Linking tropical forest function to hydraulic traits in a size-structured and trait-based model
A major weakness of forest ecosystem models is their inability to capture the diversity of responses to changes in water availability, severely hampering efforts to predict the fate of tropical forests under climate change. Such models often prescribe moisture sensitivity using heuristic response functions that are uniform across all individuals and lack important knowledge about trade-offs in hydraulic traits. We address this weakness by implementing a process representation of plant hydraulics into an individual- and trait-based model (Trait Forest Simulator; TFS) intended for application at discrete sites. The model represents a trade-off in the safety and efficiency of water conduction in xylem tissue through hydraulic traits, which then lead to variation in plant water use and growth dynamics. The model accounts for the buffering effects of leaf and stem capacitance (Cleaf and Cstem) on leaf water potential at short time scales, and cavitation-induced reductions in whole-plant conductance over longer periods of water stress. We conducted meta-analyses of drought tolerance traits (osmotic potential at full turgor πo, bulk elastic modulus ε, apoplastic fraction af) for leaves and stems in tropical forests to inform links between hydraulic trait spectra and other plant traits, such as maximum photosynthetic capacity (Amax),leaf mass per area (LMA) and wood density (WD).
Meta-analysis revealed a significant negative and positive relationship of leaf πo and ε with WD (and to a lesser extent, LMA), respectively, while WD was a poor predictor sapwood πo and ε. However, a strong negative relationship of Cstem with WD emerged, driven by predictable changes in the af and saturated sapwood moisture content with WD. Incorporating these relationships in the model greatly improved the diversity of tree response to seasonal changes in water availability as well as response to drought, as determined by comparison with sap flux and stem dendrometry measurements. Importantly, this individual- and trait-based framework provides a testbed for identifying both critical processes and functional traits needed for inclusion in coarse-scale Dynamic Global Vegetation Models, which will lead to reduced uncertainty in the future state of tropical forests.