An evolutionary framework for studying mechanisms of social behavior
NESCent Working Group on Integrative Models of Vertebrate Sociality: Evolution, Mechanisms, and Emergent Properties, Hans A. Hofmann
1, Annaliese K. Beery
2, Daniel T. Blumstein
3, Iain D. Couzin
4, Ryan L. Earley
5, Loren D. Hayes
6, Peter L. Hurd
7, Eileen A. Lacey
8, Steven M. Phelps
1, Nancy G. Solomon
9, Michael Taborsky
10,
Larry J. Young
11, and Dustin R. Rubenstein
121TheUniversityofTexasatAustin,DepartmentofIntegrativeBiologyandInstituteforCellularandMolecularBiology, 2415Speedway,Austin,TX78712,USA
2SmithCollege,DepartmentofPsychologyandPrograminNeuroscience,Northampton,MA01063,USA
3UniversityofCalifornia,DepartmentofEcologyandEvolutionaryBiology,621YoungDriveSouth,LosAngeles,CA90095-1606, USA4PrincetonUniversity,DepartmentofEcologyandEvolutionaryBiology,Princeton,NJ08644,USA
5UniversityofAlabama,DepartmentofBiologicalSciences,300HackberryLane,Box870344,Tuscaloosa,AL35487,USA
6UniversityofTennesseeatChattanooga,DepartmentofBiologicalandEnvironmentalSciences,Chattanooga,TN37403,USA
7UniversityofAlberta,DepartmentofPsychologyandCentreforNeuroscience,Edmonton,Alberta,T6G2E9,Canada
8UniversityofCaliforniaatBerkeley,MuseumofVertebrateZoologyandDepartmentofIntegrativeBiology, 3101ValleyLifeSciencesBuilding,Berkeley,CA94720-3160,USA
9MiamiUniversity,DepartmentofBiology,Oxford,OH45056,USA
10UniversityofBern,InstituteofEcologyandEvolution,DivisionofBehaviouralEcology,Wohlenstrasse50a,3032Hinterkappelen, Switzerland
11EmoryUniversity,CenterforTranslationalSocialNeuroscience,YerkesNationalPrimateResearchCenter,954GatewoodRoad, Atlanta,GA30329,USA
12ColumbiaUniversity,DepartmentofEcology,EvolutionandEnvironmentalBiology,1200AmsterdamAvenue,NewYork, NY10027,USA
Socialinteractionsarecentraltomostanimalsandhavea fundamentalimpactuponthephenotypeofanindividual.
Socialbehavior(socialinteractionsamongconspecifics) representsacentralchallengetotheintegrationofthe functionalandmechanisticbasesofcomplexbehavior.
Traditionally, studies of proximate and ultimate ele- mentsofsocialbehaviorhavebeenconducted bydis- tinctgroupsofresearchers,withlittle communication across perceived disciplinary boundaries. However, recenttechnologicaladvances,coupledwithincreased recognitionofthesubstantialvariationinmechanisms underlying social interactions, shouldcompel investi- gatorsfromdivergentdisciplinestopursuemoreinte- grative analyses of social behavior. We propose an integrative conceptual framework intended to guide researchers towards a comprehensive understanding oftheevolutionandmaintenanceofmechanismsgov- erningvariationinsociality.
Thestudyofsocialbehaviorinthe21stcentury
Allanimalsinteractwithconspecificsatsomepointintheir lives.Membersofthesamespeciestendtobeeachother’s fiercest competitorsandstrongest allies,asevidencedby the intense cooperation and conflict that characterize many intraspecific interactions [1]. These interactions are the products of genetic, epigenetic, endocrine, and neural mechanisms that – in conjunction with environ- mentalconditions–affectDarwinianfitnessandevolvevia naturalselection.BuildinguponAristotle’sfourquestions, Tinbergen [2] posited that understanding behavior requires the integration of studies of mechanism and function.Onlybyaskingquestionsbothfromaproximate perspective(i.e.,focusing oncausationanddevelopment) and an ultimate perspective (i.e., focusing on adaptive value and evolutionary descent) can behavior be fully understood. Social behavior in particular lends itself to such an integrative approach not only because it com- mandsthe attention ofmanydisciplines [3]but alsobe- cause even many behaviors commonly considered non- socialoftenoccurinasocialcontext(e.g.,mating,fighting, parentalcare).Socialbehaviorisalsospecialbecausethe selective agents areother membersof thesame species, andthisresultsinintriguingevolutionarydynamics.Nev- ertheless, in the intervening decades since Tinbergen’s
Corresponding authors: Hofmann, H.A. (hans@utexas.edu); Rubenstein, D.R.
(dr2497@columbia.edu).
Keywords:evolution;socialbehavior;complexsociality;group-living;neuralcircuits;
hormones;genomics.
581
Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-0-290994 Erschienen in: Trends in Ecology & Evolution ; 29 (2014), 10. - S. 581-589
https://dx.doi.org/10.1016/j.tree.2014.07.008
seminal work[2]studiesofbehavioralmechanismshave proceeded largely independentlyof analysesof ultimate- levelexplanationsforsocialbehavior[4].Amongthefac- tors contributing tothisdisconnect arethe challengesof applyinglaboratorymethodstofieldresearchwheremost complex social behaviors are studied, as well as long- standing differences in terminology,conceptual foci,and studytaxa[3,5–7].Progresstowardsanintegratedunder- standingoftheevolutionofsocialbehaviorhasbeenlimit- ed.
Only now, 50 years after Tinbergen’s seminal 1963 publication[8],effortstointegrateneural,genetic,epige- netic,physiological,ecological,andevolutionarystudiesof behavior are gainingincreased prominence [7,9–11,101], facilitated bymultiplefactors,includinginnovativetech- nologies (e.g., high-throughput sequencing [12]), and analytical procedures (e.g.,improved statistical methods formodelingandcomparativeanalyses[13])aswellasthe increasing ease of applicationof these advances to field studies (e.g., biotelemetry [14,15]). As a result, it is in- creasinglypossibletoaddressallfourofTinbergen’sques- tions concurrently for the same species [3,7,10,11,16], which is mosteffective when usingmodern comparative methods [13]. Such integration is crucial if studies of behavioraretocontributetosolutionstopressingbiologi- cal problems. For example, only by understanding the evolutionaryoriginsofdiversemechanismscanwebegin topredicthowspecieswillrespondtoglobalchange[17].
Similarly,athoroughunderstandingoftheadaptive con- sequencesofdiversemechanismscanhelptoidentifynovel modelsystemsforstudiesofspecificneuropsychiatricdis- orders [18]. Integrating Tinbergian levels of analysis is especially appropriate for the study of social behavior which,givenitscomplexity,mustbeapproachedfrom an integrativeperspective.
Historicalperspectives
Although most current textbooks on animal behavior prominently feature Tinbergen’s four questions [19–21], researchers have been slower to adoptthe type of truly integrative approach that Tinbergen originally proposed [2]. Indeed, studies of behavior remain to some extent dividedintoeffortstounderstandultimate-versusproxi- mate-levelreasonsforvariation insocialinteractions[3].
Eachtraditionoffersimportantimpulsesfortheintegra- tiveconceptualframeworkweoutlinebelow.
Ecologicalandevolutionarytraditions
Ethologists and behavioral ecologists have emphasized field studiesof ultimate-level aspects of social behavior.
Crucial concepts addressed by such studies include the roles of kinship and inclusive fitness in shaping social interactions, as well as the effects of specific ecological parameters onsocialstructure [8,22].Such studieshave theadvantageofdocumentingpatternsofbehaviorandthe associated adaptive consequences in the environments, andundertheselectiveregimesexperiencedbythestudy organisms.However,suchanalyseshavetendedtoignore thephysiological,neural,andgeneticmechanismsunder- lying these behavioral patterns as part of a ‘phenotypic gambit’, a heuristic construct positing that detailed
knowledge of the mechanistic bases for behavior is not requiredforanunderstandingofitsfunctionandevolution [23,24]. As a result, such studies have been typically unable to determine howunderlying mechanisms shape observed behavioralresponsestoexternal environments, including generating significant individual variation in responsetosimilarexternalenvironments.
Neuroendocrineandgeneticfoundations
Psychologists and neuroscientistsinterested insocial be- haviorhavefollowedanoftenparallelbutdistinctresearch traditionthatemphasizesitsphysiological,neuroendocrine, and genetic bases. Prominent themes have included the rolesoflearningandontogeneticexperienceonsocialinter- actions, as well as the effects ofhormone levels in both generating and mediating specific patterns of behavior.
Such studies are typically conducted under laboratory conditionsand involve alimited number of‘model’ study organisms, thereby offering important opportunities for controlledexperimentation, oftenemployingtools specific totheorganismsunderstudy.However,theseanalyseshave tendedtoemployhighlyinbredstudyorganismsthatlivein simplistic laboratory environments [25], thereby largely precludingconsiderationofthefunctionalcontextsinwhich behavior–particularlycomplexsocialbehavior–occursand has evolved [26]. As a result, studies of proximate-level mechanisms of social behavior generally cannot address thepotentialimpactsofvariableenvironmentalconditions.
Thepowerofintegration
Althoughnumerousopportunitiesexistformultidisciplin- aryresearch,atpresentwelackanappropriateconceptual framework–includingacommonlanguagefordescribing socialbehavior–todevelopanintegrativeunderstanding of the evolution of social behavior. To capitalize upon emergingopportunitiesweneedpredictivemodelsofsocial interactionsthatintegratefunction andmechanism,and thatcanbeappliedtodiversetaxaoverarangeofsocial andecologicalcontexts.Weofferheresuchanintegrative frameworkofsociality(Figure1A),onethat incorporates individualvariationsinecology,fitness,andexperienceas wellastheneural,physiological,genetic,anddevelopmen- tal mechanisms underlying social behavior. We outline ways in which researchers can use this framework to dissect mechanisms of socialbehavior in free-living ani- malsexposedtothereal-worldecologicalandevolutionary factorsthat shape suchbehavior. We do soina manner thatwillopenupinnovativeavenuesforcomparisonacross disparatetaxonomicgroups.Importantly,thisframework canbeextendedtoothertypesofcomplexbehaviors(e.g., finding food or a suitable habitat, migratory behavior, learningandmemory formation)andtherefore acts as a blueprintfortheintegrativestudyofbehavior.
Anintegrativeframework
Clearly, combining proximate and ultimate approaches to the same phenomenon generates opportunities for understanding social behavior that are not possible througheithertraditionalone.Forexample,becausethe genetic, molecular, and neural mechanisms underlying behavioraresubjecttoselectionandhaveaphylogenetic
582
(A)
(B)
Internal attributes External attributes
Neural and molecular attributes Sensory, memory, valuation, and motor centers
rJ
Neuro anatomy
Neural gene Receptor expression ligand binding
Ufe-h istory traits
Neural activity
lndividuall Individual 2 Individual n
Behavior
I>D 0 0 u UJ
"'
·o
0 Vl
Habitat Resource Predation/
structure distribution parasitism
Distribution Kinship Demography
Functional contexts of social grouping
Predator Resource avoidance acquisition
Mate Offspring Homeo- acquisition care stasis
Internal attributes Functional External attributes Neuromolecular Life-history contexts Ecological Social
SpeciesA • • • • • • • • •
SpeciesB • • • • • • • • • • • • • • • • •
SpeciesC • • • • • • • • • • • • • • • • • •
SpeciesD • • • • • • • • • • • • • • • • • •
SpeciesE • • • • • • • • • • • • • • •
TRENDS i1 Ecology & E10/tA/on
Figure 1. An integrative framework for the study of social behavior. !AI The framework incorporates external (ecology and social environment) and internal attributes (neural and molecular measures together with intrinsic life-history traits) in which individuals and populations can vary. Note that even subtle differences over time and among individuals or species in neuraVmolecular characteristics can result in functional variation, giving rise to behavioral diversity. Triangles with gradients represent! he continuous or semi-<>Ontinuous nature of the variables, indicating a range from high to low. (B) Evolutionary processes influence internal attributes (such as neural and molecular mechanisms and life-history traits) in relation to external attributes (ecological characteristics and social group traits) and determine social behavior within different functional contexts. Multivariate approaches can be used to identify co-variance patterns within and across populations or species at ecological, individual, socia I, and/or mechanistic levels. Variation in color intensity represents quantitative variation in the attributes similar to the gradients in the triangles in (A).
history, they need to be understood in a variety of social and ecological contexts within an explicitly integrative framework [1]. Conversely, understanding the nature of these mechanisms can help to reveal why responses to
variable environmental conditions take the forms that they do [27]. The study of social behavior, in particular, is uniquely positioned to benefit from such integration for several reasons. First, as noted above, social behavior is
583
nearly ubiquitous, with clear functional ties to crucial issues such as conservation and human health. Second, previousresearchhasbeenremarkablyproductiveiniden- tifyingtheecological conditions thatshape thesocialbe- havior of a wide range of taxa (e.g., [9,28,29]). Third, detailedinvestigationsofthemechanisticbasesforsocial behaviorhavebeencompletedforseveralmodelorganisms (e.g.,[30,31]),providinganimportantbaselineforstudies ofothertaxa.Althoughstudiesofsocialbehaviorarenot unique in offering such opportunities, few aspects ofor- ganismalbiologyareasclearlyandfirmlypoisedtoforge innovative and integrative perspectives on phenotypic variation [32].
Developing atruly integrativeview ofsocialbehavior requiresanappropriateconceptualframeworkthatwill(i) facilitateidentificationofgeneral,potentiallycausal,rela- tionshipsbetweenbehaviorandotheraspectsofthebiolo- gy of an organism, (ii) improve our ability to generate testable predictions regarding these relationships, and (iii)enhanceourabilitytoidentifythemostsuitablestudy systemsforagivenbehavioralattribute.Weproposehere such aframeworkthat isaimed at(i)facilitating under- standingofthediversity ofregulatoryprocesses ofsocial behavior in an ecological and evolutionary context, (ii) providing a roadmapfor generating testable predictions from existing data, and (iii) identifying suitable model systemsforsimultaneousstudyinthelaboratoryandfield.
We believethat thisframeworkservestobridge thehis- torical conceptual gap between relevant biological disci- plines, thereby pavingthe way fora comprehensiveand trulyintegratedunderstandingofsocialbehavior.
Functionalexplanationsforproximatemechanisms Causesforsocialgrouping
Social behavior occurs in many forms and contexts, but group-livingorganismsexhibit someofthemostcomplex forms of social behavior. Understanding how and why animalsformgroupsrepresentsanidealsituationinwhich to develop an integrative framework of complex social behavior becauseit involves manyforms ofpositiveand negative social interactions. Empirical studies of verte- bratesandinvertebrateshavedemonstratedthatanimals typically form groups for one or more of five functional reasons: (i) predator avoidance; (ii) resource acquisition;
(iii)mateacquisition;(iv)offspringcare;and(v)homeosta- sis [33] (Figure1A).These functionalcontexts, however, cannotalwaysbeclearlydistinguishedwhenconspecifics interact under natural conditions. Understanding the mechanisms underlying social grouping can provide insights intowhygroupsform and,perhaps moreimpor- tantly,whygroup-livinghasevolved.Forexample,weare onlybeginningtounderstandthattheneuralandmolecu- larmechanismsunderlyingsocialbehaviors–asisthecase forallphenotypes–aretheresultofinteractionsbetween genetic, environmental, developmental, and epigenetic processes [7].Comparative studieshave illuminated the behavioral,neural,andmolecularunderpinningsofsocial behavior,suggestingthatmechanismsregulatingbehavior in similar contexts might, in part, be highly conserved across diversevertebratetaxa,as hasbeensuggestedfor paternal care in mammals and teleost fishes [34]. By
contrast, similar behaviorsin different contextsor time- periods can also result from different mechanisms. For example, territorial aggression, in the context of mate acquisition, is often modulated by sex steroids, such as androgens,whereasaggressionoutsideofreproductionis oftenmodulatedbyotherhormonalmechanisms[35].Im- portantly,temporaldifferencesinneurochemicalandmo- lecular regulatory mechanisms or variation across individuals or species result in functional variation in sensory, memory, valuation, and motor centers. Thus, theexpressionofseeminglyidenticalbehavioralpatterns indifferentreproductivecontextsorseasons–orindiffer- entindividualsorspecies–mightinvolvediverseregula- tory processes. Understanding these processes in the contextofsocialbehaviorcanhelp toinform ushowand when groups form, and whether similar associations in different speciesare drivenby the sameor different un- derlyingmechanisms.
Neuralmechanismsinsocialspecies
Modernbiologyhaslongmovedbeyondthefruitlessdebate abouttherelativecontributionsofnatureversusnurture, and instead has come to the insightthat behavior – in commonwithanyotherphenotype–istheresultofinter- actionsamonggenetic,environmental,developmental,and epigeneticprocesses.Nevertheless,howtheseneuraland molecular mechanisms evolve is much less well under- stood. Four different hypotheses have been proposed [7,36]:(i)theneuralandmolecularsubstratesofbehavior might be conserved even though the resulting behavior patternshaveevolvedinparallel(deephomology[37]);(ii) independentlyevolvedmechanismsmightresultinsimilar behavioralfunctions(e.g.,[38]);(iii)molecularandneural pathwaysmightdivergethroughtimewithnoconcomitant changeinthephenotype(developmentalsystemshift[39]);
or(iv)conservedmolecularmechanismscanbecomeasso- ciatedwithdivergentfunctionsandphenotypesoverevo- lutionarytime(phenologs[40]).Theseapparentlyopposing scenarios are infact not mutually exclusive,and allcan shapedifferentbehavioralphenotypesacrosspopulations orspeciessuchthatagivenfunctionallyequivalentbehav- ioralphenotypemightarisefromseveraldifferentmecha- nisms.
Supportcanbefoundforallfourhypothesesinadiver- sityofsocialorganisms.Forexample,monogamousmating systems have evolved independently numerous times in manytaxa,buttheformationofpairbondsmightinvolve different (e.g., prairie vole vs California mouse [38]) or conserved (e.g., prairie voles and convict cichlid fish [30,41])neuroendocrinepathways.Similarly,thereissig- nificantneuroendocrinevariationintheregulationofter- ritorial aggression, but the central role of the biogenic amineserotonin appearstobeconserved across animals [42–44].A well-knownexampleofdevelopmental system drift(i.e.,developmentalpathwaysdivergeinresponseto selection, although the resulting phenotypes do not change) concerns sex-determining mechanisms, where very different underlyingmechanismsinvolving chromo- some dosage, sex-determining genes, or environmental factorssuchas temperatureorsocialstatus[45–47] give rise tomales andfemales with sex-specificbehaviors.In
584
thecontextofsocialbehavior,developmentalsystemdrift canmeanthatbehavioralresponsesorbrainregionsthat regulate behavior can be homologous even though their morphologicalsubstratesordevelopmentaloriginsarenot [7].Phenologs,bycontrast,compriseconservedgene net- workswhichbecomeassociatedwithverydifferentpheno- types over the course of evolution [40]. For example, nonapeptidesregulatepair-bondingbehaviorsacrossver- tebrates[30,41],andorthologsoftheoxytocinorvasopres- sin ancestral gene also regulate mating behavior in nematodes[48]andleeches[49].The studyofthesecon- vergent and divergent pathways in conjunction with a detailedunderstandingof the survivalvalue and fitness consequences of specific behavior patterns promises to yield insights into general principles underlying social evolutionatbothproximateandultimatelevels.
Conceptualrelationshipsbetweenmechanismsand function
A comprehensive understanding of variation in sociality requires not only the study of social behavior (i.e., the interactions among conspecifics) but alsoof reproductive behavior(i.e.,theregulationofwhomateswithwhom)and socialorganization(i.e.,thepatternsofassociationwithin and between groups) (see [22] for detailed discussion).
Moreover,atrulyintegrativeunderstandingofsocialevo- lutionrequiresthereconstructionoftheevolutionaryhisto- riesofsocialtraitsandthecharacterizationofrelationships amongthedifferentregulatorymechanismsresponsiblefor patternsofsocialbehavior.Distinctbehavioraltraitsdonot operateindependentlyandarenotacteduponbyselectionin isolationfrom oneanother,eventhough they areusually studiedinthismanner [26].Instead,suitesofbehavioral patternscommonlyco-vary,formingoverallsocialsystems andlife-historystrategiesthatcandifferwithinandamong individuals[50],aswellasacrosspopulationsandspecies [51].Similarly,behavioralpatternsgenerallyco-varywith endocrineandneuralmeasures.Forexample,acrossverte- brates,competingphenotypesoftendifferintrade-offsbe- tweentraitsthat affectfitness,including bodycoloration, aggression, and immunefunction[52,53].Strong correla- tional selection isgenerally thoughttoresult insuch co- adapted trait complexes [54], with pleiotropic hormonal systemsplayingacentralrole[55,56].Neuroendocrinesys- temsmightthuspromoteorconstraindivergenceandspe- ciationbecausetheeffectsofdisruptiveselectionononetrait aretransferredtotheothertraitineitherasynergisticor antagonisticmanner[53,56].
We propose a framework for the integrative study of complex socialbehavior that formalizes conceptualrela- tionshipsbetweenmechanismsandfunction(Box1).Spe- cifically, we propose a list of attributes, either external (e.g.,ecologicalcharacteristicsorsocialand/ordemograph- ictraitsofthegroup)orinternal(i.e.,neuralandmolecular characteristics, life-historytraits) that canbe quantified (repeatedlyandsimultaneously,ifnecessary)inmultiply- interacting individuals (Figure 1A). Importantly, these attributesaremuchbroaderthan thekindsofelemental behavior patterns (e.g., aggression towards an intruder;
dichotomousfemalemate-choice)thataretypicallyexam- inedinmostmechanisticstudiesconductedinlaboratory
settings. We also propose a multivariate approach for identifying patterns ofcovariance andfor reducingcom- plexity in such datasets (e.g., principal components at ecological,individual,social,andmechanisticlevels)with thegoalofunravelingtheprocessesthatgovernthe evo- lutionoftheneuralandmolecularmechanismsunderlying socialbehavior(Figure1B).Theseinsightsprovidequan- tifiable variables that can facilitate a thorough under- standing of, and generate testable predictions on, the causes,origins,andfunctionalconsequencesofbehavioral variationwithin andacrosspopulationsandspecies.
Externalattributes:ecologicalcharacteristicsandsocial grouptraits
The mechanismsregulatingsocialbehaviors areaffected by external conditions including the ecology and social environment of an individual (Figure 1A). Importantly, theseattributes candifferentiallyaffectgroupmembers.
For example,habitat structure, resource distribution, or risk ofpredationandparasitism,candifferentiallyinflu- ence thebehavior ofdominantandsubordinate,or male andfemale, groupmates [57].Such parameterscanalso influencethedistributionandbehaviorofonesex,whichin turn can affect the behavior of the opposite sex [58].
Likewise,thedemographicandkincompositionofapopu- lationcanaffectdecision-makinginjuveniles(Box2)[59–
61].Thecostsandbenefitsoflivingingroupscanaffectthe evolution of neural pathwaysunderlying aggressive and cooperative behaviors, which in turn mightaffect group composition and persistence, and ultimately population structure(e.g.,estrildidfinches[62]).
Internalattributes:life-historytraits
Theneuralprocessesunderlyingsocialbehaviorsarealso influenced by a variety of attributes of the individual, includingsex,reproductivestate,age,condition,andexpe- rience(Figure1A),allofwhichcanaffecttheopportunities
Box1.Anintegrativeframeworkofsociality
Ourframeworkexplainspatternsofsocialbehaviorthataremost frequentlystudied(e.g.,matingbehavior,offspringcare).Inreality, these apparently disparate behavioral patterns are linked by ecologicalfactorsatonecausallevelandacommon neuromole cularsubstrateatanother.Thus,bothultimateandproximateforces will shape and constrain behavioral strategies to vary along principal component dimensions. Similarly, there are functional relationshipsbetween individualneural and molecular attributes (e.g.,hormonelevelsarefunctionallylinkedtoreceptordensities).
Components of variation in these dimensions will reflect the organizationofpartsofthemechanismintoafunctioningwhole.
Forexample,behavioralpatternsclusterintofunctionalsets(e.g., monogamouspair bonding,parentalcare,territorialdefense,etc.).
Eachprincipalcomponentofvariationintraitssuchasneurotrans mitterandneuromodulatorexpressionandreceptioninthenucleiof the social decision making network in these organisms should relatetobiologicallymeaningfulvariationinbehavior.Areasonable startingpointistomodelaone to onecorrespondencebetweenthe principal components ofbehavior and those of the mechanistic underpinnings.Asidefromthislargeraimofidentifyingcorrelations between axes of mechanisms and axes of behavior, there is a practicalbenefittoanalyzingprincipal componentsofbehavioral variation,orvariationinmechanism:toidentifythesetofthemost robust, efficient,proxy measures for causal mechanisms and/or behavioralvariation.
585
of an individual during competition to access resources such as mates and breeding sites. Moreover, within a group, the position of an animal in a hierarchy, or its abilitytodefendresources,isoftendependentuponbody sizeandphysicalstrength[63].Age,size,condition,rank, relatedness,andexperiencecanalsoinfluencethetenden- cy tocare for offspringand toparticipate in cooperative activities[64,65].Theseandothercooperativeandantag- onisticbehaviorscanbepartlyregulated bymechanisms such as androgens [66], allostatic load [67], or receptor densities in specific brain regions [68]. In all cases, the underlyingneuralmechanismsregulatingthesebehaviors remainlargelyunknown.
Internalattributes:neural,neuroendocrine,andgenetic mechanisms
Behavioral neuroscientists haveidentified numerous en- docrine and neural mechanisms that control behavioral decision-making and hence influence life-history traits, particularly in social species (Figure 1A). For example, specific neuralcircuits suchas the dopaminergic reward system[69]andthebrainsocialbehaviornetwork[70,71]
regulate social behavior. Homologs of the nodes of this socialdecision-makingnetworkhavebeeninferredacross vertebrates, suggesting that this system is highly con- served[72,73].In general,evensubtletemporal,individ- ual, or species-level differences in neural and molecular characteristicscanresultinfunctionalvariationinsenso- ry,memory,valuationandmotorcenters,therebycontrib- utingtothebehavioraldiversityweobserveinnature.For example,neuroanatomicaldifferences inthe volumeofa particular brain region can be related to the relative behavioraldemandsonthatregion[74–76].Atthecellular level, the release of neurochemicals such as biogenic amines [43] and neuropeptides [62] into specific brain regions candirectly result inspecific behavioralprofiles.
Althoughtheexpressionpatternsofneurochemicalgenes, particularly those encoding receptors, are remarkably conserved [77],quantitativevariationinreceptordensity
and/orlevelsofneurochemicalsinspecificnetworknodesis strongly associated with diversity in social behavioral attributes between individuals and across species [68,78]. Given the development of high-throughput se- quencingtechnologies,itisnowpossibletoquantifymany oftheseneuralandmolecularcharacteristicsinemerging modelsystems(e.g.,[79–83]).
Integratingacrosstemporalandtaxonomicscales Evolutionaryprocessesinfluenceallinternalattributesin relation to a variety of external parameters. Toexplain these evolutionary processes it is important to obtain comparative quantitative data from a range of species andundermultiplephysiologicalandenvironmentalcon- ditions[84].Oncecrucialexternalandinternalattributes havebeenmeasuredfornumerousindividualsandspecies, theirinter-relationshipscanbeidentifiedviapairwiseor multivariatestatisticalanalyses [85].Byemploying such anapproach,weexpecttodiscoverfunctionalrelationships amongindividualattributes.Forexample,circulatinghor- monelevelsandreceptordensitiesarefunctionallylinked, and variation in these dimensions can be thought of as clusteringintoprincipalcomponentsreflectingfunctional units. At the same time, these mechanistic components likelyalsocorrespondtoexternalattributes.Forexample, individualattributessuchassex,reproductivestate,age, condition,andexperience arelikely toimpingeuponthe decision-makingcircuitsvianeuroendocrineandneuromo- dulator pathways [1]. The identification of co-variance patterns within and between axes representing mecha- nisms and those representing functional significance will likely reveal robust andrepresentative measures of causalmechanismsassociated withbehavioralvariation.
Unfortunately, few if any such empirical studies have beenconducted, despite that fact that we nowhave the analyticalmeanstodoso[86].
Behavioral, ecological, and neurobiological data from the same species are required to conduct this type of integrative analysis. At present, however, the most Box2.Casestudies:dispersalandaffiliativebehavior
Socialbehaviorentailsbothnegative(e.g.,aversive,aggressive)and positive(e.g.,affiliative,cooperative)interactions. Negativeinterac tionsoftenleadtodispersal(i.e.,emigrationfromnatalgroup)which hasbeenwellstudiedinnumerousspeciesandecologicalcontexts.
Although some hormonal and physiological factors have been identifiedthatrelatetodispersal(e.g.,glucocorticoidlevels,organiza tionalhormonalfactors,bodycondition[94 96]),littleisknownabout the underlying neural circuits. Integrative approachesto dispersal might include neuroendocrine profiling before and after dispersal from the natal group, comparisons between dispersers and non disperserswithinapopulation,andothertemporalchangesingroup structuresuchasimmigrationthatresultinseasonalgroupforma tion.
Studiesofaffiliativebehavior,bycontrast,havealreadybegunto integrate neurobiological and ecological data. In prairie voles, dopamine, oxytocin, and vasopressin act within the mesolimbic rewardpathwaytoestablishpair bondsbetweenmates[30].Across Microtusvolespecies,differencesinthedistributionofoxytocinand vasopressinreceptorsaswellasestrogenreceptoraareassociated withspeciesdifferencesinmatingstrategy[97,98],andstudiesthat investigate theconsequences ofthese variationsin fieldsettings
havebeenconducted[99,100].Work inseasonallysocialmeadow volesaswellasincolonialSouthAmericanrodentshassuggested parallel and potentially convergent pathways by whichoxytocin receptordensityisinvolvedinnaturalvariationinaffiliativebehavior andgroup livingoutsidethecontextofmonogamy[78].Inestrildid finches, homologouspeptide receptorsmodulategroupsize pre ferencesandaredifferentiallydistributedinspeciesthatexhibityear roundterritorialityorflocking,andthepeptideneuronsthatsupply thesereceptorsrespondselectivelytopositivesocialstimuli[62].
Thesespeciesshareecologicalandotheraspectsofsocialorganiza tion (e.g., all are monogamous and biparental), thus a major questionishowthemechanismsofpair bonding,gregariousness, andterritorialityevolveinothertaxawheretheecological,social, and functional contexts can be different. In prairie voles, for example,vasopressinreceptor expression inthe cingulatecortex predictsreproductivesuccessspecificallyinwanderers,butnotin pair bondedmales[99],indicatingthatspace usepatternsneedto beconsideredtounderstandindividualvariationinreceptorexpres sion.Tounderstandbehavioraldiversityfullywemustplacethese, and additional studies, into a broad ecological and evolutionary context.
586
detailedecologicaldatasetsoftenlackcomplementaryneu- ral measures and, conversely, ecological information is often lacking for (laboratory) species whose neural sub- strateshavebeenstudiedindetail.Infact,itappearsthat, forestablishedlaboratoryspeciesaswellasforemerging model systems, the amount of information available on ecology and/or reproductive biology might be inversely correlatedwiththeextentofneurobiologicalinformation.
Currently, mostanalyses ofneural features ofemerging modelorganismsinvolvepairwisecomparisonsofspecies orpopulations,withonlyafewinstancesofdatacollection ina broader phylogeneticcontext [75,77,87]. Thosecom- parisons that exist are typically limited to only a few measures such as gross neuroanatomy [88], circulating hormone levels [89], and gene expression for small sets ofloci[77].Althoughsuchrelativelylimitedcomparisons can provide insights into evolutionary processes, more extensive species sampling and additional fine-grained neural andmolecular measures are necessaryto gain a fullunderstandingof theevolution ofthese mechanisms andthebehaviorpatternstheyregulate.
RevisitingTinbergen’svision
Fifty years after Tinbergen defined his proverbial ‘four questions’[2],thereisatremendousopportunityforinte- grativestudiesontheultimateandproximatemechanisms ofcomplexbehaviorssuchas sociality[7,9–11,16,90].Al- though the number of animal species considered to be acceptedasbiomedicalmodel systemsisdecreasing[91], thistrendisbothparadoxicalandmisguidedgiventhatthe verynotionofamodelsystemisundergoingrapidchange andmight soon be obsolete[92], at last liberating us to (again)usethespeciesthatarebestsuitedfortheproblem in question (i.e., Krogh’s Principle [93]). Innovative re- searchprogramsindiversespeciesarenowpossiblethanks toadvancesinbehavioralecology,genomics,andneurosci- encetogetherwithnumeroustechnologicalbreakthroughs that facilitate the collection of ever-larger and more de- taileddatasetsthanwereimaginableevenafewyearsago.
Systematiceffortsarenowneeded tofillthe gapsinour understandingofsocialbehaviorforspeciesthat arenot theestablishedbiomedicalmodel systemsdiscussedhere and elsewhere. The development of new model systems thatcreatecomprehensivebehavioral,ecological,andneu- raldatasetswithintheframeworkwehaveprovidedhere willhelpustofulfillTinbergen’svisiontounderstandtruly the evolution of neuroethological mechanisms across all levelsofbiologicalorganizationandatalllevelsofanalysis.
Acknowledgments
WethanktheNationalEvolutionarySynthesisCenter (NESCent)[US National Science Foundation (NSF) EF-0905606] for supporting our working group; Jim Goodson, and Emilia Martins for stimulating discussions;andRaynaHarrisforassistancewithmanuscriptprepara- tion.H.A.H.issupportedbyNSFgrantIOS-0843712;A.K.B.issupported byNSFIOS-1257162;D.T.B.issupportedbyNSFDEB-1119660;I.D.C.is supported by NSF PHY-0848755, NSF EAGER-1251424, NSF CNH- 1211972, Officeof Naval Research N00014-09-1-1074, ArmyResearch OfficeW911NG-11-1-0385andHumanFrontierScienceProgramaward RGP0065/2012;R.L.E.issupportedbyNSFIOS-1051682andNSFIOS- 1311347; L.D.H. is supported by NSF IIA-0853719 and NSF IIA- 0901056;P.L.H.issupportedbytheNaturalSciencesand Engineering ResearchCouncilofCanada(NSERC)RGPIN249685;M.T.issupported
bySwissNationalScienceFoundationgrant310030B 138660;L.J.Y.is supportedbyUSNationalInstitutesofHealth(NIH)grantsMH064692 and P51OD11132; D.R.R. is supported by NSF IOS-1121435 and IOS-1257530.
References
1Taborsky, B. and Oliveira, R.F. (2012) Social competence: an evolutionaryapproach.TrendsEcol.Evol.27,679 688
2Tinbergen,N.(1963)Onaimsandmethodsofethology.Z.Tierpsychol.
20,410 433
3Sherman,P.W.(1988)Thelevelsofanalysis.Anim.Behav.36,616 619
4Crews,D.andMoore,M.(1986)Evolutionofmechanismscontrolling matingbehavior.Science231,121 125
5McNamara,J.M.andHouston,A.I.(2009)Integratingfunctionand mechanism.TrendsEcol.Evol.24,670 675
6Blumstein,D.T.etal.(2010)Towardanintegrativeunderstandingof socialbehavior:newmodels andnewopportunities.Front.Behav.
Neurosci.4,34
7O’Connell, L.A.and Hofmann, H.A.(2011) Genes,hormones, and circuits: an integrative approach to studythe evolution of social behavior.Front.Neuroendocrinol.32,320 335
8Barrett,L.etal.(2013)TakingnoteofTinbergen,or:thepromiseofa biologyofbehaviour.Philos.Trans.R.Soc.Lond.B:Biol.Sci.368, 20120352
9Oliveira,R.F.etal.,eds(2008)AlternativeReproductiveTactics:An IntegrativeApproach,Oxford
10 MacDougall-Shackleton,S.A.(2011)Thelevelsofanalysisrevisited.
Philos.Trans.R.Soc.Lond.B:Biol.Sci.366,2076 2085
11 Ophir,A.G.(2011)Towards meetingTinbergen’schallenge.Horm.
Behav.60,22 27
12 Hitzemann,R.etal.(2013)Genes,behaviorandnext-generationRNA sequencing.GenesBrainBehav.12,1 12
13 Pennell, M.W. and Harmon, L.J. (2013) An integrative view of phylogenetic comparative methods: connections to population genetics, communityecology, and paleobiology. Ann. N. Y. Acad.
Sci.1289,90 105
14 Houle,D.(2010)Colloquium papers:Numberingthehairs on our heads:thesharedchallengeandpromiseofphenomics.Proc.Natl.
Acad.Sci.U.S.A.107(Suppl.),1793 1799
15 Wilson,A.J.etal.(2010)Anecologist’sguidetotheanimalmodel.
J.Anim.Ecol.79,13 26
16 Healy,S.D.etal.(2009)Explanationsforvariationincognitiveability:
behaviouralecologymeetscomparativecognition.Behav.Processes 80,288 294
17 Candolin,U.andWong,B.B.M.(2012)BehaviouralResponsestoa ChangingWorld:MechanismsandConsequences,OxfordUniversity Press
18 Robinson, G.E. et al. (2010) Empowering 21st century biology.
BioScience60,923 930
19 Alcock, J. (2013) Animal Behavior: An Evolutionary Approach, SinauerAssociates
20 Dugatkin,L.A.(2013)PrinciplesofAnimalBehavior.(3rdedn),W.W.
Norton
21 Goodenough,J.etal.(2009)PerspectivesonAnimalBehavior,Wiley 22 Kappeler,P.M.etal.(2013)Constraintsandflexibilityinmammalian socialbehaviour:introductionandsynthesis.Philos.Trans.R.Soc.
Lond.B:Biol.Sci.368,20120337
23 Grafen,A.(1984)Naturalselection,kinselectionandgroupselection.
InBehavioralEcology(2ndedn)(Krebs,J.R.andDavies,N.B.,eds), pp.62 84,BlackwellScientific
24 Springer,S.A.etal.(2011)Beyondthephenotypicgambit:molecular behaviouralecologyandtheevolutionofgeneticarchitecture.Mol.
Ecol.20,2240 2257
25 Sokolowski,M.B.(2010)Socialinteractionsin‘simple’modelsystems.
Neuron65,780 794
26 Bolker,J.(2012)Modelorganisms:there’smoretolifethanratsand flies.Nature491,31 33
27 Rubenstein,D.R.(2007)Stresshormonesandsociality:integrating socialandenvironmentalstressors.Proc.Biol.Sci.274,967 975 28 Stacey,P.B.andKoenig,W.D.(1990)CooperativeBreedinginBirds:
LongTermStudiesofEcologyandBehaviour,CambridgeUniversity Press
587
