Résumé
Allometric equations relating stem diameter to tree height (H–D) and crown area (CA–D) are widely applied to estimate forest structure and biomass. Few studies, however, have assessed whether differences in H–D and CA–D models among forest types are driven by species variability or plasticity. We examined six forest types in Central Africa to test the contribution of species variability and evaluate whether models incorporating forest-type data improve predictive accuracy. Data included 845 trees (52 species, 49 genera, 17 families). Variance partitioning showed that H–D allometry varied significantly among forest types, with diameter, forest type, and species jointly explaining 80% of variance. Excluding forest type increased the variance explained by species from 9% to 14%. For CA–D allometry, predictors explained 72% of variance, with independent effects of forest type (2%) and species (5%). Removing forest type did not shift variance toward species. Models incorporating forest-type information consistently yielded lower prediction errors than generalized models. These results demonstrate that species variability is the dominant driver of allometric relationships in Central African forests, although the balance between variability and plasticity differs between height and crown dimensions. Our findings highlight the importance of forest-type-specific allometric models for accurate carbon stock estimation and remote sensing calibration.