On the potential for GWAS with phenotypic population means and allele-frequency data (popGWAS)

It is vital to understand the genomic basis of differences in ecologically important traits if we are to understand the impact of global change on biodiversity and enhance our ability for targeted intervention. This study explores the potential of a novel genome-wide association study (GWAS) approach for identifying loci underlying quantitative polygenic traits in natural populations, based on phenotypic population means and genome-wide allele frequency data as obtained e.g. by PoolSeq approaches. Extensive population genetic forward simulations demonstrate that the approach is generally effective for oligogenic and moderately polygenic traits and relatively insensitive to low heritability. However, applicability is limited for highly polygenic architectures and pronounced population structure. The required sample size is moderate with very good results being obtained already for a few dozen populations scored. When combined with machine learning for feature selection, the method performs very well in predicting population means. The data efficiency of the method, particularly when using pooled sequencing and bulk phenotyping, makes GWAS studies more accessible for research in biodiversity genomics. Moreover, in a direct comparison to individual based GWAS, the proposed method performed constistently better with regard to the number of true positive loci identified and prediction accuracy. Overall, this study highlights the promise of popGWAS for dissecting the genetic basis of complex traits in natural populations.