American Journal of Primatology 5:345-356 (1983) Attacks on Neonates by a Male Mangabey (Cercocebus atys) CURT D. BUSSE AND THOMAS P. GORDON Yerkes Regional Primate Research Center, Emory Unruersity, Atlantq Georgia Aggressive behavior by a n adult male toward selected infants and their mothers was observed during a long-term study of reproduction in a captive social group of sooty mangabeys (Cercocebus atys). The highest-ranking adult male in this group was observed to attack and bite three neonates out of a total of 13 infants born in 1982. All three attacks were directed attacks in which infants were grabbed from their mothers and bitten in the head. The first infant was fatally injured; the other two probably would have sustained fatal wounds had the male’s canines not been blunted beforehand and had observers not intervened. The attacks were preceded by a pattern in which the male persistently stalked or chased the mother-infant pairs, a behavior first observed in the hours immediately following parturition, Unlike attacks in wild baboon groups following male immigration, these attacks on infants occurred in a stable social group in which the male attacker had been a lifelong resident. This male, however, had gained alpha rank only 3 months before attacking the first infant. These attacks, in the context of other evidence of aggression and wounding, are discussed relative to current models of infanticide in primates. Key words: infanticide, sexual selection, sooty mangabeys, Cercocebus atgs INTRODUCTION The killing of infants by adult. males has been reported to occur in a variety of primate taxa [reviewed by Hrdy, 1979, 1981; Butynski, 19821. Several interpretations of this behavior have been offered, some generating controversy-particularly the suggestion that infant killing might provide reproductive advantages to males under some circumstances and thus that the behavior developed through the process of sexual selection [Hrdy, 19741. According to this interpretation, by killing an infant a male could shorten the interval until the mother is again fertile, since ovulation is suppressed in primate females during lactation [eg, Plant et al, 1980; Schallenberger et al, 19811, and could therefore gain mating access to the female significantly sooner. A second tenet of this sexual selection model is that infant killing behavior would not be directed to infants that males could have sired. Infant killing therefore Received J u n e 18, 1983; revision accepted July 29, 1983 Address reprint requests to Curt D. Busse, Yerkes Field Station, 2409 Collins Hill Rd., Lawrenceville, GA 30245. 0 1983 Alan R. Liss, Inc. 346 Busse and Gordon would be most, advantageous to males who are recent immigrants to a group and who are likely to mate with estrous females [Hrdy, 1974; Chapman & Hausfater-, 19791. An alternative point of view is that infant killing is accidental or even pathological behavior, resulting from social disturbance, social stress, or unusual levels of crowding [Curtin & Dolhinow, 1978; Boggess, 19791. Infant killing has been reported most often in one-male breeding groups following replacement of the breeding male [eg, Sugiyama, 19651. More recently, infant killing has been reported in species that form groups with multiple breeding males. For one such set of species, the savanna-dwelling baboons (Papio spp), several cases of infant killing have been observed, usually involving new immigrant males attacking infants that they could not have sired [Busse & Hamilton, 1981; Collins et al, in press]. The present paper reports male aggression toward infants, including infant killing, in another primate genus that forms multimale groups: the mangabeys (Cercocebus spp). The five mangabey species are closely related taxonomically to baboons [Napier & Napier, 19671. The social behavior of mangabeys, however, has been studied comparatively little, either in the wild [but see Chalmers, 1968; Waser, 1977; Wallis, 19831 or in captivity [eg, Bernstein, 19761. Infant killing has been reported once previously in mangabeys. During a n experimental merging of three groups, Bernstein  observed the alpha male of one group attack an adult female from another group and kill her 2-month-old infant. This attack, produced inadvertently in a controlled social environment, may be comparable to attacks on infants during encounters between free-ranging groups of baboons [Shopland, 1982; Collins et al, in press]. Such attacks are associated with high rates of fighting in which infants might be injured by accident; the attacks are usually also consistent with the sexual selection model, which postulates that the attackers could not have sired the infants. The present paper describes several attacks on infants by a male sooty mangabey in a n entirely different context; ie, involving a male who was a lifelong member of the group but who had recently attained alpha rank. These attacks were observed in the same group observed by Bernstein [19711, but after 13 yr of demographic change. To aid in the interpretation of this attack behavior, data are presented on rates of agonistic interactions between the male attacker and the mothers before, during, and after the day of birth, and on the overall rates of aggressive behavior by the three adult males in the group. In addition, data are presented on the effects of infant deaths on interbirth intervals to assess the applicability of the sexual selection model to sooty mangabeys. METHODS The study was conducted a t the Yerkes Primate Center Field Station in Lawrenceville, Georgia, using a group of sooty mangabeys (Cercocebus atys) which had been housed there since 1968. Preliminary observations began in October 1981; formal observations were conducted from February 18, 1982 to October 15, 1982. At the start of formal observations, the group consisted of 49 individuals, including three adult males, 26 adult females, and 20 immatures. One of the males (F) was a member of the original group formed in 1968; the other two males (LA and AB) were born in the group in 1970 and 1972 respectively. In the latter half of the study, seven adult females who had been out of the group for 2 yr were returned to the group (June 22, 19821, and seven juvenile males (1.5-3 yr old) were removed from the group (August 26, 1982). The only other demographic changes during the study resulted from 13 births and one death (described below). The group was housed in an outdoor compound (30 m x 30 m) Attacks on Neonate Mangabeys 347 with a n attached indoor quarters (11m x 3 m). Observations were made from a 5-m tower that provided a n unobstructed view of the outdoor enclosure. Access to the indoor quarters was prohibited during observations. The group was observed 3 h r per day, four to five times a wk, for a total of 456 hr. Each 3-hr observation was subdivided into 12 focal observations lasting 15 min each. Adult males were observed once every observation day; adult females were observed once every 3-4 d on a randomized basis. All social interactions involving focal animals were scored using a standard behavioral inventory based on the inventories of Chalmers , Bernstein , and Wallis [ 19811. All instances of adult males mounting adult females, a highly visible behavior typically punctuated by a vocalization by t h e female, were scored on a n ad libitum basis. A different observation protocol was used on days of birth. The group was observed for 3 hr in the morning, using ad libitum observations for all social interactions between the mother and any of the adult males. The group was observed for longer than 3 hr if there was any indication of antagonism between the mother and any of the adult males. Hourly rates of aggression by adult males were calculated from focal observations of each male. An aggressive interaction was scored whenever the focal male directed any contact aggression (BITE, SLAP, GRAB-PULL) or noncontact aggression (CHASE, RAISED EYEBROW, VOCAL THREAT, HEAD SHAKE) toward another individual. Hourly rates of agonistic interactions for male-female dyads were calculated from the combined focal observations of both animals. An agonistic interaction was scored whenever one individual directed any aggressive or submissive (AVOID, FLEE, SQUEAL, GRIMACE, CROUCH) behaviors toward the other. Dominance relationships were inferred from observations of one individual supplanting or eliciting submissive behaviors from another. RESULTS Relationships Among Adult Males The three adult males maintained stable dominance relationships throughout the period of formal observations; AB ranked first, LA second, F third. The dominance relationship between AB and LA had reversed in mid-November 1981, 3 months before formal data collection began. Prior to this reversal LA had maintained alpha rank for at least 24 months. Wounds Observed After the Rank Reversal At the time that AB and LA reversed ranks, the group contained five infants (0-12 months of age) and 15 juveniles (12-36 months of age). Within a month after the reversal, the two youngest infants and a juvenile received head wounds, although the causes of wounding were not known. These were the only wounds received by infants or juveniles between October 1981 and October 1982 except for those described later i n which the attacks were observed. A 2.5-month-old female (UD) was observed 6 days after the reversal (November 23, 1981) with two lacerations, each 2-3 cm long, on the crown of its head. One laceration appeared to have occurred 12-24 hr before the other, judging from the degrees of necrosis of the wounds. Both lacerations occurred while the group was indoors during cold weather. The infant was removed from the group, treated, and returned to the group 5 wk later (January 29,1982). After 2 wk (February 15,1982)the wound had reopened, and the infant was again removed for treatment. Behavioral observations were not conducted during this time; thus there was no observation of either incident of injury. Busse and Gordon 348 Male Dominance 0 0 000 0 0 Reversal MD AC 0 1981 0 0 000 HC 1982 7- M e a n Gestation infant 0 Infant 0 QA interval Attacked not Attacked Antagonism Between AB I and Mother Fig. 1. Sequence of births and attacks following the male dominance reversal. AC, MD, $A, HC = mothers. The infant was returned to the group 4 wk later and was observed for 5 hr. During the first 15 min, UD was closely followed and investigated by most of the group, behavior typically displayed on the return of any group member. Then, when the mother picked up and carried UD, male AB immediately began following and chasing the mother and infant. During the next 4 hr AB persistently chased the mother, who avoided or fled from him 94 times. Several times UD left its mother, and AB then began chasing the infant. Twice after chasing the infant, AB was mobbed by the mother, male LA, and the infant's two maternal siblings. AB's pursuit, however, appeared to be more rapid when the infant was being carried by its mother than when it was away from its mother. AB eventually stopped pursuing the mother and infant, and the infant received no further wounds during the study. The second individual to receive wounds was a 17-month-old male, who was observed 8 d after the rank reversal (November 25,1981) with a facial puncture. The juvenile was removed, treated, and returned to the group after 4 wk without further incident. The third wounded individual was a 1.5-month-old male (VD) who was observed 24 d after the rank reversal (December 11,1981)with a large laceration on the crown of its head. As in the case of UD, this infant was injured while the group was confined indoors during cold weather. The infant and mother were removed together for treatment, but the mother rejected the infant while caged; thus the infant was not returned to the group. The mother was returned without incident. Aggression Toward Mothers and Neonates Between the time of the rank reversal and the end of the study, there were 13 live births, four stillbirths, and one confirmed abortion. Figure 1 shows the sequence of live births and attacks by AB on neonates. The first three infants were conceived Attacks on Neonate Mangabeys -z v) 0 cn c3 z a 0 - F t7 LL 0 1.5 I- Iz 0 a U w W a U 48.0 1 I- - ’.O 0.0 349 MD HC 09 Other Females 1 1 BEFORE DURING AFTER DAYS RELATIVE TO BIRTH Fig. 2. Rates of agonistic interactions between male AB and two mothers whose infants he attacked compared with nine mothers whose infants he did not attack (X f SE). MD’s infant was attacked and permanently removed from the group on the day of birth. HC’s infant was attacked on the day of birth but remained with her in the group. before the rank reversal, when LA held alpha rank. AB attacked and wounded the first two of these infants (for details, see below). He showed some aggression toward the third infant, following the mother and yawning repeatedly on the morning of birth, but this behavior stopped within a few hours and no further antagonism was observed. The next three infants had been conceived near the time of the rank reversal. AB directed no aggressive behaviors toward these infants or their mothers either on the day of birth or subsequently. The last seven infants were conceived after the rank reversal, and five of these mothers received little or no aggression on the morning of birth or afterward. One female (QA) maintained an antagonistic relationship with AB both prior to and after her infant‘s birth. Finally, female HC was stalked, and her infant was bitten on the morning of birth (described below). Figure 2 shows the hourly rates of agonistic interactions between male AB and 11 mothers before, during, and after the days of birth. Female MD’s infant was attacked and removed from the group on the day of birth, but this female received no aggression from AB before or after the day of birth. Female HC’s infant was attacked on the day of birth, but was not removed from the group. Like MD, HC received little aggression from male AB before the day of birth, but unlike MD, HC received aggression from AB for several months after her infant was attacked. For nine other females, whose infants were not attacked, rates of aggression by AB were low in all three time categories. AB also directed no affiliative behaviors (eg, grooming) toward these nine females. Another female whose infant was attacked (AC) is not shown in Figure 2 because no data were collected before or during the day of birth. Like female MD, however, AC received little aggression from male AB after her infant was attacked and 350 Busse and Gordon removed (0.07 interactionsihr; 53 hr). Finally, female QA (also not shown in Fig. 2) had a n antagonistic relationship with AB throughout the study, including during pregnancy (0.23 interactionsihr; 35 hr). This rate quadrupled after QA’s infant was born (0.94 interactionsihr; 16 hr). Descriptions of Attacks Attack on AC’s infant. The first attack (February 17, 1982) occurred 3 months after the rank reversal while part of the group was being held temporarily in the indoor quarters during a routine animal maintenance procedure. By chance the middle- and low-ranking adult males (LA and F) were outside while AB was inside with a portion of the group. Once indoors AB briefly chased adult female AC, who was carrying a 2-d-old infant on her ventrum. AB grabbed the infant from the mother’s ventrum and bit it twice, inflicting puncture wounds to the skull. The infant was immediately removed for treatment, but it died as a consequence of the wounds. The mother was a high-ranking, multiparous female. This attack had been completely unexpected: no similar behavior had been observed in the group for over 12 yr [Bernstein, 19711. The attack occurred so quickly that observers were unable to intervene in time to save the infant. To minimize risk of injury to other infants, the male’s canines were surgically blunted and the group capture procedures were modified so that the adult males were kept outdoors. An alternative action, removing the male from the group, was not taken because the resultant social reorganization might have led to even greater risk of injury to animals. Also, it was decided a t this time to observe the group closely on the mornings of all subsequent births in a n attempt to prevent additional serious trauma. Attack on MD’s infant. The second attack (April 8, 1982) occurred 4.5 months after the rank reversal. On the morning of the infant’s birth, male AB stalked the mother (MD) and the newborn for 5 hr, culminating this behavior with a directed attack in which he grabbed and bit the infant. Observers intervened immediately and removed the infant, who was not seriously injured. The mother was a middleranking, primiparous female. MD was first observed with her new infant a t 0812. Doors to the indoor quarters were closed and the group was observed from 0846 to 1200. Throughout this interval AB persistently followed MD, who avoided his approach. AB approached to within 2 m of MD 40 times; she immediately walked or ran away during 31 of these approaches. During five approaches in which MD did not avoid, AB sat within 2 m, yawned repeatedly, intermittently glanced toward the infant, and repositioned himself closer to MD. Each time as AB moved closer MD walked away and AB again followed. A steady rain began a t 1200 and doors to the indoor quarters were opened. Most of the group, including MD and AB, moved indoors. AB’s pursuit slowly intensified as he continuously followed MD throughout the indoor quarters. MD avoided AB repeatedly, and four times she fled outdoors. Each time AB followed MD outdoors, and she quickly returned inside through a different door. During this time, no other group members had been involved in the antagonistic episode. Finally, after MD had fled outdoors a t 1330, AB chased her for 20 m, grabbed the infant from her ventrum, and bit it once behind the ear. The mother was squealing loudly throughout this sequence, and male LA immediately ran to her aid from the indoors. AB fled carrying the infant as LA, barking loudly, chased him. As LA lunged a t him, AB released the infant, which the mother immediately retrieved and carried indoors. Attacks on Neonate Mangabeys 351 The mother and infant were removed from the group, and the infant was treated for lacerations a t the base of both ears. Neither wound penetrated the skull, and the infant recovered quickly. The mother rejected the infant, however, and the infant was transferred to the Yerkes nursery facility. Attack on HC’s infant. The third attack (September 2, 1982) occurred 9.5 months after the rank reversal, after five infants had been born without incident. AB was again observed pursuing a mother (HC) carrying a newborn, grabbing the infant from the mother, and inflicting a minor head wound. The mother temporarily abandoned the infant before and after the attack. After being placed in a cage and sedated, the mother accepted the infant, which developed normally in the group. The mother was a high-ranking, primiparous female. HC was first observed with the newborn a t 0820. In contrast to his pursuit of MD, AB’s pursuit of HC was rapid from the outset. AB persistently chased HC, intermittently grinding his teeth during the pursuit. The yawning behavior that characterized his pursuit of MD was absent in his pursuit of HC. As AB chased HC she squealed repeatedly, which elicited support from LA five times. The more vigorous support, however, was from HC’s mother, QA, the highest-ranking female in the group. QA repeatedly threatened and charged at AB, and frequently interposed herself between AB and HC. (Despite a possible risk to her own 1-month-old infant, QA carried it continuously while supporting HC.) This pattern of pursuit by AB and support by QA continued until 0855 when HC abandoned her infant, setting it down and walking to the opposite side of the compound. QA picked up the infant, carried it ventral for 20 s, then set it down. It was then investigated by several adult females and juveniles. None of the adult males, however, approached any closer than 5 m from the infant while it was abandoned. Male AB directed no behavior toward HC or the infant during the period of separation. HC retrieved the infant after 2 min, when observers started to enter the compound, and AB immediately resumed the pattern of pursuit. HC again abandoned the infant 3 min later. Again, none of the males approached closer than 5 m from the abandoned infant. HC retrieved the infant 30 s later when observers again started to enter the compound; AB again started pursuing her. As AB chased HC she squealed, turned, and countercharged. AB grabbed the infant from her ventrum, bit i t on the top of the head (without breaking the skin), and dropped it immediately when male LA chased him away. HC retrieved the infant and carried it indoors where she and QA were separated from the group. Once inside, HC again abandoned her infant, and when she failed to retrieve it both were removed and placed in a cage. The mother continued to reject the infant. She was then tranquilized (10 m g k g ketamine hydrochloride), and the infant was placed on her ventrum. As the mother recovered from sedation, she accepted the infant. Upon being returned to the group 5 d later, the mother and infant were again pursued by AB, although less vigorously than on the day of birth. This behavior persisted for a n hour, then ceased. Female QA again supported HC by blocking AB’s approach to HC. HC frequently chased other females, and male LA often supported these females against HC. Thus, upon reintroduction HC received more aggression from LA than from AB. For the remainder of the study, however, HC maintained a close association with LA and QA and avoided any close approach by AB. Hourly Rates of Male Aggression Table I shows the hourly rates of aggressive behaviors by the three males directed toward all other group members during focal observations of the males. The 352 Busse and Gordon TABLE I. Rates of Aggressive Interactions by Adult Male Mangabeys During Focal Observations Male AB LA F Social rank Hr of observation 1 41.00 2 3 41.00 37.75 Hourly rates of aggression Contact Noncontact 0.39 0.61 0.79 1.44 2.32 1.01 alpha male (AB) exhibited the lowest rate of contact aggression and a rate of noncontact aggression that was intermediate between that of the other two males. Reproductive Patterns in Females Following Infant Death or Removal Analysis of female reproductive patterns between 1980 and 1983 shows that infant deaths or removals resulted in significantly shorter interbirth intervals. For a sample of 13 mothers whose infants survived to weaning, the average interbirth interval was 21.3 months (SE = 1.1months). For eight other mothers whose infants died or were removed before weaning, the average interbirth interval was 11.3 months (SE = 1.4 months), 10.0 months shorter than that for the above sample (Mann Whitney U = 4,P < 0.001, two-tailed test). The average age a t which these infants died or were removed was 2.1 months. Female AC, whose infant died following a n attack by AB, conceived 34 days after the death, during her first ovulation after the attack. In observations of the group during this period, AB mounted AC 25 times, including one confirmed ejaculation. LA mounted her once, 6 d before the estimated time of conception (based on maximum perineal swelling size). The third male, F, was not observed mounting AC. The resulting infant (born August 29, 1982) was stillborn. AC’s next infant (born March 23,1983) was liveborn and was not attacked. Female MD, whose infant was removed after AB attacked it, conceived 7.5 months later, during her sixth menstrual cycle after the attack. The group was not observed a t this time, hence no data are available on the possible paternity of her next infant. The infant (born May 13, 1983)was not attacked. DISCUSSION The behavior detailed herein, in which a n adult male mangabey monkey attacked and wounded three neonates in the period following his ascent to alpha rank, was observed against a background of general social stability and low rates of aggression. Serious episodes of contact aggression with resultant wounding were virtually absent during the study period except for the events described [also see Bernstein et al, 19831. The only exception was a brief period of fighting between males AB and LA that was associated with the rank reversal and terminated shortly thereafter. The rarity of these attacks is further evidenced by the fact that such behavior has been observed only one other time in the 15-yr history of this group, and that was during a group merger which produced fighting and trauma in all age/ sex categories [Bernstein, 19711. The male who attacked the infants was born into this social group and had a lifelong common social history with the other group members. The social relationships between this male and three adult females were altered dramatically following and a s a n apparent function of the females’ giving birth. Following the births the male became intensively aggressive, in marked contrast to any previously observed behavior, and this behavior culminated in directed attacks on each of the neonates. Attacks on Neonate Mangabeys 353 All three attacks would most likely have proven fatal if precautions had not been taken to limit the potential for trauma and had observers not intervened. The first attack occurred before routine observation began; thus no evidence is available as to whether AB’s behavior toward the mother was altered during the infant’s first 2 d of life. Prior to the second attack, however, the male was observed to stalk the mother and infant for 5 hr. This stalking behavior, characterized by persistent, slow following with intermittent threat behaviors, had not been described previously in mangabeys. A striking aspect of these attacks is that once the infants were removed there was little or no further antagonism between the male and the mothers. One of these females was even observed to copulate with the male soon after he had fatally wounded her infant. Thus, these cases provide no evidence that the attack behavior had any short-term or long-term negative consequences on the male’s relationships with the mothers of the infants that died or were removed from the group. The outcome of the third attack, on HC’s infant, contrasts with that of the first two attacks. Following treatment the infant was returned to the group with its mother and an antagonistic relationship, punctuated by relatively high rates of aggression, was observed between the mother and male AB for several months. The mother (HC) received frequent aid from her mother (QA) and from male LA, and this appeared to help forestall further direct attack on the infant. An interesting element in two episodes of stalking was that AB’s pursuit was more vigorous when infants were on their mothers than when they were off their mothers. This pattern was evident in AB’s pursuit of 7.5-month-old UD when she was returned to the group. Also, during AB’s stalking of HC and her newborn, HC twice abandoned the infant for short periods. In neither instance did male AB attempt to attack the abandoned infant, but both times he resumed pursuit of female HC as soon as she retrieved the infant. Thus, although the attacks were directed specifically to infants, the stimulus that produced aggression in male AB appears to have been the motheriinfant unit. This is also supported by the observations of aggression terminating immediately following removal of the infants from the group. Unlike observations of infant killing among wild baboons [Collins et al, in press], the present observations occurred in a social group with stable membership and low rates of aggression. The male who attacked the infants had lived in the group for his entire life, with the only notable change being that he had gained alpha rank 3 months before the first attack on an infant. Despite this recent change in dominance relationships, the new alpha male was not an unusually aggressive individual; his rates of aggression were comparable to those of the two other adult males as well as to five other males studied by Bernstein et a1  during 1974 and 1975. All three observed attacks were directed attacks on the infants and were not preceded by fighting among the males or other animals. For these reasons, social disturbance seems unlikely to be the basis of the infant attacks observed in this study. Unequivocal interpretation, however, is not possible at present, in part because one of the attacks occurred during an external disturbance. Also, three of the four females who received high rates of aggression from AB were from the dominant matriarchy (QA, AC, HC), suggesting a possible social conflict between the new alpha male and the highest-ranking family. Nonetheless, this would not account for aggression intensifying on the day each female gave birth. Finally, since all three attacks were by one male, it is possible that they were a randomly occurring, idiosyncratic behavior. An alternative interpretation is that infanticide in primate groups might have a biological basis, and that males might under some circumstances gain a reproduc- 354 Busse and Gordon tive advantage by killing infants. This model has a n assumption, upheld for the mangabey group under study, that females will conceive significantly earlier when their infants die than when they survive. The model predicts that males will attack only those infants that they did not sire, whereas if the attack behavior is pathological males might be expected to attack infants indiscriminantly. The critical evidence needed to test these competing hypotheses-paternity of attacked vs nonattacked infants-is unavailable, however. In studies of savanna baboons, it has been possible to verify that males killed infants they did not sire, because in most cases the males had recently immigrated to the groups [Collins et al, in press]. In the present study, male AB resided in the group when all infants that were subjects of this study were conceived. On this basis he could have sired any of the infants. The first two infants that AB attacked were conceived before he gained alpha rank. In this group alpha rank appears to confer a mating advantage as inferred by frequency of copulation [Gordon and Busse, unpublished]. This indirect evidence suggests that the infants born early during the study were more likely sired by LA than by AB. The third attacked infant was conceived after AB attained alpha rank and thus was more likely sired by AB than LA. However, for none of the three attack victims or the other ten infants that were not attacked are there direct paternity data necessary to interpret the behavior clearly. In studies of savanna baboons the most prominent condition leading to infant killing is male immigration, in which nonpaternity between attackers and victims is certain. To be consistent with the sexual selection model, the results of the present study would suggest two additional rank-residency patterns that could lead to infant killing in multimale primate groups. In both patterns a male resides in the group when a n infant is conceived but does not sire the infant. In one pattern, suggested by the attacks on AC and MD’s infants, a male does not mate with the mother because he does not hold alpha rank and other males outcompete him. Collins et a1 [in press] report a similar pattern in which a natal male baboon (Papio anubis) attained alpha rank and attacked infants conceived before he gained alpha rank. Also, Wolf [1980; detailed by Hrdy, 19811 observed a male silvered leaf-monkey (Presbytis cristutu) rise to alpha rank and kill an infant soon after the rank change. These observations suggest that periods following alpha rank reversals might be times of high risk for infants, even when the new alpha male has been a long-term resident in the group. In a second pattern, suggested by the attack on HC’s infant, a male holds alpha rank a t the time a n infant is conceived, but for some reason does not mate with the mother. For example, the male might be mating with other females that are ovulating a t about the same time. This pattern predicts that infants might be attacked occasionally, even during times of stability of male dominance ranks. This reasoning, that infant killing could be favored even if males reside in the group a t the time of conception but do not sire a n infant, also assumes that there are mechanisms by which males can discriminate their own offspring from those of other males. Among laboratory rodents, discrimination of offspring by males appears to be based on past association and copulation with the mother, rather than by direct recognition [Labov, 1980; vom Saal & Howard, 1982; Huck et al, 19821. The existence of such mechanisms in primates remains to be critically evaluated. Among some primate species that have a multimale social organization, females appear to provide unambiguous signals as to the optimal mating time. Female mangabeys, like female baboons, develop cyclical swellings of their perineal skin [Hadidian & Bernstein, 1979; Wallis, 19831. These swellings reflect the secretion of ovarian steroids [Cercocebus atys: Aidara et al, 19811 and reach a maximum size near the time of ovulation [Papiospp: Wildt et al, 1977; Shaikh et al, 19821. Perineal swellings might therefore provide information by which male confidence of paternity is mediated in these species. Attacks on Neonate Mangabeys 355 The observations reported here raise questions concerning both the interpretation of data and the investigation of potentially traumatic behavior that might nonetheless have an evolutionary basis. One question concerns the relevance of observations made in captive environments. Studies, such as this one, of behavioral phenomena in captive subjects cannot exclude the possibility that the research environment may influence the behavior under investigation. On the other hand, captive environments provide excellent observation conditions and, frequently, knowledge of antecedent events and contexts not available in natural study sites. Perhaps the fullest explication of behavioral patterns will be possible when hypotheses generated from data collected in natural settings are subject to rigorous, experimental test in controlled environments. Second, a study of behavior which may produce trauma or even fatal wounding in a captive, managed social group raises the question of whether and when to intervene; concerns with both practical and ethical implications. In this study, the first observed attack was totally unexpected. Subsequently, procedures were modified in order to continue observations while minimizing the possibility of additional serious trauma. This course was chosen because the alternative (removal of animals from the social unit) would have been socially disruptive and because the male’s attack behavior was viewed as being a potentially significant phenomenon meriting careful study. CONCLUSIONS 1. Directed attacks by an adult male on young infants, a behavior previously reported in single-male and multimale primate groups following male immigration, are here shown to occur in a captive, multimale group of Cercocebus atys with stable male membership. 2. The selective nature of the observed attacks and their occurrence following a male’s rise to alpha rank, as well as the overall low levels of aggresssion and absence of trauma in other agelsex classes, are all consistent with the hypothesis that male attacks on infants may be a biologically based behavior derived through sexual selection. 3. The present study provided little evidence to support an alternative hypothesis that such attack behavior is the consequence of either social disturbance or pathological events. Neither the sexual selection nor the social disturbance hypothesis, however, was unequivocally supported or directly refuted. The most crucial evidence in testing these hypotheses, paternity determinations for both attacked and unattacked infants, is lacking. 4. The present data do confirm that female mangabeys resume ovarian activity significantly sooner following the death or removal of a n infant. Additionally, the results show that a male mated repeatedly with a female and probably impregnated her 1 month after attacking and killing her infant. There was no evidence that the male’s attack behavior resulted in punitive behavior by the mothers. 5. The constraints on interpretations of data in a captive environment and ethical issues associated with the study of possibly traumatic behavior are discussed. ACKNOWLEDGMENTS We thank Drs. Carolyn L. Ehardt, William J. Hamilton 111, Frederick A. King, Margaret L. Walker, Patricia L. Whitten, and Mark E. Wilson for helpful comments on this manuscript. We also express our gratitude to Dr. Kim Wallen, who maintained and operated the PDP 11/23 computer on which the data were analyzed and the manuscript was prepared. The research was supported by PHS grants from NIH (RR 00165) and NIMH (1T32 M4 16543). 356 Busse and Gordon REFERENCES Aidara, D.; Badawi, M.; Tahiri-Zagret, C; Robyn, c. Changes in concentrations of serum prolactin, FSH, oestradiol and progesterone and of the sex skin during the menstrual cycle in the mangabey monkey. JOURNAL OF REPRODUCTION AND FERTILITY 62:475-481, 1981. Bernstein, I.S. The influence of introductory techniques on the formation of captive mangabey groups. PRIMATES 12:33-44,1971. Bernstein, I S . Activity patterns in a sooty mangabey group. FOLIA PRIMATOLOGICA 26:185-206,1976. Bernstein, IS.; Williams, L.; Ramsay, M. The expression of aggression in Old World monkeys. INTERNATIONAL JOURNAL OF PRIMATOLOGY 4:113-125,1983. Boggess, J. Troop male membership changes and infant-killing in langurs (Presbytis entellus). FOLIA PRIMATOLOGICA 32:65107,1979. Busse, C.; Hamilton, W.J., 111. Infant carrying by male chacma baboons. SCIENCE 21211281-1283,1981. Butynski, T.M. Harem-male replacement and infanticide in the blue monkey (Cercopithecus mitis stuhlmanni) in the Kibale Forest, Uganda. AMERICAN JOURNAL OF PRIMATOLOGY 3:l-22,1982. Chalmers, N.R. The social behaviour of free living mangabeys in Uganda. FOLIA PRIMATOLOGICA 8:263-281,1968. Chapman, M.; Hausfater, G. The reproductive consequences of infanticide in langurs: A mathematical model. BEHAVIORAL ECOLOGY AND SOCIOBIOLOGY 5:227240,1979. Collins, D.A.; Busse, C.D.; Goodall, J . Infanticide in two populations of savanna baboons, in INFANTICIDE: COMPARATIVE AND EVOLUTIONARY PERSPECTIVES. G. Hausfater, S.B. Hrdy, eds. New York, Aldine (in press). Curtin, R.; Dolhinow, P. Primate social behavior in a changing world. AMERICAN SCIENTIST 66:468-475,1978. Hadidian, J.; Bernstein, I.S. Female reproductive cycles and birth data from a n Old World monkey colony. PRIMATES 20:429442, 1979. Hrdy, S.B. Male-male competition and infanticide among the langurs (Presbytis entellus) of Abu, Rajasthan. FOLIA PRIMATOLOGICA 22119-58,1974. Hrdy, S.B. Infanticide among animals: A review, classification and examination of the reproductive strategies of females. ETHOLOGY AND SOCIOBIOLOGY 1:13-40,1979. Hrdy, S.B. THE WOMAN THAT NEVER EVOLVED. Cambridge, Harvard University Press, 1981. Huck, U.W.; Soltis, R.L.; Coopersmith, C.B. Infanticide in male laboratory mice: effects of social status, prior sexual experience, and basis for discrimination of related and unrelated young. ANIMAL BEHAVIOUR 30:1158-1165, 1982. Labov, J.B. Factors influencing infanticidal behavior in wild male house mice (Mus musculus). BEHAVIORAL ECOLOGY AND SOCIOBIOLOGY 6:297-303, 1980. Napier, J.R.; Napier, P.H. A HANDBOOK OF LIVING PRIMATES. New York, Academic, 1967. Plant, T.M.; Schallenberger, E.; Hess, D.L.; McCormack, J.T.; Dufy-Barbe, L.; Knobil, E. Influence of suckling on gonadotropin secretion in the female rhesus monkey (Macacu mulatta). BIOLOGY OF REPRODUCTION 23:760-766, 1980. Schallenberger, E.; Richardson, D.W.; Knobil, E. Role of prolactin in the lactational amenorrhea of the rhesus monkey (Macaca mulattu). BIOLOGY OF REPRODUCTION 25:370-374, 1981. Shaikh, A.A.; Celaya, C.L.; Gomez, I; Shaikh, S.A. Temporal relationship of hormonal peaks to ovulation and sex skin deturgescence in the baboon. PRIMATES 23:444-452, 1982. Shopland, J.M. An intergroup encounter with fatal consequences in yellow baboons (Papio cynocephalus). AMERICAN JOURNAL OF PRIMATOLOGY 3:263-266,1982. Sugiyama, Y. On the social change of hanuman langurs (Presbytis entellus) in their natural condition. PRIMATES 6:381-418, 1965. vom Saal, F.S.; Howard, L.S. The regulation of infanticide and parental behavior: Implications for reproductive success in male mice. SCIENCE 215:1270-1272, 1982. Wallis, S.J. The behavioural repertoire of the grey-cheeked mangabey Cercocebus albig e m johnstonii. PRIMATES 22:523-532, 1981. Wallis, S.J. Sexual behavior and reproduction of Cercocebus albigena johnstonii in Kibale Forest, Western Uganda. INTERNATIONAL JOURNAL OF PRIMATOLOGY 4~153-166,1983. Waser, P.M. Individual recognition, intragroup cohesion and intergroup spacing: evidence from sound playback to forest monkeys. BEHAVIOUR 60:28-74,1977. Wildt, D.E.; Doyle, L.L.; Stone, S.C.; Harrison, R.M. Correlation of perineal swelling with serum ovarian hormone levels, vaginal cytology, and ovarian follicular development during the baboon reproductive cycle. PRIMATES 18:261-270, 1977. Wolf, K. Social change and male reproductive strategy in silvered leaf-monkeys, Presbytis cristata, in Kuala Selangor, Peninsular Malaysia. AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 52:294, 1980.