Inferring the higher-order phylogeny of mosses (Bryophyta) and relatives using a large, multigene plastid data set.

PREMISE OF THE STUDY Investigating the early diversification of major clades requires well-corroborated and accurate phylogenetic inferences. We examined the performance of a large set of plastid genes for inferring the broad phylogenetic backbone of mosses-the second largest major clade of land plants-and their nearest relatives. METHODS We surveyed 14-17 plastid genes from a broadly representative taxonomic sampling of the major bryophyte lineages, including all major lines of non-peristomate mosses. We examined how well these new data corroborated or contradicted the findings of other studies, and investigated the effect of removing rapidly evolving characters. KEY RESULT: We inferred major clades with at least as strong support as other studies that used more taxa. We corroborated current views of overall embryophyte relationships, i.e., (liverworts, (mosses, (hornworts, tracheophytes))), with strong maximum likelihood (ML) bootstrap support, and also placed Zygnematales as the sister group of embryophytes with moderate ML bootstrap support. Within mosses, we confirmed Oedipodiaceae as the sister group of the large clade of peristomate taxa. Likelihood analysis also firmly placed Takakiaceae as the sister group of all other mosses, a strong conflict with parsimony results. Parsimony converged on the Takakia-sister result when rapidly evolving characters were removed, depending on the tree used to classify the site rates. CONCLUSIONS Our findings broadly support the utility of a 14-gene set from the plastome for future, more densely sampled phylogenetic studies of mosses and relatives, potentially complementing anticipated whole-plastome studies. Likelihood and parsimony conflicts flag possible instances of long-branch attraction, including one involving the earliest split in moss phylogeny.