COS 13-8
Modelling phenology and vegetation growth in tropical dry forest: relating coordination and trade-offs among functional traits to vegetation responses to water stress
Tropical dry forests (TDFs) account for over 40% of potential forested area in the tropics and subtropics. They are characterized by a drastic dry season. Plants in these biomes show diverse phenology and growth patterns in response to the seasonality of water availability. However, TDFs are highly underrepresented in ecosystem models. This affects model predictive power. Plant responses to water stress are ultimately driven by functional traits, which are interconnected by plant economic spectra. Here, we test the hypothesis that inclusion of coordination and trade-offs among functional traits would improve model performance in TDFs. We first examine interconnection among functional traits in tropical dry forests through meta-data analysis. We then implement a plant hydraulic module, a new phenology module and optimization-based stomatal conductance module to relate functional traits with plant responses to water stress in Ecosystem Demography 2 (ED2) model. New plant functional types (PFTs) are defined according to both in-situ observations and the coordination and trade-offs among function traits. We compare the model predictions from 2009 to 2013 in phenology and vegetation growth with field observations from TDFs in Palo Verde, Costa Rica.. We further conduct numerical experiments to show the possible biases in various ecosystem functions if the diversity in functional traits is ignored.
Results/Conclusions
We find that turgor loss point is correlated with both specific leaf area and wood density in TDFs. The relationship suggests there is a trade-off between plant drought tolerance and resource (light and water) acquisition. Combined with identified coordination and trade-offs among functional traits, the updated model simulate more realistic seasonality in leaf area index, leaf litter dynamics and diversity in responses to water stress in the TDF site. The new PFTs exhibt a wide range of phenology and vegetation growth patterns that match the observations. Generally, PFTs with low (high) wood density and high (low) specific leaf area tend to be deciduous (evergreen). The growth of deciduous-like PFTs is also more sensitive to water availability than that of evergreen-like PFTs. Our results suggest that diverse responses to water stress in TDFs can be sufficiently captured by models if coordination and trade-off among functional traits are mechanistically incorporated. Further numerical experiments also imply that not explicitly accounting for trait diversity would introduce large biases in several key land surface processes in the model.