Sex estimation from morphology in living animals and dinosaurs

Abstract

Sexual dimorphism is a prevalent feature of sexually reproducing organisms yet its presence in dinosaurs has recently been questioned. However, the inferred absence of sexual dimorphism may be a methodological artefact, rooted in the lack of systematic knowledge concerning how sexual dimorphism of living animals behaves statistically. To start building such knowledge, I re-analysed published data of 139 species of living animals that are sexually dimorphic. The previous method used for dinosaurs recognized only 5% of the living species correctly as dimorphic. This low rate is largely caused by the tilting of ordinated multivariate space due to interactions between size and shape dimorphisms, low signal/noise ratios and inclusion of outliers. The rate can be improved to 50% by modifying the method but not further, unless the information on the sex of individual specimens is used. Such information is unavailable in dinosaurs, so sexual dimorphism probably cannot be established for a large proportion of sexually dimorphic dinosaurs. At the same time, about 32% of the 139 are strongly sexually dimorphic, and can be re-sexed from shape with misclassification rates below 0.05. A reassessment of dinosaurian data suggests that sexual dimorphism likely existed at least in some species, such as Allosaurus fragilis.