Evidence for bias in estimates of local genetic structure due to sampling scheme

Traditional population genetic analyses typically seek to characterize the genetic substructure caused by the nonrandom distribution of individuals. However, the genetic structuring of adult populations often does not remain constant over time, and may vary relative to season or life-history stages. Estimates of genetic structure may be biased if samples are collected at a single point in time, and will reflect the social organization of the species at the time the samples were collected. The complex population structures exhibited by many migratory species, where temporal shifts in social organization correspond to a large-scale shift in geographic distribution, serve as examples of the importance that time of sampling can have on estimates of genetic structure. However, it is often fine-scale genetic structure that is crucial for defining practical units for conservation and management and it is at this scale that distributional shifts of organisms relative to the timing of sampling may have a profound yet unrecognized impact on our ability to interpret genetic data. In this study, we used the wild turkey to investigate the effects of sampling regime on estimates of genetic structure at local scales. Using mitochondrial sequence data, nuclear microsatellite data and allozyme data, we found significant genetic structuring among localized winter flocks of wild turkeys. Conversely, we found no evidence for genetic structure among sampling locations during the spring, when wild turkeys exist in mixed assemblages of genetically differentiated winter flocks. If the lack of detectable genetic structure among individuals is due to an admixture of social units as in the case of wild turkeys during the spring, then the F _IS value rather than the F _ST value may be the more informative statistic in regard to the levels of genetic structure among population subunits.