Corbiculate Bees (Hymenoptera: Apidae): Exploring the Limits of Morphological Data to Solve a Hard Phylogenetic Problem

Abstract Corbiculate bees comprise a distinctive radiation of animals including many familiar species, such as honey bees and bumble bees. The group exhibits a broad variety of morphologies and behaviors, including solitary, social, and cleptoparasitic lifestyles. Since corbiculate bees play a critical role for the interpretation of eusocial behaviors, understanding their phylogeny is crucial to explain patterns and mechanisms of social evolution. Despite advances to unveil corbiculate relationships employing genomic data, the drivers of conflict between molecular and morphological hypotheses are still not fully understood. Morphological datasets favor a single origin for highly eusocial behaviors (i.e., Apini + Meliponini) whereas molecular datasets favor other scenarios (e.g., Bombini + Meliponini). Explanations for this incongruence have been suggested, including quality, quantity, and source of data or methodological issues. In this work we tackled this problem by generating the most extensive morphological dataset for the corbiculate bee species by exploring characters from all body regions, including external and internal adult skeletal anatomy. We produced a matrix with 289 characters for 53 taxa of Apidae, including 24 corbiculate bees. We explored different analyses and optimality criteria including extended implied weights parsimony and two partitioning schemes for Bayesian inferences. We contrasted hypotheses with Bayesian topological tests and conducted analyses to investigate if characters were prone to concerted convergence. Our results are congruent with the conclusions of previous studies based on morphology, recovering Apini sister to Meliponini and both of them together sister to Bombini. Finally, we provide our interpretations on the corbiculate controversy and provide a conciliatory scenario about this issue.