|
Résumé :
|
Molecular paternity tests show that extra-pair fertilizations are common in many socially monogamous bird species. However, the question of why females often seek extra-pair copulations is still controversial. Competing alternative explanation,, differ in their predictions on how extra-pair young should be distributed within and among broods. Analysing these distributions may therefore help to resolve this controversy. In several species broods without extra-pair young and those with total or nearly total extra-pair paternity have been claimed to be over-represented. Consequently, extra-pair nestlings would be distributed non-randomly among nests. To compute expected frequencies (i.e. the patterns of random distribution) the Poisson distribution has frequently been used. However, the Poisson distribution is not appropriate for two reasons. (1) Impossible configurations can receive positive probabilities. (2) The Poisson distribution approaches randomness only if extra-pair fertilizations are events of low probability, which often is not the case. We show how randomness can be computed appropriately by using the multivariate hypergeometric distribution. Re-analysing published data on distributions of extra-pair young, we shoes that the result that there are more broods containing either none or many extra-pair young than expected by chance is even more pronounced than preciously thought.
|