Mitogenomic sequences effectively recover relationships within brush-footed butterflies (Lepidoptera: Nymphalidae)

Mitogenomic phylogenies have revealed well-supported relationships for many eukaryote groups. In the order Lepidoptera, 113 species mitogenomes had been sequenced (May 14, 2014). However, these data are restricted to ten of the forty-three recognised superfamilies, while it has been challenging to recover large numbers of mitogenomes due to the time and cost required for primer design and sequencing. Nuclear rather than mitochondrial genes have been preferred to reconstruct deep-level lepidopteran phylogenies, without seriously evaluating the potential of entire mitogenomes. Next-generation sequencing methods remove these limitations by providing efficiently massive amounts of sequence data. In the present study, we simultaneously obtained a large number of nymphalid butterfly mitogenomes to evaluate the utility of mitogenomic phylogenies by comparing reconstructions to the now quite well established phylogeny of Nymphalidae.

We newly obtained 30 nymphalid mitogenomes via pyrosequencing on the Roche 454 GS Junior system, and combined these sequences with publicly accessible data to provide a 70-taxa dataset covering 37 genes for a 15,495 bp alignment. Polymorphic sites were not homogeneously distributed across the gene. Two gene regions, nad6 and 3' end of nad5, were most variable, whereas the cox1 and 5' ends of rrnL were most conserved. Phylogenetic relationships inferred by two likelihood methods were congruent and strongly supported (>0.95 posterior probability; ML bootstrap >85%), across the majority of nodes for multiple partitioning strategies and substitution models. Bayes factor results showed that the most highly partitioned dataset is the preferred strategy among different partitioning schemes. The most striking phylogenetic findings were that the subfamily Danainae not Libytheinae was sister of the remaining brush-footed butterflies and that, within Limenitidini, the genus Athyma was clearly polyphyletic. None of the single-gene phylogenies recovered the highly supported topologies generated on the basis of the whole mitogenomic data.

Thirty mitogenomes were assembled with 89% completeness from the contigs of pyrosequencing-derived reads. Entire mitogenomes or higher-quality sequences could be obtained by increasing pyrosequencing read coverage or by additional Sanger sequencing. Our mitogenomic phylogenies provide robust nodal support at a range of levels, demonstrating that mitogenomes are both accurate and efficient molecular markers for inferring butterfly phylogeny.