Strategies for improving approximate Bayesian computation tests for synchronous diversification

Table 2 RMSE on Ψ and Ω varying the prior on Ψ and the sorting method

	18 taxon-pairs Ψ	18 taxon-pairs Ω	4 taxon-pairs Ψ	4 taxon-pairs Ω	3 taxon-pairs Ψ	3 taxon-pairs Ω
U_S	4.6079; 4.141	0.0298; 0.0295	0.9695; 0.974	0.1052; 0.0952	0.8000; 0.8069	0.0850; 0.087
U_U	5.1069; 4.892	0.0867; 0.0856	0.9592; 1.038	0.1242; 0.1139	0.8367; 0.7625	0.0949; 0.091
DPP_S	5.339; 5.317	0.0286; 0.0303	1.0149; 0.698	0.1158; 0.1073	0.8246; 0.7937	0.0720; 0.0845
DPP_U	5.8853; 5.058	0.0868; 0.0854	0.9; 0.92	0.1262; 0.1245	0.8124; 0.7454	0.0954; 0.092

Mean RMSE in estimation of Ψ and Ω averaged across 100 PODS for each of the three data configurations under each of the four models we tested. In the baseline text of each model “U” and “DPP” refer to the two priors on Ψ that were tested; the uniform distribution and the Dirichlet-process prior, respectively. In the subscript text “S” and “U” refer the sorting method applied; sorted and unsorted summary statistics vectors, respectively. The first value in each cell was calculated from reference tables composed of 3 × 10⁶ samples from the prior. The second value in each cell was calculated using 5 × 10⁷ samples from the prior (in bold)

ISSN: 2730-7182