The effect of vertical gradients in photosynthetic parameters on predictions of gross primary productivity of a tropical forest in terrestrial biosphere models

Abstract:

To scale model predictions of carbon and water exchange from the leaf to the whole canopy, Terrestrial Biosphere Models (TBMs) rely on assumptions of how photosynthetic traits vary vertically within canopies. Typically, the maximum carboxylation rate (V_c,max) is scaled using an exponential decrease from the top canopy (V_c,max0) to the ground. Additional parameters such as maximum electron transport rate (J_max) and dark respiration are modeled as a constant ratio of V_c,max, but others, such as those associated with stomatal conductance, are considered to be constant. It is not clear how these assumptions impact projections of Gross Primary Productivity (GPP) by TBMs in tropical forests which have particularly complex canopies and limited observational data. To address this, we performed gas exchange measurements to assess the vertical distribution of leaf traits in 10 vertical profiles in a moist evergreen forest in Panama. We focused on 11 leaf traits including photosynthetic and stomatal parameters. To increase the sampling density, we also measured the reflectance spectra of around 400 leaves from 60 different species and built models to predict traits from leaf spectra. Using these data, we modeled GPP for an average day of the dry season with the canopy photosynthesis module of the Functionally Assembled Terrestrial Ecosystem Simulator (FATES). We found vertical patterns in photosynthetic parameters with a strong variability within and across our diverse profiles. V_c,max0 and J_max0 were 20% and 10% higher than the default values implemented in FATES for evergreen tropical forests, resulting in a 50% increase in the dry season GPP. Driving the model with the observed vertical parameter distribution increased the GPP by an additional 10%. The stomatal parameters were vertically structured suggesting that top-of-canopy leaves had more water use efficient behavior. However, driving the model with the observed gradient in stomatal parameters resulted only in a small (1%) reduction in GPP prediction. These results highlight the importance of accurately estimating the top of canopy parameters, and the challenge of doing so in highly diverse tropical forests. We also show that while current representation of vertical structure could be improved, it is not the main source of variation in projected GPP.