Enrico Tordoni

Question Trait‐based ecology is built on the notion that traits impact individual performance, but trait–growth relationships have rarely been tested considering the intraspecific variation, trait interactions, and potential nonlinearity in these relationships. Location Seven tropical dry forests throughout Colombia. Methods We examined trait–growth relationships in 263 woody species of tropical dry forests in Colombia, including trait and growth information measured in the same individuals. We measured height, stem, and leaf traits related to growth and hydraulic safety–efficiency in 967 individuals in seven permanent plots (1 ha). We assessed trait–growth relationships using random forest models for each plot with different trait resolutions (individual, plot, area) and complexity of trait interactions (low, medium, high). Results Trait–growth relationships were generally weak without significant differences among trait resolutions or levels of interactions between traits. However, when considering leaf phenology, the proportion of growth variation explained in deciduous species was almost three times higher than in evergreen species. Finally, we found that the effect of traits on growth was consistent across plots and phenological strategies, with both stem (vessel area and pit diameter aperture) and leaf (leaf area and specific leaf area) traits ranked as important predictors, the relevance of which depended on the species’ leaf phenology. Conclusions Our findings indicate that growth is probably limited by different factors depending on the species’ physiology according to the leaf habit considered. However, other factors not included in the analysis, such as microenvironmental variability, might influence trait–growth relationships in tropical dry forests. Overall, our results suggest that a trait coordination approach at the whole‐plant level is needed to better understand plants’ performance and demographic rates in this ecosystem. Further studies on the traits of regional flora and consideration of longer growth periods would help to elucidate the dynamics governing trait composition in these forests, which is essential for the design of adequate forest restoration and conservation practices.