Introduction

The aim of this chapter is to evaluate the evidence for the sexual selection hypothesis for infanticide by male primates, in relation to other hypotheses for this phenomenon. The sexual selection hypothesis emerges largely vindicated. The second part of the chapter then sets the stage for the rest of this book by summarizing observed behavior during episodes of infanticide, the evidence for the decision rules used by males, and the effects of social organization on rates of infanticide by males, and by pointing out some of the remaining difficulties with the hypothesis in its present form.

Critics have repeatedly pointed out that the evidence for infanticide in primates is often circumstantial at best, and that researchers may have jumped to conclusions on the basis of premature extrapolation from snippets of observations (Boggess 1979, 1984; Bartlett et al. 1993; Sussman et al. 1995). This makes it difficult to critically evaluate the hypothesis that a male committing infanticide will on average gain reproductively by doing so. Of course, it has made it equally difficult to evaluate other hypotheses. In this chapter, I therefore evaluate the sexual selection hypothesis for infanticide by males in non-human primates by a careful review of the most appropriate sources of evidence: directly observed cases in wild primates.

The conditions in which male infanticide are expected to raise the male's fitness have been reviewed extensively before (Hausfater & Hrdy 1984).

Introduction

The chapters in this volume point to four major conclusions. First, infanticide by males in many kinds of animal is most reasonably interpreted as a reproductive strategy, although not every observed case fits neatly all the criteria specified by this hypothesis (van Schaik, Chapter 2; Palombit et al., Chapter 6; Blumstein, Chapter 8; Veiga, Chapter 9). As further evidence accumulates, the patterns appear to reinforce the sexual selection hypothesis (e.g., Borries et al. 1999b) rather than non-adaptive interpretations (see Sommer, Chapter 1). Second, there are predictable correlates of infanticide risk among primates and other mammal species, including life history factors such as long infant dependency relative to gestation, large litter size, altriciality and social factors such as the loss of protectors, in particular the rate of breeding male replacement (van Schaik, Chapters 2 and 3; Borries & Koenig, Chapter 5; Blumstein, Chapter 8; van Noordwijk & van Schaik, Chapter 14; Nunn & van Schaik, Chapter 16). Third, females in species with infanticide by males are not just passive recipients of male aggression, but have developed a broad array of behavioral and physiological strategies to reduce infanticide risk. These traits are a major focus of this volume and will be developed more fully later.

A fourth conclusion to emerge is that females may also kill infants, in some species more frequently than do males (Blumstein, Chapter 8; Veiga, Chapter 9; Digby, Chapter 17; Voland & Stephan, Chapter 18).

Introduction

It is now widely accepted that infanticide by males has affected important features of primate social organization by selecting for various social counterstrategies of females (Hrdy 1979; Smuts & Smuts 1993; van Schaik 1996; Treves & Chapman 1996; see various chapters in this book). One of these possible female counterstrategies is female secondary transfer (Marsh 1979b; Sterck 1997; Sterck & Korstjens, Chapter 13). Female transfer decisions are expected to be strongly influenced by the identity of the male with whom a female associates. Secondary transfer is expected only where female relationships are relatively weak because it means the break-up of associations and coalitions, at least temporarily. Hence, given the predominance of female philopatry in primates, only a small number of species is expected to show this behavior. Thomas's langurs (Presbytis thomasi) are one of the few species known to show female secondary transfer related to infanticide avoidance. Females leave when infants have died or when they are old and independent enough to be left behind (Sterck 1997; Steenbeek 1999a). This makes the Thomas's langur an excellent choice for investigating the possible influence of infanticide on social relationships, since in species like this female choice in relation to male characteristics and strategies will be most manifest.

Species with regular female secondary transfer are of particular interest, because where the group contains only a single male, female emigration may terminate a group's lifespan.

Introduction

The sexual selection hypothesis for the selective advantages of infanticide by males requires that certain conditions be met. Provided the male is able to locate the infant, in order to derive reproductive benefits from infanticide he must be able to kill it with limited costs, the female must resume ovarian cycling earlier or produce more offspring than she would do otherwise, and he must gain mating access to the female when she resumes cycling (Hrdy 1979; van Schaik, Chapter 2). Whether or not these conditions are met depends on life style and life history. Life style variables are the location of the infants relative to the female, the presence or absence of hiding places for infants, and the degree of predictability of female spatial position in territories. Life history variables include the degree of infant precociality, and hence their ability to escape from attacking males, and the speed of female reproduction, i.e., their ability to be pregnant and lactating at the same time.

In this chapter, I examine whether infanticide by males is concentrated in species with the expected female life history. Infanticide by males is most advantageous where lactation is long relative to gestation. In such species, postpartum mating and early pregnancy are impossible because this would produce two sets of young of different ages, different needs and different competitive power for access to milk.

Introduction

The killing of dependent young individuals by conspecifics, what ecologists have called infanticide, has been viewed as one extreme and dramatic consequence of selection favoring those behaviors that promote the direct fitness of perpetrators (e.g., Hamilton 1964a,b). However, the evolutionary scenarios of this behavior may differ considerably depending on the identity of the perpetrators. It has been proposed that infanticide by non-kin is fundamentally different from infanticide by kin, which emphasizes the fact that the latter involves the sacrificing of shared genes for some presumed compensating benefits to the perpetrator's inclusive fitness (O'Connor 1978; Mock 1984). This infanticide by kin can be further subdivided into parental infanticide, i.e., the killing of young is committed by their own parents, and siblicide, the term used when the killing is carried out by full or half-siblings. The incidence of siblicide in taxa other than birds is, however, practically unknown (see Mock 1984). Four functional hypotheses have been proposed by Hrdy (1979) to explain the different types of pay-offs that may accrue to infanticidal individuals. In other words, infanticide should have evolved in the following contexts: (1) the exploitation of the infant, mainly as a food source; (2) resource competition, either with the infant or with its parents; (3) parental manipulation, wherein the parents interrupt their investment in the offspring to maximize their reproductive success; and (4) sexual selection, wherein infanticide increases the success of killers in intrasexual competition for mates.

Introduction

The topic of infanticide has been a staple theorem of sociobiology ever since this discipline – the study of social behavior from an evolutionary perspective – was born two and a half decades ago (Wilson 1975). The killing of conspecific young is still hotly debated. Does it occur at all, does it reflect an adaptation, a pathology or even a political agenda? Infanticide – observed among such varied taxa as birds, rodents, carnivores, pinnipeds and primates (Hausfater & Hrdy 1984; Parmigiani & vom Saal 1994) – therefore remains a litmus test upon which the validity of a sociobiological interpretation of behavior depends. I attempt to trace some intellectual roots of the controversy: those of defenders of adaptationist explanations, those of critics from within the paradigm of evolutionary biology, and those of critics who operate from other paradigms such as the social sciences. My ultimate aim is to defend the adaptationist interpretation as a valid and fruitful approach, while acknowledging that its narrative is anchored in a time-dependent framework of interpretation.

Cute and brute

People are fascinated by animals, not least because people are, in their own right, animals who can empathize with similar organisms. The average viewer of a natural history documentary will feel good if a monkey mother cuddles her newborn: “It's so cute.” But different emotions flare up if, over television dinner, wild chimpanzees eat an infant of their own kind: “It's so brute.”

Introduction

Sex-biased dispersal

Dispersal is a common feature of bird and mammal societies. Philopatry, however, is generally assumed to be the most advantageous situation, because it assures knowledge of the home range and offers better opportunities for cooperation with relatives. Several benefits of foregoing these advantages have been proposed (better mating opportunities, including inbreeding avoidance; reduction of within-group competition; coercion) (Greenwood 1980; Pusey & Packer 1987a; Moore 1993). The balance between the costs and benefits of dispersal obviously differs between the sexes, because dispersal is often highly sex biased. Most bird species show a female bias in dispersal, whereas most mammal species show a male bias (Greenwood 1980; Dobson & Jones 1985; Liberg & Schantz 1985). The best explanation proposed so far for such a strong sex bias is inbreeding avoidance.

In mammals it is generally assumed that the costs of dispersal are higher for females than for males, because loss of a range or allies affects female reproductive success more than male reproductive success, explaining the prevalence of male-biased dispersal (Waser et al., 1986; Pusey & Packer 1987a; Clutton-Brock 1989a). Female dispersal, however, is found in some mammal species and a considerable number of primate species (Moore 1984; Clutton-Brock 1989a; Strier 1994).

Costs and benefits of dispersal

Different costs and benefits of female philopatry and dispersal can be distinguished. Females in gregarious animals such as diurnal primates can remain in their natal range or in their natal group.

“Quite possibly, readers ten years from now may take for granted the occurrence of infanticide in various animal species,” Glenn Hausfater and I rashly conjectured back in 1984, in a preface to the first book on this subject, “and [they] may even be unaware of the controversies and occasionally heated debate that have marked the last decade of research on this topic…”. For biologists, that projection turned out to be more or less accurate. For those with backgrounds in the social sciences, perhaps especially in my own field of anthropology, it was wildly optimistic.

Most animal behaviorists now take for granted that the killing of infants by conspecifics can be found throughout the natural world and that, for many primate species, the arrival in their group of unrelated males represents a threat to infant survival. Many anthropologists, however, remain skeptical of the proposition that a propensity to attack infants born to unfamiliar females evolved in non-human primate males because it increased their chances to breed. This would require accepting that a behavior obviously detrimental to the survival of the group or even the species could evolve in males through Darwinian sexual selection because it provided the killers with a reproductive edge in their competition with rival males.

Introduction

Female competition, especially female reproductive competition, is likely to play an important role in shaping the social systems of all mammals (Altmann 1997; Gowaty 1997b; Hrdy 1981). While this competition can often be subtle, it also can lead to what is perhaps the most extreme form of reproductive competition: infanticide.

The chapters in this volume focus primarily, if not exclusively, on the phenomenon of infanticide by males and how it may be influencing the evolution of social systems. Generally, threats from conspecific males intensify under a specific context: the presence of an unrelated male, often following immigration into an established group (e.g., Hrdy 1974). But females and their young also face threats from conspecific females, including both unfamiliar intruders and fellow group mates. The threat of infanticide by females is likely to be taxonomically more widespread and, for group-living females, potentially a more constant threat than other forms of infanticide.

The purpose of this chapter is to examine the phenomenon of infanticide by females other than the mother (hereafter “infanticide by females”) in a variety of mammalian taxa. How similar is this behavior to infanticide by unrelated males? What makes infants vulnerable to attack from female conspecifics, and what are the contexts in which female attacks on infants occur?

Introduction

Infants have the strongest incentive to avoid infanticide. Paradoxically, they are the least capable of preventing it. Primate infants are particularly vulnerable because they take a long time to develop the physical and cognitive capacity to avoid risk. Yet infants are not utterly helpless in the face of infanticidal threats. Through detection, avoidance and deterrence, infants and their caregivers prevent infanticide. If infanticide is treated as a series of steps preceding lethal injury, one can see several junctures at which animals can act to lower risk (Figure 10.1). Protectors can prevent infanticide at every step. Neonate self-protection is generally limited to the earliest steps in the process. An older infant may have more options.

In this chapter, I explore the behaviors used by primates to prevent infanticide. I begin with protectors and then turn to infant self-protection. In both sections, I search for evidence of specific adaptations to infanticide prevention. The final section specifically addresses infant transport and the coevolution of primate mothers and infants.

Protectors

Primate infants are cared for by a variety of individuals: mothers, fathers, older siblings, more distant kin, as well as some unrelated adults (Hrdy 1976; Nicolson 1987; Whitten 1987; Manson 1999; Paul 1999). Within groups, the sexes usually take different roles in infant protection. A common pattern is for females to perform direct care (e.g., transport, feeding) while males provide indirect care (e.g., guarding).

Introduction

We simultaneously know more about the proximate causes of infanticide in rodents and less about its adaptive consequences and evolution than in any other taxon. Rodents are the largest mammalian order with over 440 genera and 2021 of the 4629 described species (Musser & Carleton 1993; Wilson & Reeder 1993). Infanticide by either or both males and females may be found in 46 species from 22 genera in 3 rodent groups (see below). A number of previous reviews illustrate the detailed understanding of proximate causation, development, and function gained by studying infanticide in rodents (Sherman 1981; Brooks 1984; Elwood & Ostermeyer 1984; Huck 1984; Labov 1984; Svare et al. 1984; vom Saal 1984; Labov et al. 1985; Trulio 1987; Elwood 1992; Elwood & Kennedy 1994; Ebensperger 1998a). In this review, I take a slightly different approach to study the evolution of infanticide in rodents. I first use the taxonomic distribution of infanticide by males and infanticide by females along with available phylogenetic evidence to parsimoniously reconstruct the evolution of infanticide by males, infanticide by females, and a potential female response to infanticide by males – male-induced pregnancy termination (Bruce effects; Bruce 1960). Then, using these evolutionary reconstructions, I test specific functional hypotheses about infanticide.

A thumbnail sketch of infanticide in rodents

Rodents provide some of the richest information about infanticide because studies of the proximate causation of infanticide are integrated with studies designed to document ultimate function.

Introduction

It has often been noted that female primates tend to have extended mating periods in their ovarian cycles, tend to mate polyandrously and also tend to mate during pregnancy (cf. Hrdy 1979; Hrdy & Whitten 1987). Since females in species vulnerable to infanticide show these features to a greater extent, this behavior was interpreted as serving to confuse paternity (cf. van Schaik et al. 1999; van Noordwijk & van Schaik, Chapter 14). The extent to which such mating tactics succeed in confusing paternity depends on the outcome of an “arms race” between males and females concerning the amount of information on paternity available to males. In order to examine more closely the claim that sexual behavior in primates serves at least in part to reduce infanticide risk, we must examine the physiological basis for paternity confusion, as well as for its complement, paternity concentration. Since ovarian cycles vary considerably in detail among taxa (e.g., Short 1984), we limit this examination to primates, the best-known order in this respect. We ask therefore how the ovarian cycles of female primates are organized in relation to the need for strategies to reduce infanticide risk. Two features are examined in particular which we will argue are related to the benefits to females of unpredictability in the timing of ovulation: (1) the large variance in the length of the preovulatory or follicular phase both within and between individuals, and (2) interspecific variation in the mean length of the follicular phase.

Introduction

Sexual selection theory suggests that divergent reproductive interests of the sexes impede the evolution of enduring social bonds between males and females. Males are more likely to increase reproductive success by acquiring multiple mates whereas females enhance fitness more by discriminative choice of individual mate(s) (Darwin 1871; Trivers 1972). Consequently, insofar as postcopulatory bonds limit a male's sexual access to additional fertile females and are irrelevant to antecedent female mate choice, such bonds should be relatively rare. The mostly polygynous mammals, for example, fulfill this expectation: male mating effort generally exceeds paternal effort, and affiliative interactions between the sexes center on the period of copulation or female fertility (Clutton-Brock 1989b).

And yet, males and non-fertile or anestrous females maintain stable relationships with one another in some species. The same theoretical framework that predicts the rarity of persistent heterosexual bonds also highlights a primary context for their evolution: when a male restricts his mating to a single female, a postcopulatory relationship with her is not only less costly to him, but may also offer fitness advantages to both parties. The proposed benefit to the female is the extensive parental care she receives from a male that is now certain of paternity. Thus durable male–female relationships were originally viewed as part of a coevolved suite of behaviors including monogamy, biparental care, and, in gregarious animals, “nuclear families” of parents and offspring (Morris 1967; Wittenberger & Tilson 1980; Gubernick 1994). The primarily monogamous, biparental birds have long served as vivid examples of this system (Lack 1968).