Systematics/Phytogeography / Taxonomie/ Section
Randle, Christopher P. , Pickett, Kurt M. .
Conflation of ignorance and knowledge in the inference of clade posteriors.
The objective Bayesian approach in phylogenetics relies on the construction of prior distributions that reflect ignorance. When topologies are considered equally probable a priori, clades cannot be. Several justifications have been offered for the use of uniform topological priors in Bayesian inference. 1) They do not inappropriately influence Bayesian inference because they are uniform. 2) Although clade priors are not uniform, their undesirable influence is negated by the likelihood function, even when data sets are small. 3) The influence of non-uniform clade priors is unavoidable and appropriate. The first two justifications have been addressed previously, and found to be questionable. We address the final justification, which is inconsistent with the first two justifications and objective Bayesian philosophy itself. If uniform priors arrived at by convention do not represent beliefs, but rather are mathematical place holders to allow the estimate of posterior distributions, extension of these uniform priors to non-uniform priors of composite hypotheses is not warranted. In other words, if priors on clades must be either informative or incoherent, use of any prior distribution on clades precludes an objective interpretation of the posterior. We propose and discuss several potential solutions: 1) Bayesian inference can be abandoned in favor of other methods of phylogenetic inference. 2) The topology with the greatest posterior probability, which is also the tree of greatest marginal likelihood, can be accepted as optimal, with clade support estimated using other means. 3) The objective Bayesian philosophy can be abandoned in favor of a subjective interpretation. 4) A modified Bayes factor can be used to assess differences between prior and posterior beliefs introduced by the data at hand.
Log in to add this item to your schedule
1 - Sam Houston State University, Department of Biological Sciences, 1900 Avenue I., Huntsville, TX, 77340
2 - University of Vermont, Department of Biology, Marsh Life Science Building, 109 Carrigan Drive, Burlington, VT, 05405, USA
Presentation Type: Oral Paper:Papers for Sections
Location: Room 1/Woodward
Date: Monday, July 28th, 2008
Time: 3:15 PM