Reassessing the role of morphology in bryophyte phylogenetics: combined data improves phylogenetic inference despite character conflict

Abstract Morphological data has gained renewed attention and has been shown to be crucial in clarifying the phylogenetic relationship in a wide range of taxa. In the last decades, phylogenetic analyses of sequence-level data have radically modified the systematic schemes within bryophytes (early non-vascular land plants) and have revealed a widespread pattern of conflict with morphology-based classifications. Yet, a comprehensive evaluation of character conflict has not yet been performed in the context of combined matrices. In this study, we evaluate the impact of morphology on bryophyte phylogeny following a total-evidence approach across 10 published matrices. The analysed matrices span a wide range of bryophytes, taxonomic levels, gene sampling and number of morphological characters and taxa. Data conflict was addressed by measuring: (i) the topological congruence between individual partitions, (ii) changes in support values of the combined data relative to the molecular partition and (iii) clade stability. The association between these measures and the number of morphological characters per taxon (Nc/T ratio) and the proportion of non-fixed characters (i.e., inapplicable, polymorphic and missing data) was explored. In the individual partition analyses, the Nc/T ratio correlated positively with the topological congruence in six to seven datasets depending on the weighting scheme. The proportion of non-fixed cells had a minor influence on congruence between data partitions. The number of characters and proportion of non-fixed data varied significantly between morphological datasets that improved congruence between data types. This variation suggests that morphological datasets affect the results of combined analyses in different ways, depending on the taxa studied. Combined analyses revealed that, despite the low congruence values between partitions, integrating data types improves support values and stability. However, while non-fixed data had no negative effect on support values, stability was reduced as the proportion of non-fixed cells increased. Nc/T ratio was negatively associated with support values and it showed ambiguous responses in stability evaluations. Overall, the results indicate that adding morphology may contribute to the inference of phylogenetic relationships of bryophytes despite character conflict. Our findings suggest that merely comparing (a) morphology-based classifications with molecular phylogenies or (b) the outcome from individual data partitions can misestimate data conflict. These findings imply that analyses of combined data may provide conservative assessments of data conflict and, eventually, lead to an improved sampling of morphological characters in large-scale analyses of bryophytes.