Factors influencing phylogenetic inference: a case study using the mammalian carnivores.

Phylogenetic reconstruction has undergone numerous developments in tree selection criteria (e.g., phenetics, cladistics, maximum-likelihood), available data sources (morphology versus molecules, and subsets of the latter), and practical limits on study size. Together with study age, I examined the effects of these variables on inferences of phylogeny for the mammalian order Carnivora. The raw data comprised 274 source trees spread among 13 carnivore taxa (generally families), which I divided into categories for each variable and combined using the supertree technique matrix representation with parsimony analysis. Incongruence between the resultant tree topologies or the underlying data was assessed using four comparison measures, each with slightly different properties: the triplet measures "do not conflict" and "explicitly agree," the partition metric, and the incongruence length difference metric. Except for a few cases reflecting historical problem areas in carnivore systematics, no significant differences in incongruence levels were found among the different categories within each variable, between the variables themselves, or between the taxa. Thus, most estimates of carnivore phylogeny cannot be distinguished from one another (and may even point toward the same solution) regardless of the methodology or data source employed. This conclusion held regardless of the comparison measure used.