Strategies for GWAS include population re-structuring and regional re-sampling. (a) A schematic phylogenetic tree illustrating genetic diversity and population structure in a hypothetical sample of a species whose adaptive traits are to be investigated genetically. (b) To map the loci underlying adaptation at the broadest scale, a balanced core set of accessions is made by pruning closely related individuals from the global set. GWAS can be performed at this stage, but for traits whose variation that is confounded by population structure (Figure 1), crosses are needed. (c) To map loci underlying local adaption, the focus should be on less structured regional sub-samples that are identified in the initial sample (for example, RegMap lines in A. thaliana). GWAS can be performed on these regional samples, which have reduced allelic heterogeneity and confounding by population structure, but LD blocks are likely to be longer in the regional subsets and this will decrease mapping resolution. Regional re-sampling along an environmental cline in the field can increase the power of the association mapping.