Oak woodlands have a nearly continuous distribution through California’s Coast Ranges and Sierra Nevada foothills. This, together with their wind-borne pollen, has led researchers to expect a pattern of genetic isolation by distance in California oaks. However, isolation by environment could also play a role in their genetic differentiation, mediated for example by effects on flowering phenology or seedling survival. In addition, unrestricted cross-fertilization among closely related but ecologically differentiated oak species potentially increases the movement of genes across even broader environmental gradients.
To test the relative effects of geographic versus environmental distance on genetic structure in oaks, I used ddRAD sequencing to genotype individuals from 30 sites, which span the range of California blue oak (Quercus douglasii) and also beyond its climatic niche into sites occupied by Oregon white oak (Q. garryana) and Tucker’s oak (Q. john-tuckeri). Pairwise genetic differentiation (as Nei’s Gst) was estimated using 2000 high-quality SNP loci. These data were fitted to a non-linear matrix regression (generalized dissimilarity modeling, GDM) with geographic distance, climate distance from downscaled models (ClimateNA), and population-level taxonomic distance as predictors. My approach takes advantage of flexible models developed in community ecology, but applies them to the analysis of population genetic data from high-throughput sequencing.
Results/Conclusions
Pairwise Gst indicates moderate differentiation between sites of the same named species (0.02 – 0.10) but, in spite of hybridization, strong differentiation between species (0.15 – 0.40). This genetic differentiation is not explained by geographic distance either in analyses of single species or across species. Instead, up to a third of the genetic differentiation between sites can be explained by differences in climate even when including geographic and taxonomic distance in the model. Sites that are more climatically similar are also more genetically similar. Because ddRAD samples the genome haphazardly, the environmental effect described here is more likely explained by broad patterns of gene flow than by local adaptation. Environmental constraints on gene flow are likely to influence these oaks’ evolutionary response to climate change and the modeling framework described here can easily be extended through time to predict potential genetic exchange under future climate scenarios.