Abstract
Phylogenetic relationships among the basal orders of winged insects remain unclear, in particular the relationship of the Ephemeroptera (mayflies) and the Odonata (dragonflies and damselflies) with the Neoptera. Insect evolution is thought to have followed rapid divergence in the distant past and phylogenetic reconstruction may therefore be susceptible to problems of taxon sampling, choice of outgroup, marker selection, and tree reconstruction method. Here we newly sequenced three mitochondrial genomes representing the two most diverse families of the Ephemeroptera, one of which is a basal lineage of the order. We then used an additional 90 insect mitochondrial genomes to reconstruct their phylogeny using Bayesian and maximum likelihood approaches. Bayesian analysis supported a basal Odonata hypothesis, with Ephemeroptera as sister group to the remaining insects. This was only supported when using an optimized data matrix from which rogue taxa and terminals affected by long-branch attraction were removed. None of our analyses supported a basal Ephemeroptera hypothesis or Ephemeroptera + Odonata as monophyletic clade sister to other insects (i.e., the Palaeoptera hypothesis). Our newly sequenced mitochondrial genomes of Baetis rutilocylindratus, Cloeon dipterum, and Habrophlebiodes zijinensis had a complete set of protein coding genes and a conserved orientation except for two inverted tRNAs in H. zijinensis. Increased mayfly sampling, removal of problematic taxa, and a Bayesian phylogenetic framework were needed to infer phylogenetic relationships within the three ancient insect lineages of Odonata, Ephemeroptera, and Neoptera. Pruning of rogue taxa improved the number of supported nodes in all phylogenetic trees. Our results add to previous evidence for the Odonata hypothesis and indicate that the phylogenetic resolution of the basal insects can be resolved with more data and sampling effort.
