Constraining uncertainty in terrestrial tropical carbon flux dynamics requires capturing local biogeochemical influences on structure and function

Spatial heterogeneity in tropical forest productivity and resulting rates of carbon uptake and storage emerge from variation in ecosystem structure and functional traits reflecting differences in climate, edaphic conditions, evolutionary history, and natural and anthropogenic disturbance histories. Yet, models poorly represent this heterogeneity. Remote sensing data provide landscape-scale measures of tropical forest heterogeneity in structure and functional traits that can be used to advance terrestrial biosphere models. To examine whether forest functional traits related to photosynthetic capacity can be used to improve predictions of tropical biomass dynamics and carbon fluxes, we parameterized the Ecosystem Demography model version 2.2 (ED2.2) using canopy traits derived from visible to shortwave infrared (VSWIR) airborne imaging spectroscopy data across an edaphic gradient in Borneo. We find significant site-level differences in relationships between SLA and foliar nutrient concentrations, suggesting that remotely sensed foliar traits can be used to capture variation in photosynthetic capacity at large, edaphically varying spatial scales. We further show that plant functional types parameterized with site-constrained trait values yield more accurate predictions of canopy demography, forest productivity and above-ground biomass dynamics than simulations that depend solely on parameterization of edaphic conditions. However, the most substantial improvements result from allowing for site-level variation in background disturbance rates in the model. Our study reveals the importance of capturing tropical forest heterogeneity in terrestrial biosphere models, particularly as it relates to nutrient availability and disturbance processes.