Predicting the fate of fertiliser sulphur in grazed hill country pastures by modelling the transfer and accumulation of soil phosphorus

A nutrient transfer model developed by Saggaret al. (1990) closely predicted the amounts of soil phosphorus (P) (0–150 mm depth) that had accumulated in the soil from fertiliser and excretal inputs in 20 pastoral farmlets grazed by sheep. The 20 farm1ets, located on hill country in the North Island of New Zealand, were grazed by sheep for 8 years (1980–87) with two grazing managements (continuous and rotational). Five rates of single superphosphate (SSP) were applied within each grazing management. The predictions of the transfer model explained 90% of the variations in soil P amounts between farmlets, and 88% of the variations between slope units across all farmlets. However, the model could predict accurately (within 10% error of prediction) the amounts of accumulated soil S in only those farmlets receiving up to 288 kg S/ha during the last 8 years. At higher levels of SSP fertilisation (–288 kg S/ha), the predicted amounts of soil S were up to 40% higher than the measured amounts of soil S. The differences in the predicted and measured soil S amounts were used to quantify S loss and then pinpoint the areas of S losses. At annual applications of 36,60, and 120 kg S/ha, 25, 38, and 49% of the fertiliser S, respectively, appeared to be lost even from these highly retentive and productive soils. The major S losses occurred from campsites (0–10°) and steep slopes (31–40°); and 41°+). Losses of fertiliser S from the farmlets in this study, which were on soils with high sulphate retention properties, were lower than those measured in a similar study (Saggar et al. 1990) using farmlets on soils having low sulphate retention properties.