KATZ, E P; SCHWARZBACH, A E; VOLKERT, L G; Kent State University: The Prevalence of Heterodeigmatic (Model-Heterogeneous) Data Sets in Maximum Likelihood Phylogenetic Reasearch

In Maximum Likelihood phylogenetic studies, most analyses assume that one substitution model (e.g. HKY or GTR) is sufficient to describe the entire species-set. However, we expect species to evolve independatly from each other; potentially, any sub-clade could evolve according to its own model. A test for this new sort of model-heterogeneity was developed by Weiss and von Haeseler (2003). Here we introduce a new pair of terms (heterodeigmatic and homodeigmatic) to describe and clarify this clade-by-clade model-heterogeneity. We explore published sequence alignments from the animal and plant kingdoms to determine the prevalance of heterodeigmatic data sets, using the Weiss and von Haeseler statistic, and the AIC, BIC and LRT model-selection criteria. This phenomenon requires much future study.