Predicting diffuse light-enhancement of GPP from plant functional traits: A multi-site synthesis

Diffuse light enhances canopy-scale photosynthesis because isotropic diffuse light penetrates deeper into the canopy, involves more leaf area in photosynthesis, and prevents the top of the canopy from becoming light saturated. However, the observational and modeling communities still have little understanding of how the 'Diffuse light Enhancement Effect' (DEE) varies across plant functional types or is constrained by factors such as nitrogen availability and plant structure. So far, variability in the strength of DEE across plant functional types (PFTs) remains poorly constrained, but canopy models indicate leaf area index (LAI) is a primary controller. While the very few existing multi-site, measurement-based syntheses of the DEE have provided valuable information on the variability of the DEE across a few plant functional types, no study has correlated measured metrics of DEE magnitude with direct measurements of canopy physical traits across a wide range of plant functional types. Here we report a new metric that is suitable for quantifying the DEE in both flux measurements and land surface models. We also present, for the first time, an examination of the relationship between the DEE metric and plant functional traits. Results from our 70+ site AmeriFlux and FLUXNET synthesis indicate that LAI is the strongest controller of the DEE across sites and PFTs, with less significant influences from foliar nitrogen, canopy height, and mean annual precipitation. Our results will enable direct evaluation and improvement of remote sensing algorithms and light use efficiency models (e.g. MODIS GPP), which to this point regard diffuse light fraction as a source of noise. Additionally, improving resolution of the DEE in prognostic land surface models, such as the Community Land Model (CLM), will greatly improve our ability to forecast future feedbacks to terrestrial carbon sequestration from changes in cloudiness and aerosol amount.