How useful are genomic data for predicting maladaptation to future climate?

Area of Science:

• Ecology
• Genetics
• Climate Change Biology

Background:

• Genomic methods are increasingly used to predict population maladaptation to environmental change.
• Lack of model validation hinders the application of these genomic approaches in conservation and policy.
• Two common methods, Gradient Forests (GF_offset) and risk of non-adaptedness (RONA), require rigorous validation.

Purpose of the Study:

• To compare the validation of maladaptation estimates from GF_offset and RONA methods.
• To assess the sensitivity of these algorithms to the source of input genetic markers (GEA candidates vs. random markers).
• To evaluate the predictive power of these methods against empirical transplant experiment data.

Main Methods:

• Exome capture pool-seq data from 35-39 populations of Douglas-fir and jack pine were analyzed.
• Maladaptation estimates were generated using Gradient Forests (GF_offset) and risk of non-adaptedness (RONA).
• Model performance was validated against 2- and 5-year growth and mortality data from independent conifer transplant experiments.

Main Results:

• Both GF_offset and RONA methods generally predicted transplant performance better than climatic or geographic distances.
• Using genotype-environment association (GEA) candidate markers did not improve the performance of GF_offset or RONA models compared to random markers.
• Despite promising validation, variability in future climate projections complicates the identification of the most maladapted populations.

Conclusions:

• This study advances the understanding of the sensitivity and applicability of genomic maladaptation prediction methods.
• GF_offset and RONA show potential for predicting population adaptive capacity but require careful validation.
• Recommendations for the future use and development of these genomic approaches in conservation are discussed.