What do we really know about the signalling role of plumage colour in blue tits? A case study of impediments to progress in evolutionary biology.

ABSTRACT

Evolutionary biologists seek to explain the origin and maintenance of phenotypes, and a substantial portion of this research is accomplished by thorough study of individual species. For instance, many researchers study individual species to understand evolution of ornamental traits which appear to be products of sexual selection. I explored our understanding of sexual ornaments in a well-studied vertebrate species that may serve as a case study for research programs in evolutionary biology. I attempted to located all published papers examining plumage colour and variables related to sexual selection hypotheses in a common European songbird, the blue tit (Cyanistes caeruleus). Researchers have estimated over 1200 statistical relationships with plumage colour of blue tits in 52 studies. However, of the approximately 1000 main-effect relationships from the 48 studies that are candidates for inclusion in this meta-analysis, more than 400 were reported without details of strength and direction. Circumstantial evidence suggests that an unknown number of other estimated effects remain unpublished. Missing information is a substantial barrier to interpretation of these papers and to meta-analytic synthesis. Examination and analysis of funnel plots indicated that unpublished effects may be a biased sample of all effects, especially for comparisons of plumage colour to age and individual quality, and possibly also to measures of mate choice. Further, type I error was likely elevated by the large number of statistical comparisons evaluated, the frequent use of iterative model-building procedures, and a willingness to interpret a wide variety of results as support for a hypothesis. Type I errors were made more problematic because blue tit plumage researchers only rarely have attempted to replicate important findings in their own work or that of others. Replication is essential to drawing robust scientific conclusions, especially in probabilistic systems with moderate to weak effects or a likelihood of bias. Last, researchers studying blue tit plumage have often developed ad hoc explanations for deviations of results from their predictions. Revising hypotheses in light of data is appropriate, but these revised hypotheses were rarely tested with new data. The only highly robust conclusion supported by meta-analysis is that male blue tits have plumage that reflects more light in the ultraviolet and yellow wavelengths than the plumage of females. Various other effects, including condition-dependence of plumage colour expression and a tendency for females to adjust the sex ratio of their offspring in response to male colour, remain uncertain. These obstacles to progress in the blue tit plumage literature are likely common in evolutionary biology, and so I recommend changes to incentive structures which may improve progress towards scientific understanding in this discipline.