Report

The evolution of infanticide by males in mammalian societies

See allHide authors and affiliations

Science  14 Nov 2014:
Vol. 346, Issue 6211, pp. 841-844
DOI: 10.1126/science.1257226

Abstract

Male mammals often kill conspecific offspring. The benefits of such infanticide to males, and its costs to females, probably vary across mammalian social and mating systems. We used comparative analyses to show that infanticide primarily evolves in social mammals in which reproduction is monopolized by a minority of males. It has not promoted social counterstrategies such as female gregariousness, pair living, or changes in group size and sex ratio, but is successfully prevented by female sexual promiscuity, a paternity dilution strategy. These findings indicate that infanticide is a consequence, rather than a cause, of contrasts in mammalian social systems affecting the intensity of sexual conflict.

Why some male animals kill infants

One of the most unpleasant aspects of social life in some animal species is killing of the young by adult males. Lukas and Huchard looked at mammalian groups with a variety of social systems—from mice to mongoose and from bats to bears. Infanticidal behavior in males appeared to be a result of sexual conflict in social species with nonseasonal breeding. Killing of a female's young by a new male will speed her return to a reproductive state and allow him to raise his own young. Evolutionarily, the only successful defense in females appears to be polyandry: Females that mate with multiple males make it hard for any one male to know that he is, or is not, the father of her offspring.

Science, this issue p. 841

Infanticide by males is widespread in mammals and may be the main cause of infant mortality in some populations (1). It has long been viewed as a sexually selected strategy that increases mating opportunities for killer males by shortening postpartum infertility in the victim’s mother (2, 3). This is supported by natural observations across taxa showing that males target unrelated infants and often impregnate the victim’s mother afterward (1, 3, 4). Attempts to explain the taxonomic distribution of infanticide have, however, been mainly limited to investigations of the life-history correlates of male infanticide (58), showing that it rarely occurs where it does not accelerate the mother's return to sexual activity, as in seasonal breeders who cannot resume cycling before the next breeding season (8).

Several scenarios have linked the distribution of infanticide across species to the evolution of social organization and mating systems in mammals, and they remain largely untested or disputed (9, 10). According to the sexual selection hypothesis, the distribution of infanticide is expected to be modulated by contrasts in social systems that affect the intensity of male intrasexual competition (3, 5, 8). Male infanticide should thus be prevalent in species in which a few males mate with multiple females and monopolize most reproductive opportunities. However, females may respond to infanticide by developing counterstrategies that may refine patterns of associations across species. Social counterstrategies may include the evolution of female sociality (6, 11, 12), of permanent male-female associations (1315), or of changes in the group sex ratio (6, 11, 12, 15), because females may form coalitions with other females or with resident males to defend their progeny against male invaders. Additionally, females may mate with multiple partners to confuse paternity and dissuade infanticide (6, 16, 17), which may thus be absent in species with pronounced sperm competition. The evolutionary arms race between the sexes driven by male infanticide has generated confusion and controversy regarding the role of infanticide in the evolution of mammalian societies, calling for more integrative studies.

We used information gathered on 260 mammal species, including 119 species with and 141 without infanticide (Fig. 1), to perform phylogenetic analyses (18) to identify how variation in social organization and mating systems may have favored or prevented the evolution of infanticide by males. Specifically, we tested whether infanticide has primarily evolved in species in which females breed throughout the year and some males have high reproductive monopoly, and whether it has selected for social and sexual counterstrategies, including transitions toward female gregariousness, permanent male-female associations, changes in group size or sex ratio, or increases in female sexual promiscuity. We included infanticide records from wild populations in which the killer was unambiguously identified as an adult male. Species in which infanticide had never been observed were included only if natural observations on females and juveniles were available over more than three reproductive seasons, to minimize the risk of misclassifying them as “non-infanticidal.” The phylogenetic relatedness between species was inferred from an updated mammalian supertree (19), and analyses were also replicated in primates exclusively, which confirmed results obtained across mammals (see the supplementary materials).

Fig. 1 Distribution of male infanticide across mammals.

Male infanticide has been reported for about half of all species in our sample (open circles) and seems to have evolved independently multiple times. It mostly occurs in social species (dark gray branches), less in solitary species (light gray branches), and least in monogamous species (black branches). Animal drawings are from phylopic.org [for full credit, see (18)]

The distribution of male infanticide is closely associated with the absence of seasonal or annual breeding (defined as an average interbirth interval of 360 to 370 days). Infanticide occurs in a majority of non-annual breeders (76% of 97 species), as compared to 28% of 134 annual breeders (lambda = 0.67, t = –4.0, P < 0.001). A long lactation (relative to gestation) and the absence of a postpartum oestrus, two life-history traits affecting the time to cycling resumption in mothers of killed infants, are also associated with male infanticide, although their association disappears when controlling for annual breeding [for both traits, P as estimated by Markov chain Monte Carlo methods for generalized linear mixed models (pMCMC) >0.08, whereas pMCMC for annual breeding <0.001]. Although previous studies have used the ratio of the duration of lactation to that of gestation to estimate infanticide risk (68), our results reveal that this association is weak: Lactation exceeds gestation in 25 of 45 mammal species with infanticide and in 38 of 89 species without infanticide. Other aspects of life history were not associated with male infanticide (longevity: n = 210 species, pMCMC = 0.08; litter size: n = 230 species, pMCMC = 0.07; relative offspring weight at birth: n = 191 species, pMCMC = 0.59), suggesting that the evolution of infanticide is not determined by a species’ pace of life. Overall, the possibility for females to breed throughout the year is the only life-history trait examined that significantly explains the distribution of infanticide, confirming previous work and suggesting that infanticidal males gain extra mating opportunities due to earlier cycle resumption of the victim’s mother in non-annual breeders. Such males are thereby redirecting maternal investment from unrelated offspring toward their own future offspring.

Comparing the occurrence of male infanticide across social systems, we find that males commit infanticide more frequently in stable mixed-sex groups (66% of 112 species) than in solitary species [40% of 78 solitary species, P as estimated by phylogenetic analysis of variance (pAN) = 0.01], in species with female-only groups (23% of 31 species, pAN = 0.006), and in pair-living species (18% of 39 species, pAN < 0.001). Phylogenetic reconstructions confirm that variation in social organization is a key determinant of the evolution of male infanticide, which is more likely in stable bisexual groups than in any other social system (table S1). In addition, social groups contain, on average, 1 male per 2.5 females in species with male infanticide (SD = 1.7, n = 56 species) but only 1 male for 1.3 females in species without infanticide (SD = 3.2, n = 19 species, Fig. 2A; lambda = 0.47, t = –3.3, P = 0.002; this is not an effect of differences in group sizes, because group size is similar in both sets of species: P = 0.28). Phylogenetic reconstructions suggest that a biased group sex ratio is a determinant, rather than a consequence, of the evolution of male infanticide: The strongest female biases are observed just before or at the transition toward male infanticide, whereas there are few or no subsequent changes once infanticide has evolved (see the supplementary materials).

Fig. 2 Infanticide is associated with polygynous mating systems.

Male infanticide occurs in species in which (A) social groups contain more females per male, (B) dominant males obtain a higher share of reproduction in a given season, but (C) maintain their dominant position for shorter periods (here measured as the average number of interbirth intervals).

In line with this, infanticide occurs more frequently where reproduction is monopolized by a few males (Fig. 2B): The median percentage of offspring sired by the alpha male in a group is 67% (SD = 19.9, n = 15 species) in species with infanticide, compared to 35% (SD = 22.2, n = 5) in species without (lambda = 0.75, t = 3.9, P = 0.001). Finally, males retain their dominant position for shorter durations in species with infanticide (Fig. 2C): On average, dominance tenure covers two interbirth intervals (if the females were to successfully wean each litter) (SD = 1.0, n = 34 species), compared to four in species without infanticide (SD = 1.2, n = 6) (lambda = 0.0, t = –3.4, P = 0.001). These correlations hold after controlling for the effect of annual breeding, which may also affect male ability to monopolize females (table S2). Overall, these findings indicate that a high male monopolization potential favors transitions toward infanticide by allowing males to maximize their reproductive output in systems where the costs of holding dominance shorten their breeding tenures (20).

Male infanticide has been proposed to alter social evolution by favoring female gregariousness or permanent male-female associations or by promoting a higher number of resident males that may help mothers to protect their infants from male intruders. We found no support for any of these scenarios. Infanticide risk is low in solitary species, the ancestral mammalian social organization (9), and as such is unlikely to motivate evolutionary transitions from a solitary lifestyle to other social organizations. Phylogenetic reconstructions confirm that male infanticide is significantly more likely to evolve after transitions toward group living than in their solitary ancestors. The presence of male infanticide does not increase the likelihood of a transition from solitary living to social monogamy, to female gregariousness, or to group living (table S1). In addition, among species living in stable bisexual groups, male infanticide is not linked to changes in the degree of female sociality (5.5 ± 5.1 females per group in 56 infanticidal species as compared to 8.9 ± 4.0 in 13 non-infanticidal species; pMCMC = 0.93). These findings indicate that infanticide by males did not trigger important transitions in mammalian social evolution. Population studies have nevertheless highlighted relationships between infanticide rates and group size (21), suggesting that the social consequences of infanticide may differ within populations versus across species, or that selection exerted by infanticide is insufficient at inducing major shifts in mammalian social organization.

Infanticide risk may also lead to the evolution of sexual counterstrategies by females, who might prevent infanticide by mating with many males to dilute paternity (6, 16, 17). To test this hypothesis, we investigated evolutionary relationships between infanticide and testis size, an indicator of the intensity of sperm competition (22, 23). In support of the paternity dilution hypothesis, phylogenetic reconstructions indicate that relative testis size is larger in lineages in which more time has elapsed since the evolution of infanticide. State-dependent diversification models (24) show that increases in testis size follow rather than precede transitions to infanticide, because they are not detected on branches immediately before transitions to infanticide, suggesting that infanticide exerts directional selection on relative testis size (see the supplementary materials). In addition, infanticide is subsequently lost in lineages where testes have grown large (Fig. 3; pMCMC = 0.02; see the supplementary results), indicating that infanticide may disappear where female attempts to prevent male sexual monopolization are successful.

Fig. 3 Loss of male infanticide occurs in species with large testes.

In some lineages in which males commit infanticide, testes appear to increase in size (relative to body mass) and subsequently, male infanticide is lost when testes have become relatively large.

Our findings support key predictions of the sexual selection hypothesis: Males primarily kill the offspring of rivals in stable bisexual groups where a few males monopolize reproduction over short periods. Increased testis size after evolutionary transitions to infanticide and secondary loss of infanticide in species with large testes suggest that female paternity dilution strategy efficiently reduces infanticide risk, and emphasizes the reversible nature of infanticide, which may appear and disappear according to the evolutionary arms race between the sexes. Although past studies failed to find comparative evidence for sexual counterstrategies to infanticide (7, 16), our sample spanning all mammals, combined with phylogenetically controlled analyses, and our focus on testis size to index female sexual promiscuity may explain divergent results.

Our analyses build on five decades of field observations to understand the distribution of species differences in the occurrence of infanticide by males across mammalian societies. The evolution of infanticide is largely determined by variation in the intensity of male-male contest competition and has not promoted major switches in mammalian social organization. However, it has affected the evolution of mating systems by increasing female sexual promiscuity as a paternity dilution strategy. These findings suggest that the distribution of infanticide across mammals is a consequence of contrasts in social and mating systems and closely reflects variation in the intensity of intra- and intersexual conflict.

Supplementary Materials

www.sciencemag.org/content/346/6211/841/suppl/DC1

Data and Methods

Supplementary Text

Tables S1 to S4

References (25262)

References and Notes

  1. Materials and methods are available on Science Online.
  2. Acknowledgments: We are grateful to T. Clutton-Brock, P. Kappeler, G. Cowlishaw, D. Blumstein, and an anonymous reviewer for helpful comments on an earlier draft of this manuscript, and to the Large Animal Research Group for insightful discussions. D.L. received funding from the European Research Council (grant no. 294494-THCB2011) and E.H. from the Natural Environment Research Council (grant no. NE/RG53472). Data supporting this study are available in the supplementary materials.
View Abstract

Navigate This Article