Soil test is the most widely used tool to determine the soil fertility status and fertilizer demand for crop production. Tables of soil analysis interpretations and fertilizer recommendation based on field experiments are required for this purpose. A recommendation system normally aims at supplying crop requirements and improving soil fertility levels to a sufficient degree. To obtain information for building K interpretation and recommendation tables, seven experiments were carried out with wheat, corn and soybean in four Paraguayan Departments: two in Misiones, two in Itapua, two in Alto Paraná and one in the Department of Amanbay. The experimental design was random blocks with three replications. Treatments consisted of five K2O levels (0, 25, 50, 75, and 100 kg ha-1) broadcast on the soil surface before sowing. N and P were fertilized at constant and satisfactory rates to supply crop requirements. For wheat, corn and soybean 100 kg ha-1 P2O5 was applied; for wheat 60 kg ha-1 N and for corn 180 kg ha-1 N. Exchangeable potassium was determined in the soil by the Mehlich-1 method as well as the wheat, corn and soybean grain yield. The relative yield for each crop and site was calculated. The Mitscherlich equation was used to establish a calibration curve that describes soil K values determined by the Mehlich-1 method and the relative yield. The critical level was defined as the soil nutrient level sufficient to reach 90 % of maximum yield. The average critical K level in the soil determined by Mehlich-1 for wheat, corn and soybean under no-tillage system in Paraguay was 75 mg dm-3. To increase the K level by one mg dm-3 in the soil, 5 kg ha-1 K2O is needed. In soils with K content above 150 mg dm-3 the plant response to potassium fertilization was low.
Mehlich-1 calibration; K critical level; fertility classes; K recommendation; no-till sowing
