Aspects of fight interference in free-ranging and compound-dwelling rhesus macaques (Macaca mulatta).код для вставкиСкачать
American Journal of Primatology 12:287-298 (1987) Aspects of Fight Interference in Free-Ranging and Compound-Dwelling Rhesus Macaques (Macaca rnufatta) JAY R KAPLAN,' DENNIS K CHIKAZAWA,' AND STEPHEN B MANUCK3 'Bowman Gray School of Medicine of Wake Forest University, Winston-Salem, North Carolina, 'Yerkes Regional Primate Research Center, Emory University, Atlanta, Georgia, 3Un1versity of Pittsburgh, Pittsburgh, Pennsylvania Pat terns of fight interference (agonistic aiding) were compared among three groups of rhesus monkeys (Macaca mulatta) living in two settings: 1)two groups a t Cay0 Santiago (Caribbean Primate Center); and 2) one group a t the Yerkes Regional Primate Research Center (YRPRC). A total of 1,227 interference episodes were recorded in 1,650 hours of observation. The only significant intergroup difference was the increased tendency of males at YRPRC to aid aggressors rather than victims. Among other findings, females aided relatives, interfered against target animals dominant to themselves, aided juveniles, and aided victims more consistently and frequently than did males. Importantly, female interference became more male-like in pattern when aid was given to nonrelatives. Neither the dominant males nor males in general displayed a unique or consistent tendency to interfere in fights in a manner which could be interpreted as controlling aggression. The males' interference patterns also did not suggest they were forming coalitions to either attain or defend status rankings. It is concluded that, overall, observations of compound-dwelling and free-ranging rhesus monkeys reveal similar relationships. Further, while female rhesus monkeys interfered in fights in a manner consistent with the control of aggression and protection of kin, the motives of male interferers remain unknown; however, their behavior is consistent with the hypothesis that they were reducing intermale tensions while, a t the same time, minimizing physical risk. Key words: interference, agonistic aiding, rhesus macaques, primates, Macaca mulatta INTRODUCTION Social behavior within a species is believed to vary as a function of ecological setting or demographic conditions [eg, Gartlan & Brain, 19681; however, the degree and significance of this variability is difficult to gauge as the relevant data are most often collected by different investigators, with little uniformity in collection tech- Received Marc:h 5, 1986; revision accepted November 2, 1986. Address reprint requests t o Jay R. Kaplan, Ph.D., Bowman Gray School of Medicine, Department of Comparative Medicine, 300 S. Hawthorne Road, Winston-Salem, NC 17103. 0 1987 Alan R. Liss. Inc. 288 I Kaplan, Chikazawa, and Manuck niques or methods of analysis. Additionally, investigators typically examine only a single group within any given condition or setting, making it difficult to establish the extent of variability that reasonably may be expected within a location. It seemed especially important to us to examine variability in the occurrence of fight interference (also termed agonistic aiding or third party interference [Bernstein & Ehardt, 198511, as this behavior has been utilized to evaluate social bonding networks within groups and also to test hypotheses concerning sociobiology, dominance rank, and kinship [Trivers, 1971; West-Eberhard, 1975; Kurland, 1977; Kaplan, 1978; Bernstein & Ehardt, 19851. In the current study, aspects of fight interference are reported for three social groups of rhesus monkeys (Mucaca mulatta) living in two disparate conditions: a) two groups free-ranging at the Cay0 Santiago (CS) station of the Caribbean Primate Center; and b) one group dwelling in a 38 m x 38 m compound a t the Yerkes Regional Primate Research Center (YRPRC). The data were collected and analyzed utilizing identical techniques. One objective was to determine the extent and nature of variability in interference behavior, especially the manner in which compound-dwelling monkeys differed from their free-ranging counterparts. A further objective was to determine the robustness of fight interference patterns reported previously a t both CS and YRPRC. Such observations showed, for example, that males and females differed significantly in their patterns of aiding, with females aiding kin, and with the beneficiaries of such aiding often being victims of aggression [Kaplan, 1977, 1978; Bernstein & Ehardt, 19851. The current data were used also to consider two unresolved issues relating to interference behavior in macaques: a) the motivation of males to interfere, especially when the interference is on behalf of aggressors; and b) the existence of a “control role,” occupied by dominant males. A control role has been reported in some, but not all, groups of rhesus monkeys as well as in other macaque species. METHODS Study Sites, Animals and Data Collection The study animals consisted of groups M (termed “Cl” in the current study) and L (“CB”) from CS, and group R-9 (“Yl”) a t the YRPRC (Table I). Behavioral observations were made by one investigator (JRK) on group C1 from March, 1974 through July, 1975 (950 hours) and on group C2 from January, 1974 through July, 1974 (450 hours). Observations were made by JRK on group Y1 in July and August, 1977. These observations were continued by another investigator (DKC) from September, 1977 through February, 1978 (250 hours, total). All of these groups had been monitored previously by a number of other investigators, hence their social histories were well known [Sade et al, 1977; Chikazawa et al, 19791. Notes by JRK were taken with ad libitum sampling [Altmann, 19741 and described, in detail, all interactions occurring between individual monkeys, between groups and subgroups, and relationships between monkeys and their nonsocial environments [Kaplan, 1976, 19771. Notes by DKC were taken in abbreviated diary fashion, with the emphasis on grooming and fight interference interactions. It is recognized that the ad libitum sampling procedure does not provide a random sample of the rate of occurrence of interference or any other specific behavior; however, it is assumed that the pattern of interference observed on the part of each animal represented a random sample of all interferences performed by that animal [cf Hausfater, 19751. All notes were tabulated by JRK using the same criteria for identifying the occurrence of interference episodes. Specifically, interference was defined as a n episode in which a monkey entered a n ongoing or just-terminated agonistic bout (fight) that began as a dyadic encounter between two other monkeys [Kaplan, 1977; see Sade, 1967, 1972 for definitions of “fight,” “episode,” and “bout”]. An earlier Fight Interference in Rhesus Macaques I 289 TABLE I. AgelGender Composition of Groups Y1, C1, and C2 Group Y1 c1 c2 Infants-Juveniles (< 3Yr) Males Fema1es 18 17 14 17 10 8 Adolescents-Adults ( 2 3Yr) Males Females 16 12 17 22 15 24 study had shown that, in rhesus monkeys, the interfering animal almost always aids one of the original combatants against the other [Kaplan, 1974, 19761. Thus, three roles are represented in each interference episode: a) the interferer; b) the beneficiary of the interference; and c) the target. Additionally, beneficiaries can be distinguished by their identification as victim or aggressor in the original fight. Comparison of behavioral data taken by JRK and DKC on group Y1 revealed no significant differences in patterns of fight interference, grooming, or dominance [Kaplan & Chikazawa, 19781. Ana1y ses Previously published analyses of rhesus monkey interference behavior have been based on the summation of episode frequencies within classes of animals (eg, all males, all females, all adults) [Kaplan, 1977, 1978; Bernstein & Ehardt, 19851. This is because the data were collected ad libitum, making true rates of interaction impossible to determine. However, such analyses have the potential to distort the contributions of animals which interfere frequently, if such animals are acting idiosyncratically. Further, the use of summed frequencies generally limits analyses to nonparametric techniques (C‘Z” tests or repeated x2). Data in the current study also were collected ad libitum. However, we sought to avoid potential problems in the use of frequency data summed within classes by creating for each animal a series of four indices based on the percentages of their interference episodes which were directed toward different categories of beneficiaries or targets: a) the percentage of interference episodes in which the beneficiary was a victim in the original fight; b) the percentage of episodes in which the interferer was dominant to the target animal; c) the percentage of episodes in which (for natal animals only) the interferer was related to the target animal; and d) the percentage of episodes in which infant-juveniles were aided. These calculations reflect the four factors reported previously as significant determinants of interference patterns: a) role of the beneficiary (victim or aggressor in the original fight); b) status of the interferer relative to the target animal (dominant or subordinate);c) the genealogical relationship between the interferer and the beneficiary (related or not related); and d) age of the beneficiary (infant-juvenile, < 3 yr; adolescent-adult, 2 3 yr) [Kaplan, 1977, 19781. To be included in the primary analyses, an animal had to be an adolescent-adult and had to have interfered at least five times. The data from all animals which met these criteria were then included with equal weight in a series of analyses of variance (ANOVA), comparing male and female interferers within and between groups. Analyses of covariance were then applied to the data in order t o demonstrate that the patterns revealed by the ANOVAs were not influenced by the differing number of episodes contributed by individual monkeys. In these latter analyses, the number of interference episodes performed by each individual (range 5-49) was used as the covariate. None of the covariance analyses resulted in alteration of the original ANOVA results; the covariance analyses, thus, are not reported in the text. 290 I Kaplan, Chikazawa, and Manuck In addition to the above analyses, post-hoc tests and nonparametric evaluations were performed as needed. The inclusion of data from adults only provided for a relatively homogeneous data set, and the inclusion criterion of five episodes provided a reasonable lower limit for representation in the study. Males natal and non-natal to the social groups were considered together for all analyses except the relative vs nonrelative comparisons. Of the 1,227 recorded episodes of interference, 80% were performed by animals meeting the above inclusion criteria, 6% were performed by adults with less than five interferences each, and 14% were performed by juveniles. Of the 106 adults in the three groups, 60% contributed data to the analyses; these animals performed a n average of 15.76 ( & 1.30 SEM) interferences. While the primary analyses were performed on the percentage indices of interference, parallel x2 analyses were applied to the total adult frequency data in order to demonstrate the robust nature of the results and their comparability to previously published results. All tests of significance were two-tailed. RESULTS Genealogical Relationship Between Interferers and Beneficiaries The mean percentage of episodes in which males and females (natal only) in each group aided relatives is shown in Table 11. Females interfered on behalf of relatives considerably more frequently than did males. A Gender male,female x Group C ~ , C B , YANOVA ~ applied to the data revealed a significant main effect for Gender (F[1,52] = 53.44, P < 0.001). Neither the Group main effect nor the Group x Gender interaction was significant. Because the data violated somewhat the parametric assumption regarding homogeneity of variance, additional analyses (t-tests)were performed on the data from each group separately. These analyses showed, also, that males interfered on behalf of relatives significantly less than did females (Cl: t = 3.04, P < 0.01; C2: t = 3.68, P < 0.01; Y1: t = 6.43, P < 0.001). A comparison of the percentage of aid given by females to relatives revealed no differences among the three groups (F[12,41] = 0.63, NS). The rather consistent percentage of aid given to relatives by females was, despite the increased opportunities for relative aid among the animals from C1, where the percentage of dyads containing relatives was higher than in the other two groups (Cl: 22% dyads related; C2: 15% dyads related; Y1: 12% dyads related). Role of the Beneficiary The percentage of episodes in which male and female interferers in each group aided victims (rather than attackers) are shown in Table 111. A Gender x Group ANOVA applied to these data revealed main effects for Group (F[2,57] = 7.83, P < 0.01) and Gender (F[1,57] = 14.68, P < 0.01), as well as for the Group x Gender interaction (F[2,57] = 4.71, P < 0.02). An initial interpretation of these data is that TABLE 11. Percentage of Aid Given by Natal Males and Females to Relatives (+ SD)* Gender c1 Grom c2 Males 56 (15) n=4 85 (17) n = 15 n=4 83 (20) n = 11 Females 35 (30) Y1 38 (10) n=6 78 (14) n = 18 *ANOVA: Gender, F[1,52] = 53.4, P < 0.001; Group, F[2,52] = 2.21, P > 0.10; Group x Gender, F[2,52] = 1.09, P > 0.20. Fight Interference in Rhesus Macaques / 291 TABLE 111. Percentage of Aid ( & SD) Given by Interferers to Victims of Aggression in Groups Cl, C2, and Y1* Group Gender c1 Males 68 (16) n=8 62 (30) 30 (19) n=5 n=6 76 (16) n = 15 72 (22) n = 11 70 (14) n = 18 Females Y1 c2 *ANOVA: Gender, F[1,57] = 14.68, P < 0.001;Group, F[2,57] = 7.83, P < 0.001; Group x Gender, F[2,57] = 4.71, P = 0.01. while animals in general aided victims of aggression (67% of instances), females were more likely than males to aid victims. Further, there was variability across groups, with[ the animals in group Y1 significantly less likely than other animals to aid victims of aggression. Finally, there was a more marked male-female difference in group Y l than in the other groups. Subsequent contrasts among group means a t P < 0.05 [Kirk, 19681 suggests some qualification of this interpretation. The propensity of females to interfere on behalf of victims was significantly greater than that of males only in group Y1. This finding suggests, in turn, that most of the observed ANOVA effects resided in the Y1 males (ie, if these males were similar to the C1 and C2 males, the overall male-female difference probably would have been of only marginal significance). Given the significant influence of kinship on female interference, we tried to determine whether the percentage of aid given by females to victims might have been conditioned, in part, by genealogical relationships. Here, the analysis was limited to animals which interfered more than five times on behalf of both relatives and nonrelatives (n = 9 males, n = 6 females). It was found that when interfering on behalf of' relatives, females were significantly more likely to aid victims than were males (females: = 67% (+ 22 [SD]) aid to victims; males: =43% (+ 15); t = 2.15, P : < 0.05). However, when these same animals aided nonrelatives, the percentage of aid by females and males to victims could not be distinguished (females: X = 53% (+ 18);males; X = 53% (+ 15); t = 0, NS). x x Age of Beneficiary The percentages of aid given by adolescent-adult interferers to infant-juveniles are shown in Table IV.A Gender X Group ANOVA applied to these data revealed only a signiificant Gender effect (F[1,57] = 10.64, P < 0.01). Mean values indicated that females were significantly more likely to aid infant-adolescents than were males. While about half of the animals in groups Y1 and C1 were infant-juveniles (Table I), this was true for only 35% of the animals in group C2. Despite this variability, however, there were no group differences in propensity to aid infantjuveniles as opposed to adolescent-adults. As with aid to victims, it was hypothesized that genealogical associations might have influenced the relatively high amount of aid given by females to infant-juveniles. Again, the analysis was limited to those animals interfering on behalf of relatives and nonrelatives a t least five times (n = 9 males, n = 6 females). When interfering on behalf of relatives, females helped juveniles mlore than did males (females: X = 50% ( 5 26) aid to infant-juveniles; males: X = 23 (k 23); t = 2.24, P < 0.05). However, males and females could not be distinguished when aiding nonrelatives (females: X = 27% (+ 17); males: = 28% (+ 281, NS). x 292 I Kaplan, Chikazawa, and Manuck TABLE IV. Percentage of Aid Given by Adolescent-Adult Male and Female Interferers to Infant-Juveniles* c1 Gender Males Fema1es *ANOVA: Gender, F[1,53] = F[2,57] = 0.55, P > 0.20. Group c2 Y1 37 (20) 31 (19) n=8 n=5 58 (21) 41 (21) n = 15 n = 11 10.6, P < 0.01; Group, F[2,57] = 1.91, P > 0.10; Group x 25 (14) n=6 50 (21) n = 18 Gender, TABLE V. Percentage of Episodes (& SD) in Which Interferers Were Subordinate to Target Animals Gender c1 Group c2 Males 2 (6) n=8 38 (19) n = 15 3 (8) n=5 46 (18) n = 11 Females Y1 0 (0) n=6 46 (27) n = 18 Status of the Interferer (Dominant or Subordinate to the Target) The percentages with which males and females of each group interfered against animals which were dominant to them are shown in Table V. Inspection of this table reveals that males rarely interfered in fights if they were subordinate to the target, while females were often subordinate to the target. Because of the depressed variability among the males (with means and SDs close to 01, a series of Mann-Whitney U tests (two-tailed) were applied to the data, comparing males and females within each group. These analyses revealed that the females in all three groups interfered in fights in which they were subordinate to the target animal significantly more frequently than did males (all P values < 0.01). WhiIe males were similar across groups in not interfering in fights if they were subordinate to target animals, females were similar in the tendency for about 40% of their interferences to be performed while subordinate to the target animal. An ANOVA applied to the data involving females indicated no significant differences between groups (F[2,41] = 0.66, NS). Although females showed a slight tendency to interfere more frequently when subordinate to the target if they were related to the beneficiary than if not, the difference was not significant (females [n = 61 aiding relatives: = 37% (+ 23) subordinate to target; females aiding nonrelatives: = 30%(+ 24); t = 0.77, NS). x x Dominant Males and the “Control Role” Concept We sought to determine whether the dominant males may have, more than other animals, helped to “control” aggression in their groups. This was done by evaluating the extent to which the two highest ranking males in each group aided victims, rather than aggressors, when they interfered. Additionally, we examined the pattern with which these dominant males occupied the target and beneficiary roles in interference episodes. There was considerable intergroup variability in the degree to which dominant males aided victims rather than aggressors; within groups, however, the dominant Fight Interference in Rhesus Macaques I 293 TABLE VI. Interference Pattern of the Two Most Dominant Males in Each Group Percentage aid to No. as Percentage as No. as Percentage as Group Male interferer victims target aggressor beneficiary aggressor No. as C1 M1" M2" C2 Mlb Y1 M2" Mlb M2b 37 38 14 7 22 23 76 82 64 46 36 26 15 31 14 33 14 21 100 90 100 100 100 86 7 4 9 16 2 7 100 75 89 94 100 71 'Nonnatal to the group. bNatal to the group. males appeared comparable (Table VI). Thus, for example, the dominant males in Y 1 both aided victims less than aggressors and differed significantly from females in this pattern (cf Table 111)(P < 0.01, Mann-Whitney U). On the other hand, the dominant males in C1 aided victims more than aggressors and did not differ from females in t h i s pattern (P > 0.20, Mann-Whitney U). The two most dominant males from C2 aided victims and aggressors approximately equivalently and, although their percentage values were below the mean percentage of aid given to victims by females, the difference was not significant (0.20 > P > 0.10, Mann-Whitney U). Importantly, further analyses showed that the two highest ranking males in each group could not be differentiated from other males in their groups in the extent to which victimis or aggressors were aided (all P > 0.20, Mann-Whitney U); hence, in terms of their pattern of aid to aggressors and victims, dominant males were not uniform in their "control" of aggression, nor were they unique and distinct from other animals in this aspect of interference. Further evaluation of the data (Table VI) indicates that these six dominant males were themselves the targets of interference (almost always as aggressors) approximately one and a half times a s frequently as they helped victims. Also, while dominant males were sometimes the beneficiaries of interference behavior, this was almost always when they were aggressors rather than victims. In fact, the data show that dominant males in all groups either aided aggressors or were involved in interference episodes a s aggressors (while targets or beneficiaries) almost three times as frequently as they aided victims. Evaluation of Frequency of Interference by Subclasses Table V[I shows the summed frequency of interference a s performed by all adolescent-adult animals in each of the three social groups (including the 69 episodes performed bj7 the 43 monkeys which interfered less than five times each). A series of chi-square tests applied to these data revealed results almost identical to those obtained above with the use of the percentage indices. Thus, for example, males in general were less likely than females to aid victims of aggression (x2 = 32.0, P < 0.01), and the animals of group Y1 were less likely than were the animals of other groups to aid victims of aggression (x2 = 19.2, P < 0.01).Further analysis shows that females interfered more on behalf of victims than did males in groups C2 (x2 = 5.94, P < 0.03) and Y1 (x2 = 45.6, P < 0.011, but not in group C1 (x2 < 1.0, NS). Similar analyses showed that males in all groups interfered in fights only when dominant to the target animal (x2 = 18.4, P < 0.011, that males were less likely to aid relatives than were females (x2 = 124.9, P < 0.011, and that males showed a lesser tendency to aid infant-juveniles than did females (x2 = 34.9, P < 0.01). The 294 I Kaplan, Chikazawa, and Manuck TABLE VII. Summed Frequency of All Interference Performed by Adolescent-Adult C1. C2, C2. and Y1 Animals of C1, Group c2 c1 ~~ Y1 Male Female Male Female Male Female Aid victim Aid attacker Dominant 118 52 234 87 24 21 99 37 40 82 176 77 to target Subordinate to target Aid relative Aid nonrelative Aid juvenile Aid adult 168 201 44 72 121 140 2 45 30 61 109 120 271 50 179 142 1 8 20 14 31 64 114 22 53 82 1 39 68 28 93 113 200 53 126 127 ANOVAs reported above (on the percentage indices) allowed for a somewhat more precise description of these effects. DISCUSSION Intergroup Variability The three rhesus monkey study groups (Cl, C2, and Y1) differed in a number of ways, each of which could have influenced the expression of interference behavior. Most obviously, groups C1 and C2 were free-ranging, with animals able to avoid unwanted interactions with other animals through the use of cover or through timely departures from locations of conflict. On the other hand, the monkeys of Y 1 were in relative proximity to each other on a continuous basis. Additionally, the spontaneous migration of animals (especially males) into and out of social groups which occurred at CS [Kaplan, 19761 was precluded in Y1. As a result, the ratio of natal to non-natal males was greater in Y1 than in C1 or C2. Finally, many of the Y 1 males were, as adults, subordinate to many of the adult females. In groups C1 and C2, on the other hand, most adult males were non-natal and most were dominant to most females. These are the kinds of differences that sometimes are suggested as capable of causing signficant alterations in patterns of social behavior [Rowell, 19721. Interference behavior was relatively uniform in pattern of occurrence, given the demographic and ecologic variability among these groups. Females were especially consistent in aiding relatives and victims, while males were particularly consistent in aiding when dominant to the target and in aiding older, rather than younger, animals. Although there was behavioral variability among groups, the only significant difference in interference behavior related to the increased likelihood of Y 1 males (including the dominant males) to aid attackers rather than victims. It is tempting to suggest that such a difference resulted from the increased visibility present in the corral situation and the greater willingness of males under these conditions to join in attacks on victims rather than to defend such animals (see below). Given that this was the only significant difference among groups, it can be concluded that variability in interference behavior between groups living under ecologically disparate conditions, as well as between groups in a single location, is relatively minimal. Fight Interference in Rhesus Macaques I 295 Comparative Studies The data presented in this report go beyond providing a comparison of conspecific rhesus monkey groups living in free-ranging and compound environments. The data offer also the opportunity to evaluate the robustness of previously reported findings on rhesus monkeys, particularly those from CS and YRPRC. For example, data from CS and YRPRC indicate that aid is more often given to victims than aggressors; however, male and female patterns of interference were reported to be different, with females giving more aid overall, and aiding victims and relatives and interfering as subordinate to the target more frequently than males [Kaplan, 1977, 1978; Bernstein & Ehardt, 19851. On the other hand, males appeared to aid a t little risk to themselves (ie, when dominant to the target), and did not seem to “control” intergroup aggression, as has sometimes been suggested for other species of macaques, such as M. fuscata [Tokuda & Jensen, 1968; Watanabe, 19791 and M. nemestrina [Oswald & Erwin, 19761. The data reported in the current manuscript are largely consistent with these previous results. Thus, in C1, C2, and Y1, victims, overall, were aided significantly more often than aggressors. Additionally, four male-female differences consistent with earlier findings were observed among the three groups: a) females were more active than males in aiding relatives; b) females were more active than males in aiding against target animals dominant to themselves; c) females were more likely than males to aid infant-juveniles; and d) overall, females were somewhat more likely to aid victims of aggression than were males, though this effect was appreciable (and indeed statistically significant) only in group Y 1. Finally, the dominant males failed to show any uniform or unique tendency to occupy a control role (protecting victims in fights). Significantly, the above patterns of interference were present whether the data were evaluated on the basis of summed frequencies within subclasses, or on the basis of equivalent individual contributions (the percentage indices). As a whole, the data from CS and YRPRC suggest that the interference patterns of male and fismale rhesus monkeys are distinct. Further, these patterns are consistent with the hypothesis that a major activity of (adolescent-adult) female rhesus macaques is to control aggression, especially as it affects their kin, while males interfere in ai manner which minimizes their risk (because they almost never aid when they are subordinate to the target animals) and which does not relate specifically to the control of aggression (because relatively often they aid aggressors rather than victims:l. Similar male-female differences have been reported in pigtail and cynomolgus macaques (M. nemestrina, M. fasicularis) [Massey, 1977; de Waal, 19771. Marked gender differences in aiding patterns exist also among chimpanzees (Pan troglodytes) [de Waal, 1984; Nishida, 19831. It is worth noting that the gender differences observed a t CS and YRPRC may have been dependent, a t least in part, on genealogical relationships. Thus, it was found that when females aided nonrelatives, they dild so in a pattern that was much more male-like than when they aided relatives (males did not change their pattern of aiding as a function of genealogical associations). Based on such a finding it might be speculated that the interference patterns of females living in groups not characterized by genealogical and long-term peer associations would converge strongly with those observed among males. In fact, a recent report describes the interference behavior within artificially formed groups of adult female cynomolgus macaques as being male-like [Kaplan & Manuck, 19851. Unresolved Issues The current data leave several issues unresolved. The pattern and motivation of male interference, in particular, is problematic. For example, data in the current 296 1 Kaplan, Chikazawa, and Manuck and previous studies show that males a t CS aid victims of aggression more frequently than aggressors, while males at the YRPRC consistently aid aggressors more than victims [Kaplan, 1977, 1978; Datta, 1983; Bernstein & Ehardt, 19851. De Waal  suggested that the tendency to aid aggressors under confined circumstances might be related to the increased degree of tension presumably existing between high ranking males forced to live in close proximity; aiding each other against lower ranking animals (“scapegoating”) may help release such tensions. Alternatively, or in addition, we would suggest that the increased visibility in closely confined conditions (such as experienced by the males of group Y1) may cause males to more completely reduce risk to themselves by only becoming involved in those fights in which the target is already losing (ie, is a victim). Importantly, it does not appear that these rhesus males often use interference behavior as a means of raising their ranks relative to the targets against which they are interfering, since they are almost always already dominant to the targets. Whether some of the interference episodes of males represent coalitions helping to maintain status distinctions, such as has been suggested for baboons [Packer, 19771, awaits the collection of more detailed data. A further unresolved issue relates to the control role. Numerous authors have contended that dominant macaque males play a control role by breaking up fights and punishing aggressors [Bernstein & Sharpe, 1966; Kaufmann, 1967; Kurland, 1977; Watanabe, 19791. The data reported above do not support, among rhesus monkeys, the existence of a control role consistently filled by dominant males. An examination of the criteria used to evaluate the presence of the control role may, in part, resolve the issue. Usually, a major criterion for the presence of a control role is whether aid is given to victims rather than agressors [de Waal, 1977; Kurland, 1977; Watanabe, 19791. A less frequently used criterion is whether agonism (overall, or within specific agelsex categories) increases with the removal of the supposed control animal [Tokuda & Jensen, 1968; Oswald & Erwin, 1976; Erwin, 19791. Unfortunately, there appears to be considerable inter- and intra-species variability in both of these phenomena. However, if fight outcomes are carefully evaluated, a more consistent result may emerge. For example, de Waal  showed that although the dominant male of each of his small cynomolgus macaque groups was active in helping victims of aggression, the fight interference of the dominant males had three characteristics: a) these males were not more effective than other animals in stopping aggression; b) they were no more protective than other animals; and most importantly c) their participation in fights tended (whether helping victims or aggressors) to increase the overall level of aggression. This suggests that where dominant males are reported to occupy a control role, the effect may be more apparent than real. Further diminishing the potential importance of direct male interference behavior is its relatively low frequency in comparison to that of females [Bernstein & Ehardt, 19851. Clearly, additional, detailed studies are needed to resolve this issue. CONCLUSIONS 1. Rhesus monkeys displayed a relatively consistent pattern of interference behavior, both within and across ecological settings. Females were particularly consistent in aiding victims of aggression and relatives, while males were most consistent in aiding older animals and in aiding when dominant to the target. 2. Compound-dwelling rhesus males were much more likely to aid aggressors than were their free-ranging counterparts. This appeared to be the most significant effect of increased confinement on interference behavior. 3. Rhesus monkey females differed from males in that they were more likely than males to aid relatives and juveniles, to aid against target animals dominant to Fight Interference in Rhesus Macaques I 297 themselves, and, overall, to aid victims. However, when not aiding relatives, females tended to be more male-like in their interference patterns. 4. Rhesus monkey males interfered at little risk to themselves, and may have interfered in a way (eg, “scapegoating”) which reduced social tension among themselves. 5. There was little evidence to support the existence of a unique “control” role for dominant rhesus monkey males. These males either interfered in a way which exacerbated rather than controlled aggression, or they interfered in a manner similar to that of other males or females. ACKNOWLEDGMENTS This work was supported in part by grants from the NSF (6s 2377), NIH (HL R01 26561, IIL 14164, RR 00165), NIMH (MH-204831, and the University of Alabama in Birmingham (faculty research grant to JRK). REFERENCES Altmann, J. Observational study of behavior: Sampling methods. BEHAVIOUR 48:l-41, 1974. Bernstein, IS.; Sharpe, L.G. Social roles in a rhesus monkey group. BEHAVIOUR 26: 91104, 1966. Bernstein, I.S.; Ehardt, C.L. Agonistic aiding: Kinship, rank, age, and sex influences. AMERICAN JOURNAL OF PRIMATOLOGY 8:37-52,1985. Chikazawa, D.; Gordon, T.P.; Bean, C.A.; Bernstein, IS. Mother-daughter reversals in rhesus monkeys (Macaca mulatta). PRIMATES 20:301-304, 1979. Datta, S.B. Patterns of agonistic interference. Pp 289-297 in PRIMATE SOCIAL RELATIONSHIPS: AN INTEGRATED APPROACH. R.A. Hinde, ed. Sunderland, Massachusetts, Sinauer Associates, 1983. de Waal, F.B.:M. Organization of agonistic relations within two captive groups of Java monkeys (2Clacaca fascicularis). ZEITSCHRIFT :FUR TIERPSYCHOLOGIE 44:225-282, 1977. de Waal, F.B.M. Sex differences in the formation of co,alitions among chimpanzees. ETHOLOGY .AND SOCIOBIOLOGY5:239255, 1984. Erwin, J. Agression in captive macaques: Interaction of social and spatial factors. Pp 139-171 in CAPTIVITY AND BEHAVIOR: PRIMATES IN BREEDING COLONIES, LABORATOFLIES, AND ZOOS. J. Erwin, T.L. Maple, G. Mitchell, eds. New York, Van Nostrand Reinhold, 1979. Gartlan, J.S.; Brain, C.K. Ecology and social variability in Cercopithecus aethiops and C. mitis. Pp 253--292 in PRIMATES, STUDIES IN ADAPTATION AND VARIABILITY. P.C. Jay, ed. New York, Holt, Rinehart 8z Winston, 1968. Hausfater, G. Dominance and reproduction in baboons (Papio cynocephalus): A quantitative analysis. CONTRIBUTIONS TO PRIMATOLOGY, No. 7. Basel, S. Karger, 1975. Kaplan, J. Patterns of fight interference in rhesus monkeys. AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 41:487, 1974. Kaplan, J.R. Patterns of Interference and the Control of Aggression in a Group of Freeranging Rhesus Monkeys. Ph.D. Dissertation, University of Michigan, Ann Arbor, Michigan, UNIVERSITY MICROFILMS, 1976. Kaplan, J.R. Patterns of fight interference in free-ranging rhesus monkeys. AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 47:279-287,1977. KaDlan. J.R. Fight interference and altruism i n rhesus monkeys. AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 491241-249, 1978. Kaplan, J.R.; Chikazawa, D. Comparison of compound and free-ranging rhesus monkeys. AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY. 48:410,1978. Kaplan, J.R.; Manuck, S.B. Patterns of fight interference in one-male harem groups of cynomolgus macaques. AMERICAN JOURNAL OF PRIMATOLOGY 8:346,1985. Kaufmann, J.H. Social relations of adult males in a free-ranging band of rhesus monkeys. Pp 73-98 in SOCIAL COMMUNICATION AMONG PRIMATES, S.A. Altmann. ed. Chicago University Press, Chicago, 1967: Kirk. R.E. EXPERIMENTAL DESIGN: PROCEDURES FOR THE BEHAVIORAL SCIENCES. Belmont, California, Brooks/ Cole, 1968. Kurland, J.A. Kin selection in the Japanese monkey. CONTRIBUTIONS TO PRIMATOLOGY No. 12. Basel, S. Karger, 1977. Massey, A. Agonistic aids and kinship in a group of pigtail macaques. BEHAVIORAL ECOLOGY AND SOCIOBIOLOGY 2:31-40, 1977. Nishida, T. Alpha status and agonistic alliance in wild chimpanzees (Pan troglodytes schweinfurthii). PRIMATES 24:318-336, 1983. 298 I Kaplan, Chikazawa, and Manuck Oswald, M.; Erwin, J. Control of intragroup aggression by male pigtail monkeys (Macaca nernestrina). NATURE (LONDON) 2621656-687,1976. Packer, C. Reciprocal altruism in Pupio unubis. NATURE (LONDON) 265:441-443, 1977. Rowell, T. THE SOCIAL BEHAVIOR OF MONKEYS. Baltimore, Penguin Books, 1972. Sade, D.S. Determinants of dominance in a group of free-ranging rhesus monkeys. Pp 99-114 in SOCIAL COMMUNICATION AMONG PRIMATES. S.A. Altmann, ed. Chicago, University of Chicago Press, 1967. Sade, D.S. A longitudinal study of social behavior in rhesus monkeys. Pp 377-398 i n FUNCTIONAL AND EVOLUTIONARY BIOLOGY OF PRIMATES. R. Tuttle, ed. Chicago, Aldine Atherton, 1972. Sade, D.S.; Cushing P.; Cushing, K.; Dunaif, A.; Figueroa, A.; Kaplan, J.R.; Lauer, C.; Rhodes, D.; Schneider, J. Population dynamics in relation to social structure on Cay0 Santiago. YEARBOOK OF PHYSICAL ANTHROPOLOGY 20:251-262,1977. Tokuda, K.; Jensen, G. The leader’s role in controlling aggressive behavior. PRIMATES 9:319-322,1968. Trivers, R.L. The evolution of reciprocal altruism. QUARTERLY REVIEW OF BIOLOGY 46:35-57,1971. Watanabe, K. Alliance formation in a freeranging troop of Japanese macaques. PRIMATES 20:459-474,1979. West-Eberhard, M.J. The evolution of social behavior by kin selection. QUARTERLY REVIEW OF BIOLOGY 5O:l-33.1975.