Landscape Genomics in Populus trichocarpa: Disentangling the Contributions of Migration, Hybridization and Natural Selection

The study of genome wide data from wild individuals of an organism sampled from across its range (landscape genomics) can potentially disentangle genetic drift from natural selection and identify genes important for plant survival. To demonstrate this approach we have quantified geographical patterns of genetic variation in Populus trichocarpa, an ecologically important tree species widespread in the Pacific Northwest of North America. We genotyped a large population sample (498 individuals) from a wide geographical area (latitude 44.3° N to 59.6° N) using a Populus 34,000 SNP genotyping array (Geraldes et al., 2013). Three processes contribute to the observed patterns: (1) introgression from the sister species P. balsamifera at the range margin  (2) isolation-by-distance and (3) natural selection. Isolation-by-distance is strongly evident when the whole sampled area is considered, but in a core area (Southern British Columbia) there is no evidence of restricted gene flow. Fst outliers are numerous, both across the entire area sampled and in Southern British Columbia alone. Evidence that these Fst outliers may be due to selection comes from two additional lines of evidence. First, GO analyses reveal that Fst outliers are overrepresented in classes of genes involved in response to environment. These are circadian rhythm and response to red/far-red light genes (entire dataset) and heat response genes (core area). Secondly, we identify strong correlations between geographic/environmental variables and allele frequencies at Fst outlier loci. We conclude that local adaptation has a major effect on the distribution of genetic variation across the landscape in this species and identify candidate allelic variants that appear to underlie local adaptation.