SEARCH

SEARCH BY CITATION

Abstract Mitochondrial genomes provide a promising new tool for understanding deep-level insect phylogenetics, but have yet to be evaluated for their ability to resolve intraordinal relationships. We tested the utility of mitochondrial genome data for the resolution of relationships within Diptera, the insect order for which the most data are available. We sequenced an additional three genomes, from a syrphid, nemestrinid and tabanid, representing three additional dipteran clades, ‘aschiza’, non-heteroneuran muscomorpha and ‘basal brachyceran’, respectively. We assessed the influence of optimality criteria, gene inclusion/exclusion, data recoding and partitioning strategies on topology and nodal support within Diptera. Our consensus phylogeny of Diptera was largely consistent with previous phylogenetic hypotheses of the order, except that we did not recover a monophyletic Muscomorpha (Nesmestrinidae grouped with Tabanidae) or Acalyptratae (Drosophilidae grouped with Calliphoridae). The results were very robust to optimality criteria, as parsimony, likelihood and Bayesian approaches yielded very similar topologies, although nodal support varied. The addition of ribosomal and transfer RNA genes to the protein coding genes traditionally used in mitochondrial genome phylogenies improved the resolution and support, contrary to previous suggestions that these genes would evolve too quickly or prove too difficult to align to provide phylogenetic signal at deep nodes. Strategies to recode data, aimed at reducing homoplasy, resulted in a decrease in tree resolution and branch support. Bayesian analyses were highly sensitive to partitioning strategy: biologically realistic partitions into codon groups produced the best results. The implications of this study for dipteran systematics and the effective approaches to using mitochondrial genome data are discussed. Mitochondrial genomes resolve intraordinal relationships within Diptera accurately over very wide time ranges (1–200 million years ago) and genetic distances, suggesting that this may be an excellent data source for deep-level studies within other, less studied, insect orders.