To what extent do microsatellite markers reflect genome-wide genetic diversity in natural populations?

Microsatellite variability is widely used to infer levels of genetic diversity in natural populations. However, the ascertainment bias caused by typically selecting only the most polymorphic markers in the genome may lead to reduced sensitivity for judging genome-wide levels of genetic diversity. To test this potential limitation of microsatellite-based approaches, we assessed the degree of nucleotide diversity in noncoding regions of eight different carnivore populations, including inbred as well as outbred populations, by sequencing 10 introns (5.4-5.7 kb) in 20 individuals of each population (wolves, coyotes, wolverines and lynxes). Estimates of nucleotide diversity varied 30-fold (7.1 x 10(-5)-2.1 x 10(-3)), with densities of one single nucleotide polymorphism every 112-5446 bp. Microsatellite genotyping (10-27 markers) of the same animals revealed mean multilocus heterozygosities of 0.54-0.78, a 1.4-fold difference among populations. There was a positive yet not perfect (r(2) = 0.70) correlation between microsatellite marker heterozygosity and nucleotide diversity at the population level. For example, point estimates of nucleotide diversity varied in some cases with an order of magnitude despite very similar levels of microsatellite marker heterozygosity. Moreover, at the individual level, no significant correlation was found. Our results imply that variability at microsatellite marker sets typically used in population studies may not accurately reflect the underlying genomic diversity. This suggests that researchers should consider using resequencing-based approaches for assessing genetic diversity when accurate inference is critical, as in many conservation and management contexts.