lable at ScienceDirect

Animal Behaviour 92 (2014) 241e252

SPECIAL ISSUE: KIN SELECTION

Contents lists avai

Animal Behaviour

journal homepage: www.elsevier .com/locate/anbehav

Special Issue: Kin Selection

The evolution of multiqueen breeding in eusocial lineages withpermanent physically differentiated castes

Jacobus J. Boomsma*, Dóra B. Huszár, Jes Søe PedersenCentre for Social Evolution, Department of Biology, University of Copenhagen, Denmark

a r t i c l e i n f o

Article history:Received 29 October 2013Initial acceptance 15 January 2014Final acceptance 24 February 2014Available online 9 May 2014MS. number: ASI-13-00903

Keywords:cooperative breedingEmery’s rulegroup selectionHamilton’s rulekin selectionlife spanmodular growthpolygynysib competitiontragedy of the commons

* Correspondence: J. J. Boomsma, Centre for SociBiology, University of Copenhagen, UniversitetsparkeDenmark.

E-mail address: jjboomsma@bio.ku.dk (J. J. Booms

http://dx.doi.org/10.1016/j.anbehav.2014.03.0050003-3472/� 2014 The Authors. Published on behalf oND license (http://creativecommons.org/licenses/by-n

The hypothesis that obligate eusociality always evolved from ancestral states of strict lifetime monogamyimplies that (1) facultatively eusocial lineages had to abandon multifemale breeding to achieve per-manent morphologically differentiated castes, and (2) lineages of obligatorily eusocial insects had toindependently re-evolve multifemale breeding when that served the inclusive fitness interests of nursingworkers. Multiqueen nesting (eusocial polygyny) is known to be common across the ants, but rare in thecorbiculate (eusocial) bees, the vespine wasps and the higher termites, but we show that this differenceis mostly due to cases of obligate polygyny being restricted to the ants. This pattern is remarkably similarto the distribution of inquiline social parasites that use stealth rather than aggression to invade hostcolonies, which also repeatedly evolved in ants only. We explore the lineage-specific selection forces thathave allowed or constrained de novo evolution of stable eusocial polygyny in Hamiltonian inclusivefitness terms. We argue that perennial life histories, male survival as stored sperm rather than as lifetimemates, and sib competition are possibly sufficient to explain the general prevalence of secondarypolygyny across the obligatorily eusocial insects. We infer that obligate polygyny compromises eusocial‘soma’ and ‘germ-line’ segregation in ways known to decrease developmental stability in metazoans, andwe briefly evaluate the selection forces that reduce queen life span in highly, but probably not faculta-tively, polygynous species. We conclude that secondary polygyny in its obligate (ant) form resemblescooperative breeding with multiple ‘tragedy of the commons’ aspects, but in a peculiar manner becausebreeding females are selected to exploit the services of unmated workers rather than each other’s. Thisbreeding system has likely been maintained in ants because it allows modular extensions of colonies indirectly adjacent habitat of similar quality without the re-emergence of sexual conflicts or unproductivelocal competition with kin.� 2014 The Authors. Published on behalf of The Association for the Study of Animal Behaviour by ElsevierLtd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/

by-nc-nd/3.0/).

During the late 1970s Bert Hölldobler, George Oster, and I consid-ered changes in queen number following the rise of eusociality inants as a problem in evolutionary optimization, determined ulti-mately by kin selection and environmental selection pressures(Wilson, 1993)

Eusocial insects have an enormous variety of breeding systems,characterized by different numbers of egg-laying females withdifferent shares in reproduction (Bourke & Heinze, 1994; Choe &Crespi, 1997; Hölldobler & Wilson, 1990; Keller, 1993; Michener,1974; Ross & Matthews, 1991; Wheeler, 1928; Wilson, 1975). This

al Evolution, Department ofn 15, DK-2100 Copenhagen,

ma).

f The Association for the Study of Ac-nd/3.0/).

is not unlike the vertebrate cooperative breeders (Kappeler, 2010;Koenig & Dickinson, 2004; Solomon & French, 2007), but the maleroles in eusocial insects are fundamentally different, because theyare only present as stored sperm in the haplodiploid ants, bees andwasps, or as unusually committed monogamous breeders in thetermites (Boomsma, 2007; Boomsma, Baer, & Heinze, 2005). In theobligatorily eusocial insects (ants, corbiculate bees except theeuglossines, vespine wasps and higher termites), queens are somorphologically specialized that they can never reproduce withouthelpers, and these helpers are all irreversibly committed tomorphologically distinct caste phenotypes characterized by lifetimeunmatedness. Queens, workers and (if they occur) soldiers havethus universally lost reproductive totipotency in exchange for per-manent mutual dependence, fuelled by indirect fitness benefits(Hamilton,1964a,1964b) for the unmated castes (Beekman, Peeters,& O’Riain, 2006; Boomsma, 2013; Crespi & Yanega, 1995).

nimal Behaviour by Elsevier Ltd. This is an open access article under the CC BY-NC-

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252242

SPECIAL ISSUE: KIN SELECTION

Both inclusive fitness theory and comparative analyses indicate thatthese obligatorily eusocial lineages evolved from lifetime monoga-mous ancestors (Boomsma, 2007, 2009, 2013; Hughes, Oldroyd,Beekman, & Ratnieks, 2008) so that average relatedness to siblingsequalled relatedness to offspring when castes differentiatedtowards irreversible morphological and physiologicalcomplementarity.

The alternative of facultative eusociality may take two forms. Inthe most basal version, colonies or populations differ in havinghelpers at the nest or doing without, as happens in some halictidbees and vertebrate cooperative breeders. In the more elaboratedversion, colony life has become obligate (i.e. all nests have at leastsome helpers), which is typically accompanied by some physicaldifferentiation between breeders and helpers, but always by adultphenotypic plasticity rather than by hardwired developmental dif-ferences that are completed before pupation. Examples are polistinewasps, including the swarm-forming polybiines and, in vertebrates,the naked mole-rat. Lifetime commitment to unmatedness and apriori constrained reproductionof helpers thusnever fully applies infacultatively eusocial breeding systems, consistent with parentalmonogamy never being universal either (Boomsma, 2009, 2013).Representatives of these lineages always live in colonies where atleast some individuals express eusocial nursing or defending be-haviours, but where a variable number of them also retain repro-ductive totipotency (Beekman et al., 2006; Crespi & Yanega, 1995).

EusocFacultative (non-lifetime-committed castes)

Col

ony

fou

nd

ing

Rel

ated

que

ens

Un

rela

ted

qu

een

sSi

ngl

e q

uee

n

Exclusive colony founding; later joining by additional foundresses may increase nest survival but reduce future reproductive success; usurpation is a major risk of founding colonies alone

E.g. Many halictine bees and some polistine wasps

Shared colony founding from the start, often resulting in shared reproduction to prevent subordinates from leaving

E.g. Some carpenter bees and some polistine wasps

Shared colony founding from the start, often resulting in shared reproduction; selection pressure on subordinates to reproduce or leave is less compared to the unrelated foundress situation above because subordinates can also gain indirect fitness

E.g. Many polistine wasps including swarm foundinglineages

Figure 1. An overview of what is known about the relationship between colony founding (hof helpers, separating lineages that are facultatively eusocial (having either purely behavioufrom those with obligate eusociality (having physically differentiated castes that are irrevedefining characteristics and gives typical but nonexhaustive examples. Secondary polygyny,mated queens independent of the previous generation of queen(s), is restricted to the obli

Dominantbreeders thus obtain their status by likelihood rather thandevelopmental destiny, so nestswill have at least some helperswiththe potential to become breeders later in life, either at home orelsewhere (Boomsma, 2013; Crespi & Yanega, 1995).

In facultatively eusocial and cooperative breeders, decisions tojoin or leave nests are largely topedown in hierarchies of dominantbreeders, so that nest inheritance practices always have elements ofusurpation (Fig. 1). However, once the point of no return towardsobligate eusociality has been passed, such decisions becomecontingent on the bottomeup approval of a ‘silent’ majority ofunmated, lifetime subordinate workers (Bourke & Franks, 1995;Nonacs, 1988; Pamilo, 1991). It now seems most likely that thistype of transition happened twice independently in the bees, as thehoneybees and euglossine bees are more closely related than eachof them is to the bumblebees and stingless bees, which form theirown clade (Cardinal & Danforth, 2011). The ants and vespine waspsrepresent single origins of obligate eusociality (Hughes et al., 2008;Johnson et al., 2013), and whether there are additional transitionstowards true worker castes outside the higher termites is stilldebated (Inward, Vogler, & Eggleton, 2007).

Being precise about categories of eusociality implies that similarprecision is helpful when defining polygyny. The term is confusingto start with, because it is normally used to describe harems(i.e. situations where a single male breeds simultaneously withmultiple females). This never occurs in the eusocial insects as

ialityObligate (lifetime-committed castes)

Shared colony founding followed by:

a. Restoration of monogyny after lethal fighting between co-foundresses or their first workers b. Continued co-breeding (primary polygyny)

E.g. Some ants and vespine wasps

Secondary polygyny as later elaboration not reported

Only possible when perennial colonies produce budnests

E.g. Some termites and ants

Secondary polygyny involving newly mated reproductives seems only possible in polygynous ants; termite bud-nests contain reproductives that are all offspring of the founding pair, so do not introduce new blood into existing colonies, but nest mergers may do

Exclusive colony founding similar to the ancestral situation when permanent, physically differentiated castes evolved. The threat of usurpation has often selected for claustral colony founding when queens are not accompanied by worker swarms

E.g. Ants, corbiculate bees, vespine wasps, higher termites

Secondary polygyny may be initiated after thefoundress has reproduced for the first time

Pleo

met

rosi

sPl

eom

etro

sis

Hap

lom

etro

sis

aplometrosis or single queen; pleometrosis or multiple queens) and caste commitmentral helpers or workers and breeders that differentiate via adult phenotypical plasticity)rsibly determined before reaching the adult stage). Each of the six boxes summarizesin which physically differentiated workers make decisions to re-adopt or reject newlygatorily eusocial domain (see text for details).

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252 243

SPECIAL ISSUE: KIN SELECTION

colony-founding males are either monogamously committed (ter-mites) or fundamentally subordinate, when their role is restrictedto being represented as stored sperm in the body of a queen, eitheralone or with other ejaculates (ants, bees and wasps; Boomsmaet al., 2005). Within the eusocial insects it is further important todiscriminate between three forms of polygyny or pleometrosis: (1)primary polygyny, based on foundresses initiating a nest togetherand sharing later reproduction (common in facultatively eusocialbreeding and rare in obligatorily eusocial breeding; Bernasconi &Strassmann, 1999; Ross & Matthews, 1991); (2) primary polygyny,which only lasts until the first workers hatch, as that event triggerslethal fighting among the cofounding queens or their workers sothat only one queen remains (Holman, Dreier, & d’Ettorre, 2010;Sommer & Hölldobler, 1995); and (3) secondary polygyny, basedon workers of established nests re-adopting new cohorts of queens(mostly or exclusively colony daughters) after they have matedwith usually unrelated males (Bourke & Franks, 1995; Crozier &Pamilo, 1996; Hölldobler & Wilson, 1990). Our present review willfocus on this third form of polygyny where unmated and physicallydifferentiated workers make these re-adoption decisions (Fig. 1).

The first reviews on the evolution of polygyny in social insectsfollowing Hamilton (1964b, 1972) focused on ants (Hölldobler &Wilson, 1977) and polistine wasps (Turillazzi & West-Eberhard,1996; West-Eberhard, 1978), and early models indicated thatpolygyny can in principle be understood as a kin-selected adapta-tion (Nonacs, 1988; Pamilo, 1991; see also quote at the start of thisreview). These insights were aptly summarized, elaborated, andsupplemented with data on other social taxa by Bourke and Heinze(1994) and Keller (1993), and in some later chapters of editedvolumes (e.g. Heinze & Foitzik, 2009; Steiner, Crozier, & Schlick-Steiner, 2010). At the same time, empirical studies of polygynybegan to focus on reproductive skew (Nonacs & Hager, 2011) and onpolygynous ants that seemed to challenge kin selection orthodoxy(i.e. the most extreme, evolutionarily derived forms of polygyny inthe so-called unicolonial ants; Helanterä, Strassmann, Carrillo, &Queller, 2009). However, all previous reviews have consideredpolygyny as a continuum of multifemale nesting, without makingthe distinction between facultative and obligate eusociality that wemake in Fig. 1 (see also Boomsma, 2009, 2013). In the former,dominant females compete for a maximal share in reproductionwith subordinates that end up assuming helper roles for indirectfitness benefits and/or a statistical likelihood of gaining directfitness later in life (e.g. Leadbeater, Carruthers, Green, Rosser, &Field, 2011). In the latter, parties are fundamentally differentcaste phenotypes that are engaged in a game of mutual trust, oneparty being the adopting (permanently unmated) workers that areunder selection to favour close relatives for indirect fitness gains,the other being the newly mated queens looking for any colony inwhich they will be allowed to breed and primarily realize theirdirect fitness potential.

When we consider only the eusocial insects with physicallydifferentiated, permanent castes, the ones we call obligatorilyeusocial, it is clear that secondary polygyny is most prevalent in theants, where it has evolved many times (e.g. Bourke & Franks, 1995;Buschinger, 1974; Crozier & Pamilo, 1996; Hölldobler & Wilson,1977, 1990; Keller, 1993). In contrast, secondary polygyny invespine wasps and eusocial corbiculate bees is rare (Michener,1974; Ross & Matthews, 1991), but quantitative comparative ana-lyses have been few (but see Cronin, Molet, Doums, Monnin, &Peeters, 2013) and neither do we have formal analyses of the dif-ferences between facultative polygyny (only some nests in a pop-ulation adopt newly mated queens) and obligate polygyny (most ifnot all reproducing colonies have multiple co-breeding queens; butsee Bourke & Franks, 1991; Sundström, Seppä, & Pamilo, 2005).Although there may be intermediate situations between these

categories, it is important to acknowledge such bimodal distribu-tions of breeding systems. This is because ‘facultative’ refers to theorigin of a social trait (it has evolved and is maintained at some,usually low frequency) and ‘obligate’ refers to an elaboration inwhich the social trait has essentially become fixed in a populationor species. The evolutionary explanations for these alternativeadaptive states are often very different (Bourke, 2011), similar tothe explanations for facultative and obligate multiple mating ofeusocial Hymenoptera (Boomsma, 2013). The same applies to thedistinction between fortress defenders and life insurers (Queller &Strassmann, 1998), which is an argument of origins that has alsorecently been extended to offer an explanation for the sex ofhelpers when advanced social breeding evolves (Ross, Gardner,Hardy, & West, 2013). However, when eusociality has becomeelaborated into more advanced stages, this distinction loses itspower because helpers usually combine nursing and defencefunctions.

The first objective of this review is to explainwhy all obligatorilyeusocial lineages except ants have apparently been constrained toevolve high and potentially obligate degrees of polygyny, in spite offacultative polygyny usually occurring. Our second objective is todevelop analogies between different versions of polygyny withforms of eukaryote multicellularity, elaborating on similar ap-proaches in earlier reviews (Boomsma, 2009; Bourke, 2011). Suchanalogies are of interest because the colonies of obligatorily euso-cial insects have (super)organismal properties (Hölldobler &Wilson, 2009; Queller & Strassmann, 2009; Ratnieks & Reeve,1992), but to be fully consistent with organisms these analogiesshould include joint principles for sequestering germ-lines(Boomsma, 2009; Bourke, 2011). Our third objective is to revisitthe evolution of aggressive nest usurpers and stealthy (inquiline)social parasites to explicitly connect the latter (but not the former)to the evolution of secondary polygyny in the obligatorily eusocialdomain (Bourke & Franks, 1991, 1995; Buschinger, 1986, 2009).Finally, we will briefly dwell on evolutionary aspects of life spanand ageing when reproductives become disposable because ofsecondary polygyny. Earlier work (Keller & Genoud, 1997) empha-sized that polygynous queens are shorter-lived than monogynousqueens, but the ultimate evolutionary mechanism behind selectionfor shorter life span of reproductives after secondary polygynyevolved remains opaque (Bourke & Franks, 1995; Heinze &Schrempf, 2008; Parker, 2010). Also here we feel that it is helpfulto consider obligate and facultative eusociality separately.

TYPES AND PREVALENCES OF SECONDARY POLYGYNY

Comparative data indicate that facultative secondary polygynyhas evolved not only in the ants, but also in the corbiculate bees andvespine wasps. Monogyny always remains the most prevalentcolony structure (chi-square tests: P < 0.0001 for all three groups;Fig. 2), consistent with single queen colony founding being theancestral state (Boomsma, 2007; Hughes et al., 2008) and second-ary polygyny having evolved to varying degrees (Fisher’s exact test:P ¼ 0.0006) across lineages as later elaborations of obligate euso-ciality. In contrast, high degrees of polygyny are only found in theants where an appreciable frequency (23%) of the species have atleast some populations that are obligatorily polygynous, and for anumber of ant species this social structure appears to apply in allpopulations (see Supplementary material). Our statistical analysisdid not adjust for possible phylogenetic confounding, but webelieve it gives a reasonably accurate account because monogynyand polygyny are social traits that normally vary among pop-ulations of the same species, thus having little phylogenetic inertia.Although we have not formally analysed this, the higher termites(i.e. the major derived monophyletic termite clade with true

200

150

100

Nu

mbe

r of

sp

ecie

s

50

0Monogynous

(a) (b) (c)Nsample = 340Nknown = ca. 12500

Nsample = 88Nknown = ca. 1200

Nsample = 69Nknown = ca. 69

Polygynous Monogynous MonogynousPolygynous Polygynous

Ants Eusocial corbiculate bees Vespine wasps

Figure 2. Facultative secondary polygyny (black bars) has evolved in all three lineages of obligatorily eusocial Hymenoptera: (a) the ants, (b) the corbiculate bees and (c) the vespinewasps. It is always less common than monogyny (white bars; c2

ðaÞ ¼ 91:87, c2ðbÞ ¼ 75:41, c2

ðcÞ ¼ 34:12, P < 0.0001 for all), but its relative prevalence varies across lineages (Fisher’sexact test: P ¼ 0.0006). However, obligate polygyny (grey bar) is exclusively found in the ants. Our ant database contained 364 species, but we excluded primary polygynous,parthenogenetic and obligate socially parasitic species from further analysis. Of the 340 species analyzed 55% were monogynous, 13% facultatively polygynous and 23% obligatorilypolygynous. In the case of bees and wasps 97% and 87% were monogynous and 3% and 13% were facultatively polygynous, respectively. For a number of secondary polygynous ants(29; 8%) literature records did not allow us to assess whether polygyny was facultative or obligate (hatched bar). See Supplementary material for data and notes.

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252244

SPECIAL ISSUE: KIN SELECTION

workers; Inward et al., 2007) are similar to the corbiculate bees andvespine wasps in their degree of polygyny. Facultative polygyny hasbeen found in a number of genera, but it is usually if not alwayslinked to replacement reproductives with obligate sib mating or,rarely, colony mergers. However, very high and obligate degrees ofpolygyny have not been observed (Kobayashi et al., 2013; Vargo &Husseneder, 2011; see also Supplementary material).

Although these patterns will not come as a surprise to socialinsect researchers (e.g. Keller, 1993, has multiple such notionsacross various chapters), they have never been interpreted inrelation to specific evolutionary hypotheses. This is because polis-tine wasps also have lineages with obligate polygyny and thesewere always explicitly (e.g. Sherman, Lacey, Reeve, & Keller, 1995)or implicitly (e.g. Wilson, 1975) considered as part of a eusocialitycontinuum that also included the eusocial insects with permanentmorphologically differentiated castes. However, this continuumconcept can only be defended for the cooperative and facultativelyeusocial breeding systems (to which polistine wasps belong), but itdoes not include the obligatorily eusocial domain (Beekman et al.,2006; Boomsma, 2013; Crespi & Yanega, 1995). The explicit ques-tion of why high and potentially obligate levels of polygyny havenever evolved after lifetime ancestral monogamy secured perma-nent caste differentiation in the corbiculate bees, vespine waspsand higher termites has therefore never been addressed. To do that,it appeared to be most straightforward to consider social traits thateach of these lineages does not share with the ants: (1) annualcolony life cycles (bumblebees and vespine wasps); (2) long-distance dispersal and colony founding by swarming (honeybees,stingless bees and Provespa); and (3) the physical presence of malebreeders in colonies (higher termites).

Semelparity and Iteroparity

As it appears, there are no bumblebees and vespine wasps inwhich queens have been under selection to live longer than a year,and neither are their paper or wax nests in the soil or open air madeto last (Hansell, 1996; Michener, 1974). This likely reflects thatannual semelparity is a life-history syndrome that is difficult toabandon, a phenomenon that is known as Cole’s paradox. A radicalreduction of extrinsic (predation and disease) mortality of maturenests is needed to induce selection for postponed reproduction,longer queen life spans, and ultimately iteroparous breeding

(Charnov & Schaffer, 1973; Stearns, 1992). This transition to peren-nial iteroparity has independently been achieved by the ancestorsof the ants, the termites, the honeybees and the stingless bees(Keller & Genoud, 1997; Parker, 2010), which all tend to have betterprotected nests and probably lower disease pressure than bum-blebees and vespine wasps (Boomsma, Schmid-Hempel, & Hughes,2005). As secondary polygyny implies that workers adopt newlymated queens after their colony has reproduced for the first time(Fig. 1), it appears obvious that the annual life cycles of bumblebeesand vespine wasps may allow facultative polygyny under unusuallybenign conditions for nest survival, but not obligate polygyny.

Long-distance Dispersal by Swarming

The syndrome of obligate eusocial swarming is restricted to thehoneybees, the stingless bees and the vespine genus Provespa (i.e.to lineages whose workers are winged). It implies that queens donot found new colonies on their own, but are accompanied by agroup of permanently unmated workers to achieve long-distancedispersal. In more detail, swarming evolved in the little-studiedbasal and nocturnal Provespa branch of the vespine wasps, whereswarms contain rather few workers so that colonies can producemultiple swarms, each headed by a single newly hatched andmated queen (Archer, 2012; Matsuura, 1991). Swarming evolvedtwice independently in the eusocial bees (Cardinal, Straka, &Danforth, 2010) and is characterized by the young queen leavingwith the swarm in stingless bees and the old queen leavingwith theswarm in honeybees, although the latter may also produce after-swarms headed by young queens (Michener, 1974). Neither inannual Provespa, nor in perennial honeybees and stingless beeshave swarms been observed to have multiple queens. In contrast,multiple breeder swarms are characteristic for social insect line-ages, such as the epiponine and ropalidiine wasps, that have notpassed the transition to obligate eusociality in the sense of castebeing determined before pupation for all individuals. Here, adultfemale caste remains phenotypically plastic during a significantpart of adult life, so these swarms are conceptually incomparablewith those of Provespa, stingless bees and honeybees.

The most obvious explanation for swarm founding of obligato-rily eusocial colonies being incompatible with secondary polygynyis local resource competition (Hölldobler &Wilson,1977; Kümmerli& Keller, 2008). When competing for a fixed local resource, even

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252 245

SPECIAL ISSUE: KIN SELECTION

closely related breeders are predicted to eliminate each other if thegeneration that reared them did not already avoid producing morethan a single heir (West, Pen, & Griffin, 2002). This is the same logicas in documented or mythological human history where royalsiblings are happy to collaborate in conquering neighbouring ter-ritory, but are likely to kill each other when competing for thethrone at home.Wemeet this scenario in its simplest possible formin honeybees, where the local resource for a young queen is veryclearly defined as the home nest. Honeybee workers will oftenproduce only a single new queen, but newly hatched queens willalways look out for any sister queens that are about to hatch andwill kill them without workers interfering (Michener, 1974; Seeley,1985). Consistent with this logic, even facultative polygyny is ab-sent in honeybees, except for a brief period of possible motheredaughter coexistence during queen supersedure (Michener, 1974)or the initiation of an after-swarm, and in the unusual partheno-genetic and clonal Cape honeybee, Apis mellifera capensis (Martin,Beekman, Wossler, & Ratnieks, 2002).

The stingless bees have much higher variation in life historyand mode of resource acquisition than the honeybees (Michener,1974). However, it is also because swarms always contain a youngvirgin queen, rather than the old queen, that more variation inthe production and subsequent dispersal of young queens couldevolve (Peters, Queller, Imperatriz-Fonseca, Roubik, & Strass-mann, 1999). Swarm dispersal of single young queens impliesthat they are unlikely to compete more with sisters than withnonrelatives for the same distant nesting opportunities, consis-tent with, as far as we are aware, newly hatched sister queens ofstingless bees not having been observed to kill each other. Whenqueen killing happens, it is by workers culling surplus queensthat should have developed as workers but escaped controlbecause of mass provisioning and capping of brood cells (Bourke& Ratnieks, 1999; Wenseleers, Ratnieks, & Billen, 2003). However,stingless bee queens that do end up dispersing accompanied by aswarm of workers are expected to be equally territorial as younghoneybee queens once the swarm has left and not to toleratesister rivals in their incipient colony (Michener, 1974; Peters et al.,1999). The same applies to honeybees when colonies occasionallyproduce after-swarms (Seeley, 1985); these new queens are not athreat to the half-sister that inherited the colony earlier in theseason because they will not mate until after dispersal.

Stingless bee colonies often produce more virgin queens thanthey can endow with worker swarms. Also this appears logicalbecause these queens can disperse by themselves and compete foroccasional vacancies in unrelated colonies that have lost theirqueen, which justifies their production by the parental colony ifthey have at least some chance of succeeding (Van Oystaeyen et al.,2013). This leaves open the possibility that queens may compete forthe right to join a swarm, unless that is fully controlled by workerchoice so that directly competitive queens’ traits are useless.Hamilton (1964b) argued that full-sib relatedness will likely selectfor nonaggression between newly hatched queens, particularlywhen there are many, so that the marginal benefit of killing onerival while risking injury is low. In contrast, lone dispersal optionsare never available in honeybees, where a young queen only leavesthe company of her workers to go on her mating flight after anyswarms have left, to return to the same colony after a few hours,either the old colony where she was raised, or the new colony if shewas part of an after-swarm. Half-sib relatedness between newlyhatched queens and the possibility to kill rivals before they canfight back (Hamilton, 1964b) are additional arguments to expectqueenequeen aggression in honeybees. It is therefore not surpris-ing that facultative polygyny is occasionally found in stingless beesonly (Supplementary material), and that obligate polygyny couldnot evolve in any of the obligatorily eusocial swarm founders.

None of the arguments outlined so far apply to swarm-founding polistine wasps where caste roles are determined byadult phenotypic plasticity and where the nest-specific number ofegg-laying females may cycle but without allowing a single queenany long-term exclusive egg-laying rights (Queller, Strassmann,Solis, Hughes, & Deloach, 1993; Ross & Matthews, 1991). Neitherdo they apply to the army ants where swarming has occasionallybeen used as a term to describe colony fission, because dispersalis local (on foot). This breeding system thus retains elements oflocal resource competition, consistent with army ants beingmonogynous throughout, except for a single North Americanspecies at the northern edge of the army ant range that is so rarethat the colony survival benefits of secondary polygyny likelycame to surpass the local resource competition costs (Kronauer &Boomsma, 2007).

The Threat of Sexual Conflict

The higher termites are similar to the ants in social organizationand life history. Both clades are perennial and iteroparousthroughout, have wingless foragers and a number of lineages withimpressive nest architecture and huge colony sizes. However, thereare also fundamental differences in that the termites are hemi-metabolous and breed in lifetime monogamous pairs and the antsare holometabolous and have males that survive only as storedsperm. Any higher termite that would secondarily evolve faculta-tive polygyny would thus have to admit either sister queens andunrelated males, or brothers and unrelated virgin queens into thecolony, if it were to secure outbreeding. In contrast, ants can simplyadopt daughter queens that are inseminated by an unrelated malewithout having to accept that male himself. Importing fresh bloodthis way might be highly problematic for existing termite mounds,because termite males live as long as termite queens and have beenshaped by selection to mate with queens inside nests throughouttheir lives (Boomsma, 2007). Adopting them would thereforeunavoidably introduce some likelihood of remating promiscuitywith older queens inside existing colonies, the very trait that wascompletely abandoned in the lifetime monogamous ancestors ofthe higher termites that evolved true worker castes consisting ofindividuals that can no longer moult to become reproductives(Boomsma, 2009, 2013).

There is no a priori reason why the hypothetical polygyny/polyandry scenario outlined above could never have occurred intermites, but as far as multiple reproductives have been found, theyseem to be either sib-mating replacement reproductives or plei-ometrotic cofounders (i.e. examples of primary polygyny; Fig. 1; e.g.Hartke & Rosengaus, 2013; Kobayashi et al., 2013). This suggests thatthe reduction in offspring relatedness below the lifetime monoga-mous value of exactly 0.5 thatwould follow in thewake of acceptingnew reproductives makes secondary polygyny after outside matingflights a nonstarter in termites (Boomsma, 2013). This may bebecause the true (sterile) workers that emerge via hemimetabolousdevelopmental pathways would come under selection to abandontheir commitment to lifetime sterility as adults when the indirectfitness gains from raising siblings would no longer be equal to thedirect fitness of raising offspring. As higher termites are central-place foragers rather than log-nesters (Inward et al., 2007) such areintroduction of sexual conflict might be so profoundly damagingfor competitive performance against permanently monogamousneighbouring colonies that no developments towards secondarypolygyny have apparently been maintained over evolutionary time.As it seems, however, colonymergersmay occasionally occur also inhigher termites, which could represent exceptions to this argument(Adams, Atkinson, & Bulmer, 2007; Hartke & Rosengaus, 2013;Vargo & Husseneder, 2011).

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252246

SPECIAL ISSUE: KIN SELECTION

Positive Selection for Secondary Polygyny in Ants

We have summarized above the three general factors that mayhave precluded secondary polygyny, particularly its obligate form,but the question remains whether there have also been generalselection forces to promote secondary polygyny that might haveapplied only in the ants. It has been suggested that low survivalduring queen dispersal, due for example to low availability ofsuitable nest sites elsewhere, has been such a general force (e.g.Bourke & Franks, 1995; Bourke & Heinze, 1994; Heinze & Foitzik,2009; several chapters in Keller, 1993), but this argument seemsconceptually problematic because it tends to ignore the significantdownside of local sibling competition that wemet above. This logicis the core of Hamilton and May (1977) and consistent with theirmathematical evidence that local resource competition betweenrelatives is such a powerful force that natural selection can reducetendencies to disperse but never eliminate them. A later review(West et al., 2002) further detailed this insight, showing that sibcompetition exactly cancels relatedness incentives for local coop-eration. It thus appears that dispersal limitations per se are unlikelyto explain the prevalence of secondary polygyny in ants and the factthat these polygynous queens almost always coexist peacefully(Bourke & Franks, 1995).

We propose that options for short-distance colony budding mayhave provided the general selection regime promoting secondarypolygyny. Ants have their eggs and brood in piles so that immaturestages can easily be moved around, whereas obligatorily eusocialbees and wasps have their brood in fixed comb structures withcells. Ants further forage on foot in largely two-dimensionalhabitat, so that bud nests will be additional central-place foragingunits (Bourke & Franks, 1995) that are likely to increase a colony’stotal resource base in parallel with having more mouths to feed.Reduced dispersal will therefore not imply increased competitionwith relatives if resource requirements and foraging territory in-crease at the same rate (Gardner, Arce, & Alpedrinha, 2009). Pro-portionality of this kind is most likely when the habitat ishomogeneous, which is indeed the general rule both for obligato-rily polygynous ants and for populations of normally monogynousants that switch to predominant polygyny (Bourke & Heinze, 1994;Cronin et al., 2013; Heinze, 1993; Sundström et al., 2005). As long assuch habitats are unsaturated for the particular ant species thathappens to colonize and later monopolize it, selection will favourfemale-biased sex ratios (Gardner et al., 2009). However, antsupercolonies are usually found and studied when they havesaturated the available habitat, in which case reproductive successis likely maximized by predominantly producing dispersing males,as is often observed (e.g. Gardner et al., 2009; Holzer, Keller, &Chapuisat, 2009; Kümmerli, Helms, & Keller, 2005; Van derHammen, Pedersen, & Boomsma, 2002).

These benefits of colony budding are unavailable to eusocialcorbiculate bees and vespine wasps because (1) foraging territoriesare orders of magnitude larger than those of ants and fundamen-tally heterogeneous in three dimensions, so they will alwaysoverlap, implying that bud nests nearby will increase competitionwith relatives without enhancing resource availability (Boomsma,Schmid-Hempel, et al., 2005; Bourke & Franks, 1995), and (2) cav-ity nesting will often imply that no suitable nesting opportunitiesare available nearby (Archer, 2012; Matsuura,1991; Michener,1974;Seeley, 1985). However, higher termites are wingless, two-dimensional central-place foragers similar to ants and known tohave evolved nest budding in some lineages (Vargo & Husseneder,2011), but the reproductives dispersing to these bud nests areinbreeding full siblings. Explaining this form of polygyny is notreally challenging, as incestuously reproducing replacementreproductives are equivalent to a recombined clonal extension of

the colony germ-line that was established by the outbred foundingpair (Boomsma, 2009).

SECONDARY POLYGYNY INTRODUCES MODULARITY IN THEGERM-LINES OF EUSOCIAL COLONIES

When eusocial insect queens are singly inseminated and foundcolonies of full-sib workers that are 100% sterile, there is a perfectanalogy with metazoan bodies that start with a zygote, consistingof lifetime-committed gametes, and differentiate into a germ-lineand diverse somatic tissues (Boomsma, 2009; Bourke, 2011;Wheeler, 1928), while maintaining what Hamilton (1972)described as “the disciplined co-operation of the two genomes ofthe diploid cell”. However, whenworker castes aremerely unmatedbut retain their ovaries, as often happens in the eusocial Hyme-noptera, the colony continues to have elements of modular repro-duction. The contrast between unitary (metazoan) growth andmodular growth (e.g. plants, fungi, lineages of multicellular algae)is very fundamental (Bourke, 2011; Buss, 1987; Fisher, Cornwallis, &West, 2013), but has not been elaborated in much detail for itspossible analogies in the eusocial insects (but see Queller, 2000).

Haplodiploidy implies that worker sons are analogous to maleflowers on a body that otherwise reproduces as an animal(Boomsma, 2009), but producing worker sons does not change theprinciple of unitary colony growth as the males disperse like thepollen of male flowers do. All it means is that somatic differentia-tion is not completely terminal and still allows some gene trans-mission to future generations (Buss, 1987). However, secondarypolygyny changes this picture in a fundamental way, because thecolony will now contain mated sisters of workers who have shedtheir wings and can disperse on foot to empower budding nestswith their egg-laying capacity. Such newly adopted inseminatedqueens therefore both introduce modular growth (when the ‘so-matic’ workers use them to found bud nests) and recurrentchimerism of the colony via the newly imported, unrelated storedsperm that fertilizes their eggs. Colony budding is fundamentallydifferent from colony fission in honeybees, stingless bees and armyants as discussed in the previous section. Here, dispersal precedesthe insemination of young queens so chimerism is avoided, exceptfor a brief period in which honeybee workers may coexist with areplacement queen, a situation in which they also shift to lessaltruistic behaviours (Woyciechowski & Kuszewska, 2012). Swarm-founding Provespa appear to be an interesting exception as queensare already inseminated when they leave with their swarm (Archer,2012; Matsuura, 1991), but here the annual life cycle will precludelasting chimerism via recurrent queen adoption.

Metazoan bodies are normally protected from any later de-velopments towards chimerism by a merciless immune systemimposing very strict selfenon-self-discrimination (Bourke, 2011),and analogous nestmate recognition mechanisms apply in themany eusocial Hymenoptera where workers maintain obligatemonogyny (Lenoir, Fresneau, Errard, & Hefetz, 1999). Suchdiscrimination also seems a universal trait in the higher termites(Thorne & Haverty, 1991), which may recycle genes by incestuouslyproducing replacement reproductives, but appear to never admitnew reproductive in the way polygynous ants do (Boomsma, 2007,2009; see also previous sections). The evolution of facultativesecondary polygyny in the eusocial Hymenoptera thus represents amajor recognition transition as it relaxes this form of absolutediscrimination (Bourke & Franks, 1991) and thereby the exclusivefoundress right to colony ownership and at least the female part ofits germ-line function. However, only high degrees of secondarypolygyny came to really challenge the unitary germ-line function ofthe colony founding pair, and those elaborations appear to berestricted to the ants where colony budding may offer fitness

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252 247

SPECIAL ISSUE: KIN SELECTION

compensation in homogeneous habitats (see previous section andnotes in the Supplementary material for Fig. 2).

The problem that remains is that the inclusive fitness interestsof the old queen(s), the young queens and the workers are onlypartly aligned, as budding can only happen after further colonygrowth, and queens may disagree about staying or leaving with agroup of budding workers (Bourke & Franks, 1995; Nonacs, 1988;Pamilo, 1991). It is here that chimerism reveals its evolutionaryJanus-face that makes colony budding inherently unstable as asocial construct. While vascular plants with modular growth areclones that produce new germ-line function (flowers) after meri-stematic growth without these new germ-lines having any say inthe matter, secondary queens need to actively decide or be coercedto join budding nests. Polygynous colonies and their workers havethus become a public good that provides services, but that also canbe exploited for personal gain. Buss (1987) evaluated a remarkableparallel when he discussed the diversity of late ontogeny inmetazoans, highlighting that fully distinct somatic cells of somesponges are still able to differentiate into sperm or eggs. In clonalmetazoan bodies such diversity has generally been interpreted asundesirable for stable development, because it completely dis-appeared in the more advanced lineages of the animal phylogeny(Buss, 1987). The evolutionary stability of secondary polygyny thusseems to have many evolutionary odds against it.

SECONDARY POLYGYNY AND THE EVOLUTION OF INQUILINECHEATER MORPHS

The ants are not only unique in being the only eusocial lineagethat has repeatedly evolved obligate secondary polygyny, but theyare also the only clade that has a huge variety of inquiline socialparasites that gain access to host colonies by stealth rather thanovert aggression (Bourke & Franks, 1991; Buschinger, 1986, 2009;Hölldobler & Wilson, 1990). Similar to many aggressive usurpers,these inquilines have been hypothesized to have evolved fromancestors that practised intraspecific social parasitism, but only thestealthy inquilines have been linked to secondary polygyny as aprecursor state (Bourke & Franks, 1991, 1995; Buschinger, 1986,2009). This underlines that stealthy inquilines are fundamentallydifferent from aggressive social parasites that kill host queenseither directly or after a short period of coexistence. The latter usurplone colony founding females or queenless nests and have evolvedboth in the ants and in facultatively and obligatorily eusocial beesand wasps, while sharing with the stealthy inquilines the habit ofoften using their closest relatives as hosts (Bourke & Franks, 1991,1995; Buschinger, 1986, 2009; Wcislo, 1987). This phenomenon isknown as Emery’s rule (Hölldobler & Wilson, 1990) or Popov’s rule(Wcislo, 1987), which is increasingly interpreted to imply thatspecies pairs of social parasite and host most likely emerged viasympatric speciation (Bourke & Franks, 1995; Buschinger, 1986,2009; Savolainen & Vepsäläinen, 2003; Wcislo, 1987).

The results of our comparative analysis (Fig. 2) are consistentwith obligate or predominant secondary polygyny (rather thanfacultative polygyny; Bourke & Franks, 1991) having been a neces-sary condition for the evolution of stealthy inquiline social para-sites, as these appear to have never been found in the perennialsocieties of honeybees, stingless bees, and termites where facul-tative polygyny can occur but obligate polygyny is absent (Fig. 2).This does not imply that such inquiline species cannot maintainthemselves in host populations where secondary polygyny is onlyfacultative, but merely indicates that stealthy inquilines had littleopportunities to specialize and become reproductively isolatedunless host polygyny was obligate or at least highly prevalent forvery long time spans. The honeybees and stingless bees also appearto lack aggressive usurper social parasites, which is likely related to

founding queens being protected by their worker swarm. The samelack of social parasites apparently applies to the swarm-foundingpolistine wasps as well (Hamilton, 1972). The termites may haveother smaller termite species occupying cavities in their nestmounds, but appear completely devoid of Emery’s-rule-type socialparasites, both usurpers and stealthy ones (Wilson,1971). However,the vespine wasps and bumblebees have, respectively, 4% and 11%usurpers (Archer, 2012; Fisher, 1987), and the overall percentage ofsocial parasites (including both categories) in ants is about 2% (230/12500; Buschinger, 2009), but many inquilines probably remain tobe discovered. In all clades the number of independent evolu-tionary origins of usurping social parasites seems limited, but theknown ones have often radiated into separate genera. This seemsopposite for the stealthy inquilines of ants, which have evolvedmany times independently, but tend to show very little subsequentradiation.

To understand obligate secondary polygyny as a preconditionfor the evolution of stealthy social parasitism, it is relevant to brieflyelaborate the fundamental reproductive conflicts that characterizethis breeding system (Bourke & Franks, 1991, 1995; Hölldobler &Wilson, 1977; Nonacs, 1993; Pamilo, 1991; Rosengren, Sundström,& Fortelius, 1993). It is instructive to start at the opposite end andtake a group selection angle that would deny any such conflicts andemphasize that secondary polygyny followed by nest buddingshould be an honest queen succession and nest inheritancemechanism to maximize population-level productivity. We haveillustrated this scenario as a thought experiment in Fig. 3, assumingthat colonies are founded by a single queen, live for 6 years, butneed continuous recruitment of new queens from the second yearonwards because queens die after 2 years. Workers are assumed tolive 1 year and to be in control of the queen admission procedureafter each year’s nuptial flight (Bourke & Franks, 1995; Kümmerli &Keller, 2008; Nonacs, 1988; Pamilo, 1991). To keep the examplesimple we have not included nest budding, but that could be asurplus reproductive activity in years when colonies adopt morequeens than usual (e.g. six rather than the usual four in the exampleof Fig. 3). This null model was inspired by Hamilton (1964b), whoquotes Wynne-Edwards (1962, p. 653) for ‘the widespread practiceof attacking and persecuting strangers and relegating newcomersto the lowest social rank’; in this case the production of sterileworker helpers by newly adopted queens.

The group selection null model of Fig. 3 has all the characteristicsof a public good (the rearing services of older workers) that can beexploited for personal rather than group benefits by co-breedingqueens that are supposed to produce workers only. At least fourtypes of selfish behaviour will likely be selected for and be difficultto police for workers. First, more queens will want to be adoptedthan is beneficial for the colony in an average year. This idea wasoriginally developed by Elmes (1973) for Myrmica ants, but maywell apply also in other obligatorily polygynous breeding systems(Bourke & Franks, 1991). Second, once secondary queens areadopted, they may be selected to live longer than the 2 years of theharmonious group selectionmodel, particularly if they can continueto produce reproductives rather than workers. Third, queens maybe selected to cut down on worker production in their first year,anticipating that other queens will produce these workers and theycan somehow reallocate the saved resources towards reproductionlater in life (Bourke & Franks, 1991; Herbers, 1993; Rosengren et al.,1993). Fourth, we have so far implicitly assumed that queens willseek adoption in the colony inwhich they were raised (after matingclose by), but once a number of these cheating traits have evolved,queens would obtain larger fitness gains when they manage to beadopted in unrelated nests, avoiding that their relatives pay thecosts of their exploitation of the public good services of theworkers(Nonacs, 1993; Rosengren et al., 1993; Rüppell & Heinze, 1999).

Year 1 Year 2 Year 3 Year 4 Year 5 Year 6

Daughterworkers

Daughter gynes

Successivemother queens

Dispersing gynes

Figure 3. A simple ‘honest succession’ null model of obligate secondary polygyny. Queens are assumed to live 2 years and workers a single year. All queens produce workers in theirfirst year of tenure and reproductives (virgin queens) in their second year, so the colony becomes polygynous after the first year when its workers begin to adopt newly mated gynedaughters as additional queens. Each new queen is assumed to have stored the sperm of a single unrelated male, so worker relatedness to new queen offspring is 50% of theirrelatedness to these queens themselves. During the 6 years of the colony’s life, worker relatedness is thus diluted every year relative to what it was in the founding year (the shiftingshades towards lighter grey). As workers are shorter-lived than queens, there will always be a mix of two worker cohorts that are approximately twice as related within cohorts asbetween cohorts. When queens produce workers in their first year and virgin queens only in their final year, all workers will have equal and maximal (given the rules of this thoughtexperiment) inclusive fitness because they will always raise the sexual offspring of their mother and her sisters rather than the offspring of their own sisters to which they are lessrelated. How many new queens a colony adopts each year seems of little overall importance, but we have assumed that the number will be similar to that of colony life span (i.e. onaverage, 4 queens/year; range 2e6 queens).

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252248

SPECIAL ISSUE: KIN SELECTION

The above account illustrates that obligate secondary polygynyis fundamentally susceptible to being corrupted by selfish behav-iours of the co-breeding queens. Cheating is thus likely to berampant, consistent with many lineages having specialized on asocially parasitic life history that later became reproductively iso-lated so that inquilines that use stealth rather than aggression couldemerge according to the logic of Emery’s rule (e.g. Bourke & Franks,1991; Savolainen & Vepsäläinen, 2003; Schultz, Bekkevold, &Boomsma, 1998). Over evolutionary time, species with high andoccasionally obligate polygyny would be particularly susceptible tocheating queen morphs, as facultatively polygynous populationshave retained the option of realizing full reproductive potential inmonogynous colonies, so selection will start favouring monogynyas soon as stealthy inquiline tendencies arise. However, thisvulnerability does not mean that workers of adopting polygynouscolonies are powerless in reducing the ill effects of being exploitedby newly adopted queens. They can correct adoption mistakes byculling supernumerary queens (Keller, Passera, & Suzzoni, 1989), bymaking aggression against (or neglect of) queens conditional onlow egg-laying rates in the first year (Evesham, 1984), or they mayterminate old queens whose relatedness to younger workers hasbecome too low, all topics that remain greatly understudied.

Recent work on the polygynous ant Formica fusca (Bargum &Sundström, 2007) showed that old queens are more likely tosire reproductive brood than young queens, but that youngqueens raise their own worker brood, which may help them toproduce reproductives the year after, consistent with the type ofqueen turnover illustrated in Fig. 3. Queens having higher sharesin reproduction were also found to be more closely related to theworkers raising the brood, particularly in the critical sexualbrood-rearing season, which might imply that workers practisenepotistic discrimination (Hannonen & Sundström, 2003a, 2003b).Another study of the same ant (Ozan, Helanterä, & Sundström,2013) showed that queens somehow manage to communicatetheir identity to the workers that raise their brood, and thatworkers may allow queens to which they are more related tooviposit early in the season, which turns a higher fraction of theirfemale larvae into virgin queens rather than workers. At the sametime, workers allocated more efforts towards direct fitness ben-efits via the production of worker sons when interacting with a

lowly related queen than when they nursed the brood of a highlyrelated queen.

The possible occurrence of nepotistic discrimination in preciselythese types of eusocial breeding systems is interesting, becausenepotism is all but absent in the eusocial domain (Boomsma &d’Ettorre, 2013; Keller, 1997; Thorne & Haverty, 1991). It might bethat intermediate but highly variable relatedness due to frequentqueen turnover offers the right set of conditions for nepotism to bemaintained in F. fusca, similar to what is expected and found incooperative-breeding vertebrates (Cornwallis, West, Davis, &Griffin, 2010; Griffin & West, 2003) and social bacteria (West,Griffin, Gardner, & Diggle, 2006). Work on other Formica antsfurther showed that new queens are produced and recruited innests where queen number, and thus local resource competition, islow (Brown & Keller, 2002; Fortelius, Rosengren, Cherix, &Chautems, 1993; Kümmerli & Keller, 2008), consistent with theidea that high levels of polygyny are characterized by workersmanaging their adoption decisions to maximize inclusive fitness(for another example see Fernandez-Escudero, Seppä, & Pamilo,2001), in spite of difficulties in protecting the commons of thenest, as predicted by earlier models (Nonacs, 1988; Pamilo, 1991).Finally, F. fusca is host to a large number of usurping social parasites(Czechowski, Radchenko, & Czechowska, 2002), which may alsohave contributed to workers having unusually well-developeddiscrimination skills.

REDUCTION OF QUEEN LIFE SPAN UNDER SECONDARYPOLYGYNY

Selection on queen life span is not expected to change muchunder facultative polygyny, as this retains monogyny as a valid andoften dominant alternative mode of breeding in the same popula-tion, but this is no longer truewhen polygyny becomes dominant orobligate. As stated earlier, the border-line between facultative andobligate polygyny may not always be sharp, but the distribution ofqueen numbers across andwithin species tends to be bimodal, withfacultative polygyny usually implying that only some coloniessometimes adopt secondary queens and obligate polygyny, mean-ing that colonies are essentially unable to grow big enough toreproducewithout havingmultiple queens (Sundström et al., 2005;

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252 249

SPECIAL ISSUE: KIN SELECTION

see also Bourke & Franks,1991). Consistent with this dichotomy, thepractical separation between facultative and obligate polygyny inour comparative analysis was reasonably straightforward in spite ofcase studies usually not offering precise quantitative data on queennumber or relatedness (i.e. there were rather few cases that wecould not assign to one of the two categories with reasonableconfidence; see Fig. 2, Supplementary material).

Once polygyny has become dominant, selection for shorterqueen life span may follow rather automatically because the moresterile workers the new queens contribute in their first year afteradoption (Bourke & Heinze,1994; Ozan et al., 2013), the faster thesenew cohorts will outnumber the older workers. This will erodeloyalty to the previous generation of queens, possibly via scenariosreminiscent of a ‘virtual dominant’, an abstract shifting breedingentity to which group members have maximal average relatedness(Reeve & Jeanne, 2003). The ant Cardiocondyla obscurior hasrecently been shown to have shorter-lived queens in polygynouscolonies and longer-lived queens in sympatric monogynous col-onies (Schrempf, Cremer, & Heinze, 2011), suggesting that adjust-ments in queen life span can remain phenotypically plastic whenpolygyny varies. It may even pay to make workers smaller so moreof them can be produced for the same investment and faster,consistent with observations that workers and queens often tend tobe somewhat smaller in polygynous populations relative tomonogynous ones (Bourke & Franks, 1995; Elmes & Keller, 1993;Rüppell & Heinze, 1999).

While seemingly altruistic worker production in the first yearafter adoptionmay secure colony productivity (and solve one of thepotential cheating problems outlined in the previous section), italso increases every queen’s risk of being ‘dismissed’ earlier byyounger worker cohorts. This will relax selection for somatic repairin queens, accelerating rates of ageing (Stearns, 1992) and thusdecrease the need to store maximal numbers of sperm. Acrossspecies this would lead to the expectation that average queennumber per colony is negatively correlated with average queenlongevity and fecundity (Keller & Vargo, 1993) when adjusted forqueen life span in monogynous sister clades. Thus, in monogynouscolonies (the ancestral state of obligate eusociality) the life spandifference between workers and queens will gradually increasewith increasing colony size (i.e. queens will age more slowly andworkers relatively faster; Keller & Genoud, 1997; Kramer &Schaible, 2013; Parker, 2010), but this trend should be reversedunder obligate polygyny. This yields the expectation that queensshould be relatively short-lived in ant species that are highlypolygynous in all populations, but not in ants that have bothmonogynous and polygynous populations.

The interaction between inclusive fitness theory and evolu-tionary theories of ageing has recently been reviewed by Bourke(2007), but applying these principles to obligatorily polygynousants is complex because the interests of workers and secondaryqueens are unlikely to be aligned (Nonacs, 1988; Pamilo, 1991).Workers would benefit from having newly adopted queen co-horts producing many new workers to help raise the sexualoffspring of their mother and her sister queens, even though thiswill likely shorten the reproductive tenure of their mother.Queens on the other hand would be selected to survive andreproduce longer than is in the joint interest of the workers andto evolve strategies intended to make workers believe theybelong to a later cohort (Herbers, 1993; Nonacs, 1993; Rosengrenet al., 1993). However, more general life-history theories ofageing (Heinze & Schrempf, 2008) have to our knowledge notbeen applied in any explicit detail to the secondary polygynouseusocial insects since the initial survey by Keller and Genoud(1997). We believe that any such conceptual developmentsshould be based on elaborations of the seminal model by Charnov

and Schaffer (1973), who showed that selection for perenniality(and by implication iteroparous reproduction) depends not onabsolute mortality rates, but on the ratio between mortality inthe juvenile and adult stage.

When eusocial colonies are claustrally founded and remainmonogynous, colony life span is equal to queen life span and thetransition from the juvenile to the adult stage will then, as in allother organisms, start with sexual maturity (Stearns, 1992); that is,the end of the so-called ergonomic phase of purely somatic colonygrowth (Oster & Wilson, 1978). Such colonies are founded afterrisky mating flights and have highly vulnerable phases of initialcolony growth (Clark & Fewell, 2014; Oster & Wilson, 1978), so ju-venile mortality will always be very high. This implies that iter-oparity could probably evolve only after nests had becomefortresses against predators and diseases (Queller & Strassmann,2002) by the time reproduction started, consistent with cavitynesting in honeybees and stingless bees and quite possibly deeper-soil nesting in the ancestors of ants that made the transition toperennial eusociality early on (Boomsma, 2009; Johnson et al.,2013), similar to the termites. The most spectacular elaborationsof iteroparity based on single queen tenure are found in the antsand termites (Keller & Genoud, 1997), in comparison to whichqueen life span in honeybees and stingless bees is modest at best.This may be related to mortality of mature colonies of honeybees,stingless bees, ants and termites being of similarly low magnitude,whereas dispersal by swarming implies considerably reducedmortality of young colonies in honeybees and stingless bees(Cronin et al., 2013; Michener, 1974; Seeley, 1985). The ratio be-tween the two mortality rates would thus be increased and selec-tion for extremely long queen life spans weakened in these beelineages. This is consistent with queens of army ants being onlymoderately long-lived (Franks & Hölldobler, 1987) as their shorter-distance dispersal by colony fission also decreases mortality ofyoung colonies.

Using the same logic for obligatorily polygynous ants is moreinvolved, but seems to allow at least some sensible inferences.Mating close to the nest should reduce juvenile mortality and beingadopted rather than having to found a colony independentlyfurther reinforces that trend even though becoming part of a smalland possibly vulnerable bud nest may partly reverse this. Adult nestmortality, on the other hand may either be the same as inmonogynous ants, or different depending on colony size varyingbetween a couple of thousand ants in, for example,Myrmica species(Elmes & Keller, 1993) and Formica fusca, and millions in somepolydomous Formica species (Rosengren & Pamilo, 1983). Thissuggests that reductions in queen life span may be most pro-nounced in polygynous ants with small colonies.

There is a final fundamental issue about queen life span andageing that we would like to touch upon. It would seem obvious tothe field ecologist that monogynous eusocial queens combine highiteroparous fecundity with a long life span, but in fact it is not. AsCharnov and Schaffer (1973) and later life-history studies havenoted (Stearns, 1992), there usually is a trade-off (i.e. negativecorrelation) between fecundity and life span, so that long-livedanimals tend to be less fecund. However, the reverse applies inhoneybees (Remolina & Hughes, 2008) and in two closely relatedCardiocondyla ant species (Heinze, Frohschammer, & Bernadou,2013; Heinze & Schrempf, 2012; Schrempf et al., 2011). The sepa-ration between obligate and facultative individual commitment toeusocial caste that we applied throughout this review may offer anexplanation for this phenomenon.When lifetime unmated workersfarm physically differentiated queens as egg-laying ‘organ(s)’ of acolony, producing only the queens needed, and have them matewith local males so they never leave the colony, it seems highlyunlikely that there is a cost-of-reproduction trade-off. In metazoan

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252250

SPECIAL ISSUE: KIN SELECTION

bodies, germ-line longevity is normally positively correlated withsomatic longevity, and permanently eusocial colonies wherebreeder and worker castes can never realize any reproductivesuccess without each other appear to offer an analogous situation.No such positive correlations would be expected for queens ofpaper wasps or (in)vertebrate cooperative breeders, whose repro-ductive efforts should remain linked to aggressive dominance,which is more likely to trade-off against life span.

CONCLUSIONS

We have argued that the study of multiqueen breeding in ants,bees and wasps with eusocial colonies gains in transparency whenlineages with permanent, morphologically differentiated castes areclearly distinguished from those having castes that are eitherbehavioural or based on phenotypic plasticity after reaching theadult stage (Beekman et al., 2006; Boomsma, 2013; Crespi &Yanega, 1995). The rationale is that it matters whether femalereproductive success is determined directly in dominance hierar-chies or indirectly by lifetime unmated workers who essentially‘farm’ a shifting group of co-breeding queens. Maintaining thisseparation makes it easier to see the convergent ways in which kinselection has shaped breeding in eusocial colonies on either side ofthe lifetime monogamy window that marks the origins of obligatelifetime unmatedness of worker castes and their physical differ-entiation. Whether or not workers are often or always lifetimecommitted to unmatedness makes a fundamental difference,because they become obligate ‘soma’ for a single or multiple queengerm-line(s) by default in the latter case, but not in the former.

So where, if anywhere, is secondary polygyny heading in alonger-term evolutionary perspective? Our, admittedly crude,comparative analyses suggest that facultative polygyny can hardlyevolve in the higher termites and honeybees and that it remains arelatively marginal add-on to the life span of existing colonies insome bumblebees, vespine wasps and stingless bees. The antsappear to be the champions of this type of social system, as bothfacultative and obligate polygyny are common in populations ofmany species. Shifting to polygyny carries many potential costs, butthe unique ecology of ants seems to provide a number ofcompensating benefits that often surpass these costs. These ad-vantages appear to be ultimately related to workers foraging onfoot in two-dimensional habitat and to adopted queens beinglifetime inseminated, which is not possible in termites. It seemsunlikely, however, that the long-term ecological benefits ofpolygyny continue to increase when queen numbers keep rising.Although many ant lineages have achieved such high degrees ofpolygyny that within-colony relatedness among workers ap-proaches zero, these lineages appear to be evolutionary dead ends(Helanterä et al., 2009) and to have modest ecological footprintsunless humans start vectoring them to novel habitats where they,likely due to their specific life-history characteristics (Pedersen,2012), can become large-scale invasive pests (Rabitsch, 2011).

As it seems then, secondary polygyny is an evolutionarily suc-cessful innovation only when it remains both moderate and vari-able in degree even when essentially all nests in a population are(potentially) polygynous. Examples of such variation, among spe-cies, among conspecific populations, and among nests withinpopulations, abound in well-studied ant genera such as Formica,Myrmica, Lepthothorax/Temnothorax, Cardiocondyla, Crematogasterand probably across other ant genera as well. The swarm-foundingpolistine wasps appear to be the only other lineage that has made asimilar success of moderate but obligate polygyny but, as weargued, for fundamentally different reasons because they remainassociations of reproductively totipotent adult females whereasevery female ant is committed to a mated-breeder or an unmated-

helper status for life. Polygynous ants and swarm-founding polis-tines intrigued Bill Hamilton. They were discussed at some lengthin his seminal inclusive fitness papers (Hamilton, 1964a, 1964b,1972) and it remains truly remarkable how few hard data weresufficient for him to initiate the most fundamental innovation ofthe Darwinean paradigm in the 20th century. Fifty years later, theseobservations remain powerful illustrations of how his gene’s eyeview of evolution through natural selection is indispensable forunderstanding the general principles of social evolution. However,interpreting the kin selection forces that have shaped divergentpolygynous systems in the social Hymenoptera becomes morestraightforward when acknowledging that they belong to differentdomains of social evolution where the same inclusive fitnessprinciples produce different adaptive end points.

Acknowledgments

Wewere supported by the Danish National Research Foundation(grant DNRF57) and thank Tamara Hartke, Jürgen Heinze, DaveQueller, Alexandra Schrempf, Lotta Sundström, Bill Wcislo and ananonymous referee for discussion and constructive comments.

Supplementary Material

Supplementary material for this article is available, in the onlineversion, at http://dx.doi.org/10.1016/j.anbehav.2014.03.005.

References

Adams, E. S., Atkinson, L., & Bulmer, M. S. (2007). Relatedness, recognition errors,and colony fusion in the termite Nasutitermes corniger. Behavioral Ecology andSociobiology, 61, 1195e1201.

Archer, M. E. (2012). Vespine wasps of the world [Siri Scientific Press MonographSeries]. Manchester, U.K.: Siri Scientific Press.

Bargum, K., & Sundström, L. (2007). Multiple breeders, breeder shifts and inclusivefitness returns in an ant. Proceedings of the Royal Society B: Biological Sciences,274, 1547e1551.

Beekman, M., Peeters, C., & O’Riain, M. J. (2006). Developmental divergence:neglected variable in understanding the evolution of reproductive skew insocial animals. Behavioral Ecology, 17, 622e627.

Bernasconi, G., & Strassmann, J. E. (1999). Cooperation among unrelated in-dividuals: the ant foundress case. Trends in Ecology & Evolution, 14, 477e482.

Boomsma, J. J. (2007). Kin selection versus sexual selection: why the ends do notmeet. Current Biology, 17, R673eR683.

Boomsma, J. J. (2009). Lifetime monogamy and the evolution of eusociality. Philo-sophical Transactions of the Royal Society B: Biological Sciences, 364, 3191e3207.

Boomsma, J. J. (2013). Beyond promiscuity: mate-choice commitments in socialbreeding. Philosophical Transactions of the Royal Society B: Biological Sciences,368, 20120050. http://dx.doi.org/10.1098/rstb.2012.0050.

Boomsma, J. J., Baer, B., & Heinze, J. (2005). The evolution of male traits in socialinsects. Annual Review of Entomology, 50, 395e420.

Boomsma, J. J., & d’Ettorre, P. (2013). Nice to kin and nasty to non-kin: revisitingHamilton’s early insights on eusociality. Biology Letters, 9, 20130444.

Boomsma, J. J., Schmid-Hempel, P., & Hughes, W. O. H. (2005). Life histories andparasite pressure across the major groups of social insects. In M. D. E. Fellowes,G. J. Holloway, & J. Rolff (Eds.), Insect evolutionary ecology: Proceedings of theRoyal Entomological Society’s 22nd symposium (pp. 139e175). Wallingford, U.K.:CABI.

Bourke, A. F. G. (2007). Kin selection and the evolutionary theory of aging. AnnualReview of Ecology Evolution and Systematics, 38, 103e128.

Bourke, A. F. G. (2011). Principles of social evolution. Oxford, U.K.: Oxford UniversityPress.

Bourke, A. F. G., & Franks, N. R. (1991). Alternative adaptations, sympatric speciationand the evolution of parasitic, inquiline ants. Biological Journal of the LinneanSociety, 43, 157e178.

Bourke, A. F. G., & Franks, N. R. (1995). Social evolution in ants [Monographs inBehavior and Ecology]. Princeton, NJ: Princeton University Press.

Bourke, A. F. G., & Heinze, J. (1994). The ecology of communal breeding: the case ofmultiple-queen leptothoracine ants. Philosophical Transactions of the Royal So-ciety B: Biological Sciences, 345, 359e372.

Bourke, A. F. G., & Ratnieks, F. L. W. (1999). Kin conflict over caste determination insocial Hymenoptera. Behavioral Ecology and Sociobiology, 46, 287e297.

Brown, W. D., & Keller, L. (2002). Queen recruitment and split sex ratios in polyg-ynous colonies of the ant Formica exsecta. Ecology Letters, 5, 102e109.

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252 251

SPECIAL ISSUE: KIN SELECTION

Buschinger, A. (1974). Monogyny and polygyny in insect societies. In G. H. Schmidt(Ed.), Sozialpolymorphismus bei Insekten: Probleme der Kastenbildung im Tierreich(pp. 862e896). Stuttgart: Wissenschaftliche Verlagsgesellschaft.

Buschinger, A. (1986). Evolution of social parasitism in ants. Trends in Ecology &Evolution, 1, 155e160.

Buschinger, A. (2009). Social parasitism among ants: a review (Hymenoptera:Formicidae). Myrmecological News, 12, 219e235.

Buss, L. W. (1987). The evolution of individuality. Princeton, NJ: Princeton UniversityPress.

Cardinal, S., & Danforth, B. N. (2011). The antiquity and evolutionary history of socialbehavior in bees. PLoS One, 6, e21086.

Cardinal, S., Straka, J., & Danforth, B. N. (2010). Comprehensive phylogeny of apidbees reveals the evolutionary origins and antiquity of cleptoparasitism. Pro-ceedings of the National Academy of Sciences of the United States of America, 107,16207e16211.

Charnov, E. L., & Schaffer, W. M. (1973). Life-history consequences of natural se-lection: Cole’s result revisited. American Naturalist, 107, 791e793.

Choe, J. C., & Crespi, B. J. (Eds.). (1997). Social behavior in insects and arachnids.Cambridge, U.K.: Cambridge University Press.

Clark, R. M., & Fewell, J. H. (2014). Transitioning from unstable to stable colonygrowth in the desert leafcutter ant Acromyrmex versicolor. Behavioral Ecologyand Sociobiology, 68, 163e171.

Cornwallis, C. K., West, S. A., Davis, K. E., & Griffin, A. S. (2010). Promiscuity and theevolutionary transition to complex societies. Nature, 466, 969e972.

Crespi, B. J., & Yanega, D. (1995). The definition of eusociality. Behavioral Ecology, 6,109e115.

Cronin, A. L., Molet, M., Doums, C., Monnin, T., & Peeters, C. (2013). Recurrentevolution of dependent colony foundation across eusocial insects. Annual Re-view of Entomology, 58, 37e55.

Crozier, R. H., & Pamilo, P. (1996). Evolution of social insects colonies: Sex allocationand kin selection. Oxford series in ecology and evolution. Oxford, U.K.: OxfordUniversity Press.

Czechowski, W., Radchenko, A., & Czechowska, W. (2002). The ants (Hymenoptera,Formicidae) of Poland. Warszawa: MIZ PAN.

Elmes, G. W. (1973). Observations on density of queens in natural colonies ofMyrmica rubra L (Hymenoptera: Formicidae). Journal of Animal Ecology, 42,761e771.

Elmes, G. W., & Keller, L. (1993). Distribution and ecology of queen number in antsof the genus Myrmica. In L. Keller (Ed.), Queen number and sociality in insects(pp. 294e307). Oxford, U.K.: Oxford University Press.

Evesham, E. J. M. (1984). Queen distribution movements and interactions in a semi-natural nest of the ant Myrmica rubra L. Insectes Sociaux, 31, 5e19.

Fernandez-Escudero, I., Seppä, P., & Pamilo, P. (2001). Dependent colony founding inthe ant Proformica longiseta. Insectes Sociaux, 48, 80e82.

Fisher, R. M. (1987). Queeneworker conflict and social parasitism in bumble bees(Hymenoptera: Apidae). Animal Behaviour, 35, 1026e1036.

Fisher, R. M., Cornwallis, C. K., & West, S. A. (2013). Group formation, relatednessand the evolution of multicellularity. Current Biology, 23, 1120e1125.

Fortelius, W., Rosengren, R., Cherix, D., & Chautems, D. (1993). Queen recruitment ina highly polygynous supercolony of Formica lugubris (Hymenoptera, For-micidae). Oikos, 67, 193e200.

Franks, N. R., & Hölldobler, B. (1987). Sexual competition during colony reproduc-tion in army ants. Biological Journal of the Linnean Society, 30, 229e243.

Gardner, A., Arce, A., & Alpedrinha, J. (2009). Budding dispersal and the sex ratio.Journal of Evolutionary Biology, 22, 1036e1045.

Griffin, A. S., & West, S. A. (2003). Kin discrimination and the benefit of helping incooperatively breeding vertebrates. Science, 302, 634e636.

Hamilton, W. D. (1964a). Genetical evolution of social behaviour. I. Journal ofTheoretical Biology, 7, 1e16.

Hamilton, W. D. (1964b). Genetical evolution of social behaviour. II. Journal ofTheoretical Biology, 7, 17e52.

Hamilton, W. D. (1972). Altruism and related phenomena, mainly in social insects.Annual Review of Ecology and Systematics, 3, 193e232.

Hamilton,W.D., &May, R.M. (1977). Dispersal in a stable habitat.Nature, 269, 578e581.Hannonen, M., & Sundström, L. (2003a). Reproductive sharing among queens in the

ant Formica fusca. Behavioral Ecology, 14, 870e875.Hannonen, M., & Sundström, L. (2003b). Sociobiology: worker nepotism among

polygynous ants. Nature, 421, 910.Hansell, M. H. (1996). Wasps make nests: nests make conditions. In S. Turillazzi, &

M. J. West-Eberhard (Eds.), Natural history and evolution of paper wasps (pp.272e289). Oxford, U.K.: Oxford University Press.

Hartke, T. R., & Rosengaus, R. B. (2013). Costs of pleometrosis in a polygamoustermite. Proceedings of the Royal Society B: Biological Sciences, 280, 20122563.

Heinze, J. (1993). Habitat structure, dispersal strategies and queen number in twoboreal Leptothorax ants. Oecologia, 96, 32e39.

Heinze, J., & Foitzik, S. (2009). The evolution of queen number in ants: from one tomany and back. In J. Gadau, & J. Fewell (Eds.), Organization of insect societies: Fromgenome to sociocomplexity (pp. 26e50). Cambridge, MA: Harvard University Press.

Heinze, J., Frohschammer, S., & Bernadou, A. (2013). Queen life-span and totalreproductive success are positively associated in the ant Cardiocondyla cf.kagutsuchi. Behavioral Ecology and Sociobiology, 67, 1555e1562.

Heinze, J., & Schrempf, A. (2008). Aging and reproduction in social insects: a mini-review. Gerontology, 54, 160e167.

Heinze, J., & Schrempf, A. (2012). Terminal investment: individual reproduction ofant queens increases with age. PLoS One, 7, e35201.

Helanterä, H., Strassmann, J. E., Carrillo, J., & Queller, D. C. (2009). Unicolonial ants:where do they come from, what are they and where are they going? Trends inEcology & Evolution, 24, 341e349.

Herbers, J. M. (1993). Ecological determinants of queen number in ants. In L. Keller(Ed.), Queen Number and sociality in insects (pp. 262e293). Oxford, U.K.: OxfordUniversity Press.

Holman, L., Dreier, S., & d’Ettorre, P. (2010). Selfish strategies and honest signalling:reproductive conflicts in ant queen associations. Proceedings of the Royal SocietyB: Biological Sciences, 277, 2007e2015.

Holzer, B., Keller, L., & Chapuisat, M. (2009). Genetic clusters and sex-biased geneflow in a unicolonial Formica ant. BMC Evolutionary Biology, 9, 69.

Hughes, W. O. H., Oldroyd, B. P., Beekman, M., & Ratnieks, F. L. W. (2008). Ancestralmonogamy shows kin selection is key to the evolution of eusociality. Science,320, 1213e1216.

Hölldobler, B., & Wilson, E. O. (1977). The number of queens: an important trait inant evolution. Naturwissenschaften, 64, 8e15.

Hölldobler, B., & Wilson, E. O. (1990). The ants. Berlin: Springer.Hölldobler, B., & Wilson, E. O. (2009). The superorganism. New York, NY: W.W.

Norton.Inward, D. J. G., Vogler, A. P., & Eggleton, P. (2007). A comprehensive phylogenetic

analysis of termites (Isoptera) illuminates key aspects of their evolutionarybiology. Molecular Phylogenetics and Evolution, 44, 953e967.

Johnson, B. R., Borowiec, M. L., Chiu, J. C., Lee, E. K., Atallah, J., & Ward, P. S. (2013).Phylogenomics resolves evolutionary relationships among ants, bees, and wasps.Current Biology, 23, 2058e2062. http://dx.doi.org/10.1016/j.cub.2013.08.050.

Kappeler, P. (Ed.). (2010). Animal behaviour: Evolution and mechanisms. Heidelberg:Springer.

Keller, L. (Ed.). (1993). Queen number and sociality in insects. Oxford, U.K.: OxfordUniversity Press.

Keller, L. (1997). Indiscriminate altruism: unduly nice parents and siblings. Trends inEcology & Evolution, 12, 99e103.

Keller, L., & Genoud, M. (1997). Extraordinary lifespans in ants: a test of evolu-tionary theories of ageing. Nature, 389, 958e960.

Keller, L., Passera, L., & Suzzoni, J. P. (1989). Queen execution in the Argentine ant,Iridomyrmex humilis. Physiological Entomology, 14, 157e163.

Keller, L., & Vargo, E. L. (1993). Reproductive structure and reproductive roles incolonies of eusocial insects. In L. Keller (Ed.), Queen number and sociality ininsects (pp. 16e44). Oxford, U.K.: Oxford University Press.

Kobayashi, K., Hasegawa, E., Yamamoto, Y., Kawatsu, K., Vargo, E. L., Yoshimura, J.,et al. (2013). Sex ratio biases in termites provide evidence for kin selection.Nature Communications, 4, 7.

Koenig, W., & Dickinson, J. L. (Eds.). (2004). Ecology and evolution of cooperativebreeding in birds. Cambridge, U.K.: Cambridge University Press.

Kramer, B. H., & Schaible, R. (2013). Colony size explains the lifespan differencesbetween queens and workers in eusocial Hymenoptera. Biological Journal of theLinnean Society, 109, 710e724.

Kronauer, D. J. C., & Boomsma, J. J. (2007). Multiple queens means fewer mates.Current Biology, 17, R753eR755.

Kümmerli, R., Helms, K. R., & Keller, L. (2005). Experimental manipulation of queennumber affects colony sex ratio investment in the highly polygynous antFormica exsecta. Proceedings of the Royal Society B: Biological Sciences, 272, 1789e1794.

Kümmerli, R., & Keller, L. (2008). Reproductive parameters vary with social andecological factors in the polygynous ant Formica exsecta. Oikos, 117, 580e590.

Leadbeater, E., Carruthers, J. M., Green, J. P., Rosser, N. S., & Field, J. (2011). Nestinheritance is the missing source of direct fitness in a primitively eusocial in-sect. Science, 333, 874e876.

Lenoir, A., Fresneau, D., Errard, C., & Hefetz, A. (1999). Individuality and colonialidentity in ants: the emergence of the social representation concept. InC. Detrain, J. L. Deneubourg, & J. M. Pasteels (Eds.), Information processing insocial insects (pp. 219e237). Basel, Switzerland: Birkhauser Verlag.

Martin, S. J., Beekman, M., Wossler, T. C., & Ratnieks, F. L. W. (2002). Parasitic Capehoneybee workers, Apis mellifera capensis, evade policing. Nature, 415, 163e165.

Matsuura, M. (1991). Vespa and Provespa. In K. G. Ross, & R. W. Matthews (Eds.), Thesocial biology of wasps (pp. 232e262). Ithaca, NY: Cornell University Press.

Michener, C. D. (1974). The social behavior of the bees. Cambridge, MA: HarvardUniversity Press.

Nonacs, P. (1988). Queen number in colonies of social hymenoptera as a kin-selected adaptation. Evolution, 42, 566e580.

Nonacs, P. (1993). The effects of polygyny and colony life history on optimal sexinvestment. In L. Keller (Ed.), Queen number and sociality in insects (pp. 110e131). Oxford, U.K.: Oxford University Press.

Nonacs, P., & Hager, R. (2011). The past, present and future of reproductive skewtheory and experiments. Biological Reviews, 86, 271e298.

Oster, G. F., & Wilson, E. O. (1978). Caste and ecology in the social insects. Monographsin population biology. Princeton, NJ: Princeton University Press.

Ozan, M., Helanterä, H., & Sundström, L. (2013). The value of oviposition timing,queen presence and kinship in a social insect. Proceedings of the Royal Society B:Biological Sciences, 280, 20131231.

Pamilo, P. (1991). Evolution of colony characteristics in social insects. 2. Number ofreproductive individuals. American Naturalist, 138, 412e433.

Parker, J. D. (2010). What are social insects telling us about aging? MyrmecologicalNews, 13, 103e110.

Pedersen, J. S. (2012). The logic of hypersocial colonies. Behavioral Ecology, 23, 934e935.

J. J. Boomsma et al. / Animal Behaviour 92 (2014) 241e252252

SPECIAL ISSUE: KIN SELECTION

Peters, J. M., Queller, D. C., Imperatriz-Fonseca, V. L., Roubik, D. W., &Strassmann, J. E. (1999). Mate number, kin selection and social conflicts instingless bees and honeybees. Proceedings of the Royal Society B: BiologicalSciences, 266, 379e384.

Queller, D. C. (2000). Relatedness and the fraternal major transitions. PhilosophicalTransactions of the Royal Society B: Biological Sciences, 355, 1647e1655.

Queller, D. C., & Strassmann, J. E. (1998). Kin selection and social insects. Bioscience,48, 165e175.

Queller, D. C., & Strassmann, J. E. (2002). The many selves of social insects. Science,296, 311e313.

Queller, D. C., & Strassmann, J. E. (2009). Beyond society: the evolution of organ-ismality. Philosophical Transactions of the Royal Society B: Biological Sciences, 364,3143e3155.

Queller, D. C., Strassmann, J. E., Solis, C. R., Hughes, C. R., & Deloach, D. M. (1993).A selfish strategy of social insect workers that promotes social cohesion. Nature,365, 639e641.

Rabitsch, W. (2011). The hitchhiker’s guide to alien ant invasions. Biocontrol, 56,551e572.

Ratnieks, F. L. W., & Reeve, H. K. (1992). Conflict in single-queen hymenopteransocieties: the structure of conflict and processes that reduce conflict inadvanced eusocial species. Journal of Theoretical Biology, 158, 33e65.

Reeve, H. K., & Jeanne, R. L. (2003). From individual control to majority rule:extending transactional models of reproductive skew in animal societies. Pro-ceedings of the Royal Society B: Biological Sciences, 270, 1041e1045.

Remolina, S. C., & Hughes, K. A. (2008). Evolution and mechanisms of long life andhigh fertility in queen honey bees. Age, 30, 177e185.

Rosengren, R., & Pamilo, P. (1983). The evolution of polygyny and polydomy inmound building Formica ants. Acta Entomologica Fennica, 42, 65e77.

Rosengren, R., Sundström, L., & Fortelius, W. (1993). Monogyny and polygyny inFormica ants: the result of alternative dispersal tactics. In L. Keller (Ed.),Queen number and sociality in insects (pp. 308e333). Oxford, U.K.: OxfordUniversity Press.

Ross, L., Gardner, A., Hardy, N., & West, S. A. (2013). Ecology, not the genetics of sexdetermination, determines who helps in eusocial populations. Current Biology,23, 2383e2387.

Ross, K. G., & Matthews, R. W. (1991). The social biology of wasps. Ithaca, NY: CornellUniversity Press.

Rüppell, O., & Heinze, J. (1999). Alternative reproductive tactics in females: the caseof size polymorphism in winged ant queens. Insectes Sociaux, 46, 6e17.

Savolainen, R., & Vepsäläinen, K. (2003). Sympatric speciation through intraspecificsocial parasitism. Proceedings of the National Academy of Sciences of the UnitedStates of America, 100, 7169e7174.

Schrempf, A., Cremer, S., & Heinze, J. (2011). Social influence on age and repro-duction: reduced lifespan and fecundity in multi-queen ant colonies. Journal ofEvolutionary Biology, 24, 1455e1461.

Schultz, T. R., Bekkevold, D., & Boomsma, J. J. (1998). Acromyrmex insinuator newspecies: an incipient social parasite of fungus-growing ants. Insectes Sociaux, 45,457e471.

Seeley, T. D. (1985). Honeybee ecology [Monographs in Behavior and Ecology].Princeton, NJ: Princeton University Press.

Sherman, P. W., Lacey, E. A., Reeve, H. K., & Keller, L. (1995). The eusociality con-tinuum. Behavioral Ecology, 6, 102e108.

Solomon, N. G., & French, J. A. (Eds.). (2007). Cooperative breeding in mammals.Cambridge, U.K.: Cambridge University Press.

Sommer, K., & Hölldobler, B. (1995). Colony founding by queen association anddeterminants of reduction in queen number in the ant Lasius niger. AnimalBehaviour, 50, 287e294.

Stearns, S. C. (1992). The evolution of life histories. Oxford, U.K.: Oxford UniversityPress.

Steiner, F. M., Crozier, R. H., & Schlick-Steiner, B. C. (2010). Colony structure. InL. Lach, C. L. Parr, & K. L. Abbott (Eds.), Ant ecology (pp. 177e193). Oxford, U.K.:Oxford University Press.

Sundström, L., Seppä, P., & Pamilo, P. (2005). Genetic population structure anddispersal patterns in Formica ants: a review. Annales Zoolici Fennici, 42, 163e177.

Thorne, B. L., & Haverty, M. I. (1991). A review of intracolony, intraspecific, andinterspecific agonism in termites. Sociobiology, 19, 115e145.

Turillazzi, S., & West-Eberhard, M. J. (Eds.). (1996). Natural history and evolution ofpaper-wasps. Oxford, U.K.: Oxford University Press.

Van der Hammen, T., Pedersen, J. S., & Boomsma, J. J. (2002). Convergent devel-opment of low-relatedness supercolonies in Myrmica ants. Heredity, 89, 83e89.

Van Oystaeyen, A., Alves, D. A., Oliveira, R. C., do Nascimento, D. L., doNascimento, F. S., Billen, J., et al. (2013). Sneaky queens in Melipona beesselectively detect and infiltrate queenless colonies. Animal Behaviour, 86, 603e609.

Vargo, E. L., & Husseneder, C. (2011). Genetic structure of termite colonies andpopulations. In D. E. Bignell, Y. Roisin, & N. Lo (Eds.), Biology of termites: Amodern synthesis (pp. 321e347). Berlin: Springer.

Wcislo, W. T. (1987). The roles of seasonality, host synchrony, and behavior in theevolutions and distributions of nest parasites in Hymenoptera (Insecta), withspecial reference to bees (Apoidea). Biological Reviews of the Cambridge Philo-sophical Society, 62, 515e543.

Wenseleers, T., Ratnieks, F. L. W., & Billen, J. (2003). Caste fate conflict in swarm-founding social Hymenoptera: an inclusive fitness analysis. Journal of Evolu-tionary Biology, 16, 647e658.

West-Eberhard, M. J. (1978). Polygyny and the evolution of social behavior in wasps.Journal of the Kansas Entomological Society, 832e856.

West, S. A., Griffin, A. S., Gardner, A., & Diggle, S. P. (2006). Social evolution theoryfor microorganisms. Nature Reviews Microbiology, 4, 597e607.

West, S. A., Pen, I., & Griffin, A. S. (2002). Cooperation and competition betweenrelatives. Science, 296, 72e75.

Wheeler, W. M. (1928). The social insects. New York, NY: Harcourt Brace.Wilson, E. O. (1971). The insect societies. Cambridge, MA: Harvard University Press.Wilson, E. O. (1975). Sociobiology: The new synthesis. Cambridge, MA: Belknap Press

of Harvard University Press.Wilson, E. O. (1993). Foreword. In L. Keller (Ed.), Queen number and sociality in in-

sects (p. v). Oxford, U.K.: Oxford University Press.Woyciechowski, M., & Kuszewska, K. (2012). Swarming generates rebel workers in

honeybees. Current Biology, 22, 707e711.Wynne-Edwards, V. C. (1962). Animal dispersion in relation to social behaviour.

Edinburgh, U.K.: Oliver & Boyd.