Methods and theory of phylogenetic inference
Simmons, Mark P. , Zhang, Li-Bing , Webb, Colleen T. , Reeves, Aaron , Miller, Jeremy A. .
Parsimony vs. Bayesian likelihood for phylogenetic inference from heterogeneous datasets.
Using simulations, we compared the relative performance of parsimony and Bayesian likelihood by progressively incorporating more different sets of parameters and increasing the severity of the incorporated heterogeneity. Within the context of nucleotide characters, we examined differential rates of evolution among characters, differential character-state frequencies and character-state space, and differential relative branch lengths among characters. Overall, parsimony significantly outperformed Bayesian likelihood given differential relative branch lengths among characters and heterogeneous character-state frequencies and character-state space. In contrast, Bayesian likelihood outperformed parsimony given rate heterogeneity among sites. The higher the rate of evolution simulated, the better parsimony performed relative to Bayesian likelihood. Our results indicate that parsimony and Bayesian likelihood converge in their performance as more different sets of parameters are integrated into the simulations. Increasing rate heterogeneity among sites was found to be advantageous for phylogenetic inference using both parsimony and Bayesian likelihood. Consistent with earlier studies, parsimony-based jackknife analyses were found to be more conservative than Bayesian likelihood in that they supported significantly fewer incorrectly resolved clades.
1 - Colorado State University, Department of Biology, Fort Collins, Colorado, 80523-1878, U.S.A.
2 - Smithsonian Institution, Department of Entomology, Washington, DC, 20013, U.S.A.
Presentation Type: Symposium
Location: Alpine A (Snowbird Center)
Date: Tuesday, August 3rd, 2004
Time: 12:15 PM