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

Highlights

• We present a novel trait-based 3D functional-structural forest stand model.

• Growth of individual trees is an emergent property of their leaf traits.

• The position on the leaf economics spectrum determines long-term growth patterns.

• The emerging tree growth patterns resemble well-known functional tree types.

• The model simulates structurally realistic forests over long time spans.

Abstract

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.

Introduction

Tropical forests provide multiple social, ecological and economical services, represent the most diverse terrestrial ecosystem, and play an important role in the global carbon cycle (Hassan et al., 2005; Heywood and Watson 1995; Malhi and Grace 2000). In addition to high tree diversity, there are even more plant and animal species that directly or indirectly depend on the resources and habitat provided by trees (Erwin, 1988, Nakamura et al., 2017) Almost 9% of all vascular plant species, for instance, live as epiphytes on trees, predominantly in subtropical or tropical regions (Zotz et al., 2021). Ongoing deforestation and potential adverse effects of climate change thus pose a threat to all species associated with tropical forests (Wright 2005). To assess the impact of a changing environment on tropical biodiversity, we thus need a mechanistic understanding of the functioning of these forests, e.g., how they will respond to those changes, and how associated species respond to changing forest dynamics.

A variety of dynamic models are available to predict future changes of tropical forests and to analyze the functioning of these ecosystems. These models differ considerably in the level of detail and in temporal and spatial resolution. Among these models, dynamic global vegetation models focus on large-scale predictions of vegetation dynamics and carbon cycles, but commonly use simplistic forest structures (e.g., Cramer et al., 2001, Purves and Pacala 2008). At small to medium scales (<1 ha to >100 km²), forest gap models and forest landscape models simulate forest dynamics and tree species composition (Bugmann 2001; Petter et al., 2020). Such models represent forest structure in more detail by including stems and crowns of individual trees or cohorts, allowing simulations of within-canopy light attenuation and competition among different species or functional types of trees (Fischer et al., 2016; Köhler and Huth 1998). Even finer structural details can be simulated by functional-structural tree models (FSTMs), in which trees are represented in 3D space by interconnected structural and functional units, such as branches, leaves, or reproductive organs (Godin and Sinoquet 2005; Sievänen et al., 2014). These ‘virtual trees’ explicitly integrate complex, mechanistic interactions between tree architecture and physiological processes, for instance, light-dependent within-tree carbon acquisition and allocation at the meristem level in growing trees (Fourcaud et al., 2008; Sterck et al., 2005). FSTMs are therefore suitable tools to study structural tree growth as an emergent property of lower level ecophysiological processes. Furthermore, integrating FSTMs into forest stand models can improve our understanding of the fundamental principles that interlink tree and forest dynamics. In addition, such detailed 3D forest models could be useful for model-based studies of the spatio-temporal dynamics of canopy-dwelling plants and animals (Cabral et al., 2015; Petter et al., 2021). However, only few attempts have been made to couple FSTMs with forest stand models, and these studies focused on growth of even-age monocultures over short time periods (Feng et al., 2011; Guillemot et al., 2014). So far, species-rich forest models that are based on FSTMs and simulate all crucial demographic processes over long periods of time, i.e., regeneration, growth and mortality, are not available.

Extending FSTMs to tropical forest stands is both computationally and conceptually challenging. On the one hand, stand-scale FSTMs are computer-intensive due to their complexity and thus require efficient modelling techniques to keep simulations feasible. On the other hand, tropical forests pose particular challenges due to their high species diversity (Slik et al., 2015; ter Steege et al. 2013). In contrast to temperate forests, for which the low number of well-studied tree species allows calibrating ecophysiological parameters at the species level (e.g., Seidl et al., 2012), alternative approaches are required for species-rich forests where species-specific physiological information is scarce. Individual-based models for tropical forests typically use distinct functional groups aggregating tree species with similar characteristics (e.g., Fischer et al., 2016; Köhler and Huth 1998). In the simplest case, only light-demanding pioneers and shade-tolerant climax species are distinguished (Swaine and Whitmore 1988), but a classification into more groups has also been proposed (Chazdon et al., 2010; Gourlet-Fleury et al., 2005). While functional group approaches are useful, they retain a simplification of the continuum from fast growing, short-lived pioneer to slow growing, long-lived shade-tolerant species (Denslow 1987; Wright et al., 2003). Similar trade-offs between growth and mortality occur at the leaf scale (Wright et al., 2004): many leaf traits co-vary strongly, and this variation is largely explained by a single principal axis - the leaf economics spectrum (LES). The LES runs from leaves with high photosynthetic capacities and low life spans to leaves with low photosynthetic capacities and long life spans. Hence, a relationship between leaf traits and whole-tree performance can be assumed, and significant relationships were observed for many tropical tree species (Poorter and Bongers 2006; Sterck et al., 2006). A leaf trait-based approach should thus be a promising way to integrate the different life history strategies of trees into forest models. However, we are not aware of any study in which 3D growth over a tree´s entire life span has been modelled as an emergent property of the tree´s set of traits.

Here, we present a dynamic forest stand model in which trees are represented as 3D functionally- and structurally-explicit individuals. This model simulates the long-term forest dynamics (500–1000 years) at the plot scale (∼1 ha) with a high level of structural detail and functional diversity. Branches are considered up to the second order and leaf biomass development is modelled at a resolution of 1 m³, which allows detailed consideration of competition for light and space. Tree species are characterized by a set of leaf traits under consideration of the trade-offs and correlations between traits (LES; Wright et al., 2004). Using the principles of the pipe model theory (Shinozaki et al., 1964), the light-driven carbon assimilation at the leaf level and the within-tree carbon allocation are coupled. We hypothesize that this continuous, trait-based approach captures essential life history variations across species regarding their growth, survival, and light demand. In addition, we hypothesize that the long-term dynamics of diverse, tropical forest communities can be reproduced by coupling the FSTM with a forest stand model, in which key demographic processes and between-tree competition are simulated (e.g., competition for light and space, neutral recruitment). To test these hypotheses, we contrast emerging growth patterns at the tree level (e.g., diameter growth rates, maximum height and life span) against observations. In addition, we simulate forest stands from bare soil over 500 years and test if the model can successfully reproduce multiple attributes of Neotropical forests. Specifically, we test if a dynamic equilibrium is reached and if 12 forest attributes (e.g., basal area, net primary production, mortality rate) and several ecological patterns (e.g., diameter and height distribution, vertical leaf area distribution) resemble field observations. Our main purpose here is to present the structure, parameterization and validation of the model, and to explore how trait trade-offs at the leaf level are functionally linked to emergent patterns of tree growth and forest stand dynamics. Our model can form the basis for more detailed future studies of (successional) dynamics, distribution of functional and structural traits and tree diversity in detailed 3D forests, or spatiotemporal dynamics of canopy dwellers, making it an important tool for forest and canopy ecologists.