Correlates of data availability.

We used Bayesian logistic regression to estimate the effect of several variables on the on the probability that phylogenetic datasets were either available from a public archive (left column) or could be successfully procured by direct solicitation (right column). Specifically, for all datasets we explored the effect of the data-sharing policy of the publishing journal (scored as none, weak, strong, or JDAP membership) and the impact of funding-agency policy (NSF). For solicited datasets, we also assessed the impact of solicitor status (undergraduate, graduate, or professor). We estimated effects of these variables on our ability to successfully procure either the tree or alignment files (top panels), or both the tree and alignment files (bottom panels) for a given study. The estimated effect size for a given variable reflects its contribution to the probability of successfully acquiring the data. For each variable, the marginal distribution of its estimated effect size is summarized as a boxplot, indicating the median effect (solid line), interquartile range (box), and interquartile range (whisker) of the corresponding posterior probability distribution. Journal-policy effects are relative to the effect of a weak policy, and solicitor-status effects are relative to that of graduate student. The predictor variables and interpretation of the corresponding parameters are described in Table 1.