1. Individual-level demographic outcomes should be predictable upon the basis of traits. However, linking traits to tree performance has proven challenging likely due to a failure to consider physiological traits (i.e., hard-traits) and the failure to integrate organ-level and whole plant-level trait information. 

2. Here, we modeled the survival rate and relative growth rate of trees while considering crown allocation, hard-traits, and local-scale biotic interactions, and compared these models to more traditional trait-based models of tree performance. 

3. We found that an integrative trait, total tree-level photosynthetic mass (estimated by multiplying specific leaf area and crown area) results in superior models of tree survival and growth. These models had a lower AIC than those including the effect of initial tree size or any other combination of the traits considered. Survival rates were positively related to higher values of crown area and photosynthetic mass, while relative growth rates were negatively related to the photosynthetic mass. Relative growth rates were negatively related to a neighbourhood crowding index. Furthermore, none of the hard-traits used in this study provided an improvement in tree performance models. 

4. Synthesis. Overall, our results highlight that models of tree performance can be greatly improved by including crown area information to generate a better understanding of plant responses to their environment. Additionally, the role of the hard-traits in improving models of tree performance is likely dependent upon the level of stress (e.g. drought stress), micro-environmental conditions, or short-term climatic variations that a particular forest experiences.

Functional hard-trait data collected from adult trees in the El Yunque National Forest, Puerto Rico. Species names are given as species codes used by the Luquillo LTER.