Genebanks seek to understand the partitioning of genetic diversity among species, populations, and individuals in their collections since this informs decisions for adopting the most effective sampling strategy. Recent reports have suggested that diploid wild species have much less heterogeneity within populations than cultivated forms. We here review past empirical phenotypic trait variation data and examine previous and new DNA marker datasets. We also examine simulation datasets and calculations designed to mimic the effects of artificial biases against wild species heterozygosity due to ascertainment, ploidy, and allele frequencies. Trait data suggests large practical variation exists within populations. Similarly, DNA markers on multiple individuals within diploid wild potato species populations show that substantial heterogeneity in a species is partitioned within populations. Simulations illustrate that biases due to ascertainment, ploidy, and allele frequencies account for much of the apparent homogeneity of wild diploid potato species.