Meeting Abstract
Use of transcriptomic and genomic datasets has become increasingly common as researchers attempt to test phylogenetic hypotheses across the diversity of the tree of life. The size and complexity of these large datasets present analytical challenges such as detecting systematic errors and conflict in the data. Three approaches have been applied to phylogenomic data including supermatrix approaches that analyze all data partitions in a single matrix, coalescent-based approaches that examine individual partitions and seek to combine resultant gene trees into a species trees, and Bayesian concordance analysis (BCA). We explored the performance of these approaches using a 15 taxon transcriptomic data set for siboglinid annelids. Currently, Siboglinidae contains four major clades. Placement of Osedax has been controversial with some studies recovering it sister to Vestimenifera+Sclerolinum while others recover it sister to Frenulata. In this study, every analysis conducted supported Osedax sister to Vestimentifer+Sclerolinum clade, not Frenulata. Moreover, we recovered topologies that were largely consistent across different methods and datasets, which had different numbers of genes and varying amounts of missing data. Spirobrachia was unexpectedly recovered sister to all other siboglinids in BI supermatrix analysis and coalescent analyses of the smallest, most complete dataset (98 genes), whereas it was placed within Frenulata in ML and previous studies. Given that Spirobrachia contains the most missing data of all taxa sampled herein (75.5%), variations of placement are most likely due to lack of characters in alignment to inform placement of Spirobrachia.
