The effects of sample size on population genetic diversity estimates in song sparrows Melospiza melodia

To empirically determine the effects of sample size on commonly used measures of average genetic diversity, we genotyped 200 song sparrows Melospiza melodia from two populations, one genetically depauperate (n=100) and the other genetically diverse (n=100), using eight microsatellite loci. These genotypes were used to randomly create 10,000 datasets of differing sizes (5 to 50) for each population to determine what the effects of sample size might be on several estimates of genetic diversity (number of alleles per locus, average observed heterozygosity, and unbiased average expected heterozygosity) in natural populations of conservation concern. We found that at small sample sizes of 5 to 10 individuals, estimates of unbiased heterozygosity outperformed those based on observed heterozygosity or allelic diversity for both low‐ and high‐diversity populations. We also found that when comparing across populations in which different numbers of individuals were sampled, rarefaction provided a useful way to compare estimates of allelic diversity. We recommend that standard errors should be reported for all diversity estimators, especially when sample sizes are small. We also recommend that at least 20 to 30 individuals be sampled in microsatellite studies that assess genetic diversity when working in a population that has an unknown level of diversity. However, research on critically endangered populations (where large sample sizes are impossible or extremely difficult to obtain) should include measures of genetic diversity even if sample sizes are less than ideal. These estimates can be useful in assessing the genetic diversity of the population.