Molecular systematics of cytochrome oxidase I and 16S from Neochlamisus leaf beetles and the importance of sampling.

If a gene tree is to be judiciously used for inferring the histories of closely related taxa, (1) its topology must be sufficiently resolved and robust that noteworthy phylogenetic patterns can be confidently documented, and (2) sampling of species, populations, and pertinent biological variation must be sufficiently broad that otherwise misleading sources of genetic variation can be detected. These principles are illustrated by the complex gene tree of Neochlamisus leaf beetles that I reconstructed using 90,000 bp of cytochrome oxidase I (COI) and 16S mitochondrial DNA (mtDNA) sequences from over 100 specimens. Cytochrome oxidase I haplotypes varied up to 25.1% within Neochlamisus and up to 11.1% within the gibbosus species group, while exhibiting very low A + T bias for insect mtDNA (63%), low transition saturation, and conservative patterns of amino acid variation. 16S exhibited lower sequence divergences and greater A + T bias and transition saturation than COI, and substitutions were more constrained in stems than in loops. Comparisons with an earlier study of Ophraella leaf beetles highlighted conservative and labile elements of molecular evolution across genes and taxa. Cytochrome oxidase I parsimony and neighbor-joining analyses strongly supported a robust mtDNA genealogy that revealed the monophyly of Neochlamisus and of the gibbosus species group. Phylogeographic relationships suggested that the eastern U.S. gibbosus group derives from southwestern velutinus group ancestors. Haplotypes from individual velutinus group species clustered monophyletically, as expected. However, haplotypes from each of several gibbosus group taxa were polyphyletically distributed, appearing in divergent parts of the tree. 16S provided a less-resolved gibbosus group topology that was congruent with the COI tree and corroborated patterns of mitochondrial polyphyly. By subsampling haplotypes corresponding to particular species, populations, and ecological variants of gibbosus group taxa, I demonstrate that recovered topologies and genetic distances vary egregiously according to sampling regime. This study thus documents the potentially dire consequences of inadequate sampling when inferring the evolutionary history of closely related and mitochondrially polyphyletic taxa.