Genomic selection in forest tree breeding opens up new opportunities for accelerating selection cycles, for improving selection accuracy and for integrating explicit diversity management. Traits targeted by forest breeding programs are mostly final performances of the trees (height, circumference). However, such phenotypes are very integrative: the final performances are the results of multiple years of growth in interaction with an evolving environment. To tackle these interactions and to propose new breeding strategies including an environmental information, we model individual reaction norms using random regression models for maritime pine. We used wood ring traits describing trees growth and response to past environmental conditions and we related them with annual environmental indexes. The data came from two experimental trials located in the Landes forest in France. From 25 half-sib families, a total of 650 individuals aged of 26-years-old have been cored and genotyped using the 4TREE array. Random regression models allowed us to extract genetic parameters related to tree growth regarding several climatic indexes. We investigated the evolution of genetic parameters including breeding values in order to (i) understand the evolution of the genetic control of growth, (ii) to identify environmental variables responsible for differential responses in trees, and (iii) to propose new selection criteria taking into account the environmental information. For the last point, we aim to identify differentiated response curves of genetic values and to predict performances in unobserved environments.