Table 2.
The 56 MODELTEST models and the grasses- or Amborella-sister topology that received the higher likelihood.
| Model | equal | +I | +G | +I +G |
| JC | grasses | grasses | grasses | grasses |
| F81 | grasses | Amborella | Amborella | grasses |
| K80 | grasses | Amborella | Amborella | Amborella |
| HKY | grasses | Amborella | Amborella | Amborella |
| TrNef | grasses | Amborella | Amborella | Amborella |
| TrN | grasses | Amborella | Amborella | Amborella |
| K81 | grasses | Amborella | Amborella | Amborella |
| K81uf | grasses | Amborella | Amborella | Amborella |
| TIMef | grasses | Amborella | Amborella | Amborella |
| TIM | grasses | Amborella | Amborella | Amborella |
| TVMef | grasses | Amborella | Amborella | Amborella |
| TVM | grasses | Amborella | Amborella | Amborella |
| SYM | grasses | Amborella | Amborella | Amborella |
| GTR | grasses | Amborella | Amborella | Amborella |
The four rate-heterogeneity conditions used in these MODELTEST analyses are: 1) "equal" = equal rates across sites; 2) "+I" = estimated percentage of invariant sites; 3) "+G" = four gamma-distributed rate categories; and 4) "+I+G" = combination of invariant sites and 4 gamma-rate categories.