When good mitochondria go bad: Cyto-nuclear discordance in landfowl (Aves: Galliformes)

Mitochondrial sequences were among the first molecular data collected for phylogenetic studies and they are plentiful in DNA sequence archives. However, the future value of mitogenomic data in phylogenetics is uncertain, because its phylogenetic signal sometimes conflicts with that of the nuclear genome. A thorough understanding of the causes and prevalence of cyto-nuclear discordance would aid in reconciling different results owing to sequence data type, and provide a framework for interpreting megaphylogenies when taxa which lack substantial nuclear data are placed using mitochondrial data. Here, we examine the prevalence and possible causes of cyto-nuclear discordance in the landfowl (Aves: Galliformes), leveraging 47 new mitogenomes assembled from off-target reads recovered as part of a target-capture study. We evaluated two hypotheses, that cyto-nuclear discordance is "genuine" and a result of biological processes such as incomplete lineage sorting or introgression, and that cyto-nuclear discordance is an artifact of inaccurate mitochondrial tree estimation (the "inaccurate estimation" hypothesis). We identified seven well-supported topological differences between the mitogenomic tree and trees based on nuclear data. These well-supported topological differences were robust to model selection. An examination of sites suggests these differences were driven by small number of sites, particularly from third-codon positions, suggesting that they were not confounded by convergent directional selection. Hence, the hypothesis of genuine discordance was supported.