Infanticide Among Animals: A Review, Classification, and Examination of the Implications for the Reproductive Strategies of Females Sarah Blaffer Hrdy Peabody Museum, Harvard Unirqersity

Infanticide among animals is a widespread phenomenon Within the same species, infanticide may occur in some with no unitary explanation. Although the detrimental areas but not others, as evidenced by the variable outcome for the infant is fairly constant, individuals expression of infanticidal behavior among Hanuman responsible for infanticide may or may not benefit, and langurs. At present, the most obvious factor influencing when they gain in fitness there may be considerable facultative expression of the infanticidal trait is popu- variation in how they gain. Sources of increased fitness lation density. from infanticide include: (1) exploitation of the infant Where it occurs, sexually selected infanticide is a as a resource, (2) elimination of a competitor for re- significant cause of mortality. As such, it has important sources, (3) increased maternal survival or lifetime re- implications for the evolution of behavior, particularly productive success for either mother or father by elim- for patterns of association between males and females, ination of an ill-timed, handicapped, or supernumerary for female reproductive physiology, and for the pat- infant, and, finally, (4) increased access for individuals terning of sexual receptivity by females. It is hypothe- of one sex for reproductive investment by the other sex sized that the threat posed by infanticide is one of sev- at the expense of same-sex competitors. Predicted attri- eral pressures selecting for a shift among higher butes of the perpetrators (such as sex and degree of primates away from strictly cyclical estrous receptivity relatedness to the infant), attributes of the victim (i.e., towards socially determined or situation-dependent re- age and vulnerability), as well as schedule of gain, vary ceptivity. for each class. Under some circumstances, individuals commit infanticide which does not result in any prospect Key Words: Infanticide: Female reproductive strate- for gain: such instances are considered nonadaptive or gies: Population density. “pathological.” In those cases where infanticide does on the average increase fitness, selection pressures fa- voring it have arisen as a result of the extensive and INTRODUCTION time-consuming investment involved in production of young, and the extreme vulnerability that characterizes Offspring are vital to the continued survival of infancy in many animals. any species. At first glance, it seems surprising The scattered but nevertheless extensive occurrence to find selection for behavior that does not con- of infanticide among primates raises the question of tribute to the survival of infants, odder still to inter-specific variation. Factors such as seasonality in breeding, cooperation between individuals in defense of find selection for behavior that actually de- infants, marginal habitats, and low intrinsic rates of creases infant survivorship. Yet nature provides natural increase may outweigh other pressures, such as abundant examples, many of them paradigms short male tenure lengths, which select for infanticide. for the evolutionary process: natural selection at the individual level may-at least temporar- ily-preempt species advantage. Received January 18. 1979: revised May 31. 1979 The most widely reported cases of infanticide Address reprint requests lo: Sarah Blafkr Hrdy. Peabody Museum. Harvard University. Cambridge, MA 02138. involve adult males. but adult females and even

Ethology and Sociobiology I: 13-40 (1979) 13 @ Elsevier North Holland. Inc.. 1979 Ol62-3095/79/010013-28/502.25 14 S. B. Hrdy

hhhhhhhhhhhhhhhhhhs are sometimes implicated. In- an artifact of disturbed settings (e.g., Calhoun, fants may be killed not only by outsiders but by 1962). but rather a widely spread phenomenon members of their own group, even their parents. in natural populations. particularly among pre- As diverse as the perpetrators are the selection datory fish, rotifers. and insects which live in pressures that promote infanticidal behavior. ln- fresh water. fanticide among animals cannot be understood Rotifers such as A.splorrc/~r~m si~holtli. for ex- as a unitary phenomenon. ample, provide elegant evidence for the extent This paper attempts to classify manifold in- and duration of selective pressures from canni- stances of infanticide into categories with some balism. Male rotifers. which are smaller than explanatory and predictive potential. In general, females in this group. have evolved a variety of examples of infant-killing by either males or fe- major structural adaptations (such as proturber- males can be explained by recourse to one of antes in the body wall) and other features which five classes of explanation. The first four-ex- specifically protect them from being eaten by ploitation. resource competition. parental ma- other rotifers. including clone-mates (Gilbert, nipulation. and sexual selection-all assume that 1976). infanticidal behavior is evolved and adaptive for Lethal exploitation of an infant is not limited those individuals who perpetrate the killing. The to eating it. Among various primates, for ex- fifth class of explanation-social pathology- ample, infants may be used as “buffers” in does not assume that infant-killing is adaptive. agonistic episodes (Deag and Crook. 1971, and Cannibalism is not considered here as a separate especially. Popp, 1978) or as an object for “play category. Typically, cannibalism falls under ex- mothering” (Lancaster. 197 I ): such episodes ploitation of the infant as a resource, but it may may on rare occasions lead to injury or starva- also occur opportunistically in conjunction with tion of the infant as an incidental byproduct ohhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhf other classes of infant-killing. exploitation. Among primates. the majority of infant deaths so far attributable to adult females do not occur as a result of violence. but rather CLASSES OF INFANTICIDE starvation. This phenonmenon. known as “aunting to death.” results when nonlactating Exploitation females take an infant from its mother and for- When individuals responsible for an infant’s cibly retain it (Hrdy. 1976). As was true with death directly benefit from consumption or use the early reports of infanticide by male primates. of their victim, infanticide can be said to be a there has been a general tendency to dismiss form of “exploitation”: the infant itself becomes deaths due to “kidnap” or “aunting” as quirks a resource, either a food resource, a protective or pathologies in an otherwise benign system of buffer against aggression by third parties. or a cooperative infant care which is primarily “prop” for obtaining maternal-like experience. geared to benefit the infant. However, this view The most obvious widespread form of such accords with neither the abusive treatment exploitation is cannibalism. The victim must be sometimes dealt to borrowed infants (Hrdy, killed in order for the resource to be exploited, 1977). nor with accumulating case histories of and infants are particularly vulnerable targets. infant fatalities (Quiatt, 1977). Among some ground squirrels (Steiner. 1972). In most cases of infant transfer between fe- hyenas (Kruuk, 1972), occasionally in lions males belonging to the same group. the mother (Schaller, 1972). and other predatory creatures, is able to retrieve her infant without difficulty. particularly some fish and insects (Fox, 1975). Sometimes, however, as has been reported on cannibalism appears to be an end in itself. not multiple occasions for rhesus macaques (Hinde clearly distinguished from predation except that and Spencer-Booth, 1967: Quiatt. 1977) and immatures are more likely to be eaten than squirrel monkeys (Rosenblum. 1972) and in a adults because of the size differential among single case among wild Lowe’s guenons (CPY- animals that would otherwise be nearly equals. copirh~~crls ctrmphdli lowi) (Bourliere et al., The extensive literature on intraspecific preda- 1970) and captive colobus monkeys (Horwich tion in natural populations has been recently and Manski, 1975). the infant is taken by females reviewed by Laurel Fox (1975). Contrary to a who, because of their ability to dominate the common misconception, cannibalism is not just mother, prevent her from retrieving her infant. Infanticide among Animals 15

Where observers did not intervene. infants males may be particularly at risk. Among wild “kidnapped” by nonlactating females have died. dogs, pups born to a subordinate pack member The threat of losing an infant to a more dom- may be murdered by the dominant female, inant female may contribute to maternal pos- thereby garnering for her own offspring a larger sessiveness, and restrict the occurrence of post- proportion of the pack’s resources, including not natal infant-sharing to species with less only meat brought back by other pack members pronounced dominance hierarchies (Hrdy. 1976: but the milk of the mother who lost her pups McKenna, in press). Only among colobines, (van Lawick, 1972). Mother’s milk may also fig- where dominance relations are typically less ure in infanticide by female elephant seals who rigid than among macaques or baboons, do attack and kill infants who become separated mothers routinely give up their infants in the from their mothers: such “orphans” survive first hours after birth. only by stealing milk from other mothers and, Among colobines such as the Hanuman lan- hence. compete with the legitimate offspring of gur. mothers are virtually always able to retrieve those females (LeBoeuf and Briggs. 1977). infants from troopmates. although not necessar- Natural observations of direct attacks upon ily from females belonging to tl~“$~c~rrt troops. infants by primate females belonging to the same Intertroop kidnappings that resulted in pro- group have been confined to the great apes, and longed separation from the mother or starvation involve cannibalism. In three cases of intracom- have been reported for Dharwar. Jodhpur, and munity infanticide by female chimpanzees, the Mount Abu (Sugiyama. 1965. IY66: Mohnot, mothers of the infant killed (the same female IY74: Hrdy. 1977). “Gilka” in two cases) suffered from a disabil- ity-in both cases partial paralysis due to polio-and both females were lower in rank than Resource Competition the females who killed their infants (Goodall, This model applies to competition between in- 1977). A suspected case of cannibalism among dividuals (often females or lineages) for phys- wild gorillas (Fossey, 1976, and personal com- ical resources such as food or nest sites. Typi- munication) was similarly attributed to a domi- cally, such resources are related either directly nant female. By and large, these attacks follow (in the case of nest sites) or indirectly (as in the pattern reported for wild dogs in which a parental food supplies) to the production or rear- dominant female murders the offspring of a sub- ing of offspring. The resource competition hy- ordinate. It is of some interest therefore, that in pothesis predicts that the death of an infant will wild dogs, gorillas, and chimpanzees, females, on the average result in increased access to re- or else both males and females, move between sources for the killer and his or her descendants. communities. This contrasts with the far more Because other members of a population often typical mammalian pattern of sedentary sister- stand to gain from reduced pressure on re- hoods with migrating males (Frame, 1976: Har- sources. animals may gain from infanticide in court et al., 1976: Tutin, 1975, Wrangham, in this respect even when resource competition is press). One outcome of such a breeding struc- not the primary selective agent foJ infanticide. ture would bc lower degrees of relatedness In reviewing the literature on cannibalism, Fox, among females in wild dog packs, gorilla’troops, for example, stresses the multiple advantages or chimpanzee communities, and hence the pos- for animals which prey on conspecifics: “The sibility of harming a close relative would be re- cannibal gains a meal at the same time that it duced. eliminates a potential competitor and perhaps a Although the chimpanzee cases may repre- potential conspecific predator as well. Because sent elimination of offspring in competing li- population size is reduced, more food will be neages, it is not clear what resources are at available to each survivor. . . .” (1975, p. 97). stake. Possible additional explanations include Because of their vulnerability, infants are an exploitation of the infant as a food source, but obvious target for extermination, particularly it is not known how important the food value of among fish and insects where, parental protec- a cannibalized infant might be. tion may be absent. In mammalian species In some cases, offspring may be associated where mothers remain near and protect their with some particular resource. Paul Sherman’s young, orphans and offspring of subordinate fe- observations of Belding’s ground squirrels 16 S. B. Hrdy

(Spermophillrs beldingi) provide a classic mam- sive literature for example on brood reduction malian case in point (1976. 1978, and in press). among raptorial birds such as hawks, buzzards, Among these squirrels of the high California crows, owls, skuas, and perhaps especially, ea- Sierras, infanticide by females is the leading gles (Lockie, 1955; Schuz. 1957; von Wendland, cause of mortality. By an extremely conserva- 1958: Ingram, 1959: Meyburg, 1974: Procter, tive estimate (that is, only killings which were 1975). In several eagle species (e.g., Aguila po- directly witnessed, and none which were in- &rina, A. ~~er~~~~ltsi), two eaglets are hatched, ferred, were counted) 8% of infants born are but almost invariably only one fledges. The ex- killed by conspecifics. The major killers were traordinary aggressiveness between eaglets of adult females distantly related to their victims. the same eyrie among golden eagles (A. crysar- Females who lost infants left their seemingly tos) has led observers to the conclusion that one “unsafe” borrows. Sherman hypothesizes that eaglet was murdering the other, and the term abandoned nest sites become available to the “cainism” is used to describe this phenomenon infanticidal female. (von Wendland, 1958). In fact, however, as A number of animals, including hyenas, Bernd-Ulrich Meyburg (1974) points out in his wolves, chickens, geese, and various primates review of sibling aggression among nestling ea- (Mech, 1966; Kruuk, 1972: Bygott, 1974: South- gles, the actual killing of one eaglet by another wick et al., 1974: Wilson, 1975) exhibit hostility is probably rare. Instead one hatchling (typically towards strangers of other groups. Southwick et the firstborn) contributes indirectly to the de- al. have labelled this behavior “xenophobia.” mise of the other hatchling through intimidation Instances of infants being killed in the course of which predisposes the second eaglet to die of intergroup hostilities have been recorded for starvation or exposure. Similar patterns apply rhesus macaques at Cayo Santiago (Carpenter. to other’raptors with slight variation. In some 1942: D. Sade. personal communication) and, in cases. the dead hatchling is eaten by siblings. one instance involving an adult female among parents. or both. In his 195Y review ofjuvenile Hanuman langurs at Polonnaruwa, Sri Lanka cannibalism among birds of prey, Collingwood (S. Ripley, personal communication). In addi- Ingram pointed out that such “fratricide” is tion, Camenzind (1978) reports infanticide by made feasible by marked disparity in age and intruding females among coyotes. Specific se- size. Since staggered production of young is lection pressures favoring xenophobic infanti- probably a parental strategy to allocate repro- cide have not been identified, other than mar- ductive effort according to available resources. ginal benefits to be gained from reduced allowing parents to focus on just one infant if pressures on resources used by adjacent groups. food is in short supply, it becomes difficult in Not surprisingly. xenophobic infanticide occurs some of these instances to separate fractricidal opportunistically. An infant which becomes sep- cannibalism from parentally manipulated infan- arated from its mother during an inter-troop en- ticide. counter may be killed, almost incidentally. in the course of the conflict. By contrast, in the wild dog case cited above-which was recorded Parental Manipulation in detail in a film by van Lawick-the behavior The death of an infant and termination of paren- of the killer was systematic, goal-directed, and tal investment will sometimes improve the to an anthropocentric viewer premeditated, chances for survival of either the mother or ex- since the narrative indicated that the killer isting offspring, or otherwise lead to greater net waited for the opportunities to enter the burrow reproductive fitness of either the mother or the containing the pups unobserved by the mother. father. Such circumstances include the destruc- The paucity of reports of xenophobic killing of tion of imperfect or debilitated offspring whose alien infants may attest to the minute gain to chances of survival and reproduction would be individuals from elimination of infants in other poor: suboptimal ecological conditions or dan- groups, and hence, the weak selection pressure gerous circumstances which might make it ad- for such behavior. vantageous for a parent to defer parental in- The most common and intense competition vestment; existence of older offspring whose for resources not infrequently occurs among future would be jeopardized if parental re- members of the same family. There is an exten- sources were diverted to a new infant; and, mar- Infanticide among Animals 17 riage or mating potentialities, or patterns of re- Ultimately, most instances in which parents source use, that would make one sex of offspring destroy or consume their own offspring relate in less valuable than another. Examples of infants some way to resource availability. For example, being killed or abandoned for each set of con- if resources and parental energy were limitless, ditions can be found in the ethnographic record there would be little (if any) penalty attached to for the human species and have been reviewed rearing defective offspring. The justification for by Richard Alexander (1974) and Mildred Dicke- separating “parental manipulation” and “re- man (1975). Dickeman (in press) explores in source competition” into different explanatory depth the pressures for preferential elimination categories derives from the concept of parental of female infants in stratified human societies effort which combines utilization of material re- where hypergamy (“marrying up”) is the norm. sources with other less tangible costs to the par- In some cases, a mother confronted with ent, such as risk taking and time (Williams, 1966; poor or dangerous conditions for childrearing Low, 1978: and references therein). In poor or may abort, or, where physiologically possible, uncertain environments, continued investment reabsorb the foetus prior to birth (see Baird, by the parent in defective offspring (or alterna- 1945: Roberts and Lowe, 1975, for humans: and tively, nondefective offspring born at a bad Bruce, 1960: Stehn and Richmond, 1975; Mal- time), which had low productive value, would lory and Brooks, in preparation; and Labov. be heavily penalized not only because of the 1978, for rodents). Similarly, it has been sug- environmental resources deflected to a poor bet, gested that the marsspial style of viviparity may but because of the cost to the parent in terms of be an adaptatiun to uncertain environments lifetime allocation of effort in the production and which permits termination of investment in a rearing of offspring. Whether infanticide by par- given infant without the attendant risks of abor- ents is viewed as a subset of “resource com- tion while also providing the mothers flexibility petition” or as a separate category, some dis- in timing the discard (Low, 1978). For example, tinction is worthwhile, since attributes of both a mother kangaroo closely pursued may jettison the perpetrators of infanticide and their victim joeys from her pouch, thereby eliminating an will differ in important ways. In the case of encumbrance and perhaps simultaneously dis- parental manipulation, the victim is a close rel- tracting the predator (Ealey, 1963, cited in Low, ative and age, dependence, quality of infant, and 1978). Even where parental investment is not amount of parental investment necessary for terminated outright, nurture may be reduced, survival are all crucial, while in the case of re- and offspring neglected, as in the case of some source competition, the victim is typically un- humans (Daly and Wilson, in press). related, parents are predicted to resist if present, In rare instances, exploitation of the offspring and vulnerability of the victim is the important as a resource may increase future breeding op- attribute. portunities for the parent. Cannibalism of fertil- ized eggs by a three-spined stickleback father guarding his own eggs may be such an example Sexual Selection (Rohwer, 1978). The male stickleback benefits Competition between members of one sex for from consuming his own genetic investment by reproductive investment by the other sex may staying in competitive condition for subsequent make it advantageous for an animal (usually a brood cycles at the expense of whichever fe- male) to destroy another animal’s offspring. This male’s cytoplasmic investment is parasitized. sexual selection model relies on arguments laid Not surprisingly, parental cannibalism appears out by Charles Darwin (1871) and later refined to be related in some cases to social conditions by Robert Trivers (1972). It predicts that (1) as well as availability of resources (which are infant-killing will be directed at offspring un- often related). The success of litters of Norway likely to be direct descendents of the killer, rats (Rarrrts non~aiclrs) was found to be related thereby reducing the reproductive success of to social rank of the parents. Low ranking, competitors, and (2) that, on average, elimina- scarred females ate more than 60% of their tion of the infant increases the infanticidal’s young, while high ranking, unscarred females male’s own opportunities to breed, typically by weaned all of their offspring (Boice, 1972: cited shortening the interval until next ovulation in in Fox, 1975). the mother of the infant killed. 18 S. B. Hrdy

Typically, as in the case of parental manip- would be a critical factor in reproductive suc- ulation, infants are killed during the period of cess for both males and females who, with luck, maximal parental investment (e.g.. lactation). can produce two litters in a summer. Whereas in the case of exploitation of the infant The conclusive nature of the lemming results, as a resource, or elimination of a competitor for and especially the male’s tolerance for his own resources, vulnerability of the victim is para- offspring, are all the more remarkable given the mount, in sexually selected infanticide. age of artificial conditions of the experiment. Evidence the infant is crucial. from other taxonomic orders, while less precise. In rare instances, competition for reproduc- has been collected under more realistic condi- tive investment by the opposite sex may apply tions. Infanticide has been reported among Ser- to infanticide by females, for example in sex- engeti lions (Pcr77thc~m lao) by Schaller (1972a: reversed species where females compete for pa- 1972b) and Bertrdm (1976) and a diverse array rental investment by males (Trivers, 1972). The of wild primates including species from both the button quail (Turnip s~l~~aticm) may provide such cercopithecine and colobine branches of the Old an example. Females compete for male services World monkeys. the Great Apes, and a single and are responsible for the destruction of eggs genus (A/otrtrt/n) from the generally little-known laid by other females (Hoesch. 1959). New World monkeys (discussed in Section “ln- To date, the only published study which has terspecific Variation in Infanticide”). In both examined predictions of the sexual selection hy- lions and the primate examples. males kill in- pothesis under controlled conditions is that of fants when they enter a new breeding situation Mallory and Brooks with collared lemmings from outside it. Based on limited information (Dicros/on.va groe~rlcrft~icrts). Three sets of I6 available for Hanuman langurs (P~c.rhvris e~te/- females with newly born litters were exposed to /r/s) at Dharwar (Sugiyama, lY65) and Mount either a strange male, or to the “stud” male Abut in western India (see Map I). males ap- with whom the mothers had previously mated. parently do gain from infanticide: in three troops An additional group of females with three-day- where post-infanticide birth intervals were old infants were exposed to a strange male, and known, 70 percent of the females who lost in- finally, I6 females with new litters were allowed fants gave birth within eight months-Just over to rear their infants without interference. as a one gestation period (6.5 months) later (Hrdy. control. Forty-two percent of the day-old new- 1971). Such gains may be habitat-specific. In the borns, and 13% of the three-day-old infants ex- harsh near-desert environment at Jodhpur, the posed to a strange male were killed by him. post-infanticide birth interval from a small sam- None of the infants exposed to their father or in plc of four cases was longer (17 months). and up the control group were killed (Mallory and to 27 months in one extreme case (personal com- Brooks, 1978). Elapsed time till the next litter. munication from S. M. Mohnot). Duration of his own, averaged 71 days when the male killed interbirth intervals, as demonstrated by com- the infants, compared to an average of 28 days parisons between captive and wild primates. is if infants were not killed. As in house mice (M/ts highly dependent on food availability and other rn~sc~l~s), pregnant females exposed to strange factors: monkeys in zoos, for example, typically males may reabsorb foetuses. a phenomenon have shorter interbirth intervals than their wild known as the “Bruce effect.” Quite possibly, counterparts. More field evidence on average the Bruce effect represents a maternal strategy birth spacing in different environments and to reduce-and even partially recoup-her in- under different social conditions is needed to vestment in offspring liable to be killed: Jay specify how much. in terms of reproductive ef- Labov (1978, and in preparation) details evi- fort by females. males gain from infanticide. dence supporting this hypothesis. On a theoretical level. there can be little In the case of collared lemmings. both pre- question that males gain. provided the interval dictions from the sexual hypothesis are met. between births is shortened if an infant dies be- Male attacks are aimed exclusively at unrelated fore weaning. as has been demonstrated for wild infants and death of infants reduces the period savanna baboons (Altmann et al., lY78) and for of time till the mother gives birth again. As Mal- humans (Ring and Scragg. lY73: and other ref- lory and Brooks point out, lemmings live in the erences reviewed in Hrdy. 1977). A recent com- Arctic where the breeding season is short: time puter simulation by Chapman and Hausfater Infanticide among Animals 19 based primarily on data from Dharwar and dwellers. Nevertheless. there is preliminary ev- Mount Abu, calculates the reproductive advan- idence indicating that Thompson’s findings may tage for males who eliminate unweaned infants be relevant to the natural history of this species. at various tenure lengths (cited in Hausfater. Several primatologists have independently re- 1978; and in press). The model makes explicit ported the occurrence of male invasions and the gain to males from infanticide in the langur takeovers among wild M. fhsciwloris (Whea- breeding system under conditions such as those tley. 1978: personal communication from C. and that prevail at Dharwar and Abu where takeo- E. Shurmann). In addition, there are a number vers occur on average every 27 months.’ Ac- of different reports of infanticide for this species cording to the simulation, the infanticidal trait, in captivity (Washburn and Hamburg, 1967: once evolved should sweep through the popu- Angst and Thommen, 1977: personal communi- lation under those conditions. The simulation cation from Elizabeth Larsen of the Primate Re- also points to the existence of tenure lengths at search Colony, Kuala Lampur. Malaysia: and which infanticide will root be advantageous. N. Herrenschmidt of the Primatological Re- If infanticide is a male reproductive strategy, search Station, University of Strasbourg, the question arises of how males avoid killing France). Angst and Thommen have catalogued their own offspring. The cost to a male of killing I6 instances in which caged crab-eating ma- his own offspring would be far greater than the caque males attacked and wounded infants, re- “error” of permitting a competitor‘s offspring sulting in the death of six. Thirteen of the infants to survive. Naturally occurring experiments attacked were offspring of a preceding male; among langurs provide two lines of evidence only three were sired by the male who attacked suggesting that it is the female rather than the them. infant who provides langur males with the cue So far as the human case is concerned, there to attack or tolerate her infant. As reported by is little or no reliable evidence to support the Sugiyama (1965). Mohnot (lY74). and Hrdy hypothesis that human males have been selected ( 1977). infants are sometimes kidnapped from a to murder infants in order to decrease the re- neighboring troop and taken by females back to productive success of competitors and increase their own troop. Kidnapped offspring are unfa- their own-the most widely reported form of miliar, and almost certainly not sired by the infanticide among other primates. As discussed resident male, and yet the infant is never at- above, the majority of known human cases fall tacked while in the possession of familiar fe- into the category of parental manipulation, and males. Secondly. infants that were probably not economic constraints and sex preferences figure sired by a usurper but that were born into a importantly (Alexander, 1974). Typically, the troop after his takeover are not attacked (Sugi- parents themselves decide to terminate invest- yama, 1966: Hrdy, 1977). These observations ment in an infant. Occasionally, a male may kill are compatible with an experiment by Thomp- an infant fathered by some other male (examples son ( 1967) with caged Mnctrca .fa.sc~ic~lrlari.s, the are reported for Eskimoes, Balikci, 1970: among crab-eating macaque, indicating that only the the Tikopia, Lorrimer. 1954: and among the offspring of unfamiliar females would be at- Mundurucu, Murphy and Murphy, 1974, cited tacked. There are of course difficulties in inter- in Hrdy. 1977). In our own society, infants with preting captive observations. particularly since unrelated males in the household may be very infanticide has never been observed in the wild slightly more at risk from abuse (Gil, 1970, dis- among these evasive Southeast Asian forest cussed in Hrdy, 1976: Daly and Wilson, in press) but, by and large, developmental and economic factors in the lives of the adults involved are I It should be noted. however. that in real life there is primary (Kempe and Kempe, l978), and ade- great variation from troop to troop. For example. in the case quate evolutionary explanations for child abuse of a particularly vulnerable troop (Hillside troop, described in Hrdy 1977) the minimum rate of takeovers between I971 and remain elusive. To date, Daly and Wilson (in 1979 was at least every ten months. compared to minimum press) provide the only serious effort to under- estimates for more stable neighboring troops (the School and stand child abuse from an evolutionary perspec- Toad Rock troops) of once every 7.8 years. These estimates tive. 1 know of only one description of the actual were based on the assumption that if the same male was present from year to year, no takeover had occurred (unpub- murder of children by invading males, reported lished field notes of S. Hrdy and D. Hrdy for 1977 and 1979). by an eye-witness among the Yanamamo tribe S. B. Hrdy of Brazil (Biocca, 1971). More commonly, Yan- mals sufficiently stressed under artificial condi- amamo mothers are kidnapped, while their tions, particularly in captivity, may exhibit un- infants left behind are not killed. However (as natural, destructive, and even self-destructive Yanamamo women themselves are aware), this behaviors, to the extreme position that any de- is functionally equivalent to infanticide since structive behavior must be maladaptive (Dolhi- unweaned infants separated from their mothers now, 1977). Hence, infanticide among primates typically starve (Chagnon, 1968). Interestingly, is regarded as an epiphenomenon, a byproduct Yanamamo society is characterized by raiding of abnormally high levels of aggression and so- for women and by the trndency for a few males cial “chaos.” As it is applied by its major pro- to be disproportionately represented in the gene ponents (e.g., see Curtin and Dolhinow, 1978; pool of succeeding generations (Neel, 1970: 1979), the social pathology hypothesis predicts Chagnon, 1972). Apart from the oft-cited Yan- that infanticide should be limited to areas re- amamo case, other instances derive from liter- cently disturbed (typically by humans). Accord- ature (notably, the story of Herod who “sent ing to Curtin and Dolhinow, behaviors, and par- forth and slew all the children that were in Beth- ticularly infant deaths occurring in the course of lehem, and in all the coasts thereof, from two such “chaos,” will not be predictable. Infant years old and under, ” in Matthew 2:16, and the deaths will simply be a consequence of their command of Pharaoh that all Hebrew sons increased vulnerability. should die at birth, Exodus 1:16) or from the rumors and propaganda about genocide that ac- company war. INTERSPECIFIC VARIATION IN INFANTICIDE

Social Pathology In no species is infanticide a common event. Even evolved behaviors, which on average When it occurs, the sequence of events may are adaptive, are not necessarily so under all transpire in the space of minutes. A lethal attack circumstances. Put another way, an adaptive on an infant by an adult male langur which was strategy is probabilistic: animals sometimes witnessed by S. M. Mohnot took under three make mistakes-often for no reason more mys- minutes. Not surprisingly, records of this phe- terious than an inability to predict the future. nomenon are scanty and often incomplete. With Such “mistakes” cannot properly be considered the exception of the exemplary study by Paul pathological and belong to other categories. Sherman-a study which involved a team of re- Technically speaking, the classification *.path- searchers monitoring I.981 animals at a single ological” applies to behaviors which could not site over a period of years-much of the evi- evolve because on balance they decrease the dence on infanticide is based on isolated cases, individual and inclusive fitnesses of infanticidal and on inferences derived from witnessed at- animals. In practice, however, it is extremely tacks which were followed by disappearance of difficult to discriminate such behaviors from infants. Existing reviews of infanticide among mistakes. Take, for example, a captive animal mammals focus on the relatively well-studied who is disturbed by her keepers while giving order Primates (Angst and Thommen, 1977: birth: she responds by devouring her own off- Hrdy, 1977). The majority of these cases involve spring. Is this deranged behavior, or have some infant-killing by males and appear to fit the sex- animals been selected to respond to danger and ual selection pattern: discussion here of inter- potential threats to her infant by eliminating the, specific variation will focus on this evidence current offspring (like the pursued kangaroo only. who jettisons her joey) and deferring further pa- In the roughly 16,000 hours that Hanuman rental investment until a more opportune birth? langurs have been studied, some 46 infants have Despite this difficulty, social pathology has disappeared when males entered troops from been widely invoked to account for infanticide outside. In the type case, reported by Yukimaru among primates (Warren, 1967; Curtin, 1977; Sugiyama at Dhawar (1965a), a band of males Dolhinow, 1977; Curtin and Dolhinow, 1978: usurped a bisexual troop and ousted the resident 1979). Applications of the hypothesis range from male. Subsequently, one of the invaders drove the nearly incontestable position that wild ani- out all other males. This new “leader” bit to Infanticide among Animals 21 death unweaned infants presumably sired by the infants (Suzuki, 1971; Nishida, 1979: Goodall, former resident. Mothers resumed cycling and 1977, reviews this evidence). In virtually all copulated with the usurper (1965b). cases where details were known, the mother of By comparison with Presbytis entellus, other the infant belonged to a different community members of the subfamily Colobinae have from the males who killed her infant. Based on scarcely been studied, but even from prelimi- the reports of her field assistants, Goodall be- nary observations it is apparent that male take- lieves that “attacks which led to the killing of overs and infanticide occur in other colobines. stranger infants were aimed as much (or more Rasanayagam Rudran ( 1973) has reported adult so) at injuring the mother as her infant.” In male replacements accompanied by the wound- support of conclusion that the killings reflect ing or disappearance of infants among purple- xenophobia, Goodall cites (1) observations of a faced langurs (Presbytis senex) in Sri Lanka. chimpanzee which became separated from its Similar patterns have been reported for P. cris- alien mother and was subsequently “rescued” tutu in Malaysia (Wolf and Fleagle, 1977), and by a resident male and (2) observations of at- in a single, problematic case among the rare tacks on a strange female even after her infant Mentawei Island leaf-monkey (P. potenziani) was killed. However, on the basis of similar (personal communication from R. Tilson). A evidence from the Mahale mountains population continent away, in Africa, John Oates has re- of chimpanzees, Toshisada Nishida attributes corded the disappearance of infants coincident infanticide and cannibalism there to reproduc- with male takeovers among Colobus grrereza tive advantages gained by males (1979). (1978). The langur pattern of male takeover and in- Outside of the Colobinae, the highest inci- fanticide is paralleled almost exactly in the case dence of infanticide among wild primates is of redtailed monkeys (Cercopithecus ascanius) among the great apes. As the work of Fossey studied in the Kanyawara district of the Kibale among gorillas (Gorilla gorilla beringei) and forest in Uganda (Struhsaker, 1977; 1978) with Goodall among chimpanzees continues, these two important exceptions. First, Struhsaker’s apes are checking in with a record of infanticide site is one of the least disturbed forest habitats made even more impressive by the compara- where monkeys have ever been carefully stud- tively small numbers of births among apes. In ied. Secondly, the two infants killed were also some 7,000 hours of observation, Dian Fossey partially cannibalized by the new male. Other witnessed three cases of infanticide by adult cases of infanticide occurring after adult male males, and inferred three others. In the best replacements include Rudran’s observations documented cases, the mother subsequently among Venezuelan howler monkeys, Alouatta copulated with the male who killed her infant seniculus ( 1979: see also suspected cases among (Fossey, 1974: 1976: personal communication). A. palliata in Collias and Southwick, 1956). The gorilla reports indicate that infant killing Hence, adult male infanticide has been re- is not necessarily confined to species in which ported for all the major groups of higher pri- males drive out other males and unsurp harems, mates: the Old World monkeys, the New World since the typical pattern for gorillas is for an monkeys (in the genus Alouatta only), and the extra-troop male to “steal” or lure away young Hominoidea (apes and man). These cases appear females from an established group without tak- to primarily fit the sexual selection hypothesis, ing over the group itself. The same point is dri- with several possible exceptions. Patterns of in- ven home by accumulating evidence for infan- fanticide among chimpanzees and redtail mon- ticide and cannibalism among chimpanzees, a keys, where the victim is eaten, might also sug- species with a multimale breeding system in gest exploitation of the infant as a food resource, which females initiate transfers between com- though in neither case is there any estimate of munities. Five cases of infanticide by adult the food value of the infant for conspecifics.’ males have been reported for chimpanzees at Moreover, some of the evidence for infanticide the Gombe Stream Reserve (Bygott, 1974: among howler monkeys fails to fit predictions Goodall, 1977). Researchers at the Budongo for- est in Uganda and in the Mahale mountains of

southern Tanzania have witnessed similar in- * Fox (1975) reviews additional cases of primarily herbiv- stances of adult males killing and cannibalizing orous or granivorous animals which engage in cannibalism. 22 S. B. Hrdy

from the sexual selection hypothesis. Rudran. an infant among strictly seasonal annual breed- who first observed infanticide in this species. ers, such as rhesus macaques. At whatever point favors an explanation based on resource com- he killed the infant. the timing of the mother’s petition. Rudran hypothesizes that howlers are next conception would be roughly the same. food-limited and an incoming male is eliminating Similarly. circumstances promoting infanticide- vulnerable immatures who would subsequently for example. male takeovers resulting in re- compete with his own offspring. Rudran notes placement of one adult male by another-may for example that in several male takeovers ac- not be feasible in marginal environments where companied by infanticide, the usurper did not the energetic costs for one male to aggressively appear to gain increased reproductive opportun- exclude another are prohibitive. An example ities (1979: in preparation). In the best docu- might be the high altitude site of Solu Khumbu. mented of these cases, the usurper was lethally at 3.500 meters in the Himalayas. where invad- wounded in a fight with the former leader, and ing males do not completely drive out their com- it was the former resident who mated with fe- petitors (Boggess. 1977). By contrast. infanti- males whose infants had been killed by his com- cide would be most advantageous under petitor. Future observations should be aimed at conditions where male replacements occurred clarifying whether howler males do or do not on and the tenure of male access to females was on average gain reproductive advantages from in- average short. and where females responded to fanticide, and whether the males in Rudran’s loss of an infant by conceiving again as soon as sample were typical or just “unlucky.” Since possible, regardless of season. females as well as males move between groups One interesting question to arise from a com- in howler societies, one would predict on the parative approach is why infanticide is so rare basis of the resource competition hypothesis among savanna baboonsAspecially since in- that incoming individuals of either sex, not just fanticide can be easily brought about by the males, would kill distantly related immatures if artificial introduction of new males in another the opportunity arose. Furthermore, vulnerabil- member of the genus Ptrpio, the harem-dwelling ity of the victim rather than age should be par- hamadryas baboon (Zuckerman. 1032: Kummer amount. By contrast. in sexually selected infan- et al.. 1973: Angst and Thommen. 1977). In the ticide, the gain to the male is inversely thousands of hours that Pnpio r~r.si~~u.s. P. U~IW correlated with the age of the victim. Among his. and P. c~~r~oc~rplrtrlrrs have been studied, Presbytis enfellrrs. no male has ever been ob- there have been only a very few, isolated cases served to kill a weaned infant. of adult males injuring or killing infants (e.g.. at According to the sexual selection hypothesis, Gilgil, Kenya personal communication from N. infant-killing is a reproductive strategy whereby Nicolson: and at Amboseli. Kenya, in an inci- an invading male increases his own reproductive dent which involved a crippled, older infant, success at the expense of the former leader or personal communication from G. Hausfaterl. dominant male (who would typically be the fa- Such isolated instances have occurred when the ther of the infant killed), the mother. and of infant was some distance from its mother. and course the infant. Close relatives of the infant hence the cost to the male from adult retaliation would be expected to resist: such defense im- greatly reduced. Despite the superior fighting poses a potential cost upon the infanticidal male. abilities of male baboons relative to females. a Hence, on average, the gain to the male must wild male baboon has never been known to incur outweigh such risks if infanticide is to evolve. the risk of attacking an infant in the possession Crucial to the evolution of infanticide are factors of its mother, as wild langur. gorilla, chimpan- which augment the relative gain to a male from zee, and howler monkey males have been eliminating unweaned infants. For example. known to do. Altmann et al. (1978) have shown sexually selected infanticide depends for its ev- that baboon birth intervals are significantly olutionary feasibility on a flexible female repro- shortened by loss of an unweaned infant. Based ductive physiology such that it is both feasible on the work of the Altmanns, Hausfater, and and advantageous for a mother to conceive again others (see Hausfater, 1975; and references soon after the death of an infant. There would therein). it is apparent that male baboons rou- be little reproductive gain for a male who killed tinely move between troops and hence enter Infanticide among Animals 23 troops composed of females whose offspring element in this explanation. but it also relies were sired by some other male. upon the strong affiliative tendencies which What has countered selection pressures for characterize baboon social organization (Smuts, infanticide in these species? Explanations to in preparation). date all focus at some level on the multimale troop organization of savanna baboons. It has been suggested for example that male takeovers INTRASPECIFIC VARIATION IN THE and infanticide may be a phenomenon charac- EXPRESSION OF INFANTICIDE terizing one-male social systems (see for exam- ple, Struhsaker. 1977). However, it is clearly With the exception of savanna baboons, rhesus, insufficient to argue that a multimale breeding and Japanese macaques, infanticide by adult structure by itself precludes the evolution of males is a recurrent phenomenon among each of adult male infanticide since it has not done so those primate species which have been studied among chimpanzees or crab-eating macaques. for ten thousand or more hours (i.e., Hanuman Furthermore. some langur and howler monkey langurs, chimpanzees, gorillas, and howler mon- troops are multimale. To date. three hypotheses keys). It is apparent however that within each have been proposed to explain the absence of of those well-studied species, infanticide is not infanticide among savanna baboons. (I) It is equally likely to occur in all populations. To possible that male baboons entering troops are date, Presbytis entrllus provides the broadest following their fathers and brothers from troop documentation for such intraspecific variation. to troop: hence. the inclusive fitness of a male At Dharwar, Jodhpur, and Mount Abu, male might be reduced by infanticide. This hypothesis takeovers and infanticide appear to be routine predicts an extremely inbred population of ba- or regular occurrences (Sugiyama. 1967: boons. (3) Terrestrial, savanna-dwelling ba- Mohnot. 1971: 1974; Hrdy, 1974: 1977: Mak- boons are uniquely dependent on cooperation wana. in press). Calculations for Dharwar and from other adult males in defending their troop Mount Abu indicate that successful male inva- from predators such as leopards (as is the case sions and troop takeovers occur on the average at Amboseli. for example). If having a repro- once every 27 months. Male takeovers associ- ductive stake in the troop (such as the presence ated with the disappearance of infants have also of his own offspring) is one of the incentives for been reported for langurs at the Gir forest male defense (as suggested by Popp. 1978). then (Starin. 1978) and at Polonnaruwa (personal any male who eliminated the offspring of his communication from S. Ripley). Infanticide has rival would jeopardize effective cooperation in also been witnessed among langurs at the game troop defense. That is. any male who killed off- sanctuary of Sariska by the warden, Fateh Singh spring belonging to a male still present in the (personal communication). By contrast. neither troop would reduce the stake that male had in takeovers nor infanticide have been observed risking his life to defend the troop. Hence, all among langurs at orcha, Kaukori. or Singur, infants in a troop containing an infanticidal male North India. nor at Solu Khumbu or Melemchi (including his own) would have been less likely in the Nepal himalayas (Jay, 1965; Oppenhei- to survive. This hypothesis predicts that re- mer, 1977; Boggess, 1977; Bishop, 1975). (See moval of a male‘s infants from the troop will Table I for site descriptions). decrease his participation in troop defense. (3) Currently, there are three hypotheses to ac- Based on her field observations at Gilgil, Kenya, count for differences recorded between sites: Smuts has suggested that a baboon male would I. Populations of langurs may be gentically be forestalled from killing another male’s infants dissimilar in regard to the trait of infanticide. by the protective presence of other adult males Such differences could be a matter of geographic and females. According to Smuts. female rela- or historical accident, or else due to specific tives who mobbed the attacking male would environmental conditions which countered ev- quickly be joined in their defense of the infant olution of the trait. Differences could encompass by other adult males. including relatives, con- areas as large as eastern and western sides of sorts, or potential consorts of the mother. the subcontinent. Consider for example that in- Clearly, the presence of multiple males is an fanticide has been reported at four sites, and is hhhhhhhh

Table 1. LkscriDtion of laneur studv sites mentioned in text Population Home Altitude Rainfall (cm) density Mean troop range Site and Source 0-n) [% monsoon] (per km’) Habitat type size [range] (km*)

Melemchi. 2.442- 200 ca. 16 Himalayan 2.1 Nepal 3;050 [measured only temperate [I tzop1 (Bishop, 1975) during monsoon] forest Solu Khumbu, 2.591- I Mixed evergreen II 12.7 Nepal 3,505 and deciduous (Curtin, 1975: forest and Boggess. 1976) meadow Kaukori, U.P. 122 76-127 2.7 Dry scrub 54 7.8 (Jay, 1965) [70%-80%] forest [I troop] Jodhpur, 241 I8 Arid open Garden: Rajasthan [9& scrub [82;2] 0.6-0.96; (Mohnot, 1971a.b; 1974; Open personal communication: rocky: Makwana, in press) 0.74-1.3 Sariska. 400 80 I04 Thorn forest 64 0.6 Rajasthan (Vogel, 1971; 1973; 1976; Sankhala, 1978: Fateh Singh, personal communication) Abu, Rajasthan 1.150- 200 50 Subtropical 0.3 (Blaffer Hrdy, 1977a) 1,240 [9Ml deciduous [ I?301 forest Singur. West 7 I46 5-20 Village I3 0.04 Bengal [90%] [2 troops] (Oppenheimer. 1977: 1978 and personal communication) Gir Forest, Gujarat 226- 108 I l5- 128 Riverine 0.2 (Rahaman, 1973; Starin, 648 deciduous [i-:8] 1978; S. Berwick, personal forest communication) Orcha, M.P. 762 203 2.7-6.0 Moist 22 3.9 (Jay, 1965) [75R] deciduous forest Dharwar. SC 76-127 84-133 Dry I4 (forest) 0.2 Mysore 760 [905?cl deciduous 17 (open) (Sugiyama. 1964: forest Yoshiba, 1968) Polonnaruwa. 60 127-191 IO@200 Mixed evergreen 0.25 Sri Lanka [75%] and deciduous (Ripley, 1965: 1967b: forest personal communication)

suspected at two other sites which lie towards The main reason for rejecting an explanation the west of the Indian subcontinent, while in- based solely on genetic differentiation between fanticide has never been reported at any of the populations of langurs is the failure of morpho- five sites which lie towards the east (Fig. I). logic differences between populations to accord However, the fact that there are no known geo- with any behavioral differences. (Fig. 2 shows graphic, climatic, or population boundaries be- subspecific differentiation of Presb.vtis entelltrs tween these two areas makes the proposition based on Hill, 1939; at that time, the synonym that langurs are divided into two populations Semnopithecus was in use). Langurs at Mount unlikely. Alternatively, local circumstances Abu and Jodhpur (Presbytis entellus entelllts) could have selected or failed to select for the are easily distinguished from those at Dharwar trait in numerous small populations. (P. e. achures) by mode of tail carriage and other Infanticide among Animals 25 morphological differences (Roonwal, in press). of long-distance vocalizations (the male whoop), There are no known differences however, in the reduced estrous signaIling, increased huddling, range of behaviors exhibited by these two sub- and increased seasonality of breeding (Vogel, species. To date, the only subspecific differ- 1973; Bishop, 1975; reviewed in Bishop, in ences which parallel different behavior patterns press). are those between Himalayan langurs and the 2. Populations of langurs could be genetically rest of the langur on the subcontinent of India similar in respect to behavior, but the expression and Sri Lanka. These behavioral differences in- of specific traits is facultative. That is, expres- clude reduced frequency of aggression, in- sion of the trait occurs along a behavioral scale creased male-male tolerance, reduced incidence (Wilson, 1971; 1975) in response to specific en-

Figure 1. Langur study sites; note that sites where takeovers have been reported lie to the west of line, while sites without takeovers lie to the east.

\ I .Simla \ 1 l Mussoorie

‘Bhimtal \i .Melemchi

.Sariska

Jaipur Jodhpur, l

Mount Abu .

l P’olonrGruwa 0, , , , , 5yO Ym .B?nda:a

- c \ 26 S. B. Hrdy

S. h. l llua S. h. aonoamf

S. h. durwmiorl~

S. h hypobucor,

Figure 2. Distribution of races within the genus “Semnopithecus.” according to Hill (1939). Note that a new taxonomic classification of Prrshyris r~~,//r,s is being prepared by C. Vogel.

vironmental or demographic conditions. Again portance of a nontypological perspective and the variation could be due to environmental differ- need to focus on the strategies of individuals ences on a large scale, as between eastern and which may differ greatly even between closely western sides of the subcontinent. or, as seems related animals belonging to the same species. more probable, to differences between sites or At present. the chief candidate for determin- even between populations or groups at the same ing behavioral differentiation between popula- site. Markedly different social organization in tions or groups of langurs are human disturbance adjacent groups of acorn woodpeckers (Mala- and population density. Several authors, includ- nerpres formiciwrous) provides a case in point ing Sugiyama (1967). Eisenberg et al. (1972), (Stacey and Bock, 1978), and stresses the im- Rudran ( 1973). and Hrdy (1974) have suggested Infanticide among Animals 27 that the frequency of male takeovers among leaf- on local predators. As shown in Table 3, the eating monkeys is a function of population den- mean score for human disturbance at five sites sity. As Rudran points out: “there are reasons without male takeovers was not significantly dif- to indicate that male replacements occur both at ferent (using a Mann-Whitney U-test) from the high and low densities, but the frequency of average score of five sites where male takeovers aggressive interactions between males which have been reported and where infanticide was lead to male replacements varies with the den- either observed or suspected. Assuming that the sity” (1973, p. 179). Not only would the occa- measures used do indeed reflect the degree of sion for aggression be increased, but frequent human disturbance in the habitat, these findings encounters between troop and non-troop males indicate that other factors besides habitat dis- would wear down males in troops, making them ruption are at issue. Far more promising is a more vulnerable to ouster. Furthermore, in most comparison of mean population density at five cases. there will be a larger absolute number of sites without male takeovers (7.31 langurs per extra-troup males at high densities, and larger square kilometer) compared with five sites average sizes for male bands: several authors where male takeovers were reported (89.6 per have suggested that size of male bands, which square kilometer) (see Table 4). Mean popula- travel over wide areas and contact a number of tion density was more than twelve times higher different bisexual troops, may be related to suc- at sites with takeovers. and this difference was cess rates for takeovers (reviewed in Hrdy, highly significant (p < 0.01 using a Mann- Whit- 1977). Any of these factors, singly or in combi- ney U-test). nation, would contribute to a shorter duration Because Christian Vogel’s field site at Sar- of male tenures in bisexual troops at high dens- iska was not scored for human disturbance in ities. the Bishop study, it was omitted from the above Alternatively, it has been argued that the key analysis. Nevertheless, information from lan- variable determining the occurrence of the take- gurs living at this game sanctuary in western over infanticide pattern is disturbance of the Rajasthan are consistent with the finding that habitat by humans, which creates an “abnor- population density rather than habitat disturb- mal” environment for the primates living in it. ance is correlated with the takeover-infanticide That is, these are the “disturbing results when complex. Sariska is a protected sanctuary, and (a species) is pushed beyond the range of its one of the few areas in India where tigers are flexibility” (Curtin and Dolhinow, 1978: 1979). still common. Other predators include leopards One difficulty with testing this widely held belief and jackals (Sankhala, 1978). According to has been a tendency in the literature to define Vogel (1971, 1973, 1976), population densities at as “crowded,” “abnormal,” or “disturbed” Sariska are approximately 104 langurs per sites where infanticide has been reported, with square kilometer. roughly equivalent to the no attempt at objective classification of study densities recorded by Sugiyama at Dharwar in sites based on tangible features of the habitat 1960 and about twice the density recorded for (see Curtin. 1977: Dolhinow. 1977). Hence. Mount Abu in the early seventies (Hrdy. 1974). Dharwar, Jodhpur, and Abu are characterized [Since these estimates, the population of langurs as “crowded,” “stressed.” “disturbed,” and at Dharwar has decreased, while that at Mount “harassed” populations, while Kaukori, Orcha, Abu has increased (Sugiyama and Parthasara- and the two high-altitude sites at Melemchi and thy, 1979; S. and D. Hrdy, unpublished field Solu Khumbu are referred to as “normal.” notes for 1979). Although 100 langurs per square Only objective measures of human disturb- kilometer is at the high end of the distribution ance, however, will allow us to ask whether of population densities for this species, it is male takeovers are indeed associated-much questionable whether the designation “crowded” less caused-by habitat disturbance. A first ef- (Cut-tin. 1977: Curtin and Dolhinow, 1979) is an fort towards such a classification is provided by appropriate one. Oates ( 1974) found population Bishop et al. (in press). In Table 2, ten langur densities of roughly 100 animals per square kil- study sites are scored according to the degree ometer among the related species Colobrrs grwr- of human influence on the habitat, harassment r:cr living in the Kanyawara forest, part of the of the animals by humans, habituation of the Kibale Forest Reserve in Uganda, one of the animals to humans, and finally, human influence least disturbed areas monkeys have ever been 28 S. B. Hrdy

Table 2. Qualitative rating of degree of human disturbance at ten langur study sites”

Disturbance to Disturbance to Habituation Presence Site habitatb anima& to humansd of predato& Mean score

Singur 4 4 4 3 3.75 Abu town 4 4 4 3 3.75 Jodhpur 3 3 3 3 3.00 Kaukori 3 2 3 3 2.15 Polonnaruwa 2 2 3 3 2.50 Dharwar forest 2 2 2 3 2.25 (16.2-30 km. from Dharwar) Melemchi 2 2 2 2 2.00 Solu Khumbu 2 2 2 2 2.00 Gir Forest 2 2 2 I 1.15 Orcha Forest I I I I 1.00

a AU scores are from Bishop et al. (in press). C. Vogel’s sites at Sariska and Bhimtal were not included because the authors were nor personally familiar with those sites, nor able 10 discuss them directly with the German investigators. b I. Undisfurbedforesr-primary or mature secondary forest, no human habitations or economic activities: no livestock or dogs. 2. Disrurbed Joresr-moderate lo extensive exploitation of habitat (e.g.. woodcutters or tourists): some livestock may be present in the habitat. 3. lnrermediore hobifor-habitat essentially man-made (i.e.. fields, cleared areas) and forested areas adjacent to human settlement. 4. Urban-monkeys live in continuous close association with human habitations or activities. c 1. Animals ore undisturbed-contact between humans and monkeys rare. 2. Minimal harossmenf-monkeys are chased or harassed only when they enter fields lo raid crops. 3. Occasion-specific horossmenr-monkeys are harassed when they steal food and when they enter area of human habitations. 4. Infense harossmenf-harassment occurs frequently, on a daily basis. d I. Wild-monkeys flee when humans appear. 2. Moderately wild-monkeys move away if humans actually approach; they may crop raid, but are not provisioned. 3. Habifucrred-monkeys are accustomed to human presence, although they will usually not tolerate approach unless food is thrown to them. 4. Habituated and routinely commensol-monkeys do not move away ar benign human approach and typically live on human refuse and provisioning. ’ I. Fu// complemenr ofpredarors-large cats and other predators are protected. 2. Partial complemenr of predators-predators are represented by a few individuals of some species. but numbers are diminished due to hunting or habitat destruction. 3. Impoverished complemenf of predators-most major predators eliminated; village dogs may harass primates. 4. No predators-neither dogs nor humans present a threat IO monkeys.

Table 3. Mean score for human disturbance in the habitat and mean population density for langur study sites with and without male takeovers

Human Population disturbance density

e I. Singur 3.15” 12.5 per km2 P 2 2. Kaukori 2.75 2.7 X score = 2.3 Y 3. Orcha Forest I .o 4.35 i density = 7.31 5 4. Melemchi 2.0 16.0 7” 5. Solu Khumbu 2.0 I.0

e 6. Abu town 3.15 50.0 9 1. Jodhpur 3.0 18.0 i score = 2.65 8 8. Polonnaruwa 2.5 150.0 X density = 89.60 I 9. Dharwar Forest 2.25 108.5 b IO. Gir Forest I.75 121.5

a Score taken from Bishop et al. Infanticide among Animals 29

Table 4: Number of months troops of langurs were monitored at sites with and without reported takeovers

No. of Sites where no takeovers detected troops X Months

I. Kaukori (Jay, 1965) 1 X 4 z 4 2. Orcha (Jay, 1965) 3 X II a 33 3. Melemchi (Bishop, 1975) 1 X II = 11 4. Singur (Oppenheimer, 1977) 2 X 20 = 40 5. Solu Khumbu (Boggess. 1977) 2 X 29 = 58 146 troop months i = 29.2 if the rate were once every 27 months, we would expect to see it.

Sites where takeovers detected 6. Dharwar (Sugiyama, 1967; 38 X 23 = 874 Yoshiba. 1968) 7. Jodhpur (Mohnot. 1971) 20 X 30 = 600 8. Mount Abu (Hrdy, 1977) 5 X 48 = 248 9. Polonnaruwa(Ripley, in prep.) II X 36 = 396 10. Gir Forest (Stat-in, 1978) 2 X 2 = 4 2,122 troop months X = 424.4 troop months

studied on a long-range basis. Oates also re- ulation density and takeovers is the increased viewed the literature for population densities numbers of extra-troop males traversing the among other colobines in Africa and Asia. Com- home ranges of the bisexual troops at high dens- parable densities (1 animal per hectare) have ities. One striking common denominator be- been reported for Presbytis melalophos, P. ob- tween colobines such as Colobus guereza and sqmu and P. entelh. Somewhat higher dens- Presbytis senex, as well as to some extent P. ities (as high as 2 animals per hectare) have been enfellus, and the other monkeys in which the reported by Eisenberg et al. (1972) for Presbytis male takeover-infanticide complex appears senex. Although Colobus gltereza (a more foli- strongly entrenched (such as Alouatta seniculus vorous monkey than P. entellus, but similar in and Cercopithecus ascanius) is the relatively body weight and social organization) share the high proportion of leaf material in their diets forest at Kanyawara with seven other species of (Hladik, 1978; Kay and Hylander, 1978). An primates, no ecologist to date has suggested that obvious question for future investigators is the monkeys in this habitat are “crowded.” Hence, extent to which a large folivorous component in Szriska would appear to provide an example of the diet allows a population to reach high pop- a relatively undisturbed langur population where ulation densities and to maintain itself at a level a high (but not abnormally high) density of lan- close to the carrying capacity of the environ- gurs is maintained, and where infanticide is ment for extended periods. If hypothesis 2 is known to occur. These findings fail to support correct, this capacity would intensify a tendency the prediction of the social pathology hypothesis towards male takeovers.3 As early as 1972, and that infanticide only occurs in disturbed or un- before it was apparent just how widespread the naturally crowded areas (see also “Reply” in male takeover pattern was among monkeys, Ei- Dolhinow, 1977). senberg et al. pointed out that folivorous pri- These findings are instead consistent with the mates were prone to live in one-male groups. hypothesis proposed by Rudran, Sugiyama, In the light of these speculations, recent de- Yoshiba, Hrdy, and others that the incidence of mographic information from Dharwar (Sugi- male takeovers among leaf-monkeys is higher at high population densities. The mechanism un- J Note that an associaGon between infanticide and high derlying this association is still not known, but population densities is equally compatible with explanations as suggested above, the most obvious link (orig- based on sexual selection or resource competition. This is inally suggested by Yoshiba, 1967) between pop- going to further complicate separation of the two models. S. B. Hrdy

yama and Parthasarathy, 1979). Jodhpur overs and infanticide may occur with equal (Mohnot et al., 19791, and Mount Abu (S. and probability at all sites but have been missed due D. Hrdy, unpublished field notes for 1979) take to differences in observation conditions. on special significance. By and large, the pro- A corollary of high population density at a portion of one-male versus multimale troops in study site is the opportunity for researchers to a given population of langurs is a good index of simultaneously monitor a number of different how frequently male takeovers occur. Typically. troops. Hence. the mean number of troop- when there are frequent male takeovers, the months langurs were monitored at five low-den- majority of troops have only one full adult male sity sites where no takeovers were ever reported (Hrdy, 1977). Whereas population densities at was 29.2. far lower than the mean number of Jodhpur and a predominantly one-male troop troop-months langurs were followed at sites organization have remained constant during the where takeovers were reported (424.4 troop- decade between 1968 and 1978, population dens- months: see Table 5). On the basis of sample ities at Dharwar and Mount Abu have changed. size alone, the chances of an observer recording The population at Mount Abu has grown by a takeover would be higher at high densities. 12.3% in the years between 1971 and 1979 and Nevertheless, examination of Table 5 makes male takeovers continue at a high rate. By con- clear that even though only a few troops were trast, the population of langurs at Dharwar in observed at Solu Khumbu and Singur. the du- 1978-79 had declined by 54.5% since 1961 and ration of these studies should have been suffi- the troops nearest to the town of Dharwar have cient to record takeovers if they were occurring disappeared. Based on an interim census at at a rate comparable to those reported at Dhar- Dharwar in 1976, Sugiyama and Parthasarathy war and Abu (every 27 months). On these concluded that the most serious decreases had grounds, I believe we can discount the possibil- occurred in the last three years and were due ity that takeovers are occurring at the same rate largely to felling of forests and the planting of at all sites. eucalyptus trees, which are not usable for food Having followed these detailed arguments, by langurs.d Although the data gathered by Su- the reader is perhaps entitled to my own opin- giyama and Parthasarthy are not yet fully ana- ion, namely that the second hypothesis is the lyzed, it is apparent that the proportion of mul- most promismg and that, so far as langurs on timale troops has increased, while levels of the Indian peninsula are concerned, animals inter-troop aggression and the number of all- from Singur, Orcha. Abu, and Dharwar reared male bands have decreased. Based on hypoth- in similar environments would exhibit the same esis 2 presented here, one would predict that range of behavior. I would not, however, rule male takeovers and infanticide at Dharwar have out the possibility that Himalayan langurs who also decreased, even though the level of human have evolved under extreme environmental con- disturbance has increased dramatically since Su- ditions at the margin of the species’ range will giyama’s initial study in the early 1960s. be shown to be genetically different and exhibit 3. According to a third hypothesis, male take- corresponding differences in behavior.

* Firmly entrenched in the langur literature is the notion that as the habitat is destroyed. animals become more densely THE IMPLICATIONS OF INFANTICIDE FOR packed togethefin remnant forest. For example. Curtin and FEMALE REPRODUCTIVE STRATEGIES Dolhinow write that “The forest near Dharwar [referring to the early ‘6Os] had recently been cleared and the langars Infanticide has been reported for a broad array concentrated in what little remained, with the result that their of animals in a variety of circumstances. For no population density reached 134 langurs per km’.” (1978, p. species are there precise data on mortality at- 471). Although such a “refugee effect” may occur immedi- ately after felling of a forest, no habitat can continue to sup- tributable to infanticide. though we have a port a population which exceeds the carrying capacity of the known minimum of 8% of infants killed by con- environment. Substantial portions of forest still existed at specifics at Sherman’s ground squirrel study Dharwar through the ’60s and into the early ’70s. and these site, and an estimated minimum of 83% in one forests were capable of supporting high densities of langurs. langur troop at Mount Abu which happened to When extensive destruction of the forest did take place, the population dropped rapidly rather than increased as usable be particularly vulnerable to male invasions. In habitat was destroyed. some cases, chiefly those involving resource Infanticide among Animals 31 competition, the killers may be animals of either ately during interunit interactions” (Harcourt et sex, and age of the victim secondary to vulner- al., 1976. p. 227). ability. In other circumstances, particularly sex- Similarly, Goodall has postulated that chim- ually selected infanticide, males are implicated, panzee females stay close to adult males after and targets are primarily unweaned infants. In the birth of offspring to gain protection against all cases however, a situation permitting infan- attacks on their infants by other females (Good- ticide is liable to be uncommon (i.e., the pres- all, 1977, p. 276). Analysis of association pat- ence of infants coincident with the opportunity terns at the Gombe Stream feeding area indicate to kill them and pressures to do so). This makes that mothers tended to associate more fre- it all the more striking that scientists witness quently with males after the birth of an infant such fleeting episodes as often as they do (see than they did before. Such increases are appar- for example Struhsaker. 1977). One must as- ently due to the mother rather than the male. A sume that although uncommon, infanticide is a parallel pattern has been more tentatively sug- widely recurrent behavior with drastic conse- gested for orangutans by Joel Cohen in his anal- quences for the fitness of individuals. As such, ysis of data collected by John MacKinnon. Typ- infanticide-like classical predation and other ically, an orang mother travels on her own. recurrent perils-is likely to have played a sig- Nevertheless, a female orang is statistically more nificant role in the evolution of infanticidal spe- likely to be seen travelling with a male if she has cies. an infant (Cohen, 1975). Whereas a gorilla male In the case of primates, studies are now avail- may help to care for weaned infants (Harcourt, able for a number of species belonging to the 1977), orang males do not. Neither of these large same family and even for members of the same apes is in much danger from any predator other species inhabiting different locations. This com- than man. Although there are a variety of ad- parative material permits us to search for ad- vantages (as well as costs) to group living (see aptations to the threat which infanticide poses Alexander, 1974; Clutton-Brock and Harvey, for individuals. In this section, 1 will outline l976), protection of infants from conspecifics three areas-social organization, the physiology must be one of the important determinants of of reproduction, and the patterning of female those other animals with which females choose sexual receptivity-which may have been sub- to associate with (Wrangham, in press, makes ject to selection pressure from infanticide. this same point). In situations where there are frequent shifts in group memberships, mothers with infants Social Organization may exhibit clear preferences for one male while What sort of effects will the threat of intraspe- avoiding another. After a takeover, Hanuman cific damage to infants have on patterns of as- langur females may leave their natal troop tem- sociation? Grouping patterns of the great apes porarily to travel with ousted males or even, in provide some of the clearest examples. Based on an extreme case, attempt to follow a male who the work of Fossey, Harcourt, and Stewart at has left one troop to take over another (Hrdy, the Karisoke Research Center in Rwanda, we 1977, pp. 51 and 266). In the gorilla case, female now know that female gorillas, like chimpan- preferences may be influenced by male care, but zees, move between groups. The main attraction this probably is not a factor among langurs, for transferring females, and the main cohesive where the main contribution of males to the focus of gorilla groups generally, is the silver- survival of offspring (at least at high density back male of each group (Harcourt, 1977). Har- sites) is to protect infants from other males. court and others have postulated that one source Infanticide may also have implications for rela- of continuing male attractiveness is his ability to tions among females as well as between the protect the female and any offspring she pro- sexes. As pointed out by Sherman (1978), the duces. That is, “subsequent decisions to stay or threat to offspring from aliens may increase the transfer might rest on the female’s success at benefits of remaining near relatives. Further- raising her offspring in the unit: two mothers more, a hitherto little-considered advantage for who successfully raised their offspring in a new a female’s high rank would be to reduce inter- unit stayed; three who did not, left. The infants ference by other females in her own childrearing of at least two of the latter were killed deliber- while permitting her to harass her competitors. 32 S. B. Hrdy

Among wild dogs and the great apes for exam- other large, infanticidal mammals, such as lions, ple, it was the offspring of subordinate females do abort when new males replace residents (Ber- who were killed by other females in the group. tram, 1975). Because of the greater relative in- Similarly, Mallory and Brooks (in preparation) vestment of a monkey or ape mother per infant, report that caged Dicrostonyx groenlandicus abortion in primates would be a far costlier ma- mothers are better able to defend their one-day ternal strategy than it is among lions. If abortion old litters from the depradations of nonpregnant, of a nondefective infant is ever selected for nonlactating females introduced into the cage if among primates one would expect it to be an the mother is significantly heavier than the option of “last resort”-hence a threefold ques- strange female.5 tion might be the appropriate one for future re- search: (1) Does the risk of abortion increase when a new male is introduced? (2) Does the Female Reproductive Physiology rate of abortion under such conditions differ for The capacity of females to conceive- again infanticidal and noninfanticidal species? (3) Is soon after losing an infant is a crucial preadap- the likelihood of abortion affected by past sexual tation for the evolution of infanticide through encounters’? That is, is a pregnant female less sexual selection. Although it might be advanta- likely to abort if she mated with the usurping geous for species if females delayed ovulation, male prior to takeover’? Although highly specu- and hence “penalized” infanticide, it would be lative, I feel it is important to ask these ques- disadvantageous to the individual female. If any- tions to encourage the collection and publication thing, selection will operate on individual fe- of relevant data from field and laboratory stud- males to ovulate even sooner after the loss of ies. As a parallel example, it is worth noting that an infant. Females can, however, be selected to prior to the 1970s. when considerable attention reduce investment, or even abandon an infant was focused on the phenomenon of infanticide seriously endangered by infanticide. In the case in primates, apparent cases of infanticide went of a number of small mammals, including Ml/s unreported. mliscrrlus. Perornyscrfs manicrtlat~rs. Microtrrs agrestis. M. penn.syl~~anicl:s. M. orchrogaster. and Clethrionom>~s glareolus (reviewed in Mal- Female Sexual Behavior lory and Brooks, in preparation and Labov, in Among mammals where estrous behavior and preparation), females exposed to a strange male imminent ovulation are less well correlated than early in their pregnancy may reabsorb the foe- is the case among rodents, counter-strategies tus, and it seems possible that this is a strategy less costly to females may be feasible. If males to reduce the risk of infanticide when the pups are in fact using past history of sexual relations are born (Stehn and Richmond, 1975; Hrdy, with the mother as a cue to either attack or 1977, p. 308: Mallory and Brooks. 1978: and in tolerate the infant, females capable of exhibiting preparation; Labov, 1979. and in preparation.). estrous behavior at all stages of the reproductive Most of these strategies however mean consid- cycle possess some control over information erable loss in investment for females. available to males concerning possible paternity. To date no one has investigated for any pri- Provided extra-troop males cannot detect ovu- mate species whether or not introduction of a lation. it would behoove females to enter into new male alters the risk of abortion. Absence Of consortships with potential usurpers. Such a information from field studies, at this point, can strategy, for example, could account for both only be taken as evidence for the difficulty of the behavior of female langurs who leave their collecting information on incomplete pregnan- troop to travel and mate with males in all-male cies under field conditions. As Labov (in prep- bands (Hrdy, 1977; personal communication aration) has pointed out, females among some from Moore and Mohnot), and for the onset of semicontinuous estrous behavior when male in- vaders approach or enter langur troops. At such J Mallory and Brooks (work in preparation) tentatively times, females without infants may solicit males attribute this infanticide to competition among female Mus regardless of their cyclical state, and even if musculus for territories. As in the case of Belding‘s ground squirrels. a female who loses her litter moves away to a new they are already pregnant. Unlike baboons, and nest site. other primates with sexual sweliings or colora- Infanticide among Animals 33 tion keyed to the menstrual cycle, a langur fe- tive outside of that period, even though males male signals receptivity by presenting to a male attempted to mount. In contrast, among pig- and shuddering her head. Except for the possi- tailed macaques (Macaca nemeslrina), female bility of pheromones (which have never been receptivity did not change across the cycle, and studied in langurs), the male apparently depends the mean number of intromissions did not vary on behavior signals. Female langurs are then in significantly across the cycle. What changed, a relatively good position (as are human females) however, was the “attractiveness” of females, to confuse the issue of paternity, particularly since males inspected the vaginal area and eja- with extra-troop males.6 culated more frequently at midcycle. For both The hypothesis that pseudo-estrus is a strat- pig-tailed macaques, and rhesus macaques (Mi- egy to forestall infanticide predicts that (1) males chael and Zumpe, l970), the only measure that who have copulated with a female prior to her has been reliably correlated with a particular giving birth will be less likely to subsequently stage of a female’s reproductive cycle is fre- attack an infant associated with her, and that quency of ejaculation, which was highest at mid- (2) females will solicit extra-troop males both at cycle. Data from free-ranging rhesus macaques ovulation when conception is possible, and at corroborate the finding that females in a rela- other times, when it is unlikely or not possible. tively normal social setting copulate throughout This second prediction, already partially sub- the cycle (Conaway and Koford, 1965: Loy, stantiated for langurs, is an important one be- 1970). Similarly, both captive and wild gelada cause it allows us to exclude competing hypoth- baboons copulate throughout the cycle (Dunbar eses which would explain female promiscuity, and Dunbar, 1974; Smith and Credland, 1977). namely that females are increasing their options Unlike the macaques, the gelada males were not to choose between males by preventing a dom- more likely to ejaculate at midcycle than at other inant animal from monopolizing them, or sec- times, but females were significantly more likely ondly, that females copulate with a number of to solicit males in the ten days or so when the different males to insure fertility, or thirdly, that vesicles on the bare skin of the gelada chest and such promiscuity is a mechanism to insure out- perineal region were maximally swollen and pig- breeding. mented than they were in the ten days surround- In the course of primate evolution, there has ing menstruation (Dunbar, 1978). Among cap- apparently been a trend away from strictly hor- tive orangutans (Nadler, 1977) and captive monal determination of receptivity. Based on an chimpanzees (Lemmon and Allen, 1978) females experimental study, Eaton (1973) provides an copulated on any day of the cycle, although instructive comparison of the invariant relation- there were increases in frequency at midcycle. ship between ovarian hormones and mating in On rare occasions, a wild chimpanzee has also a nocturnal African prosimian (Galago crassi- been known to lapse from cyclicity into semi- caudarus) and the more flexible breeding of continuous receptivity (Tutin, 1975, p. 59). higher primates. Among galagos, Eaton found Among humans, females are more or less con- a mean cycle length of 44 days, with a period of tinuously receptive: tendencies towards in- estrus (as determined by the presence of comi- creased sexual activity at midcycle (comparable fied epithelial cells in vaginal smears) of I2 days. to those reported for macaques and some apes) Within these I2 days, females were receptive have been reported among women in both Wes- (i.e., permitted intromission) for a mean of only tern and gathering-hunting societies (Udry and 5.8 days, and no galago female was ever recep- Morris, 1968; Adams et al., 1978; Worthman, 1978). Research by Adams et al. (1978) is par- ticularly significant because the study was spe- cifically designed to focus on female-initiated e Resident males. however, are in a position to monitor sexual activity. Results from earlier studies have the intensity of solicitations from cycle to cycle and to com- pare estrus at ovulation with pseudo-estrus, just as a human been blurred by the confounding effects of male- observer watching langurs can: this may be one of several initiated sexual relations in the human case. reasons why resident males are more discriminating in their The lack of any strict correlation between mating activities. A soliciting female ignored by a resident ovulation and sexual receptivity in higher pri- male is frequently sought after by extra-troop males (Hrdy. 1977. pp. 137-141; l&mm film by D. Hrdy, S. Hrdy, and J. mates has been known since the turn of the Bishop, entitled “Stolen Copulations”). century when Walter Heape (1900) presented his 34 S. B. Hrdy exhaustive report on estrus in mammals. Heape mates, including humans, ovulation is best dis- concluded that in respect to estrous behavior guised among this group. Female marmosets monkeys stood in an intermediate position be- copulate at low level throughout the cycle, ex- tween other mammals and man. It has been left hibit no visible signs of ovulation. and do not for modern workers belatedly to agree. Pressure menstruate (Hearne, 1978). to do so was applied by Thelma Rowell in her Nevertheless. the Morris proposition is review of primate reproductive cycles (1972) clearly insufficient to account for all the evi- which emphasized that there was no necessary dence, and particularly for the precursors of correlation between copulation and stage of the continuous receptivity among a number of non- menstrual cycle. human primates such as langurs (Hrdy, 1977) These findings raise important issues for pri- and vervets (Struhsaker, 1967) which are char- mate evolution. Under what conditions do fe- acterized by neither pair bonds, unusually help- males shift away from cyclical determination of less young, nor male care. Furthermore, the hy- receptivity towards socially determined or situ- pothesis based on female choice cannot account ation-determined receptivity’? And. in particu- for the widespread occurrence of post-concep- lar, under what conditions do female primates tion estrus among langurs. Japanese and rhesus benefit from copulating at times when they will macaques. patas monkeys, gorillas. Sykes mon- not conceive’? keys. and chimpanzees (reviewed in Hrdy. 1977: There are at least three possible ways in pp. 7X4-285) because in such cases male genetic which females might benefit from concealing contribution to offspring is not at issue. although ovulation and from mating throughout the cycle. his subsequent behavior to offspring will be. according to circumstances with any of several In addition to female choice. the cultivation or a number of partners. Briefly. these involve: of male investment, and inhibition of infanticide, increasing options from which to choose an ad- there is of course a fourth possible explanation vantageous genotype (Trivers. 1978): cultivating for continuous receptivity, namely that it is an additional male investment-as is the case for endocrinological byproduct of hormonal changes barbary macaques, MUC~IUISJ~IYIIIO. where care which evolved for some other reason. This al- from several different males is essential for the ternate hypothesis will be difficult to rule out. survival of infants (Taub, in press): and, fore- However. the hypothesis will be weakened if a stalling infanticide (Hrdy. 1974). Since females female primate is more likely to switch from may copulate independently of ovulation in mul- cyclical to continuous receptivity under specific timale troops (e.g.. macaques). one-male harems social conditions. such as those conducive to (e.g., langurs), and in monogamous pairs (e.g., infanticide. Barring endocrinological accident, marmosets), and since this phenomenon occurs the strongest hypothesis to account for the in both species with and without extensive pa- emergence of continuous receptivity among pri- ternal care, it would be an error to insist on any mates is that it first evolved in social contexts unitary explanation for continuous receptivity. where it was beneficial for females to confuse According to the classic anthropological for- paternity, as in the case of infanticidal species. mulation proposed by Morris ( 1967). continuous This sexuality would constitute the physiol- receptivity evolved in human females to main- ogical heritage that prehominid females brought tain the pair bond and to insure paternal care to th revolutionary new lifestyle that distin- necessary to rear particularly altricial human guished our hominid ancestors from other pri- young (see also Barash, 1977). This proposi- mates-a life centered about a home base, char- tion-largely untestable-may well be true. Cer- acterized by division of labor into gathering and tainly. it accords with such findings as an in- hunting, and an emphasis on food sharing (Isaac. crease in libido among human females during lY78). Out of gathering-hunting economics the first, and particularly the second trimester would grow the new dimensions to the pair bond of pregnancy (Masters and Johnson, IY66, p. postulated by Sahlins ( IYSY), Morris ( 1967). and l54- 159), and it accords with some of the com- others. The helplessness of human young would parative evidence for other primates. For ex- place strong selective pressures on males to help ample, marmosets are a group of New World rear their young. But continuous. situation-de- monkeys characterized by both monogamy and pendent sexual receptivity would have predated extensive male care of offspring. Of all the pri- those developments. Infanticide among Animals 35

I thank Richard Alexander, Naomi Bishop, John Ei- Bishop, N., Hrdy, S.B., Moore, J. Teas, senberg. Paul Harvey. Dan Hrdy, and especially Jim J. Qualitative and quantitative definitions of Moore for valuable criticisms. Dan Hrdy prepared human influence in habitats of South Asian Figure 2. 1 am particularly indebted to Rasanayagam monkeys. Manuscript submitted for publica- Rudran and to Paul Sherman for discussingwith me tion. their ideas on the relationship between infanticide and Boggess, J. The social behavior of the Himala- competition for resources: both workers have ex- yan langur (Presbyris etzrellus) in Eastern plored this topic in greater depth than I have. and their Nepal. Ph.D. dissertation, Univ. of Califor- findings have broadened my perspective. I am also nia, Berkeley, 1976. grateful to Glenn Hausfater. Jay Labov, Frank Mal- Bourliitre, F.. Hunkeler. C., Bertrand, lory, Jim Moore, Nancy Nicolson, M. L. Roonwal, R. M. Ecology and behavior of guenon (Cer- Rudran. Barb Smuts, and Ron Tilson for sharing with copirhectts cclmpbelli lobvei) in the Ivory me unpublished observations and/or manuscripts. I Coast. In Old World Monkeys, J.R. Napier thank Bob Trivers for years of inspired teaching and and P.H. Napier (Eds.). New York: Aca- for the button quail reference. Presentation of an demic Press, 1970. abridged version of this paper at the Vllth Interna- Bruce, H.M. A block to pregnancy in the house tional Congress of the Primatological Society in Ban- mouse caused by the proximity of strange galore, India was made possible by the Smithsonian males. Journul of Reprodtrcliotl nnd Fertility Institution. I: 96- 103 (1960). Bygott, D. Cannibalism among wild chimpan- zees. Ntrtttre 238: 410-41 I (1972). REFERENCES Bygott, D. Agonistic. behaviour and dominance in wild chimpanzees. Ph.D. thesis, Univ. of Adams. D.B.. Gold, A.R. Burt, A.D. Rise in Cambridge, 1974. female-initiated sexual activity at ovulation Calhoun, J. Population density and social pa- and its suppression by oral contraceptives. thology. Sci. Am. 206: 139- 148 (1962). Nut, Ettglutld Jotrrnul of Medicitlc 229(2 I): Camenzind. F.J. Behavioral ecology of coyotes ll45-1150(1978). on the National Elk Refuge, Jackson, Wyo- Alexander. R.D. The evolution of social behav- ming. In Coyotes: Biology, BehuL1ior und ior. Annual Rrlieic, of Ecology and Sgstetn- Manugemenr. M. Bekoff (Ed.). New York: atics 5: 325-383 (1974). Academic Press, 1978. Altmann, J., S. Altmann, S., Hausfater, Carpenter. C.R. Societies of monkeys and apes. G. Primate infant’s effects on mother’s fu- Biol. Symposia 8: 177-204 (1942). ture reproduction. Science 201: lO28- 1029 Chagnon, N.A. Yunotncrmij, the Fierce People. (1978). New York: Holt, Rinehart, and Winston, Angst, W., Thommen. D. New data and a dis- 1968. cussion of infant killing in Old World mon- Chagnon, N.A. Tribal social organization and keys and apes. Folitr Pritnmt. 27: 198-229 microdifferentiation. In The Struclure oj (1977). Human Popularions, G.A. Harrison and A.J. Baird, Sir D. Social and economic factors on Boyce (Eds.). Oxford: Clarendon Press, stillbirths and neonatal deaths. J. of Obsrrr- 1972. rics and Gyncrecolog~ of the British Empire Clutton-Brock, T.H., Harvey, P.H. Evolution- 52: 217-234 (1945). ary rules and primate societies. In Growing Balikci, A. The Nersilik Eskimo. New York: Points in Ethology, P.P.G. Bateson and R.A. Natural History Press. 1970. Hinde (Eds.). Cambridge: Cambridge Univ. Barash, D. Sociobiology and BehccL*ior. New Press, 1976. York: Elsevier North Holland, 1977. Bertram, B. Kin selection in lions and in evo- Cohen, J. The size and demographic composi- lution. In Gro~viq Poitirs iti Ethology, P.P.G. tions of social groups of wild orangutans. Bateson and R.A. Hinde (Eds.). Cambridge: AnOnal Behuliottr 23: 543-550 (1975). Cambridge Univ. Press, 1976. Collias, N.. Southwick, C.H. A field study of Biocca, E. Yanoama. New York: Dutton, 1971. population density and social organization in Bishop, N. Social behavior of langur monkeys howling monkeys. Proceedings of rhe Amer- (Presbyris entelltts) in high altitude environ- icun Philosophical Society, 96: 143- 156 ment. Ph.D. dissertation, Univ. of California, (1952). Berkeley. 1975. Conaway, C.H., Koford, C.B. Estrous cycles Bishop, N. Himalayan langurs: temperate co- and mating behavior in a free-ranging band lobines. Jotrrnul of Human Evolution (in of rhesus monkeys. Journul of Mammalogy press). 45: 577-588 (1965). S. B. Hrdy

Curtin, R. Langur social behavior and infant Fossey, D. The behavior of the mountain go- mortality. Kroeber Anthropological Society rilla. Ph.D. thesis, Cambridge University, Papers 50: 27-36 (1977). 1976. Curtin, R., Dolhinow, P. Primate social behav- Fox, L.R. Cannibalism in natural populations. ior in a changing world. Am. Sci. 66: 468-475 Annual Review of Ecological Systems 6: 87- (1978). 106 (1975). Curtin, R., Dolhinow, P. Infanticide among lan- Frame, L., Frame, G. The female African wild gurs-a solution to overcrowding? Science dogs emigrate. Nature 263: 277-229 (1976). Today (published Bombay, India), 13(7): 35- Gil, D. Violence Against Children. Cambridge, 41 (1979). Mass.: Harvard University Press, 1970. Daly, M., Wilson, M.I. Abuse and neglect of children in evolutionary perspective. In Nat- Gilbert, J. Sex-specific cannibalism in the roti- ural Selection and Social Behavior, R.D. fer Asplanchna sieboldi. Science 194: 730- Alexander and D.W. Tinkle (Eds.). In press. 732 (1976). Darwin, C. On the Origin of Species. New Goodall, J. Infant killing and cannibalism in York: Atheneum, 1859 (1967 fascimile). free-living chimpanzees. Folia Primat. 28: Darwin, C. The Descent of Man and Selection 259-282 (1977). in Relation to Sex. London: John Murray, Harcourt, A.H. Social relationships of wild 1871. mountain gorillas (Gorilla gorilla beringei). Deag, J.H., Crook, J. Social behavior and Ph.D. thesis, Cambridge University, 1977. ‘agonistic buffering’ in the wild barbary ma- Harcourt, A.H., Stewart, K.S. Fossey, D. Male caque Macaca s$\,anas. Folia Primat. 15: emigration and female transfer in wild moun- 183-200 (1971). tain gorilla. Nature 263: 226-228 (1976). Dickeman, M. Demographic consequences of Hausfater, G. Dominance and reproduction in infanticide in man. Annual Reliew ofEcology baboons (Papio cynocephalus) Contrib. to and Systematics 6: 107- 137 ( 1975). Primato1og.v 7. Basel: S. Karger, 1975. Dickeman, M. Female infanticide and the re- Hausfater, G. Life history strategies in long- productive strategies of stratified human so- lived species. Paper presented at Symposium cieties: a preliminary model. In El~ollltionary on Natural Selection and Social Behavior at Biology and Human Social Behavior. N.A. the University of Michigan, Ann Arbor, Oc- Chagnon and W.G. Irons (Eds.). North Sci- tober 16-18, 1978. tuate, MA: Duxbury Press, in press. Heape, W. The “sexual season” of mammals Dolhinow, P. The North Indian langur. In Pri- and the relations of the “pro-oestrum” to mate Patterns, P. Dolhinow (Ed.). New menstruation. Quart. J. Microsc. Sci. 44: I- York: Holt, Rinehart, and Winston, 1972. 70 (1900). Dolhinow, P. Normal monkeys?A,n. Sci. 6: 266 Hearne, J.P. The endocrinology of reproduc- (letter) ( 1977). tion in the common marmoset Callithrix jac- Dunbar, R.I.M. Sexual behavior and social re- chas. In The Biology and Conservation of the lationships among the gelada baboon. Animal Callithricidae. D. Kleiman (Ed.), Washing- Behaliour 26: 167- 168 (1978). ton, D.C.: Smithsonian Institution, 1978. Dunbar, R.I.M., Dunbar, P. The reproductive Hill, W.C.O. An annotated systematic list of the cycle of the gelada baboon. Animal Beha,*- leaf-monkeys. Ceylon Journal of Science iorrr 22: 203-210 (1974). (Colombo), ser. B., 21: 277-305 (1939). Eaton, G. Social and endocrine determinants of Hinde, R., Spencer-Booth, Y. The behaviour of sexual behavior in simian and prosimian fe- socially living rhesus monkeys in their first males. Symp. IVth Int. Congr. Primat. (Vol. two and a half years. Animal BehalYorlr 15: 2: Reproducti\ne Beha,,ior). Basel: S. Karger, 169- 196 (1967). 1973, pp. 20-35. Hladik, C.M., Adaptive strategies of primates Eisenberg, J., Muckenhern, N., Rudran, R. The in relation to leaf-eating. In The Ecology of relationship between ecology and social Arboreal Folilvres, G.G. Montgomery (Ed.). structure in primates. Science 176: 864 Washington: Smithsonian Institution Press, (1972). 1978. Fossey, D. Development of the mountain go- Hoesch, W. von Zur Biologie des siidafrikan- rilla (Gorilla gorilla beringei) through the first ischen Laufhtihnchens Tarnix sylvaticu le- thirty-six months. Paper presented at Berg purana. J. Orn. IOO(3): 341-349 (1959). Wartenstein symposium no. 62, The Behav- Horwich, R., Manski, D. Maternal care and in- ior of the Great Apes, Wenner-Gren Foun- fant transfer in two species of colobus mon- dation for Anthropological Research, 1974. keys. Primates, 16: 49-73 (1975). Infanticide among Animals 37

Hrdy, S. Male-male competition and infanti- receptivity in the female chimpanzee (Pan cide among the langurs (Presbytis enfellus) of troglodytes). Folia Primat. 30: 80-88 (1978). Abu, Rajasthan. Folia Primat. 22: 19-58 Lockie, J.D. The breedings and feeding ofjack- (1974). daws and rooks with notes on carrion crows Hrdy, S. The care and exploitation of nonhu- and other Corvidae. Ibis 97: 341-366 (1955). man primate infants by conspecifics other Low, B.S. Environmental uncertainty and the than the mother. In Advances in the Study of parental strategies of marsupials and placen- Beha\fior, J. Rosenblatt, R. Hinde, C. Beer, tals. Amer. Nat. 112: 197-213 (1978). and E. Shaw (Eds.). New York: Academic Loy, J. Peri-menstrual behavior among rhesus Press, 1976, Vol. 6. monkeys. Folia Primat. 13: 286-297 (1970). Hrdy, S. The Langurs of Abu: Female and Makwana, S.C. Infanticide and social change in Male strategies of reproduction. Cambridge, two groups of the Hanuman langur, Presbytis MA: Harvard Univ. Press, 1977. entellus, at Jodhpur. Primates, in press. Hrdy, S., Hrdy, D., Bishop, J. Stolen copula- Mallory, F.F. Infanticide in Mus musculus: in- tions (16-mm film). fluence of genetic and environmental factors Ingram, C. The importance of juvenile canni- (in prep.). balism in the breeding biology of certain birds Mallory, F.F. Infanticide and pregnancy fail- of prey. Auk 76: 218-226 (1959). ure: reproductive strategies in female col- Isaac, G. The food-sharing behavior of proto- lared lemming (Dicrostonyx groenlandicus) hominids. Sci. Am. 238(4): 90- 106 (1978). (in prep.). Jay, P. The common langur of North India. In Mallory, F.F., Brooks, R.J. Infanticide and Primate Behabpior, 1. DeVore (Ed.). New other reproductive strategies in the collared York: Holt, Rinehart, and Winston, 1965. lemming, Dicrostonyx groenlandicus. Nature Kay, R.F., Hylander, W.L. The dental struc- 273: 144- 146 (1978). ture of mammalian folivores with special ref- Masters, W.H., Johnson, V.E. Human Sexual erence to primates and phalangeroidea (Mar- Response. Boston: Little, Brown, and Co., supialia). In The Ecology of Arboreal 1966. Foli\jores (G.G. Montgomery, Ed.). Wash- McKenna, J. Origins of multiple infant-caregiv- ington, D.C.: The Smithsonian Institution ing behavior among Colobine monkeys: func- Press, 1978. tion and opportunism in evolution. Amer. Kempe, R.S., Kempe, C.H. Child Abuse. Cam- Anthrop. (in press). bridge, MA: Harvard Univ. Press, 1978. Mech, L.D. The Woh*es of Isle Royale. Fauna Kruuk, H. The Spotted Hyena. Chicago: Univ. of the National Parks of the United States, of Chicago Press, 1972. Fauna. Series 7. Washington, D.C.: U.S. Kummer, H., Gotz, W., Angst, W. Triadic dif- Govt. Printing Office, 1966. ferentiation: an inhibitory process protecting Meyburg, B.-U. Sibling aggression and mortal- pairbonds in baboons. BehaiVour 49: 62 ity among nestling eagles. Ibis 116: 224-228 (1974). (1974). Labov, J.B. Pregnancy blocking in house mice Michael, R.P., Zumpe, D. Rhythmic changes in (Mus muscuhts) and other mammals: socio- the copulatory frequency of rhesus monkeys biological implications and adaptive strate- (Macaca mulatta) under laboratory condi- gies for females. Ph.D. thesis, Univ. of tions. ./. Endocrin. 41: 231-246 (1970). Rhode Island, Kingston, 1979. Mohnot, S.M. Some aspects of social change Lancaster, J. Play-mothering: the relations be- and infant-killing in the Hanuman langur, tween juvenile females and young infants Presbytis entellus (Primates: Cercopithecae) among free-ranging vervet monkeys (Cercop- in Western India. Mammalia 35: 175-198 ithecus aethiops). Folia Primat. 4: 257-264 (1971). (1971). Mohnot, S.M. Ecology and behavior of the Lawick, H. van Solo: The Stop of an African common Indian langur, Presbytis entellus Wild Dog Puppy and Her Pack. London: Dufresne. Ph.D. dissertation, Univ. of Jodh- Collins, 1973. pur, 1974. LeBoeuf, B.J. Male-male competition and re- Mohnot, S.M., Makwana, S.C., Gadgil, productive success in elephant seals. Amer. M. Dynamics of langur populations at Jodh- Zool. 14: 163- 176 (1974). pur. Satellite Symposium on Primate Popu- LeBoeuf, B.J., Briggs, K.T. The cost of living lations, VIIth Int. Congress of Primatology, in a seal harem. Mammalia, 41(2): 167-195 Bangalore, India, 1979. (1977). Morris, D. The Naked Ape. New York: Mc- Lemmon, W.B., Allen, M.L. Continual sexual Graw Hill, 1967. 38 S. B. Hrdy

Nadler, R.D. Sexual behavior of captive oran- Roonwal. M.L. Field study of geographical. gutans. Arch. Sex. Brlro~~. 6: 457-475 ( 1977). subspecific and clinal variations in tail car- Nishida. T. Predatory behavior among wild riage in the Hanuman langur. Pwsh~ris ctt- chimpanzees of the Mahale mountains. Pri- tel/tt.s (Primates). in South Asia. in press. mates 20(I): I-20 (1979). Rosenblum. L. Infant attachment in monkeys. Oates, J.F. The ecology and behaviour of the In Tire Ori,qitr.s c?/‘ Httttrtrtt Socitrl Rrltrtiotts black-and-white colobus monkey (Co/ohr/s (R. Schaffer. Ed.). New York: Academic gl,c’rc’:o Ruppell) in East Africa. Ph.D. the- Press, 1971. sis, Univ. of London, 1974. Rosenblum. L. Sex and age differences in re- Oates, J.F. The social life of a black-and-white sponse to infant squirrel monkeys. Brrtitt, colobus monkey. Colobl~s gttrrcyu. Z. TicJrp BcItoIhr, E:lw/ttfiott. 5: 30-40 ( 1972). .s~c/Io/. 45: l-60 ( 1978). Rowell, T. Female reproductive cycles and so- Oppenheimer, J.R. Preshyris r~rrellrls, the Han- cial behavior in primates. At/\,. itt I/I~ Sfttc!\ uman langur. In Pritttatr Cottsrr~~crtiotf. o/‘Beltctl~ior. 4: hY- IO5 ( I Y72). H.S.H. Rainier and G. Boume (Eds.). New Rudran. R. Adult male replacement in one-male York: Academic Press. IY77. troops of purple-faced langurs (Prc.rhJ;fi.s Oppenheimer, J.R. Differential use of space and .scttc.\- .setto.\-) and its effect on populatton time by Hanuman langurs (Prcshytis ctttel- structure. Fdirr Prittftrf. IY: l66- I92 (lY73). 1rl.r). Paper presented at the 2nd Annual Rudran. R. Infanticide in red howlers (Alortrrtrtr Meeting of the American Society of Prima- sc,tticrtltt.s) of Northern Venezuela Paper pre- tologists, Atlanta. Georgia. 1978. sented in Vllth Int. Cong. of Primatology. Popp, J. Male baboons and evolutionary prin- Bangalore, India. Jan. X- 12. lY7Y. ciples. Ph.D. thesis. Harvard Univ.. Cam- Saayman. G.S. Behaviour of the adult males in bridge. Mass., 1978. a troop of free-ranging chacma baboons Procter. D.L.C. The problem of chick loss in (Ptrpio r/r-.sit1tt.s 1. Fditr Pritmc. 15: 36- 57 the South polar skua C’trtlrtr~trc(cr /rmccor- (lY71). ttticki. Ibis 117: 452-459 (lY75). Sahlins. M. The social life of monkeys. apes Quiatt. D. Aunts and mothers and selection and and primitive man. Htttmttt Biol.. 3 I( I ): 54- adaptation. Paper presented at Inaugural 73 (IYSYJ. Meeting of American Society of Primatolo- Sankhala. K. Tiger. London: Collins, IY7X. gists. Seattle. 1977. Schaller. G. T/w Scwttgc,fi Liott. Chicago: Rahaman, H. The langurs of the Gir sanctuary Univ. of Chicago Press, lY72a. (Gujarat)-a preliminary survey. J. Bottthtr~ Schaller. G. The sociable kingdom. In T/w ,Mrrr- NUI. His,. SW. 70: 295-3 I4 (1973). ids oj’ Attittrnl Bclrctl~ior (P. Marler. Ed. ). Ripley, S. The ecology and social behavior of Washington. D.C.: National Geographic the Ceylon gray langur. Pwsbytis r11rc4l11.s Society, lY72b. tltrrsitrs. Ph.D. dissertation. Univ. of Cali- Schiiz. E. Das verschlingen eigener Junger fornia, Berkeley, 1965. (“Kronismus”) bei Vogeln und seine Bedeu- Ripley. S. Intertroop encounters among Ceylon tung. \/o,qchwrrc IY: 2. I- I.‘, ( IY57). gray langurs (Pwshytis c~ttrrlltt.sI. In Syciul Sherman, P. Natural selection among some Cottttnrtttic,rrriott uttzottg Prittttrtes (S. Alt- group-living organisms. Ph.D. dissertation. mann, Ed.). Chicago: Univ. of Chicago Univ. of Michigan, IY76. Press, 1967. Sherman. P. Intraspecific killing and nepotistic Ripley, S. Infanticide in gray langurs and man: defense. Paper presented at Symposium on reproductive competition population regula- Natural Selection and Social Behavior. Ann tion, genetic polymorphism insurance or so- Arbor. Michigan. October l6- IX, IY78. cial pathology’? Paper presented at Hudson Skutch. A.F. Ptrlrtrr Birds cod tltrir. YortttS. Symposium on Biosocial Mechanisms of Austin: Univ. of Texas, 1976. Population Regulation. S.U.N.Y. at Platts- Smith, R.R., Credland, P.F. Menstrual and burgh, April. 1978. copulatory cycles in the gelada baboon Tltrr- Ring, A., Scragg, R. A demographic and social opitltf~crts g~ltrtlrr. In ltzrc~rtf.%ool. Yiwrhook. study of fertility in rural New Guinea. J. (t/‘ J.S. Olney (Ed.). London: Zoological Soci- Biosociul Science, 5: 8Y- I21 ( IY73). ety. 1977. Roberts, C., Lowe, C. Where have all the con- Southwick. C.H.. Siddiqi. M.F., Farooqui, ceptions gone’? Lctttwf I: 498-499, 1975. M.Y., Pal, B.C. Xenophobia among free- Rohwer, S. Parent cannibalism of offspring and ranging rhesus groups. In Pritttutc Aggrrs- egg-raiding as a courtship strategy. Ant. NNI. siotl. Trrritorirrlity. rrttd Xctwplrohiu. Ralph 112: 429-440 (1978). Infanticide among Animals 39

Holloway (Ed.). New York: Academic Press, among wild barbary macaques. (Mucaca syl- 1974. IYJIUIS L.). In The Mucuyrres: Studies in Ecol- Stacey, P.B., Bock, C.E. Social plasticity in ogy, Beha~~ior und ELvolution (D. Lindburg, acorn woodpeckers. Science 202: 1298- 1300 Ed.). New York: Van Nostrand-Reinhold (1978). Co., in press. Starin, D. A preliminary investigation of home Thompson. N.S., Primate infanticide: a note and range use in the Gir Forest langur. Primutes request for information. Laboratonl Primate 19(3): 551-568 (1978). Newsletter 6(3): l8- 19 (1967). Stehn, R.A., Richmond, M.E. Male-induced Trivers, R.L. Parental investment and sexual pregnancy termination in the prairie vole. selection. In Sexual Selection und the De- Microtrrs ochrogaster. Science 187: I2 I I- scent ofMan 1871-1971 (B. Campbell, Ed.), 1231 (1975). Chicago: Aldine, 1972, pp. l36- 176. Steiner, A.L. Mortality resulting from intras- Trivers. R.L. The logic of female choice. Paper pecific fighting in some ground squirrel pop- presented at the Annual Meeting of the Ani- ulations. J. oJ‘ Mammalogy 53(3): 601-603 mal Behavior Society. Seattle, Washington, (1972). June, 1978. Struhsaker, T. Behavior of vervet monkeys Tutin, C. Sexual behaviour and mating patterns (Cercopithecus uefhiops). Univ. of California in a community of wild chimpanzees (Pan Public. in Zool. 82: l-74 (1967). troglodytes troglodytes). Ph.D. thesis, Uni- Struhsaker. T. Infanticide and social organiza- versity of Edinburgh, 1975. tion in the redtail monkey (Ccrcopithecus us- Udry, J.R., Morris. N.M. Distribution of coitus cunius schmidti) in the Kibale Forest, in the menstrual cycle. Nature 220: 593-596 Uganda. Z. Tierpsvcltol. 45: 75-84 (1977). (1968). Struhsaker, T. Infaniicide in the redtail monkey Vogel. C. Behavioral differences in Presbytis (Cercopithecirs ascunirrs schmidti). In Pro- rrzfelllls in two different habitats. Proceed- ceedings oJ‘ the 6th Congress oJ‘ Int. Primu- ings of‘ the 3rd lnt. Congr. of Prirnutology, tological Society. D. Chivers (Ed.). New Zurich, 1970. 3: 41-47 (1971). York: Academic Press, 1978. Vogel, C. Hanuman as a research object for Sugiyama. Y. Group composition, population anthropologists--field studies of social be- density and some sociological observations havior among the gray langurs of India. In of hanuman langurs (PresbFtis entellus). Pri- German Scholars on Indiu, Chowkhamba mates 5: 7-38 (1964). Sanskrit Services Office, Varanasi, India, Sugiyama, Y. Behavioral development and so- 1973, Vol. I. cial structure in two groups of hanuman lan- Vogel, C. Okologie, Lebensweise und Sozial- gurs (Presbytis entellus). Primutes 6:313-247 ~~erhulten der gruuen Languren in Iverschie- (1965a). denen Biotopen Indiens. Journal oJ’ Cornpar- Sugiyama, Y. On the social change of hanuman uti\.e Ethology. Suppl. l7-Advances in langurs (Presbytis enrr//~r.s) in their natural Ethology. Berlin: Paul Parey, 1976 conditions. Primutes 6: 381-417 (1965b). Warren, J.M. Discussion of social dynamics. In Sugiyama, Y. An artificial social change in a Sociul Commrrnicution umong Primates (S. hanuman langur troop (Presbytis eatellr~s). Altmann, Ed.). Chicago: Univ. of Chicago Primates 7: 41-72 (IY66). Press, 1967. Sugiyama, Y. Social organization of hanuman Washburn, S., Hamburg. D. Aggressive behav- langurs. In Sociul Com,nlinicatiori umong ior in Old World monkeys and apes. In Pri- Primates, S. Altmann (Ed.). Chicago: Univ. mutes P. Jay (Ed.). New York: Holt, Rine- of Chicago Press, 1967. hart. and Winston, 1968. Sugiyama, Y. Characteristics of the ecology of Wendland. V. von. Zum problem des vorzeiti- the Himalayan langurs. J. oj‘ Hurnrm E~wlrr- gen Sterbens von jugen Greifviigeln und tion. 5: 249-277 (1975). Eulen. Die Vo~geh~urte 19(3): l86- I91 (1958). Sugiyama. Y., Parthasarathy, M.D. Hanuman Wheatley, B. Adult male replacement in crab- langur populations of South India. Paper pre- eating macaques (Mucuca fascicaluris). sented at the Satellite Symposium on Primate Paper presented at the 47th Ann. Meeting of Populations, VIIth Int. Congr. of the Primat. American Assoc. of Physical Anthropolo- Society, Bangalore, India, 1979. gists, Toronto, 1978. Suzuki, A. Carnivority and cannibalism among Williams, G.C. Aduptution und Natural Selec- forest-living chimpanzees. J. Anthrop. Sot. tion. Princeton: Princeton Univ. Press, 1966. Nipponese, 79: 30-48 (1971). Wilson. E.O. Competitive and aggressive be- Taub. D. Female choice and mating strategies havior. In Mun and Beust. J.H. Eisenberg hh

,and D. Ripley (Eds.). Washington, D.C.: D.A. Hamburg and J. Goodall (Eds.). Lon- Smithsonian Institution Press, 1971. don: Staples/W.A. Benjamin, in press. Wilson, E.O. Sociobiology: The New Synrhe- Wrangham, R.W. On the evolution of ape social sis. Cambridge, MA: Belknap Press, 1975. systems. In Sociobiology and the Social Sci- Wolf, K., Fleagle, J. Adult male replacement in ences, 1. DeVore (Ed.). Chicago: Aldine, in a group of silvered leaf monkeys ‘(Presbytis press. crisrara) at Kuala Selangor, Malaysia. Pri- Yoshiba, K. Local and intertroop variability in mutes, 18(4): 949-955 (1977). ecology and social behavior of common In- Worthman, C. Psychoendocrine study of human dian langurs. In Primates, P.C. Jay (Ed.). behavior: some interactions of steroid hor- New York: Holt, Rinehart, and Winston, mones with affect and behavior in the !Kung 1967. San. Ph.D. thesis, Harvard Univ., 1978. Zuckerman, Sir S. The Social Life of Monkeys Wrangham, R.W. Sex differences in chimpan- and Apes. London: Routledge and Kegan zee dispersion. In Perspectives on Human Paul. 1932. Evolution: The Behaviour of Great Apes,