Understanding local adaptation is critical for conservation management under rapidly changing environmental conditions. Local adaptation inferred from genotype-environment associations may show different genomic patterns depending on the spatial scale of sampling, due to differences in the slope of environmental gradients and the level of gene flow. We compared signatures of local adaptation across the genome of mountain ash (Eucalyptus regnans) at two spatial scales: a species-wide dataset and a topographically-complex sub-regional dataset. We genotyped 367 individual trees at over 3700 single-nucleotide polymorphisms (SNPs), quantified patterns of spatial genetic structure among populations, and used two analytical methods to identify loci associated with at least one of three environmental variables at each spatial scale. Together, the analyses identified 549 potentially adaptive SNPs at the sub-region scale, and 435 SNPs at the range-wide scale. Thirty-nine genic or near-genic SNPs, associated with 28 genes, were identified at both spatial scales, although no SNP was identified by both methods at both scales. We observed that non-genic regions had significantly higher homozygote excess than genic regions, possibly due to selective elimination of inbred genotypes during stand development. Our results suggest that strong environmental selection occurs in mountain ash, and that the identification of putatively adaptive loci can differ substantially depending on the spatial scale of analyses. We also highlight the importance of multiple adaptive genetic architectures for understanding patterns of local adaptation across large heterogenous landscapes, with comparison of putatively adaptive loci among spatial scales providing crucial insights into the process of adaptation.

We used next-generation sequencing data of mountain ash (Eucalyptus regnans) to compare patterns of putative local adaptation at two spatial scales of sampling. Detailed methods are provided in the published manuscript.

All bioinformatic and R scripts used in the analysis are provided as part of this dataset. ANGSD output, scripts, and GDM data are provided in separate folders, with separate metadata spreadsheets and unfiltered genotypes (filtering steps provided in the main R script).

See README file for all details.

All sequence data in BAM file format has been uploaded to the NCBI Sequence Read Archive (PRJNA560448).