Abstract: Plant mitochondrial protein-coding genes are slow evolving, and therefore are less subject to substitution saturation, and therefore perhaps more suitable for resolving deep relationships of high taxonomic categories. Plant mitochondrial genes hold, however, hundreds of RNA editing sites, involving mostly non-synonymous substitutions in the first and second codon positions, which has been reported to affect phylogenetic reconstructions. We have previously identified ca. 4700 mitochondrial RNA editing sites within a group of liverworts representing the ordinal diversity of liverworts, allowing us to critically evaluate the impacts of RNA editing sites on phylogenetic reconstructions in liverworts. Our phylogenetic inferences are mostly congruent on topology inferred from the original mitochondrial gene dataset, dataset with RNA editing sites corrected, and dataset with RNA editing sites excluded. The RNA editing site excluded dataset recovered a topology identical to that of the RNA editing site corrected dataset, supporting the sister relationship of Ptilidium and Jungermanniales, whereas the original dataset supported the sister relationship of Ptilidium and Jungermanniidae. The controversial placements of Ptilidium could be explained by site-wise log-likelihood analysis, as the majority of liverwort RNA editing sites supported the sister relationship of Ptilidium and Jungermanniidae, hence the correction or exclusion of the RNA editing sites changed the tree topology and supported the sister relationship of Ptilidium and Jungermanniales. Our study shows that RNA editing sites potentially impact phylogenetic analyses, suggesting that both genome and transcriptome derived data should be used with caution for phylogenetic reconstruction with genes hosting vast numbers of RNA editing sites such as plant organellar genes.

Key words: Jungermanniales, Jungermanniidae, phylogenetic inferences, Ptilidium, site likelihood