Abstract
Based upon data from five years of field research on yield response by ‘Valencia’ orange (Citrus sinensis) to N, P, and K application, the diagnostic utility of sufficiency ranges, DRIS, DRIS using log‐transformed ratio data, log‐transformed concentrations, and concentration data interpreted similarly to the Kenworthy approach are compared. The accuracy of the sufficiency range approach was 75% for N diagnoses and 90% for P diagnoses, compared to 50% or less for other methods due to high incidence of false positive (F+) diagnoses. All methods showed 60% accuracy for K diagnoses, with 40% F+ diagnoses. Total net yield effects explained by sufficiency ranges were greater than 120 kg/tree for each of N and P due mainly to true negative diagnoses. Other diagnostic methods accounted for net increases of less than 30 kg/tree or net yield decreases as great as 80 kg/tree due to yield losses associated with F+ diagnoses. All methods accounted for nearly 500 kg/tree net yield increases attributable to K application. Under these conditions, the more conservative sufficiency range approach avoided F+ diagnoses, resulting in greater accuracy and yield effects than DRIS and its derivatives.