Fig. 1.

Assessing prior sensitivity in analyses of an example dataset. PrioriTree allows users to assess the prior sensitivity of discrete-geographic analyses performed using BEAST. Here, we explore the prior sensitivity of the average dispersal rate parameter, μ. Panels summarize estimates of μ under two priors; a CTMC-rate reference prior (used as the default in BEAST, left) and a hierarchical exponential prior (right). Within each panel, each boxplot summarizes the marginal distribution for μ inferred for different numbers of data clones (x-axis); the prior is inferred without data (green), the posterior is inferred from a single copy (purple) and the data-cloning posteriors are inferred from datasets with 5 or 20 copies (gray). Each pair of boxplots represents replicate analyses (to assess MCMC performance)