Simulating Guinea Grass Production: Empirical and Mechanistic Approaches

Tropical grasses are economically important for cattle production in Brazil, and accurate simulation models for tropical pastures can benefit forage researchers and farm managers by improving tropical forage production systems. This research calibrated and validated four modeling approaches of contrasting complexity to simulate mass production of Mombaca Guinea grass (Panicum maximum Jacq.). The models included three empirical agro-climatic models (i.e., using cumulative degree days, photothermal units, and a climatic growth index) and a biophysical simulation model, Agricultural Production Systems Simulator (APSIM)-Growth. Data sets for calibration and independent validation included frequent records of aboveground dry matter production during the 2005-2006 and 2010-2011 growing seasons from three trials. All models performed well during calibration (R-2 = 0.78-0.86; coefficient of variation = 26-32.1%). During model validation, the R-2 varied between 0.69 and 0.78, the agreement index was between 0.88 and 0.93, the coefficient of variation between 37.6 and 50.2%, and the mean bias error was between 6 and 470 kg ha(-1). Even though all models were in agreement between simulated and observed results, APSIM-Growth was able to simulate Guinea grass production across broader climatic, soil, and management (e.g., N fertilization) conditions.