Reference populations for evaluation of the nutritional status of coffee by DRIS

Abstract

The Diagnosis and Recommendation Integrated System (DRIS) was tested on Coffea arabica with the objective of defining an efficient reference population. The data bank contained results on leaf samples taken from about 800 production fields during the summer in the states of Sao Paulo, Parana, and Minas Gerais, Brazil. From the original population, a second one, called normal, was defined by excluding boron, zinc, and copper data when these elements were considered as contaminants. Three populations were selected and tested as references: (a) the original medium yield, (b) the normal high yield, and (c) the normal medium yield with balanced nutrient concentrations. DRIS indices for macro- and micronutrients from 15 selected samples were tested by comparison with the sufficiency range approach. For the original population there were a low number of correct diagnoses indicating the need to exclude contaminated samples. Use of the high yield population resulted in a poor assessment of N and K nutrition due to nutrient accumulation or removal in years of low and high yields, respectively. The high percentage of correct diagnostics with the normal medium yield nutrient equilibrated population suggested it was the best option to calculate DRIS indices for coffee.