Data sets, partitions, and characters: Philosophies and procedures for analyzing multiple data sets

We compared four approaches for analyzing three data sets derived from staphylinoid beetles, a superfamily whose known species diversity is roughly comparable to that of vertebrates. One data set is derived from adult morphology and the two molecular data sets are from 12S ribosomal RNA and cytochrome b mitochondrial DNA. We found that taxonomic congruence following conditional data combination, herein called compatible evidence (CE), resolved more nodes compatible with an initial conservative hypothesis than did total evidence (TE), conditional data combination (CDC), or taxonomic congruence (TC). CE sets a base of nodes obtained by CDC analysis and then investigates what further agreement may arise in a universe where these nodes are accepted as given. We suggest that CE75-75 may be appropriate for future studies that aim to both generate a well-corroborated tree and investigate conflicts between data sets, partitions, and characters. CE75-75 is a 75% bootstrap consensus CDC tree followed by combinable-component consensus of a 75% bootstrap consensus of each homogeneous set of partitions having hierarchical structure.