wait. i assumed that you meant that STRUCTURE lacked a mechanism for inferring the number of “clusters”. this would be understandable.
if you think that STRUCTURE actually clusters currently living subjects into distinct groups, you are just profoundly mistaken. it models living subjects as descendants of once-distinct populations which, over the course of centuries, intermixed according to standard inheritence models. it is too simple for some applications, but in principle there is nothing controversial about its assumptions. it’s just a model of mutation and chromosomal crossover under non-selective conditions. that literally means that the model assumes that the alleles involved are not connected to any expressed positive or negative trait.
