The problem is that grouping individuals -> species is highly nontrivial. Even your constraint fails with bidirectional admixture. Imagine individuals from spec

The problem is that grouping individuals -> species is highly nontrivial. Even your constraint fails with bidirectional admixture. Imagine individuals from species A interbreed with people of species B. The admixture survives in B for a couple hundred thousand years and then back-admixture occurs, producing descendants in A who have intermediate ancestors of a different species. This is a common pattern in genus Homo, which is part of why human taxonomy is insanity. Bonus points if you can explain how to decide which species name to give to the immediate descendants of the admixture event from A->B. There's an older paper by Holliday on this subject that you might find interesting: https://doi.org/10.1086/377663 https://doi.org/10.1086/377663