In many plant species, spatial genetic autocorrelation (AKA spatial genetic structure) is created by restricted seed and pollen movement.  However, spatial genetic structure can be reduced by long-distance gene flow or by self-thinning processes.  Trees, due to their longevity and relatively high potential for LDD, tend to exhibit less genetic structure than do other plant species.  This has important implications for inbreeding and the potential for local adaptation.  This study focuses on Northern Red Oak (Quercus rubra), comparing two secondary forest stands in North Carolina, one located in the Piedmont (Duke Forest), the other in the Southern Appalachians (Coweeta LTER).   These sites differ in topography, climate, and seedling density (0.17 and 1.14/ m², respectively).   We hypothesized that spatial genetic structure would be more pronounced in seedlings than in adult trees and that seedlings in the Appalachian stand would exhibit more spatial genetic structure than those in the Piedmont stand, where densities are an order of magnitude lower.  Permanent seedling census plots were established at both sites.  Sampled seedlings and all adults were genotyped at six microsatellite loci, and allelic diversity was high at both sites (>15 alleles/locus for all stages).  Spatial genetic structure was assessed using the coefficient r developed by Smouse & Peakall (1999).

Results/Conclusions

At the Piedmont site, spatial genetic structure was not observed in either adults or seedlings.   This contrasts with the analysis by Aldrich et al (2005) of an old-growth red oak stand, which found significant genetic structure even in large adult trees.  The lack of genetic structure at our site may be due to several factors.  First, seed sources for this second-growth stand are scattered across the landscape, which would increase genetic diversity in the adult population while reducing genetic autocorrelation.   Second, low seedling survival is known to rapidly reduce genetic structure.   A preliminary parentage analysis indicates that many of these seedlings originated from outside the stand.  At the Appalachian site, genetic structure was significant at fine spatial scales (<10 m); this is consistent with clumping or co-dispersal of siblings.  When thinning is simulated by randomly removing individuals from the data set, spatial genetic structure is reduced.  Unlike the Piedmont site, large, dense patches of oak seedlings are observed at this site, often in close proximity to large adult trees.  We hypothesize that future parentage analyses will show these patches to be clumps of siblings or half-siblings derived from adjacent adult trees.