Table 4.
Model specification: JC69 analysis
| Data set | Models | P-value | Conf. interval | ||
|---|---|---|---|---|---|
| ML score | Parameters | Lower | Upper | ||
| Buckley | GTR+G | GTR+G | 0.241 | 0.215 | 0.269 |
| Buckley | JC69 | GTR+G+I | <0.001** | <0.001 | 0.004 |
| Dixon | HKY+G | HKY+G | 0.012* | 0.004 | 0.026 |
| Dixon | JC69 | GTR+G+I | <0.005** | <0.005 | 0.007 |
| Dunn | GTR+G+I | GTR+G+I | <0.01** | <0.01 | 0.036 |
| Dunn | JC69 | GTR+G+I | <0.01** | <0.01 | 0.036 |
| Edwards | GTR+G+I | GTR+G+I | <0.01** | <0.01 | 0.036 |
| Edwards | JC69 | GTR+G+I | <0.01** | <0.01 | 0.036 |
| Liu | GTR+G+I | GTR+G+I | <0.01** | <0.01 | 0.036 |
| Liu | JC69 | GTR+G+I | <0.01** | <0.01 | 0.036 |
| Sullivan | GTR+G+I | GTR+G+I | 0.290 | 0.204 | 0.389 |
| Sullivan | JC69 | GTR+G+I | <0.01** | <0.01 | 0.036 |
| Wang | GTR+G+I | GTR+G+I | 0.026* | 0.014 | 0.044 |
| Wang | JC69 | GTR+G+I | <0.005** | <0.005 | 0.007 |
Notes: Model 1 represents the model used for likelihood inference (i.e., searching and scoring both the original and simulated data sets). Model 2 was used to estimate parameter values and simulate data sets. Separating the models for scoring and simulation has been suggested as a correction for a liberal bias present in the SOWH test. Using a model for scoring with fewer parameters free to vary, such as JC69, here resulted in a more liberal test. All hypotheses were rejected when JC69 was used as Model 1. * indicates p-values less than 0.05; ** indicates less than 0.01.