Mitochondrial DNA sequences and multiple data sets: a phylogenetic study of phytophagous beetles (Chrysomelidae: Ophraella).

This paper presents the phylogenetic infrastructure for an integrated historical and experimental study of host use evolution in the chrysomelid leaf beetle genus Ophraella. We report the collection of sequence data from the 16S ribosomal RNA (446 bp) and the cytochrome oxidase subunit I (420 bp) mitochondrial genes from 12 species of Ophraella and two outgroups. Sequence analysis revealed a strong A + T nucleotide bias, high interspecific COI sequence divergences (up to 21.4%) that greatly exceeded those for 16S (up to 5.9%), high intraspecific COI divergences (up to 3.8%), a dearth of amino acid substitutions in COI, and differing substitution patterns in ribosomal stems and loops. Intraspecific variation in COI haplotypes generally supported the genealogical coherence of Ophraella lineages, while suggesting two cases of paraphyletic species. Separate phylogenetic analyses of 16S and COI data sets yielded largely congruent trees. A combined 16S + COI analysis yielded a single shortest tree under maximum parsimony that was identical to trees provided by successive approximations, neighbor-joining, and maximum-likelihood methods. This topology proved robust to various forms of weighting and most nodes were highly supported (by bootstrap analysis). Separate parsimony analyses of mtDNA and previously collected morphological and electromorphic data sets revealed congruent estimates of all cladistic relationships except those within one clade. Analysis of the pooled data sets in a combined approach additionally provided support for the basal placement of two species from this clade, although the topology for the remaining species was weakly supported and incongruent with the mtDNA tree. Each data set contained significantly structured phylogenetic signal with respect to this clade, and data sets exhibited limited conflict (character incongruence) with each other. The combined data set, however, was found to lack phylogenetic signal. These observations may imply that pooling heterogeneously evolving classes of data obscured the phylogenetic signal in each, a potential limitation of the combined approach.