Performance of a fertiliser management algorithm to balance yield and nitrogen losses in dairy systems

To demonstrate the use of a previously developed fertilisation algorithm and to determine its potential effects on nitrogen (N) losses from grazed pastoral systems, a simulation study was performed using the Agricultural Production Systems Simulator (APSIM). The study considered a dairy system with irrigated ryegrass pasture on a silt loam soil in the Canterbury region of New Zealand. Firstly, the algorithm was parameterised for each month based on pasture yield and N contents from simulation run over 20 years using a wide range of N fertilisation rates. The algorithm was then used in the simulation of fertilisation management of a hypothetical dairy farm under different scenarios where its performance for increasing pasture yield with more efficient N use was tested. The scenarios were based on different yield targets for the proposed algorithm (50, 75, 90 or 100% of the average maximum yield) and included scheduled fertilisation to mimic more typical management. For more realistic evaluation, the simulations took into account changes in stocking rates and N flows in the farm resulting from the different fertiliser management. The simulations also considered the uneven return of urinary N by grazing animals, which are crucial to determine N losses in these systems. Both pasture yield and N losses were in general agreement with available measured data from similar systems and with comparable N inputs. Thus providing support for the simulation study as a valid way to demonstrate the potential effects of changing fertiliser management. The average of simulations run over 10 years showed that direct losses from the fertiliser were lower when the fertilisation was controlled by the proposed algorithm compared with scheduled fertilisation at similar N rates. However, with animals in the paddock and thus including the effects of urine patches, N losses were not significantly different. As there was an increase in pasture yield and consequent stocking numbers, the area receiving urinary N increased, counter balancing the increased N use efficiency when using the algorithm. Nonetheless, the larger yield lead to greater farm productivity, and this resulted in about 13% reduction in N losses per unit of milk production.