Development of the Pasture Simulation Model for assessing livestock production under climate change

To simulate climate change impacts on pastures and domestic herbivores as well as feedbacks to the atmosphere in terms of greenhouse gas emissions, we have improved a process-based biogeochemical pasture model, PaSim. The overall aim was to simulate the meat and milk production of cattle (suckler cows with their calves, dairy cows and heifers) in response to climate and management, as well as feedbacks to the atmosphere through enteric methane emissions. Herbage intake at grazing was calculated from animal characteristics, herbage availability, diet digestibility and air temperature. With suckler cows, milk production and changes in daily liveweight and body condition were calculated from net energy balance. The net energy intake of dairy cows and their body reserves at turnout to pasture were used to simulate milk production at pasture, daily liveweight and body condition changes, taking into account cow energy requirements and intake capacity. Heifer growth was determined from heifer net energy intake and liveweight. Net energy intake was used to assess enteric methane production through a conversion factor, which depends both on the energy level of the diet and on the herbivore type. The model was assessed against experimental data for animal performance and methane emissions at grazing. Predictions show good agreement with observations. On average, the root mean square error was 6.5, 4 and 2.5% for the liveweights of suckler cows, suckler calves and heifers, respectively, 18% for dairy milk production and 12% for enteric methane emissions. By comparing this new version of the PaSim model to the previous version, we show that a greater accuracy in animal performance modelling improves the sensitivity of the model to interannual climate variability. However, long term (30 years) projections of climate change impacts on grasslands and of radiative feedbacks to the atmosphere are not significantly modified. The originality and the validity domain of the model are discussed.