Natural plant populations often exhibit marked differences in gene expression patterns that can reflect heterogeneity in selective pressures. Analyzing gene expression as a quantitative trait provides a unique opportunity to evaluate the underlying genomic basis of a plethora of traits and their interactions in driving adaptive evolution. We investigated patterns and processes driving expression differentiation under conditions mimicking future climates by combining common garden experiments with transcriptome-wide datasets obtained from hybrid populations of Pinus strobiformis and P. flexilis. We found strong signals of genotype-environment interactions (GEI) at the individual transcript and the co-expression module levels suggesting a marked influence of drought related variables on adaptive evolution. Overall, survival was positively associated with P. flexilis ancestry, but it exhibited an environment-specific pattern. Co-expression modules exhibiting strong associations with survival and genomic ancestry were representative of similar functional categories across both gardens. Using network topology measures, putatively adaptive garden-specific expression traits were pleiotropic and belonged to modules exhibiting high population differentiation yet low preservation across gardens. Overall, our study suggests the presence of substantial genetic variation underlying univariate and multivariate traits in novel climates that may enable populations of long-lived forest trees to respond to rapid shifts in climatic conditions in early seedling stages when mortality tends to be the highest. Our finding of pleiotropic trait architectures underlying adaptive traits, however, implies rapid adaptive responses to changing selection pressures depend on whether trait covariances align with the direction of change in selection pressures.