Résumé
Behavioural ecologists have for decades investigated the adaptive value of extra‐pair copulation (EPC) for females of socially monogamous species. Despite extensive effort testing for genetic benefits, there now seems to be a consensus that the so‐called ‘good genes’ effects are at most weak. In parallel the search for direct benefits has mostly focused on the period surrounding egg laying, thus neglecting potential correlates of EPC that might be expressed at later stages in the breeding cycle. Here we used Bayesian methods to analyse data collected over four years in a population of blue tits Cyanistes caeruleus, where no support was previously found for ‘good genes’ effects. We found that broods with mixed paternity experienced less brood failure at the nestling stage than broods with single paternity, and that females having experienced complete brood failure in their previous breeding attempt had higher rates of mixed paternity than either yearling or previously successful females. To better understand these observations we also explored relationships between extra‐pair mating, male and female phenotype, and local breeding density. We found that in almost all cases the sires of extra‐pair offspring were close neighbours, and that within those close neighbourhoods extra‐pair sires were older than other males not siring extra‐pair offspring. Also, females did not display consistent EPC status across years. Taken together our results suggest that multiple mating might be a flexible female behaviour influenced by previous breeding experience, and motivate further experimental tests of causal links between extra‐pair copulation and predation.