In previous studies we reported that while core populations of Sitka spruce [Picea sitchensis (Bong.) Carr] have little within-population genetic structure, peripheral populations are strongly spatially structured at distances up to 500 m. Here we explore the implications of this difference in structure on ex situ gene conservation collections and estimates of genetic diversity from research collections. We test the effects of varying the number of individuals sampled and the total area they are sampled across on capture of neutral genetic variation in collections from core, continuous versus peripheral, disjunct populations. Bivariate response surface analysis of genetic marker data for eight sequence tagged site loci from core and peripheral populations suggest that a population sample from 150 trees covering at least 225 ha would be adequate for capturing 95% of the genetic diversity (as measured by allelic richness or expected heterozygosity) in core populations. However, a larger sample of 180 individuals from an area of at least 324 ha is needed in peripheral populations to capture the same proportion of standing variation because of stronger within-population spatial genetic structure. Standard population sampling protocols for estimating among and within-population genetic diversity would significantly underestimate the within-population allelic richness and expected heterozygosity of peripheral but not core populations, potentially leading to poor representation of genetic variation in peripheral populations as well as erroneous conclusions about their genetic impoverishment.
