Bayesian tests of topology hypotheses with an example from diving beetles.

Area of Science:

• Phylogenetics
• Bayesian statistics
• Evolutionary biology

Background:

• Bayes factors are commonly used for model testing in phylogenetics.
• Standard methods for Bayes factor tests can be misleading due to model-specific priors.
• Assessing topological hypotheses requires careful consideration of prior tree spaces.

Purpose of the Study:

• To review Bayesian approaches for model testing, particularly for phylogenetic hypotheses.
• To identify and address the misleading nature of standard Bayes factor tests.
• To propose refined methods for evaluating evolutionary relationships using Bayesian statistics.

Main Methods:

• Review of Bayesian model testing principles.
• Development of methods to ensure appropriate tree space priors for Bayes factors.
• Application of refined Bayes factor tests to an empirical case study.
• Comparison of standard versus refined Bayes factor analyses.

Main Results:

• Standard Bayes factor tests for phylogenetic hypotheses can be misleading due to prior dependencies.
• Refined methods, including constrained searches or posterior filtering, improve Bayes factor accuracy.
• Posterior model odds can be more informative when appropriate tree sets are difficult to define.
• Empirical analysis strongly supports the monophyly of Suphrodytes within Hydroporus.

Conclusions:

• Bayesian phylogenetic hypothesis testing requires careful prior specification.
• Refined Bayes factor approaches provide more reliable assessments of evolutionary relationships.
• The genera Suphrodytes and Hydroporus should be synonymized based on strong phylogenetic evidence.