SRH

SRH, Stationary, Reversible, and Homogeneous models are commonly used in phylogenetic inference to model sequence evolution. However, the extent and impact of violations of these assumptions are poorly understood. This study introduces and applies maximal matched-pairs tests of homogeneity to assess the scale and impact of SRH model violations on a large dataset of 3,572 partitions from 35 published phylogenetic data sets.

Key findings:

1. Nearly one-quarter (23.5%) of the analyzed partitions reject the SRH assumptions.

2. For 25% of the data sets, tree topologies inferred from all partitions differ significantly from those inferred using only those that do not reject the SRH assumptions.

3. The proportion of differing tree topologies increases when comparing trees inferred from partitions that reject the SRH assumptions to those that do not.

The results suggest that SRH model violations in phylogenetics may be substantial and can significantly affect the accuracy of inferred tree topologies and edge lengths. The authors emphasize the importance of testing for model violations and potentially excluding partitions that violate the assumptions before tree reconstruction. They also highlight the need to develop models that do not require SRH assumptions to improve the accuracy of phylogenomic inference.