Counter aggression and reconciliation in Assamese macaques (Macaca assamensis).код для вставкиСкачать
American Journal of Primatology 56:215–230 (2002) Counter Aggression and Reconciliation in Assamese Macaques (Macaca assamensis) MATTHEW A. COOPER* AND IRWIN S. BERNSTEIN Department of Psychology, University of Georgia, Athens, Georgia Patterns of aggressive and affiliative behavior, such as counter aggression and reconciliation, are said to covary in the genus Macaca; this is referred to as the systematic variation hypothesis. These behavior patterns constitute a species dominance style. Van Schaik’s  socioecological model explains dominance style in macaques in terms of withinand between-group contest competition. Dominance style is also said to correlate with phylogeny in macaques. The present study was undertaken to examine phylogenetic and socioecological explanations of dominance style, as well as the systematic variation hypothesis. We collected data on counter aggression and reconciliation from a habituated group of Assamese macaques (Macaca assamensis) at the Tukeswari Temple in Assam, India. The proportion of agonistic episodes that involved counter aggression was relatively low. Counter aggression, however, occurred more often among males than among females, and it was most common when females initiated aggression against males. The conciliatory tendency for this group of Assamese macaques was 11.2%. The frequency of reconciliation was low for fights among males and for fights among females, but reconciliation was particularly rare for opposite-sexed opponents. Female social relationships were consistent with the systematic variation hypothesis, and suggest a despotic dominance style. A despotic dominance style in Assamese macaques weakens the correlation between dominance style and phylogeny in macaques, but it is not inconsistent with the socioecological model. Male–female relationships were not well explained by the despotic-egalitarian framework, and males may well have more tolerant social relationships than do females. Sex differences need to be considered when categorizing species according to dominance style. Am. J. Primatol. 56:215–230, 2002. © 2002 Wiley-Liss, Inc. Key words: Macaca assamensis; reconciliation; counter aggression; dominance style; sex differences Contract grant sponsor: National Geographic Society; Contract grant number: 5862-97. *Correspondence to: Matthew Cooper, Center for Brain Science and Health, Georgia State University, 24 Peachtree Center Ave. NE, Atlanta, GA 30303. E-mail: firstname.lastname@example.org Received 28 February 2000; revision accepted 30 January 2002 DOI: 10.1002/ajp.1076 Published online in Wiley InterScience (www.interscience.wiley.com). © 2002 Wiley-Liss, Inc. 216 / Cooper and Bernstein INTRODUCTION Dominance is recognized by an asymmetry in the pattern of aggression and submission in agonistic encounters [Bernstein, 1981]. Physical abilities and social skills both influence dominance relationships. Consequently, dominance hierarchies are rarely perfectly linear. Even among macaques, where dominance relationships are pervasive, subordinates contest aggression in a variety of species (M. arctoides [de Waal & Luttrell, 1989], M. mulatta [Bernstein & Ehardt, 1985], M. nigra [Petit et al., 1997], and M. tonkeana [Thierry, 1985]). Bidirectional aggression in some cases may reflect the influence of alliances, testing by the subordinate, and the process of rank acquisition among immature animals. Bidirectional aggression may also indicate social tolerance. For example, bidirectional aggression has been related to a lower intensity of aggression and less exclusion around a limited water source [Thierry, 1985; de Waal & Luttrell, 1989]. If dominants are forced to “tolerate” subordinates when they are physically incapable of preventing the subordinates’ retaliation, their “social tolerance” is not tolerance at all. Alternatively, a dominant animal may tolerate insubordination from valuable social patterns such as kin, grooming partners, sexual partners, alliance partners, or individuals that aid against an external enemy, such as another group or predators. Social tolerance is related not only to the frequency of bidirectional aggression, but also to the intensity of aggression, the expression of reconciliation, kin biases in affiliative behavior (such as grooming), and patterns of agonistic aiding and infant care [Thierry, 2000]. In despotic societies, subordinates should readily submit and rarely contest clumped resources. Affiliative relationships among dominants and subordinates should be weak, and reconciliation rare. Animals may affiliate and form alliances with kin, and mothers should be restrictive with their offspring. In species with a more tolerant dominance style, access to resources should be less strongly linked to dominance rank and subordinate animals should be less likely to immediately submit. Relationships among dominants and subordinates should permit high levels of affiliative contact and reconciliation. Social bonds with kin may be less important, and mothers should be less restrictive. These behavior patterns should covary and constitute a species’ dominance style (the systematic variation hypothesis) [Thierry, 2000; Castles et al., 1996; de Waal & Luttrell, 1989; Thierry, 1985]. These behavior patterns are interrelated, so factors influencing one type of behavior should influence the rest. Cords and Aureli  suggested that, at least for reconciliation, variation depends on the value, security, and compatibility of the relationship. While the systematic variation hypothesis predicts the covariation of behavior that comprises dominance style, two theories have been proposed to explain interspecies variation in dominance style. Van Schaik’s  socioecological hypothesis states that the distribution of resources influences the form of resource competition [Sterck et al., 1997]. When resources occur in small clumps, and access within groups is resolved by contest competition, subordinates may learn to submit immediately to dominant individuals whenever there is any possibility of an agonistic encounter. This leads to despotic dominance relationships marked by a high frequency of submissive signaling [de Waal & Luttrell, 1989; Thierry, 1985]. When resources occur in larger clumps, priority of access to incentives within groups may only sometimes be resolved by contest competition, and between-group competition may become important. In this case, subordinates may wait for some aggressive signal from a dominant before submitting or withdrawing. This is considered to indicate greater tolerance and less despotic relation- Counter Aggression and Reconciliation / 217 ships [Preuschoft & van Schaik, 2000]. When resources are not clumped, and within-group competition is mainly of a scramble variety, then egalitarian relationships should prevail. In this case, subordinates show initial resistance to a dominant, dominants use milder forms of aggression that do not elicit prompt withdrawal, or dominance relationships do not exist. The phylogenetic hypothesis states that closely related macaques will show similar dominance styles [Matsumura, 1999; Thierry, 2000]. This hypothesis predicts a less immediate effect of ecology. It predicts that dominance styles should reflect the ecological pressures for greater or lesser contest competition within and between groups in the niches of the subgenera in the genus Macaca. The socioecological and phylogenetic hypotheses are not mutually exclusive. While the socioecological hypothesis advocates the importance of ecological factors that select for variation in dominance style, the phylogenetic hypothesis stresses the importance of phylogenetic constraints on the expression of dominance style. No doubt both ecological and phylogenetic processes are at work. This study examined the dominance style of Macaca assamensis, the Assamese macaque. Since M. assamensis belongs to the Sinica group of macaques, and members of this group are hypothesized to have more tolerant social relationships than members of the Fascicularis group, which includes M. fascicularis, M. mulatta, M. fuscata, and M. cyclopis (Fig. 1), the phylogenetic hypothesis predicts tolerant social relationships. Alternatively, since our study group lived in a temple environment where food was clumped and restricted, and where between group competition was virtually nonexistent, the socioecological hypothesis predicts despotic social relationships. Regardless of the role of phylogenetic and ecological factors, the systematic variation hypothesis predicts that the behavior patterns that constitute dominance style vary as a set. Thus, we predicted that levels of counter aggression would correspond to levels of reconciliation, regardless of whether those levels indicated despotic or tolerant social relationships. METHODS Study Site This study was conducted on a group of Assamese macaques living at the Tukeswari temple in the district of Goalpara in the state of Assam, India. The group lived on the temple grounds and on the steep, boulder-strewn hill behind the temple. The hill was sparsely forested and surrounded by farmland. The main temple was at the base of the hill and a second, smaller temple was at the top of the hill, 110 m above the temple grounds. The monkeys raided nearby rice fields and stole food from shops and houses. Local priests fed the monkeys daily at each temple site, and the monkeys received food from worshipers as well. Feeding was intense and competitive at the temple sites. Sporadic offerings could extend feeding bouts for up to an hour. Subjects The group consisted of 64 animals, all well habituated and individually recognizable. No births or deaths occurred during the study period. Fourteen females and 14 males were judged to be over 4 years of age (sexually mature). Eight females and 5 males were estimated to be 3 years of age (pubertal), and the remainder of the group consisted of 16 juveniles and seven yearlings. Age 218 / Cooper and Bernstein Fig. 1. The phylogenetic tree is taken from Delson . Dominance style is indicated above species for which data are available. T = tolerant, D = despotic, I = intermediate, and ? = a degree of uncertainty. The dominance style for Sulawesi macaques is based on M. maurus, M. nigra, and M. tonkeana. This classification differs from Thierry  in that he considered Sulawesi macaques more tolerant than all other macaques, and M. fascicularis intermediate. References: 1Abegg et al. , 2 Aureli et al. , 3Castles et al. , 4Chaffin et al. , 5Matsumura , 6Matsumura , 7Ogawa , 8Petit et al. , 9Preuschoft et al. , 10 Silk , 11Silk and Samuels , 12Silk et al. , 13Taub , 14Thierry , 15de Waal and Luttrell , 16de Waal and Ren , and 17Zhao . Counter Aggression and Reconciliation / 219 estimates were made at the start of the study and were based on size and the eruption of canines in males. Aside from mother–infant relationships, kinship was unknown. Procedures A team of three observers collected data from October 1997 to February 1998. We began data collection after establishing better than 85% agreement in focal sampling using a pre-established ethogram. Only M.C. and I.B. collected the post-conflict (PC) samples, and M.C. collected all focal samples. We conducted focal observations on all 28 sexually mature animals. During focal samples, we recorded the identity of the initiator and recipient, the type of behavior, and the time when the behavior occurred. The focal sample ethogram included affiliative behavior, sexual behavior, and agonistic behavior. Our ethogram was modeled after the behavior of stumptail macaques (M. arctoides) [Bertrand, 1969]. We also included the lip-grin, as described for Assamese macaques [Hill & Bernstein, 1969]. Other behavioral definitions match those used by de Waal and Luttrell  and Castles et al. . Aggressive behavior included, in order of greater intensity: open-mouth threat, lunge, charge, chase, contact aggression, bite, and severe bite. We scored all agonistic responses individually for third-party aggression, indicating the direction of support. Counter aggression was defined as the recipient of aggression responding with aggressive behavior of any intensity. Counter aggression did not require that opponents use aggressive signals simultaneously, nor did it require that the fight end in a draw. Each subject was observed for 4 hr (approximately 24 samples of 10 min each), and the order of selection was randomized. Focal observations were restricted to periods when the monkeys were not receiving food from priests, worshipers, or local people. An effort was also made to reduce human interference by not performing focal observations when the monkeys were actively engaged with people. Prior to the start of each focal sample we recorded the identity of every animal in view and its distance from the focal subject. Distances from the focal subject were divided into three categories: within 5 m, between 5 and 25 m, and over 25 m. We also recorded data on aggression ad libitum. We recorded data in the same fashion as during the focal samples. To be sure we accurately identified the initiator and recipient of aggression in the ad libitum data, we analyzed only those episodes for which we were confident we observed the beginning of the encounter. We based dominance relationships on the presence of clear submissive behavior, such as silent bared-teeth display, avoid, and flee. When submissive types of behavior were accompanied by counter aggression, the episode was not used for determining dominance relationships. We performed a 10-min PC focal sample after all agonistic episodes between sexually mature animals and between sexually mature and pubertal animals that involved a charge or more intense aggression. The PC sample began immediately following the last agonistic response in the conflict. If fighting started again within 2 min the PC sample was extended. We used the same data collection protocol for PC samples as for focal samples. We collected PC samples on all dyads in polyadic fights, although the two main opponents were also identified. The main victim was the animal initially attacked, and the main aggressor was the animal that attacked most intensively and longest. For fights among sexually mature animals, priority as focal subject was given to the animal for which we had the least amount of data. 220 / Cooper and Bernstein The sexually mature animal was always chosen as the focal subject in fights with pubertal animals. We attempted to sample each subject an equal number of times as an initiator and recipient of aggression. Complete counterbalancing, however, could not be achieved because some animals were rarely the recipients of aggression. Although methods used to measure reconciliation differ, studies often employ a matched control (MC) sample that can be compared to the PC sample. The ideal MC sample should resemble the corresponding PC sample in every way, but should not occur shortly after aggression. In captivity a planned MC sample is typically performed the following day at the same time as the corresponding PC sample, but is postponed if the subject was recently engaged in an agonistic interaction. This procedure controls for the daily fluctuation in group activity associated with husbandry. In field studies the appropriate MC sample is selected from baseline data, while correcting for the availability or distance between former opponents [Aureli, 1992]. In this study, MC samples were selected from the pool of focal samples. The appropriate focal sample was selected by matching the distance between subjects at the start of the focal and PC sample. In the case of a tie, the focal sample closest to the date of the PC sample was chosen. Focal samples were not performed when the animals received food from humans. Consequently, MC samples do not represent bouts of intense feeding. In contrast, PC samples include fights during feeding and nonfeeding conditions. We felt the MC samples were appropriate comparisons for the PC samples in the feeding condition because the amount of affiliative contact between former opponents during PC samples was not significantly different in the feeding and nonfeeding conditions (Wilcoxon: Z (18) = 1.73, P > .05; see discussion for further explanation). Analysis Counter aggression. In the analysis of counter aggression, we used only aggressive encounters that involved charges or more intense forms of aggression. The victim’s response to aggression was divided into four types: 1) submit immediately; 2) hold ground and confront the opponent, but neither submit nor respond with aggression; 3) respond with aggression and eventually submit; and (4) respond with aggression and never submit. In category 1 the recipient of aggression issued a bared-teeth display or squeal, and/or fled. In category 2 the recipient of aggression oriented toward the aggressor, walked backward if the aggressor approached, and lip-grinned. In general, category 2 included ambiguous responses that could not be scored as clearly submissive or aggressive. Categories 3 and 4 combined measured counter aggression. We analyzed the responses to aggression as weighted means. We calculated the percentage of responses for each subject in each category, and then calculated a mean and standard deviation (SD) for all animals. The total number of responses in each category is shown in Table I. We used two-tailed Mann-Whitney U-tests and Wilcoxon matched-pairs tests for statistical analysis. To partially compensate for the large number of planned comparisons, we set the significance level at P < .01. Reconciliation. The occurrence of reconciliation was analyzed using the PC/ MC method as developed by de Waal and Yoshihara , and the time-rule method as developed by Aureli et al.  and Aureli and van Schaik . For the PC/MC method, the minute during which former opponents initiated affiliative contact was compared in the PC and MC samples. Each PC-MC pair Counter Aggression and Reconciliation / 221 TABLE I. Sample Sizes for Responses to Aggression No. of subjects Response total Submit Hold ground Aggression & submit Aggression & no submit All All males M-M F-M All females M-F F-F 41 1858 1377 205 59 19 827 485 148 33 19 658 417 127 24 18 169 68 21 9 22 1031 892 57 26 22 637 554 33 20 21 394 338 24 6 217 161 90 71 56 30 26 The four responses to aggression include: 1) submit immediately, 2) hold ground and confront opponent, but neither submit nor respond with aggression, 3) respond with aggression and eventually submit, and 4) respond with aggression and never submit. Individual subjects were pooled and the total frequency of each response type is given. The order of each dyad represents the initiator and recipient of aggression, respectively. For example, F-M means a female initially attacked a male, and the male’s response is shown in the table. was summarized as attracted, dispersed, or neutral depending on whether affiliative contact between former opponents occurred earlier in the PC, earlier in the MC, or at the same time, respectively. If affiliative contact did not occur during one of the samples, the sample was assigned the maximum value. For each focal individual, let a be the number of attracted samples, let d be the number of dispersed samples, and let t be the total number of samples (including neutral samples). Then the conciliatory tendency equals (a – d)/t [Veenema et al., 1994]. Conciliatory tendency is a measure of reconciliation that takes into account baseline levels of affiliation, and is appropriate for within- and betweenspecies comparisons. A conciliatory tendency was calculated for each focal subject, regardless of who initiated reconciliation. For the time-rule method we recorded the minute during which former opponents initiated the first affiliative contact in the PC and MC samples. We compared the distribution of first affiliative contacts in the PC and MC samples using the Kolmogorov-Smirnov test. If this test was significant we performed a Wilcoxon matched-pairs test over the time period during which the PC and MC distributions differed. The Wilcoxon matched-pairs test accounted for individual variability, and ensured that the Kolmogorov-Smirnov test was not biased by the extreme behavior of a few individuals. We used the results of the time-rule analysis to make further comparisons. For example, the number of fights reconciled (according to the time rule) was compared for male–male and female– female opponents. Mann Whitney U-tests or Wilcoxon matched-pairs tests were used for statistical analysis. When sample sizes were too small for individual analysis, data were pooled and we used the z-test approximation to the binomial. All tests were two-tailed, and the significance level was P < .05. We had 208 PC-MC pairs when one sample was taken per fight, and 247 PC-MC pairs when multiple samples were included from polyadic fights. Compared to fights for which only one PCMC sample was taken (17.2%, SD = 10.9), including additional PC-MC samples from polyadic fights (16.7%, SD = 9.9) did not significantly alter the amount of affiliative contact between former opponents in the PC samples (Wilcoxon: Z (18) = 0.47, P >.05). Consequently, all 247 PC-MC pairs were used for analysis. Fights among the 41 mature and pubertal animals were distributed among 19 mature subjects. The number of PC-MC samples per subject ranged from 8 to 21. One aggressor–victim dyad was sampled 10 times, and these two subjects also had 222 / Cooper and Bernstein the most PC-MC samples. The sample sizes are shown for each subdivision of the data set in Table II. RESULTS Counter Aggression We observed 1,858 aggressive encounters. The victim of aggression submitted immediately in a mean of 72.5% of encounters (SD = 23.7, n = 41) (see Table I for the raw data). Victims held ground in a mean of 10.9% of encounters (SD = 10.9, n = 41), fought back but eventually submitted in a mean of 3.2% of encounters (SD = 3.2, n = 41), and fought back without submission in a mean of 13.4% of encounters (SD = 17.9, n = 41). The mean percentage of counter aggression was 16.6% (SD = 18.6, n = 41), i.e., 3.2% + 13.4%. Sex differences. Males were the victims of 827 aggressive episodes. They received aggression from males and females in 658 and 169 episodes, respectively. Females were the victims of 1,031 aggressive episodes, and they received aggression from males and females in 637 and 394 episodes, respectively. Figure 2 shows sex differences in the mean percentage of the four types of responses to aggression. Females submitted immediately in a mean of 87.5% of aggressive encounters (SD = 8.0, n = 22), while males submitted immediately in a mean of 55.1% of aggressive encounters (SD = 24.4, n = 19) (Mann-Whitney: U (21, 18) = 23.0, P < .001). Males held ground in a mean of 17.7% of encounters (SD = 12.6, n = 19), while females held ground in 5.0% (SD = 3.3, n = 22) (Mann-Whitney: U (21, 18) = 75.5, P < .001). Males also fought back against their opponent without submission in a mean of 23.1% of encounters (SD = 22.7, n = 19), while females fought back in 5.1% (SD = 4.2, n = 22) (Mann-Whitney: U (21, 18) = 40.0, P < .001). Overall, males showed more counter aggression (mean = 27.2%, SD = 22.2, n = 19) than did females (mean = 7.5%, SD = 5.6, n = 22) (Mann-Whitney: U (21, 18) = 37.0, P < .001). Figure 3 shows the responses to aggression for the four different sex-class combinations. Fights between males were more likely than fights between females to result in the victim holding ground and confronting its opponent, and the victim counter-attacking its opponent and winning (confront: male mean = 18.1%, SD = 14.4, n = 19, female mean = 5.2%, SD = 5.5, n = 21; Mann Whitney: U (20, 18) = 81.0, P = .001; counter aggression: male mean = 18.2%, SD = 23.5, n = 19, female mean = 5.5%, SD = 6.9, n = 21; Mann Whitney: U (20, 18) = 101.0, P < .01). Males, nonetheless, showed immediate submission in a mean of 60.1% of encounters with other males (SD = 27.0, n = 19). Males were more likely to submit when they were the victims of male aggression (mean = 60.1%, SD = 27.0, n = 19) than the victims of female aggression (mean = 35.6%, SD = 35.5, n = 18) (Wilcoxon: Z (17) = 2.58, TABLE II. Reconciliation Sample Sizes Type of PC-MC pairs No. of subjects No. of PC-MC pairs Mean (SD) Total Feeding Non-feeding Male-male dyads Female-female dyads Male-female dyads 19 19 19 9 10 17 247 125 122 106 58 83 13.0 (4.5) 6.6 (3.0) 6.4 (2.1) 11.8 (3.5) 5.8 (2.3) 4.9 (2.1) Two subjects did not have any fights with oppositely sexed opponents. The number of PC-MC pairs is pooled for all subjects. The mean represents the mean number of PC-MC samples per subject. Counter Aggression and Reconciliation / 223 Fig. 2. The mean percentage for each of the four types of responses is shown for male and female subjects (mean ± SD). P = .01). Likewise, counter aggression occurred more often when males were attacked by females (mean = 55.5%, SD = 39.0, n = 18) than by males (mean = 21.8%, SD = 23.5, n = 19) (Wilcoxon: Z (17) = 3.10, P < .01). In contrast to males, when females received aggression they were more likely to submit immediately than to show the other three types of responses combined, regardless of the sex of the opponent (aggression received from females (mean = 88.2%, SD = 11.9, n = 21): Wilcoxon: Z (20) = 4.03, P < .001; aggression received from males (mean = 87.4%, SD = 8.0, n = 22): Wilcoxon: Z (21) = 4.11, P < .001). 224 / Cooper and Bernstein Fig. 3. The mean percentage for each of the four types of responses to aggression is shown for different types of dyads (mean ± SD). The order of the dyad indicates the direction of the initial aggression; for example, female–male means female initiates aggression against male. Thus, the response is shown for the second member of the dyad. Reconciliation Demonstration of reconciliation. Former opponents were more likely to make affiliative contact during the PC sample (16.7%, SD = 9.9) than during the MC sample (5.8%, SD = 7.3) (Wilcoxon: Z (18) = 2.98, P < .01). The PC-MC method indicated that the proportion of attracted PC-MC pairs (15.7%, SD = 9.8) was greater than the proportion of dispersed PC-MC pairs (4.6%, SD = 6.1) (Wilcoxon: Z (18) = 3.0, P < .01). The overall mean conciliatory tendency was 11.2% (SD = 12.6, n = 19). The time-rule analysis indicated that former opponents contacted sooner after the conflict than they did during the control period, and that the maximum difference between the PC and MC distributions occurred within the first 2 min (Kolmogorov-Smirnov: D = .422, n = 19, P < .05). This result was also confirmed at the individual level. Subjects contacted former opponents more often in the Counter Aggression and Reconciliation / 225 first 2 min of the PC sample (10.1%, SD = 6.8) than in the first 2 min of the MC sample (0.9%, SD = 2.7) (Wilcoxon: Z (18) = 3.22, P < .01). For subsequent comparisons, any fight that was followed by affiliative contact between the former opponents in the first 2 min of the PC sample was considered reconciled. The PC-MC method yielded very similar results on the following comparisons, but for simplicity this is not shown. Selective attraction. Affiliation after a conflict was highly selective. Former opponents accounted for 20.0% (SD = 13.8) of affiliative partners in the PC samples, while they accounted for only 5.5% (SD = 8.5) during the MC samples (Wilcoxon: Z (18) = 2.62, P < .01). If the increased affiliation between former opponents after a fight was the result of a general tendency for opponents to act affiliatively, then there should have been no specific increase in affiliation between former opponents. Initiation of reconciliation. We investigated whether recipients of aggression initiated reconciliation more often than did original aggressors, and whether recipients of aggression did so more often than they initiated other affiliative interactions. The analysis was limited to those fights in which the initiation of aggression was unambiguous. Owing to the small sample size, individual data points were pooled together. Aggressors initiated reconciliation in 33 of 44 interactions, while victims initiated reconciliation in the remaining 11 interactions (P = .006, binomial probability). Former aggressors initiated affiliative contact during the MC sample in 11 of 14 interactions, while victims initiated affiliative contact in the remaining three interactions (P = .029, binomial probability). Recipients of aggression initiated reconciliation less often than aggressors, and recipients of aggression did not initiate affiliative contact proportionally more often than they did during baseline. Severity of aggression. We analyzed the proportion of fights reconciled according to the severity of aggression. Former opponents reconciled 12.9% of 116 fights involving a charge or chase, 11.8% of 76 fights involving manual contact aggression, and 8.5% of 47 fights involving biting. We categorized severe biting separately, and none of these eight fights were reconciled. Thus, the severity of aggression did not affect the frequency of reconciliation. Feeding and nonfeeding conditions. Fights that occurred when food was offered by (or taken from) priests, worshipers, or local people were compared to fights that occurred in other contexts, which we refer to as a nonfeeding context. The nonfeeding context included fights that occurred while foraging on natural foods, but these fights were considered different because of the intense and competitive situation created by provisioning. Animals reconciled 10.3% (SD = 9.0) of their fights in the nonfeeding context, and 10.1% (SD = 11.9) in the feeding context (Wilcoxon: Z (18) = .28, P > .05). The mean percentage of fights reconciled was not significantly different. Sex differences. Male–male, female–female, and male–female opponents reconciled 14.9% (SD = 6.2), 10.2% (SD = 14.6), and 5.1% of their fights (SD = 13.1) (Kruskal-Wallis: [χ2 (2, n = 19) = 9.52, P < .01]). A Mann-Whitney U-test indicated that male–male opponents reconciled more often than did male–female opponents. Male–female opponents rarely reconciled, but, interestingly, three of the four fights reconciled by opposite-sexed opponents involved the alpha male. DISCUSSION Counter Aggression The proportion of counter aggression in this study (16.6%) is considered low to moderate for macaques. Although this is the only data of its kind for this species, it suggests that Assamese macaques may have a relatively rigid domi- 226 / Cooper and Bernstein nance hierarchy. de Waal and Luttrell  found that the proportion of counter aggression in stumptail macaques (19%) was greater than in rhesus (9%), and interpreted this difference as evidence of relaxed dominance relationships in stumptails. More recently, counter aggression has been found to range from 1% of aggressive episodes in pigtail macaques (M. nemestrina) [Castles et al., 1996] to 56% in crested black macaques (M. nigra) [Petit et al., 1997]. On the other hand, intraspecies variation in counter aggression can be as large as the difference between stumptails and rhesus, e.g., counter aggression in Japanese macaques ranges from 12% [Chaffin et al., 1995] to 20% [Petit et al., 1997]. This highlights the caution necessary when making a species-wide generalization for Assamese macaques. Sex differences in counter aggression suggest considerable variation in dominance relationships in Assamese macaques. Males counter-attacked opponents in 27% of their encounters, while females did so in only 8% of their encounters. Males also held their ground in fights more often than did females, and this type of confrontation was most common in fights among males. These male–male conflicts were characterized by the dominant animal charging the subordinate, and the subordinate facing his opponent, backing up, and teeth chattering. The teeth chatter was not clearly a submissive signal. At times both the advancing and retreating males simultaneously teeth chattered. Teeth chattering was also used to enlist support [but see Aureli et al., 1994]. The tendency for males to face their opponent is similar to the pattern of aggression describe for stumptail macaque males. In stumptail macaques, males are wounded in their forequarters more frequently than are females [Whitten & Smith, 1984]. In this study, counter aggression occurred most frequently when females initiated aggression against males. Females often received aid from the alpha or beta male, but when they did not the male victim tended to counter attack. Dominance relationships for females therefore seemed much more strict than for males. Reconciliation This group of Assamese macaques had a conciliatory tendency of 11.2%, which is similar to the 9% in rhesus [de Waal & Ren, 1988; recalculated by Veenema et al., 1994], 12% in non-kin in Japanese macaques [Aureli et al., 1997], and 14% in non-kin in long-tailed macaques [Aureli et al., 1997]. This group of Assamese macaques, although not a member of the Fascicularis group, had a conciliatory tendency similar to species of that group. Their conciliatory tendency was far below that of stumptail, Barbary (M. sylvanus), liontailed (M. silenus), and crested black macaques (41% [de Waal & Ren, 1988; recalculated by Veenema et al., 1994]; 31% non-kin [Aureli et al., 1997]; 63% [Abegg et al., 1996]; and 25% [Petit et al., 1997], respectively). Reconciliation is a relatively good index of dominance style [Castles et al., 1996], and the low conciliatory tendency in this group of Assamese macaques suggests that they have a despotic dominance style. Consistent with this interpretation is the finding that the recipients of aggression initiated reconciliation less often than the initiators of aggression, a characteristic of species with a despotic dominance style [de Waal, 1993; Castles & Whiten, 1998a]. Additional research will be necessary to determine whether the conciliatory tendency found here is characteristic of the species. Although not statistically significant, intense bouts of feeding decreased affiliative contact among former opponents in the PC samples. Since fights in the feeding condition were compared to MC samples from the nonfeeding condition, we may have underestimated the level of reconciliation during feeding bouts. Counter Aggression and Reconciliation / 227 We are confident that fights during feeding bouts were reconciled at least as often as fights in other contexts. Castles and Whiten [1998a] found that fights over food were reconciled less often than were fights in other contexts, although they also found that fights over food produced the same amount of behavioral indicators of stress in the opponents as did fights in other contexts [Castles & Whiten, 1998b]. This is consistent with the hypothesis that fights over food damage social relationships, and that reconciliation is delayed while feeding takes priority. We agree that fights over food damage social relationships, but suggest that reconciliation may not always be delayed. Conditions such as the distribution of food, the proximity of foragers, and the duration of a feeding bout probably influence whether reconciliation is delayed. Phylogenetic and Socioecological Hypotheses The low-to-moderate level of counter aggression, and the low frequency of reconciliation observed in this study suggest that this group of Assamese macaques have a relatively despotic dominance style. A despotic dominance style does apply to female social relationships. Females reconciled their fights infrequently and rarely initiated counter aggression. Insofar as the correlation in macaques between dominance style and phylogeny is based on female social relationships, these data on Assamese macaques weaken that correlation. A phylogenetic analysis suggests that tolerant social relationships are the ancestral condition for macaques, and that only recently has the Fascicularis group shifted to despotic social relationships (Fig. 1) [Matsumura, 1999; Thierry et al., 2000]. Despotic dominance styles, however, can develop outside the Fascicularis group. For example, a despotic dominance style in pigtail macaques does not easily fit into a phylogenetic framework (Fig. 1). Intermediate dominance styles have been suggested to help account for patterns in species such as pigtail macaques [Thierry, 2000]. Assamese macaques may provide another example of the evolution of relatively despotic female social relationships outside of the Fascicularis group. Our knowledge of the role of habitat in shaping dominance style is still incomplete. Data on long-tailed macaques suggest that reconciliation in wild groups is similar to that in captive groups [Aureli, 1992; Aureli et al., 1989]. Certainly not all captive groups of macaques have despotic social relationships. So the immediate effect of clumped resources on dominance style is limited. The long-term effects of a temple environment, however, are unknown. We do not know when the Tukeswari Temple was constructed, but the religious site is hundreds of years old. The presence of Assamese and rhesus macaques in the area, and the Hindu custom of offering food to monkeys suggest that some monkeys have been receiving food at this location for many generations. How animals that have lived commensally with humans for generations differ from their wild counterparts might provide another means of investigating the socioecological hypothesis. At present, we cannot be sure whether the despotic social relationships in our study group are due to the temple habitat or are representative of the species. Systematic Variation The low-to-moderate proportion of counter aggression, and the low frequency of reconciliation observed in this study are consistent with the systematic variation hypothesis. Female social relationships in particular support the systematic variation hypothesis. Male–female social relationships, however, are not easily explained by dominance style. If affiliative and agonistic behavior covary at a dyadic 228 / Cooper and Bernstein level, then those dyads that reconcile frequently should have high levels of counter aggression. In Assamese macaques, when females initiated aggression against males, males counter-attacked females, and the sexes rarely reconciled. If counter aggression was due to social tolerance, then females were tolerant of males. Counter aggression between males and females probably reflects a power imbalance. Sexual dimorphism in macaques makes counter aggression against females a low-risk competitive strategy for males. In contrast, counter aggression against males would be a high-risk strategy for females. A similar power imbalance was observed in a group of Japanese macaques in which all seven observed episodes of counter aggression occurred when a subordinate female attacked an adult male [Chaffin et al., 1995]. In our group, females appeared to resist the presence of adult males in the group. Adult males remained in the group by both aggressive assertion and high rates of grooming directed toward females [Cooper & Bernstein, 2000]. In bonnet (M. radiata), Barbary, and stumptail macaques, males frequently affiliate and appear to have behavioral mechanisms that reduce social tension. [Silk, 1994; Paul et al., 1996; Estrada, 1984]. Assamese macaque males also appear to have tolerant social relationships. Males in this study reconciled their fights, and they also frequently groom and mount each other [Bernstein & Cooper, 1999; Cooper & Bernstein, 2000]. Hill  suggested that low sex ratios make affiliative relationships among males more likely. The fact that these four species live in groups with relatively even sex ratios may, in part, relate to the tolerant social relationships among males. In addition, the males in these species are single-mount ejaculators and may have reduced or delayed dispersal [Caldecott, 1986; Pusey & Packer, 1987; Moore, 1992]. The despotic social relationships observed among females, and the tolerant social relationships observed among males confuse species-wide generalizations about dominance style. Sex differences in affiliative and aggressive behavior need to be considered when categorizing species according to dominance style. For example, in Barbary macaques, males tend to ignore one another in competitive situations while females tend to compete directly for limited resources [Preuschoft et al., 1998]. The authors thought this tolerant response to aggression by males resulted from males attempting to avoid injury, rather than from males having strong social bonds. ACKNOWLEDGMENTS This research was conducted in cooperation with the Indo-U.S. Primate Project, principal investigator Professor S.M. Mohnot. A special thank-you goes to Arun Srivastava, Scientist-in-Charge of the Northeast Indo-U.S. team; Prabal Sarkar, who was working on his own dissertation on Assamese macaques at the Tukeswari temple; and Mohibul Haque, who helped us collect the data. We are indebted to the people of the Goalpara district, who cooperated with us in so many ways, officially and unofficially. We are also grateful to Filippo Aureli, Carolyn Ehardt, and three anonymous reviewers for their helpful comments on an earlier version of the manuscript. REFERENCES Abegg C, Thierry B, Kaumanns W. 1996. Reconciliation in three groups of lion tailed macaques. Int J Primatol 17:803– 816. Aureli F, van Schaik CP, van Hooff JARAM. 1989. Functional aspects of reconciliation among captive long-tailed macaques (Macaca fascicularis). Am J Primatol 19:39–51. Aureli F, van Schaik CP. 1991. Post-conflict behavior in long-tailed macaques (Macaca fascicularis): II. Coping with uncertainty. Ethology 89:101–114. Counter Aggression and Reconciliation / 229 Aureli F. 1992. Post-conflict behavior among wild long-tailed macaques (M. fascicularis). Behav Ecol Sociobiol 31:329–337. Aureli F, Das M, Verleur D, van Hooff JARAM. 1994. Postconflict social interactions among barbary macaques (Macaca sylvanus). Int J Primatol 15:471–485. Aureli F, Das M, Veenema HC. 1997. Differential kinship effect on reconciliation in three species of macaques (Macaca fascicularis, M. fuscata, and M. sylvanus). J Comp Psychol 111:91–99. Bernstein IS. 1981. Dominance: the baby and the bathwater. Behav Brain Sci 4:419–457. Bernstein IS, Ehardt CL. 1985. Intragroup agonistic behavior in rhesus monkeys (Macaca mulatta). Int J Primatol 6:209–226. Bernstein IS, Cooper MA. 1999. Dominance in Assamese macaques (Macaca assamensis). Am J Primatol 48:283–289. Bertrand M. 1969. The behavioral repertoire of the stumptail macaque. Basel: Karger. p 68–129. Caldecott JO. 1986. Mating patterns, societies and the ecogeography of macaques. Anim Behav 34:208–220. Castles DL, Whiten A. 1998a. Post-conflict behavior of wild olive baboons. I. Reconciliation, redirection, and consolation. Ethology 104:126–147. Castles DL, Whiten A. 1998b. Post-conflict behavior of wild olive baboons. II. Stress and self-directed behavior. Ethology 104: 148–160. Castles DL, Aureli F, de Waal FBM. 1996. Variation in conciliatory tendency and relationship quality across groups of pigtail macaques. Anim Behav 52:389–403. Chaffin CL, Friedlen K, de Waal FBM. 1995. Dominance style of Japanese macaques compared with rhesus and stumptail macaques. Am J Primatol 35:103–116. Cooper MA, Bernstein IS. 2000. Social grooming in assamese macaques (Macaca assamensis). Am J Primatol 50:77–85. Cords M, Aureli F. 2000. Reconciliation and relationship qualities. In: Aureli F, de Waal FBM, editors. Natural conflict resolution. Berkeley: University of California Press. p 177–198. Delson E. 1980. Fossil macaques, phyletic relationships and a scenario of development. In: Lindberg DG, editor. The macaques: Studies in ecology, behavior, and evolution. New York: Van Nostrand Reinhold. p 10–30. de Waal FBM, Yoshihara D. 1983. Reconciliation and redirected affection in rhesus monkeys. Behaviour 85:224–241. de Waal FBM, Ren R. 1988. Comparison of the reconciliation behavior of stumptail and rhesus macaques. Ethology 78:129–142. de Waal FBM, Luttrell LM. 1989. Toward a comparative socioecology of the genus Macaca: different dominance styles in rhesus and stumptail monkeys. Am J Primatol 19:83–109. de Waal FBM. 1993. Reconciliation among primates: a review of empirical evidence and unresolved issues. In: Mason WA, Mendoza SP, editors. Primate social conflict. Albany: State University of New York Press. p 111–144. Estrada A. 1984. Male-infant interactions among free-ranging stumptail macaques. In: Taub TM, editor. Primate paternalism. New York: Van Nostrand Reinhold. p 56–87. Hill DA. 1994. Affiliative behavior between adult males of the genus Macaca. Behaviour 130:293–308. Hill WCO, Bernstein IS. 1969. On the morphology, behavior and systematic status of the Assam macaque (Macaca assamensis McClelland, 1939). Primates 10:1–17. Matsumura S. 1996. Post-conflict contacts between former opponents among wild moor macaques (Macaca maurus). Am J Primatol 38:211–219. Matsumura S. 1998. Relaxed dominance relations among female moor macaques (Macaca maurus) in their natural habitat, South Sulawesi, Indonesia. Folia Primatol 69:346–356. Matsumura S. 1999. The evolution of egalitarian and despotic social systems among macaques. Primates 40:23–31. Moore J. 1992. Dispersal, nepotism, and primate social behavior. Int J Primatol 13: 361–378. Ogawa H. 1995. Bridging behavior and other affiliative interactions among male Tibetan macaques (Macaca thibetana). Int J Primatol 16:707–729. Paul A, Kuester J, Arnemann J. (1996). The sociobiology of male-infant interactions in Barbary macaques, Macaca sylvanus. Anim Behav 51:155–170. Petit O, Abegg C, Thierry B. 1997. A comparative study of aggression and conciliation in three cercopithecine monkeys (M. fuscata, M. nigra, P. papio). Behaviour 134:415–432. Preuschoft S, Paul A, Kuester J. 1998. Dominance styles of female and male Barbary macaques (Macaca sylvanus). Behaviour 135:731–755. Preuschoft S, van Schaik CP. 2000. Dominance and communication: conflict management in various social settings. In: Aureli F, de Waal FBM, editors. Natural conflict resolution. Berkeley: University of California Press. p 77–105. Pusey AE, Packer C. 1987. Dispersal and phylopatry. In: Smuts BB, Cheney DL, 230 / Cooper and Bernstein Seyfarth RM, Wrangham RW, Struhsaker TT, editors. Primate societies. Chicago: University of Chicago Press. p 250–266. Silk JB, Samuels A, Rodman PS. 1981. Hierarchical organization of female Macaca radiata in captivity. Primates 22:84–95. Silk JB, Samuels A. 1984. Triadic interactions among Macaca radiata: passports and buffers. Am J Primatol 6:373–376. Silk JB. 1994. Social relationships of male bonnet macaques: male bonding in a matrilineal society. Behaviour 130:271–291. Sterck EHM, Watts DP, van Schaik CP. 1997. The evolution of female social relationships in nonhuman primates. Behav Ecol Sociobiol 41:291–309. Taub DM. 1980. Testing the “agonistic buffering” hypothesis. Behav Ecol Sociobiol 6:187–197. Thierry B. 1985. Patterns of agonistic interactions in three species of macaque (Macaca mulatta, M. fascicularis, M. tonkeana). Aggres Behav 11:223–233. Thierry B. 2000. Covariation of conflict management patterns across macaque species. In: Aureli F, de Waal FBM, editors. Natural conflict resolution. Berkeley: University of California Press. p 106–128. Thierry B, Iwanink AN, Pellis SM. 2000. The influence of phylogeny on the social behavior of macaques (Primates: Ceropithecidae, genus Macaca). Ethology 106: 713–728. van Schaik CP. 1989. The ecology of social relationships amongst female primates. In: Standen V, Foley R, editors. Comparative socioecology: the behavioural ecology of humans and other mammals. Boston: Blackwell. p 105–218. Veenema HC, Das M, Aureli F. 1994. Methodological improvements for the study of reconciliation. Behav Process 31:29–38. Whitten PL, Smith EO. 1984. Patterns of wounding in stumptail macaques (M. arctoides). Primates 25:326–336. Zhao QK. 1996. Male–infant–male interactions in Tibetan macaques. Primates 37: 135–143.