Integrating environmental gradients into breeding: application of genomic reactions norms in a perennial species

Abstract

Global warming threatens the productivity of forest plantations. We propose here the integration of environmental information into a genomic evaluation scheme using individual reaction norms, to enable the quantification of resilience in forest tree improvement and conservation strategies in the coming decades. Random regression models were used to fit wood ring series, reflecting the longitudinal phenotypic plasticity of tree growth, according to various environmental gradients. The predictive performance of the models was considered to select the most relevant environmental gradient, namely a gradient derived from an ecophysiological model and combining trunk water potential and temperature. Even if the genotype ranking was preserved over most of the environmental gradient, strong genotype x environment interactions were detected in the extreme unfavorable part of the gradient, which includes environmental conditions that are very likely to increase in the future. Combining genomic information and longitudinal data allowed to predict growth in unobserved environments: considering an equivalent phenotyping effort, the cross-validation scenarios led to predictive performances ranging from 0.25 to 0.59 highlighting the importance of phenotypic data allocation. Genomic reaction norms are useful for the characterization and prediction of the function of genetic parameters and facilitate breeding in a climate change context.