Modelling the long-term dynamics of tropical forests: from leaf traits to whole-tree growth patterns

Tropical forests are the most diverse terrestrial ecosystems and home to numerous tree species competing for resources in space and time. Functional traits influence the ecophysiological performance of tree species, yet the relationship between traits and emergent long-term growth patterns is poorly understood. Here, we present a novel 3D forest stand model in which growth patterns of individual trees and forest stands are emergent properties of leaf traits. Individual trees are simulated as 3D functional-structural tree models (FSTMs), considering branches up to the second order and leaf dynamics at a resolution of one cubic meter. Each species is characterized by a set of leaf traits that corresponds to a specific position on the leaf economics spectrum and determines light-driven carbon assimilation, respiration and mortality rates. Applying principles of the pipe model theory, these leaf scale processes are coupled with within-tree carbon allocation, i.e., 3D tree growth emerges from low-level processes. By integrating these FSTMs into a dynamic forest stand model, we go beyond modern stand models to integrate structurally detailed internal physiological processes with interspecific competition and interactions with the environment in diverse tree communities. For model calibration and validation, we simultaneously compared a large number of emergent patterns at both the tree and forest levels in a pattern-oriented modelling framework. At the tree level, the specific leaf area and correlated leaf traits determined the maximum height and age of trees, as well as their size-dependent growth rate and shade tolerance. Trait variations along the leaf economics spectrum led to a continuous transition from fast-growing, short-lived and shade-intolerant to slow-growing, long-lived and shade-tolerant trees. These emerging patterns resemble well-known functional tree types, indicating a fundamental impact of leaf traits on long-term growth patterns. At the forest level, a large number of patterns taken from lowland Neotropical forests were reproduced, indicating that our forest model simulates structurally realistic forests over long time spans. Our ecophysiological approach improves the understanding of how leaf level processes scale up to the tree and the stand level, and facilitates the development of next-generation forest models for species-rich forests in which tree performance emerges directly from functional traits.