An Evaluation of Processes Critical to Predicting the Carbon Sink of Natural Tropical Forests in a Demographic Vegetation Model

The direct effects of rising atmospheric CO2 concentrations on tropical forests have been the focus of a large body of research including manipulative experiments, observational studies and model estimation. The work presented here seeks to evaluate the processes involved in modelling forest dynamics under changes in atmospheric CO2, and ascertain the strengths and deficiencies of these representations. To do this, the Ecosystem Demography Model 2 (ED2) and the Community Land Model (CLM 4.5-BGC) are used to simulate the vegetation dynamics of an old-growth Central Amazonian forest through the next century, and are compared with flux and inventory data. Using default calibrations (regional specificity), both models were found to overestimate mortality rate and biomass increment (by 1.4 and 0.8 Mg ha-1 yr-1 in ED2 and CLM respectively). This comparison has lead to a closer examination of mortality, the allocation of assimilated carbon and the phasing of plant competition. An analysis of model output and literature review corroborate that tree mortality in old growth tropical forests is complex and is driven by a variety of mechanisms. We find that mortality parameterizations used in earth system models may benefit from simplicity until a more comprehensive mechanistic understanding of mortality and its drivers becomes available. An analysis of field data also showed that a significant fraction of mature trees in the upper canopy were exhibiting no increment in growth. It is not immediately clear if these trees are exhibiting decreased net primary production, or alternatively, how these trees have shifted their resource usage strategy. Demographic ecosystem models such as ED2 provide a means to represent and test these alternative hypotheses as they emerge.