When different samples have the same richness, a second criteria of optimization is used.
Three criteria are proposed :
What's interesting with Nei and Shannon indices is that they are both based on classes frequency, which is immediatly get owing to AxijFreq matrix.
Moreover they permit not to favour the more frequent classes when drawing competing accessions.
Indeed, subcore collections are favourised when they contain fair-distributed classes, because these two indices are leveled-up with frequency similarity.