Monogamy with a purpose.

Humans are fascinated by animal monogamy. In the 1960s, Konrad Lorenz idealized the lifelong pair-bonds of geese until one of his students pointed out some infidelities and suggested that geese may be “only human” (1), and Desmond Morris (2) speculated about the advantages of the pair-bond for early humans. Even though many theories have since been proposed, human monogamous inclinations have also been questioned, including the pervasive idea that it all boils down to provisioning by fathers. Biologists prefer to place monogamy in a broad comparative perspective to determine what factors may have promoted its evolution. Why is monogamy ten-times more common in birds than mammals? Additionally, even though relatively common in primates, why are there no primates—other than humans—in which multiple reproductive pairs live together? Primate monogamy generally entails territoriality, with both the male and the female repelling outsiders of their own sex. Recently, two independent British teams have addressed these issues by analyzing variation in candidate traits that may have pushed species toward monogamy. Both studies used Bayesian and maximum-likelihood statistics to explore three traits: (i) paternal care, (ii) female sociality, and (iii) infanticide. The researchers used different databases, however. Kit Opie’s University College London team compared data on 230 primate species (3), whereas Dieter Lukas and Tim Clutton-Brock of Cambridge University covered no less than 2,545 mammalian species, including 330 primates (4). The teams further classified mating systems differently, with one team criticizing the classification of the other (4). The teams agreed on one crucial point, though, which is that paternal care is more likely a consequence of monogamy—an evolutionary afterthought with benefits—than the key to its existence. This conclusion left the other two factors, female sociality and infanticide, as possible drivers, and it is here that the teams parted ways. Opie at al. (3) were so convinced that infanticide was the primary cause that they put it in their title. That males kill youngsters sired by other males is known of many species, and widely viewed as a way of speeding up a female’s reproduction to the advantage of the infanticidal male (5, 6). The primate literature is rife with theories about how the need to curb infanticide shaped social organization, including a tendency of males to accompany and protect a female carrying their progeny (7). The low infanticide rate in monogamous species noted by both research teams appears to fit this idea, but does not necessarily imply a causal connection. Is monogamy effective at preventing infanticide or was there perhaps never much infanticide in these species to begin with? The Cambridge team (4) points out that most monogamous animals do not fit the typical mold of infanticidal species. It is only when the duration of lactation exceeds that of gestation that the killing of another male’s offspring benefits males. However, this is hardly a common characteristic of monogamous animals. The Cambridge team argues against infanticide as a cause, therefore, and concludes that monogamy likely began as a way for males to monopolize isolated females. Whenever food competition drove females apart, males ended up defending one female at a time because they could not claim several females at once. Once a male had settled on guarding a single female, the defense of her offspring and provisioning of food were logical extensions. The contrast in conclusions by both teams is disturbing given the tight statistical connection between monogamy and discrete female ranges. Both teams found this connection to exceed the one between monogamy and infanticide (table T1 in ref. 3 and table S2 in ref. 4). So, why did Opie et al. (3) place so much emphasis on infanticide? Were the authors swayed by prior theorizing (7) or by sound statistical arguments? Opie et al. estimated transition rates in three stochastic models, each including a mating system and one other factor; their results show that transitions from polygyny to pair-bonding can happen at similar rates whether female ranges overlap or not (figure 1B in ref. 3). Estimated transition rates further suggested that to move to pair-bonding required passing through a stage with high infanticide (figure 1C in ref. 3). However, data on infanticide is notoriously hard to obtain, and when the authors restricted their analysis to the subset of studies with the most reliable data, the connection with infanticide fell away (figure S2 in ref. 3). To us, the results are entirely consistent with an emphasis on female ranges, hence in line with the Frans B. M. de Waal.