Current Biology 17, R673–R683, August 21, 2007 ª2007 Elsevier Ltd All rights reserved DOI 10.1016/j.cub.2007.06.033

ReviewKin Selection versus Sexual Selection:Why the Ends Do Not Meet

Jacobus J. Boomsma

I redevelop the hypothesis that lifetime monogamy isa fundamental condition for the evolution of eusociallineages with permanent non-reproductive castes,and that later elaborations — such as multiply-matedqueens and multi-queen colonies — arose withoutthe re-mating promiscuity that characterizes non-social and cooperative breeding. Sexually selectedtraits in eusocial lineages are therefore peculiar,and their evolution constrained. Indirect (inclusive)fitness benefits in cooperatively breeding verte-brates appear to be negatively correlated with pro-miscuity, corroborating that kin selection and sexualselection tend to generally exclude each other. Themonogamy window required for transitions from sol-itary and cooperative breeding towards eusocialityimplies that the relatedness and benefit-cost vari-ables of Hamilton’s rule do not vary at random, butoccur in distinct and only partly overlapping combi-nations in cooperative, eusocial, and derived euso-cial breeding systems.

Let’s start with the fact that sexual selection is to dowith social relations within species

H. Cronin (1991), The Ant and the Peacock

Sex is an antisocial force in evolutionE.O. Wilson (1975), Sociobiology, The New Synthesis

IntroductionThe second half of the 20th century saw major advancesin our understanding of evolutionary conflicts. Hamil-ton’s [1,2] ‘gene-centered’ view of evolution madereproductive altruism and reproductive conflict under-standable [3,4] and allowed the fundamental intergen-erational conflicts between parents and offspring tobe formulated [5]. Likewise the study of sexually se-lected conflicts was firmly reinstated in Darwin’s origi-nal adaptive framework [6–8], while also the Fisherianrunaway null-model of sexual selection was greatly re-fined [9]. More recently, Haig [10] combined insightsfrom inclusive fitness and sexual selection theory to for-mulate the concept of genomic imprinting. Althoughexceptional in its mechanisms, the discovery of geno-mic imprinting underlines the general principle thatgene-level interactions may both be cooperative andselfish [11]. The notion that there is no form or level ofbiological cooperation that is not somehow threatenedby internal corruption is now increasingly penetratingother fields of biology and medical science [12,13].

An interesting paradox is that selfish-gene ap-proaches have allowed us to make progress in under-standing the true general nature of cooperation,

Institute of Biology, University of Copenhagen, 2100

Copenhagen, Denmark.

Email: JJBoomsma@bi.ku.dk

because they forced us to stare the omnipresent po-tential of conflict in the face. However, the actual per-ception of which interactions are cooperative andwhich are conflict-prone has changed as more dataon paternity and family structure became available.The original emphasis on cooperation in matechoice and breeding — two unrelated individuals col-laborating to produce and sometimes raise offspringtogether — has gradually been replaced by conceptsemphasizing battles between the sexes and antago-nistic co-evolutionary arms races [14,15]. The para-digm shifted because life-long monogamy, the crucialcondition for avoiding conflict between mating part-ners, turned out to be rare in birds and mammals[16]. However, explicit studies of mating systems ofsocial insects have shown the opposite, i.e. that multi-ple paternity of queen offspring is much rarer thanpreviously assumed [17–19]. The advanced eusocialinsects have thus turned out to be fundamentally mo-nogamous, whereas most of the remaining free-livinganimal world is now confirmed to be mostly promiscu-ous (see Box 1 for a glossary of relevant terms used inthis review, including promiscuity).

As the quotes above illustrate, the conceptual inter-face between sexual selection and social evolution hasbeen confusing (but see [20]). While reproductive con-flict thinking has been equally crucial for students ofboth fields, their research agendas have diverged(Figure 1). Apparently, very few animal model systemswith spectacular sexually selected traits also displayaltruistic helping behaviour. Likewise, pinnacles of eu-social evolution have stimulated little effort to investi-gate sexually selected traits. Thus, the ‘ant’ and the‘peacock’ that have determined much of the agendain evolutionary ecology during the last 40 years [7] doin fact rarely meet. My main argument in this reviewwill be that this is not just a remarkable coincidenceof researchers in adjacent fields being blind for eachothers merits, but a logical and fundamental conse-quence of the way in which the sexes interact [6].

I will start by noting that promiscuity discouragespermanent commitments to reproductive altruism inthe same way as it corrupts cooperation between thesexes. I will then combine this logic with the currentlyknown mating system characteristics of the eusocialants, bees, wasps and termites to develop the hypoth-esis that strict lifetime monogamy (Box 1) was the sin-gle pervasive condition for the evolution of perma-nently sterile helper castes. In subsequent sections Iwill address some of the implications:

1. The three terms in Hamilton’s rule (br > c; Box 1)were unlikely to have varied at random when eu-sociality evolved, and marginal ergonomic (b/c)benefits of becoming a reproductive helper musthave sufficed to tip the balance away from inde-pendent, solitary reproduction when parentswere constrained to be monogamous for life.

Current Biology Vol 17 No 16R674

Box 1

Glossary.

Caste: A social phenotype that becomes permanently specialized for reproduction (queen, king), defence (soldier) or labour

(worker), before reaching maturity [55].

Cooperative breeding: A form of social organisation characterised by cooperative brood care, usually between one or both

parental breeders and one to several offspring helpers, but without any of them having fixed caste characteristics.

Eusociality: An advanced state of social organization characterized by generation overlap, cooperative brood care and fixed

castes with division of labour, of which at least one (workers or soldiers) has irreversibly specialized on a subset of the original

behavioural or physiological spectrum so that direct fitness is reduced (i.e. the caste has lost totipotency) [55].

Evolutionary (reproductive) conflict: Any situation in which the reproductive interests of genes in females, males and/or their

offspring are not completely aligned, such that manipulative traits may evolve that preferentially serve the interests of genes

expressed in specific focal individuals.

Hamilton’s rule: The inequality (br > c) specifying the conditions under which the expression of reproductive altruism can be

promoted by natural selection. r is the relatedness of the recipient of help to the donor, b is the increment in reproductive success

of the recipient owing to the help received, and c is the decrease in personal reproductive success of the donor [100].

Haplodiploidy hypothesis: The idea that relatedness of 0.75 between female full-siblings, which occurs in animals with

a haplodiploid sex determination system (e.g. the Hymenoptera), would by itself increase the likelihood for eusocial helper castes

to evolve.

Kin selection: Process by which traits are favoured because of their beneficial effects on the fitness of relatives [88].

Monogamy: Pair-bonding for life between a single male and female, here defined in its strictest possible sense of exclusive mating

between that male and female, for the purpose of their reproduction only. In this context, a stored ejaculate (e.g. as found in

Hymenopteran queens) is equivalent to a live mate.

Promiscuity: All mating systems that are not strictly monogamous. Promiscuity includes the following sub-categories:

Polygamy: Lifetime simultaneous pair-bonding between a single female and multiple males or their stored sperm (polyandry),

or a single male or his sperm and multiple females (polygyny). The latter is unlikely in eusocial insects, because males cannot

defend harems. Here, the term polygyny therefore refers to multiple queens breeding in the same nest or colony, each of them

mated to a different male (or sometimes different multiple males).

Re-mating promiscuity or promiscuity sensu stricto: Change of mates during a focal breeder’s lifetime, such that younger

offspring or later clutches are half siblings rather than full siblings. This includes serial monogamy, i.e. changing monogamous

partner so that clutches of mixed parentage are avoided, because overlapping generations of helpers and breeders will usually

imply that helpers should raise at least some half siblings.

Reproduction in eusocial insects: The production of sexual offspring: virgin queens, which are usually winged and often disperse

during mating flights, and males, which almost always disperse on the wing. The production of workers is normally not considered

to be reproduction, but colony growth.

Sexual selection: Process by which traits are favoured that increase competitive abilities for mating opportunities among males or

efficiency of partner selection by females. The roles are reversed in some species.

2. Secondaryelaborations ofeusocial matingsystemsaway from strict lifetime monogamy have evolvedonly when re-mating promiscuity could be avoided.

3. The operation of sexual selection in eusocial sys-tems with multiply mating queens is constrained,so that unusual adaptations are expected, whichallow for testing aspects of sexual selection the-ory that are normally not accessible in promiscu-ous mating systems.

4. Monogamy as a necessary condition for the evo-lution of eusociality is expected to apply in allnon-clonal eusocial arthropod lineages, but ex-plicit mating system studies are needed to verifythe generality of this principle.

5. Confirmation of this hypothesis would establishthat eusociality can evolve from cooperativebreeding only via an intermediate monogamousbreeding system.

6. None of the cooperatively breeding vertebrateshave evolved mating systems without re-matingpromiscuity, with the possible exception of theancestors of the two species of mole rats whichare often characterized as being eusocial.

7. Re-mating promiscuity and indirect fitness bene-fits seem to be negatively correlated acrosscooperatively breeding birds, suggesting thatalso in vertebrates the more extreme forms ofsexual selection preclude kin selection and viceversa.

8. Monogamy as a working hypothesis for the evolu-tion of sterile castes may help to put recent dis-cussions on the merits of kin selection for explain-ing social evolution to rest.

Lifetime Monogamy as the Key Conditionfor Becoming EusocialRing chromosomes and inbreeding cycles have beenproposed as key factors for the evolution of eusocialityin the diplo-diploid termites, because of their potentialto increase sibling relatedness beyond 0.5, but bothturned out to be problematic as they were not typicalfor the lower termites [21–23]. However, the crucialpoint is not whether relatedness becomes elevated,but rather that breeding systems must ensure that relat-edness among siblings does not drop below the value

Special IssueR675

BE

BES

– +

+

–

Sexual selection

Num

ber

of p

aper

s

Current Biology

0

20

40

60

80

100

120

140

Kin se

lectio

n

Figure 1. ‘Kin selection’ and ‘sexual selec-tion’ in the literature.

The number of times that the terms ‘kin se-lection’ and ‘sexual selection’ were foundin the titles or key words of articles pub-lished in two representative journals,Behavioral Ecology and Sociobiology(BES; light grey) and Behavioural Ecology(BE; dark grey) in 2004. More than 50% ofthe articles contained at least one of thesekey words but only ca. 3% containedboth. Relative to the expected frequen-cies, articles addressing both kin selectionand sexual selection are clearly underrep-resented (c2 = 5.54, P = 0.019 for BES and4.75, P = 0.029 for BE).

of 0.5 that applies to own offspring if reproductive altru-ism is to evolve de novo in the easiest possible way.

Throughout the termites, the 0.5 relatedness condi-tion is achieved by obligate lifetime monogamy [24].Even a low probability of some ancestral termite off-spring being half-siblings rather than full-siblingswould have required a step-wise, non-gradual b/c ad-vantage for starting evolution towards eusocial workercastes. In fact, we now know that the sister-group ofthe termites, the nest-building cockroach Cryptocer-cus, is also biparentally monogamous and sharesmany of the cellulose-digesting bacterial symbiont lin-eages that may have contributed to the ergonomic (b/c)fitness benefits of eusocial organization [22–29].However, symbiont acquisition by immature individ-uals does not preclude adult dispersal, which sug-gests that initial cooperative breeding in the lowertermites could only evolve into eusociality becausemonogamy was maintained. This implies that themultiple origins of soldiers and permanent workersin termites are easy to conceptualize, because ther-term in Hamilton’s rule was fixed at 0.5.

It has gradually become clear that lifetime monog-amy is also the default breeding system in the ants, eu-social bees and eusocial wasps, which are all haplo-diploid, but this fact is less widely appreciated. Bythe time inclusive fitness theory [1,2] became influen-tial, the belief had somehow become established thatmultiple mating was common enough to discredit thegenerality of the conceptual kin-selection framework(for example, see [26,30]), even though Hamilton[1,2], Wilson [25] and Andersson [31] had argued thatmultiple mating would not be a problem for explainingthe origins of eusociality across the Hymenoptera, if itcould be proven that such mating systems developedsecondarily after workers had irreversibly lost the ca-pacity to mate. Over the decades that followed, thenecessary comparative data did become graduallyavailable and showed that strict monogamy is highlylikely to have been the basal condition in all three majorclades of eusocial Hymenoptera [17–19]. This implies

that in ancestral lineages the average relatedness ofhelpers to female and male siblings under outbreedingand 1:1 Fisherian sex allocation was predictably 0.5 —the average between a relatedness of 0.75 to sistersand a relatedness of 0.25 to brothers — equivalent tothe relatedness to own offspring [21].

The current empirical evidence for monogamy as anancestral trait of eusocial taxa can be summarized asfollows: The vespid wasps [32] and the corbiculatebees [33] seem both highly likely to have had single-queen monogamous ancestors. In ants, both multiplemating [17,19] and multi-queen colonies [34] are verylikely to be derived, so that monogamous ancestry isa logical inference. Explicit mating system studies inother relevant taxa are not available, but indirect evi-dence suggests that basal clade monogamy is likelyto apply also in the single known ‘eusocial’ sphecidwasp [35], the thrips with soldiers [36], the social am-brosia beetles [37,38], as well as the snapping shrimpsand other social Crustacea [39,40]. The clonal aphidswith a soldier caste are an exception that proves therule, as parthenogenesis instead of outbred monog-amy doubles the inclusive fitness benefit of helpingand eliminates the variance [41].

In retrospect, it seems surprising that monogamyhas not been pursued more forcefully as a crucial con-dition for the evolution of eusociality, because it is soobviously consistent with Hamilton’s inclusive fitnesstheory [1,2]. Part of this is likely due to the unfortunateinitial emphasis on the elevated haplodiploid related-ness to full sisters, although the notion that the 0.75 re-latedness to sisters is compensated by 0.25 related-ness to brothers was established in formal modelsdecades ago [42,43]. However, both before and afterthe demise of the haplodiploidy hypothesis for the evo-lution of eusociality [44], references to monogamywere usually brief or implicit [2,7,21,23,45–48]. Themost transparent acknowledgment of the general sig-nificance of monogamy for the evolution of obligate re-productive altruism that I have been able to find is inthe endnotes of the second edition of The Selfish

Current Biology Vol 17 No 16R676

Gene, where Dawkins [3] states that self’s monoga-mous mother is worth as much reproductively as selfor, as Cronin [7] reiterates, as valuable as an identicaltwin. The reason is that, under outbreeding and 1:1sex allocation, all such mothers will predictably pro-duce offspring to whom self is related by 0.5 on aver-age. This principle should, thus, be at the core of anygeneral argument about the evolution of sterile workercastes.

The monogamy hypothesis is consistent with allevolutionary developments towards eusociality havingfollowed the so-called ‘subsocial’ (parent-offspring)route, so that the parasocial route (multiple same gen-eration foundresses), which has been quoted as an al-ternative for decades, would be rendered invalid, con-sistent with evidence already available a decade ago[21,23,46,47]. Also today, we know of no single casein which communal or parasocial breeding of unre-lated females has selected for the emergence ofa worker caste that has lost the capacity to mate andexpress full reproductive potential [47,49]. This is con-sistent with reproductive skew models which predictthat subordinates in matrifilial associations will giveup reproduction, whereas this will only partially takeplace in sibling associations with the same relatedness[50,51]. Comparative data confirm this prediction [50].Moreover, communal and eusocial breeding appear tobe mutually exclusive across the latest phylogeny ofthe Halictid bees — a clade in which eusociality is evo-lutionarily labile [52].

The transition to a eusocial state with helper castesof permanently reduced fecundity is a discontinuouspoint of no return in social evolution [53] that deservesto be added to seven other examples of irreversibleevolution [54]. Once eusociality has progressed be-yond the facultative stage, the helper castes arealways maintained except in some social parasites inwhich queens came to depend on workers of a differ-ent species, but without ever returning to a morphologyand behavioral syndrome reminiscent of the ancestralsolitary state [25,26]. The strict definition of eusociality(Box 1) [55] uses this irreversibility of worker castes,but not soldier castes, as a defining trait for extant eu-sociality. The hypothesis outlined here suggests thatthese same worker castes could only arise via strictlifetime monogamy.

The monogamy hypothesis implies that the threevariables in Hamilton’s rule are unlikely to vary at ran-dom when they assume values that fulfil the conditionsfor an evolutionary transition towards eusociality.When relatedness to siblings is predictably 0.5 be-cause of lifetime monogamy, any marginal gain in thebenefit/cost ratio of rearing siblings suffices to startan evolutionary process towards eusocial organisation[2,56]. This is fundamentally different in cooperativebreeders in which strict monogamy is not required,such that significantly positive b/c ratios are crucialto maintain temporary helping behaviour. The socialspiders are an illustrative example. Inbreeding ap-pears to have been a precursor of cooperative breed-ing in many lineages and b/c ratios are likely to be pos-itive [57]. However, none of the social spiders everbecame eusocial, which is consistent with the monog-amy hypothesis, as new colonies are founded by more

individuals than a single lifetime committed pair andre-mating promiscuity is part of normal social life.

The arguments above imply that students of socialevolution would benefit from loosening their focuson, or aversion to, relatedness per se and try to under-stand the fundamental characteristics of the matingand breeding systems that produce specific related-ness values in colonies. As long as there is no evidencefor anything else than lifetime monogamous matingsystems and outbreeding during pair formation in thebasal lineages of extant eusocial clades, it is most par-simonious to consider inbreeding and the variousergonomic conditions that contribute to the mainte-nance of established eusocial systems as secondarydevelopments. A corollary of the monogamy hypothe-sis is that there is only room for relaxing the lifetimemonogamy condition after a transition towards perma-nent reproductive altruism has been completed.

Eusocial Insects May Evolve to Be Polygamousbut Not to Change MatesBreeding systems where queens always mate withmore than a single male (polyandry) have evolved inseveral derived clades of ants, but appear to be re-stricted to very few taxa in the bees, essentially onlythe honeybees, and wasps, essentially only some gen-era of vespine wasps. In the eusocial Hymenoptera,polyandry merely means that more than one, some-times many, males commit their lifetime reproductivesuccess to the sperm storage organ of a single orsometimes a few queen(s) who will never re-mate[58]. The result is an inseminated queen ‘matriarch’maintaining a ‘harem of deceased males’ that surviveas her internalized private sperm bank. This early ac-quired sperm store is all she will ever have. Evenarmy ant queens, the last standing case where re-mating has been deemed likely, were recently shownnot to be able to re-mate effectively, as was in factpredicted by inclusive fitness theory [47,59]. Thus, asfar as the eusocial Hymenoptera have secondarilyevolved non-monogamous mating systems, theyhave become polygamous (i.e. polyandrous) withoutbeing promiscuous in the strict sense of changingmates.

As far as we know, virtually all termites have main-tained their ancestral, strictly monogamous matingsystem, the only functional difference to monogamousants being that the termite male (king) lives as long asthe queen and has exclusive lifetime mating rights be-cause he shares a royal nest chamber with his mate[58]. One study suggests that some termites mayhave secondarily become facultatively polygamouswith multiple unrelated queens coexisting in thesame nest, possibly with the same number of males.However, this occurs only as a rare alternative to de-fault monogamy [60] and re-mating promiscuity hasnot been confirmed with genetic markers. A morecommon elaboration is the retention of reproductiveneotenic offspring in the nest and having them estab-lished in additional royal nest chambers where theymate incestuously with a sibling [22,23,28]. Also thiselaboration does not violate the principle that termitecolonies are genetically closed systems that may gothrough a few inbreeding cycles after having been

Special IssueR677

Ave

rage

rel

ated

ness

to s

iblin

gs

Maximal tenure orlife-span of helpers

A

B

C

Death Death Death

Full-sibling relatedness (A)

Full- and half-sibling relatedness among a single queen’s offspring (B)

Sibling and more distant relatedness among multiple queen’s offspring (C)

Full- and half-sibling relatedness when one parent breeds with a new partner

Current Biology

0.5

1.0

0

Figure 2. Drastic sibling relatedness effects of re-mating promiscuity.

Diagram illustrating the effect of re-mating promiscuity in cooperative breeders on relatedness of helpers towards siblings under theassumption that a single dead parent (‘Death’) is replaced by an unrelated floater (thick black line). Such changeovers imply that helperswill experience sudden transitions from raising full-siblings (r = 0.5) towards raising half-siblings (r = 0.25). This may make them disperseto breed on their own if their b/c ratio of helping is <2, so that the new breeding pair will have to produce their own offspring before theycan recruit new helpers. For comparison, also the relatedness among offspring of lifetime monogamous parents has been drawn(dashed line A at 0.5), as well as an example of the expected average relatedness among female offspring of a eusocial multiply matedqueen (in large colonies, variation is negligible when stored sperm of different males is mixed; dashed line B) and among femaleoffspring of a eusocial group of co-breeding queens (fluctuations are expected to be larger as reproductive skew may vary overtime, not least because new offspring queens may be adopted as breeders; dotted line C).

founded by an outbred monogamous pair, a habit theyappear to have in common with some ambrosia bee-tles [37,61] and the naked mole rat. However, no newgenes ever seem to enter an established termite soci-ety, which implies that the termites may have remainednon-promiscuous throughout.

In ants and some eusocial wasps and bees, multiplequeens are often found breeding in the same nest (po-lygyny [62]). This normallydoesnot comeabout throughinbreeding with full siblings, although lesser degrees ofinbreeding are common in ants [48]. It is important tonote that also these derived breeding systems do notinvolve re-mating promiscuity. Multiple reproducingqueens will reduce the relatedness betweennestmates,but — as in single-queen societies — each queen car-ries the lifetime committed sperm of a single male or,more rarely, multiple males in her sperm storage organ.Re-mating later in life is not known to take place, so thatcooperatively breeding (polygynous) queens can nei-ther select new mates nor exploit opportunities for extrapair copulations, options that are normally available incooperatively breeding vertebrates. Thus, unlike thetermites, new genes do enter established polygynouscolonies of eusocial Hymenoptera, but they do so inmutually committed pairs of queens and stored ejacu-lates such that there is no re-mating promiscuity.

Neither polyandry nor polygyny induce the samehigh-amplitude shifts in inclusive fitness benefits forhelper castes that characterize cooperative breedingsystems with re-mating promiscuity (Figure 2). Multi-ple males mated to the same queen may contribute un-equal amounts of sperm, but sperm mixing normally

assures that average offspring relatedness in a colonyvaries very little over time. Multiple nest queens maypartition reproduction according to similar rules asdo cooperative breeders that have not become euso-cial [63]. However, reproductive skew in cooperativebreeders is always a matter of settled or unsettled re-productive conflict among totipotent breeders of sim-ilar status and is thus often moderate [50]. However,changes in vigour of established queens, and theadoption of new queens, may induce variation in relat-edness to siblings over time (Figure 2).

The fact that males of eusocial Hymenoptera areonly posthumously part of colony life may have beenan important factor for the evolution of secondary so-cial elaborations — such as polyandry and polygyny —because already stored sperm can no longer ‘becomepromiscuous’ in the strict sense of re-mating. Suchelaborations have not or hardly happened in the ter-mites, quite possibly because multiple live queensand kings in a colony would make promiscuous behav-iour inevitable and would thus have eliminated the nec-essary lifetime inclusive fitness benefits for immaturehelpers to become permanent workers or soldiers. Inother words, polygamy was a secondary option onlyin those clades where it could be implemented withoutre-mating promiscuity. Lifetime sperm storage se-cured this by default in the Hymenoptera, but not inthe termites.

The monogamy hypothesis for the evolution of euso-ciality, its subsequent further extensions towards mul-tiple queen mating and multi-queen colonies, andits connections with Hamilton’s rule are illustrated in

Current Biology Vol 17 No 16R678

Life

-tim

e ne

stm

ate/

offs

prin

g re

late

dnes

s

Hamiltonian benefit-cost ratio (b/c)

10.1 100.2 0.5 2 5

Eusocial breeding and restricted

promiscuity

Eusocial inbreeding cycles

Solitary/communal breeding and normal promiscuity

Cooperative breeding and normal promiscuity

Inbreeding

1

0.1

2

0.5

0.2

Current Biology

Figure 3. The monogamy hypothesis forthe evolution of eusociality.

The axes represent the crucial variables inHamilton’s rule over the lifetime of an indi-vidual: the ratio of relatedness (r) towardsnestmate siblings versus outbred off-spring (y) and the ratio of benefits versuscosts of helping (x). Both axes are logarith-mic such that ratios become linearised.The grey area towards the upper right rep-resents all x,y combinations where Hamil-ton’s condition for reproductive altruism(br > c) is fulfilled across the lifetime ofthe individual such that permanent helpercastes are expected to evolve. The highestvalue of y is fixed at a value of 2, becauseinbreeding cannot increase nestmate (sib-ling) relatedness beyond 1, i.e. twice therelatedness to outbred offspring. Arrowsindicate that all transitions towards euso-ciality are hypothesized to pass via a nar-row lifetime monogamy window (the blackcircle in the centre) and that it is not possi-ble to cross the diagonal elsewhere.

Figure 3. The white area has the solitary and communalbreeders in the lower left quadrant — where neither re-latedness nor ergonomic benefits are sufficient tomeet Hamilton’s rule — and the cooperative breedersin the right hand triangle — where b/c ratios favourhelping behaviour for some of the individual’s lifetime,but where total lifetime relatedness benefits are insuf-ficient for permanent helper castes to evolve. Matingsystem transitions can occur freely throughout thewhite area and will usually imply re-mating promiscu-ity. Transitions towards eusociality — both from soli-tary/communal breeding and from cooperative breed-ing — are hypothesized to occur only via the central(black circle) monogamy window where the sibling/off-spring relatedness ratio is equal to 1, so that initial b/cratios only need to be marginally >1 to make the tran-sition into the grey eusociality area. This contrasts withprevailing — often implicit — interpretations of Hamil-ton’s rule that assume the diagonal towards eusocial-ity can be crossed anywhere, i.e. with many differentcombinations of relatedness and b/c ratios. After eu-sociality has been irreversibly established, relatednessratios may remain equal to 1 (e.g. termites) or may de-crease to lower values because of polyandry and po-lygyny, while b/c ratios will tend to increase relativeto values that applied to the ancestors that made thetransition. Cyclic inbreeding as a secondary elabora-tion in termites (and naked mole rats) has been indi-cated by an arrow with positive slope. Inbreeding asan ancestral trait (the white area towards the upperleft) is not known to have given rise to eusocial devel-opments independent of monogamy, but was almostcertainly the precursor of cooperative breeding in sev-eral clades of spiders where it was not coupled to mo-nogamy [57].

Constrained and Unusual Forms of SexualSelection in Eusocial InsectsObligate multiple mating of queens must have evolvedvia facultative multiple mating — some queens mateonce, others multiple times — but the extant genera

of eusocial Hymenoptera with multiple mating belongto either the facultative or the obligate category[17,19]. All these multiple mating systems had monog-amous ancestors with only the most basal form of sex-ual selection, i.e. assortative mating based on overallvigour [26,64,65]. To the extent that more elaboratesexually selected traits did evolve in clades with multi-ply mated queens, we may thus expect them to be in-dependently novel and possibly convergent. They arealso likely to differ from those in non-social breedingsystems, because of the unusual lack of re-matingpromiscuity.

Male mating efforts in ants, eusocial bees and euso-cial wasps may either result in exclusive paternitywhen a mating plug prevents additional matings andmakes females lose interest [66–68] or in shared pater-nity when females mate with a number of males inquick succession [69–71]. Males may actively competefor access to females in bees and wasps [72], but thisis uncommon or absent in ants and honeybees [58,71].Females of many social bees and wasps are able to re-ject unwanted mates [72], but this seems more excep-tional in ants in which queens are less agile flyers andhave often been selected to minimize the duration oftheir mating flight to reduce exposure to predatorsand diseases [58]. Thus, as far as pre-mating sexuallyselected behaviours were present, they seem to havebeen largely lost in the advanced eusocial taxa withperennial societies.

Lethal male fighting has evolved in connection withincestuous mating within the nest, but it occurs injust a few genera of ants and bears many similaritiesto male fighting in fig wasps [73]. Mating in the nestprovides an unusual selection arena in which maleshave access to multiple females without having tomake much of an effort, whereas rivals tend to be rel-atively few and cannot escape. The evolution of malefighting, therefore, illustrates that relatedness betweencontenders is not decisive for preventing aggressionwhen competition is local [74]. Breeding systems ofthis kind have been extensively studied in the ant

Special IssueR679

genus Cardiocondyla [58,75]. Some lineages of theseants have males with sable-shaped mandibles whokill rivals that eclose after them [76]. These winglessfighting males often coexist with winged brotherswith normal (harmless) mandibles that disperse tomate in neighbouring colonies [77]. Most remarkable,the Cardiocondyla fighting males are the only antmales known that mature sperm throughout their adultlives, as ant males normally hatch with a fixed amountof sperm that is determined in the pupal stage [76].However, just like any other ant, the Cardiocondylamating system does not appear to have female re-mat-ing promiscuity. Whether singly or multiply insemi-nated [78], and whether they leave the colony or stay,also queens of Cardiocondyla are receptive for onlya very short time and do not re-mate later in life.

Also post-mating sexual selection in eusocial Hyme-noptera with multiple paternity is peculiar. At firstsight, the possibilities for sperm competition appearto be substantial because ejaculates normally overlapalmost completely in time and space. However, of theseven mechanisms of sperm precedence listed bySimmons [79] only three — sperm loading, spermstratification and sperm selection — may apply in eu-social Hymenoptera. Three of the four remainingmechanisms — sperm flushing, sperm displacementand sperm removal — are unknown and the fourth —sperm incapacitation — is unlikely.

Sperm loading, i.e. males transferring differentamounts of sperm when mating with the same femaleto bias what would otherwise be a fair sperm storageraffle, is probably common in eusocial Hymenoptera.However, after the brief mating window has passed,transferred sperm is never threatened by rival sperm.Even during this window, it merely risks being diluted,not being displaced. Every time that multiple matingevolved from monogamous ancestors, some selectionpressure may have arisen for ejaculates to reduce theshare in paternity of non-related sperm. However, themechanisms available for the evolution of such traitsare constrained [58,68]. Manipulative traits that wouldnegatively affect queen fitness before the onset of re-production would likely be selected against, as newlyfounded colonies produce only sterile workers duringthe first months in most bees and wasps and for sev-eral years in most ants. It is, therefore, difficult to con-ceive of direct mechanisms by which stored ejaculatescould obtain significant selfish benefits in the repro-ductive phase, without suffering large collective costsin the ergonomic phase of colony development. Ingrowth phases, it would in fact increase male fitnessto have eggs fertilized by rival sperm, so that less ofthis is left in storage when reproduction starts. Even af-ter reaching the size to reproduce, perennial ant colo-nies will continue to produce at least an order of mag-nitude more sterile workers than virgin queens, so thatincapacitation of stored rival sperm may still backfirewhen it affects colony productivity. This implies thatthe evolution of sperm traits that damage queen fit-ness is less likely than in non-eusocial insects andthat any such traits that did evolve are expected tohave relatively mild effects [58].

Targets for sexual selection that remain available inpolyandrous eusocial Hymenoptera are traits that

affect sperm mobility and viability. These topics haverecently been reviewed elsewhere [58,80]. They areall affected by the unusual circumstances that long-lived queens — in particular those of ants that maylive for decades — are assumed to potentially use alltheir stored sperm before dying, such that sperm sup-ply ultimately becomes the limiting factor for lifetimereproductive success. Even if this assumption wouldturn out not to be strictly true, it would still be a fair ap-proximation as sperm becomes a limiting resourcewhen re-mating later in life is not an option. This im-plies that we should expect specific adaptations tokeep sperm maximally viable during storage and tominimize sperm waste during fertilization [81], bothof which are not expected in mating systems with re-mating promiscuity. The relevant questions related tosexual selection, sperm competition and cryptic fe-male choice in polyandrous eusocial Hymenopteratherefore often differ from those addressed in non-eusocial mating systems, but their unusual charactermay allow interesting experimental tests of the gener-ality of sexual selection theory that cannot be per-formed in more promiscuous model systems [58,80].

Sexual Selection and Kin Selectionin Cooperatively Breeding VertebratesThe comparative study of cooperatively breeding ver-tebrates has progressed significantly during recentdecades [82–85]. An important synthesis was recentlyprovided by Griffin and West [86] who analyzed com-parative data from 11–15 species of birds as well astwo to three species of mammals and showed thatthere is a strong positive correlation between the de-gree to which helping is preferentially directed towardskin and the extent to which helping results in increasedproduction or survival of offspring of dominantbreeders. The necessity for helpers to discriminate ac-cording to degree of kin is expected to be directly re-lated to the turnover of breeders after death or usurpa-tion and the extent of promiscuity of the breedingfemale (Figure 2). A closer comparative look at the mat-ing systems of cooperatively breeding birds and mam-mals is therefore worthwhile, in order to evaluate howkin selection and sexual selection may interact in sys-tems that are not constrained to be monogamous andthat have helpers that can disperse to reproduceindependently.

A recent study by Griffith et al. [87] reviewed theavailable data on extra pair paternity across 129 spe-cies of birds and categorised them in five mutually ex-clusive breeding systems. A simple analysis of thedata shows that 61% (11/18) of the cooperativelybreeding species show some degree of extra-pairpaternity, whereas this proportion is 75% (83/111) forall other, non-cooperative breeding systems. Thoughnon-significant (c2 = 1.46; P = 0.227), this differenceis in the expected direction of cooperative breedershaving fewer offspring with extra-pair paternity, analo-gous with the evolution of permanent helper castes re-lying on strict monogamy. A more specific evaluationof the ranked proportions of offspring with extra-pairpaternity in cooperative versus non-cooperativebreeders gave a marginally significant result (t =1.5463; one tailed P = 0.0624; Wilcoxon two-sample

Current Biology Vol 17 No 16R680

test; Figure 4). As it appears, the category of below10% extra-pair paternity is equally common in cooper-ative and non-cooperative breeders, but geneticallydocumented monogamy is overrepresented andmore excessive promiscuity underrepresented amongthe cooperative breeders. This analysis is obviouslyvery crude as phylogenetic effects have not beentaken into account, but some of the deviations fromthe overall trend suggest that more detailed and bettercontrolled future studies might actually produce moresignificant effects.

One of the four cooperative breeders in the highestextra-pair paternity class, the white-browed scrubw-ren, is just above the 10% cut-off point with 12.4% off-spring with extra-pair paternity, but the other threespecies have much higher percentages and werepart of the comparative data set analysed by Griffinand West [86]. One of these, the superb fairy wren(with 71.6% offspring with extra-pair paternity) is inthe very bottom left corner of their regression plot(see Figure 2 in [86]; see also Figure 5 in [88]), indicat-ing that it neither obtains kin-selected inclusive fitnessbenefits from helping, nor directs its helping effortspreferentially towards kin. When the Wilcoxon rankscore test is repeated without this outlier, the resultis actually significant (t = 1.987; one tailed P = 0.025).The other two highly promiscuous species (westernbluebird and Seychelles warbler) are further towardsthe right in Griffin and West’s plot [86], indicating thatthey obtain, respectively, some and substantial inclu-sive fitness benefits from helping. Interestingly, thesetwo species also have the two highest positive resid-uals for kin discrimination. This implies that they areconsiderably better in directing their help towards

–EPP +EPP(<10%) +EPP(>10%)

51 43350

0.05

0.10

0.15

0.20

0.25

0.30

Coo

pera

tive

bree

ders

Oth

er b

reed

ing

syst

ems

Current Biology

Figure 4. Cooperative breeding and promiscuity in birds.

The frequency of no extra-pair paternity (-EPP), some extra-pairpaternity (0% <+EPP<10%), and ample extra-pair paternity(+EPP>10%) among cooperative breeding birds, relative tobirds with other breeding systems. Data are from Griffith et al.[87]. The separation of the +EPP data at 10% is arbitrary, buthas the merit of achieving roughly equal sample sizes amongthe three categories (written in the bars). The horizontal line isthe average ratio of cooperative breeders versus other breedingsystems in the 129 species for which data on % offspring withEPP were available and deemed reliable enough to be usedfor comparative analysis by Griffith et al. [87].

close kin than predicted from the overall regression,which seems an appropriate way to partially eliminatethe negative inclusive fitness consequences of paren-tal promiscuity.

The cooperatively breeding mammal data included inGriffin and West’s analysis [86] are too few to evaluatesystematically, but it is interesting to note that the twomongoose species (dwarf mongoose and meerkat)show moderate inclusive fitness benefits of helpingand intermediate tendencies to preferentially directhelping towards close kin. Some promiscuity has beenreported for both species [85,89,90], which contrastswith the naked and Damaraland molerats, where mostindividuals are lifetime helpers raising only very closekin. The negative correlation between sexual selectionand kin selection that made monogamy a requirementfor the evolution of sterile castes in insects, thus, seemsto occur also in cooperative breeding birds and mam-mals where options for re-mating promiscuity are pres-ent. This suggests that kin selection and sexual selec-tion also work in opposite directions in cooperativelybreeding vertebrates, but that becoming strictly mo-nogamous is a much more difficult transition to makethan in insects. This is consistent with only two speciesof mole rats having come close to being eusocial, al-though their helper castes are not irreversible as, for ex-ample, in ants, honeybees or higher termites.

As it turns out, the two species of mole rats areinteresting cases to evaluate the general validity ofthe monogamy criterion for eusociality (Table 1). Theircolonies have a single breeding female and one tothree dominant males that sire offspring [91]. As faras it is strictly monogamous, this breeding system iscomparable with termites where a single live king islifetime committed to a single queen and where in-breeding cycles may occur after colony establishmentby an outbreeding pair to prolong colony life span.However, naked mole rat colonies with multiple malebreeders seem functionally more comparable toarmy ants — having multiple males lifetime committedas stored sperm [59] — where colonies multiply byfission and by sending out drones to mate with unre-lated queens [92].

Following the logic of the monogamy hypothesis,a key criterion for classifying the naked and Damara-land mole rats as eusocial would be that dispersingmales — a special morph predicted by Dawkins [3]and found in the naked mole rat by Braude [93] —should not breed with an already established dominantfemale but only found new colonies with unrelated fe-males that have not bred before. If this were the casethe two mole rat species would fulfil the same non-re-mating promiscuity criterion that characterizes euso-cial insects. It would make their breeding systems fullyanalogous to those of honeybees, stingless bees, andarmy ants, where colonies that lose their queen mayraise a virgin replacement queen that will mate with un-related males from other colonies, or to those of ter-mites where a replacement or additional queen matesincestuously with a brother. It is interesting to notethat, while the naked mole rat has secondarily evolvedinbreeding cycles reminiscent of termites, the Damara-land mole rat has retained the ancestral state of avoid-ing inbreeding and having colony fission after the

Special IssueR681

death of a reproductive, reminiscent of army ants [92].Naked mole rats have much larger colonies than Dam-araland mole rats, which may imply that selection foraccepting a permanent helper role in order to maintainalready established colonies has been more pro-nounced [91]. Similarly, increasing colony size hasbeen an important factor during the elaboration of eu-social organization in insect societies [94].

ConclusionsThe relative impact of sexual selection among femalesand males is determined by their parental investments.In this review, I have extended this fundamental insightby Trivers [6] by arguing that lifetime bi-parental com-mitment to a single family characterizes all eusocialbreeding systems with permanent castes such thatsexual selection is greatly constrained. My central ar-gument has been that re-mating promiscuity, in spiteof all the direct fitness benefits that it may bring to par-ents, is a form of social cheating from the perspectiveof offspring when they care for younger siblings with-out being able to discriminate between full and halfsiblings. Permanent helper castes that rely mostly orexclusively on indirect (inclusive) fitness benefits will,therefore, not evolve unless lifetime monogamy se-cures these benefits. Cooperative breeders do nothave this ancestral monogamy constraint, but the ex-tent to which helpers in birds and mammals obtain in-clusive fitness benefits appears to be similarly affectedby parental promiscuity.

Although supported by a fair amount of evidence,monogamy as a general prerequisite for the evolutionof permanent eusocial castes is a hypothesis thatawaits further testing. Its generality would be refutedif cases came to light where non-monogamous cladeshave produced species with permanently sterilecastes. This issue, and other predictions derivedfrom the monogamy hypothesis (Table 1) should befurther tested both in the eusocial Hymenoptera andtermites, and in the presumed eusocial ambrosia bee-tles, gall-forming thrips with soldiers, snappingshrimps and mole rats. Focused comparative studiesin cooperative breeders and even human societiescould further shed important light on the generality ofthe negative correlation between inclusive fitness ben-efits and re-mating promiscuity.

The monogamy hypothesis implies that eusocial andcooperative breeding are distinct social systems be-cause the three variables of Hamilton’s rule [1,2] tendto combine into different strategy sets. This is easiestto illustrate when using life-time relatedness and ben-efit-cost ratios, as in Figure 3. This also makes itstraightforward to use Wilsonian group selection ter-minology to describe the same phenomena [95–97].All cases of strong altruism where worker and soldiercastes rely on indirect fitness benefits are concen-trated in the grey area of Figure 3. In contrast, casesof weak altruism that are driven by direct benefitsmay occur throughout the diagram, i.e., both in thewhite area in the Figure referring to independent or co-operative breeders and in the grey area when multipleeusocial queens breed in the same colony. Parentsmaking life-time commitments to their partner(s)when founding colonies are, therefore, also a condition

for strong reproductive altruism [95]. This underlinesthat recent challenges to kin selection as the crucialmechanism for the evolution of eusociality (for exam-ple, [53]) are unsustainable not only because of con-ceptual problems [49,96,97], but also because theywould require the monogamy hypothesis to be falsifiedfor every extant eusocial clade.

The hypothesized high (0.5) relatedness and low(just >1) b/c ratio conditions for the early evolution ofeusociality contrast with the low relatedness andhigh b/c ratios (relative to ancestors that made thetransition towards eusociality) that often characterizederived eusocial systems with multiple mating or mul-tiple queens. This underlines the need to distinguishbetween scenarios for the early evolution and laterelaboration of social behaviour. Additional variablessuch as coercion and punishment are now known tobe important as secondary developments both incooperatively breeding and eusocial clades [98,99].However, although these mechanisms help to stabilizesocieties, their establishment rarely implied thatrelatedness became unimportant.

Acknowledgements

I thank Boris Baer, Trine Bilde, Tim Clutton-Brock, Sylvia

Cremer, Bernie Crespi, Patrizia D’Ettorre, Alan Grafen, David

Hughes, Laurent Keller, David Nash, Gosta Nachman, Jes

Pedersen, Joan Strassmann and Stu West for discussion and

comments, David Nash for editing the figures and the Danish

National Research Foundation for support.

References1. Hamilton, W.D. (1964). The genetical evolution of social behaviour,

I & II. J. Theor. Biol. 7, 1–52.

2. Hamilton, W.D. (1972). Altruism and related phenomena, mainly insocial insects. Annu. Rev. Ecol. Syst. 3, 193–232.

3. Dawkins, R. (1976). The Selfish Gene, Second Edition, 1989(Oxford, UK: Oxford University Press).

4. Dawkins, R. (1982). The Extended Phenotype: The Gene as the Unitof Selection (Oxford, UK: Oxford University Press).

5. Trivers, R.L. (1974). Parent-offspring conflict. Amer. Zool. 14, 249–264.

6. Trivers, R.L. (1972). Parental investment and sexual selection. InSexual Selection and the Descent of Man, 1871-1971, B. Campbell,ed. (Chicago, USA: Aldine-Atherton), pp. 136–172.

Table 1. Testable predictions that follow from the Monogamy

Hypothesis.

1. Clades with eusocial species are characterized by strict

lifetime monogamy as ancestral state.

2. Clades where dominant breeders have retained the

option to change sexual partner(s) do not have lifetime

committed worker/soldier morphs (castes) of irreversibly

reduced fertility.

3. Established queens of eusocial Hymenoptera do not

naturally re-mate later in life in ways that allow them to

store significant amounts of new sperm from unrelated

males so that their tenure as breeder is prolonged.

4. Polygamy of, or promiscuity among, unrelated

reproductives in termites is rare or absent, or merely

happens in accidentally merged senescent colonies that

have no reproductive future.

5. Cooperative breeders in which helpers obtain larger

indirect fitness benefits should be less promiscuous.

6. Manipulative traits of ejaculates that harm the

reproductive fitness of multiply mating eusocial queens

are uncommon and have relatively mild effects.

Current Biology Vol 17 No 16R682

7. Cronin, H. (1991). The Ant and the Peacock (Cambridge, UK:Cambridge University Press).

8. Andersson, M. (1994). Sexual Selection (Princeton, New Jersey:Princeton University Press).

9. Lande, R. (1981). Models of speciation by sexual selection on poly-genic traits. Proc. Natl. Acad. Sci. USA 78, 3721–3725.

10. Haig, D. (1997). The social gene. In Behavioural Ecology: An Evolu-tionary Approach, Fourth Edition, J.R. Krebs and N.B. Davies, eds.(Oxford, UK: Blackwell Scientific Publications), pp. 284–304.

11. Burt, A., and Trivers, R. (2006). Genes in Conflict: The Biology ofSelfish Genetic Elements (Cambridge, Massachusetts: HarvardUniversity Press).

12. Stearns S.C., ed. (1999). Evolution in Health and Disease (Oxford,UK: Oxford University Press).

13. Denison, R.F., Kiers, E.T., and West, S.A. (2003). Darwinian agricul-ture: When can humans find solutions beyond the reach of naturalselection? Q. Rev. Biol. 78, 145–168.

14. Chapman, T., Arnqvist, G., Bangham, J., and Rowe, L. (2003). Sex-ual conflict. Trends Ecol. Evol. 18, 41–47.

15. Bernasconi, G., Ashman, T.-L., Birkhead, T.R., Bishop, J.D.D.,Grossniklaus, U., Kubli, E., Marshall, D.L., Schmid, B., Skogsmyr,I., Snook, R.R., et al. (2004). Evolutionary ecology of the prezygoricstage. Science 303, 971–975.

16. Barash, D.P., and Lipton, J.E. (2001). The Myth of Monogamy:Fidelity and Infidelity in Animals and People (Oxford, UK: W.H.Freeman).

17. Boomsma, J.J., and Ratnieks, F.L.W. (1996). Paternity in eusocialHymenoptera. Philos. Trans. R. Soc. London Ser. B 351, 947–975.

18. Strassmann, J. (2001). The rarity of multiple mating by females inthe social Hymenoptera. Insectes Soc. 48, 1–13.

19. Boomsma, J.J., Kronauer, D.J.C., and Pedersen, J.S. (2008). Theevolution of social insect mating systems. In Organization of InsectSocieties – Integrated Research Projects, J. Gadau and J. Fewell,eds. (Cambridge, Massachusetts: Harvard University Press),In press.

20. Queller, D.C. (1994). Male-female conflict and parent-offspringconflict. Am. Nat. 144, S84–S99.

21. Bourke, A.F.G. (1997). Sociality and kin selection in insects. In Be-havioural Ecology: An Evolutionary Approach, Fourth Edition, J.R.Krebs and N.B. Davies, eds. (Oxford, UK: Blackwell Scientific Pub-lications), pp. 203–227.

22. Shellman-Reeve, J.S. (1997). The spectrum of eusociality in ter-mites. In Social Behavior in Insects and Arachnids, J.C. Choe andB.J. Crespi, eds. (Cambridge, UK: Cambridge University Press),pp. 52–93.

23. Thorne, B.L. (1997). Evolution of eusociality in termites. Annu. Rev.Ecol. Syst. 28, 27–54.

24. Nalepa, C.A., and Jones, S.C. (1991). Evolution of monogamy in ter-mites. Biol. Rev. 66, 83–97.

25. Wilson, E.O. (1971). The Insect Societies (Cambridge, Massachu-setts: Belknap Press of Harvard University Press).

26. Wilson, E.O. (1975). Sociobiology, the New Synthesis (Cambridge,Massachusetts: Belknap Press of Harvard University Press).

27. Nalepa, C.A., and Bell, W.J. (1997). Postovulation parental invest-ment and parental care in cockroaches. In Social Behavior in In-sects and Arachnids, J.C. Choe and B.J. Crespi, eds. (Cambridge,UK: Cambridge University Press), pp. 26–51.

28. Roisin, Y. (1999). Philopatric reproduction, a prime mover in theevolution of termite sociality? Insectes Soc. 46, 297–305.

29. Inward, D., Beccaloni, G., and Eggleton, P. (2007). Death of anorder: a comprehensive molecular phylogenetic study confirmsthat termites are eusocial cockroaches. Biol. Lett. Published online. DOI: 10.1098/rsbl.2007.0102.

30. Brockmann, H.J. (1984). The evolution of social behaviour ininsects. In Behavioural Ecology: An Evolutionary Approach,Second Edition, J.R. Krebs and N.B. Davies, eds. (Oxford, UK:Blackwell Scientific Publications), pp. 340–361.

31. Andersson, M. (1984). The evolution of eusociality. Ann. Rev. Ecol.Syst. 15, 165–189.

32. Foster, K.R., and Ratnieks, F.L.W. (2001). Paternity, reproductionand conflict in vespine wasps: a model system for testing kin selec-tion predictions. Behav. Ecol. Sociobiol. 50, 1–8.

33. Danforth, B. (2007). Bees. Curr. Biol. 17, R156–R161.

34. Holldobler, B., and Wilson, E.O. (1977). The number of queens: animportant trait in ant evolution. Naturwissenschaften 64, 8–15.

35. Matthews, R.W. (1991). Evolution of social behavior in sphecidwasps. In The Social Biology of Wasps, K.G. Ross and R.W.Matthews, eds. (Ithaca, New York: Cornell University Press).

36. Crespi, B.J., and Mound, L.A. (1997). Ecology and evolution of so-cial behavior among Australian gall thrips and their allies. In Social

Behavior in Insects and Arachnids, J.C. Choe and B.J. Crespi, eds.(Cambridge, UK: Cambridge University Press), pp. 166–180.

37. Kirkendall, L.R., Kent, D.S., and Raffa, K.A. (1997). Interactionsamong males, females and offspring in bark and ambrosia beetles:the significance of living in tunnels for the evolution of social behav-ior. In Social Behavior in Insects and Arachnids, J.C. Choe and B.J.Crespi, eds. (Cambridge, UK: Cambridge University Press), pp.181–215.

38. Peer, K., and Taborsky, M. (2007). Delayed dispersal as a potentialroute to cooperative breeding in ambrosia beetles. Behav. Ecol.Sociobiol. 61, 729–739.

39. Duffy, J.E. (2003). The ecology and evolution of eusociality insponge-dwelling shrimp. In Genes, Behavior and Evolution in So-cial Insects, T. Kikuchi, N. Azuma, and S. Higashi, eds. (Sapporo,Japan: Hokkaido University Press), pp. 217–252.

40. Crespi, B.J. (2007). Comparative evolutionary ecology of social andsexual systems: Water-breathing insects come of age. In Evolu-tionary Ecology of Social and Sexual Systems: Crustaceans asmodel Organisms, J. Emmett Duffy and M. Thiel, eds. (Oxford,UK: Oxford University Press), pp. 442–460.

41. Stern, D.L., and Foster, W.A. (1997). The evolution of sociality inaphids: a clone’s-eye view. In Social Behavior in Insects and Arach-nids, J.C. Choe and B.J. Crespi, eds. (Cambridge, UK: CambridgeUniversity Press), pp. 150–165.

42. Trivers, R.L., and Hare, H. (1976). Haplodiploidy and the evolutionof the social insects. Science 191, 249–263.

43. Craig, R. (1979). Parental manipulation, kin selection, and the evo-lution of altruism. Evolution 33, 319–334.

44. Strassmann, J.E., and Queller, D.C. (2007). Insect societies as di-vided organisms: The complexities of purpose and cross-purpose.Proc. Natl. Acad. Sci. USA 104, 8619–8626.

45. Alexander, R.D. (1974). The evolution of social behavior. Ann. Rev.Ecol. Syst. 5, 325–383.

46. Alexander, R.D., Noonan, K.M., and Crespi, B.J. (1991). The Biologyof the Naked Mole Rat, P.W. Sherman, J.U.M. Jarvis, and R.D.Alexander, eds. (Princeton, New Jersey: Princeton UniversityPress), pp. 3–44.

47. Bourke, A.F.G., and Franks, N.R. (1995). Social Evolution in Ants(Princeton, New Jersey: Princeton University Press).

48. Crozier, R.H., and Pamilo, P. (1996). Evolution of Social InsectColonies (Oxford, UK: Oxford University Press).

49. Helantera, H., and Bargum, K. (2007). Pedigree relatedness, notgreenbeard genes, explains eusociality. Oikos 116, 217–220.

50. Reeve, H.K., and Keller, L. (1995). Partitioning of reproduction inmother-daughter versus sibling associations: a test of optimalskew theory. Am. Nat. 145, 119–132.

51. Stubblefield, J.W., and Charnov, E.L. (1986). Some conceptualissues in the origin of eusociality. Heredity 57, 181–187.

52. Schwarz, M.P., Richards, M.H., and Danforth, B.N. (2007). Chang-ing paradigms in insect social evolution: insights from Halictineand Allodapine bees. Annu. Rev. Entomol. 52, 127–150.

53. Wilson, E.O., and Holldobler, B. (2005). Eusociality: origin andconsequences. Proc. Natl. Acad. Sci. USA 102, 13367–13371.

54. Bull, J.J., and Charnov, E.L. (1985). On irreversible evolution.Evolution 39, 1149–1155.

55. Crespi, B.J., and Yanega, D. (1995). The definition of eusociality.Behav. Ecol. 6, 109–115.

56. Queller, D.C. (1989). The evolution of eusociality: reproductivehead starts of workers. Proc. Natl. Acad. Sci. USA 86, 3224–3226.

57. Aviles, L. (1997). Causes and consequences of cooperation andpermanent-sociality in spiders. In Social Behavior in Insects andArachnids, J.C. Choe and B.J. Crespi, eds. (Cambridge, UK:Cambridge University Press), pp. 476–498.

58. Boomsma, J.J., Baer, B., and Heinze, J. (2005). The evolution ofmale traits in social insects. Annu. Rev. Entomol. 50, 395–420.

59. Kronauer, D.J.C., and Boomsma, J.J. (2007). Do army ant queensre-mate later in life? Insectes Soc. 54, 20–28.

60. Atkinson, L., and Adams, E.S. (1997). The origins and relatednessof multiple reproductives in colonies of the termite Nasutitermescorniger. Proc. Roy. Soc. Lond. B 264, 1131–1136.

61. Peer, K., and Taborsky, M. (2005). Outbreeding depression, but noinbreeding depression in haplodiploid ambrosia beetles with regu-lar sibling mating. Evolution 59, 317–323.

62. Keller, L. (1995). Social life: the paradox of multiple-queen colonies.Trends Ecol. Evol. 10, 355–360.

63. Reeve, H.K., and Keller, L. (2001). Tests of reproductive-skewmodels in social insects. Annu. Rev. Entomol 46, 347–385.

64. Darwin, C. (1871). The Descent of Man and Selection in Relation toSex (New York: Appleton).

Special IssueR683

65. Shellman-Reeve, J.S. (1999). Courting strategies and conflicts ina monogamous, biparental termite. Proc. Roy. Soc. Lond. B 266,137–144.

66. Baer, B., Morgan, E.D., and Schmid-Hempel, P. (2001). A non-spe-cific fatty acid within the bumblebee mating plug prevents femalesfrom remating. Proc. Natl. Acad. Sci. USA 98, 3926–3928.

67. Mikheyev, A.S. (2003). Evidence for mating plugs in the fire antSolenopsis invicta. Insectes Soc. 50, 401–402.

68. Baer, B., and Boomsma, J.J. (2004). Male reproductive investmentand queen mating frequency in fungus growing ants. Behav. Ecol.15, 426–432.

69. Reichardt, A.K., and Wheeler, D.E. (1996). Multiple mating in the antAcromyrmex versicolor: a case of female control. Behav. Ecol.Sociobiol. 38, 219–225.

70. Wiernasz, D.C., Perroni, C.L., and Cole, B.J. (2004). Polyandry andfitness in the western harvester ant, Pogonomyrmex occidentalis.Molec. Ecol. 13, 1601–1606.

71. Koeniger, N., Koeniger, G., Gries, M., and Tingek, S. (2005). Dronecompetition at drone congregation areas in four Apis species.Apidologie 36, 211–221.

72. Thornhill, R., and Alcock, J. (1983). The Evolution of Insect MatingSystems (Cambridge, Massachusetts: Harvard University Press).

73. Hamilton, W.D. (1979). Wingless and fighting males in fig waspsand other insects. In Sexual Selection and Reproductive Competi-tion in Insects, M.S. Blum and N.A. Blum, eds. (New York: Aca-demic Press), pp. 167–220.

74. West, S.A., Murray, M.G., Machado, C.A., Griffin, A.S., and Herre,E.A. (2001). Testing Hamilton’s rule with competition between rela-tives. Nature 409, 510–513.

75. Heinze, J., Trindl, A., Seifert, B., and Yamauchi, K. (2005). Evolutionof male morphology in the ant genus Cardyocondyla. Mol. Phyl.Evol. 37, 278–288.

76. Heinze, J., and Holldobler, B. (1993). Fighting for a harem ofqueens: physiology of reproduction in Cardiocondyla male ants.Proc. Natl. Acad. Sci. USA 90, 8412–8414.

77. Cremer, S., and Heinze, J. (2003). Stress grows wings: environmen-tal induction of winged and dispersal males in Cardiocondyla. Curr.Biol. 13, 219–223.

78. Lenoir, J.C., Schrempf, A., Lenoir, A., Heinze, J., and Mercier, J.L.(2007). Genetic structure and reproductive strategy of the antCardiocondyla elegans: strictly monogynous nests invaded byunrelated sexuals. Molec. Ecol. 16, 345–354.

79. Simmons, L.W. (2001). Sperm Competition and Its EvolutionaryConsequences in the Insects (Princeton, NJ: Princeton UniversityPress).

80. Baer, B. (2005). Sexual selection in Apis bees. Apidologie 36, 187–200.

81. Tschinkel, W.R., and Porter, S.D. (1988). Efficiency of sperm use ofqueens of the fire ant, Solenopsis invicta (Hymenoptera Formici-dae). Ann. Entomol. Soc. Am. 81, 777–781.

82. Emlen, S.T. (1997). Predicting family dynamics in social verte-brates. In Behavioural Ecology: An Evolutionary Approach, FourthEdition, J.R. Krebs and N.B. Davies, eds. (Oxford, UK: BlackwellScientific Publications), pp. 228–253.

83. Koenig, W., and Dickinson, J. (2004). Ecology and Evolution of Co-operative Breeding in Birds (Cambridge, UK: Cambridge UniversityPress).

84. Cockburn, A. (2003). Cooperative breeding in oscine passerines:does sociality inhibit speciation? Proc. Roy. Soc. Lond. B 270,2207–2214.

85. Clutton-Brock, T.H., and Isvaran, K. (2006). Paternity loss in con-trasting mammalian societies. Biol. Letters 2, 513–516.

86. Griffin, A.S., and West, S.A. (2003). Kin discrimination and the ben-efit of helping in cooperatively breeding vertebrates. Science 302,634–636.

87. Griffith, S.C., Owens, I.P.F., and Thuman, K.A. (2002). Extra pairpaternity in birds: a review of interspecific variation and adaptivefunction. Mol. Ecol. 11, 2195–2212.

88. West, S.A., Griffin, A.S., and Gardner, A. (2007). Evolutionary expla-nations for cooperation. Curr. Biol. 17, R661–R672.

89. Keane, B., Waser, P.M., Creel, S.R., Creel, N., Elliott, L.F., and Min-chella, D.J. (1994). Subordinate reproduction in dwarf mongooses.Anim. Behav. 47, 65–75.

90. Griffin, A.S., Pemberton, J.M., Brotherton, P.N.M., McIlrath, G.,Gaynor, D., Kansky, R., O’Riain, J., and Clutton-Brock, T.H.(2003). A genetic analysis of breeding success in the cooperativemeerkat (Suricata suricatta). Behav. Ecol. 14, 472–480.

91. Faulkes, C.G., and Bennett, N.C. (2001). Family values: group dy-namics and social control of reproduction in African mole-rates.Trends Ecol. Evol. 16, 184–190.

92. Bennett, N.C., and Faulkes, C.G. (2000). African Mole-Rats: Ecol-ogy and Eusociality (Cambridge, UK: Cambridge University Press).

93. Braude, S. (2000). Dispersal and new colony formation in wildnaked mole-rats: evidence against inbreeding as the system ofmating. Behav. Ecol. 11, 7–12.

94. Bourke, A.F.G. (1999). Colony size, social complexity and repro-ductive conflict in social insects. J. Evol. Biol. 12, 245–257.

95. Wilson, D.S. (1990). Weak altruism, strong group selection. Oikos59, 135–140.

96. Foster, K.R., Wenseleers, T., and Ratnieks, F.L.W. (2006a). Kinselection is the key to altruism. Trends Ecol. Evol. 21, 57–60.

97. West, S.A., Griffin, A.S., and Gardner, A. (2007). Social semantics:altruism, cooperation, mutualism, strong reciprocity and group se-lection. J. Evol. Biol. 20, 415–432.

98. Clutton-Brock, T.H., and Parker, G.A. (1995). Punishment in animalsocieties. Nature 373, 209–216.

99. Wenseleers, T., and Ratnieks, F.L.W. (2006). Enforced altruism ininsect societies. Nature 444, 50.

100. Grafen, A. (1984). Natural selection, kin selection and group selec-tion. In Behavioural Ecology: An Evolutionary Approach, SecondEdition, J.R. Krebs and N.B. Davies, eds. (Oxford, UK: BlackwellScientific Publications), pp. 62–84.