Increasing Functional Diversity in a Global Land Surface Model Illustrates Uncertainties Related to Parameter Simplification

Abstract

Simulations of the land surface carbon cycle typically compress functional diversity into a small set of plant functional types (PFT), with parameters defined by the average value of measurements of functional traits. In most earth system models, all wild plant life is represented by between five and 14 PFTs and a typical grid cell (≈100 × 100 km) may contain a single PFT. Model logic applied to this coarse representation of ecological functional diversity provides a reasonable proxy for the carbon cycle, but does not capture the non‐linear influence of functional traits on productivity. Here we show through simulations using the Energy Exascale Land Surface Model in 15 diverse terrestrial landscapes, that better accounting for functional diversity markedly alters predicted total carbon uptake. The shift in carbon uptake is as great as 30% and 10% in boreal and tropical regions, respectively, when compared to a single PFT parameterized with the trait means. The traits that best predict gross primary production vary based on vegetation phenology, which broadly determines where traits fall within the global distribution. Carbon uptake is more closely associated with specific leaf area for evergreen PFTs and the leaf carbon to nitrogen ratio in deciduous PFTs.Plain Language SummaryPlants play a critical role in the global carbon cycle, and diversity has been shown to influence vegetation productivity. However, when the land surface is simulated in a global model all wild plant life is reduced to a small number of plant functional types. Here we estimate how incorporating diversity influences ecosystem carbon uptake in 15 globe spanning landscapes. We find that diversity has a strong influence on modeled productivity, particularly in the arctic and tropics. Further, we find that whether plants shed their leaves annually has a strong influence on where traits fall within the global distribution and thus how traits and productivity interact.Key PointsWe implemented distributions of leaf economic spectrum traits in a land surface model in 15 diverse landscapesTrait variation has a substantial influence on gross primary production (GPP)Phenology plays a key role in guiding where traits fall within the global trait distribution and hence trait‐GPP relationships