Fig. 2
From: Structure is more robust than other clustering methods in simulated mixed-ploidy populations
The degree of correct and biased clustering as a function of the simulated population differentiation (Fst), for different clustering methods and three types of genetic data. The results are based on 10 replicate simulations for each of eight different migration rates, with 100 loci, a mutation rate of u = 0.0001, and 400 sampled individuals (100 diploids and 100 tetraploids from each of the two populations). Green dots show βpop, which indicates correct clustering by population; red dots show βploidy, which indicates biased clustering by ploidy level. Note that for Admixture and FastStructure, bias cannot be calculated for co-dominant data with known dosage (hence no red dots shown)
