Sexual dimorphism and sex-biased sampling influence analyses of trait evolution and diversity in a major songbird clade

Abstract Avian functional traits are widely studied in ecology and evolutionary biology, and these traits are often quantified from museum study skins. Across such studies, the sex ratio of specimens is highly variable or often unreported, despite knowledge that many species are sexually dimorphic in size and shape and that museum collections are sex-biased toward male specimens. Our study explicitly quantified whether sex-biased specimen sampling influences the answers to primary research questions about avian functional trait diversity and evolution. We focused on the songbird clade Emberizoidea, using an existing phylogenetic tree for this clade and a large male-specific dataset for 10 functional traits. After inter-measurer calibration, we collected a matching female-specific dataset for 408 species to assess the degree of size and shape dimorphism across the clade. We found that emberizoids are significantly dimorphic in both size and shape. We then quantified 3 metrics of functional trait diversity and found that estimated diversity differed significantly between male, female, and pooled datasets for all metrics. We found that all traits have strong significant phylogenetic signal that was of comparable magnitude despite differing significantly between sex-biased samples. We also modeled branch-specific rates of trait evolution and found these inferences to be more nuanced. Estimated rate shifts were largely congruent between sex-biased datasets, but branch-specific rates of size evolution were strongly impacted by both the sex ratio of the dataset and the choice of tool that quantified them. Thus, we found that sex-biased sampling can influence the answers to primary research questions in ecology and evolutionary biology. These issues may be avoided by sampling sex ratios consistently across taxa in these studies. Further, we recommend that these ratios should be clearly reported, and that researchers sample specimens in a way that makes sense given their questions and the biology of the organisms being studied.