Compositional heterogeneity in true bug mitochondrial phylogenomics

Mitochondrial phylogenomics is often controversial, in particular for inferring deep relationships. The recent rapid increase of mitochondrial genome data provides opportunities for better phylogenetic estimates and assessment of potential biases resulting from heterogeneity in nucleotide composition and mutation rates. Here, we gathered 76 mitochondrial genome sequences for Heteroptera representing all seven infraorders, including 17 newly sequenced mitochondrial genomes. We found strong heterogeneity in base composition and contrasting evolutionary rates among heteropteran mitochondrial genomes, which affected analyses with various datasets and partitioning schemes under site-homogeneous models and produced false groupings of unrelated taxa exhibiting similar base composition and accelerated evolutionary rates. Bayesian analyses using a site-heterogeneous mixture CAT + GTR model showed high congruence of topologies with the currently accepted phylogeny of Heteroptera. The results confirm the monophyly of the six infraorders within Heteroptera, except for Cimicomorpha which was recovered as two paraphyletic clades. The monophyly of Terheteroptera (Cimicomorpha and Pentatomomorpha) and Panheteroptera (Nepomorpha, Leptopodomorpha and Terheteroptera) was recovered demonstrating a significant improvement over previous studies using mitochondrial genome data. Our study shows the power of the site-heterogeneous mixture models for resolving phylogenetic relationships with Heteroptera and provides one more case showing that model adequacy is critical for accurate tree reconstruction in mitochondrial phylogenomics.