Evolution of aposematic coloration in Xylocopa: An unsupervised machine learning approach

Large carpenter bees (genus Xylocopa), are a species-rich, easily recognizable genus resembling bumble bees in size and appearance. Carpenter bees are found on every continent except Antarctica, and display a diverse array of color patterns. In some lineages many species display aposematic coloration, often consisting of brightly colored bands or patches of pubescence on the dorsum of the bee. Some species also exhibit sexual dichromatism, in most cases the female is the brightly colored sex while the male is often drab and covered in yellow or brown pubescense. As part of this study, I am imaging the dorsum of a variety of aposematic carpenter bees species from museum collections and using an unsupervised machine learning approach to cluster species with similar aposematic patterns. Each image will also be overlaid onto a standardized bee template, both the original photographs and templates will be analyzed separately and compared. The evolution of color patterns will then be investigated by testing for the presence of phylogenetic signal based on a phylogeny of Xylocopa constructed as part of this project. In addition, color pattern clusters will be correlated with Xylocopa species distributions to test whether the ranges of members within a cluster overlap, and co-occurring species may have converged on similar color patterns to reinforce the warning signal to predators through Mullerian mimicry. This study's goal is to identify potential drivers underlying the evolution of color patterns across Xylocopa and lay the groundwork for future studies on additional factors influencing present patterns.