DRIS diagnoses of soybean nitrogen, phosphorus, and potassium status are unsatisfactory

Abstract

Previous research has shown that the Diagnosis and Recommendation Integrated System (DRIS) is useful to identify the nutrient most likely to limit yield of soybean (Glycine max). However, recent work with other crops has shown that DRIS diagnoses are sometimes unsatisfactory due to large numbers of false positve (F+) diagnoses, which would lead to recommendation of unnecessary fertilizer application. This paper reports a reconsideration using the prescient diagnostic analysis approach of data previously used to establish the diagnostic utility of DRIS for soybean. In addition to previously‐published evaluation criteria, the approach is extended by defining an efficiency rating which accounts for differences in the incidence of sufficient and deficient cases. Using this evaluation approach, DRIS phosphorus (P) and potassium (K) diagnoses of soybean using numerous sources of norms and methods to calculate function values and nutrient indices led to unacceptable efficiency ratings (<67%), even though accuracies often exceeded 90%. In one test, diagnoses of nitrogen (N), P, and K by both DRIS and sufficiency ranges were low in accuracy and efficiency ratings. In several other tests, poor ratings for P diagnoses often followed from low accuracy among deficient cases; i.e. false negative (F‐) diagnoses were excessive. In contrast, K diagnoses were often poor due to low accuracy among cases of sufficiency: false positive (F+) diagnoses were excessive. Despite much modification, DRIS diagnoses of soybean N, P, and K status among data sets of known response to fertilizer application are not sufficiently reliable to support routine adoption of DRIS for diagnosis.