Fertiliser history is a useful predictor of soil fertility status

A Fertiliser index is described which consists of summing total superphosphate applied to a site after using an annual discount to past applications. This index is related to relative pasture yield for 3 major New Zealand super-phosphate withholding trials using a Mitscherlich function. A discount factor of 0.15 was chosen. Accuracy of relative pasture yield prediction was at least as good as for an Olsen P predictor and better where sulphur responses were implicated in the response to super-phosphate (r-$0.77 to 0.93). The effect on relative pasture yield predictions of measurement errors in Olsen P and Fertiliser index were similar using the Whatawhata site for comparison. A 20% error in predictor resulted in a 3 unit error in relative pasture yield estimation. Measurement errors in the Fertiliser index are minimal if the last 10 years' fertiliser history are known. Where this knowledge is not available, 5 years of known fertiliser history and use of national average fertiliser statistics beyond that may give acceptable estimates of relative yield. The advantage of the model for onfarm use is that intended fertiliser use directly affects the index and hence prediction of relative pasture yield. This facilitates economic analysis of options. An analysis using the model to investigate production consequences of fertiliser use at the industrylevelshowscontinuationofcurrentfertiliser levels on hill country farms will result in a drop in farm output of 13% in the next 10 years. Keywords super-phosphate, discount, Olsen P, policy, regional production, pasture yield, fertiliser strategy