TABLE 1.
Summary of the various diversification models tested.
| Model and descriptions | Method used | Data used | Results |
| Null models: constant rates and pure birth | γ statistics in LASER (ML models testing; Rabosky, 2006) | Dataset 3 and 1,000 simulated trees based on Dataset 1 | Null models rejected, supporting the hypothesis of a slowdown in diversification |
| Diversity-dependent (DD) model: rates vary as a function of species density | DDD (ML models testing; Etienne et al., 2012) | 100 simulated trees based on Dataset 1 | DD models rejected, supporting the alternative diversity-independent model |
| Models in which rates vary among clades and time | BAMM (Bayesian models testing; Rabosky, 2014) | Dataset 1 | No significant rate shift detected; speciation decreased through time |
| Time-dependent (TD) model: rates vary discretely as a function of time | LASER and RPANDA (fit_bd) (ML models testing; Morlon et al., 2016) | Dataset 1 and Dataset 2 | Speciation rate is dramatically decreased and extinction rate is constant |
| Palaeoenvironment-dependent model: rates vary continuously as a function of both time and environmental condition | RPANDA (fit_env) (ML models testing; Morlon et al., 2016) | Dataset 2 | Speciation is positively correlated to the drop of atmospheric pCO2 from the late Miocene to present |
| Trait-dependent model: rates vary as a function of character states | DIVERSITREE (BiSSE) (ML models testing; FitzJohn, 2012) | Dataset 3 | Higher CAM associated speciation rate and 10-fold extinction rate to C3 |
| Trait-dependent model: rates vary as a function of hidden character states | HISSE (ML models testing; Beaulieu and O’Meara, 2016) | Dataset 1 | Photosynthetic pathways explain most of the diversification heterogeneity |