Trading up: the fitness consequences of divorce in monogamous birds.

Social and genetic mating systems play an important role in natural and sexual selection, as well as in the dynamics of populations. In socially monogamous species different genetic mating patterns appear when individuals mate outside the breeding pair within a breeding season (extra-pair mating) or when they change partners between two breeding seasons (widowing or divorce). Divorce can be defined as having occurred when two previously paired individuals are alive during the next breeding season and at least one of them has re-mated with a new partner. In socially monogamous birds divorce is widespread, but it is not clear whether it is a behavioural adaptation to improve the quality of a mating decision or whether, alternatively, it results as a non-selected consequence of other processes: existing studies suggest a heterogeneous set of results with respect to this central question. This heterogeneity could result from a number of factors, ranging from the methodological approaches used, to population- or species-specific characters. In this review we use phylogenetic meta-analyses to assess the evidence that divorce is adaptive (in terms of breeding success) across 64 species of socially monogamous birds. Second, we explore biological and methodological reasons for the heterogeneity in the results of previous studies. Results of our analyses supported the hypothesis that divorce is, in general, an adaptive behavioural strategy as: (1) divorce is triggered by relatively low breeding success; (2) there is a positive change in breeding success as a result of divorce. More specifically, while controlling for methodological moderators, we show that: (i) earlier stages of breeding are better predictors of divorce than later stages (r = 0.231; 95% CI: 0.061-0.391 for clutch size; similar for laying date); (ii) females benefited from divorce more than males in terms of increasing breeding success between successive breeding attempts, with different stages of the breeding cycle improving at different rates (e.g. r = 0.637; 95% CI: 0.328-0.817 for brood-level measures). We show that the effect size was dependent on the methodological approach used across studies and argue that research on the adaptive nature of divorce should be cautious when designing the study and interpreting the results. Altogether, by providing strong evidence that divorce is an adaptive strategy across monogamous birds, the results of our analysis provide a firm ground for further exploration of external covariates of divorce (e.g. demographic factors) and the mechanisms underlying the differences in the effect sizes of the proximal fitness causes and consequences of divorce.