An exploratory evaluation of APSIM to simulate growth and yield processes for winter cereals in rotation systems in South Australia

The Agricultural Production Systems Simulator (APSIM) suite of models was used to predict dynamics in water and nitrogen in soil, as well as the growth and yield of sequential crops of wheat and barley in pasture–wheat–barley rotations, between 1995 and 1997 at Roseworthy, South Australia. The NWHEAT model satisfactorily predicted above-ground dry matter, leaf area index and grain yields for both crops in rotations with either grassy (Grass) or medic (Medic) pastures, including the lack of significant response of yield to nitrogen fertiliser applied to wheat at sowing. Simulation data for soil water, from SOILWAT2, was consistent with measured data. Simulation with SOILN2, however, largely underestimated soil nitrogen, due to excessive uptake by the simulated wheat during the season when nitrogen was abundant and water supply readily available. Thus, the soil nitrate had to be reset at sowing for the following barley crop; simulated soil nitrate agreed with the measured data in this season when this nutrient was low. For most variables of crop growth and soil water, the simulated data were mostly within 2 standard errors of the measured means. Prediction of grain protein was underestimated in all cases, including where nitrogen in the shoot was overestimated. This was possibly due to inadequate remobilisation of nitrogen from the straw and roots to the grain by the simulated crop. A satisfactory prediction of dry matter, grain yield and grain weight was obtained for wheat when the models were extended to other trials at Roseworthy (Lower North), Minnipa (Upper Eyre Peninsula) and Wunkar (Murray Mallee), based on limited soil data. Long-term simulations of wheat yields showed that, with early sowing in the Lower North, median wheat yield increased by 50 kg/ha for every kilogram of nitrogen applied at sowing, up to a maximum nitrogen rate of 50 kg/ha. In the drier districts of the Upper Eyre Peninsula and the Murray Mallee, nitrogen fertiliser of no more than 25 kg/ha, applied at sowing, was enough to achieve yield benefits in any given season. At these drier sites, crop failures occurred in 5% (Upper Eyre Peninsula) and 10% (Murray Mallee) of the seasons simulated. Median sowing dates from these simulations were 15 May for the Lower North, 30 May for the Upper Eyre Peninsula and 24 May for Murray Mallee. This suggested that sowing could be conducted at least a week earlier than currently practised in the 3 environments. This study demonstrated the capability of APSIM to predict growth and grain yield of wheat and barley, as well as the associated dynamics of soil water in the main cereal belts of South Australia