Table 1.
Bayesian selection of the relaxed-clock model
| Data | Model | log mL  S.E |
Pr |
|---|---|---|---|
| 1g, 20s | AR |
|
0.993 |
| IR |
|
0.007 | |
| STR |
|
0 | |
| 1g, 40s | AR |
 026  0.050 |
1 |
| IR |
 258  0.040 |
0 | |
| STR |
 601  0.047 |
0 | |
| 1g, 85s | AR |
 864  0.059 |
1 |
| IR |
 492  0.069 |
0 | |
| STR |
 348  0.043 |
0 | |
| 5g, 20s | AR |
 810  0.035 |
0.999 |
| IR |
 580  0.030 |
0.001 | |
| STR |
 840  0.022 |
0 | |
| 20g, 20s | AR |
 470  0.043 |
0.998 |
| IR |
 010  0.058 |
0.002 | |
| STR |
 540  0.038 |
0 | |
| 20g, 40s | AR |
 300  0.181 |
1 |
| IR |
 800  0.218 |
0 | |
| STR |
 500  0.130 |
0 |
Notes: Data list each treatment with the number of genes (g) and species (s) for each alignment. Models tested include AR, autocorrelated rates; IR, independent rates; STR, strict clock. Log mL + SE is the log-marginal likelihood for the model with standard error for the log-likelihood estimate. Pr is the posterior model probability (assuming equal prior probabilities for models), calculated as in dos Reis et al. (2018, Appendix 2).