Modelling of tropical pasture growth using DairyMod: Model parameterisation and validation across multiple environments

Abstract

Tropical forages are the primary feed source for livestock production in tropical and subtropical regions. Biophysical modelling has been an effective tool to explore the likely performances of forage species under different edaphoclimatic and agronomic management practices. The existing models are lacking in parameterised and validated tropical pastures, thus hindering their use for tropical and subtropical regions. The aims of this study were to parameterise the DairyMod, a mechanistic biophysical pasture model and robustly validate the species-specific parameters for the prediction of the growth of the three tropical pastures; Megathyrsus maximus ‘Gatton Panic’ (GP), Brachiaria ruziziensis x B. decumbens x B. brizantha ‘Brachiaria Mulato II’ (BM), and Chloris gayana ‘Rhodes grass cv. Reclaimer’ (RR). The model was calibrated using measurements of biomass components, canopy structure, and carbon assimilation collected from a field experiment at the Gatton Research Dairy Farm Queensland, Australia. The model was tested extensively using the published data from a diverse set of environments and management practices (16 data sets, 32 experiments, 14 different locations across South America, North America, Australia, and Africa). In the model parameterisation stage, DairyMod model predicted the aboveground biomass with good agreement for all tropical pastures with a high R2 of 0.92, 0.98, 0.74 and low RMSE of 341, 583, 848 kg DM ha–1 for BM, GP and RR, respectively. The model agreement was good for the validation data with R2 of 0.86, 0.80, 0.87 and RMSE of 954.5, 790.5, 633.2 kg DM ha–1 for the BM, GP and RR, respectively. The model predicted leaf and stem partitioning relatively poorly, and the model also struggled to simulate realistic pasture growth in Mediterranean and desert environments (R2 < 0.50). The study has improved the robustness and accuracy of DairyMod in relation to tropical pastures under tropical and subtropical climate conditions. Our robust and widely tested model can be successfully used for broader explorations of tropical pastures for improving livestock production systems in the tropics and subtropics, while calling for future improvement of the model accuracy in the areas of tropical pasture biomass partitioning, stubble dynamics and low temperature stress recovery function.