RDAforest: Identifying Environmental Drivers of Polygenic Adaptation.

Area of Science:

• Ecological genomics
• Population genetics
• Bioinformatics

Background:

• Identifying environmental factors that drive genetic adaptation is crucial in ecological genomics.
• Previous methods often focus on single nucleotide polymorphisms (SNPs) and linear relationships, potentially missing complex adaptive patterns.

Purpose of the Study:

• To introduce RDAforest, a novel methodology for identifying environmental gradients influencing genetic adaptation.
• To leverage random forest regression to model the polygenic basis of adaptation and complex environmental interactions.

Main Methods:

• RDAforest utilizes random forest regression to explain genetic covariance structure based on multiple environmental predictors.
• The method detects non-linear and non-monotonous dependencies, including interactions between predictors.
• It incorporates predictor selection for correlated variables and jackknifing for uncertainty assessment.

Main Results:

• RDAforest successfully explains genetic covariance structure by relating it to environmental variables.
• The methodology identifies and maps 'adaptive neighborhoods,' highlighting areas with distinct local adaptation.
• Demonstrated effectiveness in simulated data and a real-world case study of North American grey wolves.

Conclusions:

• RDAforest offers a powerful, flexible approach to understanding the genetic basis of adaptation in response to environmental gradients.
• The generated adaptive neighborhood maps provide valuable tools for conservation planning and ecological restoration.
• This method advances the field of ecological genomics by accounting for complex ecological and genetic interactions.