Understanding forest responses to climate change requires moving beyond mean-based trait–environment relationships and explicitly quantifying how climate affects intraspecific trait variability. Trait dispersion within populations may represent a key component of adaptive capacity, particularly for hydraulic traits linked to drought resistance and tree survival.

We analysed five functional traits — P50, Kmax, Huber Value, specific leaf area and wood density — measured on >1,100 individual trees from 111 Gene Conservation Units across 11 European tree species within the FORGENIUS network. We applied hierarchical Bayesian location–scale models to test whether climatic gradients affect trait means and trait dispersion independently. Predictors included interannual variability in water deficit, temperature seasonality, mean annual precipitation, climatic moisture index range and soil water availability. We further tested whether trait dispersion follows Taylor’s Power Law and whether climate modulates mean–variance scaling.

Climate effects on trait means were generally weak.  In contrast, trait dispersion showed an ecologically interpretable climatic responses. P50 dispersion increased with interannual water-deficit variability (β = 0.102, 95% CI [0.035, 0.170]), indicating that climatically unpredictable sites maintain broader hydraulic safety strategies within populations. Conversely, Huber Value dispersion decreased with temperature seasonality (β = −0.094, 95% CI [−0.189, −0.001]), suggesting stronger filtering of allocation strategies under thermally variable climates.

Across traits, dispersion scaled positively with trait means, consistent with Taylor’s Power Law. Scaling was trait-specific, with SLA showing the steepest relationship (b ≈ 1.45), Kmax a lower slope (b = 0.80, 95% CI [0.62, 0.98]) and P50 a moderate slope (b = 1.05, 95% CI [0.05, 2.06]). Climate modulation of mean–variance scaling was mainly detected for hydraulic traits: climatic moisture index range weakened scaling in Kmax (β = −1.107, 95% CI [−1.690, −0.530], p = 0.004) and Huber Value (β = −0.130, 95% CI [−0.240, −0.020], p = 0.020).

These results show that climate may restructure the distribution of functional strategies within tree populations without necessarily shifting their average trait values, highlighting the need to incorporate trait dispersion into assessments of forest vulnerability and adaptive potential.