Multivariate models of animal sex: breaking binaries leads to a better understanding of ecology and evolution.

'Sex' is often used to describe a suite of phenotypic and genotypic traits of an organism related to reproduction. However, these traits - gamete type, chromosomal inheritance, physiology, morphology, behavior, etc. - are not necessarily coupled, and the rhetorical collapse of variation into a single term elides much of the complexity inherent in sexual phenotypes. We argue that consideration of 'sex' as a constructed category operating at multiple biological levels opens up new avenues for inquiry in our study of biological variation. We apply this framework to three case studies that illustrate the diversity of sex variation, from decoupling sexual phenotypes to the evolutionary and ecological consequences of intrasexual polymorphisms. We argue that instead of assuming binary sex in these systems, some may be better categorized as multivariate and nonbinary. Finally, we conduct a meta-analysis of terms used to describe diversity in sexual phenotypes in the scientific literature to highlight how a multivariate model of sex can clarify, rather than cloud, studies of sexual diversity within and across species. We argue that such an expanded framework of 'sex' better equips us to understand evolutionary processes, and that as biologists it is incumbent upon us to push back against misunderstandings of the biology of sexual phenotypes that enact harm on marginalized communities.