Assessing nitrogen fluxes from dairy farms using a modelling approach: A case study in the Moe River catchment, Victoria, Australia

Assessment of nitrogen (N) loss forms and pathways from farming systems is important for improved understanding of potential off-farm impacts on high value environmental assets. The objective of this study was to estimate N losses in different pathways in dairy systems across the range of climate, soil and farm management by using west Gippsland (Victoria, Australia) as case study area, and to characterise the sensitivity of the adopted model parameters. We combined the point scale models DairyMod and Howleaky to estimate dissolved N (DN) and particulate N (PN) loads in runoff, and N leaching (LN) in deep drainage from representative dairy farms in west Gippsland. Monte Carlo error propagation with Latin hypercube sampling was performed to identify sensitive model parameters and assess potential uncertainty in N load predictions. The combined model was capable of simulating climate-soil-animal-pasture management interactions and estimating DN, PN and LN at an annual scale; which were estimated at up to 18 kg-N ha−1, 15 kg-N ha−1 and 312 kg-N ha−1, respectively. The combined model demonstrated that more intensive feeding as mixed ration, and nutrient budgeting that takes into account the fertiliser equivalent of recycled nutrients can achieve an increase in milk production by up to 13% and a decrease in N loads by up to 31% compared to the intensive system in the case study catchment. Soil type and farm management explained much of the variability (up to 76%) observed in LN and DN loads, whereas climate and soil type had significant influence on PN loads (62-77%). Year-to-year variation, particularly under dry conditions had a marked influence on N loads. Soil N, vegetation cover, rooting depth and soil maximum drainage rate must be well characterised in order to reduce potentially high uncertainty in the estimation of N losses in heterogeneous catchments.