Taxon sampling to address an ancient rapid radiation: a supermatrix phylogeny of early brachyceran flies (Diptera)

Early diverging brachyceran fly lineages underwent a rapid radiation approximately 180 Ma, coincident in part with the origin of flowering plants. This region of the fly tree includes 25 000 described extant species with diverse ecological roles such as blood‐feeding (haematophagy), parasitoidism, predation, pollination and wood‐feeding (xylophagy). Early diverging brachyceran lineages were once considered a monophyletic group of families called Orthorrhapha, based on the shared character of a longitudinal break in the pupal skin made during the emergence of the adult. Yet other morphological and molecular evidence generally supports a paraphyletic arrangement of ‘Orthorrhapha’, with strong support for one orthorrhaphan lineage – dance flies and relatives – as the closest relative to all higher flies (Cyclorrhapha), together called Eremoneura. In order to establish a comprehensive estimate of the relationships among orthorrhaphan lineages using a thorough sample of publicly available data, we compiled and analysed a dataset including 1217 taxa representing major lineages and 20 molecular markers. Our analyses suggest that ‘Orthorrhapha’ excluding Eremoneura is not monophyletic; instead, we recover two main lineages of early brachyceran flies: Homeodactyla and Heterodactyla. Homeodactyla includes Nemestrinoidea (uniting two parasitic families Acroceridae + Nemestrinidae) as the closest relatives to the large SXT clade, comprising Stratiomyomorpha, Xylophagidae and Tabanomorpha. Heterodactyla includes Bombyliidae with a monophyletic Asiloidea (exclusive of Bombyliidae) as the closest relatives to Eremoneura. Reducing missing data, modifying the distribution of genes across taxa, and, in particular, removing rogue taxa significantly improved tree resolution and statistical support. Although our analyses rely on dense taxonomic sampling and substantial gene coverage, our results pinpoint the limited resolving power of Sanger sequencing‐era molecular phylogenetic datasets with respect to ancient, hyperdiverse radiations.