Plant trait-based assessment of the Pasture Simulation model

We used a functional trait-based plant classification to improve the plant module of the biogeochemical grassland model PaSim. Based on four main classes (A, B, C, D) and two derived types (b, d) covering a gradient from high to low productive/fertile grassland vegetation types, we derived new classes of model plant parameters representing an evolution of a previous parameterization obtained by calibration without considering any plant diversity. Illustrative results are presented for the French grassland site of Laqueuille, by comparing two grazing management treatments: intensive (type B) and low animal stocking rate extensive (type b). [br/] Model performances (reflected by root mean square error and coefficient of determination metrics) showed that accounting for plant traits may help predicting carbon-water fluxes (actual evapotranspiration, gross primary productivity, ecosystem respiration and net ecosystem exchange) and soil variables (temperature and water content). Whether the pattern of results (yet complex) generally supported the validity of the plant trait-based approach to derive model parameters, a substantiation is required by assessing model performances on a range of sites (as listed in the paper) covering a wide variety of condition