Microsatellite analysis of spatial structure among seedlings in populations of Pinus strobus (pinaceae)

Abstract

In a detailed analysis of how limited seed dispersal can create spatial structuring of genetic variation, several nuclear microsatellites were assayed in seedlings from two forests of Pinus strobus, one old growth (OG) and the other (second site, SS) logged in ca. 1900. By using loci with a large number of alleles and new statistical methods on averaged spatial correlation coefficients, unusually precise estimates of spatial genetic structure were obtained, even though the structure was expected to be very weak. This high precision allowed the spatial patterns to be contrasted across loci and populations. At the OG site, the average spatial correlation coefficient for short distances (<15 m) exceeded its random expected value by 0.035, providing an indirect estimate of ca. 230 for Wright's neighborhood size. The value is similar to that estimated in a previous study of adult trees at OG and probably represents the natural level of spatial structure. A very similar value, 0.030, was obtained for seedlings at SS, despite the fact that unlike OG, genotypes of adults are randomly distributed, a likely result of logging. The results show that a single cycle of limited seed dispersal recreated the natural level of spatial structuring. In addition, one microsatellite, Rps50, had far greater amounts of allele variation, likely implicating it as having a higher mutation rate. The spatial structure of Rps50 also was significantly reduced, in a way that could be consistent with theoretical effects of high mutation rates (up to μ = 10(-2)). The choice of markers may influence estimates of spatial genetic structure. For example, if Rps50 is omitted the values are nearly doubled to 0.058 and 0.051 for SS and OG, respectively, both indicating a much smaller neighborhood size of ca. 100.