McComb & D. Reby

isolated example is provided by the studies of Cheney and colleagues (Cheney

et al., 1995; Ch. 25) on baboons Papio cynocephalus ursinus, which demonstrated

that receivers distinguish between appropriate and anomalous vocal exchanges

between dominants and subordinates in their group. However, lack of evidence

for attention to the vocal interaction itself may be more apparent than real. Few

researchers other than Cheney and colleagues have conducted the appropriate

experiments specifically to examine this phenomenon.

Finally, while it is certainly the case that a system of mammal communication

linking signallers to several receivers has some properties of a network, it lacks

others. A network can be viewed simply as ‘a system of interconnected people

or things’. However, advanced networks such as human social networks or the

Internet are generally understood to involve the passage of information from one

remote part of the network to another, via intermediate recipients that pass on

information to other individuals. Non-human vocal communication (including

bird and mammal communication systems) appears to fall short of this and we

will consider possible explanations for this in the course of the review.

Loud calls and social behaviour

Availability of information in a simple territorial system

The typical nature of mammal loud calls that are used to mediate so-

cial behaviour suggests that selection for concealing information from unwanted

receivers has not been paramount. Their high sound-pressure level, abrupt on-

set and broadband (often noisy) nature are properties that would be expected to

make them easy to detect and locate by listeners (e.g. Brown et al., 1979, 1980).

Given that such calls often function to advertise territory ownership, it would

usually be advantageous for them to attract the attention of any conspecifics in

the vicinity. There is evidence from a range of mammals that individuals can

distinguish between familiar and unfamiliar callers, even where receivers are

separated from signallers by large distances relative to the size of the animal:

for example pikas Ochotona princeps (Conner, 1985), cotton-top tamarins Sanguinus

oedipus (Snowdon et al., 1983) mangabeys Cercocebus albigena (Waser, 1977), rhesus

macaques Macaca mulatta (Rendall et al., 1996), wolves Canis lupus (Tooze et al., 1990),

and lions (McComb et al., 1993; Grinnell & McComb, 2001).

Where calls characteristics are adapted for long-distance transmission and are

easy to locate, the only mechanism for withholding information from unwanted

receivers would be to suppress calling altogether. The potential for loud calls to

attract unwanted attention may well be considerable in social mammals that use

loud calls not only in territorial defence but also to maintain contact with widely

spaced social companions: for example wolves (Harrington & Mech, 1979), lions



(McComb et al., 1994; Grinnell & McComb, 1996), and chimpanzees Pan troglodytes

(Mitani & Nishida, 1993). Here particular classes of individual that cannot afford

to risk escalated encounters with competitors that might hear their calling could

benefit by remaining silent even though, by doing so, they would forfeit the ben-

efits of coordinating their movements with members of their own social group.

The behaviour of free-ranging nomadic male lions in the Serengeti National Park

is consistent with these predictions (Grinnell & McComb, 2001).

Prides of African lions consist of matrilineal kin groups of females, their depen-

dent offspring and a coalition of resident males that enter the pride from outside

(Packer et al., 1988). In the pride, both sexes use loud calls (roaring) to advertise

ownership of a territory and to stay in contact with other members of their so-

cial group (Schaller, 1972; McComb et al., 1994; Grinnell et al., 1995; Grinnell &

McComb, 1996). At any one time, however, a high proportion of male lions in the

population are not in possession of a pride. These ‘nomads’wander widely, passing

through pride ranges singly or in coalitions until they are successful in taking over

a pride of their own (Hanby & Bygott, 1987; Pusey & Packer, 1987). It is crucial for

nomadic males to gain and maintain social bonds with their companions while

they wander because success in competition for prides is strongly dependent on

group size (Bygott et al., 1979; Grinnell et al., 1995). Roaring provides a means by

which nomadic males might coordinate their movements with coalition partners

or recruit new ones (see also McComb et al., 1994; Grinnell et al., 1995). However,

if nomads used this loud, long-distance signal to communicate with social com-

panions, they would also advertise their position to resident males in the area.

Nomadic males are likely to pay high costs if they attract the attention of

resident males in the area. Resident males have been consistently shown to ap-

proach aggressively playbacks of roaring from strange males that are broadcast in

their territories (Grinnell et al., 1995; Grinnell & McComb, 2001) and intercoalition

encounters can be fatal (Schaller, 1972; Grinnell et al., 1995). Given these costs, no-

mads might benefit by reducing their rate of roaring or even abandoning roaring

altogether and concealing their presence – despite the detrimental effects that

this would be likely to have on their ability to maintain contact with coalition

partners and attract potential mates. Grinnell & McComb (2001) found that in the

Serengeti population only male lions that were resident in a pride ever roared.

Nomadic males were never observed roaring when they were followed at night,

even when they became separated from their coalition partners. They also failed

to roar when played recordings of unfamiliar males roaring. In contrast, resident

males maintained a high rate of roaring in both these circumstances.

There are two possible explanations for why nomadic male lions fail to roar:

first, non-resident males gain no benefits from roaring and so never do so; second,

non-resident males could benefit from roaring, particularly by enhancing their



ability to recruit and maintain contact with coalition partners (McComb et al.,

1994; Grinnell et al., 1995), but the costs of engaging in this behaviour outweigh

the benefits. If the first explanation is true, then nomadic males should never roar

under any circumstances, whereas the second explanation predicts that nomadic

males will only roar when the probability of incurring costs, specifically attracting

the attention of resident males in the area, is low. Observational studies at other

field sites suggest that the second explanation is correct (Grinnell & McComb,

2001). Funston (1999), working on nomadic male lions in Kruger National Park,

South Africa, found that nomadic males do sometimes roar, but at greatly reduced

rates in comparison with resident males. Of the three nomadic coalitions that he

followed, one was explicitly noted to spend most of their time in an area without

resident males and thus where the social costs to roaring would be reduced. In

addition, observations by Grinnell in Pilanesberg National Park, South Africa sug-

gested that here, too, nomadic males roar when local resident males are unlikely to

hear them. Pilanesberg is an ancient volcanic caldera that contains valleys which

are acoustically isolated from each other by mountainous ridges. A non-resident

male coalition was observed roaring in one of these valleys that was not occupied

by resident males or females (Grinnell & McComb, 2001). It is also important to

note that, while nomadic males in the Serengeti study did not roar, males that

had been nomadic were seen to begin roaring as soon as they launched a chal-

lenge for ownership of a pride (Grinnell & McComb, 2001). This emphasizes that

roaring is a flexible behaviour that signallers may have been selected to adjust ac-

cording to the potential costs and benefits of revealing information on location to

listeners.

There are reports from other species of low signalling rates, or suppression of

signals altogether, in situations in which conspicuous signals could attract the

attention of potential aggressors. Chimpanzees have been observed to remain un-

usually quiet during excursions into the territories of other communities (Goodall,

1986) and, when they hunt monkeys (Colobus and Cercopithecus spp.), are reported

to fall silent on hearing the prey’s calls (Boesch & Boesch-Achermann, 2000). Lone

wolves howl less than do territorial pairs and packs (Harrington & Mech, 1979).

Similarly, transient coyotes Canis latrans howl at greatly reduced rates compared

with residents when passing through others’ territories (Gese & Ruff, 1998). It is

interesting that resident male lions may also adopt an apparently stealthy strat-

egy when ranging outside their territory. Grinnell & McComb (2001) noted that

resident males that had ventured well beyond their territory boundaries never

roared even when missing male companions. Long-distance signalling may well

be controlled in similar ways in other social species where eavesdroppers can

impose high costs on signalling (see also Ch. 4). Recent work on transient killer

whales (Orcinus orca) suggests that these animals adjust their calling behaviour to



minimize the costs of being detected by their acoustically sensitive mammalian

prey (Deecke, 2003).

Availability of information in mammals with fluid social systems

Above we have presented evidence that information on caller character-

istics such as identity is potentially available over quite long distances in species

that use loud calls for social communication within territorial systems. Nonethe-

less, degradation of calls with distance from the source will eventually result in

such information being engulfed in background noise and lost to receivers. In

certain mammal social systems, however, receivers are not limited to learning

only the calls of individuals in their own group or of particular territorial neigh-

bours within their hearing range. Some mammal societies are highly fluid, with

individuals and social units moving freely with respect to each other and ranging

widely. In these fluid societies, individuals pass through the signalling ranges of a

much larger number of conspecifics and are provided with opportunities to learn

to recognize the vocalizations of many more signallers than just their immediate

neighbours (see also discussions in Chs. 20 and 25). If mental capacities for storing

information on the identity of conspecifics’ signals are adequate, these circum-

stances would provide individuals with opportunities to become familiar with

the signals of many different conspecifics that form part of a widespread popula-

tion. Thus in mammals with fluid social systems, the unusually high encounter

rates that individuals have with conspecifics should interact with long-distance

signalling abilities to increase greatly the opportunities that receivers have for

learning to recognize the vocalizations of other individuals in the population. A

number of large mammals, including some primates (e.g. chimpanzees: Boesch

& Boesch-Achermann, 2000), cetaceans (e.g. sperm whales Physeter macrocephalus:

Whitehead et al., 1991) and African elephants use long-distance signals for social

communication and have fluid social systems.

In African elephants, the closest social relationships exist between members

of a family unit, typically composed of adult females that are matrilineal rela-

tives and their immature offspring, and between bond groups of families that

associate frequently and often greet one another when they meet (Moss & Poole,

1983). However, individual family units move freely with respect to one another

and range widely, frequently coalescing with other family units in the popula-

tion as they move and feed, thus forming highly fluid fission–fusion societies

(Moss & Poole, 1983). In a population of elephants in Amboseli National Park,

Kenya, with known life histories and ranging patterns, the extent to which fe-

male subjects were capable of recognizing others in the population through

long-distance contact calls was evaluated from playback experiments (McComb

et al., 2000). These experiments demonstrated that female African elephants not



only give a characteristic reaction to the contact calls of family or bond group

members but also can discriminate between the calls of less-frequent associates,

distinguishing the calls of individuals in this category with whom they have

higher association indices from those with whom they have lower association

indices. Based on the association indices involved, McComb et al. (2000) estimated

that subjects would have to be familiar with the contact calls of a mean of 14

different families (including about 100 adult females) in order to perform this

discrimination.

Empirical studies of the extent of networks of vocal recognition in cetaceans

and primates, which are currently lacking, may reveal similar patterns. Networks

of vocal recognition are likely to be particularly extensive where individuals

are long lived and social knowledge can be accumulated over considerable time

periods.

Loud calls and sexual behaviour

Many large mammals have loud calls that function to attract individuals

of the opposite sex and advertise resource-holding potential to competitors in

the vicinity (e.g. Clutton-Brock & Albon, 1979; Tyack, 1981 ; McElligott et al., 1999).

These calls are often very conspicuous and seem specifically adapted for attracting

the attention of a wide audience. The loud reproductive calls of polygynous deer,

which typically serve several functions, provide some of the best examples.

Loud mating calls in deer

Male red deer Cervus elaphus roar at high rates during the autumn breeding

season or rut and these loud vocalizations are known not only to affect the outcome

of contests between males (Clutton-Brock & Albon, 1979; Reby & McComb, 2003b)

but also to influence mate attraction (McComb, 1991) and advance ovulation in

females (McComb, 1987). There are consequently several receivers to whom male

roars might be relevant, including other males, the signaller’s own harem of fe-

males and other potential mates within hearing range. Video footage of red deer

stags orientating their responses to the roars of several neighbours with distinct

spatial locations clearly indicates that they take the complex spatial distribution

of callers around them into account (D. Reby & K. McComb, personal observation).

Moreover, it has been shown that female red deer are able to discriminate between

the roars of their own stag and those of surrounding males (Reby et al., 2001).

Finally, there is some evidence that when signallers would benefit by advertising

the outcome of their interactions, they use particularly conspicuous vocalizations.

Roaring bouts given during roaring competitions with rival stags and after chasing

hinds tend to contain a high proportion of ‘harsh roars’,which are unusually loud



and easy to locate and have an acoustic structure that emphasizes the caller’sbody

size (Reby & McComb, 2003b).

Groaning in fallow deer bucks Dama dama also appears adapted for more than

one category of receiver and each of these must be considered when modelling the

vocal behaviour of callers (McElligott & Hayden, 1999, 2001; McElligott et al., 1999).

McElligott et al. (1999) found that the bucks that achieved most matings were those

who had initiated vocal activity early in the season and who had remained vocal on

most days. This led the authors to conclude that females may discriminate between

males on the basis of long-term cumulative investment in vocal activity. However,

although rates of groaning were higher when females were present, males with

females exhibited higher groaning rates in the presence of nearby vocal males,

suggesting that the signal was also a threat aimed at male rivals (McElligott &

Hayden, 1999).

Given the several functions of deer vocalizations, it is clear that there are situa-

tions in which signals that would be beneficial in one context may be costly in an-

other: for example, when an individual male could gain reproductive advantages

by signalling to attract mates but in doing so would invite escalated contests with

male competitors. We have observed that young red deer stags (four to five year

old) who have gained access to a harem of females while the mature harem holder

is temporarily absent, and who have started to roar, will rapidly fall silent when

the harem holder returns, often dropping their heads to feed as he approaches

(K. McComb & D. Reby, personal observation). Similarly, playback experiments on

fallow deer (Komers et al., 1997) have shown that immature males decrease their

rate of groaning in response to playbacks of groans from mature males, whereas

mature males increase their groaning rates in this situation. Fallow deer bucks

may, therefore, adjust groaning rate in relation to several receivers, responding to

the complex balance between the benefits of deterring other males and displaying

to females and the costs of inviting contests with potentially stronger males in

the vicinity.

Since red and fallow deer rutting calls are individually distinct (McComb, 1988;

Reby, 1998; Reby et al., 1998), females and males may be able to recognize individual

callers from their vocalizations and accumulate knowledge on both a signaller’s

short-term vocal interactions with others and it’s long-term calling behaviour. Re-

search on red deer has revealed that females can discriminate between the roars

of their own stag and those of neighbouring harem holders (Reby et al., 2001). It

is possible that red deer hinds could receive information from roaring exchanges

and move between harems accordingly. Stags may also attend to contests between

other males for information on the body size and motivational state and adjust

their decisions to challenge harem holders on this basis. In this context, all re-

ceivers, whether they are directly involved in an interaction with the caller or not,



would benefit from attending to cues to resource-holding potential madeavailable

in this way. What is now required is empirical work to investigate directly the ex-

tent to which receivers attend to interactions in which they are not themselves

involved.

Loud mating calls in other mammals

Loud and acoustically complex sexual songs produced by humpback

whales Megaptera novoeangliae (Payne & McVay, 1971) and fin whales Balaenoptera

physalus (Croll et al., 2002) have the potential to travel unprecedented distances

underwater (e.g. Croll et al., 2002). While it is known that the individuals who give

these songs are male, the intended receivers have not yet been unambiguously

identified; they may be rival males, potential mates or both (Tyack, 1983; Mobley

et al., 1988; Noad et al., 2000; Croll et al., 2002). What is clear is that such vocaliza-

tions are detectable over vast tracts of ocean and may reach a much larger audience

than the sexual calls of terrestrial mammals discussed above. Male pinnipeds also

have loud sexual advertisement calls (e.g. Northern elephant seals Mirounga angu-

stirostris (Shipley et al., 1981, and common seals Phoca vitulina (van Parijs et al., 2000))

and calling interactions between males on land or underwater may be attended

to by rival males, potential mates or both. It remains to be seen whether receivers

alter their subsequent behaviour on the basis of which male dominates in a calling

interaction (see fuller discussion in Ch. 18).

Mammal anti-predator calls

In contrast to long-distance social and sexual calls given by large mam-

mals, alarm calls typically have acoustic features that would be expected to make

them difficult to locate. While these calls may be delivered at moderate ampli-

tudes, the information that they contain is likely to be available over shorter dis-

tances. Despite this, they are clearly attended to by a range of receivers, including

members of other mammal species (Schaller, 1967; Oda, 1997; Zuberbuhler, 2002).

In responding to the alarm calls of Campbell’smonkeys, Diana monkeys attend not

only to the referent of the alarm call, responding with their own species-specific

alarm call for the same predator, but also appear sensitive to the detailed composi-

tion of the alarm-calling sequence. In situations where the presence of a predator

is less threatening, Campbell’s monkeys emit a pair of ‘boom’ calls before their

alarm calls. Playbacks of Campbell’s alarm calls with booms did not elicit alarm

calls from Diana monkey subjects (Zuberbuhler, 2002).

Some anti-predator calls may have an even wider audience. A study of roe

deer Capreolus capreolus revealed that barks, previously identified as ‘alarm calls’,

in fact function to elucidate the cause of disturbance (Reby et al., 1999a). In this



communication system, calls inform any predator that might be present that it has

been detected and simultaneously reveal the caller’sidentity and status to any con-

specific (whether the latter is the cause of the disturbance or not). The likelihood of

barking in response to a predator-like disturbance is independent of the presence

of (related or unrelated) conspecifics in the close vicinity, demonstrating that it is

not an alarm call (Reby et al., 1999a). However, barking is contagious, with one in-

dividual’s barks often being followed by antiphonal calling behaviour from up to

seven neighbouring individuals of both sexes (Reby et al., 1999a). Since the acoustic

structure of the vocalization carries information on the sex, age and identity of

the caller (Reby et al., 1999b), barking may enable roe deer to identify and locate

each other, and possibly assess dominance status (particularly during counter-

barking sessions involving several animals). Playback experiments supported the

hypothesis that although barking may have initially evolved as an anti-predator

strategy it is also a signal attended to by conspecific receivers, in particular other

males during the territorial period (Reby et al., 1999a). Therefore, when a roe deer

barks, irrespective of the stimulus that elicits it (predator or conspecific), it reveals

its location, identity and status to a diverse audience of receivers, the composi-

tion of which will have marked effects on the costs and benefits associated with

calling.

Encoding of information on individuality and size

Within a network, the ability of individuals to determine each other’s

identity, physical status or internal state from signals dramatically increases the

level of functionally relevant information that is potentially exchanged. Whereas

in some cases identity may be inferred from the location of the caller or by using

visual or olfactory signals, acoustic cues are likely to be of primary importance

when individuals range widely. Such cues can provide receivers with instanta-

neous information on the location and attributes of the caller and may represent

the only effective signalling modality in nocturnal or forest-dwelling species. There

is a considerable body of evidence indicating that the vocalizations of terrestrial

mammals contain information on the identity and physical attributes of the caller

(see below).

In principle, individual differences can be present at several levels in the acous-

tic structure of the call. When mammals give voiced calls, the resultant sound

is the product of a source signal, generated in the larynx, that is subsequently

filtered in the cavities of the vocal tract (Fant, 1960). The source–filter theory of

voice production separates the source components, generated by the vibration of

the vocal folds, from the filter components, generated when certain frequencies

in the source spectrum are selectively amplified or filtered as the signal passes



through the supralaryngeal vocal tract. The characteristics of the source include

the duration of the call, its fundamental frequency, the periodicity of the signal,

its spectral slope and the presence of phenomena associated with non-linear dy-

namics, such as subharmonics, biphonation and deterministic chaos (Wilden et al.,

1998). Differences in these characteristics of call structure arise from variation in

subglottal pressure and in the length and shape of the vocal folds and their stress

and tension. All of these parameters can vary between individuals, either as a re-

sult of differences in the way the larynx is operated or simply because of random

variation in the morphology of callers. In comparison, the key characteristics of

the filter are the position and bandwidths of the formant frequencies, which de-

scribe the shape of the spectral envelope. Formant frequencies are determined by

the length and shape of the cavities of the vocal tract, namely the pharynx, mouth

and nasal cavities. Individual differences in formant frequencies can arise from

differences in vocal tract morphology or from variation in the way the shape of

the vocal tract is actively modified during vocalization (e.g. the extent of mouth

opening, lip rounding and vocal tract extension).

Variation in source and filter characteristics both appear to be important in

encoding individual identity in a range of large mammals. Differences in the

fundamental frequency contour have been identified as important in broadcast-

ing information on identity in wolves (Tooze et al., 1990) and elephants (McComb

et al., 2003), while individuality in formant frequencies has been demonstrated in

fallow deer (Reby et al., 1998), roe deer (Reby et al., 1999b), red deer (McComb, 1988;

Reby, 1998), elephants (McComb et al., 2003) and rhesus macaques (Rendall et al.,

1998). Filter characteristics, in particular the frequency spacing between succes-

sive formants, provide the most reliable cues to body size (Fitch, 1997; Riede &

Fitch, 1999; Reby & McComb, 2003a). In contrast, source characteristics, in partic-

ular fundamental frequency values, provide relatively poor information on size

(Masataka, 1994; Reby & McComb, 2003b) but are better indicators of age and sex

and may, therefore, reflect important variation in vocal fold length between sexes

and throughout the lifetime (Reby & McComb, 2003b).

It is important to appreciate that source and filter characteristics that have the

potential to provide receivers with information on caller identity can be distorted

or lost as distance from the signaller increases. Even where calls can theoretically

be transmitted over long distances because they possess acoustic characteristics

that are well adapted for sound transmission in a particular environment, it is

unsafe to conclude that receivers can extract socially relevant information from

degraded calls at these distances. In female African elephants, playback exper-

iments and re-recordings indicate that abilities for social recognition through

long-distance contact calls become limited when frequency components around

115 Hz become immersed in background noise (McComb et al., 2003). This typically



occurs at distances of 1–2 km from the caller, which is considerably shorter than

the propagation distances that have been proposed for calls with infrasonic funda-

mental frequencies (McComb et al., 2003). This finding highlights the importance

of considering the distances over which vocal signals within communication net-

works can propagate without losing their intelligibility to receivers, which are

not necessarily equivalent to the distances over which such signals are physically

detectable (see also Ch. 20).

Acquiring and storing information on vocal characteristics

Little is known about the factors that influence how effectively individ-

uals acquire and store information about their social companions, although it is

known that social knowledge, particularly that used in vocal recognition between

mothers and offspring, can be retained for several years even when individuals do

not encounter each other (Insley, 2000; McComb et al., 2000). In African elephants,

where adult females are familiar with the contact calls of a large proportion of

the population around them (see above), the key factor that affects social discrim-

ination abilities is the age of the oldest female in the group (McComb et al., 2001).

Playback experiments revealed that families with older matriarchs were signifi-

cantly better at discriminating the contact calls of genuine strangers from those

of more familiar associates than were families with younger matriarchs (McComb

et al., 2001). While families with older matriarchs were several thousand times

more likely to bunch into defensive formation when played the calls of families

they have encountered only rarely than when played the calls of families they

frequently associate with, families with younger matriarchs were only marginally

more likely to bunch (McComb et al., 2001). Log-linear analysis revealed that vari-

ables such as the number of other females present in the group, and their respec-

tive ages, did not affect vocal discrimination abilities. An additional factor that did

appear to be important was the rate at which subjects encountered other families

in the population. An elephant family unit directly encounters, on average, 25

other families over the course of the year in Amboseli National Park, and passes

within 1–2 km of 35, providing family members with plenty of opportunity to

become familiar with the calls of others. Recent analyses suggest that having a

high encounter rate with others in the population can enhance a family’sability to

identify the calls of genuine strangers, and that this may be particularly beneficial

for families with younger matriarchs (K. McComb, unpublished data).

The above results suggest that the age of one crucial individual, the oldest

female or matriarch, can affect the social knowledge of the group as a whole. Age

and experience are likely to affect abilities to acquire and store information on

vocal signals in other societies where animals are long lived and remain part of



a social network for many years. The social systems of some whales have strong

parallels with those of elephants (e.g. Pennisi, 2001) and Ford et al. (1994) noted

that in killer whales the death of the oldest female, from whom many of the

individuals are usually descended, may destabilize a pod. The effects of age and

other factors on abilities to recognize the vocalizations of conspecifics has not been

investigated for species that defend individual territories (rather than sharing a

range with matrilineal relatives as in elephants and some whales) and studies

of this sort are now required. Moreover, we as yet know nothing of the extent

to which large mammals develop knowledge of the mating calls of others in the

population and the factors that affect the acquisition of this knowledge.

The fitness consequences of attending to the calls of others

There is some evidence that attending to the exchange of social calls

between other individuals can confer fitness benefits on receivers. In African ele-

phants, where the matriarch appears to act as a repository for information on

the calls of others in the population (see above), families with older matriarchs

have greater reproductive success, at least some of which appears to derive from

superior social knowledge (McComb et al., 2001). The fitness consequences of at-

tending to vocal interactions involving sexual calls have never been quantified

for mammals but are likely to be highly significant. Acquiring information on

resource-holding potential by monitoring the outcome of vocal contests may al-

low receivers to assess rivals and potential mates much more accurately, and to

benefit from better decisions made as a consequence.

Summary and back to definitions

In light of the examples discussed above, the term communication net-

work can be usefully employed to identify sets of links between individuals (not

necessarily contiguous) that are known to each other through vocal signals or

that acquire information about each other’s interactions through vocalizations.

However, animal communication networks appear to be limited in a number of

important respects (see the Introduction to this chapter). In their typical form, they

describe overlapping lattices each consisting of three individuals: the signaller,

the intended receiver and an extraneous listener. In such systems, extraneous lis-

teners do not normally interact – in particular they do not pass on information

that they gain from attending to interactions. This is in stark contrast with hu-

man communication networks, where information can be transferred from one

remote part of the network to another and where intermediate recipients may

not be the ultimate receivers. Several key constraints on vocal communication



may prevent non-human mammals from sharing information in this more com-

plex way. These include the limited size of mammalian vocal repertoires, the

rarity of fully referential calls and the limited productivity possible in the ab-

sence of duality of patterning – a unique feature of human language whereby

phonemes can be combined into words and words into sentences (Hockett, 1960;

Pinker, 1994). These characteristics are likely to have been selected for in the

course of the massive expansion in sociality and social fluidity that occurred

during human evolution, creating an environment where the ability to use sym-

bolism and syntax to communicate about displaced activities would be of great

importance. It is important to appreciate that once these abilities had evolved,

the fitness benefits of attending to the calls of others would increase by orders of

magnitude.

Acknowledgements

The research described was funded by grants from BBSRC, NERC, the Royal Society (all

to K.M.) and INRA (to D.R.). David Reby was supported by Fyssen and E. U. Marie Curie Fellowships

and the University of Sussex. We also thank Vincent Janik, Peter McGregor and an anonymous

referee for helpful comments on the first draft of the chapter.

References

Boesch, C. & Boesch-Achermann, H., 2000. The Chimpanzees of the Tai Forest. Oxford:


Brown, C. H., Beecher, M. D., Moody, D. B. & Stebbins, W. C. 1979. Locatability of vocal

signals in Old World monkeys: design features for the communication of

position. Journal of Comparative Physiological Psychology, 5, 806–819.

1980. Localization of noise bands by Old World monkeys. Journal of the Acoustical

Society of America, 68, 127–132.

Bygott, J. D., Bertram, B. C. R. & Hanby, J. P. 1979. Male lions in large coalitions gain

reproductive advantages. Nature, 282, 839–841.

Cheney, D. L., Seyfarth, R. M. & Silk, J. B. 1995. The responses of female baboons (Papio

cynocephalus ursinus) to anomalous social interactions: evidence for causal

reasoning? Journal of Comparative Psychology, 109, 134–141.

Clutton-Brock, T. H. & Albon, S. D. 1979. The roaring of red deer and the evolution of

honest advertisement. Behaviour, 69, 145–169.

Conner, D. A. 1985. Analysis of the vocal repertoire of adult pikas: ecological and

evolutionary perspectives. Animal Behaviour, 33, 124–134.

Croll, D. A., Clark C. W., Acevedo, A. et al. 2002. Only male fin whales sing loud songs.

Nature, 417, 809.

Deecke, V. B. 2003. The vocal behaviour of transient killer whales (Orcinus orca):

communicating with costly calls. Ph.D. Thesis, University of St Andrews.



Fant, G. 1960. Acoustic Theory of Speech Production. The Hague: Mouton.

Fitch, W. T. 1997. Vocal tract length and formant frequency dispersion correlate with

body size in rhesus macaques. Journal of the Acoustical Society of America, 102,

1213–1222.

Ford, J. K. B., Ellis, G. M. & Balcomb, K. C. 1994. Killer Whales. Vancouver: UBC Press;

Seattle, WA: University of Washington Press.

Funston, P. J. 1999. Predator–prey relationships between lions and large ungulates in

the Kruger National Park. Ph.D. Thesis, University of Pretoria.

Gese, E. M. & Ruff, R. L. 1998. Howling by coyotes (Canis latrans): variation among

social classes, seasons, and pack sizes. Canadian Journal of Zoology, 76,

1037–1043.

Goodall, J. 1986. The Chimpanzees of Gombe. Cambridge, MA: Belknap.

Grinnell, J. & McComb, K. 1996. Maternal grouping as a defense against potentially

infanticidal males: evidence from field playback experiments on African lions.


2001. Roaring and social communication in African lions: the limitations imposed

by listeners. Animal Behaviour, 62, 93–98.

Grinnell, J., Packer, C. & Pusey, A. E. 1995. Co-operation in male lions: kinship,

reciprocity or mutualism? Animal Behaviour, 49, 95–105.

Hanby, J. P. & Bygott, J. D. 1987. Why do subadult lions leave their natal pride? Animal

Behaviour, 35, 161–169.

Harrington, F. H. & Mech, L. D. 1979. Wolf howling and its role in territory

maintenance. Behaviour, 68, 207–249.

Hockett, C. F. 1960. The origin of speech. Scientific American, 203, 88–96.

Insley, S. J. 2000. Long-term vocal recognition in the northern fur seal. Nature, 406,

404–405.

Komers, P. E., Pelabon, C. & Stenstrom, D. 1997. Age at first reproduction in male

fallow deer: age-specific versus dominance-specific behaviors. Behavioral Ecology, 8,

456–462.

Masataka, N. 1994. Lack of correlation between body size and frequency of

vocalisations in young female Japanese macaques (Macaca fuscata). Folia

Primatologica, 63, 115–118.

McComb, K. 1987. Roaring by red deer stags advances the date of oestrus in hinds.

Nature, 330, 648–649.

1988. Roaring and reproduction in red deer, Cervus elaphus. D. Phil. Thesis,

University of Cambridge.

1991. Female choice for high roaring rates in red deer, Cervus elaphus. Animal


McComb, K., Pusey, A., Packer, C. & Grinnell, J. 1993. Female lions can identify

potentially infanticidal males from their roars. Proceedings of the Royal Society of


McComb, K., Packer, C. & Pusey, A. E. 1994. Roaring and numerical assessment in

contests between groups of female lions, Panthera leo. Animal Behaviour, 47,

379–387.



McComb, K., Moss, C., Sayialel, S. & Baker, L. 2000. Unusually extensive networks of

vocal recognition in African elephants. Animal Behaviour, 59, 1103–1109.

McComb, K., Moss, C., Durant, S. M., Baker, L. & Sayialel, S. 2001. Matriarchs act as

repositories of social knowledge in African elephants. Science, 292, 491–494.

McComb, K., Reby, D., Baker, L., Moss, C. & Sayialel, S. 2003. Long-distance

communication of social identity in African elephants. Animal Behaviour, 65,

317–329.

McElligott, A. G. & Hayden, T. J. 1999. Context-related vocalization rates of fallow

bucks, Dama dama. Animal Behaviour, 58, 1095–1104.

2001. Postcopulatory vocalisations of fallow bucks: who is listening? Behavioral

Ecology, 12, 41–46.

McElligott, A. G., O’Neill, K. P. & Hayden, T. J. 1999. Cumulative long-term investment

in vocalization and mating success of fallow bucks, Dama dama. Animal Behaviour,

57, 1159–1167.


Evolution of Acoustic Communication in Birds, ed. D. E. Kroodsma and E. H. Miller.


Mitani, J. C. & Nishida, T. 1993. Contexts and social correlates of long-distance calling

by male chimpanzees. Animal Behaviour, 45, 735–746.

Mobley, J. R., Herman L. M. & Frankel A. S. 1988. Responses of wintering humpback

whales (Megaptera novaeangliae) to playback of recordings of winter and summer

vocalizations and of synthetic sound. Behavioral Ecology and Sociobiology, 23,

211–223.

Moss, C. J. & Poole, J. H. 1983. Relationships and social structure of African elephants.

In: Primate Social Relationships: An Integrated Approach, ed. R. A. Hinde. Oxford:

Blackwell Scientific, pp. 315–325.

Noad, M. J., Cato, D. H., Bryden, M. M. et al., 2000. Cultural revolution in whale songs.

Nature, 408, 537.

Oda, R. 1997. The response of Verreaux’s sifakas to anti-predator alarm calls given by

sympatric ring tailed lemurs. Folia Primatologica, 69, 357–360.

Ogutu, J. O. & Dublin, H. T. 1998. The response of lions and spotted hyenas to sound

playbacks as a technique for estimating population size. African Journal of Ecology,

36, 83–95.

Packer, C., Herbst, L., Pusey, A. E. et al. 1988. Reproductive success of lions. In

Reproductive Success, ed. T. H. Clutter-Brock. Chicago, IL: University of Chicago

Press, pp. 363–383.

Payne, R. S. & McVay, S. 1971. Songs of humpback whales. Science, 173, 585–597.

Pennisi, E. 2001. Elephant matriarchs tell friend from foe. Science, 292, 417–418.

Pinker, S. 1994. The Language Instinct. New York: William Morrow.

Pusey, A. E. & Packer, C. 1987. The evolution of sex-biased dispersal in lions. Behaviour,

101, 275–310.

Reby, D. 1998. Etude de la communication sonore chez les cervides. Prise en compte de

la dimension individuelle. Ph.D. Thesis, Universite Paul Sabatier, Toulouse,

France.



Reby, D. & McComb, K. 2003a. Anatomical constraints generate honesty: acoustic

cues to age and weight in the roars of red deer stags. Animal Behaviour, 65,

519–530.

2003b. Vocal communication and reproduction in deer. Advances in the Study of

Behaviour, 33, 231–264.

Reby, D., Joachim, J., Lauga, J., Lek, S. & Aulagnier, S. 1998. Individuality in the groans

of fallow deer (Dama dama) bucks. Journal of Zoology (London), 245, 79–84.

Reby, D., Cargnelutti, B. & Hewison, A. J. M. 1999a. Contexts and possible functions of

barking in roe deer. Animal Behaviour, 57, 1121–1128.

Reby, D., Cargnelutti, B., Joachim, J. & Aulagnier, S. 1999b. Spectral acoustic structure

of barking in roe deer: sex-, age-, and individual-related variations. Comptes Rendus

de l’Academie des Sciences, 322, 271–279.

Reby, D., Izquierdo, M., Hewison, A. J. M. & Pepin, D. 2001. Red deer (Cervus elaphus)

hinds discriminate between the roars of their current harem holder stag and

those of neighbouring stags. Ethology, 107, 951–959.

Rendall, D., Peter S. R. & Emond R. E. 1996. Vocal recognition of individual and kin in

free ranging rhesus monkeys. Animal Behaviour, 51, 1007–1015.

Rendall, D., Owren, M. J. & Rodman, P. S. 1998. The role of vocal tract filtering in

identity cueing in rhesus monkey (Macaca mulatta) vocalizations. Journal of the


Riede, T. & Fitch, W. T. 1999. Vocal tract length and acoustics of vocalization in the

domestic dog, Canis familiaris. Journal of Experimental Biology, 202, 2859–2869.

Schaller, G. B. 1967. The Deer and the Tiger: A Study of Wildlife in India. Chicago, IL:

University of Chicago Press.

1972. The Serengeti Lion. Chicago, IL: University of Chicago Press.

Semple, S & McComb, K. 1996. Behavioural deception. Trends in Ecology and Evolution,

11, 434–437.

Shipley, C., Hines, M. & Buchwald J. S. 1981. Individual differences in threat calls of

northern elephant seal bulls. Animal Behaviour, 29, 12–18.

Snowdon, C. T. Cleveland, J. & French, J. A. 1983. Responses to context- and

individual-specific cues in cotton top tamarin long-calls. Animal Behaviour, 31,

92–101.

Tooze, Z. J., Harrington, F. H. & Fentress, J. C. 1990. Individually distinct vocalisations

in timber wolves, Canis lupus. Animal Behaviour, 40, 723–730.

Tyack, P. 1981. Interactions between singing Hawaiian humpback whales and

conspecifics nearby. Behavioral Ecology and Sociobiology, 8, 105–116.

1983. Differential response of humpback whales, Megaptera novaeangliae, to playback

of song or social sounds. Behavioral Ecology and Sociobiology, 13, 49–55.

van Parijs, S. M., Hastie, G. D. & Thompson, P. M. 2000. Individual and geographic

variation in display behaviour of male harbour seals in Scotland. Animal Behaviour,

59, 559–568.

Waser, P. M. 1977. Individual recognition, intragroup cohesion and intergroup

spacing: evidence from sound playback to forest monkeys. Behaviour, 60, 28–74.



Whitehead, H., Waters, S. & Lyrholm, T. 1991. Social organisation of female sperm

whales and their offspring: constant companions and casual acquaintances.


Wilden, I., Herzel, H., Peters, G. & Tembrock, G. 1998. Subharmonics, biphonation and

deterministic chaos in mammal vocalisation. Bioacoustics, 9, 171–196.

Zuberbuhler, K. 2002. A syntactic rule in forest monkey communication. Animal

Behaviour, 63, 293–299.


18

Underwater acoustic communicationnetworks in marine mammals

v i nc e n t m . j a n i k

University of St Andrews, St Andrews, UK

Introduction

Communication in networks has received considerable research attention

over the last few years (Naguib & Todt, 1997; Otter et al., 1999; Peake et al., 2001;

Mennill et al., 2002; Ch. 1). This is true for two types of network interaction, one

in which several receivers react to the signal of just one individual and a more

complex one in which receivers eavesdrop on the signal exchange of two or more

individuals and use the information they gained in their own decision making

(McGregor & Dabelsteen, 1996; Ch. 2). If we think about communication, the fact

that there often is more than just one individual receiving any given signal is not

surprising. One reason that many studies on more transient signals concentrated

on only one signaller and one receiver was a methodological problem. It is notori-

ously difficult to sample behaviour from more than one or two individuals at a

time, especially if interactions are rapid and involve movements of individuals.

Recently, the simultaneous tracking of several individuals in a large group has

become feasible. This has led to an increase in studies investigating the effects

of signals on several receivers in rapid communication interactions. Many such

studies have concentrated on the acoustic domain, a modality that is inherently

transient. Signals rarely last more than a few seconds and usually provide a variety

of different messages within a single signal.

While there is ample evidence from terrestrial environments that more than

one individual can receive and use information from a call or a calling interaction,

data on acoustic communication networks in marine environments are sparse.

The marine environment imposes constraints on, and presents opportunities for,

communication in networks that are different from those found in the terrestrial


390


Underwater acoustic communication in marine mammals 391

world (Janik, 1999). Light only penetrates a few tens of metres into the ocean from

the surface. This limits the use of colouration or movement signals in the ocean.

Signals that use active light sources are more successful but also cost more to

develop. Sound is a much cheaper option and travels much further than in air.

However, a variety of parameters, especially pressure and temperature gradients,

influence sound propagation at different depths.

The loudest and best-known biological sound sources in the ocean are marine

mammals. Pinnipeds and cetaceans use underwater sound in very similar ways.

Many features of their social lives make them interesting subjects for the study of

communication networks. For example, sound transmission characteristics of the

sea allow individuals to stay in acoustic contact over very long distances (Tyack,

1998). Several species live in large aggregations or fission–fusion societies in which

group composition changes frequently while individuals maintain preferences

for certain associates (Wells et al., 1999). Many marine mammals are capable of

vocal learning, which helps them to produce a variety of different sounds (Janik &

Slater, 1997). Finally, many of their cognitive abilities rival those of the great apes

(Herman, 1987; Kako, 1999; Schusterman & Kastak, 2002). This combination of

environmental conditions and social skills is unique among mammals. Each of

them affects how animals communicate and will have profound effects on the

characteristics of communication networks. In this chapter, I summarize what we

know about acoustic communication networks in marine mammals.

Size and characteristics of marine mammal communication

networks at sea

Payne & Webb (1971) suspected that cetacean communication networks

are among the largest in the world. To identify the potential size of a communica-

tion network, we need information on the density of receivers and the active space

of a signal. The active space is the area in which another individual can perceive

the calls of a conspecific (Brenowitz, 1982). Active space can either be measured

directly by playing back calls of known source level and observing a predicted

reaction of the receiver, or through theoretical calculations using call source lev-

els, perception thresholds of individuals and models of sound propagation. Direct

measurement and theoretical calculation have advantages and disadvantages. For

example, an advantage of direct measurement using playback is that the sound

reached the animal through its actual environment and no assumptions about

propagation loss are necessary. Such assumptions can be a weak point of theoret-

ical calculations since propagation is influenced by several parameters, many of

which can change from one minute to the next. A disadvantage of direct measure-

ment using playback is that the active space is likely to be underestimated because


392 V. M. Janik

receivers may extract information from a call but not show an immediate reaction.

An advantage of theoretical calculations of active space is that empirical data on

sound perception thresholds in different noise conditions can be used to predict

when an animal can perceive a sound. However, an ability to perceive a sound of

a particular frequency does not necessarily mean that the signal is recognizable

as a call of a conspecific.

Marine mammal sound propagation in the sea

Sound propagation in the sea differs greatly from that in air. In addition

to normal spreading loss, underwater a sound of 1 kHz loses around 0.04 dB/km

through absorption while the same sound in air loses 4 dB/km (Richardson et al.,

1995). The result is that marine mammal calls have a much larger active space

than those of most terrestrial animals. Given that animals can usually detect sig-

nals at, or greater than, the level of background noise (Ch. 20), as can we using

microphones, it is safe to assume that animals can perceive sounds if we are able

to record them. If such recordings are made with passive acoustic localization

techniques (e.g. Watkins & Schevill, 1972; Clark & Ellison, 2000; Janik et al., 2000),

we are able to determine the distance to the sound source. Studies using such

equipment have confirmed that many marine mammals produce signals that can

be detected from more than 10 km away (Table 18.1). These distances coincide

roughly with those at which reactions to calls have been observed. A fin whale

Balaenoptera physalus has been observed to start swimming towards a vocalizing

group 20–25 km away (Watkins, 1981). Fin whales also refrain from using certain

sound types if there are no other whales within a 20 km radius (Watkins, 1981).

Humpback whales Megaptera novaeangliae respond to sounds 9 km away (Tyack &

Whitehead, 1983). Such responses are indications that the communication net-

work includes animals at such distances. However, the actual network may be

much larger. Recent use of ocean-wide microphone arrays offshore has allowed

researchers to record baleen whales over several hundred kilometres (Table 18.1).

At large distances, it is unlikely that an animal shows an immediate reaction to a

single call. Nevertheless, marine mammals may use distant sounds that indicate

the location of other individuals to find breeding or foraging grounds.

Detailed theoretical calculations of active space have been conducted for bot-

tlenose dolphins Tursiops truncatus, sperm whales Physeter macrocephalus and killer

whales Orcinus orca. Figure 18.1 shows the active space of bottlenose dolphin whis-

tles. Given the fact that most bottlenose dolphin whistles do not have much energy

below 3 kHz, whistles should be detected over distances of 20–25 km at maximum

source levels. The average source level of 158 dB re 1 µPa measured by Janik (2000a)

still gives an active space of 9–16 km in calm seas. However, transmission loss in-

creases with frequency, which means that high-frequency whistles or parts of



Table 18.1. Maximum distances from which marine mammal calls can be detected

Species Frequency range of Recorded from Source

vocalizations (kHz)a distance (km)

Bearded seal, Erignathus

berbatus

0.02–6 25 Cleator et al., 1989

Harp seal, Phoca

groenlandica

< 0.1–16+ 30 Watkins & Schevill,

1979

Ringed seal, Phoca

hispida

0.4–16 1 Richardson et al., 1995

Delphinids 0.1–27.3 16 Barlow et al., 2001

Peale’s dolphin,

Lagenorhynchus

australis

0.3–12 0.02 Schevill & Watkins,

1971

Sperm whale, Physeter

macrocephalus

0.1–30 37 Barlow & Taylor, 1997

Bowhead whale,

Balaena mysticetus

0.025–3.5 17 Clark et al., 1986

Humpback whale, 0.02–8.2 15 Helweg et al., 1992

Megaptera 160 Clark, 1995

novaeangliae

Fin whale, Balaenoptera

physalus

0.01–0.75 > 20 Watkins, 1981;

Watkins et al., 1987

Blue whale, 0.012–0.39 600 Stafford et al., 1998

Balaenoptera musculus 1600 Clark, 1995

aTaken from overview in Richardson et al., 1995.

whistles would not have the same active space as low-frequency components. Cur-

rently, we know little about how this would affect the information that is available

to the receiver. With experience, bottlenose dolphins can identify individually dis-

tinctive signature whistles of conspecifics even if they only hear parts of the whistle

(Caldwell et al., 1990). However, subtle variations in whistle parameters can carry

additional information (Janik et al., 1994), which could be lost in such cases. Using

similar methods, Miller (2004) found that killer whale calls have an active space of

up to 26 km in calm seas and Madsen et al. (2002) calculated 60 km for slow clicks

and 16 km for usual clicks of sperm whales. Again, the maximum source levels

were used for these calculations and the same restrictions for high frequencies

apply.

While these active spaces seem very large, there are some marine mammal

signals that are much quieter and do not travel nearly as far (Table 18.1). Most

marine mammals produce sounds at various different source levels many of which


394 V. M. Janik

0

10

20

30

40

1 2 3 4 5 6 7 8 9 10 11 12

Frequency (kHz)

Ran

ge (

km)

0

10

20

30

40

1 2 3 4 5 6 8 10 12

Frequency (kHz)

Ran

ge (

km)

(b)

(a)

Fig. 18.1. The estimated radius of the active space of dolphin whistles without

frequency modulation at different frequencies for sea state 0 (�, calm seas, no wind)

and sea state 4 (�, moderate breeze of 13–18 miles/h). (a) Whistles produced at

maximum source level of 169 dB re 1 µPa. (b) Whistles produced at mean source level

of 158 dB re 1 µPa. Transmission loss in a habitat of homogeneous temperature and

10 m depth (source and receiver at 5 m depth) was calculated following Marsh &

Schulkin (1962) and Urick (1983). Ambient noise was taken from Knudsen et al. (1948).

Data for auditory thresholds and critical ratios of Tursiops truncatus were taken from

Johnson (1967, 1968). (After Janik, 2000a.)



would only be audible to conspecifics within a 100 m or less. Furthermore, several

species like harbour porpoises Phocoena phocoena (Busnel & Dziedzic, 1966) and

Hector’s dolphins Cephalorhynchus hectori (Dawson, 1991), rarely use low-frequency

sounds (i.e. <20 kHz) but instead use clicks to communicate. Their signals are

subject to much larger transmission loss. Communicative clicks are very similar to

echolocation clicks and travel only a few hundred metres, making the active space

of these species relatively small. Communication networks in these species are,

therefore, much smaller and more comparable to those found in some terrestrial

species.

The active space calculations for bottlenose dolphins and killer whales used

empirical models of sound propagation to predict transmission loss in shallow

water. While this is a useful method for this estimate, several other factors influ-

ence transmission in different parts of the water column. Shallow water trans-

mission is greatly influenced by reflections off the surface and the bottom. This

leads to reverberation that can make acoustic signals unrecognizable. Perception

experiments using degraded signals would help us to understand how degrada-

tion affects signal detection and recognition in marine mammals. Furthermore,

other parameters like transmitter depth or frequency can have a strong effect on

active space (Mercado & Frazer, 1999). Consequently, the loudest marine mammal

signals are not necessarily the ones that transmit the furthest.

In deep water, temperature and pressure profiles give the propagation path a

unique shape that is very different from those found in terrestrial environments

(Richardson et al., 1995). The speed of sound increases with depth and tempera-

ture. In summer, when the surface layer is warmer than the water below, sound is

refracted downwards, leading to a shadow zone ahead of the sound source. As the

sound travels deeper, temperature does not change much but pressure increases.

This leads to refraction towards the surface. As a result, sound travels up and

down through the water column as it travels away from the source (Fig. 18.2). If

the surface layer is mixed or shows little temperature layering, as is often the case

in winter, sound travels more easily through the upper layers. However, some en-

ergy still leaks into lower layers and travels in the same ray pattern as in summer.

The result of these conditions is that animals at the surface enter and leave con-

vergence zones of the ray paths of a sound produced at great distance. Therefore,

to locate a calling animal, an individual needs to consider the special propagation

path. If it listened at the surface in one of the convergence zones, it may encounter

an area that appears to have a caller in its centre with sound energy decreasing in

all directions from it. However, the centre does not have the calling animal in it. In-

stead, the sound was produced several kilometres away and has travelled through

deep waters before returning to the surface. The result is that animals cannot

use changes in received levels at the surface to locate a distant caller. However,


0

500

1000

1500

2000

2500

3000

3500

4000

4500

50001500 1510 1520 1530 1540 1550 1560

Dep

th (

m)

Sound speed (m/s)

(a)

(b)

Fig. 18.2. Changes in the speed of sound and ray paths with depth. (a) Typical profile

of speed of sound versus depth for temperate or tropical seas. (b) Calculated ray paths

for a 20 Hz signal produced at a depth of 50 m in an environment with the speed of

sound profile of (a). Ray paths were calculated using a parabolic equation model.

White blocks indicate attenuation of ≤ 60 dB; black blocks indicate attenuation of

≥100 dB. Note the convergence zone (shown by arrows) near the surface at ranges of

6.5, 130 and 190 km.

396



following parts of the actual sound path underwater may allow an individual to

recognize that it is listening in a convergence zone rather than being close to the

original caller.

Another interesting aspect of sound transmission at sea is the deep sound

channel or SOFAR channel. This is a layer at approximately 600–1200 m in which

sound is trapped and travels almost horizontally with much less transmission

loss (because of the shorter travel path and no losses from surface or bottom

reflections). It can be found in the layer with minimum sound speed. Little is

known on whether marine mammals use this channel, but it has been suggested

that whales may use it for long-distance communication (Payne & Webb, 1971).

This is only possible for a few species that travel routinely to this depth, for example

elephant seals Mirounga spp. (Le Boeuf et al., 1989; Hindell et al., 1991) or northern

bottlenose whales Hyperoodon ampullatus (Hooker & Baird, 1999). However, in Arctic

waters, where the minimum sound speed (and thus the SOFAR channel) can be at

much shallower depths, it may be within reach of more species.

The conditions described here are idealizations assuming little variation in

other parameters. They describe general patterns but the actual situation faced

by a marine mammal changes with location and time. One conclusion from these

patterns is that it must be difficult to estimate range from a caller using parame-

ters such as sound intensity. However, they may be able to use other parameters

to determine their distance from a caller. Premus & Spiesberger (1997) analysed

fin whale sounds recorded in the Gulf of California. They found that the signal

arrived several times at each hydrophone, which is typical if the sound takes sev-

eral different paths to reach the receiver. Longer paths result in later arrivals, and

at great distances these time delays can be substantial. However, Premus & Spies-

berger (1997) noted that the first arrival of a fin whale call was much sooner than

expected even if it was taking the shortest route available through the water. This

fast sound transmission could only be explained if the first arrival represented

sound energy that entered and travelled in the sediment, where sound speed is

much higher than in water. If such multipath arrivals through different media

are common, whales may be able to tell the distance of the caller by listening to

the differences in the time of arrival of the sound travelling through the sediment

and that travelling through water. This sound path may even allow them to listen

to individuals on the other side of an island. Another way in which distance in-

formation could be extracted is by listening to the extent of sound degradation.

Again, we know little about the abilities of marine mammals to use such features

to judge distance to the caller.

The number of animals in a communication network

The second variable that determines communication network size is the

number of animals within the transmission range of a signal. This varies greatly


398 V. M. Janik

Table 18.2. Examples of animal density and average group sizes for selected sites

Species Location Density Average Source

(animals/100 km2) group size

Bottlenose

dolphin

Gulf of Mexico

(US coast)

6–480 2–15 Shane et al., 1986

Harbour

porpoise

Northwest

Europe

10–80 1.49 Hammond et al.,

2002

White-beaked

dolphin

Northwest

Europe


2002

Minke whale Northwest

Europe


2002

Vocalizing fin

whales

Hawaii 0.0027;

maximum 0.0081

1 McDonald &

Fox, 1999

and depends on the area, species and behaviour of a marine mammal. Bottlenose

dolphins, for example, can be found in groups of hundreds (Saayman et al., 1973)

or even thousands (Scott & Chivers, 1990) offshore, while individuals in coastal

areas may at times find themselves acoustically isolated from all conspecifics if

they enter small inlets in which sound is blocked by land. Furthermore, many ma-

rine mammal species, especially delphinids, live in fission–fusion societies where

group composition and size can change rapidly. Finally, if we consider that in-

dividuals are capable of restricting signal transmission to specific receivers (see

below), it becomes clear that network size is difficult to assess. On an evolutionary

scale, however, it is interesting to look at how many potential receivers there are

for any given signal. This might help us to understand the relationship between

network size and specific strategies to direct or restrict signals. Because of the lack

of information on average transmission distances of marine mammal sounds, we

can currently only look at data on the number of animals in an area rather than

calculate network sizes. Ultimately, to calculate network sizes, population den-

sities and the average active space of a signal from the same area need to be

combined.

The average group size of a marine mammal species is a good indicator of the

most commonly found minimum network size (Table 18.2). If group size is very

large, as in some oceanic dolphin species, the transmission range of a signal can be

limited by masking noise from conspecifics and the actual network would contain

fewer animals than are in the group (for similar considerations in anurans, see

Ch. 13). In most cases, however, the network will be larger than the average group

size because of the large active space of marine mammal calls. Population-density

data can be used to estimate average network sizes for unrestricted signals that



travel beyond the group’s boundaries. Table 18.2 gives population densities and

average group sizes for a few selected marine mammal species. Some studies have

assessed animal density using acoustic surveys. This means that only vocalizing

animals are registered. While this gives a good density estimate of signallers, it

represents a minimum rather than a representative average estimate of network

size. However, population density studies usually look at very large areas that are

less relevant for estimates of communication network sizes. One tight group of 20

dolphins in 1000 km2, for example, will yield a very low population density but still

represents a communication network of 20 animals. Like sound propagation in the

sea, population density is a highly dynamic variable. Therefore, the size of marine

mammal communication networks is likely to vary greatly on a temporal as well

as spatial scale. We can expect that territorial and, therefore, relatively stationary

species display more stability in network size, but studies on the dynamics of such

network sizes are still lacking.

Directing and restricting signals

In a network, we can expect to find two different kinds of signal: those

that are directed at all receivers within range and others aimed at only one or a few.

Callers directing signals at specific individuals benefit from adding information

that indicates who they are addressing. This is even more important if, like in

marine mammals, the network can be large, locations of individuals are difficult

to predict (e.g. if animals are not territorial) and if only one sensory modality is

available. Concurrently, such conditions render it more difficult for the sender

to identify who is within range as a potential receiver that it is worth calling to.

One way of solving this problem for the caller is to give unequivocal information

about its own identity or group membership. This makes it more likely to be

recognized by other group members or close associates within range. While it

might be disadvantageous to broadcast one’s identity or location if predators use

such cues to find prey, signalling this information can be evolutionarily stable

if it improves information transmission for the sender to the required receiver

(Johnstone, 1997).

Most animal species cannot avoid providing identity information through indi-

vidually specific voice cues. Such cues result from individually specific genetic and

environmental influences on the morphology of the vocal apparatus during devel-

opment. Similarly, genetically related individuals may share a voice feature that

can be used in kin or even group recognition if related individuals stay together.

However, voice cues are relatively subtle and can be difficult to decode over long

distances or in high background noise. There are several ways to improve the en-

coding of information on identity. First, if groups are genetically isolated, genetic


400 V. M. Janik

drift can increase group differences. Even between sympatric groups, this is pos-

sible if their members do not interbreed. Another solution is to use call types

that are shared by all animals in a population at higher or lower rates than the

rest of the population. Finally, animals can develop group- or individual-specific

call types. This last solution can usually only be achieved through vocal learning

or invention (Janik & Slater, 2000). Nevertheless, several of these influences can

act together to create individual differences. For example, limited skills in vocal

learning that only allow a slight change in the fundamental frequency of a given

signal may be used to enhance individual differences caused by environmental

influences during development.

Many marine mammal species show pronounced differences between groups of

animals. Weddell seals Leptonychotes weddelli in breeding colonies only 20 km apart

have been found to use colony-specific call types and show differences in usage

of shared call types (Morrice et al., 1994). Similar geographic variation over much

larger distances has been described for leopard seals Hydrurga leptonyx (Thomas &

Golladay, 1995), bearded seals Erignathus berbatus (Cleator et al., 1989), harp seals

Phoca groenlandica (Terhune, 1994) and harbour seals Phoca vitulina (van Parijs et al.,

2000a). However, in these cases it is possible that individuals from different sites

are geographically isolated. Humpback whales in the Atlantic and the Pacific, for

example, sing very different songs (Winn et al., 1981). Since they cannot encounter

each other, these differences are not necessary for group recognition. Thus, the

occurrence of differences between the calls of groups of animals is not evidence

for a specific adaptation for group recognition.

Killer whales (Ford & Fisher, 1983), sperm whales (Rendell & Whitehead, 2003)

and blue whales Balaenoptera musculus (Stafford et al., 2001) also have distinctive

group calls, but here these groups overlap in their geographic ranges. In these

cases, the distinctiveness in the repertoire may be more important for directing

signals than in geographically isolated groups. However, the calls in these exam-

ples are not individually specific.

Individual specificity may not be necessary for animals that live in stable fam-

ily groups like killer whales. If group composition is less stable though, more

unequivocal signals may be required for individual recognition. Bottlenose dol-

phins, for example, associate preferentially with specific individuals, but their

daily ranging behaviour results in regular changes of group composition and

short-term associations (Wells et al., 1987). This organization is often referred to

as a fission–fusion society. Bottlenose dolphins develop individually distinctive

signature whistle types (Fig. 18.3) that are used while animals are out of visual

contact (Caldwell et al., 1990; Janik & Slater, 1998). These signals have a much

larger interindividual variability than isolation calls of other animal species and

thus transmit individual identity more reliably (Tyack, 2000). Signature whistles



Fig. 18.3. Three randomly chosen spectrograms (columns; FFT size, 1024; time

resolution, 20.5 milliseconds; frequency resolution, 50 Hz; number of FFT steps, 200;

weighting function, Hanning window) of signature whistles from each of four

different individual bottlenose dolphins (rows). Background noise and harmonics

have been removed on all spectrograms to show the pronounced difference in the

shape of the fundamental frequency of signature whistles of the different individuals.

(After Janik & Slater, 1998).


402 V. M. Janik

have been reported to be stable over more than 12 years in female bottlenose dol-

phins in the Bay of Sarasota, Florida, USA (Sayigh et al., 1990). Tyack (1997) found

that vocal learning influences whistle development. In fact, learning or innova-

tion may be the only way to develop such signals that have to be different from

those of a large number of conspecifics within a fission–fusion society. Janik &

Slater (1997) suggested that individual or group recognition might have been one

of the main selection pressures on the evolution of vocal learning in cetaceans.

Evidence for similar individually distinctive signature signals exists for common

dolphins Delphinus delphis (Caldwell & Caldwell, 1968), Pacific white-sided dolphins

Lagenorhynchus obscurus (Caldwell & Caldwell, 1971), spotted dolphins Stenella pla-

giodon (Caldwell et al., 1973), Pacific humpback dolphins Sousa chinensis (van Parijs &

Corkeron, 2001) and sperm whales (Watkins & Schevill, 1977).

While such shared calls may facilitate recognition in general, calls can also

be directed at specific individuals through vocal matching. In vocal matching, an

individual responds to the signal of a caller by producing a signal of the same type.

Vocal matching can be used without the existence of individual-specific calls as

long as other individuals can copy calls or if they have a repertoire of shared calls.

Many species of cetaceans have been observed to produce calls of the same kind

in response to a call of a conspecific, but such anecdotal reports cannot exclude

the possibility of matching occurring by chance. If individuals share a repertoire

and produce sounds independently, by chance alone two different individuals

can produce signals of the same type in close succession. However, this does not

necessarily mean they interact vocally. The proportion of such interactions has to

be larger than expected by chance to represent evidence for vocal matching. True

vocal matching has been demonstrated for bottlenose dolphins (Janik, 2000b) and

for killer whales (Miller et al., 2004). In bottlenose dolphins, signature whistles

can be copied by another individual in such matching interactions (Janik & Slater,

1998). Tyack (1991) raised the interesting possibility that bottlenose dolphins may

use signature whistles of other individuals to initiate contact with the ‘owner’ of

the signature whistles. However, in all reported cases in which signature whistle

matching has been observed and the identities of the calling individuals were

known, the ‘owner’ of the signature whistle called first (Janik & Slater, 1998).

A different strategy is changing the directionality of calls. The fundamental

frequencies of most marine mammal calls are usually transmitted in a relatively

omnidirectional pattern (Evans et al., 1964; Lammers & Au, 2003). However, clicks

are highly directional (reviewed by Au, 1993) and have the potential to be used

in addressing specific individuals. Dolphins use clicks in echolocation as well as

communication. Several species (see above) rely on clicks for communication and

do not produce any whistles at all (Dawson, 1991; reviewed by Herman & Tavolga,

1980). It is possible that these species use the directionality of their clicks to direct



or restrict signals in social interactions. Even in whistles, the high-frequency parts

are highly directional (Lammers & Au, 2003) so that the same signal may carry dif-

ferent types of information with some of it only available to animals ahead of the

caller. This could be achieved by filtering high-frequency components in specific

patterns, similar to formants in human signals. Such filtering would not be dis-

cernible from listening to the low-frequency component alone. Dolphins have con-

trol over the filtering of higher-frequency harmonics (Fig. 1 in Janik et al., 1994), but

the significance of such changes is unclear. Miller (2002) found that killer whales

have call types with high-frequency components that show higher directionality

than the low-frequency parts of the same call or calls without these components.

High- and low-frequency components of the same killer whale call are not har-

monically related. Therefore, the modulation pattern of one component cannot

be discerned from the other one. This makes withholding information even easier.

Miller (2002) suggested that killer whales might use calls with high-frequency

components to indicate their direction of movement. Alternatively, they may

be used to direct signals at specific individuals and withhold information from

others.

Another way of restricting the spread of signals through a communication net-

work is by decreasing the source level so that they do not carry as far. Pinnipeds

and cetaceans produce the same call types at a variety of different source levels

(reviewed by Richardson et al., 1995). Many species of odontocetes also have sig-

nals of very different frequency in their repertoires. High-frequency signals are

attenuated much more rapidly than low-frequency sounds. Odontocetes may be

able to restrict transmission range by choosing high-frequency clicks rather than

lower-frequency whistles even though they are produced with the same source

level. However, calling depths have different optimal frequencies for long-range

signal transmission (Mercado & Frazer, 1999). Higher frequencies can sometimes

travel further than lower ones, especially in relatively shallow water (i.e. less than

100 m deep). Further studies are needed to explore the possible use of source-level

adjustments and frequency selection in directing and restricting signals.

Eavesdropping

Peake (Ch. 2) has distinguished two types of eavesdropping; interceptive

eavesdropping (e.g. predators locating prey by listening to prey vocalizations) and

social eavesdropping (extracting information from a signalling interaction). Brad-

bury & Vehrencamp (1998) also used the term ‘cue’ for prey signals that are used

by predators to locate prey. While it is arguable whether such interactions can be

called communication, effects of calls on predators and their prey are an interest-

ing ecological variable that can influence the design of communication systems.


404 V. M. Janik

Several studies have looked at the impact of killer whale signals on other ma-

rine mammal species that are potential prey. The diet of killer whales can vary

considerably from one location to another. In British Columbia, Canada, some

killer whales only eat fish while others take marine mammals as prey. Fish eaters

are known as resident killer whales since they have smaller ranging patterns than

the so-called transient killer whales that feed on marine mammals. Deecke et al.

(2002) conducted playback experiments and inferred from diving patterns that

harbour seals in British Columbia avoided sounds made by transient killer whales

but they did not react to sounds of resident killer whales. Transients and residents

use different call types and individual killer whale pods have repertoires of up

to 17 call types (Ford, 1989). Deecke et al. (2002) carefully selected specific sound

types for each comparison to ensure that the discrimination performed by the

seals could not be based on just one or two call types. Harbour seals also avoided

playbacks of sounds from Norwegian killer whales. These whales concentrate on

herring as prey for at least part of the year, which makes it unlikely that fish-

eating killer whales share voice features that identify them as harmless to seals.

It is unclear how harbour seals distinguish between known residents and other

killer whales. There are genetic differences between killer whale populations and

even between sympatric residents and transients of British Columbia (Hoelzel

et al., 1998). Perhaps residents share a voice feature that affects all their calls and

makes them recognizable. Alternatively, the seals may have learned all call types

used by resident killer whales and avoid all other call types.

Grey whales Eschrichtius robustus (Cummings & Thompson, 1971) and beluga

whales Delphinapterus leucas (Fish & Vania, 1971) have been found to avoid loca-

tions from which killer whale sounds had been played. Unfortunately, it is not

clear whether the sounds used in these studies came from mammal-eating or fish-

eating killer whales. Belugas (Schevill 1964; Fish & Vania, 1971) and grey whales

(Cummings & Thompson, 1971) also ceased vocalizing when exposed to killer

whale sounds, another well-known reaction of cetaceans to any unusual stimulus

(Herman & Tavolga, 1980). Other examples are pilot whales Globicephala melaena

falling silent when hunted (Schevill, 1964) and bottlenose dolphins (Caldwell &

Caldwell 1967) and Peale’s dolphins Lagenorhynchus australis (Schevill & Watkins,

1971) falling silent when captured or when approached by a boat. Interestingly,

transient killer whales appear to use fewer echolocation clicks than resident

whales while they forage (Barrett-Lennard et al., 1996). This may be a counter-

strategy to avoid early detection by their prey.

Several aspects of marine mammals make it difficult to establish whether social

eavesdropping occurs in this group (i.e. whether information has been extracted

from a signalling interaction). Individuals often approach callers, for example.

Groups of surface-active humpback whales produce a variety of sounds that can



attract males that are several kilometres away (Tyack, 1983) and similar results

have been found for southern right whales Eubalaena glacialis (Clark & Clark, 1980).

However, it is difficult to determine who is interacting (i.e. whether individuals in

such groups signal to each other or to individuals outside the group) and, there-

fore, whether there is potential for eavesdropping. Distant individuals may extract

information from the calls of animals interacting in the group and decide to ap-

proach (which would qualify as social eavesdropping), or they may be attracted by

calls that are directed at distant animals (which is a good example of communi-

cation in a network but not for eavesdropping). It will be difficult to distinguish

between these possibilities experimentally; furthermore, these two scenarios are

not mutually exclusive.

One context in which social eavesdropping has been demonstrated is in song

interactions between birds (Naguib & Todt, 1997; Otter et al., 1999; Peake et al.,

2001; Mennill et al., 2002). Many marine mammal species also produce song dur-

ing the mating season and some, like Weddell seals (Bartsh et al., 1992) and har-

bour seals (van Parijs et al., 2000b), establish underwater territories. By analogy

with songbirds, social eavesdropping by marine mammals may be found in such

circumstances. However, other singing species of marine mammals are less sta-

tionary. For example, while singing humpback whales are spaced further apart

than non-singers and singers often avoid each other (Frankel et al., 1995), individ-

uals can rarely be found in the same location from one day to the next (Clapham,

2000). Clapham termed this arrangement a floating lek, in which females are able

to listen to several males but males are not stationary. Given the apparent lack of

direct vocal interactions outside of the surface-active groups that form when sev-

eral males start to escort a female, eavesdropping is less likely to be of importance

here. However, further studies relating vocal displays to movement of individuals

are needed before we can assess the relevance of eavesdropping in this context.

Another very different context in which the term eavesdropping has been used

is echolocation (Xitco & Roitblat, 1996). These authors found that a bottlenose dol-

phin could extract information about the location and shape of an object without

having to produce echolocation sounds itself; it did so by listening to the echoes

of echolocation clicks produced by another individual. This might be a common

feature of echolocating animals. Bats have been found to be attracted by feeding

buzzes of conspecifics (Barclay, 1982; Balcombe & Fenton, 1988). However, the stud-

ies on bats could not determine whether feeding buzzes are generally attractive,

like food calls of non-echolocating animals, or whether they can provide informa-

tion about the exact location and shape of the target to the eavesdropper. In the

study of Xitco & Roitblat (1996), the eavesdropping animal was very close to the

echolocating one and such close proximity may be a prerequisite for gathering

such target-specific information. This form of eavesdropping can be defined as


406 V. M. Janik

extracting information from an interaction between another individual’s echolo-

cation signal and an echolocation target. While this might explain some of the

swimming formations that dolphins use during foraging, it is of less relevance to

the topic of communication networks.

Conservation implications

A major concern in marine mammal conservation is the impact of noise

made by human activity in the sea. There has been an increasing amount of indus-

trial, shipping and seismic survey noise over the last century. For example, engine

noise of ships in the busy shipping lanes of the North Atlantic increases the average

ambient noise levels below 500 Hz by 10–40 dB (Urick, 1983). Ross (1976) estimated

that shipping led to a 10 dB increase in ambient noise in these areas from 1950

to 1975. The issue of such noise has been discussed recently in the context of

the Acoustic Thermometry of Ocean Climate (ATOC) study and the low-frequency

active sonar systems deployed by the military (Richardson et al., 1995). These tech-

niques can potentially harm marine mammals because of high source levels and

signals that are similar to those of some marine mammal species.

The main concern in noise exposure has been potential physical damage to the

animals. For example, several Cuvier’s beaked whale Ziphius cavirostris strandings

occurred at the same time as military exercises (e.g. Frantzis, 1998; Balcomb &

Claridge, 2001) and Jepson et al. (2003) reported acute and chronic tissue damage

caused by gas bubbles in whales stranded during such exercises. Weddell seals

exposed to underwater blasts showed severe damage to their inner ears (Bohne

et al., 1986). Another form of impact is a change in the animal’s behaviour. This

can have the same consequences as physical damage since isolation from group

members or the exclusion from feeding grounds can easily lead to the death of an

animal. There are many studies showing short-term avoidance by marine mam-

mals of sound sources (review in Richardson et al., 1995). Examples are killer whales

(Morton & Symonds, 2002) and harbour porpoises (Johnston, 2002) avoiding areas

ensonified by acoustic harassment devices deployed to reduce seal predation on

fish farms; beluga whales avoiding ice-breakers by as much as 80 km for up to

48 hours (Finley et al., 1990; Foote et al., 2004); and bottlenose dolphins in Florida

avoiding specific feeding grounds on weekends when boat activity is highest (Allen

& Read, 2000).

Another response to noise of human origin is a change in calling behaviour.

Such responses can involve a change in temporal or structural parameters of a

call (e.g. Au et al., 1985; Foote at al., 2004) or lead to animals changing call rates

or ceasing to vocalize (Terhune et al., 1979; Bowles et al., 1994). Such changes

could either indicate a direct disruption of communication or be a by-product of a



general change in behaviour if animals stop activities that involve specific calling

rates or types of signal to avoid a sound source.

A source of human-derived noise may also affect the behaviour of an animal

by masking marine mammal sounds, thereby disrupting communication. This

could be a serious problem for animals in a number of circumstances. First, many

marine mammals use acoustic signals to maintain contact between mothers and

calves (e.g. Renouf, 1984; Smolker et al., 1993) and noise can shorten the range

over which they are able to hear each other. Second, if information is gathered

by eavesdropping on interactions of more distant individuals, noise could mask

such interactions. Erbe (2002) found that the noise of a fast-moving boat can mask

quiet killer whale sounds if the vessel is 14 km from the listening animal. Simi-

lar calculations predict that icebreaking noise can mask quiet beluga sounds if

the icebreaker is up to 71 km from the animal that is listening (Erbe & Farmer,

2000). While we do not know to what extent information gathered through eaves-

dropping is used by marine mammals, masking certainly has an effect on signals

designed to reach more distant receivers, as in marine mammal song. Therefore,

apart from inflicting physical damage, noise could have a severe effect by disrupt-

ing acoustic contact between individuals.


While we have data on maximum transmission distances for some marine

mammal sounds, it is still unclear to what extent acoustic signals from distant

animals provide valuable information to a conspecific. If the active space of the

signal is particularly large, as seems to be the case for many marine mammals,

the information from distant animals may not be of much use. For example, it

is of only limited value for a predator to know that an animal is foraging 20 km

away if a long time is needed to travel that distance. For marine mammals, most

aggregations of prey species are very dynamic and either move quickly or only

last for brief periods of time. Consequently, a large active space may just increase

noise for distant receivers and could have contributed to the evolution of redun-

dancy and distinctiveness in communication signals: two features that can help to

improve information transmission and that are pronounced in marine mammal

communication systems.

One way of addressing the question of the value of distant signals would be

to compare reactions to distant marine mammal calls with reactions to artificial

broadband noise at similar levels. If conspecifics only add noise to the commu-

nication channel, responses should be the same. Only at a closer, more relevant

distance should reactions differ. Yet communication over large distances may

help in mate attraction or coordination of behaviour patterns. In that case, the


408 V. M. Janik

animals should show specific reactions to distant signals of conspecifics, for exam-

ple specific changes in movement direction. Any studies investigating how marine

mammals react to distant signals would be extremely valuable.

Two related issues are the effect of degradation and how marine mammals

judge the distance to a caller. The distinctiveness of signature whistles in bot-

tlenose dolphins, for example, suggests that at least the identity information

encoded is relatively resistant to degradation. However, what happens to more

subtle cues? The auditory system of marine mammals is adapted to detect and

identify marine mammal signals. Therefore, it would be difficult to make predic-

tions from experiments with artificial test signals. How receivers estimate their

distance from a sound source has been studied extensively in birds and humans

(reviewed by Naguib & Wiley, 2001). The most important parameter appears to

be the degree of reverberation. However, other parameters such as overall and

frequency-dependent attenuation or amplitude fluctuations can also help in the

assessment of distances if the receiver has some experience with the source signal

and the environment. Marine mammals may also use additional cues like time

delays of multiple arrivals via different sound paths (Premus & Spiesberger, 1997)

or changes in signal composition of the same received signal at different receiver

depths (Mercado & Frazer, 1999). Whether and how such information is used by

marine mammals is still unknown.

Eavesdropping on interactions of conspecifics in marine mammals is still vir-

tually unstudied. Territorial seal species would probably be the best starting point

for such studies as interactions between neighbours and intruders are the most

likely source of relevant information that could be obtained through eavesdrop-

ping. However, to simulate such interactions experimentally we need to know

the acoustic parameters that identify a successful or unsuccessful animal in such

contests. Furthermore, we need to investigate whether individuals can recognize

other individuals by general voice features. Without voice recognition, it is diffi-

cult to explain how an animal would recognize an individual that it previously

eavesdropped on. Studying how marine mammals address specific individuals can

also help to understand how relevant eavesdropping is. If marine mammals not

only address specific individuals by matching or the use of signature signals but

also actively exclude potential receivers through the selective use of highly direc-

tional signals, eavesdropping might have been a factor in the evolution of such

strategies.

Theoretical estimates of maximum signal transmission distance and commu-

nication network sizes are useful but they need verification in the real world.

Most likely such extremes are rarely relevant for communicating in everyday life.

Nevertheless, marine mammal communication networks are clearly among the

largest that can be found. As we have seen, this opens up interesting opportunities



but also imposes further constraints by increasing background noise. Future stud-

ies that investigate the dynamics of marine mammal signalling will improve our

understanding of how underwater sound transmission helped to shape their com-

munication systems and to what extent marine mammals use the extra informa-

tion provided by such large active spaces in their communication networks.

Acknowledgements

I would like to thank Peter McGregor for valuable comments on earlier drafts of this

chapter. The chapter was written with support from a Royal Society University Research Fellow-

ship. Figure 18.1 has been reprinted from Janik (2000a) with permission from Springer Verlag.

Figure 18.3 has been reprinted from Janik & Slater (1998) with permission from Elsevier Science.

References

Allen, M. C. & Read, A. J. 2000. Habitat selection of foraging bottlenose dolphins in

relation to boat density near Clearwater, Florida. Marine Mammal Science, 16,

815–824.

Au, W. W. L. 1993. The Sonar of Dolphins. New York: Springer.

Au, W. W. L., Carder, D. A., Penner, R. H. & Scronce, B. L. 1985. Demonstration of

adaptation in beluga whale echolocation. Journal of the Acoustical Society of America,

77, 726–730.

Balcomb, K. C. & Claridge, D. E. 2001. A mass stranding of cetaceans caused by naval

sonar in the Bahamas. Bahamas Journal of Science, 8, 2–12.

Balcombe, J. P. & Fenton, M. B. 1988. Eavesdropping by bats: the influence of

echolocation call design and foraging strategy. Ethology, 79, 158–166.

Barclay, R. M. R. 1982. Interindividual use of echolocation calls: eavesdropping by bats.


Barlow, J. & Taylor, B. 1997. Acoustic census of sperm whales in the eastern temperate

North Pacific. Journal of the Acoustical Society of America, 102, 3213.

Barlow, J., Rankin, S., Oswald, J. N. & Mellinger, D. K. 2001. Long-range detection of

delphinid whistles using a towed hydrophone array. In Proceedings of the 14th

Biennial Conference on the Biology of Marine Mammals, Vancouver, p. 17.

Barrett-Lennard, L. G., Ford, J. K. B. & Heise, K. A. 1996. The mixed blessing of

echolocation: differences in sonar use by fish-eating and mammal-eating killer

whales. Animal Behaviour, 51, 553–565.

Bartsh, S. S., Johnston, S. D. & Siniff, D. B. 1992. Territorial behaviour and breeding

frequency of male Weddell seals (Leptonychotes weddelli) in relation to age, size, and

concentration of serum testosterone and cortisol. Canadian Journal of Zoology, 70,

680–692.

Bohne, B. A., Bozzay, D. G. & Thomas, J. A. 1986. Evaluation of inner ear pathology in

Weddell seals. Antarctic Journal, 21, 208.


410 V. M. Janik

Bowles, A. E., Smultea, M., Wursig, B., DeMaster, D. P. & Palka, D. 1994. Relative

abundance and behavior of marine mammals exposed to transmissions from the

Heard Island feasibility test. Journal of the Acoustical Society of America, 96,

2469–2482.



Brenowitz, E. A. 1982. The active space of red-winged blackbird song. Behavioral Ecology


Busnel, R.-G. & Dziedzic, A. 1966. Acoustic signals of the pilot whale Globicephala

melaena and of the porpoises Delphinus delphis and Phocoena phocoena. In: Whales,

Dolphins, and Porpoises, ed. K. S. Norris. Berkeley, CA: University of California Press,

pp. 607–646.

Caldwell, M. C. & Caldwell, D. K. 1967. Intraspecific transfer of information via the

pulsed sound in captive odontocete cetaceans. In: Animal Sonar Systems: Biology and

Bionics, ed. R. G. Bullock. Jouy-en-Josas: Laboratoire Physiologie Acoustique,

pp. 879–936.

1968. Vocalization of naive captive dolphins in small groups. Science, 159, 1121–1123.

1971. Statistical evidence for individual signature whistles in Pacific whitesided

dolphins, Lagenorhynchus obliquidens. Cetology, 3, 1–9.

Caldwell, M. C., Caldwell, D. K. & Miller, J. F. 1973. Statistical evidence for individual

signature whistles in the spotted dolphin, Stenella plagiodon. Cetology, 16, 1–21.

Caldwell, M. C., Caldwell, D. K. & Tyack, P. L. 1990. Review of the signature-

whistle-hypothesis for the Atlantic bottlenose dolphin. In: The Bottlenose Dolphin,

ed. S. Leatherwood & R. R. Reeves. San Diego, CA: Academic Press, pp. 199–234.

Clapham, P. J. 2000. The humpback whale: seasonal feeding and breeding in a baleen

whale. In: Cetacean Societies: Field Studies of Dolphins and Whales, ed. J. Mann, R. C.

Connor, P. L. Tyack & H. Whitehead. Chicago, IL: University of Chicago Press,

pp. 173–196.

Clark, C. W. 1995. Application of US Navy underwater hydrophone arrays for

scientific research on whales. Reports of the International Whaling Commission, 45,

210–212.

Clark, C. W. & Clark, J. M. 1980. Sound playback experiments with southern right

whales (Eubalaena australis). Science, 207, 663–665.

Clark, C. W. & Ellison, W. T. 2000. Calibration and comparison of the acoustic location

methods used during the spring migration of the bowhead whale, Balaena

mysticetus, off Pt Barrow, Alaska, 1984–1993. Journal of the Acoustical Society of

America, 107, 3509–3517.

Clark, C. W., Ellison, W. T. & Beeman, K. 1986. A preliminary account of the acoustic

study conducted during the 1985 spring bowhead whale, Balaena mysticetus,

migration off Point Barrow, Alaska. Reports of the International Whaling Commission,

36, 311–316.

Cleator, H. J., Stirling, I. & Smith, T. G. 1989. Underwater vocalizations of the bearded

seal (Erignathus barbatus). Canadian Journal of Zoology, 67, 1900–1910.

Cummings, W. C. & Thompson, P. O. 1971. Gray whales, Eschrichtius robustus, avoid the

underwater sounds of killer whales, Orcinus orca. Fishery Bulletin, 69, 525–530.



Dawson, S. M. 1991. Clicks and communication: the behavioural and social contexts of

Hector’s dolphin vocalizations. Ethology, 88, 265–276.

Deecke, V. B., Slater, P. J. B., Ford, J. K. B. 2002. Selective habituation shapes predator

recognition in harbor seals. Nature, 420, 171–173.

Erbe, C. 2002. Underwater noise of whale-watching boats and potential effects on

killer whales (Orcinus orca), based on an acoustic impact model. Marine Mammal

Science, 18, 394–418.

Erbe, C. & Farmer, D. M. 2000. Zones of impact around icebreakers affecting beluga

whales in the Beaufort Sea. Journal of the Acoustical Society of America, 108,

1332–1340.

Evans, W. E., Sutherland, W. W. & Beil, R. G. 1964. The directional characteristics of

delphinid sounds. In: Marine Bio-acoustics, ed. W. N. Tavolga. Oxford: Pergamon

Press, pp. 353–370.

Finley, K. J., Miller, G. W., Davis, R. A. & Greene, C. R. 1990. Reactions of belugas,

Delphinapterus leucas, and narwhals, Monodon monoceros, to ice-breaking ships in

the Canadian high Arctic. Canadian Bulletin for Fisheries and Aquatic Sciences, 224,

97–117.

Fish, J. F. & Vania, J. S. 1971. Killer whale, Orcinus orca, sounds repel white whales,

Delphinapterus leucas. Fishery Bulletin, 69, 531–535.

Foote, A. D., Osborne, R. W. & Hoelzel, A. R. 2004. Whale-call response to masking boat

noise. Nature, 428, 910.

Ford, J. K. B. 1989. Acoustic behaviour of resident killer whales (Orcinus orca) off

Vancouver Island, British Columbia. Canadian Journal of Zoology, 67, 727–745.

Ford, J. K. B. & Fisher, H. D. 1983. Group-specific dialects of killer whales (Orcinus orca)

in British Columbia. In: Communication and Behavior of Whales, ed. R. Payne.

Boulder, CO: Westview Press, pp. 129–161.

Frankel, A. S., Clark, C. W., Herman, L. M. & Gabriele, C. M. 1995. Spatial distribution,

habitat utilization, and social interactions of humpback whales, Megaptera

novaeangliae, off Hawai’i, determined using acoustic and visual techniques.


Frantzis, A. 1998. Does acoustic testing strand whales? Nature, 392, 29.

Hammond, P. S., Berggren, P., Benke, H. et al. 2002. Abundance of harbour porpoise

and other cetaceans in the North Sea and adjacent waters. Journal of Applied

Ecology, 39, 361–376.

Helweg, D. A., Frankel, A. S., Mobley, J. R. & Herman, L. M. 1992. Humpback whale

song: our current understanding. In: Marine Mammal Sensory Systems, ed. J. Thomas,

R. Kastelein & S. Supin. New York: Plenum Press, pp. 459–483.

Herman, L. M. 1987. Receptive competencies of language-trained animals. Advances in

the Study of Behavior, 17, 1–60.

Herman, L. M. & Tavolga, W. N. 1980. The communication systems of cetaceans. In:

Cetacean Behavior: Mechanisms and Functions, ed. L. M. Herman. New York: John

Wiley, pp. 149–209.

Hindell, M. A., Slip, D. J. & Burton, H. R. 1991. The diving behaviour of adult male and

female southern elephant seals, Mirounga leonina (Pinnipedia: Phocidae).

Australian Journal of Zoology, 39, 595–619.


412 V. M. Janik

Hoelzel, A. R., Dahlheim, M. & Stern, S. J. 1998. Low genetic variation among killer

whales (Orcinus orca) in the Eastern Pacific and genetic differentiation between

foraging specialists. Journal of Heredity, 89, 121–128.

Hooker, S. K. & Baird, R. W. 1999. Deep-diving behaviour of the northern bottlenose

whale, Hyperoodon ampullatus (Cetacea: Ziphiidae). Proceedings of the Royal Society of


Janik, V. M. 1999. Origins and implications of vocal learning in bottlenose dolphins.

In: Mammalian Social Learning: Comparative and Ecological Perspectives, ed. H. O. Box &

K. R. Gibson. Cambridge, UK: Cambridge University Press, pp. 308–326.

2000a. Source levels and the estimated active space of bottlenose dolphin (Tursiops

truncatus) whistles in the Moray Firth, Scotland. Journal of Comparative Physiology A,

186, 673–680.

2000b. Whistle matching in wild bottlenose dolphins (Tursiops truncatus). Science, 289,

1355–1357.

Janik, V. M. & Slater, P. J. B. 1997. Vocal learning in mammals. Advances in the Study of


1998. Context-specific use suggests that bottlenose dolphin signature whistles are

cohesion calls. Animal Behaviour, 56, 829–838.

2000. The different roles of social learning in vocal communication. Animal


Janik, V. M., Dehnhardt, G. & Todt, D. 1994. Signature whistle variations in a

bottlenosed dolphin, Tursiops truncatus. Behavioral Ecology and Sociobiology, 35,

243–248.

Janik, V. M., van Parijs, S. M. & Thompson, P. M. 2000. A two-dimensional acoustic

localization system for marine mammals. Marine Mammal Science, 16,

437–447.

Jepson, P. D. Arbelo, M., Deaville, R. et al. 2003. Gas-bubble lesions in stranded

cetaceans. Nature, 425, 575.

Johnson, C. S. 1967. Sound detection thresholds in marine mammals. In: Marine

Bio-acoustics, Vol. 2, ed. W. N. Tavolga. Oxford: Pergamon Press, pp. 247–255.

1968. Masked tonal thresholds in the bottlenosed porpoise. Journal of the Acoustical


Johnston, D. W. 2002. The effect of acoustic harassment devices on harbour porpoises

(Phocoena phocoena) in the Bay of Fundy, Canada. Biological Conservation, 108,

113–118.

Johnstone, R. A. 1997. Recognition and the evolution of distinctive signatures: when

does it pay to reveal identity? Proceedings of the Royal Society of London, Series B, 264,

1547–1553.

Kako, E. 1999. Elements of syntax in the systems of three language-trained animals.

Animal Learning and Behavior, 27, 1–14.

Knudsen, V. O., Alford, R. S. & Emling, J. W. 1948. Underwater ambient noise. Journal of

Marine Research, 7, 410–429.

Lammers, M. O. & Au, W. W. L. 2003. Directionality in the whistles of Hawaiian

spinner dolphins (Stenella longirostris): a signal feature to cue direction of

movement? Marine Mammal Science, 19, 249–264.



Le Boeuf, B. J., Naito, Y., Huntley, A. C. & Asaga, T. 1989. Prolonged, continuous, deep

diving by northern elephant seals. Canadian Journal of Zoology, 67, 2514–2519.

Madsen, P. T., Wahlberg, M. & Møhl, B. 2002. Male sperm whale (Physeter macrocephalus)

acoustics in a high-latitude habitat: implications for echolocation and

communication. Behavioral Ecology and Sociobiology, 53, 31–41.

Marsh, H. W. & Schulkin, M. 1962. Shallow-water transmission. Journal of the Acoustical


McDonald, M. A. & Fox, C. G. 1999. Passive acoustic methods applied to fin whale

population density estimation. Journal of the Acoustical Society of America, 105,

2643–2651.



Ithaca, NY: Comstock, pp. 409–425.



Mercado III, E. & Frazer, L. N. 1999. Environmental constraints on sound transmission

by humpback whales. Journal of the Acoustical Society of America, 106, 3004–3016.

Miller, P. J. O. 2002. Mixed-directionality of killer whale stereotyped calls: a direction

of movement cue? Behavioral Ecology and Sociobiology, 52, 262–270.

Miller, P. J. O. 2004. Estimated active space of killer whale sounds. Journal of

Comparative Physiology A, in press.

Miller, P. J. O., Shapiro, A. D., Tyack, P. L. & Solow, A. R. 2004. Call-type matching in

vocal exchanges of free-ranging resident killer whales (Orcinus orca). Animal

Behaviour, 67, 1099–1107.

Morrice, M. G., Burton, H. R. & Green, K. 1994. Microgeographic variation and songs in

the underwater vocalization repertoire of the Weddell seal (Leptonychotes weddelli)

from the Vestfold Hills, Antarctica. Polar Biology, 14, 441–446.

Morton, A. B. & Symonds, H. K. 2002. Displacement of Orcinus orca (L.) by high amplitude

sound in British Columbia, Canada. ICES Journal of Marine Science, 59, 71–80.

Naguib, M. & Todt, D. 1997. Effects of dyadic interactions on other conspecific




62, 825–837.

Otter, K. A., McGregor, P. K., Terry, A. M. R. et al. 1999. Do female great tits (Parus major)



Payne, R. & Webb, D. 1971. Orientation by means of long range acoustic signaling in

baleen whales. Annals of the New York Academy of Sciences, 188, 110–140.




Premus, V. & Spiesberger, J. L. 1997. Can acoustic multipath explain finback (B.

physalus) 20-Hz doublets in shallow water? Journal of the Acoustical Society of America,

101, 1127–1138.


414 V. M. Janik

Rendell, L. E. & Whitehead, H. 2003. Acoustic clans in sperm whales (Physeter

macrocephalus). Proceedings of the Royal Society of London, Series B, 270, 225–231.

Renouf, D. 1984. The vocalization of the harbour seal pup (Phoca vitulina) and its role in

the maintenance of contact with the mother. Journal of Zoology, 202, 583–590.

Richardson, W. J., Greene, C. R., Jr, Malme, C. I. & Thomson, D. H. 1995. Marine

Mammals and Noise. San Diego, CA: Academic Press.

Ross, D. 1976. The mechanics of underwater noise. New York: Pergamon Press.

Saayman, G. S., Taylor, C. K. & Bower, D. 1973. Diurnal activity cycles in captive and

free-ranging Indian Ocean bottlenose dolphins (Tursiops truncatus Ehrenburg).

Behaviour, 44, 212–233.

Sayigh, L. S., Tyack, P. L., Wells, R. S. & Scott, M. D. 1990. Signature whistles of

free-ranging bottlenose dolphins, Tursiops truncatus: mother–offspring

comparisons. Behavioral Ecology and Sociobiology, 26, 247–260.

Schevill, W. E. 1964. Underwater sounds of cetaceans. In: Marine Bio-Acoustics, Vol. 1, ed.

W. N. Tavolga. Oxford: Pergamon Press, pp. 307–316.

Schevill, W. E. & Watkins, W. A. 1971. Pulsed sounds of the porpoise Lagenorhynchus

australis. Breviora, 366, 1–10.

Schusterman, R. J. & Kastak, D. 2002. Problem solving and memory. In: Marine Mammal

Biology: an Evolutionary Approach, ed. A. R. Hoelzel. Oxford: Blackwell, pp. 371–387.

Scott, M. D. & Chivers, S. J. 1990. Distribution and herd structure of bottlenose

dolphins in the eastern tropical Pacific Ocean. In: The Bottlenose Dolphin, ed. S.

Leatherwood & R. R. Reeves. San Diego, CA: Academic Press, pp. 387–402.

Shane, S. H., Wells, R. S. & Wursig, B. 1986. Ecology, behavior and social organization

of the bottlenose dolphin: a review. Marine Mammal Science, 2, 34–63.

Smolker, R. A., Mann, J. & Smuts, B. B. 1993. Use of signature whistles during

separations and reunions by wild bottlenose dolphin mothers and infants.


Stafford, K. M., Fox, C. G. & Clark, D. S. 1998. Long-range acoustic detection and

localization of blue whale calls in the northeast Pacific Ocean. Journal of the


Stafford, K. M., Nieukirk, S. L. & Fox, C. G. 2001. Geographic and seasonal variation of

blue whale calls in the North Pacific. Journal for Cetacean Research and Management,

3, 65–76.

Terhune, J. M. 1994. Geographical variation of harp seal underwater vocalizations.


Terhune, J. M., Stewart, R. E. A. & Ronald, K. 1979. Influence of vessel noises on

underwater vocal activity of harp seals. Canadian Journal of Zoology, 57, 1337–1338.

Thomas, J. A. & Golladay, C. L. 1995. Geographic variation in leopard seal (Hydrurga

leptonyx) underwater vocalizations. In: Sensory Systems of Aquatic Mammals, ed. R. A.

Kastelein, J. A. Thomas & P. E. Nachtigall. Woerden: De Spil, pp. 201–221.

Tyack, P. 1983. Differential response of humpback whales, Megaptera novaeangliae, to

playback of song or social sounds. Behavioral Ecology and Sociobiology, 13, 49–55.

1991. If you need me, whistle. Natural History, 8, 60–61.



1998. Acoustic communication under the sea. In: Animal Acoustic Communication:

Sound Analysis and Research Methods, ed. S. L. Hopp, M. J. Owren & C. S. Evans. Berlin:

Springer, pp. 163–220.

1997. Development and social functions of signature whistles in bottlenose

dolphins Tursiops truncatus. Bioacoustics, 8, 21–46.

2000. Dolphins whistle a signature tune. Science, 289, 1310–1311.

Tyack, P. & Whitehead, H. 1983. Male competition in large groups of wintering

humpback whales. Behaviour, 83, 132–154.

Urick, R. J. 1983. Principles of Underwater Sound, 3rd edn. New York: McGraw-Hill.

van Parijs, S. M. & Corkeron, P. J. 2001. Evidence for signature whistle production by a

Pacific humpback dolphin, Sousa chinensis. Marine Mammal Science, 17, 944–949.

van Parijs, S. M., Hastie, G. D. & Thompson, P. M. 2000a. Individual and geographic

variation in display behaviour of male harbour seals in Scotland. Animal Behaviour,

59, 559–568.

van Parijs, S. M., Janik, V. M. & Thompson, P. M. 2000b. Display area size, tenure and

site fidelity in the aquatic mating male harbour seal. Canadian Journal of Zoology,

78, 2209–2217.

Watkins, W. A. 1981. Activities and underwater sounds of fin whales. Scientific Reports

of the Whales Research Institute, 33, 83–117.

Watkins, W. A. & Schevill, W. E. 1972. Sound source location by arrival times on a

non-rigid three-dimensional hydrophone array. Deep-Sea Research, 19, 691–706.

1977. Sperm whale codas. Journal of the Acoustical Society of America, 62, 1485–1490.

1979. Distinctive characteristics of underwater calls of the harp seal, Phoca

groenlandica, during the breeding season. Journal of the Acoustical Society of America,

66, 983–988.

Watkins, W. A., Tyack, P., Moore, K. E. & Bird, J. E. 1987. The 20-Hz signals of finback

whales (Balaenoptera physalus). Journal of the Acoustical Society of America, 82,

1901–1912.

Wells, R. S., Scott, M. D. & Irvine, A. B. 1987. The social structure of free-ranging

bottlenose dolphins. In: Current Mammalogy, ed. H. H. Genoways. New York:


Wells, R. S., Boness, D. J. & Rathbun, G. B. 1999. Behavior. In: Biology of Marine Mammals,

ed. J. E. Reynolds III & S. A. Rommel. Washington, DC: Smithsonian Institute Press,

pp. 324–422.

Winn, H. E., Thompson, T. J., Cummings, W. C. et al. 1981. Song of the humpback

whale: population comparisons. Behavioral Ecology and Sociobiology, 8, 41–46.

Xitco, M. J. & Roitblat, H. L. 1996. Object recognition through eavesdropping: passive

echolocation in bottlenose dolphins. Animal Learning and Behavior, 24, 355–365.


19

Looking for, looking at: social control,honest signals and intimate experiencein human evolution and history

j o h n l . l o c k e

City University of New York, USA

Introduction

Recently, Hauser et al. (2002) argued that if we are to understand human

language, several disciplines must work cooperatively. Predictably, these include

linguistics and certain areas within psychology and anthropology as well as some

relative newcomers: biology and animal behaviour. However, if collaboration can

facilitate the investigation of language, long held to be a uniquely human faculty,

it is surely indispensable to the study of human communication, for which a

number of homologous or analogous processes exist in other species.

In the case of language, a behaviour with countless social benefits, researchers

have tended to focus on dyadic interactions. In the typical model, the ‘sender’ is a

rational human being who has information. As a social being, the sender wishes to

share it. The ‘receiver’, equally rational and social, wants to hear it; so the receiver

listens and makes an appropriate response. ‘Communication occurs,’according to

one authoritative source, ‘when one organism (the transmitter) encodes informa-

tion into a signal which passes to another organism (the receiver) which decodes

the signal and is capable of responding appropriately’ (Ellis & Beattie, 1986, p. 3).

Dyadic interactions such as these occur, of course, and deserve linguists’ theo-

retical attention. However, in a gregarious species such as ours – and this is a major

point of divergence between social communication and linguistic interaction –

dyads are often embedded in aggregations of individuals, in various arrange-

ments (communication networks in the sense of this book), and these will usually

include one or more perceptual bystanders. If thought to be unobserved, dyads

tend to behave in an unguarded way, making them unusually interesting, and


416


Humans: control, signalling and intimate experience 417

their behaviour unusually informative, to uninvited viewers and listeners. Of

course, the perceptual target may be alone, acting without recourse to the dis-

plays or material objects that are normally used to project his more public self.

These solitary behaviours will be less veiled than the dyadic interactions, mak-

ing them especially useful to anyone who stands to benefit from prediction-grade

social knowledge.

What I am suggesting is that behaviour which is neither donated by the actor

nor observed with his awareness is likely to be unusually high in reliability. It is

also likely to be intimate: that is, sufficiently personal that the actor might like

to shield it from prying eyes and ears. Reliability and intimacy give prospective

observers two rather compelling reasons to sample such behaviour, but it will

usually be impossible for them to do so overtly. This gives rise to eavesdropping,

a form of information gathering that in humans occurs only by stealth.

In animals, research has addressed two broad areas of observation. One relates

to the information that is obtained when animals look for other animals. When

non-human primates do this type of looking, their focus is typically on the loca-

tion and activities of outsiders, including predators and competitors from other

groups. Typically, this is referred to as vigilance. Other types of information are

obtained when animals look at the constituents of their own groups. This type

of looking, social observation, is addressed later in the chapter and elsewhere in

this volume (e.g. Ch. 25). Predictably, these discriminable functions are associated

with different benefits.

Looking for

Animal vigilance

When animals look for other animals, their tendency is to scan territorial

boundaries in order to detect encroachment of predators or competitors. Early

detection alerts individuals and, through their reactions, other group members

to the need for evasive or defensive action. The perceptual act is performed from

the naturally exposed position of group-living animals. Vigilance appears to be

a form of perceptual alertness that occurs in anticipation of important events,

rather than a form of observation per se, and may even be discontinued when

those events occur. The observing itself is performed as sporadic interruption of

other activities, rather than a circumscribed commitment of looking time (an

exception, ‘sentinels,’ will be discussed below).

Much of the research on vigilance involves non-human primates. They, like

other animals, need to look out for predators. Vigilance thus produces valuable

information, but it comes at a price. Red colobus Procolobus badius tephrosceles and

redtail monkeys Cercopithecus ascanius schmidtii typically spend over 50% of their


418 J. L. Locke

time just visually scanning (Treves, 1998). Similar figures have been obtained for

chacma baboons Papio cynocephalus ursinus (Cowlishaw, 1998). This does not count

all the time these primates spend in a vigilant state, a figure that may approach the

totality of free-ranging animals’ waking time.

In several different species of monkey, animals that were physically isolated

spent more time looking than monkeys that were near a group member (Steenbeek

et al., 1999; Treves et al., 2001). This difference may reflect a greater fear of predation

on the part of solitaires. Where attacks are unlikely, however, animals might be

expected to devote more of their attention to members of their group. Research on

brown capuchin monkeys Cebus apella in Iguazu Falls National Park in Argentina,

where the annual rate of predation is extremely low, suggests that this may be so

(Hirsch, 2002).

Predictably, there are also variations in time spent looking among the mem-

bers of a single group. Subordinate animals tend to look more than dominant

ones, largely because they spend a great deal of time watching the dominant

animals themselves. These rank differences were suggested some years ago by

Chance (Chance, 1967; Chance & Jolly, 1970) and have since been confirmed in

a number of species, including long-tailed macaques Macaca fascicularis (Pitcairn,

1976), talapoins Miopithecus talapoin (Dixson et al., 1975; Keverne et al., 1978) and

brown capuchins (Hirsch, 2002). In each group, subordinates more often look at

dominant animals than the reverse arrangement.

In many studies, there have also been sex effects. A male vigilance bias has been

witnessed in various primate groups in at least seven different studies (Cheney &

Seyfarth, 1981; Fragaszy, 1990; Baldellou & Henzi, 1992; Rose, 1994; Rose & Fedigan,

1995; Gould et al., 1997; also see reviews by Quenette, 1990; Steenbeek et al., 1999). A

great deal of male vigilance appeared to be directed outside the group, presumably

to predators or sexual competitors, but perceptual targets are notoriously difficult

to identify in free-ranging animals.

There is one exception to the usual hierarchical pattern. High-ranking males

frequently assume unusual responsibilities for vigilance (Rose & Fedigan, 1995).

In cooperative groups such as vervets and marmosets, as well as baboons, they – or

some other large male – may even take on the role of sentinel and adopt a superior

vantage point. These individuals then become the focus of attention for group

members, who monitor the sentinel instead of looking for predators themselves

(Hall, 1960; Horrocks & Hunte, 1986; Koenig, 1994).

The behaviour of sentinels has been described in detail. The tendency is for the

sentinel to ascend a tree or rock, mainly so foraging animals can achieve visual con-

tact, making auditory warnings unnecessary (Horrocks & Hunte, 1986). In green

monkeys Cercopithecus aethiops sabaeus, this works so well that when a sentinel

detects approaching humans, his quietly visible movements may enable the troop



to disperse without detection (Poirier, 1972). Therefore, in this species at least, the

sentinel seemed to ‘tip off ’ the foragers with subtle visual behaviours instead of

warning them with loud barks. In the case of baboons, the vigilance itself involves

repeated 180◦ head turns, which take about five seconds (Hall, 1960), and in this

group detection produces barks. In every report I have read, the only ‘predators’of

possible concern to the sentinel were human and the ongoing feeding behaviours

best characterized as ‘raids’ on a plantation (Maples et al., 1976). Consequently the

use of sentinels may be predator and context specific.

Much of the time, the targets of vigilant males are other males, including

interlopers from outside conspecific groups (Rose & Fedigan, 1995). Therefore, the

motivation for male vigilance may be more closely linked to a self-oriented control

function than a contribution to group welfare. This issue will assume a broader

significance below when we see some related sex differences in our own species.

Levels of vigilance vary, affected by a variety of internal factors such as mating

periods, births and infant excursions away from their mothers. In vervet monkeys

Cercopithecus aethiops, males are more vigilant than females, especially during the

breeding season (Baldellou & Henzi, 1992). In a study of black howler monkeys

Alouatta pigra, female vigilance rates increased after the birth of infants (Treves

et al., 2001). In squirrel monkeys Saimiri boliviensis, tape recordings of infant vocal-

izations increased the time that adult females spent looking for predators fivefold

(Biben et al., 1989).

In several species, it has been shown that visual obstruction alters the usual

benefits of herding and flocking. The first to study this effect was Underwood

(1982). He noted that African antelopes frequently interrupted their foraging to

look around, but when grazing in tall grass they spent even more time lifting their

heads to look at distant areas. Metcalfe (1984a,b) observed a similar pattern in two

different species of shorebirds in western Scotland. He found that in both species

the time devoted to vigilance rose with increases in the density of obstructions such

as rocks, boulders and banks of seaweed. He also noticed that obstructions broke

up the usual relationship between flock size and vigilance. Metcalfe reasoned that

obstructed animals were in a vulnerable position, unable to see if predators or

potentially protective neighbours were nearby.

Recapitulating, the primary functions of primate vigilance appear to be defen-

sive when looking is externally directed and the threat of predation is high, and

social when looking is internally directed and the threat of predation is low. In

the latter, vigilance enables animals to evaluate dominance relations – a criti-

cal function in primate societies – and resource-holding potential. In the species

studied, the primary sensory modality has been visual. In non-human primates,

males and subordinates generally devote more time to vigilance than females and

dominants. The target of review is a physical area, such as the perimeter of an


420 J. L. Locke

occupied territory, or other animals that are either within or outside the group.

The observer may make no attempt either to conceal or to expose his position.

Human vigilance

Humans have not been reluctant to engage in the ancient and deeply

ingrained behaviour of vigilance. Some of the benefits derived from these activities

resemble those enjoyed by our evolutionary ancestors, broadly understandable as

social knowledge and social control. As we will see, our species critically relies

on information that can only be obtained through these one-way processes, and

yet there is little record of empiricism. To be sure, there have been psychological

studies of vigilance – usually defined as the detection of prespecified perceptual

targets that occur infrequently, irregularly and weakly – in relation to a range of

military and industrial issues. Currently, there is concern with baggage scanning

in relation to airport security screening.

It is difficult, however, to find reports of research conducted within an ethologi-

cal framework. Few investigators have asked how humans exercise vigilance with

respect to strangers or potentially aggressive intruders. Yet, in societies wishing

to guard against crime and terrorism, citizens are concerned with precisely this

issue.

Until about 20 000 years ago, our ancestors spent much of their time follow-

ing herds of large animals from place to place. But when the herds dissipated,

nomads began to hunt smaller game, to fish and to gather. This shift enabled

the new sedentists to spend more time in their resting places before seasonal

changes precipitated the next round of migration. At this stage in history, one

assumes that human and non-human primates behaved rather similarly with re-

spect to vigilance. Since human groups were several times larger, there were more

individuals that had to be monitored, but this was obviously manageable as our

premodern ancestors lived almost as openly as the other primates.

The nature of their encampments is implied both by archaeological evidence

and the behaviour of an existing group whose way of life is thought to replicate

ancient patterns of living (Lee, 1979). This group is the !Kung, a population of

largely egalitarian hunter–gatherers who inhabit the Kalahari Desert of Botswana

and southwest Africa. Although their way of life is changing, in the mid- to late-

twentieth century, when they were studied fairly intensively, most of the !Kung

lived in bands of 50 or 60. These bands periodically dispersed into still smaller

groups or concentrated into larger ones, as suited their needs.

The typical camp was laid out in concentric circles. In the centre was a public

gathering place. Rimming this plaza were the bandsmen’s grass huts, which were

used mainly for storage. These were packed very closely together, enabling bands-

men to perceive and react to the earliest and subtlest acts of an antisocial nature.



‘If a person is angry,’ wrote Draper (1978, p. 47), ‘someone, if not everyone, will

soon know about it.’

Significantly, the !Kung rarely if ever entered their huts to escape scrutiny. The

reason is that it was considered improper for anyone to withdraw from the sociality

of camp life, either physically or psychologically. ‘To seek solitude,’ according to

Lee (1979, p. 32), ‘is regarded as bizarre behaviour.’But this attitude toward privacy

was not unique to the !Kung. There are several other openly living groups.

1. The Baktaman of New Guinea. ‘There are no recognised and respected

ways in which the public gaze can be cut off, no way of separating

oneself out from others present’ (Barth, 1975, p. 24).

2. The Mehinacu of Central Brazil. ‘Wherever a person goes in the village

he can be seen or heard. When he speaks there is a chance that a third

person is listening, and that in a short time everyone else will know

what he said. Even the most intimate details of his sex life often become

a matter of public knowledge’ (Gregor, 1970, p. 238).

3. The Nayaka of southern India. ‘They remain sited by their respective

fire-places, and talk across space from fire to fire . . . they rarely try to

conceal their domestic activities’ (Bird-David, 1994, pp. 590–591).

4. The Samoans. They ‘live most of their lives in a very public arena. The

more private aspects of experience are strongly discouraged by the

absence of walls in a Samoan house, and by powerful norms of social

life, which keep people in almost constant social interaction’ (Shore,

1982, p. 148).

5. The Sakalava people of Madagascar. ‘To stay alone in the house is

considered a sure sign of evil intent.’ (Feeley-Harnik, 1980, p. 568). A

house with curtains on the outside doors, or fences and walls, was also

seen as a threat to normal sociality. Even the house itself could pose

problems, Feeley-Harnik wrote, since it is meant to remove the

occupants from the larger social order. ‘Secrecy and separation,’ she

continued, ‘indicate at best a lack of generosity, a suspiciously

anti-social striving for distinction’ (Feeley-Harnik, 1980, p. 581).

6. Villagers in the mountainous Zinacantan region of southern Mexico.

They too have also been suspicious of too much domestic privacy. The

typical home is fenced in, and village folk are forbidden from passing

through the fence without prior approval. However, staying indoors, or

closing the house door, is considered ‘a gross and open admission of

being up to no good’ (Haviland & Haviland, 1983, p. 347).

Many of these cultures that have opposed privacy and favoured social visibil-

ity were egalitarian; according to Bailey (1971, p. 19), ‘equality is the reward for


422 J. L. Locke

constant vigilance’. In parallel, there was also a suspiciousness of structures and

behaviours that reduced visibility, since these would surely foil the only proven

means of keeping the group together and under control.

I consider that there are three benefits of vigilance and eavesdropping: social

control, honest signals and intimate experience. Let us now examine the first of

these, which is closely linked to vigilance. Honest signals and intimate experience

are tied to privacy, thus to eavesdropping, and we will address these benefits in

that section. Since there is little in the way of relevant research, my treatment of

social control will necessarily be historical, discursive and somewhat speculative.

Social control

Some things that occur in private are intended to be secret. They may be

offensive, morally wrong or even criminal. There are good reasons for humans to

observe this activity, too, but in some cases there may be little benefit in doing so

covertly. The reason is that looks, if interpreted as gazes, can also send messages of

their own. Some are confrontations that vary in intensity from ‘I see you’ to ‘I’m

keeping an eye on you’ and, in the extreme case, ‘Back off ’.

On the community level, the protective function of surveillance has always

been clear. ‘If by chance some good-for-nothing appeared in the neighbourhood,’

wrote Yves Castan (1989a, p. 49) in reference to French villages, ‘there were plenty

of eyes to survey his movements’. In Victorian England, the rich and powerful

lived side by side with the poor and powerless. This made it possible for each

group to observe the other and particularly for establishment figures to keep an

eye on potentially troublesome subordinates. ‘The middle-classes desired privacy

for themselves,’ wrote Olsen (1974, pp. 275–276) ‘but wished the lives of the lower

orders to be lived in the full blaze of publicity. Street improvements and slum

clearance schemes were designed to bring the poor out into the open, where they

could be observed, reproved and instructed by their superiors.’

On an individual level, vigilance also enables humans to avoid quarrelsome

or dangerous people, our equivalent of predators. In large cities, one is forced to

acquire ‘street smarts’, an awareness of menacing strangers in relation to oneself,

and the relation of one’s own location to places of safety. Predictably, the best

security – as criminologists have shown – is the presence of some reasonable

number of non-predatory people on the street. For a city to be safe, ‘there must be

eyes upon the street,’ wrote Jacobs (1961, p. 45), ‘eyes belonging to those we might

call the natural proprietors of the street’. Research in the ensuing years has been

supportive of this view (Kelling & Coles, 1996).

Primates in hierarchically organized groups spend more time looking at each

other when they could be looking for predators or food (Caine & Marra, 1988)

and animals do more social looking within mixed than in homogeneous groups

(Treves, 1999). These findings are relevant to Putnam’s (1993) study of provincial



self-government in Italy. Provinces that lacked trust, he found, spent a great deal

of time keeping an eye on each other.

Watching humans

With vision alone, non-human primates pick up cues to sex, age and rank,

the last inferred from dominance and submissive displays. We humans care about

these things, too, and also transmit much of our information visually, through

physical alterations and adornments. Among the Kayapo of the Amazon forest,

visible affectations include pierced ears, lip plugs, penis sheaths and body painting

(Turner, 1980). These adornments convey messages about status as well as personal

roles and significance and do so just as surely as verbal signals. In modern societies,

hairstyle, cosmetics, jewellery, eyeglasses, tattoos and body rings – to say nothing

of cell phones, water bottles, clothes, shoes, handbags, briefcases, fanny packs,

shopping bags and backpacks – send visual signals about who we are or how we

wish to be perceived.

The desire to enhance personal images goes back at least 28 000 years. Studies

of the ‘Venus’ figurines and burial sites indicate that women many millennia ago

were already wearing hats, dresses and various bodily adornments (Soffer et al.,

2000). This suggests that, before they were securely and privately housed, our

historical ancestors already had some sense of self, a matter to which we will

return shortly.

We are not, of course, merely intelligible through our clothes and other objects

of material culture. Like other species, humans have a number of ritualized action

patterns that presuppose visualization (cf. Smith, 1977). These include the facial

and bodily displays that emerge in infancy, are seemingly universal (Schiefenhovel,

1997) and occur in blind as well as sighted infants (Eibl-Eibesfeldt, 1973). Under

the influence of culture, humans take on additional gestures – some functioning

as salutations, others signalling transition points in verbal engagements (Kendon,

1990) – and learn rules of proxemics that suggest possible ranges of interpersonal

distance (Hall, 1966). Personal status and relational intimacy are also revealed by

touching (Hall, 1996) and, in the case of single women in America, hair flips and

head tosses (Moore, 1985).

The eyes send many different types of social and emotional signal. We saw

earlier that in primates socially dominant individuals receive more gazes than

subordinate ones. This relationship also holds in humans. At any given moment

in time, the person who is being looked at is usually the person who is talking and

that will typically be the person with the highest status (Bales et al., 1951; Fisek &

Ofshe, 1970; Exline et al., 1975; Abramovitch, 1976; Kalma, 1991).

With all these visible signals, it would be surprising indeed if people did not

create opportunities to be looked at and to do so on their own terms. In 1800,

Parisians began to put their public selves on parade. That is when pavements


424 J. L. Locke

came to Paris and merchants adaptively repositioned their shops and displays.

In the new pavement cafes, the chairs were ‘always placed towards the street,’

wrote an urbanologist, ‘as the chairs in a theatre are placed towards the stage’

(Oosterman, 1992, p. 161).

Promenades were once expected to achieve an instructive or regulatory func-

tion. When a family went out for an evening stroll, it was assumed that the husband

would ‘see himself as others saw him,’ according to Cranz (1980, p. S80), ‘the head

of a family, wife on arm, children in tow, all in Sunday best. Reformers reasoned

that he would experience this as pleasurable and resolve to make it the mainstay

of his life.’ In 1890, the commissioners of Boston’s parks department saw public

viewings as a course of moral instruction. The mere sight of families was expected

to exert ‘a wholesome influence’on other patrons, and to do so far more effectively

than laws and police ever could (Cranz, 1980, p. 581).

In contemporary America, recreational vigilance is largely carried out in parks

and malls. In a survey conducted in the early 1970s, a fifth or fewer of the patrons

of two parks in Portland, Oregon said they went to the parks to walk, eat, talk,

read, engage in crafts or hobbies, or exercise. Far more patrons, fully 55%, said

they went to the parks in order to watch other people (Love, 1973). In a survey

conducted in the Los Angeles area a decade later, adolescents said that the main

reason they went to a particular mall, after shopping, was to look for members of

the opposite sex (Anthony, 1985).

There are several circumstances in which vigilance is exercised in relation to

intimate relationships. Buss (1988, 1997) surveyed American couples to see how

frequently they reported the use of vigilance in order to control intimate relation-

ships. Items in the survey included unexpectedly calling and dropping by a place

to see if the partner was there and remaining nearby, or at least in visual contact,

during social engagements. Men and women reported equal levels of vigilance,

but there was a significant correlation for men, and not women, between levels

of vigilance and ratings of partner attractiveness.

Control and intimacy are also conjoined in many cases of the crime ‘stalking’.

Since stalking is usually defined as an unwelcome act of ‘perceptual following’

that is overt or blatant, it qualifies as vigilance. In one study, 57% of stalkers had

previously been in an intimate relationship with the victim (Hall, 1998). In another

study, approximately a third of all stalkers were considered intimacy seekers. Most

lived alone and had never had a romantic partner (Mullen et al., 2000).

Ethological studies of human vigilance

In non-human primates, individuals tend to look up less often when a

group member is nearby (Hirsch, 2002). A similar trend has been found in humans.

Observing students in a university snack bar, Barash (1972) found that cumulative



looking-up frequency was significantly higher in solitaires than individuals in

groups. In a similar study conducted in Germany, Wirtz & Wawra (1986) observed

university students having lunch in a refectory. Each sat alone or with one to four

other students. In these subjects, it was found that the time spent looking away

from the table steadily decreased as the number of number of people at the table

increased, possibly because this increased the proximity of others.

In Wirtz & Wawra (1986), male students spent significantly more time looking

away from their table than females. This fits with primate research, reviewed

earlier, that revealed a male looking bias, particularly for distant areas. It also

agrees with Aiello (1972, 1977), who found that men looked significantly longer

at each other than women did when seated 10 feet (3.2 m) or more apart, a trend

that was reversed for shorter distances.

Paradoxically, most of the work on vigilance in humans involves detection

of signals, whereas in animals, vigilance involves attention to the existence and

behaviour of individuals (also see studies of social monitoring and comparison).

The disposition of females to look longer at near individuals may be linked to a

tendency to rely on the support of group members, while the disposition of males

to look longer at distant individuals may be associated with the need to address the

threats posed by strangers. Stripped to the basics, here are two issues – intimacy

(the network ‘glue’) and control – that concern human women and men. These

issues, as we will see, have been connected to sex differences in social monitoring

for the last six or seven centuries of recorded history.

Looking at

Eavesdropping in animals

Much information is acquired by social observation: looking at con-

specifics. For example, male Mallee dragon lizards Ctenophorous fordi produce sig-

nificantly more ejaculate and spend 60% more time copulating with a female

previously seen copulating with another male than do males not having this prior

perceptual experience (Olsson, 2001). In this example, the source of information

did not involve signals. However, an important subset of observational informa-

tion comes from the signals of others. Such information is gathered by eavesdrop-

ping, a behaviour that is defined in animals as ‘the use of information in signals by

individuals other than the primary target’ (Ch. 2). The context for eavesdropping

is a communication network (Ch. 1); therefore it is not surprising that this volume

discusses at length the evidence for eavesdropping by animals (e.g. Chs. 2 and 5).

There have also been a number of recent reviews of eavesdropping (e.g. McGregor &

Peake, 2000).


426 J. L. Locke

Among the primates, maintenance of societies – including kin and power rela-

tions – requires that individuals spend a certain amount of time gathering infor-

mation about other group members. These internal appraisals, which in human

research are usually called social comparison and in non-human primates are con-

sidered a form of social vigilance (Hirsch, 2002), are required if animals are to alter

or maintain their status. Since Cheney & Seyfarth (Ch. 25) describe research in this

area, I will limit my own review to studies that expose links to our own species.

Primates’resources include cooperative relationships, which may involve high-

ranking animals. As Whiten (1993, p. 719, italics his) has pointed out, ‘simply to be

seen by others grooming with high-ranking A, or chatting with high-status B, is worth

something to the individual because of what this advertises with respect to future

coalition.’ An individual that has these kinds of social resource is considered to be

rich in ‘social attention holding potential’ (Gilbert, 1989). Animals may also look

within the group for individuals with valued physical resources. For example, a

perceptual target of so-called ‘scroungers’ is the foraging success of other animals

(Beauchamp, 2001).

There are variations between species in social vigilance, partly because of dif-

ferences in social organization. Consider squirrel monkeys, which live in large

groups that are characterized by cliques, subgroups and dominance hierarchies,

and cotton-top tamarins Saguinus oedipus, who live in more egalitarian family

groups known for cooperation, sharing and relative peacefulness. In a compar-

ison of social looking during foraging, the congenial tamarins devoted 17% of

their time to within-group vigilance. The more competitive squirrel monkeys, by

contrast, devoted 45% of their time attending to group members (Caine & Marra,

1988).

A feature of eavesdropping by animals is that it is usually carried out by isolated

individuals who do not subsequently share their perceptual intake with others,

although in a densely populated area other observers may individually sample

the same activity on their own. The perceptual target of eavesdropping is often a

pair or small group of individuals, which provides the observer with interactive or

relational information connected with fighting or sex, and it does so with minimal

risk or expenditure of effort.

Eavesdropping in humans

In this section, I follow accepted semantic practice and use the term eaves-

dropping only where the act of observation occurs surreptitiously. I also include

cases of social vigilance that do not involve signals or interactions (as do Cheney &

Seyfarth in Ch. 25).

I noted earlier that domestic vigilance has gone largely unstudied ethologi-

cally, but when it comes to human eavesdropping there is no record of empiricism



whatsoever. To be sure, there are publications that use the word eavesdropping,

but these typically describe government anti-crime programmes that include wire-

tapping and surveillance. What is missing is research on the behaviour of social

eavesdropping in a naturalistic context.

The lack of research on human eavesdropping seems odd, since the practice

is neither rare nor lacking in benefits. When asked, people usually admit that

they have eavesdropped in the past, or even do so habitually. Frequently, the

admissions are offered shyly, occasionally with embarrassment, but I have yet to

find anyone who denies ever having engaged in this practice. This is not to say

that everyone peeks through keyholes. Most of us ‘tune in’ less adventurously.

When in a restaurant or waiting room, for example, we tend to accomplish our

perceptual business in a number of optical stabs, interrupted by bogus glances at

other features of the physical or social landscape. If the subject suddenly looks

up, the invasion may be disguised by a slow and smooth deflection, as though a

continuous sweep was in progress when the ‘interruption’ occurred.

If people are naturally inclined to penetrate the private spaces of others, and

just as naturally resist such intrusions themselves, one might expect historical

evidence of these dispositions, perhaps in art or literature. In fact, there was activ-

ity in both media in the seventeenth century, from the paintings of Dutch artist

Nicolaes Maes to the novels of Le Sage and Hawthorne and the plays of Marivaux.

These depictions suggest that our historical ancestors were acutely aware of eaves-

dropping and may even have approved of it. However, there are also church and

court documents going back three centuries earlier, in several different cultures,

and these tell us something about the relative frequency of eavesdropping as a

behaviour, and a crime.

The !Kung hunter–gatherers, as we have seen above, welcomed round-the-clock

surveillance and intentionally subjected themselves to a panoptical living arrange-

ment (cf. Bentham, 1791). This made vigilance easy, but for the same reason it made

covert eavesdropping impossible (P. Wiessner, personal communication).

One assumes the !Kung’s residential arrangement was somewhat representa-

tive of historically earlier ways of living, when variations in the availability of

food required individuals constantly to relocate. With the advent of agriculture,

however, the new sedentists departed from the hunter–gatherer pattern, build-

ing huts that could be lived in – not just used for storage – and spacing them

more widely. This necessitated an aggressive form of perceptual intrusion. At the

same time, groups began to expand, and strangers grew more numerous. In a brief

space of time, a lifestyle that was two million years old – open living, with visual

monitoring – began to unravel.

Before the development of structural privacy as it is enjoyed in modern soci-

eties, some degree of solitude was achieved behaviourally. Bird (1983) reported that


428 J. L. Locke

the Naiken people of India exhibit nachika, literally a ‘shyness’ or ‘reticence’ that

protects them from direct encounters with others. Since the Naiken live openly,

reticence provides relief against what Bird called ‘involuntary intimacy.’

Similar observations were made by Fejos (1943), who studied the Yagua people

of northeastern Peru in the early 1940s. All the families of a clan, which ranged

from 25 to 50 members, lived communally in one large house. Fejos noted that

although there were no partitions, members could achieve privacy at any time

simply by turning away. ‘No one in the house,’ wrote Fejos (1943, p. 87), ‘will look

upon, or observe, one who is in private facing the wall, no matter how urgently

he may wish to talk to him’.

Note that the privacy achieved by these individuals was in each case negotiated

with, and conferred by, others. It began when the privacy-seeking individual gave

an observable sign. The observers, out of respect for the person, then reduced or

suspended evaluation. Goffman (1963) called this ‘civil inattention’.

Perhaps these behavioural means of securing privacy were sufficient, for even

with inclement weather, social competitors and wild animals, little interest was

shown in domestic walls (Carpenter, 1966; Rapoport, 1969). Consequently one is

curious about the residents of more hospitable climes who nonetheless chose

to live behind walls. The reasons for these exceptions to climatic determinism,

Rapoport (1969) pointed out, may have had something to do with religion, status

or some ‘other’ factor. One candidate for the ‘other’ factor, according to Wilson

(1988), would have been the desire to escape constant scrutiny. But there is another

possibility.

We have already seen that the !Kung sat in full view of each other during their

time in camp. If new members continually join such arrangements, eventually

something has to give. Individuals who cut back on their looking time will dis-

cover that the machinations of an increasingly complex, if not Machiavellian,

society have left them frightfully unaware and out of step. Alternatively, those

who continue to crank up their looking time in step with population growth will

soon have no time to do anything but look. Therefore, the critical factor may have

been the need to minimize the time that they, as members of burgeoning groups,

had to spend surveying the social landscape.

Little wonder that the desire for privacy grew as people became accustomed to

domestic life. In the 1960s, the Sarakatsani were a small group of shepherds who

alternately, by season, inhabited the Zagori Mountains and plains of Greece. To

them, a hut was inviolate. ‘Whatever takes place within the sanctuary of its walls

is private and sacred to the members of the family’ wrote Campbell (1964, p. 292).

‘No stranger may invade it without an invitation.’

Occupants could only be safe by assuming that they, like birds (e.g. Metcalfe,

1984a,b), grazing animals (e.g. Underwood, 1982) or isolated monkeys (e.g. Treves



et al., 2001) were ineligible for assistance. Therefore, the need was for walls

that were not merely visually obstructive but also secure. Eavesdropping became

the only way to restore information that had once been available. However,

eavesdropping was not merely restorative, for with increasing privacy, this very

penetrant means of observation became the only way to obtain the newest

and highest grade of personal information, one that had never been available

previously.

Honest signals

As people spent more time behind walls, direct sensory information about

them became less available. This posed problems for the community, but the

experience of privacy also altered the people themselves. In time, the most honest

and reliable information about individuals was only available to those who were

behind the walls, or in personal relationships, with them. Others were excluded.

Therefore, behaving as trained ethologists, eavesdroppers attempted to conceal

themselves in order to avoid detection, which would alter or discontinue the flow

of desired information.

People behave differently when they believe others are unable to see them.

When shielded from public view, they have the opportunity literally to compose

themselves – to decide who they are and how they would like to be perceived

by others. When they plunge into the social world again, they may then do so

appropriately dressed and ornamented, presenting others with the image they

would most like to convey.

Earlier I referred to a low-grade sense of self that antedated, or occurred early in

the development of, domestic life. While the hominids may have had some level of

self-awareness, along with the other primates (Hauser, 2000), every domestically

living human now has two selves: one public, the other private. The public self is

the way we are in the presence of others. Our private self is on view only when

individuals are alone or with intimate friends. There is a telling fact about the

private self, in connection with the process of perceptual theft. It is, as Baumeister

(1986, p. v, italics mine) said, ‘the way the person really is’.

The dishonest signals that are issued in public are not worthless, of course.

These may provide information as to the way a person really is not. For example, a per-

son who is making a conspicuous display of wealth may be ‘financially strapped’–

not wealthy at all – and also seeking to hide this fact for a reason, one that with

further analysis may be discovered. Still, people in private are likely to act in ways

that are, as Baumeister said, more ‘them’. This fact, by itself, increases the reliabil-

ity of private behaviour. However, private behaviour is also privileged. This gives

others reasons to want it, for as humans they have the inherited dispositions of

evolutionary ancestors whose survival was dependent on the ability to observe


430 J. L. Locke

behaviour and infer intentions, and they enjoy vicariously experiences that are

not issued for the benefit of observers.

Intimate experience

After nearly two million years of watchfulness, walls enabled the chroni-

cally wary Homo sapiens to cast its senses inwards for a change. Dwellers could pay

undivided attention to compelling tasks not just for three seconds, but for three

hours. Free from the stares and queries of villagers, the new residents could begin

to examine their own lives and think about how they differed from others. With

shielding, they could create or discover the existence of a deeper and more reflec-

tive form of themselves and begin to contrast this with their public presentation.

The time that domestication liberated from external vigilance could be devoted

to matters that were occurring – or with additional attention could be initiated –

on the inside. These would have included personal and communal activities. Fam-

ily members could, at last, devote their undivided attention to each other. By

creating an ‘outside’, individuals found ways in which more intimate relations

could be developed with members of the family on the ‘inside’. Consequently, to

look inside a house was to get unprecedented glimpses of intimate behaviour.

If someone peeks through a crack or keyhole, how will this come to the attention

of researchers? Eavesdroppers are no more likely to be detected by an ethologist

than by their perceptual prey, nor would they be likely to describe their activities

truthfully to an interviewer. The situation seems hopeless, and yet we do know

something about eavesdropping, especially the kind that occurred many centuries

ago when the threat of privacy was new and attempts to breech it were frequent,

adaptive and perhaps even honourable. In sixteenth century England, there was

a law against adultery and it required eye-witness testimony. Court records have

been preserved, revealing the testimony and identity of witnesses. Frequently, the

lead or sole witness was a woman who had peeked through a door from within

the house, or crack in the wall from an adjoining house. In a case that occurred in

London in 1598, a housewife named Margaret Browne watched a tryst involving

the woman who lived next door, her looking ‘bout’ – like the adulterous activity

itself – lasting for an entire afternoon (Crawford & Gowing, 2000).

I have inspected many cases involving this sort of domestic eavesdropping.

Although I kept no detailed count, it is clear that the typical perpetrator was

female. One might suppose that this is because women were merely home more

often, but in the sixteenth and seventeenth centuries, the husband was often

somewhere about the house, too. When Margaret Browne saw what was happening

next door, she called her husband to the crack to confirm her observations.

Mr Browne took a brief look and left, but Margaret remained at the crack,

taking mental notes. Her courtroom testimony two weeks later was extraordinarily

detailed, down to the exact words and phrases of the lovers and details of their



various sex acts, as well as the colour of her neighbour’s underwear. Clearly, this

was a memorable experience for Mrs Browne.

In early fourteenth century France, they had something more ominous than

adulterers: heretics. In Montaillou, a small village in the Pyrenees, there was a

group of Cathars that were actively working to oppose the Catholic Church. In

order to get needed evidence, the church requested parishioners to bring in eye-

witness testimony, which was subsequently used in court (Le Roy Ladurie, 1978).

These church records were preserved, testimony revealing that the women of Mon-

taillou, in general, were unusually active in the more subtle form of eavesdropping

that involves listening at keyholes and looking through holes in domestic doors

and walls. The men ‘were inquisitive enough,’ wrote Le Roy Ladurie (1978, p. 257),

‘but their curiosity was nothing beside that of the women’ (italics mine).

Centuries later, in a completely different context, a similar comment was made

about the women of Italy. These women, too, were ‘curious by nature’ according

to Nicole Castan (1989b, p. 417). ‘Women of the lower orders shamelessly admitted

it.’ One confessed that ‘she was “obliged” to follow the movements of a passer-by,

another that she could not help overhearing a conversation or lying in wait for a

neighbor’.

While the courts welcomed eavesdropping as eyewitness testimony, they pun-

ished cases of eavesdropping when it proved to be disruptive to community life. In

England, it was a crime to ‘listen under walls or windows, or the eaves of a house

to hearken after discourse, and thereupon to frame slanderous and mischievous

tales’. Data analysed by McIntosh (1998) revealed that for a good 200 years, begin-

ning in the 1370s, eavesdropping made up about 8% of all social crimes. But here

we find a sex reversal; during this period, about 80% of the courts having some

incidence of eavesdropping happened to hear male cases only.

Why such a high percentage of men? McIntosh (1998) suggested that the men

who were caught listening under eaves were actually attempting to control their

communities by investigating the possibility of domestic misbehaviour. If so, many

of the arrests for eavesdropping may well have been instances of vigilance, an

activity that in other primates also favours males. The irony is that much of the

eavesdropping – a misdemeanour – was undertaken by people who may have been

attempting to prevent domestic misbehaviour.

When the English eavesdroppers witnessed moral transgressions, the obvious

next step was to broadcast what they had seen and this is what many did. However,

recall that the second part of the eavesdropping law involved framing ‘slanderous

and mischievous tales’. The fourteenth and fifteenth century English were still in

the process of privatizing and so felt ambivalent about publicizing the results of

perceptual invasion.

The male eavesdroppers may have been attempting to police their communi-

ties, but they could also have been attempting to control individuals. I base this,


432 J. L. Locke

in part, on the fact that stalking, as discussed earlier, is widely understood as a

means of controlling the life of another and it also has a near-identical sex bias to

medieval English eavesdropping. In a large American survey, 87% of the stalkers

were male (Tjaden & Thoennes, 2000) and similar statistics are available for other

cultures (Mullen et al., 2000).

If honest signals, social control and intimate experience are interconnected, it

would not be surprising if visual monitoring sometimes leads to eavesdropping. In

Mineville, a town of 1000 inhabitants in America’s Rocky Mountains, Blumenthal

(1932, p. 103) noted that peoples’attempts to live privately merely inflamed the cu-

riosities of others. Some, he wrote, became ‘more thoroughly known than would

have been the case had they not tried so obviously to guard their privacies, for

in doing so they made themselves mysterious, and thus stimulated the curios-

ity of the people so that more than ordinary attention was given to discovering

something about them’.

Because of such interconnections, it may be difficult to carry out a motivational

analysis on anything but the initial bout of observation. In the daily parade of

public selves, people in search of honest signals have been forced to invade private

spaces, thereby accessing the intimate experience that occurs there, finally finding

themselves in possession of knowledge of the kind that leads to social control.

Looking at and for: a functional comparison

Cheney & Seyfarth (Ch. 25) have described the need of primates to moni-

tor their fellow group members, but do non-human primates actually engage in

dictionary-definition eavesdropping: that is, observe under conditions of stealth?

It is not clear that researchers have asked this question, and yet it appears that an-

imals sometimes secure conditions of perceptual privacy – a circumstance that

favours eavesdropping – before undertaking certain behaviours. For example,

when subordinate males approach females in oestrus, they look around, evidently

to see if they and their intended partners are under review. This is evidently be-

cause the sight of a presenting female is arousing and may produce unwanted

competition (Hall & De Vore, 1965). Females do the same. Kummer (1968, p. 41) de-

scribed the attempt by adult female baboons to copulate with young males ‘behind

the backs of their leaders’ and Smuts (1987) presented photographic evidence of a

rhesus female checking to see if she and an extragroup male were being watched

before they commenced mating activities.

Whether primates ever undertake within-group evaluation from obscure posi-

tions remains to be demonstrated, but some males take measures that make this

unlikely. I refer to consortship, the practice whereby a male browbeats a female

into following his exodus from the group, for mating purposes, sometimes over

considerable distances and for extended periods of time (Goodall, 1986; McGinnis,



1979; Tutin, 1979). These bouts of absenteeism are of particular interest in rela-

tion to eavesdropping, since they suggest an awareness, on some level, that some

activities are better pursued in perceptual privacy.

As groups enlarge beyond some optimal size, any savings in predator vigilance

may be mitigated by new observational needs within the group. Other things being

equal, the larger the group the more competition there will be for food and other

resources (van Schaik et al., 1983). This increases conspecific threat and, with it,

alliances, which also must be visually monitored (Treves, 2000) and personally

serviced (Dunbar, 1993).

In red colobus and redtail monkeys, Treves (1999) found that if vigilance was

needed for external activity it came out of the time that would otherwise be

devoted to within-group looking. This, he speculated, might explain the fact that

in primates there has been little evidence for the hypothesis that looking time

decreases as group size increases (Pulliam, 1973; Elgar, 1989). For it is difficult to

see how total looking time could decrease if individuals are forced to keep an eye

on individuals in their own group.

Indeed, there is a tension between the time devoted to vigilance outside one’s

group and the time spent looking within it. In white-faced capuchin males, as

indicated above, external competition increases males’attention to outside males,

at the expense of internal vigilance (Rose & Fedigan, 1995), although presumably

reducing cuckoldry (Gould et al., 1997). In one study, redtail monkeys glanced at

associates more often when in the presence of red colobus monkeys than in purely

conspecific groups (Treves, 1999).

In various species, focus of attention is susceptible to rapid and dynamic shifts

from family and alliance members to strangers and predators. Some types of moni-

toring of the physical and social environment are carried out openly – even demon-

stratively – while other types may be effected with stealth. These shifts require a

dynamic model that recognizes the continuous interplay of multiple variables.

Indeed, the optimal paradigm would seem to be one that flexibly admits all types

of observation.

Toward a unified model

Is it possible to achieve a model that accounts for core principles asso-

ciated with vigilance, social observation and eavesdropping? In both human and

non-human primates, individuals appear to spend less time looking for predators

and competitors, and more time looking at each other, if group members are

nearby. In hierarchically organized societies – whether inherently complex squir-

rel monkey groups or highly politicized human societies – a great deal of internal

attention appears to be needed if individuals are to keep or to feel adequately

informed. In Machiavellian societies, inference would seem to play an exaggerated


434 J. L. Locke

role. In such societies, there may be greater need of observation, particularly of

the surreptitious kind.

It should be noted that when one looks at fellow group members, information

becomes available not only about them but also about the events to which they

may be reacting. In our own species, for example, when the young enter novel

situations they often monitor caregivers’ facial activity, which serves as a reliable

index of danger. Thus, internal observation offers both within-group information

needed for social comparison and extra-group information about predators and

competitors.

In both human and non-human primates, subordinate individuals spend more

time watching dominant individuals than the reverse arrangement. In Marivaux’s

plays, according to Trapnell (1987), keyholes enabled the young social climber ‘to

distinguish between his ally and his enemies’, to ‘observe the terrain on which he

must manoeuver, assess the efforts his ambition will require, determine the appro-

priate strategy and gauge his chances of success’. Were it not for eavesdropping,

Trapnell (1987, p. 109) wrote, the world would be ‘inaccessible and even unknown’

to people born without special advantages and privileges.

In Victorian England, as we saw above, the upper classes in many instances lived

beside the lower ones. This arrangement gave the ruling classes unobstructed vi-

sion of the individuals they wished to control, but it also gave the lower classes

a regular view of behaviours they had reason to emulate. In nineteenth century

America, upwardly mobile men and women had limited perceptual access to the

upper class behaviours they needed to absorb. To compensate, they used biogra-

phies as ‘handbooks’. They did so, according to Casper (1999), in the belief that

the difference between public success and failure lay in the private habits that

defined one’scharacter, or true self: the stuff of which compelling biographies are

made.

Earlier, we saw that, in non-human primates, males do more looking than fe-

males, presumably in an attempt to detect competition and danger. This vigilance

is very clearly tied to control and defence. In our own species, too, males seem to

have performed in a vigilant capacity more than females, boldly standing under

domestic eaves and then broadcasting the perceptual ‘take’.

Female networks are more extensive and stronger both in non-human primates

(Dunbar, 1988) and in our own species. There is evidence of a strong female ad-

vantage in human grooming (Sugawara, 1984, 1990) as well as touching (Jones &

Yarbrough, 1985) and concerted social action (Motz, 1983). There also is evidence

of a female preference for gossip – the use of speech to discuss mutual acquain-

tances not physically present – that spans cultures and most decades of the twen-

tieth century (Bischoping, 1993). There are indications, additionally, that when

peers offend young women in various cultures, the victims respond by working



indirectly through female friends (Bjorkqvist et al., 1994; Galen & Underwood,

1997; Crick & Bigbee, 1998).

We saw earlier that vigilance levels in non-human primates are affected by

transient factors such as mating and birth. In humans, too, vigilance levels are

clearly influenced by environmental change, including increases in population,

economic competition and terrorism. Surveillance cameras proliferated in the

USA after 11 September 2001. Although there had been a long-standing fear of

surveillant societies of the type envisaged in Orwell’s book 1984, the terrorist

attacks on New York and Washington seemed to have had the opposite effect. The

lack of objection noted by several newspaper columnists suggests that citizens may

have derived solace from the knowledge that government officials were looking

for and screening out ‘predators’.

In non-human primates, as we have seen, animals may devote as much as

half their waking hours to looking. Dunbar (1993) reported that animals spend as

much as 20% of their time grooming in some primate groups. There are indications

that crowding increases grooming (Nieuwenhuijsen & de Waal, 1982; Novak et al.,

1992; Judge & de Waal, 1997) and grooming has been found to decrease within-

group monitoring (Maestripieri, 1993; Hirsch, 2002). In future work, it would be

interesting to look at within-group vigilance and grooming in the same animals

as a function of density and predational threat.

Merely by comparing species, certain common patterns emerge, but more can

be done, beginning with the resolution of definitional issues. If interdisciplinary

collaboration facilitates the study of processes by which ‘senders’ use language to

donate information, as Hauser et al. (2002) have argued, it will surely affect the

investigation of processes by which ‘receivers’ of widely ranging communicative

abilities use their senses to extract it.

Acknowledgements

This chapter developed from a paper delivered to the Konrad Lorenz Institute in Al-

tenberg, Austria in December of 2001. Portions coevolved with a larger work in progress about

eavesdropping. The author wishes to acknowledge helpful comments by Adrian Treves, Eric

Salzen, Michael Studdert-Kennedy, Polly Wiessner and Ben Hirsch.

References

Abramovitch, R. 1976. The relation of attention and proximity to rank in preschool

children. In: The Social Structure of Attention, ed. M. R. A. Chance & R. R. Larsen.

London: John Wiley, pp. 153–176.


436 J. L. Locke

Aiello, J. R. 1972. A test of equilibrium theory: visual interaction in relation to

orientation, distance and sex of interactants. Psychonomic Science, 27, 335–336.

1977. Visual interaction at extended distances. Personality and Social Psychology

Bulletin, 3, 83–86.

Anthony, K. H. 1985. The shopping mall: a teenage hangout. Adolescence, 20, 307–312.

Bailey, F. G. 1971. Gifts and poison. In: Gifts and Poison: The Politics of Reputation, ed. F. G.

Bailey. Oxford: Basil Blackwell, pp. 1–25.

Baldellou, M. & Henzi, P. 1992. Vigilance, predator detection and the presence of

supernumerary males in vervet monkey troops. Animal Behaviour, 43, 451–461.

Bales, R. F., Strodtbeck, F. L., Mills, T. M. & Roseborough, M. E. 1951. Channels of

communication in small groups. American Sociological Review, 16, 461–468.

Barash, D. P. 1972. Human ethology: the snack-bar security syndrome. Psychological

Reports, 31, 577–578.

Barth, F. 1975. Ritual and Knowledge Among the Baktaman of New Guinea. New Haven, CT:

Yale University Press.

Baumeister, R. F. 1986. Preface. In: Public Self and Private Self, ed. R. F. Baumeister. New

York: Springer, pp. v–viii.

Beauchamp, G. 2001. Should vigilance always decrease with group size? Behavioral


Bentham, J. 1791. Panopticon: Postscript; Part I: Containing Further Particulars and Alterations

Relative to the Plan of Construction Originally Proposed; Principally Adapted to the Purpose

of a Panopticon Penitentiary-House. London: Printed for T. Payne, at the Mews-Gate.

Biben, M., Symmes, D. & Bernhards, D. 1989. Vigilance during play in squirrel

monkeys. American Journal of Primatology, 17, 41–49.

Bird, N. 1983. Conjugal units and single person: an analysis of the social system of the

Naiken of the Nilgiris (South India). D. Phil. Thesis, University of Cambridge.

Bird-David, N. 1994. Sociality and immediacy: or, past and present conversations on

bands. Man, 29, 583–603.

Bischoping, K. 1993. Gender differences in conversation topics, 1922–1990. Sex Roles,

28, 1–18.

Bjorkqvist, K., Osterman, K. & Lagerspetz, K. M. J. 1994. Sex differences in covert

aggression among adults. Aggressive Behaviour, 20, 27–33.

Blumenthal, A. 1932. Small Town Stuff. Chicago, IL: University of Chicago Press.

Buss, D. M. 1988. From vigilance to violence: tactics of mate retention in American

undergraduates. Ethology and Sociobiology, 9, 291–317.

1997. From vigilance to violence: mate retention tactics in married couples. Journal

of Personality and Social Psychology, 72, 346–361.

Caine, N. G. & Marra, S. L. 1988. Vigilance and social organization in two species of

primates. Animal Behaviour, 36, 897–904.

Campbell, J. K. 1964. Honour, Family and Patronage: A Study of Institutions and Moral Values

in a Greek Mountain Community. Oxford: Clarendon Press.

Carpenter, E. 1966. Image making in Arctic art. In: Sign, Image, Symbol, ed. G. Kepes.

New York: George Braziller, pp. 206–225.

Casper, S. E. 1999. Constructing American Lives: Biography and Culture in Nineteenth-Century

America. Chapel Hill, NC: University of North Carolina Press.



Castan, N. 1989b. The public and the private. In: A History of Private Life. III. Passions of the

Renaissance, ed. R. Chartier. Cambridge, MA: Harvard University Press, pp. 403–446.

Castan, Y. 1989a. Politics and private life. In: A History of Private Life. III. Passions of the

Renaissance, ed. R. Chartier. Cambridge, MA: Harvard University Press, pp. 21–67.

Chance, M. R. A. 1967. Attention-structure as the basis of primate rank orders. Man, 2,

503–518.

Chance, M. R. A. & Jolly, C. J. 1970. Social Groups of Monkeys, Apes and Men. London:

Jonathan Cape.

Cheney, D. L. & Seyfarth, R. M. 1981. Selective forces affecting predator alarm calls of

vervet monkeys. Behaviour, 76, 25–61.

Cowlishaw, G. 1998. The role of vigilance in the survival and reproductive strategies of

desert baboons. Behaviour, 135, 431–452.

Cranz, G. 1980. Women in urban parks. Signs: Journal of Women in Culture and Society, 5,

S79–S95.

Crawford, P. & Gowing, L. 2000. Women’s World in Seventeenth-Century England: A

Sourcebook. London: Routledge.

Crick, N. R. & Bigbee, M. A. 1998. Relational and overt forms of peer victimization: a

multiinformant approach. Journal of Consulting and Clinical Psychology, 66, 337–347.

Dixson, A. F., Scruton, D. & Herbert, J. 1975. Behaviour of the talapoin monkey

(Miopithecus talapoin) studied in groups in the laboratory. Journal of Zoology,

(London), 176, 177–210.

Draper, P. 1978. The learning environment for aggression and anti-social behavior

among the !Kung. In: Learning Non-Aggression: The Experience of Non-Literate Societies,

ed. M. F. A. Montagu. Oxford: Clarendon Press, pp. 31–53.

Dunbar, R. I. M. 1988. Primate Social Systems. Ithaka, NY: Cornell University Press.

1993. On the origins of language: a history of constraints and windows of

opportunity. Behavioral and Brain Sciences, 16, 720–735.

Eibl-Eibesfeldt, I. 1973. The expressive behaviour of the deaf- and -blind-born. In: Social

Communication and Movement: Studies of Interaction and Expression in Man and

Chimpanzee, ed. M. von Cranach & I. Vine. London: Academic Press, pp. 163–194.

Elgar, M. A. 1989. Predator vigilance and group size in mammals and birds: a critical

review of the empirical evidence. Biological Review, 64, 13–33.

Ellis, A. & Beattie, G. 1986. The Psychology of Language and Communication. New York:

Guilford Press.

Exline, R. V., Ellyson, S. L. & Long, B. 1975. Visual behaviour as an aspect of power role

relationships. In: Nonverbal Communication of Aggression 2, ed. P. Pliner, L. Krames &

T. Alloway. New York: Plenum press, pp. 21–52.

Feeley-Harnik, G. 1980. The Sakalava house (Madagascar). Anthropos, 75, 559–585.

Fejos, P. 1943. Viking Fund Publications in Anthropology, No. 1: Ethnography of the Yagua.

New York: Viking Fund.

Fisek, M. H. & Ofshe, R. 1970. The process of status evolution. Sociometry, 33, 327–346.

Fragaszy, D. M. 1990. Age and sex differences in the organization of behavior in

wedge-capped capuchins, Cebus olivaceus. Behavioral Ecology, 1, 81–94.

Galen, B. R. & Underwood, M. K. 1997. A developmental investigation of social

aggression among children. Developmental Psychology, 33, 589–600.


438 J. L. Locke

Gilbert, P. 1989. Human Nature and Suffering. London: Erlbaum.

Goffman, E. 1963. Behavior in Public Places. New York: Free Press.

Goodall, J. 1986. The Chimpanzees of Gombe: Patterns of Behavior. Cambridge, MA: Harvard

University Press.

Gould, L., Fedigan, L. M. & Rose, L. M. 1997. Why be vigilant? The case of the alpha

animal. International Journal of Primatology, 18, 401–414.

Gregor, T. 1970. Exposure and seclusion: a study of institutionalized isolation among

the Mehinacu Indians of Brazil. Ethnology, 9, 234–250.

Hall, D. M. 1998. The victims of stalking. In: The Psychology of Stalking: Clinical and

Forensic Perspectives, ed. J. R. Meloy. San Diego, CA: Academic Press, pp. 115–137.

Hall, E. T. 1966. The Hidden Dimension. New York: Doubleday.

Hall, J. A. 1996. Touch, status, and gender at professional meetings. Journal of Nonverbal


Hall, K. R. L. 1960. Social vigilance, behaviour of the chacma baboon, Papio ursinus.

Behaviour, 16, 261–294.

Hall, K. R. L. & De Vore, I. 1965. Baboon social behavior. In: Primate Behavior: Field Studies

of Monkeys and Apes, ed. I. De Vore. New York: Holt, Rinehart & Winston. pp. 53–110.

Hauser, M. 2000. Wild Minds: What Animals Really Think. London: Allen Lane.

Hauser, M., Chomsky, N. & Fitch, W. T. 2002. The faculty of language: what is it, who

has it, and how did it evolve? Science, 298, 1569–1579.

Haviland, L. K. & Haviland, J. B. 1983. Privacy in a Mexican Indian village. In: Public and

Private in Social Life, ed. S. I. Benn & G. F. Gaus. London: Croom Helm, pp. 341–361.

Hirsch, B. T. 2002. Social monitoring and vigilance behavior in brown capuchin

monkeys (Cebus apella). Behavioral Ecology and Sociobiology, 52, 458–464.

Horrocks, J. A. & Hunte, W. 1986. Sentinel behaviour in vervet monkeys: who sees

whom first. Animal Behaviour, 34, 1566–1567.

Jacobs, J. 1961. The Death and Life of Great American Cities. London: Penguin Books.

Jones, S. E. & Yarbrough, A. E. 1985. A naturalistic study of the meanings of touch.

Communication Monographs, 52, 19–56.

Judge, P. G. & de Waal, F. B. M. 1997. Rhesus monkey behaviour under diverse

population densities: coping with long-term crowding. Animal Behaviour, 54,

643–662.

Kalma, A. 1991. Hierarchisation and dominance assessment at first glance. European

Journal of Social Psychology, 21, 165–181.

Kelling, G. L. & Coles, C. M. 1996. Fixing Broken Windows: Restoring Order and Reducing

Crime in our Communities. New York: The Free Press.

Kendon, A. 1990. Conducting Interaction: Patterns of Behavior in Focused Encounters.

Cambridge: Cambridge University Press.

Keverne, E. B., Leonard, R. A., Scruton, D. M. & Young, S. K. 1978. Visual monitoring in

social groups of talapoin monkeys (Miopithecus talapoin). Animal Behaviour, 26,

933–944.

Koenig, A. 1994. Random scan, sentinels or sentinel system? A study in captive

common marmosets (Callithrix jacchus). In: Current Primatology, Vol. 11: Social

Development, Learning and Behaviour, ed. J. J. Roeder, B. Thierry, J. R. Anderson & N.

Herrenschmidt. Strasbourg: University Louis Pasteur, pp. 69–76.



Kummer, H. 1968. Social Organization of Hamadryas Baboons. Chicago, IL: University of

Chicago Press.

Lee, R. B. 1979. The Dobe !Kung. New York: Holt, Rinehart & Winston.

Le Roy Ladurie, E. 1978. Montaillou: Cathars and Catholics in a French Village. London:

Scolar Press.

Love, R. L. 1973. The fountains of urban life. Urban Life and Culture, 2, 161–209.

Maestripieri, D. 1993. Vigilance costs of allogrooming in macaque mothers. American

Naturalist, 141, 744–753.

Maples, W. R., Maples, M. K., Greenhood, W. F. & Walek, M. L. 1976. Adaptations of

crop-raiding baboons in Kenya. American Journal of Physical Anthropology, 45,

309–316.

McGinnis, P. R. 1979. Sexual behavior in free-living chimpanzees: consort

relationships. In: The Great Apes, ed. D. A. Hamburg & E. R. McCown. Menlo Park,

CA: Benjamin/Cummings, pp. 429–439.



McIntosh, M. K. 1998. Controlling Misbehavior in England, 1370–1600. Cambridge:

Cambridge University Press.

Metcalfe, N. B. 1984a. The effects of mixed-species flocking on the vigilance of

shorebirds: who do they trust? Animal Behaviour, 32, 986–993.

1984b. The effects of mixed-species flocking on the vigilance of shorebirds: is

visibility important? Animal Behaviour, 32, 981–995.

Moore, M. M. 1985. Nonverbal courtship patterns in women: context and

consequences. Ethology and Sociobiology, 6, 237–247.

Motz, M. F. 1983. True Sisterhood: Michigan Women and their Kin 1820–1920. Albany, NY:

State University of New York Press.

Mullen, P. E., Pathe, M. & Purcell, R. 2000. Stalkers and their Victims. Cambridge:


Nieuwenhuijsen, K. & de Waal, F. B. M. 1982. Effects of spatial crowding on social

behavior in a chimpanzee colony. Zoo Biology, 1, 5–28.

Novak, M. A., O’Neill, P. & Suomi, S. J. 1992. Adjustments and adaptations to indoor

and outdoor environments: continuity and change in young adult rhesus

monkeys. American Journal of Primatology, 28, 125–138.

Olsen, D. J. 1974. Victorian London: specialization, segregation, and privacy. Victorian

Studies, 17, 265–278.

Olsson, M. 2001. ‘Voyeurism’ prolongs copulation in the dragon lizard Ctenophorous

fordi. Behavioral Ecology and Sociobiology, 50, 378–381.

Oosterman, J. 1992. Welcome to the pleasure dome: play and entertainment in urban

public space – the example of the sidewalk cafe. Built Environment, 18, 155–164.

Pitcairn, T. K. 1976. Attention and social structure in M. fascicularis. In: The Social

Structure of Attention, ed. M. R. A. Chance & R. R. Larsen. London: John Wiley

pp. 51–81.

Poirier, F. E. 1972. The St Kitts green monkey (Cercopithecus aethiops sabaeus): ecology,

population dynamics, and selected behavioral traits. Folia primatologica, 17,

20–55.


440 J. L. Locke

Pulliam, H. R. 1973. On the advantages of flocking. Journal of Theoretical Biology, 38,

419–422.

Putnam, R. D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton,

NJ: Princeton University Press.

Quenette, P.-Y. 1990. Functions of vigilance behaviour in mammals: a review. Acta

Œcologica, 11, 801–818.

Rapoport, A. 1969. House Form and Culture. Englewood Cliffs, NJ: Prentice-Hall.

Rose, L. M. 1994. Benefits and costs of resident males to females in white-faced

capuchins (Cebus capucinus). American Journal of Primatology, 32, 235–248.

Rose, L. M. & Fedigan, L. M. 1995. Vigilance in white-faced capuchins, Cebus capucinus,

in Costa Rica. Animal Behaviour, 49, 63–70.

Schiefenhovel, W. 1997. Universals in interpersonal interactions. In: Nonverbal

Communication: Where Nature Meets Culture, ed. U. Segerstrale & P. Molnar. Mahwah,

NJ: Lawrence Erlbaum, pp. 61–79.

Shore, B. 1982. Sala ‘Ilua: A Samoan Mystery. New York: Columbia University Press.

Smith, W. J. 1977. The Behavior of Communicating: An Ethological Approach. Cambridge,

MA: Harvard University Press.

Smuts, B. B. 1987. Sexual competition and mate choice. In: Primate Societies, ed. B. B.

Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham & T. T. Struhsaker. Chicago,

IL: University of Chicago Press, pp. 385–309.

Soffer, O., Adovasio, J. M. & Hyland, D. C. 2000. The ‘Venus’ figurines: textiles, basketry,

gender, and status in the Upper Paleolithic. Current Anthropology, 41, 511–537.

Steenbeek, R., Piek, R. C., van Buul, M. & van Hooff, J. A. R. A. M. 1999. Vigilance in

wild Thomas langurs (Presbytis thomasi): the importance of infanticide risk.


Sugawara, K. 1984. Spatial proximity and bodily contact among the Central Kalahari

San. African Study Monographs, 3, 1–43.

1990. Interactional aspects of the body in co-presence: observations on the Central

Kalahari San. In: Senri Ethnological Studies 27: ‘Culture Embodied,’ ed. M. Moerman &

M. Nomura. Osaka, Japan: National Museum of Ethnology, pp. 79–122.

Tjaden, P. & Thoennes, N. 2000. Prevalence and consequences of male-to-female and

female-to-male intimate partner violence as measured by the National Violence

Against Women Survey. Violence Against Women, 6, 142–161.

Trapnell, W. H. 1987. Eavesdropping in Marivaux. Geneve: Librairie Droz.

Treves, A. 1998. The influence of group size and neighbors on vigilance in two species

of arboreal monkeys. Behaviour, 135, 453–481.

1999. Within-group vigilance in red colobus and redtail monkeys. American Journal of


2000. Theory and method in studies of vigilance and aggregation. Animal Behaviour,

60, 711–722.

Treves, A., Drescher, A. & Ingrisano, N. 2001. Vigilance and aggregation in black

howler monkeys (Alouatta pigra). Behavioral Ecology and Sociobiology, 50, 90–95.

Turner, T. S. 1980. The social skin. In: Not Work Alone: A Cross-Cultural View of Activities

Superfluous to Survival, ed. J. Cherfas & R. Lewin. London: Temple Smith,

pp. 112–140.



Tutin, C. E. G. 1979. Mating patterns and reproductive strategies in a community of

wild chimpanzees (Pan troglodytes schweinfurthii). Behavioral Ecology and Sociobiology,

6, 29–38.

Underwood, R. 1982. Vigilance behaviour in grazing African antelopes. Behaviour, 79,

81–107.

van Schaik, C. P., van Noordwijk, M. A., Warsono, B. & Sutriono, E. 1983. Party size and

early detection of predators in Sumatran forest primates. Primates, 24, 211–221.

Whiten, A. 1993. Social complexity: the roles of primates’ grooming and people’s

talking. Behavioral and Brain Sciences, 16, 719.

Wilson, P. J. 1988. The Domestication of the Human Species. New Haven, CT: Yale University

Press.

Wirtz, P. & Wawra, M. 1986. Vigilance and group size in Homo sapiens. Ethology, 71,

283–286.


442

P1: KPB/... P2: KOD0521823617agg.xml CU1917B/McGregor 0 521 582361 7 April 12, 2005 12:11

Part IV I N T E R F A C E S W I T H O T H E R

D I S C I P L I N E S

443


444


Introduction

Communication has a history of addressing topics of interest to other

disciplines, both in biology and more generally. The interface between disciplines

has been long recognized to generate paradigm shifts and the same has been

true of interfaces with communication. The interface between communication

and neurobiology provides a good example. The discovery that the brain nuclei

controlling song production varied in size seasonally (Nottebohm, 1981) was a

finding that overturned accepted notions of the stability of brain architecture and

triggered studies of evolutionary plasticity in brain structure (e.g. Sherry, 1998).

An important question then is whether the communication network approach

enhances communication’s interest to, and interfaces with, other disciplines.

This section shows that the answer is an emphatic yes, it does. In part, this is

shown by the wide range of topics addressed: from perception and physiology,

through aspects of cognition to the evolution of altruism. However, it is in the

details of the chapters that the value of the approach becomes apparent, as does

an enthusiasm about the further research possibilities.

Perception

The extent of a communication network is often an important issue and

is discussed by several chapters in this book. Network size is related to the distance

at which signals can be received and this distance is influenced by several factors.

These factors include the distorting and attenuating effects of the environment

through which the signal travels and the level of interference from the signals of

others. A key factor that is often overlooked is the sensory abilities of the receiver.

Such abilities can be extraordinary; for example, some bird species respond to a


445


446 Part IV

signal even when that signal is embedded in noise that is louder than the signal.

Ulli Langemann and Georg Klump (Ch. 20) discuss such perceptual abilities and

how they relate to aspects of signal structure and transmission in the acoustic

modality. Their chapter, therefore, covers the interface between communication

networks, psychophysics and physics.

Endocrinology

Hormones play an established role in determining when animals com-

municate, by controlling annual and circadian rhythms. They are also known to

be involved in mediating the response to signals; for example, raising the level

of oestradiol can induce female songbirds to perform copulation solicitation dis-

plays in response to song in the absence of the male singer (Searcy & Yasukawa,

1996). In Ch. 21, Rui Oliveira uses examples mainly from bony fishes to show that

androgens can be affected by the social environment as well as modulating be-

haviours that partly create the social environment. Establishing this reciprocal

link between hormones and behaviour (comparable to that for stress hormones)

has obvious consequences for both endocrinology and communication behaviour.

Cooperation and altruism

Altruism between unrelated individuals (that can include cooperation)

has long been considered an evolutionary puzzle. Recent mathematical models

and experiments with humans have shown that altruism can evolve through an

increase in the altruist’s ‘prestige’or ‘image’ in the eyes of others (e.g. Wedekind &

Milinski, 2000). Redouan Bshary and Arun D’Souza point out in Ch. 22 that these

recent advances in cooperation theory are a specific instance of a communication

network because others not directly involved in the altruistic interaction must

observe it. They then investigate the evolution and maintenance of altruistic be-

haviour, tactical deception and spiteful behaviour using data gathered in the field

from interactions between coral reef cleaner fish and their clients.

Semiochemicals

Information is obviously a key concept in communication, but as Brian

Wisenden and Norm Stacey make clear in Ch. 23, communication is a subset of

information. Aquatic animals can obtain information to guide reproductive and

predator-avoidance behaviour from chemicals released as a by-product of other

processes. Information in such semiochemicals can have striking effects; for ex-

ample chemicals released by predators attacking or digesting their prey can change


Interfaces with other disciplines 447

the behaviour, life history and morphology of potential prey (e.g. Brancelj et al.,

1996). After considering the use of chemical information by fishes in predator–

prey and sexual interactions, Wisenden and Stacey set out the case for communica-

tion networks as a subset of information networks and the potential evolutionary

routes for the origins of network behaviours such as eavesdropping. Their sug-

gestions include the possibility that a process analogous to eavesdropping can

precede the origin of communication, rather than such information gathering

following the development of communication interactions.

Cognition

The role of communication in cognition has ensured a lively interface

with the cognitive sciences, often concerning the extent to which non-human

communication can be considered a language (e.g. Hauser, 1996). Irene Pepper-

berg considers the cognitive abilities of birds in Ch. 24, particularly the issue of

transitive inference, using information from two different approaches. The first

approach uses communication to explore the cognitive abilities of parrots, with

human speech being used as the tool in much the same way as it is in explorations

of human cognitive abilities. The second approach uses the results of field ex-

periments investigating social eavesdropping to indicate the cognitive abilities of

territorial songbirds.

Several chapters in other sections of this book touch upon the cognitive abilities

of animals communicating in networks (e.g. the extent to which individual identi-

fication is a prerequisite of eavesdropping in Ch. 16). Dorothy Cheney and Robert

Seyfarth expand this theme in Ch. 25. They review the evidence for eavesdropping

in primates, concentrating on species living in large, permanent social groups,

often with complex social relationships. They then suggest a framework for assess-

ing the occurrence in other animal groups of social intelligence (i.e. mechanisms

such as transitive inference used to gather information relevant to social interre-

lationships). This framework, together with the information presented in other

chapters of this book, provides an opportunity for a taxonomically wide-ranging

comparative approach to the issue of social intelligence.

Mathematical models

Mathematical models have provided insights in many areas of biology,

including communication. However, most models of communication have not

dealt with networks, at least partly because of the difficulty in applying tractable

analytical models to networks. In Ch. 26, Andrew Terry and Rob Lachlan describe

models that capture two important aspects of networks by being spatially explicit


448 Part IV

and based on individual behaviour. These simulation models of anuran acoustic

choruses and eavesdropping strategies generate different results from more tra-

ditional models applied to the same questions. Incorporating strategic communi-

cation decisions into such spatially realistic individually based models seems an

approach that is likely to generate the kind of testable predictions about commu-

nication in networks that are needed to stimulate further research.

Applied aspects

Communication is central to animals’ lives and as such can be used to

modify their behaviour for human ends, such as in pest control, animal welfare and

conservation. A chapter jointly written with Tom Peake was planned for this book;

it would have explored the possible interfaces between communication networks

and aspects of applied biology. However, the relative newness of the network

perspective means that there are, as yet, few concrete examples. We decided,

therefore, to outline here some of the interfaces that we think are promising

and to draw attention to chapters elsewhere in the book that have mentioned

applications.

One way to judge the welfare of animals is to assess the extent to which ani-

mals in captivity, including those on public display in zoos and aquaria, are able to

display the full range of behaviours shown by free-ranging animals under natural

conditions. Many of the chapters in this book have argued that communication in

a network is such a natural condition and, therefore, the ability to communicate as

part of a network could be regarded as a feature of adequate captive provision. Ad-

verse effects on breeding performance have been noted when group-living species

are kept in small groups, and communication networks that are much smaller

than occur in the wild could underlie this effect. Attempts to increase apparent

group size (e.g. use of mirrors with captive flamingos (Whitfield, 2002)) have met

with mixed success, perhaps because the manipulations did not adequately mimic

communication networks. It may be easier to create an apparent communication

network for species that are widely spaced and possibly territorial, because at long

range it is only signals that are detected and signal playback is straightforward, at

least with acoustic signals. Tom Peake has suggested that interactive acoustic sig-

nals could be provided as a type of environmental enrichment in communication

for captive animals. For example, zoo visitors could interact with vocal species

such as gibbons Hylobates spp. via playback from remotely sited loudspeakers trig-

gered from a control panel in the cage’s viewing area (obviously there would have

to be safeguards to ensure that the nature and extent of interaction did not exceed

natural levels). Communication in a network may be equally difficult if a popula-

tion is held at an unnaturally high density for production reasons (e.g. fish farms),


Interfaces with other disciplines 449

because the network is overloaded. From the discussions above, it can be expected

that removing the ability to signal will have adverse effects on welfare (e.g. removal

of rodent scent marks by cage cleaning (Gray & Hurst, 1995)). Signal removal can

also have additional network-wide effects. Removing the major chela of male fid-

dler crabs Uca tangeri clearly prevents individual males from signalling visually in

their usual manner. In addition, it may skew the sex ratio apparent to males and

females because males without a major chela appear female to both sexes, and the

response to the apparent sex ratio could accelerate population declines (Oliveira

et al., 2000).

There are several established applications of communication in conservation,

such as identifying individuals from features of their vocalizations (e.g. McGregor

et al., 2000). The network perspective emphasizes the possibility that anthro-

pogenic noise could have adverse effects by disrupting or restricting the size of

acoustic communication networks (e.g. McGregor & Dabelsteen, 1996). Vincent

Janik considers this applied aspect in more detail in relation to marine mammals

in Ch. 18. In the terrestrial environment, road noise may similarly restrict the

acoustic communication networks of other taxa such as songbirds. Habitat frag-

mentation is another way of disrupting songbird communication networks and

it is discussed by Ken Otter and Laurene Ratcliffe in Ch. 7. In Ch. 8, Alexandra

Lang and colleagues discuss how the need to use signals with limited range and

detectability to predators may have the side effect of reducing the effective popula-

tion size of katydids, with associated increased susceptibility to random extinction

processes.

We think that these brief examples illustrate that communication networks can

have relevance to applied biology and that often the implications are not straight-

forward. We suggest that those researching communication networks have an

obligation to explore the applications of their findings. Arguably the best way to

make applied biologists aware of the relevance of network research is to make

suggestions on how best to modify current practice to incorporate new findings.

Future directions

The chapters in this section clearly show how several areas of research

interface with communication networks. An obvious question is whether this will

also be true for interfaces with disciplines that are not represented in this section. A

number of chapters (e.g. Chs. 12, 14 and 26) mention the possibility of fruitful links

with ecology, or more specifically spatial ecology, suggesting that this would seem

to be a good interface to explore. Also, given that we humans consider ourselves

to be supreme communicators and often do so in a network environment, it is

possible that many aspects of social psychology and sociology could interface


450 Part IV

fruitfully with a communication network approach. These brief considerations

suggest that the answer to the question posed above is that interfaces between

communication networks and many other disciplines can be sources of insight

and inspiration in the future.

References

Brancelj, A., Celhar, T. & Sisko, M. 1996. Four different head shapes in Daphnia hyalina

(Leydig) induced by the presence of larvae of Chaoborus flavicans (Meigen).

Hydrobiologia, 339, 37–45.

Gray, S. & Hurst, J. L. 1995. The effects of cage cleaning on aggression within groups of

male laboratory mice. Animal Behaviour, 49, 821–826.


McGregor, P. K. & Dabelsteen, T. 1996. Communication networks. In Ecology and



McGregor, P. K., Peake, T. M. & Gilbert, G. 2000. Communication behaviour and

conservation. In: Behaviour and Conservation, ed. L. M. Gosling & W. J. Sutherland.


Nottebohm, F. 1981. A brain for all seasons: cyclical anatomical changes in song

control nuclei of the canary brain. Science, 214, 1368–1370.

Oliveira, R. F., Machado, J. L., Jordao, J. M. et al. 2000. Human exploitation of male

fiddler crab claws: behavioural consequences and implications for conservation.

Animal Conservation, 3, 1–5.




Sherry, D. F. 1998. The ecology and neurobiology of spatial memory. In: Cognitive

Ecology: The Evolutionary Ecology of Information Processing and Decision Making, ed. R.

Dukas. Chicago, IL: Chicago University Press, pp. 261–296.


Science, 288, 850–852.

Whitfield, J. 2002. Mirrors to help birds mate. Nature Science Update, 19 March:

http://www.nature.com/news/2002/020318/full/020318-2.html.

P1: KPB/... P2: KOD0521823617c20.xml CU1917B/McGregor 0 521 582361 7 April 12, 2005 12:27

20

Perception and acousticcommunication networks

u l r i k e l a ng e m a n n & g e o rg m . k l u m p

University of Oldenburg, Germany

Introduction

Traditionally, the analysis of acoustic communication has been based on

a model system composed of a sender, the transmission channel and a receiver

(Shannon & Weaver, 1949). Since the early 1990s, this view has been extended to

communication networks, in which several signallers and receivers are involved

(e.g. McGregor & Peake, 2000). Two general approaches have been adopted in order

to investigate communication behaviour. First, measurements of physical modifi-

cations to the signal during transmission (e.g. Wiley & Richards, 1978; Dabelsteen

et al., 1993; Holland et al., 1998) have been used to assess the feasibility of com-

munication (e.g. estimating maximum communication distances) or to evaluate

which features of signals might be adaptive in a certain context. Second, playback

studies have been used to conclude which features may be of importance for signal

discrimination: different behavioural responses can be elicited by playback of sig-

nals that have been modified by physical properties of the environment or by the

experimenter. Often the physical properties of signals are manipulated in ways

that are informed by studies of signal transmission in the animal’s environment.

However, behavioural responses can be understood more fully if the animal’s per-

ceptual abilities are taken into account (Wiley & Richards, 1982; Klump, 1996).

Perception includes the transduction process by the animal’s sensory organs and

the subsequent processing by the nervous system. However, perception can only

be inferred indirectly from the animal’s responses. An animal’s failure to respond

differentially to playback can either mean that the animal was not motivated

to discriminate within the experimental context (e.g. because the modified sig-

nal deviated too far from species-specific signals) or that the animal’s auditory


451


452 U. Langemann & G. M. Klump

system could not ‘resolve’ modifications of the signal and, therefore, they were

not perceived. In the latter case, perceptual constraints render a behavioural re-

sponse impossible. If more than one signal parameter was modified in playback

experiments, the results might be even more difficult to interpret. Studies in the

laboratory focusing on perceptual mechanisms allow us to control for motivation

of an animal, help to conclude which modifications can be exploited by the ani-

mal and make it possible to determine the perceptual resolution of the animal’s

sensory system.

In this chapter, we will explain how the current knowledge on perceptual

mechanisms offers a better understanding of animal communication, especially

in the context of a communication network. For example, new results from per-

ceptual studies now give us a more complete and accurate picture of how animals

parse signals from noise and from different sources in acoustic scenes (e.g. Feng &

Ratnam, 2000; Hulse, 2002). We will demonstrate this by showing how knowledge

on perceptual masking can aid in explaining signal assessment by the animal. An

understanding of how animals segregate several sources of signals in a commu-

nication network requires knowledge of how accurately signals are localized and

how the spatial arrangement of sources affects masking (e.g. Klump, 2000). Finally,

the behaviour of an animal in a communication network often requires an abil-

ity to range a signal, that is, assess the distance of a sound source (e.g. Naguib &

Wiley, 2001). Explaining results from playback studies of ranging requires an un-

derstanding of the perception of degradation. Providing a comprehensive review

of all the topics mentioned here is beyond the scope of this chapter. Instead, we

will present specific examples from the animal behaviour literature and discuss

them in light of knowledge of the physiology of perception.

Detection and recognition

Spectral aspects of masking and signal detection

Communication in any context requires signal detection. This would

hardly be a problem in a silent world. However, the environment is noisy, perhaps

particularly so in a communication network. Therefore, signal detection needs

to be considered in relation to the level of the background noise. Environmen-

tal noise originates from biotic sources such as calling insects (e.g. grasshoppers

and cicadas (Waser & Waser, 1977; Ryan & Brenowitz, 1985)) and calling frogs

(Wollerman & Wiley, 2002) or singing birds (especially during the dawn chorus,

e.g. Staicer et al. (1996)). Biotic environmental noise constitutes an especially severe

problem in large assemblies of individuals of the same species (e.g. roosts, breed-

ing colonies, choruses) since the masking noise matches the frequency spectrum

of the signal (as the signals of conspecifics constitute most of the masking noise


Perception and acoustic communication networks 453

experienced by an individual). There are other biotic sources that may contribute

to background noise. For example, the rustling of the foliage and the movement

of twigs and branches in a deciduous wood provide a substantial level of back-

ground noise (e.g. Klump, 1996) that increases with wind speed (Fegeant, 1999).

A similar increase in background noise with wind speed can be observed in open

grassland habitats. For example, wind with a moderate speed of 5 m/s produces

sound pressure levels of more than 60 dB in the one-third-octave band at 20 Hz for

at least 95% of the time (Boersma, 1997). Much lower levels of background noise are

observed at frequencies above 500 Hz. Coastal environments and running waters

may produce a substantial amount of non-biotic noise (Dubois & Martens, 1984;

Douglas & Conner, 1999).

The difference between the frequency spectrum of the signal and the back-

ground noise and their relative amplitudes determine the amount of masking.

The amplitude of sounds can either be specified in terms of their overall sound-

pressure level or in terms of the amplitude contained in the individual compo-

nents. Commonly, the components are resolved into frequency bands 1 Hz wide

and the sound pressure level (which is a measure of the amplitude relative to the

standard reference pressure of 2 × 10−5 Pa) in each 1 Hz band is determined. This

amplitude measure is conventionally called the spectral density or the spectrum

level (e.g. Moore, 2003). The difference between the signal amplitude and the am-

plitude of the background can be described by the signal-to-noise (S/N) ratio. The

S/N ratio can be either expressed as the difference between the overall signal level

and the overall level of the noise or as the difference between the signal spectral

density and the spectral density of the noise. The S/N ratio allows us to estimate

whether a signal can be detected or not (see Box 20.1). The same rules for signal

detection apply for single receivers as well as for individuals in a communication

network. Each ‘node’ in a network (a receiver sitting at a different place), however,

might experience quite different S/N ratios for the same signal. Behavioural ex-

periments in the laboratory can determine how random background noise affects

absolute auditory sensitivity for tonal signals or signals with a distinctive peak in

the spectrum.

The value that denotes the shift in auditory sensitivity when random (wide-

band) noise is present is called the critical masking ratio or critical ratio (CR).

The CR is simply the S/N ratio at detection threshold in random wideband noise

expressed as the difference between the level of a tonal signal (which is identical

to its spectral density) and the spectral density of the noise (N0). The CR is usu-

ally independent of the level of random background noise. However, the CR is

frequency dependent, increasing at about 3 dB per octave (overview in Fay, 1988).

The CR also provides a rough estimate of the bandwidth of auditory analysis filters

(CR filter bandwidth in Hz is given by 10CR/10; e.g. Yost, 1994). There are numerous



Box 20.1 Determining the signal-to-noise (S/N) ratio forsignal detection

Calculating the S/N ratio for signal detection in animal studies is relatively

straightforward in the laboratory environment with suitable equipment and

all parameters well controlled. Field measurements are more difficult.

Nevertheless, as shown here, it is possible to get an estimate of the S/N ratio

for signal detection in the natural environment.

Which equipment to use

Preferably, a sound-level meter should be used in the field to

determine signal level directly. Often it is possible to record sounds through

the microphone of the sound-level meter (e.g. on a DAT recorder or directly

with a note-book computer) for subsequent analysis, but any microphone

and recording equipment that allows signals to be recorded with an

accurately defined gain is suitable. The directional characteristics of the

microphone should be adapted to the question of interest. For example, an

omnidirectional microphone is the best choice when the general level of

background noise is being measured. If the goal is to measure a signal

originating from a specific source, it is recommended to use a directional

microphone and approach the source as close as possible. The

frequency-transfer function of the microphone should be as flat as possible

to avoid a bias in the later analysis. Recording a calibration signal of known

sound-pressure level with the signal of interest makes it subsequently

possible to compute the absolute power spectra of the recorded signals (the

calibration signal can be recorded at 1 m from the microphone, at the

location from which the animal was recorded or by using a calibrator placed

on the microphone). Even inexpensive sound-level meters can be used for

accurate field measurements if they are calibrated against a high-quality

instrument in the laboratory. Spectral analysis can be carried out with

specialized spectrum analyser hardware or (if the signals are digitized and

stored on a computer) with suitable software, e.g. freeware.

Computing signal-to-noise ratios

First, one needs to decide which of the two common measures of the

S/N ratio should be used: comparing overall sound-pressure levels or

comparing the sound-pressure levels with reference to spectral density (i.e.

compute the ratio between signal and noise spectrum level N0; for further

details, see text). The frequency spectrum and the sound-pressure level of



both the signal of interest and the background noise have to be known in

order to determine the S/N ratio correctly.

Step 1: measuring sound-pressure levels of signal and background noise

It is important that the sound-pressure level of the signal of interest

is at least 10 dB above the sound pressure level of other sound making up the

background noise. If the difference is less than 10 dB, the sound-pressure

level of the signal cannot be measured independently from the

sound-pressure level of the background noise and, therefore, a S/N ratio

cannot be computed accurately (the ratio would then be the level of signal

plus noise divided by the level of the noise). Strategies to achieve the 10 dB

minimum difference include getting close enough to the signal source,

getting as far away as possible from sources of noise or using a highly

directional microphone. The distance between the microphone and the

sound source(s) should be reported. This allows an estimate of the signal

level at a specified distance from the source (e.g. by using the rule of thumb

that the signal level is reduced by 6 dB for every doubling of the distance

and, if more accurate estimates are required, the estimates should include

effects of excess attenuation).

Sound-level meters frequently offer at least two types of filter setting: the

A and C settings. The A filter has a low-frequency cut-off of about 800 Hz, a

high-frequency cut-off of about 9 kHz and emphasizes the intermediate

frequencies in this range. The C filter has a low-frequency cut-off of

approximately 30 Hz, a high-frequency cut-off of approximately 8 kHz and

has a flat unbiased frequency response. To be sure that an appropriate filter

is used, the signals of interest must fall within the frequency range of the

filter. For example, measuring ambient background noise over a wide

frequency range is only possible with a C filter setting. The integration time

constant of the sound-level meter ideally should match the integration time

of the auditory system of the study species or at least the duration of its

signals. The ‘fast’ integration time is suitable for measuring the

sound-pressure level of animal signals of 125 milliseconds or longer

duration. Using a very long integration time (1 second at the ‘slow’ setting of

the sound-level meter) will underestimate the level of signals that are

composed of brief components. If shorter signals than 125 milliseconds need

to be analysed, the ‘impulse’ or ‘peak’ settings are more suitable. These

provide integration times of 35 and 0.05 milliseconds, respectively, for

fast-rising signal levels and a very slow decay. The integration times used to

determine root mean square sound pressure with a sound-level meter allow

only approximations of the real sound pressure (that can be calculated from



the calibrated digitized signal) or the perceived sound pressure (for which

one should apply the integration time of the animal’s auditory system).

When digitizing the recordings for further analysis with a computer, the

sampling rate must be at least twice the highest frequency of the signals to

avoid serious sampling errors.

Step 2: calculating signal-to-noise ratios

If the signal and the background noise have the same frequency

spectra (e.g. frog calls in a dense chorus of conspecifics) the S/N ratio is

simply the difference between the overall signal level and the overall level of

the background noise, both measured in decibels. If the signal and the

background noise have different bandwidths and thus different frequency

spectra (e.g. a tonal bird vocalization in a wideband background noise), a S/N

ratio based on measurements of the spectrum level is easier to interpret. The

spectrum level can be calculated from the overall level by subtracting the

bandwidth (in decibels) of the frequency filter used in the measurement

from the overall sound-pressure level measured through this filter, i.e.

spectrum level (dB) = overall level (dB) − 10 × log10(bandwidth)

If the signal has a smaller bandwidth than the frequency filter throughwhich its sound-pressure level was determined, the bandwidth of the signal

must be used instead of the bandwidth of the filter. Then the S/N ratio is

calculated as the difference between the spectrum level of the signal and of

the background noise measured in decibels. It must be noted that this

engineering-type measure of the S/N ratio provides only an approximation of

stimulus characteristics relevant for perception. The filters relevant for the

perception of the animal are the auditory analysis filters. Furthermore, the

measure of the S/N ratio as calculated here does not incorporate temporal

aspects of the structure of the signal and background noise. The physiology

of the auditory system will determine how temporal aspects affect the

perceived S/N ratio. If one wants to know more accurately how the acoustic

environment is perceived by an animal, the measurements should be

interpreted using a model of the physiology of the animal’s auditory system.

studies determining the CR values at various frequencies of the hearing range in

fish, amphibians, birds and mammals (summarized by Fay, 1988). The average CR

in birds is about 23 dB at 1 kHz, 24.5 dB at 2 kHz, 27.5 dB at 4 kHz and 36 dB at 8 kHz

(Klump, 1996). Typical CR values in mammals are 22.5 dB at 1 kHz, 25.5 dB at 2 kHz,

29.5 dB at 4 kHz and 32 dB at 8 kHz (average for cat, rat, chinchilla and human,

respectively (Fay, 1988)). The average slope of 3 dB per octave that is often described

probably does not hold for frequencies below 500 Hz and the CR will decrease much



less for such frequencies (e.g. Moore & Glasberg, 1983). There are some exceptions

from the usual pattern of an increasing values of CR with increasing frequency.

In mammals, the greater horseshoe bat Rhinolophus ferrumequinum has the lowest

CR values at the frequencies of their ultrasonic echolocation calls (Long, 1977).

In birds, great tits Parus major show relatively little change in CR with frequency.

At high frequencies, CR values for great tits are much lower than those of other

bird species (e.g. 25.9 dB at 8 kHz). This may be an adaptation that makes the high-

pitched communication sounds of great tits much less susceptible to masking

by environmental noise in their deciduous forest habitat (Klump & Curio, 1983;

Langemann et al., 1998). Signal-detection thresholds in background noise, called

masked thresholds, thus depend on the level of background noise and on the

CR. Estimates of masked auditory thresholds (MAT) are commonly calculated by

adding the CR to the spectrum level of the noise (N0 + CR = MAT). Knowing both

CR and the level of the background noise allows us to estimate approximately how

background noise in a specific communication setting will influence the auditory

sensitivity of animals.

We will use an example to demonstrate how such calculations can estimate

the distance over which communication is possible: that is, the extent of the

communication network (see also Box 20.1). The detection distance for a typical

great tit song element with a spectral peak frequency of 2 kHz and signal amplitude

of 90 dB can be estimated to be 331 m (Table 20.1). This estimate takes into account

the great tit’s absolute threshold at 2 kHz (6.5 dB), spherical spreading of the song

(6 dB per doubling of distance for all frequencies) and the habitat-dependent signal

attenuation (excess attenuation, e.g. Marten & Marler, 1977; Dabelsteen et al., 1993).

However, the main acoustic energy of background noise in a deciduous forest

occurs at lower frequencies (Klump, 1996; Fegeant, 1999) and the 2 kHz signal

will be masked by background noise of 10 dB, and the perceptual threshold of a

great tit receiver in noise (masked auditory threshold) will be much worse than

6.5 dB: in fact the masked threshold of a 2 kHz signal would be 35.6 dB (N0+ CR, i.e.

10 + 25.6 dB; Table 20.1). Hence, a great tit can only perceive song elements of 2 kHz

as long as the sound level does not drop below the great tit’s masked threshold,

giving a maximum detection distance of 124 m. At this distance, the sound level

of the element equals the value of the masked threshold (35.6 dB). In contrast, a

‘seeet’ alarm call indicating the presence of an aerial predator (peak frequency

around 8 kHz, mean amplitude 60.1 dB) experiences much less masking by the

forest background noise. At 8 kHz, the background noise spectrum level is −5.2 dB

and since great tits’CR at 8 kHz is similar to the CR at 2 kHz, the masked threshold

is very close to great tits’ absolute threshold. In consequence, the perception of

this alarm call is mainly limited by great tits’ absolute auditory sensitivity.

Infrasound communication in elephants is assumed to be a communica-

tion network extending over several kilometres (Larom et al., 1997; Ch. 17).



Table 20.1. An example of estimates of masked thresholds (i.e. the signal-detection

threshold in background noise) and perceptual distances (i.e. the maximum distance where

behavioural responses would be expected) of single great tit song elements with spectral

peak frequencies between 2 and 8 kHz or the aerial predator ‘seeet’ alarm call

(of approximately 8 kHz)

Signal frequency (kHz)

2 4 6.3 8 ‘Seeet’ call

Spherical spreading (dB/dd) 6 6 6 6 6

Excess attenuation (dB/100 m) 10.0 13.7 18.0 21.1 21.1

Source level (dB) 90 90 90 90 60.1

Absolute hearing threshold (dB) 6.5 9.1 12.8 18.1 18.1

How far in quiet (m) 331 242 179 137 56

Noise level N0 (dB) 10.0 4.3 −1.7 −5.2 −5.2

Critical ratio (dB) 25.6 23.8 25.9 25.9 25.9

Masked auditory threshold (dB) 35.6 28.1 24.2 20.7 20.7

Detection distance in noise (m) 124 138 130 128 49

Recognition threshold (dB) 38.6 31.1 27.2 23.7 23.7

Perceptual distance in noise (m) 107 124 118 117 41

dd, doubling of distance; CR, critical ratio;

Signal detection is a function of physical signal properties (signal frequency, source level), of

the environment (background noise level N0, intensity loss from excess attenuation), of physics

(intensity loss from spherical spreading) and of physiological constraints set by the auditory

system. The critical ratio denotes the shift in auditory sensitivity from the absolute hearing

threshold (in quiet) to the masked auditory threshold when random wideband noise is present.

For calculating perceptual distances, random noise is used as an approximation of the ambient

background noise. Note that the background noise level N0 is expressed with reference to spectral

density of the noise in order to calculate masked thresholds (N0 + CR) Spectral density is the sound-

pressure level of each 1 Hz wide frequency component relative to the standard reference pressure

of 2 × 10−5 Pa). Detection is only the first step in perception; the second step is recognition (or

discrimination), which is also influenced by noise. Recognition thresholds may be estimated to be

an additional 3 dB higher in signal-to-noise ratio than detection thresholds. Values for thresholds

in great tits are from Langemann et al. (1998), noise level from Klump (1996), source level of

‘seeet’ calls from Klump & Shalter (1984). An excess attenuation of 10 dB/100 m was assumed for

2 kHz; the excess attenuation above 2 kHz was increased by 1.85 dB/100 m for every additional

1 kHz. The amount (A) to which the original sound pressure level of the source (S) can drop to

be just detectable in quiet (i.e. to the absolute threshold AAT), in noise (the masked auditory

threshold AMAT) or recognized in noise (the recognition threshold ART) is one variable used for

estimating the perceptual distance (m). The other components include the amplitude decrease

from spherical spreading (20 log10 m) and the decrease from excess attenuation (EA/100 m). Solving

the following equation for m yields the perceptual distance (see also Marten & Marler, 1977):

A(dB) = S (dB) − 20 log10 m − m(E A/100).



Communication distances of nearly 10 km are obtained by assuming that absolute

thresholds are limiting and that wind speeds are low at night. However, during the

day, wind speed is much higher and it is likely that wind-induced noise provides

sufficient masking to reduce communication distances. Using measurements of

the sound-pressure level of wind-induced noise in grassland (Boersma, 1997) and

assuming a CR of 10 dB (low-frequency CR values can only be extrapolated from

studies in humans, e.g. Moore & Glasberg, 1983) gives a masked threshold of 73 dB.

This is much higher than the absolute threshold of 50 dB that was used in the

calculation by Larom et al. (1997) and suggests that infrasound communication

networks may be less extensive than previously thought.

Some caveats must be borne in mind when estimating detection distances

or transmission distances from CR values. First, CR is measured by presenting a

narrowband signal in wideband masking noise. If the frequency spectrum of the

signal and the noise are rather similar (e.g. detection of an individual calling frog

in the masking noise provided by a chorus of thousands of conspecific frogs), then

the masked threshold calculated from the CR will be overestimated. With signals

of similar frequency spectra, the S/N ratio is probably close to, or even below, 0 dB,

i.e. signals can be detected when their level is equal to, or even below, the level of

the background noise. In this case, the task resembles an increment detection in

overall sound amplitude when the signal is added to the background noise (e.g.

Miller, 1947). Second, the CR may not provide a good estimate of masked thresholds

when the temporal structure of the background noise has very pronounced slow

envelope fluctuations (see below). In this case, masked thresholds may be up to

20 dB more sensitive than would be expected from the CR. Third, it should be

remembered that communication signals are often broadcast repeatedly, or at

least some signal elements are repeated, whereas in laboratory studies the test

signals for detection will often be presented only once. Detection sensitivity is

known to improve by the square root of the number of independent observations

(e.g. Swets et al., 1950), suggesting that repetitive signals may be detected more

readily.

Separating sounds by exploiting temporal patterns

Environmental background noise will usually not resemble the random

noise with a steady-state envelope that is often employed in the laboratory as

a masker. Animals in communication networks appear to be adapted to exploit

temporal envelope fluctuations in the background noise that are typical of the

natural environment (Klump, 1996; Nelken et al., 1999). Signallers have often been

observed to call or sing during periods of reduced amplitude of the background

noise. For example, many frog species call when the nearest signalling neighbours

are silent (Klump & Gerhardt, 1992; Ch. 13). Some frog species go even further and



(a)

(b)

(c)

Time (s)

Fig. 20.1. Waveforms illustrating different noise envelopes. (a) Unmodulated

wideband random noise exhibiting a Gaussian distribution of amplitude values. This

type of noise exhibits little variation in the temporal envelope. Unmodulated

wideband noise is normally used in determining critical masking ratios.

(b) Coherently amplitude modulated noise. Note the slow envelope fluctuations. This

noise was synthesized by multiplying random wideband noise by a low-passed noise

with a bandwidth of 12.5 Hz. It has the same overall bandwidth as the unmodulated

noise in (a). This type of noise resembles more closely the structure of environmental

noise. (c) A dawn chorus recorded in a European deciduous forest. It has a waveform

with pronounced slow amplitude fluctuations that are more similar to the

fluctuations in the envelope of coherently modulated noise (b) than of unmodulated

noise (a).

will use call timing strategies to mask their neighbours’ calling (e.g. Gerhardt &

Huber, 2002). Receivers may also benefit from temporal patterns in signals and

in background noise because the separation of sounds from different sources is

improved if their amplitude patterns differ considerably. The separation of sounds

originating from different sources into different ‘auditory streams’ is also known

as sound segregation (Bregman, 1990). The ‘unmasking’ effect associated with

sound segregation is well documented in laboratory studies with humans and

other animals (e.g. Moore, 1990; Klump & Langemann, 1995; Nelken et al., 1999;

Klump & Nieder, 2001; Pressnitzer et al., 2001). It should be noted that unmasking



Table 20.2. Masking release observed in the laboratory for detection of pure tone signalsa

Species Masking Amplitude Study

release (dB) factor

Human, Homo sapiens 12 4 e.g. Schooneveldt & Moore, 1989

European starling, Sturnus

vulgaris

12 4 Klump & Langemann, 1995

Gerbil, Meriones unguiculatus 17 7 Klump et al., 2001

Chinchilla, Chinchilla laniger 6 2 Niemiec, 2001; A. J. Niemieac,

personal communication

Cat, Felis catus 5 2 Budelis et al., 2002; b B. J. May,

personal communication

aMasking release is the masked threshold in wide-band coherently amplitude-modulated back-

ground noise compared with the masked threshold in unmodulated background noise of the

same overall signal energy and bandwidth. The ‘gain’ in signal detection can be expressed either

in decibels or as an amplitude factor (10dB/20). Similar masker envelopes were used in all species,

i.e. the noise bands had dominant frequencies of envelope fluctuations below 50 Hz.bTaking into account only the masking release across auditory analysis filters.

does not mean that masking is absent, rather it means that a partial release from

masking can be observed.

For example, this unmasking effect is shown by the ability of European starlings

Sturnus vulgaris to detect a tone (i.e. a signal of a particular frequency) in noise

(e.g. Klump & Langemann, 1995). Their tone-detection threshold in the type of

wideband random noise commonly used in studies of the CR (Fig. 20.1a) is up to

20 dB worse than in noise of the same overall signal energy and bandwidth that

has been coherently amplitude modulated (Fig. 20.1b) to resemble more closely

typical environmental noise (e.g. bird song dawn chorus; Fig. 20.1c).

The amount of masking release resulting from the modulation of noise de-

pends on the bandwidth of the masking noise: masking release decreases with

decreasing bandwidth of the masking noise. For bandwidths of the size of an au-

ditory analysis filter (about 10–20% of the centre frequency), the masking release

is reduced. For example, in the starling, the masking release is reduced to about

5 dB if 2 kHz signals have to be detected in 200 Hz wide noise centred at the sig-

nal frequency (Klump et al., 1998). Slow rates of envelope fluctuation result in a

larger masking release than fast rates of fluctuation (Klump & Langemann, 1995).

Masking release in amplitude-modulated background noise has been observed in

all species studied so far, although to a different extent (Table 20.2). In humans, an

unmasking effect is also observed if the signal and the masking background noise

are both amplitude-modulated noise bands (no other species has been tested with



this paradigm). If the noise-band signal has a pattern of envelope fluctuation that

is different from the masking noise bands, the detection of the noise band sig-

nal is improved in comparison with stimuli that have similar correlated envelope

fluctuations in both signal and masking bands. For signals with durations of more

than 200 milliseconds, detection differences of about 8 dB have been observed (e.g.

McFadden & Wright, 1990).

The unmasking effects described in the previous paragraph are likely to be

relevant in animal communication systems (e.g. Klump, 1996). Signals that are

broadcast from a sender will often be amplitude modulated during transmission

(Richards & Wiley, 1980). The same signal travelling along different paths will ex-

perience different modulation patterns since the modulation results from local

turbulence in the atmosphere (arising from temperature gradients or wind). This

applies also for signals from different sources that travel along different paths

to a receiver. Thus, receivers should be able to exploit the different modulation

patterns imposed on signals by the natural environment to gain an advantage in

signal detection. In addition, signallers themselves create amplitude modulation

patterns when broadcasting calls or songs. For example, king penguins Aptenodytes

patagonicus appear to be able to utilize amplitude modulations in their calls to im-

prove their sensitivity when searching for their mate or chick in an assembly of

hundreds of individuals (e.g. Aubin & Jouventin, 1998). The contact calls of the

emperor penguin Aptenodytes forsteri and of the king penguin exhibit a syllable

structure with pronounced amplitude modulations, and their two-voice mode of

call generation creates amplitude beats (Aubin et al., 2000; Lengagne et al., 2001).

Both species of penguin thus produce calls with a distinctive envelope-modulation

pattern that facilitates unmasking effects. The observation by Aubin & Jouventin

(1998) that king penguin chicks are able to detect their parents’ calls within the

colony background noise with a S/N ratio of about −6 dB can be adequately ex-

plained by masking release in amplitude-modulated background noise.

Spatial release from masking

Signal detection also depends on the spatial arrangement of the sound

sources. If the sources of signal and masking noise are well separated, such as by

territorial songbirds, a considerable masking release may be observed. Hine et al.

(1994) measured the detection thresholds of ferrets Mustela putorius for 500 Hz tones

masked by narrowband noise. Signal and noise were presented either from the

same direction (+90◦, i.e. from the subject’s right side) or the signal was presented

from −90◦ and the noise bilaterally from +90◦ and −90◦. Signal detection was

improved by about 10 dB in the bilateral case. Signal detection did not improve in

animals that were only allowed monaural listening (Hine et al., 1994), indicating



that the release from masking was a result of binaural processing of signal and

noise. Dent and colleagues (1997) replicated the experiment with budgerigars

Melopsittacus undulatus using wideband noise as the masker and tone signals of

different frequencies. The amount of masking was, on average, 7.5 dB less for

bilaterally presented masking noise versus unilateral masking noise presented

from the same direction as the tone signal (Dent et al., 1997). In an additional

experiment, the authors demonstrated that the directional characteristics of the

birds’ auditory system are sufficient to explain the amount of unmasking. By

keeping the masking noise source constant at 0◦ azimuth and moving the signal

source around the animal in 30◦ steps, they observed a masking release of up to

10 dB (Dent et al., 1997; see also the review by Klump, 1996). Binaural processing

will thus contribute to signal detection in communication networks as well as in

other natural situations. However, such a large masking release has not been found

in all species that have been tested. For example, in the green treefrog Hyla cinerea,

spatial unmasking of up to 3 dB has been observed by phonotaxis experiments

with separate noise and signal sources (Schwartz & Gerhardt, 1989).

Recognition of signals

Detecting a signal is the first step in perception. Individuals in an acoustic

communication network may become alert when they detect a signal, but further

reaction will depend on the specific message: that is, the signal needs to be rec-

ognized. Signal recognition in the sense of statistical separation of signals may be

explained best by an everyday example. When listening to the radio while driving

a car, an individual may tune to a programme and can just detect ‘some signal’

or even ‘human speech’. By turning up the volume, and thus increasing the S/N

ratio between the speech and the engine noise of the car, words and, therefore, the

meaning of the message can be recognized. More theoretically, detecting a signal

means observing the occurrence of a signal, i.e. the addition of some signal to a

(possibly noisy) background. However, recognition implies the ability to classify

a detected signal as a member of a set of many (e.g. Green et al., 1977; Wiley &

Richard, 1982). We have little direct evidence on how the S/N ratio for recognition

compares with the S/N ratio required for detection of a signal.

Lohr et al. (2003) presented budgerigars and zebra finches Taeniopygia guttata

with contact calls of three species (zebra finch, budgerigar, canary Serinus canarius)

in order to determine thresholds for the detection of the calls in noise. They also

determined each species’ ability to discriminate between different call types of

zebra finches or budgerigars in the same masking noise. Birds were thus forced in

noisy background conditions to ‘hear out’ and recognize a deviant call in a series

of repeating reference calls. Thresholds for the discrimination of both conspecific



and heterospecific calls in noise were about 3 dB worse than thresholds for call

detection: the S/N ratio for recognition had to be, on average, 3 dB higher than the

S/N ratio for mere detection (Lohr et al., 2003).

There is no information from field studies that allows us to compare the S/N

ratio necessary for detection of signals with that for recognition of signals in the

natural environment. This is because field studies use playback to elicit natural

species-specific responses and, therefore, the subjects have both detected and rec-

ognized the signal (e.g. Brenowitz, 1982a,b; Aubin & Jouventin, 1998). Brenowitz

(1982a,b) studied the reaction of territorial male red-winged blackbirds Agelaius

phoeniceus to song signals that were either played alone or with wideband random

noise added to the playback song. Playback elicited more high-intensity song and

visual display when the S/N ratio was increased from 0 to 3 dB (as measured in

the 4 kHz octave band that contained most of the spectral energy of the song).

Since this male response required the recognition of the signal, one can conclude

that 3 dB is a conservative estimate of the S/N ratio necessary for signal recogni-

tion. Using data on auditory signal detection in red-winged blackbirds obtained

in the laboratory (Hienz & Sachs, 1987), Klump (1996) calculated that the S/N ratio

necessary for detection should be about 8 dB less than the S/N ratio necessary for

recognition. In king penguins, the S/N ratio necessary for recognition appears to

be lower than in the red-winged blackbird (Aubin & Jouventin, 1998). However,

as suggested above, this may be because of the distinctive envelope-modulation

pattern of penguin calls.

The sender’s adaptations for maximizing signal transmission

Behavioural observations of signalling birds suggest that senders have

evolved mechanisms to modify signal production in order to improve detection

by the receiver. Holland et al. (1998) concluded from measurements of broadcast

song of the wren Troglodytes troglodytes that higher song posts could optimize sound

transmission. Broadcasting song from high perches where the vegetation is less

dense also possibly improves the ability to detect responses of conspecifics in

wrens (Holland et al., 1998) and blackbirds Turdus merula (Dabelsteen et al., 1993).

However, it should be noted that changing location alone may improve perception

by allowing sequential integration of acoustic information, for example in the

context of sound localization.

Another strategy employed by a sender to increase information transfer within

a communication network is to adapt its sound output to the level of the back-

ground noise. This requires that a sender constantly monitors the level and spec-

tral composition of the background noise interfering with its own vocalization.

The increase in sound-pressure level of vocal output at times of increased levels of



background noise is called the Lombard effect and it has been well investigated

in humans (e.g. Pick et al., 1989). Sinnott et al. (1975) demonstrated that trained

monkeys Macaca fascicularis and M. nemestrina spontaneously increased their call

amplitude if noise bands of the same fundamental frequencies as their call masked

them, but not if noise bands were of much higher frequencies. The monkeys’voice

amplitude increased by about 2 dB for every 10 dB of masking noise. Also budgeri-

gars significantly increase the level of their vocalization in response to noise of the

same frequency spectrum as their contact calls but not to noise outside this spec-

tral range (Manabe et al., 1998). The Lombard effect has also been reported in zebra

finches (Cynx et al., 1998) and nightingales Luscinia megarhynchos (Brumm & Todt,

2002). The first animal species in which the Lombard effect has been shown un-

der field conditions is the blue-throated hummingbird Lampornis clemenciae (Pytte

et al., 2003). The authors observed that naturally occurring and experimentally

controlled amplitude changes of the ambient noise level induced change in am-

plitude of the birds’ territorial advertisement call.

Localization and distance perception

Perceiving the direction of a sound source

For participants in communication networks, it is advantageous to be

able to identify the location of the signal source. For example, the pattern of

alarm calls in a bird community could provide a good estimate of the path taken

by a predator (e.g. McGregor & Dabelsteen, 1996). Also, in territorial interactions,

birds appear to combine information from the song signal and the direction from

which it is heard to evaluate the potential threat by a competitor (e.g. McGregor &

Avery, 1986). There is considerable variation in the accuracy of sound localization

between species. Furthermore, each species’ ability to localize sound depends on

the physical characteristics of the sound, such as the frequency spectrum of the

sound or its temporal characteristics.

The accuracy and mechanisms of sound localization have been reviewed in

frogs (e.g. Rheinlaender & Klump, 1988), in birds (e.g. Klump, 2000) and in terres-

trial mammals (e.g. Gourevitch, 1987; Brown, 1994). Two cues are used in sound

localization: the difference in the time of arrival (or the phase difference) and the

intensity differences between the spectral components of the sound impinging

on the two ears. Figure 20.2a shows an example of interspecific variation in the

accuracy of the localization of tones in the horizontal plane (azimuth) for two

birds of prey, barn owl Tyto alba (Knudsen & Konishi, 1979) and sparrowhawk Ac-

cipiter nisus (G. M. Klump & E. Kretzschmar, unpublished data), and for four species

of small birds, great tit (Klump et al., 1986), zebra finch, budgerigar and canary



(a)

(b)

Fig. 20.2. The accuracy of azimuth sound localization. (a) Minimum detectable angles

of pure-tone stimuli in relation to frequency as determined in the laboratory in

two avian predators (open symbols, Knudsen & Konishi, 1979; G. M. Klump &

E. Kretzschmar, unpublished data) and four species with smaller interaural distances

than the raptors (filled symbols; Klump et al., 1986; Park & Dooling, 1991).



(Park & Dooling, 1991). The superior accuracy of the two birds of prey may be

explained at least partly by the physical properties of their auditory system pro-

viding larger interaural cues and by specializations in processing the interaural

sound differences in the auditory pathway (see Klump, 2000).

When discussing the accuracy of sound localization in the natural environ-

ment, data from field studies are relevant (e.g. Nelson & Stoddard, 1998). Figure

20.2b shows the sound localization accuracy of a trained male sparrowhawk in

the laboratory and in the field (G. M. Klump & E. Kretzschmar, unpublished data).

The data in both sets were obtained with the same stimulus paradigm and operant

procedure and so can be compared directly. The sparrowhawk’ssound localization

accuracy was considerably reduced in the field. Localization accuracy for natural

sounds was similar to the accuracy for tones of comparable frequency. The data

shown in Fig. 20.2 were obtained by forcing the bird to localize a single signal

presented at an unpredictable time. This procedure ensures that the bird is only

using open-loop sound localization (Klump, 1995) and cannot use strategies to in-

tegrate information over several signal presentations or maximize binaural cues

in some other way. It is to be expected that field studies presenting several signals

before the subject responds will result in more accurate sound localization (e.g.

Nelson & Stoddard, 1998).

Perceiving the distance of a sound source

To assess the location of a signal source, knowledge of distance is as impor-

tant as information about the direction from which the signal is heard. Assessing

distance information is often referred to as ranging (Morton, 1982). In the past,

distance assessment has almost exclusively been investigated in birds by simulat-

ing territorial intrusions and studying the behavioural response of the territory

owner (reviewed by Naguib & Wiley, 2001). There are also many other common

contexts in which distance is assessed, for example animals maintaining contact

with mates or flock members when moving through dense vegetation.

Fig. 20.2 (cont.) (b) Minimum angles detectable by a sparrowhawk Accipiter nisus

determined either in an anechoic chamber in the laboratory (open symbols, thick grey

lines) or in a natural deciduous forest (closed symbols, thick black lines) using a

two-alternative forced-choice procedure that had been established for measuring the

accuracy of sound localization in small birds (Klump, 1995). Circles show the

minimum detectable angle for pure tones (open circles are the same data as in (a).

Diamonds indicate the minimum detectable angle for the ‘seeet’ aerial predator call.

Horizontal lines represent the frequency range and the sparrowhawk’s localization

accuracy for great tit mobbing calls (dashed thick lines) and scolding calls (solid thick

lines). (G. M. Klump & E. Kretzschmar, unpublished data.)



Two recent studies have provided data on the accuracy of distance percep-

tion as revealed by the approach responses of birds in the natural environment.

Nelson & Stoddard (1998) measured the accuracy of the approach by Eastern

towhees Pipilo erythrophthalmus to a loudspeaker playing back the species’ calls.

The error in distance assessment was determined from the birds’ closest approach

to the loudspeaker. The birds’ initial distance to the sound source was approxi-

mately 10, 20 or 30 m and resulted in average distance errors of 2.3, 3.5 and 3.4 m,

respectively. In additional experiments, the actual distance to the sound source

did not match the characteristics of the playback signal. For example, a signal rere-

corded after being transmitted over a distance of 20 m was played from an actual

distance of 10 m resulting in a simulated distance of 30 m. About half of the ex-

perimental birds responded by flying the actual distance towards the loudspeaker

and the other half by flying the simulated distance, suggesting that some cues for

distance assessment are derived from signal characteristics and some from the

actual location of the sound source (Nelson & Stoddard, 1998).

Simulated distances have also been used by Naguib et al. (2000) to study distance

assessment by chaffinches Fringilla coelebs. Unlike previous field studies, Naguib

et al. (2000) manipulated song signals by simulating their transmission in a vir-

tual forest with the help of a computer. This allowed control over the amount of

reverberation imposed and control over frequency-dependent attenuation. Simu-

lated transmission distances ranged from 0 (original source signal) to 120 m. The

approach response to playback of a single song with different virtual distances

showed that chaffinches mainly discriminated between playback signals simu-

lating shorter distances (0, 20 and 40 m) and playback signals simulating longer

distances (80 and 120 m) but did not discriminate within short- or long-distance

categories. This means that chaffinches exhibit a categorical response to simulated

intruders close to, versus more distant from, their territory, rather than gauging

their approach to the virtual distance of the sound source (Naguib et al., 2000).

Laboratory studies of birds’ perception allow us to evaluate the salience of the

cues that may be used for distance assessment. Cues suggested by field studies in-

clude the overall signal amplitude, the frequency-dependent excess attenuation,

the amplitude modulation of the signal envelope imposed by atmospheric turbu-

lence along the transmission path, the addition of noise to the signal, and rever-

beration resulting from echoes overlapping or trailing the signal (e.g. Dabelsteen

et al., 1993; McGregor, 1994). Overall signal amplitude has been identified as

a useful cue for distance assessment in psychoacoustic studies. Phillmore et al.

(1998) trained zebra finches and black-capped chickadees Poecile atricapillus with

operant procedures to distinguish calls and songs from either species that were

recorded in a woodland habitat at distances of 5, 25, 50 and 75 m from the



loudspeaker. After training with a set of signals from all four distances, the birds

could discriminate between the signals of the set and also between unknown

songs from these recording distances. The stimuli representing various distances

differed in amplitude at least as much as the known intensity-difference limen (the

just-noticeable intensity difference) in birds (Dooling & Saunders, 1975; Klump &

Baur, 1990). Removing amplitude cues made discrimination considerably worse

(Phillmore et al., 1998), indicating its potential role in distance assessment. How-

ever, in the field, amplitude alone may not be a reliable indicator of distance,

since head movements of the singing bird can lead to amplitude differences at the

receiver’s position (e.g. Larsen & Dabelsteen, 1990).

A number of field studies suggest reverberation as an important cue for distance

assessment (e.g. McGregor, 1994; Naguib & Wiley, 2001). Echos that are imposed

on each signal element during transmission ‘degrade’ (i.e. distort) its original

amplitude and time pattern: the study by Holland et al. (2001) suggested that

wrens can extract cues that allow distance assessment from the echo tail trailing

the signal. Wrens responded to songs consisting of undegraded elements with

added trailing echo tails (from degraded elements) in the same way as to degraded

songs (i.e. degraded element and echo tail). Songs consisting of degraded elements

without echo tails elicited a response that was intermediate between that to an

undegraded and that to a degraded song with echo tail, stressing the salience of the

echo tail as a cue. Psychoacoustic studies of humans also indicate that echo tails

provide an important cue for distance assessment; we appear to use the direct-

to-reverberant energy ratio (Zahorik, 2002). This cue may allow us to estimate

sound-source distance independent of the sound level, which in turn may allow

the loudness of the source to be inferred (Zahorik & Wightman, 2001).

SINDSCAL: an analysis method for perceptual distances

Laboratory studies in the great tit also indicate that echoes alone may

provide a sufficient cue for distance assessment. In this final part, we would like

to describe how perceptual differences can be examined in trained animals. We

present a multidimensional scaling procedure (SINDSCAL: symmetric individual

difference scaling; e.g. Arabie et al., 1987) that is especially suited to investigate

which signal modifications are salient to the animals. A virtual forest (Naguib

et al., 2000) was used to impose reverberation on synthetic great tit song signals

equivalent to sound transmission distances of between 5 and 320 m (Fig. 20.3).

The sound pressure of the song signals was then adjusted to the same overall root

mean square amplitude so that the reverberation pattern imposed on the signal

remained the only possible cue to distance. Great tits were then trained in an oper-

ant Go/NoGo procedure with repeating background to discriminate between song



Fig. 20.3. An example of stimuli used to estimate perceptual differences in laboratory

experiments with great tits. The spectrogram shows a two-element great tit song at a

virtual distance of 5 m and the waveforms show the same song elements at virtual

distances of 5, 80 and 160 m. Further details in text.



elements in which echo patterns alone indicated different virtual distances (e.g.

Dooling & Okanoya, 1995). The response latencies of the great tits to all possible

reference–test differences were recorded (Fig. 20.4a) and analysed by a SINDSCAL

(e.g. Arabie et al., 1987) model with log-transformed data. The result of SINDSCAL is

a three-dimensional object space (Fig. 20.4b), providing a kind of ‘perceptual map’

in which the response latencies are translated into relative distances between

data points. The distance between data points in the perceptual space provides a

measure of the perceived similarity of the acoustic signals. Data points that are

close to each other indicate that large response latencies were observed in the

discrimination (i.e. the difference was not salient to the birds). Short response

latencies indicate a salient differences that will lead to a large spread of the data

points. A three-dimensional SINDSCAL model accounts for 86–90% of the variance

in response latencies. Since the first dimension explained most of the total vari-

ance, the latency data for the different experimental songs were reanalysed with a

one-dimensional SINDSCAL model to allow easier comparison between song types.

Perceptual space coordinates (from the one-dimensional model) of the four differ-

ent two-note songs that were tested are shown in Fig. 20.5 as a function of the

virtual distance of each signal. As in Fig. 20.4b, close perceptual space coordinates

indicate signals that have been perceived as being more similar. There was no sig-

nificant difference between the different song types tested (two-way ANOVA with

distance and song type as factors and subsequent Tukey-tests, F3,24 < 0.001; p = 1).

However, within the factor ‘virtual distance’, significant differences were obvious

(F6,21 = 73.7; p ≤ 0.001). Space coordinates for ‘long’ virtual distances (160 and

320 m) differed significantly from those for the other distances but not from each

other (i.e. ‘long’ virtual distances were similar for all the birds’). Space coordinates

for the virtual distance of 80 m significantly differed from those of the three ‘short’

distances (5, 10 and 20 m). ‘Short’ virtual distances of 5, 10 and 20 m were treated

as similar by the birds (i.e. no significant difference was found). Therefore, great

tits in the laboratory provided with echo patterns as the only available distance

assessment cue showed a categorical response that was similar to the response

observed in field experiments with chaffinches (Naguib et al., 2000).

Auditory scene analysis

So far we have discussed basic perceptual mechanisms involved in rela-

tively simple auditory detection and discrimination tasks. Real-world situations,

however, require receivers to analyse sounds from a mixture of simultaneously

active sources: that is, to perform auditory scene analysis (Bregman, 1990). For ex-

ample, this applies to communication networks of birds during the dawn chorus,

when a receiver has to analyse streams of song elements from each individual



500

1000

1500

2000

2500

0 50 100 150 200 250 300 350

Virtual distance difference (m)

rs = −0.839 p < 0.001

Song 5

Perceptual space(song 5)

5 m10 m

20 m

80 m

160 m

320 m

40 m

(a)

(b)

Res

po

nse

late

ncy

(m

s)

−

−

−

−

−

−

−

−

−

−

−

−

−

−

−

−

Fig. 20.4.



Fig. 20.5. One-dimensional perceptual space coordinates as a function of the virtual

distance (i.e. the distance simulated by imposing reverberations upon the signal) for

four test songs. Similar perceptual space coordinates indicate that the differences in

the echo pattern of the signals are not very salient to the great tits, and large

differences between data points indicate salient differences have been detected

between the respective echo modifications. There are no salient differences between

signals of ‘short’ virtual distances of 5, 10 and 20 m or between signals of ‘long’ virtual

distances of 160 and 360 m. Differences between signals of ‘short’ virtual distances

and ‘long’ virtual distances are very salient to the birds. Virtual distances of 40 and

80 m lie in a transition range. This pattern occurs for all songs that were tested (see

ANOVA results in the text).

Fig. 20.4 (cont.) Results of great tits scaling differences between songs manipulated to

represent various virtual distances. The signal variants consisted of the same song

elements that differed only in echo pattern, which simulated sound transmission

distances of 5 m to 320 m. (a) Average response latencies of four great tits as a function

of the difference in virtual distance (e.g. the virtual-distance difference between

signals of simulated transmission distances of 320 and 80 m is 240 m). Larger response

latencies indicate less-salient differences in the cues; shorter response latencies

indicate more-salient differences. (b) Three-dimensional object space or ‘perceptual

map’ (SINDSCAL model with log-transformed data, see text; Arabie et al., 1987)

demonstrating the salience of reverberation for great tits. Distances between data

points in perceptual space reflect response-latency differences in discriminating

between the signal variants. Small distances indicate that signals are treated as being

similar, and large distances indicate that salient differences between signals have

been perceived. (U. Langemann, U. Pander & G. M. Klump, unpublished data.)



singer. Laboratory experiments have shown that animals form auditory streams

and analyse auditory scenes in a similar way to humans (e.g. Hulse et al., 1997;

Feng & Ratnam, 2000; Moss & Surlykke, 2001; Hulse, 2002). However, some of

the basic mechanisms of masking discussed above also contribute to an auditory

scene analysis that is characterized by an improved segregation of overlapping

signals. For example, we have shown that common modulation of components

of sounds aids signal segregation and results in reduced masking of one sound

by another. Similarly, common onsets or offsets of signal components lead to the

formation of auditory objects (e.g. Geissler & Ehret, 2002) that can be analysed

separately from other objects in the same auditory scene. Spatial separation of

sources will also aid auditory object formation, and the spatial release from mask-

ing discussed above may be partly a result of improved signal segregation from the

background.

Summary

In this chapter, we have illustrated how sensory abilities of individuals af-

fect auditory perception and thus acoustic communication. How does this relate

to communication between individuals in networks? In a network, individuals

are distributed in space. The relative position of any ‘node’ in this network, the

distance between individuals, profoundly determines a receiver’s ability to detect

and recognize acoustic signals. Because of the spatial distribution of signallers

and receivers, the same propagated signal may result in quite different percep-

tion at different places in a communication network. Acoustic signals will be

modified along their transmission path and will be masked by acoustical energy

from other sources. On the one hand, masking is certainly the most important fac-

tor severely impairing the detection of acoustic communication signals. Spectral

aspects and the temporal patterns of masking sounds affect the amount of mask-

ing that is exerted and the spatial distribution of concurrent sound sources (or

individuals) contributes to masking efficiency. On the other hand, receivers may

exploit changes imposed on a signal during transmission. For example, reverbera-

tion patterns will allow distance assessment of sound sources and, together with

binaural cues, render it possible to gain insights into the spatial distribution of

the individuals in a network.

Our current knowledge from perceptual studies will provide a better under-

standing of animal behaviour within acoustically complex communication net-

works. In addition, an approach that takes the receiver’s perception into account

will allow a better evaluation of communication behaviour in the field than ap-

proaches that rely mainly on physical properties of signals and their transmission

(McGregor et al., 2000).



Acknowledgements

The research was supported by grants from the Deutsche Forschungsgemeinschaft (SFB

204, FOR 306). Ulrike Pander provided data from experiments with great tits that were reanalysed

for this study.

References

Arabie, P., Carroll, J. D. & DeSarbo, W. S. 1987. Three-Way Multidimensional Scaling and

Related Techniques. Newbury Park, CA: Sage.

Aubin, T. & Jouventin, P. 1998. Cocktail-party effect in king penguin colonies.


Aubin, T., Jouventin, P. & Hildebrand, C. 2000. Penguins use the two-voice system to

recognize each other. Proceedings of the Royal Society of London, Series B, 267,

1081–1087.

Boersma, H. F. 1997. Characterization of the natural ambient sound environment:

measurements in open agricultural grassland. Journal of the Acoustical Society of

America, 103, 2104–2110.

Bregman, A. S. 1990. Auditory Scene Analysis: The Perceptual Organization of Sound.

Cambridge, MA: MIT Press.

Brenowitz, E. A. 1982a. The active space of red-winged blackbird song. Journal of

Comparative Physiology A, 147, 511–522.

1982b. Long-range communication of species identity by song in the red-winged

blackbird. Behavioral Ecology and Sociobiology, 10, 29–38.

Brown, C. H. 1994. Sound localization. In: Comparative Hearing: Mammals, ed. R. R. Fay &

A. N. Popper. New York: Springer, pp. 57–96.

Brumm, H. & Todt, D. 2002. Noise-dependent song amplitude regulation in a

territorial songbird. Animal Behaviour, 63, 891–897.

Budelis, J., Fishbach, A. & May, B. J. 2002. Behavioral assessment of comodulation

masking release in cats. Abstracts of the Association for Research in Otolaryngology, 25,

211.

Cynx, J., Lewis, R., Tavel, B. & Tse H. 1998. Amplitude regulation of vocalizations in

noise by a songbird, Taeniopygia guttata. Animal Behaviour, 56, 107–113.

Dabelsteen, T., Larsen, O. N. & Pedersen, S. B. 1993. Habitat-induced degradation of

sound signals: quantifying the effects of communication sounds and bird

location on blur ratio, excess attenuation, and signal-to-noise ratio in blackbird

song. Journal of the Acoustical Society of America, 93, 2206–2220.

Dent, M. L., Larsen, O. N. & Dooling, R. J. 1997. Free-field binaural unmasking in

budgerigars (Melopsittacus undulatus). Behavioral Neuroscience, 111, 590–598.

Dooling, R. J. & Okanoya, K. 1995. Psychophysical methods for assessing perceptual

categories. In: Methods in Comparative Psychoacoustics, ed. G. M. Klump, R. J. Dooling,

R. R. Fay & W. C. Stebbins. Basel: Birkhauser, pp. 307–318.

Dooling, R. J. & Saunders, J. C. 1975. Auditory intensity discrimination in the parakeet

(Melopsittacus undulatus). Journal of the Acoustical Society of America, 58, 1308–1310.



Douglas, H. D. & Conner, W. E. 1999. Is there a sound reception window in coastal

environments? Evidence from shorebird communication systems.

Naturwissenschaften, 86, 228–230.

Dubois, A. & Martens, J. 1984. A case of possible vocal convergence between frogs and a

bird in Himalayan torrents. Journal fur Ornithologie, 125, 455–463.

Fay, R. R. 1988. Hearing in Vertebrates: a Psychophysics Databook. Winnetka, IL: Hill-Fay.

Fegeant, O. 1999. Wind-induced vegetation noise. Part II: field measurements.

Acustica/Acta Acustica, 85, 241–249.

Feng, A. S. & Ratnam R. 2000. Neural basis of hearing in real-world situations. Annual

Review of Psychology, 51, 699–725.



Geissler, D. B. & Ehret, G. 2002. Time-critical integration of formants for perception of

communication calls in mice. Proceedings of the National Academy of Sciences, USA, 99,

9021–9025.

Gourevitch, G. 1987. Binaural hearing in land mammals. In: Directional Hearing, ed.

W. A. Yost & G. Gourevitch. New York: Springer, pp. 226–246.

Green, D. M., Weber, D. L. & Duncan, J. E. 1977. Detection and recognition of pure

tones in noise. Journal of the Acoustical Society of America, 62, 948–954.

Hienz, R. D. & Sachs, M. B. 1987. Effects of noise on pure-tone thresholds in blackbirds

(Agelaius phoeniceus and Molothrus ater) and pigeons (Columba livia). Journal of


Hine, J. E., Martin, R. L. & Moore, D. R. 1994. Free-field binaural unmasking in ferrets.

Behavioral Neuroscience, 108, 196–205.

Holland, J., Dabelsteen, T., Pedersen, S. B. & Larsen, O. N. 1998. Degradation of wren

Troglodytes troglodytes song: implications for information transfer and ranging.


Holland, J., Dabelsteen, T., Bjørn, C. P. & Pedersen, S. B. 2001. The location of ranging

cues in wren song: evidence from calibrated interactive playback experiments.

Behaviour, 138, 189–206.

Hulse, S. H. 2002. Auditory scene analysis in animal communication. Advances in the

Study of Behaviour, 31, 163–200.

Hulse, S. H., MacDougall-Shackleton, S. A. & Wisniewski, A. B. 1997. Auditory scene

analysis by songbirds: stream segregation of birdsong by European starlings

(Sturnus vulgaris). Journal of Comparative Psychology, 111, 3–13.

Klump, G. M. 1995. Sound localization studies in nonspecialized birds. In: Methods in

Comparative Psychoacoustics, ed. G. M. Klump, R. J. Dooling, R. R. Fay & W. C.

Stebbins. Basel: Birkhauser, pp. 171–182.

1996. Bird communication in the noisy world. In: Ecology and Evolution of Acoustic

Communication in Birds, ed. D. E. Kroodsma & E. H. Miller. Ithaca NY: Cornell

University Press, pp. 321–338.

2000. Sound localization in birds. In: Comparative Hearing: Birds and Reptiles, ed. R. J.

Dooling, R. R. Fay & A. N. Popper. New York: Springer, pp. 247–305.

Klump, G. M. & Baur, A. 1990. Intensity discrimination in the European starling

(Sturnus vulgaris). Naturwissenschaften, 77, 545–548.



Klump, G. M. & Curio, E. 1983. Warum liegen viele Lautaußerungen von Singvogeln

oberhalb des Bereichs besten Horens? Verhandlungen der Deutschen Zoologischen

Gesellschaft, 76, 182.

Klump, G. M. & Gerhardt, H. C. 1992. Mechanisms and function of call-timing in

male–male interactions in frogs. In: Playback and Studies of Animal Communication,

ed. P. K. McGregor. London: Plenum Press, pp. 153–174.

Klump, G. M. & Langemann, U. 1995. Comodulation masking release in a songbird.

Hearing Research, 87, 157–164.

Klump, G. M. & Nieder, A. 2001. Release from masking in fluctuating background

noise is represented in a songbird’s auditory forebrain. Neuroreport, 12,

1825–1829.

Klump, G. M. & Shalter, M. D. 1984. Acoustic behaviour of birds and mammals in the

predator context: I. Factors affecting the structure of alarm calls. II. The

functional significance of alarm signals and their evolution. Zeitschrift fur

Tierpsychologie, 66, 189–226.

Klump, G. M., Windt, W. & Curio E. 1986. The great tit’s (Parus major) auditory

resolution in azimuth. Journal of Comparative Physiology, 158, 383–390.

Klump, G. M., Langemann, U. & Nieder, A. 1998. Mechanisms that improve signal

detection in noise: a study of comodulation masking release in a songbird. In:

Psychophysical and Physiological Advances in Hearing, ed. A. R. Palmer, A. Rees, A. Q.

Summerfield & R. Meddis. London: Whurr, pp. 270–276.

Klump, G. M., Kittel, M., Wagner, E. 2001. Comodulation masking release in the

Mongolian gerbil. Abstracts of the Association for Research in Otolaryngology, 24, 84.

Knudsen, E. I. & Konishi, M. 1979. Mechanisms of sound localization in the barn owl

(Tyto alba). Journal of Comparative Physiology, 133, 13–21.

Langemann, U., Gauger, B. & Klump, G. M. 1998. Auditory sensitivity in the great tit:

perception of signals in the presence and absence of noise. Animal Behaviour, 56,

763–769.

Larom, D., Garstang, M., Lindeque, M. et al. 1997. Meteorology and elephant

infrasound at Etosha National Park, Namibia. Journal of the Acoustical Society of

America, 101, 1710–1717.

Larsen, O. N. & Dabelsteen, T. 1990. Directionality of blackbird vocalization.

Implications for vocal communication and its further study. Ornis Scandinavica,

21, 37–45.

Lengagne, T., Lauga, J. & Aubin, T. 2001. Intra-syllabic acoustic signatures used by the

king penguin in parent-chick recognition: an experimental approach. Journal of

Experimental Biology, 204, 663–672.

Lohr, B., Wright T. F. & Dooling, R. J. 2003. Detection and discrimination of natural

calls in masking noise by birds: estimating the active space of a signal. Animal

Behaviour, 65, 763–777.

Long, G. R. 1977. Masked auditory thresholds from the bat, Rhinolophus ferrumequinum.

Journal of Comparative Physiology A, 116, 247–255.

Manabe, K., Sadr, E. I. & Dooling, R. J. 1998. Control of vocal intensity in budgerigars

(Melopsittacus undulatus): differential reinforcement of vocal intensity and the

Lombard effect. Journal of the Acoustical Society of America, 103, 1190–1198.



Marten, K. & Marler, P. 1977. Sound-transmission and its significance for animal

vocalization. I. Temperate habitats. Behavioral Ecology and Sociobiology, 2,

271–290.

McFadden, D. & Wright, B. A. 1990. Temporal decline of masking and comodulation

detection differences. Journal of the Acoustical Society of America, 88, 711–724.

McGregor, P. K. 1994. Sound cues to distance: the perception of range. In: Perception and

Motor Control in Birds: An Ecological Approach, ed. M. N. O. Davies & P. R. Green.

Berlin: Springer, pp. 74–94.

McGregor, P. K. & Avery, M. I. 1986. The unsung songs of great tits (Parus major):

learning neighbours’ songs for discrimination. Behavioral Ecology and Sociobiology,

18, 311–316.


Evolution of Acoustic Communication in Birds, ed. D. E. Kroodsma & E. H. Miller. Ithaca

NY: Cornell University Press, pp. 409–425.



McGregor, P. K., Otter, K. A. & Peake, T. M. 2000. Communication networks: receiver

and signaller perspectives. In: Animal Signals. Signalling and Signal Design in Animal

Communication, ed. Y. Espmark, T. Amundsen & G. Rosenqvist. Trondheim: Tapir


Miller, G. A. 1947. Sensitivity to changes in the intensity of white noise and its

relation to masking and loudness. Journal of the Acoustical Society of America, 19,

609–619.

Moore, B. C. J. 1990. Co-modulation masking release: spectro-temporal pattern analysis

in hearing. British Journal of Audiology, 24, 131–137.

2003. An Introduction to the Psychology of Hearing. Amsterdam: Academic Press.

Moore, B. C. J. & Glasberg, B. R. 1983. Suggested formulae for calculating

auditory-filter bandwidths and excitation pattern. Journal of the Acoustical Society of

America, 74, 750–753.

Morton, E. S. 1982. Grading, discreteness, redundancy, and motivation–structural

rules. In: Acoustic Communication in Birds, Vol. I, ed. D. E. Kroodsma, E. H. Miller & H.

Ouellet. New York: Academic Press, pp.183–212.

Moss, C. F. & Surlykke, A. 2001. Auditory scene analysis by echolocation in bats. Journal

of the Acoustical Society of America, 110, 2207–2226.



62, 825–837.

Naguib, M., Klump, G. M., Hillmann, E., Griessman, B. & Teige, T. 2000. Assessment of

auditory distance in a territorial song bird: accurate feat or rule of thumb? Animal

Behaviour, 59, 715–721.

Nelken, I., Rotman, Y. & Bar Yosef, O. 1999. Responses of auditory-cortex neurons to

structural features of natural sounds. Nature, 397, 154–157.

Nelson, B. S. & Stoddard, P. K. 1998. Accuracy of auditory distance and azimuth

perception by a passerine bird in natural habitat. Animal Behaviour, 56, 467–477.



Niemiec, A. J. 2001. The effects of increasing masker temporal regularity on

co-modulation masking thresholds in chinchillas. Abstracts of the Association for

Research in Otolaryngology, 24, 85.

Park, T. J. & Dooling, R. J. 1991. Sound localization in small birds: absolute localization

in azimuth. Journal of Comparative Psychology, 105, 125–133.

Pick, H. L., Siegel, G. M., Fox, P. W., Garber, S. R. & Kearney, J. K. 1989. Inhibiting the

Lombard effect. Journal of the Acoustical Society of America, 85, 894–900.

Phillmore L. S., Sturdy, C. B., Ramsay, S. M. & Weisman, R. G. 1998 Discrimination of

auditory distance cues by black-capped chickadees (Poecile atricapillus) and zebra

finches (Taeniopygia guttata). Journal of Comparative Psychology, 112, 282–291.

Pressnitzer, D., Meddis, R., Delahaye, R. & Winter, I. M. 2001. Physiological correlates

of comodulation masking release in the mammalian ventral cochlear nucleus.

Journal of Neuroscience, 21, 6377–6386.

Pytte, C. L., Rusch, K. M. & Ficken, M. S. 2003. Regulation of vocal amplitude by the

blue-throated hummingbird, Lampornis clemenciae. Animal Behaviour, 66, 703–710.

Rheinlaender, J. & Klump, G. M. 1988. Behavioural aspects of sound localization. In: The

Evolution of the Amphibian Auditory System, ed. B. Fritsch, M. J. Ryan, W. Wilczynski,

T. E. Hetherington & W. Walkowiak. New York: John Wiley, pp. 297–305.

Richards, D. G. & Wiley, R. H. 1980. Reverberations and amplitude fluctuations in the

propagation of sound in a forest: implications for animal communication.


Ryan, M. J. & Brenowitz, E. A. 1985. The role of body size, phylogeny and ambient noise

in the evolution of bird song. American Naturalist, 126, 87–100.

Schooneveldt, G. P. & Moore, B. C. J. 1989. Comodulation masking release (CMR) as a

function of masker bandwidth, modulator bandwidth, and signal duration.


Schwartz, J. J. & Gerhardt, H. C. 1989. Spatially mediated release from auditory

masking in an anuran amphibian. Journal of Comparative Physiology A, 166, 37–41.

Shannon, C. E. & Weaver, W. 1949. The Mathematical Theory of Communication. Urbana, IL:

University of Illinois Press.

Sinnott, J. M., Stebbins, W. C. & Moody, D. B. 1975. Regulation of voice amplitude by

the monkey. Journal of the Acoustical Society of America, 58, 412–414.

Staicer, C. A., Spector, D. A. & Horn, A. G. 1996. The dawn chorus and other dial

patterns in acoustic signalling. In: Ecology and Evolution of Acoustic Communication in

Birds, ed. D. E. Kroodsma & E. H. Miller. Ithaca NY: Cornell University Press,

pp.426–453.

Swets, J. A., Shipley, E. F., McKey, M. J. & Green, D. M. 1950. Multiple observations of

signals in noise. Journal of the Acoustical Society of America, 31, 514–521.

Waser, P. M. & Waser, M. S. 1977. Experimental studies of primate vocalization:

specializations for long-distance propagation. Zeitschrift fur Tierpsychologie, 43,

239–263.






1982. Adaptations for acoustic communication in birds: sound transmission and

signal detection. In: Acoustic Communication in Birds, Vol. I, ed. D. E. Kroodsma, E. H.

Miller & H. Ouellet. New York: Academic Press, pp. 131–181.

Wollerman, L. & Wiley, R. H. 2002. Background noise from a natural chorus alters

female discrimination of male calls in a Neotropical frog. Animal Behaviour, 63,

15–22.

Yost, W. A. 1994. Fundamentals of Hearing: An Introduction. San Diego CA: Academic Press.

Zahorik, P. 2002. Direct-to-reverberant energy ratio sensitivity. Journal of the Acoustical


Zahorik, P. & Wightman, F. L. 2001. Loudness constancy with varying sound source

distance. Nature Neuroscience, 4, 78–83.

P1: GFZ-IYP/... P2: KOD0521823617c21.xml CU1917B/McGregor 0 521 582361 7 April 11, 2005 19:12

21

Hormones, social contextand animal communication

r u i f. o l i v e i r a

Instituto Superior de Psicologia Aplicada, Lisbon, Portugal

Introduction

The views on the role that hormones play in the control of behaviour

have changed progressively with time. Hormones were classically seen as causal

agents of behaviour, acting directly on the display of a given behaviour. This view

was mainly supported by early studies of castration and hormone-replacement

therapy, which showed that some behaviours were abolished by castration and

restored by exogenous administration of androgens (Nelson, 2001). Later this view

shifted towards a more probabilistic approach and hormones started to be seen

more as facilitators of behaviour than as deterministic factors (Simon, 2002). Ac-

cording to this new view, hormones may increase the probability of the expression

of a given behaviour by acting as modulators of the neural pathways underlying

that behavioural pattern. For example, the effects of androgens on the expres-

sion of aggressive behaviours in mammals are mediated by modulatory effects

on central serotonergic and vasopressin pathways (Simon, 2002). Yet, it is also

known that the social environment (i.e. network of interacting individuals) also

feeds back to influence hormone levels (Wingfield et al., 1990), suggesting a two-

way type of interaction between hormones and behaviour. In this chapter, I will

develop the hypothesis that social modulation of androgens is an adaptive mech-

anism through which individuals adjust their motivation according to the social

context that they are facing. Thus, the social interactions within a given social

network would stimulate the production of androgens in the individuals and the

individual levels of androgens would be a function of the perceived social status

and the stability of the social environment in which the animal is living. According

to this view, androgens may play a key role as endocrine mediators of the effects


481


482 R. F. Oliveira

Fig. 21.1. Interplay between androgens, social behaviour and social context.

Androgens influence the production of a number of social behaviours involved in

communication interactions between animals. In turn, these social interactions

among a network of individuals will shape the social context in which these animals

live, which subsequently will modulate their androgen levels.

of social context on the expression of social behaviour, allowing the animal to

adjust its social behaviour to the context by modulating sensory, cognitive and

motor neural mechanisms underlying animal communication (Fig. 21.1).

In this chapter, I only consider vertebrates because they have a remarkably

similar endocrine system, whereas that of invertebrates can be very different (e.g.

the androgenic hormone in crustaceans is a peptide not a steroid as in vertebrates

(Hasegawa et al., 2002)). Within the vertebrates, I mainly use examples from bony

fishes, a group with wide diversity in mating and parental care systems that makes

up about half the existing vertebrate species (Nelson, 1994). I have also concen-

trated on androgens and male behaviour because I argue that the social network in

which the individual lives modulates its neuroendocrine system, which, in turn,

adjusts the expression of behaviour according to social context. Stress hormones

are, by definition, affected by the social environment and a number of reviews on

social stress and hypothalamus–pituitary–adrenal axis have been published since

the early 1990s (e.g. Sapolsky, 1992). Consequently, an additional benefit of this

chapter is to claim that, like stress hormones, androgens (and perhaps also other

hormones) respond in an adaptive way to the social context, preparing the animal

for the social interactions that it has to face in its everyday life.

Hormones and communication I: the dyadic view

Conceptually, the neurochemical pathways modulated by hormones can

be part of one of three major functional compartments of the nervous system:

sensory, central processing and effector systems (Nelson, 2001). If we translate

this rationale to the communication paradigm, one can consider that hormones

may affect communication by modulating the production of the signal in the


Hormones, social context and animal communication 483

Signal

Input/perception

Motoroutput

Centralprocessing

Hormones

Receiver

Input/perception

Motoroutput

Centralprocessing

Hormones

Somatic releasers

Sender

Fig. 21.2. Flow of information in a communication dyad. The arrows indicate the

direction of circulation of the information within and between individuals. In the

sender, sensory information received will influence central processing mechanisms

in the central nervous system (CNS), which control, at a higher level, the behavioural

motor output systems that produce the signals. Hormones may modulate signal

production by the sender by acting on central mechanisms, on motor output

mechanisms or by modifying somatic structures that affect the emission of the signal

(i.e. somatic releasers). In the receiver, the signal will be detected by sensory systems

and after peripheral processing will be forwarded to central processing systems in the

CNS. Hormones may affect signal reception and processing in the receiver by acting

directly on the sensory systems that perceive it and/or by acting at a higher level on

the central processing mechanism of the CNS. The central processing of the signal by

the limbic system (and other structures involved in motivational mechanism) may

feed back on hormone levels. The boxes delimit the two organisms and within the

boxes the grey elliptical areas represent the nervous system.

sender, the perception of the signal by the receiver or the central processing of

the message in both senders and receivers (Fig. 21.2).

Hormonal modulation of effector pathways

In senders, hormones may modulate the effector pathways that are in-

volved in the motor circuits underlying the production of the signal. In this way,

hormones can affect the expression of visual displays, vocalizations or pheromone

production and/or release. From the numerous examples in the literature, I have

selected the following, which are intended to cover different communication chan-

nels in different vertebrate taxa.

Androgens and the production of acoustic signals

In songbirds, circulating levels of testosterone are higher at the peak of

the breeding season when singing behaviour reaches its maximum (e.g. Rost, 1990,


484 R. F. Oliveira

1992; Smith et al., 1997). Moreover, song production is substantially reduced after

castration and is restored after androgen-replacement therapy (e.g. Arnold, 1975;

Heid et al., 1985). Finally, both androgen and oestrogen receptors have been local-

ized in the song control nuclei of the bird brain: the former in the high vocal cen-

tre, the robust nucleus of the archistriatum, the lateral part of the magnocellular

nucleus of the anterior neostriatum and the dorsomedial part of the intercollic-

ular nucleus; the latter in the high vocal centre and the intercollicular nucleus

(Balthazart et al., 1992; Brenowitz & Arnold, 1989, 1992; Gahr et al., 1987, 1993). Song

is produced by the coordinated contraction of respiratory, syringeal and cranio-

mandibular muscles (Suthers et al., 1999). The activity of syringeal muscles regu-

lates both the timing and the fundamental frequency of the sound (Suthers et al.,

1999). Therefore, by acting directly on the activity and development of syringeal

muscles, hormones may affect song production. For example, in zebra finches

Taenopygia guttata, androgens inhibit the activity of the enzyme cholinesterase,

which breaks down the neurotransmitter acetylcholine in the neuromuscular

junctions of the syrinx. This results in a longer lifetime for the neurotransmitter

in the synaptic cleft, which will affect the syringeal contraction pattern and, sub-

sequently, song output and/or structure (Luine et al., 1980). Testosterone is also

known to increase both syringeal muscle mass (Luine et al., 1980) and the density

of acetylcholine receptors in syringeal muscles, suggesting that circulating levels

of testosterone may increase the size and number of endplates in neuromuscular

junctions (Bleisch et al., 1984). Also, in non-oscine birds, testosterone is effective in

inducing changes in call structure by acting on the motor vocal structure underly-

ing these calls. In grey partridges Perdix perdix, male mating calls used by females

in mate choice are affected by testosterone treatment, which induces a thickening

of the external tympanic membranes that are known to be the main sound source

in galliforms (Beani et al., 1995).

These effects of androgens on motor systems underlying the production of vocal

signals are not exclusive to birds. Many fish species also use sounds to communi-

cate. Male toadfish are among the most vocal fish, producing loud humming calls

to attract females to their nest site (e.g. plainfin midshipman Porichthys notatus;

Brantley & Bass, 1994). Also in toadfish, the exogenous administration of andro-

gens promotes the development of the sonic muscles, for example the oyster toad-

fish Opsanus tau (Fine & Pennymaker, 1986) and the plainfin midshipman (Brantley

et al., 1993). Another example comes from amphibians, in which vocal behaviour

is sexually dimorphic in most species and thus potentially androgen dependent

(Kelley, 2002). In the African frog Xenopus laevis, males produce mating calls charac-

terized by fast trills that attract females (Wetzel & Kelley, 1983; Kelley, 2002). The

call-production organ of X. laevis is the larynx and all sounds are produced under-

water (Kelley, 2002). The sex differences in vocal behaviour observed in this species



are mostly a result of sex differences in adult laryngeal synapses. Male larynx motor

neurons release less neurotransmitter, which will produce lower postsynaptic po-

tentials than in females, allowing the fibres to reach a spike threshold (Tobias et al.,

1995). This synaptic facilitation in male motor neurons allows modulation of the

amplitude of the trills, a characteristic of the call that is used by females when

assessing the males as potential mates (Tobias et al., 1995). Contrary to most cases

of sexual dimorphism in which the default situation is female, the sex differences

in postsynaptic response emerge in females under the influence of oestradiol,

with the default being the slow neurotransmitter release typical of males (Tobias

& Kelley, 1995). However, other sex differences in this vocal system are androgen

dependent, namely the differentiation of laryngeal motor neurons, muscles fibres

and laryngeal cartilage (Kelley, 2002).

Androgens and pheromone production and/or release

A very large number of mammals use chemical signals (i.e. pheromones)

in intraspecific communication. These pheromones can be produced by specific

scent glands or are released into the environment in the urine or in other body

fluids (Bradbury & Vehrencamp, 1998). Most mammals use marking behaviour to

release these pheromones, a behaviour that is sexually dimorphic ( Johnson, 1973;

Brown & McDonald, 1985; Chs. 11 and 16). There are classic examples of marking

behaviour, such as the scent marking of reindeer Rangifer tarandus, with preorbital,

caudal and tarsal glands as well as with urine (see Brown & MacDonald, (1985)

for other examples and detailed references). Scent marks are also widespread in

rodents such as mice Mus musculus, hamsters Mesocricetus auratus and rats Ratus spp.

(Hurst, 1990 Chs. 11 and 16).

In general, both pheromone production and its release (i.e. scent marking)

are androgen dependent in males, as shown by castration and testosterone-

replacement therapy experiments for example hamsters Mesocricetus auratus

(Gawienowsky et al., 1976), meadow voles Microtus pennsylvanicus (Ferkin & Johnston,

1993), tree shrews Tupaia belangeri (Holst & Eichman, 1998) and Wistar rats Manzo

et al., 2002); however, see Lepri & Randall (1983) and Randall (1986) for an ex-

ception regarding the endocrine control of sandbathing in male kangaroo rats

Dipodomys spp. The scent-marking behaviour decreases after castration and is

restored after treatment with testosterone (e.g. rats: Brown, 1978; Taylor et al.,

1987; Manzo et al., 2002). Interestingly, in many species, testosterone is the pro-

hormone for this effect, because it needs to be metabolized in specific brain

areas into oestradiol or dihydrotestosterone in order to become biologically

active, for example rabbits Oryctolagus cuniculus (Gonzalez-Mariscal et al., 1993),

gerbils Meriones unguiculatus (Yahr & Stephens, 1987) and Wistar rats (Manzo

et al., 2002).


486 R. F. Oliveira

Chemical communication is also widespread in urodeles, playing a major role

in sex recognition and mate attraction (e.g. European newts Triturus spp.: Cedrini

& Fasolo, 1970; Malacarne et al., 1984; Belvedere et al., 1988). One of the best-studied

species is the Japanese red-bellied newt Cynops pyrrhogaster. In this species, males

produce a female-attracting pheromone (sodefrin) with the abdominal glands,

which is released by the cloaca of the male (Kikuyama et al., 1995, 1997). Both

castration and hypophysectomy reduced the sodefrin content of the abdominal

glands and testosterone administration restored it (Yamamoto et al., 1996).

Androgens and visual displays

Many species of vertebrates use complex visual displays in intraspecific

communication, both in the context of conflict resolution (i.e. aggressive displays)

and for mate attraction (i.e. courtship displays) (Bradbury & Vehrencamp, 1998).

The evolution of stereotypic species-specific movements suggests that specific

neuromuscular systems (i.e. motoneurons and their target muscles) may have

evolved specifically for the production of these behaviours.

In some bird species, courtship displays involve coordinated wing and leg move-

ments with the individuals on the ground, on perches or in the air (Schlinger et al.,

2001). These visual displays are usually sexually dimorphic. Because sex steroids,

including androgens, have been shown to play a major role in secondary sex dif-

ferentiation in most vertebrate species studied so far, they are also potential can-

didates for a key role in the control of these displays. In wild golden-collared

manakins Manacus vitellinus, a tropical arena bird, males perform a courtship dis-

play that consists of a sequence of jumps and wing snaps (i.e. upward flips of

the bird’s wings that produce an acoustic signal). The feathers involved in the

production of these wing snaps are the primary and secondary wing feathers

(Schlinger et al., 2001), which are sexually dimorphic (Chapman, 1935). Also the

muscles controlling the wing movements and/or feather position and the jump

often associated with the wing snap are hypertrophied in male manakins (Lowe,

1942). The muscles involved in the wing-snap movement also show sex differ-

ences when examined in more detail (e.g. in fibre diameter, metabolic enzyme

activity and myosin isoform expression), which suggests that they are specialized

for greater force generation and speed of contraction (Schultz et al., 2001). These

sex differences in this neuromuscular system are not present in species in which

males do not use these muscles in courtship displays (e.g. zebra finch), although

they are still functional for other activities (e.g. for raising and lowering of the

wings during flying). These muscles are innervated by motor neurons that accu-

mulate [3H]-testosterone in their soma in the spinal cord, suggesting a role for

androgens in the control of these behavioural mechanisms (Schultz & Schlinger,

1999).



Another example of an androgen-dependent display is the amplexus behaviour

displayed by amphibian males to clasp females during mating. The forelimb mus-

cle involved in this behaviour (i.e. the flexor carpi radiali), is androgen sensi-

tive (Dorlochter et al., 1994). Castration induces atrophy and testosterone treat-

ment of castrated males causes hypertrophy of some regions of this muscle;

immunocytochemistry techniques have identified the presence of androgen re-

ceptors (Dorlochter et al., 1994). Adult males have slower acetylcholine recep-

tor kinetics than females, which facilitates slow and tonic muscle contractions

appropriate for the function of this behaviour (Brennan & Henderson, 1995).

Moreover, testosterone has been shown to act both at the pre- and postsynap-

tic level in these neuromuscular junctions, which may be viewed as an adap-

tation for a more flexible modulation of this behaviour (Nagaya & Herrera,

1995).

Finally in fish, androgens induce the development of somatic structures used in

visual signalling such as the elongation of the dorsal and anal fins used in lateral

displays and the thickening of the jaw used in mouthfighting (e.g. Mozambique

tilapia Oreochromis mossambicus: Oliveira & Almada, 1998).

Androgens and electrocommunication signals

There are two orders of fish that produce weak electric signals with an

electric organ located in their tails: the Gymnotiformes from South America and

the Mormyriformes from Africa (Zakon & Smith, 2002). The evolution of these

weak electric signals most probably occurred independently in the two orders

because they are phylogenetically distant (Alves-Gomes, 1999). Nevertheless, in

both orders, this electric sense is used for the same two functions: electrolocation

(i.e. locating objects in the environment) and intraspecific communication (Zakon

& Smith, 2002). Electrical signals are perceive by the receivers with specialized

electroreceptors mainly located in the midline of the fish (Zakon & Smith, 2002).

There are two types of electric organ discharges: pulse type and wave type. Each

species only produces one or the other (Zakon & Smith, 2002). Within species, there

are marked sex differences in the electric organ discharge. In most Gymnotiform

species that generate wave-type discharges, the males produce signals of lower

frequency than females. For example, in Sternopygus macrurus males produce an

electric discharge of 50–90 Hz while female signals range from 100 to 150 Hz

(Hopkins, 1972). Sex steroids, in particular androgens, seem to be important in

the determination of electric organ discharge frequency. In male S. macrurus, cir-

culating levels of androgens are negatively correlated with frequency (Zakon et al.,

1991) and when their reproductive axis was challenged with human chorionic go-

nadotrophin, they responded with an increase in circulating 11-ketotestosterone

levels and a decrease in the frequency of the discharge from their electric organs


488 R. F. Oliveira

(Zakon et al., 1990). Moreover, treatment of wave gymnotiforms with androgens

induces a masculinization of the waveform (i.e. higher wave frequency and in-

creased duration (Meyer, 1983; Mills & Zakon, 1987; Dunlap & Zakon, 1998)). Inter-

estingly, 11-ketotestosterone increased the frequency of electric organ discharge

(Meyer et al., 1987) in species in which the discharge pattern is sex reversed, that

is males generate higher-frequency discharges than females (e.g. brown ghost,

Apternotus leptorhynchus, Hagedorn & Heiligenberg, 1985). In all pulse-type species,

both mormyriforms and gymnotiforms, the treatment of juveniles, females, cas-

trated males or non-reproductive males with androgens masculinizes the pulse

form (Bass & Hopkins, 1983, 1985; Hagedorn & Carr, 1985; Bass & Volman, 1987;

Landsman & Moller, 1988; Freedman et al., 1989; Landsman et al., 1990; Herfeld

& Moller, 1998). The effects of androgens on the frequency and/or duration of

electric organ discharges may be mediated by their effects on the morphology of

the electric organ (i.e. size and/or shape of electrocytes) or by an influence on the

ionic currents of the electromotor system (e.g. Bass et al., 1986; Bass & Volman,

1987; Mills & Zakon, 1991). Apart from its influence on electric organ discharge

parameters, testosterone also activates the onset of electric signalling in weakly

electric fish (Landsman & Moller, 1988).

Hormonal effects on signal reception

A literature search revealed fewer studies of androgen modulation of

sensory perception than of the effects of androgens on effector mechanisms. The

four studies below are examples of effects on perception.

In many cyprinid fishes, females produce a sex pheromone that elicits male

courtship behaviour. The response of males to the female pheromones can be

measured either behaviourally or electrophysiologically, by placing electrodes

in the olfactory epithelium and measuring the potentials evoked by the expo-

sure of the epithelium to different odorants (i.e. electroolfactograms: Stacey &

Sorensen, 2002). In the tinfoil barb Puntius schwanenfeldi, females release a sex

pheromone (15-ketoprostaglandin-2α) that stimulates male courtship behaviour

(Cardwell et al., 1995). This response is greatest during the breeding season in sexu-

ally mature males; such males have visible breeding tubercules, dermal structures

that are known to be androgen dependent (Smith, 1974). Moreover, juveniles im-

planted with androgens (either 11-ketotestosterone or methyltestosterone) show

both an increased electroolfactogram response to 15-ketoprostaglandin-2α and

increased sexual behaviours directed towards stimuli fish (i.e. juveniles injected

with 15-ketoprostaglandin-2α (Cardwell et al., 1995)). These results clearly demon-

strate a peripheral effect of androgens on olfactory sensitivity. Other species also

show increased olfactory sensitivity to such stimuli during the breeding season

when androgen levels are also higher (e.g. electroolfactogram responsiveness to



testosterone in the Atlantic salmon Salmo salar (Moore & Scott, 1991)), which

suggests that the effect described above may be a general phenomenon in fish

olfaction.

In addition, electroreception in weakly electric fish seems to be modulated

by androgens (Keller et al., 1986; Sisneros & Tricas, 2000). Testosterone not only

affects the frequency of discharge from electric organs (as described above) but

also shifts the maximum receptivity of the electroreceptor to the new frequency

produced (Meyer & Zakon, 1982; Bass & Hopkins, 1984). Thus, androgens keep

the electroreceptors of a given individual fine-tuned to its own electric organ

discharge, which might be viewed as an adaptation for electrolocation.

A third example comes from studies of auditory sensitivity in the plainfin mid-

shipman. As mentioned above, in this species type I males produce a humming

call during the breeding season that is used to attract spawning females to their

nests (Ibara et al., 1983; Brantley & Bass, 1994). Male reproductive success must

depend heavily on their calling behaviour because females are choosy regarding

call parameters of the ‘hum’ signal (McKibben & Bass, 1998). Female reproductive

success is also expected to depend on their ability to locate and choose males

based on their acoustic signals. Recently, it has been demonstrated that, during

the summer when females need to exert their mate choice preferences based on

the male call, the auditory saccular units in the females increase their temporal

encoding capacity up to 340 Hz, compared with only 100 Hz in winter females

(Sisneros & Bass, 2003). This seasonal plasticity of the peripheral auditory system

is most probably driven by sex steroids, because it follows the seasonal variation

in steroid profiles (Forlano et al., 2003) and because expression of the oestrogen

receptor β has been identified recently in auditory hair cells (P. M. Forlano & A. H.

Bass, unpublished data). Therefore, an increase in sex steroids at the beginning of

the breeding season may induce changes in the frequency sensitivity of these hair

cells in a similar way to androgen-dependent changes in electroreceptor tuning

described above.

Finally, there are suggestions that sex steroids may also be involved in the

modulation of visual perception in teleost fish. In the three-spined stickleback Gas-

terosteus aculeatus, sexually active females prefer to mate with males with redder

bellies (e.g. Milinski & Bakker, 1990). Using optomotor responses, Cronley-Dillon &

Sharma (1968) have demonstrated that the sensitivity of the female visual system

to red wavelengths increases during the breeding season, suggesting a potential

role for female sex hormones. In this example, it can be argued that the effect

found could be acting either at the level of the sensory organ or at the level of

visual information processing by the central nervous system (i.e. optic tectum). In-

terestingly, aromatase activity has been found in fish retina, indicating that these

cells are actively metabolizing sex steroids (Callard et al., 1993) and supporting


490 R. F. Oliveira

the idea that the steroid modulation of visual sensitivity to key colours may occur

in the periphery.

These four studies taken together suggest that sex steroid modulation of sen-

sory perception is a common phenomenon in different sensory modalities.

Hormonal modulation of motivational and memory mechanisms

Androgens can also affect central mechanisms of information processing

both in senders and receivers. At this level, the modulatory action of hormones

may affect signalling behaviour by acting either on motivational neural circuits

underlying decision-making mechanisms, or on learning and memory systems

(Schulkin, 2002; Dohanich, 2002). By acting on central mechanisms, androgens

may set up the subject to perceive stimuli and to behave in particular ways,

for example by increasing the likelihood of the expression of a given behaviour,

ranging from food ingestion to maternal behaviour or aggression. For example,

androgens modulate central mechanisms of chemical perception in male ham-

sters. In this species, vaginal secretions stimulate male sexual behaviour after

male anogenital investigation of the female ( Johnston, 1975; Ch. 16). These se-

cretions are detected by two different sensory systems, the olfactory mucosa and

the vomeronasal organ, that use different neural pathways converging in three

central areas: the medial nucleus of the amygdala, the bed nucleus of the stria

terminalis and the medial preoptic area (Scalia & Winans, 1975). Androgen re-

ceptors are found in all these three areas (Wood et al., 1992) and direct androgen

implantation here restores sexual behaviour in castrated males (Lisk & Bezier,

1980).

Usually the effects of steroids, including androgens, on motivational mecha-

nisms involves the regulation of neuropeptide gene expression in the limbic sys-

tem, namely of arginine-vasopressin (or its homologue arginine-vasotocin in non-

mammalian vertebrates), which subsequently influence central states that con-

trol the behavioural output (Herbert, 1993). There are numerous examples of this

principle. In hamsters, testosterone enhances the effects of arginine-vasopressin

infused in the bed nucleus of the stria terminalis on scent-marking behaviour

(Albers et al., 1988). In male prairie voles, testosterone also promotes the expres-

sion of parental behaviour by increasing arginine-vasopressin synthesis and by

preventing the apoptosis of responsive neurons (De Vries, 1995). Finally, in amphib-

ians, sex steroids control both female egg-laying behaviour and male courtship via

arginine-vasotocin modulation (Moore et al., 1992).

The potential effects of sex steroids on learning, memory and other cognitive

functions have been addressed using two main approaches: (a) by documenting

the distribution of androgen and oestrogen receptors in brain areas known to

be involved in these functions, and (b) by testing hormone-treated subjects in



cognitive tasks. There is a much larger body of literature on oestrogens than on

androgens regarding this topic. The available data on androgens will be summa-

rized below.

Androgen receptors are found in the hippocampus of mammals and birds (Kerr

et al., 1995; Saldanha et al., 1999) and in the homologue dorsolateral telencephalon

of fish (Northcutt & Davis, 1983; Gelinas & Callard, 1997). These brain areas are

involved in relational memory processes, namely in spatial memory (Eichenbaum

et al., 1992; Squire, 1992). Androgen receptors have also been found in pyramidal

cells of the cortex in rats, monkeys and humans (Pomerantz & Sholl, 1987; Kerr

et al., 1995; Tohgi et al., 1995). These results set the stage for a potential functional

direct effect of androgens on memory mechanisms. The occurrence of oestrogen

receptors together with aromatase (an enzyme that metabolizes androgens into

oestrogens) also suggests a potential alternative route for aromatizable androgens

to affect cognitive function (e.g. Gelinas & Callard, 1997).

There are numerous examples of sex differences in spatial memory tasks, with

males outperforming females, which suggests a role for sex steroids in spatial

memory mechanisms (reviewed by Dohanich, 2002). Early androgen exposure ap-

parently has organizational effects on adult spatial abilities, and the masculiniza-

tion of spatial learning involves the aromatization of androgens into oestrogens

in rodents (Williams et al., 1990; Roof & Havens, 1992; Roof, 1993). In humans,

early exposure to androgens masculinizes spatial function, as is suggested by

data on girls suffering from congenital adrenal hyperplasia. These girls are ex-

posed to androgens in utero as a result of hypertrophy of the adrenal glands and

are born with virilized genitalia. When compared with their unaffected sisters,

girls with congenital adrenal hyperplasia have better performances in mental

object-rotation tests designed to measure spatial ability (Resnick et al., 1986; see

Kimura (1996) for further references). In adults, the relationship between circu-

lating androgen levels and spatial ability is not linear. Lower testosterone levels

in males, and higher testosterone levels in females, are associated with better

performances in an object-rotation task, which suggests an optimum circulating

level of testosterone to excel in this task (Moffat & Hampson, 1996). As regards

other cognitive mechanisms, in general the administration of androgens to birds

and mammals outside the critical period of development fails to affect learn-

ing and memory tasks (Dohanich, 2002). However, social memory is an exception

to this rule in rats and zebra finches (Sawyer et al., 1984; Cynx & Nottebohm,

1992).

Hormones and somatic releasers

There are a number of somatic structures that act as sign stimuli (sensu

Tinbergen, 1951) evoking a behavioural response in conspecifics. The classic


492 R. F. Oliveira

example of these releasers is the red belly of the male three-spined stickleback,

which elicits aggressive responses in other male sticklebacks (Tinbergen, 1951).

Since initially proposed by Tinbergen, these social releasers have been described

in many other species and can range from nuptial colouration patterns in

fish and birds to dermal appendages in fish (e.g. dermal tubercules), birds (e.g.

combs, elongated tail feathers) and reptiles (e.g. dewlap membrane in Anolis

spp.). The development of at least some of these somatic structures with a re-

leaser function is under hormonal control. There are various examples in the

teleosts. First, male nuptial colouration in African cichlids is suppressed in cas-

trated males and restored in castrates and females by exogenous administration

of testosterone (Levy & Aronson, 1955; Reinboth & Rixner, 1972; Wapler-Leong &

Reinboth, 1974; Fernald, 1976). Also in male sticklebacks, the nuptial colouration

can be suppressed by castration (Ikeda, 1933) or by the exogenous administration

of an anti-androgen (cyproterone acetate) (Rouse et al., 1977). Finally, in the sex-

role-reversed peacock blenny Salaria pavo, in which some ‘sneaker’ males mimic

female nuptial colouration, androgens (i.e. 11-ketotestosterone) inhibit the ex-

pression of female nuptial colouration in these sneaker males (Oliveira et al.,

2001a).

However, nuptial colouration is not the only releaser to be androgen depen-

dent in fish. The development of the sword as an extension of the caudal fin in

male swordtail fish Xiphophorus helleri and the development of the dermal breed-

ing tubercules in male cyprinids are both also induced by testosterone (Baldwin

& Goldin, 1939; Smith, 1974). Therefore, another way for hormones to affect com-

munication is by affecting the expression of somatic releasers in senders.

Social modulation of androgen levels

As shown above, androgens can be viewed, on the one hand, as causal

agents of behaviour, including signalling behaviour among animals in a commu-

nication network. On the other hand, the endocrine system is responsive to the

network of social relationships in which the animal is involved. Several studies

have shown the effects of social interactions on the short-term modulation of

androgen levels. In the early 1940s, it was established that male mice that lost

an agonistic interaction had lower levels of androgens than winners (Ginsberg &

Allee, 1942). This pattern has been found repeatedly in other vertebrate taxa from

fish (e.g. Hannes, 1984, 1986) to primates, including humans (e.g. Rose et al., 1971,

1975; Bernstein et al., 1974; Booth et al., 1989; see Mazur & Booth, 1998 for more

references). This set of results led to the proposal of the ‘challenge hypothesis’ by

John Wingfield and co-workers (Wingfield, 1984; Wingfield et al., 1987, 1990), ac-

cording to which the social interactions involving the subject determine androgen

levels. This hypothesis gives a conceptual framework for the study of the interplay



between social factors and endocrine responses and generates a number of testable

predictions.

1. Androgen levels should be higher during periods of social instability

when social interactions are more frequent and more intense. In fact, in

bird species in which a clear breeding cycle can be recognized,

testosterone levels are higher during the period of territory

establishment than when territories are established (Hegner &

Wingfield, 1987a; see Wingfield et al., 1999, 2000 for more examples).

2. Territorial and dominant males are expected to show higher androgen

levels than non-territorial or subordinate males because territorial males

have to defend their territories from intruders and dominant males

have actively to maintain their status. Again the available evidence

supports this hypothesis (e.g. see Oliveira et al. (2002) for a review of

teleost fish and Wingfield et al. (1999, 2000) for reviews of birds).

3. Populations of the same species breeding under different

population-density regimes should also show differences in the average

androgen levels of breeding males as a result of a different probability of

territory intrusions. This prediction should be taken with caution

because in a population with increased density, physiological and/or

behavioural mechanisms may be present to avoid aggression.

Nevertheless, positive correlations have been found between density of

breeding territories and androgen levels both in fish and in birds (e.g.

Ball & Wingfield, 1987; Beletsky et al., 1990, 1992; Pankhurst & Barnett,

1993).

Interestingly, during periods of social inertia, the levels of social interaction fall

to a baseline and androgen levels become decoupled from social behaviour. These

results have been interpreted as an adaptation (or an exaptation sensu Gould &

Vrba (1982), depending on the underlying historical evolutionary pathway) for

the individuals to adjust their behaviour (motivation) to the social milieu that

they are currently experiencing. Thus, social interactions would stimulate the

production of androgens and androgen levels would be a function of the stability

of the social environment in which the animal is living (Wingfield et al., 1990,

1999, 2000; Oliveira et al., 2002).

It is interesting to note here that it is the perception that the individual has

of the interaction in which it is involved or which it is observing that activates

the endocrine response and not the objective structure of the situation per se. To

investigate this idea we have recently tested the effect of mirror-elicited aggres-

sion on androgen levels in a cichlid fish (L. Carneiro & R. F. Oliveira, unpublished

data). The mirror image stimulation test is widely used in fish ethology to assess


494 R. F. Oliveira

aggressiveness (Rowland, 1999), but some inconsistencies have been found in the

relationship between social status and the aggressive score of an individual in

this test (Ruzzante, 1992). In Mozambique tilapia, we showed that androgen levels

before the fish were grouped were not good predictors of social status, but andro-

gen levels at the end of the time spent in a group were highly correlated with the

social status of each individual, suggesting that androgens are being modulated

by the social interactions experienced by the grouped individuals (Oliveira et al.,

1996). In the mirror image stimulation test, the individual is placed in a very pe-

culiar situation. Because fish do not recognize as themselves the image reflected

by the mirror, they respond to it as an intruder and attack. In our experiment,

males reacted aggressively to their own images in the mirror and escalated the

interaction using more overt aggressive behaviours (e.g. biting) as time went by.

However, because the mirror reflects exactly the same behaviours that the experi-

mental fish is displaying, the interaction has no outcome (winning versus losing).

Therefore, if the endocrine response to the social interaction is triggered by the

behavioural output during the interaction (e.g. number of displays or time spent

displaying), a variation in androgen levels is predicted. However, if it depends

on behavioural feedback received from the opponent, then no androgen varia-

tion is predicted in the test. In our mirror image stimulation experiment with

Mozambique tilapia, we found a strong behavioural response but a complete lack

of an androgen response (L. Carneiro & R. F. Oliveira, unpublished data), which

suggests that the endocrine system responds to a clear perception of the outcome

of the social interaction. This result is also interesting because it shows that it is

the communication component of the social interaction that may affect hormone

levels.

Hormones and communication II: the network view

In the previous section, the interrelationship between hormones (i.e. an-

drogens) and social behaviour was considered at the dyadic level. First, the mech-

anisms through which androgens may affect communication between a pair of

individuals were described. Second, the ways in which these androgen levels might

be modulated by the social environment (network) in which the animal is living

were described.

If we now consider an interaction that occurs within a social network (or a

communication network sensu (McGregor, 1993)), with possibilities for other in-

dividuals in the network to eavesdrop on the interaction (e.g. Ch. 2) and for the

interacting pair to adjust their behaviour according to the presence of an audi-

ence (Ch. 4), the complexity of the interrelationship between hormones and be-

haviour/communication mechanisms could increase substantially. The presence



Input/perception

Motoroutput

Centralprocessing

Hormones

Receiver

Signal

Input/perception

Motoroutput

Centralprocessing

Hormones

Somatic releasers

Sender

Input/perception

Motoroutput

Centralprocessing

Hormones

Bystander

Fig. 21.3. Flow of information in a communication network. This figure is similar to

Fig. 21.2 with the following extra elements. The presence of a third individual (the

bystander) may be perceived by both the sender and the receiver and affect their signal

production and signal reception mechanisms, respectively. The perception of the

presence of the bystander may also affect central (motivational) mechanisms in the

central nervous system in both senders and receivers, which may, in turn, modulate

hormone levels in both individuals. The perception of the signal by the bystander

would also affect its central processing of information at the level of motivational

mechanisms, which could affect its hormone levels.

of a bystander that could act both as an eavesdropper and as an audience (Fig. 21.3),

may affect androgen levels in both the sender and the receiver and subsequently

affect their androgen-modulated communication behaviour in the same ways as

described above (see also Fig. 21.2). The androgen levels of the bystander itself may

respond to the observed interaction, which, in turn, will affect its own subsequent

social behaviour.

Consequently, androgens may play a key role as physiological mediators of the

modulation of behaviour by the social context. A number of social phenomena (e.g.

winner–loser effects, bystander effect, dear enemy effect) that have been described

in social networks may then be physiologically mediated by changes in hormone

levels, especially androgens.


496 R. F. Oliveira

Adjusting behaviour to the social context: a role for androgens?

Winner–loser effects

There is an extensive literature (including most of the chapters of this

book) that clearly shows that animals use information on relative competitive

abilities in the social network in which they are placed to adjust their behaviour

accordingly. They may obtain this information by direct assessment of their peers

by interacting with one another in a dyadic fashion and then adjusting their

behaviour in subsequent interactions depending on the outcome of previous in-

teractions. For example, individuals that win an interaction increase their prob-

ability of winning a subsequent interaction and vice versa for losers. In this case,

although only two individuals have to be present during the initial interaction,

unless there were other individuals with whom the interactants subsequently

interacted, there would be no winner–loser effect. Therefore, this effect is better

understood within the framework of social networks than with a dyadic approach.

This winner–loser effect may last from a few minutes up to several hours or even

days and has been reported for several taxa, for example invertebrates (Alexander,

1961; Otronen, 1990; Whitehouse, 1997), fish (McDonald et al., 1968; Frey & Miller,

1972; Bakker & Sevenster, 1983; Francis, 1983, 1987; Abbott et al., 1985; Beaugrand

& Zayan, 1985; Beacham & Newman, 1987; Franck & Ribowski, 1987; Beacham,

1988; Bakker et al., 1989; Beaugrand et al., 1991, 1996; Chase et al., 1994; Hsu &

Wolf, 1999), reptiles (Schuett, 1997) and birds (Drummond & Osorio, 1992). The

winner effect is usually of shorter duration than the loser effect (e.g. Chase et al.,

1994), and when integrating prior social experiences more recent outcomes are

more effective in predicting the probability of winning a subsequent interaction

than previous ones (Hsu & Wolf, 1999). Another interesting characteristic of this

effect is that it is more effective when winning or losing against a well-matched

opponent than when there is a large asymmetry in resource-holding potential

(sensu Parker, 1974) between the two individuals (Beaugrand & Goulet, 2000).

The behavioural mechanism proposed to explain the winner effect is based on

the fact that initiators of interactions have higher probabilities of winning and that

winners of recent encounters become more likely to initiate future interactions

( Jackson, 1991). This is especially true for the initiators of attacks (Hsu & Wolf,

2001).

It is conceivable that by winning an interaction an individual raises its androgen

levels, which, in turn, increases its willingness to initiate future interactions and

the probability of winning the next interaction in which it participates. The reverse

would be predicted for losers.

We are conducting an ongoing literature survey to collect data on the two

steps of this endocrine hypothesis for the winner–loser effect: (a) that winners



Table 21.1. Literature survey of reported differences in male androgen levels among

vertebrates according to social status and the phase of the sexual cycle and of the effects of

androgen treatment on aggressive behaviour

Taxa Androgens and Androgens and Effect of androgen

social status phase of treatment on

sexual cycle aggressive behaviour

D = S or MP < PP or No

No. D > S D < S No. MP > PP MP = PP No. Effect effect

Fish 12 9 3 9 8 1 12 7 5

Amphibians 1 0 1 2 2 0 0 – –

Reptiles 3 2 1 0 – – 3 3 0

Birds 10 4 6 48 47 1 10 6 4

Mammals 18 15 3 4 3 1 9 5 4

Total 44 30 14 63 58 5 34 21 13

Male androgen levels in; D, dominant; S, subordinate; MP, mating phase; PP, parental care phase.

K. Hirschenhauser & R. F. Oliveira, unpublished data.

have higher androgen levels than losers; (b) that androgens increase aggressive

behaviour and hence the probability of victory in a subsequent interaction. As

there are not enough studies that we can find that measured the androgen vari-

ations in response to a social interaction to address the first step, it was decided

to search for correlational data in the form of reported androgen differences be-

tween dominant and subordinate individuals. We found 44 published studies,

68% of which confirmed that androgen levels were higher in dominants than in

subordinates (Table 21.1). Our literature survey revealed that 62% of the studies

confirmed that administration of androgens increased aggressive behaviour in

different taxa (Table 21.1), thereby supporting the second step of the endocrine

hypothesis. Although the majority of the studies supported the assumptions of the

proposed hypothesis, the percentages do not provide overwhelming support and

so we decided to test this hypothesis experimentally with the Mozambique tilapia

(A. Silva & R. F. Oliveira, unpublished data). After staging a first fight between two

males, the winner and the loser fought two independent, naive individuals (i.e.

males that have not been involved in social interactions recently) (Fig. 21.4a). As ex-

pected, our preliminary data showed that winners of the first encounter won the

majority of the interactions with the naive fish and vice versa for losers (Fig. 21.4b).

When winners were treated with an anti-androgen (cyproterone acetate) between

the two interactions (which were two hours apart), the winner effect was no longer

detectable in the second fight with the neutral fish, suggesting an involvement


498 R. F. Oliveira

0102030405060708090

100

Winner Loser

control

Per

cent

age

treated

(a) (b)

Winner LoserNeutralmale

Neutralmale

t1

t2t2

Fig. 21.4. Androgens and the winner–loser effect. (a) Experimental-set up: four

Mozambique tilapia Oreochromis mossambicus males were introduced to individual

compartments separated by opaque partitions in an aquarium. After a period of

acclimation (t1), the central partition was removed and the two individuals in the

central compartments were allowed to interact until a winner and a loser could be

recognized. Then the partition was put back in place. Two hours later (t2), the two

lateral partitions were simultaneously removed and the winner and the loser of the

previous interaction were allowed to interact with the neutral males that were placed

in the end compartments. This second interaction went on until a winner and a loser

could be recognized. (b) Percentage of second interactions won by winners or losers of

the first interaction. Four groups of experimental animals were compared: t1 winners

treated with an androgen inhibitor (cyproterone acetate); control t1 winners (treated

with a placebo saline solution); t1 losers treated with an androgen

(11-ketotestosterone); control t1 losers (treated with a placebo saline solution). Twelve

replicates of the experiment were run. (A. Silva & R. F. Oliveira, unpublished data.)

of androgens in the winner effect (Fig. 21.4b). However, the loser effect was not

inhibited in the second interaction by treating losers with exogenous androgens,

which suggests that, although a fall in androgens is observed in losers, it is not

the underlying causal mechanism for the loser effect. Although this result may

seem paradoxical at first sight, it makes some sense; androgen variations induced

by social interactions occur in the short term and so do winner effects; however,

the loser effect may last up to several days depending on a number of factors.

Consequently, other neuroendocrine mechanisms must be involved in the loser

effect. One of the best candidates for this role is the serotonergic system. The fol-

lowing evidence from studies using different fish species seems to support this

hypothesis: (a) losers experience increased brain levels of serotonin and subordi-

nate individuals have chronically elevated brain levels of serotonin (Winberg &



Nilsson, 1993a,b; Winberg et al., 1997; Winberg & Lepage, 1998); and (b) serotonin

appears to be inhibitory to behavioural responsiveness in general and to inhibit

aggressive behaviour in particular (Winberg & Nilsson, 1993a,b; Adams et al., 1996;

Edwards & Kravitz, 1997). Therefore, losers would display a marked behavioural in-

hibition, with increased attack latencies in subsequent interactions, which would

prevent them from winning these interactions and would reinforce their subordi-

nate role. Interestingly, the administration of a precursor of dopamine (L-dopa) to

individuals that had lost an interaction two days before induced lower serotoner-

gic activity and reduced the attack latency in subsequent interactions, suggesting

that the dopaminergic system counteracts the serotonin-mediated effects of social

subordination (Hoglund et al., 2001).

Bystander effects

Information on the relative competitive ability of conspecifics within a

social network can also be gathered using indirect methods, namely by extract-

ing information from watching conspecific interactions that the subject uses in

subsequent interactions with the observed individuals (eavesdropping (McGre-

gor, 1993; McGregor & Peake, 2000) and social eavesdropping sensu Peake, Ch. 2).

This sort of information gathering on the relative ability of conspecifics has been

demonstrated in a number of species (see Ch. 2), for example fish ( Johnsson &

Åkerman, 1998; Oliveira et al., 1998; Earley & Dugatkin, 2002; Ch. 5) and birds

(Hogue et al., 1996; McGregor et al., 1997; Naguib et al., 1999; Peake et al., 2001),

and has the advantage of avoiding the costs associated with fighting (e.g. McGre-

gor, 1993; McGregor & Peake, 2000; Dugatkin, 2001). Some authors consider that

there is a difference between bystander and eavesdropping effects: eavesdropping

implies an active gathering of information by bystander individuals that will be

used in future interactions within the social network (McGregor, 1993; McGregor

& Peake, 2000), while the bystander effect was originally described as a priming

of aggressive motivation in bystanders of agonistic interactions (Hogan & Bols,

1980; Bronstein, 1989). Therefore, from the point of view of the required cognitive

abilities, eavesdropping is expected to be more demanding than a mere priming

response. However, both phenomena are adaptive because they might increase the

probability of eavesdroppers/bystanders of winning their next social interaction

(Clotfelter & Paolino, 2003; Hollis et al., 1995; Peake & McGregor, 2004).

The priming response associated with the bystander effect is another phe-

nomenon that could be mediated by androgens. To investigate if bystanders

experience an increase in their androgen levels, we conducted an experiment

with Mozambique tilapia in which a bystander fish had visual access through a

one-way mirror to two conspecific neighbours separated by an opaque partition


500 R. F. Oliveira

Male A

Male C

One way mirror

−12−10−8−6−4−2

0246

30 min 2 h 6 h

Time after exposure to stimuli

11-K

eto

test

ost

ero

ne

vari

atio

n

(ng

/ml)

p < 0.05

p < 0.01p < 0.01

Male B

(a)

(b)

Fig. 21.5. Social modulation of androgen levels in bystander male Mozambique tilapia

Oreochromis mossambicus. (a) Experimental-set up: three males were introduced to

individual compartments in an aquarium. Males A and B were separated by an opaque

partition and male C was separated from the other two males by a one-way mirror

that allowed it to observe the other two males without being observed. After a period

of acclimation, two conditions were created. (i) In the experimental group, the opaque

partition separating males A and B was removed (grey arrow) and the two individuals

were allowed to interact for 20 minutes while male C observed the interaction

(bystander). (ii) In the control group, the opaque partition separating males A and B

was not removed and the bystander individual observed its two neighbours resting or

swimming around for 20 minutes. Urine samples were collected from male C at

regular intervals (just before the start of the test, 30 minutes after, two hours after

and six hours after the experiential situation) and assayed for androgens using

radioimmunoassays. (b) Androgen (11-ketotestosterone) variation (i.e. urine

concentrations after the experiential situation minus the urine concentrations just

before the experiential situation) in bystander males of the experimental (black bars)

and control (white bars) groups. (Adapted from Oliveira et al., 2001b.)

(Fig. 21.5; Oliveira et al., 2001b). After a period of familiarization, in the exper-

imental treatment the opaque partition between neighbours was removed and

the bystander was allowed to observe the agonistic interaction between its neigh-

bours. In the control group after the same period of familiarization, the opaque

partition between neighbours remained in place and the bystander could see its



two neighbours resting or swimming around in their respective compartments. As

predicted, androgen levels (both 11-ketotestosterone and testosterone) increased

significantly in the experimental group of bystanders after watching their neigh-

bours fighting and no effect was detected in the control group (Oliveira et al.,

2001b). This result has an interesting parallel in humans. It has been demon-

strated that sports fans experience variations in testosterone levels depending on

the outcome of the game they have attended, both for college basketball and for

soccer. Fans of the winning team display an increase in salivary testosterone levels

and there is a decrease in testosterone levels in fans of the losing team (Bernhardt

et al., 1998).

Audience effects

The term audience effect was first used in the ethological literature to de-

scribe the facilitation effect of the presence of other individuals on the production

of food calls or alarm calls in response to food items or a predator, respectively

(Gyger et al., 1986; Marler et al., 1986; Evans & Marler, 1994). Here the term will be

used in a more restricted way, following the definitions provided by McGregor &

Peake (2000) and by Matos (2002) (see also Ch. 4): individuals participating in an in-

teraction may also manipulate the information available to others and adjust their

signalling behaviour according to the presence and composition of an audience

of conspecifics. These audience effects have been demonstrated in different ver-

tebrate taxa, including fish (e.g. Doutrelant et al., 2001; Matos & McGregor, 2002),

birds (e.g. Searcy et al., 1991; Baltz & Clark, 1997) and mammals (e.g. Hector et al.,

1989), and have involved different social contexts, from agonistic interactions in

Siamese fighting fish Betta splendens (Doutrelant et al., 2001; Matos & McGregor,

2002) to extra-pair copulations in male budgerigars Melopsittacus undulatus (Baltz &

Clark, 1997).

The audience effect has been interpreted as a way for the individual to ma-

nipulate the information broadcast to its social network, which may influence

subsequent social interactions in which it will have to participate. Therefore, it

can be predicted that subjects behave more promptly and aggressively towards

an intruder when a male audience is present. Again it is predicted that this ef-

fect may be mediated by increased androgen levels in the interacting individuals

induced by the presence of the audience. This aggressive priming effect of an au-

dience has already been established in Siamese fighting fish (Matos, 2002) but its

androgen-mediation remains to be tested.

Dear enemy effects

In territorial systems, residents react less aggressively towards famil-

iar opponents than to intrusions by strangers, a phenomenon called the dear


502 R. F. Oliveira

enemy effect (Ydenberg et al., 1988; Temeles, 1994). In evolutionary terms, this

phenomenon can be viewed as an adaptation for the individual to adjust its ter-

ritorial behaviour according to the threat posed by the intruder (Temeles, 1994):

having a dear enemy neighbour allows the resident individual to defend its ter-

ritory against unfamiliar intruders with the same efficiency as if they were the

only competitors in the area, which reduces the costs of territory defence (Leiser

& Itzkowitz, 1999; Whiting, 1999).

The dear enemy phenomenon can be explained in terms of proximate mech-

anisms by an ability of the resident male to discriminate between familiar and

unfamiliar intruders together with a habituation to the neighbours, which would

explain the lower response that they elicit, for example visual habituation to neigh-

bours in Siamese fighting fish (Bronstein, 1994) and habituation to neighbours’

calls in frogs (Owen & Perrill, 1998). Therefore, it can be predicted that resident

males will react more aggressively towards strangers than towards familiar intrud-

ers and that the increase in androgen levels expected from the social challenge

experienced will be higher in the case of intrusions by strangers. Moreover, it is

also predicted that, for repeated intrusions by neighbouring males, the androgen

response should be higher in the first trials and decrease with the number of trials

(i.e. habituation). These two predictions remain to be tested.

The adaptive value of social modulation of hormones: a cost–benefit

analysis of androgen levels

As stated above, the main adaptive reason for androgens to respond to

the social environment is to allow individuals to fine-tune the expression of their

behaviours in a context-dependent fashion. For example, this mechanism would

allow subordinate individuals to downregulate the expression of their aggressive

behaviour and thus avoid the initiation of agonistic encounters that they have

low probabilities of winning. In the long run, this mechanism can be seen as

an opportunity for individuals to adopt a behavioural tactic that suits best their

relative competitive ability. As a result, androgen-mediated behavioural tuning

to the social environment may result in either a continuous or a discrete vari-

ation of behavioural phenotypes. For example, even small changes in androgen

levels induced by social interactions in electric fish, can affect the pulse duration,

resulting in dominant males with more masculinized discharges from their elec-

tric organs than subordinates (e.g. Brienomyrus brachyistius: Carlson et al., 2000).

Also, in the Mozambique tilapia, the acquisition of dominant status induces the

exaggeration of male morphological traits, an effect that has been shown to be

mediated by androgens (Oliveira & Almada, 1998). By comparison, in a number



of teleost species, individuals of lower competitive ability can adopt frequency-

or condition-dependent alternative reproductive tactics (Taborsky, 1994) or even

change sex (Grober, 1998).

However, it can be argued that, instead of having their androgen levels open to

social influences, selection could have favoured animals that permanently keep

their androgen levels at an optimum high value in order to optimize their social

behaviour at all times. It follows that there must be costs associated with maintain-

ing high levels of androgens that counteract the social benefits of high androgen

levels. Therefore, a cost–benefit analysis is needed to establish the adaptive value

of the social modulation of androgens.

Potential benefits of high androgen levels

Among the potential benefits of increasing androgen levels at periods of

social challenge, one can think of androgen effects both on aggressive motivation

and on cognitive tasks that would promote the success of the animals in social

interactions.

The available data on the effects of androgens on aggressive motivation has

already been review above, and in most studies an effect has been found (Ta-

ble 21.1). Sex steroids, including androgens, are known to play a major role in

cognitive processes such as social attention, learning and memory in a variety

of vertebrate taxa (e.g. Andrew, 1991; Cynx & Nottebohm, 1992) and so they may

help the animal to be prepared for a competitive context (see text above for more

references).

We have recently tested the effects of androgens on social attention in Siamese

fighting fish. Eavesdropping has already been demonstrated in this species and

male Siamese fighting fish are known to spend time observing conspecific inter-

actions (Oliveira et al., 1998). So we designed an experiment to assess the effect of

the administration of exogenous androgens on the time males spend observing

social interactions between conspecific males. Not surprisingly, androgen-treated

males spent more time observing social interactions than controls, suggesting an

effect of androgens on selective attention to the social environment (R. F. Oliveira

& L. Carneiro, unpublished data).

Another potential benefit that androgens may convey in a competitive situa-

tion is an increased probability of the expression of risk-taking behaviours, which

might be adaptive in a competitive situation. A nice example of this phenomenon

has recently been published (Kavaliers et al., 2001). Male mice were pre-exposed

to the odour of an oestrous female and subsequently exposed to the odours of

predators (cat and weasel). Mice that were only exposed to the predator odour,

simulating a situation of increased predation risk, showed increased circulating


504 R. F. Oliveira

levels of corticosterone and decreased levels of testosterone. The pre-exposure to

the female odour attenuated this response to predator odour, which might reflect

a greater tendency for risk taking in the presence of predators (Kavaliers et al.,

2001).

Potential costs of high androgen levels

Elevated androgen levels have been shown to have associated costs;

consequently, one would expect high circulating levels to be restricted to

periods of social challenge. The following potential costs have been dis-

cussed in the literature: (a) increased energy consumption; (b) impairment of

immunocompetence; (c) higher rates of injuries and reduced survival; (d) inter-

ference with parental care; and (e) potential oncogenic effects (Wingfield et al.,

1999). Of all these potential costs, two will be analysed in more detail below:

the potential negative effects of androgens on metabolic rate and on parental

behaviour.

Metabolic costs of high androgen levels

Studies of the metabolic effects of androgens have produced contradic-

tory results. In bird species, testosterone treatment increased the basal metabolic

rate in house sparrows Passer domesticus (Buchanan et al., 2001), reduced it in white-

crowned sparrows Zonotrichia leucophrys (Wikelsky et al., 1999) and had no effect in

dark-eyed juncos Junco hyemalis (Deviche, 1992). However, in juncos, an indepen-

dent study found an association between high testosterone levels and increased

lipid catabolism and nocturnal body temperature (Vezina & Thomas, 2000). In the

lizard Sceloporus jarrovi, testosterone treatment increased the maximal metabolic

rate but had no effect on basal metabolic rate (Marler et al., 1995) and male tilapia

treated with 11-ketotestosterone showed an increase in the resting metabolic rate

and in metabolic scope but a non-significant increase in the basal metabolic rate

(Ros et al., 2004). This discrepancy in the results can be attributed to methodolog-

ical variations among studies, including the choice of the measures taken, the

timespan of the experiment, the season, etc. Nevertheless, androgens failed to

affect the metabolic measure used in only two studies, and in one case the data

are contradicted by a subsequent study on the same species. In the other four

studies, androgens affected different metabolic measures. Consequently, it can be

said that androgens may affect metabolism in a non-linear way and a metabolic

cost associated with higher levels of androgens should not be excluded.

Parental care trade-off with androgens

One of the predictions of the challenge hypothesis is that male andro-

gen levels above a breeding baseline are incompatible with male parental care



(Wingfield et al., 1990). If androgen levels increase as a result of social challenges,

males will invest less time in paternal activities, and thus a trade-off between

social interactions and paternal care, mediated by androgens emerges. In many

bird species with male parental care, the experimental increase of circulating

testosterone in parental males suppressed paternal behaviour and promoted ago-

nist interactions (Silverin, 1980; Hegner & Wingfield, 1987b; Ketterson et al., 1992;

Beletsky et al., 1995). Moreover, several studies on the seasonal variation of andro-

gen levels in birds show that during the breeding season male androgen levels

are higher during the mating phase than during the parental phase (Wingfield et

al., 1987). To document this trade-off further, we have gathered published data on

androgen levels in vertebrate species with respect to paternal care: out of the 63

species of vertebrates for which data are available, 92% show the expected pattern

of lower circulating androgen concentrations during the parental phase (Table

21.1). In summary, keeping high levels of androgens at all times is detrimental to

the individual in many ways and so the stage is for the evolution of a flexible

system modulated by the social environment.


Androgen modulation by social context and the subsequent role of an-

drogens in the activation of expression of social behaviour have been proposed in

this chapter to explain the mechanisms underlying experiential effects. However,

this hypothesis does not exclude explanations of the phenomena described, in

terms of associative learning mechanisms. Cognitive abilities such as individual

recognition and discrimination would explain some of the described behavioural

responses to social context (e.g. McDonald et al., 1968) and winning or losing can

be seen as having reinforcing properties. For instance, male Siamese fighting fish

will perform an operant response to have access to an opponent that they can

subsequently fight (Hogan, 1967; Bols, 1977). Similar results have been reported

for mice (Tellegen et al., 1969), suggesting that the opportunity to interact with

an opponent may be a universal positive reinforcer in vertebrates. However, the

two explanations (i.e. endocrine modulation and associative learning) should not

be seen as mutually exclusive but as complementary, and it is even possible that

they represent two levels of analysis that are tightly interconnected. Condition-

ing of the endocrine response by social stimuli is a possibility that remains to

be tested, and there are already examples of androgen modulation of learning

mechanisms (e.g. in castrated zebra finches testosterone facilitates conspecific

song discrimination (Cynx & Nottebohm, 1992)). Therefore, the interrelationship

between androgens and associative learning mechanisms is certainly a key topic

for future research in this area.


506 R. F. Oliveira

Acknowledgements

I wish to thank all the present and past members of the ‘Mackerel Academy’ (what

other name could a fish ethology group have in a Psychology School?) for all the reading club

discussions and the Monday seminars that generated hypotheses, shaped experimental designs

and dissected the results of some of the research reported here. They are in alphabetical order: R.

Andrade, K. Becker, L. Carneiro, N. Castro, T. Fagundes, D. Goncalves, K. Hirschenhauser, J. Jordao,

M. Lopes, T. Oliveira, A. Ros, J. Saraiva and A. Silva. I also thank P. McGregor, R. Bshary and two

anonymous reviewers for providing helpful comments on an earlier version of the manuscript.

I would like to dedicate this chapter, as a posthumous expression of thanks, to the memory of

the late Luis Carneiro (b. 18 January 1969, d. 12 September 2002). More than a PhD student with

a promising career, Luis was a beloved friend and his humour and attitude towards life made

him an example of how intensely life can, and should, be lived. The unpublished studies reported

here were funded by two ongoing research grants from Fundacao para a Ciencia e a Tecnologia –

FCT (PRAXIS XXI/P/BIA/10251/1998 and POCTI/BSE/38395/2001). The writing of this chapter was

partially funded by the Plurianual Program from FCT (UI&D 331/94).

References

Abbott, J. C., Dunbrack, R. L. & Orr, C. D. 1985. The interaction of size and experience

in dominance relationships of juvenile steelhead trout (Salmo gairdneri). Behaviour,

92, 241–253.

Adams, C. F., Liley, N. R. & Gorzalka, B. B. 1996. PCPA increases aggression in male

firemouth cichlids. Pharmacology, 53, 328–330.

Albers, H. E., Liou, S. Y. & Ferris, C. F. 1988. Testosterone alters the behavioural

response of the medial preoptic–anterior hypothalamus to microinjection of

arginine vasopressin in the hamster. Brain Research, 456, 382–386.

Alexander, R. D. 1961. Aggressivness, territoriality and sexual behaviour in field

crickets. Behaviour, 17, 130–223.

Alves-Gomes, J. 1999. Systematic biology of gymnotiform and mormyriform electric

fishes: phylogenetic relationships, molecular clocks and rates of evolution in the

mitochondrial rRNA genes. Journal of Experimental Biology, 202, 1167–1183.

Andrew, R. J. 1991. Testosterone, attention and memory. In: The Development and

Integration of Behaviour, ed. P. Bateson. Cambridge, UK: Cambridge University Press,

pp. 171–190.

Arnold, A. P. 1975. The effects of castration and androgen replacement on song,

courtship and aggression in zebra finches (Poephilia guttata). Journal of Experimental

Zoology, 191, 309–326.

Bakker, T. C. M. & Sevenster, P 1983. Determinants of dominance in male sticklebacks

(Gasterosteus aculeatus L.). Behaviour, 86, 55–71.

Bakker, T. C. M., Bruijn, E. & Sevenster, P. 1989. Asymmetrical effects of prior winning

and losing on dominance in sticklebacks (Gasterosteus aculeatus). Ethology, 82,

224–229.



Baldwin, F. M., Goldin, H. S. 1939. Effects of testosterone propionate on the female

viviparous teleost, Xiphophorus helleri Heckel. Proceedings of the Society for

Experimental Biology and Medicine, 42, 813–819.

Ball, G. F. & Wingfield, J. C. 1987. Changes in plasma luteinizing hormone and sex

steroid hormones in relation to multiple-broodness and nest-site density in male

starlings. Physiological Zoology, 602, 191–199.

Balthazart, J., Foidart, A., Wilson, E. M. & Ball, G. F. 1992. Immunocythochemical

localization of androgen receptors in the male songbird and quail brain. Journal of

Comparative Neurology, 317, 407–420.

Baltz, A. P. & Clark, A. B. 1997. Extra-pair courtship behaviour of male budgerigars and

the effect of an audience. Animal Behaviour, 53, 1017–1024.

Bass, A. H. & Hopkins, C. D. 1983. Hormonal control of sexual differentiation: changes

in electric organ discharge wave-form. Science, 220, 971–974.

1984. Shifts in frequency tuning in electroreceptors in androgen-treated mormyrid

fish. Journal of Comparative Physiology A, 155, 713–724.

1985. Hormonal control of sex differences in the electric organ discharge (EOD) of

mormyrid fishes. Journal of Comparative Physiology A, 156, 587–604.

Bass, A. H. & Volman, S. F. 1987. From behavior to membranes: testosterone-induced

changes in action potential duration in electric organs. Proceedings of the National


Bass, A. H., Denizot, J. P. & Marchaterre, M. A. 1986. Ultrastructural features and

hormone-dependent sex differences of mormyrid electric organs. Journal of

Comparative Neurology, 254, 511–528.

Beacham, J. L. 1988. The relative importance of body size and aggressive experience as

determinants of dominance in pumpkinseed sunfish, Lepomis gibbosus. Animal

Behaviour, 36, 621–623.

Beacham, J. L. & Newman, J. A. 1987. Social experience and the formation of

dominance relationships in the pumpkinseed sunfish, Lepomis gibbosus. Animal

Behaviour, 35, 1560–1562.

Beani, L., Panzica, G., Briganti, F., Persichella, P. & Dessı-Fulgheri, F. 1995.

Testosterone-induced changes of call structure, midbrain and syrinx anatomy in

partridges. Physiology and Behavior, 58, 1149–1157.

Beaugrand, J. P. & Goulet, C. 2000. Distinguishing kinds of prior dominance and

subordination experiences in males of green swordtail fish (Xiphophorus helleri).


Beaugrand, J. P. & Zayan, R. 1985. An experimental model of aggressive dominance in

Xiphophorus helleri (Pisces, Poeciliidae). Behavioural Processes, 10, 1–52.

Beaugrand, J. P., Goulet, C. & Payette D 1991. Outcome of dyadic conflict in male green

swordtail fish, Xiphophorus helleri: effects of body size and prior dominance. Animal

Behaviour, 41, 417–424.

Beaugrand, J. P., Payette, D. & Goulet, C. 1996. Conflict outcome in male green

swordtail fish dyads (Xiphophorus helleri): interaction of body size, prior

dominance/subordination experience, and prior residency. Behaviour, 133,

303–319.


508 R. F. Oliveira

Beletsky, L. D., Orians, G. H. & Wingfield, J. C. 1990. Steroid hormones in relation to

territoriality, breeding density, and parental behavior in male yellow-headed

blackbirds. The Auk, 107, 60–68.

1992. Year-to-year patterns of circulating levels of testosterone and

corticosterone in relation to breeding density, experience, and reproductive

success of the polygynous red-winged blackbird. Hormones and Behavior, 26,

420–432.

Beletsky, L. D., Gori D. F., Freeman, S. & Wingfield, J. C. 1995. Testosterone and

polygyny in birds. In: Current Ornithology, Vol. 12, ed. D. M. Power. New York:


Belvedere, P., Colombo, L., Giacoma, C., Malacarne G. & Andreoletti, G. E. 1988.

Comparative ethological and biochemical aspects of courtship pheromones in

European newts. Monitore Zoologico Italiano, 22, 397–403.

Bernhardt, P. C., Dabbs, J. M., Fielden, J. A. & Lutter, C. D. 1998. Testosterone changes

during vicarious experiences of winning and losing among fans at sports events.

Physiology and Behavior, 65, 59–62.

Bernstein, I. S., Rose, R. M. & Gordon T. P. 1974. Behavioral and environmental events

influencing primate testosterone levels. Journal of Human Evolution, 3,

517–525.

Bleisch, W., Luine, V. N. & Nottebohm, F. 1984. Modification of synapses in

androgen-sensitive muscle. I. Hormonal regulation of acetylcholine receptor

number in the songbird syrinx. Journal of Neuroscience, 4, 786–792.

Bols, R. J. 1977. Display reinforcement in the Siamese fighting fish, Betta splendens:

aggressive motivation or curiosity? Journal of Comparative and Physiological


Booth, A., Shelley, G., Mazur, A., Tharp, G. & Kittok, R. 1989. Testosterone, and winning

and losing in human competition. Hormones and Behavior, 23, 556–571.



Brantley, R. K. & Bass, A. H. 1994. Alternative male spawning tactics and acoustic

signals in the plainfin midshipman fish, Porichthys notatus (Teleostei,

Batrachoididae). Ethology, 96, 213–232.

Brantley, R. K., Marchaterre, M. A. & Bass, A. H. 1993. Androgen effects on vocal muscle

structure in a teleost fish with inter- and intra-sexual dimorphisms. Journal of

Morphology, 216, 305–318.

Brenowitz, E. A. & Arnold, A. P. 1989. Accumulation of estrogen in a vocal control

brain region of a duetting songbird. Brain Research, 359, 364–367.

1992. Hormone accumulation in song regions of the canary brain. Journal of

Neurobiology, 23, 871–880.

Brennan, C. & Henderson, L. P. 1995. Androgen regulation of neuromuscular junction

structure and function in a sexually dimorphic muscle of the frog Xenopus laevis.

Journal of Neurobiology, 27, 172–188.

Bronstein, P. M. 1989. The priming and retention of agonistic motivation in male

Siamese fighting fish, Betta splendens. Animal Behaviour, 37, 165–166.



1994. On the predictability, sensitisation, and habituation of aggression in male

bettas (Betta splendens). Journal of Comparative Psychology, 98, 421–431.

Brown, R. E. 1978. Hormonal control of odor preferences and urine-marking in male

and female rats. Physiology and Behavior, 20, 21–24.

Brown, R. E. & MacDonald, D. W. 1985. Social Odours in Mammals. Oxford: Clarendon

Press.

Buchanan, K. L., Evans, M. R., Goldsmith, A. R., Bryant, D. M. & Rowe, L. V. 2001.

Testosterone influences basal metabolic rate in male house sparrows: a new cost

of dominance signalling? Proceedings of the Royal Society of London, Series B, 268,

1337–1344.

Callard, G. V., Drygas, M. & Gelinas, D. 1993. Molecular and cellular physiology of

aromatase in the brain and retina. Journal of Steroid Biochemistry and Molecular

Biology, 44, 541–547.

Cardwell, J. R., Stacey N. E., Tan, E. S. P., McAdam, D. S. O. & Lang, S. L. C. 1995.

Androgen increases olfactory receptor response to a vertebrate sex pheromone.

Journal of Comparative Physiology A, 176, 55–61.

Carlson, B. A., Hopkins, C. D. & Thomas, P. 2000. Androgen correlates of socially

induced changes in the electric organ discharge waveform of a mormyrid fish.


Cedrini, L. & Fasolo, A. 1970. Olfactory attractants in sex recognition of the crested

newt: an electrophysiological research. Monitore Zoologico Italiano, 5, 223–229.

Chapman, F. M. 1935. The courtship of Gould’s manakin (Manacus vitellinus vitellinus) on

Barrado Colorado Island, Canal Zone. American Museum of Natural History Bulletin,

68, 472–521.

Chase, I. D., Bartolomeo, C. & Dugatkin, L. A. 1994. Aggressive interactions and


393–400.

Clotfelter, E. & Paolino, A. 2003. Winner effect is induced in bystanders to aggressive

contests between male fighting fish. Animal Behaviour, 66, 343–347.

Cronley-Dillon, J. & Sharma, S. C. 1968. Effect of season and sex on the photopic

spectral sensitivity of the three-spined stickleback. Journal of Experimental Biology,

49, 679–687.

Cynx, J. & Nottebohm, F. 1992. Testosterone facilitates some conspecific song

discriminations in castrated zebra finches Taeniopygia guttata. Proceedings of the


De Vries, G. J. 1995. Studying neurotransmitter systems to understand the

development and function of sex differences in the brain: the case of vasopressin.

In: Neurobiological Effects of Sex Steroid Hormones, ed. P. E. Micevych & R. P. Hamer Jr.


Deviche, P. 1992. Testosterone and opioids interact to regulate feeding in a male

migratory songbird. Hormones and Behavior, 26, 394–405.

Dohanich, G. 2002. Gonadal steroids, learning and memory. In: Hormones, Brain and

Behavior, Vol. 2, ed. D. W. Pfaff, A. P. Arnold, A. M. Etgen, S. E. Fahrbach & R. T.

Rubin. New York: Academic Press, pp. 265–327.


510 R. F. Oliveira

Dorlochter, M., Astrow, S. H. & Herrera, A. A. 1994. Effects of testosterone on a sexually

dimorphic frog muscle: repeated in vivo observations and androgen receptor

distribution. Journal of Neurobiology, 25, 897–916.




Drummond, H. & Osorio, J. L. 1992. Training siblings to be submissive losers:

dominance between booby nestlings. Animal Behaviour, 44, 881–893.



Dunlap, K. D. & Zakon, H. H. 1998. Behavioral actions of androgens and androgen

receptor expression in the electrocommunication system of an electric fish,

Eigenmannia virescens. Hormones and Behavior, 34, 30–38.




Edwards, D. H. & Kravitz, E. A. 1997. Serotonin, social status and aggression. Current

Opinion in Neurobiology, 7, 812–819.

Eichenbaum, H., Otto, T. & Cohen, N. J. 1992. The hippocampus: what does it do?

Behavioral and Neural Biology, 57, 2–36.

Evans, C. S. & Marler, P. 1994. Food calling and audience effects in male chickens,

Gallus gallus: their relationships to food availability, courtship and social

facilitation. Animal Behaviour, 47, 1159–1170.

Ferkin, M. H. & Johnston, R. E. 1993. Roles of gonadal hormones in control of five

sexually attractive odors in meadow voles (Microtus pennsylvanicus). Hormones and

Behavior, 27, 523–538.

Fernald, R. D. 1976. The effect of testosterone on the behavior and coloration of adult

male cichlid fish, Haplochromis burtoni. Hormone Research, 7, 172–178.

Fine, M. L. & Pennymaker, K. R. 1986. Hormonal basis for sexual dimorphism of the

sound-producing apparatus of the oyster toadfish. Experimental Neurology, 92,

289–298.

Forlano, P. M., Sisneros, J. A., Knapp, R. & Bass, A. H. 2003. Reproductive periodicity of

gonadal steroid hormone levels in the vocal plainfin midshipman, Porichthys

notatus. Hormones and Behavior, 44: 49.

Francis, R. C. 1983. Experiential effects on agonistic behavior in the paradise fish,

Macropodus opercularis. Behaviour, 85, 292–313.

1987. The interaction of genotype and experience in the dominance success of

paradise fish (Macropodus opercularis). Biology of Behaviour, 12, 1–11.

Franck, D. & Ribowski, A. 1987. Influences of prior agonistic experiences on

aggression measures in the male swordtail (Xiphophorus helleri). Behaviour, 103,

217–240.

Freedman, E. G., Olyarchuk, J., Marchaterre, M. A. & Bass, A. H. 1989. A temporal

analysis of testosterone-induced changes in electric organs and electric organ

discharges of mormyrid fishes. Journal of Neurobiology, 20, 619–634.



Frey, D. F. & Miller, R. J. 1972. The establishment of dominance relationships in the

blue gourami Trichogaster trichopterus (Pallas). Behaviour, 42, 8–62.

Gahr, M., Flugge, G. & Guttinger, H. R. 1987. Immunocytochemical localization of

estrogen-binding neurons in the songbird brain. Brain Research, 402, 173–177.

Gahr, M., Guttinger, H. R. & Kroodsma, D. A. 1993 Estrogen receptors in the avian

brain: survey reveals general distribution and forebrain areas unique to

songbirds. Journal of Comparative Neurology, 327, 112–122.

Gawienowsky, A. M., De Nicola, D. B. & Stacewicz-Sapuntzakis, M. 1976. Androgen

dependence of a marking pheromone in rat urine. Hormones and Behavior, 7,

401–405.

Gelinas, D. & Callard, G. V. 1997 Immunolocalization of aromatase and androgen

receptor-positive neurons in the goldfish brain. General and Comparative


Ginsberg, B. & Allee, W. C. 1942. Some effects of conditioning of social dominance and

subordination in inbred strains of mice. Physiological Zoology, 15, 485–506.

Gonzalez-Mariscal, G., Melo, A. E., Zavala, A. & Beyer, C. 1993. Sex steroid regulation of

chin-marking behavior in male New Zealand rabbits. Physiology and Behavior, 54,

1035–1040.

Gould, S. J. & Vrba, E. S. 1982. Exaptation: a missing term in the science of form.

Paleobiology, 8, 4–15.

Grober, M. S. 1998. Socially controlled sex-change: integrating ultimate and proximate

levels of analysis. Acta Ethologica, 1, 3–17.

Gyger, M., Karakashian, S. J. & Marler, P. 1986. Avian alarm calling: is there an

audience effect? Animal Behaviour, 34, 1570–1572.

Hagedorn, M. & Carr, C. 1985. Single electrocytes produce a sexually dimorphic signal

in South American electric fish Hypopomus occidentalis (Gymnotiformes,

Hypopomidae). Journal of Comparative Physiology A, 156, 511–523.

Hagedorn, M. & Heiligenberg, W. 1985. Court and spark: electric signals in the

courtship and mating of gymnotid fish. Animal Behaviour, 32, 254–265.

Hannes, R. P. 1984. Androgen and corticoid levels in blood and body extracts of high

and low ranking swordtail males (Xiphophorus helleri) before and after social

isolation. Zeitschrift fur Tierpsychologie, 66, 70–76.

1986. Blood and whole-body androgen levels of male swordtails correlated with

aggression measures in a standard-opponent test. Aggressive Behavior, 12, 249–254.

Hasegawa, Y., Okuno, A. & Nagasawa, H. 2002. Immunohistochemical study of

androgenic gland hormone: localization in the male reproductive system and

species specificity in the terrestrial isopods. General and Comparative Endocrinology,

125, 218–225.

Hector, A. C. K., Seyfarth, R. M. & Raleigh, M. 1989. Male parental care, female choice

and the effect of an audience in vervet monkeys. Animal Behaviour, 38, 262–271.

Hegner, R. E. & Wingfield, J. C. 1987a. Social status and circulating levels of hormones

in flocks of house sparrows, Passer domesticus. Ethology, 76, 1–14.

1987b. Effects of experimental manipulations of testosterone levels on parental

investment and breeding success in male house sparrows. The Auk, 104, 462–469.


512 R. F. Oliveira

Heid, P., Guttinger, H. R. & Prove, E. 1985. The influence of castration and testosterone

replacement on the song architecture of canaries (Serinus canarius). Zeitschrift fur

Tierpsychologie, 69, 224–236.

Herbert, J. 1993. Peptides in the limbic system: neurochemical codes for coordinated

adaptive responses to behavioural and physiological demand. Neurobiology, 41,

723–791.

Herfeld, S. & Moller, P. 1998. Effects of 17-alfa-methyltestosterone on sexually

dimorphic characters in the weakly electric fish Brienomyrus niger (Gunther, 1866)

(Mormyridae): electric organ discharge, ventral body wall indentation, and

anal-fin ray bone expansion. Hormones and Behavior, 34, 303–319.

Hogan, J. A. 1967. Fighting and reinforcement in Siamese fighting fish Betta splendens.

Journal of Comparative and Physiological Psychology, 64, 356–359.

Hogan, J. A. & Bols, R. J. 1980. Priming of aggressive motivation in Betta splendens.


Hoglund, E., Kolm, N. & Winberg, S. 2001. Stress-induced changes in brain

serotonergic activity, plasma cortisol and aggressive behavior in Arctic charr

(Salvelinus alpinus) is counteracted by L-DOPA. Physiology and Behavior, 74,

381–389.




Hollis, K. L., Dumas, M. J., Singh, P. & Fackelman, P. 1995. Pavlovian conditioning of

aggressive behavior in blue gourami fish (Trichogaster trichopterus): winners become

winners and losers stay losers. Journal of Comparative Psychology, 109, 123–133.

Holst, D. V. & Eichmann, F. 1998. Sex-specific regulation of marking behavior by sex

hormones and conspecifics scent in tree shrews (Tupaia belangeri). Physiology and

Behavior, 63, 157–164.

Hopkins, C. D. 1972. Sex differences in signalling in an electric fish. Science, 176,

1035–1037.




Behaviour, 61, 777–786.

Hurst, J. L. 1990. Urine marking in populations of wild mice Mus domesticus Rutty. III.

Communication between the sexes. Animal Behaviour, 40, 233–243.

Ibara, R. M., Penny, L. T., Ebeling, A. W., van Dykhuizen, G. & Cailliet, G. 1983. The

mating call of the plainfin midshipman fish, Porichthys notatus. In: Predators and

Prey in Fishes, ed. D. G. L. Noakes, D. G. Lindquist, G. S. Helfman & J. A. Ward. The

Hague: Junk Press, pp. 205–212.

Ikeda, K. 1933. Effect of castration on the sexual characters of anadromous

three-spined stickleback, Gasterosteus aculeatus aculeatus (L.). Japanese Journal of

Zoology, 5, 135–157.

Jackson, W. 1991. Why do winners keep winning? Behavioral Ecology and Sociobiology, 28,

271–276.

Johnson, R. P. 1973. Scent marking in mammals. Animal Behaviour, 21, 521–535.



Johnsson, J. I. & Åkerman, A. 1998. Watch and learn: preview of the fighting ability of

opponents alters contest behaviour in rainbow trout. Animal Behaviour, 56,

771–776.

Johnston, R. E. 1975. Sexual excitation function of hamster vaginal secretions Animal

Learning and Behavior, 3, 161–166.

Kavaliers, M., Choleris, E. & Colwell, D. D. 2001. Brief exposure to female odors

‘emboldens’ male mice by reducing predator-induced behavioural and hormonal

responses. Hormones and Behavior, 40, 497–509.

Keller, C. H., Zakon, H. H. & Sanchez, D. Y. 1986. Evidence for a direct effect of

androgens on electroreceptor tuning. Journal of Comparative Physiology A, 158,

301–310.

Kelley, D. C. 2002. Hormonal regulation of motor output in amphibians: Xenopus laevis

vocalizations as a model system. In: Hormones, Brain and Behavior, Vol. 2, ed. D. W.

Pfaff, A. P. Arnold, A. M. Etgen, S. E. Fahrbach & R. T. Rubin. New York: Academic

Press, pp. 445–468.

Kerr, J. E., Allore, R. J., Beck, S. G. & Handa, R. J. 1995. Distribution and hormonal

regulation of androgen receptors (AR) and AR messenger ribonucleic acid in the

rat hippocampus. Endocrinology, 136, 3213–3221.

Ketterson, E. D., Nolan, V., Jr, Wolf, L. & Ziegenfus, C. 1992. Testosterone and avian life

histories: effects of experimentally elevated testosterone on behavior and

correlates of fitness in the dark-eyed junco, Junco hyemalis. American Naturalist, 140,

980–999.

Kikuyama, S., Toyoda, F., Ohmiya, Y. et al. 1995. Sodefrin: a female-attracting

pheromone in newt cloacal glands. Science, 267, 1643–1645.

Kikuyama, S., Toyoda, F., Yamamoto, K., Tanaka, S. & Hayashi, H. 1997. Female-

attracting pheromone in newt cloacal glands. Brain Research Bulletin, 44, 415–422.

Kimura, D. 1996. Sex, sexual orientation and sex hormones influence human

cognitive function. Current Opinion in Neurobiology, 6, 259–263.

Landsman, R. E. & Moller, P. 1988. Testosterone changes the electric organ discharge

and external morphology of the mormyrid fish Gnathonemus petersii

(Mormyriformes). Experientia, 44, 900–903.

Landsman, R. E., Harding, C. F., Moller, P. & Thomas, P. 1990. The effects of androgens

and estrogen on the external morphology and electric organ discharge waveform

of Gnathonemus petersii (Mormyriformes, Teleostei). Hormones and Behavior, 24,

532–553.

Leiser, J. K. & Itzkowitz, M. 1999. The benefits of dear enemy recognition in

three-contender convict cichlid (Cichlasoma nigrofasciatum) contests. Behaviour, 136,

983–1003.

Lepri, J. & Randall, J. A. 1983. Hormonal regulation of sandbathing in male kangaroo

rats (Dipodomys merriami). Behavioral and Neural Biology, 37, 256–264.

Levy, M. & Aronson, L. R. 1955. Morphological effects of castration and hormone

administration in the male cichlid fish Tilapia macrocephala. Anatomical Records,

122, 450–451.

Lisk, R. D. & Bezier, J. L. 1980. Intrahypothalamic hormone implantation and

activation of sexual behavior in the male hamster. Neuroendocrinology, 30, 220–222.


514 R. F. Oliveira

Lowe, P. 1942. The anatomy of Gould’s manakin (Manacus vitellinus) in relation to its

display. Ibis, 6, 50–83.

Luine, V., Nottebohm, F., Harding, C. & McEwen, B. S. 1980. Androgen affects

cholinergic enzymes in syringeal motor neurons and muscle. Brain Research, 192,

89–107.

Malacarne, G., Bottoni, L., Massa, R. & Vellano, C. 1984. The abdominal gland of the

crested newt: a possible source of courtship pheromones. Monitore Zoologico

Italiano, 18, 33–39.

Manzo, J., Garcia, L. I., Hernandez, M. E., Carrillo, P. & Pacheco, P. 2002.

Neuroendocrine control of urine-marking behavior in male rats. Physiology and


Marler, C. A., Walsberg, G., White, M. I. & Moore, M. 1995. Increased energy

expenditure due to increased territorial defense in male lizards after phenotypic

manipulation. Behavioral Ecology and Sociobiology, 37, 225–231.

Marler, P., Dufty, A. & Pickert, R. 1986. Vocal communication in the domestic chicken.

II. Is a sender sensitive to the presence and nature of a receiver? Animal Behaviour,

34, 194–198.

Matos, R. J. 2002. Social influences on signalling: audience effects and communication

networks. Ph.D. Thesis, University of Copenhagen.

Matos, R. J. & McGregor, P. K. 2002. The effect of the sex of an audience on male–male

displays of Siamese fighting fish (Betta splendens). Behaviour, 139, 1211–1221.

Mazur, A. & Booth, A. 1998. Testosterone and dominance in men. Behavioral and Brain

Sciences, 21, 353–397.

McDonald, N. W., Heimstra, A. L. & Damkot, D. K. 1968. Social modification of

agonistic behaviour in fish. Animal Behaviour, 16, 437–441.




McGregor, P. K. & Peake, T. 2000. Communication networks: social environments for

receiving and signalling behaviour. Acta Ethologica, 2, 71–81.



McKibben, J. R. & Bass, A. H. 1998. Behavioral assessment of acoustic parameters

relevant to signal recognition and preference in a vocal fish. Journal of the Acoustic


Meyer, J. H. 1983. Steroid influences upon the discharge frequency of a weakly electric

fish. Journal of Comparative Physiology A, 153, 29–38.

Meyer, J. H. & Zakon, H. H. 1982. Androgens alter the tuning of electroreceptors.

Science, 217, 635–637.

Meyer, J. H., Leong, M. & Keller, C. H. 1987. Hormone-induced and ontogenetic changes

in electric organ discharge and electroreceptor tuning in the weakly electric fish

Apteronotus. Journal of Comparative Physiology A, 160, 385–394.

Millinski, M. & Bakker, T. C. M. 1990. Female sticklebacks use male coloration in mate

choice and hence avoid parasitised males. Nature, 344, 330–333.



Mills, A. C. & Zakon, H. H. 1987. Coordination of EOD frequency and pulse duration in

a weakly electric wave fish: the influence of androgens. Journal of Comparative

Physiology A, 161, 417–430.

1991. Chronic androgen treatment increases action potential duration in the

electric organ of Sternopygus. Journal of Neuroscience, 11, 2349–2361.

Moofat, S. D. & Hampson, E. 1996. A curvilinear relationship between testosterone and

spatial cognition in humans: possible influence of hand preferences.

Psychoneuroendocrinology, 21, 323–337.

Moore, A. & Scott, A. P. 1991. Testosterone is a potent odorant in precocious male

Atlantic salmon (Salmo salar L.) parr. Philosophical Transactions of the Royal Society of


Moore, F. L., Wood, R. E. & Boyd, S. K. 1992. Sex steroids and vasotocin interact in a

female amphibian (Taricha granulosa) to elicit female-like egg-laying behavior or

male-like courtship. Hormones and Behavior, 26, 156–166.

Nagaya, N. & Herrera, A. A. 1995. Effects of testosterone on synaptic efficacy at

neuromuscular junctions in a sexually dimorphic muscle of male frogs. Journal of

Physiology, 483, 141–153.



Series B, 266, 537–542.

Nelson, J. S. 1994. Fishes of the World, 3rd edn. New York: John Wiley.

Nelson, R. 2001. An Introduction to Behavioural Endocrinology, 2nd edn. Sunderland, MA:

Sinauer.

Northcutt, R. G. & Davis, R. E. 1983. Telencephalic organization in ray-finned fishes. In:

Fish Neurobiology, Vol. 2: Higher Brain Areas and Functions, ed. R. E. Davis & R. G.

Northcutt. Ann Arbor, MI: University of Michigan Press, pp. 203–236.

Oliveira, R. F. & Almada, V. C. 1998. Androgenization of dominant males in a cichlid

fish: androgens mediate the social modulation of sexually dimorphic traits.

Ethology, 104, 841–858.

Oliveira, R. F., Almada, V. C. & Canario, A. V. M. 1996. Social modulation of sex steroid

concentrations in the urine of male cichlid fish Oreochromis mossambicus. Hormones

and Behavior, 30, 2–12.




Oliveira, R. F., Carneiro, L. A., Goncalves, D. M., Canario, A. V. M. & Grober, M. S.

2001a. Effects of 11-ketotestosterone in sneaker males of the peacock blenny, a

species with male alternative mating tactics. Brain, Behavior and Evolution, 58,

28–37.

Oliveira, R. F., Lopes, M., Carneiro, L. A. & Canario, A. V. M. 2001b. Watching fights

raises fish hormone levels. Nature, 409, 475.

Oliveira, R. F., Hirschenhauser, K., Carneiro, L. & Canario, A. V. M. 2002. Social

modulation of androgen levels in male teleost fish. Comparative Biochemistry and

Physiology B, 132, 203–215.


516 R. F. Oliveira

Otronen, M. 1990. The effect of prior experience on the outcome of fights in the

burying beetle, Nicrophorus humatio. Animal Behaviour, 40, 980–1004.

Owen, P. C. & Perrill, S. A. 1998. Habituation in the green frog, Rana clamitans.


Pankhurst, N. W. & Barnett, C. W. 1993. Relationship of population density, territorial

interaction, and plasma levels of gonadal steroids in spawning male demoiselles

Chromis dispilus (Pisces, Pomacentridae). General and Comparative Endocrinology, 90,

168–176.

Parker, G. A. 1974. The assessment strategy and the evolution of fighting behaviour.

Journal of Theoretical Biology, 47, 223–243.


Behavior 32, 114–121.




Pomerantz, S. M. & Sholl, S. A. 1987. Analysis of sex and regional differences in

androgen receptors in fetal rhesus monkey brain. Brain Research, 433, 151–154.

Randall, J. A. 1986. Lack of gonadal control of the dorsal gland and sandbathing in

male and female Bannertail kangaroo rats (Dipodomys spectabilis). Hormones and

Behavior, 20, 95–105.

Reinboth, R. & Rixner, W. 1972. Verhalten des Kleinen Maulbruters Hemihaplochromis

multicolor nach Kastration und Behandlung mit Testosteron. [Film C 1019/, 1970].

Gottingen: Institut fur den Wissenschaftlichen Film.

Resnick, S. M., Berenbaum, S. A., Gottesman, I. J. & Bouchard, T. J. 1986. Early

hormonal influences on cognitive functioning in congenital adrenal hyperplasia.

Developmental Psychology, 22, 191–198.

Roof, R. L. 1993. Neonatal exogenous testosterone modifies sex difference in radial

arm and Morris water maze performance in prepubescent and adult rats.

Behavioral and Brain Research, 53, 1–10.

Roof, R. L. & Havens, M. D. 1992. Testosterone improves maze performance and induces

development of a male hippocampus in females. Brain Research, 572, 310–313.

Ros, A. F. H., Becker, K., Canario, A. V. M. & Oliveira, R. F. 2004. Androgen levels and

energy metabolism in Oreochromis mossambicus. Journal of Fish Biology, 65,

895–905.

Rose, R. M., Bernstein, I. S. & Holaday, J. W. 1971. Plasma testosterone, dominance

rank and aggressive behavior in a group of male rhesus monkeys. Nature, 231,

366.

Rose, R. M., Berstein, I. S. & Gordon, T. 1975. Consequences of social conflict on plasma

testosterone levels in rhesus monkeys. Psychosomatic Medicine, 37, 50–61.

Rost, R. 1990. Hormones and behavior: a joint examination of studies on seasonal

variation in song production and plasma levels of testosterone in the great tit

Parus major. Journal of Ornithology, 131, 403–411.

1992. Hormones and behavior: a comparison of studies on seasonal changes in song

production and testosterone plasma levels in the Willow tit Parus montanus. Ornis

Fennica, 69, 1–6.



Rouse, E. F., Coppinger, C. J. & Barnes, P. R. 1977. The effect of an androgen inhibitor

on behavior and testicular morphology in the stickleback Gasterosteus aculeatus.


Rowland, W. J. 1999. Studying visual cues in fish behavior: a review of ethological

techniques. Environmental Biology of Fishes, 56, 285–305.

Ruzzante, D. E. 1992. Mirror image stimulation, social hierarchies, and population

differences in agonistic behavior: a reappraisal. Canadian Journal of Fisheries and

Aquatic Sciences, 49, 1966–1968.

Saldanha, C. J., Clayton, N. S. & Schlinger, B. A. 1999. Androgen metabolism in the

juvenile oscine forebrain: a cross-species analysis at neural sites implicated in

memory function. Journal of Neurobiology, 40, 397–406.

Sapolsky, R. M. 1992. Neuroendocrinology of the stress-response. In: Behavioral

Endocrinology, ed. J. B. Becker, S. M. Breedlove & D. Crews. Cambridge, MA: MIT

Press, pp. 287–324.

Sawyer, T. F., Hengehold, A. K. & Perez, W. A. 1984. Chemosensory and hormonal

mediation of social memory in male rats. Behavioral Neuroscience, 98, 908–913.

Scalia, F. & Winans, S. S. 1975. The differential projections of the olfactory bulb and

accessory olfactory bulb in mammals. Journal of Comparative Neurology, 161, 31–56.

Schlinger, B. A., Schultz, J. D. & Hertel, F. 2001. Neuromuscular and endocrine control

of an avian courtship behavior. Hormones and Behavior, 40, 276–280.

Schuett, G. W. 1997. Body size and agonistic experience affect dominance and mating

success in male copperheads. Animal Behaviour, 54, 213–224.

Schulkin, J. 2002. Hormonal modulation of central motivational states. In: Hormones,

Brain and Behavior, Vol. 1, ed. D. W. Pfaff, A. P. Arnold, A. M. Etgen, S. E. Fahrbach &

R. T. Rubin. New York: Academic Press, pp. 663–657.

Schultz, J. D. & Schlinger, B. A. 1999. Widespread accumulation of [3H]testosterone in

the spinal cord of a wild bird with an elaborate courtship display. Proceedings of the


Schultz, J. D., Hertel, F., Bauch, M. & Schlinger, B. A. 2001. Adaptations for rapid and

forceful contraction in wing muscles of the male golden-collared manakin: sex

and species comparisons. Journal of Comparative Physiology A, 187, 677–684.

Searcy, W. A., Eriksson, D. & Lundberg, A. 1991. Deceptive behavior in pied flycatchers.


Silverin, B. 1980. Effects of long-acting testosterone treatment on free-living pied

flycatchers, Ficedula hypoleuca, during the breeding period. Animal Behaviour, 28,

906–912.

Sisneros, J. A. & Bass, A. H. 2003. Seasonal plasticity of peripheral auditory frequency

sensitivity. Journal of Neuroscience, 23, 1049–1058.

Sisneros, J. A. & Tricas, T. C. 2000. Androgen-induced changes in the response

dynamics of ampullary electrosensory primary afferent neurons. Journal of

Neuroscience, 20, 8586–8595.

Simon, N. G. 2002. Hormonal processes in the development and expression of

aggressive behavior. In: Hormones, Brain and Behavior, Vol. 1, ed. D. W. Pfaff, A. P.

Arnold, A. M. Etgen, S. E. Farbach & R. T. Rubin. New York: Academic Press,

pp. 339–392.


518 R. F. Oliveira

Smith, G. T., Brenowitz, E. A., Beecher, M. D. & Wingfield, J. C. 1997. Seasonal changes

in testosterone, neural attributes of song control nuclei, and song structure in

wild songbirds. Journal of Neuroscience, 17, 6001–6010.

Smith, R. J. F. 1974. Effects of 17-methyltestosterone on the dorsal pad and tubercles of

fathead minnows (Pimephales promelas). Canadian Journal of Zoology, 52, 1031–1038.

Squire, L. R. 1992. Memory and the hippocampus: a synthesis from findings with rats,

monkeys, and humans. Psychological Reviews, 9, 195–231.

Stacey, N. & Sorensen, P. 2002. Hormonal pheromones in fish. In: Hormones, Brain and

Behavior, Vol. 2, ed. D. W. Pfaff, A. P. Arnold, A. M. Etgen, S. E. Fahrbach & R. T.

Rubin. New York: Academic Press, pp. 375–434.

Suthers, R. A., Goller, F. & Pytte, C. 1999. The neuromuscular control of bird song.

Philosophical Transactions of the Royal Society of London, Series B, 354, 927–939.

Taborsky, M. 1994. Sneakers, satellites, and helpers: parasitic and cooperative

behaviour in fish reproduction. Advances in the Study of Behaviour, 23, 1–100.

Taylor, G. T., Bartko, G. & Farr, S. 1987. Gonadal hormones and conspecific marking in

male rats. Hormones and Behavior, 21, 234–244.

Tellegen, A., Horn, J. M. & Legrand, R. G. 1969. Opportunity for aggression as a

reinforcer for mice. Psychonomic Science, 14, 104–105.

Temeles, E. J. 1994. The role of neighbours in territorial systems: when are they ‘dear

enemies’? Animal Behaviour, 47, 339–350.

Tinbergen, N. 1951. The Study of Instinct. Oxford: Clarendon Press.

Tobias, M. L. & Kelley, D. B. 1995. Sexual differentiation and hormonal regulation of

the laryngeal synapse in Xenopus laevis. Journal of Neurobiology, 28, 515–526.

Tobias, M. L., Kelley, D. B. & Ellisman, M. 1995. A sex-difference in synaptic efficacy at

the laryngeal neuromuscular junction of Xenopus laevis. Journal of Neuroscience, 15,

1660–1668.

Tohgi, H., Utsugisawa, K., Yamagata, M. & Yoshimura, M. 1995. Effects of age on mRNA

expression of glucocorticoid, thyroid hormone, androgen, and estrogen receptors

in postmortem human hippocampus. Brain Research, 700, 245–253.

Vezina, M. & Thomas, D. W. 2000. Social status does not affect resting metabolic rate

in wintering dark-eyed junco (Junco hyemalis). Physiological and Biochemical Zoology,

73, 231–236.

Wapler-Leong, D. C. Y. & Reinboth, R. 1974. The influence of androgenic hormone on

the behaviour of Haplochromis burtoni (Cichlidae). Fortschritte der Zoologie, 22,

334–339.

Wetzel, D. M. & Kelley, D. B. 1983. Androgen and gonadotrophin effects on male mate

calls in South African clawed frogs, Xenopus laevis. Hormones and Behaviour, 17,

388–404.

Whitehouse, M. A. 1997. Experience influences male–male contests in the spider

Argyrodes antipodiana (Theridiidae: Aranea). Animal Behaviour, 53, 913–923.

Whiting, M. J. 1999. When to be neighbourly: differential agonistic responses in the

lizard Platysaurus broadleyi. Behavioral Ecology and Sociobiology, 46, 210–214.

Wikelsky, M., Lynn, S., Breumer, C., Wingfield, J. C. & Kenagy, G. J. 1999. Energy

metabolism, testosterone and corticosterone in white-crowned sparrows. Journal

of Comparative Physiology A, 185, 463–470.



Williams, C. L., Barnett, A. M. & Meck, W. H. 1990. Organizational effects of early

gonadal secretions on sexual differentiation in spatial memory. Behavioral

Neuroscience, 104, 84–97.

Winberg, S. & Lepage, O. 1998. Elevation of brain 5-HT activity, POMC expression, and

plasma cortisol in socially subordinate rainbow trout. American Journal of

Physiology, 274, R645–R654.

Winberg, S. & Nilsson, G. E. 1993a. Roles of brain monoamine neurotransmitters in

agonistic behaviour and stress reactions, with particular reference to fish.

Comparative Biochemistry and Physiology C, 10, 597–614.

1993b. Time course of changes in brains serotonergic activity and brain tryptophan

levels in dominant and subordinate juvenile Arctic charr. Journal of Experimental

Biology, 179, 181–195.

Winberg, S., Winberg, Y. & Fernald, R. D. 1997. Effect of social rank on brain

monoaminergic activity in a cichlid fish. Brain, Behavior and Evolution, 49,

230–236.

Wingfield, J. C. 1984 Environmental and endocrine control of reproduction in the

song sparrow, Melospiza melodia. II. Agonistic interactions as environmental

information stimulating secretion of testosterone. General and Comparative


Wingfield, J. C., Ball, G. F., Dufty, A. M., Hegner, R. E. & Ramenofsky, M 1987.

Testosterone and aggression in birds: tests of the challenge hypothesis. American

Scientist, 75, 602–608.

Wingfield, J. C., Hegner, R. E., Dufty, A. M. & Ball, G. F. 1990. The ‘challenge hypothesis’:

theoretical implications for patterns of testosterone secretion, mating systems,

and breeding strategies. American Naturalist, 136, 829–846.

Wingfield, J. C., Jacobs, J. D., Soma, K. et al. 1999. Testosterone, aggression, and

communication: ecological bases of endocrine phenomena. In: The Design of

Animal Communication, ed. M. D. Hauser & M. Konishi. Cambridge, MA: MIT Press,

pp. 257–283.

Wingfield, J. C., Jacobs, J. D., Tramontin, A. D. et al. 2000. Toward an ecological

basis of hormone–behavior interactions in reproduction of birds. In: Reproduction

in Context, ed. K. Wallen & J. E. Schneider. Cambridge, MA: MIT Press,

pp. 85–128.

Wood, R., Brabec, R. K., Swann, J. M. & Newman, S. W. 1992. Androgen and estrogen

concentrating neurons in chemosensory pathways in the male Syrian hamster

brain. Brain Research, 596, 89–98.

Yahr, P. & Stephens, D. R. 1987. Hormonal control of sexual and scent marking

behaviors of male gerbils in relation to the sexually dimorphic area of the

hypothalamus. Hormones and Behavior, 21, 331–346.

Yamamoto, K., Toyoda, F., Tanaka, S., Hayashi, H. & Kikuyama, S. 1996.

Radioimmunoassay of a newt sex pheromone, sodefrin, and the influence of

hormones on its level in the abdominal gland. General and Comparative


Ydenberg, R. C., Giraldeau, L. A. & Falls, J. B. 1988. Neighbours, strangers and the

asymmetric war of attrition. Animal Behaviour, 36, 343–347.


520 R. F. Oliveira

Zakon, H. H. & Smith, G. T. 2002. Weak electric fish: behavior, neurobiology and

neuroendocrinology. In: Hormones, Brain and Behavior, Vol. 2, ed. D. W. Pfaff, A. P.

Arnold, A. M. Etgen, S. E. Fahrbach & R. T. Rubin. New York: Academic Press,

pp. 349–374.

Zakon, H. H., Yan, H. Y. & Thomas, P. 1990. Human chorionic gonadotropin-induced

shifts in the electrosensory system of the weakly electric fish, Sternopygus. Journal

of Neurobiology, 21, 826–833.

Zakon, H. H., Thomas, P. & Yan, H. Y. 1991. Electric organ discharge frequency and

plasma sex steroid levels during gonadal recrudescence in a natural population

of the weakly electric fish Sternopygus macrurus. Journal of Comparative Physiology A,

169, 493–499.


22

Cooperation in communicationnetworks: indirect reciprocity ininteractions between cleaner fishand client reef fish

r e d o ua n b s h a r y 1 & a r u n d ’s o u z a 2

1University of Cambridge, Cambridge, UK2University of Wurzburg, Wurzburg, Germany

Introduction

The aim of this chapter is twofold. First, to outline how recent develop-

ments in cooperation theory are so similar to the communication network concept

(McGregor, 1993) that a unified terminology would be useful to facilitate exchange

of ideas. Second, we argue that the communication network concept provides an

evolutionary framework to predict the widespread occurrence of phenomena that

until now have been discussed in the context of highly developed cognitive abili-

ties. This creates a problem: as it stands, there appear to be no words in cooperation

theory that were not developed in the human context and hence do not include

a cognitive component. We have to use definitions that only constitute the func-

tional aspects of phenomena (like tactical deception and indirect reciprocity) and

ignore the mechanistic aspects (i.e. theory of mind, intentionality) that are com-

monly part of the definitions. We ask readers always to keep in mind that our

definitions never imply any specific cognitive abilities. We illustrate our ideas

with data on the mutualism between the cleaner wrasse Labroides dimidiatus and

its ‘client’ reef fish, which visit cleaners to have ectoparasites removed from their

surface, gills and mouth (reviewed by Losey et al., 1999; Cote, 2000).

Cooperation provides a challenge to evolutionary theory because it often in-

volves apparently altruistic behaviour. Hamilton (1964) provided a framework to

understand why altruism between kin can be evolutionary stable; specifically, an

altruist gains indirect fitness benefits from its action. However, there are plenty of


521


522 R. Bshary & A. D’Souza

examples where recipients of altruistic acts are unrelated to the helping individual

(Dugatkin, 1997). Trivers (1971) pointed out that such altruism might be evolu-

tionarily stable if the recipient later reciprocates, an idea that was formalized

by Axelrod & Hamilton (1981). They used an iterated version of the prisoner’s

dilemma game, a two-player game in which opponents can either cooperate or

defect. Defection yields a higher payoff than cooperation independently of the

partner’s action, but if both players cooperate they receive a higher payoff than

if both defect, hence the dilemma. In a computer tournament in which several

strategies competed with each other, a simple strategy called ‘tit-for-tat’ emerged

as a cooperative solution to the game. Tit-for-tat players start by being cooperative;

then in the next round they play what their opponent did in the previous round.

Thus, tit-for-tat players can reap the benefits of mutual cooperation while avoid-

ing exploitation by an uncooperative opponent (except for the first occasion on

which an opponent defected). Several new strategies that are similar but appar-

ently superior to tit for tat have been tested since the first computer tournament

(reviewed by Dugatkin, 1997).

Despite the intuitive appeal of reciprocal altruism and behavioural strategies

similar to tit for tat to ensure cooperative behaviour, few empirical examples have

been reported and some that have are contentious (Dugatkin, 1997). In addition,

Alexander (1987) pointed out that many examples of human altruistic behaviour

do not fit an iterated prisoner’s dilemma game: humans often help individuals

who are highly unlikely to ever reciprocate. He proposed that humans might help

others simply to increase their own image within the society. Nowak & Sigmund

(1998) explored this idea by developing a game theory model in which direct

reciprocity on altruistic acts was excluded. Instead, an altruist gained an increase

in his image score. If an individual’s image score was linked to the probability that

others were willing to provide help when needed, cooperation readily emerged

and was evolutionarily stable (Nowak & Sigmund, 1998). Individuals helped in

order to be helped themselves during future interactions with current bystanders

(for further theoretical developments see Lotem et al. (1999, 2003) and Leimar &

Hammerstein (2001)). An experiment with first-year students confirmed a crucial

prediction of the models: students that helped more than average received more

help and, therefore, a final payoff that was above the average (Wedekind & Milinski,

2000). This new approach to the evolution of cooperation is a specific instance of

a communication network: the interactions between individuals do not happen

in a social vacuum but in the presence of other individuals who eavesdrop and

thereby extract relevant information for own future interactions with the actors

(McGregor, 1993). As a consequence of eavesdropping, it pays individuals to alter

their behaviour, either as a general unconditional response (Johnstone, 2001) or

specifically in situations where bystanders are present (audience effects; see Ch. 4).


Indirect reciprocity in interactions in fish 523

Nowak & Sigmund (1998) proposed that altruism based on indirect reciprocity

is a hallmark of human evolution. Although they do not specify why this should

be so, their proposal implies that non-human animals either do not live in so-

cial environments that would favour the evolution of altruism through indirect

reciprocity or lack some of the cognitive abilities required (but see Zahavi, 1995;

Roberts, 1998). In contrast, we suggest that the communication network concept

has the advantage of coming from a purely functional perspective rather than

trying to explain apparently maladapted human behaviour. Early discussions of

communication networks (McGregor, 1993) argued that behaviours such as eaves-

dropping and audience effects should evolve in the context of a network, with-

out detailed consideration of the underlying cognitive mechanisms. By ignoring

mechanisms, we use the communication network concept to predict eavesdrop-

ping and audience effects in potentially cooperative contexts without worrying

about cognitive constraints. Experimental evidence for the existence of eaves-

dropping has been provided for a wide variety of taxa (Ch. 2). Whereas human

subjects can be asked about their behaviour, eavesdropping in other animals has

to be inferred from the eavesdroppers’ subsequent behaviour towards individu-

als observed interacting. Differences in individuals’ roles must elicit differences

in subsequent eavesdropper behaviour towards them. It was thus an implicit as-

sumption of communication-network studies that eavesdroppers attribute some

sort of image score to observed individuals and that this score governs their own

future behaviour towards those individuals. Scoring an individual’s tendency to

help is just one type of image score. Fighting ability, aggressiveness and mating

success with regard to female choice are the image scores typically studied in

communication networks (e.g. Ch. 5).

That eavesdroppers adjust their own behaviour to what they have witnessed has

important implications for the behaviour of individuals that are observed. While a

classical approach would suggest that individuals maximize payoffs in each single

interaction (with the exception of reciprocal altruism and punishment, where ben-

efits are delayed), any occurrence of eavesdropping implies that selection favours

individuals that optimize current actions by integrating both immediate payoffs

and future consequences of their behaviour. Within the framework of coopera-

tion theory, it may pay individuals to be altruistic if this increases the probability

of meeting more cooperative eavesdroppers in the future; in contest theory, it

may pay to be more aggressive in the presence of potential challengers if winning

a fight results in fewer attacks from eavesdroppers. Individuals can respond to

eavesdropping in two ways. First, they can alter their behaviour in any interaction

in relation to the average probability that eavesdroppers are present. In this case,

all individuals behave in the same way (with respect to eavesdroppers) in all inter-

actions. Second, individuals can pay attention to specific cues that eavesdroppers



are present for a particular interaction and alter their behaviour accordingly. In

the latter, individuals show a flexible behavioural pattern. Communication net-

work models have until now dealt with the first scenario (Nowak & Sigmund, 1998;

Johnstone, 2001; Leimar & Hammerstein, 2001). There is, however, increasing ev-

idence that animals adjust their behaviour in a particular interaction according

to the presence or absence of eavesdroppers (Doutrelant et al., 2001; Bshary, 2002;

Ch. 4).

Altruism towards unrelated individuals has been linked to positive reciprocity,

be it direct (Trivers, 1971) or indirect (Nowak & Sigmund, 1998): individuals help

because they will receive help in return. However, this is not necessarily the case.

An alternative is that an individual helps in order to raise its image and uses

its image to exploit recipients or eavesdroppers, which will behave cooperatively

because of this high image. If all individuals in a population exploited eavesdrop-

pers, then image scoring would break down. However, as long as image scoring

yields an overall benefit, either because most altruistic acts are honest in that

an individual’s willingness to cooperate is revealed or because the benefits of co-

operation exceed the costs of being exploited, altruism may be used both as an

honest and as a deceptive signal. Therefore, altruism may sometimes serve as a

signal out of context, causing other individuals to react in the signaller’s favour

and to their own disadvantage. This is the functional definition of tactical decep-

tion (Hauser, 1998). In communication-network terms, it may pay individuals to

be altruistic if this signal is misinterpreted by eavesdroppers in a way that allows

future exploitation of them.

Such a functional approach to tactical deception is in strong contrast to the

traditional cognitive approach. Though such behaviour has been described, for

example in birds (Munn, 1986), tactical deception is often seen as a hallmark

of primate ‘Machiavellian intelligence’ (Byrne & Whiten, 1988): the notion that

most primate species have been strongly selected for the cognitive abilities to

cope with their social environment (see references in Byrne & Whiten (1988) and

Ch. 25). The ability to use tactical deception has, therefore, been linked to the

concept of theory of mind (Premack & Woodruff, 1978): the ability to speculate

how another individual might perceive a certain situation. However, Heyes (1998)

cautioned that any observations of tactical deception do not imply the existence of

particularly high cognitive abilities. Instead, originally animals might have made

an error (i.e. produced a signal out of context) but it may have had a favourable

outcome for the signaller. As a result the signaller may associate this error with

a reward and consequently would be more likely to produce the signal again in

this context. The notion that simple associative learning might suffice to produce

signals that fit the functional definition of tactical deception offers the possibility

of using a much more functional, rather than mechanistic, approach to the topic



(see also Hauser, 1998) – and the appropriate framework for the study of tactical

deception is communication networks.

Interactions between cleaner fish and clients

In the remainder of this chapter, we present data on interactions between

the cleaner wrasse and client reef fish to illustrate the arguments outlined above.

Data were collected in the Red Sea, at Ras Mohammed National Park, Egypt. Meth-

ods of data collection are described in detail elsewhere (Bshary, 2001, 2002), so here

we will keep this kind of information to a minimum. All data are field observa-

tions; therefore, experimental proof is still lacking. However, these data illustrate

that it is worthwhile searching for potential examples of positive indirect reci-

procity and tactical deception with a functional perspective rather than worrying

about cognitive constraints.

Clients regularly visit the cleaners at their small territories called ‘cleaning

stations’ (cleaning mutualism reviewed by Losey et al., 1999; Cote, 2000). As in-

dividual cleaner wrasse may have more than 2000 interactions per day (Grutter,

1995), interactions often take place in the presence of other potential visitors.

Such bystanders can eavesdrop and evaluate the cleaner’s service quality. While

the cleaner fish eat parasites, in particular gnathiid isopods (Grutter, 1996), they

also feed on client mucus and scales (Randall, 1958; Grutter, 1997). Feeding on

healthy client tissue is correlated with the occurrence of client ‘jolts’, an observ-

able short shake of the client’s body, in response to mouth contact by the cleaner

fish (Bshary & Grutter, 2002a). The frequency of client jolts correlates negatively

with parasite load; therefore, client jolts are not a byproduct of parasite removal.

Rather, jolts are an easily observable correlate of cleaner fish cheating (Bshary, &

Grutter, 2002a). Note that only non-predatory clients (i.e. species that could not eat

cleaner fish) jolt on a regular basis, while jolts of predatory clients are infrequent

(Bshary, 2001). Therefore, we will only present data on non-predatory clients. In

response to a jolt, clients often dart off or chase the cleaner, depending on their

strategic options. Client species with large home ranges that cover several cleaning

stations (‘choosy clients’) usually make use of their choice options and swim off

and visit a different cleaning station for their next inspection (Bshary & Schaffer,

2002), as predicted by biological market theory (Noe et al., 1991; reviewed by Noe,

2001). In contrast, client species with small territories or home ranges, and hence

with access to only one cleaning station (‘resident clients’),tend to punish cleaners

by chasing them (Bshary & Grutter, 2002a). Both darting off and punishment could

readily provide bystanders with the information that a cleaning service was bad.

In contrast, if an observed interaction ends without apparent conflict, then the

service had probably been good. Therefore, clients could easily attribute an image



score to a cleaner fish, and cleaners could adjust their behaviour to the presence

of eavesdropping potential clients.

Why should clients attribute an image score to cleaners?

Attributing an image score to an individual and basing behavioural de-

cisions during interactions with that individual on that score only makes sense

if the score has some predictive power about how that individual will behave. In

potentially cooperative interactions, a positive image score should be attributed

to an individual only if cooperation on one occasion is usually followed by coop-

eration on the next occasion. In the context of cleaning mutualism, this implies

that there must be consistent variation in cheating rates either between indi-

vidual cleaners or within individual cleaners. Indeed, Bshary (2002) found that a

minority of cleaners cheated more frequently than the rest. These ‘biting cleaners’,

compared with normal cleaners, specifically targeted larger non-predatory clients,

both residents (median client jolt rate was 12/100 seconds in interactions with bit-

ing cleaners compared with 2/100 seconds in interactions with normal cleaners)

and choosy clients (18/100 seconds compared with 3/100 seconds), while there was

no evidence for increased cheating of predatory clients (0/100 seconds compared

with 0/100 seconds) or small resident clients (6/100 seconds compared with 6/100

seconds) (Bshary, 2002). These data suggest that it would pay larger non-predatory

clients to avoid interactions with such biting cleaners. One way they could do this

is to extract information from ongoing interactions and attribute an image to a

cleaner.

There is another reason why constant image scoring of cleaner behaviour is

advantageous for clients. Data from one biting cleaner fish revealed that cheating

rates changed considerably over a period of six weeks. Some of the 99% confi-

dence intervals around observed daily jolt rates of choosy clients did not overlap,

suggesting that the variation is significant (Fig. 22.1). This individual was a female,

as were all other biting cleaners that have been observed (n = 7). Cheating of non-

predatory choosy clients peaked at the two periods of full moon that occurred

during the observation period, and full moon coincided with repeated spawning

with her male partner. After the second spawning period, the male disappeared

and cheating rates fell to very low values. The cleaner wrasse is a protogynous

hermaphrodite; that is, individuals start their reproductive career as females and

eventually switch sex to become males (Robertson, 1972). Males have a larger repro-

ductive output because they often have a harem. Therefore, females face a trade-off

between investing in current reproductive effort through the production of eggs

and investing in growth to become a male. If the energy requirements for egg

production are maximal close to spawning, females needed extra energy in order

to avoid compromising growth too much. We suggest that the females’ switch to a



Cho

osy

clie

nt jo

lts/1

00 s

Fig. 22.1. Jolt rates (with 99% confidence intervals) of choosy clients (species with large

home ranges that cover several cleaning stations) when interacting with one

particular female cleaner on nine different days, based on one hour of observations

on each day.

temporarily deceptive strategy yields short-term energetic advantages. In aquaria,

clients jolt more frequently when interacting with hungry cleaners that when in-

teracting with satiated cleaners (A. S. Grutter, unpublished data). The benefits of

cheating, therefore, seem to be condition dependent, and the client control mech-

anisms like punishment (Bshary & Grutter, 2002a) and partner switching (Bshary

& Schaffer, 2002) only work most of the time.

It even appears that the same individual can switch back and forth between a

cooperative and a biting strategy within seconds. Another biting female, observed

over a six-week period, cheated clients frequently during the spawning period

but client jolt levels remained high after that. Her male partner tolerated her

presence at his cleaning station only during spawning but not thereafter and

chased her off repeatedly. The female spent about equal amounts of time at her

own cleaning station on the other side of the reef patch and on excursions to

the male’s cleaning station. When the female was at the male’s cleaning station,

her resident and choosy clients jolted significantly more frequently than when

interacting with the female at her own cleaning station (residents (n = 8): t = 1.5,

p = 0.021; choosy clients (n = 7): t = 0, p = 0.016; Wilcoxon matched-pair signed-

ranks tests; Fig. 22.2). The variation both within and between cleaners apparent

from these examples means that client image scoring is a profitable strategy to

avoid (temporarily) cheating cleaners.

Do clients attribute image scores to cleaners?

As shown elsewhere (Bshary, 2002), clients use information about the out-

come of ongoing interactions when visiting a cleaning station. To appreciate fully

what is happening, it is important to note that clients usually do not ‘hang out’ at

cleaning stations but visit them only when they seek an inspection by a cleaner.



Jolts

/100

s

C C

p p

Fig. 22.2. Jolt rates (medians and interquartile ranges) of eight resident species (clients

without choice; n = 8) and seven client species with choice (defined in text; n = 7)

during interactions with one particular female cleaner depending on the location of

the interaction (at her own station (•) or at the male’s station (�); see text for further

details).

Therefore, clients can gather information on how cleaners treat other clients only

when they visit the station themselves and only if a cleaner is busy inspecting

another client while they approach. So visiting clients can base their decision to

invite inspection on current information only if they can observe another client

being inspected. The newly arrived individual can attribute a positive image score

to the cleaner if the current interaction ends without apparent conflict and a

negative image score if the current interaction ends with the client darting off

or chasing the cleaner. If another client is not present when the client arrives,

no current information is available and the image score might be neutral. This is

what the data suggest. If an ongoing interaction ended without apparent conflict,

clients that had arrived during the interaction invited inspection by the cleaner in

almost 100% of observed interactions. In contrast, if the interaction ended with an

apparent conflict, clients hardly ever invited inspection (Bshary, 2002). When no

information about a cleaner’s previous interaction was available, clients invited

inspection with intermediate probability and the actual outcome of the previous

interaction (that was unobserved by the client) had no significant effect. When

clients do not invite inspection, they often exhibit an ambiguous response; they

let the cleaner approach and inspect but do not stop coordinated swimming move-

ments before the interaction starts (they may stop afterwards). They may also flee

from the approaching cleaner. Fleeing most often results in no inspection and fre-

quently happens when clients are approached immediately after an interaction

has ended with a conflict. In contrast, fleeing hardly ever occurs after a positive

interaction had just ended or if the previous interaction had ended a while ago

(Fig. 22.3). These observations are consistent with the statement above that clients

only visit cleaning stations to seek an inspection. The decision to invite inspection

is only altered if they observe a negative interaction.



P

P N

C

Fig. 22.3. The frequency of fleeing (accelerating away from approaching cleaners) by

resident client species (•) and choosy client species (defined in text, �) arriving at a

cleaning station in four different situations: the previous interaction ended either ≤5 s or ≤ 5 s ago and had positive (without conflict) or negative (with conflict)

outcomes. Values are median and interquartile ranges. The letters b and b′ above the

values for fleeing in the situation where the previous interaction had ended

negatively ≤ 5 s ago indicate a significant difference to the other three situations,

which are not statistically significant between each other, as indicated by using the

same letters a and a′.

Response of ‘normal’ cleaner fish to image-scoring clients

If clients attribute image scores to cleaners, one would expect that clean-

ers adjust their behaviour and cheat current clients less frequently if bystanders

are present than when no bystanders are present. Such audience effects should

be particularly common if bystanders have access to several cleaning stations, as

these species (see above) might not only delay their interaction with the cleaner

but also swim to another cleaning station. Resident bystanders can only delay

their interaction or avoid interactions altogether and remain uncleaned. To look

for such effects, we assumed that all individuals within 50 cm of cleaner–client

interactions at the beginning and at the end of each interaction were able to

collect information about the ongoing interaction and all individuals ≥ 10 cm

total length were potential next clients. We quantified the number of all such

individuals and their species identity for 12 cleaners. For each client species and

cleaner station, we calculated correlations between the frequency of jolts and the

number of bystanders. We analysed four (partly overlapping) classes of bystander;

conspecific, heterospecific, resident species and choosy species. We only calculated

correlations for observations where only one of these classes of bystander was

present. For each client species and bystander category, we compared the number



No.

spe

cies

C Heterospecifics R C

pp

Fig. 22.4. The influence of the presence (within 50 cm at the beginning and the end of

an interaction) of four bystander classes on the jolt rate of clients during interactions

with cleaners. The histograms show the number of client species for which the

correlation between jolt rate and the number of bystanders of a category was either

negative (black) or positive (white).

of positive and negative correlations and scored a plus if the majority was positive

and a minus if the majority was negative. Thus, we had one data point for each

client species and bystander category and evaluated any significant impacts of by-

stander categories on client jolt rates using sign tests. We did not find a significant

effect of conspecific bystanders on client jolt rates (n = 13; x = 6; NS) while the pres-

ence of heterospecific bystanders had a significantly negative effect (n = 23; x = 6;

p = 0.034; Fig. 22.4). The effect of heterospecific bystanders was mainly owing to

choosy bystanders (n = 17; x = 1; p < 0.001) while resident bystanders did not have

a significant effect on client jolt rates (n = 15; x = 6; NS; Fig. 22.4).

While the data presented above are in line with the hypothesis that client image

scoring influences cleaner fish behaviour, there is an alternative explanation. It

could be that when more clients are present it is easier for cleaners to pick the few

obvious parasites from each of them and the reduction in client jolt rate is a result

of an optimal foraging decision of cleaners rather than caused by bystander image

scoring. In favour of the optimal foraging interpretation, it is known that choosy

clients are, on average, larger than resident clients (Bshary 2001), which could

explain the stronger effect of their presence on the current clients’ jolt rates than

the smaller residents. However, optimal foraging cannot explain our observation

that choosy bystanders have different effects on cleaner fish behaviour, depending

on whether they are the same species as the interacting client or whether they are

a different species. We can explore this effect further by considering only inter-

actions in which individuals of one choosy species, the sergeant major Abudefduf

vaigiensis, were bystanders. We picked 13 cleaning stations for data collection on

the basis that these clients were frequent visitors. Sergeant majors may visit as

single individuals or as large shoals of 20–50 individuals. For 11 out of 12 client

species, we found more negative than positive correlations between the number

of sergeant majors present and client jolt rates (sign test: n = 12; x = 1; p < 0.01).

This result is the opposite of the effects of sergeant major bystanders on the jolt



1–40Bystanders

p

p

Clie

nt (

jolts

/100

s)

≥ 5

Fig. 22.5. The influence of the number of sergeant major Abudefduf vaigiensis

bystanders present (within 50 cm at the beginning and the end of an interaction) on

the jolt rate of the sergeant major client.

rates of sergeant majors being cleaned: in comparison to interactions in which no

bystanders were present, small numbers of bystanders did not have any detectable

effect and the presence of large numbers led to an increase in jolt rates (Friedman

test: n = 12; χ2 = 9.9; df = 2; p < 0.01; Fig. 22.5).

Response of biting cleaners to image scoring clients

Biting cleaners have more interactions that end with a conflict and clients

approaching their cleaning station more often avoid them than normal cleaners

(Bshary, 2002). As explained above, it seems likely that the latter observation is the

result of client image scoring rather than previous direct experience of clients.

Do biting cleaners still have some means to improve their image? In this respect,

it is important to note that biting cleaners behave very differently from normal

cleaners, not only with respect to jolt rates of large clients but also with respect to

their behaviour towards small resident clients. Biting cleaners often ride above the

small residents’dorsal area and provide tactile stimulation with their pectoral and

pelvic fins. While this behaviour is part of every cleaner’s repertoire, about 50% of

the interactions between biting cleaners and small residents consisted of tactile

stimulation only (Bshary, 2002). Providing tactile stimulation is incompatible with

foraging; hence interactions that consist of tactile stimulation only are clearly

costly to cleaners. Usually, cleaners provide tactile stimulation in response to the

behaviour of the client; for example, manipulating clients that are unwilling to

interact. The manipulation serves to slow down the clients, allowing the cleaners

to forage on the clients’ surface (Bshary & Wurth, 2001). As tactile stimulation of

small residents did not appear to provide the cleaners with any direct benefits from

the recipients, Bshary (2002) proposed that it may serve as a signal to attract image-

scoring clients, which can then be exploited. In line with this argument, it was

found that interactions that consisted of tactile stimulation only were followed



C

B N

p

Fig. 22.6. Frequencies with which biting and normal cleaners ignored the invitations

for inspection of small resident clients. Values are the median and interquartile

ranges for five biting and eleven normal cleaners. (The p value is derived from

Mann–Whitney U-test.)

by interactions ending with a conflict immediately after a client jolt more often

than expected. It appears that tactile stimulation of small residents is a signal out

of context that attracts image-scoring clients to their own disadvantage (they will

be cheated) and to the cleaners’ advantage, fulfilling the functional definition of

tactical deception (Hauser, 1998).

The presence or absence of bystanders was not noted, so it remains unclear

whether biting cleaners seek small residents in particular when larger clients are

nearby or whether they start such interactions independently of the presence of

bystanders. The latter scenario is more plausible, as larger clients, in particular the

choosy ones, are not willing to queue for inspection (Bshary & Schaffer, 2002) and

would, therefore, swim off despite the cleaner’spositive image. So, cleaners appear

unable to time interactions with small residents for maximal effects. However,

some evidence suggests that tactile stimulation of small residents is part of biting

cleaners’ strategies to improve their image. Cleaner fish sometimes ignore clients

that invite inspection, in particular small resident clients (Bshary & Wurth, 2001).

These clients do not offer a large food source and do not have the option to visit

another cleaner if ignored. When approached by small residents, biting cleaners

ignore them significantly less frequently than normal cleaners (Mann-Whitney

U-test: n = 11; m = 5; U = 2; p = 0.004; Fig. 22.6).

Discussion

We have provided a description of behavioural patterns in interactions

between cleaner fish and client reef fish that emphasizes the importance of the

communication-network framework in understanding the dynamics of coopera-

tive interactions and the occurrence of tactical deception. Cleaning interactions

often occur in the presence of other potential clients of cleaners. These bystanders

eavesdrop on ongoing interactions, and the information that they collect appears



to be crucial for their decision to invite inspection or to avoid the cleaner. In

response, it appears that normal cleaners reduce cheating frequencies in the pres-

ence of eavesdroppers, in particular if these eavesdroppers have access to several

cleaning stations. Data of this kind are still missing for biting cleaners. Biting clean-

ers frequently engage in costly (or at least non-profit) interactions with small resi-

dents that appear to serve to attract larger image-scoring clients, which can then

be exploited. The results have important implications for theoretical approaches

to indirect reciprocity. Existing models predict that image scoring drives altruistic

behaviour towards fixation (Nowak & Sigmund, 1998; Lotem et al., 1999; Leimar

& Hammerstein, 2001). Cheating individuals can only reinvade an image-scoring

population after genetic drift has led to an increase in non-discriminatory altru-

ists. This scenario does not fit the cleaner fish mutualism very well. Image scoring

of clients mediates cooperative behaviour of cleaners, but this cooperative be-

haviour may be an honest or a deceptive signal. Cheating individual cleaners use

one class of clients for altruistic behaviour to produce a signal that allows them

to exploit, through image scoring, another class of clients. Image scoring thus

works for the receiver of altruistic behaviour but it does not always work for the

eavesdropper.

The major difference between the cleaner fish system and the models might

concern the payoff matrix. While it is assumed in the models that payoffs are the

same in every interaction, payoffs are variable for cleaners. First, an advantage

of cheating clients with access to several cleaning stations rather than resident

clients is that the former just swim off after being cheated, while the latter chase

the cleaner fish around (Bshary & Grutter, 2002a), so the cleaner loses some of the

energy it has just gained. Second, cleaners can probably gain very little from in-

teractions with small clients anyway, no matter whether they cooperate or cheat.

This contrasts with interactions with large clients, which have more parasites but

also more mucus and a larger surface for the cleaner to scrape along with its lower

jaw. This gives the opportunity for cleaners to behave altruistically when payoffs

are low and to be exploitative when payoffs are high, as long as the altruism in-

creases the frequency of high payoff interactions. Image scoring would not persist

if it did not yield a benefit to its performer, but it also provides an opportunity

for individuals to perform altruistic acts in order to gain access to and exploit

image-scoring individuals. Therefore, image scoring in communication networks

may explain both the evolution of altruistic behaviour and the occurrence of

tactical deception. The commonness of dishonest signals that nevertheless still

fool observers has yet to be evaluated. While verbal arguments predicted low fre-

quencies (Dawkins & Krebs, 1978), game theoretic models indicate that this is

not necessarily the case (Johnstone & Grafen, 1993; Szamado, 2000). In particu-

lar, if the benefits of finding a cooperative partner largely outweigh the cost of



interacting with a cheating partner, tactical deception may occur at quite high

frequencies.

Future work with the cleaner system

Several important points of the cleaner fish mutualism still have to be

clarified. First, we need experimental evidence for both client image scoring and

cleaner fish audience effects. There is increasing evidence that a reduction in client

jolt rates in the presence of bystanders reflects a more cooperative behaviour by

cleaners (Bshary & Grutter, 2002b) and that such behaviour is indeed more altru-

istic. Grutter & Bshary (2003) offered cleaners the choice between equal amounts

of mucus, gnathiid isopods and monogeneans attached to plexiglas plates and

found that cleaners ate mucus more often than parasites, in particular gnathiids.

Assuming that the results reflect the items’quality as a food source, cleaners profit

even more from feeding on mucus when interacting with real clients as mucus is

abundantly spread over the clients’ surface whereas parasites have to be searched

for. In conclusion, while the experiment did not quantify energy intake, it makes

it very plausible that feeding on mucus yields a higher energy gain than feeding

on parasites. Another point that needs to be addressed is the biting cleaners’ be-

haviour with respect to small and large clients. Is it really true that interactions

with small clients generally offer low payoffs compared with interactions with

larger clients, and that the margin between the benefits from cooperation and

cheating increase with client size? Does image scoring of clients indeed inhibit

the cleaners’ tendency to cheat in low-payoff interactions but not in high-payoff

interactions? Currently, no data are available to evaluate these questions. Finally,

one might expect that clients should respond to the biting cleaners’ behaviour

by fine-tuning their image scoring, paying less attention to the outcome of inter-

actions between cleaners and small residents. Pooling of existing data indicate

that this is indeed the case. Invitation for inspection (i.e. spreading the pectoral

fins and stopping coordinated swimming movements) occurred more frequently

if cleaners interacted with choosy clients than with small residents. This prelim-

inary result has to be tested with a larger data set that allows statistical analysis

based on behaviour of individual client species.

Cognitive aspects

While indirect reciprocity and tactical deception were considered to be

a hallmark of human evolution (Nowak & Sigmund, 1998) and primate Machi-

avellian intelligence (Byrne & Whiten, 1988), the data on cleaner–client interac-

tions suggest that, on a purely descriptive level ignoring underlying mechanisms,

these phenomena are more widespread. We have argued that they should occur



frequently in social networks. These phenomena should occur if it pays to alter

the optimal behaviour in a situation in order to alter one’s image, which will, in

turn, produce benefits during future interactions with bystanders that exceed the

momentary costs. With respect to aggression, a game theoretic model shows that

it may even pay individuals to act spitefully towards a partner (in the sense that

the spiteful act will not lead to any benefits gained from future behaviour of

the recipient) if this spiteful act reduces, for example, the threat of attack from

bystanders (Johnstone & Bshary, 2004; Ch. 10).

With respect to cognition, the data generally support the view of Heyes (1998)

that we need to establish what kind of information animals use for their decision

making to find out what cognitive abilities are involved in a given phenomenon. In-

direct reciprocity and tactical deception may be something ‘smart’in some species

and simple conditioning in others. Cleaners certainly have ideal conditions to de-

velop their behaviour through conditioning. They have more than 2000 interac-

tions per day (Grutter, 1995), making it easy to connect altruistic behaviour with

reward (i.e. the invitation from bystanders to inspect) and cheating with punish-

ment (i.e. evasive actions of bystanders when approached by the cleaner). In the

absence of decisive experiments, it could even be possible that parts of cleaner

fish and client behaviour may be governed by endocrine responses rather than

through learning (Ch. 21). A good candidate for an endocrine-mediated behaviour

might be the good service that cleaners provide to predatory clients: there might

be an innate programme to recognize predators, and the presence of a predator

might trigger a stress response that, in turn, may inhibit cheating behaviour. Al-

ternatively, one could also generalize the Machiavellian intelligence hypothesis

and predict that a complex social network should have similar effects on cognitive

abilities in all species (Byrne & Whiten, 1988). It may turn out that cleaners have

relatively high cognitive abilities, as their large interspecific social network is at

least in part based on individual recognition (Tebbich et al., 2002) and demands

the solving of a variety of problems (Bshary et al., 2002). In this context, it is worth

pointing out that both the biting females which were observed over longer time

periods showed considerable variation in their behaviour, as so did their male part-

ners. One male often prevented his female from interacting with clients while the

other did not (Fisher test: n = 34; p = 0.003; Table 22.1). The preventive male was

also almost significantly more likely to chase his female when their client darted

off after a jolt, while the other male often followed the client to provide tactile

stimulation (Fisher test: n = 12; p = 0.053; Table 22.1). This observation of flexi-

bility of both males and females is important as it was the careful description of

individual-specific strategies in primates that eventually led to a cognitive, rather

than genetic, approach towards behaviour (Strum et al., 1998).



Table 22.1. The behaviour of two males that were partners of biting

females

Male responses to female biting

Tactile Preventative

stimulation Chasing No obvious chasing of

of clienta femalea reactiona femaleb

First male 5 2 3 1

Second male 0 5 2 16

a Reaction to clients darting off following cheating by female.b Keeping female away from clients.

Summary

In summary, we think that the concept of communication networks has

major implications for our understanding of the evolution and maintenance of

altruistic behaviour, tactical deception and spiteful behaviour. Because of its func-

tional approach, the communication-network framework may help to demystify

phenomena that are often considered to demand high cognitive abilities, opening

the way to focus on the underlying mechanisms and the complexity of information

processing and decision rules in order to illuminate cognitive differences between

species (for a parallel discussion, see Ch. 24). Game theory models should help to

generate testable predictions of the circumstances in which altruism, tactical de-

ception and spiteful behaviour may yield fitness benefits within communication

networks. In particular, it is time to develop cognitive models rather than genetic

models, allowing individuals to process information about their social environ-

ment before making a behavioural decision (see Stephens & Clements (1998) for a

first approach towards cognitive game theory).

Acknowledgements

We thank Peter McGregor for inviting us to write this chapter. We are grateful to the

EEAA in Cairo for the permit to work in the Park and to Alain de Grissac, the Park rangers and Ingo

Riepl for their support at the Park. The study was supported by the Deutsche Forschungsgemein-

schaft (grants BS 2/2-1 to BS 2/2-4) and written while RB was on a Marie Curie Fellowship of the EU.

We want to thank Wolfgang Wickler, Karin Bergmann and Barbara Knauer for additional support.

The chapter was greatly improved by comments from Peter McGregor, Rui Oliveira, Alexandre

Roulin, Sabine Tebbich and an anonymous referee.


Bibliography 537

References

Alexander, R. D. 1987. The Biology of Moral Systems. New York: Aldine de Gruyter.

Axelrod, R. & Hamilton, W. D. 1981. On the evolution of co-operation. Science, 211,

1390–1396.

Bshary, R. 2001. The cleaner fish market. In: Economics in Nature, ed. R. Noe, J. A. R. A. M.

van Hooff & P. Hammerstein. Cambridge, UK: Cambridge University Press,

pp. 146–172.

2002. Biting cleaner fish use altruism to deceive image scoring clients. Proceedings of

the Royal Society of London, Series B, 269, 2087–2093.

Bshary, R. & Grutter, A. S. 2002a. Experimental evidence that partner choice is the

driving force in the payoff distribution among cooperators or mutualists: the

cleaner fish case. Ecology Letters, 5, 130–136.

2002b. Asymmetric cheating opportunities and partner control in a cleaner fish

mutualism. Animal Behaviour, 63, 547–555.

Bshary, R. & Schaffer, D. 2002. Choosy reef fish select cleaner fish that provide high

service quality. Animal Behaviour, 63, 557–564.

Bshary, R. & Wurth, M. 2001. Cleaner fish Labroides dimidiatus manipulate client reef

fish by providing tactile stimulation. Proceedings of the Royal Society of London, Series

B, 268, 1495–1501.

Bshary, R., Wickler, W. & Fricke, H. 2002. Fish cognition: a primate’s eye view. Animal

Cognition, 5, 1–13.

Byrne, R. W. & Whiten, A. 1988. Machiavellian Intelligence. Oxford: Clarendon Press.

Cote, I. M. 2000. Evolution and ecology of cleaning symbioses in the sea. Oceanography

and Marine Biology Annual Review, 38, 311–355.

Dawkins, R. & Krebs, J. R. 1978. Animal signals: information or manipulation? In:

Behavioural Ecology: An Evolutionary Approach, ed. J. R. Krebs & N. B. Davies. Oxford:

Blackwell, pp. 282–309.


intrasexual communication in male Siamese fighting fish, Betta splendens.


Dugatkin, L. A. 1997. Cooperation among Animals: A Modern Perspective. Oxford: Oxford

University Press.

Grutter, A. S. 1995. Relationship between cleaning rates and ectoparasite loads in

coral reef fishes. Marine Ecology Progress Series, 118, 51–58.

1996. Parasite removal rates by the cleaner wrasse Labroides dimidiatus. Marine Ecology

Progress Series, 130, 61–70.

1997. Spatio-temporal variation and feeding selectivity in the diet of the cleaner fish

Labroides dimidiatus. Copeia, 1997, 346–355.

Grutter, A. S. & Bshary, R. 2003. Cleaner wrasse prefer client mucus: support for

partner control mechanisms in cleaning interactions. Proceedings of the Royal

Society of London, Series B, Biology Letters Supplement, 2, 242–244.

Hamilton, W. D. 1964. The genetical evolution of social behaviour. Journal of Theoretical

Biology, 7, 1–52.



Hauser, M. D. 1998. Minding the behaviour of deception. In: Machiavellian Intelligence II,

ed. A. Whiten & D. W. Byrne. Cambridge, UK: Cambridge University Press,

pp. 112–143.

Heyes, C. M. 1998. Theory of mind in non human primates. Behavioral and Brain Sciences,

21, 101–148.

Johnstone, R. A. 2001. Eavesdropping and animal conflict. Proceedings of the National


Johnstone, R. A. & Bshary, R. 2004. The evolution of spite through indirect reciprocity.


Johnstone, R. A. & Grafen, A. 1993. Dishonesty and the handicap principle. Animal

Behaviour, 46, 759–764.

Leimar, O. & Hammerstein, P. 2001. Evolution of cooperation through indirect


Losey, G. C., Grutter, A. S., Rosenquist, G., Mahon, J. L. & Zamzow, J. P. 1999. Cleaning

symbiosis: a review. In: Behaviour and Conservation of Littoral Fishes, ed. V. C. Almada,

R. F. Oliveira & E. J. Goncalves. Lisbon: Instituto Superior de Psicologia Aplicada,

pp. 379–395.

Lotem, A., Fishman, M. A. & Stone, L. 1999. Evolution of cooperation between

individuals. Nature, 400, 226–227.

2003. From reciprocity to unconditional altruism through signaling benefit.





Munn, C. A. 1986. Birds that ‘cry wolf’. Nature, 319, 143–145.

Noe, R. 2001. Biological markets: partner choice as the driving force behind the

evolution of cooperation. In: Economics in Nature, ed. R. Noe, J. A. R. A. M. van Hooff

& P. Hammerstein. Cambridge, UK: Cambridge University Press, pp. 92–118.

Noe, R., van Schaik, C. P. & van Hooff, J. A. R. A. M. 1991. The market effect: an

explanation for pay-off asymmetries among collaborating animals. Ethology, 87,

97–118.


Nature, 393, 573–577.

Premack, D. & Woodruff, G. 1978. Does the chimpanzee have a theory of mind?

Behavioral and Brain Sciences, 4, 515–526.

Randall, J. E. 1958. A review of labrid fish genus Labroides, with descriptions of two

new species and notes on ecology. Pacific Scientist, 12, 327–347.

Roberts, G. 1998. Competitive altruism: from reciprocity to the handicap principle.


Robertson, D. R. 1972. Social control of sex reversal in a coral-reef fish. Science, 177,

1007–1009.

Stephens, D. W. & Clements, K. C. 1998. Game theory and learning. In: Game theory and

Animal Behaviour, ed. L. A. Dugatkin & H. K. Reeve. Oxford: Oxford University Press,

pp. 239–260.



Strum, S. C., Forster, D. & Hutchins, E. 1998. Why machiavellian intelligence may not

be machiavellian. In: Machiavellian Intelligence II, ed. A. Whiten & D. W. Byrne.


Szamado, S. 2000. Cheating as a mixed strategy in a simple model of aggressive

communication. Animal Behaviour, 59, 221–230.

Tebbich, S., Bshary, R. & Grutter, A. S. 2002. Cleaner fish Labroides dimidiatus recognise

familiar clients. Animal Cognition, 5, 139–145.

Trivers, R. L. 1971. The evolution of reciprocal altruism. Quarterly Review of Biology, 46,

35–57.


Science, 288, 850–852.

Zahavi, A. 1995. Altruism as a handicap : the limitations of kin selection and

reciprocity. Journal of Avian Biology, 26, 1–3.

P1: IYP/... P2: IYP0521823617c23.xml CU1917B/McGregor 0 521 582361 7 April 12, 2005 12:49

23

Fish semiochemicals and the evolutionof communication networks

b r i a n d . w i s e n d e n 1 & no r m a n e . s t ac e y 2

1Minnesota State University, Moorhead, USA2University of Alberta, Edmonton, Canada

Introduction

The concept that animals typically communicate in networks (involving

at least one signaller and more than one receiver) derives from the active space of

signals and social spacing of conspecific and heterospecific receivers (McGregor

& Peake, 2000; Ch. 1). The ecological and evolutionary consequences of such net-

works have been explored most thoroughly for visual (e.g. Ch. 12) and acoustic

signals (e.g. Otter et al., 1999; Ch. 2), although it is clear that chemical signalling

also can involve networks (Chs. 11 and 16). Research on aquatic communication

networks has so far been limited to the context of visual and acoustic signalling

(e.g. Oliveira et al., 1998; Chs. 5 and 18). Semiochemicals (i.e. chemicals that trans-

fer information within and/or between species) exert important and diverse ef-

fects on the behaviour and physiology of aquatic animals (Liley, 1982; Chivers &

Smith, 1998; Kats & Dill, 1998; Sorensen & Stacey, 1999; Stacey & Sorensen, 2002;

Wisenden, 2003). Studies of two key aspects of fish chemical ecology (predator–

prey and reproductive interactions) have revealed great differences in the sources

and nature of the semiochemicals released, their active spaces and their biologi-

cal functions. These studies also provide sufficient information to assess, in fish,

the existence and function of semiochemical information networks, which we de-

fine more fully below as a general category of network that includes not only

communication networks employing specialized signals but also other networks

employing unspecialized cues. Here, we briefly describe well-studied examples of

predator–prey and reproductive semiochemicals to explore the applicability of cur-

rent communication network theory to aquatic chemical information networks


540


Fish semiochemicals 541

and consider how their function and evolution might differ from those employing

other sensory modalities and information-transmission media.

Research on intra- and interspecific transfer of chemical information in ter-

restrial species has generated a bewildering terminology related not only to the

nature, actions and functions of the chemicals but also to the concept of com-

munication (Hauser, 1996; Beauchamp, 2000; Hasson, 2000; McClintock, 2002).

Although we do not presume to clarify such a complex terminological problem in

this brief paper, it is imperative that we begin by clearly defining key terms, par-

ticularly as much of our subject matter appears to be distinct from that typically

discussed in the context of communication networks.

Definition of terminology

Semiochemicals include allomones and pheromones that, respectively, trans-

mit interspecific and intraspecific information. We will consider fish pheromones

and allomones involved in predator–prey interactions (p. 544) separately from

those involved in reproduction (p. 549). We define a pheromone as ‘a substance, or

mixture of substances, which is released by an individual and that evokes a specific

and adaptive response in conspecifics’ (Stacey & Sorensen, 2002). This definition is

more inclusive than the original definition of pheromone (Karlson & Luscher, 1959)

because, for reasons explained below, it omits any requirement that pheromones

be involved in communication. We use the terms releaser and primer not to classify

pheromones but only to describe their rapid behavioural and slower physiological

actions, respectively, for the simple reason that ‘it is quite possible for the same

pheromone to be both a releaser and a primer’ (Wilson & Bossert, 1963), as is the

case for sex pheromones of goldfish Carassius auratus (p. 549).

Central to our terminological schema is the concept (Sorensen & Stacey, 1999;

Stacey & Sorensen, 2002) that evolution of chemical communication progresses

through a series of three functional phases: ancestral, spying and communication. In

the ancestral phase, individuals (originators) release a chemical(s) that does not in-

fluence receivers (Fig. 23.1). This primitive, prepheromonal condition progresses

to spying if receivers evolve the ability to detect and respond adaptively to the

originator’s released chemical(s), now termed a pheromonal or allomonal cue(s).

In spying, originators may or may not benefit from the receiver’s response but,

importantly, remain in an unspecialized state with respect to production and re-

lease of pheromonal cues. Finally, spying progresses to communication if there is a

mechanism for receiver responses to select for specialization in production and/or

release of the detected cue(s), now termed a pheromonal or allomonal signal(s) and

released by a signaller. Signals evolve through natural selection because of fitness

benefits the signaller receives by manipulating the behaviour or physiology of

receivers. In many cases, signal senders and receivers form a mutualism in which


542 B. D. Wisenden & N. E. Stacey

Fig. 23.1. The evolution of communication from the ancestral state, where the

originator does not possess specializations for synthesis and release of

semiochemicals, to spying, where receivers possess specializations for detecting

semiochemicals but originators do not possess specializations, to true

communication, in which both originator (now signaller) and receiver possess

specializations for semiochemical exchange of information.

signals coevolve with the sensory biology of receivers. Once established, however,

mutually beneficial communicative relationships could be susceptible to deceit-

ful signal manipulations by signallers, which reduce the receiver fitness, as seen

in visual, acoustic and chemical signals (e.g. Lloyd, 1965; Møller, 1989; Paxton &

Tengo, 2001).

The ancestral state applies to released chemicals not currently functioning in

spying or communication. We restrict the terms signal and communication to

those situations in which there is clear evidence for signal specialization, such

as tissue hypertrophy or discrete structures for signal production; in contrast to

the situation in terrestrial insects and vertebrates, where pheromone-producing



glandular structures are common, such specializations in fish appear to be the ex-

ception rather than the rule (e.g. Laumen et al., 1974; Colombo et al., 1980; van den

Hurk & Resink, 1992). Consequently, we regard the great majority of fish predator–

prey and reproductive semiochemicals to function in spying, which is, in effect,

the default condition for cases where there is no evidence for specialization in

semiochemical production or release and/or where the social system apparently

precludes selection for signal specialization. It is to be expected that future re-

search may reveal some putative examples of spying to be true communication

because they involve previously undetected signal specialization.

Although we believe the distinction between cues and signals is fundamental

to an understanding of the function and evolution of semiochemical systems,

fish olfactory systems evidently do not make this distinction and process cues

and signals through similar mechanisms, which differ considerably from those

processing food odours (amino acids). Therefore, in comparison with food odours,

semiochemical cues and signals are detected by more sensitive and specific ol-

factory receptor mechanisms and generate neuronal activity that is processed in

distinct arrays (glomeruli) in the olfactory bulbs, is conducted to the brain by

distinct nerve bundlets (olfactory tracts), and is projected to distinct brain areas

(Sorensen et al., 1998; Hamdani et al., 2000, 2001; Brown et al., 2001; Stacey &

Sorensen, 2002). The distinction between chemical spying (via cues) and chemical

communication (via signals) highlights a dichotomy relevant not only to our un-

derstanding of semiochemicals (the functional relationships among originators,

signallers and receivers; evolutionary origins of species-specific cues and signals:

Sorensen & Stacey (1999)) but also to the concept of communication networks

(McGregor & Peake, 2000). In particular, first, how might networks involving com-

munication differ from those involving spying and, second, can the concept of

eavesdropping, defined as ‘extracting information from signalling interactions

between others’ (McGregor and Peake, 2000) be applied to information networks

that do not involve signalling?

Transfer of chemical information

Propagation of chemical information differs fundamentally from propa-

gation of visual and acoustic information. In general, visual and acoustic sig-

nals are propagated with predictable speed and direction, and they generate pre-

dictable active spaces throughout which much of the temporal information con-

tained in the signal’s initial pattern can be retained. In contrast, semiochemi-

cals are released into fluid media (air or water) in which local variation in flow

typically creates turbulent odour plumes, which not only distort or destroy tem-

poral pattern but also make the position, shape and size of the chemical’s active



space highly unpredictable (Weissburg, 2000). Moreover, semiochemicals can per-

sist in the environment for considerable time (e.g. Wisenden et al., 1995; Sorensen

et al., 2000; Polkinghorne et al., 2001), and thus can become disassociated from

originators/signallers either when currents carry away a transiently released

semiochemical or when the originator/signaller moves to a new location. Although

there is considerable information on the mechanisms by which some invertebrates

(e.g. crustaceans and moths) navigate in physically characterized odour plumes,

this complex issue is poorly understood in fish (Vickers, 2000). Finally, it is im-

portant to realise that semiochemical function in water also can be influenced

by additional solutes that affect olfactory response, such as heavy metals (Hansen

et al., 1999) and organics (Hubbard et al., 2002).

The olfactory system is similar to other sensory systems in being functionally

delimited by the sensitivity and specificity of its sensory neurons, but it differs

in the nature of the information it processes. Visual and acoustic systems process

linear arrays of light and sound frequencies in spectra common to many species,

particularly if they are related; olfactory systems process information from odor-

ants that cannot be arranged in a linear dimension by means of receptors that

are sensitive to one or a few chemicals. These differences have two important im-

plications for the nature and evolution of semiochemicals. First, whereas visual

and acoustic signals usually encode species-typical information in frequency and

temporal pattern, semiochemicals encode this information through the presence,

absence or ratio of specific odorants. Second, whereas visual and acoustic signals

are potentially detectable by all individuals and species sensitive to the emitted

spectra, semiochemical detection will be restricted to individuals with olfactory

receptors sensitive to the odorant(s). Thus, large differences in semiochemical

production and detection can occur with only small changes either in chemical

metabolism and release or in olfactory receptor specificity.

Assessment of predation risk

Natural selection strongly promotes attendance to cues that reduce the

probability of predation. Consequently, temporal and spatial variation in pre-

dation risk governs much of animal behaviour. Chemicals reliably inform about

predation risk because they are carried well in water, persist for ecologically appro-

priate amounts of time, transmit information through turbid or highly structured

habitat and darkness, and provide types of information not contained in visual

and acoustic modalities.

To apply communication-network theory to chemical assessment of predation

risk, we must first determine whether use of chemical information for the pur-

poses of risk assessment involves spying (via cues) or communication (via signals).



Fig. 23.2. Semiochemicals associated with predation. Predation escalates from initial

detection (top) to attack (middle) and, finally, ingestion (bottom). Chemical cues (solid

arrows) released at each stage inform nearby prey (conspecific and heterospecific) and

predators of the presence and extent of the interaction between predator and prey.

Known and hypothesized benefits are indicated by dotted arrows.

We conclude in the discussion below that, despite a plethora of semiochemically

mediated mechanisms for predator avoidance, evidence for signals is not com-

pelling. Although these information networks may not be communication networks

per se, there is evolutionary opportunity for communication networks and eaves-

dropping to evolve because receivers have evolved the ability to detect and respond

to many types of semiochemical (see below).

Chemicals correlated with predation

The literature concerning chemicals linked to predation has been re-

viewed elsewhere (Smith, 1992; Chivers & Smith, 1998; Kats & Dill, 1998; Wisenden,

2000, 2003; Chivers & Mirza, 2001) and only a brief overview will be presented here.

Several classes of chemical compound inform prey about predation risk. Gener-

ally, these cues are released passively before, during and after a predation event

(Fig. 23.2). Before an attack is initiated, prey can detect and respond to three types

of chemical cues: (a) odour of disturbed (startled but uninjured) prey (Chivers &

Smith, 1998; Wisenden, 2003), (b) species-specific kairomones (a predator’s natural

odour) (Kats & Dill, 1998) and (c) injury-released alarm cues of prey that leak from



the gut of the predator (Chivers & Mirza, 2001). When a predator attacks and in-

jures a prey organism, damaged prey tissues release chemical compounds that are

released only in this context; consequently, these cues reliably indicate risk and

elicit intense anti-predator behaviour (Chivers & Smith, 1998). These are alarm

cues. Most aquatic taxa exhibit anti-predator behaviour in response to alarm cues

(Chivers & Smith, 1998; Wisenden, 2003) in ways that reduce the probability of

predation (Hews, 1988; Mathis & Smith, 1993; Wisenden et al., 1999; Gazdewich &

Chivers, 2002). Ingested prey release alarm cues, or their metabolites, from the gut

of their predators (Chivers & Mirza, 2001). The ecological reality is undoubtedly

more complex than the interactions depicted in Fig. 23.2. Additional interactions

arise from variation in (a) diet breadth and overlap among predators, (b) relative

threat from each predator species over time and space, (c) interacting ontogenies

of prey and predator species, and (d) learned behavioural responses to correlates

of alarm cues.

The vast majority of chemical information used by aquatic prey to assess preda-

tion risk appears to be opportunistic use of chemical information mediated by un-

specialized chemical cues. This information is of great fitness benefit to receivers,

but receiver response, with one notable exception discussed below, generally

has not been shown to accrue benefit to the originator/signaller. Chemically me-

diated predation risk might be described most parsimoniously as an information

network, where a suite of prey species spy on the foraging activities of a suite of

predator species.

Ostariophysan alarm substance cells

For passively released chemical cues to qualify as signals, specializations

for their synthesis and/or release must occur that plausibly have been selected

for by benefits accruing to the originator/signaller. This condition appears to be

met in fishes of the superorder Ostariophysi (minnows, tetras, catfishes, suckers

and sundry others). This large group of vertebrates (> 5500 species) makes up ap-

proximately 27% of the global ichthyofauna and 64% of all freshwater fish species

(Nelson, 1994). In addition to successful occupation of a diverse array of habitats,

they are often the numerically dominant vertebrates in aquatic ecosystems.

Ostariophysans possess specialized epidermal cells that contain a potent alarm

chemical(s), termed schreckstoff or alarm substance (von Frisch, 1941; Pfeiffer, 1977;

Smith, 1992); this appears to activate components of the olfactory system that also

are activated by sex pheromones (Hamdani et al., 2000). It is not known how much

skin area is typically damaged during a predatory attack, but homogenates of

1 cm2 skin can create active spaces of 10 000 litres (zebrafish Danio rerio; Gandolfi

et al., 1968) to 58 000 litres (fathead minnow Pimephales promelas; Lawrence &

Smith, 1989), equivalent to spheres 2.6–4.8 m in diameter. The active ingredient in



ostariophysan alarm cells is likely, at least in part, to be hypoxanthine 3N-oxide,

a compound first isolated from European minnows Phoxinus phoxinus (Argentini,

1976; after Smith, 1999). Subsequent work demonstrated that hypoxanthine 3N-

oxide elicits anti-predator behaviour from a characin (Pfeiffer et al., 1985) and from

fathead minnows at concentrations as low as 0.4 nmol/l (Brown et al., 2001). How-

ever, efforts to detect hypoxanthine 3N-oxide in fathead minnow skin with high

performance liquid chromatography have not been successful (Smith, 1999) and

fractionation of skin extract indicates the biologically active component is found

with the polypeptides with molecular weights greater than 1100 (Kasumyan &

Ponomarev, 1987) rather than with the small molecules such as hypoxanthine 3N-

oxide. It is possible that hypoxanthine 3N-oxide is associated with protein while

within skin cells and remains associated with protein once released. Heat-treated

skin extract of fathead minnows loses 70% of its protein and its ability to elicit

alarm (N. L. Korpi, L. D. Louisiana, J. J. Provost & B. D. Wisenden, unpublished data).

A protein–hypoxanthine association would be consistent with cross-species alarm

reactions that decline with phylogenetic distance (Schutz, 1956). Whatever the

active ingredient(s) of alarm cue might be, their biological potency (Lawrence &

Smith, 1989; Brown et al., 2001) suggests selection for olfactory sensitivity similar

to that seen with sex pheromones (Stacey & Sorensen, 2002).

Is ostariophysan alarm substance a passively released cue or a specialized sig-

nal? Although the epidermal cells appear to be structures specialized for infor-

mation transfer of alarm, selection for signal specialization via benefits to the

originator (i.e. the individual that released the substance) is not immediately ap-

parent. There has been much speculation over the historical and current selection

benefits to individuals that invest in these cells (Smith, 1992, 1997; Williams,

1992; Magurran et al., 1996; Henderson et al., 1997). Smith (1992) summarized 16

hypotheses by which signallers may benefit from alarm signalling. One of these,

attraction of secondary predators (Fig. 23.2), has empirical support. Laboratory

and field experiments have demonstrated that predators are attracted to minnow

skin extract containing alarm substance cells over minnow skin lacking alarm

substance cells or skin extract from non–ostariophysan species (Mathis et al.,

1995; Wisenden & Thiel, 2002). Interruption of a predation event by the arrival of

a second predator allows prey an opportunity to escape (Chivers et al., 1996), a ben-

efit that elevates passively released cues to signal status. In this context, however,

this signal is not an alarm signal, but an attractant signal because the signaller

benefits because the from the responses of secondary predators, rather than from

responses of conspecific and heterospecific members of the prey community.

Is this a communication network? Are members of the prey community eaves-

droppers on the predator-attractant signal? Assessment of predation risk via chem-

ical cues does not depend on receiver (secondary predators) response (i.e. not social



eavesdropping) but interceptive eavesdropping (defined in Ch. 2) on this attrac-

tant signal provides highly salient temporal and contextual information about

predation risk. Therefore, from the perspective of the general non-ostariophysan

prey community, detection of injury-released chemical compounds may be con-

sidered as (interceptive) spying; however for the ostariophysan fishes, spying on

the predator-attractant signal might best be considered as a case of interceptive

eavesdropping (Ch. 2).

Evolutionary opportunities for communication networking

Several lines of evidence suggest potential for communication networks

in the ostariophysan alarm cue system. First, fishes frequently survive predatory

attacks (Smith & Lemly, 1986). An originator/signaller that survives an attack may

benefit from group behavioural responses of the prey community (Smith, 1992; Fig.

23.2) increased shoal cohesion and dashing or skittering behaviour that confuse

predators and reduce attack efficiency. Second, minnows associate alarm cues

with correlates of predation such as predator appearance and odour (reviewed

by Chivers & Smith, 1998). Alarm cues enable conspecifics and heterospecifics

to acquire predator recognition after a single simultaneous or non-simultaneous

encounter with a novel indicator of risk (Suboski, 1990; Suboski et al., 1990; Chivers

& Smith, 1994; Hall & Suboski, 1995; Korpi & Wisenden, 2001). Therefore, a second

benefit to the signaller could be providing shoalmates with an opportunity to learn

predator identity, as a shoalmate trained in this way may detect that predator in

the future and alert the signaller. If a shoal contains individuals related to the

signaller, then a third benefit might accrue to the signaller’s inclusive fitness

through kin selection.

In this context, eavesdropping on signaller–group communication in the os-

tariophysan system could occur when minnows observe the visual stimulus of

anti-predator behaviour of an alarmed shoal (Verheijen, 1956; Magurran, 1989;

Suboski et al., 1990; Brown et al., 1999).

Although evolutionary ecologists have focused on cells producing alarm sub-

stances in ostariophysans, these fishes are not unique in possessing specialized

epidermal cells (Smith, 1992). The epidermal layer of freshwater perch, wall-

eye and darters (superorder Acanthopterygii, order Perciformes, family Percidae:

Smith, 1979, 1982; Wisenden, 2003), Australian bullies (order Perciformes, family

Eleotridae: Kristensen & Closs, 2004) and poeciliids (superorder Acanthopterygii,

Order Cyprinodontiformes, family Poeciliidae: Bryant, 1987) all possess epidermal

club cells with similar histological properties. The tropical marine and freshwater

fishes in the Gobiidae (order Perciformes, 1875 species) possess epidermal vacuo-

late cells but have an inconsistent behavioural response to skin extract (Smith,



1992). This leaves open the possibility that analogous communication networks

for assessment of predation risk occur among other fish taxa.

In summary, it is parsimonious, based on current knowledge of chemical alarm

cues, to conclude that for aquatic taxa, including most fishes, chemically mediated

risk assessment does not constitute a true communication network because it is

not based on specialized signals. However, the ostariophysan alarm semiochemical

system appears to be a good candidate for an incipient communication system (see

p. 558). Future research may reveal signaller–group communication of alarm and

potentially uncover communication networks and eavesdropping.

Sex pheromones in information networks

In addition to alarm responses discussed in the previous section,

pheromonal cues and signals of teleost fish influence many diverse non-

reproductive (migration, parent–young interactions, schooling and related social

behaviours: Liley, 1982) and reproductive (Stacey et al., 1986; Stacey & Sorensen,

2002) phenomena. Best understood are those cases (the great majority being re-

productive) in which chemical identification has allowed study of pheromone

production, detection and biological effects under controlled and repeatable

conditions. Since Colombo et al. (1980) first proposed that the male black goby,

Gobius niger, releases a conjugated steroid (etiocholanolone glucuronide) to func-

tion as a pheromone that attracts the female to his nest for spawning, many studies

have reported putative pheromonal roles for steroid and prostaglandin hormones,

and their precursors and metabolites (hereafter termed hormonal pheromones) in a

variety of fish (reviewed by Sorensen & Stacey, 1999; Stacey & Sorensen, 2002).

Indeed, we expect the use of hormonal pheromones might be universal among

fish, given that information-rich hormones and hormonal metabolites are neces-

sarily released into the same water medium bathing the olfactory systems of con-

specifics. Here, we briefly discuss two species in which identification of distinctly

different reproductive pheromones has led to an understanding of pheromone

function germane to concepts of chemically mediated information networks.

Goldfish

The hormonal pheromones of goldfish are currently the best under-

stood of any fish and have recently been reviewed in detail (Sorensen & Stacey,

1999; Kobayashi et al., 2002; Stacey & Sorensen, 2002); therefore, we provide a

brief summary before considering aspects that appear directly related to con-

cepts of semiochemical information networks. Goldfish live in mixed-sex, appar-

ently unstructured, groups, undergoing gonadal growth during the winter and

spawning a number of times in spring and summer. At ovulation, which occurs


Fig. 23.3. Nature and actions of goldfish hormonal pheromones released by

periovulatory females (see Stacey & Sorensen (2002) for additional details and original

sources). (a) Female periovulatory events. An afternoon surge of pituitary (P)

gonadotrophin II (GTH-II) release induces follicular synthesis of

17α,20ß-dihydroxy-4-pregnen-3-one (17,20ß-P), which induces final maturation

(completion of arrested meiosis) in mature oocytes. When ovulation occurs

approximately 12 hours later, oocytes in the oviduct stimulate synthesis of

prostaglandin F2α (PGF2α), which remains at high concentrations in the blood until

ovulated oocytes are shed. (b) Preovulatory pheromone. During the GTH-II surge,

females release a changing mixture of three steroids: 17,20ß-P and androstenedione

(AD), which are released together across the gills, and a sulphated 17,20ß-P metabolite

(17,20ß-P-S), which is released in urine pulses. Peak release of AD (which inhibits

endocrine response to 17,20ß-P) occurs early in the GTH-II surge, followed by peaks of

17,20ß-P and 17,20ß-P-S release. The preovulatory steroid acts on specific and sensitive

(picomolar detection threshold) olfactory receptors, both inducing male behavioural

responses and, by the time of ovulation, increasing the quantity and quality of sperm

550



near dawn, groups of males vigorously compete for spawning access as females

repeatedly enter aquatic vegetation to oviposit adhesive, undefended eggs over

a period of several hours. In such a promiscuous mating system, where male

reproductive success likely depends only on the number of eggs fertilized, we

believe sperm competition has been a major selective force in the evolution of

male reproductive tactics.

The cascade of events leading to spawning begins when exogenous factors (in-

creased water temperature and aquatic vegetation) trigger an afternoon surge

release of pituitary gonadotrophin II (GTH-II), which stimulates follicular syn-

thesis of the oocyte maturation-inducing steroid 17α,20ß-dihydroxy-4-pregnen-

3-one (17,20ß-P) (Fig. 23.3a). Ovulation occurs approximately 12 hours later; at

which point females become sexually active for the several hours that eggs in the

oviduct stimulate synthesis of prostaglandin F2α (PGF2α), a behavioural hormone

(Fig. 23.3c). During the approximately 15 hours between the onset of the

GTH-II surge and completion of spawning, females sequentially release a preovula-

tory steroid pheromone (Fig. 23.3b) and a postovulatory prostaglandin pheromone

(Fig. 23.3c), which dramatically affect male physiology and behaviour.

The preovulatory steroid pheromone (Fig. 23.3b) is a dynamic mixture in which

the primary components appear to be 17,20ß-P, its sulphated metabolite (17,20ß-

P-S) and androstenedione (a testosterone precursor). Although the nature and ac-

tions of the preovulatory pheromone are complex (Stacey & Sorensen, 2002), it

induces in males both releaser effects on socio-sexual behaviours (e.g. Poling et al.,

2001) and a dramatic primer effect: a rapid increase in blood GTH-II that increases

both the quantity and quality of releasable stores of milt (sperm and seminal

fluids) in the sperm ducts prior to ovulation and spawning (e.g. Zheng et al., 1997).

At ovulation, females terminate release of the preovulatory steroid pheromone

and begin to release the postovulatory prostaglandin pheromone (PGF2α and its

more potent metabolite 15-keto-PGF2α) (Fig. 23.3c). The prostaglandin pheromone

not only triggers male courtship and attracts the male to the ovulated female

(anosmic males do not spawn) but also activates non-endocrine and endocrine

Fig. 23.3 (cont.) stores in the ducts (inducing GTH-II release, which stimulates

testicular 17,20ß-P synthesis). (c) Postovulatory pheromone. Entry of ovulated oocytes

to the oviduct stimulates synthesis of PGF2α, which acts in the brain (B) to stimulate

female sexual behaviours. PGF2α and its more potent metabolite 15-keto-PGF2α are

released in urinary pulses and act on olfactory receptors to trigger male sexual

behaviours. Sexual interactions then stimulate movement of sperm to the ducts by

two mechanisms: an endocrine mechanism distinct from that mediating testicular

response to the preovulatory pheromone; and a rapid and apparently non-endocrine

mechanism that begins to increase sperm stores within 15 minutes.



mechanisms (different from those mediating responses to the preovulatory

pheromone) that further increase the volume of releasable milt.

In summary, male goldfish first increase their potential fertility through en-

docrine responses to reliable chemical indicators of imminent ovulation (17,20ß-P

and 17,20ß-P-S) and then use reliable indicators that ovulation has occurred (PGF2α

and 15-keto-PGF2α) to locate the female and maintain sperm stores. For a number

of reasons (Stacey & Sorensen, 2002), most notably a lack of evidence for specialized

pheromone production and release, we regard these components of the goldfish

hormonal pheromone system as an example of male spying on female chemical

cues. Although it is difficult to exclude the possibility that domestication has in-

fluenced the goldfish pheromone system, it appears remarkably similar to those

of the closely related Crucian carp Carassius carassius and common carp Cyprinus

carpio (Irvine & Sorensen, 1993; Stacey et al., 1994; Bjerselius et al., 1995). Further-

more, it is likely that other cyprinids (Family Cyprinidae; > 2000 species) possess

similar pheromone systems given that olfactory detection of 17,20ß-P-like steroids

and prostaglandins is widespread among this taxon (Stacey & Sorensen, 2002).

The effects of goldfish pheromones described above have been studied in

the context of dyadic interactions between female originators of hormonal

pheromone cues and their male receivers (Fig. 23.3b,c). However, given the prox-

imity of individuals in aggregations, and the size of pheromonal active spaces

estimated from release rates and olfactory detection threshold (Sorensen et al.,

2000), it is obvious that these ovulatory cues normally operate in an information

network, where a female’s preovulatory steroids can potentially be detected by

many males and her postovulatory prostaglandins are the proximate trigger pro-

moting sperm competition at spawning. Moreover, the network activated by the

preovulatory pheromone evidently includes not only the ovulatory female and her

potential spawning partners but also additional females and males not directly

exposed to her preovulatory cues (Fig 23.4.).

The evidence for female interactions is based on the finding that low concen-

trations of water-borne 17,20ß-P induced ovulation in goldfish (Kobayashi et al.,

2002), suggesting a mechanism for the ovulatory synchrony observed in the field

and laboratory. The female benefit(s) of ovulatory synchrony is not known but

may involve predator swamping, amplification of preovulatory cues that stimu-

late male fertility, or (perhaps counter-intuitively) reduction of male to female

ratios at spawning (high ratios can result both in ‘forced’ egg release away from

suitable spawning substrate and skin damage through abrasion by the male’s

breeding tubercles or ‘pearl organs’).

Interactions among males appear more complex because they both decrease

(Fig. 23.4.a) and increase sperm stores in response to unidentified cues from other

males (Stacey et al., 2001; Fraser & Stacey, 2002). For example, males isolated from


m

fm

m

f

2

1

2

Tem

pera

ture

+su

bstr

ate

Fii

i

ii

(a)

(b)

iii

3

?

?

Fig.

23.4

.N

etw

ork

acti

vati

onby

ph

erom

ones

.(a)

Inth

ep

rese

nce

ofm

atu

refe

mal

esth

atar

en

otov

ula

tory

(f),

mat

ure

mal

es

rem

ain

ina

‘bas

al’e

nd

ocri

ne

stat

e(m

)an

dre

leas

eu

nkn

own

cues

that

sup

pre

sssp

erm

stor

es.(

b)If

exog

enou

sst

imu

li(i

ncr

ease

d

wat

erte

mp

erat

ure

and

aqu

atic

vege

tati

on)i

nd

uce

anov

ula

tory

gon

adot

rop

hin

IIsu

rge

inon

efe

mal

e(s

had

edgr

ey),

rele

ase

of

the

pre

ovu

lato

ryp

her

omon

eca

nd

irec

tly

stim

ula

teen

doc

rin

ech

ange

sin

one

mal

e(1

)th

atin

crea

seh

issp

erm

stor

e;h

eth

an

rele

ases

un

know

ncu

esin

du

cin

gsi

mil

arch

ange

sin

add

itio

nal

mal

es(2

).In

add

itio

n,t

he

pre

ovu

lato

ryp

her

omon

eca

nin

du

ce

ovu

lati

onin

add

itio

nal

fem

ales

(F),

wh

ose

pre

ovu

lato

ryp

her

omon

ere

leas

ew

ill

acti

vate

add

itio

nal

net

wor

ksof

dir

ectl

y(i

)an

d

ind

irec

tly

(ii)

stim

ula

ted

mal

es.I

tis

not

know

nif

mal

es(2

and

ii)t

hat

rece

ive

ind

irec

tin

form

atio

nth

atov

ula

tion

isim

min

ent

intu

rnin

flu

ence

add

itio

nal

mal

es(3

and

iii)

.

553



a male group dramatically increase sperm stores within 24 hours, indicating they

normally suppress their potential for milt production in response to an inhibitory

male cue(s). However, if sperm stores of one of a group of males are increased (either

by gonadotrophin injection or exposure to 17,20ß-P), untreated males in the group

also increase their stores. Therefore, it appears that, in the absence of cues from

preovulatory or ovulated females, a mature male goldfish is both originator and

receiver of unknown cues that suppress sperm stores in other males by maintain-

ing basal GTH-II and steroids (Fig. 23.4.a). This stable, negative-feedback situation is

rapidly and transiently perturbed, however, when exogenous stimuli trigger a pre-

ovulatory GTH-II surge in females, resulting in release of the preovulatory steroid

pheromone. Males and non-ovulatory females encountering this stimulatory cue

in turn increase their GTH-II, amplifying and disseminating the original cue(s) and

promoting synchronous final maturation (ovulation and increased sperm stores)

of individuals within the network (Fig. 23.4.b).

Numerous unresolved questions make it difficult to compare the complex re-

productive interactions of goldfish with the classical visual and acoustic commu-

nication networks that have been studied in terrestrial species. Perhaps the key

issue is whether the pheromonal interactions known among goldfish involve only

responses of receivers to unspecialized cues, or whether some are mediated by

specialized signals. There is no evidence that female preovulatory and postovu-

latory pheromones are specialized signals to males or other females (Sorensen

& Stacey, 1999; Stacey & Sorensen, 2002). Nor is it obvious how the male’s adap-

tive endocrine–testicular response to female preovulatory cues (Fig. 23.3b) would

also be shaped by selection to include the release of a specialized signal that ev-

idently increases the fertility of his competitors. Indeed, it seems more probable

that the indirect responses of males (2 and ii in Fig. 23.4.b) to female preovula-

tory cues are mediated by spying on unspecialized cues released as by-products of

the endocrine responses of males (1 and i in Fig. 23.4.b) directly stimulated by a

preovulatory female. In a species where males are territorial and where females

mate with several males on their territories, the interactions depicted in Fig. 23.4.b

might be expected to have arisen from female tactics to promote sperm competi-

tion. However, given that the female goldfish cannot control the number of males

competing for fertilization attempts, that virtually all her eggs can be fertilized

by a single male (Zheng et al., 1997) and that, as noted above, additional males may

disrupt spawning activity, we feel it most probable that the interactions depicted

in Fig. 23.4.b result solely from male competition.

We hope that our proposal that the hormonal pheromones of goldfish function

in spying interactions will stimulate discussion of how this and similar systems

can be integrated into current theoretical concepts of information networks based



on true communicative interactions. To open such a discussion, can we speak of

eavesdropping in spying networks (e.g. males 2 and ii in Fig 23.4.), given that

eavesdropping appears to be restricted to interactions mediated by specialized

signals (McGregor & Peake, 2000; Ch. 2) or do we require new terms and/or new

definitions?

Sea lamprey Petromyzon marinus

Anadromous sea lamprey Petromyzon marinus spend most of their life as

stream-dwelling, filter-feeding ammocoete larvae before undergoing a dramatic

metamorphosis, migrating to the ocean or large lakes and feeding parasitically

on large fish, whose unpredictable movements can carry the lamprey far from

their natal streams. After approximately a year, the parasites cease feeding, be-

gin to mature sexually and search for a spawning stream, guided by a potent

pheromone that serves as a reliable indicator of suitable larval habitat. Since gain-

ing access to the American Great Lakes from the Atlantic Ocean about a century

ago, the sea lamprey has seriously depleted many of these lakes’ fisheries. Based

on preliminary evidence (Teeter, 1980) that larvae release a pheromone attracting

migrating adults and that spawning adults employ sex pheromones, sea lamprey

pheromones have been extensively studied in the hope of identifying semiochemi-

cals for use in biological control, as has successfully been achieved for many insects

(Chapman, 2000).

Larval pheromone attracting migratory adults

Both field and laboratory studies provide compelling evidence that mi-

gratory adult lamprey do not return preferentially to natal streams but instead

locate suitable spawning habitat by responding to a pheromone released by stream-

dwelling larvae. Historical capture records show that estimated numbers of mi-

grating adults fall by up to 50% following application of larvicides to remove

larval populations (reviewed by Sorensen & Vrieze, 2003). Evidence that such re-

ductions in migrant numbers result from removal of larval odour comes from

studies of captive migrants in large two-choice mazes (Vrieze & Sorensen, 2001).

Water from streams without larvae is much less attractive to adults than is water

from larva-bearing streams, but water from streams without larvae becomes at-

tractive following addition of low concentrations of larval odour. The potency of

larval odour is such that a single larva (weighing only several grams) creates an

active space of 400–4000 l/h, sufficient to account for the attractive properties of

streams with larvae. These studies also reveal that spawning-stream selection is

based on more than larval odour alone: migratory adults prefer stream water (even



without larval odour) to lake water, suggesting the presence of unknown stream

odorants that act synergistically with larval odour (Vrieze & Sorensen, 2001).

The larval pheromone attracting adult migrants has been fully character-

ized and shown to be a mixture of chemicals (Vrieze & Sorensen, 2001; P. W.

Sorensen, personal communication); two of the primary components are the novel

bile acids, allocholic acid (ACA; 3α,7α,12α-trihydroxy-5α-cholan-24-oic acid) and

petromyzonol sulphate (PS; 3α,12α,24-trihydroxy-5α-cholan-24-sulphate). PS may

be a unique lamprey product and is synthesized by the liver of larvae but not by the

parasitic or adult phases (Polkinghorne et al., 2001). Because larvae undergo gall

bladder and bile duct atrophy at metamorphosis and also cease synthesis of PS and

ACA (Polkinghorne et al., 2001), these compounds should be specific indicators of

streams containing favourable spawning and nursery habitat. PS and ACA, which

are released primarily in larval faeces (Polkinghorne et al., 2001), are detected by

the olfactory organ of migratory adults (Li & Sorensen, 1997) not only with great

specificity, but also with a sensitivity (1 pmol/l olfactory detection threshold) that

would account for behavioural responsiveness at the low concentrations estimated

to occur in spawning streams (Polkinghorne et al., 2001). Furthermore, these bile

acids attract migratory adults (but not parasites) in maze tests (Bjerselius et al.,

2000; Vrieze & Sorensen, 2001).

Taken together, the results indicate that a suite of conspecific cues regulate

stream selection and upstream migration of maturing adult lamprey, and that

response to larval odour is adaptive in so far as it increases the likelihood of lo-

cating habitat suitable for larval growth. Moreover, because there is no evidence

at this time that larval production and release of PS and ACA are specialized for

functions other than digestion (Polkinghorne et al., 2001), and no evident mecha-

nism whereby adult response could select for specialized signalling functions for

these compounds, we regard these components of the migratory pheromone as

cues involved in chemical spying. Unlike the transient pheromonal steroid and

prostaglandin cues of goldfish, which are released only at specific stages of repro-

duction, however, the bile acid cues of lamprey appear to be released not only

during the period of peak adult migration in May but throughout the extended

period (April–August) of larval feeding (Sutton & Bowen, 1994; Polkinghorne et al.,

2001). In addition, whereas the transient pheromonal cues of female goldfish

are estimated to generate only small active spaces (Sorensen et al., 2000), PS and

ACA released by larval lamprey are estimated to create very large active spaces

sufficient to serve effectively as an upstream attractant given that larval popu-

lations can contain hundreds of thousands of individuals (Polkinghorne et al.,

2001). Perhaps the greatest departure from the goldfish situation, however, is

that, whereas goldfish pheromonal cues promote interactions of originators and

receivers within a small social unit, the lamprey larval pheromone functions in



a vast network of dispersed originators and receivers that do not interact be-

haviourally.

Sex pheromones

During upstream migration, adult male and female lamprey undergo

final maturation (spermiation and ovulation), lose behavioural responsiveness

to the larval pheromone and develop behavioural responsiveness to the odour

of mature conspecifics of the opposite sex (Bjerselius et al., 2000; Li et al., 2002).

Although the described behavioural responses of mature adults (positive rheotaxis,

increased locomotory behaviours) are rather non-specific, they are appropriate to

mediate upstream movement to spawning grounds and facilitate male–female

interactions, although this has not been demonstrated experimentally. However,

the traditional use of mature males to trap females (Fontaine, 1938; discussed in

Teeter, 1980) supports the existence of a potent male attractant, which is the only

lamprey sex pheromone to be studied intensively. This pheromone, estimated to

have a large active space (> 106 l/h per adult male (Li et al., 2002)), is proposed to

function in attracting females to mature males, which are reported to precede

females to the spawning grounds.

Major components of the pheromone released by spermiated male lam-

prey are proposed to be 3-keto-petromyzonol-sulphate (3-keto-PS; 7α,12α,24-

trihydroxy-3-one-5α-cholan-24-sulphate) and 3-keto-allocholic acid (3-keto-ACA;

7α,12α-dihydroxy-5α-cholan-3-one-24-oic acid) (Li et al., 2002; Yun et al., 2003). Al-

though both these compounds are detected by the lamprey olfactory system, only

3-keto-PS has been investigated for pheromonal activity. As with the odour of sper-

miated males, 3-keto-PS when added to a two-choice maze both attracts ovulated

females (but not preovulatory females or males) and stimulates their searching

behaviours (Li et al., 2002). Moreover, whereas non-spermiated males (whose odour

does not attract ovulated females in the maze) do not release appreciable quanti-

ties of 3-keto-PS, spermiated males release large quantities of 3-keto-PS (approxi-

mately 500 g/h) (Li et al., 2002; Yun et al., 2002).

As with the bile acid pheromone of larval lamprey (Polkinghorne et al., 2001),

3-keto-PS has been found in the liver of spermiated males (Li et al., 2002). However,

unlike the larval pheromone, which is released primarily in faeces, the pheromone

from spermiated males appears to be released by the gills, which in mature males

(but not females) develop glandular cells (Pickering, 1977) that evidently are spe-

cialized for pheromone release (Siefkes et al., 2003).

The current information on male lamprey pheromone suggests its synthesis

occurs through a subtle shift in bile acid metabolism that results in the larval

pattern of PS and ACA production changing to 3-keto-PS and 3-keto-ACA in spermi-

ating males (presence of the 3-keto acids in livers of ovulating females appears not



to have been examined). Furthermore, because fully mature adults are exposed to

larval and adult bile acids in spawning streams, it is expected that the lamprey ol-

factory system has been selected to discriminate larval (3-hydroxy acid) and adult

(3-keto acid) odours, although this remains to be examined.

The identified sex pheromones of lamprey and goldfish are similar in that

they operate within a complex network of originators/signallers and receivers,

although they differ fundamentally both in the interactions between genders

and in the ancestral (prepheromonal) functions of the cues and signals. Moreover,

the evidence for signal specialization in production and release of male lamprey

pheromone suggests a true communicatory interaction, which is unlikely in

goldfish.

Synthesis

Current theory about the function of animal communication networks

(e.g. McGregor & Peake, 2000) has been heavily influenced by studies of acoustic and

visual systems, where it seems clear that true communication between specialized

signallers and receivers has arisen through the bilateral benefits resulting from

their reciprocal interactions. Although studies of fish semiochemicals also provide

evidence of specializations indicative of communication, the specific functions of

such specialized semiochemicals within networks are not well understood.

In sea lamprey, for example, both the large active space of the proposed male

sex pheromone 3-keto-PS and apparent male-specific gill structure facilitating its

release (Li et al., 2002) suggest specializations for increased amplitude of a spe-

cialized tonic signal. The proposed function of this male lamprey signal appears

analogous to the aggregate signal produced by chorusing male anurans (Ch. 13),

in so far as the combined odour of many males induces the upstream movement

of many females. However, it remains to be determined if attracted female lam-

prey also use the male pheromone in mate choice and if this might have been the

pheromone’s original function.

Also, in the black goby, non-spermatogenic portions of the testes appear spe-

cialized for synthesis of a steroid pheromone, etiocholanolone glucuronide, orig-

inally proposed simply to attract ovulated females to the male’s nest (Colombo

et al., 1980). In the round goby Neogobius melanostomus, however, both males and fe-

males respond behaviourally to etiocholanolone glucuronide (Murphy et al., 2001),

suggesting that the pheromone functions in a more complex network involving

both intra- and intersexual communication.

Given that semiochemical communication appears to have evolved in sea lam-

prey and gobies, and perhaps in some other fish such as blennies (Laumen et al.,

1974; Goncalves et al., 2002) and African catfish Clarias gariepinus (van den Hurk &



Resink, 1992), these species may communicate in semiochemical networks analo-

gous to those seen in terrestrial systems involving acoustic and visual signals. How-

ever, other fish semiochemicals, such as the alarm cues of ostariophysans and the

sex pheromones of goldfish, appear to function not in communication but rather

in spying, where specialization for information transfer evidently is restricted to

receivers. Nonetheless, these semiochemical cues also operate in complex infor-

mation networks in which semiochemicals can influence several conspecifics both

directly (through exposure) and indirectly (through changes induced in exposed

individuals) (e.g. Figs. 23.2 and 23.4.b)

Because such fish semiochemical networks based on unspecialized cues have

the potential to give rise to true communication networks, they should not only

extend the scope of current network theory but also raise important issues rele-

vant to the evolutionary processes by which such communicatory networks evolve.

To cite just one example, when discussion of information networks is restricted

to those that involve communication, it might seem reasonable to assume that

eavesdropping arises only after communicative interaction has been established.

However, the ability of male goldfish to derive information indirectly about female

cues by spying on the responses of exposed males (e.g. Figs. 23.2 and 23.4.b) demon-

strates that a process analogous (and possibly homologous) to eavesdropping can

precede the origin of communication.

To promote discussion of the functional and evolutionary relationships among

spying, eavesdropping and communication, we propose two hypothetical schemes.

One is based on the intraspecific interactions induced by the goldfish preovulatory

steroid pheromone (Fig. 23.5a); the second involves both intra- and interspecific

predator–prey interactions in ostariophysan fishes (Fig. 23.5b), and both are de-

rived from our general model for the evolution of communication (Fig. 23.1).

In goldfish, spying by male receivers (R) on an unspecialized steroid cue released

by female originators (O; Fig. 23.5a1) could lead to communication (Fig. 23.5a2)

if male response to heritable variation in cue production leads to differential

female fitness. If this occurs, females would then be signallers (S) releasing a

specialized pheromonal signal and the male’s role would change (R1), as he now

influences, and is influenced by, signal evolution. As we emphasize in this chapter,

however, the goldfish preovulatory pheromone mediates more than the simple

dyadic spying event depicted in Fig. 23.5a1. The pheromone directly stimulates

behavioural and endocrine–testicular responses in more than one male (R) and

also induces a distinct response (ovulation) in females (Fig. 23.5a3). In addition,

the pheromone indirectly stimulates males (R2) via cues released by pheromone-

exposed males (Fig. 23.5a4).

In the ancestral condition of predator-induced prey chemical alarm cues, preda-

tor (P) attack releases general cues from the originator (O) that can be received both



spying

communication

R1

2

S

3

R

R

OO R

1

OO

R2

R

4

OOO

IE

R1

5

S

SE

R1

6

S

(a)

1

spying

communication

(b)R

OO

R

P2

Rr

3 4

OO P2

OC+

OO R

2 6 SE

S P3

IE

S P3

IE

R1

OC+

5

PPP

P P P

S

Fig. 23.5. Theoretical evolutionary pathways of the transition between spying and

communication networks involving semiochemicals used in reproductive (a) and

predator–prey (b) interactions. Thin solid and dashed arrows indicate spying

functions; thick, opposed, black and white arrows indicate communicative functions,

and large white arrows indicate transitions between proposed stable states. O,

originator; S, signaller; R, receiver (r, heterospecific receiver); IE, interceptive

eavesdropper; SE, social eavesdropper; P, predator; C, alarm cue. See text for further

explanation.

as an alarm cue by conspecific prey (R) and as a feeding cue by secondary preda-

tors (P2; Fig. 23.5b1). If interference by secondary predators benefits originators and

leads to alarm cue specialization, originators become signallers (S), the secondary

predator’s role changes (P3), and receiving conspecific prey become interceptive

eavesdroppers (IE) in a communication network (Fig. 23.5b2). As with the goldfish

pheromone (Fig. 23.5a3,4), predator-induced alarm cues can exert complex effects

prior to the evolution of communication. For example, alarm cues are used to



associate risk with stimuli (C) correlated with predation, which later serve as in-

dicators of predation risk (Fig. 23.5b3). This latter system may become elevated to

that of a communication network without involvement of a secondary predator

if an originator’s shoalmates learn to recognize a novel indicator of risk and later

alert the surviving originator to the presence of risk through early response to dan-

ger (Fig. 23.5b5). In direct relevance to the evolution of eavesdropping, alarm cues

can also affect predator–prey interactions indirectly through social facilitation

(social spying?) of alarm behaviour both in conspecifics (R2) and in heterospecifics

(r; Fig. 23.5b4).

If it is reasonable to assume that sex and alarm pheromone communication

evolves from spying, as depicted in Figs. 23.1, 23.5a1,2 and 23.5b1,2, then it also

seems reasonable to ask whether and how communication networks evolve from

spying networks. We, therefore, propose two general scenarios, which differ pri-

marily in the evolutionary origins of eavesdropping. In the first scenario, a simple

dyadic communication (Figs. 23.5a2 and 23.5b2) could lead to the evolution of

interceptive or social eavesdropping (Ch. 2) if receivers evolve adaptive responses

either to the signalling behaviour per se (interceptive eavesdropper (IE): Figs. 23.5a5

and 23.5b6) or to the signalling interaction (social eavesdropper (SE): Figs. 23.5a6

and 23.5b6). In this scenario, where the evolution of communication precedes that

of eavesdropping, eavesdropper functions (interceptive and social) are analogous

to the various receiver functions in spying networks (Figs. 23.5a3,4 and 23.5b3,4).

In the second scenario, incipient eavesdropping arises in spying networks, either

as direct (Figs. 23.5a3 and 23.5b4) or indirect (Figs. 23.5a4 and 23.5b4) spying by

receivers on originators and is retained as interceptive and social eavesdropping,

respectively, following the evolution of communication. In this scenario, receivers

in spying networks are homologous to eavesdroppers in communication networks.

In all the scenarios shown in Fig. 23.5, we depict eavesdropping in its proposed

initial state: that is, spying via a cue that is not specialized for transmission to

eavesdroppers, despite being a signal specialized for information transfer to the

primary target (Ch. 2). At this early stage, the network functions of eavesdropper

and primary target differ in kind. However, if subsequent selection by eavesdrop-

pers leads to signal specialization specific to the eavesdropping interaction, and

thus forming a communicative relationship between eavesdropper and signaller,

functions of eavesdroppers and receivers will come to differ only in degree. Studied

examples of eavesdropping (Ch. 2) typically appear to involve costs or benefits to

signallers that would be expected to modify signal function; consequently, it will

be important to determine whether, as has been suggested for sex pheromone

function in fish (Fig. 23.5a1), various forms of eavesdropping in communication

networks (Figs. 23.5a5,6 and 23.5b2,5,6) can persist as spying. Moreover, it will be

important to document covariance in the relative proportions of spying versus



communicative eavesdropping and the ecological and social factors that lead to

the spying–communication transition.

Acknowledgements

The authors gratefully acknowledge support from MSUM College of Social and Nat-

ural Sciences, MSUM Dille Fund for Excellence, MSUM Alumni Foundation, MnSCU Learning

by Doing (B. Wisenden) and the Natural Sciences and Engineering Research Council of Canada

(N. Stacey).

References

Argentini, M. 1976. Isoleirung des Schreckstoffes aus der Haut der Erlitz Phoxinus

phoxinus (L.). Ph.D. Thesis, University of Zurich.

Beauchamp, G. K. 2000. Defining pheromones. In: The Monell Connection, Fall 2000, ed.

L. J. Stein. Philadelphia, PA: Monell Chemical Senses Center, p. 2.

Bjerselius, R., Olsen, K. H. & Zheng, W. 1995. Endocrine, gonadal and behavioral

responses of male crucian carp (Carassius carassius) to the hormonal pheromone

17α,20β -dihydroxy-4-pregnen-3-one. Chemical Senses, 20, 221–230.

Bjerselius, R., Li, W., Teeter, J. H. et al. 2000. Direct behavioral evidence that unique

bile acids released by larval sea lamprey (Petromyzon marinus) function as a

migratory pheromone. Canadian Journal of Fisheries and Aquatic Sciences, 57,

557–569.

Brown, G. E., Godin, J.-G. J. & Pedersen, J. 1999. Fin-flicking behaviour: a visual

anti-predator alarm signal in a characin fish, Hemigrammus erythrozonus. Animal

Behaviour, 58, 469–475.

Brown, G. E., Adrian, J. C., Jr & Shih, M. L. 2001. Behavioural responses of fathead

minnows to hypoxanthine-3N-oxide at varying concentrations. Journal of Fish

Biology, 58, 1465–1470.

Bryant, P. B. 1987. A study of the alarm system in selected fishes of Northern

Mississippi. Ph.D. Thesis, University of Mississippi.

Chapman, R. F. 2000. Entomology in the twentieth century. Annual Reviews of

Entomology, 45, 261–285.

Chivers, D. P. & Mirza, R. S. 2001. Predator diet cues and the assessment of predation

risk by aquatic vertebrates: a review and prospectus. In Chemical Signals in

Vertebrates, ed. A. Marchlewska-Koj, J. Lepri & D. Muller-Schwarze. New York:


Chivers, D. P. & Smith, R. J. F. 1994. The role of experience and chemical alarm

signalling in predator recognition by fathead minnows, Pimephales promelas

Rafinesque. Journal of Fish Biology, 44, 273–285.

1998. Chemical alarm signalling in aquatic predator–prey systems: a review and

prospectus. Ecoscience, 5, 338–352.

Chivers, D. P., Brown, G. E. & Smith, R. J. F. 1996. The evolution of chemical alarm

signals: attracting predators benefits alarm signal senders. American Naturalist,

148, 649–659.



Colombo, L., Marconato, A., Belvedere, P. C. & Frisco, C. 1980. Endocrinology of teleost

reproduction: a testicular steroid pheromone in the black goby, Gobius jozo L.

Bolletino di Zoologia, 47, 355–364.

Fontaine, M. 1938. La lamproie marine. Sa peche et son importance economique.

Bulletin de la Societe Oceanographique de France, 17, 1681–1687.

Fraser, E. J. & Stacey, N. 2002. Isolation increases milt production in goldfish. Journal of

Experimental Zoology, 293, 511–524.

Gandolfi, G., Classon, L. J. & Rossi, A. C. 1968. The fright reaction of zebra fish. Atti

Societa Italiana di Scienze Naturali, 107, 74–88.

Gazdewich, K. J. & Chivers, D. P. 2002. Acquired recognition by fathead minnows:

influence of habitat characteristics on survival. Journal of Chemical Ecology, 28,

439–445.

Goncalves, D. M., Barata, E. N., Oliveira, R. F. & Canario, A. V. M. 2002. The role of male

visual and chemical cues on the activation of female courtship behaviour in the

sex-role reversed peacock blenny. Journal of Fish Biology, 61, 90–105.

Hall, D. & Suboski, M. D. 1995. Visual and olfactory stimuli in learned release of alarm

reactions by zebra danio fish (Brachydanio rerio). Neurobiology of Learning and

Memory, 63, 229–240.

Hamdani, E. H., Stabell, O. B., Alexander, G. & Døving, K. B. 2000. Alarm reaction in the

crucian carp is mediated by the medial bundle of the medial olfactory tract.

Chemical Senses, 25, 103–109.

Hamdani, E. H., Kasumyan, A. & Døving, K. B. 2001. Is feeding behaviour in crucian

carp mediated by the lateral olfactory tract? Chemical Senses, 26, 1133–1138.

Hansen, J. A., Rose, J. D., Jenkins, R. A., Gerow, K. G. & Bergman, H. L. 1999. Chinook

salmon (Oncorhynchus tshawytscha) and rainbow trout (Oncorhynchus mykiss) exposed

to copper: neurophysiological and histological effects on the olfactory system.

Environmental Toxicology and Chemistry, 18, 1979–1991.

Hasson, O. 2000. Knowledge, information, biases and signal assemblages. In:

Animal Signals: Signalling and Signal Design in Animal Communication, ed.

Y. Espmark, T. Amundsen & G. Rosenqvist. Trondheim: Tapir Academic Press,

pp. 119–139.

Hauser, M. 1996. The Evolution of Communication. Cambridge, MA: MIT Press.

Henderson, P. A., Irving, P. W. & Magurran, A. E. 1997. Fish pheromones and

evolutionary enigmas: a reply to Smith. Proceedings of the Royal Society of London,

Series B, 264, 451–453.

Hews, D. K. 1988. Alarm response in larval western toads, Bufo boreas: release of larval

chemicals by a natural predator and its effect on predator capture efficiency.


Hubbard, P. C., Barata, E. N. & Canario, A. V. M. 2002. Possible disruption of

pheromonal communication by humic acid in the goldfish, Carassius auratus.

Aquatic Toxicology, 60, 169–183.

Irvine, I. A. S. & Sorensen, P. W. 1993. Acute olfactory sensitivity of wild common carp,

Cyprinus carpio, to goldfish sex pheromones is influenced by gonadal maturity.


Karlson, P. & Luscher, M. 1959. ‘Pheromones’: a new term for a class of biologically

active substances. Nature, 183, 55–56.



Kasumyan, A. O. & Ponomarev V. Y. 1987. Biochemical features of alarm pheromone in

fish of the order Cypriniformes. Journal of Evolutionary Biochemistry and Physiology,

23, 20–24.

Kats, L. B. & Dill, L. M. 1998. The scent of death: chemosensory assessment of predation

risk by prey animals. Ecoscience, 5, 361–394.

Kobayashi, M., Sorensen, P. W. & Stacey, N. E. 2002. Hormonal and pheromonal control

of spawning behaviour in the goldfish. Fish Physiology and Biochemistry, 26, 71–84.

Korpi, N. L. & Wisenden, B. D. 2001. Learned recognition of novel predator odour by

zebra danios, Danio rerio, following time-shifted presentation of alarm cue and

predator odour. Environmental Biology of Fishes, 61, 205–211.

Kristensen, E. A. & Closs, G. P. 2004. Anti-predator response of naive and experienced

common bully to chemical alarm cues. Journal of Fish Biology, 64. 643–652.

Laumen, T. J., Pern, U. & Blum, V. 1974. Investigations on the function and hormonal

regulation of the anal appendices in Blennius pavo (Risso). Journal of Experimental

Zoology, 190, 47–56.

Lawrence, B. J. & Smith, R. J. F. 1989. Behavioral response of solitary fathead minnows,

Pimephales promelas, to alarm substance. Journal of Chemical Ecology, 15, 209–219.

Li, W. & Sorensen, P. W. 1997. Highly independent olfactory receptor sites for

conspecific bile acids in the sea lamprey, Petromyzon marinus, Journal of Comparative

Physiology A, 180, 429–438.

Li, W., Scott, A. P., Siefkes, M. J. et al. 2002. Bile acid secreted by male sea lamprey that

acts as a sex pheromone. Science, 296, 138–141.

Liley, N. R. 1982. Chemical communication in fish. Canadian Journal of Fisheries and

Aquatic Sciences, 39, 22–35.

Lloyd, J. E. 1965. Aggressive mimicry in Photuris: firefly femme fatales. Science, 149,

653–654.

Magurran, A. E. 1989. Acquired recognition of predator odour in the European

minnow (Phoxinus phoxinus). Ethology, 82, 216–223.

Magurran, A. E., Irving, P. W & Henderson, P. A. 1996. Is there a fish alarm pheromone?

A wild study and critique. Proceedings of the Royal Society of London, Series B, 263,

1551–1556.

Mathis, A. & Smith, R. J. F. 1993. Chemical alarm signals increase the survival time of

fathead minnows (Pimephales promelas) during encounters with northern pike

(Esox lucius). Behavioral Ecology, 4, 260–265.

Mathis, A., Chivers, D. P. & Smith, R. J. F. 1995. Chemical alarm signals: predator

deterrents or attractants? American Naturalist, 146, 994–1005.

1996. Cultural transmission of predator recognition in fishes: intraspecific and

interspecific learning. Animal Behaviour, 51, 185–201.

McClintock, M. 2002. Pheromones, odors, and vasanas: the neuroendocrinology of

social chemosignals in humans and animals. In: Hormones, Brain, and Behavior,

Vol. 2, ed. D. Pfaff, D. Arnold, A. Etgen, S. Fahrbach & R. Rubin. New York:






Møller, A. P. 1989. Deceptive use of alarm calls by male swallows, Hirundo rustica: a new

paternity guard. Behavioral Ecology, 1, 1–6.

Murphy, C. A., Stacey, N. E. & Corkum, L. D. 2001. Putative steroidal pheromones in the

round goby, Neogobius melanostomus: olfactory and behavioural responses. Journal

of Chemical Ecology, 27, 443–470.

Nelson, J. S. 1994. Fishes of the World, 4th edn. New York: Wiley-Interscience.




Otter, K. A., McGregor, P. K., Terry, A. M. R. et al. 1999. Do female great tits (Parus major)



Paxton, R. J. & Tengo, J. 2001. Double duped males: the sweet and sour of the orchid’s

bouquet. Trends in Ecology and Evolution, 16, 167–169.

Pfeiffer, W. 1977. The distribution of fright reaction and alarm substance cells in

fishes. Copeia, 1977, 653–665.

Pfeiffer, W., Riegelbauer, G., Meir, G. & Scheibler, B. 1985. Effect of

hypoxanthine-3(N)-oxide and hypoxanthine-1(N)-oxide on central nervous

excitation of the black tetra Gymnocorymbus ternetzi (Characidae, Ostariophysi,

Pisces) indicated by dorsal light response. Journal of Chemical Ecology, 11, 507–524.

Pickering, A. D. 1977. Sexual dimorphism in the gills of the spawning river lamprey,

Lampetra fluviatilis L. Cell and Tissue Research, 180, 1–10.

Poling, K. R., Fraser, E. J. & Sorensen, P. W. 2001. The three steroidal components of the

goldfish preovulatory pheromone signal evoke different behaviors in males.

Comparative Biochemistry and Physiology B, 129, 645–651.

Polkinghorne, C. N., Olson, J. N., Gallaher, D. G. & Sorensen, P. W. 2001. Larval sea

lamprey release two unique bile acids to the water at a rate sufficient to produce

detectable riverine pheromone plumes. Fish Physiology and Biochemistry, 24, 15–30.

Schutz, F. 1956. Vergleichende Untersuchungen uber die Schreckreacktion bei Fischen

und Verbreitung. Zeitschrift fur Vergleichende Physiologie, 38, 84–135.

Siefkes, M. J., Scott, A. P., Zielinski, B. et al. 2003. Male sea lampreys Petromyzon marinus,

excrete a sex pheromone from gill epithelia. Biology of Reproduction, 69, 125–132.

Smith, R. J. F. 1979. Alarm reaction of Iowa and johnny darters (Etheostoma, Percidae) to

chemicals from injured conspecifics. Canadian Journal of Zoology, 57, 1278–1282.

1982. Reaction of Percina nigrofasciata, Ammocrypta beani, and Etheostoma swaini

(Percidae, Pisces) to conspecific and intergeneric skin extracts. Canadian Journal of

Zoology, 17, 2253–2259.

1992. Alarm signals in fishes. Reviews in Fish Biology and Fisheries, 2, 33–63.

1997. Does one result trump all others? A response to Magurran, Irving and

Henderson. Proceedings of the Royal Society of London, Series B, 264, 445–450.

1999. What good is smelly stuff in the skin? Cross function and cross taxa effects in

fish ‘alarm substances’. In: Advances in Chemical Signals in Vertebrates, ed. R. E.

Johnston, D. Muller-Schwarze & P. W. Sorensen. New York: Kluwer

Academic/Plenum Press, pp. 66–84.



Smith, R. J. F. & Lemly, A. D. 1986. Survival of fathead minnows after injury by

predators: the possible role in the evolution of alarm signals. Environmental Biology

of Fishes, 15, 147–149.

Sorensen, P. W. & Stacey, N. E. 1999. Evolution and specialization of fish hormonal

pheromones. In: Advances in Chemical Signals in Vertebrates, ed. R. E. Johnston, D.

Muller-Schwarze & P. W. Sorensen. New York: Kluwer Academic/Plenum Press,

pp. 15–47.

Sorensen, P. W. & Vrieze, L. A. 2003. The chemical ecology and potential application of

the sea lamprey migratory pheromone. Journal of Great Lakes Research, 29 (Suppl. 1),

66–84.

Sorensen, P. W., Christensen, T. A. & Stacey, N. E. 1998. Discrimination of pheromonal

cues in fish: emerging parallels with insects. Current Opinion in Neurobiology, 8,

458–467.

Sorensen, P. W., Scott, A. P. & Kihslinger, R. L. 2000. How common hormonal

metabolites function as relatively specific pheromonal signals in goldfish. In:

Proceedings of the Sixth International Symposium on the Reproductive Physiology of Fish,

ed. B. Norberg, O. S. Kjesbu, G. L. Taranger, E. Andersson & S. O. Stefansson.

Bergen: John Grieg, pp. 125–128.

Stacey, N. E. & Sorensen, P. W. 2002. Hormonal pheromones in fish. In: Hormones, Brain,

and Behavior, Vol. 2, ed. D. Pfaff, D. Arnold, A. Etgen, S. Fahrbach & R. Rubin. New

York: Academic Press, pp. 375–435.

Stacey, N. E., Kyle, A. L. & Liley. N. R. 1986. Fish reproductive pheromones. In: Chemical

Signals in Vertebrates, Vol. 4, ed. D. Duvall, D. Muller-Schwarze & R. M. Silverstein.

New York: Plenum Press, pp. 117–133.

Stacey, N. E., Zheng, W. B. & Cardwell, J. R. 1994. Milt production in common carp

(Cyprinus carpio): stimulation by a goldfish steroid pheromone. Aquaculture, 127,

265–276.

Stacey, N. E., Fraser, E. J., Sorensen, P. W. & van der Kraak, G. J. 2001. Milt production

in goldfish: regulation by multiple social stimuli. Comparative Biochemistry and

Physiology C, 130, 467–476.

Suboski, M. D. 1990. Releaser-induced recognition learning. Psychology Review, 97,

271–284.

Suboski, M. D., Brian, S., Carty, A. E. et al. 1990. Alarm reaction in acquisition and

social transmission of simulated predator recognition by zebra danio fish

(Brachydanio rerio). Journal of Comparative Physiology, 104, 101–112.

Sutton, T. M. & Bowen, S. H. 1994. Significance of organic detritus in the diet of larval

lampreys in the Great Lakes basin. Canadian Journal of Fisheries and Aquatic Sciences,

51, 2380–2387.

Teeter, J. 1980. Pheromone communication in sea lampreys (Petromyzon marinus):

implications for population management. Canadian Journal of Fisheries and Aquatic

Sciences, 37, 2123–2132.

van den Hurk, R. & Resink, J. W. 1992. Male reproductive system as sex pheromone

producer in teleost fish. Journal of Experimental Zoology, 261, 204–213.

Verheijen, F. J. 1956 Transmission of fright reaction amongst a school of fish and the

underlying sensory mechanisms. Experientia, 12, 202–204.



Vickers, N. J. 2000. Mechanisms of animal navigation in odor plumes. Biological Bulletin

198, 203–212.

von Frisch, K. 1941. Uber einen Schreckstoff der Fischaut and seine biologische

Bedeutung. Zeitschrift Fur Vergleichende Physiologie, 29, 46–145.

Vrieze, L. A. & Sorensen, P. W. 2001. Laboratory assessment of the role of a larval

pheromone and natural stream odor in spawning stream localization by

migratory sea lamprey (Petromyzon marinus). Canadian Journal of Fisheries and Aquatic

Sciences, 58, 2374–2385.

Weissburg, M. J. 2000. The fluid dynamical context of chemosensory behavior.

Biological Bulletin, 198, 188–202.

Williams, G. C. 1992. Natural Selection: Domains, Levels and Challenges. Oxford: Oxford

University Press.

Wilson, E. O. & Bossert, W. H. 1963. Chemical communication among animals. In:

Recent Progress in Hormone Research, Vol. 19, ed. G. Pincus. New York: Academic

Press, pp. 673–716.

Wisenden, B. D. 2000. Scents of danger: the evolution of olfactory ornamentation in

chemically mediated predator–prey interactions. In: Animal Signals: Signalling and

Signal Design in Animal Communication, ed. Y. Espmark, T. Amundsen & G.

Rosenqvist. Trondheim: Tapir Academic Press, pp. 365–386.

2003. Chemically mediated strategies to counter predation. In: Sensory Assessment of

the Aquatic Environment, ed. S. P. Collin & N. J. Marshall. New York: Springer,

pp. 236–251.

Wisenden, B. D. & Thiel, T. A. 2002. Field verification of predator attraction to minnow

alarm substance. Journal of Chemical Ecology, 28, 433–438.

Wisenden, B. D., Chivers, D. P., Brown, G. E. & Smith, R. J. F. 1995. The role of experience

in risk assessment: avoidance of areas chemically labelled with fathead minnow

alarm pheromone by conspecifics and heterospecifics. Ecoscience, 2, 115–122.

Wisenden, B. D., Cline, A. & Sparkes, T. C. 1999. Survival benefit to anti-predator

behavior in the amphipod Gammarus minus in response to injury-released

chemical cues from conspecifics and heterospecifics. Ethology, 105, 407–414.

Yun, S.-S., Scott, A. P., Siefkes, M. & Li, W. 2002. Development and application of an

ELISA for a sex pheromone released by the male sea lamprey (Petromyzon marinus

L.). General and Comparative Endocrinology, 129, 163–170.

Yun, S.-S., Scott, A. P. & Li, W. 2003. Pheromones of the male sea lamprey, Petromyzon

marinus L: structural studies on a new compound, 3-keto allocholic acid, and

3-keto petromyzonol sulfate. Steroids, 68, 297–304.

Zheng, W., Strobeck, C. & Stacey, N. E. 1997. The steroid pheromone

17α,20β -dihydroxy-4-pregnen-3-one increases fertility and paternity in goldfish.

Journal of Experimental Biology, 200, 2833–2840.


24

Cognitive aspects of networks andavian capacities

i r e n e m . p e p p e r b e rg

Brandeis University, Waltham, USA

Introduction

The natural world is an extremely complicated place of myriad interac-

tions – some obvious, some hidden – but all of critical importance if one is to

understand its workings. Information must be processed, sorted, ignored or acted

upon by all creatures, even though the levels of processing ability vary across

species. Scientists, although well aware of these complexities and eager to make

sense of them, often begin by reducing interactions to their simplest form, un-

der the assumption that one can gain an understanding of more complex issues

by first gaining full knowledge of the simplest. Consequently, in most scientific

endeavours, initial studies examine the effect of a single stimulus on an entity:

in physics, how light waves interact with a single atom, or how two atoms might

interact; in child psychology, the reaction of an infant to a caretaker’ssmile or to a

novel toy; in animal behaviour laboratories, the effect of a shock on the behaviour

of a rat’smovement in a simple laboratory maze or the effect of a tape loop of song

on a bird in a sound isolation box. In each instance, however, the data obtained

provide only a small glimmer of the complexity that exists in the real world, and

in many cases inferences drawn from data in such experiments truly explain only

the specific laboratory situation being studied. To expand to a larger system and a

broader base often requires – and triggers – the development of more sophisticated

tools, be they mathematical theories (e.g. the Nash equilibrium), more powerful

computers for handling data or more sophisticated equipment for gathering data

(e.g. complex recording arrays). Sometimes, however, what is first needed is simply

the capacity to think outside of traditional forms of experimentation.


568


Cognitive aspects of networks and avian capacities 569

The study of animal (and particularly avian) cognition was one such paradigm

shift, and the consequent realization that animals needed to – and could – process

several sets of information (e.g. Pepperberg, 1990) was a logical outcome of such a

shift. The story begins with the so-called cognitive revolution (Hulse et al., 1968),

when researchers began to accept that levels and types of intelligence in non-

humans formed a continuum with those of humans, and to investigate a wide

range of behaviour and its development through many techniques in various

species. Most resulting studies, however, simply adapted material from human

cognitive experiments, often focused on a single set of tasks (i.e. not forcing sub-

jects to choose an appropriate set of responses from several possibilities, as they

must do in nature), continued to use a small number of species (predominantly

monkeys, rats, and pigeons) and made fairly sweeping conclusions as to the rel-

ative intelligence of all animals on the basis of these data (review in Pepperberg,

2001). Researchers, mostly in the laboratory but even in the field, initially failed

to examine species’ innate predispositions, evolutionary histories or ecological

constraints and, possibly most important of all, focused on how mostly social ani-

mals reacted in situations of social isolation. Although laboratory tasks presented

to animals may indeed have been cognitively complex, many animals failed to

demonstrate advanced capacities because of the specific nature of the task and

the situation in which the task was presented (for examples see Menzel & Juno,

1982).

Moreover, researchers often allowed their prejudices about animal capacities

to influence their hypotheses. The phylogenetic closeness of primates to humans

(e.g. Sarich & Cronin, 1977) and the large brains of cetaceans (e.g. Russell, 1979)

led scientists to anticipate and accept that their communicative and cognitive

capacities would be comparable to those of humans. (It should be noted however,

that Morgane et al. (1986) expressed concern that the dolphin brain, although large

even with respect to body size, may lack some of the complexity found in primates.)

Yet experimenters rarely expected analogous abilities in birds and failed to search

for such capacities. For many years, researchers argued that cognitive capacity

was likely a consequence of relative cortical size, and that birds, lacking much in

the way of cortical development, had to be inferior to mammals and primates (e.g.

Premack, 1978). My own research on the cognitive and communicative capacities

of the African grey parrots Psittacus erithacus is a particularly striking example of

these issues (Pepperberg, 1999): birds once thought to be capable merely of mind-

less mimicry have demonstrated, under appropriate experimental conditions,

referential use of elements of English speech and cognitive abilities (e.g. concepts

of category, number, bigger/smaller, same/different, absence) comparable to

those of a human child aged four to six years. Similarly, songbirds once thought

merely to be emitting sets of innately predisposed vocal patterns acquired during


570 I. M. Pepperberg

a brief stage in their lives, have demonstrated vocal behaviour indicating various

levels of cognitive processing and extensive memory. There are a number of good

examples. White-crowned sparrows Zonotrichia leucophrys, assumed to acquire only

their species-specific song during a limited sensitive period (Marler, 1970), show

flexibility in learning elements of other species’songs when living in a complex so-

cial environment, not only in the laboratory but also in areas of sympatry (Baptista

& Catchpole, 1989). Marsh wrens Cistothorus palustris actively choose which song in

their 100–400 song repertoire to use in competitive countersinging so as to match,

anticipate and possibly ‘jam’ the next song in their neighbours’ 100–400 song

series (Kroodsma, 1979; Kroodsma & Byers, 1998). Nightingales Luscinia megarhyn-

chos learn their 100 or so songs in chunks, much like humans learning long lists

(Todt & Hultsch, 1998). Numerous avian species recognize subtle variations that

differentiate their neighbours’ songs from those of strangers (Stoddard, 1996) and

some even remember neighbours’ songs from one year to the next (Godard, 1991).

Finally, if a song sparrow Melospiza melodia does not have an appropriate song type

in its repertoire for an exact match in a countersinging bout, it selects one that

is most similar (e.g. with the same introductory section,showing some level of

same/different comprehension (Burt et al., 2002)). However, even these examples

generally have focused on a single individual or one-on-one interactions and have,

therefore, to some extent ignored the real world: that these birds are actually part

of a larger network and that a countersinging bird, for example, would interact

over time with usually at least two or three individuals (i.e. all its territorial

neighbours), processing and storing all that information. Interestingly, advances

in field techniques, both in recording and playback (e.g. McGregor et al., 1992;

Naguib & Todt, 1997; Burt 2000), have not only allowed researchers to examine all

the information available to their subjects but have also led these researchers to

appreciate the complex cognitive processes that birds must be using to make sense

of this information. A specific avian case involves the relationship between the

complex cognitive task called ‘transitive inference’ and the natural situational

behaviour of ‘eavesdropping’ among networks of songbirds (Dabelsteen et al.,

1997; McGregor et al., 1997; Naguib & Todt, 1997; Naguib et al., 1999; Otter et al.,

1999; Peake et al., 2001, 2002; Mennill et al., 2002). A discussion of the complexity of

transitive inference, including the advances and pitfalls of laboratory work on the

topic and a brief review of data on eavesdropping (for details, see Ch. 2 and several

other chapters in this volume) will demonstrate some extent of avian cognitive

capacities.

Transitive inference

Transitive inference is one of several psychological tasks that engender

incredible amounts of discussion both as to the actual mechanism by which it is



performed and as to whether animals (particularly birds) are indeed capable of its

performance (reviewed by Zentall, 2001). The task, as originally stated for human

children, goes something like ‘Sam is taller than Bob. Bob is taller than Jack. Is

Jack shorter than Sam?’ (e.g. McGonigle & Chalmers, 1984a); sometimes the prob-

lem is given as ‘Bob is taller than Jack. Sam is taller than Bob. Who is shortest?’.

No training or rewards are involved when the task is given to children and the

child is assumed to understand the concept of taller (or bigger, stronger, etc.).

The child is thought to succeed by being able to integrate the two different vocal

pieces of information, including the reversal from taller to shorter, in a conscious,

cognitive manner; the complexity of the task derives from this integration and

reversal and the strong likelihood that the process requires a mental represen-

tation of the integrated pieces of information for success. In one instance, adult

subjects were given only five seconds to solve each of a series of transitive infer-

ence problems of various forms (‘Triangle is above circle. Square is below circle. Is

triangle above square?’ ‘Circle is darker than square. Circle is lighter than triangle.

Is triangle darker than square?’)and were then asked to report their reasoning pat-

terns (Egan, 1983). Subjects who used different reasoning strategies (e.g. ordering

the objects on a linear scale, which they then ‘scanned’, versus making individual

images of the objects, which they then compared) made different amounts and dif-

ferent types of reasoning error (the linear thinkers had approximately 10% errors

whereas the imaging group had approximately 38% errors). Use of a particular rea-

soning strategy was affected by aptitude for visualizing spatial transformations

of figures and the context in which reasoning problems were posed, but each

strategy involved some form of representation and integration: that is, cognitive

processing. Whatever the strategy, the connection is evident between such a task

and real-world knowledge of dominance hierarchies for any species living in a

network of individuals; therefore, researchers assumed that demonstrating this

understanding in non-human animals would be straightforward. Such, however,

has not been the case.

The task, as presented to non-humans, usually differs in a number of ways. First,

the number of contrasting pairs involved usually is at least five (note that some

studies such as those of McGonigle & Chalmers (1984b) use comparable numbers

for children). Second, a hungry animal undergoes extensive training on pairwise

comparisons where one of the pair is reinforced by a food reward (designated by +);

the other is not reinforced (designated by −); the amount and type of food re-

ward never varies. So the animal is trained to criterion on one pair (A+/B−),

then to criterion on the next pair (B+/C−), likewise for subsequent pairs (C+/D−,

D+/E−), and finally tested on an internal novel pair such as B/D to see which it

will choose. The elements of the pair to be tested have never individually been

shown to be the best or the worst and their relative worth has never been trained.

Third, the animal is not specifically cued, as humans generally are, that the task



involves relative judgements. Remember, humans are given specific verbal cues

that could be seen as the equivalent of ‘four chocolates if you chose A, three if

you choose B, two if you choose C, one if you choose D and none if you choose

E; now, do you prefer B over D?’ So in some sense the animals are given a task

that is more difficult than that initially given to children: the animal has not only

to learn and remember a series of comparisons but also to understand the point

of the query when presented with a novel pair, each member of which had ac-

tually led to reward in some instances. Rules that might have initially assisted

during training (e.g. ‘choose the familiar item in a new pair even if it was not

previously rewarded’) are of no use during testing, and additional rules developed

during pairwise training (‘choose what was most recently rewarded’) would be

misleading. Moreover, because the tests are not rewarded, the subject may not

even be able to learn through successive iterations. Note, however, that the ani-

mal does not need to engage in reversal (i.e. the taller-to-shorter change mentioned

above) and whether the animal is indeed cognitively engaged is unclear. Several

researchers argue that non-cognitive mechanisms based simply on reward might

be sufficient to explain the results of the pigeon subjects tested, which received

only a reward/no-reward condition within pairs of items (i.e. given no reason to

expect explicit relative relationships; see Couvillon & Bitterman, 1992; Wynne,

1997). At issue is the fact that the speed of acquisition and thus the number of

trials would differ somewhat for each of the pairs, and consequently more errors

would be made to some elements than others (see Zentall, 2001). Other researchers

argue for cognitive processes based on spatial mapping (e.g. Weaver et al., 1997),

such that the animals form some kind of linear set or mental representation to

which they can retrospectively refer during testing. Interestingly, pigeons given

explicit size cues to assist in forming a linear hierarchy did not learn any faster

nor were they more accurate than those without such cues (von Ferson, 1989),

suggesting that a linear model was not necessary for success. Possibly a form of

‘value transfer’ is involved, in that B, although never rewarded when given with A,

accrues some of A’s value of 100% reward, whereas D, although always rewarded

with E, loses some value by being connected to C, which accrues only 50% reward,

and E, which is never rewarded (Weaver et al., 1997). Although a number of differ-

ent species, from pigeons (Zentall, 2001) to chimpanzees (Gillan, 1981; Boysen et

al., 1993), appear to succeed, the mechanisms used by animals and humans might

differ.

The transitive inference studies my students and I are attempting with African

grey parrots, although not involving problems faced in the natural world, should

avoid these issues; I present the material to clarify how an experiment on transi-

tive inference could be performed in a laboratory to test whether a bird is using

a representational, cognitive mechanism. The oldest parrot, Alex, already vocally



designates the bigger/smaller of an object pair with respect to mass (Pepperberg &

Brezinsky, 1991) and can quantify collections of up to six objects with vocal

English number labels (Pepperberg, 1994); he is learning to label Arabic numerals

so that we can determine if he can combine these abilities to rank order number

symbols using transitive inference. Will he understand, without specific training

to associate Arabic numerals directly with their physical values, that the symbol

‘5’ is greater than the symbol ‘3’? Only a task using equivalence relations and

transitive inference can test this ability: he must use the commonality of English

to correlate (form equivalence relations between) quantity and Arabic numerals,

then use a form of transitive inference to identify the colour of one of a pair of

Arabic numbers that is bigger or smaller (e.g. a blue 3, red 5). To succeed, he must

base choice of ‘Arabic numeral X bigger/smaller than Arabic numeral Y’ on de-

ductions and on inferences: deduce that an Arabic numeral has the same value

as a vocal label, compare representations of quantity (mass) for which the numeral

stands, infer rank ordering based on these representations (transitive inference)

and then vocally report the result. Specific stimuli within pairs are not associated

with reward (Wynne, 1997), and by requiring colour, not number label responses,

rote replies cannot be used for a given pair. He has had no explicit training on

‘more/less than’rankings of individual elements (Arabic numerals) to be tested. The

task involves use of both working and long-term memory (Geary et al., 2000). Note,

too, that Alex was not trained to associate numbers with quantities sequentially:

He first learned 3 and 4, then 2 and 5, then 6 (Pepperberg, 1987), and he does

not produce vocal strings of number labels (i.e. does not say ‘one, two, three . . . ’);

therefore, he has had no training in rank ordering numbers whatsoever. Initial

trials are encouraging; he scored 15/18 in probes.

Although Alex’s task will provide some intriguing information on transitive

inference, the procedure still does not allow us to equate animal and human

studies. Consequently, given the differences between the standard animal and

human tasks, of particular interest are data collected when tasks more like those

given to the animals were given to adult humans. When adult humans were given

a non-vocal transitive task based on a computer game and not told that the game

involved transitive inference, only 70% succeeded (Siemann, 1993). In a different

experiment, when adult college students were given the exact same task as the

animals (Werner et al., 1992), their accuracy for the transitive pair was impressive

(approximately 95%), but only about two-thirds could explicitly state how they

solved the problem. When two groups of humans were given the non-vocal animal

task, with only one having been told that the task was inferential (Greene et al.,

2001), both groups succeeded, again suggesting that the processing need not be

conscious. Nevertheless, the informed group performed slightly better than those

in the uninformed group, many of whom, after testing, indicated that they had



inferred the hierarchical relationship. Finally, when human subjects are given the

vocal human task with five elements even without reversal (‘John is taller than

Bob. Bob is taller than Jim. Jim is taller than Richard. Richard is taller than David.

Who is taller, Bob or Richard?’), humans often fail because the information given

in such a task exceeds their memory capacity; many repetitions of the information

are necessary before they succeed (Delius & Siemann, 1998; see also Woocher

et al., 1978). Therefore, one can argue that the emphasis on overt hierarchical

presentation in the standard human vocal task (A > B, B > C, A ? C) provides strong

cues for its solution, and that in the longer case (A > B, B > C, C > D, D > E, B ? D)

some form of learning is likely involved. When the task involves specific sequential

training on abstract pairs and trial-and-error learning, however, subjects – be they

pigeons (and by inference birds in general) or humans – do not truly demonstrate

transitive inference; rather their results appear to be merely an artefact of the

training and reward situation. Furthermore, transitive inference involving groups

larger than three, even with explicit instruction as to the hierarchical nature of

the task, appears to be more difficult than expected.

The real issue, then, is not whether a pigeon can be taught something that has

the surface appearance of transitive inference, but whether birds (a) are indeed

capable of a task that has the same cognitive complexity of the human task and,

probably more importantly, (b) are faced with, and able to solve, such a task in the

real world. Although my research on the former issue is only in the earliest stages,

my findings suggest that a parrot, with vocal and cognitive capacities that resemble

those of very young children, is a good candidate for such a task (Pepperberg, 1999;

see comment in Delius et al., 2000). Some data on chickens (Hogue et al., 1996) and

recent field research on avian song (e.g. Peake et al., 2002), however, have provided

intriguing information that birds use a network of information to solve at least

simple transitive inference problems in their daily lives. Such data demonstrate

a level of cognitive processing unexpected in a creature with a brain not much

bigger than the size of a pea, or at most a shelled walnut, and a brain that is

organized so differently from that of humans (cf. Jarvis & Mello, 2000). Excluding

the chickens because they watched actual physical interactions, I will concentrate

on songbirds, where the data involve decisions based on vocalizations.

Songbirds and transitive inference

Songbirds, as noted above, live in a noisy environment of vocalizations

and other sounds of numerous species. Even if we make simplifying assumptions

that they rarely need to attend to the songs of other species or that they may not

need to learn to ignore such songs, they still need to learn the repertoires of their

neighbours. By so doing, they can determine, for example, whether, a territorial



encroachment is by a neighbour with whom they can resolve the boundary dispute

fairly quickly (e.g. Kroodsma, 1979; Beecher et al., 1996; Stoddard, 1996) or by a

stranger who may pose a serious threat (Stoddard, 1996). Recent research suggests

that songbirds also process and remember how their neighbours fare in territorial

disputes with other neighbours and strangers and, by eavesdropping, they become

aware of the relative dominance hierarchy of these birds and react with respect

to that information, a case of transitive inference. Note that, for example, an

unfamiliar floater male passing through a given area is quite likely to challenge

several residents; knowing one’s relationship to one’s neighbours and how one’s

neighbours have fared in such interactions could be advantageous.

Now, eavesdropping by itself does not provide direct evidence for transitive

inference, but it clearly sets the stage. Female black-capped chickadees Poecile

atricapillus, for example, attend to the vocal duels between males and make their

reproductive choices based on the outcomes (Mennill et al., 2002); they actively seek

extra-pair copulations when their high-ranking mate has lost an interaction with

a simulated intruder, and rarely if he has won. Data from such experiments (Ch. 7)

show that the females are capable of processing information from at least two

sources and making comparisons: A has beaten B, so B is less appealing. However,

such data merely suggest that rankings can be made on the basis of several com-

parisons; for transitive inference, the question is whether females rank a number of

different males and, if so, must they use overt interactions or can they interpolate

(true transitive inference)? Note, too, that several females will be competing for

the winning male in nature and female quality must also be taken into account.

An interesting study would be to see how females judge their relative quality and

whether they use some form of transitive inference.

At a different level are results from nightingales using a simulated playback

between two rivals; the target male noted which rival was overlapping the other

(a sign of dominance) and proceeded to respond more strongly to the overlapper

(Naguib & Todt, 1997; Naguib et al., 1999). The targeted male processed the inter-

action it heard, apparently viewed the overlapper as the greater threat and chose

to react as though he needed to establish dominance over the overlapper: that is,

given that A has beaten B, I had better beat A because he poses the greater threat;

the untested inference is that B will be attending and will not have to be dealt with

independently. Additional compelling data come from a study on male great tits

Parus major (Peake et al., 2002; Ch. 2), which also appear to base their interactions

with a simulated stranger on how that stranger has fared with a neighbour of

known rank with respect to themselves.

Another interesting case involves female great tits, who appear to decide

whether to enter a male neighbour’s territory based on eavesdropping upon ex-

perimentally manipulated interactions between a stranger and her mate and the



same stranger and said neighbouring male. The female makes her decision by

inferring the ranking of the two resident males based on their respective abilities

in dealing with the same intruder and is much more likely to enter the terri-

tory of the neighbour if he is inferred to be dominant to her mate (Otter et al.,

1999; Chs. 2 and 7; see also Fig. 2.1b, p. 20). Interestingly, because the relative

ranking of males chosen for the experiment was unknown and the choice was

random as to whether a given playback would simulate a dominant or a subordi-

nate interaction, the information might counter what she knows about previous

interactions between her mate and her neighbour. Even if her response was merely

to obtain more information, the experiment shows how much attention is paid

to such interactions. Of course, her mate might act differently overall after los-

ing a simulated encounter, and no one has yet observed natural interactions of

this type, although experiments are underway to examine this possibility (K. A.

Otter, personal communication; note that Mennill et al. (2002) did not observe any

post-playback behavioural differences in chickadees).

However, how do these interactions demonstrate cognitive complexity? Is the

level of complexity as great as it is in the human case? Is cognition involved at all,

or are some other mechanisms at play, as in the case of the trained pigeons?

Discussion

These field studies did not test several levels of inference as did occur in

the laboratory studies: that is, the birds were not exposed to a large number of

different interactions among a simulated intruder and several different neighbours

whose rankings were known and asked to rank the simulated intruder with respect

to an untested comparison with these birds. Yet the field studies did demonstrate

an interesting level of cognitive complexity. The male birds were asked to place

themselves inside the rankings and to determine how they were likely to fare in

a previously untested situation; the female birds appeared to act on an inference

based on their observations. To determine the specific complexity of the situation,

let us deconstruct two of the field tasks.

In one task involving male responses, the bird judges the relative worth of the

intruders in order to decide how to respond (Peake et al., 2002; see also Fig. 2.1e,

p. 20). First, the bird must recognize that stranger A is in its territory, duel with

it, then store its own rank with respect to that stranger. It must then attend to

an interaction outside its territory between stranger A and a second stranger, B,

and determine which has the higher rank. Subsequently, it must listen to one of

those strangers, determine that it was B and not A (with whom it had previously

interacted), remember this stranger’s rank with respect to the bird A with whom



it did interact, remember its rank with respect to A, and then infer whether it has

a chance against B or whether it should avoid B.

In the task involving female responses, a bird judges the relative worth of her

mate and neighbours based on the males’ interactions with a simulated intruder

in order to decide how to respond (Otter et al., 1999; see also Fig. 2.1b, p. 20). First,

the female has to distinguish her various neighbours from her mate and has likely

stored the relative worth of her mate M, and each neighbour N. She must identify

a new male, S, listen and determine his rank in a contest versus M, and then in

another contest versus N. She must store and compare these two rankings and

then infer the relative ranking of M and N based on their rankings with S, possibly

updating her stored original memory. Although not examined, of interest would

be whether she would try to search out S if he beat both M and N.

Conceivably, birds, like humans (Duchaine et al., 2001), have an easier time

making decisions that involve a social, familiar setting than they do if the same

decision is required in an abstract context (Greene et al., 2001). That is, these tasks

are somewhat simplified by being explicitly important to the bird’s survival and

its reproductive success. Nevertheless, both situations may involve reversal and

neither involves specific pairwise rewards; consequently, the likelihood of the

results being merely some experimental artefact as in the case of, for example,

the pigeon studies is unlikely.

Clearly, of future interest would be the addition of simulated interactions by

more intruders, C and D with A or B, and with the subjects’ other neighbours to

determine how many dominance relationships a bird might encode. Chickadees

may present an interesting case: their sense of overall ranking at winter feeding

stations may be settled well in advance of their daily interactions, because the

signals between any two birds landing at a feeding station are somewhat cursory

(Popp et al., 1990). Given that such flocks involve approximately a dozen birds, the

data suggest that individual birds may have some general understanding of their

rank on a global basis, that is, via transitive relationships (K. A. Otter, personal

communication). Could an experiment be designed to test whether (or how) a

subject could rank others independent of their relationships to himself, or would

a bird be able to rank others only in relationship to its need to avoid or engage in a

direct confrontation? Possibly the rankings with respect to self actually complicate

the issue, in that the bird must demonstrate some level of self-awareness as to

where it fits into the hierarchy.

Self-awareness is a separate but related issue in terms of animal abilities (e.g.

Griffin, 1998) and merits some discussion in the present case, at least for clar-

ification. Self-awareness, as used here, is distinguishable from ‘consciousness’:

the full-blown central monitoring of sensory inputs and mental states, executive



control of decision making and voluntary action, awareness of one’sown thoughts

(being aware that one is aware (Carruthers, 1992)) and attribution of mental states

to others (see discussion in Pepperberg & Lynn, 2000). Here, awareness describes

a state of higher-order cognition in which information is represented, processed

and used to control behaviour (e.g. Pepperberg, 1992). Thus the male tit is aware of

its relationship to A, aware of A’s relationship to B, and makes a decision based on

processing of these pieces of information; likewise, the female tit is aware of its re-

lationships to M and N, the relationships between these males and S, and makes a

decision based on the processing of these pieces of information. The point of evok-

ing awareness is that the tits will probably have a mental representation of these

pieces of information (e.g. Saidel, 2002) and use this representation in a ‘mind’s

eye’ view to make a decision; a researcher would be hard pressed to character-

ize the tits’ processing in any of the non-cognitive, non-representational manners

used to characterize the laboratory-based pigeon studies described above. The tits,

however, may not be consciously aware of their use of these representations (e.g.

be reacting to the specific situation in which, for example, its relationship to B

is unclear by consciously weighing all the possible risks and future benefits on a

personal basis and imputing the same to B, rather than reacting by chance based

on lack of information). Devising a test to uncover conscious processing would be

difficult.

Summary

In sum, at least some birds appear capable of solving transitive inference

tasks when dealing with a network of information, thus demonstrating complex

cognitive processing requiring the formation of several representations, exten-

sive memory for these various representations, and the ability to make inferences

based on a hierarchical organization of these representations. The situations pre-

sented to great tits (Otter et al., 1999; Peake et al., 2002) are at least as complicated as

those presented vocally to young children, and the results are not likely artefacts

of experimental manipulation. Should Alex’s preliminary data hold, African grey

parrots will also have succeeded in transitive inference. Although many objections

exist to evaluating animal intelligence and cognition based on human tasks (see

Pepperberg, 2001), the issue of importance here is that animals in nature are often

faced with the same types of task as their human counterparts (both at present and

historically) and so have been faced with the same evolutionary pressures on cog-

nitive development. In such circumstances, evaluating their competence on what

at first appears to be a human task does not ignore their natural behaviour, their

motivations, their ecological niches or their sensorimotor competence. Rather,

when animals are given tasks that are fully comparable to those given to humans



with respect both to ecological validity and experimental control, we can make

clear comparisons of animal and human abilities.

Acknowledgements

Writing of this manuscript was supported by the MIT School of Architecture and Plan-

ning and a grant from the American Foundation. Research on African grey parrots was supported by

NSF (IBN 96–03803) and REU supplements, the John Simon Guggenheim Foundation, the Kenneth

A. Scott Charitable Trust, the Pet Care Trust, the University of Arizona Undergraduate Biology

Research Program and many donors to the Alex Foundation.

References

Baptista, L. F. & Catchpole, C. K. 1989. Vocal mimicry and interspecific aggression in

songbirds: experiments using white-crowned sparrow imitation of song sparrow

song. Behaviour, 109, 247–257.


matching between neighbouring song sparrows. Animal Behaviour, 51, 917–923.

Boysen, S. T., Berntson, G. G., Shreyer, T. A. & Quigley, K. S. 1993. Processing of

ordinality and transitivity by chimpanzees (Pan troglodytes). Journal of Comparative


Burt, J. M. 2000. Use of a radio microphone array to study banded wren song

interactions at the neighborhood level. Journal of the Acoustical Society of America,

108, 2583.

Burt, J. M., Bard, S. C., Campbell, S. E. & Beecher, M. D. 2002. Alternative forms of song

matching in song sparrows. Animal Behaviour, 63, 1143–1151.

Carruthers, P. 1992. The Animals Issue: In Search of a Fundamental Theory. New York:


Couvillon, P. A. & Bitterman, M. E. 1992. A conventional conditioning analysis of

‘transitive inference’ in pigeons. Journal of Experimental Psychology: Animal Behavior

Processes, 18, 308–310.

Dabelsteen, T., McGregor, P. K., Holland, J., Tobais, J. & Pedersen, S. B. 1997. The signal

value of overlapping singing in male robins (Erithacus rubecula). Animal Behaviour,

53, 249–256.

Delius, J. D. & Siemann, M. 1998. Transitive responding in animals and humans:

exaptation rather than adaptation? Behavioural Processes, 42, 107–137.

Delius, J. D., Jitsumori, J. & Siemann, M. 2000. Stimulus equivalencies through

discrimination reversals. In: The Evolution of Cognition, ed. C. Heyes & L. Haber.

Cambridge, MA: MIT Press, pp. 103–122.

Duchaine, B., Cosmides, L. & Tooby, J. 2001. Evolutionary psychology and the brain.

Current Opinion in Neurobiology, 11, 225–230.

Egan, D. E. 1983. Retrospective reports reveal differences in people’s reasoning. Bell

System Technical Journal, 62, 1675–1697.



Geary, D. C., Hamson, C. O. & Hoard, M. K. 2000. Numerical and arithmetical

cognition: a longitudinal study of process and concept deficits in children with

learning disability. Journal of Experimental Child Psychology, 77, 236–263.

Gillan, D. J. 1981. Reasoning in the chimpanzee: II. Transitive inference. Journal of

Experimental Psychology: Animal Behavior Processes, 7, 150–164.

Godard, R. 1991. Long-term memory of individual neighbours in a migratory songbird.

Nature, 350, 228–229.

Greene, A. J., Spellman, B., Dusek, J. A., Eichenbaum, H. B. & Levy, W. B. 2001.

Relational learning with and without awareness: transitive inference using

nonverbal stimuli in humans. Memory & Cognition, 29, 893–902.

Griffin, D. R. 1998. From cognition to consciousness. Animal Cognition, 1, 3–16.




Hulse, S. H., Fowler, H. & Honig, W. K. (eds.) 1968. Cognitive Processes in Animal Behavior.

Hillsdale, NJ: Erlbaum.

Jarvis, E. D. & Mello, C. V. 2000. Molecular mapping of brain areas involved in parrot

vocal communication. Journal of Comparative Neurology, 419, 1–31.

Kroodsma, D. E. 1979. Vocal dueling among male marsh wrens: evidence for ritualized

expressions of dominance/subordinance. The Auk, 96, 506–515.

Kroodsma, D. E. & Byers, B. E. 1998. Songbird song repertoires: an ethological approach

to studying cognition. In: Animal Cognition in Nature, ed. R. P. Balda, I. M.

Pepperberg & A. C. Kamil. London: Academic Press, pp. 305–336.

Marler, P. 1970. A comparative approach to vocal learning: song development in

white-crowned sparrows. Journal of Comparative and Physiological Psychology, 71,

1–25.

McGonigle, B. & Chalmers, M. 1984a. The selective impact of question form and input

mode on the symbolic distance effect in children. Journal of Experimental Child


1984b. Are children any more logical than monkeys on the five-term series

problem? Journal of Experimental Child Psychology, 37, 355–377.








Menzel, E. W., Jr & Juno, C. 1982. Marmosets (Saguinus fuscicollis): are learning sets

learned? Science, 217, 750–752.

Morgane, P. J., Jacobs, M. S. & Galaburda, A. 1986. Evolutionary morphology of the

dolphin brain. In: Dolphin Cognition and Behavior: A Comparative Approach, ed. R. J.

Schusterman, J. A. Thomas & F. G. Wood. Hillsdale, NJ: Erlbaum, pp. 5–29.






leaders of simulated dyadic interactions. Proceedings of the Royal Society London,

Series B, 266, 537–542.



Proceedings of the Royal Society London, Series B, 265, 1045–1049.


eavesdrop on simulated male-to-male interaction. Proceedings of the Royal Society

London, Series B, 268, 1183–1187.


gathered by eavesdropping? Proceedings of the Royal Society London, Series B, 269,

1925–1929.

Pepperberg, I. M. 1987. Evidence for conceptual quantitative abilities in the African

grey parrot: labeling of cardinal sets. Ethology, 75, 37–61.

1990. Cognition in an African grey parrot (Psittacus erithacus): further evidence for

comprehension of categories and labels. Journal of Comparative Psychology, 104,

41–52.

1992. Proficient performance of a conjunctive, recursive task by an African grey

parrot (Psittacus erithacus). Journal of Comparative Psychology, 106, 295–305.

1994. Numerical competence in an African grey parrot. Journal of Comparative


1999. The Alex Studies: The Cognitive and Communicative Abilities of Grey Parrots.

Cambridge, MA: Harvard University Press.

2001. Evolution of avian intelligence. In: The Evolution of Intelligence, ed. R. Sternberg &

J. Kaufman. Mahwah, NJ: Erlbaum, pp. 315–337.

Pepperberg, I. M. & Brezinsky, M. V. 1991. Relational learning by an African grey parrot

(Psittacus erithacus): discriminations based on relative size. Journal of Comparative


Pepperberg, I. M. & Lynn, S. K. 2000. Possible levels of animal consciousness with

reference to grey parrots (Psittacus erithacus). American Zoologist, 40, 893–901.

Popp, J. W., Ficken, M. S. & Weise, C. M. 1990. How are agonistic encounters among

black-capped chickadees resolved? Animal Behaviour, 39, 980–986.

Premack, D. 1978. On the abstractness of human concepts: why it would be difficult to

talk to a pigeon. In: Cognitive Processes in Animal Behavior, ed. S. H. Hulse, H. Fowler &

W. K. Honig. Hillsdale, NJ: Erlbaum, pp. 421–451.

Russell, I. S. 1979. Brain size and intelligence: a comparative perspective. In: Brain,

Behavior and Evolution, ed. D. A. Oakley & H. C. Plotkin. London: Methuen,

pp. 126–153.

Saidel, E. 2002. Animal minds, human minds. In: The Cognitive Animal, ed. M. Bekoff,

C. Allen & G. M. Burghardt. Cambridge, MA: Bradford Books and MIT Press,

pp. 53–57.

Sarich, V. M & Cronin, J. E. 1977. Generation length and rates of hominid evolution.

Nature, 269, 354–355.

Siemann, M. 1993. Transitive Inferenz: Experimentelle Untersuchung einer

kognitiven Leistund. [Transitive inference: experimental investigation of a



cognitive performance.] Ph.D. Thesis, University of Konstanz. [Cited in Delius &

Siemann, 1998.]

Stoddard, P. K. 1996. Vocal recognition of neighbors by territorial passerines. In:

Ecology and Evolution of Acoustic Communication in Birds, ed. D. E. Kroodsma & E. H.


Todt, D. & Hultsch, H. 1998. Hierarchical learning, development and representation of

song. In: Animal Cognition in Nature, ed. R. Balda, I. M. Pepperberg & A. C. Kamil.

London: Academic Press, pp. 275–303.

von Ferson, L. 1989. Kognitive Prozesse bei Tauben (Columba livia). [Cognitive processes in

pigeons (Columba livia).] Pfaffenweiler: Centaurus.

Weaver, J. E., Steirn, J. N. & Zentall, T. R. 1997. Transitive inference in pigeons: control

for differential value transfer. Psychonomic Bulletin & Review, 4, 113–117.

Werner, U. B., Koeppl, U. & Delius, J. D. 1992. Transitive Inferenz bei nicht-verbaler

Aufgabendarbietung. [Transitive inference in nonverbal task presentation.]

Zeitschrift fur Experimentelle und Angewandte Psychologie, 39, 662–683.

Woocher, F. D., Glass, A. L. & Holyoak, K. J. 1978. Positional discriminability in linear

orderings. Memory & Cognition, 6, 165–173.

Wynne, C. D. L. 1997. Pigeon transitive inference: tests of simple accounts of a

complex performance. Behavioural Processes, 39, 95–112.

Zentall, T. R. 2001. The case for a cognitive approach to animal learning and behavior.



25

Social complexity and the informationacquired during eavesdropping byprimates and other animals

d o ro t h y l . c h e n e y & ro b e r t m . s e y fa r t h

University of Pennsylvania, Philadelphia, USA

Introduction

In many of the studies reviewed in this book, eavesdropping takes the

following form: a subject has the opportunity to monitor, or eavesdrop upon, an

interaction between two other animals, A and B. The subject then uses the informa-

tion obtained through these observations to assess A’s and B’s relative dominance

or attractiveness as a mate (e.g. Mennill et al., 2002; Ch. 2). For example, Oliveira

et al. (1998) found that male fighting fish Betta splendens that had witnessed two

other males involved in an aggressive interaction subsequently responded more

strongly to the loser of that interaction than the winner. Subjects’behaviour could

not have been influenced by any inherent differences between the two males, be-

cause subjects responded equally strongly to the winner and the loser of compet-

itive interactions they had not observed. Similarly, Peake et al. (2001) presented

male great tits Parus major with the opportunity to monitor an apparent competi-

tive interaction between two strangers by simulating a singing contest using two

loudspeakers. The relative timing of the singing bouts (as measured by the degree

of overlap between the two songs) provided information about each ‘contestant’s’

relative status. Following the singing interaction, one of the ‘contestants’ was

introduced into the male’s territory. Males responded significantly less strongly

to singers that had apparently just ‘lost’ the interaction (see also McGregor &

Dabelsteen, 1996; Naguib et al., 1999; Ch. 2).

What information does an individual acquire when it eavesdrops on others?

In theory, an eavesdropper could acquire information of many different sorts:

about A, about B, about the relationship between A and B, or about the place of


583


584 D. L. Cheney & R. M. Seyfarth

A’s and B’s relationship in a larger social framework. The exact information ac-

quired will probably reflect the particular species’ social structure. For example,

songbirds like great tits live in communities in which six or seven neighbours

surround each territory-holding male. Males appear to benefit from the knowl-

edge that certain individuals occupy specific areas (e.g. Brooks & Falls, 1975), that

competitive interactions between two different neighbours have particular out-

comes, and that these outcomes are stable over time. We would, therefore, expect

an eavesdropping great tit not only to learn that neighbour A was dominant to

neighbour B, for example, but also to form the expectation that A was likely to

defeat B in all future encounters. More speculatively, because the outcome of ter-

ritorial interactions are often site specific (reviewed by Bradbury & Vehrencamp,

1998), we would expect eavesdropping tits to learn further that A dominates B

in some areas but B dominates A in others. In contrast, the information gained

from monitoring neighbours’ interactions would unlikely be sufficient to allow

the eavesdropper to rank all of its neighbours in a linear dominance hierarchy,

because not all neighbouring males would come into contact with one another.

Such information would be difficult if not impossible to acquire; it might also be

of little functional value.

In contrast, species that live in large, permanent social groups have a much

greater opportunity to monitor the social interactions of many different indi-

viduals simultaneously. Monkey species such as baboons Papio cynocephalus, for

example, typically live in groups of 80 or more individuals, which include several

matrilineal families arranged in a stable, linear dominance rank order (Silk et al.,

1999). Offspring assume ranks similar to those of their mothers, and females main-

tain close bonds with their matrilineal kin throughout their lives. Cutting across

these stable long-term relationships based on rank and kinship are more tran-

sient bonds: for example, the temporary associations formed between unrelated

females whose infants are of similar ages, and the ‘friendships’ formed between

adult males and lactating females as an apparent adaptation against infanticide

(Palombit et al., 1997, 2001). In order to compete successfully within such groups, it

would seem advantageous for individuals to recognize who outranks whom, who

is closely bonded to whom, and who is likely to be allied to whom (Harcourt, 1988,

1992; Cheney & Seyfarth, 1990; see below). The ability to adopt a third party’s per-

spective and discriminate among the social relationships that exist among others

would seem to be of great selective benefit.

In this chapter, we review evidence for eavesdropping in selected primate

species and we consider what sort of information is acquired when one individual

observes or listens in on the interactions of others. We then compare eavesdrop-

ping by primates with eavesdropping in other animal species, focusing on both

potential differences and directions for further research.


Social complexity and eavesdropping 585

B1 threat-grunt + E1 scream B and E

B1 threat-grunt + D scream B and E

A threat-grunt + C scream B and E

Playback sequence Subjects

Test

Control 1(dominant’s kin)

Control 2(no kin)

Fig. 25.1. The protocol for playback experiments testing baboon females’ recognition

of other individuals’ kin. B, the more dominant of the subjects; E, the more

subordinate; B1 and E1, the subjects’ close kin; A, C and D, signallers unrelated to

either subject.

Knowledge about other animals’ kin

Some of the first evidence that monkeys recognize other individuals’ so-

cial relationships emerged as part of a relatively simple playback experiment de-

signed to document individual vocal recognition in vervet monkeys Cercopithecus

aethiops (Cheney & Seyfarth, 1980). We had noticed that mothers often ran to sup-

port their juvenile offspring when these individuals screamed during aggressive

interactions. This observation, like many others (e.g. Hansen, 1976; Gouzoules

et al., 1984), suggested that mothers recognized the calls of their offspring. To

test this hypothesis, we designed a playback experiment in which we played the

distress scream of a juvenile to a group of three adult females, one of whom

was the juvenile’s mother. As expected, mothers consistently looked toward the

loudspeaker for longer durations than did control females. Even before she had re-

sponded, however, a significant number of control females looked at the mother.

In so doing, they behaved as if they recognized not only the identity of signallers

unrelated to themselves but also associated those individuals with specific adult

females (Cheney & Seyfarth, 1980, 1982).

In an attempt to replicate these results, we carried out a similar set of ex-

periments on free-ranging baboons in the Okavango Delta of Botswana. In these

experiments, two unrelated female subjects were played a sequence of calls that

mimicked a fight between their close relatives (Fig. 25.1). The females’ immediate

responses to the playback were videotaped and both subjects were followed for

15 minutes after the playback to determine whether their behaviour was affected

by the calls they had heard. In separate trials, the same two subjects also heard two

control sequences of calls (Fig. 25.1). The first sequence mimicked a fight involving

the dominant subject’s relative and an individual unrelated to either female; the

second mimicked a fight involving two individuals who were both unrelated to

either female (for details see Cheney & Seyfarth, 1999).



0

1

2

3

4D

urat

ion

of lo

okin

g (s

) Both kinDominant kinNo kin

Dominant female(B)

Subordinate female(E)

Fig. 25.2. The duration that the subject looked at the other female following each type

of playback sequence. Histograms show means for 26 dyads in each of the three

conditions.

After hearing the test sequence, a significant number of subjects looked toward

the other female (Fig. 25.2), suggesting that they not only recognized the calls of

unrelated individuals but also associated these individuals with their kin (or close

associates). Females’ responses following the test sequence differed significantly

from their responses following control sequences. Following the first control se-

quence, when only the dominant subject’s relative appeared to be involved in the

fight, only the subordinate subject tended to look at her partner (Fig. 25.2). Fol-

lowing the second control sequence, when neither of the subjects’ relatives was

involved, neither subject looked at the other (Fig. 25.2). Finally, following a sig-

nificant proportion of test sequences, the dominant subject approached and sup-

planted (a mild form of aggression) the subordinate (Fig. 25.3). In contrast, when

the two subjects approached each other following the two control sequences, the

dominant rarely supplanted the subordinate (Fig. 25.3).

Taken together, these experiments suggest that baboons and vervet monkeys

recognize the individual identities of group members unrelated to themselves

and that they recognize the social relationships that exist among these animals.

Such knowledge can only be acquired by observing, or eavesdropping, on social

interactions in which the observer is not involved and making the appropriate

deductions.

Other studies provide additional evidence of monkeys’ability to distinguish the

close associates of other individuals. For example, in an experiment performed on

captive long-tailed macaques Macaca fascicularis, Dasser (1988a) trained a female

subject to choose between slides of one mother–offspring pair from her social



0

10

20

30

40

50

Dominantsupplants

Per

cent

age

of tr

ials

Both kinDominant kinNo kin

Dominantapproaches

Subordinateapproaches

Fig. 25.3. The percentage of subjects’ first interactions with each other that took

various forms following each playback sequence. Histograms show means for 26 dyads

in each condition. Dominant supplants indicates that the dominant subject

approached and supplanted the more subordinate subject. Dominant approaches

indicates that the dominant subject approached the subordinate subject without

supplanting her and/or interacted with her in a friendly manner. Subordinate

approaches indicates that the subordinate subject approached the dominant subject

and/or interacted with her in a friendly manner.

group and slides of two unrelated individuals from her group. Having learned to

respond to one mother–offspring pair, the subject was then tested with 14 novel

slides of familiar mothers and offspring paired with an equal number of novel

slides of familiar unrelated animals matched for age and sex. In all tests, she

correctly selected the mother–offspring pair. In so doing, she appeared to use an

abstract category to classify pairs of individuals that was analogous to our concept

of ‘mother–child affiliation’. Dasser (1988a) was able to exclude the possibility that

mothers and offspring were matched according to physical resemblance, because

subjects were unable to match unfamiliar mothers and offspring. Instead, indi-

viduals appeared to be classified according to their degree of association. Again,

such knowledge of other individuals’ close associates can only be obtained by

monitoring, or eavesdropping upon, their social interactions.

Under natural conditions, it is difficult to determine whether animals distin-

guish between different categories of social relationships. Do monkeys recognize,

for example, that mother–offspring bonds are distinct from sibling bonds or friend-

ships even when all are characterized by high rates of interaction? In perhaps the

only test of monkeys’ ability to recognize different categories of social affiliation,

Dasser (1988b) trained a long-tailed macaque to identify a pair of siblings from



her social group and then tested her ability to distinguish novel slides of familiar

sibling pairs from familiar mother–offspring pairs, familiar pairs of less-closely-

related matrilineal kin and familiar unrelated pairs. Although the subject did

distinguish siblings from unrelated pairs and pairs of less-closely-related individ-

uals, she was unable to discriminate between siblings and mothers and offspring.

This failure may have occurred because the same female had previously been re-

warded for picking the mother–offspring pair. It is also possible, however, that

she did not distinguish between different kinship categories and simply chose the

pair that was more closely affiliated.

Natural patterns of aggression also reflect the knowledge that monkeys have

of their group’s social network. In many monkey species, an individual who has

just threatened or been threatened by another animal will often ‘redirect aggres-

sion’ by threatening a third, previously uninvolved, individual. Judge (1982) was

the first to note that redirected aggression in rhesus macaques Macaca mulatta

does not always occur at random. Rather than simply threatening any nearby

individual, animals will instead specifically target a close matrilineal relative of

their recent opponent. Similar kin-biased redirected aggression occurs in Japanese

macaques Macaca fuscata (Aureli et al., 1992) and vervets (Cheney & Seyfarth, 1986,

1989). Kazem & Aureli (Ch. 10) further discuss the relationship between redirected

aggression and communication networks.

Knowledge about other animals’ dominance ranks

Dominance ranks offer another opportunity to test whether non-human

primates gain information about other animals’relationships by eavesdropping on

their social interactions. Like matrilineal kinship, linear, transitive dominance re-

lations are a pervasive feature of social behaviour in groups of Old World monkeys.

A linear, transitive rank order might emerge because individuals simply recognize

who is dominant or subordinate to themselves. In this case, a linear hierarchy

would occur as an incidental outcome of paired interactions and there would be

no evidence to suggest that animals eavesdropped on others’interactions. Alterna-

tively, a linear hierarchy might emerge because individuals genuinely recognize

the transitive dominance relations that exist among others: a middle-ranking in-

dividual, for example, might know that A is dominant to B and B is dominant

to C and, therefore, conclude that A must be dominant to C. Like knowledge of

matrilineal kin, such knowledge could only be acquired through eavesdropping

on the interactions of others.

In many species of Old World monkeys, female dominance ranks are deter-

mined by the rank of an individual’smatriline (Walters & Seyfarth, 1987; Chapais,

1988). Knowledge of another female’s rank cannot, therefore, be obtained by



attending to absolute attributes such as age or size; instead, it demands the mon-

itoring of other individuals’ interactions. Several observations and experiments

suggest that monkeys do recognize the rank relations that exist among other

females in their group. For example, dominant female baboons often grunt to

mothers with infants as they approach the mothers and attempt to handle or

touch their infants. Grunts seem to function to facilitate social interactions by ap-

peasing anxious mothers, because an approach accompanied by a grunt is signifi-

cantly more likely to lead to subsequent friendly interaction than is an approach

without a grunt (Cheney et al., 1995a). Occasionally, however, a mother will utter a

submissive call, or ‘fear bark’, as a dominant female approaches. Fear barks are an

unambiguous indicator of subordination; they are never given to lower-ranking

females.

To test whether baboons recognize that only a more dominant animal can

cause another individual to give a fear bark, we designed a playback experiment

in which adult female subjects were played a causally inconsistent call sequence

in which a low-ranking female apparently grunted to a high-ranking female and

the higher-ranking female apparently responded with fear barks. As a control, the

same subjects heard the same sequence of grunts and fear barks made causally

consistent by the inclusion of additional grunts from a third female who was dom-

inant to both of the other signallers. For example, if the inconsistent sequence was

composed of female 6’sgrunts followed by female 2’sfear barks, the corresponding

consistent sequence might begin with female 1’s grunts, followed by female 6’s

grunts and ending with female 2’s fear barks. Some subjects were higher-ranking

than the signallers; others were lower ranking. Regardless of their own relative

ranks, subjects responded significantly more strongly to the causally inconsistent

sequences, suggesting that they recognize not only the identities of different sig-

nallers but also the rank relations that exist among others in their group (Cheney

et al., 1995b).

Further suggestion that monkeys recognize other individuals’ ranks comes

from observations on competition among adult female vervet monkeys for access

to a grooming partner (Seyfarth, 1980). Such competition occurs when one fe-

male approaches two that are grooming, supplants one of them and then grooms

with the female that remains. Interestingly, in those cases when a female ap-

proaches two groomers who are both subordinate to her, the lower ranking

of the two groomers typically moves away, while the higher ranking remains

(Cheney & Seyfarth, 1990). By remaining seated, the higher ranking of the two

groomers acts as if she recognizes that, although they are both lower ranking

than the approaching female, she is the higher ranking. Though not definitive,

these observations suggest that females recognize not only their own status rela-

tive to other individuals but also other individuals’ status relative to each other.



In other words, they appear to recognize a rank hierarchy (Cheney & Seyfarth,

1990).

The ability to rank other group members is perhaps not surprising, given the

evidence that captive monkeys and apes can be taught to rank objects according to

an arbitrary sequential order (D’Amato & Colombo, 1989; Treichler & van Tilberg,

1996), the amount of food contained within a container (Gillan, 1981), their size

or the number of objects contained within an array (e.g. Matsuzawa, 1985; Hauser

et al., 1996; Brannon & Terrace, 1998). What distinguishes the social example, how-

ever, is the fact that, even in the absence of human training, female monkeys seem

able to construct a rank hierarchy and then place themselves at the appropriate

location within it.

Knowledge about more transient social relationships

All of the studies discussed so far focus on interactions among females

in groups where matrilineal kin usually retain close bonds and similar ranks

throughout their lives. It might seem, therefore, that an individual could simply

memorize the close associates and relative ranks of other females and thereafter

navigate easily through a predictable network of social relationships. Not all social

and rank relationships, however, are as stable as those among matrilineal kin.

Some types of social bond are relatively transient, and some rank relationships –

particularly among adult males – change often. Nonetheless, there is evidence

that non-human primates also recognize these more transient associations.

For example, under natural conditions, male and female hamadryas baboons

Papio hamadryas form close, long-term bonds that can last for a number of years.

Potential rivals appear to recognize the ‘ownership’ of specific females by other

males and refrain from challenging those males for their females (Kummer et al.,

1974). Experiments conducted in captivity have shown that rival males assess the

strength of other males’ relationships with their females before attempting to

challenge them. They do not attempt to take over a male’s female if the pair ap-

pears to have a close social bond (Bachmann & Kummer, 1980). Although similar

experiments have not yet been conducted with savannah baboons, observational

data suggest that these baboons, too, recognize the temporary bonds, or ‘friend-

ships’, that are formed between males and lactating females (Palombit et al., 1997).

For example, Smuts (1985) observed that males who had recently been threatened

by another male often redirected aggression toward the female friends of their op-

ponent (see Dunbar (1983) for similar observations on gelada baboons Theropithecus

gelada).

Monkeys also seem to recognize the bonds that exist between males and particu-

lar infants. In Tibetan macaques Macaca thibetana, males are often closely affiliated



with a particular infant in the group. Competitive interactions between males are

mediated by the carrying of infants and a male will frequently carry an infant and

present it to another male. In a study of such carrying (or ‘bridging’) behaviour,

Ogawa (1995) observed that males more frequently provided other males with

those males’ affiliated infants than with other, non-affiliated infants.

Finally, there is evidence that monkeys recognize even very transient dom-

inance relations among others. Dominance among male vervets, baboons and

macaques is determined primarily by age, fighting ability, and, in some popula-

tions, the presence of alliance partners. As a result, rank relations among males are

considerably less stable than they are among females (Walters & Seyfarth, 1987).

In a study of a large social group of captive bonnet macaques Macaca radiata, Silk

(1993, 1999) found that males formed linear, transitive dominance hierarchies that

remained stable for only short periods of time. As in other primate species, males

occasionally attempted to recruit alliance support during aggressive interactions

(approximately 12% of all aggressive encounters). Significantly, males consistently

solicited allies that outranked both themselves and their opponents. Males did not

simply solicit the highest-ranking individual in the group or choose allies that out-

ranked only themselves. Instead, soliciting males seemed to recognize not only

their own rank relative to a potential ally but also the rank relation between the

ally and their opponent. If dominance ranks remained stable, this might not have

been a difficult task. However, over the course of one year, approximately half

of the 16 males changed dominance rank each month (data from Table 3 in Silk,

1993). The males’ apparent ability to keep track of such highly transient rank re-

lations suggests that they carefully monitored all aggressive interactions among

other males, constantly updated their list of relative ranks and placed themselves

accurately into each new list.

Eavesdropping by other mammals

Data from dolphins Tursiops truncatus and hyaenas Crocuta crocuta suggest

that non-human primates are not the only mammals in which individuals acquire

information about many different individuals’social relationships (for other mam-

mals see Chs. 17 and 18). When competing over access to females, male dolphins

form dyadic and triadic alliances with selected other males, and allies with the

greatest degree of partner fidelity are most successful in acquiring access to fe-

males (Connor et al., 1992, 1999, 2001). The greater success of high-fidelity alliances

raises the possibility that males in newly formed alliances, or in alliances that have

been less stable in the past, recognize the strong bonds that exist among others

and are more likely to retreat when they encounter rivals with a long history of

cooperative interaction.



Like many species of Old World monkeys, hyaenas live in social groups compris-

ing matrilines in which offspring inherit their mothers’ dominance ranks (Smale

et al., 1993; Engh et al., 2000). Holekamp et al. (1999) played recordings of cubs’

‘whoop’ calls to mothers and other breeding females. As with vervet monkeys

and baboons, hyaena females responded more strongly to the calls of their own

offspring and those of close relatives than to the calls of unrelated cubs. In con-

trast to vervets and baboons, however, unrelated females did not look at the cubs’

mothers. One explanation for these negative results is that hyaenas are unable to

recognize third-party relationships, despite living in social groups that are super-

ficially similar to those of many primates. It also remains possible, however, that

hyaenas are simply uninterested in the calls of unrelated cubs.

In fact, hyaenas’ patterns of alliance formation suggest that they do monitor

other individuals’ interactions and extrapolate information about other animals’

relative ranks from their observations. During competitive interactions over meat,

hyaenas often solicit alliance support from other, uninvolved individuals. When

choosing to join ongoing skirmishes, hyaenas that are dominant to both of the

contestants almost always support the more dominant of the two individuals

(Engh et al., 2004). Similarly, when the ally is intermediate in rank between the

two opponents, it inevitably supports the dominant individual. These data provide

the first evidence in a non-primate species that alliance partners may be chosen on

the basis of both the allies’and the opponents’relative ranks (Harcourt, 1988, 1992).

They are consistent with the hypothesis that hyaenas are able to infer transitive

rank relations among other group members.

Possible differences between primates and other animals

Do primates differ from other animals in their ability to infer third-party

social relationships through eavesdropping? We can identify at least three com-

peting hypotheses.

The first hypothesis argues that primates are in fact more intelligent than non-

primates. This intelligence is reflected not only in tests of captive animals but also

in primates’ superior ability to keep track of complex social relationships. The

difference between primates and non-primates is qualitative and fundamental

and will be corroborated by future research.

The second hypothesis maintains that selection has favoured the ability to rec-

ognize other individuals’ relationships in all species that live in large, complex

social groups. According to this hypothesis, monkeys only appear to have a greater

capacity to recognize third-party social relationships because they have received

more attention than non-primates living in similarly large groups. Once this im-

balance in research has been redressed, differences between primates and other



animals will disappear, to be replaced by a difference that depends primarily on

group size and composition.

The third hypothesis claims that neither phylogeny nor group size and composi-

tion have influenced animals’ ability to gain information about other individuals’

social relationships. It argues, in effect, that there are no species differences in

‘social intelligence’. Monkeys and hyaenas, for example, only appear to excel in

their ability to recognize the relative ranks of allies and opponents because their

large social groups allow them to display this knowledge. In contrast, studies of

species that live in small social groups have to date focused primarily on observers’

ability to assess the dominance of only two individuals. Once monogamous and

even solitary species have been given the opportunity to reveal what they know

about the social relationships of many different individuals, they will be shown

to possess a level of social intelligence that is no different from that found among

animals living in large social groups.

At present, it is difficult to test these alternative hypotheses; below we review

some information that may be relevant.

Hypothesis 1: primates have greater social intelligence than other species

Primates have larger brains for their body size than other vertebrates

(Martin, 1983). Dunbar (2000) argued that this arises because primate social groups

are not only larger but also more complex than those of other taxa. Primate groups

are typically composed of many reproductively active males and females, and

individuals interact regularly with both kin and non-kin, with whom they must

simultaneously cooperate and compete for resources. Such social complexity may

place strong selective pressure on the ability to recognize close associates of other

individuals.

To date, only monkeys and possibly dolphins have been shown to recognize

the affiliative relationships that exist among other group members. In monkey

groups, closely bonded individuals are usually matrilineal kin, but this is not

always the case. The ability to classify other individuals into matrilineal or closely

bonded subgroups is likely to be relatively complex, for several reasons.

Matrilineal kin groups vary in size and not all individuals within a kin group

interact at the same rate or in the same way. Moreover, no single behavioural mea-

sure underlies the associations between individuals and there is no threshold or

defining criterion for a ‘close’ social bond. For example, females in many monkey

species form the majority of their alliances with matrilineal kin, and high-ranking

kin usually form alliances at higher rates than low-ranking kin (reviewed by Silk,

1987; Walters & Seyfarth, 1987). There is no evidence, however, that other group

members more easily recognize the kin (or close associates) of high-ranking in-

dividuals than the kin of low-ranking individuals. Similarly, female kin usually



occupy adjacent dominance ranks. This rule of thumb, however, cannot reliably be

used to classify females into kin groups, because not all adjacently ranked females

are kin. We do not yet know whether monkeys discriminate among different types

of social bond: whether they distinguish, for example, among the bonds formed

by mothers and offspring, sisters, or friends. Moreover, the degree to which there

is a quantitative or qualitative threshold for learning to recognize that two other

individuals share a close bond is not known.

Furthermore, some social relationships among monkeys are transitive, while

others are not. For example, if infant A1 and juvenile A2 both associate at high

rates with a particular adult female A, it is usually correct to infer that the ju-

venile and infant are also closely bonded. Similarly, if A is dominant to B and B

is dominant to C, it is usually true that A is dominant to C. In other cases, how-

ever, transitivity cannot be assumed. If infant baboon A1 and juvenile baboon A2

both associate at high rates with the same adult female and she associates with

an adult male ‘friend’, we can infer that the male is probably also closely allied

to the infant. However, it would incorrect to assume that he is equally closely

allied to the juvenile, who may instead be more closely allied to another male

who was previously the mother’s friend (Seyfarth, 1978; Smuts, 1985; Palombit

et al., 1997). Baboon females from the same matriline often form friendships with

different males; conversely, the same male may form simultaneous friendships

with females from two different matrilines. In the latter case, the existence of a

close bond between a male and two females does not predict a close bond between

the two females. In fact, their relationship is likely to be as competitive as it is

friendly (Palombit et al., 2001).

Finally, as group size increases, the challenge of monitoring other individuals’

social relationships and dominance ranks increases exponentially. In a group of

80 animals (not an unusual size for many monkey species), each individual con-

fronts 3160 different possible dyadic combinations and 82 160 different triadic

combinations of individuals: numbers that may place considerable demands on

the observer’s memory and inferential abilities.

Preliminary evidence suggests that monkeys are able to monitor and remember

the social ranks and relationships of many individuals simultaneously. Despite

the lack of a consistent criterion for determining which individual is likely to be

closely bonded with which others, monkeys appear to be able to distinguish the

close associates of other group members. They appear to view their social groups

not just in terms of the individuals that constitute them but also in terms of a

web of social relationships in which certain individuals are linked with several

others.

Some learning experiments with captive animals support the view that pri-

mates are generally more adept than non-primates at classifying items according



to their relative relations. In oddity tests, for example, a subject is presented with

three objects, two of which are the same and one of which is different, and asked

to choose the object that is different. Monkeys and apes achieve high levels of ac-

curacy in such tests even when tested with novel stimuli (Harlow, 1949; D’Amato

et al., 1985; see also reviews by Tomasello & Call, 1997; Shettleworth, 1998). Baboons

and chimpanzees can also learn to make abstract discriminations about relations

between relations, matching patterns containing repeated samples of the same

item with similar ‘same’ patterns (Premack, 1983; Oden et al., 1988; Fagot et al.,

2001). In all cases, subjects’ performances suggest the use of an abstract hypoth-

esis, because concepts like ‘odd’ specify a relation between objects independent

of their physical features. In a similar manner, the concept ‘closely bonded’ can

be applied to any two individuals and need not be restricted to specific pairs that

look alike.

Judgements based on relations among items have been demonstrated more

often in non-human primates than in other taxa, and primates seem to recognize

abstract relations more readily than at least some other animals. Although it is

possible, for example, to train pigeons to recognize relations such as ‘same’, the

procedural details of the test appear more critical for pigeons than they are for

monkeys, and relational distinctions can easily be disrupted (Herrnstein, 1985;

Wright et al., 1988; Wasserman et al., 1995). Rather than attending to the relations

among stimuli, pigeons seem predisposed to focus on absolute stimulus properties

and to form item-specific associations (reviewed by Shettleworth, 1998). Similarly,

in tests of transitive inference, monkeys and apes appear to acquire a representa-

tion of series order that allows them to rank items even when some items in the

list are missing. In contrast, pigeons seem to attend primarily to the association

between adjacent pairs, which limits their ability to add or delete items from a list

(D’Amato & Colombo, 1989; von Fersen et al., 1991; Treichler & van Tilberg, 1996;

Zentall et al., 1996).

Hypothesis 2: differences in ‘social intelligence’ are related to group sizeand complexity

If, as has been hypothesized, the recognition of third-party relationships

confers a selective advantage because it allows individuals to remember who as-

sociates with whom, who outranks whom and who is allied to whom, we should

expect to find evidence for this ability not just in non-human primates but also

in any animal species that lives in large social groups composed of individuals of

varying degrees of dominance rank and genetic relatedness. We would also pre-

dict that selection should have acted less strongly on this ability in solitary species

and species living in small, egalitarian groups that are composed primarily either

of close kin or of unrelated individuals. Thus, the ability to recognize the close



associates of others should be evident in non-primate species such as hyaenas

and lacking or less evident in some ape species, including gorillas Gorilla gorilla

and orangutans Pongo pygmaeus. Although recent evidence that hyaenas recognize

other individuals’ relative ranks lends support to this hypothesis, other compara-

tive data are lacking. For example, no study has yet attempted to determine the

extent to which any ape species is able to recognize the social relationships of

other group members.

Within the Primate order, species that live in large groups have a relatively

larger neocortex than those that are solitary or live in small groups (Barton &

Dunbar, 1997). A similar relation is found in carnivores (Barton & Dunbar, 1997)

and toothed whales (Connor et al., 1998a,b; Marino, 1998), supporting the hypoth-

esis that sociality has favoured the evolution of large brains (see also Jolly, 1966;

Humphrey, 1976; Cheney & Seyfarth, 1990). Indeed, differences in social complex-

ity may exert their effect even in species that lack a cortex entirely. In paper wasps

Polistes dominulus, for example, there is a significant increase in the size of the

antennal lobes and collar (a substructure of the calyx of the mushroom body) in

females that nest colonially – with other queens – as opposed to solitary breeders

(Ehmer et al., 2001). This increase in neural volume may be favoured because so-

ciality places increased demand on the need to discriminate between familiar and

unfamiliar individuals and to monitor other females’ dominance and breeding

status. Clearly, therefore, neural correlates of sociality need not be restricted to

higher mammals.

Further supporting this argument are data from some other laboratory studies

suggesting fewer differences between primates and other animals in the ability

to make relational distinctions. For example, Alex, an African grey parrot Psitta-

cus erithacus, is reported to make explicit same/different judgements about sets

of objects (Pepperberg, 1992, Ch. 24). Similarly, sea lions Zalophus californianus

(Schusterman & Krieger, 1986; Schusterman & Gisiner 1988) and dolphins (Herman

et al. 1993; Mercado et al. 2000) have been taught to respond to terms such as ‘left’

and ‘bright’, which require the animals to assess relations among a variety of dif-

ferent objects. Finally, a number of species, including parrots (Pepperberg, 1994)

and rats (Church & Meck, 1984; Capaldi, 1993), are able to assess quantities, sug-

gesting that relatively abstract concepts of numerosity and transitivity may be

pervasive among animals (reviewed by Shettleworth, 1998).

Hypothesis 3: there are few differences in ‘social intelligence’ across species

Recent research on social eavesdropping (Ch. 2) by birds and fish indi-

cates that even animals living in small social groups are capable of acquiring

detailed information about other individuals’ relative dominance or attractive-

ness as a mate. Often, this information is of necessity restricted to a few other



individuals. For territorial species living in small family groups, questions about

the ability to track social relationships among many other individuals are largely

moot, because the opportunity to monitor interactions among all possible neigh-

bours rarely arises. Eavesdropping on the competitive singing duets of strangers,

for example, allows territorial songbirds to extract information about the two

contestants’ relative dominance. Whether these birds would also be capable of

recognizing a dominance hierarchy involving numerous individuals remains un-

clear. Although many species of songbirds form flocks during the winter, little

is known about the social interactions that take place within such flocks, or the

degree to which flock members recognize other individuals’ relative ranks (but

see Popp, 1987).

Recently, Bond et al. (2003) tested the prediction that socially living birds will

display enhanced abilities to make transitive inferences by comparing the per-

formance of highly social pinyon jays Gymnorhinus cyanocephalus with relatively

non-social western scrub jays Aphelocoma californica. Using operant procedures,

subjects were required to order a set of arbitrary stimuli by inference from a se-

ries of dyadic comparisons. Subjects of both species learned the sequence order,

but pinyon jays did so more rapidly and more accurately than scrub jays. Although

not conclusive, these results lend support to the hypothesis that social complexity

may be correlated with superior performance in tasks involving the ranking of

multiple stimuli (see also Hogue et al. (1996) for experiments with flock-dwelling

domestic chickens Gallus domesticus).

As yet, very little is known about the ability of non-primate mammals or birds

to recognize social relationships of other individuals. Colonial white-fronted bee-

eaters Merops bullockoides offer one example of an avian society in which there

would appear to be strong selective pressure for the recognition of the kin groups

of other individuals. Observational evidence suggests that bee-eaters may recog-

nize other individuals and kin groups and associate these groups with specific

feeding territories (Emlen et al., 1995), although this has not yet been tested

experimentally.

Clearly, more data are needed from both natural and laboratory studies before

we can draw any definitive conclusions about cognitive differences between pri-

mates and other animals, or between species living in large as opposed to small

groups. It remains entirely possible that apparent species differences between

primates and other animals in the recognition of third-party social relationships

result more from differences in the social context in which eavesdropping occurs

than from any cognitive differences in the ability to monitor social interactions.

Given the opportunity to evaluate the social relationships of many different indi-

viduals, species living in small family groups and even primarily solitary species

may well be shown to have similar abilities to those living in large social groups. It



is to be hoped that future research will attempt to investigate the extent to which

gregarious species in taxa other than primates are capable of recognizing the

close associates and allies of other group members, and to determine the neural

correlates, if any, of this ability.

Summary

Non-human primates are skilled voyeurs. By observing or listening to the

interactions of others, they acquire information about the social relationships

of other individuals and learn to place these relationships within a larger social

framework, such as a group of ranked, matrilineal families. Given the large, com-

plex societies in which monkeys cooperate and compete, the adaptive value of

such eavesdropping seems clear. At present, however, we do not know whether

the information acquired by eavesdropping in primates differs significantly from

the information acquired by individuals in other species. Primates (and a few

other mammals) may be qualitatively different from other species in their ability

to monitor the social relationships of many other individuals. Alternatively, the

societies of birds, fish and other non-primate species – often superficially simpler

than those of primates – may have led us to underestimate the information that

individuals acquire about others. Finally, both hypotheses may have some valid-

ity. There may be qualitative differences in social intelligence between different

taxonomic groups, but within each group the information acquired from eaves-

dropping may increase in sophistication with increasing social complexity. The

chapters in this volume demonstrate that eavesdropping is widespread among

animals. They set the stage for comparative research that examines differences

between species in the information acquired about others.

References

Aureli, F., Cozzolino, R., Cordischi, C. & Scucchi, S. 1992. Kin-oriented redirection

among Japanese macaques: an expression of a revenge system? Animal Behaviour,

44, 283–291.

Bachmann, C. & Kummer, H. 1980. Male assessment of female choice in hamadryas

baboons. Behavioral Ecology and Sociobiology, 6, 315–321.

Barton, R. A. & Dunbar, R. 1997. Evolution of the social brain. In: Machiavellian

Intelligence II: Extensions and Evaluations, ed. A. Whiten & R. W. Byrne. Cambridge,

UK: Cambridge University Press, pp. 240–263.

Bond, A. B., Kamil, A. C. & Balda, R. P. 2003. Social complexity and transitive inference

in corvids. Animal Behaviour, 65, 479–487.





Brannon, E. & Terrace, H. 1998. Ordering of the numerosities 1 to 9 by monkeys.

Science, 282, 746–749.

Brooks, R. J. & Falls, J. B. 1975. Individual recognition by song in white-throated

sparrows. III. Song features used in individual recognition. Canadian Journal of

Zoology, 53, 1749–1761.

Capaldi, E. J. 1993. Animal number abilities: implications for a hierarchical approach

to instrumental learning. In: The Development of Numerical Competence, ed. S. T.

Boysen & E. J. Capaldi. Hillsdale, NJ: Lawrence Erlbaum, pp. 191–209.

Chapais, B. 1988. Rank maintenance in female Japanese macaques: experimental

evidence for social dependency. Behaviour, 104, 41–59.

Cheney, D. L. & Seyfarth, R. M. 1980. Vocal recognition in free-ranging vervet monkeys.


1982. Recognition of individuals within and between groups of free-ranging vervet

monkeys. American Zoologist, 22, 519–529.

1986. The recognition of social alliances among vervet monkeys. Animal Behaviour,

34, 1722–1731.

1989. Reconciliation and redirected aggression in vervet monkeys. Behaviour, 110,

258–275.

1990. How Monkeys See the World: Inside the Mind of Another Species. Chicago, IL:

University of Chicago Press.

1999. Recognition of other individuals’ social relationships by female baboons.


Cheney, D. L., Seyfarth, R. M. & Silk, J. B. 1995a. The role of grunts in reconciling

opponents and facilitating interactions among adult female baboons. Animal

Behaviour, 50, 249–257.

1995b. The responses of female baboons (Papio cynocephalus ursinus) to anomalous

social interactions: evidence for causal reasoning? Journal of Comparative


Church, R. M. & Meck, W. H. 1984. The numerical attribute of stimuli. In: Animal

Cognition, ed. H. L. Roitblat, T. G. Bever & H. S. Terrace. Hillsdale, NJ: Lawrence

Erlbaum, pp. 445–464.

Connor, R. C., Smolker, R. A. & Richards, A. F. 1992. Dolphin alliances and coalitions.

In: Coalitions and Alliances in Humans and other Animals, ed. A. H. Harcourt & F. de

Waal. Oxford: Oxford University Press, pp. 415–443.

Connor, R. C., Mann, J., Tyack, P. L. & Whitehead, H. 1998a. Social evolution in toothed

whales. Trends in Ecology and Evolution, 13, 228–232.

1998b. Reply. Trends in Ecology and Evolution, 13, 408.

Connor, R. C., Heithaus, R. M. & Barre, L. M. 1999. Superalliance of bottlenose

dolphins. Nature, 371, 571–572.

2001. Complex social structure, alliance stability, and mating access in a bottlenose

dolphin ‘super-alliance’. Proceedings of the Royal Society of London, Series B, 268,

263–267.

D’Amato, M. & Colombo, M. 1989. Serial learning with wild card items by monkeys

(Cebus apella): implications for knowledge of ordinal rank. Journal of Comparative




D’Amato, M., Salmon, P. & Colombo, M. 1985. Extent and limits of the matching

concept in monkeys (Cebus apella). Journal of Experimental Psychology: Animal Behavior

Processes, 11, 35–51.

Dasser, V. 1988a. A social concept in Java monkeys. Animal Behaviour, 36, 225–230.

1988b. Mapping social concepts in monkeys. In: Machiavellian Intelligence: Social

Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans, ed. R. W. Byrne

and A. Whiten. Oxford: Oxford University Press, pp. 85–93.

Dunbar, R. 1983. Structure of gelada baboon reproductive units. III. The male’s

relationship with his females. Animal Behaviour, 31, 565–575.

2000. Causal reasoning, mental rehearsal, and the evolution of primate cognition.

In: Vienna Series in Theoretical Biology: The Evolution of Cognition, ed. C. Heyes & L.

Huber. Cambridge, MA: MIT Press, pp. 205–219.

Ehmer, B., Reeve, H. K. & Hoy, R. R. 2001. Comparison of brain volumes between single

and multiple foundresses in the paper wasp Polistes dominulus. Brain, Behavior and

Evolution, 57, 161–168.

Emlen, S. T., Wrege, P. H. & Demong, N. J. 1995. Making decisions in the family: an

evolutionary perspective. American Scientist, 83, 148–157.

Engh, A. L., Esch, K., Smale, L. & Holekamp, K. E. 2000. Mechanisms of maternal rank

‘inheritance’ in the spotted hyaena, Crocuta crocuta. Animal Behaviour, 60, 323–332.

Engh, A. L., Siebert, E., Greenberg, D. & Holekamp, K. 2004. Primate-like cognition in a

social carnivore. Animal Behaviour, in press.

Fagot, J., Wasserman, E. A. & Young, M. 2001. Discriminating the relation between

relations: the role of entropy in abstract conceptualization by baboons (Papio

papio) and humans (Homo sapiens). Journal of Experimental Psychology: Animal Behavior

Processes, 27, 316–328.

Gillan, D. J. 1981. Reasoning in the chimpanzee, II. Transitive inference. Journal of

Experimental Psychology: Animal Behavior Processes, 7, 150–164.

Gouzoules, S., Gouzoules, H. & Marler, P. 1984. Rhesus monkey (Macaca mulatta)

screams: representational signalling in the recruitment of agonistic aid. Animal

Behaviour, 32, 182–193.

Harlow, H. F. 1949. The formation of learning sets. Psychological Review, 56, 51–65.

Hansen, E. W. 1976. Selective responding by recently separated juvenile rhesus

monkeys to the calls of their mothers. Developmental Psychobiology, 9, 83–88.

Harcourt, A. H. 1988. Alliances in contests and social intelligence. In: Machiavellian

Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans,

ed. R. W. Byrne & A. Whiten. Oxford: Oxford University Press, pp. 132–152.

1992. Coalitions and alliances: are primates more complex than non-primates? In:

Coalitions and Alliances in Humans and Other Animals, ed. A. H. Harcourt & F. de Waal.

New York: Oxford University Press, pp. 86–99.

Hauser, M. D., MacNeilage, P. & Ware, M. 1996. Numerical representations in primates.


Herman, L. M, Pack, A. A & Morrel-Samuels, P. 1993. Representational and conceptual

skills of dolphins. In: Comparative Cognition and Neuroscience, ed. H. L. Roitblat, L. M.

Herman & P. E. Nachtigall. Hillsdale, NJ: Lawrence Erlbaum, pp. 403–442.



Herrnstein, R. J. 1985. Riddles of natural categorization. Philosophical Transactions of the


Hogue, M. E., Beaugrand, J. P. & Laugue, P. C. 1996. Coherent use of information by

hens observing their former dominant defeating or being defeated by a stranger.


Holekamp, K. E., Boydston, E. E., Szykman, M. et al. 1999. Vocal recognition in the

spotted hyaena and its possible implications regarding the evolution of

intelligence. Animal Behaviour, 58, 383–395.

Humphrey, N. 1976. The social function of the intellect. In Machiavellian Intelligence:

Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans, ed. R. W.

Byrne and A. Whiten. Oxford: Oxford University Press, pp. 13–26.

Jolly, A. 1966. Lemur Behavior. A Madagascar Field Study. Chicago, IL: University of

Chicago Press.

Judge, P. 1982. Redirection of aggression based on kinship in a captive group of pigtail

macaques. International Journal of Primatology, 3, 301.

Kummer, H., Goetz, W. & Angst, W. 1974. Triadic differentiation: an inhibitory process

protecting pair bonds in baboons. Behaviour, 49, 62–87.

Marino, L. 1998. Quantifying brain–behavior relations in cetaceans and primates.

Trends in Ecology and Evolution, 13, 408.

Martin, R. D. 1983. Primate Origins and Evolution: A Phylogenetic Reconstruction. Princeton:

Princeton University Press.

Matsuzawa, T. 1985. Use of numbers by a chimpanzee. Nature, 315, 57–59.


Evolution of Acoustic Communication in Birds, eds. D. E. Kroodsma & E. H. Miller.




Mercado, E., Killebrew, D. A., Pack, A. A., Macha, I. & Herman, L. M. 2000.

Generalization of ‘same-different’ classification abilities in bottlenosed dolphins.




Series B, 266, 537–542.

Oden, D. Thompson, R. & Premack, D. 1988. Spontaneous transfer of matching by

infant chimpanzees (Pan troglodytes). Journal of Experimental Psychology: Animal

Behavior Processes, 14, 140–145.

Ogawa, H. 1995. Recognition of social relationships in bridging behavior among

Tibetan macaques (Macaca thibetana). American Journal of Primatology, 35, 305–310.




Palombit, R. A., Seyfarth, R. M. & Cheney, D. L. 1997. The adaptive value of ‘friendships’

to female baboons: experimental and observational evidence. Animal Behaviour,

54, 599–614.



Palombit, R. A., Cheney, D. L. & Seyfarth, R. M. 2001. Female–female competition for

male ‘friends’ in wild chacma baboons (Papio cynocephalus ursinus). Animal

Behaviour, 61, 1159–1171.


eavesdrop on simulated male–male vocal interactions. Proceedings of the Royal


Pepperberg, I. M. 1992. Proficient performance of a conjunctive, recursive task by an

African grey parrot (Psittacus erithacus). Journal of Comparative Psychology, 106,

295–305.

1994. Numerical competence in an African gray parrot (Psittacus erithacus). Journal of


Popp, J. W. 1987. Choice of opponents during competition for food among American

goldfinches. Ethology, 75, 31–36.

Premack, D. 1983. The codes of man and beast. Behavioral and Brain Sciences, 6, 125–167.

Schusterman, R. J. & Gisiner, R. 1988. Artificial language comprehension in dolphins

and sea lions: the essential cognitive skills. Psychological Record, 38, 311–348.

Schusterman, R. J. & Krieger, K. 1986. Artificial language comprehension and size

transposition by a California sea lion (Zalophus californianus). Journal of Comparative


Seyfarth, R. M. 1978. Social relationships among adult male and female baboons. II.

Behavior throughout the female reproductive cycle. Behaviour, 64, 227–247.

Seyfarth, R. M. 1980. The distribution of grooming and related behaviors among adult

female vervet monkeys. Animal Behaviour, 28, 798–813.

Shettleworth, S. 1998. Cognition, Evolution, and Behaviour. Oxford: Oxford University

Press.

Silk, J. B. 1987. Social behavior in evolutionary perspective. In: Primate Societies, ed. B. B.

Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham & T. Struhsaker. Chicago, IL:

University of Chicago Press, pp. 318–329.

1993. Does participation in coalitions influence dominance relationships among

male bonnet macaques? Behaviour, 126, 171–189.

1999. Male bonnet macaques use information about third-party rank relationships

to recruit allies. Animal Behaviour, 58, 45–51.

Silk, J. B., Seyfarth, R. M. & Cheney, D. L. 1999. The structure of social relationships

among female savanna baboons. Behaviour, 136, 679–703.

Smale, L., Frank, L. G. & Holekamp, K. E. 1993. Ontogeny of dominance in free-living

spotted hyaenas: juvenile rank relations with adult females and immigrant

males. Animal Behaviour, 46, 467–477.

Smuts, B. 1985. Sex and Friendship in Baboons. Chicago, IL: Aldine.

Tomasello, M. & Call, J. 1997. Primate Cognition. Oxford: Oxford University Press.

Treichler, F. & van Tilburg, D. 1996. Concurrent conditional discrimination tests of

transitive inference by macaque monkeys: list linking. Journal of Experimental

Psychology, Animal Behavior Processes, 22, 105–117.

von Fersen, L., Wynne, C., Delius, J. & Staddon, J. 1991. Transitive inference formation

in pigeons. Journal of Experimental Psychology: Animal Behavior Processes, 17, 334–341.



Walters, J. R. & Seyfarth, R. M. 1987. Conflict and cooperation. In: Primate Societies, ed.

B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham & T. Struhsaker.

Chicago, IL: University of Chicago Press, pp. 306–317.

Wasserman, E. A., Hugart, J. A. & Kirkpatrick-Steger, K. 1995. Pigeons show

same-different conceptualization after training with complex visual stimuli.

Journal of Experimental Psychology: Animal Behavior Processes, 21, 248–252.

Wright, A., Cook, R. & Rivera, J. 1988. Concept learning by pigeons: matching to sample

with trial-unique video picture stimuli. Animal Learning and Behavior, 16, 436–444.

Zentall, T., Weaver, J. & Sherburne, L. 1996. Value transfer in concurrent-schedule

discriminations by pigeons. Animal Learning and Behavior, 24, 401–409.


26

Communication networks ina virtual world

a n d r e w m . r . t e r r y 1 & ro b e r t l ac h l a n 2

1University of Copenhagen, Denmark2University of North Carolina, Chapel Hill, USA

Introduction

When one individual is signalling, or two individuals are interacting, they

do so within a network of potential receivers (see McGregor, 1993; McGregor &

Dabelsteen, 1996; Ch. 1). As the other chapters in this book show, the decisions

that both signallers and receivers make about their future behaviour are thus

contingent not only on each other’s behaviour but also on a wider network of

individuals (McGregor & Peake, 2000). This view is finding support in empirical

studies showing that individuals use information that could only be extracted

from network interactions (e.g. Oliveira et al., 1998; Peake et al., 2001, 2002; Ch. 2).

These empirical findings also have implications for the theoretical study of sig-

nalling strategies (e.g. Nowak & Sigmund, 1998; Johnstone, 2001). For example,

an individual’s signalling strategy may no longer be predicted solely from the re-

sponses of an opponent. We consider that the signalling strategies of individuals

will only be explored realistically by models that include the potential responses

of signallers to other individuals. In this chapter, we ask whether current mod-

elling approaches can be adapted to include networks or whether new modelling

techniques need to be considered.

The aim of creating a model is to advance our conceptual understanding of a

system and create empirically testable hypotheses (Wilson, 2000; Hemelrijk, 2002)

by simplifying the real world using words or mathematical expressions. Most hy-

potheses start with a verbal model and develop into mathematical models, which

more precisely specify limiting conditions and assumptions and often provide

a deeper understanding of the logic underlying the hypothesis. Models vary in


604


Communication networks in a virtual world 605

their complexity and the level of explanatory power they provide: often, simple

models provide a deeper and general understanding of underlying dynamics but

also contain more restrictive (or unrealistic) assumptions. For example, simple

biological models, such as the Lotka–Volterra equations of predator–prey interac-

tions, provided a powerful heuristic insight into the cyclic nature of population

abundances but contained highly restrictive assumptions, primarily concerning

the heterogeneity of populations and environments, that limited their applica-

tion to specific cases (Maynard Smith, 1982; Begon et al., 1990; Wilson, 2000). In

population genetics models, assumptions are typically made about population

sizes and the pattern of distribution of traits. Nevertheless, such assumptions of-

ten do not qualitatively affect the conclusion (e.g. Turelli & Barton, 1994). Simpler

models are likely to be mathematically tractable, allowing a more complete analy-

sis and understanding of the processes underlying the system being investigated.

Complexity increases as more realistic assumptions are incorporated and the pos-

sibility of mathematical analysis becomes more remote. However, in recent years,

the rapid rise in computer power has allowed theoretical methods to acquire new

levels of complexity, mostly through the use of simulation modelling (Grimm,

1999). Simulation models allow as many variables to be included as the investiga-

tors have imagination, programming skills and time. The downside to this is that

the models are harder to generalize (it is harder to ensure that all the variables

are realistic for a wide range of conditions) and that it is more difficult to iso-

late precisely the factors that are causing an effect of interest (Wilson, 2000). An

attractive solution to this dilemma is to use a variety of modelling techniques, dif-

fering in how many assumptions are required (Dieckman, 1997), with the specific

aim of identifying the parameters, variables and assumptions that are critical in

explaining the behaviour of the system under investigation (Wilson, 2000).

In this chapter, we discuss the role modelling has played in the conceptual

development of communication networks. In doing this, we also examine which

features of communication network models are especially important, often by

identifying which unrealistic assumptions are likely to change qualitatively the

conclusions reached about behaviour in networks. Finally, we examine how theo-

ries of communication that include networks are likely to differ from those that

do not.

Conceptualizing networks

Network structure in existing models

A network is an association of nodes connected to each other by some

means. In animal communication, the nodes are individual animals and the con-

nections (or links) are patterns of communication between them. For every signal,


606 A. M. R. Terry & R. Lachlan

there are likely to be several possible receivers, leading to the concept of the

communication network. This concept modifies our understanding of the costs

and benefits of signalling (e.g. McGregor & Peake, 2000). For example, individuals

winning an interaction may be more willing to publicize the interaction than

those losing it. How does this view of communication fit with standard methods

used to model animal signalling and do different methods need to be developed?

Most current models of animal communication consider only dyadic encoun-

ters between individuals. For example, game theory models of communication

normally study the evolution of strategies through the responses of two interac-

tants to each other (examples in Maynard Smith, 1982). When analytical models

have been used to study interactions within large groups, they typically make

restrictive assumptions that may limit their ability to assess the evolutionary

pressures of communicating within networks. For example, individuals may be

drawn at random from the population to interact or they may have perfect knowl-

edge about the behaviour of all other individuals (e.g. Nowak & Sigmund, 1998;

Johnstone, 2001). It seems that, whereas traditional evolutionary game theory

models are well suited to the study of dyadic encounters and contests, a network

of individuals all gathering information from each other and using this in sig-

nalling interactions, which may or may not be directed at other receivers as well

as the primary receiver, may prove too complicated to be tractable (although see

p. 617). It is with this in mind that we consider the role of simulation modelling

as a tool for studying communication in networks.

Simulation models are widely used in ecology as they allow the user to incor-

porate an unlimited number of variables and parameters in the model, and they

are being increasingly used in behavioural studies. A recent trend has favoured

individually based and spatially explicit models, which contain a discrete popula-

tion of individual animals within a defined spatial environment. Individuals are

governed by a series of movement and behavioural rules (Grimm, 1999). These

models, therefore, allow an increase in biological realism (Wilson, 2000) by re-

placing the inaccurate assumptions about the random or structured patterns of

interaction within a population found in simpler models with assumptions that

better capture spatial structuring. These types of model have been used to show

how complex collective behaviours can arise from the interaction of individu-

als obeying simple behavioural rules (Hemelrijk, 2002). For example, Hemelrijk

(2000) created an individually based model of dominance interactions within a

social group to show that increased aggression caused the emergence of ‘selfish

herd’ organization in the group. Previous theoretical studies of selfish herding

had difficulties equating the complex movement rules needed to make individ-

uals aggregate in the models (i.e. with tight clustering of dominants surrounded

by subordinate individuals (Hamilton, 1971)) with observations of this herding



behaviour in the field (Morton et al., 1994; Viscido et al., 2002). Individuals in the

model of Hemelrijk (2000) were governed by two rules; the tendency to aggregate

and the tendency to enter into dominance interactions. When an individual lost

a dominance interaction, it would flee and the winner would move to occupy its

space. In the model, it became possible for weak individuals to get ‘caught in a rut’

and lose several interactions in a row causing them to move to the periphery of

the group, whereas the strongest individuals would remain in the centre. This is

an example of a model that made no assumptions about the way in which individ-

uals chose to aggregate and yet patterns of social organization that were similar

to those observed in the wild emerged as products of aggressive interactions and

individual differences.

Small-world network analysis

As explained above, most models of communication have not considered

the role of individuals other than the immediate interactants; when they have,

they placed unrealistic assumptions on the information gathered. A further con-

sideration in the theoretical analysis of communication and information gather-

ing in a network is whether the way in which the network is organized affects

its function. In recent years, considerable research has focused on the structure

and organization of networks. Most networks associated with social interactions

may be physically limited. For example, the interactions in a territorial system are

mostly restricted to neighbouring individuals. However, a consideration of com-

munication networks means that a larger group of individuals must be considered

in the analysis as the long-range signals most commonly used for advertising or ag-

gressive interactions usually travel much further than an individual’s immediate

neighbours (McGregor & Peake, 2000).

The analysis of networks in fields as diverse as metabolic pathways in eukaryote

cells (Jeong et al., 2000), food webs (Williams et al., 2000; but see Dunne et al., 2002)

and links in the World Wide Web (Albert et al., 1999) has found that they show a

number of similar structural properties. As a result of these similarities, such net-

works are referred to as ‘small-worlds’ or ‘scale-free’ networks (Watts & Strogatz,

1998; Barabasi & Albert, 1999). Fundamentally, these networks are dynamic: a

new node joining the network is likely to attach preferentially to certain exist-

ing nodes. As a result, networks arise that contain tight clustering around some

highly connected nodes, called ‘hubs’.There is no one centralized dominant node;

consequently, organization is spread between the few highly connected hubs. This

organization is both the network’s main asset and its Achilles heel, as it means

that the behaviour of most nodes has little impact on the network at large, but

removing one of the hubs can have a critical impact on the flow of information

through the network (Albert & Barabasi, 2002).



A key consequence of this structure is that these networks have short path

lengths between any two nodes (i.e. any two individuals within the population

can be linked by a small number of connections). For example, Kudo & Dunbar

(2001) found that primate groups remained socially interconnected despite be-

ing fragmented into small cliques, with males possibly serving the role as hubs,

connecting female groups. The implication of this phenomenon is that informa-

tion exchanged between individuals may spread widely and rapidly throughout

the network. Watts & Strogatz (1998) considered a simple model of a contagion

that examined how a disease might spread through a small-world network. They

found, rather frighteningly, that only a few hubs were required for the disease to

spread rapidly through the population. This model can be directly compared to

a communication situation, where the probability of an individual producing a

signal is affected by the number of others it perceives signalling. The conclusion

would be that, in a small-world network, a bout of signalling could spread very

rapidly throughout an entire population (see Ch. 12 on information cascades).

In summary, small-world analyses suggest that, when studying the pattern of

communication within a population, it may be important to identify how asym-

metric the communication networks of individuals are. In this context, asymme-

try means the number of connections each individual has with other individuals.

In a random network, all individuals would have, on average, the same num-

ber of connections. However, in a small-world network, the few hub individuals

have the majority of connections, while the rest of the population has very few.

Asymmetry could be imposed by environmental features (a hub individual could

occupy a more central position within a forest or a more open spot from which his

signal could propagate further) or by social roles (e.g. male primates connecting

female groups; or possibly ‘floating’juveniles connecting territorial adults in bird

species).

Models of communication network dynamics from

signallers’ perspectives

There are a number of potential costs and benefits associated with sig-

nalling in a network. Signallers must compete with each other to make their sig-

nals detectable by receivers. They must also balance the benefits of the intended

receiver perceiving the signal with the costs of other receivers doing the same.

These costs can range from heterospecific predators or parasites to competing

conspecifics. Signalling within a network may also coordinate behaviour among

individuals within larger social groups. Although many experimental studies have

shown the effects of signalling within networks of several individuals, there have

been far fewer studies modelling the effects of networks on signals and signalling

dynamics (but see Chs. 2, 5 and 13). Here we highlight some of the benefits of



modelling signalling within groups with three case studies: the chorus behaviour

of acoustic insects and frogs, the social coordination of ant foraging, and territory

establishment. The case studies share similarities in the methods they employ and

the conclusions they draw. In each case, the examples show that complex patterns

of behaviour can arise from simple individual decision patterns.

Signal dynamics in acoustic choruses

Choruses are of interest when studying behaviour in communication

networks from both experimental and theoretical perspectives as they can show

tightly coordinated patterns of signalling within large groups (Chs. 12 and 13). Indi-

viduals within an acoustic chorus must deal with a complex acoustic background

generated by other signallers within which they must maximize the efficacy

of their signals in terms of transmission and female attraction. Females can show

preferences for specific temporal features of male signals; for example, they may

respond less to calls that are overlapped (e.g. gray tree frogs Hyla versicolor: Schwartz

et al., 2001) or may prefer leading calls (Snedden & Greenfield, 1998; Greenfield,

2002). Thus, female preferences for certain temporal features may have led to pat-

terns of synchronous and alternating choruses in anuran and insect species. In

general, when a species calls at a rapid rate, choruses tend towards synchrony; as

the call period becomes longer, choruses are more likely to alternate (Grafe, 1999).

Anuran and insect choruses are amenable to experimental studies of net-

work behaviour because the whole network can be controlled and manipulated

(Schwartz et al., 2002) and individual behaviour within the network can be mea-

sured. This level of experimental control allows the predictions of models to be

tested. However, to date, there have been few theoretical considerations of sig-

nalling dynamics within choruses. We discuss two different approaches that have

been used to model signalling within choruses (Brush & Narins, 1989; Greenfield et

al., 1997). Both models consider mechanisms individuals may use to control their

call timing and hence avoid interference in a chorus. The mechanism used in each

case is a form of inhibitory resetting. Each individual has an internal mechanism

that increases from a basal state to a peak where it initiates a call. If, before call-

ing, the individual perceives another individual’s call, the mechanism is reset to

its basal level and it begins to increase again. Such mechanisms have been shown

to exist for many anuran and insect species (Zelick & Narins, 1985; Greenfield

et al., 1997).

Brush & Narins (1989) adapted models of computer networks to study whether

choruses in the Puerto Rican treefrog Eleutherodactylus coqui were controlled by this

inhibitory resetting mechanism. Computer-network models simulate the flow of

data between interconnected computers and study how a shared resource (i.e.

bandwidth) is partitioned between them. Individual computers are linked to each

other via data lines and send packets of information through the network. Before



sending a packet, each computer checks if the data line is clear. If it is not, the

computer waits for a random time period before checking the line again (Brush &

Narins, 1989). However, even when the line is free, computers have a probability of

deferring the transfer. Also in some cases, the terminals must transmit informa-

tion even if the lines are busy. In the model, frogs represented the terminals; the

data lines represented the communication network between them, and their calls

were the packets of information. Although their model was an individually based

simulation, it was not spatially explicit and, therefore, distance between individ-

uals was not a factor in the analysis. The frogs in this model used their inhibitory

resetting to avoid being jammed by other individuals; if they detected another frog

calling during their refractory period, they delayed the next call by a randomly

chosen period and then returned to the standard refractory period for the next

call (Brush & Narins, 1989). Using this model, Brush & Narins (1989) showed that

this mechanism would lead to fewer calls being overlapped and that there was an

optimal chorus size of between three and four individuals at which information

transfer was maximized (i.e. overlap was minimized). The results of the model

were also corroborated by field data collected on the treefrog, which showed that

choruses occurred in small groups and when group size was large (five or six

individuals) males showed selective attention to one or two neighbours. These

results are similar to those of Greenfield & Rand (2000) who show that tungara

frogs Physalaemus pustulosus paid attention to a subset of the potential signallers

in artificially generated choruses.

Greenfield et al. (1997) developed a linear model of chorus signalling in the

katydid Neoconocephalus spiza, which initially was based on dyadic interactions.

Individuals in their model used an inhibitory resetting mechanism similar to that

of Brush & Narins (1989). They modelled two individuals signalling at the same

time and measured signal overlap from a receiving female’s perspective. They

showed that individuals overlapped far less and avoided producing following calls

(i.e. calls that were initiated after the onset of another male’s call) when using an

inhibitory resetting mechanism, and that signals tended towards synchrony or

alternation depending on the speed at which the mechanism reached its peak

level (i.e. at call initiation). If males within a given call period could return to

their peak level quickly from inhibition (‘rebound’), alternation would arise as

males could quickly begin calling again. However, if the rebound took almost as

long as the call cycle, males would fall into bouts of synchrony (Greenfield et al.,

1997). They also considered the case of a larger number of calling males. This

model was an individually based and spatially explicit simulation, with males

randomly placed on a 20 m × 20 m grid. Simulated females were also randomly

located around this grid; the model assessed the level of call overlapping from the

female’s perspective. Male attractiveness was assessed as the number of their calls



that were overlapped. When they ran simulations for a number of males (2–10),

Greenfield et al. (1997) found that males had to pay attention to a subset of the

chorus for inhibitory resetting to function as a chorus mechanism, but when they

did, patterns of synchrony and alternation emerged.

The theoretical analyses of call timing in choruses of signalling anurans and

acoustic insects have shown that simple mechanisms that control signal timing,

which may have arisen in response to female preferences for leading signals, can

generate the complex patterns of alternation and synchrony observed in nature.

An emergent feature of these models was that, given the existence of such a re-

setting mechanism, males could only pay attention to a subset of the individuals

in their chorus when determining their call timing. Choruses, in general, would

seem to represent a tractable means of studying signalling dynamics in networks.

Although the individuals within the chorus are signalling to females, males in-

directly compete with each other for acoustic space and directly compete over

actual resources, all within the scope of the chorus. Therefore, it is surprising that

chorus dynamics have received such limited theoretical attention given that they

represent such an amenable study system for simulation models.

Swarm intelligence and self-organization in social insects

Social insects are not noted for their individual cognitive abilities, yet

they are famous for their ability collectively to ‘solve’ problems of how best to ex-

ploit food resources in their environment. Such coordination requires communi-

cation between individuals; for example, the honeybee Apis mellifera uses a waggle

dance that indicates the location of food sources to other individuals (von Frisch,

1967). Similarly, ants use pheromone trails to lead colony members to food. The

collectively adaptive processes that arise out of these interactions are examples

of self-organization. Key ‘ingredients’ of self-organization (Bonabeau et al., 1999,

p. 9) are positive feedback (e.g. one forager recruits more bees to a food source by

dancing) and multiple communicative interactions.

The field of foraging strategies in social insects has a rich empirical background.

Moreover, several recent models of such behaviour (Deneubourg & Goss, 1989;

Deneubourg et al., 1990; Camazine & Sneyd, 1991; Seeley et al., 1991; reviewed by

Bonabeau et al., 1999) have shed light on the much simpler underlying individual

behaviour patterns. In these models, the type of foraging problem that is faced

typically structures the social and communication networks of the population.

For example, individuals that are laying a pheromone trail for a food source that

is nearby will overlay that trail more frequently than those laying a trail for a food

source further away, simply because they move along it more frequently. Such a

model can explain the ability of ants preferentially to use the nearest food source

first (Resnick, 1994).



Most of these models have implicitly incorporated communication networks.

For example, Deneubourg et al. (1990) investigated the ability of ants to choose

one of two paths to a food source. Experimental evidence suggests that Argentine

ants Linepithema humile quickly decide to use one of the two paths. Deneubourg

et al. (1990) fitted this behaviour to a simple positive-feedback model in which

the more ants had travelled along a path, the more likely it was to be chosen by

another ant. Interestingly, the best fit between model parameters and empirical

data occurred when each extra ant travelling along a path had a greater impact on

others’ behaviour (e.g. the twentieth ant had a disproportionately larger impact

than the first). This is yet another example of asymmetry in networks, where the

most recent information has a greater impact than old, and possibly outdated,

information. A similar situation is found in the chemical over-marking signals of

mice (e.g. Rich & Hurst 1999; Ch. 11).

Territory establishment

To compare and contrast the different approaches to modelling behaviour

and communication, we use an example from models of territory establishment.

Models of territory establishment are predominantly based on game theory, where

a series of dyadic encounters occur over indivisible areas. These models contain

‘winner-takes-all’ assumptions, in that whoever wins the contest takes the re-

source, and contests cannot end in draws or with division of the area.

Stamps & Krishnan (1999) developed an individually based spatial simulation

model of territory establishment. In their model, individuals moved around a spa-

tially heterogeneous area containing patches of different size. At each time step,

they assessed the attractiveness of the patches around them and moved into the

one with the highest attractiveness. The attractiveness of an area was based on two

key parameters: positive and aggressive experiences. Positive experiences occurred

when an individual entered a patch and did not become involved in an aggres-

sive interaction. Positive experiences increased the attractiveness of the patch and

thus increased the likelihood that the individual would return to the patch in the

future. Aggressive experiences occurred when two individuals entered the same

patch. They would then enter into a costly interaction, which, for this model,

would end in a draw (i.e. there would be no clear winner). Aggressive encounters

would discourage individuals from returning to that patch. The model predicted

that individuals would gradually build territories by incorporating novel patches

into their home range. Individuals would show periods of sustained aggression

when fighting over familiar (i.e. repeatedly visited) sites and they could also take

over sites by repeatedly entering into aggressive interactions. The net result of

the model was that a pattern of stable territories would be generated through

repeated interactions where there was no clear winner taking a resource.



Game theoretic models make similar predictions to those of Stamps & Krishnan

(1999, 2001), albeit for different reasons (Sih & Mateo, 2001). Similar predictions

include the increased benefits of being a resident, prolonged encounters between

two residents and the more desperate attempts of newcomers to claim territorial

space when they have few remaining options. In game theory models, the territory

has an intrinsic value, which is greater for residents than newcomers, and the

stable strategy that evolves is one of territory choice (i.e. when to stay and when

to move on).

As with the previous case studies, the models of Stamps & Krishnan (1999, 2001)

do not explicitly study networks; however, they contain network-like features, as

the behaviour of individuals are affected by their previous encounters with several

other individuals. The model could be extended to include some explicit network

effects such as eavesdropping (e.g. Ch. 2). For example, an individual’s decision

to enter an area could, in part, be based on previous encounters it had observed

take place there. This would allow it to discriminate between two novel areas, one

hotly contested by other individuals and one that was not. This form of modelling

shows that individually based simulation models can create predictions similar

to those of game models but without the same restrictive assumptions. As in this

case, they can also extend the predictions that the models can make.

Summary

In this section, we have chosen examples of communication within net-

works that are very different. While these studies have not explicitly used the

concept of communication networks, they would not be possible without such a

viewpoint. They also show how the network may influence a signaller’s behaviour

and show that complex patterns of organization and behaviour within networks

may be possible through the implementation of simple rules. The territory estab-

lishment models also show how the simulation models can be compared with

game theoretic models and generate similar predictions while making fewer re-

strictive assumptions.

Models of communication networks from receivers’ perspectives

In the previous section, we emphasized the consequences of communi-

cation networks for signalling behaviour and how it influenced population level

patterns of signalling and organization. In this section, we examine how receivers

can use the network environment to extract information about others and modify

their own behaviour. We deal primarily with the role of eavesdropping (Ch. 2) as a

means of information gathering. Eavesdropping is one source of information that

becomes available to receivers within a network and it is the one that has received



the most empirical study. Here we describe two case studies that reflect the new ap-

proaches being taken to the theoretical analysis of communication networks; the

game theory simulation model (R. Lachlan & A. M. R. Terry, unpublished data) stud-

ies the evolution of eavesdropping as a strategy and the spatial simulation model

(R. L. Earley, S. Brosnan & J. Bragg, unpublished data) determines how eavesdrop-

ping may shape the formation of linear dominance hierarchies. Both models are

spatial simulations and emphasize the point made throughout this chapter: that

consideration of the spatial nature of communication is required for the success-

ful modelling of communication networks. In this section, we also discuss how

game theory has been used to model communication in groups and why tradi-

tional game theory methods may not be best suited to the study of networks. We

conclude that significant advances may be made through the combined use of

both proximate-based simulation models and strategic decision making through

game theory models.

Game theory and eavesdropping

Evolutionary game theory encompasses a well-established set of tech-

niques for determining which strategies prove most effective in interactions be-

tween individuals. The aim is to establish whether a given strategy can be invaded

by any other strategy; if not, it can be called an evolutionarily stable strategy (ESS,

Maynard Smith, 1982). There have been several studies using traditional game the-

ory in which individuals use communication networks to predict the behaviour

of others. The studies adapted well-established game scenarios and investigated

how a given eavesdropping strategy would fare in the game. Johnstone (1998)

developed a model of signal detection that aimed to determine whether ‘con-

spiratorial whispers’ (low-cost and inconspicuous signals (Krebs & Dawkins, 1984))

could be evolutionarily stable in cooperative communication systems. The idea of

conspiratorial whispers implicitly recognizes the role of eavesdroppers and hence

communication networks. Here eavesdroppers were modelled in the most general

sense of the word (i.e. both conspecific and heterospecifics receivers, see Ch. 2) and

it was assumed that it was generally costly to be overheard by an eavesdropper.

Johnstone’s (1998) model maintained that even when signalling was cooperative,

expenditure was required to make the signal detectable for a receiver and that

this creates a conflict of interest between signallers and receivers in the face of

the costly eavesdroppers. However, in many situations it may not be costly to be

overheard. Pollock & Dugatkin (1992) investigated eavesdropping in the famous

Prisoner’s Dilemma game. The Prisoner’s Dilemma is an extreme abstraction of

many cooperative situations where individuals have the option of either cooper-

ating or defecting in any given round of the game (for a review, see Dugatkin,

1998). Cooperators benefit if they play one another, compared with two defectors



playing one another. However, if a cooperator plays a defector, the former does

badly, while the latter does well (i.e. the defector exploits the cooperator’s benefi-

cence). Therefore, in a single round of the game, individuals should always defect

as they will benefit the most; if, however, the game is repeated over a number

of rounds, a strategy of reciprocation or ‘tit-for-tat’ becomes most beneficial and

stable (Axelrod & Hamilton, 1981; Stephens et al., 2002).

Since the publication of the classic paper by Axelrod and Hamilton (1981), many

revisions have been published, but versions of tit-for-tat are still regarded as the

most successful strategy. Although individuals playing tit-for-tat copy the strategy

their opponent played in the previous round, in reality this information may not

be present (if they only play each other once) or might be unreliable (if individ-

uals update their strategy frequently). In these situations, individuals could use

eavesdropping as a way of obtaining up-to-date information about their opponents’

strategies. Pollock & Dugatkin (1992) found that their so-called ‘observer tit-for-tat’

strategy was sometimes successful when tit-for-tat itself was not an evolutionarily

stable strategy (although observer tit-for-tat was out-competed by tit-for-tat un-

der many conditions). Sigmund & Nowak (1998) examined a similar situation and

found that indirect reciprocity through ‘image scoring’ (i.e. cooperating with in-

dividuals that had a record of cooperation) was a successful strategy. In this case,

individuals gain an increase to their image score each time they cooperate and a

decrease when they do not (for more details, see Ch. 22).

One of the main criticisms of image scoring was that observing an individual’s

image score did not take into account the behaviour of that individual’sopponent.

Thus, an observer would react the same way to an individual that defected against a

notoriously uncooperative opponent as to one that defected against a good cooper-

ator (Leimar & Hammerstein, 2001). To remedy this, Leimar & Hammerstein (2001)

investigated the evolutionary stability of the ‘good standing’ strategy (Sugden,

1986), in which individuals strive to maintain their good social standing. Under

this strategy, individuals could improve their standing by cooperating and could

have it damaged by defecting, but defecting against a player that was uncooper-

ative was not punished. Leimar & Hammerstein (2001) found that this strategy

was very successful. Milinski et al. (2001) investigated whether humans used good

standing in cooperative games but found that the simpler image scoring tended to

be used. They concluded that a strategy using good standing might ask too much

of working memory as individuals would have to remember each opponent’s pre-

vious interactions and when there were errors in perception of the roles adopted

in the interaction, image scoring predominated (Milinski et al., 2001).

The models discussed above studied the evolution of cooperation through

indirect reciprocity, something which takes place within communication net-

works (e.g. Ch. 22); however, they do not really examine how individuals interpret



interactions between others, a prerequisite for eavesdropping. In a more explicit

model of eavesdropping, Johnstone (2001) adapted the hawk–dove game to assess

the use of eavesdropping within such a framework of contests. The hawk–dove

model is a simple game that models some of the essential features of animal con-

tests (Maynard Smith, 1982; Riechert, 1998). In this game, individuals can play

one of two strategies; aggressive hawks escalate contests into fights while peaceful

doves rely on non-aggressive displays. Hawks, therefore, always beat doves but risk

damage if they face another hawk. Doves always lose to hawks but suffer no cost to

meeting another dove. With a high risk of damage from fights, the evolutionar-

ily stable strategy of this game is a mixture of hawks and doves (Maynard Smith,

1982).

In Johnstone’s (2001) model, the success of a third strategy, eavesdropper was

investigated in comparison to the pure strategies. The eavesdropper chose to play

dove if its opponent had won its previous encounter, and hawk otherwise. To repli-

cate ‘error’ in an eavesdropper’s assessment of the outcome of an interaction,

there was a chance that individuals could misinterpret their eavesdropping, for

example, by playing hawk against a winner. The model showed that eavesdropper

was most common when there was a high cost to fighting. When eavesdropping

errors were more frequent, eavesdroppers reached their peak frequency at a lower

fighting cost. However, eavesdropping never spread to fixation within the popu-

lation. Johnstone (2001) suggests that this is because eavesdroppers are unable to

assess the strategy that other eavesdroppers will adopt, because it may change each

round. Therefore, eavesdroppers are reducing the fitness benefit upon which they

are based. When rare, they have the advantage of predicting the correct role in

most cases. Consequently, a mixed equilibrium evolves with eavesdroppers at low

frequencies. A surprising result of the model was that eavesdropping promoted

increased aggression. This is because the model includes a form of ‘winner–loser’

effect: an individual that won in one round is more likely to win in the next as

eavesdroppers will chose a submissive role to play.

In the models described above, individuals were restricted to obtaining infor-

mation about only a small part of the interaction between two others. For example,

Johnstone’s (2001) model focused on the outcomes of interactions (who won and

who lost) and not on the roles adopted in the interactions themselves. This repre-

sents one form of eavesdropping (interceptive eavesdropping, see Ch. 2); however, it

is likely that real eavesdroppers also consider the roles that the interactants play

(who played hawk or dove). Finally, except for the simple viscosity factor of Pollock

& Dugatkin (1992), which defined a probability that an individual would have

eavesdropped on his opponent’s last contest (Nowak & Sigmund (1998), include

a similar factor), the models did not consider how structuring the communica-

tion networks might be important. However, as stressed throughout this chapter,



spatial structuring of populations is likely to have a significant effect on the pre-

dictions of theoretical models. A spatial simulation model of the hawk–dove game,

in which strategies could evolve in response to all aspects of the interaction, has

been developed by R. Lachlan and A. M. R. Terry (unpublished data) to analyse both

these factors. In a hawk–dove game between two individuals, A and B, player A

could be involved in one of six different types of interaction:

1. A plays hawk, defeats B, which also plays hawk

2. A plays hawk, loses to B, which also plays hawk

3. A plays dove, defeats B, which also plays dove

4. A plays dove, loses to B, which also plays dove

5. A plays hawk, defeats B, which plays dove

6. A plays dove, loses to B, which plays hawk.

An eavesdropper on the interaction, who would eventually play A, could, therefore,

obtain one of six pieces of information about A.

The spatial simulation model (R. Lachlan & A. M. R. Terry, unpublished data)

examines how eavesdroppers should respond to obtaining one of these pieces of

information. The model is a spatially explicit game theoretic simulation: individ-

uals within the population were placed in a 40 × 40 two-dimensional grid (i.e.

population size of 1600). In each round of the simulation, individuals engaged

in one contest with each of their four neighbours. The strategy adopted by each

individual was determined by a vector containing six values (varying between 0

and 1). Each value represented the probability of playing either hawk or dove hav-

ing just witnessed the opponent in one of the six situations mentioned above.

For example an individual with the vector {0.9 0.8 0.2 0 0.7 0.2} would follow an

image-scoring strategy as it would tend to play hawk, having observed its oppo-

nent play hawk in a previous round (refer to the six types of information listed

above).

Mortality and reproduction were modelled by having individuals periodically

update their strategy by choosing one of their neighbours’ strategies. This choice

was determined by the success of individuals: that is, their total payoff after play-

ing the game during the previous round. We investigated two types of inheritance;

either the strategy of the most successful neighbour was inherited, or the probabil-

ity of inheritance was directly proportional to the neighbours’relative success. The

difference between these conditions was that only the most successful strategies

were rewarded in the first, whereas in the second, moderately successful strategies

could also be inherited. A mutation rate was also included, which would create

novel strategies in new individuals.

Over a range of conditions, a variety of eavesdropping strategies evolved, but

only two groups of strategies were found to arise regularly. The first strategy is



Fig. 26.1. The pattern of aggression in a simulated population of 1600 individuals.

Each square represents an individual and the square’s shade of grey represents how

many times the individual was attacked over the previous 20 rounds of interaction:

attacked every time (black); never attacked (white). (a) The pattern for an image-scoring

strategy (individuals copy the hawk or dove behaviour of their opponents) is a similar

level of aggression to that of neighbours (the pattern is a fairly even shade of grey) but

darker areas show local waves of aggression. (b) The pattern for a reputation-scoring

strategy, in which individuals copied the behaviour of their opponent’s opponent (i.e.

if an individual was witnessed being attacked, it was subsequently more likely to be

attacked), shows that squares are usually either black or white.

somewhat similar to the image-scoring (or observer tit-for-tat) strategy: play hawk if

you witness your opponent playing hawk, and dove if he played dove. However, under

a wider range of conditions, especially when only the most successful individuals

could pass their strategies on, a second group of strategies was most successful.

This group consisted of two extreme strategies and a range of intermediates be-

tween them. The first extreme corresponded to Johnstone’s (2001) eavesdropper

strategy (i.e. play hawk if you eavesdrop on situations 2, 4 or 6; play dove otherwise,

see Fig. 26.1a). The more common extreme (which we call reputation scoring),

however, consisted of a novel strategy in which individuals essentially copied the

behaviour of their opponent’sopponent (i.e. play hawk in situations 1, 2 and 6; play

dove otherwise). An anthropomorphism of this strategy would be that individuals

paid attention to an individual’s reputation rather than its deeds. The reputation-

scoring strategy was successful because it tended to lead to neighbours ‘ganging

up’ on the same individuals (Fig. 26.1b); as a result, some individuals were highly

successful, and others were very unsuccessful (Fig. 26.2). The overall mean level of

success for the image-scoring and reputation-scoring strategies was actually rather



(b)

0

300

600

900

0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 3

Success (arbitrary units)

Fre

qu

ency

(a)

0

300

600

900

0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 3

Success (arbitrary units)

Fre

qu

ency

Fig. 26.2. The distribution of individual success after five rounds of interaction (five

lines on graph). Ten runs of a simulation of 1600 individuals were carried out for each

case. (a) The distribution for an image-scoring strategy shows one sharp peak,

indicating that most individuals have a similar, moderate level of success. (b) The

distribution for a reputation-scoring strategy has two peaks, indicating that

individuals were either successful or unsuccessful.

similar (1.78 versus 1.73), but the higher frequency of very successful individuals

means that the reputation-scoring strategy out-competes the image-scoring strat-

egy if the most successful individuals within a local neighbourhood monopolize

their success at replication (Fig. 26.2). The reputation-scoring strategy is the closest



approximation to social eavesdropping (Ch. 2), because it requires eavesdroppers

to pay attention to the interaction and the role of each individual. The model’s

combination of a spatial simulation limiting individuals to information gathered

from the interactions going on around them, with a cost–benefit approach of de-

termining the optimal strategy to follow should provide a deeper understanding

of the evolution of eavesdropping as a strategy for gathering information.

Network effects on the formation of linear dominance hierarchies

The formation of dominance hierarchies represents another area where

information from interactions between individuals can be used by observers to

influence future encounters (e.g. Dugatkin, 2001). A linear dominance hierarchy

is defined by the number of component triads (trios) within the group that form

transitive relationships (i.e. if A beats B, and B beats C, then A also beats C); as the

number of transitive relationships increases, linearity increases. Here we consider

a simulation model of the effects of communication networks on the formation

of linear dominance hierarchies (R. L. Earley, S. Brosnan & J. Bragg, unpublished

data).

Linear dominance hierarchies are established via overt aggressive interactions

and their establishment leads to the unequal distribution of resources among

dominant and subordinate individuals. Hierarchies also cause a general decrease

in the overall aggression levels within the group. Dominance hierarchy forma-

tion has been studied in a wide range of taxa; however, the factors involved in

their formation remain contentious. Conceptual models attribute the formation

of linear hierarchies either to some aspect related to fighting ability (e.g. Slater,

1986; Jackson & Winnegrad, 1988) or to social effects such as winner effects, loser

effects and eavesdropping (e.g. Chase, 1980; Bonabeau et al., 1996; Dugatkin, 1997,

2001). The individual-based spatially explicit simulation model is being developed

(R. L. Earley, S. Brosnan & J. Bragg, unpublished data) to study how social eaves-

dropping may influence the dynamics of hierarchy formation in groups of virtual

animals.

In each simulation, a group of 10 individuals are allowed to interact for a

predetermined period of time. At each time step, one individual can initiate an

interaction with another and, if the other individual responds an aggressive in-

teraction begins. Individuals can interact through displays or they can escalate

the contest to fighting. At the conclusion of the contest, individuals update their

estimate of their own fighting ability. These updates mimic the winner and loser

effects where dominant animals increase and subordinate animals decrease their

perception of their own fighting ability (Hsu & Wolf, 2001). A certain proportion

of individuals close to the interaction (eavesdroppers) can observe it and, in conse-

quence, update their estimates of the interactants’ fighting abilities. The estimate



of an opponent’sfighting ability determines whether an individual will initiate an

interaction and to what extent it will be pursued. The outcome of the simulation

is a dominance matrix and the degree of linearity is determined using Landau’s

index (0 < h < 1; Landau, 1951). If Landau’s h > 0.9, the hierarchy is considered to

be linear (Chase, 1974).

In an initial assessment of the model, the social effects were kept symmetrical

(i.e. winner effects = loser effects); there was no initial variation in fighting abili-

ties, and all individuals could eavesdrop on each interaction within their network.

When social factors were excluded from the model, display and escalated inter-

actions occurred with equal frequency and non-linear hierarchies emerged. Also

when winner or loser effects operated alone, the linearity index remained low

and did not increase greatly when the magnitude of the winner/loser effects was

increased. However, the model showed that eavesdropping, when it was included,

acted to increase the estimates of the fighting abilities of winning individuals

and to decrease those of losers. When included with winner–loser effects, eaves-

dropping caused the formation of strongly linear hierarchies. The most important

factor was the inflation of the winner’s estimated fighting ability (decreasing the

loser’s estimate had less effect). As with our model of a spatial hawk–dove game,

this model provided a simulation of social eavesdropping (Ch. 2) as individuals

paid attention to both the interactants. The resulting modification of the esti-

mates of fighting ability depended not only on each of the interactants but also

on the relative differences in fighting ability when the two individuals met in the

interaction. This model is a first attempt to study the implications of networks

in the formation of dominance hierarchies. Future studies could investigate the

relationship between asymmetries in both fighting abilities and eavesdropping in

promoting or hindering linear dominance hierarchies.

Summary

The models detailed here have provided the first theoretical studies of the

role of eavesdropping in communication networks. Eavesdropping is one of the

potential sources of information available to receivers and has received the most

experimental attention (McGregor & Peake, 2000, Ch. 2). Experimental data from

different taxa have shown that individuals can and do pay attention to the inter-

actions of conspecifics and that these interactions will modify the behaviour of

individuals in future encounters (e.g. Oliveira et al., 1998; Peake et al., 2001, 2002). It

is likely that eavesdropping is a common behaviour. Although originally thought

to be a cost-free source of information (McGregor, 1993), it is likely that eavesdrop-

ping has costs associated with the partitioning of cognitive processes required to

follow interactions and the fact that individuals may have to abstain from per-

forming other behaviours to witness interactions. Initial theoretical treatments



of eavesdropping showed that it was only stable as a minority strategy because

eavesdroppers could not predict how other eavesdroppers would behave and, as

a result, it would lead to more aggressive encounters (Johnstone, 2001). However,

when eavesdroppers were able to follow the interactions of individuals around

them, as opposed to being randomly drawn into interactions, it became apparent

that eavesdropping was a viable strategy (R. Lachlan & A. M. R. Terry, unpublished

data). Further analyses will address how individuals balance the costs and benefits

of eavesdropping against other sources of information or other behaviours.

Summary

When communication is considered to occur within a network, new pos-

sibilities emerge for individuals to broadcast and receive information that will

affect their behaviour in future interactions. Networks also extend the considera-

tion of the costs and benefits of signalling to include other signallers and receivers

that may or may not be apparent to the respective interactants. Current models

of animal communication have been dominated by game theory, which is well

suited to the analysis of strategies used by individuals in dyadic encounters. How-

ever, when applied to networks, pure game theory models allow only a superficial

analysis of the costs and benefits of signalling. While consideration of the strategic

nature of communication in networks remains important, we feel that it must

be combined with more process-based approaches that place fewer restrictions on

individual behaviour. Individually based simulation models are becoming increas-

ingly common in behavioural ecology and a combination of these more proximate

level models with game theory approaches will give a greater understanding of the

evolution of communication networks. In particular, in the examples we have de-

scribed (anuran acoustic choruses and eavesdropping strategies), spatially realistic

individually based models have generated different results from more traditional

techniques applied to the same question. In this chapter, we have emphasized the

importance of studying realistically structured populations and have identified

the spatial nature of communication in networks as an important feature of any

theoretical consideration. The importance of the spatial nature of networks may

also extend to network organization. There are likely to be asymmetries in the

extent to which individuals contribute to the flow of information in a network:

some individuals providing more information than others. One future avenue

for both theoretical and experimental research may be to determine how com-

munication networks are organized and whether hub individuals act as routers

through which most information flows. The nature of the flow of information

through networks, whether it is a signal spreading through a chorus or an alarm



call spreading through a population, will undoubtedly have implications for the

evolution of signalling within that system.

Although there is a wide literature on the theoretical aspects of strategic deci-

sions when communicating and cooperating in groups, there is little experimen-

tal evidence to support these models (for exceptions see Wedekind & Milinski,

2000; Milinski, et al., 2001; Stephens et al., 2002). Work on cooperation in groups

has shown that modelled strategies may not consider the complex combination

of assumptions, working rules and limitations that individuals face when decid-

ing how to respond in interactions with known individuals (Milinski et al., 2001;

Stephens et al., 2002). We suggest that future models incorporating both strategic

and simulation aspects will be able to model more closely the dynamics involved

in observing and taking part in repeated interactions, and this, in turn, will lead to

a better understanding of the strategies and behaviours that individuals employ

when communicating in networks.

Acknowledgements

We are very grateful to Ryan Earley for allowing us to discuss and present his model. We

would also like to thank several people whose comments helped make this chapter more readable;

Ryan Earley, Ricardo Matos, Tom Peake and Denise Pope. A. T. was funded by the Zoological

Institute at the University of Copenhagen during the writing of this chapter.

References

Albert, R. & Barabasi, A. L. 2002. Statistical mechanics of complex networks. Reviews of

Modern Physics, 74, 47–97.

Albert, R., Jeong, H. & Barabasi, A. L. 1999. Diameter of the World-Wide Web. Nature,

401, 130–131.

Axelrod, R. & Hamilton, W. D. 1981. The evolution of cooperation. Science, 211,

1390–1396.

Barabasi, A. L. & Albert, R. 1999. Emergence of scaling in random networks. Science,

286, 509–512.

Begon, M., Harper, J. L. & Townsend, C. R. 1990. Ecology: Individuals, Populations and

Communities, 2nd edn. Oxford: Blackwell.

Bonabeau, E., Theraulaz, G. & Deneubourg, J.-L. 1996. Mathematical model of

self-organizing hierarchies in animal societies. Bulletin of Mathematical Biology, 58,

661–717.

Bonabeau, E., Dorigo, M. & Theraulaz, G. 1999. Swarm Intelligence: From Natural to

Artificial Systems. Oxford: Oxford University Press.



Brush, J. S. & Narins, P. M. 1989. Chorus dynamics of a neotropical amphibian

assemblage: comparison of computer-simulation and natural behavior. Animal


Camazine, S. & Sneyd, J. 1991. A model of collective nectar source selection by

honey-bees: self-organization through simple rules. Journal of Theoretical Biology,

149, 547–571.

Chase, I. D. 1974. Models of hierarchy formation in animal societies. Behavioral Science,

19, 374–382.

1980. Social process and hierarchy formation in small groups: a comparative

perspective. American Sociology Review, 45, 905–924.

Deneubourg, J. L. & Goss, S. 1989. Collective patterns and decision making. Ethology,


Deneubourg, J. L., Aron, S., Goss, S. & Pasteels, J. M. 1990. The self-organizing

exploratory patterns of the Argentine ant. Journal of Insect Behaviour, 3, 159–168.

Dieckman, U. C. 1997. Can adaptive dynamics invade? Trends in Ecology and Evolution,

12, 128–131.

Dugatkin, L. A. 1997. Winner and loser effects and the structure of dominance

hierarchies. Behavioral Ecology, 8, 583–587.

1998. Game theory and cooperation. In: Game Theory and Animal Behavior, ed. L. A.

Dugatkin & H. K. Reeve. Oxford: Oxford University Press, pp. 38–63.

2001. Bystander effects and the structure of dominance hierarchies. Behavioral

Ecology, 12, 348–352.

Dunne, J. A., Williams, R. J. & Martinez, N. D. 2002. Food-web structure and network

theory: the role of connectance and size. Proceedings of the National Academy of

Sciences, USA, 99, 12917–12922.

Grafe, T. U. 1999. A function of synchronous chorusing and a novel female preference

shift in an anuran. Proceedings of the Royal Society of London, Series B, 266, 2331–2336.



Greenfield, M. D. & Rand, A. S. 2000. Frogs have rules: selective attention algorithms

regulate chorusing in Physalaemus pustulosus (Leptodactylidae). Ethology, 106,

331–347.



1355–1361.

Grimm, V. 1999. Ten years of individual-based modelling in ecology: what have we

learned and what could we learn in the future? Ecological Modelling, 115, 129–148.

Hamilton, W. D. 1971. Geometry of the selfish herd. Journal of Theoretical Biology, 31,

295–311.

Hemelrijk, C. K. 2000. Towards the integration of social dominance and spatial

structure. Animal Behaviour, 59, 1035–1048.

2002. Understanding social behaviour with the help of complexity science. [Invited

article.] Ethology, 108, 655–671.



Hsu, Y. Y. & Wolf, L. L. 2001. The winner and loser effect: what fighting behaviours are

influenced? Animal Behaviour, 61, 777–786.

Jackson, W. M. & Winnegrad, R. L. 1988. Linearity in dominance hierarchies: a second

look at the individual attributes model. Animal Behaviour, 36, 1237–1240.

Jeong, H., Tombor, B., Albert, R., Oltval, Z. N. & Barabasi, A. L. 2000. The large-scale

organization of metabolic networks. Nature, 407, 651–654.

Johnstone, R. A. 1998. Conspiratorial whispers and conspicuous displays: games of

signal detection. Evolution, 52, 1554–1563.

2001. Eavesdropping and animal conflict. Proceedings of the National Academy of

Sciences, USA, 98, 9177–9180.

Krebs, J. R. & Dawkins, R. 1984. Animal signals: mind-reading and manipulation. In:

Behavioural Ecology: An Evolutionary Approach, 2nd edn, ed. J. R. Krebs & N. B. Davies.

Oxford: Blackwell, pp. 380–402.

Kudo, H. & Dunbar, R. I. M. 2001. Neocortex size and social network size in primates.


Landau, H. G. 1951. On dominance relations and the structure of animal societies: I.

Effect of inherent characteristics. Bulletin of Mathematical Biophysics, 13, 1–19.

Leimar, O. & Hammerstein, P. 2001 Evolution of cooperation through indirect


Maynard Smith, J. 1982. Evolution and the Theory of Games. Cambridge, UK: Cambridge

University Press.



Society, Series B, 340, 237–244.






Milinski, M., Semmann, D., Bakker, T. C. M. & Krambeck, H.-J. 2001. Cooperation

through indirect reciprocity: image scoring or standing strategy? Proceedings of the


Morton, T. L., Haefner, J. W., Nugala, V. & Decino, R. D. 1994. The selfish herd revisited:

do simple movement rules reduce relative predation risk? Journal of Theoretical

Biology, 167, 73–79.


Nature, 393, 573–577.









2002. Do great tits assess rivals by combining information gathered by


Pollock, G. & Dugatkin, L. A. 1992. Reciprocity and the evolution of reputation. Journal

of Theoretical Biology, 159, 25–37.

Resnick, M. 1994. Turtles, Termites and Traffic Jams. Cambridge, MA: MIT Press.

Rich, T. J. & Hurst, J. L. 1999. The competing countermarks hypothesis: reliable

assessment of competitive ability by potential mates. Animal Behaviour, 58,

1027–1037.

Riechert, S. E. 1998. Game theory and animal contests. In: Game Theory and Animal

Behaviour, ed. L. A. Dugatkin & H. K. Reeve. Oxford: Oxford University Press,

pp. 64–93.

Schwartz, J. J., Buchanan, B. W. & Gerhardt, H. C. 2001. Female mate choice in the gray

treefrog (Hyla versicolor) in three experimental environments. Behavioral Ecology


2002. Acoustic interactions among male gray treefrogs, Hyla versicolor, in a chorus

setting. Behavioral Ecology and Sociobiology, 53, 9–19.

Seeley, T. D., Camazine, S. & Sneyd, J. 1991. Collective decision making in honey-bees:

how colonies choose among nectar sources. Behavioral Ecology and Sociobiology, 28,

277–290.

Sigmund, K. & Nowak, M. A. 1998. The complexity of cooperation: agent-based models

of competition and collaboration. Nature, 392, 457–457.

Sih, A. & Mateo, J. 2001. Punishment and persistence pay: a new model of territory

establishment and space use. Trends in Ecology and Evolution, 16, 477–479.

Slater, P. J. B. 1986. Individual differences and dominance hierarchies. Animal

Behaviour, 34, 1264–1265.

Snedden, W. A. & Greenfield, M. D. 1998. Females prefer leading males: relative call

timing and sexual selection in katydid choruses. Animal Behaviour, 56, 1091–1098.

Stamps, J. A. & Krishnan, V. V. 1999. A learning-based model of territory establishment.

Quarterly Review of Biology, 74, 291–318.

2001. How territorial animals compete for divisible space: a learning-based model

with unequal competitors. American Naturalist, 157, 154–169.

Stephens, D. W., McLinn, C. M. & Stevens J. R. 2002. Discounting and reciprocity in an

iterated prisoner’s dilemma. Science, 298, 2216–2218.

Sugden, R. 1986. The Economics of Rights, Co-operation and Welfare. Oxford: Blackwell.

Turelli, M. & Barton, N. H. 1994. Genetic and statistical analyses of strong selection

and polygenic traits: what, me normal? Genetics, 138, 913–941.

Viscido, S. V. & Wethey, D. S. 2002. Quantitative analysis of fiddler crab flock

movement: evidence for ‘selfish herd’ behaviour. Animal Behaviour, 63, 735–741.

von Frisch, K. 1967. The Dance Language and Orientation of Bees. Cambridge, MA: Harvard

University Press.

Watts, D. J. & Strogatz, S. H. 1998. Collective dynamics of ‘small-world’ networks.

Nature, 393, 440–442.


Science, 288, 850–852.



Williams, R. J. & Martinez, N. D. 2000. Simple rules yield complex food webs. Nature,

404, 180–183.

Wilson, W. 2000. Simulating Ecological and Evolutionary Systems in C. Cambridge, UK:


Zelick, R. & Narins, P. M. 1985. Characterization of the advertisement call oscillator in

the frog Eleutherodactylus coqui. Journal of Comparative Physiology A: Sensory Neural and

Behavioral Physiology, 156, 223–229.

P1: IRK/KFO P2: KOD0521823617ind.xml CU1917B/McGregor 0 521 582361 7 April 12, 2005 11:14

Index

Abudefduf vaigiensis see sergeant major fish

Acheta domesticus see house cricket

Accipiter nisus see sparrowhawk

acoustic choruses, models of signalling

dynamics 609–611

see also anuran choruses; insect choruses

acoustic communication

caller identity and status information 381

factors affecting evolution of 152–153

problem of masking interference 157–162

see also signal production; sound

transmission

Acrocephalus arundinaceus see great reed

warbler

advertising signals, to facilitate

eavesdropping 48–49, 52

African catfish (Clarias gariepinus),

semiochemicals 558–559

African elephant (Loxodonta africana)

benefits of attending to others’ social calls

384

contact call discrimination and memory

383–384

fluid social systems and long-distance

signalling 377–378

group social knowledge and age of

matriarch 383–384

individuality in fundamental frequency

contour in vocalizations 382

infrasound communication 457–459

intelligible distance of calls 372, 382–383

social knowledge related to age and

experience 383–384

African grey parrot (Psittacus erithacus),

cognitive and communicative capacities

569, 572–573

Agelaius phoeniceus see red-winged blackbird

Agelenopsis aptera see funnel-web spider

aggression 2

and social instability in fishes 96

androgens and 481–482

in models of territory establishment

612–613

modelling linear dominance hierarchy

formation 620–621

physiological costs 191–192

song overlapping in songbirds 304,

305–306

victory displays 11

see also redirected aggression

aggressive calls, variety in male anurans

279–280

aggressive priming, audience effects

75–78

aggressive signal exchanges see

eavesdropping; signalling interactions

agonistic contests see aggression

Alcelaphus buselaphus see hartebeeste

allomones see fish semiochemicals

Alouatta pigra see black howler monkey

alpine accentor (Prunella collaris), quiet singing

53

628


Index 629

altruism

and cooperation in communication

networks 446

audience effects 30–31

dishonest signals used for tactical

deception 524–525

eavesdropping and indirect reciprocity

522–523

evolution and maintenance 536

functional rather than cognitive approach

522–523

image scoring and evolution of 533–534

reciprocal altruism and behavioural

strategies 521–522

Alytes spp. see midwife toads

Amolops tormotus see Chinese frog

androgens

adaptive value of social modulation 497,

502–505

and aggressive behaviour 481–482, 496–498,

499

and bystander priming response 499–500,

501

and electrocommunication signals

487–488

and pheromone production and/or release

485–486

and sex differences in spatial memory

490–491

brain receptors 490–491

costs and benefits of high levels 497,

502–505

effects of early exposure (critical period)

490–491

effects of high levels on male parental care

497, 504–505

effects of population density 492–494

effects on cognitive functions 490–491

effects on expression of somatic releasers

491–492

effects on pheromone production in

urodeles 485–486

effects on singing behaviour of songbirds

483–485

effects on visual displays in vertebrates

486–487

effects on vocal structures of amphibians

484–485

effects on vocal structures of birds 483–484

effects on vocal structures of toadfish 484

interactive effects with social environment

482

interrelationship with associative learning

mechanisms 505

levels during social instability 492–494

levels in dominant and subordinate males

492–494, 496–498, 499

modulation of behaviour in

communication networks 494–502

modulation of central mechanisms

affecting motivation 490

modulation of sensory perception 488–489

possible mediation of audience effects 501

possible mediation of dear enemy effects

501–502

role in winner–loser effects 198, 492–494,

496–498, 499

social modulation of androgen levels

492–494

social modulation of behavioural effects

481–482

stimulation by social interactions 492–494

Anolis aeneus see lizard

anonymity, used to counter eavesdropping

56–57

Anser anser see greylag goose

antbirds (Thamnophilidae), sound

characteristics 49

Antilocapra americana see pronghorn antelope

anuran amphibians

communication networks 2, 248

effects of androgens 484–485, 486–487

use of private signalling 291

victory displays 118, 122

anuran choruses 263–264

adaptations for acoustic competition

278–279

alternation of signalling 280–282


630 Index

anuran choruses (cont.)

auditory systems and background noise 292

call-timing, evolution and maintenance

282–284

conflicting demands of signalling 279–280

entrained calling 280–282

features of network communications

291–292

female acoustic responses 283

female attraction to larger choruses 283

female call preferences 264, 283–284,

285–286, 292

female choices in trials and natural

choruses 287–289

female cognitive requirements for social

eavesdropping 289

female eavesdropping on male–male

interactions 289

female mate sampling techniques 288–289

female sensory abilities and signal

discrimination 285

fine-scale patterns of signal timing 280–283

graded or discrete calls 279

leading and lagging roles 280–282

male aggressive calls 279–280

male calling energy costs 284–285

male interceptive eavesdropping 289–291

male–male vocal competition 278–279

males spatial distribution and selective

attention 284

models of signalling dynamics 609–610, 611

network view 277–278

‘off response’ call initiation 280–282

repertoire size of signallers 279

selection pressures on signals 277, 283–284

signal overlap avoidance in males 280–282

suitability for studying communications

networks 292–293

synchronous calling 280, 281, 282

two-part advertising call 279–280

use of both ‘on-response’ and ‘off-response’

calling 282–283

Aphelocoma californica see western scrub jay

Apis mellifera see honeybee

Aptenodytes forsteri see emperor penguin

Aptenodytes patagonicus see king penguin

Apternotus leptorhynchus see brown ghost

Argentine ant (Linepithema humile), foraging


arthropods, victory displays 117–118,

122

associative learning

and tactical deception 524–525

interrelationship with androgen effects

505

Atlantic salmon (Salmo salar), androgen effects

on olfactory sensitivity 488–489

audience effects 10, 30–31

and altruism 522–523

and eavesdropping 66

and victory displays 122–123

clarification of terminology 66–67, 68

definition 65–66

human behaviour 74–75

in redirected aggression 209–210

male–male aggressive signalling 68–72

male parental behaviour 72–74

on cheating behaviour 529–530, 531

on signalling 65–66

possible influences on scent marking

362–363

possible mediation by androgens 501

pre-exposure and male aggression 75–78

review of evidence for 67–75

audiences

and signal evolution 79

apparent 65

definition of 64

evolutionary 64–65

terminology 66–67, 68

see also eavesdropping

auditory systems

analysis by receivers 471–474

coping with background noise 292

hormonal modulation of sensitivity 489

Australian bushcricket (Elephantodeta nobilis),

evidence for eavesdropping 291

autocommunication, and eavesdropping 18


Index 631

avian cognition, research history and

prejudices 568–570

baboon (Papio cynocephalus)

knowledge of other animals 585–586, 587,

589

social structure and relationship

recognition 584

background noise

and functioning of auditory systems 292

effects of lunar phase 158, 163–164

see also masking interference; masking

release

Baktaman people (New Guinea) 421

Balaenoptera musculus see blue whale

Balaenoptera physalus see fin whale

banded wren (Thryothorus pleurostictus)

behaviour after dawn chorus 327

bout structure during and after dawn

chorus 329–331

dawn chorus as an interactive network 322,

337

dawn chorus features 326–327

dawn chorus singing and male quality 338

dawn chorus structure 336

daytime song-delivery patterns 326

male interactions at dawn chorus 335,

337–338

movement patterns around dawn chorus

330, 332

multiple song matching 335

overlapping and matching songs 331, 334,

335–336

recording methods 327–329, 330, 339–340

singing behaviour during and after dawn

chorus 323, 329–336

song behaviour 325–327

song matching as indicator of conflict

333–335

song matching during and after dawn

chorus 330, 332–335

song overlapping as an aggressive signal

331, 334, 335–336

song rates during dawn chorus 330, 331

song repertoire 326

territorial conflict and song matching

333–335

territorial behaviour 325–326, 327

variation in dawn chorus singing of

individual males 336

barking treefrog (Hyla gratiosa), female

assessment of males 137, 288–289

barn owl (Tyto alba)

nestling interactions 176

sound localization ability 465–467

barnacle goose (Branta leucopsis), victory

display 115–116

bats (various species)

attraction to echolocation signals 405

predation on katydids 154, 156–157

bearded seal (Erignathus berbatus), distinctive

group calls 399–400

behaviour, reciprocal link with hormones

481–482

beluga whale (Delphinapterus leucas)

avoidance of ice-breakers 406–407

reactions to killer whale sounds 403–404

victory display 118

Betta splendens see Siamese fighting fish

binaural processing, and acoustic signal

masking release 462–463

birds

effects of androgens on vocal structures

483–484

equating human and avian cognitive

studies 573–574

signal distance assessment 467–468

victory displays 115–117

visual displays and effects of androgens 486

see also songbirds; individual species

black-capped chickadee (Poecile atricapillus)

dawn chorus and male condition 141, 338

eavesdropping and transitive inference (TI)

574–576

female mate choice 140–141, 142

habitat change and song networks 143–144

habitat quality and song output 144

ranking process 577


632 Index

black-capped chickadee (cont.)

signal amplitude as a distance cue 468–469

use of social eavesdropping 21–22

black goby (Gobius niger)

male sex pheromone 549–558

semiochemical communication 558–559

black howler monkey (Alouatta pigra), factors

affecting levels of vigilance 419

black swan (Cygnus atratus), victory display

115–116

black-throated diver (Gavia immer), victory

display 116–117

blackbird (Turdus merula)

courtship interruption by neighbours 41

dawn chorus singing 337

male quality and dawn chorus singing 338

sound characteristics and attenuation

49–50

use of high perches 43, 44, 464–465

use of quiet song 50, 53–55, 56

blackcap (Sylvia atricapilla)

sound characteristics and attenuation 49

use of high perches 43, 44

blenny semiochemical communication

558–559

blue-throated humming bird (Lampornis

clemenciae), Lombard effect 464–465

blue tit (Parus caeruleus), dawn chorus singing

and male quality 338

blue whale (Balaenoptera musculus), distinctive

group calls 400

bonnet macaque see macaques

Boophis madagascariensis see Madagascar

treefrog

bottlenose dolphin (Tursiops truncatus)

active space of signals 392–394, 395

avoidance of feeding grounds with boat

noise 406–407

eavesdropping on other’s echolocation

clicks 18, 405–406

fission–fusion societies 400–402

individual signature whistle types 392–395,

400–402

numbers of animals in networks 397–399

recognition of others’ social relationships

591

silence when captured or near a boat

403–404

victory display 118

vocal learning and individual whistle

development 400–402

vocal matching to signal a specific

individual 402

boubou (Laniarius aethiopicus), victory display

116

Branta canadensis see Canada goose

Branta leucopsis see barnacle goose

Brienomyrus brachyistius see electric fish

Broadley’s painted reed frog (Hyperolius

marmoratus broadleyi)

female call preferences 286, 287–288

‘off-response’ call initiation 280–282

simultaneous mate choice in females

289

brown capuchin monkey (Cebus apella)

demands of social monitoring

32

predation levels and vigilance 417–418

time spent looking by subordinates

417–418

brown ghost (Apternotus leptorhynchus),

androgen effects on electric signals

487–488

brown-headed cowbird (Molothrus ater),

nestling begging 177

budgerigar (Melopsittacus undulatus)


masking release and binaural processing

462–463

signal-to-noise ratios for recognition and for

detection 463–464


bystanders

and redirected aggression 192–194, 200,

201–203

costs and benefits of information gathering

131–132

distinction from eavesdroppers 84–85


Index 633

effects on dominance hierarchies 97–99

influence of eavesdropping on behaviour

86–91, 93

influence of previous experience on

behaviour 88–89

postconflict attacks on losers 196–197

priming response and androgens 499–500,

501

social eavesdropping 10–11

taking advantage of ‘loser effects’ 198–200

see also eavesdroppers

Canis latrans see coyote

Canis lupus see wolf

Campbell’s monkey (Cercopithecus campbelli),

attention to alarm calls by another

species 373, 380

Canada goose (Branta canadensis), victory

display 115–116

canary (Serinus canarius), sound localization

ability 465–467

Capreolus capreolus see roe deer

Carassius auratus see goldfish

Carassius carassius see Crucian carp

cat, feral ( Felis catus), victory display 118–119,

121–122

Cebus apella see brown capuchin monkey

Cebus capucinus see white-faced capuchin

monkey

Cephalorhyncus hectori see Hector’s dolphin

Cercocebus albigena see mangabey

Cercopithecus aethiops see vervet monkey

Cercopithecus ascanius schmidtii see redtail

monkey

Cercopithecus campbelli see Campbell’s monkey

Cercopithecus diana see Diana monkey

Cervus elaphus see red deer

cetaceans, vocal matching to signal a specific

individual 402

chacma baboon (Papio cynocephalus ursinus)

attending to signal interactions 373–374

awareness of dominance relationships

199

time spent visually scanning 417–418

chaffinch (Fringilla coelebs), signal distance

assessment 467–468

chemical communication

correlation with predation 545–546

evolution through different functional

phases 541–542, 543

predation risk assessment 544–549

scent marking 131

signal propagation 543–544

see also fish semiochemicals; scent marking

chimpanzee (Pan troglodytes)

choice of targets for aggression 208


signalling 377

situations where calling is suppressed

374–375, 376–377

Chinese frog (Amolops tormotus) acoustic signal

repertoire 279

chorusing interactions

and female preference for leading signals

264

and signal competition 263–264

in anurans and insects 263–264

precedence effect 264

to avoid signal masking 264

see also anuran choruses

Cichlasoma nigrofasciatum see convict cichlid

cichlid see Mozambique tilapia

Cistothorus palustris see marsh wren

Clamator glandarius see great spotted cuckoo

Clarias gariepinus see African catfish

claw waving see fiddler crabs

cleaner wrasse (Labroides dimidiatus)

behaviour towards clients 534

cheating behaviour 525–526

client image scoring and wrasse behaviour

526–528, 529–530, 531

cognitive abilities used in tactical deception

535, 536

effects of bystander types on cheating

behaviour 529–530, 531

interactions with client reef fish 525–526

mutualism with client reef fish 521

possible endocrine-mediated response 535


634 Index

cleaner wrasse (cont.)

preferred food 534

response of biting cleaners to image scoring

clients 531–532

tactile stimulation 531–532

variable payoffs from different clients

533–534

variations in cooperative and cheating


cognition in animals, research history and

prejudices 568–570

cognitive aspects of communication networks

250–251

cognitive capacity

and cortical size 569

of birds 568–570

requirements for tactical deception 524–525

cognitive sciences, interface with

communication networks 447

common carp (Cyprinus carpio), pheromone

system 552

common dolphin (Delphinus delphis), individual

signature signals 400–402

common seal (Phoca vitulina)

avoidance of killer whale sounds 403

distinctive group calls 399–400

loud sexual advertisement calls 380

territorial behaviour 404–406

communication, dyadic view 2

and hormones 482–483, 494

limitations of 9

nestling begging 171–173

social context 78–79

communication networks 2, 9

androgen-modulated behaviour in 494–502

animal versus human 384–385

application to welfare of captive animals

assessing social intelligence 447

banded wren dawn chorus 337

chemical assessment of predation risk

544–549

cognitive requirements for participants

250–251, 447

concept of true individual recognition

363–366

context effects 1, 78–79, 129–132

cooperation and altruism 446, 536

defining properties 13

effects on signalling and receiving 180–185

evolution of spiteful behaviour 536

evolution of tactical deception 536

evolutionary process 558–559

game theory models 536

habitat alteration effects 143–144

hormones and communication 494–502

impact on signalling errors 184–185

implications for theoretical study of

signalling strategies 604

individual recognition mechanism 363–366

information cascades 270, 607–608

interfaces with other disciplines 1, 445–450

mathematical modelling 447–448

models of dominance hierarchy formation

620–621

models of eavesdropping by receivers

613–622

models of effects on signallers and

signalling dynamics 608–613

models of structure and organization

607–608

ostariophysan alarm system 546, 547–549

possible links with applied biology 448–450

receiver’s perception 445–446, 451–452, 474

semiochemicals 446–447, 540–541

size and extent 250

spatial distribution effects on acoustic

signals 474

structure influences signals and signalling

286–287

structure within a nest 179

use of nestling begging to study 179–180

see also modelling communication

contest behaviour see aggression; redirected

aggression; victory displays

contest behaviour in fishes 85

effects of eavesdropping 85–86, 94–95

environmental influences 95–97


Index 635

‘loser effect’ 88–89

opportunities to eavesdrop 86, 87

physical effects of observing fights 90–91,

93–94

potential costs 86

‘winner effect’ 88–89

convict cichlid (Cichlasoma nigrofasciatum),

eavesdropping 17

cooperation and altruism, in communication

networks 446, 521–525

cooperation theory 521

cotton-top tamarin (Saguinus oedipus)

ability to distinguish unfamiliar callers 374

courtship interactions, interruption by

eavesdroppers 40–41

‘social organization and vigilance 426

coyote (Canis latrans)

potential for redirected aggression 205,

206


376–377

victory display 118–119, 122

crabs (various species), similarities to fiddler

crabs 258

crickets (various species), victory displays 118,

122

critical (masking) ratio 459

Crocuta crocuta see hyaena

Crucian carp (Carassius carassius), pheromone

system 552

Ctenophorous fordii see Mallee dragon lizard

Cuculus canorus (European cuckoo), nestling

begging 177

Cuvier’s beaked whale (Ziphius cavirostris),

strandings and noise pollution 406–407

Cygnus atratus see black swan

cyprinid fishes

androgen effects on olfactory sensitivity

488–489

androgen effects on somatic releasers

491–492

pheromone systems 552

Cynops pyrrhogaster see Japanese red-bellied

newt

Cyprinus carpio see common carp system

552

Dama dama see fallow deer

Danio rerio see zebrafish

dark-eyed junco (Junco hyemalis), metabolic

effects of high androgen levels 504

dawn chorus

characteristics 320

communication network view 322

comparison with known daytime singing

interactions 324

directed song matching 338–339

environmental explanations for 320–321

female eavesdropping to assess males

321–322

hypotheses to explain 320–322

indicator of male condition 141

interactions in relation to functions 324–325

meaning of song matching and timing 307

mediating changes in social status 339

multi-way male interaction 335, 337–338

possible network structures 322–323

recording methods 324–325, 327–329

singing and male quality 338

social dynamics hypothesis 321, 339

structure in relation to functions 324

temporal patterns of singing behaviour 323,

329–336

see also banded wren

dear enemy effects, possible androgen

mediation 501–502

Delphinapterus leucas see beluga whale

Delphinus delphis see common dolphin

diademed sifaka (Propithecus diadema), scent

over-marking 357

Diana monkey (Cercopithecus diana), attention

to alarm calls of another species 373,

380

Dipodomys spp. see kangaroo rats

Docidocercus gigliotosi see katydid

dolphins

filtering of high–frequency signal

components 402–403


636 Index

dolphins (cont.)

use of directionality of clicks to target

signals 402–403

see also bottlenose dolphin

domestic fowl (Gallus domesticus)

audience effects on calls 66–67, 68

eavesdropping and dominance 28

dominance hierarchies

among nestlings 179

and social eavesdropping 28

eavesdropping effects on 97–99, 620–621

effects of prior experience (winner/loser

effects) 97–99

in non-human primates 588–589

modelling formation of 620–621

dominance interactions, simulation

modelling 606–607

dominance status, problems in poor habitats

144

dunnock (Prunella modularis)


quiet singing 53

dwarf mongoose (Helogale undulata), group

scent marking 363

dyadic view see communication, dyadic view

Eastern towhee (Pipilo erythrophthalmus), signal

distance assessment 467–468

eavesdroppers

alternative terms for 14

and mate choice 142–143

attending to outcomes of conflicts 198–200

attention to asymmetries in songbird vocal

interactions 313–314

awareness of dominance relationships

198–200

behavioural responses of those observed

523–524

costs and benefits 31–32, 39–40, 45–47, 211

distinction from bystanders 84–85

image scoring allows exploitation by cheats

532–533

image scoring of observed individuals

523

information from scent counter-marking

345, 346

interruption of courtship interactions

40–41

selection pressures caused by 30–31,

523–524

strategies for effective eavesdropping 42

use of song overlapping information 304,

305–306

see also bystanders, eavesdropping

eavesdropping

and altruism 522–523

and audience effects 66

and autocommunication 18

and dominance hierarchies 97–99

and predation risk 45

and secrecy 13–14

and transitive inference 574–576

as a type of bystander effect 499

at dawn chorus 338

bystander behaviour and social instability

96

comparison of primates with other animals

592–598

costs and benefits for signallers 30–31,

40–41, 48–49

countering with private signalling 52–53, 55

definitions of 3, 10, 13–15

effects on bystanders’ behaviour 85, 86–91,

93, 94–95

effects on female mate choice 100–103

effects on interactions 30–31

environmental influences on 95–97, 107

evidence for 39

evolution in semiochemical

communication 542, 559–560, 562

facilitation by interactants 51–52

factors influencing 97, 107–108

for song repertoire information 45–47, 76

identifying different types 28–29

image scoring and dishonest signals

524–525

image scoring in client reef fish 526–528,

529


Index 637

implications and future research 57–58

in anuran choruses 289

in communication networks 2–3

in non-primate mammals 591, 592

in relation to scent marking 235–237

indicators of dominance 52–53

individual differences 106–107

information not shared in animals 426

knowledge about other animals’ dominance

ranks 588–590

knowledge about transient social

relationships 590–591

knowledge of other animals’ kin 585–587,

588

models of how receivers use networks

613–622

ostariophysan alarm system 545, 547–549

physical effects of observing fights 90–91,

93–94

predation pressures 95–96

quiet song as response to risks 55

reliability and intimacy of information

416–417

risks for interactants 52

scent over-marking and mate choice 359,

360–361

signalling in different modalities 14, 28–29

social observation in animals 425

social structure and types of information

available 87, 583–584

sound transmission in natural habitats

42–43, 44

state dependency of bystander effects

95

strategies for effective eavesdropping

44–48

strategies for private signalling 52–53

to assess potential mates 141–142

to take advantage of ‘loser effects’ 198–200

true recognition or simple association

363–366

use of advertising to facilitate 48–49

use of high perches to improve reception

43, 44

used to assess fighting ability 85–86

using anonymity to counter 56–57

see also audiences; interceptive

eavesdropping; social eavesdropping;

mate copying

electric fish (Brienomyrus brachyistius),

androgen levels and dominance signals

502–505

electrocommunication signals, effects of

androgens 487–488, 489

elephant see African elephant

elephant seals (Mirounga spp.)

loud sexual advertisement calls 380

use of deep sound channel 397

Elephantodeta nobilis see Australian bushcricket

Eleutherodactylus coqui see Puerto Rican treefrog

emperor penguin (Aptenodytes forsteri),

amplitude modification in calls 462

endocrine response, importance of

individual’s perception of event 493–494

endocrine system

interface with communication networks

446

vertebrates 482

environment, influences on signals and

signalling 286–287 see also habitat

alteration

Erignathus berbatus see bearded seal

Erithacus rubecula see robin

Eschrichtius robustus see grey whale

Eubalaena glacialis see southern right whale

Eudyptula minor see little blue penguin

European cuckoo (Cuculus canorus), nestling

begging 177

European minnow (Phoxinus phoxinus), alarm

substance 546–548

European newts (Triturus spp.), chemical

communication and mate attraction

485–486

European starling (Sturnus vulgaris),

unmasking effect of sound segregation

460, 461

European treefrog (Hyla arborea), energy costs

of male calling 284–285


638 Index

fallow deer (Dama dama)

costs and benefits of groaning 379

discrimination of individual male callers

379–380

functions of groaning in males 379

individuality in formant frequencies in

vocalizations 382

fathead minnow (Pimephales promelas), active

space of alarm substance 546–548

Felis catus see cat, feral

female mate choice

assessment by eavesdropping 141–142

call preferences in anurans 264, 283–284,

285–286, 292

desirable male attributes 134

effects of observing interactions 78, 100–103

effects on male trait distribution 99–100,

101, 105–106

emergence of strategies 106

environmental influences 107

future work on 144–145, 146

genetic-based preferences 99–100,

101

in communication networks 129, 133–134

inferences from female movement patterns

135–136, 137

influence of differences in females 106–107

influence of particular signals 134

influence of predation risk 107

information from dawn chorus 141

instigation of male–male interactions

139–140, 141

male trait preference versus mate copying

103–105

measures of male quality 137–141

movement patterns infer assessment of

males 135–136, 137

preference for leading signals 264, 285–286

scent mark assessment in house mice

225–227

secondary mate choice 133, 135

simultaneous assessment 135, 136

song preferences, costs and benefits for

males 152–153

state-dependent influences 106–107

strategies for assessing males 135–136, 137

transmission distance and multiple signals

138–140

use of dawn chorus to assess males 321–322

use of network information 141–143

see also mate choice; mate copying

ferret (Mustela putorius), masking release and

binaural processing 462–463

Ficedula hypoleuca see pied flycatcher

fiddler crabs (Ocypodidae, Uca spp.)

biology 253–256, 257, 258

claw waving display 254–256, 261–262

communication networks 247–248,

252–253, 258

competition for burrow ownership 268–269

conspecific interceptive eavesdroppers

266–267, 268–271

costs and benefits of interceptive

eavesdropping 271–272

costs and benefits of signalling 271–272

courtship displays 254–256

detection distance for conspecifics

259–260

estimating density 258–259

female assessment of male quality 264–266,

267

gross signal timing among males 263

male response to rivals’ waving 268–271

neighbouring eavesdroppers 268, 269–271

primary and secondary receivers 266–271

range and functioning of visual system

259–260

reaction distances for conspecifics 260–261

signals other than claw waving 257

strategies for information gathering

266–271

strategies for signal competition 263–266

synchronous waving in males 264–266

wandering females as target receivers 266,

267

wandering males as eavesdroppers 268–269

waving rate as indication of male quality

264–266


Index 639

field cricket (Gryllus bimaculatus), audience

effects on males 72

fin whale (Balaenoptera physalus)

call response distances 392–395

loud sexual songs 380

use of multipath signal arrivals to locate

callers 397

fish

androgen-induced development of somatic

display structure 487

eavesdropping on visual interactions 24–27,

28

information transfer within shoals 84

vocal sounds 484

fish semiochemicals

ancestral phase 541–542, 543

correlation with predation 545–546

distinction between cues and signals

541–542, 543

eavesdropping 543

evolution into communication networks

541–542, 543, 558–559

evolution into spying 541–542, 543

evolution through different functional

phases 541–542, 543

fish taxa with specialized epidermal cells

548–549

fitness benefits for alarm signallers 545,

548–549

hormonal pheromones in information

networks 549–558

hypoxanthine N-oxide 540–541

occurrence of true communication

networks 545, 547–549

ostariophysan alarm substance 546–548

processing by fish olfactory system 543–544

signal propagation 543–544

spying and communication in information

networks 542, 545, 553, 558–560, 562

terminology 541–543

Fringilla coelebs see chaffinch

frogs, habitat influence on signalling 286–287

funnel-web spider (Agelenopsis aptera), victory

display 117–118, 122

Gallinago media see great snipe

Gallus domesticus see domestic fowl

game theory models 536, 606, 612–613,

614–618, 619, 620

Gasterosteus aculeatus see three-spined

stickleback

Gavia immer see black-throated diver

genetic relatedness and nestling begging

177–178

gerbil (Meriones unguiculatus), effects of

androgens on scent marking behaviour

485

ghost crabs (Ocypode spp.) 258

Globicephala melaena see pilot whale

Gobius niger see black goby

golden-collared manakins (Manacus vitellinus),

effects of androgens on visual display 486

golden hamster (Mesocricetus auratus)

androgen effects on central motivational

mechanisms 490

androgen effects on scent marking

behaviour 485

female counter-marking

female preference for top-scent males 360

flank marking 346–347

male preference for top-scent females 360

mechanisms to distinguish top and bottom

scents 347–357

over-marking and territory defence 358–359

persistence of scent marks 344

preferential memory for top scents 347–350

process of counter-marking 344–345

scent marking and social environment 362

targeted over-marking in males 358

true individual recognition 364–365

use of geometric relationships to determine

top scent 352, 356

vaginal marking 347

goldfish (Carassius auratus)

hormonal pheromones and spawning

549–551, 552

hormonal pheromones in an information

network 550–551, 552–553, 555

Gorilla gorilla beringei see mountain gorilla


640 Index

grasshopper (Ligurotettix coquilletti), female

attraction to larger choruses 283

great reed warbler (Acrocephalus arundinaceus),

female mate assessment 47–48

great snipe (Gallinago media), female

assessment of males 139

great spotted cuckoo (Clamator glandarius),

nestling begging 177

great tit (Parus major)

attention to asymmetries in vocal

interactions 314

calculation of maximum detection distance

454–456, 457, 458, 459

critical ratios 457

dawn chorus and male condition 141

eavesdropping and transitive inference

575–576

female mate choice and eavesdropping

141–142

information acquired from eavesdropping

19–21, 22–24, 583–584

signal reverberation as a distance cue

469–471, 472–473


use of unmodulated sounds 49

greater horseshoe bat (Rhinolophus

ferrumequinum), critical ratios 457

green frog (Rana clamitans), victory display 118,

122

green swordtail (Xiphophorus helleri)

eavesdropping on visual displays 25,

26–27

social interactions 84, 85

influence of eavesdropping on bystander


opportunities to eavesdrop 86, 87

potential costs of combat 86

green treefrog (Hyla cinerea), spatial

unmasking of signals 462–463

grey partridge (Perdix perdix), effects of

androgens on vocal structures 483–484

grey treefrog (Hyla versicolor)

energy costs of male calling 284

female call preferences 284, 288

grey whale (Eschrichtius robustus), reactions to

killer whale sounds 403–404

greylag goose (Anser anser)

kin-oriented redirected aggression 204

victory display 115–116

Gryllus bimaculatus see field cricket

gulf toadfish (Opsanus beta), acoustic signalling

99

guppy (Poecilia reticulata)

eavesdropping and mate choice 100–103

mate copying versus male trait preference

103–105

social interactions 84, 85

Gymnorhinus cyanocephalus see pinyon jay

Gymnotiformes, androgen effects on weak

electric signals 487–488

habitat alteration

and female mate assessment 144

effects on song transmission 143–144

see also environment

habitat quality, and dominance effects

144

hamadryas baboon (Papio hamadryas),


relationships 590

hamster see golden hamster

harbour porpoise (Phocoena phocoena)

avoidance of noise of human activity

406–407

use of clicks for communication 395

harbour seal see common seal

harp seal (Phoca groenlandica), distinctive


hartebeeste (Alcelaphus buselaphus), victory

display 118

Hector’s dolphin (Cephalorhyncus hectori), use of

clicks for communication 395

Helogale undulata see dwarf mongoose

Hemideina spp. see wetas

honeybee (Apis mellifera), communication and

social coordination 611–612

hooded warbler (Wilsonia citrina), female

assessment of males 141


Index 641

hormones and communication

adaptive value of social modulation 497,

502–505

dyadic view 482–483, 494

effects on cognitive functions 490–491

effects on communication 482–483

effects on learning and memory 490–491

effects on signal reception 488–490

expression of somatic releasers 491–492

modulation of central mechanisms

affecting motivation 490

modulation of effector pathways 483–488

network view 483, 494–495, 502

pheromones 549–558

reciprocal link with behaviour 481–482

role in control of behaviour 481–482

social modulation of androgen levels

492–494

house cricket (Acheta domesticus)

eavesdropping 17

male attraction to rivals’ signals 269

house mouse (Mus musculus domesticus)

effect of androgens on scent marking 485

female reproductive priming through scent

231–232

genetic sources of individual scent

signatures 232–234

information from age of scent marks

228–229

information in urine about owner 220–222

kin and group member recognition by scent

234–235

major urinary proteins present in urine

222–223, 233–234

male dominance structures 229–231

male territorial scent marking 221,

223–225, 227–228

MHC odour types and individual scent

signatures 232–233, 234–235

pheromones in urine 231–232

reproductive priming and the Bruce effect

232

scent mark assessment of males by females

225–227

scent mark assessment of males by other

males 227–228

scent mark detection and the vomeronasal

organ 219–220, 223, 229, 232

scent marking patterns 220–222, 228

scent marks of subordinate males 229–231

scent over-marking and mate choice 360

urine scent marking 220–237

volatile and non-volatile components of

urine 222–223, 228–229

see also mice

house sparrow (Passer domesticus), metabolic

effects of high androgen levels 504

human communication

dyadic view 416–417

network view 416–417

human eavesdropping 426–432

achieving privacy by behavioural means

427–428

and female curiosity 430–431

as a result of increasing privacy 428–429

caused by the need for privacy 428–430, 432

everyday occurrence of 427

historical evidence for 427, 430–432

honest signals and private behaviour

416–417, 429–430


lack of research on 426–427

male attempts to control others 431–432

privacy and intimate experience 430

social observation time costs in large

groups 428–429

stalking as a means of control 431–432

humans

altruism and indirect reciprocity 522–523

ancestral way of life 420

androgen effects on spatial memory

490–491

animal networks compared with human

networks 384–385

audience effects on behaviour 74–75

benefits from vigilance 420

common behaviour patterns with other

primates 433–435


642 Index

humans (cont.)

community benefits of vigilance 422–423

congenital adrenal hyperplasia 490–491

creating opportunities to be looked at

423–424

demands of social monitoring 31–32

domestication and freedom from external

vigilance 430

domestication and intimate behaviour 430

enhancing personal image with visual cues

423

equating human and non-human cognitive

studies 573–574

ethological studies of vigilance 424–425

factors affecting vigilance 435

individual benefits of vigilance 422

information gained from observation 422,

433–434

Lombard effect 464–465

male vigilance and control 434

parading in front of other people 423–424

privacy and self-awareness 429–430

redirected aggression 208–209

sex differences in social monitoring

424–425

sex differences in spatial memory 490–491

signal echo tail as a distance cue 469–471,

472–473

social comparisons 425–426

social control and vigilance 421–423, 424

stalking 424, 431–432

strength and features of female networks

434–435

suspicion of private behaviour in openly

living groups 420–422

transitive inference task 570–571

unifying model for vigilance, social

observation and eavesdropping 433–435

unmasking effects of amplitude-modulated

background noise 461–462

using vigilance to control 424


vigilance 420–425

visual cues in movements and gestures 423

humpback whale (Megaptera novaeangliae)

call response distances 392–395

distinctive group calls 399–400

loud sexual songs 380

possible social eavesdropping 404–406

hyaena (Crocuta crocuta)

ability to infer rank among other group

members 592

intelligible distance of loud calls 372

kin-oriented redirected aggression 204

persistence of scent marks 344

redirected aggression 191

redirected aggression and target diversion

201

Hydrurga leptonyx see leopard seal

Hyla arborea see European treefrog

Hyla cinerea see green treefrog

Hyla gratiosa see barking treefrog

Hyla microcephala see neotropical treefrog

Hyla versicolor see grey treefrog

Hyperolius marmoratus broadleyi see Broadley’s

painted reed frog

Hyperolius marmoratus marmoratus see South

African painted reed frog

Hyperoodon ampullatus see northern bottlenose

whale

image scoring

and tactical deception 531–532, 533–534

and evolution of altruistic behaviour

533–534

benefits for client reef fish 526–527, 528

cheats exploit eavesdroppers 532–533

evidence in client reef fish 527–528, 529

indirect reciprocity

and cheating behaviour 532–534

cognitive abilities involved 535

occurrence in social networks 534–535

information cascades 270, 607–608

information gathering see bystanders;

eavesdroppers

information networks

and sex pheromones 549–558

fish semiochemicals 540–541


Index 643

information transfer

hierarchies 18–19

within fish shoals 84

infrasound communication, in African

elephants 457–459

insect choruses 2, 263–264

female preference for leading signals 264

models of signalling dynamics 609, 610–611

insects

foraging strategies 611–612

self-organization among social insects

611–612

interceptive eavesdropping

among invertebrates 346

among marine mammals 403

compared with social eavesdropping 14–15

interspecific 16

intraspecific 17–18

ostariophysan alarm system 547–548

signaller payoff 15–16, 18


invertebrates, interceptive eavesdropping

346

Japanese macaque see macaques

Japanese medaka (Oryzias laticeps), mate

copying 103

Japanese red-bellied newt (Cynops pyrrhogaster),

androgens and pheromone production

485–486

Junco hyemalis see dark-eyed junco

kangaroo rats (Dipodomys spp.), endocrine

control of sandbathing in males 485–486

Kassina fusca see savannah running frog

Kassina kuvangensis see Kuvangu running frog

Kassina senegalensis see Senegal running frog

katydids (various species)

acoustic communication 152–153

activity patterns and predator avoidance

153–154

chorus signalling dynamics 610–611

cryptic signalling mode 164–165

lunar phase effects 158, 162–164

masking interference in acoustic

communication 157–162

predation of 154, 156–157

predator avoidance 154–157, 164–165

Kayapo people physical adornments 423

killer whale (Orcinus orca)

active space of signals 392–395

avoidance of human noise 406–407

call suppression near prey 376–377

distinctive group calls 400

filtering of high-frequency signal

components 403

importance of oldest female 384

mammal- and fish-eating groups 403

resident and transient groups 403

sound avoidance by potential prey 403


individual 402

king penguin (Aptenodytes patagonicus)

amplitude modifications in calls 462

signal-to-noise ratio for recognition

463–464

kingfishers (Coraciiformes), interspecific

aggression 191

kin-oriented redirected aggression 203–204

klipspringer (Oreotragus oreotragus), persistence

of scent marks 344

!Kung people, openly living groups 420–421,

427

Kuvanga running frog (Kassina kuvangensis)

adjustment of male call 282

call response types 282–283

Labroides dimidiatus see cleaner wrasse

Lagenorhynchus australis see Peale’s dolphin

Lagenorhynchus obscurus see Pacific white-sided

dolphin

Lampornis clemenciae see blue-throated

humming bird

Laniarius aethiopicus see boubou

Lemur catta see ring-tailed lemur

Leptonychotes weddelli see Weddell seal

leopard seal (Hydrurga leptonyx), distinctive



644 Index

Ligurotettix coquilletti see grasshopper

Linepithema humile see Argentine ant

lion (Panthera leo)

ability to distinguish unfamiliar callers

374

benefits and costs of loud calling 374–376

intelligible distance of loud calls 372

male suppression of loud calling 374–376,

376–377

little blue penguin (Eudyptula minor), victory

display 116, 121–122

lizards (various species)


metabolic effects of high androgen levels

504


longtailed macaque (Macaca fascicularis)

ability to distinguish social relationships

586–588



postconflict behaviour 194–204

time spent looking by subordinates

417–418

losers see also winner–loser effects

increased receipt of aggression 196–197

loser effects in victims 197–198

physiological changes from conflict

197–198

postconflict changes 195–196, 201–202

role of androgens 496–498, 499

serotonin-related behavioural inhibition

498–499

Luscinia megarhynchos see nightingale

Loxodonta africana see African elephant

Macaca fascicularis see longtailed macaque

Macaca fuscata see macaques

Macaca mulatta see rhesus macaque

Macaca nemestrina see macaques

Macaca radiata see macaques

Macaca thibetana see macaques

macaques (various species)

dominance relationships 194–195

kin-orientated redirected aggression

203–204, 588


relationships 586–588, 590–591



see also longtailed and rhesus macaques

Madagascar treefrog (Boophis madagascariensis),

acoustic signal repertoire 279

male parental behaviour, audience effects

72–74

male signal traits

and exposure to predators 152–153

and female preferences 152–153

male traits

distribution effects of mate copying

105–106

effects of female mate choice 99–100,

101

preference versus mate copying 103–105

male-male aggressive signalling, audience

effects 68–72

Mallee dragon lizard (Ctenophorous fordii),

social observation affects sexual

behaviour 425

mammals

acquiring and storing social knowledge

383–384

animal versus human networks 384–385

anti-predator calls 380–381

benefits of attending to others’ social calls

384

caller identity and status information in

acoustic signals 381

communication network view of loud

calling 373–374

contact call discrimination and memory

383–384

distinguishing unfamiliar callers 374

effects of androgens on scent marking

behaviour 485

filter characteristics of vocalizations 382

fluid social systems increase receivers of

loud calls 373


Index 645

high densities increase receivers of loud

calls 373

high encounter rates and long-distance

signalling 377–378

information availability in fluid social

systems 377–378

information availability in territorial

systems 374–377

intelligible distance of calls 372, 382–383

loud calls 249, 372, 374–376, 378–380

recognition of vocalizations from

conspecifics 377–378


374–377

scent marking in small terrestrial

mammals 249

source characteristics of vocalizations 382

source-filter theory of voice production

381–382


see also marine mammals

Manacus vitellinus see golden-collared manakins

mangabey (Cercocebus albigena), ability to

distinguish unfamiliar callers 374

marine environment and acoustic

communications 390–391

marine mammals

acoustic communication networks 249


caller identity information in signals

399–400

determining the distance of a caller

395–396, 397

disruption by human noise 406–407

directional high-frequency signal

components 402–403

eavesdropping 403–406, 408

features of communications 391

fission–fusion societies 397–399

group identity information in signals

399–400

maximum call detection distances 393

methods of restricting and directing signals

402–403

numbers of animals in networks 397–399

population density and network size

397–399

predator–prey interceptive eavesdropping

403

restricting range by selecting

high-frequency signals 403

restricting signal by decreasing source level

403

size of communication networks 391–394,

399, 408–409

sound propagation in the sea 392–394, 396,

397, 408

value of distant signals 407–408

vocal learning in cetaceans 400–402


individual 402

see also mammals

marsh wren (Cistothorus palustris), cognitive

abilities 569–570

masking interference 157–162

masking release

and binaural processing 462–463

and spatial separation of sound sources

462–463

in amplitude-modulated background noise

459–461, 462

mate choice see also female mate choice

and eavesdropping 141–143

and scent over-marking 359, 360–361

in communication networks 129

mate copying 100–103

costs and benefits for different females

106–107

effects on male trait distribution 105–106

emergence of strategies 106

environmental influences 107

evolutionary consequences 105–106

influence of predation risk 107

state-dependent influences 106–107

versus male trait preference 103–105

mathematical modelling, interface with

communication networks 447–448 see also

modelling communication


646 Index

meadow vole (Microtus pennsylvanicus)

effects of androgens on scent marking

behaviour 485

female preference for top-scent males 354,

355, 359

mechanism to distinguish top and bottom

scents 351–356

over-marking and territory defence

358–359

overlap of scent marks to determine top

scent 349, 352, 354–356

preferential memory for top scents 350

process of counter-marking 344–345

Megaptera novaeangliae see humpback whale

Mehicacu people, openly living groups

421

Melopsittacus undulatus see budgerigar

Melospiza melodia see song sparrow

Meriones unguiculatus see gerbil

Mesocricetus auratus see golden hamster

mice (Peromyscus spp.), studies of redirected

aggression 208–209

Micronycteris spp. see bats

Microtus montanus see montane vole

Microtus ochrogaster see prairie vole

Microtus pennsylvanicus see meadow vole

midwife toads (Alytes spp.)

evidence for eavesdropping 291

female acoustic responses 283

Miopithecus talapoin see talapoin 417–418

Mirounga spp. see elephant seals

modelling communication

comparing eavesdropping strategies

614–618, 619, 620

complex behaviour from simple rules

606–607

conceptualizing networks 605–608

cooperation strategies 614–615

development and assumptions 604–605

eavesdropping by receivers in networks

613–622

effects of female preferences on signalling

in choruses 611

effects of hubs in a network 607–608

effects of networks on signal and signalling

dynamics 608–613

emergence of ‘selfish herd’ organization

606–607

game theory and eavesdropping 614–618,

619, 620

game theory compared with simulations

612–613

game theory models of dyadic encounters

606

hawk–dove game 615, 618, 619

implications of communication networks

604

individually based spatially explicit

simulations 606–607, 612–613

information cascades in networks

607–608

mechanisms to control call timing in

choruses 611

network effects on linear dominance

hierarchy formation 620–621

network structure in existing models

605–607

new possibilities with network approach

622–623

signal dynamics in acoustic choruses

609–611

simulation modelling 605

small-world (scale-free) network analysis

607–608

swarm intelligence and self-organization in

social insects 611–612

territory establishment 612–613

Molothrus ater see brown-headed cowbird

montane vole (Microtus montanus), studies of

redirected aggression 208–209

Morymyriformes, androgen effects on weak

electric signals 487–488

mountain gorilla (Gorilla gorilla beringei)

absence of redirected aggression in females

208

postconflict attacks on losers

196–197

mouse see house mouse


Index 647

Mozambique tilapia (Oreochromis mossambicus)

androgen effects 487, 491–492, 499–500,

501, 502–505

effects of eavesdropping 27–28

endocrine response to social interaction

493–494

hormonal effects of aggressive priming

78


504

winner–loser effects 496–498, 499

Mus musculus see house mouse

Mustela putorius see ferret

Myotis lucifugus see brown bat

Naiken people, achieving privacy by

behavioural means 427–428

Nayaka people, openly living groups

421

Neoconocephalus spiza see katydids

Neogobius melanostomus see round goby

neotropical tree frog (Hyla microcephala)

female attraction to larger choruses 283

fine adjustment of male calls 282

male selective attention to neighbours

284

nestling begging

and genetic relatedness 177–178

as communication network 179–180,

185–186

costs and benefits to the brood 178

distinguishing signalling from physical

competition 174–175

dyadic communication approach 171–173

evolution of 171–172, 184

future work 185–186

heat loss and signalling behaviour 178–179

importance of signalling first 183

in interspecific brood parasites 177

influence of nestmates 173, 175–176, 178

locatability of calls 181–182

nestling signalling interactions 175–176

nestlings as a communication network 130,

170–171, 174–179, 180

parents’ responses to 172–173, 188

physical competition and dominance

hierarchies 179

reliability as a signal of need 172–173

signal costs and reliability 171–172

signalling errors 184–185

signalling to catch receiver attention

180–183

suppressing competitors’ signals 183

use of locatable signals 181–182

nestling gape 181

network view of communication 2

nightingale (Luscinia megarhynchos)

background noise and song output level

51

cognitive abilities 569–570

eavesdropping and transitive inference (TI)

574–576


male vocal interactions 302

solo and interactive singing 51

song matching 303–304, 307

use of social eavesdropping 20–21, 22

non-human primates

common behaviour patterns with humans

433–435

consortship behaviour 432–433

defensive and social functions of vigilance

417, 419–420, 425–426

factors affect levels of vigilance 419

male bias for vigilance outside the group

417–418

male vigilance and social control 419

proportion of time spent in vigilance

417–418

securing perceptual privacy for some


sentinel behaviour among high-ranking

males 418–419

time spent looking 417–418, 433

unifying model for vigilance, social

observation and eavesdropping 433–435

vigilance 417

see also primates


648 Index

northern bottlenose whale (Hyperoodon

ampullatus), use of deep sound channel

397

Nycticebus pygmaeus see pygmy loris

Ochotona princeps see pika

odontocetes, restricting range by selecting

high-frequency signals 403

oestrogens, receptors in the brain 490–491

olfactory system

hormonal modulation of sensitivity

488–489

signal detection and encoding 543–544

olive baboon (Papio anubis)

absence of redirected aggression in females

208

effects of redirected aggression 191–192

postconflict attacks on losers 196–197

Onchorhynchus mykiss see rainbow trout

Opsanus beta see gulf toadfish

Opsanus tau see oyster toadfish

Orcinus orca see killer whale

Oreochromis mossambicus see Mozambique

tilapia

Oreotragus oreotragus see klipspringer

Oryctolagus cuniculus see rabbit

Oryzias laticeps see Japanese medaka

oyster toadfish (Opsanus tau), effects of

androgens on vocal structures 484

Pacific humpback dolphin (Sousa chinensis),

individual signature signals 400–402

Pacific white-sided dolphin (Lagenorhynchus

obscurus), individual signature signals

400–402

Pan troglodytes see chimpanzee

Panthera leo see lion

paper wasp (Polistes dominulus), neural

development in colonial females

595–596

Papio anubis see olive baboon

Papio cynocephalus see baboon

Papio cynocephalus ursinus see chacma

baboon

Papio hamadryas see hamadryas baboon

parent birds

comparison of nestlings’ signals 183

information from behaviour of nestlings

174–175

parental care, effects of high androgen levels

in males 497, 504–505

Paroaria gularis see red-capped cardinal

parrots (Trichoglossus spp.), victory displays

116

Parus atricapillus see black-capped chickadee

Parus caeruleus see blue tit

Parus major see great tit

Passer domesticus see house sparrow

peacock blenny (Salaria pavo), androgen effects

on somatic releasers 491–492

Peale’s dolphin (Lagenorhynchus australis),

silence when captured or near a boat

403

Perdix perdix (grey partridge), effects of

androgens on vocal structures

483–484

Petromyzon marinus see sea lamprey

pheromones

fish semiochemicals 541–543

hormonal pheromones 549–558

releaser and primer effects 541

reproductive priming effects of mouse

urine 231–232

Phoca groenlandica see harp seal

Phoca vitulina see common seal

Phocoena phocoena see harbour porpoise

Phoxinus phoxinus see European minnow

Physalaemus pustulosus see tungara frog

Physeter macrocephalus see sperm whale

physics, interface with communication

networks 445–446

pied flycatcher (Ficedula hypoleuca)

female assessment of males 135–136,

137


pigtail macaque see macaques

pika (Ochotona princeps), ability to distinguish

unfamiliar callers 374


Index 649

pilot whale (Globicephala melaena), falling

silent when hunted 403

Pimephales promelas see fathead minnow

pinyon jay (Gymnorhinus cyanocephalus), ability

to rank multiple stimuli 596–598

Pipilo erythrophthalmus see Eastern towhee

plainfin midshipman (Porichthys notatus)

auditory sensitivity modulation by sex

steroids 489


484

Poecile atricapillus see black-capped chickadee

Poecilia latipinna see sailfin molly

Poecilia reticulata see guppy

Polistes dominulus see paper wasp

Porichthys notatus see plainfin midshipman

prairie vole (Microtus ochrogaster)

androgen effects on central motivational

mechanisms 490

studies of redirected aggression 208–209

precedence effect, female preference for

leading signals 285–286

predation risk

and eavesdropping 14, 16, 17, 45, 95–96

and katydid activity patterns 154–157

and katydid roost site selection

154–157

and male signal traits 152–153

and signal detection 159, 160

avoidance in communication networks

130

chemical assessment of 544–549

influence on female mate choice 107

influence on mate copying 107

interceptive eavesdropping among marine

mammals 403

mammal anti-predator calls 380–381

quiet song as response to 55

predators

information from prey alarm calls 380–381

use of locatable calls to find prey 181–182

Procolobus badius tephrosceles see red colobus

monkey

Propithecus diadema see diademed sifaka

primates

ability to classify objects based on abstract

concepts 594–595

ability to rank objects 590

awareness of social rank relationships 28,

588–591

benefits of associations with high-ranking

animals 426

complexities of recognizing affiliative

relationships 593–594

eavesdropping abilities versus that in other

animals 592–598

effects of social organization on vigilance

426

‘greater intelligence’ hypothesis for

eavesdropping abilities 592, 593–595, 598

interconnected groups with males as ‘hubs’

608

knowledge of other animals’ kin 585–587,

588

‘large social groups’ hypothesis for

eavesdropping abilities 592–593,

595–596, 598

monitoring social relationships as group

size increases 594

‘no species difference’ hypothesis about

eavesdropping abilities 593, 596–598

social resources 426

types of information acquired from

eavesdropping 584

see also non-human primates

priming response

and male aggression 75–78

androgen effects on bystanders 499–500,

501

private signalling, in male anurans 291

see also quiet song

pronghorn antelope (Antilocapra americana),

scent over-marking 361

Prunella collaris see alpine accentor

Prunella modularis see dunnock

Psittacus erithacus see African grey parrot

psychophysics, interface with communication

networks 445–446


650 Index

Puerto Rican treefrog (Eleutherodactylus coqui)

diphasic advertisement call 280

male selective attention to neighbours

284

models of chorus signalling dynamics

609–610

‘off response’ call initiation 280–282

Puntius schwanenfeldi see tinfoil barb

pygmy loris (Nycticebus pygmaeus), scent

over-marking and mate choice 360–361

pygmy Mulga monitor lizard (Varanus gilleni),

victory display 118

quiet song

as aid to anonymity 56–57

as response to predation risks 55

reasons for variability 56–57

used for private signalling 53–55

rabbit (Oryctolagus cuniculus), effects of


485

rainbow trout (Onchorhynchus mykiss)

eavesdropping and dominance 28

effects of redirected aggression 191–192

Rana clamitans see green frog

Rangifer tarandus see reindeer

rank-order fights see contest behaviour

rat (Rattus norvegicus), effects of redirected

aggression 191–192

receivers

auditory scene analysis 471–474

comparison of signals in a network 183–184

precedence effects 183

signals designed to catch attention 180–183

see also signal detection

reciprocal altruism, and behavioural


red-capped cardinal (Paroaria gularis),


red colobus monkey (Procolobus badius

tephrosceles)

time spent looking and group size 433

time spent looking 417–418

red deer (Cervus elaphus)

discrimination of individual male callers

379–380

individuality in formant frequencies in

vocalizations 382

loud mating calls 378–379

red-winged blackbird (Agelaius phoeniceus),

signal-to-noise ratios for recognition and

for detection 463–464

redirected aggression

and postconflict attacks on losers 196–197,

200

and reconciliation with aggressor 192–194,

200, 202–203

and the ‘fight or flight’ response 191–192

as an outlet for ‘frustration’ 191–192

as possible target diversion 201

as audience effect 202–203, 209–211

benefits for losers 200, 201–202


in non-human primates 192–194, 195

in species other than primates 204–205, 206

in winners 196

influence on bystanders’ behaviour

192–194, 200

intraspecific aggression 206, 208–209

kin-oriented 203–204, 588

‘loser effects’ 195–196, 198–200

possible use a signal 201–202, 204–205, 208

summary of functions 210–211

testing occurrence and function 193,

209–210

to attenuate endocrine stress response

191–192

to signal postconflict condition to

bystanders 201–202

redtail monkey (Cercopithecus ascanius schmidtii)

time spent looking 417–418, 433

redwing (Turdus iliacus), quiet singing 53

reindeer (Rangifer tarandus), androgen effects

on scent marking behaviour 485

relational distinction

abilities of non-primates 596–598

abilities of primates 594–595


Index 651

reptiles, victory displays 118

rhesus macaque (Macaca mulatta)


374

kin-biased redirected aggression 588

individuality of frequencies in vocalizations

382


see also macaques

Rhinolophus ferrumequinum see greater

horseshoe bat

ring-tailed lemur (Lemur catta), scent

over-marking 362–363

robin (Erithacus rubecula)


eavesdropping 17

female assessment of males 141

quiet singing 53

rodents, redirected aggression 208–209

roe deer (Capreolus capreolus)

barking call 380–381

individuality of vocal frequencies 382

round goby (Neogobius melanostomus),

semiochemical communication 558–559

rufous-collared sparrow (Zonotrichia capensis),

song changes with habitat 143–144

Saguinus oedipus see cotton-top tamarin

sailfin molly (Poecilia latipinna), mate copying

103, 104–105

Saimiri boliviensis see squirrel monkey

Sakalava people, openly living groups 421

Salaria pavo see peacock blenny

Salmo salar see Atlantic salmon

Samoan people, openly living groups 421

Sarakatsani people, privacy of the hut 428

savannah baboon, knowledge about transient

social relationships 590

savannah running frog (Kassina fusca)

female call preference 286

variation in call response types 282–283

Sceloporus jarrovi see lizards

scent marking

adjacent marking 344–345

amount of scent and top scent

discrimination 349, 353–354

and communication networks 131

and mate choice 359, 360–361

androgen effects 485

as a visual signal 362–363

as broadcast signals 235–237, 344

discriminating individual odours

350–351

functions of 345, 354, 355, 358–361

for network communication 219, 362–363


information available to eavesdroppers 345,

346

information in spatial and temporal

distributions 221, 223–225

in social contexts 220

mechanisms for targeted over-marking

357–358

mechanisms to distinguish top and bottom

scents 347–357

olfactory detection 219–220, 223

persistence of signals 219, 344

possible olfactory consequences 345

relative freshness and top scent

discrimination 351–353

reliability of signals 221, 223–225

to advertise competitive ability 221,

223–225, 227–228

to advertise territory ownership 221,

223–225, 227–228

use of geometric relationships to determine

top scent 352, 356

volatile and non-volatile components

219–220

sea lamprey (Petromyzon marinus)

larval pheromone attracts migrating adults

555–557

life cycle 555

male sex pheromone in communication

networks 557–558

possible specialization in signal production

and release 557–559

search for biological control 555


652 Index

selection pressures

female song preferences 152–153

from eavesdroppers 14, 16, 30–31

imposed by audiences 10, 64–65, 79

on interactive singing strategies 309

on primate brain size 28

on signalling interactions 18–19, 79

on victory displays 122–123

self-awareness in animals 577–578

semiochemicals

information from 446–447

see also fish semiochemicals

Senegal running frog (Kassina senegalensis),

synchronous calling 282

sergeant major fish (Abudefduf vaigiensis),

bystander effects on cleaner fish

behaviour 530–531

Serinus canarius see canary

serotonin

and aggressive behaviour 197–198

and behavioural inhibition in losers

498–499

sex steroids, modulation of sensory

perception 488–490

Siamese fighting fish (Betta splendens)

androgen effects on eavesdropping 503–504


eavesdropping on visual displays 24–27

female mate choice and eavesdropping 142

priming and male aggression 75–78

signal detection

amplitude as a distance cue 468–469

auditory scene analysis by receivers 471–474

biological background noise 152

calculation of maximum detection distance

454–456, 457, 458, 459

compared with signal recognition 463–464

critical (masking) ratio for various animals

453–457, 459

decision tree learning to discriminate

signals 158, 160–162

distance assessment 467–468, 470, 471,

472–473

female song preferences 152–153

high background noise 157–160

hormonal effects on 488–490

in a complex environment 152

perceptual mechanisms of receivers

451–452

precedence effect 183, 285–286

reverberation as a distance cue 469–471,

472–473

signal ranging 467–471, 472–473

signal-to-noise ratio 453–456, 457

simultaneous comparison by receiver 183

spatial separation of sound sources 462–463

use of ‘biological microphone’ 157–160

see also receivers

signal production

amplitude modification 462

anonymous signalling 56–57


broadcasting from high perches 464–465

catching receiver attention 180–183

competition and cooperation with other

signallers 63

competition in chorusing interactions

263–264

cryptic signalling 164–165

environmental temperature effects on

signallers 153

high output to overcome interference 181

impact of signal errors in communication

networks 184–185


predation risk 159, 160

sender adaptations to maximize

transmission 464–465

signal evolution and selection pressures

79

suppressing competitors’ signals 183

targeting a specific receiver 63–64

use of multimodal components 182

using locatable signals 181–182

signal transmission

background noise sources and levels 52–53,

452–453

environmental influences on 286–287


Index 653

frequency and amplitude masking effects

49–50, 453

influence of communication network

structure 286–287

information transfer hierarchies 29

in natural habitats 42–43, 44

masked auditory thresholds 457

masking release in amplitude-modulated

background noise 459–460, 461, 462

private signalling 32–53

signaller proximity and masking

interference 152

sound segregation unmasking effect

460–461

sound types 49–51

use of high perches 44, 50

using background noise fluctuations to

reduce masking 459–460, 461, 462

signal recognition, compared with signal

detection 463–464

signal-to-noise ratio

and signal detection 453–456, 457

determination 454–456

for detection and for recognition 463–464

signalling interactions

between nestlings 175–176

influence of shared fate of nestlings

178

information gathered from 29–30, 38–39

models of effects of networks on signalling

dynamics 608–613

selection pressures on 18–19


signalling modalities

and audience effects 10

acoustic signalling in 99

multimodal signalling 99, 182

use of different modalities 28–29

signals, distinguishing from physical

competition in nestlings 174–175

simulation models see modelling

communication

SINDSCAL multidimensional scaling analysis

469–471, 472–473

small-world (scale-free) network analysis

607–608

Smilisca sila see treefrog

social complexity, possible selection pressure

on brain development 595–596

social context

influence on eavesdropping effects 97

influence on hormone levels 481–482,

492–494

of communication 1

of communication networks 78–79

of dyadic interactions 78–79

social eavesdropping 10, 18–32

and dominance hierarchies 28

and victory displays 122–123

cognitive requirements in female anurans

289

compared with interceptive eavesdropping

14–15

form of bystander behaviour 10–11


in territorial songbirds 19–21, 24

in territorial systems 18–19

information gathered from 38–39

visual interactions in fish 24–27, 28

see also bystanders; eavesdropping

social instability, and bystander decisions 96

social intelligence, assessment of 447

social modulation of androgen levels 492–494

song matching

during and after dawn chorus 330,

332–335

value as directional signal at dawn chorus

338–339

song output, and habitat quality 144

song overlapping, as an aggressive signal 331,

334, 335–336

song repertoire size, and individual

identification 56–57

song sparrow (Melospiza melodia)


individual recognition of neighbours 364

song repertoire matching 303

victory display 116, 117, 121–122


654 Index

song transmission, effects of habitat on

143–144 see also sound transmission

songbirds eavesdropping

and female mate assessment 19–21, 22,

47–48, 314, 577

attention to interaction asymmetries

313–314

cognitive complexity of male ranking

judgements 576–577

courtship interruptions by eavesdroppers

40–41

for repertoire size information 45–47

male eavesdropping 313–314

use of quiet singing 53–55

songbirds vocal interactions

among neighbours 309–310, 311–312

among territorial males 301–302

and territorial behaviour 301–302, 312–313


androgen effects 483–485

asymmetries in vocal interactions 301, 309,

310–311, 313–314

cognitive processes in a communication

network 569–570, 574–576

evolution of vocal interaction strategies

314–315

function of the dawn chorus 248–249

functions of specific singing strategies 308,

309–310

in dialogues 302–303

in various contexts 301

interactive dimension of vocal interactions

301, 302–303

memory capacity 569–570

maintenance of territorial spacing pattern

311–312

precise timing and interpretation of

interactions 306

resident–intruder vocal interactions

309–311

selection pressures on singing strategies

309

self-awareness 577–578

site-specific dominance in neighbours

311–312

social contexts of vocal interactions

307–308, 310

solo versus interactive singing 51–52

song alternating 305, 306


song overlapping 304, 305–306

song production relative timing 304,

305–306

territorial settlement by multiple


territories as communication networks

248–249

sound localization

cues used for 465

variation in ability between species

465–467

sound transmission see signal transmission

source-filter theory of voice production

381–382

Sousa chinensis see Pacific humpback dolphin

South African painted reed frog (Hyperolius

marmoratus marmoratus), female call

preferences 287

southern right whale (Eubalaena glacialis),

possible social eavesdropping

404–406

spadefoot toad (Spea multiplicata), male

attraction to rivals’ signals 269

sparrowhawk (Accipiter nisus), sound

localization ability 465–467

spatial memory, sex differences and effects of

androgens 490–491

Spea multiplicata see spadefoot toad

sperm whale (Physeter macrocephalus)


distinctive group calls 400


signalling 377

individual signature signals 400–402

spiteful behaviour, evolution and

maintenance 534–535, 536

spotted dolphin (Stenella plagiodon), individual

signature signals 400–402

spotted hyaena see hyaena

spying see fish semiochemicals


Index 655

squirrel monkey (Saimiri boliviensis)

factors affect levels of vigilance 419

social organization and vigilance 426

Stenella plagiodon see spotted dolphin

Sternopygus macrurus see weakly electric fish

stress hormones, effects of social environment

482

Sturnus vulgaris see European starling

swordtail (Xiphophorus helleri)

androgen effects on sword development

491–492

audience effects on males 70

communication via private channels 79

Sylvia atricapilla see blackcap

Sylvia communis see whitethroat

Tachycineta bicolor see tree swallow

tactical deception

cognitive abilities required 524–525, 535

concept of theory of mind 524–525

evolution and maintenance 536

exploitation of eavesdroppers 524–525

image scoring and evolution of 533–534

occurrence in social networks 534–535

Taeniopygia guttata see zebra finch

talapoin (Miopithecus talapoin), time spent

looking by subordinates 417–418

terrestrial mammals see mammals

territorial behaviour

and eavesdropping 17–18


relative timing of song production 305–306

resident–intruder vocal interactions

309–311

scent marking in house mice 220–222,

223–225, 227–228


vocal interactions among neighbours

309–310, 311–312


territorial systems

and communication networks 143–144

availability of information from loud calls

374–377

dear enemy effects 501–502

modelling of territory establishment

612–613

social eavesdropping in 18–19

testosterone levels, increase in winners 198

three-spined stickleback (Gasterosteus aculeatus)

androgen effects on somatic releasers

491–492

visual perception modulation by sex

steroids 489–490

Thryothorus pleurostictus see banded wren

Tibetan macaque see macaques

tilapia see Mozambique tilapia

Tinbergen, Niko 1

tinfoil barb (Puntius schwanenfeldi), androgen

effects on olfactory sensitivity 488–489

toadfish, effects of androgens on vocal

structures 484

Tonatia silvicola see bats

transitive inference

cognitive abilities of birds 578–579

equating human and avian studies 573–574,

578–579

mechanisms used by animals 572

presentation to humans 570–571

presentation to non-humans 571–572

songbirds in communication networks

574–576

studies with parrots 572–573

tree shrew (Tupaia belangeri), effects of


485

tree swallow (Tachycineta bicolor), nestling

signalling errors 184–185

treefrog (Smilisca sila), synchronous calling

282, 283–284

tremulation, as cryptic signalling mode

164–165

Trichoglossus spp. see parrots

Triturus spp. see European newts

triumph ceremonies see victory displays

Troglodytes troglodytes see wren

tungara frog (Physalaemus pustulosus)

adaptations for acoustic competition 278

male selective attention to neighbours 284

Tupaia belangeri see tree shrew


656 Index

Turdus iliacus see redwing

Turdus merula see blackbird

Tursiops truncatus see bottlenose dolphin

Tyto alba see barn owl

urodeles, androgen effects on pheromone

production 485–486

Uca spp. see fiddler crabs

Varanus gilleni see pygmy Mulga monitor

lizard

vertebrate endocrine systems 482

vervet monkey (Cercopithecus aethiops)

bystander effects on aggression 202–203

factors affecting levels of vigilance

419


knowledge of other animals 585, 589–590


victory displays 11, 114

amphibians 118

androgen effects on vertebrates 486–487

arthropods 117–118, 122

birds 115–117

categories of 121–122

distinctive features of 120

distinguished by context 119–120

effects within the communication network

120–121, 122–123

functions within the winner–loser dyad

120–122

humans 118–119

mammals 118–119

occurrences of 119

reptiles 118

Vidua spp. see whydahs

vigilance

in animals 417–420

in humans 420–425

visual interactions in fish, eavesdropping on

24–27, 28

visual system, hormonal modulation

489–490

vocal learning in cetaceans 400–402

vocal matching, by cetaceans to signal a

specific individual 402

vomeronasal organ 219–220, 223, 229,

232

weakly electric fish (Sternopygus macrurus),

androgen effects on electroreception 489

Weddell seal (Leptonychotes weddelli)

colony-specific call types 399–400

ear damage from underwater noise 406–407


western scrub jay (Aphelocoma californica),

ability to rank multiple stimuli 596–598

wetas (Hemideina spp.), victory displays 118,

122

white-crowned sparrow (Zonotrichia leucophrys)



504

white-faced capuchin monkey

(Cebus capucinus)

male vigilance and social control 419

time spent looking as group size increases

433

white whale see beluga whale

whitethroat (Sylvia communis)


private signalling 55

signalling in different modalities 29

use of high perches to improve reception

42–43

whydahs (Vidua spp.), nestmate signal

suppression 183

Wilsonia citrina see hooded warbler

winner–loser effects

role of androgens 496–498, 499

serotonin-related behavioural inhibition in

losers 498–499

see also losers; winners

winners

increase in androgen levels 198

physiological changes following victory 198

postconflict aggression 196

‘winner effects’ following victory 198


Index 657

Wistar rat, effects of androgens on scent

marking behaviour 485

wolf (Canis lupus)


374

individuality in vocal frequency 382

scent marking 363


376–377

victory display 118–119

wren (Troglodytes troglodytes)

signal echo tail as distance cue 469–471,

472–473

use of high perches 43, 44, 464–465

use of low-frequency sounds 49

Xenopus laevis see African frog

Xiphophorus helleri see green swordtail;

swordtail

Yagua people, achieving privacy by

behavioural means 427–428

zebra finch (Taeniopygia guttata)


483–484


signal-to-noise ratios 463–464

signal amplitude as a distance cue 468–469


zebrafish (Danio rerio), active space of alarm

substance 546–548

Zinacantan people, suspicion of private

behaviour 421–422

Ziphius cavirostris see Cuvier’s beaked whale

Zonotrichia capensis see rufous-collared

sparrow

Zonotrichia leucophrys see white-crowned

sparrow

Animal Communication Networks

Documents

Transcript of Animal Communication Networks