Unstable taxa in cladistic analysis: identification and the assessment of relevant characters.

A common problem in phylogenetic analysis is the presence of unstable taxa that are depicted in multiple positions in optimal topologies. These uncertainties are reflected in strict consensus trees with polytomies that hamper the interpretation of the phylogenetic results. We propose a protocol for detecting unstable branches (either terminal taxa or clades) and identifying particular characters related to their instability in cladistic analysis. This procedure is based on an iterative evaluation of the agreement of triplets among the optimal topologies (i.e. most-parsimonious trees, MPTs) and examination of character optimizations on these trees. Different types of characters underlying the unstable behaviour of taxa are detected: those with conflicting scorings that support alternative positions of problematic taxa and those with missing data in the unstable taxa that could reduce their instability if they are scored. The entire process is automated through a TNT script that provides a list of characters related to the instability of each unstable taxon. The outcome of this procedure can be used as a guide for further research efforts focused on the revision or addition of (morphological or molecular) phylogenetic data for elucidating the affinities of unstable taxa.  © The Willi Hennig Society 2009.