Comparative cladistics: identifying the sources for differing phylogenetic results between competing morphology-based datasets

Competing morphology-based phylogenetic analyses are routinely compared and evaluated only by a posteriori phylogenetic topology and group support, with little or no analysis of a priori data sources responsible for differing results. Although discordant characters and character-state scores are usually key to differing results (more so than variation in terminal taxa), programs currently do not exist to facilitate even simple two-way comparisons of morphology-based datasets, despite the impracticality and imprecision of manual assessment for datasets involving hundreds of characters. This paper, a first step to remedy this circumstance, presents methods (within TNT) to identify, compile and evaluate differences in characters and character states between datasets that yield different trees and degrees of group support. These apparently simple and urgently needed computer-assisted routines involve conceptual and computational challenges, even when competing morphology-based datasets are grossly similar. Example two-way comparisons are presented using pairs of similar morphology-based datasets for hominin and basal dinosaur radiations.