Simulating Amazon forest carbon cycling using an individual- and trait-based model.

Abstract:

The Amazon forest, a regional and global regulator of climate and store of enormous biodiversity, is an incredibly complex ecosystem. Just one ha of forest can contain 300 different species of tree, with an estimated 16,000 tree species present in the region. Different tree species, and even different individuals of a species, vary in their functional traits, influencing how they behave in response to the environment. Dynamic global vegetation models (DGVMs) are commonly used to simulate the response of the Amazon forest to global environmental change. Yet, such DGVMs typically use a plant functional type (PFT) approach where variation between individuals and species are not represented, which inherently limits the range of outcomes for Amazonia under climate change. Here, we report on recent advances in an alternative approach to tropical forest modeling that represents the size structure and variation of traits within a community, which we term the Trait-based Forest Simulator (TFS). As originally proposed, TFS was strictly a steady-state model and here we present an extension of TFS which includes full forest dynamics, and has been evaluated with data collected from intensive carbon cycling inventory plots from the GEM (Global Ecosystems Monitoring) network. Specifically, we compare the model output to stand-level data on productivity and respiration of the canopy, stems and roots. The model development process has highlighted ecological tradeoffs that are necessary to integrate into trait-based models, such as a shorter leaf lifetime with a lower leaf mass per area. The adapted TFS model simulates carbon cycling in forest plots, including variation in productivity between sites. These results lend confidence to the ability of next-generation vegetation models to accurately simulate forest sensitivity to future changes.