Data curation and modeling of compositional heterogeneity in insect phylogenomics: a case study of the phylogeny of Dytiscoidea (Coleoptera: Adephaga)

Diving beetles and their allies are a virtually ubiquitous group of freshwater predators. Knowledge of the phylogeny of the adephagan superfamily Dytiscoidea has significantly improved since the advent of molecular phylogenetics. However, despite recent comprehensive phylogenomic studies, some phylogenetic relationships among the constituent families remain elusive. In particular, the position of the family Hygrobiidae remains uncertain. We address these issues by re-analyzing recently published phylogenomic datasets for Dytiscoidea, using approaches to reduce compositional heterogeneity and adopting site-heterogeneous mixture models. We obtained a consistent, well-resolved, and strongly supported tree, robust to analyses of various sizes of datasets. Consistent with previous studies, the monophyly of the geographically disjunct Aspidytidae is strongly supported. Our analyses support that Aspidytidae are the sister group of Amphizoidae, and more importantly, Hygrobiidae are sister to the diverse Dytiscidae, as convincingly demonstrated by morphology-based phylogenies. Our new results are congruent with recent morphology-based phylogenies. The phylogeny of Dytiscoidea can be resolved by reducing the effect of among-site compositional heterogeneity and adopting a better-fitting model accommodating site-specific amino acid preferences. Our analyses provide a backbone phylogeny of Dytiscoidea, which lays the foundation for better understanding the evolution of morphological characters, life habits, and feeding behaviors of dytiscoid beetles.