A simulation-based method for genotype-environment association analysis

Genotype-environment association (GEA) analyses aim to identify genetic loci responsible for local adaptation by evaluating the correlation between allele frequencies and environmental factors across a species' range. A significant hurdle in this approach is separating genuine adaptive signals from spurious associations caused by population structure.

Various methods have been devised to address population structure, yet they may lag in statistical power or generate false positives under certain conditions. In response, the researchers introduce a novel GEA method called SimGEA. Essentially, SimGEA derives a neutral evolutionary model that mirrors the population structure found in real-world data and employs this model to generate neutral alleles.

By applying the same GEA statistic to both empirical and simulated data, SimGEA gauges the significance of observed associations against neutral expectations that consider population structure. The performance of SimGEA was evaluated against existing GEA methods, such as LFMM2 and BayPass, using simulations of local adaptation in two-dimensional space.

The outcome showed that SimGEA consistently maintained the false discovery rate without significantly compromising statistical power across the tested scenarios. These findings suggest that calibrating statistics using neutral simulations offers a robust and adaptable approach for addressing population structure in GEA analyses.