Building an expert system to advise on the use of reactive phosphate rock on Australian pastures

Abstract

Summary. An expert system has been developed, using the results from the National Reactive Phosphate Rock Project, to determine whether reactive phosphate rock is likely to be an effective substitute for water-soluble superphosphate fertiliser for a given pasture environment. The evaluation is made from site information [annual rainfall, pasture composition and the likelihood of phosphorus (P) leaching], and soil information (pH, Colwell P, soil colour and field texture). The expert system can determine the effectiveness of both highly reactive and moderately reactive phosphate rocks. Observed substitution values of triple superphosphate for the highly reactive North Carolina phosphate rock (ratio of the respective P levels required to produce 50% of the maximum observed yield response to triple superphosphate) were closely related to values predicted by the expert system (r = 0.92); the relationship between observed and predicted substitution values of single superphosphate for the moderately reactive Hemrawein phosphate rock was also close (r= 0.86). The expert system gives one of 4 different recommendations for reactive phosphate rock based on the magnitude of the predicted substitution values. These are ‘immediately effective’, ‘effective in the medium term’, ‘marginally effective’, and ‘not effective’. The system was validated using the results from independent field experiments that provided measures of the effectiveness of reactive phosphate rock at different pasture sites.