Heterogeneity and Spatial Autocorrelation for Chloroplast Haplotypes in Three Old Growth Populations of Northern Red Oak

In eastern North America, evidence for cryptic northern refugia could contribute to resolving Reid’s Paradox, the disparity between the rate of oak recolonization indicated by pollen deposition and the rate indicated by contemporary seed dispersal studies. Severe anthropogenic disturbance of oak-dominated forests throughout eastern North America followed by regeneration from isolated patches and deliberate planting in some regions could obscure the signal of discontinuity expected from small, cryptic refugia. In this study of northern red oak, Quercus rubra L., the dominant representative of Quercus section Lobatae in the eastern United States, we address the question of appropriate sample size for accurate detection of the biogeographical distribution of chloroplast haplotype diversity in Q. rubra. We examined chloroplast DNA (cpDNA) variation in all Q. rubra over 17 cm in diameter (310 trees) in three forest fragments with documented histories of minimal disturbance for the last 100-190 years. cpDNA polymorphisms in three intergenic regions revealed different haplotype frequencies between the two local fragments located within 1 km of each other and complete discontinuity for the predominant haplotype between these two sites and a site 207 km distant. Haplotypes displayed positive spatial autocorrelation over 10-40 meter distances. Sample sizes of 10 or fewer taken at 50 meter intervals along a linear transect yielded poor estimates of haplotype frequencies and did not accurately detect haplotype richness.