LAUSR

Abstract Ant societies are primarily composed of females, whereby labor is divided into reproductive, or queen, and non‐reproductive, or worker, castes. Workers and reproductive queens can differ greatly in behavior, longevity, physiology, and morphology, but queen–worker differences are usually modest relative to the differences in males. Males are short‐lived, typically do not provide the colony with labor, often look like a different species, and only occur seasonally. It is these differences that have historically led to their neglect in social insect research, but also why they may facilitate novel phenotypic variation – by increasing the phenotypic variability that is available for selection. In this study, worker variation in multivariate size–shape space paralleled male–queen variation. As worker variation increased within species, so did sexual variation. Across species in two independent genera, using head width as a proxy for body size, sexual size dimorphism correlated with worker polymorphism regardless of whether the ancestral condition was large or small worker/sexual dimorphism. Mounting molecular data support the hypothesis that queen–worker caste determination has co‐opted many genes/pathways from sex determination. The molecular evidence, coupled with the observations from this study, leads to the hypothesis that sexual selection and selection on colony‐level traits are non‐independent, and that sexual dimorphism may even have facilitated the evolution of the distinct worker caste.