Adaptive relationships between traits and the environment are often inferred from observational data by regressing community-weighted mean (CWM) traits on environmental gradients. However, trait‒environment relationships are better understood as the outcome of trait‒abundance and environment‒abundance relationships, and the interaction between traits and the environment. Accounting for this functional structure and for interrelationships among traits should improve our ability to accurately describe general trait‒environment relationships. Using forest understory communities in Wisconsin, we applied a generalized mixed model (GLMM) incorporating this structure. We identified a simple hierarchy of trait‒environment relationships dominated by a strong positive effect of mean temperature on plant height. Compared to the traditional CWM approach, the GLMM was more conservative in identifying significant trait‒environment relationships, and also detected important relationships that CWM regressions overlooked. This work highlights the need to consider the complexity underlying trait‒environment relationships in future analyses
