Fig. 2.

The performance of species delimitation under the multispecies coalescent when the data are generated under the protracted speciation model, from simulations run for a fixed duration of time (5 units) under different species conversion rates. Speciation initiation rate was fixed at 0.5, while extinction rates were either 0.0 and 0.2 (the plot does not distinguish between these different extinction rates, because these had no meaningful effect on the main results or our argument). (A) Shown is the number of species per replicate inferred at a 0.95 probability vs. the number of true species on the input tree. (B) Shown is the number of species per replicate vs. the number of lineages (i.e., both true species as well as lineages representing incipient species or population structure). Generally, across all conversion rates, BPP tends to overestimate the number of species. However, what is striking is that BPP does not track species, as seen in A, but, rather, tracks structure of any sort, whether incipient species or true species, as seen in B.