Sex and stress: a comment on Moore et al.

Why do secondary sexual characters play such an important a role in sexual selection when choosy females pay significant costs in terms of time and energy for their mate choice? Moore et al. (2016) raise the interesting question of whether there is a general link between physiological stress response and sexual selection. They use an extensive database on corticosterone levels and various measures of sexual selection in a meta-analytic framework to test for such effects. The authors found no evidence that sexual signals reliably reflect individual stress. On the other hand, by building on a statistically significant test based on studies of one species, they conclude that less-stressed individuals are preferred as mates. It is thus claimed that sexual preferences depend on stress of potential mates, but sexual signals do not mediate such preferences. Some methodological aspects of the study by Moore et al. (2016) deserve close consideration and raise questions about the conclusions. The methods used in phylogenetic meta-analyses are almost 20 years old (Møller and Thornhill 1998; Verdú and Traveset 2005). It is imperative that meta-analyses are based on a random selection of studies and that sources of random or systematic error are accounted for. Thus, inclusion of unpublished studies is a potential cause of bias. A standard issue in meta-analysis that forms the basis for any biologically meaningful analysis is empirical tests for methodological rigor, either because of recent progress in research methods or because of differences in approach. The study by Moore et al. (2016) suffers from at least 3 methodological issues: