Evolution of Coalitionary Killing - University of Nebraska ... · and Wilson, 1988; Bueno de Mesquita, 1981, 1985;BuenodeMesquitaandLalman,1992). To some extent, therefore, patterns

Evolution of Coalitionary KillingRICHARD W. WRANGHAMDepartment of Anthropology, Peabody Museum Harvard University,Cambridge, Massachusetts 02138

KEY WORDS chimpanzee; lethal raiding; warfare; assessment

ABSTRACT Warfare has traditionally been considered unique to hu-mans. It has, therefore, often been explained as deriving from features thatare unique to humans, such as the possession of weapons or the adoption of apatriarchal ideology. Mounting evidence suggests, however, that coalitionalkilling of adults in neighboring groups also occurs regularly in other species,including wolves and chimpanzees. This implies that selection can favorcomponents of intergroup aggression important to human warfare, includinglethal raiding. Here I present the principal adaptive hypothesis for explainingthe species distribution of intergroup coalitional killing. This is the ‘‘imbalance-of-power hypothesis,’’ which suggests that coalitional killing is the expressionof a drive for dominance over neighbors. Two conditions are proposed to beboth necessary and sufficient to account for coalitional killing of neighbors: (1)a state of intergroup hostility; (2) sufficient imbalances of power betweenparties that one party can attack the other with impunity. Under theseconditions, it is suggested, selection favors the tendency to hunt and kill rivalswhen the costs are sufficiently low. The imbalance-of-power hypothesis hasbeen criticized on a variety of empirical and theoretical grounds which arediscussed. To be further tested, studies of the proximate determinants ofaggression are needed. However, current evidence supports the hypothesisthat selection has favored a hunt-and-kill propensity in chimpanzees andhumans, and that coalitional killing has a long history in the evolution of bothspecies. Yrbk Phys Anthropol 42:1–30, 1999. r 1999 Wiley-Liss, Inc.

TABLE OF CONTENTS

Coalitionary Killing among Chimpanzees and Other Species .............................................. 4Species distribution of coalitionary killing ......................................................................... 5The lethal-raiding problem ................................................................................................. 5Lethal raiding by chimpanzees ........................................................................................... 5

Territorial defense ............................................................................................................ 6Border patrols .................................................................................................................. 7Deep incursions ................................................................................................................ 7Coalitionary attacks ......................................................................................................... 8Coalitionary kills .............................................................................................................. 8

The Imbalance-of-Power Hypothesis .................................................................................... 11Explaining chimpanzee violence ....................................................................................... 11The significance of power imbalances ............................................................................... 12Origins of power imbalances ............................................................................................. 12Group territoriality and the benefits of lethal raiding ..................................................... 14Sex differences in territoriality and aggressiveness ........................................................ 16Bonobos: exceptions that support the rule? ...................................................................... 17The imbalance-of-power hypothesis and the evolution of human warfare ...................... 18

Challenges to the Imbalance-of-Power Hypothesis ............................................................. 20

YEARBOOK OF PHYSICAL ANTHROPOLOGY 42:1–30 (1999)

r 1999 WILEY-LISS, INC.

Uncertainty in the chimpanzee data ................................................................................ 20The claim that biology is irrelevant for human warfare .................................................. 21

Implications of the Imbalance-of-Power Hypothesis ........................................................... 22Chimpanzee and human psychology ................................................................................. 22The complexity of war ........................................................................................................ 23The relation between lethal raiding and hunting ............................................................ 24Morality .............................................................................................................................. 25

Conclusion ............................................................................................................................. 26Acknowledgments ................................................................................................................. 26Literature Cited .................................................................................................................... 27

Two related but distinct hypotheses haveproposed that warfare has its origins inpre-human violence. The first is no longersupported. This was the so-called ‘‘killerape’’ hypothesis, which stated that warfaresprings from an aggressive instinct thatbegan among australopithecines and contin-ued into humans (Dart 1953; Ardrey, 1961,1966; Lorenz, 1966; Tiger, 1969). RaymondDart based this idea on South African homi-nid fossils, which he interpreted with in-creasing pessimism after the Second WorldWar until eventually concluding that Austra-lopithecus africanus not only hunted othermammals but also killed adult conspecifics(Dart, 1953; Dart and Craig, 1959; Cartmill,1993). At the time of these ideas, intraspe-cific killing was considered to be absent inother wild mammals (including chimpan-zees Pan troglodytes) (Lorenz, 1966). There-fore, killing by australopithecines was con-sidered part of a uniquely hominid suite ofcharacteristics. Lorenz (1966) lent ethologi-cal authority to Dart’s ideas by suggestinghow killing could evolve. He proposed, forexample, that the use of weapons, such aspebble tools, could overcome natural inhibi-tions against killing conspecifics. Lorenz(1966) thus followed Dart in arguing thathuman warfare had evolved from australopi-thecine aggressive instincts.

The killer ape hypothesis provoked vigor-ous attacks (e.g., Ashley Montagu, 1968).Much criticism was directed at theoreticalcomponents, such as the claim that humanshave an innate aggressive drive that needsperiodic expression. It was empirical evi-dence that felled it, however. Most impor-tantly, the fossils suggestive of intraspecificviolence were convincingly reinterpreted in

terms of predation by carnivores and tapho-nomic processes, and the supposed bone andhorn weapons were better explained as frag-ments produced by carnivores chewing bone(Brain, 1981; Cartmill, 1993). The killer apehypothesis fell into general disrepute in the1970s, even though the notion that warfareevolved out of complex hunting patterns hasnot completely died (Morris, 1977; Ferrill,1985).

In the 1980’s a second, unrelated, set ofideas arose, which I collectively call thechimpanzee violence hypothesis (CVH). Likethe killer ape hypothesis, the CVH proposesthat human warfare is built on pre-humantendencies. In contrast to the killer apehypothesis, however, the CVH does not posita prior history of hunting, nor an aggressiveinstinct. These and other differences makethe killer ape hypothesis irrelevant to theCVH (Table 1). The remainder of this paperis concerned with the CVH, and not with thekiller ape hypothesis with which it hassometimes been confused (Sussman, 1997).

The CVH proposes that selection has fa-vored a tendency among adult males toassess the costs and benefits of violence, andto attack rivals when the probable net ben-efits are sufficiently high. It suggests thatthis tendency occurs as a result of similarconditions in the lives of chimpanzee andhuman ancestors, including a fission-fusionsystem of grouping, and intergroup hostility.It also raises the question of whether lethalraiding had a common origin in the ancestorof chimpanzees and humans around 5–6mya, or whether it evolved later and indepen-dently in each line.

The CVH was stimulated by observationsof male chimpanzees collaborating to kill or

2 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

brutally wound other adults (Goodall et al.,1979). Most such attacks were directed to-ward members of neighboring communities,in patterns reminiscent of human war raids(Goodall, 1986). As a result, various authorsraised the possibility of functional parallelsand/or evolutionary continuities linkingchimpanzee violence and human warfare(Trudeau et al., 1981; Otterbein, 1985, 1997;Goodall, 1986;Alexander, 1987, 1989; Wrang-ham, 1987, 1999b; Ghiglieri, 1988; van Hooff,1990; Hamburg, 1991; Knauft, 1991; Man-son and Wrangham, 1991; Boehm, 1992; vander Dennen, 1995; Wrangham and Peterson,1996; Boesch and Boesch, 1999). For ex-ample, Otterbein (1997, p 253) noted thatsimilarities between chimpanzee communi-ties and human bands suggest that ‘‘earlyman . . . is likely to have been organized intolocalized groups of related males, groupsthat engaged in intergroup conflict.’’ If so,Otterbein concluded, warfare has been con-tinuous in human and pre-human ancestry

for at least 5 million years. This idea of anancient origin of warfare is supported by therarity of coalitionary lethal violence towardadult conspecifics in other primates, and byevidence that subsequent to the split withgorillas Gorilla gorilla (Pilbeam, 1996), chim-panzees and humans share a common ances-tor around 5–6 mya.

In the first part of this paper, the evidencefor coalitionary killing by chimpanzees andthe nature of their intergroup aggressionare examined. The principal adaptive expla-nation linking chimpanzee and human vio-lence is then reviewed. This is the imbalance-of-power hypothesis, which states thatcoalitionary kills occur because of two fac-tors: intergroup hostility, and large powerasymmetries between rival parties. Afterconsidering separately the costs and ben-efits of lethal raiding among chimpanzees,how the imbalance-of-power hypothesis alsoapplies to bonobos (Pan paniscus) and tohumans is assessed.

In the third part, objections and problemsare considered. Arguments are discussedthat deny the relevance of biological argu-ments for understanding human warfare,including the following claims: warfare iswholly cultural (e.g., Keeley, 1996); modernwar is too complex for individual aggressionto be important (Hinde, 1993); and, nothinguseful can be learned by studying speciesother than humans, because humans arealready known to be violent (Leach, 1968;Gould, 1996).

Criticisms directed specifically at the im-balance-of-power hypothesis are also dis-cussed. These include concerns about thevalidity of the chimpanzee data (e.g., Power,1991; Sussman, 1997) or about the interpre-tation of human data (e.g., Knauft, 1991;Sponsel, 1996), and claims that data onbonobos undermine the use of chimpanzeesas a reference species for early hominidancestry (Zihlman, 1997; Stanford, 1998a).They also include accusations of geneticdeterminism (e.g., Regal, 1998; Sussman,1997) or societal bias (e.g., Sussman, 1997).

These are important considerations, butthey do not invalidate the comparative ap-proach. Therefore this article ends with abrief discussion of the implications of the

TABLE 1. Comparison of the killer ape andchimpanzee violence hypotheses

Killer apehypothesis

Chimpanzeeviolence

hypothesis

Lethal violence Important inhuman evolu-tionary history

Important inhuman evolu-tionary history

Significance ofhunting

Critical pre-cursor to intra-specific vio-lence

Possible conse-quence ofintraspecificviolence; not anecessary pre-cursor

Mechanism ofaggression

Instinct Strategic assess-ment

Putative reasonfor intraspe-cific violence

Inadvertentbreakdown ofnatural inhibi-tions

Assessment thatcosts of elimi-nating rivalare low

Chimpanzeesconsidered tobe

Nonviolent Strategically vio-lent

Killing amonganimals otherthan hominids

Assumed to beabsent

Known to occur

Reliance on fossilevidence

Critical Relevant, notcritical

First ancestor ofhumans sup-posed to havecoalitionaryviolence

Australopithe-cines

Unknown

Evolutionarymechanismsfavoring vio-lence

Include groupselection

Group selectionappearsunnecessary

3EVOLUTION OF COALITIONARY KILLINGWrangham]

imbalance-of-power hypothesis for humanpsychology, warfare, and morality.

COALITIONARY KILLING AMONGCHIMPANZEES AND OTHER SPECIES

Species distribution of coalitionary killing

Contrary to initial assumptions (Lorenz,1966), research in recent decades has re-vealed that intraspecific killing occurs in avariety of species, commonly following pat-terns explicable by natural selection theory.For example, among primates infanticide iswidely reported, typically committed by non-relatives (Hausfater and Hrdy, 1984; Palom-bit, 1999). Among spiders, killing of adultsoccurs predictably when resources of highvalue are at stake (Austad, 1983). Amongants, large imbalances of power increase theprobability of lethal intercolony aggression(Holldobler, 1981; Adams, 1990). Obviousparallels can be found among humans (Dalyand Wilson, 1988; Bueno de Mesquita, 1981,1985; Bueno de Mesquita and Lalman, 1992).To some extent, therefore, patterns of hu-man killing appear to follow the ordinarypatterns of lethal aggression found in otherspecies.

Not so ordinary, however, is the way thathuman killing occurs. Among humans mostkilling occurs in warfare, where the predomi-nant style of violence is coalitionary. In mostanimals, by contrast, even where aggressionoccurs at high rates, lethal violence is dyadic(one versus one) rather than coalitionary(many vs. one, or many vs. many). Duringrut-fighting among male pronghorn ante-lope (Antilocapra americana), for example,12% of 82 fights over mating rights to es-trous females led to the death of one or bothmales (Byers, 1997). Likewise, in differentpopulations of red deer Cervus elaphus, 13–29% of adult male mortality came fromrut-fighting (Clutton-Brock et al., 1982).Many similar examples occur. Deaths tendto occur in intensely escalated contests inwhich both opponents expose themselves tohigh risk of injury, typically because ‘‘amajor part of a contestant’s lifetime repro-ductive success is at stake’’ (Enquist andLeimar, 1990). But killing is never coalition-ary in these species.

The explanation for the widespread ab-sence of coalitionary violence is trivial. Most

species never form coalitionary alliances inany context. But coalitions are not a suffi-cient condition for coalitionary killing. Thusmany primates form coalitions without anyevidence of adult-killing [e.g., Cercopithecusaethiops (Cheney et al., 1988)] or with fatalfighting known only from dyadic interac-tions [e.g., Cebus capucinus] (Miller, 1998).

Indeed, the only nonprimate mammal forwhich coalitionary violence is known to becommonly responsible for adult deaths isthe wolf Canis lupus. In at least three sites,adults are known to kill other adults at highrates [Denali (Alaska), Isle Royale (Michi-gan), and Minnesota] (Mech et al., 1998).For example in Denali, 39–65% of adultmortality was due to intraspecific killing,based on 22 intraspecific killings recordedfrom 17–20 packs (Mech et al., 1998). This isthe least disturbed study site of wolves. Innortheastern Minnesota, 43% of wolves notkilled by humans were killed by other wolves.Killings tended to occur in buffer zones(where territories met), which wolves mostlyavoided (Mech, 1994). These data suggestedto Mech et al. (1998) that intraspecific kill-ing is a normal consequence of wolf territori-ality.

Occasional coalitional killing of adult con-specifics in neighboring groups has also beenrecorded among other social carnivores [li-ons Panthera leo, spotted hyenas Crocutacrocuta, cheetahs Acinonyx jubatus (Kruuk,1972; Caro and Collins, 1986; Goodall, 1986;Packer et al., 1988; Grinnell et al., 1995)]and at least one group-territorial bird [Tas-manian native hen Gallinula mortierii (Put-land and Goldizen, 1998; A. Goldizen per-sonal communication)]. However, thefrequency of killing has not been reportedfor these species. Outside of mammals, onlysocial insects are known to kill conspecificsregularly with coalitional aggression (vander Dennen, 1995).

These data suggest that animals can bedivided into three major categories. Firstare species in which intraspecific killing ofadults is rare (e.g., less than 1% of all adultdeaths). Most species fall into this category.Second are those where killing occurs morefrequently [often 10% of more of deaths(Enquist and Leimar 1990)], but entirely indyadic interactions. In these species, fatal

4 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

fighting is dangerous. In the third category,killing is also frequent, but differs by beingpolyadic (coalitionary). Furthermore, as ar-gued below, fatal fighting need not be danger-ous for the killers.

Ants dominate this last category, whichcontains probably less than 10 mammalianspecies and perhaps no other vertebrates.Chimpanzees and humans (or at least, cer-tain populations of these species) are theonly primates known to be frequent coalition-ary killers. A possible additional candidateis the western red colobus monkey (Colobusbadius), for which at least two and possiblyfour coalitionary kills were recorded by Sta-rin (1994). In these cases, coalitions of fe-males attacked and killed males attemptingto enter their groups.

The lethal-raiding problem

In chimpanzees, humans, and some otheranimals, coalitionary killing can occur in thecontext of lethal raiding. Lethal raids are anunusual form of aggression because they donot escalate from a conflict. Instead, partiesof allied males collectively invade a neighbor-ing territory, seek one or more vulnerableneighbors, apparently assess the probabilityof making a successful attack, conduct a‘‘surprise’’ attack that leaves one or morevictims dead or dying, then return to theirown territory. ‘‘Surprise’’ refers to the attackoccurring without any initial conflict, with-out escalation from a lower level, and with-out the victim interacting with the oppo-nents until the attack starts.

Thus, lethal raids indicate an appetite forhunting and killing rivals that is akin topredation. By contrast, most animal con-flicts escalate in a stepwise manner thatallows both opponents to assess each otherand to withdraw when the risks of losingappear too high (Archer and Huntingford,1994). The ‘‘appetite for lethal raiding’’ there-fore requires explanation in different termsfrom escalated conflicts.

Among humans, lethal raids are wide-spread in all forms of warfare. For example,Keeley (1996) regards small raids and am-bushes as ‘‘the commonest form of combatemployed in primitive warfare’’ (see alsoTurney-High, 1949; van der Dennen, 1995;Maschner and Reedy-Maschner, 1998).

Abundant evidence routinely attests to theblood-lust of the participants.

Against the notion that men have a readyappetite to attack their enemies, combat-ants in modern warfare are often reluctantto fight (Hinde, 1993; Ehrenreich, 1997;Grossman, 1999). This front-line lack ofaggressiveness is understandable becausein modern warfare, unlike intergroup aggres-sion in primates, soldiers are organized hier-archically and are ordered into battle bytheir superiors, regardless of their personalmotivation. Participation in raids amongpre-state societies, however, is normally vol-untary (Keeley, 1996). Thus, reluctance ofsoldiers under orders does not underminethe more widespread phenomenon of maleeagerness for fighting.

It is likely that lethal raids also occur insome of the carnivore species that engage inintraspecific killing of adults. For example,lethal raiding is suggested by the report ofMech et al. (1998) that neighbors killedthree members of a wolf pack, two othersdisappeared unseen, and the defeated pack’sterritory was taken over by the killers. Thereare also reports of spotted hyenas makingincursions into neighboring territories toattack neighbors (Goodall, 1986). Amonginvertebrates, patterns similar to lethal raid-ing occur in a variety of ants (Holldobler,1981; van der Dennen, 1995).

Among species other than humans, how-ever, lethal raids have been most clearlyreported in chimpanzees. Evidence of chim-panzee raiding has been fundamental forthe development of the CVH. I thereforereview the chimpanzee data in detail.

Lethal raiding by chimpanzees

Although lethal raiding among chimpan-zees has been described more clearly thanfor any other mammal, few cases have beencompletely observed. Furthermore, all thedetailed observations come from a singlesite, Gombe National Park in Tanzania(Goodall, 1986). Of course, lethal raids areexpected to be rare, as they must be in anylong-lived, slowly reproducing species. Nev-ertheless, the concentration of observationalevidence at Gombe has suggested to criticsof the CVH that lethal raiding may havebeen induced in Gombe by unnatural condi-

5EVOLUTION OF COALITIONARY KILLINGWrangham]

tions such as reduced habitat, or provision-ing, and therefore that lethal raiding isuncharacteristic of chimpanzees more gener-ally (Power, 1991).

There are six study sites in which habitu-ation of chimpanzees is sufficiently good toallow multi-hour observations of known indi-viduals traveling throughout the home range(Tables 2 and 3). These include four studiesof the eastern subspecies (P. t. schwein-furthii), none of the two central subspecies(P. t. troglodytes and P. t. vellerosus), and twoof the western chimpanzee (P. t. verus). Inone of the P. t. verus studies, at Bossou, thestudy community is ‘‘semi-isolated’’ by lossof habitat, separated by several kilometersfrom the home ranges of its closest neigh-bors (Sugiyama, 1989; Sugiyama et al.,1993). Consequently there is no possibilityof territorial behavior or intercommunityinteraction from Bossou. This leaves fivestudies of chimpanzees, varying in durationfrom 8 to 38 years, that permit observationof intergroup interactions (Table 3).

Six components of chimpanzee intergroupaggression are especially relevant to lethalraiding: territorial defense (showing evi-dence of hostile intergroup relationships),border patrols, deep incursions, coalitionaryattacks, coalitionary kills, and border avoid-ance (Table 2).

Territorial defense. This has been re-ported in all studies in which intercommu-nity relationships have been described, basedon some combination of: counter-calling be-tween parties of neighboring males; rapidtravel toward a site where an opposing partyhas been detected; avoidance of opposingparties that were obviously larger; chargingdisplays directed toward an opposing partyof males; or one party chasing another(Nishida, 1979; Goodall, 1986; Boesch andBoesch, 1999; V. Reynolds personal commu-nication; Wrangham et al., in preparation).For example on nine occasions chimpanzeesin Taı have been seen in ‘‘back-and-forthattacks’’ with neighbors, in which all males

TABLE 2. Territorial behavior in chimpanzees and bonobos1

P. t. schweinfurthii P. t. verus P. paniscus

Gombe Mahale Kibale Budongo Taı Wamba Lomako

Territorial defense 1 1 1 1 1 1 1Border patrols 1 1 1 ? 1 2 2Deep incursions 1 1 1 ? 1 2 2Coalitionary attacks 1 1 ? ? 1 2 2Coalitionary kills 1 1 1 1 2 2 2Border avoidance 1 ? 1 ? 1 2 2Peaceful intercommunity association 2 2 2 2 2 1 1

1 All long-term studies are included except for Bossou, where the community is isolated by agricultural land from its nearestchimpanzee neighbors. ‘‘Coalitionary kills’’ refers to adult victims only. ‘‘Coalitionary attacks’’ means non-lethal attacks by severalmales on a single victim. Data are from: Gombe (Goodall, 1986); Mahale (Nishida, 1979, 1986; Nishida et al. 1985); Kibale (Chapmanand Wrangham, 1993); Budongo (V. Reynolds, personal communication); Taı (Boesch and Boesch, 1999); Wamba (Kano, 1992;Hashimoto et al., 1998); Lomako (White, 1996).

TABLE 3. Chimpanzees and bonobos: lethal violence by site1

Site Subspecies Adult deaths Infanticides Years of study

Gombe P. t. schweinfurthii 6 (3) 6 (3) 38Mahale P. t. schweinfurthii 1 (6) 4 (6) 33Kibale P. t. schweinfurthii 2 (0) 1 (0) 11Budongo P. t. schweinfurthii 1 (1) 1 (1) 8Bossou P. t. verus 0 (0) 0 (0) 22Taı P. t. verus 0 (0) 0 (0) 19Wamba P. paniscus 0 (0) 0 (0) 24Lomako P. paniscus 0 (0) 0 (0) 15Total 10–20 12–22 1701 Numbers include (for adult deaths) kills recorded on the basis of direct observation and/or fresh bodies as well as (in parentheses)those from suspicious disappearances (see Table 5), or (for infanticides) kills observed directly or (in parentheses) inferred from context(Arcadi and Wrangham 1999, updated for Mahale by Nishida (personal communication). ‘‘Years of study’’ is number of years frombeginning of continuous study until 1998. Bonobo studies have been intermittent.

6 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

rush toward the opponents, giving loud at-tack calls, and opponents may flee up to 400m. In a further nine cases, males in the frontline of attack were supported by loud callsfrom females in a rear line (Boesch andBoesch, 1999).

Border patrols. These are an intrinsiccomponent of lethal raiding, because theyput a party of individuals in a position tostalk and to hunt a neighboring victim.Border patrols are visits to a peripheralsector of a home range by a party of malesthat monitors the area. They are initiatedwithout any immediate contact with mem-bers of the neighboring community, and areoften undertaken with little or no feeding.According to Goodall (1986) and Boesch andBoesch (1999), they include some or all ofthe following features: (1) cautious and slowtravel around or across the border, includinglong periods of gazing toward the neighbor-ing home range; (2) nervousness shown to-ward unexpected sounds; and (3) inspectionof signs of other chimpanzees, such as dis-carded food wadges, feces, nests, or aban-doned termite-fishing tools.

Border patrols were first reported atGombe in 1971 by JD Bygott, who was thefirst researcher to conduct regular all-dayobservations of individual males (Bygott,1979). Most patrols at Gombe were by malesof the principal study group, the Kasekelacommunity, but the Kahama males pa-trolled also (Bygott, 1979; Wrangham, 1975).Border patrols do not occur every time aparty reaches the boundary area. In Gombe,for example, patrols occurred during 28% of134 boundary visits made by parties fromthe Kasekela community from 1977 to 1982(calculated from Goodall, 1986, Table 17.1).Their frequency appears to vary as a func-tion of relations between particular commu-nities. Thus, during and after the 1974–1977 period, during which the Kasekelacommunity killed the males of the neighbor-ing Kahama community, Kasekela borderpatrols were disproportionately directed to-wards the Kahama territory (Goodall, 1986).The recorded frequency of border patrols byKasekela males was highest in 1972–1973,when 13 border patrols occurred in 58 daysof observation, i.e., a rate of 82 per year

(Wrangham, 1975, Table 5.9). By contrast,in the 5 years following the extinction of theKahama community (1978–1982), border pa-trols continued at a rate of 18 per individualper year (range 9–27; calculated from Good-all, 1986, Table 17.1).

Table 2 shows that border patrols havebeen reported also from Mahale, Kibale, andTaı. Border patrols have not been describedin detail from Mahale, but key elements ofborder patrols have been reported—includ-ing scouting, and silent and cautious travel,mainly by males, in border areas (Nishida,1979, 1990; Nishida et al., 1985). The Kibaleevidence comes from border patrols seen atKanyawara, involving parties of males inter-mittently checking their territorial bound-aries (Wrangham et al., in preparation). InTaı, border patrols occurred in 29% of 129territorial actions, normally involving atleast four males, and included all of theelements listed above (Boesch and Boesch,1999).

Deep incursions. These were included aspart of border patrols by Goodall (1986), butI distinguish them because they involvedeliberate travel into the neighboring terri-tory rather than merely checking of theborder area. Deep incursions are character-ized by (1) substantial penetration into theneighboring territory, e.g., for one kilometeror more; (2) silent and cautious travel dur-ing periods of moving outwards from ownterritory; and (3) noisy and vigorous dis-plays on return to their own territory. Deepincursions have been well described atGombe and Taı. Boesch and Boesch (1999)found that ‘‘many patrols were probablyaimed at finding and attacking strangers. . .. . . (they) were impressive by the intensitywith which the males searched for strang-ers, not only entering deep (into) their terri-tory, but once even heading backwards tofind the neighbors.’’ Deep incursions lastedup to 6 hours, and on 5 of 129 territorialinteractions in Taı, led to attacks. All incur-sions were led by males.

Coalitionary attacks. These are interac-tions in which observers assess that theintent of those in the aggressive party is tohurt or to kill one or more victims. They can

7EVOLUTION OF COALITIONARY KILLINGWrangham]

occur within the territory of either the ag-gressors or the victims, or in the boundaryarea. In Gombe, coalitionary attacks haveincluded both interactions between the well-habituated Kasekela and Kahama communi-ties, as well as other communities. In late1974, for example, a party of three Kahamamales encountered a male and female foundto the south of their territory (the ‘‘Kalande’’community). Two of the Kahama malesgrabbed and attacked the Kalande male, buthe escaped without serious injury (Goodall,1986).

In Mahale, cases included both attacks byM-group toward K-group males, and viceversa. For example, Nishida (1979) recordedthree K-group males chasing an M-groupmale for 200 m. Two of the pursuers gave up,but the third caught the victim, forced himto the ground, bit his thigh, stamped on him,chased him as he tried to escape, but thensuddenly gave up.

In Taı, Boesch and Boesch (1999) reporteda category of attack which they called ‘‘com-mando,’’ in which a party of males (with orwithout females) penetrated into the neigh-boring range and attacked one or moreneighbors, i.e., a combination of a ‘‘deepincursion’’ with a coalitionary attack. TheTaı males sometimes ‘‘waited and listenedsilently for hours before they attacked. Wetwice saw the study community being victimof a commando attack, in one of whichMacho (an adult male) escaped with 19wounds’’ (Boesch and Boesch, 1999). A sec-ond form of coalitionary attack was the‘‘lateral attack’’ (seen six times), in which aparty moved laterally while looking in thedirection of strangers, then approached,chased, and on at least one occasion caughtand attacked one of the opponents. Thisappeared to be a tactic for increasing theimbalance of power by isolating a victimfrom the rest of the party (Boesch andBoesch, 1999).

In Kibale, no complete coalitionary at-tacks have been seen, but twice parties ofneighbors have silently charged toward iso-lated males of the Kanyawara study commu-nity, then veered off on seeing observers,suggesting that coalitionary attacks wereaverted by the presence of humans. In athird case, five Kanyawara males attacked

an adolescent male and nulliparous femalenear the territorial border, but later re-treated when confronted by four adult malesfrom the neighboring community, who chasedthe aggressors for 700 m (M. Muller, per-sonal communication).

Coalitionary attacks also occur within com-munities. For example, a bullying and insub-ordinate young adult male (Jilba) in Ma-hale’s M-group was attacked so severely bysix males and two females that it took him 3months of traveling alone before he recov-ered sufficiently to rejoin the community(Nishida, 1994).

Coalitionary kills. No cases have beenreported of dyadic violence leading to thedeath of an adult chimpanzee. However,lethal coalitionary attacks on adults havebeen reported from all four study sites of theeastern subspecies (Table 4). Table 4 listsknown and inferred cases. The largest sam-ple comes from Gombe and includes fiveobserved brutal attacks followed by the dis-appearance of the victim (four male, onefemale), and one case of a fresh corpse of anunidentified female, considered to have beenkilled by the Kahama males. Goodall (1986)summarizes the observations as follows: theattacks lasted at least 10 min each; thevictim was always held to the ground by oneor more of the assailants while others at-tacked; the victim was dragged in at leasttwo directions, eventually gave up resisting,and was essentially immobilized by the endof the attack.

In Mahale, Nishida et al. (1985) recordedthe deaths of all six adult males of K-groupcommunity between 1969 and 1980. In Nishi-da’s words, the observers ‘‘speculate that atleast some adult males, particularly So-bongo and Kamemanfu, were killed by M-group’s chimpanzees. Severe fighting wasoccasionally witnessed between males of K-group and M-group in (their area of over-lap). . . . M-group’s males were sometimesseen to penetrate into the core area of K-group’s range from 1974 onwards’’ (Nishidaet al., 1985, p 288). The males who disap-peared were all healthy, not senile. In onecase, M-group males were known to be verynear to K-group males; there were manyoutbursts of calls; and the next day another

8 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

K-group male was missing (Kasonta) (TNishida, personal communication).

In Kibale, after 3 days during which malesfrom the Kanyawara and Rurama communi-ties had been counter-calling at each otherin a hostile manner, a Kanyawara male(Ruwenzori) was found freshly dead in theborder area. His body, huddled face down atthe bottom of a slope around which thevegetation had been beaten down, showedclear evidence of a violent attack by chimpan-zees. In an unrelated incident, theKanyawara males were followed by observ-ers to the fresh corpse of an individual froma neighboring community (Sebitole) who hadapparently been killed by chimpanzees theprevious evening. There were numerouswounds on the front of his body, his tracheahad been ripped through, and both testicles

had been removed. Nine Kanyawara maleshad been patrolling the border on the previ-ous evening, all of whom were present thenext morning (17 hours later), and several ofwhom beat on the victim’s body and draggedit about (M. Muller, personal communica-tion).

Like coalitionary attacks, coalitionary killsalso occur within communities. At Budongoin 1998, an adult male was killed by othermales of his own community (K. Fawcett,personal communication). Intracommunitykilling is also thought to have occurred inMahale, where adult male Ntologi was founddead in the center of M-group’s territorywith numerous wounds on his body. Hisdeath followed several coalitionary attackson him by his former subordinates—afterhis defeat as alpha-male of M-group

TABLE 4. All reported intraspecific kills of adult chimpanzees1

Result Site Date

Aggressor’s Victim’s

Victim’s ID Ref.Community Party Community Party

Death Gombe 1974 Kasekela 7M, 1F Kahama 1M Godi Goodall (1986, p.50)

Death Gombe 1974 Kasekela 3M, 1F Kahama 3M, 1F De Goodall (1986, p.50)

Death Gombe 1975 Kasekela 5M Kahama 1M Goliath Goodall (1986, p.50)

Death Gombe 1977 Kasekela 6M Kahama 1M Sniff Goodall (1986, p.51)

Death Gombe 1975 Kasekela 4M Kahama 1F Madam Bee Goodall (1986, p.51)

Death Mahale 1995 M-gp gang M-gp 1M Ntologi Nishida (1996)Death Budongo 1998 Sonso gang Sonso 1M Zesta (12

injured)K. Fawcett (per-

sonal commu-nication)

Death Kibale 1992 Rurama gang Kanyawara #1M Ruwenzori KCPDeath Kibale 1998 Kanyawara gang Sebitole $1M Unknown KCPDeath Gombe 1972 Kahama ? Kalande? $1F 1F Wrangham

(1975)Death Gombe 1977 Kasekela 5M Kahama $1M Charlie Goodall (1986, p.

50)Death? Mahale 1996 M-gp ? M-gp 1M Jilba M. Huffman

(personal com-munication),Hofer et al.(1998)

Death? Mahale 1970-83 M-gp ? K-gp ? Some males? Nishida et al.(1985)

Death? Gombe 1981 Kalande ? Kasekela ? Humphreykilled?

Goodall (1986, p.51)

Attack Gombe 1974 Kahama 3 M Kalande 1M, 1F M attacked,caught by 2,escaped

Goodall (1986, p.49)

Attack Gombe 1980 Kalande ? Kasekela 1F Passion(inferred)

Goodall (1986, p.51)

Attack Mahale 1974 K-group 3M M-group 1M 1 on 1 fight Goodall (1986, p.51)

1 Parties for aggressors and victims show the number of adult males (M) and females (F). Letters in bold show the victim’s sex. KCP(citation for Kibale deaths) is records of the Kibale Chimpanzee Project.

9EVOLUTION OF COALITIONARY KILLINGWrangham]

(Nishida, 1996; T. Nishida, personal commu-nication).

The chimpanzee data are summarized inTable 5. Tables 4 and 5 also list suspiciousdisappearances. These are cases where ob-servers believed the most likely explanationfor disappearances was that they were killedby neighbors, because: (1) those who disap-peared were healthy and not senescent; (2)other causes of death appeared improbable;and (3) the disappearances occurred at atime and place where there were patentlyhostile relationships with a neighboring com-munity.

Although the reported episodes of lethalcoalitionary violence are still few, the kill-ings are noteworthy because they have beenreported from four sites and, in relation tototal observed adult deaths, they appear tobe demographically significant. In Gombe,data reported by Goodall (1986) indicatethat for adult males in Kasekela and Ka-hama, the proportion of adult male mortal-ity from intraspecific coalitionary aggres-sion was 30–40%. Although fewer kills havebeen seen elsewhere, it seems likely thatthis variation is partly a function of observa-

tion time. Thus, Figure 1 shows that inrelation to observation time, the number ofobserved and suspected kills appears simi-lar in the four schweinfurthii study sites.The idea that as observation years accumu-late, more killing will be seen, is supportedby the data on infanticides, which show asimilar trend (Fig. 1). In summary althoughthe evidence needs to be substantiated bycontinuing observation, current evidence isthat in all four populations of the easternsubspecies, adults kill each other occasion-ally through coalitionary violence. Figure 1suggests a rate of approximately 0.25 adultskilled per year.

On the other hand, there is no evidence oflethal intraspecific aggression toward eitheradults or infants from either of the studies ofthe western subspecies, i.e., from Taı orBossou. Because the Bossou community hasno neighbors and few males, low rates ofaggression are not surprising. However, asdemonstrated by Figure 1, lethal coalition-ary aggression would be expected to havebeen seen at Taı: the study is now 21 yearsold and should therefore have produced evi-dence of about five killings each of adults

Fig. 1. Chimpanzee intraspecifickillings by study site. S, P. t. schwein-furthii; v, P. t. verus. See Tables 3 and 4for data and sources.

TABLE 5. Chimpanzee coalitionary kills of adults1

SiteAggressor

communityVictim’s

community

Kill seen,or freshcorpse

Suspiciousdisappearance Reference

Gombe Kasekela Kahama 5 2 Goodall (1986)Gombe Kahama Kalande 1 Goodall (1986)Gombe Kalande Kahama 1 Goodall (1986)Kibale Rurama Kanyawara 1 KCPKibale Kanyawara Sebitole 1 KCPMahale M-group K-group 6? Nishida et al. (1985)Mahale M-group M-group 1 1 T. Nishida (personal communication)Budongo Sonso Sonso 1 K. Fawcett (personal communication)Total 10 10?1 Intra-community aggression is indicated by the victim and aggressors being in the same community. KCP, records of the KibaleChimpanzee Project.

10 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

and infants if this population conformed tothe schweinfurthii pattern. Although thesample sizes are small, the fact that Taıchimpanzees show all components of lethalraiding but no coalitionary kills suggest thatthe nature of aggressive relationships dif-fers between Taı and the eastern popula-tions (below).

Finally, border avoidance is expected ifindividuals are aware that the border isphysically dangerous. Low frequency of useof border areas has been documented byBoesch and Boesch (1999), who found that inTaı 75% of time was spent in the central 35%of the range. When Gombe or Kibale(Kanyawara) chimpanzees do visit borderareas, they tend to do so in parties that arerelatively large (Gombe) and contain a higherproportion of males than normal(Kanyawara) (Bauer, 1980; Chapman andWrangham, 1993). Finally, high prey densi-ties in border areas have been reported forthe main prey species of chimpanzees, redcolobus (Colobus badius), both in Gombe(Stanford, 1998b) and Kibale [Ngogo (D.Watts, personal communication)]. Similarlyamong wolves, prey densities are higher inborder areas between territories, a result ofavoidance of those areas by wolves (Mech etal., 1998). Border avoidance by territoryholders has not been reported in other non-human species, but is presumably wide-spread in humans.

In summary, there are five study sites(Gombe, Mahale, Taı, Kibale, Budongo) inwhich chimpanzees have neighbors andwhere intercommunity interactions havebeen at least partly described or observed(Table 2). At the three best-documented sites(Gombe, Mahale, Taı) patterns of territorialinteraction appear similar: all of them showterritorial defense and border patrols byadult males, with violent coalitionary at-tacks on neighbors. All these patterns aresimilar to the data on wolves, which is theonly other nonhuman mammal with fission-fusion grouping and group territories inwhich intergroup interactions have beenwell described. These points suggest that allthe major elements of lethal raiding areroutinely present in populations of chimpan-zees.

The known kills at Gombe occurred be-tween 1973 and 1977, during a period ofintense hostility between two communitiesthat had recently split from a single commu-nity, and which were each dominated by twoalpha-males with mutually hostile relations(Goodall, 1986). Border patrols by Kasekelamales were directed mostly toward the Ka-hama community during this period, whichended with the extinction of the Kahamacommunity (Goodall, 1986). Thus, unusualdemographic and social conditions appliedto elicit this particular bout of lethal raiding.

Therefore, nonlethal raiding is a routinecomponent of the chimpanzee behavioralrepertoire. Coalitionary killing is less com-mon. However it has been recorded in fourout of five sites. This raises the question ofwhy chimpanzees have an appetite for ago-nistic interactions with members of neighbor-ing communities, and why they sometimeskill opponents.

THE IMBALANCE-OF-POWERHYPOTHESIS

Explaining chimpanzee violence

Many reasons have been advanced to ac-count for chimpanzee lethal raiding, includ-ing: male–male bonds, hostility toward out-siders, cooperative group living, cooperativehunting skills, power imbalances when par-ties from neighboring communities meet,large and overlapping home ranges, highcognitive ability, and innate killing potential(reviewed by van der Dennen, 1995). Theonly attempt at a cost-benefit analysis thatexplains the species distribution of lethalraiding, however, is the imbalance-of-powerhypothesis. This hypothesis was implied byGoodall (1986), then elaborated by Mansonand Wrangham (1991), Wrangham and Pe-terson (1996), and Wrangham (1999b). Theimbalance-of-power hypothesis proposes thatthe function of unprovoked intercommunityaggression (i.e., deep incursions and coali-tionary attacks) is intercommunity domi-nance. By wounding or killing members ofthe neighboring community, males from onecommunity increase their relative domi-nance over the neighbors. According to theimbalance-of-power hypothesis, the proxi-mate benefit is an increased probability ofwinning intercommunity dominance con-

11EVOLUTION OF COALITIONARY KILLINGWrangham]

tests (nonlethal battles); this tends to lead toincreased fitness of the killers through im-proved access to resources such as food,females, or safety.

The imbalance-of-power hypothesis con-trasts with proposals that chimpanzees areexceptionally capable of conducting attacks,or win particularly large rewards from inter-group competition.

The significance of power imbalances

Both within and between primate groups,contests tend to be won by the larger of twocoalitions, though variables such as domi-nance rank and geographic location are alsoimportant (Cheney, 1986; Chapais, 1995).Coalition size appears even more importantfor interactions among chimpanzees fromdifferent communities. In the four longeststudies of chimpanzees, the principal deter-minant of the nature of intercommunityinteractions is not the geographic locationbut the relative size and composition ofparties when they encounter each other.This conclusion is based on direct observa-tions at Gombe, Kibale, Mahale, and Taı(Bygott, 1979; Nishida, 1979; Goodall, 1986;Boesch and Boesch, 1999; Wrangham et al.,in preparation), as well as playback experi-ments at Kibale (M Wilson et al., personalcommunication). For example, Boesch andBoesch (1999) found that small parties ofmales (1–3) mainly checked for the presenceof strangers by drumming and listening tothe response (67% of 18 occasions). Middle-sized parties (4–6 males) tended to makeincursions into the neighboring territorymore often (37% of 76 observations). Largeparties (7–9 males) tended to attack thestrangers (63% of 30 observations). Moregenerally, at all sites, the probability that aparty will advance, exchange displays, orretreat appears to be well predicted bywhether it is larger than, equal to, or smallerthan the opposing party (Boesch and Boe-sch, 1999). Relative party size is also acritical variable among lions (McComb et al.,1994; Grinnell et al., 1995).

The evidence therefore suggests that chim-panzee parties are bolder when they containrelatively more males. In addition, playbackexperiments at Kibale support the hypoth-esis that males are more likely to attack

when a party of three or more males encoun-ters a lone victim (M Wilson et al., personalcommunication), supporting the observa-tional data from Gombe. These observationsmake sense because to date, there appear tobe no records of any aggressors receivingserious wounds.

In light of such data, several authors haveproposed that it is the ability of a gang ofthree or more males to overwhelm a lonevictim, at low risk of injury to themselves,which at least partly explains why chimpan-zees are so ready to attack (Goodall, 1986;Manson and Wrangham, 1991; Wranghamand Peterson, 1996; Boesch and Boesch,1999). The logic is that a victim can be helddown or otherwise disabled by two or more,while another aggressor can impose damageat will. This idea that the low cost of lethalaggression elicits lethal raiding is central tothe imbalance-of-power hypothesis (Mansonand Wrangham, 1991; Wrangham and Peter-son, 1996).

Origins of power imbalances

All chimpanzee populations have fission-fusion grouping patterns, with individualssometimes alone and sometimes in parties(Fig. 2), and adult males more gregariousthan mothers (Wrangham, 1999a). Demo-graphic, social and ecological variables influ-ence party size (Boesch, 1996). For example,party size increases both with the number offemales having sexual swellings, and withthe amount of fruit in the habitat (Nishida,1979; Goodall, 1986; Boesch, 1996; Wrang-ham, 1999a). Parties appear to be con-strained by fruit availability as a result ofscramble competition, with larger partiesformed more when fruit is sufficiently abun-dant to allow gregariousness (Chapman etal., 1995).

Neighboring communities can experiencemarkedly different levels of fruit supply, aresult of differences in fruit-tree density, orin fruiting success (Chapman et al., 1997).This means that, on occasion, neighboringcommunities may contain parties of differ-ent mean size. The community in whichparties are able to be larger can then makelow-risk raids to attack neighbors. This socio-ecological connection has been observed inMahale, when the M-group community

12 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

would make seasonal forays into the terri-tory of the K-group community, supplantingK-group parties and sometimes attackingthem (Nishida, 1979).

Why are chimpanzees (compared to otherspecies) particularly vulnerable to the pres-sures of scramble competition that lead to afission-fusion, rather than a stable-troopsystem of grouping? Wrangham et al. (1996)argued that the important characteristic ofchimpanzees is that even when fruits arescarce, individuals continue to search forthem. Consistent with this hypothesis, chim-panzees (unlike gorillas) are restricted toareas that contain year round fruits, andspend significantly greater proportion oftheir feeding time eating ripe fruits than dosympatric frugivorous monkeys (Wranghamet al., 1998). This strategy of constant fruitsearch is presumably forced on chimpanzeesby species-specific digestive adaptations,such as the rate of food passage through thegut and the ability to ferment long-chaincarbohydrates (Milton, 1987). Whatever itsorigins, it exposes them to relatively intensescramble competition.

The imbalance-of-power hypothesis statesthat violence is facilitated by vulnerability,because lone individuals can be vigorously

attacked by a coalition at low risk to theaggressors. This means that populations (orseasons) with fewer encounters between soli-taries and large groups should have fewerviolent interactions. Data on wolves at De-nali (Alaska) offer a test of this prediction.The ratio of the number of parties contain-ing three or more wolves to the number ofsolitary individuals was higher during win-ter (5.2) than summer (0.1). This means thatsolitaries were much more likely to encoun-ter a large party during winter than sum-mer. As expected, winter was also the seasonwhen intraspecific kills were more likely (asevenfold increase in probability, from 22dated kills) (data calculated from Figs 5.4and 5.7, Mech et al., 1998).

If a similar effect applies to chimpanzeepopulations, and if the fact that Taı has hada low kill rate is meaningful (rather thanstochastic), Taı should have larger, less fis-sioned parties than at Gombe, Mahale, orKibale. Preliminary data suggest this predic-tion is qualitatively correct, because Boesch(1996) found that among six chimpanzeepopulations, the mean party size was high-est at Taı (8.3, compared to a mean of5.2 6 0.8 at the other five sites, includingGombe, Mahale, and Kibale). The percent-age of lone individuals was also lowest in Taı(4%, compared to 14% 6 3% for the threepopulations with data, Boesch, 1996, Table8.2). These data thus indicate consistentlylarger parties in Taı than elsewhere, compat-ible with evidence that Taı is a relativelyproductive habitat (Boesch and Boesch,1999).

Further data will test whether party sizeand the frequency of high-intensity aggres-sion do indeed co-vary among sites, and howoften chimpanzee populations tend to havesmall parties and high rates of aggression.Two points suggest that the high frequencyof intense aggression seen at Gombe andsuggested by the other eastern chimpanzeepopulations may be unusual for the speciesas a whole. First, skeletal trauma indicativeof intraspecific aggression has been found athigher rates in a sample of chimpanzeecrania from Gombe than from elsewhere(Jurmain, 1997). Second, the four P. t. sch-weinfurthii study sites (Gombe, Mahale,Kibale, and Budongo) are all located at the

Fig. 2. Party size distribution among chimpanzeesand bonobos. Data sources: eastern chimpanzee P. t.schweinfurthii: Kibale, Kanyawara community 1994–1996. Western chimpanzee P. t. verus: Taı (Boesch,1996); Mt. Assirir (Tutin et al., 1983). Bonobo P. panis-cus; Wamba (Kuroda, 1979); Lomako (White, 1988). Allpopulations show substantial variation in party sizeover time and between communities. Comparable datafor Gombe and Mahale were not found.

13EVOLUTION OF COALITIONARY KILLINGWrangham]

extreme east of the species geographicalrange, where dry seasons are relatively long.These eastern populations may therefore beliving under relatively harsh conditions offood availability compared to more westernsites.

In summary, chimpanzees are vulnerableto particularly intense scramble competi-tion, apparently because of their digestiveadaptations to ripe fruit. This competitionforces them to travel alone or in smallparties when fruits are scarce. Patchy fruitdistribution can mean that one communityhas abundant supplies, while its neighborshave few. Demographic differences betweencommunities (i.e., differences in the numberof adult males) may also mean that partiesin one community can be dominant overthose in the neighboring territory. Such fac-tors can account for differences betweenpopulations or communities in the numberof males in parties, and hence for differencesin their vulnerability to attack by coalitionsof neighbors.

Group territoriality and the benefitsof lethal raiding

Understanding the selective advantage ofaggression is more complicated for intercom-munity than interindividual relationships,because any fitness benefits gained by a risein intercommunity dominance are sharedamong individuals within the community.This might be expected to favor free-riders(individuals who would benefit from lethalraiding without taking part), which wouldlead to a suboptimal level of collective action(van Schaik, 1996; Nunn, 1999). In fact,however, there is no evidence of defectionamong raiding chimpanzees (Goodall, 1986;Wilson et al., personal communication) [(or,for that matter, among lions in similar inter-community contexts (Grinnell et al., 1985)].

How lethal raiding escapes the free-riderproblem is not understood. One possibility isthat intercommunity conflict has been sointense that selection has occurred at thebetween-group level (Boehm, 1999). How-ever, this is unlikely because it would re-quire very frequent group extinctions withfew survivors. Another is that free-riders arepoliced by others in the community (Boehm,1999), but no evidence suggests that this

occurs in chimpanzees. Others are (1) thatthe benefits of raiding rise so steeply withincreased party size that it pays individualsto participate for selfish reasons or (2) thatchimpanzees have evolved exceptional coop-erative abilities in contexts other than le-thal raiding (C. van Schaik, personal commu-nication). This remains an importantproblem, as it does for much of humanbehavior (Boehm, 1999).

On the other hand, the occurrence ofterritoriality among chimpanzee communi-ties, and of occasional imbalances of powerbetween parties from neighboring communi-ties, are easily explained. First, currenttheory suggests that home ranges are eco-nomically defensible if individuals can eas-ily cross their home range in a day. Chimpan-zees can do so (Lowen and Dunbar, 1994; seealso van Schaik, 1996). Second, a system offission-fusion grouping can explain why loneindividuals occasionally encounter largercoalitions, and are therefore vulnerable toattack.

But neither long day ranges nor fission-fusion grouping can explain lethal raiding,in which individuals seek opportunities toattack (as opposed to responding to inva-sion, escalated contest for resources, etc.).As Goodall noted, for example, there havebeen three major invasions at Gombe andMahale. ‘‘Kasekela males took over Kahamarange, Kalande males pushed deep intoKasakela range, and M-group moved intoK-group range. During all these invasionsadult males (and some females) were killedor disappeared. Even if it is argued that theKasekela males were merely trying to re-claim an area to which they previously hadfree access, the assertion does not explainthe northward thrust of the Kalande commu-nity or the takeover by the M-group atMahale’’ (Goodall, 1986, p. 528).

Two kinds of hypothesis have been pre-sented to account for such incursions, proxi-mate competition and dominance drive.First, aggression may be proximately elic-ited by resource competition, such as formates, food, or land (Manson and Wrang-ham, 1991). This hypothesis is strongly sup-ported by some observations, such as theincursions by Mahale’s M-group into theK-group range. These occurred in a seasonal

14 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

rhythm coincident with fruit shortages inM-group’s range and abundant fruits inK-group’s range (Nishida, 1979). Again, incertain circumstances raiding might helpmales to recruit young females: this possibil-ity is suggested by evidence at Gombe thatsevere attacks on the mothers of nulliparousfemales in neighboring communities aresometimes followed by the young femalesjoining the aggressor’s community (Goodall,1986).

Alternatively, aggression may be elicitedmerely by the opportunity to reduce thecoalitionary power of the neighbors (Mansonand Wrangham, 1991; Wrangham and Peter-son, 1996). According to this ‘‘dominancedrive’’ hypothesis, no resources need be inshort supply at the time of the raid. Instead,unprovoked aggression is favored by theopportunity to attack ‘‘economically,’’ that is,at low personal risk. If raiding leads to thewounding or death of a neighboring male,the neighboring community’s competitiveability is substantially reduced. For ex-ample, if the neighboring community has 10males, its fighting power is reduced by 10%.This reduction lasts for a considerable time,because the system of male philopatry meansthat a dead male can be replaced only viabirths within the community, which is a slowprocess. The aggressors’ probability of win-ning future intercommunity contests(battles, not raids) will therefore be signifi-cantly increased by killing a neighboringmale. The increase in relative fighting powercan be expected to enable a community toenlarge its territory, as suggested by prelimi-nary evidence of a correlation between thenumber of males and territory size at Gombe(Stanford, 1998b). Over the long term, there-fore, if fitness is correlated with territorysize, successful raiding is expected to in-crease the raiders’ fitness.

This ‘‘between-community dominancedrive’’ hypothesis for explaining aggressionbetween groups is therefore similar to the‘‘within-community dominance drive’’ hy-pothesis, which contributes to explainingpatterns of aggression between individualswithin dominance hierarchies (Popp andDeVore, 1979; Chapais, 1995). Within chim-panzee communities, for example, male ag-gression occurs predictably over status;

weakness of a dominant due to aging, wound-ing, loss of allies or loss of confidence in-creases the rate of attack by a subordinate(Bygott, 1979; de Waal, 1982; Goodall, 1986;Nishida, 1994). Selection is expected to fa-vor the effort to rise in dominance becausedominant individuals (or groups) tend tohave high fitness, and accordingly, individu-als opportunistically take advantage of anyperception of changes in power asymmetry(Chapais, 1995).

Sometimes, admittedly, the expectationthat higher dominance leads to higher fit-ness is not met. Thus, in around half of thestudies between dominance and reproduc-tive success within primate groups, therewas no relationship. However, in the otherhalf, dominants had higher fitness thansubordinates (Harcourt, 1987; de Ruiter andvan Hooff, 1993; Ellis, 1995). This meansthat even though increased dominance doesnot always lead to higher fitness, it pays onaverage.

Therefore, according to the dominance-drive hypothesis, a necessary and sufficientcondition for intercommunity aggression isa perception that an opponent is sufficientlyvulnerable to warrant the aggressor(s) at-tacking at low risk to themselves.

The dominance drive hypothesis appearsuseful for explaining why carnivores sharelethal coalitionary violence with chimpan-zees. As expected by both the proximatecompetition and dominance drive hypoth-eses, fission-fusion grouping and intergrouphostility occur in these species (Table 6). Theproximate competition hypothesis also pre-dicts, however, that the type of food supplies,mating system and/or coalitionary bondshould be similar in allowing benefits to begained from raiding or killing neighbors.However, the four species of carnivores inwhich lethal coalitionary violence has beenrecorded show various combinations of mat-ing systems and coalitionary bonds, all differ-ent from those found in chimpanzees (Table6). For example, intergroup transfer is insome species primarily by males, in othersprimarily by females. Therefore, lethal vio-lence cannot be uniformly explained as re-sulting from competition over females. Be-cause the carnivore species in Table 6 varyin the type of benefits to be gained by

15EVOLUTION OF COALITIONARY KILLINGWrangham]

intergroup dominance, the dominance drivehypothesis explains similarities in their ten-dency to use lethal violence more easily thanthe proximate competition hypothesis.

The proximate competition and domi-nance drive hypotheses are closely related,because in both cases, the ultimate benefitsof dominance are increased success in re-source competition. The proximate competi-tion hypothesis is favored if raiding is elic-ited by the presence of stealable resources,or if benefits accrue immediately after araid. On the other hand, unprovoked deepincursions and attacks on males withoutany obvious reward are better explained bythe dominance drive hypothesis. To differen-tiate these hypotheses more clearly, data areneeded on the proximate stimuli that elicitaggression. Since current information sug-gests that chimpanzee raids are often initi-ated without the raiders perceiving mates orfood sources, the dominance drive hypoth-esis appears relevant to explaining the tim-ing and direction of lethal raiding.

Sex differences in territorialityand aggressiveness

Among chimpanzees, males have to datebeen the only observed killers and aggres-sors in intergroup interactions, and malesare also more likely than females to bevictims (Table 4). Among humans, warriorsare also overwhelming male (Adams, 1983).This contrasts with spotted hyenas, wherefemales are more aggressive than males(Kruuk, 1972; Frank, 1986; East and Hofer,1991; Henschel and Skinner, 1991); andwith wolves, where both sexes are killed at

high rates, and there is no evidence of a sexdifference in aggressiveness (Mech et al.,1998). Why, therefore, are males the princi-pal perpetrators of aggression in chimpan-zees and humans?

Traditional explanations are that malesare more expendable, or that males havemore to gain simply because they have highervariance in fitness than females do (re-viewed by van der Dennen, 1995). However,such general explanations do not account forspecies variation in the intensity of femaleparticipation. Nor does the degree of sexualdimorphism in body size, because amongnonhuman primates, sexual dimorphism inbody size is not correlated with female in-volvement in intergroup aggression (Man-son and Wrangham, 1991).

Male bonding, which is especially pro-nounced among chimpanzees and humans,has often been proposed to be an importantinfluence (reviewed by van der Dennen,1995). This idea is supported by the fact thatboth in humans and nonhuman primates,populations with more patrilocal residence(or male philopatry) have relatively greatertendency for aggressors to be male (Adams,1983; Manson and Wrangham, 1991).

The ultimate origins of male bonding arestill debated. In chimpanzees, males aremore gregarious than mothers, possibly be-cause, as a result of carrying and waiting forinfants, mothers travel slowly (Wrangham,1999a). The relative mobility and gregarious-ness of males means that they can use alliesto dominate access to their home ranges,excluding other males and thereby forcingmale philopatry.As a result, a system evolves

TABLE 6. Intergroup aggression in fission-fusion species with group territoriality1

Chimpanzee Human Wolf LionSpottedhyena Cheetah

Battles Y Y Y Y Y ?Kill adults Y Y Y Y Y YLethal raid Y Y Y Y? Y? ?Food supplies Dispersed Variable Clumped Clumped Clumped Dispersed/

clumpedCoalitionary

bonds amongMales Males Pair 1 Helpers Females; Males Females Males

1 ‘‘Kill adults’’ is shown separately from ‘‘Lethal raid’’ because, in hyenas and lions, it is not clear if killing of neighbors occurs withlethal raids, or merely when invaders are discovered and killed by residents. ‘‘Dispersed’’ food supplies imply that individual food-patches are not defensible, whereas ‘‘clumped’’ foods can be individually defended (e.g., carcasses). Coalitionary bonds are bonds inwhich adults support each other in aggression against others. Sources for mating system and coalitionary bonds are Kruuk (1972)(spotted hyenas), Grinnell et al. (1995), and McComb et al. (1994) for lions, and Mech et al. (1998) for wolves. For other sources, seetext.

16 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

in which it pays to eradicate males fromneighboring communities (Wrangham,1999a,b).

According to this logic, therefore, malebonding has two effects. First, it contributesto the development of male philopatry andthe benefits of excluding nongroup males,thereby raising the stakes in territorial en-counters. Second, it makes available alliesthat enable a larger party to dominate asmaller party.

None of this means, however, that maleparticipation is a necessary condition for theevolution of lethal violence in territorialinteractions. As wolves, spotted hyenas andants show, coalitionary territoriality can becarried out by both sexes, or even primarilyby females. The comparative evidence, there-fore, suggests that lethal raiding in chimpan-zees and humans cannot be attributed to thefact that bonds among adults are primarilyamong males.

Bonobos: exceptionsthat support the rule?

Intercommunity relations among bonobossometimes involve fights between large par-ties, but as a species they appear to besubstantially less hostile to each other thanare chimpanzees (reviewed by Wranghamand Peterson, 1996; Stanford, 1998a). First,bonobos have never been seen to engage inlethal raiding, nor indeed in any compo-nents of such behavior (Table 2). Second,they can include markedly peaceful interac-tions, in which individuals from neighboringcommunities rest, travel, copulate, play, andgroom together (Idani, 1991; White, 1996).In contrast, peaceful interactions involvingmales of neighboring communities have notbeen seen among chimpanzees.

In view of the anatomical and phyloge-netic similarities between chimpanzees andbonobos these differences are remarkable.Three kinds of explanation suggest them-selves.

First, the facts may be misleading; moreprolonged observation may reveal lethalraiding in bonobos (Stanford, 1998a). How-ever, this eventuality appears unlikely be-cause at Gombe, Mahale, Taı, and Kibalecomponents of lethal raiding were seenshortly after individuals were observed near

the community range borders. Among bono-bos, by contrast, interactions are seen atrange borders without the components oflethal raiding.

Second, relatively peaceful intercommu-nity relationships in bonobos may be anincidental result of a reduction in the level ofwithin-community violence compared tochimpanzees. For example, fewer violentbehavioral interactions of all kinds occuramong bonobos: no sexual coercion, no infan-ticide, no brutal fights among males or fe-males competing for dominance, no malebeatings of females (Wrangham and Peter-son, 1996; Furuichi et al., 1998). The same istrue in captivity (de Waal and Lanting,1997; Stanford, 1998a). Collateral evidencecomes from a survey of cranial and postcra-nial skeletal trauma that concluded that inchimpanzees (and gorillas), but not in bono-bos, there was evidence of serious risk frominterindividual aggression (Jurmain, 1997)and from the generally less robust and lesssexually dimorphic morphology of bonobosthan chimpanzees (Zihlman and Cramer,1978; Shea, 1984). Pending further data,bonobo males consistently appear to be lessviolent than chimpanzees.

Accordingly, a possible hypothesis is thatselection may have favored a generally lessaggressive male temperament in bonobos,as opposed to a loss of motivation specificallyfor lethal raiding. The reasons why malebonobos are generally less aggressive thanmale chimpanzees could derive from thedominance of males by powerful female–female coalitions, or the greater importanceof mothers than other males as allies forindividual males, or other social dynamicsoccurring within communities (Kano, 1992;Parish, 1996; Wrangham and Peterson, 1996;de Waal and Lanting, 1997). The importantpoint is that the reduced tendency for lethalraiding would be viewed as an incidentalconsequence of a more general reduction inmale aggression. This hypothesis is chal-lenged, however, by species that have aggres-sive intergroup interactions despite havingpeaceful relationships within groups (e.g.,female lions). Furthermore, the tendency toengage in lethal raiding seems unlikely to beselectively neutral, considering its poten-tially large effects on dominance relation-

17EVOLUTION OF COALITIONARY KILLINGWrangham]

ships between groups and the time andeffort spent on raids. I therefore concludethat the low tendency for lethal raiding inbonobos is not merely an incidental conse-quence of the benefits of within-communitypeacefulness.

The third kind of explanation is thatamong bonobos, important components oflethal raiding has been specifically selectedagainst. Under what circumstances couldthis occur? According to the imbalance-of-power hypothesis, lethal raiding is favoredby a combination of coalitionary territorial-ity and imbalances of power sufficient toallow one party to kill victims of the rivalcommunity with impunity. Since lethal raid-ing is absent even though coalitionary terri-toriality occurs among bonobos, the imbal-ance-of-power hypothesis predicts thatcompared to chimpanzees, bonobos mustexperience greatly reduced power imbal-ances between rival parties.

In general, variance in bonobo party sizeis less than among chimpanzees, even whenaverage party size is similar (Chapman etal., 1994). This is as expected from theimbalance-of-power hypothesis. However,the more critical question is how often bono-bos are forced to travel alone, because indi-viduals can be killed (at minimal cost to theaggressors) only when they are found aloneby a rival party. The two principal bonobostudy sites both indicate that lone travel israrely forced by ecological pressures. InWamba, bonobos usually range as one or twolarge mixed parties averaging more than tenindividuals (Hashimoto et al., 1998). In Lo-mako, where feeding competition appearsmore intense than in Wamba, females re-main in multi-female parties even duringthe seasons when fruit is least available(White, 1998). Such observations suggestthat compared to chimpanzees, the intensityof feeding competition is substantially re-duced among bonobos (Chapman et al., 1994;Wrangham et al., 1996; White, 1998). As aresult, extreme imbalances of power appearunlikely to occur between parties meetingfrom neighboring communities.

Whether the apparent difference in theintensity of feeding competition betweenchimpanzees and bonobos is sufficient tosatisfy the imbalance-of-power hypothesis,

however, remains to be proven. In Lomako,male bonobos spend increased time aloneduring periods of fruit scarcity (White, 1998).White (1998) suggests that these maleschoose to travel alone in order to track theincreasingly dispersed female parties, butaccording to the imbalance-of-power hypoth-esis, it should be dangerous for them to doso. However, if solitary travel is a socialoption rather than a strategy dictated byecological pressures, it may be possible forbonobos to restrict their solitary periods totimes and locations when they can assessthat they are safe. More quantitative datawill be needed to test such ideas.

One of the only other species of primate inwhich lethal raiding might be expected fromthe imbalance-of-power hypothesis are spi-der monkeys Ateles spp., because spidermonkeys have a fission-fusion grouping sys-tem like chimpanzees: individuals some-times travel alone and sometimes in largerparties. Furthermore, males are more gre-garious than mothers, and they form coali-tionary bonds with each other against neigh-boring groups (Chapman et al., 1995). Onthe other hand, spider monkeys are whollyarboreal, which may reduce their ability touse coalitionary aggression. Among baboonsPapio anubis, for example, coalitions of low-ranking males are effective in defeating asingle higher-ranking male on the ground,but not in trees (Smuts, 1986). Further dataon the effect of arboreality on power asymme-tries between coalitions and solitaries istherefore desirable.

The imbalance-of-power hypothesisand the evolution of human warfare

Peace is the normal human condition, inthe sense that most human groups, for mostof the time, are not at war (Ferguson, 1989;Sponsel, 1996). Nevertheless, ethnographicand historical records clearly show that war-fare is a frequent practice (Keeley, 1996;Manson and Wrangham, 1991; van der Den-nen, 1995). Increasingly, archaeological datasuggest that violence has often been a statis-tically important source of death, and it issometimes possible to infer that the violencewas coalitionary (Keeley, 1996; Larsen,1997). In small-scale societies, the common-est form of war interaction is a raid (e.g.,

18 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

Turney-High, 1949; Keeley, 1996; Maschnerand Reedy-Maschner, 1998). Even if hu-mans are routinely peaceful, therefore, warneeds to be explained. Because warfare in-cludes a variety of types of interaction (suchas raids and battles), it will require multipleexplanations.

The myriad hypotheses proposed to ex-plain why humans practice raids and otherforms of warfare fall into three generalclasses. Maladaptive hypotheses suggestthat warfare results from an originally adap-tive aggressive tendency that, as a result ofsubsequent developments such as the inven-tion of weapons, became disadvantageouseven to the winners (e.g., Lorenz, 1966).They suffer from the theoretical problemthat if warring tendency is indeed disadvan-tageous, it has such large effects that itshould be selected against rapidly. However,it seems unlikely that winners fare badly.

Neutral hypotheses are currently morepopular, though they also suffer from theselective-disadvantage problem. They sug-gest that warfare should be regarded asderiving merely from a capacity, or poten-tial, resulting from our cognitive creativity.Neutral hypotheses consider warfare to beelicited by environmental and social stimulithat have no evolutionary significance (e.g.,Bock, 1980; Keeley, 1996; Gould, 1996). Theyare often based on the (erroneous) premisethat behaviors that vary among populationscannot be explained in terms of naturalselection without assuming genetic differ-ences between populations (see Discussion).

The imbalance-of-power hypothesis exem-plifies a third kind of hypothesis that viewswarfare as adaptive and rooted in geneticpredispositions. It suggests that raiding de-rives from the advantages of gaining inter-group dominance and an ability to assesspower imbalances in an environment of inter-group hostility and power imbalances be-tween parties from neighboring communi-ties.As in chimpanzees, it raises the questionof why territories are defended by malesrather than females. By analogy with theargument for chimpanzees, male rather thanfemale territoriality derives from the highcost of travel experienced by mothers(Wrangham, 1999a). This cost reduces moth-

ers’ ability both to defend a range and toform alliances.

Both intergroup hostility and a fission-fusion grouping system are universal incontemporary human populations, whethertribal or nation-state (Rodseth et al., 1991).Whether these features were characteristicof humans in prehistory is unknown. How-ever, with the exception of brain size, humanmorphology has changed relatively little dur-ing the last 1.9 million years (Wolpoff, 1998),suggesting that the essential ecology of hu-man prehistory may been rather stable priorto agriculture. The essence of theories aboutfission-fusion grouping in chimpanzees isthat fission is a response to high costs ofscramble competition (Chapman et al., 1995);scramble competition is expected to be moreintense in species that depend on rare, high-quality foods (Janson and Goldsmith, 1995),and humans appear adapted to high-qualityfoods (Milton, 1987; Leonard and Robertson,1997). Following this line of argument, fis-sion-fusion grouping is expected to havebeen characteristic of human evolutionaryhistory.

Based on the ubiquity of xenophobia andingroup-outgroup bias in contemporarypopulations, intergroup hostility is normallyassumed to have been routine in humanprehistory. The likelihood of intergroup hos-tility in prehistory is supported also by itsprevalence among nonhuman primates (Che-ney, 1986). The form of hostility can beinferred as being territorial, because amongprimates, territories tend to be found inspecies with long day ranges in relation tohome ranges, and are predicted to occurmore easily where groups are split intomultiple parties (Mitani and Rodman, 1979;Lowen and Dunbar, 1994). The long dayranges of contemporary forager men [e.g., 9km (Bailey, 1991)] and the probability offission-fusion foraging suggest that territori-ality would have been possible where homeranges were not immense.

The imbalance-of-power hypothesis is thuscompatible with conventional views of hu-man prehistory. It can in theory be chal-lenged by evidence that recent prehistoricancestors foraged in stable parties, or hadways of reducing power imbalances betweenrival parties, or had little intergroup hostil-

19EVOLUTION OF COALITIONARY KILLINGWrangham]

ity, though such evidence would in practicebe difficult to obtain.

On cladistic grounds, various authors havehypothesized that lethal raiding in humansand chimpanzees shared a common originaround 5–6 mya, and has been presentcontinuously in the subsequent evolution ofeach species (Wrangham, 1987; Ghiglieri,1988; Wrangham and Peterson, 1996; Otter-bein, 1997). This hypothesis is currentlyuntestable. A key issue for human ancestryis whether australopithecine ancestors ofhumans foraged in temporary parties (i.e.,with fission-fusion) or in stable groups. An-swers to such questions are needed beforewe can be confident whether lethal raidingin chimpanzees and humans represents asynapomorphy or a homoplasy.

CHALLENGES TO THEIMBALANCE-OF-POWER HYPOTHESIS

Uncertainty in the chimpanzee data

The evidence of lethal raiding in chimpan-zees comes from few cases and a smallnumber of populations, some of which haveexperienced significant anthropogenic influ-ences. This has led to doubts about theimportance of lethal raiding as a speciestrait among chimpanzees (Power, 1991; Suss-man, 1997).

For example, Power (1991) accepts thatthe descriptions of chimpanzees in Gombe,Mahale, and Kibale as violent and status-striving are accurate, but regards the behav-iors as nonadaptive consequences of exces-sive ecological stress. A central concern forPower (1991) is why there was a shift in theperception of chimpanzee society from peace-ful to violent, beginning in the 1970s basedon observations after the first 5 years ofGoodall’s study (1960–1965). She argues thatthe 1960s view of chimpanzees living in apeaceful society was the ‘‘correct’’ one, andthat subsequent observations of violencereflect a social environment stressed by vari-ous kinds of human-induced disturbance. Insupport, she cites the fact that intense ag-gression was seen rarely in early chimpan-zee studies, especially at Gombe, Budongo,and Kibale (Ngogo community). She notesthat in studies where violence has beenreported, either humans have provided foodfor the chimpanzees (Gombe, Mahale), or

the forest has been disturbed by encroach-ment or logging (Kibale, Ngogo, andKanyawara).

However, the relative lack of observationsof violence in the early years of chimpanzeestudies cited by Power (1991) is easily under-stood without reference to the effects ofdisturbance. Where there was no provision-ing, early observations were relatively fewcompared to later years, and they weremostly of poorly habituated individuals, moreconcerned about humans than each other[e.g., contrast early observations by Ghi-glieri (1984), Kibale, Ngogo community, Isa-birye-Basuta (1989), Kibale, Kanyawaracommunity, Reynolds and Reynolds (1965),Budongo, and Sugiyama (1973), Budongo]with those based on well-habituated indi-viduals by Watts (1999) (Kibale, Ngogo com-munity), Wrangham et al. (1992) (Kibale,Kanyawara community), and Newton-Fisher(1997) (Budongo). With increased observa-tion of habituated individuals, studies of theKanyawara, Ngogo, and Budongo communi-ties conform to the essential Gombe-Mahalemodel of dominance-motivated and strategi-cally violent males. Power (1991) appearsnot to have appreciated the difficulty ofobserving dominance behavior and violenceamong poorly habituated and little knownindividuals.

Provisioning complicates the issue be-cause it confounds increased observabilitywith a concentrated food resource that isliable to promote aggression. In the case ofGombe, Goodall’s introduction of bananafeeding in 1962 led to the chimpanzeesbecoming habituated quickly, so individualswere watched at close quarters during thefirst decade. Once the chimpanzees werehabituated, they were observed almost en-tirely in the banana-feeding area (ca. 50 3 50m, approximately 1/5,000th of their terri-tory of 12 sq km or more), and not followedtoward territorial boundaries. Much aggres-sion during intense banana-feeding years of1965–1969 was clearly directed toward ob-taining bananas (Wrangham, 1974). During1969, banana feeding was reduced, and feweraggressive incidents occurred in the banana-feeding area (Wrangham, 1974).

Power (1991) argued that a particularlyimportant feature of the banana-feeding sys-

20 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

tem was that, after 1965, chimpanzees werefrustrated by their lack of control. She char-acterized the observation period 1960–1965as ‘‘naturalistic’’ (implying undisturbed byhumans) because the bananas were givenfreely, i.e., they were never withheld. From1965 onward, in attempts to reduce thebanana-induced aggression of the 1962–1965 era, bananas were made available inmetal boxes equipped with doors that werecontrolled in various ways by observers. Itwas this system that Power argued causedchimpanzees to express their potential foraggression, because it frustrated them.

In the absence of controlled experiments,no hypothesis can be rejected. Two pointsrelevant to the imbalance-of-power hypoth-esis can be made, however. First, even if thefrustration-aggression hypothesis is correctin explaining why the Kasekela communityattacked the Kahama community, it doesnot explain why coalitional lethal aggres-sion was elicited relatively easily in thesechimpanzees, or why it occurs in males,whereas it has not been seen in any otherspecies faced with similarly frustrating con-texts. [For example, baboons obtained ba-nanas regularly at the banana-feeding-area,though observers tried to prevent them(Wrangham, 1974). There has been no hintof any behavior resembling lethal raiding inthese baboons, despite intense study in sub-sequent years]. Thus, as Power (1991) her-self says, whether or not feeding frustrationcontributed to the social tensions at Gombe,there remains a problem to be explained.Why are male chimpanzees easily promptedto adopt intense coalitionary violence as asolution to social problems?

Second, the frustration-aggression hypoth-esis has much against it. The idea is that‘‘frustration causes a distinct behavioralchange in the condition of an organism’’(Power, 1991, p 3). This implies that undernatural conditions, chimpanzees are notnaturally frustrated, which is clearly nottrue. For example, intense aggressive compe-tition occurs regularly in all study sites overprized foods, such as meat (Goodall, 1986;Boesch and Boesch, 1989). On the otherhand, whatever behavioral change occurredamong the post-1965 Kasekela chimpanzees(compared to chimpanzees at other sites) is

not easily viewed as ‘‘distinct.’’ The series ofattacks that began in January 1974 oc-curred almost a decade after the start of theproblematic banana-feeding-system, andtook place several kilometers from the feed-ing station. Power (1991) implies that thislong delay can be accommodated by thehypothesis of a permanent behavioral/psy-chological reorganization, but clearly thebehavior must also be viewed in the contextof ongoing social tensions (see Goodall, 1986for an account of the relationships betweenthe alpha-males of the Kasekela and Ka-hama communities that may have helpedprecipitate the aggression).

The incidence of aggression in the banana-feeding-area was closely related to the num-ber of bananas that chimpanzees obtained.Party size increased in the feeding areacompared to the natural habitat, but withinminutes of the chimpanzees leaving thefeeding area the expected party size wasrestored (Wrangham, 1994). These and simi-lar results show that there were indeedshort-term influences related to the availabil-ity of bananas, but no long-term influenceshave been detected (Wrangham, 1974; Good-all, 1986). The accumulation of data fromother, nonprovisioned sites continues to chal-lenge the view that chimpanzees are natu-rally averse to violence.

The claim that biology is irrelevantfor human warfare

Some critics reject evolutionary explana-tions of warfare out of hand, based on themisconception that the only behavioral pat-terns explicable by biology are ‘‘instincts,’’i.e., behaviors that are obligatory and/orinvariable. According to this logic, sincewarfare is not ‘‘instinctual,’’ biological adap-tations cannot explain the propensity forwar (Keeley, 1996; Sussman, 1997; Regal,1998). This error seems remarkable, be-cause behavioral ecologists have longstressed that psychological adaptations areexpected to respond in a contingent way toappropriate contexts (e.g., Hrdy, 1990;Barkow et al., 1995; Krebs and Davies,1997). In the words of Otterbein (1997,p 272), ‘‘Man is neither, by nature, peacefulnor warlike. Some conditions lead to war,others do not.’’

21EVOLUTION OF COALITIONARY KILLINGWrangham]

The imbalance-of-power hypothesis is en-tirely compatible with the observations thatmany people live wholly peaceful lives, thatsome cultures have periods of peace lastingfor several generations, and that some chim-panzee populations have no lethal raidingfor long periods. Indeed, such variation is tobe expected. The imbalance-of-power hypoth-esis conforms well to some theories of peace,such as the idea that nonviolence is anadaptive response by societies to violence bystronger neighbors (Dentan, 1992). Admit-tedly, males are expected by this hypothesisto take advantage of power over neighbors,especially when unfettered by social or cul-tural constraints. They are also expected toprobe for weaknesses in perceived oppo-nents, and to be willing to fight in a widevariety of circumstances where elevated sta-tus is predictable or perceived opponentswill be wounded or destroyed at low cost.But the essence of the imbalance-of-powerhypothesis, like other behavioral hypoth-eses for large-brained mammals, is thatexpression of the behavior depends on con-text. Whether or not an individual employsviolence is expected to depend on the proxi-mate stimuli, about which we still knowlittle. What leads individuals to classifyothers as ‘‘opponents’’? How do social andideological pressures affect the ease withwhich men, or women, respond to incite-ments to violence? How is ‘‘dominance’’ per-ceived? How are risks perceived, for in-stance, when Ego is embedded within ahierarchy of alliances? How do institutionalwar relationships influence individual neu-roendocrinology, and vice versa? Such ques-tions are critical for understanding whobecomes violent, and when.

In the current context, the imbalance-of-power hypothesis suggests that selectionhas favored certain emotional predisposi-tions in males that cause aggressive behav-ior to be elicited relatively easily undercertain circumstances. The challenge of de-fining the eliciting circumstances is the prov-ince of disciplines that probe local varia-tions, including not only biologicalanthropology, but also social psychology, be-havioral ecology, social anthropology, cul-tural ecology, or social ecology. Those disci-plines, together with biological anthropology,

are the places to seek answers about popula-tion variations.

The assumption that evolutionary analy-sis implies genetic determinism seems oddlyold-fashioned in an era when we are begin-ning to understand the nuances of psycho-neuroendocrinological adaptations of differ-ent species and sexes. To Keegan (1993, p 3),war reaches into ‘‘the most secret places ofthe human heart, places where self dissolvesrational purpose, where pride reigns, whereemotion is paramount, where instinct isking.’’ If psychology can describe those se-cret places, it is the task of evolutionaryanthropology to explain how they arose.

IMPLICATIONS OF THEIMBALANCE-OF-POWER HYPOTHESIS

Chimpanzee and human psychology

Different versions of the CVH pay variedattention to such factors as cognitive ability,weapons, brain size, male-bonding, territori-ality, sexual dimorphism, and imbalances ofpower (van der Dennen, 1995). They areunited, however, in providing adaptive ra-tionales for chimpanzee and human vio-lence, and therefore in proposing that lethalraiding has a substantial evolutionary his-tory, possibly since our split from a commonancestor with chimpanzees.

The implication is that there has beenselection for a male psyche that, in certaincircumstances, seeks opportunities to carryout low-cost attacks on unsuspecting neigh-bors. The psychological mechanisms thatwould make such a complex function pos-sible have not been studied, but a partial listmight include: the experience of a victorythrill, an enjoyment of the chase, a tendencyfor easy dehumanization [or ‘‘dechimpiza-tion,’’ (Goodall, 1986), i.e., treating nongroupmembers as equivalent to prey], and deindi-viduation (subordination of own goals to thegroup), ready coalition formation, and sophis-ticated assessment of power differentials.Sex differences can be expected in at leastsome of these traits, unless developmentalconstraints interfere. Some features of alethal-raiding psychology are not easily pre-dicted, such as the mechanisms by whichexpected costs and benefits are assessed.

A sharp alternative to the CVH is thestandard social science model (SSSM), that

22 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

human males have no inherent propensityto take advantage of power differentials.Instead, according to the SSSM, humansmerely have a capacity for violence, andsince ‘‘the range of possible cultural resultsis not explicable by natural selection’’ (Bock,1980; p. 76; cf. Gould, 1996), evolutionaryhistory is claimed to be irrelevant. This lineof thinking has several problems. It does notaccount for the species distribution of coali-tional aggression. It treats biology and cul-tures as alternatives, rather than as mutu-ally interacting influences. It does notaccount for the predictability of human ag-gressive patterns, and it is easily subsumedunder adaptive theories of violence, whichcan account both for the fact that individu-als choose to manipulate others (whetherthrough ideology or other ways) and for thefact that they are so easily manipulated.

The complexity of war

Even in the complex human world, someof the processes that regulate aggressionamong large groups are analogous to thosethat occur at the individual or face-to-facelevel (Hinde, 1993). Both at the large groupand the individual level, for example, per-sonal relations between leaders of opposinggroups can play an important role, with athreat to the interests or values of the actorbeing capable of instigating aggression. Pla-catory signals or actions (e.g., donations) areused to deter aggression. Aggressivenesscan be augmented by a greater asymmetryof power, or reduced by a probability ofpunishment. Hinde (1993) found more than20 such analogies, of varying significanceand distinctness, linking the behavioral in-teractions among individuals and largegroups. This suggests that, in some ways,the logic of aggressive interaction amongindividuals can be applied to large groups.

Nevertheless, so many cultural and lin-guistic novelties complicate warfare that theconnection to biology can appear tenuous atbest. Among factors such as the number ofmilitary and strategic options available, theability to discuss options and manipulateothers, the adoption of cultural goals, andthe unpredictable potential for shifting alli-ances, a particularly important trait distin-guishing humans from chimpanzees is that

human groups incorporate more levels ofsocial dynamics (Hinde, 1993). At each suchlevel, ‘‘level-specific properties’’ influence andare influenced by adjacent levels. For ex-ample, group processes and institutionalinfluences modify the motivational ideals ofindividuals. In modern nation-states, themilitary-industrial-scientific complex tendsto precipitate and maintain war, e.g., byproducing increasingly sophisticated weap-onry. War as an institution can in theory bemaintained by the inertia of subinstitutions,such as the belief that capitalism needsmilitarism for its continued growth (Hinde,1993). Dynamics like these mean that apropensity for lethal raiding cannot be trans-lated directly into an explanation of thecomplexities of human warfare.

Is a propensity for lethal raiding of thechimpanzee type at all relevant to humanwarfare? Among people living in small politi-cally independent groups, lethal raiding ap-pears strikingly similar to the patternsamong chimpanzees. In both cases, smallparties of males aim to make undetectedincursions into the ranges of neighbors, at-tack unsuspecting victims, and retreat with-out being drawn into a battle (Turney-High,1949; Chagnon, 1992; Keeley, 1996). Al-though the psychological processes remainundescribed, the imbalance-of-power hypoth-esis might suggest that selection has fa-vored various complex traits, such as atendency to classify others as in-group orout-group, to regard members of out-groupsas potential prey, to be alert to (or search for)power asymmetries between in-group andout-group parties, and to be ruthless inattacking out-group parties when the per-ceived power asymmetry is sufficiently great.A list of traits such as these can in theorydescribe an evolutionarily selected ‘‘propen-sity for lethal raiding.’’

Such traits appear obviously relevant tosome aspects of intergroup relations (e.g.,the planning and execution of military en-gagements). Among humans, the complexityof society means that individual propensi-ties sometimes have less direct impact onsocial outcomes than among chimpanzees.Nevertheless, the imbalance-of-power hy-pothesis may explain why culturally derivedinformation is used in certain ways. For

23EVOLUTION OF COALITIONARY KILLINGWrangham]

example, as a result of cultural beliefs orsocial pressure, individuals can eitherbroaden or contract their concept of wherean in-group/out-group boundary falls, or ofhow important it is. Idealogues can per-suade their followers that sufficient powerasymmetry exists to make attacks on anoutgroup worthwhile. Culture can thus ma-nipulate the information an individual usesto assess whether an attack is desirable.

The imbalance-of-power hypothesis canbe reconciled with the power of culture;therefore, if human males have a tendencyto search for, and take advantage of, powerasymmetries sufficient to enable them tosafely kill rivals, while social pressuresmodify the concept of ‘‘rival,’’ ‘‘ally,’’ and‘‘sufficient power asymmetry.’’ It accordinglysuggests an explanation for why humanmales become dangerous when they obtain,or believe they have, large power advan-tages over others. (Whether, in novel circum-stances, they use such power adaptively isan open question.) It also suggests the impor-tance of systems that reduce power asymme-try, such as intergroup alliances throughtrade, marriage or treaty.

When large power asymmetries do notoccur, relationships between groups are of-ten peaceful, as expected from the imbalance-of-power hypothesis (Knauft, 1991; Buenode Mesquita and Lalman, 1992; van derDennen, 1995). Even when there is a bal-ance of power, however, lethal battles andwars can occur among humans, in contrastto the pattern among chimpanzees (Singer,1989, Boehm, 1992). Coalitionary aggres-sion occurring between opponents with bal-anced power requires other kinds of explana-tion than the imbalance-of-power hypothesis,such as the cultural exaggeration of motivat-ing forces or the development of self-decep-tive assessment strategies (Boehm, 1992;van der Dennen, 1995; Wrangham, 1999b).

The relation between lethal raidingand hunting

Both lethal raiding and hunting are car-ried out primarily by adult males acting incoordinated groups: both involve otherwiseunusual actions such as searching for largeprey, stalking, chasing, seizing, wounding,and killing; both are more elaborated in

humans and chimpanzees than in otherprimates. Furthermore, the behaviors shownby chimpanzees toward mammalian preyare partly similar to those they show towardconspecific victims, including quiet stalkingduring a hunt, intense arousal during theattack phase (pilo-erection, intimidation dis-plays), and ambivalence toward the victim.In contrast, the behaviors shown by special-ized carnivores toward their prey are notlike those directed toward conspecifics. Forexample, social carnivores do not show signsof excitement when killing prey, and tend touse a killing bite (van der Dennen, 1995).

Such observations suggested to Eibl-Eibesfeldt (1975) and Goodall et al. (1979)that among chimpanzees similar motiva-tional factors may be involved in intraspe-cific killing and hunting. Eibl-Eibesfeldt(1975) specifically proposed that, ‘‘Motiva-tionally, hunting behavior in chimpanzeeshas probably been derived from intraspecificaggression’’ (translated and quoted by vander Dennen, 1995, p 192). The essential logicis that if hunting had arisen independently,it should be expected to show more similari-ties to the patterns displayed by social carni-vores. Van Hooff (1990) agreed, suggestingthat if selection favored the ability to huntand kill conspecifics, the psychologicalmechanisms that evolved would be easilyco-opted toward obtaining meat.

Note that these ideas are opposite to thekiller ape hypothesis. The killer ape hypoth-esis suggested that intraspecific violenceevolved from hunting, whereas Eibl-Eibes-feldt (1975), Goodall (1986), and van Hooff(1990) proposed that hunting evolved fromintraspecific violence. As van der Dennen(1995) notes, the relationship between intra-specific killing and hunting probably nowinvolves multiple directions. For example, insome human populations hunting may pro-vide practice for warfare (Otterbein, 1997).Disentangling these relationships will there-fore not be easy.

Nevertheless the idea that violence begathunting is useful because it suggests a newway to solve a puzzle about bonobos, namely,that bonobos show no evidence of monkeyhunting. Thus, no monkey hunting or mon-key eating has been recorded at the long-term bonobo sites (Wamba and Lomako), or

24 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

in the shorter-term studies of Yalosidi, LakeTumba, or Lilungu, even though these stud-ies have provided sufficient data to record,for example, termite-eating in four of thesites (Thompson, 1997). Bonobos do eat meatoccasionally, however, which they obtain in-dividually by seizing young antelope. Thelack of monkey hunting by bonobos is strik-ing given that they not only prey on terres-trial mammals, but also complete with eachother to eat meat, and sometimes interactsocially with monkeys, in grooming andplay. On three occasions, they have evenkidnapped young monkeys during play, butnot eaten them (reviewed by Wrangham andPeterson, 1996).

Stanford (1998b) suggested that the rea-son why male bonobos hunt rarely is thathunting has a low pay-off, because they tendto lose meat to females. Against this, low-ranking male chimpanzees often lose meatto high-ranking males, but still hunt fre-quently (Goodall, 1986). Furthermore it ismonkey hunting, rather than meat eating,that appears to be lacking in bonobos, notonly among males but also among females.Stanford’s proposal is therefore not sup-ported.

Another possible explanation for the lackof observations of monkey hunting by bono-bos is stochastic. Hunting traditions mightvary among bonobo populations. If so, mon-key hunting may be observed in the future,in populations that have not yet been stud-ied. However, there is no evidence that anychimpanzee population fails to hunt mon-keys, provided monkeys are present. Chim-panzees prey on monkeys in at least 12 sites,including all the long-term sites [Gombe,Mahale, Taı, Kibale, and Budongo (Goodall,1986)] as well as seven lesser-known popula-tions [Chambura, (B. Fahey, personal com-munication; Kahuzi-Biega, DRC (Basasoseand Yamagiwa, 1997), Lope, Gabon (Tutinand Fernandez, 1993), Mt. Assirik, Senegal(McGrew et al., 1979), Outamba-Kilimi, Si-erra Leone (Alp, 1993), Sapo, Liberia (Ander-son et al., 1983), Tongo, DRC (A. Lanjouw,personal communication). There is, there-fore, a strong contrast between the wide-spread occurrence of monkey hunting inchimpanzees and its absence in bonobos.Contrary to the stochastic hypothesis, this

suggests that compared to chimpanzees,bonobos have a weaker motivation to huntmonkeys.

As a third possibility, therefore, Wrang-ham and Peterson (1996) noted that the lackof monkey hunting among bonobos might beexplained as a consequence of their lowinterest in intraspecific killing. For ex-ample, if bonobos evolved from a chimpanzee-like ancestor, they began with a tendency forlethal raiding which was lost or inhibitedwhen they acquired relatively stable par-ties. Accordingly, the evolution of inhibitionsagainst lethal raiding may have inadver-tently caused monkey hunting to be inhib-ited also, if the two patterns are indeedmotivationally related (Eibl-Eibesfeldt, 1975;van Hooff, 1990). This proposal implies thatmonkey hunting is motivationally more simi-lar to lethal raiding, and relatively distinctfrom the killing of terrestrial ungulates.Important similarities between monkeyhunting and lethal raiding could include thenecessity for coordination and planning andthe ability to assess an adequate powerimbalance between predators and prey.

In summary, it is admittedly speculativeto propose that monkey hunting has beenlost in bonobos as a result of selection againstpropensities relevant to lethal raiding. How-ever, this idea appears to explain the factsbetter than alternative hypotheses. It sug-gests a correlation between group huntingand lethal raiding in humans, chimpanzeesand bonobos that challenges traditionalthinking, and draws attention to the needfor further data.

Morality

This paper suggests that violent propensi-ties of a particular kind have been positivelyselected among male chimpanzees and hu-mans. Biologically, this is unsurprising. Like-wise, propensities for particular types ofaltruistic and cooperative behavior haveprobably also evolved through selection, andare neither more nor less important biologi-cally than violence.

But anthropologists’ views on violence tendto be interpreted politically. For exampleOtterbein (1997) labeled anthropologists as‘‘Hawks’’ or ‘‘Doves’’ according to whetherthey consider evolutionary biology relevant

25EVOLUTION OF COALITIONARY KILLINGWrangham]

or irrelevant to warfare (Otterbein, 1997).Although Doves (e.g., Power, 1991; Suss-man, 1997) suggest that Hawks (e.g., Ghi-glieri, 1984; Goodall, 1986; Wrangham andPeterson, 1996) are culturally biassed inthinking that male chimpanzees strive ag-gressively for status and use violence adap-tively in intergroup interactions, there is noevidence for a positive correlation betweenanthropological Hawkishness (in the Otter-bein sense) and political beliefs. Indeed,some notable ‘‘anthropological Hawks’’ havebeen prominent in the search for peace. Forexample, Hamburg (1991) argued for theimportance of biological similarities in chim-panzee and human violence, and in the samespirit co-chaired a multi-year effort to re-duce the frequency and intensity of interna-tional violence (Carnegie Commission, 1997).There is no moral high ground to be held byvirtue of being an anthropological Dove.

Admittedly, any theory of violence hasmoral implications, because biological analy-ses can be misused. But no theory, howeverbenign or malevolent or whether based onbiology, psychology, or culture, is immune toco-option by ideologues and propagandists.While German military philosophy wasbacked by Darwinism in the First WorldWar, French military philosophy was backedby Bergson’s theory of creative evolution(Tuchman, 1962). On either side, opposingtheories of evolution were used to bolster thewaging of war. Military organizations can beexpected to deceive themselves and theirfollowers using any available materials(Wrangham, 1999b).

I see no better course than to follow Dar-win (1871, p 405): ‘‘. . .we are not here con-cerned with hopes or fears, only with thetruth as far as our reason permits us todiscover it.’’ Lethal violence appears strik-ingly frequent among chimpanzees and hu-mans, and appears explicable by relativelysimple adaptive rules. Current evidence sug-gests it has been a major selective pressurefor significant periods of chimpanzee andhuman evolution. Until lethal violence isshown to be a strange new phenomenon, weshould consider it sufficiently ancient tohave influenced the temperaments of bothspecies, particularly of males, in ways thatshould be taken seriously. The more we

understand about the evolutionary originsand persistence of intergroup violence, thebetter we can predict and avert it.

CONCLUSION

Despite some important unsolved prob-lems, chimpanzee lethal raiding appearsgenerally well explained by the imbalance-of-power hypothesis, which states that success-ful attacks on rivals are favored becausethey increase the dominance status of theaggressors. A combination of three pointslikewise suggests that selection has favoredunprovoked intergroup violence in humanmales: the prevalence of human war raiding,the similarities of chimpanzee and humanlethal raiding, and the ability of the imbal-ance-of-power hypothesis to explain themammalian distribution of lethal violence.Until an alternative model exists, chimpan-zees and humans are, therefore, best re-garded as species in which a dominancedrive by male groups has been positivelyselected.

If this conclusion has merit, anthropologyhas given inadequate consideration to coali-tionary violence as a force in human evolu-tion. As anthropologists, we have a duty toacknowledge the horrors of our evolutionarypast, partly for the sake of truth, and partlyto consider how such behavior can be avoidedin the future. By combining primatological,paleontological and behavioral-ecological ev-idence, anthropologists can provide espe-cially rich tests of evolutionary hypotheses.These will offer a solid base from whichevolutionary anthropology can work withother disciplines to understand culturalvariation and the proximate stimuli thatelicit violence.

ACKNOWLEDGMENTS

I thank Clark Larsen and the AmericanAssociation of Physical Anthropologists forinviting me to speak at the AAPA annualmeeting in Salt Lake City, April 1998, whichprompted this paper. Irven DeVore, RobertHinde, Bill McGrew, and Karen Strier gener-ously offered extended comments. For help-ful critiques I am grateful also to Chris-topher Boehm, Christophe Boesch, NancyDeVore, Martin Muller, Vernon Reynolds,Carel van Schaik, and Michael Wilson. Chris-

26 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

tophe Boesch, Katie Fawcett, Martin Muller,Toshisada Nishida, Vernon Reynolds, andMichael Wilson kindly gave access to unpub-lished material, Michael Huffman aided withreferences, and Johan van der Dennen im-proved the article by providing his impor-tant book.

LITERATURE CITED

Adams DB. 1983. Why there are so few women warriors.Behav Sci Res 18:196–212.

Adams ES. 1990. Boundary disputes in the territorialant Azteca trigona: effects of asymmetries in colonysize. Anim Behav 39:321–328.

Alexander RD. 1987. The biology of moral systems.Hawthorne, NY: Aldine de Gruyter. p. 1–301.

Alexander RD. 1989. Evolution of the human psyche. In:Mellars P, Stringer C, editors. The human revolution:Behavioral and biological perspectives on the originsof modern humans. Princeton, NJ: Princeton Univer-sity Press. p 455–513.

Alp R. 1993. Meat-eating and ant-dipping by wild chim-panzees in Sierra Leone. Primates 34:463–468.

Anderson JR, Williamson EA, Carter J. 1983. Chimpan-zees of Sapo Forest, Liberia: density, nests, tools, andmeat-eating. Primates 24:594–601.

ArcadiAC, Wrangham RW. 1999. Infanticide in chimpan-zees: review of cases and a new within-group observa-tion from the Kanyawara study group in KibaleNational Park. Primates 40:337–351.

Archer J, and Huntingford F. 1994. Game theory modelsand escalation of animal fights. In: Potegal M, Knut-son JF, editors. The dynamics of aggression. Hillsdale,NJ: Lawrence Erlbaum. p 3–31.

Ardrey R. 1961. African genesis: A personal investiga-tion into the animal origins and nature of man. NewYork: Atheneum. p 1–330.

Ardrey R. 1966. The territorial imperative. New York:Atheneum. p 1–416.

Ashley Montagu MF. 1968. Man and aggression. Lon-don: Oxford University Press. p 1–178.

Austad S. 1983. A game theoretical interpretation ofmale combat in the bowl and doily spider (Frontinellapyramitela). Anim Behav 31:59–73.

Bailey RC. 1991. The behavioral ecology of Efe pygmymen in the Ituri Forest, Zaıre. Ann Arbor, MI: Univer-sity of Michigan Press. p 1–143.

Barkow J, Cosmides L, Tooby J. 1995. The adaptedmind: Evolutionary psychology and the generation ofculture. New York: Oxford University Press. p 1–668.

Basasose N, Yamagiwa J. 1997. Predation on mammalsby chimpanzees in the montane forest of Kahuzi,Zaire. Primates 38:45–55.

Bauer HR. 1980. Chimpanzee society and social domi-nance in evolutionary perspective. In: Omark DR,Strayer FF, Freedman D, editors. Dominance rela-tions: Ethological perspectives of human conflict. NewYork: Garland. p 97–119.

Boehm C. 1992. Segmentary ‘‘warfare’’ and the manage-ment of conflict: comparison of East African chimpan-zees and patrilineal-patrilocal humans. In: HarcourtAH, de Waal FBM, editors. Coalitions and alliances inhumans and other animals. Oxford: Oxford Univer-sity Press. p 137–173.

Boehm C. 1999. The natural selection of altruistic traits.Human Nature: 10.

Bock K. 1980. Human nature and history: A response tosociobiology. New York: Columbia University Press. p1–241.

Boesch C. 1996. Social grouping in Taı chimpanzees. In:McGrew WC, Marchant LF, Nishida T, editors. Greatape societies. Cambridge: Cambridge University Press.p 101–113.

Boesch C, Boesch H. 1989. Hunting behavior of wildchimpanzees in the Taı National Park. Am J PhysAnthropol 78:547–573.

Boesch C, Boesch H. 1999. The chimpanzees of the TaıForest: Behavioral ecology and evolution. New York:Oxford University Press, in press.

Brain CK. 1981. The hunters or the hunted?An introduc-tion to African cave taphonomy. Chicago: University ofChicago Press, p 1–365.

Bueno de Mesquita B. 1981. The war trap. New Haven,CT: Yale University Press. p 1–223.

Bueno de Mesquita B. 1985. The war trap revisited. Arevised expected utility model. Am Polit Sci Rev79:156–177.

Bueno de Mesquita B, Lalman D. 1992. War and reason:Domestic and international imperatives. New Haven,CT: Yale University Press. p 1–322.

Byers J. 1997. American pronghorn: Social adaptationsand the ghosts of predators past. Chicago: Universityof Chicago Press. p 1–300.

Bygott JD. 1979. Agonistic behavior, dominance, andsocial structure in wild chimpanzees of the GombeNational Park. In: Hamburg DA, McCown ER, edi-tors. The great apes. Menlo Park, CA: Benjamin/Cummings. p 405–428.

Carnegie Commission on Preventing Deadly Conflict.1997. Preventing deadly conflict: Final report. Wash-ington DC: Carnegie Commission on PreventingDeadly Conflict. p 1–257.

Caro TM, Collins DA. 1986. Male cheetahs of theSerengeti. Nat Geogr Res 2:75–86.

Cartmill M. 1993. A view to a death in the morning:Hunting and nature through history. Cambridge, MA:Harvard University Press. p 1–331.

Chagnon NA. 1992. Yanomamo: The last days of Eden.New York: Harcourt Brace Jovanovich. p 1–309.

Chapais B. 1995. Alliances as a means of competition inprimates: evolutionary, developmental and cognitiveaspects. Yearbk Phys Anthropol 38:115–136.

Chapman CA, Wrangham RW. 1993. Range use of theforest chimpanzees of Kibale: implications for theunderstanding of chimpanzee social organization. AmJ Primatol 31:263–273.

Chapman CA, White FJ, Wrangham RW. 1994. Partysize in chimpanzees and bonobos: A reevaluation oftheory based on two similarly forested sites. In:Wrangham RW, McGrew WC, de Waal FBM, editors.Chimpanzee cultures. Cambridge, MA: Harvard Uni-versity Press. p 41–58.

Chapman CA, Wrangham RW, Chapman LJ. 1995.Ecological constraints on group size: an analysis ofspider monkey and chimpanzee subgroups. BehavEcol Sociobiol 32:199–209.

Chapman CA, Chapman LJ, Wrangham R, Isabirye-Basuta G, Ben-David K. 1997. Spatial and temporalvariability in the structure of a tropical forest. Afr JEcol 35:287–302.

Cheney DL. 1986. Interactions and relationships be-tween groups. In: Smuts BB, Cheney DL, SeyfarthRM, Wrangham RW, Struhsaker TT, editors. Primatesocieties. Chicago: University of Chicago Press. p227–239.

Cheney DL, Seyfarth RM, Andelman SJ, Lee PC. 1988.Reproductive success in vervet monkeys. In: Clutton-Brock TH, editor. Reproductive success: Studies ofindividual variation in contrasting breeding systems.Chicago: University of Chicago Press. p 384–402.

Clutton-Brock TH, Guinness FE, Albon SD. 1982. Red

27EVOLUTION OF COALITIONARY KILLINGWrangham]

deer: Behavior and ecology of two sexes. Chicago:University of Chicago Press. p 1–378.

Daly M, Wilson M. 1988. Homicide. Hawthorne, NY:Aldine de Gruyter. p 1–328.

Dart RA. 1953. The predatory transition from ape toman. Intern Anthropol Ling Rev 1:201–218.

Dart RA, Craig D. 1959. Adventures with the missinglink. New York: Harper. p 1–255.

Darwin C. 1871. The descent of man and selection inrelation to sex. London: John Murray. p 1–475.

Dentan RK. 1992. The rise, maintenance, and destruc-tion of peaceable polity: a preliminary essay in politi-cal ecology. In: Silverberg J, Gray JP, editors. Aggres-sion and peacefulness in humans and other primates.Oxford: Oxford University Press. p 214–270.

de Waal FBM. 1982. Chimpanzee politics: Power and sexamong apes. New York: Harper and Row. p 1–223.

de Waal FBM, and Lanting F. 1997. Bonobo: The forgot-ten ape. Berkeley, CA: University of California Press.p 1–210.

de Ruiter JR, and van Hooff JARAM. 1993. Male domi-nance rank and reproductive success in primategroups. Primates 34:513–524.

East ML, Hofer H. 1991. Loud-calling in a femaledominated mammalian society. II. Behavioral con-texts and functions of whooping of spotted hyenas,Crocuta crocuta. Anim Behav 42:651–669.

Ehrenreich B. 1997. Blood rites: Origins and history ofthe passions of war. New York: Metropolitan. p 1–292.

Eibl-Eibesfeldt I. 1975. Krieg und Frieden aus der Sichtder Verhaltensforschung. Munich: Piper Verlag. p1–315.

Ellis L. 1995. Dominance and reproductive successamong nonhuman animals: A cross-species compari-son. Ethol Sociobiol 16:257–333.

Enquist M, Leimar O. The evolution of fatal fighting.Anim Behav 39:1–9.

Ferguson RB. 1989. Anthropology and war: theory,politics, ethics. In: Turner PR, Pitt D, editors. Theanthropology of war and peace: Perspectives on thenuclear age. London: Greenwood Press. p 141–159.

Ferrill A. 1985. The origins of war from the stone age toAlexander the Great. London: Thames and Hudson. p1–240.

Frank LG. 1986. Social organization of the spottedhyaena (Crocuta crocuta). I. Demography. Anim Be-hav 35:1500–1509.

Furuichi T, Idani G, Ihobe H, Kuroda S, Kitamura K,Mori A, Enomoto T, Okayasu N, Hashimoto C, Kano T.1998. Population dynamics of wild bonobos (Panpaniscus) at Wamba. Int J Primatol 19:1029–1043.

Ghiglieri MP. 1984. The chimpanzees of Kibale Forest: Afield study of ecology and social structure. New York:Columbia University Press. p 1–226.

Ghiglieri MP. 1988. Sociobiology of the great apes andthe hominid ancestor. J Hum Evol 16:319–358.

Goodall J. 1986. The chimpanzees of Gombe: Patterns ofbehavior. Cambridge, MA: Harvard University Press.p 1–673.

Goodall J, Bandora A, Bergmann E, Busse C, MatamaH, Mpongo E, Pierce A, Riss D. 1979. Intercommunityinteractions in the chimpanzee population of theGombe National Park. In: Hamburg DA, McCown ER,editors. The great apes. Menlo Park, CA: Benjamin/Cummings. p 13–54.

Gould SJ. 1996. The Diet of Worms and the defenestra-tion of Prague. Natur Hist, Sept: 18–67.

Grinnell J, Packer C, Pusey AE. 1995. Cooperation inmale lions: kinship, reciprocity or mutualism? AnimBehav 49:95–105.

Grossman D. 1996. On killing: The psychological cost oflearning to kill in war and society. Boston, MA: Little,Brown. p 1–367.

Hamburg DA. 1991. An evolutionary perspective onhuman aggression. In: Bateson P, editor. The develop-ment and integration of behavior: Essays in honor ofRobert Hinde. Cambridge: Cambridge UniversityPress. p 419–458.

Harcourt AH. 1987. Dominance and fertility amongfemale primates. J Zool Lond 213:471–487.

Hashimoto C, Tashiro Y, Kimura D, Enomoto T, Ingman-son EJ, Idani G, Furuichi T. 1998. Habitat use andranging of wild bonobos (Pan paniscus) at Wamba. IntJ Primatol 19:1045–1060.

Hausfater G, and Hrdy SB, editors. 1984. Infanticide:Comparative and evolutionary perspectives. Haw-thorn, NY: Aldine de Gruyter. p 1–596.

Henschel JR, Skinner JD. 1991. Territorial behavior bya clan of spotted hyenas Crocuta crocuta. Ethology88:223–235.

Hinde RA. 1993.Aggression and war: individuals, groups,and states. In: Tetlock PE, Husbands JL, Jervis R,editors. Behavior, society, and international conflict.Oxford: Oxford University Press. p 8–70.

Hofer A, Huffman MA, Zeisler G. 1998. Mahale: Begeg-nung mit Schimpansen. Verlag Navalon. p 1–159.

Holldobler B. 1981. Foraging and spatiotemporal territo-ries in the honey ant Myrmecocystus mimicus Wheeler(Hymenoptera: Formicidae). Behav Ecol Sociobiol9:301–314.

Hrdy SB. 1990. Sex bias in nature and in history: a late1980s reexamination of the ‘‘biological origins’’ argu-ment. Yearbk Phys Anthropol 33:25–37.

Idani G. 1991. Cases of inter-unit group encounters inpygmy chimpanzees at Wamba, Zaıre. In: Ehara A, etal., editors. Primatology Today: Proceedings of theXIIIth Congress of the International PrimatologicalSociety. Amsterdam: Elsevier. p 235–238.

Isabirye-Basuta G. 1989. The ecology and conservationstatus of the chimpanzee Pan troglodytes schwein-furthii in Kibale Forest, Uganda. Ph.D. dissertation,Makerere University, Kampala, Uganda. p 1–280.

Janson CH, Goldsmith ML. 1995. Predicting group sizein primates: foraging costs and predation risks. BehavEcol 6:326–336.

Jurmain R. 1997. Skeletal evidence of trauma in Africanapes, with special reference to the Gombe chimpan-zees. Primates 38:1–14.

Kano T. 1992. The last ape: Pygmy chimpanzee behaviorand ecology. Stanford, CA: Stanford University Press.p 1–248.

Keegan J. 1993. A history of warfare. New York: AlfredA. Knopf. p 1–432.

Keeley LH. 1996. War before civilization. New York:Oxford University Press. p 1–245.

Knauft B. 1991. Violence and sociality in human evolu-tion. Curr Anthropol 32:391–428.

Krebs JR and Davies NB. 1997. Behavioral ecology: anevolutionary approach. Oxford: Blackwell. p 1–456.

Kruuk H. 1972. The spotted hyena: A study of predationand social behavior. Chicago: Chicago University Press.p 1–335.

Kuroda S. 1979. Grouping of the pygmy chimpanzees.Primates 20:161–183.

Larsen CS. 1997. Bioarchaeology: Interpreting behaviorfrom the human skeleton. Cambridge: CambridgeUniversity Press. p 1–461.

Leach E. 1968. Don’t say ‘‘Boo’’ to a goose. In: AshleyMontagu MF, editor. Man and aggression. London:Oxford University Press. p 65–73.

Leonard WR, Robertson ML. 1997. Comparative pri-mate energetics and hominid evolution. Am J PhysAnthropol 102:265–281.

Lorenz K. 1966. On aggression. Translated by MarjorieLatzke. London: Methuen. p 1–273.

28 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999

Lowen C, Dunbar RIM. 1994. Territory size and defend-ability in primates. Behav Ecol Sociobiol 35:347–354.

Manson JH, Wrangham RW. 1991. Intergroup aggres-sion in chimpanzees and humans. Curr Anthropol32:369–390.

Maschner HDG, Reedy-Maschner KL. 1998. Raid, re-treat, defend (repeat): the archaeology and ethnohis-tory of warfare on the North Pacific rim. J AnthropolArchaeol 17:19–51.

McComb K, Packer C, Pusey A. 1994. Roaring andnumerical assessment in contests between groups offemale lions, Panthera leo. Anim Behav 47:379–387.

McGrew WC, Tutin CEG, Baldwin PJ. 1979. New dataon meat-eating by wild chimpanzees. Curr Anthropol20:238–239.

Mech LD. 1977. Productivity, mortality, and populationtrends of wolves in north-eastern Minnesota. J Mam-mal 58:559–574.

Mech LD. 1994. Buffer zones of territories of gray wolvesas regions of intraspecific strife. J Mammal 75:199–202.

Mech LD, Adams LG, Meier TJ, Burch JW, Dale BW.1998. The Wolves of Denali. Minneapolis, MN: Univer-sity of Minnesota Press. p 1–225.

Miller LE. 1998. Fatal attack among wedge-cappedcapuchins. Folia Primatol 69:89–92.

Milton K. 1987. Primate diets and gut morphology:implications for hominid evolution. In: Harris M, RossEB, editors. Food and evolution: Toward a theory ofhuman food habits. Philadelphia: Temple UniversityPress. p 93–116.

Mitani JC, Rodman PS. 1979. Territoriality: relation ofranging patterns and home range size to defendabil-ity, with an analysis of territoriality among primatespecies. Behav Ecol Sociobiol 5:241–251.

Morris D. 1977. Manwatching: A field guide to humanbehavior. New York: Harry N. Abrams, Inc. p 1–318.

Newton-Fisher NE. 1997. Tactical behavior and decision-making in wild chimpanzees. Ph.D. dissertation, Uni-versity of Cambridge. p 1–325.

Nishida T. 1979. The social structure of chimpanzees ofthe Mahale Mountains. In: Hamburg DA, McCownER, editors. The great apes. Menlo Park, CA: Benja-min/Cummings. p 73–121.

Nishida T. 1990. A quarter century of research in theMahale Mountains: An overview. In: Nishida T, editor.The chimpanzees of the Mahale Mountains: Sexualand life history strategies. Tokyo: University of TokyoPress. p 3–36.

Nishida T. 1994. Review of recent findings on Mahalechimpanzees: implications and future research direc-tions. In: Wrangham RW, McGrew WC, de Waal FBM,Heltne PG, editors. Chimpanzee cultures. Cambridge,MA: Harvard University Press. p 373–396.

Nishida T. 1996. The great chimpanzee ruler killed by acoalition of previous group mates: cruel political dy-namics in wild chimpanzees. Asahi-Shimbun, Janu-ary 31, 1996, evening edition. p 1.

Nishida T, Haraiwa-Hasegawa M, Takahata Y. 1985.Group extinction and female transfer in wild chimpan-zees in the Mahale National Park, Tanzania. Z Tierpsy-chol 67:284–301.

Nunn CL. 1999. Collective benefits, free-riders, andmale extragroup conflict. In: Kappeler P, editor. Maleprimates. Cambridge: Cambridge University Press. p192–204.

Otterbein KF. 1985. Preface to Second Edition. In: Theevolution of war: A cross-cultural study. New Haven,CT: Human Relations Area Files. p 1–165.

Otterbein KF. 1997. The origins of war. Crit Rev 11:251–277.

Packer C, Herbst L, Pusey AE, Bygott JD, Hanby JP,Cairns SJ, Mulder MB. 1988. Reproductive success of

lions. In: Clutton-Brock TH, editor. Reproductive suc-cess: Studies of individual variation in contrastingbreeding systems. Chicago: University of ChicagoPress. p 363–383.

Palombit RA. 1999. Infanticide and the evolution of pairbonds in nonhuman primates. Evol Anthropol 7:117–129.

Parish A. 1996. Female relationships in bonobos (Panpaniscus): evidence for bonding, cooperation, and fe-male dominance in a male-philopatric species. HumNature 7:61–96.

Pilbeam D. 1996. Genetic and morphological records ofthe Hominoidea and hominid origins: A synthesis. MolPhylogenet Evol 5:155–168.

Popp J, DeVore I. 1979. Aggressive competition andsocial dominance theory. In: Hamburg DA, McCownER, editors. The great apes. Menlo Park, CA: Benja-min/Cummings. p 317–338.

Power M. 1991. The Egalitarians—Human and chimpan-zee: An anthropological view of social organization.Cambridge: Cambridge University Press. p 1–290.

Prost JH. 1985. Chimpanzee behavior and models ofhominization. In: Kondo S, editor. Primate morpho-physiology, locomotor analyses, and human bipedal-ism. Tokyo: University of Tokyo Press, p 289–303.

Putland DA, Goldizen AW. 1998. Territorial behavior inthe Tasmanian native hen: group and individualperformance. Anim Behav 56:1455–1463.

Regal P. 1998. Violence and sex. Q Rev Biol 73:473–476.Reynolds V, Reynolds F. 1965. Chimpanzees of the

Budongo Forest. In: De Vore I, editor. Primate behav-ior. New York: Holt, Rinehart and Winston. p. 368–424.

Rodseth L, Wrangham RW, Smuts BB, Harrigan A.1991. The human community as a primate society.Curr Anthropol 32:221–254.

Shea BT. 1984. An allometric perspective on the morpho-logical and evolutionary relationships between pygmy(Pan paniscus) and common (Pan troglodytes) chim-panzees. In: Susman RL, editor. The pygmy chimpan-zee. New York: Plenum Press. p 89–130.

Singer JD. 1989. The political origins of internationalwar: a multifactorial review. In: Groebel J, Hinde RA,editors. Aggression and war: Their biological andsocial bases. Cambridge: Cambridge University Press.p 202–220.

Smuts BB. 1986. Sex and friendship in babboons. Haw-thorne, NY: Aldine de Gruyter. p 1–303.

Sponsel LE. 1996. The natural history of peace: apositive view of human nature and its potential. In:Gregor T, editor. A natural history of peace. Nashville,TN: Vanderbilt University Press. p 95–125.

Stanford C. 1998a. The social behavior of chimpanzeesand bonobos: A critical review. Curr Anthropol 39:399–420.

Stanford CB. 1998b. Chimpanzee and red colobus: Theecology of predator and prey. Cambridge, MA: Har-vard University Press. p 1–296.

Starin ED. 1994. Philopatry and affiliation among redcolobus. Behaviour 130:253–270.

Sugiyama Y. 1973. Social organization of wild chimpan-zees. In: Carpenter CR, editor. Behavioral regulatorsof behavior in primates. Lewisburg, PA: BucknellUniversity Press. p 68–80.

Sugiyama Y. 1989. Population dynamics of chimpanzeesat Bossou, Guinea. In: Heltne PG, Marquardt LA,editors. Understanding chimpanzees. Cambridge, MA:Harvard University Press. p 134–145.

Sugiyama Y, Kawamoto S, Takenaka O, Kumazaki K,Miwa N. 1993. Paternity discrimination and inter-group relationships of chimpanzees at Bossou. Pri-mates 34:545–552.

Sussman RW, editor. 1997. The biological basis of hu-

29EVOLUTION OF COALITIONARY KILLINGWrangham]

man behavior: A critical review. New York: PrenticeHall. p 1–382.

Thompson JAM. 1997. The history, taxonomy and ecol-ogy of the bonobo (Pan paniscus Schwartz, 1929) witha first description of a wild population living in aforest/savanna mosaic habitat. D. Phil. thesis, OxfordUniversity, p 1–315.

Tiger L. 1969. Men in groups. New York: Random House.p 1–254.

Trudeau MB, Bergmann-Riss E, Hamburg DA. 1981.Towards an evolutionary perspective on aggressivebehavior: the chimpanzee evidence. In: Hamburg DA,Trudeau MB, editors. Biobehavioral aspects of aggres-sion. New York: Alan Liss. p 27–40.

Tuchman BW. 1962. The guns of August. New York:Macmillan. p 1–511.

Turney-High HH. 1949. Primitive war: Its practice andconcepts. Columbia, SC: University of South CarolinaPress. p 1–288.

Tutin CEG, Fernandez M. 1993. Composition of the dietof chimpanzees and comparisons with that of sympat-ric lowland gorillas in the Lope Reserve, Gabon. Am JPrimatol 30:195–211.

Tutin CEG, McGrew WC, Baldwin PJ. 1983. Socialorganization of savanna-dwelling chimpanzees, Pantroglodytes verus, at Mt. Assirik, Senegal. Primates24:154–173.

van Hoof JARAM. 1990. Intergroup competition andconflict in animals and man. In: Dennen JMG van der,Falger VSE, editors. Sociobiology and conflict: Evolu-tionary perspectives on competition, cooperation, vio-lence and warfare. London: Chapman and Hall. p23–54.

van Schaik CP. 1996. Social evolution in primates: therole of ecological factors and male behavior. Proc BritAcad 88:9–31.

van der Dennen JMG. 1995. The origin of war: Theevolution of a male-coalitional reproductive strategy.Groningen, Netherlands: Origin Press. p 1–861.

Watts DP. 1999. Coalitionary mate-guarding by malechimpanzees at Ngogo, Kibale National Park, Uganda.Behav Ecol Sociobiol 44:43–55.

White FJ. 1988. Party composition and dynamics in Panpaniscus. Int J Primatol 9:179–193.

White FJ. 1996. Comparative socio-ecology of Pan panis-cus. In: McGrew WC, Marchant LF, Nishida T, editors.Great ape societies. Cambridge: Cambridge Univer-sity Press. p 29–41.

White FJ. 1998. Seasonality and socioecology: the impor-tance of variation in fruit abundance to bonobo social-ity. Int J Primatol 19:1013–1027.

Wolpoff M. 1998. Paleoanthropology. New York: AlfredKnopf. p 1–370.

Wrangham RW. 1974. Artificial feeding of chimpanzeesand baboons in their natural habitat. Anim Behav22:83–93.

Wrangham RW. 1975. The behavioral ecology of chimpan-zees in Gombe National Park, Tanzania. Ph.D. disser-tation, Cambridge University. p 1–280.

Wrangham RW. 1987. The significance of African apesfor reconstructing human social evolution. In: KinzeyWG, editor. Primate models of hominid evolution.Albany, NY: SUNY Press. p 51–71.

Wrangham RW. 1999a. Why are male chimpanzees moregregarious than mothers? A scramble competitionhypothesis. In: Kappeler P, editor. Male primates.Cambridge: Cambridge University Press. p 248–258.

Wrangham RW. 1999b. Is military incompetence adap-tive? Evol Hum Behav 20:3–17.

Wrangham RW, Peterson D. 1996. Demonic males: Apesand the origins of human violence. Boston, MA: Hough-ton Mifflin. p 1–350.

Wrangham RW, Clark AP, Isabirye-Basuta G. 1992.Female social relationships and social organization ofthe Kibale Forest chimpanzees. In: Nishida T, Mc-Grew WC, Marler P, Pickford M, de Waal FBM,editors. Topics in primatology: Human origins (Vol-ume 1). Tokyo: University of Tokyo Press. p 81–98.

Wrangham RW, Chapman CA, Clark AP, Isabirye-Basuta G. 1996. Social ecology of Kanyawara chimpan-zees: implications for understanding the costs of greatape groups. In: McGrew WC, Marchant LF, Nishida T,editors. Great ape societies. Cambridge: CambridgeUniversity Press. p 45–57.

Wrangham RW, Conklin-Brittain NL, Hunt KD. 1998.Dietary response of chimpanzees and cercopithecinesto seasonal variation in fruit abundance: I. Antifeed-ants. Int J Primatol 19:949–970.

Zihlman AL. 1996. Looking back in anger. Nature384:35–36.

Zihlman AL, Cramer DL. 1978. Skeletal differencesbetween pygmy (Pan paniscus) and common chimpan-zees (Pan troglodytes). Folia Primatol 29:86–94.

30 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 42, 1999