By how much do we underestimate species diversity of liverworts using morphological evidence? An example from Australasian Plagiochila (Plagiochilaceae: Jungermanniopsida).

As a framework for revisionary study of the leafy liverwort Plagiochila in Australia, two methods for species delimitation on molecular sequence data, General Mixed Yule Coalescence model (GMYC) and Automatic Barcode Gap Discovery (ABGD) were applied to a dataset including 265 individuals from Australia, New Zealand, and the Pacific. Groups returned by GMYC and ABGD were incongruent in some lineages, and ABGD tended to lump groups. This may reflect underlying heterogeneity in the history of diversification within different lineages of Plagiochila. GMYC from trees calculated using three different molecular clocks were compared, in some lineages different primary species hypotheses were returned by analyses of trees estimated under different clock models, suggesting clock model selection should be a routine component of phylogeny reconstruction for tree-based species delimitation methods, such as GMYC. Our results suggest that a minimum of 71 Plagiochilaceae species occur in Australasia, 16 more than currently accepted for the region, comprising 8 undetermined species and 8 synonyms requiring reinstatement. Despite modern taxonomic investigation over a four decade period, (1) real diversity is 29% higher than currently recognized; and (2) 12 of 33, or 36%, of currently accepted and previously untested Australasian species have circumscription issues, including polyphyly, paraphyly, internal phylogenetic structure, or combinations of two or more of these issues. These both reflect the many challenges associated with grouping decisions based solely on morphological data in morphologically simple yet polymorphic plant lineages. Our results highlight again the critical need for combined molecular-morphological datasets as a basis for resolving robust species hypotheses in species-rich bryophyte lineages.