Incongruence among different mitochondrial regions: a case study using complete mitogenomes.

Mitochondrial sequences have long been used to examine vertebrate phylogenetic relationships. The extensive use of mitochondrial data reflects the ease of obtaining mitochondrial sequences and its relatively rapid coalescence time. Mitochondrial genomes typically do not undergo recombination, so the entire mitogenome should have the same underlying gene tree. Thus, given appropriate analyses, conflict among estimates of phylogeny from different mitochondrial regions should not exist. However, estimates of phylogeny based upon different mitochondrial regions can exhibit incongruence. Conflict in phylogenetic signal among mitochondrial regions has been observed in galliform birds for the position of the Odontophoridae (New World quail). To explore this, we expanded sampling to 47 galliform mitogenomes, adding six new mitogenomes, which included representatives of two previously unsampled families. Analyses of complete mitogenomes recovered a well-supported topology that was congruent with expectations from multi-locus studies. However, when analyzing individual regions, we found conflicting positions for the Odontophoridae and several other relationships at multiple taxonomic levels. We tested multiple analytical strategies to reduce incongruence among regions, including partitioning by codon position, using mixture and codon-based models, RY coding, and excluding potentially misleading sites. No approach consistently reduced the conflict among mitochondrial regions at any taxonomic level. The biological attributes of both strongly misleading and non-misleading sites were essentially identical. Increasing taxa actually appeared to increase conflicting signal, even when taxa were selected to break up long branches. Collectively, our results indicate that analyzing mitochondrial data remains difficult, although analyzing complete mitogenomes resulted in a good estimate of the mitochondrial gene tree.