Pasture growth simulation and sensitivity analysis using APSIM-related models in a tropical silvopastoral system

Abstract

Silvopastoral systems integrate trees, pastures, and animals aiming for sustainable livestock production. The APSIM model has been used to simulate silvopastoral systems but the estimates of light transmission through the tree canopies, belowground water competition, and their effects on pasture production, have not been well adjusted. Thus, we improved the simulation of pasture growth within a silvopastoral system by integrating the APSIM-Tropical Pasture and APSIM-Slurps models and calibrated the coupled models to simulate light transmission through the trees along with water competition among species. Observed data was obtained from a silvopastoral system established in São Carlos, SP, Brazil, during two experimental periods that went from Dec-2014 until May-2016 and from Apr-2018 until May-2019. The calibration of the combination of the models was considered overall accurate. The APSIM-Slurp model simulated well (R² = 0.64, NSE = 0.60) the radiation interception by trees. The APSIM-Tropical Pasture model showed a good performance to simulate pasture (R² = 0.90 and NSE = 0.72), leaf (R² = 0.82 and NSE = 0.44), and stem (R² = 0.82 NSE = 0.75) biomass and an acceptable performance for pasture LAI (R² = 0.76 and NSE = 0.58). With that, a multifactorial simulation was performed to test the sensitivity of pasture production as a function of levels of nitrogen, tree leaf area index (LAI), tree root water extraction coefficient, post grazing residual biomass, and season of the year, using 39 years of weather data from the same location of the field experiment as input. The sensitivity analysis showed that the combination of the models was sensitive to variations of the factors with the greatest sensitivity occurring for the tree LAI. The assembling of APSIM-Tropical Pasture and APSIM-Slurp models well simulated pasture production in a silvopastoral system with different managements, indicating the potential of its use as a decision-making tool for silvopastoral systems´ design.