Testing hypotheses about the sister group of the passeriformes using an independent 30-locus data set

Although many phylogenetic studies have focused on developing hypotheses about relationships, advances in data collection and computation have increased the feasibility of collecting large independent data sets to rigorously test controversial hypotheses or carefully assess artifacts that may be misleading. One such relationship in need of independent evaluation is the position of Passeriformes (perching birds) in avian phylogeny. This order comprises more than half of all extant birds, and it includes one of the most important avian model systems (the zebra finch). Recent large-scale studies using morphology, mitochondrial, and nuclear sequence data have generated very different hypotheses about the sister group of Passeriformes, and all conflict with an older hypothesis generated using DNA-DNA hybridization. We used novel data from 30 nuclear loci, primarily introns, for 28 taxa to evaluate five major a priori hypotheses regarding the phylogenetic position of Passeriformes. Although previous studies have suggested that nuclear introns are ideal for the resolution of ancient avian relationships, introns have also been criticized because of the potential for alignment ambiguities and the loss of signal due to saturation. To examine these issues, we generated multiple alignments using several alignment programs, varying alignment parameters, and using guide trees that reflected the different a priori hypotheses. Although different alignments and analyses yielded slightly different results, our analyses excluded all but one of the five a priori hypotheses. In many cases, the passerines were sister to the Psittaciformes (parrots), and taxa were members of a larger clade that includes Falconidae (falcons) and Cariamidae (seriemas). However, the position of Coliiformes (mousebirds) was highly unstable in our analyses of 30 loci, and this represented the primary source of incongruence among analyses. Mousebirds were united with passerines or parrots in some analyses, suggesting an additional hypothesis that needs to be considered in future studies. There was no clear evidence that base-compositional convergence, saturation, or long-branch attraction affected our conclusions. These results provide independent evidence excluding four major hypotheses about the position of passerines, allowing the extensive studies on this group to be placed in a more rigorous evolutionary framework.