1

Theevolutionofinfanticidebyfemalesinmammals

DieterLukas1,2*&EliseHuchard1,3

1)DepartmentofZoology,UniversityofCambridge,DowningStreet,CB23EJCambridge,U.K.

2)DepartmentofHumanBehaviour,Ecology,andCulture,MPIforEvolutionaryAnthropology,DeutscherPlatz6,04103Leipzig,Germany

3)InstitutdesSciencesdeL'EvolutiondeMontpellier,UMR5554,CNRS,UniversitédeMontpellier,Montpellier,France

*)Authorforcorrespondence:dieter_lukas@eva.mpg.de

ORCID:DieterLukas:0000-0002-7141-3545;EliseHuchard:0000-0002-6944-449X

Keywords:socialcompetition;sexualselection;phylogeneticcomparison;sociality;kinship

Abstract

Inmostmammalianspecies,femalesregularlyinteractwithkin,anditmaythusbedifficultto

understand the evolution of some aggressive and harmful competitive behaviour among

females,suchasinfanticide.Here,weinvestigatetheevolutionarydeterminantsofinfanticide

byfemalesbycombiningaquantitativeanalysisofthetaxonomicdistributionofinfanticide

withaqualitativesynthesisofthecircumstancesofinfanticidalattacksinpublishedreports.

Ourresultsshowthatfemaleinfanticideiswidespreadacrossmammalsandvariesinrelation

tosocialorganizationandlife-history,beingmorefrequentwherefemalesbreedingroups

andinvestmuchenergyintoreproduction.Specifically,femaleinfanticideoccurswherethe

proximity of conspecific offspring directly threatens the killer’s reproductive success by

limitingaccesstocriticalresourcesforherdependentprogeny,includingfood,shelters,care

orasocialposition. Incontrast, infanticide isnot immediatelymodulatedbythedegreeof

kinship among females, and females occasionally sacrifice closely related juveniles. Our

findingssuggest that thepotentialdirect fitness rewardsofgainingaccess to reproductive

resourceshaveastrongerinfluenceontheexpressionoffemaleaggressionthantheindirect

fitnesscostsofcompetingagainstkin.

2

Introduction

Recent work has emphasized that competitive strategies of female mammals are often

strikinglysymmetricaltothoseobservedinmales,includingdisplaysandornaments,fighting

andweaponry, dominance hierarchies, and reproductive suppression by evicting rivals or

suppressing their reproduction [Clutton-Brock2007,2013;Stockley&Bro-Jørgensen2011;

Clutton-Brock& Huchard 2013]. Interactions among conspecificmalemammals are often

contextualandtemporallylimitedtocompetitionoveraccesstomatingpartners,andstudies

ofmalesocialityhavethereforerevolvedaroundmodelsof intrasexualselection[Emlen&

Oring1977;Connor&Krützen2015].Incontrast,femalemammalsaretypicallyphilopatric,

with neighbouring females living in adjacent homeranges or sharing the same group

throughouttheirlives,sothatinteractionsamongfemalescommonlyinvolvekinandoccur

across extended periods and multiple settings [Clutton-Brock 2016]. Females may thus

competeoveradiversityofresources-includingfood,resourcesnecessarytobreed(burrows,

homerange),oroffspringcare[Clutton-Brock2007]-andoftendosowiththeirrelatives.It

has therefore proven difficult to identify the determinants of overt female-female

competition,whichcanvaryacrossspeciesandcontexts,especiallyinthecaseofextremely

harmfulbehavioursuchasinfanticide[Clutton-Brocketal.2001;Youngetal.2006].

Thekillingofrivals’offspringrepresentsaviolentmanifestationofintrasexualcompetition,

andasignificantsourceofjuvenilemortalityinsomepopulations[Palombit2012],withadults

ofbothsexescommittinginfanticide.Ithasbeenintenselystudiedinmalemammals,where

fiftyyearsoffieldresearchhaveshownthatithasevolvedasasexuallyselectedstrategyover

accesstomatingpartners.Incaseswherethepresenceofadependentoffspringpreventsthe

motherfrombecomingpregnantagain,committinganinfanticideallowsthekillertocreate

3

extrareproductiveopportunities inpolygynoussocietieswhereoneorafewalphamale(s)

monopolizematingopportunitiesovershortperiodsbeforelosingdominancetoothers[van

Schaik& Janson2000; Lukas&Huchard2014; Palombit 2015]. In contrast, little is known

aboutthedeterminantsandconsequencesofinfanticidebyfemalesotherthanthemother,

although itmaybemoreprevalent than infanticidebymales,bothwithinandacross taxa

[Blumstein2000;Digby2000].Unlikemales,femalekillersdonotbenefitfromextramating

opportunities[Agrelletal.1998;Digby2000],becausemalemammalsgenerallydonotinvest

intooffspringcaretotheextentthatitwouldpreventthemfrommatingwithotherfemales

[Kleiman&Malcom1981]. Ifanything,killingadependentjuvenilemayexacerbatefemale

matingcompetitionbyspeeding-uptheresumptiontofertilityforthemotherofthevictim.

Several plausible scenarios explaining the occurrence of infanticide by females have been

proposed based on a synthesis of natural observations [reviewed by Digby 2000].

Symmetricallytothepatternsobservedformaleinfanticide,predationfornutritionalgains

(H1:‘exploitation’hypothesis)maynotprovideageneralexplanationforfemaleinfanticide

askillershaverelativelyrarelybeenobservedtoconsumevictimspartiallyorentirely (e.g.

[Goodall 1986; Blumstein 2000]). Instead, hypotheses regarding the adaptive benefits of

femaleinfanticidehavefocusedonhowkillingsmightfacilitateaccesstoresourcesthatare

critical to successful reproduction (H2: ‘resource competition’ hypotheses) [Digby 2000].

Femalekillersmightbedefendingaccesstoanexclusiveterritoryorshelterwhentheytarget

victimsoutsidetheirhomerange(H2.1:‘breedingspace’hypothesis)(asinblack-tailedprairie

dogs [Hoogland 1985] or Belding’s ground squirrels [Sherman 1981]). In species where

femalesonlyassociatetemporallytobreed,killersmaydefendaccesstotheirownmilk,by

discouragingattemptstosucklefromunrelatedjuveniles(H2.2:‘milkcompetitionhypothesis)

(asinNorthernelephantseals:[LeBoeufetal.1973]). Inspecieswhobreedcooperatively,

4

killersmaydefendaccesstoextraoffspringcarebygroupmatesotherthanthemotherby

alteringthehelper-to-pupratiointheirowngroup(H2.3:‘allocare’hypothesis)(asinmeerkats

[Clutton-Brocketal.2001;Youngetal.2006],bandedmongooses[Gilchrist2006;Cantetal.

2014],ormarmoset [Digby1995]).Finally, inspecieswho live instablegroups,killersmay

defend their offspring’s future social status (in species with stable hierarchies) or group

membership (in species with forcible evictions) by eliminating future rivals (H2.4: ‘social

status’hypothesis)(asinsomeOldWorldprimates[Hrdy1976;Digby2000]).

Here, we present an investigation of the distribution and circumstances of infanticide by

femalemammals,basedondatafrom289speciescollectedfromtheprimaryliterature.The

combinationofaquantitative synthesisof the taxonomicdistributionof infanticidewitha

qualitativeanalysisofthecircumstancesof infanticidalattacks(includingtraitsofthekiller

and victim) can contribute to reveal the ecological, life-history or social determinants of

female reproductive competition across mammalian societies, and their relevance to the

occurrence of female associations and interactionswithin and amongmatrilines.We first

summarizethesocialorganisationandlife-historiesofspeciesinwhichinfanticidebyfemales

hasbeenobserved,inordertoevaluatetheconditionsunderwhichreproductivecompetition

among females appears to be particularly intense (Table 1). Next, we perform specific

phylogenetic analyses to test core predictions generated by each hypothesis to assess

whetherfemaleshavebeenobservedtocommitinfanticideinspeciesinwhichtheyaremost

likely to benefit from such killings. In addition, we investigate whether population-level

informationonthetraitsofkillersandvictimsarecompatiblewithpredictionsgeneratedby

eachhypothesis.AllourpredictionsandtestsaresummarizedinTable2.Wefirstshowthat,

acrossallspecies,thedistributionandoccurrenceofinfanticidebyfemalesisbetterexplained

5

byresourcecompetitionthanbyexploitation.Wenexttestwhetherinstanceswherefemales

killoffspringinneighbouringranges(‘breedingspacehypothesis’)aremostlikelyexplainedby

competitionover breeding space; instanceswhere females kill offspringborn in the same

breeding association by competition over milk (‘milk competition hypothesis’); instances

where females kill offspring in groups where usually only a single female reproduces by

competition over offspring care (‘allocare hypothesis’) ; and instances where females kill

offspringborn ingroupswithmultiplebreedersbycompetitionoversocialstatusorgroup

membership (‘social status hypothesis’). This analytical framework relies on a rough

categorizationofthecircumstancesoffemaleinfanticide,andtheoccurrenceofinfanticidein

agivenspeciesmaybeexplainedbymultiplereasons,butouraimistoprovideastartingpoint

fortheinvestigationofthelikelycausesandsituationsunderwhichfemaleinfanticidemight

occur.

6

Materials&Methods

FollowingDigby[2000],weuseabroaddefinitionofinfanticideas‘anactbyoneormorenon-

parentsthatmakesadirectorsignificantcontributiontotheimmediateorimminentdeathof

conspecificyoung’.Thisdefinitionexcludesmatricidesandincludescaseswhereinfantsdieas

theresultofthephysicalaggression(directinfanticide)aswellascaseswheretheenforced

neglect of an infant, such as kidnapping, ultimately causes death (indirect infanticide).

Althoughthelattercasesareoftenexcludedfromstudiesofinfanticideduetotheirproximate

formof‘overzealous’allomaternalcare[Hrdy1976],theirultimateconsequence-infantdeath

- contributes to shape their evolution as infanticidal behaviour. We included infanticide

records from both wild and captive populations for which the killer was unambiguously

identifiedasanadultfemale.Speciesforwhichnocaseofinfanticidehaseverbeenobserved

wereincludedonlyifdetailedobservationsonindividualfemalesandjuvenileswereavailable,

eitherfromrepeatedcaptiveobservationsorfromfieldstudiesoccurringacrossatleastthree

reproductiveseasons,tominimizetheriskofmisclassifyingthemas“non-infanticidal”.For

eachspecies,werecordedwhetherobservationsoccurredinacaptivesettingorundernatural

conditions.Datawere collected from systematic searches through the scientific literature,

startingwithmajorreviewsonthetopicoffemaleinfanticide[Wasser&Barash1983;Agrell

etal.1998;Ebensperger1998;Blumstein2000;Digby2000;Ebensperger&Blumstein2007;

Stockley&Bro-Jorgensen2011;Clutton-Brock&Huchard2013]andperformingbackwardand

forwardcitationsearchestoidentifyrelevantobservations.

For thecomparativeanalyses,weextracteddata foreachspecies inoursampleon: social

organisation (classified as: solitary breeders [home-ranges of breeding females do no

completelyoverlapwithanyotherbreedingindividual],pairbreeders[home-rangescontain

7

asinglebreedingfemaleandasinglebreedingmaleoverlapbutmaycontainadditionalnon-

breeding individuals],associatedbreeders [femalessharethesamespace forbreedingbut

associationsareunstableandtendnottolastbeyondthebreedingseason],orsocialbreeders

[severalbreedingfemalessharethesamehome-rangeacrossmultiplebreedingseasons])[

Lukas & Clutton-Brock 2017]; female philopatry and dispersal (whether most breeding

femaleshavebeenbornintheircurrentlocality/grouporelsewhere)[Lukas&Clutton-Brock

2011];carnivory(whetherthedietofaspeciesincludesmeatornot)[Wilmanetal.2014];

infanticidebymales(whethermaleshavebeenobservedtokillconspecificyoung)[Lukas&

Huchard2014];environmentalclimaticharshness(aprincipalcomponent,withhighvalues

indicatingthatrainfallislowandtemperaturesarecoldandunpredictableacrosstheknown

rangeofaspecies)[Boteroetal.2014];maternalinvestment(meanbodysizeofoffspringat

weaning times mean number of offspring per year divided by mean body mass of adult

females)[Siblyetal.2014];theuseofburrowsornestholesforbreeding(informationwas

takenfromthepapersusedtoextractinformationontheabsenceorpresenceofinfanticide

byfemales);littersize(numberofoffspringperbirth);offspringmassatbirth(grams);weaning

age (age in days at which offspring are independent); inter-birth interval (time between

consecutivebirths indays)Jonesetal.2009];energeticvalueofmilk (MJ/mlbasedonthe

protein,sugar,andfatcomposition)[Langer2008;Barton&Capellini2011;Hinder&Milligan

2011]; offspring care by fathers and/or non-parental group members (whether offspring

receivesmilk or food, or are being regularly carried, by groupmemberswho are not the

mother) [Lukas & Clutton-Brock 2017]; dominance hierarchies and mechanisms of rank

acquisition in social groups (whether all adult females can be arranged in a dominance

hierarchyandifso,whetheranindividual’srankisinfluencedbyageand/ornepotism);and

forcibleevictions(whetherfemalesuseaggressiontoexcludeotherfemalesfromtheirown

8

socialgroup).Foreachspeciesinwhichfemaleshadbeenobservedtokillconspecificyoung,

we used the primary literature to record as much information as possible regarding

characteristics of the killer (age and reproductive state) and of the victim (age, sex, and

relatednesstokiller)totestspecificpredictions.ThefulldatasetisprovidedinSupplementary

Table1(comparativedata)andSupplementaryTable2(individualcharacteristicsdata),with

allreferencesfordataspecificallycollectedhereinSupplementaryFile1.

In addition to performing comparisons assessing contrasts in the presence or absence of

infanticideacrossallspeciesinoursample,weclassifiedspeciesintodifferenttypesaccording

to each of the four resource competition hypotheses (Table 2): for the breeding space

hypothesis, we only included instances of infanticide in which females did not share a

homerangewiththemotherofthevictims;forthemilkcompetitionhypothesis,werestricted

the sample to associated breeders; for the allocare hypothesis, we only included pair

breeders;andforthesocialcompetitionhypothesis,weonlylookedatsocialbreeders.

For the comparative analyses, the phylogenetic relatedness between specieswas inferred

fromtheupdatedmammaliansupertree[Rollandetal.2014].Wefittedseparatephylogenetic

modelsusingMCMCglmm[Hadfield&Nakagawa2010]toidentifytheextenttowhicheach

of thepredictedvariables (Table2)explains thepresenceof infanticideby femalesacross

species (binary response,assumingacategorical familyof traitdistribution).Following the

recommendationsofHadfield[2010],wesetthepriorsusinganuninformativedistribution

(withvariance,V,setto0.5andbeliefparameter,nu,setto0.002).Eachmodelwasrunthree

timesfor100,000iterationswithaburn-inof20,000,visuallycheckedforconvergenceand

foragreementbetweenseparateruns.

9

Results

Socialorganisationandinfanticidebyfemales

Infanticidebyfemaleshasbeenobservedin89(31%)ofthe289mammalianspeciesinour

sample(Table1).Femaleinfanticide(ofanytype)varieswiththesocialorganisationandis

more frequentwhen females breed in groups (Figure 1): it has been observed in 43% of

associatedbreeders,in36%ofpairbreeders,andin30%ofsocialbreeders,butonlyin18%

ofsolitarybreeders.Acrossallspecies,femalesareequallylikelytokilloffspringwhenthey

arephilopatric(47of135species,34%)thanwhentheydispersetobreed(17of59species,

29%)(effectoffemaledispersalonpresenceofinfanticidebyfemales:-10.1,95%CI-39.3–

11.3,p=0.34)buttherearedifferencesfortwotypesofsocialorganisation:acrossassociated

breedersinfanticideonlyoccursinphilopatricspecies;whileacrosspairbreedersinfanticide

ismorelikelytooccurinspeciesinwhichfemalesdisperse(Table1).Acrossallgroup-living

species (associated breeders, pair breeders with helpers, social breeders), there is no

relationship between levels of average relatedness among female group members and

whether infanticidebyfemalesofoffspringborninthesamegroupdoes(medianlevelsof

averagerelatednessacross10species0.09,range0.01-0.38)ordoesnotoccur(medianlevels

ofaveragerelatednessacross24species0.21,range-0.03-0.52)(effectoflevelsofaverage

relatednessonpresenceofinfanticidebyfemales:-21.1,95%CI-81.4–10.1,p=0.18).Across

species inwhichgroupsaremore stable (excludingassociatedbreederswheregroupsare

sometimesdifficulttodefineandcanbeverylarge),levelsofaveragerelatednessareslightly

higherwheninfanticideoccurs(seealso[Lukas&Clutton-Brock2018]).Thepopulation-level

informationshowinstancesofkillersbeingclosekinofthevictimin33%ofspecies(22/65),

witheithergrandmotherskillingtheirgrandchildrenorauntskillingtheirnieces.

10

Life-historiesandinfanticidebyfemales

Energetic investment into reproduction bymothers is higher in species with any form of

infanticidebyfemalescomparedtotheremainingspecies,withspecieshavinglargeroffspring

massatbirth,shortertimetoweaningandbetweenbirths,andlargerlittersizes(Table1).

However,thesepatternsencounteredacrossallspeciesdonotreflectageneralassociation

between infanticideandall thesetraits,but rather the fact thateachdifferentscenarioof

infanticideisassociatedwithspecificlifehistories,characterizedbyoneoracoupleofthese

traits.Femalesininfanticidalspecieswherethereisusuallyasinglebreedingfemaleperhome

range (extraterritorial infanticide and infanticide in pair breeders) have larger litters than

females inspecieswithout infanticide(Table1).Species inwhichfemaleskilloffspring ina

breedingassociationarecharacterizedbyfast-growingoffspring,whileoffspringarerelatively

smallatbirthinspeciesinwhichfemaleskilloffspringinstablegroups(Table1).

11

Old W

orld Primates

Hares

Squirrels

MarmotsMiceMarsupialsBats

Mongooses

Felids

Canids

Bears

Seals & sealions

Martens

Ungulates

New World monkeys

Great apes

Lemurs

extraterritorial infanticideinfanticide: associated breederinfanticide: pair breederinfanticide: social breeder

solitary breederpair/coop. breederassociated breedersocialbreeder

Figure1:Thedistributionofthedifferentformsofinfanticidebyfem

alesinrelationtothesocialorganisationacrossthemam

malianspeciesinour

sample(forpicturecreditseeSupplem

entaryFile2).

12

Table1:Thesocialandlife-historyconditionsassociatedwiththedistributionofinfanticidebyfem

ales

Femalesarem

orelikelytocommitinfanticideingroup-livingspecies,butw

hetherbreedingfemalesarephilopatricordispersedoesnotappearto

beassociatedwiththedistributionofinfanticide.Ingeneral,speciesw

ithinfanticidebyfemalesarecharacterizedbyhighm

aternalinvestment,

withfem

aleshavinglargelitters,shortlactationalandinterbirthperiods,andrelativelylargeoffspring.

Sample of species

Type of infanticide

Infanticide isabsent

presentabsent

presentabsent

presentabsent

presentabsent

presentSam

ple size2

00

89

25

33

33

41

92

91

69

22

0

Fe

ma

les s

olita

ry (%

sp

ecie

s)

38

%1

6%

39

%7

6%

NA

NA

NA

NA

NA

NA

Fe

ma

le p

hilo

pa

tric (%

sp

ecie

s)

73

%6

8%

70

%7

0%

65

%1

00

%4

8%

31

%6

4%

63

%

Ma

tern

al in

ve

stm

en

t (pe

r ye

ar re

lativ

e to

bo

dyw

eig

ht)

39

%1

18

%4

7%

15

0%

48

%8

6%

11

0%

16

9%

29

%4

5%

Litte

r siz

e1

.62

.61

.73

.91

.52

.11

.94

1.3

1.3

Inte

rbirth

inte

rva

l (da

ys)

36

51

80

34

01

04

30

82

12

30

81

81

45

35

22

Ag

e a

t we

an

ing

(da

ys)

12

76

11

13

35

10

75

91

05

42

22

92

73

Offs

prin

g w

eig

ht (re

lativ

e to

bo

dyw

eig

ht)

3.6

%3

.9%

3.9

%2

.8%

3.5

%5

.9%

5.1

%2

.8%

5.1

%3

.0%

allall

associated breederspair breeders

social breeders

any extraterritorial

within-group

within-group

within-group

13

Table2:Testingthecorepredictionsgeneratedbythedifferenthypothesesproposedtoexplainthedistributionofinfanticidebyfemales

Foreachofthemainhypotheses,w

etestedtwocorepredictionsinphylogeneticcom

parisonsandtwopredictionsabouttheindividualtraitsfrom

thefieldobservations.Forthecomparisons,w

elistthesampleofspeciesincluded.

Type of infanticide /Predictions

Sample of species

on individual traitsany form

of infanticide - prim

arily in carnivoresN

okiller: any reproductive state

No

across all species - infanticide also by m

alesN

ovictim

: any ageN

oany form

of infanticide - harsher environm

entsYes

killer: gestating/lactatingYes

across all species - higher m

aternal investment

Yesvictim

: dependent on careYes

extraterritorial infanticide - m

ore burrow use

Yeskiller: gestating/lactating

Yesacross all species

- exclusive home-ranges

Yesvictim

: unweaned

Yesw

ithin group infanticide - higher energetic m

ilk contentN

okiller: lactating

Yes

across associated breeders - faster offspring grow

thM

aybevictim

: unweaned/killed during

nursing attempt

Yes

within group infanticide

- allocarers presentYes

killer: gestating/lactatingYes

across pair breeders - higher offspring to carer ratio

Yesvictim

: dependent on careYes

within group infanticide

- nepotistic hierarchy presentYes

killer: high social rankYes

across social breeders - evictions occur

Yesvictim

: any ageYes

H2.4: over social status

Core prediction(s)

Support?Support?

Hypothesis

H1: exploitation

H2: resource com

petition

H2.1: over space

H2.2: over m

ilk

H2.3: over allocare

14

-1.5(benign)

2.5(harsh)

environmental harshness

infanticide by females

absentpresent

5% 25% 100% 1000%

maternal energetic investmentinfanticide by fem

alesabsent

present

a)b)

Figure2:Factorsassociatedwithfem

alecompetitionandthedistributionofinfanticidebyfem

ales

Speciesinwhichfem

aleinfanticideispresentare,onaverage,characterizedby(a)livinginharsherenvironments(low

erandmoreunpredictable

rainfallandtemperatures)andby(b)higherm

aternalenergeticinvestment(totalm

assofweanedoffspringproducedperyearrelativetom

aternal

mass).Blacklinesindicatethem

edianacrossthespeciesinthesample,boxescontain75%

ofthevalues,andwhiskersextendtotheextrem

es.

15

H1:Exploitation

Wefindnosupportforpredictionssuggestingthatfemaleskillconspecificoffspringprimarily

forexploitation (Table2).Across species, infanticideby females isas likely tooccur in the

absenceofinfanticidebymales(44of147species,30%)asinitspresence(43of135species,

32%)(effectofpresenceofmaleinfanticideonpresenceoffemaleinfanticide2.1,95%CI-

13.5–16.9,p=0.74).Similarly,carnivorousspeciesarenotmorelikelytoshowinfanticideby

females(18of56species,32%)thaninspeciesinwhichmeatdoesnotconstituteanimportant

partofthediet(59of192species,31%)(effectofcarnivoryonpresenceoffemaleinfanticide

-0.7,95%CI-11.2–10.0,p=0.89).Theageofvictimsvariesfrombirthtobeyondindependence

acrossspecies,butismorehomogeneouswithineachtypeofinfanticide(seebelow),sothat

killingsdonotappearsimplyopportunistic.

H2:Resourcecompetition

Infanticide by females appearsmore likely to occur where competition over resources is

expectedtobemoreintense(Table2).Theclimaticenvironmentsofspeciesinwhichfemales

commit infanticide are harsher (as estimated by a principal component reflecting low

unpredictablerainfallandcoldseasons)thantheenvironmentsofspeciesinwhichinfanticide

hasnotbeenobserved(effectofenvironmentalharshnessonpresenceoffemaleinfanticide

7.0,95%CI -0.2–14.7,p=0.03,54specieswith infanticideand193without)(Figure2a). In

specieswhere females commit infanticide, they invest substantiallymore energy into the

productionofoffspring,beingable toproduce theequivalentof1.0 times theirownbody

massinoffspringmassperyear(numberofoffspringtimesmassofweanedoffspring;median

across 41 species, range 0.05 – 12.1 times) compared to 0.33 times in species in which

infanticidehasnotbeenobserved (medianacross77species, range0.03–11.1) (effectof

16

maternal energetic investment onpresenceof female infanticide 25.8, 95%CI 1.5 – 58.7,

p<0.001)(Figure2b).

H2.1:Competitionoverbreedingspace

Thirty-twoofthe33speciesinwhichfemaleskilljuvenilesoutsidetheirownhome-rangekeep

theiroffspringinburrowsorholes,comparedto93ofthe163speciesinwhichinfanticideby

femalesappearsabsent(effectofburrowuseonthepresenceofinfanticidebyfemales14.4,

95% CI 6.0 – 22.9, p<0.001). The exception is Semnopithecus entellus, where “females

occasionally steal infants from a neighboring troop” [Hrdy 1976; Mohnot 1980]. In most

species in which females kill offspring in neighbouring home-ranges (25 of 33), females

generallyappearnottotolerateotherbreedingfemalesclosebyandmosthome-rangesonly

containasinglebreedingfemale(solitaryorcooperativelybreedingspecies),whileinmost

otherspecies females formassociationsorgroups (home-rangescontaina singlebreeding

femalein105outof268species)(effectofpresenceofasinglebreedingfemaleperhome-

rangeonpresenceofinfanticidebyfemales12.4,95%CI0.3–30.1,p=0.007).Inallcases,the

killerwaseitherpregnantorhaddependantyoungofherown(17specieswithobservations),

andalloffspringthatwerekilledwerenotyetweaned(17species).

H2.2:Competitionovermilk

Acrossassociatedbreeders, theenergycontentofmilkproducedbymothers in species in

which females have been observed to kill offspring within the same breeding space (2.8

MJ/100ml,medianacross8 species, range1.2-4.7) doesnotdiffer from thatof species in

whichsuchkillingshavenotbeenobserved(2.1MJ/100ml,medianacross7species,range

0.8-5.7)(meaneffectofmilkenergyonpresenceofinfanticidebyfemales2.0,95%CI-66.3–

17

83.1,p=0.97).Inassociatedbreederswithfemaleinfanticide,offspringdonotseemtohave

greater growth rates (they gain on average 0.28% of their adult bodymass per day until

weaning,medianacross9species,range0.04%-1.72%)thaninspeciesinwhichfemaleshave

notbeenobservedtokilljuveniles(offspringgainonaverage0.17%oftheiradultbodymass

perdayuntilweaning,medianacross11species,range0.05%-0.73%)(effectofpresenceof

infanticidebyfemalesonoffspringgrowthrate6.1,95%CI-16.2–35.4,p=0.59).Killersare

eitherpregnantorhavedependantyoung(21outof21species)andareold,withmostreports

suggesting that they had been observed to give birth in previous years (they are not

primiparous).Allvictimswerereportedtobeunweaned(22outof22species).In7outof14

infanticidereportsfromassociatedbreeders,victimswerekilledastheyattemptedtobreast-

feedfromthekiller.

H2.3Competitionoverallomaternalcare

Infanticidebyfemalesinpairbreedersoccursonlywhenfathersprovidecare(all16species)

whilefatherscareforoffspringinonly16ofthe39pairbreedingspeciesinwhichthisformof

infanticideisabsent.In15ofthe16pairbreederswithfemaleinfanticide,additionalhelpers

arepresent(cooperativebreeders),whilethereareonlyafurther7cooperativelybreeding

species in which females have not been observed to kill juveniles from their own group

(femaleinfanticideoccursin68%[15/22]ofcooperativebreedersversusin4%[1/23]ofpair

breeders in which there are no other helpers). Across species in which offspring receive

allocare,thenumberofpotentialallocarersishigherinspecieswith(3allocarerspergroup,

medianacross15species,range2-23)comparedtospecieswithout(2allocarerspergroup,

medianacross13species,range1-20)femaleinfanticide(effectofnumberofallocarerson

presenceofinfanticide25.3,95%CI1.9–48.6,p=0.02).Thekillerwasusuallythedominant

18

breeder(aswasregularlythecasein9of12species)andwaspregnantorwithdependent

infantsinallcases.Inmostinstancesthekillerandthevictimbelongedtothesamegroup(11

of14species),andwereconsequentlyrelated(10outof13species).Victimswereoftenafew

daysoldandalldependent.

H2.4:Competitionoversocialstatus

Ofthe27socialbreederswithfemalephilopatry(andavailabledata),femalegroupmembers

donotformsocialhierarchiesinthreespecies,hierarchicalrankisdeterminedbyageinfive

species,andrankisinfluencedbynepotismin19species.Femaleshavebeenobservedtokill

offspringborntoothergroupmembersineightoftheselatter19species,butinnoneofthe

specieswherenepotismdoesnotinfluencefemalerank(effectofpresenceofnepotisticrank

acquisition on presence of infanticide by females 134.1, 95% CI 28.8 – 238.1, p<0.001).

Femalesaremorelikelytokilloffspringborntootherfemalesinsocialbreedersinwhichthey

alsoaggressivelyevictotherfemalesfromtheirgroup(infanticidehasbeenobservedin6of

10specieswithevictionsand7of35withoutevictions)(effectofoccurrenceofevictionson

presenceofinfanticidebyfemales7.8,95%CI-0.2–14.4,p=0.02).Inalltwelvesocialbreeders

inwhichinfanticideeventshavebeenobserved,killerswereoldandhigh-ranking.Killerswere

neverpregnant, but in all caseshaddependant youngof their own.Victimswerenot yet

weaned,andin5outof12speciesvictimsmightberelatedtothekiller.Thereisonlyone

specieswherethedatasuggestthatfemalesmightpreferentiallykilloffspringofonesex:in

Macaca radiata (a species with female philopatry), female offspring appear to be the

predominantvictims.

19

Discussion

Ourfindingsestablishthatfemalecompetitioniswidespreadacrossmammalsandthatitis

frequentlyexpressedasintenselyascompetitionamongmales.Femaleshavebeenobserved

tokillconspecificjuvenilesinvariousspeciesandourcomparativeanalysesprovidesupport

to the idea that this behaviour may be adaptive under a wide range of circumstances.

Infanticide ismore likely tooccur in species inwhichmultiple adult females liveorbreed

togetherthanwherefemalesbreedsolitarily,andinfanticideappearsmostfrequentinspecies

where females only associate temporarily to breed. This may reflect the fact that

opportunitiestocommitandtoobserve infanticidesmaybegreaterwherefemales liveor

breedtogether.Withineachtypeofsocialorganisation,wedohoweverfindthatfemales,like

males,appeartocommitinfanticidewhenthepresenceofthevictimmightotherwiselimit

theirownreproductivesuccess.Whileinfanticidebymaleshasevolvedinresponsetoasingle

cause -mate competition - acrossmammals [Lukas&Huchard 2014; Palombit 2015], the

evolutionarydeterminantsofinfanticidebyfemalesareapparentlymorecomplex,asfemales

maycompeteovermultipleresources.

Several lines of evidence support this adaptive scenario of resource competition for the

occurrenceof infanticideby females.First, infanticideappearsassociatedwithvariation in

ecologyand life-history. Specifically, it ismost frequentlyobserved in species facingharsh

climatic conditions and making the greatest reproductive efforts; it is unlikely that such

associationsareduetovariationsinopportunitiestoobserveorcommitinfanticidesacross

species. Rather, the potential costs of sharing critical resourcesmight outweigh the risks

associatedwithcommittinginfanticideinsuchcircumstances.

20

Second, specific ecological determinants of female infanticide identified at the population

level by field studies also seem to predict its distribution across species. Extraterritorial

infanticideswerefoundtobemostfrequentinsolitaryspecieswherefemalesuseburrowsto

givebirthandterritoriestoraiseoffspring,allowingkillerstofree-upreproductivespacefor

theirownoffspring.Ourfindingsfurthershowthatfemaleinfanticideoccursinpairbreeders

wherehelpers–fathersoradditionalgroupmates-arepresent.Finally,patternsareslightly

more complex in social breeders. There, infanticidepreferentially occurs in specieswhere

aggressivecompetitionamongfemalesleadstotheevictionofsomeindividuals–generally

youngadults-fromthegroup,especiallyattimeswhengroupsizeincreases(e.g.[Kappeler

&Fichtel2012]).Insuchcases,killingunrelatedjuvenilesmaylimitfuturecompetitionand

therelatedriskofbeingevictedforthekiller’soffspring.Inaddition,insocialbreederswhere

femalesarephilopatric,infanticidewasonlyfoundtooccurwherefemalerankacquisitionis

nepotistic,ahierarchicalsystemwhereeachadditionaloffspringmaycontributetostrengthen

thesocialstatusofamatriline–andwhereinfanticidemayconsequentlyweakencompeting

matrilinesonthelongterm.

Anecdotalreportsoffemalepinnipedskillingorphansastheyattemptedtobreast-feedfrom

them inspired the hypothesis that females compete overmilk in specieswhere they only

associatetobreed[Digby2000].Whileourcomparativeanalysesdidnotrevealanydifference

in theenergycontentofmilkofassociatedbreeders inwhich infanticide ispresentversus

absent, associated breeders nevertheless comprise the species with the highest energy

contentofmilkandthefastestgrowthrates,andwefurtherfoundthatoffspringareweaned

atanearlierageinassociatedbreederswithinfanticidecomparedtothosewithoutit.The

lackofsupport forthemilkcompetitionhypothesis inouranalysesmaybeexplainedbya

21

noisydataset,wheretheabsenceofinfanticideinsomespeciesmaybeduetothefactthatit

goes undetected if it is hard to observe, or to the evolution of counter-adaptations that

protectoffspringagainstinfanticide.Alternatively,milkisnottheonlyresourceoverwhich

these femalescompete.Forexample, in the largebreedingcoloniesofpinnipeds, space is

sometimesveryrestricted[Baldietal.1996],especiallyintheimmediatevicinityoftheharem

leaders.Thesebullsoftenprotecttheirfemalesandcalvesfromattacksbyyoungermales,and

mayrepresentanothersourceofcompetitionforlactatingfemales.

It is likely that, in any given species, infanticide may be triggered by more than one

determinant-includingsomethatmaynotbeconsideredhere.Akillermayaccordinglyget

multiplebenefitsfromoneinfanticideevent,butmayalsocommitinfanticidesinmorethan

one context. For example, half of the species of pair breeders committing intra-group

infanticides also commit extraterritorial infanticides. It is therefore possible that different

types of female infanticide – following our classification - have followed a common

evolutionarypath.Specifically, it ispossible that infanticidalbehaviour initiallyemerged in

response to one particular pressure (e.g., competition over access to allocare) in a given

species,whichsubsequentlystartedtoextenditsexpressiontoothercompetitivecontexts

(e.g., competition over breeding territories). However, the limited number of species for

whichobservationaldataoninfanticideareavailable,aswellasheterogeneitiesinthesample

– such as an over-representation of group-living species – introduce uncertainty when

attemptingtoreconstitutetheevolutionaryhistoryofthetrait.Itisconsequentlyhardtoinfer

theancestralstate,whethereachinfanticidetypehasevolvedindependently,orhowmany

timesinfanticidalbehaviourhasemergedacrossmammals.

22

Inadditiontothenatureoftheresourcesthatmaydirectlylimitfemalereproductivesuccess

in various types of social organisations, contrasts in the occurrence of infanticide across

species reveal other broad patterns on female reproductive competition in mammals. In

particular,thelackofassociationbetweenfemaleinfanticideandphilopatryacrossspecies

(Table1),aswellasasynthesisofobservationsrevealingthatkillersandvictimsarecommonly

relatedinsomecontexts,suchasinpairbreederswherereproductivesuppressioniscommon

[Lukas&Clutton-Brock2018],suggestthatmatrilinealityandsubsequentincreasesinaverage

kinshipamongassociated femalesdoesnotnecessarily lead toa reduction in competition

among females. Some previous work suggested that mammalian females might be

predisposedtobehavepositivelyandcooperativelywithkin[diFiore&Rendall1994],such

thatspecieswithfemalephilopatrywouldbecharacterizedbystablesocialbonds[Silk2007].

However,thefactorsleadingtolimiteddispersalandthespatialassociationofkinfrequently

alsoresultinhighlocalcompetition[Frank1998]whichcanovercomethepotentialbenefits

ofcooperationamongkin [Westetal.2002].Studiesofcompetitionamongmales insuch

circumstanceshaveshownthatcontrastsinlevelsofaggressioncanbeexplainedbyvariation

inthepotentialdirectfitnessbenefitsofwinning[Westetal.2001],anditislikelythatthis

alsoappliestotheobservedpatternofinfanticidebyfemales–wherethedirectbenefitsof

infanticide in terms of increased access to a critical resource might outweigh its costs,

includingtheindirectfitnesscostsassociatedwithkillingrelatedjuveniles.

Ourstudycompilesfivedecadesofbehaviouraldataacrossspeciesandwithinpopulationsto

elucidate the determinants of infanticide by mammalian females, which are less well

understood than those of male infanticide. Our analyses suggest that the distribution of

female infanticide across species is a consequence of contrasts in social organisation;

23

infanticide is most frequent in species that breed in groups, which probably have more

opportunities for killings and also face greater breeding competition. Female infanticide

occurswheretheproximityofconspecificoffspringdirectlythreatensthekiller’sreproductive

successby limitingaccess to critical resources forherdependentprogeny, including food,

shelters, care or a social position. Finally, these data support the idea that female killers

occasionally sacrifice related juvenile conspecifics, andmay therefore actively harm their

indirectfitnessinordertomaximizetheirdirectfitness.

Acknowledgements

We thank Tim Clutton-Brock, Peter Kappeler, Oliver Höner and Eve Davidian for useful

discussions.

DataAccessibility

All data are provided in the supplementarymaterial and are deposited at the Knowledge

NetworkforBiocomplexitydoi:10.5063/F1ZG6QFR.

Authors’Contributions

Bothauthorscontributedtotheconceptionanddesignofthestudy,collectedandinterpreted

thedata,wrotethearticle,andgavefinalapproval;theanalysesweredonebyDLwithinput

fromEH.

CompetingInterests

Wehavenocompetinginterests.

24

Funding

DLwassupportedbytheEuropeanResearchCommission(grant294494-THCB2011toProf.T.

Clutton-Brock)andtheMaxPlanckSociety.EHwasfundedbytheANR(grantANR-17-CE02-

0008)duringthewriting-upofthismanuscript.

SupplementaryMaterial

This submission includes all data in Supplementary Tables 1 (comparative data) and 2

(population observations), with references for observations on infanticide by females in

SupplementaryFile1,andcreditsfortheanimaldrawingsusedinFigure1inSupplementary

File2.

References

1. Clutton-BrockT.2007SexualSelectioninMalesandFemales.Science318,1882–1885.(doi:10.1126/science.1133311)

2. Clutton-BrockT,HuchardE.2013Socialcompetitionanditsconsequencesinfemalemammals.J.Zool.289,151–171.(doi:10.1111/jzo.12023)

3. Clutton-BrockTH,HuchardE.2013Socialcompetitionandselectioninmalesandfemales.Philos.Trans.R.Soc.BBiol.Sci.368.(doi:10.1098/rstb.2013.0074)

4. StockleyP,Bro-JørgensenJ.2011Femalecompetitionanditsevolutionaryconsequencesinmammals.Biol.Rev.Camb.Philos.Soc.86,341–366.(doi:10.1111/j.1469-185X.2010.00149.x)

5. EmlenST,OringLW.1977Ecology,sexualselection,andtheevolutionofmatingsystems.Science197,215-223.(doi:10.1126/science.327542)

6. ConnorRC,KrützenM.2015MaledolphinalliancesinSharkBay:changingperspectivesina30-yearstudy.Anim.Behav.103,223-235.(doi:10.1016/j.anbehav.2015.02.019)

7. Clutton-BrockTH.2016MammalianSocieties.NewYork:WileyandSons.

8. YoungAJ,Clutton-BrockT.2006Infanticidebysubordinatesinfluencesreproductivesharingincooperativelybreedingmeerkats.Biol.Lett.2,385–387.(doi:10.1098/rsbl.2006.0463)

25

9. Clutton-BrockTHetal.2001Cooperation,Control,andConcessioninMeerkatGroups.Science291,478–481.(doi:10.1126/science.291.5503.478)

10.PalombitR.2012Infanticide:malestrategiesandfemalecounter-strategies.InTheEvolutionofPrimateSocieties,pp.432–468.Chicago:MitaniJC,CallJ,KappelerPM,PalombitRA,SilkJB.

11.vanSchaikCP,JansonCH.2000InfanticidebyMalesandItsImplications.CambridgeUniversityPress.

12. LukasD,HuchardE.2014Theevolutionofinfanticidebymalesinmammaliansocieties.Science346,841–844.(doi:10.1126/science.1257226)

13.PalombitRA.2015InfanticideasSexualConflict:CoevolutionofMaleStrategiesandFemaleCounterstrategies.ColdSpringHarb.Perspect.Biol.7.(doi:10.1101/cshperspect.a017640)

14.DigbyL.2000Infanticidebyfemalemammals:implicationsfortheevolutionofsocialsystems.InInfanticidebymalesanditsimplications,vanSchaikCPandJansonCH.

15.BlumsteinD.2000Theevolutionofinfanticideinrodents:acomparativeanalysis.pp.178–198.Cambridge:vanSchaikCPandJansonCH.

16.AgrellJ,WolffJO,YlönenH.1998Counter-StrategiestoInfanticideinMammals:CostsandConsequences.Oikos83,507–517.(doi:10.2307/3546678)

17.KleimanDG,MalcolmJ.1981TheEvolutionofMaleParentalInvestmentinMammals.InParentalcareinmammals,GübernickDJ;KlopferPH.

18.GoodallJ.1986ThechimpanzeesofGombe:patternsofbehavior.BelknapPressofHarvardUniversityPress.

19.HooglandJL.1985Infanticideinprairiedogs:lactatingfemaleskilloffspringofclosekin.Science230,1037–1040.(doi:10.1126/science.230.4729.1037)

20.ShermanP.1981ReproductivecompetitionandinfanticideinBelding’sgroundsquirrelsandotheranimals.InNaturalandSocialBehavior:RecentResearchandNewTheory,AlexanderRD,TinkleDW.

21.BoeufBJL,WhitingRJ,GanttRF.1973PerinatalBehaviorofNorthernElephantSealFemalesandTheirYoung.Behaviour43,121–156.(doi:10.1163/156853973X00508)

22.GilchristJS.2006Femaleeviction,abortion,andinfanticideinbandedmongooses(Mungosmungo):implicationsforsocialcontrolofreproductionandsynchronizedparturition.Behav.Ecol.17,664–669.(doi:10.1093/beheco/ark012)

23.CantMA,NicholsHJ,JohnstoneRA,HodgeSJ.2014Policingofreproductionbyhiddenthreatsinacooperativemammal.Proc.Natl.Acad.Sci.111,326–330.(doi:10.1073/pnas.1312626111)

26

24.DigbyL.1995Infantcare,infanticide,andfemalereproductivestrategiesinpolygynousgroupsofcommonmarmosets(Callithrixjacchus).Behav.Ecol.Sociobiol.37,51–61.(doi:10.1007/BF00173899)

25.HrdySB.1976CareandExploitationofNonhumanPrimateInfantsbyConspecificsOtherThantheMother.InAdvancesintheStudyofBehavior(edsJSRosenblatt,RAHinde,EShaw,CBeer),pp.101–158.AcademicPress.(doi:10.1016/S0065-3454(08)60083-2)

26.WasserSK,&BarashDP.1983Reproductivesuppressionamongfemalemammals:implicationsforbiomedicineandsexualselectiontheory.Quart.Rev.Biol.58,513-538.(doi:10.1086/413545)

27.AgrellJ,WolffJO&YlönenH.1998Counter-strategiestoinfanticideinmammals:costsandconsequences.Oikos83,507-517.(doi:10.2307/3546678)

28.EbenspergerLA.1998Strategiesandcounterstrategiestoinfanticideinmammals.Biol.Rev.73,321-346.

29.EbenspergerLA&BlumsteinDT.2007Nonparentalinfanticide.In:Rodentsocieties:anecologicalandevolutionaryperspective(eds.JOWolff&PSherman),pp.267-279.UniversityofChicagoPress,Chicago.

30.StockleyP&Bro-JørgensenJ.2011Femalecompetitionanditsevolutionaryconsequencesinmammals.Biol.Rev.86,341-366.

31.HrdySB.1976CareandExploitationofNonhumanPrimateInfantsbyConspecificsOtherThantheMother.InAdvancesintheStudyofBehavior(edsJSRosenblatt,RAHinde,EShaw,CBeer),pp.101–158.AcademicPress.(doi:10.1016/S0065-3454(08)60083-2)

32. LukasD,Clutton-BrockT.2017Climateandthedistributionofcooperativebreedinginmammals.R.Soc.OpenSci.4,160897.(doi:10.1098/rsos.160897)

33. LukasD,Clutton-BrockTH.2011Groupstructure,kinship,inbreedingriskandhabitualfemaledispersalinplural-breedingmammals.J.Evol.Biol.24,2624–2630.(doi:10.1111/j.1420-9101.2011.02385.x)

34.WilmanH,BelmakerJ,SimpsonJ,delaRosaC,Rivadeneira,MM,JetzW.2014EltonTraits1.0:Species-levelforagingattributesoftheworld'sbirdsandmammals.EcologicalArchivesE095-178.Ecology95,2027-2027.(doi:10.1890/13-1917.1)

35.BoteroCA,DorR,McCainCM,SafranRJ.2014Environmentalharshnessispositivelycorrelatedwithintraspecificdivergenceinmammalsandbirds.Mol.Ecol.23,259–268.(doi:10.1111/mec.12572)

36.SiblyRM,GradyJM,VendittiC,BrownJH.2014Howbodymassandlifestyleaffectjuvenilebiomassproductioninplacentalmammals.Proc.Roy.Soc.B281,20132818.(doi:10.1098/rspb.2013.2818)

37. JonesKEetal.2009PanTHERIA:aspecies-leveldatabaseoflifehistory,ecology,andgeographyofextantandrecentlyextinctmammals.Ecology90,2648–2648.(doi:10.1890/08-1494.1)

27

38. LangerP.2008Thephasesofmaternalinvestmentineutherianmammals.Zoology111,148–162.(doi:10.1016/j.zool.2007.06.007)

39.BartonRA,CapelliniI.2011Maternalinvestment,lifehistories,andthecostsofbraingrowthinmammals.Proc.Natl.Acad.Sci.U.S.A.108,6169–6174.(doi:10.1073/pnas.1019140108)

40.HindeK,MilliganLA.2011Primatemilk:proximatemechanismsandultimateperspectives.Evol.Anthropol.20,9–23.(doi:10.1002/evan.20289)

41.RollandJ,CondamineFL,JiguetF,MorlonH.2014FasterSpeciationandReducedExtinctionintheTropicsContributetotheMammalianLatitudinalDiversityGradient.PLOSBiol.12,e1001775.(doi:10.1371/journal.pbio.1001775)

42.HadfieldJD,NakagawaS.2010Generalquantitativegeneticmethodsforcomparativebiology:phylogenies,taxonomiesandmulti-traitmodelsforcontinuousandcategoricalcharacters.J.Evol.Biol.23,494–508.(doi:10.1111/j.1420-9101.2009.01915.x)

43.HadfieldJ.2010MCMCglmm:MarkovchainMonteCarlomethodsforgeneralizedlinearmixedmodelsretrievedfromcran.uvigo.es/web/packages/MCMCglmm/vignettes/Tutorial.pdf

43. LukasD&Clutton-BrockT.2018Socialcomplexityandkinshipinanimalsocieties.Ecol.Lett.21,1129-1134.(doi:10.1111/ele.13079)

44.MohnotSM.1980IntergroupinfantkidnappinginHanumanlangur.FoliaPrimatol.(Basel)34,259–277.(doi:10.1159/000155958)

45.KappelerPM,FichtelC.2012FemalereproductivecompetitioninEulemurrufifrons:evictionandreproductiverestraintinaplurallybreedingMalagasyprimate.Mol.Ecol.21,685–698.(doi:10.1111/j.1365-294X.2011.05255.x)

46.BaldiR,CampagnaC,PedrazaS,LeBoeufBJ.1996SocialeffectsofspaceavailabilityonthebreedingbehaviourofelephantsealsinPatagonia.Anim.Behav.51,717–724.(doi:10.1006/anbe.1996.0075)

47.DiFioreA,RendallD.1994Evolutionofsocialorganization:areappraisalforprimatesbyusingphylogeneticmethods.Proc.Natl.Acad.Sci.U.S.A.91,9941–9945.(doi:10.1073/pnas.91.21.9941)

48.SilkJB.2007SocialComponentsofFitnessinPrimateGroups.Science317,1347–1351.(doi:10.1126/science.1140734)

49.Frank,SA.1998FoundationsofSocialEvolution.PrincetonUniv.Press,Princeton.

50.WestSA,PenI,GriffinAS.2002Cooperationandcompetitionbetweenrelatives.Science296,72–75.(doi:10.1126/science.1065507)

51.WestSA,MurrayMG,MachadoCA,GriffinAS,HerreEA.2001TestingHamilton'srulewithcompetitionbetweenrelatives.Nature409,510-513.(doi:10.1038/35054057)