On the use of through-fall exclusion experiments to filter model hypotheses.

Abstract:

One key threat to the continued existence of large tropical forest carbon reservoirs is the increasing severity of drought across Amazonian forests, observed both in climate model predictions, in recent extreme drought events and in the more chronic lengthening of the dry season of South Eastern Amazonia. Model comprehension of these systems is in it’s infancy, particularly with regard to the sensitivities of model output to the representation of hydraulic strategies in tropical forest systems. Here we use data from the ongoing 14 year old Caxiuana through-fall exclusion experiment, in Eastern Brazil, to filter a set of representations of the costs and benefits of alternative hydraulic strategies. In representations where there is a high resource cost to hydraulic resilience, the trait filtering CLM4.5(ED) model selects vegetation types that are sensitive to drought. Conversely, where drought tolerance is inexpensive, a more robust ecosystem emerges from the vegetation dynamic prediction. Thus, there is an impact of trait trade-off relationships on rainforest drought tolerance. It is possible to constrain the more realistic scenarios using outputs from the drought experiments. Better prediction would likely result from a more comprehensive understanding of the costs and benefits of alternative plant strategies.