The killing and abuse of young by conspecifies is a widespread phenomenon. Parental and nonparental infanticide have been reported in almost 100 species of mammals, most of which are terrestrial (Hausfater and Hrdv, 1984; Parmigiani and vom Saal, 1994). Infant killing can be the direct outcome of a violent interaction or can result from the indirect neglect of a young or an accident. This article focuses on violent, nonparental forms of infanticide in aquatic mammals. Parental killing in this group is apparently restricted to the indirect effects of maternal neglect (see Le Boeuf and Cam-pagna, 1994) and will not be treated here. Infant abuse is a much more common behavioral occurrence than infanticide. It may imply active violence or passive neglect, and it does not necessarily involve the intended death of the victim. Death in the context of abuse is usually perceived as accidental, a byproduct that often follows a process of infection and starvation (Le Boeuf and Campagna, 1994). Infant or young refers to a lactatmg or recently weaned pup, calf, or cub.
Except for otariids and phocids, data on killing and abusing young are sparse for aquatic mammals. Infanticide is an event that may pass unobserved or unreported. Spotty research coverage, with some species being well known and others virtually unstudied, suggests that the relevance and diversification of abuse and killing of young may be more widespread than reported here. Explanations of the well-documented cases of abuse and infanticide in aquatic mammals rarelv support the adaptive hypotheses that would account for similar episodes in terrestrial species.
I. Abuse and Killing of Young by Males
Violent behavior toward young was described in four out of seven sea lion species (with the closely related Zalophus cali-fornianus, Z. japonicus and Z. wollebacki being the exceptions). Subadult and juvenile males of the South American sea lion, Otariaflavescens, abduct (seize), abuse, and kill pups during the breeding season (Campagna et al, 1988). The behavior was observed in coastal Patagonia (Campagna et al, 1988), Uruguay (Vaz Ferreira, 1965), Chile (H. Paves Hernandez and C. Espinoza, personal communication), Peru (Harcourt, 1993; P. Majluf and K. Soto, personal communication), and the Falkland Islands (C. Duck and D. Thompson, personal communication). At Punta Norte, Peninsula Valdes, Argentina, 163 successful abductions were recorded in four breeding seasons. More than 20% of the 400 pups born each season were abducted by males. In a typical abduction, a juvenile or subadult male approached the breeding area alone or as part of a group raid (Campagna et al, 1988), dashed toward a pup, and grabbed it. The pups were then abducted away from the breeding group and some were carried out to sea (11% of the abductions), whereas others were released and held close to the abductor. Pups that attempted to escape were shaken violently from side to side, tossed in the air, held crushed against rocks, or submerged. Males defended their abducted pup from other males. Some abductors mounted pups, fully covering them with their massive bodies. About 6% of the pups abducted and 1.3% of the pups born during a season died as a consequence of physical abuse. Dead pups showed tooth puncture wounds and extensive hematomas.
Australian (Neophuca cinerea) and Hooker’s (Phocaretos hookeri) sea lions abduct and abuse pups in a similar fashion described for O. flavescens, with the important difference that adult Hooker’s cannibalize the killed pups. Adult male Australian sea lions grab pups that may be alone or with the mother and bite, shake, and toss them several times (Higgins and Ted-man, 1990). Eight attacks recorded in two breeding seasons resulted in four dead pups (5% of the pups observed) and accounted for 19% of pup mortality in the looker)’ (Higgins and Tedman, 1990). Adult and subadult male Hooker’s sea lions grab pups by the neck, violently thrash them from side to side, and sometimes carry them out to sea and drown them (Wilkinson et al, 2000). Adult abductors were also observed eating pups. Opportunistic observations on Hooker’s sea lions report males abducting pups on two occasions and eating them on nine occasions (Wilkinson et al, 2000). After thrashing the victim repeatedly from side to side, males bit the flesh off the carcass and consumed it. This is the only otariid species for which cannibalism has been described. Immature males do not apparently kill pups, although thev may try to keep them under control and occasionally mount them.
Steller sea lions, Eumetopias jubatus, may trample or crush pups or push them over a cliff as an indirect consequence of territorial disputes. In some instances, however, pups are killed as a direct violent action by males (B. Porter, personal communication).
Episodes of violent behavior toward pups are rare or absent among fur seals. Juvenile male northern fur seals, Callorhinus ursinus, occasionally abduct conspecific pups in a context that suggests a form of mate substitution (R. Gentry, personal communication). Male Antarctic fur seals, Arctocephalus gazella, rarely respond to pups, even to the extent that they will fail to respond if they happen, apparently accidentally, to lie on top of a pup (I. Boyd, personal communication). Pups may be killed accidentally by males if they come between fighting individuals, a relatively common occurrence in otariids.
Among the other pinnipeds, infant abuse and killing were described in at least four phocids and the walrus. Male northern and southern elephant seals, Mirounga spp., of different age classes, kill suckling pups and weanlings (Le Boeuf and Campagna, 1994). Pups are trampled accidentally by bulls in the context of male-male competition and may then die of internal injuries. Weaned pups are abused by pubertal males that attempt to mate with them and, in the process, injure and kill individuals of both sexes (Rose et al., 1991). At the time of departure, 30-50% of northern elephant seal (M. angu-stirostris) weaned pups show signs of having been mounted by a male that range from neck bites, scraps, cuts, and puncture wounds to deep gashes exposing blubber and profuse bleeding. An adult southern elephant seal, M. leonina, male from the Patagonian colony of Peninsula Valdes killed and apparently ate pups (J. C. Lopez, personal communication). He grabbed weaned pups from the beach, dragged them out to sea, kept them underwater until struggling ceased, and then tore off chunks and consumed them. The cannibal returned to the same place at least during two consecutive breeding seasons and killed dozens of weanlings. Male gray seals, Hali-choerus grypus, occasionally shake, toss, bite, mount, and kill pups (D. Boness and P. Pomeroy, personal communication). There is also evidence of cannibalism in this species (Bedard et al, 1993; Kovacs, 1996). An adult male was involved in the killing and eating of pups during three breeding seasons. In a similar modality to the southern elephant seal cannibal, the gray seal male grabbed his victims by the hind flippers, dragged them into the water, and drowned them. He later tore off chunks of the pup’s body with a biting-shaking action and consumed the blubber, skin, and muscle. Hawaiian monk seals (Monachus schauinslandi) males mount pups, suffocate, or drown them (Hiruki et al 1993; M. Craig, personal communication). Some individuals persist in this behavior and may kill many pups. Finally, adult male, female, and immature walruses, Odobemis rosmams, can jab a pup with their tusks and cause lethal injuries.
In summary, adult, subadult, and juvenile males of several pinniped species injure, abuse, or kill suckling and recently weaned pups in the following contexts: (a) accidentally, often as an indirect outcome of trampling and crushing during dominance, female defense, and territorial disputes; (b) as a direct or indirect consequence of misdirected sexual assault, such as during abductions and abuse by pubertal males; (c) as a direct or indirect consequence of misdirected aggressive behavior with no clear sexual component, such as attack of pups by territorial males not associated to mounting, herding, or harassing; and (d) as an apparent source of food (cannibalism). The age class involved in the abuse and killing varies with the species. In Australian and Hooker’s sea lions, adults are the most aggressive toward pups, but subadult and juvenile males also sequester pups and engage in biting, mounting, and holding them underwater. In the South American sea lion, subadult and juvenile males do most of the abductions; adults are rarely involved in pup abuse. Among phocids, young males seem to be involved more often in abuse than adults; adults may cause pup death or injury as an epiphenomenon of male-male competition.
Reports of violence toward young in the rest of the aquatic mammals are rare. Male polar bears, Ursus rnaritimus, occasionally kill and eat cubs, a behavior that is apparently generalized throughout the Arctic (Taylor et al, 1985). Indirect evidence suggests infanticide in the bottlenose dolphin, Tursiops truncatus (Patterson et al, 1998). Stranded dolphin calves were found with internal injuries that included contusions around the head and thorax, bone fractures, and lacerated organs compatible with violent behavior. The interactions that may have caused the death of the calves were not observed. However, an adult dolphin was observed to interact violently with a dead conspecific calf, and dolphins were also seen to chase and hit harbor porpoises, Phocoena phocoena, hard enough to toss them into the air. Stranded harbor porpoises had evidence of trauma similar to that reported for the stranded dolphin calves. Additional indirect evidence of con-specific killing in T. truncatus is available for a population of the southeast Virginia coast (D. Dunn, personal communication). Nine bottlenose dolphins within their first year of life, thus still dependent on their mothers, stranded with multiple skeletal fractures, hematomas, organ lacerations, contusions, and hemorrhages, indicating multidirectional trauma. External signs of trauma were absent, an observation incompatible with predation, boat strike, and fisheries interactions, but similar to antemortem injuries reported for harbor porpoises and dolphins (Patterson et al, 1998).
II. Abuse and Killing of Young by Females
Adult pinniped females repel alien young in the context of aggressive protection of resources intended for their own pup. In the northern elephant seal, females aggressively reject alien pups that approach them (references in Le Boeuf and Campagna, 1994). They shake, throw, and viciously bite unrelated pups. Attacks may be violent enough to cause extensive wounds or fractures, with subsequent infection and death. Orphans attempt to nurse from any female, thus being particularly vulnerable to attack and injury.
An unusual behavior involving females and resulting in the death of unrelated pups was described for the South American sea lion rookery at Islas Ballestas (Peru; K. Soto, personal communication). During the 1997-1998 El Nino breeding season, virtually all pups born starved to death. The following year, only about one-quarter of the females gave birth. These mothers had to defend their newborn pups from the sustained attempts of neighboring females to abduct the latter. It often occurs that otariid females close to parturition attempt to bring alien pups near them. However, the particular breeding context of the post-El Nino year resulted in an unusually high incidence of a behavior that may be related to confusing alien pups with their own pup. Abductions occurred at a rate of one pup every 2.7 hr of observation. Females did not nurse the abducted pups, which were later abandoned. Those that failed to reunite with their mother died from starvation or were killed by young males. Almost 300 female abductions resulted in 11 pup deaths, and the incidence of pup mortality due to male abductions increased from the regular 1.5 to 8.0% of the pups bom.
III. Male Violent Behavior toward Mature Females and Interspecific Pups
From a behavioral standpoint, abuse and killing of conspecific young by male pinnipeds resemble male violent behaviors directed toward mature females of the same species and toward females and pups of other species. Attributes that allow males to physically overpower competitors would also promote aggressive sexual behaviors related to achieving access and maintaining control of breeding females. For example, adult and subadult O. flavescens males abduct females from established harems (Campagna et al, 1988). Abductions involve grabbing, tossing, herding, mounting, and biting. Some females are badly injured and killed in the process. Male harassment of conspecific females may be relatively common in phocids: it has been reported for both species of elephant seals: the Hawaiian monk seal and the gray seal (Mesnick and Le Boeuf, 1991; Hiruki et al, 1993; Boness et al, 1995).
Strong and large pinniped males with an indiscriminate sexual urge often injure and kill females of other species. Males killing interspecific females during mating attempts were reported in all sea lions (Miller et al., 1996). O. flavescens males kill A. australis females, and E. jubatus kill Z. californianus females and even males. Mating attempts with dead females of the same and of a different species occur in some otariids and phocids, such as the South American sea lion and the elephant seal. Abnormal escalation of aggressive sexual behaviors may lead to instances such as a Steller sea lion male killing at least 84 California sea lion females and 12 males over three seasons (see references in Miller et al, 1996) or a southern elephant seal male killing more than 100 A. pusillus breeding female over successive breeding seasons (Best et al, 1981).
Sea lion predation of pups of other otariid species typically involves grabbing a pup bv the neck, shaking it from side to side, tossing and recovering, dragging it to sea. submerging and drowning, biting off flesh, and consuming it. It has been described for at least three species. Steller sea lion prey on northern fur seal neonates (pups under 5 months of age; Gentry and Johnson. 1981). Adult South American sea lions prev on South American for seal Arctocephalus australis, pups (Harcourt, 1993). Hooker’s sea lions, a species for which cannibalistic behaviors have been described (Wilkinson et al, 2000). prey on New Zealand fur seals. A. forsteri, and on Antarctic and sub-antarctic fur seals. A. tropicalis.
In the South American sea lion, interspecific predation and conspecific abuse may be particular’ associated. Juvenile and subadult sea lion males abduct and kill A. australis pups but do not consume them. Interspecific pup abduction was observed in Peru (Harcourt, 1993) and in Uruguay (Vaz Ferreira and Bianco. 1987), where sea lion and fur seals live sympatricallv. Males grab a fur seal pup, take it to a neighboring beach and toss and shake the pup. However, instead of killing and eating the pup, as adult male behaving as predators would do, these younger males defend them from other sea lions, mount them, and behave as they do with pups of their own species. Fur seal pups may be killed as an indirect consequence of violent treatment, but are not consumed by their abductors.
In summary, pup killing in some species (e.g.. Steller sea lions) is more common in the context of interspecific predation, whereas in others (e.g.. South American and Australian sea lions) it occurs more often in a sexual or aggressive social context. In general terms, pup abuse follows a similar pattern as female abuse, with the most aggressive species toward pups being also violent toward conspecific and interspecific females.
IV. Adaptive Meaning of Infant Abuse and Killing
Several hypotheses have been proposed to account for infanticide in terrestrial mammals (Hausfater and Hrdy 1984): (1) exploitation or predation, young are killed for nutritional benefits: (2) resource competition, adults kill unrelated young to increase access to food or breeding space for themselves or their offspring; (3) sexual selection, males kill unrelated offspring to achieve reproductive access to females; (4) parental manipulation, a parent reduces litter size by eliminating all or part of a litter: and (5) social pathology, a maladaptive behavior. Adaptive explanations for the killing of young in aquatic mammals have been suggested for bottlenose dolphins (sexual selection; Connor et al. 1996) and Hooker’s sea lion (cannibalism: Wilkinson et al. 2000).
Cannibalism is exceptional among aquatic mammals, and social pathology would be involved in cases such as the cannibal adult male gray seal and the subadult male southern elephant seal described earlier. However, cannibalism in P. hookeri was suggested to fit the food resource hypothesis. Several males kill and eat pups in a fashion similar to that described for the same species preying on fin’ seals. Pups are easy targets for males and may supply calories in excess of the daily energy requirement of a male, as has been suggested to explain the predation of O. flavescens on A. australis (Harcourt, 1993). Cannibalism in polar bears appears to occur as carrion feeding and as attacks by males on cubs. There is also evidence of a polar bear male feeding on an adult female (I. Stirling, personal communication), but this is a rare observation of difficult interpretation.
Most instances of infant abuse and killing in pinnipeds are better understood as epiphenomena of indiscriminate sexual and aggressive behaviors (Le Boeuf and Campagna. 1994). Social context, sexual dimorphism, and sexually selected behaviors would set the context for the occurrence of injury and death of young. Pinniped colonies are often dense, parental investment is limited to females, males are large relative to pups and females, and male movements are frequent in the vicinity of pups. At times during the breeding season, pups may be the most abundant age class in a rookeiv, increasing the opportunity of social interaction. Reproductive females are aggressive toward conspecifics in general and alien pups in particular. Female aggressive behavior in this context would be explained by the cost of producing milk for individuals that are fasting while nursing. A large proportion of the breeding males do not have sexual access to females and males have an indiscriminate sexual behavior. Pups, particularly those close to being weaned. may be almost as large as young mature females and are often confused as females. South American sea lion and northern and southern elephant seal males would kill pups in the context of misplaced sexual behavior. Abducted O. flavescens pups, for example, are treated as female substitutes, perhaps a practice of herding or harem keeping (Campagna et al, 1988). Pups born in a harem are not likely to be the offspring of the dominant male, as they were sired the previous season. Behavioral mechanisms that can protect pups from direct and indirect violence (e.g., being crushed during male fights) would not then be under selective pressure. Infanticide in the Australian sea lion would be the consequence of misdirected aggression. It was suggested that territorial males may perceive pups as a threat. After killing a pup, males return to their usual position in the territories (Higgins and Tedman, 1990).
It is not yet clear to what extent the abuse and killing of conspecific pups may have on a common evolutionary substrate with violent behaviors directed to mature females of the same or other species and toward young of other species. Examples among otariids suggest that a circular gradation may exist from simple predation to infanticide to cannibalism. Steller sea lions kill pups of other species as predators but rarely or never abuse conspecific pups as abductors; South American sea lions prey on pups of other species when adults and abduct (but do not eat) pups of the same and other species when young; and Hooker’s sea lions abduct, abuse, kill, and eat conspecific and interspecific pups. It remains to be determined if this progression is actual or deceptive. It is possible, however, that in the behavioral similarities among these phenomena may underlay a key to understanding the evolutionary origin of abuse.
