Rare genomic characters do not support Coelomata: RGC_CAMs

Recently there has been a lively debate about a new class of rare genomic characters, RGC_CAMs, and their implications for deep bilaterian phylogeny. Most recently, nine bilaterian species were analyzed along with subsets of six outgroups (Rogozin et al. 2007b), and support for a coelomate clade reported. The authors suggested that our previously reported support for an ecdysozoan clade (Irimia et al. 2007) reflected (i) one outgroup, Nematostella vectensis, being too closely related to bilaterians and (ii) lack of rigorous statistical analysis. Here, we report further studies of these characters. First, we discuss general issues of outgroup choice. Second, we point out that an argument used by Rogozin et al. against backmutation is not statistically significant. Third, we point out that the statistical method of Rogozin et al. fails to incorporate backmutations, leading to systematic placement of the long-branch taxon as the outgroup. A simple modification of the method yields very different results: 51 of 63 outgroup combinations favor Ecdysozoa, inlcuding 51 of 52 with at least eight phylogenetically informative characters, and all 19 with statistically significant signal. These results indicate that the Coelomata signal is a long-branch artifact.