The global land tropics support a third of terrestrial productivity. Although pantropical analyses indicate that mean annual temperature (MAT) and precipitation (MAP) explain about half the variation in aboveground net primary production (ANPP), the source of remaining spatial variation is poorly understood. Here we test the hypothesis that vertical leaf-area profiles explain spatial variation in ANPP when MAT and MAP are held constant. To do this, we combined measurements from airborne light detection and ranging (LiDAR) with stochastic radiative transfer theory to quantify vertical leaf-area profiles over 18 0.5 ha plots in an old-growth Neotropical rain forest landscape in the Atlantic lowlands of Costa Rica. We developed a predictive model for ground-based ANPP (annual wood and litter production in field plots) from vertical leaf-area profiles using a partial least squares regression. We compared the predictive power of vertical leaf-area profiles to that of two alternate metrics, 1) total leaf area and 2) the vertical distribution of LiDAR point measurements.
Results/Conclusions
We found that variation in ANPP not explainable by MAT and MAP can be predicted by vertical leaf-area profiles quantified by LiDAR, but not by simpler measurements of total leaf area or vertical distribution of canopy height. Vertical leaf-area profiles explain 42% of the variation in ANPP among plots when MAT and MAP are held constant within a landscape, a number 1.75 times greater than the variation explained by total leaf area without information about its vertical arrangement, and > 4.6 times greater than models based on the vertical distribution of point measurements from LiDAR. Additionally, only models using vertical leaf-area profiles were unbiased. Our results indicate that developing predictive models of ANPP incorporating biotic factors derived from remote sensing – as opposed to only using abiotic factors like MAT and MAP – can improve forecasting aboveground carbon fluxes in evergreen broadleaf forests.