EvolutionandHumanBehavior29(2008)402–415 ☆ The multiple dimensions of male social status in an Amazonian society ⁎ Christopher von Ruedena, , Michael Gurvena, Hillard Kaplanb aDepartmentofAnthropology,UniversityofCalifornia,SantaBarbara,CA93106,USA bDepartmentofAnthropology,UniversityofNewMexico,Albuquerque,NM87131,USA Initialreceipt7January2008;finalrevisionreceived2May2008 Abstract While social-status hierarchies are common to all human societies, status acquisition is relatively understudied in small-scale societies lacking significant material wealth or intergenerational inheritance. Among the Tsimane of Bolivia, a small-scale Amazonian society, we employaphoto-rankingmethodologytodeterminetheimportantpredictorsoffourmeasuresofmalesocialstatus:successindyadicphysical confrontation,gettingone'swayinagroup,community-wideinfluence,andrespect.Thepredictorsevaluatedincludeage,physicalsize,skill infoodproduction,levelofacculturation,prosocialpersonality,andsocialsupport.Wefindthatphysicalsizebestpredictsrankingsofdyadic fighting ability while social support best predicts getting one's way in a group, community-wide influence, and respect. Level of acculturation,furthermore,isanindependentpredictorofinfluencebutnotrespect,andskillinfoodproductionisanindependentpredictorof respectbutnotinfluence.Thelackofalinearrelationshipbetweenageandthepolyadicsocial-statusmeasuresisevaluatedinlightofthe increasingexposureoftheTsimanetomarketeconomiesandpubliceducationamongrecentagecohorts.Toourknowledge,thisstudyisthe first multivariate analysis of socialstatus that considers different determinants of statussimultaneously. © 2008Elsevier Inc.All rights reserved. Keywords: Socialstatus;Stratification;Hierarchy;Acculturation;Forager-Horticulturalists;Tsimane 1. Introduction consumemoreresources,getthebestpickofmates,andtakea more central role in group decision making (Boehm, 1999; “Behaviour that's admired is the path to power among Trigger, 1985; Wiessner, 1996). Whether implicit or overt, peopleeverywhere.” classification by social status is a human universal. While -Beowulf, a new verse translation by Seamus Heaney womenaswellasmencompeteforstatus(Campbell,2002; (2000,p.5) Hess & Hagen, 2006; Hrdy, 1999; Rucas et al., 2006), this articlefocusesexclusivelyonmale-statushierarchies. In all human societies, individuals differ in social status Socialstatuscanbedefinedasrelativeaccesstoresources dependingupontheirageandpersonalability(Sahlins,1958; withinasocialgroup(Henrich&Gil-White,2001).Apriority Service,1971).Inlaboratory-basedsmallgroupstudies,status of resource access is granted to high-status individuals, we hierarchies emerge spontaneously (Bass, 1954; Campbell et argue,duetoagroup-wideperceptionthattheseindividuals al.,2002;Kalma,1991).Evenamong“egalitarian”foragers, haveagreaterrelativeabilitytoinflictcosts(i.e.,dominance) who are characterized by widespread resource sharing or confer benefits(i.e., prestige) on others. Group members (Kaplan & Gurven, 2005; Winterhalder, 1986) and some acquiesce to higher-status individuals because they believe degreeofstatusleveling(Cashdan,1980),certainindividuals they will avoid harm or gain some benefit from their deference. Status hierarchies, therefore, are not necessarily pure zero-sum arrangements. In part, status hierarchies ☆ FundingforTsimaneresearchwasprovidedbygrantsfromtheNational represent agreements, maintained by deference signals, to Science Foundation (#BCS-0136274 and BCS-0422690) and National facilitate exchange or to avoid costs of repeated contest InstitutesofHealth/NationalInstituteonAging(#1R01AG024119-01). ⁎ competition, as modeled by the war of attrition (Maynard Correspondingauthor. E-mailaddress:[email protected](C.vonRueden). Smith&Price,1973). 1090-5138/$–seefrontmatter©2008ElsevierInc.Allrightsreserved. doi:10.1016/j.evolhumbehav.2008.05.001 C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 403 Statushierarchiesarenotstatic,however.Ifanindividual therewardsofpursuitmaybedelayedandvarywiththeform becomeslessdependentupontheservicesofparticularhigh- ofstatusacquired. statusgroupmembers,thatindividualisperhapsmorelikely This study explores male-status hierarchies among the to challenge power inequities between them (Emerson, Tsimane forager-horticulturalists of Bolivia. Since status is 1962). Acquiescence to those of high status may also vary mostpalpableinthecontextofcontestcompetition,wetreat with subordinates' ability to migrate and their degree of the following measures as manifestations of social status relatedness with high-status individuals (Vehrencamp, among Tsimane adult males: (a) success in dyadic physical 1983). Subordinates can also form coalitions against confrontation, (b) getting one's way in the context of a higher-status individuals. In many traditional human socie- conflictwithinagroup,and(c)influenceinthecontextofa ties, exploitive leaders are often ridiculed, ostracized, or community-wide dispute. We also investigate (d) respect killed (Boehm,1999). because this term is cross-culturally associated with social Amongnonhumanprimates,malesocialstatusisinlarge status. These four measures of social status were chosen part tantamount to the ability to inflict costs; physically because they represent distinct assemblages of social costs dominant individuals usurp or maintain priority access to and benefits within small-scale societies, assay relative foodandmates(Boeschetal.,2006;Cowlishaw&Dunbar, resource access at different social scales, and lend them- 1991; de Waal, 2000). While alliances can be important selvestoempiricalevaluationandcross-culturalcomparison. among nonhuman primates, they are largely to support the None of these measures should be considered equivalent to aboveusesofdominance(deWaal,2000;Duffyetal.,2007; dominance or prestige; in all likelihood, each of the status Nishida & Hosaka, 1996). Among humans, however, the measureswillreflectelementsofboth.Furthermore,ourfour cooperative sharing of food, information, labor, and other measures are not meant to be exhaustive of the range of resources is extensive (Kaplan & Gurven, 2005). Priority social status among the Tsimane. access to resources is only sometimes obtained through While prior ethnographies have demonstrated a link dominance and is often mediated by voluntary transfers of betweenoneparticulartraitandsocialstatus(e.g.,Stearman, inalienable commodities. Henrich and Gil-White (2001) 1989), this study employs a multivariate approach to contrastdominancewithprestige,whichtheydescribeasthe compare several traits as predictors of status in different deferencethataccruestoindividualswhopossessortransmit social domains. If the predictors of one of our four status publicly esteemed skills. Their discussion of prestige, measures differfrom those of the other statusmeasures, we however, is narrower than the potential range of non- could conclude that status hierarchies among Tsimane are agonisticsocialstatus.Forexample,amanmayachievehigh multidimensional.Thepredictorsofmalestatusweevaluate status in the market for mates by offering “good genes” or includeage,physicalsize(e.g.,flexedbicepcircumference), materialgoodsinexchangeforsexualaccess,orhemaygain skill in food production (e.g., hunting ability), level of status in the market for influence by offering physical acculturation (e.g., Spanish fluency), prosocial personality strength or coordinative leadership topotential allies. traits (e.g., generosity in meat sharing), and social support Since humans have lived in hunter-gatherer societies for (e.g.,numberofallies).Toquantifythesevariablesaswellas themajorityoftheirexistence,modernforagercommunities thesocial-statusmeasures,asampleofTsimanemenphoto- canhelpelucidatetheselectiveforcesresponsibleforhuman rankedtheirfellowvillagers.Wehypothesizedthatbodysize males' status-seeking behavior. Forager societies typically wouldbestpredictwinningadyadicphysicalfightwhilethe lack major wealth accumulation or formal political or legal otherpredictorsofstatus,especiallyratingsofsocialsupport, institutions (Kelly, 1995). As a result, physical dominance would better predict the other social-status measures that may play a principal role in acquiring male status across involve n-personinteractions. socialdomains.Amongforagerswhopracticesomedegreeof The organization of this article is as follows: Section 2 horticulture and who engage in intergroup raiding, warrior- discusses the socioecology of the Tsimane to further ship is a primary avenue to community-wide influence motivate our choice of the four status measures; Section 3 (Yanomami:Chagnon,1988;Achuar:Patton,2000;Waorani: presentsourhypothesesconcerningthepredictorsofthefour Robarchek & Robarchek, 1998). On the other hand, status social-status measures; Section 4 describes our methods; acquisition in forager societies has also been linked to Section5presentsourresults;Section6interpretstheresults attributes like hunting ability (Tsimane: Gurven & von anddescribesthestatuscharacteristicsoftworecentTsimane Rueden,2006;Ache:Kaplan&Hill,1985;Hadza:Marlowe, leaders; andSection 7 concludes. 2000; Mbuti: Turnbull, 1965) and generosity (Achuar: Patton,2005;Yuqui:Stearman, 1989).Itisunclearwhether male-status hierarchies in forager societies are truly multi- 2. The socioecology of the Tsimane of Bolivia dimensional:dotraitspredictiveofdyadicdisputeoutcomes, for example, contrast or even trade off with those traits The Tsimane inhabit areas of lowland Bolivia along the predictive of polyadic influence? Understanding the pursuit Maniqui River and in adjacent forests. While families may of status in different social contexts is crucial precisely spend weeks or months on hunting or field cultivation trips becausetheadaptivenessofthispursuitisoftenunclearand away from settled villages, the Tsimane are semisedentary 404 C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 andliveincommunitiesrangingfrom30to500individuals. Tsimane entrepreneurs operate small businesses where they Most food the Tsimane consume derives from horticulture, purchase goods in San Borja and then resell them to other fishing, hunting, and gathering activities. They cultivate villagemembers.ManyTsimanevillagesnowhaveaccessto plantains,rice,corn,andsweetmaniocinsmallswiddensand public schooling for their children. One of the more regularlyfishandhuntformeat.Polygynyisrare,thoughit acculturated Tsimane villages, Ton'tumsi, was the location does occur at low frequencies (∼10%) in more remote ofthisstudy.Ton'tumsiisoneofthelargestTsimanevillages communities. While exclusive priority of access for indivi- and is about a 2-h drive, via logging road and highway, to duals or small groups to certain rights and resources is SanBorja.Dataarenotavailableatpresenttocomparestatus minimal, land close to village centers is de facto privately hierarchies in Ton'tumsi with status hierarchies in less owned.Disputesoverlandaccessforhorticultural purposes acculturated Tsimane villages. Nevertheless, the process of are common, especially between neighboring families. acculturation is mosaic, and individuals in Ton'tumsi vary Successingroupconflicts—oneofourfourstatusvariables greatly in education and income. Among the Tsimane in —isinlargepartameasureofsuccessininterfamilycompe- general, the mean levels of wealth inequality are relatively titionforlandorotherresources. high for a small-scale society (Godoy et al., 2004). As The Tsimane are particularly relevant to the study of traditional societies everywhere change under the forces of statushierarchiesbecausetheylackintergroupwarfare;they globalization, it is imperative to better understand how offeravaluablecomparisontoothersmall-scalesubsistence increases in private wealth within a society shape human societiesinlowlandSouthAmerica,suchastheYanomamo social motivations. (Chagnon, 1988), for whom physical dominance and warriorship beget community-wide influence. Tsimane society is not pacifist, however. Violence between adult 3. Hypotheses males is not uncommon, typically in the form of dyadic fights resulting from sexual jealousy, theft, or stinginess. The type of disputes particular to the dyad, group, and Dispute resolution is typically left to the parties directly community will affect which variables best predict social involved or, on rarer occasions, adjudicated by an informal status in those contexts. Among the Tsimane, dyadic fights gathering of adult men. We include success in dyadic often involve alcohol and are usually motivated by physicalconfrontationas one ofour status measures. accusations of theft, stinginess, or sexual jealousy. Size For much of Tsimane history, older adult males and and strength are primary determinants of successful fight shamans (cocojsi) wielded the most community-wide outcomes (Archer, 1988), and flexed bicep circumference influenceas a result oftheirability tocommune with forest among U.S. college students was recently found to predict spirits and ancestors (Daillant,1994). Shamans have all but lifting strength and self-reported success in conflicts (Sell, disappearedamongtheTsimane,dueinparttotheinfluence 2005). Although resource holding potential, in addition to of Catholic and evangelical missionaries. In the late 20th motivation, determines specific dyadic contest outcomes, century, regional political pressure led to the establishment context-dependent motivation does not influence dyadic of elected village leaders (corregidores). These leaders are relationships relative to fighting ability over the long term principally representatives to outside political bodies, and (Lewis, 2002). With respect to rankings of fighting ability, theygenerallyhaveshorttenureandlittlecoerciveauthority we anticipate that physical-size measures, such as flexed within their villages. Community-wide meetings are com- bicep circumference, chest circumference, height, and mon in Tsimane villages; they often concern disputes over weight, predict winning dyadic fights better than any other thesaleofcommunitylumberorparticipationingovernment variables.Wesuspectthatbicepandchestcircumferencewill or NGO-sponsored development projects. Influence within bestrongerdeterminantsoffightingabilitythanheightsince the context of community-wide disputes, the third of our tall,skinnymenmaybelesslikelytowinafightthanshort, statusmeasures,willoftenaccruetotheindividualswhoare brawny men. most persuasive during community meetings. Elected Group conflicts in Ton'tumsi often involve two or more leaders are not necessarily the most influential individuals familiescompetingoverlandorotherresources,ortheymay intheirvillages,noraretheynecessarilythemostrespected. be the progressive fallout from prior dyadic conflicts. The Respectismoreopaqueinitsmeaningthanourotherstatus typical numbers of individuals involved remain small measuresbutisusedbytheTsimanetodescribeindividuals enough that physical size should still play a role in group worthyof admiration. conflictoutcomes,butsocialsupportasaresultofallieswill TheTsimaneareundergoingrapidacculturation,whichis become important as well. Allies may be family members, potentially placing premiums on market-related skills and friends,orexchangepartners,includingthosewhoexchange private wealth in addition to more traditional correlates of deferenceforaccesstotheknowledgeandwealthofothers. social status. In Tsimane villages, especially those located Among the Xavante of Brazil, men's status stems from the near the town of San Borja (population ∼14,000), incipient in-group social support engendered from one's athleticism, cattle ownership, wage labor with loggers and farmers, and oratory skill, hunting ability, sense of humor, and other produce sales to local markets are on the rise. Several attributes(Maybury-Lewis,1974).FortheKayapoofBrazil, C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 405 socialstatushingesonaccesstoexoticwealthitemsthrough and conspicuous leisure items like watches and radios extra-group social contacts (Werner, 1981). Among the (Godoy et al., 2006). HadzaofTanzania,amancanclaimpoliticalauthority,albeit Prosocial personality traits should predict influence limited, if he has contacts with outsiders and their trade independently of one's level of acculturation. Certain goods(Woodburn, 1979). traits, such as trustworthiness, may confer influence from Kinshipinsmall-scalesocietiesisofparticularimportance their direct benefits to others while other personality traits, ingeneratingsocialsupport.Efemenformaffiliativebonds such as sense of humor, may confer influence more from withconsanguinealmalekintogeneratealliesinthefaceof their signaling of fitness or intelligence (Miller, 1999). competitive social situations (Bailey & Aunger, 1989). Among the Tsimane, work ethic is often described as a Yanomamo men with larger intravillage kinship networks valued trait in a man because it indicates long-term, aremorelikelytobepolygynousandhighstatus(Chagnon, intrinsic motivation to provide for one's family and to 1988). Marriage is a common strategy for constructing engage in collective action for the sake of the community. alliances. Hughes (1988) documents several ethnographic In Ton'tumsi, men occasionally assemble to clear com- examples,includingtheNuerofSudanandtheTodaofSouth munity trails and their soccer field or to engage in co- India,wherehigh-statusmenareindividualsonwhomboth operative fishing. affinal and consanguineal relatedness are concentrated. Generosity is likely to vary as a source of influence Among the Coast Salish of the Pacific Northwest, social depending upon local resource availability, the level of status was associated with intervillage connections, estab- group-wide sharing, and the opportunity to recruit and lishedthroughmarriage(Elmendorf,1971).FortheTsimane, maintain political allies through gift giving (Patton, 2005). interactionwithclosekinconsumesalargepartofmen'stime Thecostlysignalingofcooperativeintentthroughgenerous spentinleisureandresource-productionactivities.Marriage donationsoffood,money,andservicesmaybeanimportant oftenoccursbetweencross-cousins,butmanymenwillmove means of being recognized as a valuable potential social todistantcommunitiestomarrywiveswithwhomtheyhave partnerorally(Frank,1988;Gintisetal.,2001;Gurvenetal., nopriorkinrelation. 2000).ThepotlatchofthePacificNorthwest(Barnett,1938) Physical size is unlikely to be an important predictor of andmokaofhighlandNewGuinea(Strathern,1971)enabled community-wideinfluenceinTon'tumsi,especiallysincethe chiefsandbigmentoflaunttheirmaterialandsocialcapital ability to marshal allies in a conflict, that is, “derived through grandiose displays of generosity. Among the dominance”(Hand,1986),islikelytotrump“intrinsic,”size- Gitksan, individuals would move to new households after based dominance in most polyadic contexts. However, potlatches where they felt their leader was not as generous, heightandmusclemassmayalsoindicatehealth,attractive- hence powerful, as others (Adams, 1973). Among Amazo- ness, athletic performance, and resource production, which nian societies such as the Yuqui (Stearman, 1989) and the willallincreasethevalueofanindividualtopotentialmates Achuar (Patton, 2005), meat sharing is predictive of social or coalition partners. A meta-analysis revealed that 97% of status.AmongtheTsimane,sharingisgenerallyrestrictedto applicable studies found significant, positive relationships closekininextendedfamilyhouseholdclusters;intrafamily between height and socioeconomic status in urban nation- generositymaythereforehavelessrelevanceasagroup-wide states (Ellis, 1994). Case and Paxson (2006) suggested that predictor of influence than less frequent acts of extra- the cross-cultural relationship between height and socio- householdexchange. economic status may be driven by cognitive performance. Weanticipatethatsocialsupportwillmediatemuchofthe Betternutritionleadstobothtallnessandintelligence,andit effects of the size, acculturation, and personality variables isthelatterthatproducessocioeconomicsuccess.Cognitive upon community-wide influence. People seek social proxi- performance tests, however, were not administered to the mity to the strong, skilled, wealthy, or generous because of Tsimane forthis study. the knowledge or material goods they might acquire In Ton'tumsi, community-wide dilemmas have, in the (Henrich & Gil-White, 2001) and because of the indirect past, concerned allocation of government benefits, sale of social value of association with powerful individuals. In community lumber, internal conflicts and those with exchangeforsocialproximitytohigh-statusgroupmembers, colonistsorotherexploiters,andparticipationincommunity individuals offer their deference and support. Since many development and anthropological projects. These commu- conflictsarewithkin(BorgerhoffMulder,2007),thenumber nity debates often require interaction with Bolivian ofone'salliesmaybeabetter indicatorofsocialsupportin nationals, and thus, literacy, Spanish fluency, experience conflicts than thesize ofone's kinnetwork. working with loggers or ranchers, and familiarity with the WealsoinvestigatetheTsimaneunderstandingofrespect market town of San Borja are of particular importance to sincethetermiscommonlyviewedbytheTsimaneandother community-wide influence. Tsimane men demonstrate their culturesassynonymouswithsocialstatus.Respectwillvary market acumen, in part, through the conspicuous consump- cross-culturally in meaning, but here, we posit that being tion of market goods. A recent study among the Tsimane respected connotes other people's acknowledgment of an foundthatthosemenwhoearnmoremoneydevoteagreater individual'ssocialstatus:hisorherrelativeabilitytoinflict percentage of their income to the purchase of prestigious costsorconferbenefitsonothers.However,respectislikely 406 C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 more than mere acknowledgment of social status. Exploita- afterwards asmen physically senesce(Gurven etal.,2006). tion of others is likely to erode how well one is respected. About 40% of Tsimane men survive to age 60 and beyond Iftheexerciseofpowerbythoseofhighstatusisillegitimate (Gurven et al., 2007); these individuals are not as likely to or unjustified in the eyes of the community, then respect is winadyadicfightorgettheirwayinagroupasareyounger, lost.Theperceivedlegitimacyofsocialstatusiscrucialtoits stronger individuals. Older individuals, however, may rank maintenance and expansion, particularly social status highly in community influence since they are likely to be generated via social alliances. Bass (1981) found that soughtafterforadvice,thatis,wisdom,andhavemoresocial legitimateuseofpowerwillincreaseacceptanceofdemands supportintheformofdirectdescendants.Wisdomhasbeen made by high-status individuals. Blau (1964) describes the defined as a high level of contextual and procedural processofpowerlegitimation:(a)powerfulindividualswho knowledge regarding life's problems and an ability to are viewed as magnanimous and fair (b) receive the formulate appropriate judgments in the face of uncertainty collective approval (i.e., respect) of group members, which (Baltes&Smith,1990).Withincreasingage,influencemay (c) leads to social norms compelling compliance. Whether plateauorshow diminishing returns ratherthan decline. On the result of formal laws or informal norms, legitimacy is the other hand, older males have had limited access to bestviewedasaconstraintonortargetofsocialstatusrather market-related skills, which are likely an increasingly than asa form of status itself. importantpredictor ofinfluenceinTon'tumsi. Themajority InTon'tumsi,severalaggressive,dominantmenareprone of schools in Tsimane villages have existed for less than todrunkenbrawlsandspousalabuse.Anecdotally,thesemen 30years, and hence, individuals currently above age 50 are areviewedunfavorablybyothers.Personalgainbasedsolely less likely to be literate or to speak Spanish. Maxwell and on violence or the threat of violence is, in most cultures, Silverman (1970) conjecture that rapid institutional change, consideredillegitimate(Harsanyi,1966;Riches,1986).While leadingtoinformationobsolescence,translatesintoreduced theabilitytowinaphysicalfightdoesnotguaranteerespect, prestige for the elderly. We anticipate that the social-status neither does polyadic influence. The more educated and measures will increase with age until the 40s and decline materiallywealthymenofTon'tumsiareknowntomanipulate thereafter;the ability to win a dyadic fight should show the community discussions and decision making to their advan- strongest quadraticrelationship with age. tage.Thosewithlesseducationormarketacumenoftenresent theinfluenceoftheirmoreacculturatedpeers.Peoplearemore wary of what they do not understand. Thus, respect is less 4. Methods likely to accrue to men who excel in novel, market-related 4.1. Photo-ranking skillsthantomenwhoexcelinmoretraditionalskills,suchas horticulturalknowledgeorhuntingability. Allanalysesareoftheentireadultmalepopulationfrom Hunting prowess is the archetypal male skill among one of the more acculturated Tsimane villages, Ton'tumsi. foragers and correlates positively with social status almost There were 57 adult men over 18 years old among a total everywherethe relationshiphas beentested(Gurven& von village population of approximately 300 individuals. At the Rueden,2006; Smith,2004;Wiessner,1996).Oratoryskill, timeofdatacollection,only8ofthe57menwereunmarried expert tool manufacture, skill in warfare, and shamanic andno one had more than onewife. knowledgeareothervalued,traditionalskillsinAmazonian To generate rankings of our status measures and their small-scale societies, though the latter two skills are predictors, we asked Tsimane males to photo-rank other currently of low importance among the Tsimane. Tool males in their community. Status hierarchies are less manufacture is probably of decreasing importance in determined by the actual distribution of status within a Ton'tumsi since much of the resource-production techno- community than by people's perceptions of its distribu- logy, including rifles and cooking pots, is purchased in San tion. We believe that this justifies the measurement of Borjaorfromtravelingmerchants.However,manymenstill status in this study as the quantified rankings by Tsimane huntandfishwith bowsandarrows andtravelinself-made raters of other community members' social standing. wooden canoes. Furthermore, subjective impressions are based on years of Cross-culturally,olderindividuals aretypicallyofhigher interpersonal relationships and, thus, are probably better statusbecausetheyhavehadmoretimetoaccrueknowledge, metrics of personal attributes than researchers' observa- skill,wealth,orsocialsupport.Inmostsmall-scalesocieties, tions. Observation and subjective impression do tend to older men (but not necessarily the oldest men) exercise the correlate, however. Among the Ache, there is a strong greatest social power and receive the most deference positive correlation between a hunter's actual meat returns (Silverman & Maxwell, 1978; Simmons, 1945). Given the and his ranking of hunting success by other Ache (Hill & strengthrequirementsofmanystatus-relatedactivities,such Hurtado, 1996). Among the Shuar, one's work effort as as dyadic contest competition, age may not linearly predict perceived by others is highly correlated with actual work thesocial-statusmeasures.Strengthinmaleforagerstendsto effort (Price, 2006). peak in the 20s (Walker et al., 2002). Among the Tsimane, Arandomsampleof29Tsimaneadultmaleswasusedas hunting kill rate does not peak until age 40 but declines raters of their fellow villagers. The raters represented most C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 407 ages and all extended families within Ton'tumsi. Status 4.2. Interviews and anthropometrics rankings of the men in Ton'tumsi by two individuals not Alldemographicdatausedtoageindividualsanddescribe pickedasratersindicatedthattheratersamplewasnotbiased kinship relations come from extensive reproductive history towardslowerorhighersocialstanding.Irrespectiveofthese interviews done byM.G. during2002–2004.Foradescrip- precautions, no evidence that raters overestimated the tionofmethods,seeGurvenetal.(2007).Demographicdata qualitiesoffamilymemberswasfound.Sonsactuallytended inTon'tumsiwereupdatedduringthe2005fieldseasonwhen torate theirfathersbelow theirfathers' average scores! the photo-ranking took place. The demographic data allow The raters answered “yes” or “no” to questions concern- calculation of each male's degree centrality within the ing the presence or absence of a trait for other men in kinship network of Ton'tumsi. This measure is the total Ton'tumsi. These traits are hunting ability, being a hard number of households in Ton'tumsi in which a man or his worker, being funny, keeping promises, trustworthiness, wifecanclaimafullsibling,parent,oroffspring.Weassumed generosity in meat sharing, generosity in lending money, 100% paternity certainty. Our kinship measure purposely giving good advice, and how often visited. These variables conflates affinal and consanguineal kin in light of Hughes werechosenfromthetheoreticaldiscussionabove,personal (1988),whofindsthatleadersinsmall-scalesocietiespossess intuition, and ethnographic experience as a result of focus both high consanguineal and high affinal relatedness with group interviews with several Tsimane men and women othergroupmembers. concerningwhattheyconsiderimportanttraitsinadultmen. Data on commerce income (i.e., sales of horticultural Twenty-eightoftheraterseachevaluatedphotographsof16 produce) and wage labor income were collected over the other men in the village, and 1 randomly selected rater summer and fall of 2005 through weekly interviews of evaluated photographs of only 8 other men in the village. Ton'tumsi households. Flexed bicep circumference (of the Since the photographs were counterbalanced using a block dominant arm), chest circumference, height, literacy, design, this rating procedure ensured that the 57 men's Spanish fluency, and highest completed school level were photographswereevaluatedeighttimesforeverytrait,each also recorded for each adult male in the sample. A tape of the eight times by a different rater. A subject's score on measure was used for bicep and chest circumference, while thesemeasuresrangesfrom0to8andindicatesthenumber height was recorded with a Seca 214 Portable Stadiometer. ofraterswho answered “yes”tothe question. Literacy was determined by the ability to read a For fighting ability, whether the individual gets his way standardized sentence in Tsimane; the men were recorded in a group dispute, level of influence in the community, as unable to read, able to read poorly, or able to read well. whether the individual is well respected, and whether the Each man was questioned about his Spanish fluency and individual is likely to have more allies in the event of a could indicate no knowledge of Spanish, minimal know- conflict, a different rating procedure was employed. For ledge, or fluency. each of these traits, a rater was shown an array of eight photographs of Tsimane men and asked to rank them from 4.3. Data analysis highest to lowest, with a score of 8 assigned to the man ranked highest by the rater and 1 to the lowest. Twenty- Since we can analyze more than 20 predictors of status eight of the raters evaluated two arrays of photographs per but our sample size is limited to the 57 adult men in question, while one randomly selected rater evaluated one Ton'tumsi, factor analysis was employed to improve the array of photographs per question. A block design was subjects-to-variables ratio (Costello & Osborne, 2005). We employed so that no two subjects appeared together in the used a maximum likelihood factor extraction, which allows same array more than once. Thus, each of the 57 men goodness-of-fit tests, and an oblimin factor rotation, which was ranked eight times by eight different evaluators, simplifiesthefactorstructuresuchthatvariablesloadhighly yielding a range in scores from 8 (lowest) to 64 (highest). ononefactorandlessonothers.Wechoseanobliquefactor Fighting ability was ranked last so as not to bias a rater's rotation methodsince orthogonal rotation methods generate previous rankings. uncorrelated factors, resulting in a loss of valuable In general, the raters found the photo-ranking highly information if the factors are indeed correlated. Of the engaging and intuitive. Each rater photo-ranked his peers predictor variables analyzed, five contained nonsystematic with no one else present but C.V.R. The raters were made missing values, which were replaced with the sample aware of the confidentiality of their individual rankings. means. To reduce our errors of inference due to small PhotoswerePolaroidsofthetop-halfofeachman'sbody,set sample size, we divided our predictor variables into two against as neutral a background as possible. All photo- smaller groups before performing a factor analysis on each rankedmeasuresweretranslatedintoTsimanefromSpanish group. The first group, the “size and skills” variables, and then, as a test of the accuracy of translation, back- yielded three factors: a physical-size factor [on which, the translated into Spanish by Tsimane men from outside the following were loaded: (a) bicep circumference, (b) chest community of Ton'tumsi. All interviews of the raters were circumference, (c) height, and (d) weight], a food-produc- conducted in Tsimane. For definitions of the photo-ranked tion factor [on which, the following were loaded: (a) measures, seeAppendix1 in theSupplementary Material. hunting ability, (b) hardworking, and (c) commerce 408 C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 Table1 Standardizedβvaluesfrommultipleregressionanalysisofthesocial-statusmeasures(n=57) Winsdyadicfights Getswayingroup Communityinfluence Respect Socialsupporta Age .972⁎ .305 .199 −.716 .860 Age2 −1.049⁎⁎ −.221 −.026 .882 −.601 Physicalsizeb .536⁎⁎⁎ .285⁎⁎ .093 .027 .269⁎⁎ Foodproductionb .046 −.065 .102 .360⁎⁎ .112 Acculturatedb .196 .010 .301⁎⁎ .056 .546⁎⁎⁎ Prosocialpersonalityb −.093 −.013 .069 .159 .263⁎⁎ Socialsupportb .265⁎⁎ .468⁎⁎⁎ .628⁎⁎⁎ .400⁎⁎ - AdjustedR2 .711 .351 .762 .388 .517 ⁎⁎⁎ ⁎⁎⁎ ⁎⁎⁎ ⁎⁎⁎ ⁎⁎⁎ F(7,49) 20.656 5.323 26.585 6.071 10.989 a Initiallyapredictorvariablebutsubsequentlyanalyzedasadependentvariableduetoitsstrongrelationshipwiththeothersocial-statusmeasures. b Predictorsderivedfromfactoranalysis(seeSection 4). ⁎ pb.10. ⁎⁎ pb.05. ⁎⁎⁎ pb.01. income], and an acculturation factor [on which, the Supplementary Material). Additionally, height produces a following were loaded: (a) literacy, (b) Spanish fluency, smaller bivariate correlation with success in dyadic fights (c) education level, and (d) wage labor income]. The size (r=.571, n=53, pb.001) than does bicep circumference factor captured 26% of the variance in the data, the food- (r=.715, n=57, pb.001), chest circumference (r=.595, production factor captured 31% of the variance, and the n=57,pb.001),orweight(r=.779,n=53,pb.001). acculturation factor captured 11% of the variance. The Thesocial-supportfactorisalsoasignificantpredictorof factor extraction fit the data well [χ2(25)=17.906, p=.846]. winningdyadicfights.Marginallysignificantaretheageand The second group of predictors, the “social” variables, age2 terms. The food-production, acculturation, and proso- yielded two factors: a prosocial personality factor [on cialpersonalityfactorsdonotindependentlypredictwinning which, the following were loaded: (a) keeps promises, (b) dyadic fights inthe regression model. trustworthy, (c) gives good advice, (d) lends money, (e) generouslysharesmeat,(f)funny,and(g)visitedoften]and 5.2. Getting one's way in a group a social-support factor [on which, the following were Theregressionmodelissignificantandexplains35.1%of loaded:(a)numberofalliesand(b)kinnetworkcentrality]. thevarianceingettingone'swayinagroup.Socialsupport The prosocial personality factor captured 52% of the is the strongest predictor of getting one's way in a group, variance in the data, and the social-support factor captured thoughphysicalsizeisalsosignificant.Neithertheageterms 13% of the variance. The factor extraction fit the data well nor the other status-predictor factors even marginally [χ2(19)=19.114, p=.450]. See Appendix 2 in the Supple- approachsignificance. mentary Material for the rotated factor loadings. To determine the best predictors of status in Ton'tumsi, 5.3. Community-wide influence we linearly regressed our five factors on the four status measures: (a) winning a dyadic fight, (b) gets his way in a Theregressionmodelissignificantandexplains76.2%of group, (c) community-wide influence, and (d) being res- the variance in community-wide influence. The physical- pected. To each least-squares regression model, we added size, food-production, and prosocial personality factors or age and age2 as controls. See Table 1 for the regression the age and age2 terms did not independently predict results. Standardized β values are reported for ease of influenceintheregressionmodel.Therelationshipbetween comparison of the strength of each factor in predicting the social support and influence is the strongest of all the statusmeasure. regressionresultsinthisstudy.Thenumberofone'salliesis probablyagreatercontributortoinfluence(aswellastothe otherstatus measures)than isnumberofclosekin.Number 5. Results of allies alone explains 40% of the variance in winning a dyadic fight, 28% of the variance in getting one's way in a 5.1. Success in dyadic physical confrontation group, 78% of the variance in community influence, and Theregressionmodelissignificantandexplains71.1%of 43%ofthevarianceinrespect(adjustedR2values).Number the variance in winning dyadic fights. As anticipated, the of allies, furthermore, loads more heavily on the social- strongest predictor of winning fights is the physical-size support factor than does kin network centrality (see factor.Heightisprobablyleastresponsibleforthisrelation- Appendix 2in theSupplementaryMaterial). shipsinceitloadstheleastonthephysical-sizefactoramong The acculturation factor is an independent predictor of the four physical-size variables (see Appendix 2 in the community-wide influence in the regression model. Unlike C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 409 Table2 Pearsoncorrelationcoefficientsamongthesocial-statusmeasuresandtheirpredictors(n=57) Physical Food Prosocial Social Winsdyadic Getsway Community Age size productiona Acculturateda personalitya supporta fights ingroup influence Respect Age 1 .036 .205 −.665⁎⁎⁎ −.096 −.078 −.179 .049 −.070 .146 Physicalsizea 1 .292⁎⁎ .234⁎ .184 .518⁎⁎⁎ .781⁎⁎⁎ .524⁎⁎⁎ .535⁎⁎⁎ .328⁎⁎ Foodproductiona 1 −.289⁎⁎ .350⁎⁎⁎ .225⁎ .242⁎ .151 .244⁎ .459⁎⁎⁎ Acculturateda 1 .334⁎⁎ .474⁎⁎⁎ .424⁎⁎⁎ .250⁎ .497⁎⁎⁎ .115 Prosocialpersonalitya 1 .541⁎⁎⁎ .301⁎⁎ .278⁎⁎ .545⁎⁎⁎ .455⁎⁎⁎ Socialsupporta 1 .654⁎⁎⁎ .602⁎⁎⁎ .863⁎⁎⁎ .548⁎⁎⁎ ⁎⁎⁎ ⁎⁎⁎ ⁎⁎ Winsdyadicfights 1 .588 .675 .419 ⁎⁎⁎ ⁎⁎ Getswayingroup 1 .525 .339 ⁎⁎⁎ Communityinfluence 1 .653 Respect 1 a Predictorsderivedfromfactoranalysis(see Section4). ⁎ pb.10. ⁎⁎ pb.05. ⁎⁎⁎ pb.01. wage labor income, we surmise that commerce income (r=−.028, n=56, p=.837), hunting ability appears to drive neither predicts influence nor loads on the acculturation thefood production–respectrelationship. factor because horticultural commerce is a less rewarding Thefood-productionandacculturationfactorsnegatively enterpriseanddoesnotrequiretremendousinitialinvestment covary (see Table 2), which suggests that investments in in one's education. Compared to horticultural commerce, community-wide influence trade off with investments in wagelaboroffersTon'tumsimenbothhigheraverageweekly respect. Controlling for age, however, attenuates the earnings (US$4.37 vs. US$3.43) and potentially higher inverse relationship between food production and accul- absolute earnings (wage labor income σ2=27.50 vs. com- turation (partial r=−.213, n=57, p=.115). Furthermore, merceincomeσ2=12.80).Wagelabor,suchasworkingasa acculturation is neither a significant negative nor positive ranch-hand, logger, teacher, or agro-forestry consultant for bivariate predictor of respect. Table 2 presents the cross- Bolivian nationals, correlates highly with Spanish fluency correlations among all the status predictors and measures (r=.454, n=56, p=.001), literacy (r=.425, n=56, p=.001), of status. and influence (r=.256, n=56, p=.056), while commerce 5.5. Social support income is unrelated to Spanish fluency (r=−.073, n=56, p=.593),literacy(r=−.040,n=56,p=.768),orinfluence(r= The social-support factor is probably a strong mediator −.109, n=56, p=.426). The question remains, however, of the other status predictor variables; hence, we performed whetherliteracyandSpanishfluencyaretheresultsofwage a regression to gauge which predictor variables most labor or are, in fact, generating wage opportunities. associate with social support (see Table 1, column 6). The Controlling for age and age2, completed grade level regression model is significant and explains 51.7% of the (standardized β=.597, p=.001) and not wage labor income variance in social support. The acculturation factor is the (standardizedβ=−.050,p=.678) predictsliteracy inamulti- strongest predictor of the social-support factor, but the pleregression[F(4,38)=9.832,pb.001].Similarly,completed physical-size factor and the prosocial personality factor are grade level (standardized β=.882, pb.001) and not wage also significant. labor income (standardized β=−.024, p=.861) predicts Since the prosocial personality factor correlates with all Spanishfluency[F(4,38)=16.879,pb.001]. four status measures, especially influence and respect (see Table2),theeffectsofprosocialityonstatusacquisitionare 5.4. Respect likely mediated by social support. The mediating effect of The regression model is significant and explains 38.8% socialsupportislargelyduetonumberofallies.Prosociality of the variance in respect. While the physical-size factor, correlates with alliance strength (r=.508, n=57, pb.001) acculturationfactor,prosocialpersonalityfactor,andtheage butnotwithkinnetworkcentrality(r=.188,n=57,p=.161). and age2 terms do not predict respect in the regression Likewise, level of acculturation correlates with alliance model, social support and food production are strong strength (r=.473, n=57, pb.001) but not with kin network independent predictors of respect. Hunting ability loads the centrality (r=−.019, n=57,p=.887). highestonthefood-productionfactor(seeAppendix2inthe Among the prosocial personality variables, giving good Supplementary Material), and hunting ability correlates adviceloadshighestontheprosocialpersonalityfactor(see stronglywithrespect(r=.469,n=57,pb.001).Sincerespect Appendix2intheSupplementaryMaterial)andproducesthe isnotasstronglycorrelatedwithreputationasahardworker strongest correlation with community-wide influence (r=.286, n=57, p=.031) or horticultural commerce income (r=.507, n=57, pb.001). Generous meat sharing and how 410 C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 evenstrongerweretheoldestmaleinTon'tumsinotarecent immigranttothecommunity. When controlling for the acculturation factor, age does producesignificantpartialcorrelationswithgettingone'sway in a group (partial r=.270, n=57, p=.045), respect (partial r=.281, n=57, p=.036), and community influence (partial r=.398, n=57, p=.002). Wins dyadic fights produces no partialcorrelationwithage(partialr=.136,n=57,p=.318).In Ton'tumsi,thelackofalinearageeffectonthepolyadic-status measuresisinlargepartanartifactofnovelconditions. 6. Discussion 6.1. Interpretation of the results Among Tsimane men of Ton'tumsi, physical size is the primarydeterminantofdyadicfightoutcomes.Socialsupport isslightlymorepredictivethanphysicalsizeofgettingone's way in a group. Greater social support and acculturation result in community-wide influence, and greater social support and skill in food production (i.e., hunting ability) Fig.1.NonparametricLOWESSsmoothedcurvesfittedtothestatus-measure generaterespect.Largerphysicalsize,greateracculturation, rankingsbyage(n=57). and prosocial behavior are independently associated with moresocialsupport,andsocialsupportmediatestheireffects on the status measures. Age predicts status quadratically. often one is visited load nearly as high on the prosocial Fig.2illustratesthemostsignificantlinearpredictorsofmale personality factor. Among the prosocial variables, meat sharingisthestrongestbivariatepredictorofrespect(r=.450, n=57, pb.001). 5.6. Social status and age Fig.1displaysnonparametricLOWESSsmoothedcurves of each status measure by age. Fig. 1 suggests that the social-status measures relate to agequadratically,thatis,showamidlifepeak.Theabilityto winadyadicfightproducesthestrongestquadraticrelation- shipwithage[F(2,54)=6.843,p=.002],andrespectproduces theweakestquadraticrelationshipwithage[F(2,54)=1.418, p=.251]. None of the status measures produce a linear relationship with age (see Table 2). With the exception of gettingone'swayinagroup,statuspeaksinthe30sforthe average Tsimane male, which is earlier than we predicted. Respect showsonly slight declineswithold age. InTon'tumsi,thereareonly10menolderthanage50.On aggregate,theseindividualsarenotrecognizedashighstatus, and they lack many of the traits that are crucial to status acquisition. Age does not linearly predict the physical-size, food-production, prosocial personality, and social-support factors(see Table 2).Age negatively predicts theaccultura- tionfactor.Olderindividualsare,ingeneral,notovercoming deficits in wage labor income (r=−.406, n=56, p=.002) by Fig.2.Schematicoftheindependent,linearpredictorsoffourmeasuresof pursuingalternative incomestrategies.Ageshowsnolinear socialstatusamongtheadultmalesofTon'tumsi(n=57).Duetothesmall relationshipwithcommerceincome(r=.065,n=56,p=.636). samplesize,thisfigurepresentsresultsfromfourseparatemultipleregression While older men cannot claim more allies (r=−.104, n=57, analyses(seeTable1)ratherthananoverallpathanalysis.Arrowsindicate multipleregressionweightswithstandardizedvaluesN.25andpvaluesb.05; p=.442),theymaypossessrelativelymoresupportfromkin arrowsinboldfaceindicatemultipleregressionweightswithstandardized (r=.234, n=57, p=.079). This latter relationship would be valuesN.45andpvaluesb.01. C.vonRuedenetal./EvolutionandHumanBehavior29(2008)402–415 411 social status in Ton'tumsi, given the results of our multiple distributionofatraitiscomparedtoitsuniformdistribution. regressionanalysis. Gini coefficient values range from 0 (perfect equality) to Our results suggest that male fighting ability is deter- 1 (perfect inequality). Gini coefficients for wage labor minedmore bymusclesize andweightthan by height. Ina income(.59),commerceincome(.53),andliteracy(.35)are study of U.S. college students, flexed bicep circumference amongthehighestforallthepredictorsofstatusanalyzedin predicts lifting strength and self-reported conflict outcomes thisstudy.Forcomparison,the2005U.S.householdincome better than height, weight, or chest circumference (Sell, Ginicoefficientwas.47(DeNavas-Waltetal.,2006)andthe 2005). However, in a study of Indian men, height, weight, 2003EUhouseholdincomeGinicoefficientwas.31(Central and flexed bicep circumference all significantly correlate Intelligence Agency, 2008). Unequal distribution of a trait with aggression, but height and weight produce higher doesnotnecessarilymeanthattraitisstronglytiedtosocial bivariate correlations with aggression than does bicep size status, however. The Gini coefficients for Spanish fluency (Archer & Thanzami, 2007). Since the cross-correlations (.18), number of allies (.16), and bicep circumference (.04) amongallofthesevariablesareashighintheIndianandU.S. are relatively low. Most of the measures in this study, samples above as among the Tsimane, the different results including all four social-status measures, produce Gini reportedbythesestudiesmaybemoreapparentthanreal. coefficients of .20 or less. The other measures with Gini Independent of physical size, social support shows a coefficients above .20 are kinship network centrality (.38), strong relationship with winning a dyadic fight. Our money lending (.31), and meat sharing(.25). interpretation is that dyadic fighting ability is perceived as The strong relationship between hunting ability and indistinguishable from one's ability to elicit social support. respectisnoteworthy.InTon'tumsi,menhuntforanaverage The downstream consequences of any dyadic fight may of7hperweek,andhuntingreturnscontribute,onaverage, involve retaliation by the disputing parties' coalitions, and 22% of men's total daily food production (Gurven & von large size is a predictor of more social support within Rueden,2006).DespiteitsproximitytoSanBorja,Ton'tumsi Ton'tumsi. Alternatively, having more allies may simply isontheperipheryofold-growthforestwheregameanimals make one seem more physically formidable even when remain relatively abundant. In a less acculturated, more coalitional retaliationisnot anticipated. remote Tsimane village, however, men hunt for 11 h per For group dispute outcomes, coalitional support is para- week, and hunting returns contribute, on average, 50% of mount,butphysicalsizestillplaysanimportantindependent men's total daily food production (Gurven & von Rueden, role. Thus, getting one's way in a group is a form of social 2006). Perhaps the reduced contribution of hunting to food status among Tsimane males that is intermediate between production in Ton'tumsi explains in part why the food- winningadyadicfightandcommunity-wideinfluence. production factor does not predict the other social-status In generating community-wide influence, physical size measures. Nevertheless, hunting remains culturally vener- hasnoeffectindependentofsocialsupport.Individualswho ated, asgaugedbyitsrelationshipwithrespect. Analysisof aremoreprosocialhavemoreinfluencebut,again,onlyasa unpublished data also suggests Tsimane women in both result of their social support. Within one's social group, Ton'tumsi and a more remote community place similar individuals make decisions concerning how much of their emphasisonhuntingabilityinaprospectivehusband. resourcestoshareandwithhowmanyothers(Gurven,2004). As the Tsimane become more integrated in Bolivian Sharing decisions that optimize resource consumption via nationalsociety,theallocationofrespectmayremainhighly reciprocalaltruismmighttradeoffwithsharingdecisionsthat conservative relative to the more labile nature of the optimizestatusacquisitionviaallianceformation. predictors ofcommunityinfluence.Theacculturationfactor The regression results for community-wide influence showsnobivariateormultivariaterelationshipwithrespect; reveal the growing impact of acculturation within Tsimane it is possible that novel skills like literacy and the wealth it society.Independentoftheireffectsonsocialsupport,skills brings are perceived as a less legitimate, though an gained through formal education are of increasing impor- increasingly more successful, means of achieving social tancetocommunity-wideinfluencebecausethey(a)provide status. In any society, it seems that the nouveau riche are exclusiveaccesstoknowledgegermanetocommunity-wide seldom accorded respect by their peers. debates and/or (a) increase opportunities to gain and flaunt Socialsupportisthesinequanonofsocialstatusamong material wealth. Differences in time discounting among theTsimane.Ourresultssuggestthatsocialsupportiseither Tsimane males may determine who reaps more wealth and mediating or outcompeting the other status predictors in influencelaterinlife.InarecentstudyoftheTsimane,more generatingpolyadicsocialstatus.Thestrongestdeterminant patient individuals reported more years of schooling, and ofsocialsupportand,inparticular,one'snumberofalliesis 4 years later, they had earned greater wage labor income level of acculturation. Men with labor-market skills and (Reyes-Garciaet al.,2007). wage labor income may attract allies because, in part, they Measures of inequality in the distribution of income and can afford to be more prosocial. Income positively predicts literacy show that relatively few individuals in Ton'tumsi the number of times a Tsimane household gave gifts of currently benefitfrom the effectsof acculturation.The Gini money, food, or labor to other households (Godoy et al., coefficient is one measure of inequality in which the actual 2007). On the other hand, community-wide market