Intraspecific genetic variation in foundation species such as aspen (Populus tremuloides Michx.) shapes their impact on forest structure and function. Identifying genes underlying ecologically important traits is key to understanding that impact. Previous studies, using single‐locus genome‐wide association (GWA) analyses to identify candidate genes, have identified fewer genes than anticipated for highly heritable quantitative traits. Mounting evidence suggests that polygenic control of quantitative traits is largely responsible for this “missing heritability” phenomenon. Our research characterized the genetic architecture of 30 ecologically important traits using a common garden of aspen through genomic and transcriptomic analyses. A multilocus association model revealed that most traits displayed a highly polygenic architecture, with most variation explained by loci with small effects (likely below the detection levels of single‐locus GWA methods). Consistent with a polygenic architecture, our single‐locus GWA analyses found only 38 significant SNPs in 22 genes across 15 traits. Next, we used differential expression analysis on a subset of aspen genets with divergent concentrations of salicinoid phenolic glycosides (key defense traits). This complementary method to traditional GWA discovered 1243 differentially expressed genes for a polygenic trait. Soft clustering analysis revealed three gene clusters (241 candidate genes) involved in secondary metabolite biosynthesis and regulation. Our work reveals that ecologically important traits governing higher‐order community‐ and ecosystem‐level attributes of a foundation forest tree species have complex underlying genetic structures and will require methods beyond traditional GWA analyses to unravel.