Rate variation & hybridization inference
Simulating how lineage-rate variation biases summary tests of introgression
Popular summary statistics for detecting introgression (e.g., D-statistics and related f-branch tests) assume a molecular clock. In this project, I simulated networks with varying degrees of substitution rate variation across lineages to test how robust these methods are when that assumption is violated.
I found that most commonly used summary tests have high type-1 error in the face of lineage-rate variation — meaning they can mistake rate heterogeneity for a signal of hybridization. This work is described in (Frankel & Ané, 2023).
