Rare genomic characters do not support Coelomata: intron loss/gain

Recently, a new phylogenetic method employing intron positionsharing across species was proposed and support for a Coelomateclade reported (Zheng et al. 2007. A rigorous analysis of thepattern of intron conservation supports the Coelomata cladeof animals. Mol Biol Evol. 24:2583–2592.). Here, we showthat the previous analysis depends on: 1) an idiosyncratic definitionof "conserved" introns, 2) exclusion of all phylogeneticallyinformative introns present in outgroups, 3) incorrect inferenceof change along the critical branch, and 4) lack of variationin rates across branches. The method thus seems unlikely togive accurate results. In addition, we address differences inrates of loss across intron sites, which Zheng et al. claimedinvalidates our previous analysis that supported Ecdysozoa (Royand Gilbert. 2005a. Resolution of a deep animal divergence bythe pattern of intron conservation. Proc Natl Acad Sci USA.102:4403–4408.). Instead, we show that our conclusionsare likely to be robust to such concerns.