PRIMARY RESEARCH PAPER

A comparative study of predator-induced phenotypein tadpoles across a pond permanency gradient

Alex Richter-Boix Æ Gustavo A. Llorente ÆAlbert Montori

Received: 20 April 2006 / Revised: 15 September 2006 / Accepted: 14 October 2006 / Published online: 26 February 2007� Springer Science+Business Media B.V. 2007

Abstract In a field survey the distribution of

pond-breeding anuran species and their potential

large predators was investigated along a freshwa-

ter habitat gradient, ranging from ephemeral pools

to permanent ponds. In a laboratory experiment

predator-induced plasticity was examined for all

tadpole species to test whether the plastic

response of ephemeral and temporary pond spe-

cies differs from that of permanent pond species.

Desiccation and predation pose conflicting de-

mands; reduced activity lowers the risk of death by

predation but increases the risk of death by

desiccation. It was expected that species from

time-constrained habitats would display a mor-

photype that would reduce vulnerability to inver-

tebrate predators, thus allowing these species to

maintain a high level of activity, whereas species

from permanent ponds would avoid predation

both morphologically and behaviourally. Species

distribution and predator composition along the

hydroperiod gradient differed. Variations be-

tween ephemeral and temporary ponds can be

attributed to hydroperiod differences and the

presence of large invertebrate predators in tem-

porary ponds, whereas the contrasts between

temporary and permanent ponds can only be

attributed to the hydroperiod, since the presence

and abundance of top predators are similar in both

habitat types. With the exception of bufonids, all

species showed predator-induced plasticity in

agreement with previous studies. Tadpole species

differed in the integration of the phenotypic traits

measured, but differences observed between

species could not be attributed only to habitat.

Species from temporary habitats showed an

expected response, with a low reduction of activity

in comparison with the rest of the species. The lack

of general patterns in the morphological changes

suggests that species within the same habitat type

did not converge on similar phenotypes, perhaps

due to functional constraints on differences in

microhabitat use in the water column.

Keywords Behavioural plasticity �Mediterranean area � Morphological

plasticity � Phenotypic integration � Predation

risk gradient � Spatial distribution � Temporary

pond � Trade-offs

Introduction

Models of community structure in lentic systems

suggest that species composition is determined by

Handling editor: K. Martens

A. Richter-Boix (&) � G. A. Llorente �A. MontoriDepartament Biologia Animal, Facultat de Biologia,Universitat de Barcelona, Av. Diagonal 645, 08028Barcelona, Spaine-mail: arichterboix@ub.edu

123

Hydrobiologia (2007) 583:43–56

DOI 10.1007/s10750-006-0475-7

a trade-off between pond permanency and pre-

dation risk (Schneider & Frost, 1996; Wellborn

et al., 1996). Changes in predators are related to

pond permanency (Woodward, 1983; Werner &

McPeek, 1994; Gunzburger & Travis, 2004). This

model is particularly applicable to amphibian

assemblages. These two antagonistic gradients

create different selective forces for tadpole spe-

cies in ponds from different positions along the

gradient (Wellborn et al., 1996; Richardson,

2001a, b). Resource acquisition in tadpoles may

be affected by both factors: the need for timely

development in temporary ponds and the need to

avoid predators along the hydroperiod gradient.

Activity levels and foraging rates are typically

positively correlated, thus increased activity leads

to an increased predation risk (Skelly, 1994;

Anholt & Werner, 1995; Eklov & Halvarsson,

2000). Activity level is therefore expected to

reflect the differences in habitat type along the

pond permanency gradient (Woodward, 1983;

Skelly, 1996; Wellborn et al., 1996; Richardson,

2001b).

Several mesocosm and laboratory studies have

also demonstrated that tadpole species have

evolved the ability to asses predation risk and

to react in a way that reduces the likelihood of

being preyed upon (Skelly & Werner, 1990;

Chovanec, 1992; Anholt et al., 1996; McCollum

& Leimberger, 1997; Lardner, 2000; Van Buskirk,

2002; Relyea, 2004). Typical phenotypic plasticity

in the presence of predators includes a reduction

in time spent foraging and changes in morphol-

ogy which make individuals less vulnerable to

predators. All of these changes affect growth and

development, causing individuals to grow and

develop more slowly when exposed to predators

than in the absence of predators (Lardner, 2000;

Relyea, 2002; Van Buskirk, 2002). However,

species that persist in temporary ponds com-

monly face time constraints that limit their ability

to delay growth and development (Altwegg,

2002).

Due to the behavioural trade-offs that exist

along the pond permanence gradient, it can be

expected that levels of phenotypic plasticity in

response to predation pressure––like activity

levels––will differ among species evolving under

different selective regimes, with varying integra-

tion of the different traits (behavioural and

morphological) to fine-tune the plastic phenotype

response depending on species ecology and his-

tory (Relyea, 2004). Species from ephemeral and

temporary ponds can be expected to exhibit a

predominantly morphological, predator-induced

plasticity in order to allow the species to maintain

high levels of activity and to develop more rapidly

whilst also reducing predation risk (Relyea &

Werner, 1999; Anholt et al., 2000). In most

permanent ponds, morphological changes in spe-

cies can be accompanied by behavioural changes

in order to reduce encounters with and detection

by predators (Chovanec, 1992; Anholt et al.,

2000).

To date, only three studies have tested the

distribution and phenotypic plasticity of more

than two closely related species (Relyea &

Werner, 2000; Richardson, 2001b; Van Buskirk,

2002). The study made by Relyea & Werner

(2000) focuses only on the morphological changes

of four species along a predation risk gradient and

does not incorporate measures of behavioural

activity. In contrast, Richardson (2001b) studies

only activity levels and their plasticity without

considering morphological plasticity, whereas

Van Buskirk (2002) studies both plastic traits

but from an adaptive point of view of phenotypic

plasticity that gives no consideration to the trade-

offs that species face along the pond permanency

gradient. None of these studies address how

behavioural and morphological traits are inte-

grated along the hydroperiod gradient. The pres-

ent study evaluates the links between activity

behaviour, predator phenotypic plasticity and

habitat use of species along the pond permanency

gradient and, ultimately, explains distribution

patterns. Consequently, the distributions of six

tadpole species are compared in a field survey

with regard to what are commonly considered the

two principle sources of mortality: pond perma-

nency and predation risk. This is followed by a

laboratory experimental procedure to assess

predator-induced plasticity (morphology and

activity level) and its effects on development.

44 Hydrobiologia (2007) 583:43–56

123

Materials and methods

Site

The field study was carried out in protected

natural areas near Barcelona (NE Iberian Penin-

sula), including the Natural Park of Garraf and

the Metropolitan Park of Collserola. The climate

in this zone is Mediterranean, with hot, dry

summers, mild winters and two rainy periods,

one in spring and the other in autumn. The

amount of precipitation varies considerably from

year to year in this region (see Richter-Boix et al.,

2006b for a well description and localization of

the areas of study). The amphibian community in

this zone comprises seven anuran species (Alytes

obstetricans (Laurenti 1768), Pelobates cultripes

(Cuvier 1829), Pelodytes punctatus (Daudin

1802), Bufo calamita (Laurenti 1768), Bufo bufo

(Linne 1758), Hyla meridionalis (Boettger 1874)

and Rana perezi (Seoane 1885)) and Salamandra

salamandra (Linne 1758). We focused our study

on the anuran community but excluded Pelobates

because it is a rare species in the area and few

data are available.

Field sampling methods

We evaluated amphibian larvae and their poten-

tial invertebrate predators in four sampling peri-

ods during the spring and summer, from March to

August of 2002 in a total of 193 isolate ponds.

These localities span the range of aquatic habitats

in which the species studied breed in the area of

study, including ephemeral pools, temporary and

permanent ponds. Sampling time periods were

dictated by preliminary sampling and accounted

for temporal differences in breeding activity

among species (Richter-Boix et al., 2006b) and

ensured that all species breeding were captured.

Due to variation in hydroperiod not all sites could

be surveyed in all sampling periods, thus sample

sizes of ponds were not uniform. Both, amphibian

larvae and predaceous invertebrates were sam-

pled with dip-net sweeps (30 cm · 40 cm) to

obtain relative species densities. This is a stan-

dardized technique used to sample both groups

(e.g. Schaffer et al., 1994; Babbitt et al., 2003). A

minimum of 5–10 dip-net sweeps were taken in

each possible tadpole microhabitat following

standard techniques determined by pond size

(Schaffer et al., 1994). All tadpoles were identi-

fied in the field and photographed with a grid

background; the number of individuals of each

species were counted and finally returned to

water. Predaceous invertebrates shown in previ-

ous studies (e.g. Woodward, 1983; Travis et al.,

1985; Cronin & Travis, 1986) to prey on tadpoles

were identified, counted and photographed with a

grid background. Three types of insects were

considered as potential predators: Odonata larvae

(considering as predators aeshnid and libellulid

naiads), Heteroptera (notonectids and Nepa spp.)

and Coleoptera (larvae and adults diving beetles).

Fish presence was determined through visual

surveys in addition to dip-net captures. Indepen-

dently, for tadpole surveys, we visited ponds

every month throughout the year to establish the

data of drying and determine the position of the

pond across the hydroperiod gradient. In this

manner, we were able to evaluate the number of

days (in 30 · 30 days) that ponds retain water.

We defined hydroperiods in terms of three

categories: (1) ephemeral pools, (2) temporary

ponds and (3) permanent ponds. Ephemeral pools

(ponds containing surface water for a maximum

of two months) refilled after each rainfall period.

Temporary ponds were flooded by spring and

autumn rainfall. The shallowest temporary ponds

often dried out in winter, whereas the deepest

ponds held water until the end of spring or early

summer. It is important to consider that a pond’s

position on the gradient of the hydroperiod could

be dynamic, and especially that some temporary

ponds change position along this gradient annu-

ally as a result of climatic conditions. We did not

include a category for permanent ponds with or

without fish as in previous studies (Babbitt et al.,

2003; Stocks & McPeek, 2003; Van Buskirk,

2003), because in the study area only 6 of the

193 ponds contained fish, and these were not

included in the study, leaving 187 localities for

analysis.

Total counts for each amphibian species and

predaceous invertebrate captured in each pond

were divided by the number of dip-net sweeps

taken in each pond (Babbitt et al., 2003). This

yielded an abundance based on catch per unit

Hydrobiologia (2007) 583:43–56 45

123

effort which could be compared across the three

different habitats considered. Body lengths of

predaceous invertebrates were measured from

pictures on a computer and divided into two

groups: (1) small invertebrates from 5 to 15 mm

long, and (2) large invertebrate predators with a

body length of over 15 mm. We conducted

analyses of variance ANOVAs to determine if

amphibian species and predator group abun-

dances and relative densities varied among hy-

droperiod categories. Ponds were considered as

the units in all analyses and variables log-trans-

formed before analyses. Predation risk at which

each amphibian species was exposed was com-

pared with analysis of variance. Also, larval

amphibian species abundance was correlated with

invertebrate predator abundance, and a partial

set correlation analysis was performed to remove

the effect of hydroperiod.

Experimental procedure: effect of predation

risk on larval morphology and activity

Morphological and behavioural anti-predator

responses of the six species to a common predator

were tested in a short-term laboratory experi-

ment. Clutches of the six species were collected

from natural ponds. In all cases egg masses were

taken during species amplexus period to reduce

the time of exposition of embryos to predators to

avoid phenotypic plasticity induction on hatch-

ings. For Pelodytes egg samples were taken from 5

egg masses and hatched separately. For Hyla we

recollected 15 egg masses from three different

ponds. In the case of bufonids, fragments of three

different egg strings were taken for each species

(B. bufo and B. calamita). In the case of Rana,

tadpoles were obtained from 5 clutches from two

different ponds. The Alytes larvae had recently

hatched and just been deposited in the pond by

males when we recollected it. As males of

Alytes have parental care of eggs, negligible

predation on hatchings should take place during

the egg period and no defences should have been

developed.

The experiment was conducted in laboratory

tanks (30 l) filled with dechlorinated water under

approximately the same conditions as the natural

photoperiod and at a temperature of around

22�C, during the spring of 2001. Thirty tadpoles

were randomly drawn from a mixture of all

clutches and transferred from hatching aquariums

to experimental tanks when they reached Gosner

25 life stage. During the entire experiment tad-

poles were fed rabbit pellets ad libitum. Two

transparent cylindrical predator cages were

placed on both sides of each tank to prevent

predators capturing tadpoles but permitting the

flow of chemical signals. Both treatments (pred-

ator presence and absence of predator) were

replicated several times (6 for Alytes, 10 for

Pelodytes, 10 for Hyla, 7 for B. calamita, 6 for

B. bufo and 6 for Rana). In the predator presence

treatment aeshnid odonate naiads larvae were

used as predators. Aeshnid larvae are known to

feed on tadpoles and trigger defences. Each

predator was fed every day with tadpoles reared

in separate containers.

After 2 weeks from the start of the experiment

tadpole activity was measured. We sampled

activity behaviour by counting the number of

tadpoles moving in each tub the instant the tub

was first viewed (Skelly, 1995). Each tub was

observed every half hour for a total of 30

replicates during 2 weeks. This protocol was

repeated for all six species.

In order to obtain data on tadpole morphology,

4 weeks after the start of the experiment all the

tadpoles were individually photographed with a

grid background. We noted Gosner developmen-

tal stage and measured six traits which had been

taken as indications of plasticity in previous

studies (Van Buskirk & Relyea, 1998; Van

Buskirk, 2002): body length, body depth, tail fin

length, tail fin depth, tail muscle length and tail

muscle depth.

Experiment statistical analyses

To measure the behavioural response of tadpoles

in the presence of a predator the mean proportion

of active tadpoles per tub was calculated. These

proportional data were arc-sine square root

transformed previously to test the hypothesis that

the mean number of active tadpoles would differ

between treatments (presence or absence of

predator) using ANOVA analyses. Developmen-

tal stage data were log transformed, and differ-

46 Hydrobiologia (2007) 583:43–56

123

ences between treatments within species were

analysed with ANOVA analyses.

Before morphological plasticity response anal-

yses, tadpole measurements were corrected for

variation in body size. To generate size-corrected

measures, we used the residuals of the morpho-

logical measurements of log-transformed traits

after regressions against body size. We used

centroid size, obtained from landmarks, as a

measure of body size (Loy et al., 1993). Coordi-

nates of these landmarks were collected using the

TPSDIG computer program, version 1.30 (Rohlf,

2001). The centroid size, the square root of the

sum of squared distances of a set of landmarks

from their centroid (Bookstein, 1991), was calcu-

lated for each specimen and used to represent

size. After performing this correction, tadpole

morphology for each species in the two treat-

ments was tested, first with multivariate analysis

for all traits together and second with a univariate

analysis for each variable.

We measured differences in plasticity among

species (both activity and morphological traits)

and their effect on development by examining

changes in traits that occurred between treat-

ments divided by the mean value of the trait in

the absence of predator treatment ([presence of

predator – absence of predator]/absence of pred-

ator) (Van Buskirk, 2002). In the presence of

predator, positive values of plasticity reflect an

increase in the value of the trait, whereas negative

values show a decrease in this trait. The magni-

tude of plasticity in traits can be compared among

species because they are all represented in unit-

less measures of proportional change. We calcu-

lated a overall proportional change in

morphology as the mean of the absolute values

of proportional changes in the six size-corrected

traits considered (Van Buskirk, 2002). If the traits

measured affect species performance within a

habitat type (temporality and predation risk),

then species from distinct habitats should differ in

trait values. Therefore, we first tested for differ-

ences among species regardless of the kind of

habitat their occupied. We used univariate and

multivariate analyses on activity, developmental

stage and on morphological traits to determine

whether phenotypes of the species were affected

by the presence of predator. To consider the

effects of habitat we ran analyses of variance

using habitat as a fixed factor and species nested

within habitat as a random factor. Species were

nested within habitats following field data. As we

expected that the plasticity response would be in

function of habitat temporality, we considered

two groups of species: temporary ponds breeders

(species which preferably occupied ephemeral

and temporary ponds) and permanent pond

breeders (species from more predictable season-

ally and permanent water).

As species cannot be considered independent

data points because trait values could be influ-

enced by shared common ancestry (Felsenstein,

1985; Richardson, 2001a), we tested whether the

distribution of a particular species in terms of

phenotypic plasticity is correlated with its phy-

logeny. We compare proportional change in

activity for each species and the overall propor-

tional change in morphology depending on the

phylogenetic topology of the six species studied.

The phylogenetic relationships between the six

species were reconstructed using the combined

data set of three genes: 12S, 16S and cyt b.

Sequences were obtained from specimens in a

personal collection (collected and sequenced by

S. Carranza) and from the GenBank database. All

sequences were compiled, aligned and refined

manually using Sequence Navigator. Observed

distances in pairwise comparison were obtained

using PAUP software. We tested phylogenetic

independence of larval traits with the computer

program ‘‘Phylogenetic Independence 2.0’’ (Re-

eve and Abouheif, 2003). Tests For Serial Inde-

pendence (TFSI) on continuous data were

performed using the phylogenetic topology and

node distances obtained from molecular recon-

struction. Topology was randomly rotated

10,000 times to build the null hypothesis.

Results

Quantification of species abundances and

predation risk along the gradient

We surveyed 49 ephemeral pools, 85 temporary

ponds and 53 permanent ponds, with a total of

24,938 tadpoles counted during field sampling.

Hydrobiologia (2007) 583:43–56 47

123

Species richness ranged from 0 to 6 species per

locality, without significant statistical differences

in the three habitats considered (F2, 184 = 2.537;

P = 0.0818; Fig. 1a). Total tadpole abundance did

not vary significantly among hydroperiod catego-

ries (F2, 184 = 1.227; P = 0.2954; Fig. 1a). Relative

abundances of larval amphibians, predators taxo-

nomically groups and the predator size categories

varied among hydroperiod categories (Figs. 2a, b,

c). Abundances of most amphibian species were

determined by the hydroperiod category, showing

a segregation of species along the hydroperiod

gradient. Pelodytes and B. calamita were signifi-

cantly related to ephemeral and temporary ponds

(F2, 184 = 4.69; P = 0.0102 and F2, 184 = 12.42;

P < 0.001 respectively); Alytes, Hyla and B. bufo

were more abundant in temporary and permanent

ponds (F2, 184 = 12.44; P < 0.001; F2, 184 = 9.81;

P < 0.001 and F2, 184 = 4.35; P = 0.014 respec-

tively); and Rana mainly occupied permanent

ponds (F2, 184 = 34.10; P < 0.001). Based on these

preferences we classified B. calamita and Pelo-

dytes as typical temporary species (Habi-

tat = Temporary), and Alytes, B. bufo, Hyla and

Rana as permanent species (Habitat = Perma-

nent) to conduct nested analyses.

The groups of invertebrate predators also

differed along the gradient. Only 8 of the 49

ephemeral ponds (16.3%) were predator-free,

whereas none of the temporary and permanent

ponds were free of predators. Heteroptera and

small predators were more abundant in tempo-

rary ponds (F2, 184 = 56.48; P < 0.001 and

F2, 184 = 13.02; P < 0.001 respectively), whereas

the abundance of Odonata and large predators

increased along the hydroperiod gradient, with

higher values in permanent ponds (F2, 184 = 53.19;

P < 0.001 and F2, 184 = 34.10; P < 0.001 respec-

tively) (Fig. 2b). No differences between habitats

was found for Coleoptera (F2, 184 = 2.44;

P = 0.089). Abundance of predators at which

each amphibian species or species grouped by

habitat (temporary vs. permanent) was exposed

in field varies (F5, 369 = 4.14; P = 0.001 and

F1, 369 = 18.11; P < 0.001 respectively), but not

for species nested within habitat (F4, 369 = 0.06;

P = 0.505). Species from permanent habitats were

exposed to more predators than species from

temporary habitats (Fig. 1b).

Larval amphibian species abundance was cor-

related with invertebrate predator abundance.

Pelodytes and B. calamita were negatively corre-

lated with invertebrate predators (r = –0.168;

P = 0.021; N = 187 and r = –0.271; P < 0.001;

N = 187 respectively), whereas Hyla and Rana

were positively correlated with large predators

(r = 0.188; P = 0.01; N = 187 and r = 0.229;

P = 0.002; N = 187 respectively). Partial set cor-

relation analysis don’t showed any correlation

between tadpole abundance and predator abun-

dance after removing the effect of hydroperiod

(Pelodytes [b = –0.019, T = –0.275, P = 0.783],

Fig. 1 (a) Larval amphibian abundances using catch perunit effort (black boxes) and larval amphibian speciesrichness (white boxes) in the three hydroperiods consid-ered. (b) Abundance of potentially predators at ponds inwhich tadpoles of the different amphibian species were

found. Mean and standard error were represented.Ao = Alytes obstetricans, Pp = Pelodytes punctatus,Bb = Bufo bufo, Bc = Bufo calamita, Hm = Hyla merid-ionalis, Rp = Rana perezi

48 Hydrobiologia (2007) 583:43–56

123

B. calamita [b = –0.044, T = –0.643, P = 0.521],

Hyla [b = 0.141, T = 1.892, P = 0.060] and Rana

[b = –0.015, T = –0.244, P = 0.808]. These results

suggest that pond temporality and their quantity

and quality of resources affect both: tadpole

species and predator guild composition.

Effects of predation risk on larval activity,

development and morphology

Species differ in their activity levels in the

absence of predator (Table 2). Species from

temporary habitats showed higher values of

activity than species from the most permanent

extreme of the hydroperiod gradient (Table 2,

Fig. 3d). In the presence of the predator activity

decreased in all species with the exception of the

two bufonids (Table 1). However, reduction of

activity was higher in permanent species than in

temporary species (Table 2, Fig. 3d).

Species showed morphological plasticity to the

presence of the predator for several of the traits

measured with the exception of the two bufonids

(Table 1). Major morphological changes were

detected in tail fin and body traits (Fig. 3a–c).

The experiment illustrated numerous phenotypic

(a)

(b)

(c)

Rel

ativ

e ab

un

dan

ce o

f t

adp

ole

sp

ecie

s (%

)R

elat

ive

abu

nd

ance

of

inve

rteb

rate

pre

dat

ors

(%

)

Ephemeral pools(49 ponds)

Temporary ponds(85 ponds)

Permanent pond(53 ponds)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Ao

Pp Bb

Bc

Hm

Rp

PpAo

Bb

Hm

Rp

PpBc

Hm Rp

AoBb

Bc

Het

Het

Het

Col >15 mm

Col Col

Odo

Odo

Odo

<15 mm

<15 mm

<15 mm

>15 mm

>15 mm

Ao Pp Bb Bc Hm Rp Het Col Odo0.0

0.2

0.4

0.6

0.8

1.0E T P

Rel

ativ

e ab

un

dan

ce (

%)

Tadpole species Predators

< 15 mm > 15mm

Fig. 2 (a) Relative abundance of tadpole species and (b)relative abundance of invertebrate predators in a taxo-nomically subdivision and grouped by size for eachhydroperiod category considered. In (a) and (b) meanand standard error were represented. (c) Relative abun-dance of amphibian species and invertebrate predators in

ephemeral (E), temporary (T) and permanent ponds (P).Ao = Alytes obstetricans, Pp = Pelodytes punctatus,Bb = Bufo bufo, Bc = Bufo calamita, Hm = Hyla merid-ionalis, Rp = Rana perezi. Het = Heteroptera,Col = Coleoptera, Odo = Odonata, <15 mm = small pre-dators, >15 mm = large predators

Hydrobiologia (2007) 583:43–56 49

123

differences among species and for species nested

within the appropriate habitat affiliation (Ta-

ble 2), with higher values of morphological

changes for temporary habitat species. Species

were highly variable in predator-induced mor-

phological phenotypic plasticity but with a com-

mon response in activity, which varied in

magnitude depending on the species and habitat

occupied by species.

Also, developmental rate differed between

treatments within species with the exception of

Pelodytes (Table 1). Species and the two groups

considered in function of habitat differed in

proportional developmental stage (Table 2). Spe-

cies from permanent habitats showed a great

reduction of development whereas development

of species from temporary ponds remains unal-

tered. Proportional developmental stage change

was correlated with proportional change in activ-

ity (r = 0.835; P = 0.039; N = 6; Fig. 4b), but not

with overall proportional morphological change

(r = –0.327; P = 0.526; N = 6; Fig. 4a).

The TFSI showed no significant phylogenetic

autocorrelation among species for the morphol-

ogy data set (C-statistics = 0.0930; P = 0.2820)

(Fig. 4c), or for proportional change in activity

(a) (b)

(d)(c)

Bc

PpBb

HmAo

Rp

Activity in absence of predator (%)

Act

ivity

in p

rese

nce

of p

reda

tor

(%)

Pp

Bb

Rp

Hm

Ao

Bc

0.1

0.08

0.06

0.04

0.02

0

-0.02-0.01 0 0.01 0.02 0.03 0.04

Proportional change in body length

Pro

port

iona

l cha

nge

in b

ody

dept

hPp

Rp

Bb

Bc

Ao

Hm

-0.06 -0.04 -0.02 0 0.02 0.04

0.04

0.03

0.02

0.01

0

-0.01

-0.02

0.05

Pp

RpAoHm

Bc

Bb

-0.01 0 0.01 0.02 0.03 0.04

0.1

0.08

0.06

0.04

0.02

0

-0.02

Proportional change in tail fin length

Pro

port

iona

l cha

nge

in ta

il fin

dep

th

Proportional change in tail muscle length

Pro

port

iona

l cha

nge

in ta

il m

uscl

e de

pth

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

Fig. 3 Predator-induced plasticity in the body (a), tail fin(b) and tail muscle (c) shape of amphibian species. Figureshows changes in shape due to the presence of predator(mean and standard error). Dashed lines indicate the casein which there was no plasticity. Fig. (d) shows proportionof time active in the presence and absence of predator.

Species near the dashed line showed little behaviouralresponse to predator. Ao = Alytes obstetricans, Pp = Pe-lodytes punctatus, Bb = Bufo bufo, Bc = Bufo calamita,Hm = Hyla meridionalis, Rp = Rana perezi. Open circlesrepresent Temporary habitat species, full circles Perma-nent habitat species

50 Hydrobiologia (2007) 583:43–56

123

(C-statistics = 0.1451; P = 0.2882) (Fig. 4d).

These results indicated little correlation between

these traits and their phylogenetic story for the six

species studied.

Discussion

Quantitative field data showed a turnover in

anuran larvae species along the hydroperiod

Table 1 Summary of univariate and multivariate ANOVA on plasticity in activity, developmental stage and morphology intadpoles of the six species

Source ofvariation

Univariate ANOVA on

Activity level Developmental stage

d.f. F P d.f. F P

Alytes 1, 9 27.69 0.0005 1, 9 7.2 0.0251Pelodytes 1, 17 17.79 0.0005 1, 17 0.1 0.7454B. bufo 1, 10 0.72 0.4136 1, 10 6.2 0.0318B. calamita 1, 12 2.18 0.1655 1, 12 15.2 0.0021Hyla 1, 17 14.31 0.0014 1, 17 9.8 0.0061Rana 1, 9 29.37 0.0004 1, 9 14.41 0.0042

Source ofvariation

d.f. MANOVA onmorphology

Univariate ANOVA on

F P Bodylength

Bodydepth

Tail finlength

Tail findepth

Tail musclelength

Tail muscledepth

Alytes 6, 4 2.08 0.2485 8.15* 2.14 1.19 0.49 1.61 12.97*Pelodytes 6, 12 39.46 <0.0001 19.13* 23.99* 38.61* 76.58* 14.99* 48.97*B. bufo 6, 5 3.27 0.1068 0.86 10.41* 0.88 1.39 0.87 0.81B. calamita 6, 7 1.63 0.2663 4.32 0.41 0.24 0.59 2.93 0.02Hyla 6, 12 20.48 <0.0001 2.27 0.01 1.47 97.87* 0.03 0.45Rana 6, 4 3.83 0.1069 0.43 0.78 14.77* 33.88* 0.30 0.91

Entries for individual morphological traits are F-values in univariate ANOVA, asterisk indicates statistical significance atalpha < 0.05

Table 2 ANOVA of the eight standardized variables (unitless magnitude of the phenotypic plasticity) and overall pro-portional change in morphology between species, habitats and with species nested within habitat

Variable Source of variation

Species (SP) (df = 5)(error = 36)

Habitat (HAB) (df = 1)(error = 40)

Species (SP[HAB]) (df = 4)(error = 36)

Activity level (absence ofpredator)

13.788** 42.097** 4.463*

Proportional change in activitylevel

9.985** 9.439* 8.380**

Developmental stage 12.938** 11.780* 10.445**Body length 17.076** 14.824** 13.139**Body depth 23.274** 13.346** 19.562**Tail fin length 46.138** 13.491** 40.857**Tail fin depth 61.708** 0.575 75.913**Tail muscle length 16.418** 8.183* 15.508**Tail muscle depth 73.985** 21.765** 56.720**Overall proportional change in

morphology73.879** 51.557** 73.420**

Notes: Six species, and Ephemeral-Temporary (Pelodytes and B. calamita) versus Temporary-Permanent habitat species(Alytes, B. bufo, Hyla and Rana) are compared (Habitat). Table entries are F ratios; * P < 0.05; ** P < 0.001

Hydrobiologia (2007) 583:43–56 51

123

habitat gradient as documented in previous stud-

ies (Snodgrass et al., 2000; Babbitt et al., 2003).

Our results differ significantly with respect to the

mechanisms that shape community structure

along the freshwater gradient proposed by Well-

born et al. (1996), where it was assumed that

permanence transition (between temporary hab-

itats and permanent habitats) also coincides with

the shift from communities without predators to

communities with large invertebrate top preda-

tors. The present study showed that the relative

abundance of large invertebrate predators is

greater in permanent ponds, but temporary ponds

could not be considered predator-free habitats.

The transition towards communities dominated

by large invertebrate predators occurs between

ephemeral pools and temporary ponds (Skelly,

1996; Wilbur 1997; Stoks & McPeek, 2003).

Tadpole assemblage may differ in the first tran-

sition due to the high risk of ephemeral pools

drying up and the different predator assemblages

in the two habitats: ephemeral and temporary

ponds. In the second transition detected in the

system studied (from temporary to permanent

ponds) the hydroperiod also seems to be the most

important factor. In the Mediterranean system of

isolated ponds studied, differences between tad-

pole assemblages could be attributed to factors

other than an increase in the abundance of

predators. Tadpole species assemblages can be

explained by the interplay of species differences

in developmental plasticity in response to pond

drying (Richter-Boix et al., 2006a), differences in

phenological breeding between species (Richter-

Boix et al., 2006b), strong interspecific interac-

tions (Richter-Boix et al., 2007) and vulnerability

Proportional change in morphology0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Pro

port

iona

l cha

nge

inde

velo

pmen

tals

tage

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10

Bc

Bb

PpHm

Ao

Rp

Pp

Bc

Bb

AoHm

Rp

(b)

(c) (d)

Ao

Pp

Hm Bc

Bb

Rp Ao

Pp

Hm Bc

Bb

Rp

(a)

Proportional change in activity-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1

Fig. 4 (a) Relationship between morphological plasticityand change in developmental stage, and (b) betweenbehavioural plasticity and change in developmental stage.All proportions are calculated in comparison to thepredator-free treatment, bars indicate standard error.Hypothesized phylogenetic relationships among the 6anuran species, where branch lengths did not correspond

to the real molecular data estimated. Sizes of full circlesrepresent magnitude of plasticity for morphology (c) andfor behaviour (d). Ao = Alytes obstetricans, Pp = Pelo-dytes punctatus, Bb = Bufo bufo, Bc = Bufo calamita,Hm = Hyla meridionalis, Rp = Rana perezi. Open circlesrepresent Temporary habitat species, full circles Perma-nent habitat species

52 Hydrobiologia (2007) 583:43–56

123

to predation. This system contains a low number

of predator-free ponds and, with the exception of

B. calamita and Pelodytes, the other species never

occur in predator-free ponds, although high vari-

ations were observed in predator composition and

concentration between ponds. The high spatial

and temporal variability in predator communities

to which tadpole species were subjected (Van

Buskirk & Arioli, 2005) supports the theory that

phenotypic plasticity is developed and maintained

in species or populations exposed to heteroge-

neous and unpredictable environments (Via et al.,

1995; Schlichting & Pigliucci, 1998; DeWitt &

Scheiner, 2004).

The laboratory experiments demonstrated that

four of the six anuran species exhibited predator-

induced plasticity in behavioural and morpholog-

ical traits, results that are consistent with previous

studies (e.g. Lardner, 2000; Van Buskirk, 2002).

The results of the experiments, in conjunction

with field observations of species habitat prefer-

ences and predation risk, confirm in part the

expected link between phenotypic plasticity and

habitat use according to temporality and preda-

tion risk. Species from the most temporary part of

the hydroperiod range, commonly exposed to the

development/predation trade-off, exhibited dif-

ferent integration of the plastic traits evaluated to

species from the most permanent ponds. Tempo-

rary species showed higher values of morpholog-

ical changes (Relyea & Werner, 2000) and low

activity reduction. Morphological changes (dee-

per and larger tails) were consistent with the

adaptive changes observed in several larval

amphibian species (Van Buskirk et al., 1997;

Van Buskirk & Relyea, 1998). Species could

therefore maintain a high level of activity while

responding morphologically rather than behavio-

urally to the presence of predators. Presumably,

these species must maintain activity levels in

order to achieve sufficient growth and develop-

mental rates to allow them to metamorphose

before ponds dry up (Laurila & Kujasalo, 1999;

Anholt et al., 2000). The positive correlation

between developmental change and activity

response supports this idea. In ephemeral ponds

the cost of greater activity (more rapid growth

and development) is a smaller increase in preda-

tion risk compared to the situation in more

permanent ponds, in which predator densities

are higher (Skelly, 1995). Unlike species from

temporary habitats, those from the most perma-

nent part of the range may tip the balance in

favour of predation risk as they are able to remain

in the aquatic habitat as larvae for a longer period

before metamorphosis (Anholt et al., 2000). As a

result, they showed a greater plasticity in activity

than typical temporary pond species, which in

turn leads to a greater reduction in developmen-

tal rates. Development stage change was strongly

correlated with activity levels but not with mor-

phological changes, which suggests that morpho-

logical changes have little impact on

developmental rates. Contrary to the results of

the present study, previous works have suggested

that morphological defences in response to pre-

dators have come at the cost of decreased

developmental rates (Relyea, 2002; Teplitsky

et al., 2005) and growth rates (Van Buskirk,

2000; Relyea, 2002; Teplitsky et al., 2005). Unfor-

tunately no data about growth rates was available

to estimate the effect of plasticity on growth

among species and to make comparisons with

other studies. However, the detected impact of

morphological phenotypic plasticity on develop-

ment is weak with respect to activity in the work

of Relyea (2002) after Bonferroni corrections,

thus supporting our results. Teplitsky et al. (2005)

did not measure activity and attributed all devel-

opmental cost to morphological changes without

including behavioural information, when is well

known that tadpoles exhibit both morphological

and behavioural inducible defences in response to

predators (e.g. Lardner, 2000; Van Buskirk, 2002)

and it is possible that part of the developmental

cost observed was related to changes in activity.

Tadpole species differed in the integration of

the phenotypic traits measured, but differences

observed between species could not be attributed

only to habitat. It is possible that taxonomy and

habitat interact in determining anti-predator

responses (Richardson, 2001a; Van Buskirk,

2002). This may reflect the presence of alternative

strategies available for persistence in a given

habitat type (Richardson 2001b, 2002a). Our

results did not reveal any phylogenetic correla-

tion between the traits studied, although the small

number of species considered and the great

Hydrobiologia (2007) 583:43–56 53

123

phylogenetic distances between them may ob-

scure any possible relationship. In addition, the

similar response in activity and morphology

between the two bufonid species suggest some

phylogenetic constraints. Previous studies with a

greater number of species have demonstrated

potential constraint in the evolution of predator-

induced plasticity in anurans, with some phyloge-

netic inertia in behavioural and morphological

trait values (Richardson, 2001b, 2002a; Van Bus-

kirk, 2002). Species from the same habitat type do

not always converge on similar morphological or

behavioural phenotypes, which suggests that lin-

eages in this system can evolve to one of many

different evolutionary solutions (Richardson,

2001a; Van Buskirk, 2002). The absence of or

weak responses to the presence of a predator

detected in bufonids suggest that Bufo use an

anti-predator mechanism other than reduction of

activity levels or morphological changes, such as

chemical deterrents or the formation of cohesive

swarms (Watt et al., 1997). Differences in mor-

phological changes between species from a sim-

ilar habitat (Alytes, Hyla and Rana) could in part

reflect differences in microhabitat use in the

water column and biomechanical constraints

(Richardson, 2002b). Microhabitat use differs

between these species (Dıaz-Paniagua, 1987a, b)

and it is plausible that morphological changes

affect swimming speed, stability and manoeuvra-

bility. Studies of the phenotypic integration of

morphological traits in different environments

and the consequences and ecology of these traits

are needed to understand differences in morpho-

logical plasticity between species.

A potential weakness of this study in terms of

its ability to resolve possible trade-offs between

species along the pond permanency and predator

risk gradients is the lack of consideration given to

the effects of pond drying on predator-induced

phenotype. We know that species differ in their

developmental phenotypic plasticity in response

to pond drying (Richter-Boix et al., 2006a) and in

their morphological and behavioural plasticity in

response to a predation risk (present study). To

date only three studies have addressed the non-

lethal presence of predators under time con-

straints (Laurila & Kujasalo, 1999; Altwegg, 2002;

Bridges, 2002) and a discrepancy exists between

the findings of these studies. Rana temporaria, a

typical temporary pond breeder, reduced their

predation reaction when exposed to a drying

environment (Laurila & Kujasalo, 1999), whereas

the anti-predator responses of Rana lessonae and

Pseudacris triseriata do not alter under time

constraints (Altwegg, 2002; Bridges, 2002). It is

possible that species from temporary ponds with

fewer predators prioritise the desiccation risk

over the predation risk, whereas species from

permanent ponds and exposed to a greater

abundance of predators perceive predation as a

greater immediate threat than pond drying and

consequently do not take greater risks. Future

experimental research into the joint effects of

pond drying and predators on the development

and phenotypic plasticity of tadpoles is required

in order shed light on these relationships and the

distribution of species along the hydroperiod

gradient.

Acknowledgements We sincerely thank Nuria Garrigafor help during laboratory experiments and for usefulcomments. We thank Laura Perez and Santi Escartınfor assisting with field surveys, and S. Llacuna andF. Llimona for providing the corresponding permits andfieldwork facilities in the Natural Parks. We also thankS. Carranza for their assistance in the phylogenetictopology reconstruction of the species object of studyand providing personal data for this purpose. We extendour gratitude to two anonymous referees for providinguseful comments and suggestions. Permission to capturewas granted by the Departament de Medi Ambient de laGeneralitat de Catalunya.

References

Altwegg, R., 2002. Predator-induced life-history plasticityunder time constraints in pool frogs. Ecology 83:2542–2551.

Anholt, B. R. & E. E. Werner, 1995. Interaction betweenfood availability and predation mortality mediated byactivity. Ecology 76: 2235–2239.

Anholt, B. R., D. K. Skelly & E. E. Werner, 1996. Factorsmodifying antipredator behaviour in larval toads.Herpetologica 52: 301–313.

Anholt, B. R., E. E. Werner & D. K. Skelly, 2000. Effect offood and predators on the activity of four larval ranidfrogs. Ecology 81: 3509–3521.

Babbitt, K. J., J. B. Matthew & T. L. Tarr, 2003. Patternsof larval amphibian distribution along a wetlandhydroperiod gradient. Canadian Journal of Zoology81: 1539–1552.

54 Hydrobiologia (2007) 583:43–56

123

Bookstein, F. L., 1991. Morphometrics Tool for LandmarkData: Geometry and Biology. Cambridge UniversityPress, Cambridge.

Bridges, C. M., 2002. Tadpoles balance foraging andpredator avoidance: effects of predation, pond drying,and hunger. Journal of Herpetology 36: 627–634.

Chovanec, A., 1992. The influence of tadpole swimmingbehaviour on predation by dragonfly nymphs.Amphibia-Reptilia 13: 341–349.

Cronin, J. T. & J. Travis, 1986. Size-limited predation onlarval Rana aerolata (Anura: Ranidae) by two speciesof backswimmers (Heteroptera: Notonectidae). Her-petologica 42: 171–174 .

DeWitt, T. J. & S. M. Scheiner, 2004. Phenotypic Plastic-ity: Functional and Conceptual Approaches. OxfordUniversity Press, New York, USA.

Dıaz-Paniagua, C., 1987a. Estudio en cautividad de laactividad alimenticia de las larvas de siete especies deanuros. Revista Espanola de Herpetologıa 2: 189–197.

Dıaz-Paniagua, C., 1987b. Tadpole distribution in relationto vegetal heterogeneity in temporary ponds. Herpe-tological Journal 1: 167–169.

Eklov, P. & C. Halvarsson, 2000. The trade-off betweenforaging activity and predation risk for Rana tempo-raria in different food environments. Canadian Jour-nal of Zoology 78: 734–739.

Felsenstein, J., 1985. Phylogenies and the comparativemethod. American Naturalist 125: 1–13.

Gunzburger, M. S. & J. Travis, 2004. Evaluating predationpressure on green treefrog larvae across a habitatgradient. Oecologia 140: 422–429.

Laurila, A. & J. Kujasalo, 1999. Habitat duration, preda-tion risk and phenotypic plasticity in common frog(Rana temporaria) tadpoles. Journal of Animal Ecol-ogy 68: 1123–1132.

Lardner, B., 2000. Morphological and life historyresponses to predators in larvae of seven anurans.Oikos 88: 169–180.

Loy, A., M. Corti & L. F. Marcus, 1993. Landmarks data:size and shape analysis in systematics. A case studyon old world talpidae (Mammalia, Insectivora). InMarcus, L. F., E. Bello & A. Garcıa-Valdecasas (eds),Contributions to Morphometrics. Monografıas delMuseo Nacional de Ciencias Naturales, Madrid,215–239.

McCollum, S. A. & J. D. Leimberger, 1997. Predator-induced morphological changes in an amphibian:predation by dragonflies affects tadpole shape andcolor. Oecologia 109: 612–615.

Reeve, J. & E. Abouheif, 2003. Phylogenetic Indepen-dence. Version 2.0, Department of Biology, McGillUniversity. Distributed freely by the authors onrequest.

Relyea, R.A., 2002. Cost of phenotypic plasticity. Amer-ican Naturalist 159: 272–282.

Relyea, R. A. 2004. Integrating phenotypic plasticity whendeath is on the line: insights from predator–preysystems. In Pigliucci, M. & K. Preston (eds), Pheno-typic Integration: Studying the Ecology and Evolutionof Complex Phenotypes. Oxford University Press,New York, USA.

Relyea, R. A. & E. E. Werner, 1999. Quantifying therelation between predator-induced behavior andgrowth performance in larval anurans. Ecology 80:2117–2124.

Relyea, R. A. & E. E. Werner, 2000. Morphologicalplasticity in four larval anurans distributed along anenvironmental gradient. Copeia 2000: 178–190.

Richardson, J. M. L., 2001a. The relative roles of adapta-tion and phylogeny in determination of larval traits indiversifying anuran lineages. American Naturalist 157:282–299.

Richardson, J. M. L., 2001b. A comparative study ofactivity levels in larval anurans and response to thepresence of different predators. Behavioural Ecology12: 51–58.

Richardson, J. M. L., 2002a. A comparative study of pheno-typic traits related to resource utilization in anurancommunities. Evolutionary Ecology 16: 101–122.

Richardson, J. M. L., 2002b. Burst swim speed in tadpolesinhabiting ponds with different top predators. Evolu-tionary Ecology Research 4: 627–642.

Richter-Boix, A., G. A. Llorente & A. Montori, 2006a. Acomparative analysis of the adaptive developmentalplasticity hypothesis in six Mediterranean anuranspecies along a pond permanency gradient. Evolu-tionary Ecology Research 8: 1139–1154.

Richter-Boix, A., G. A. Llorente & A. Montori, 2006b.Breeding phenology of an amphibian community in aMediterranean area. Amphibia-Reptilia 27: 544–559.

Richter-Boix, A., G. A. Llorente & A. Montori, 2007.Hierarchical competition in pond-breeding anuranlarvae in a Mediterranean area. Amphibia-Reptilia28: in press.

Rohlf, F. J., 2001. Thin-Plate Spline (TPS) Software. Freeavailable on:http://life.bio.sunysb.edu/morph/ (lastmodified October 31, 2005).

Schaffer, H. B., R. A. Alford, B. D. Woodward, S. J.Richards, R. G. Altig & C. Gascon, 1994. Quantitativesampling of amphibian larvae. In Heyer, W. R., M. A.Donnelly, R. W. McDiarmid, L.C. Hayek & M. S.Foster (eds), Measuring and Monitoring BiologicalDiversity Standard Methods for Amphibians. Smith-sonian Institution Press, Washington, 130–141.

Schlichting, C. & M. Pigliucci, 1998. Phenotypic Evolution:A Reaction Norm Perspective. Sinauer, Sunderland,400 pp.

Schneider, D. W. & T. M. Frost, 1996. Habitat duration andcommunity structure in temporary ponds. Journal ofthe North American Benthological Society 15: 64–86.

Skelly, D. K., 1994. Activity level and the susceptibility ofanuran larvae to predation. Animal Behaviour 47:465–468.

Skelly, D. K., 1995. A behavioural trade-off and itsconsequences for the distribution of Pseudacris tree-frog larvae. Ecology 76: 150–164.

Skelly, D. K., 1996. Pond drying, predators, and thedistribution of Pseudacris treefrog tadpoles. Copeia1996: 599–605.

Skelly, D. K. & E. E. Werner, 1990. Behavioral and life-historical responses of larval American toads to anodonate predator. Ecology 71: 2313–2322.

Hydrobiologia (2007) 583:43–56 55

123

Snodgrass, J. W., A. L. Bryan & J. Burger, 2000.Development of expectations of larval amphibianassemblage structure in southeastern depression wet-lands. Ecological Applications 10: 1219–1229.

Stocks, R. & M. A. McPeek, 2003. Predators and lifehistories shape Lestes damselfly assemblages along afreshwater habitat gradient. Ecology 84: 1576–1587.

Teplitsky, C., S. Plenet & P. Joly, 2005. Cost and limits ofdosage response to predation risk: to what extent cantadpoles invest in anti-predator morphology? Oeco-logia 145: 364–370.

Travis, J., W. H. Keen & J. Juilianna, 1985. The role ofrelative body size in a predator–prey relationshipbetween dragonfly naiads and larval anurans. Oikos45: 59–65.

Van Buskirk, J., 2000. The cost of an inducible defense inanuran larvae. Evolution 81: 2813–2821.

Van Buskirk, J., 2002. A comparative test of the adaptiveplasticity hypothesis: relationships between habitatand phenotype in anuran larvae. American Naturalist160: 87–102.

Van Buskirk, J., 2003. Habitat partitioning in Europeanand North American pond-breeding frogs and toads.Diversity and Distributions 9: 399–410.

Van Buskirk, J. & M. Arioli, 2005. Habitat specializationand adaptive phenotypic divergence of anuranpopulations. Journal of Evolutionary Biology 18:596–608.

Van Buskirk, J. & R. A. Relyea, 1998. Natural selectionfor phenotypic plasticity: predator-induced responsesin tadpoles. Biological Journal of the Linnean Society65: 301–328.

Van Buskirk, J., S. A. McCollum & E. E. Werner, 1997.Natural selection for environmentally-induced phe-notypes in tadpoles. Evolution 52: 1983–1992.

Via, S., R. Gomulkiewicz, G. De Jong, S. Scheiner, C. D.Schlichting & P. H. Van Tienderen, 1995. Adaptivephenotypic plasticity: consensus and controversy.Trends in Ecology and Evolution 10: 212–217.

Watt, P. J., S. F. Nottingham & S. Young, 1997. Toadtadpole aggregation behaviour: evidence for a pred-ator avoidance function. Animal Behaviour 54:865–872.

Wellborn, G. A., D. K. Skelly & E. E. Werner, 1996.Mechanism creating community structure across afreshwater habitat gradient. Annual Review in Ecol-ogy and Systematics 27: 337–363.

Werner, E. E. & M. A. McPeek, 1994. Direct and indirecteffects of predators on two anuran species along anenvironmental gradient. Ecology 75: 1368–1382.

Wilbur, H. M., 1997. Experimental ecology of food webs:complex systems in temporary ponds. Ecology 78:2279–2302.

Woodward, B. D., 1983. Predator–prey interactions andbreeding-pond use of temporary-pond species in adesert anuran community. Ecology 64: 1549–1555.

56 Hydrobiologia (2007) 583:43–56

123