Yield responsiveness and response curvature as essential criteria for the evaluation and calibration of soil phosphate tests for wheat

Abstract

In a study using data from 57 wheat field experiments on the central-western slopes of New South Wales, eight soil phosphate tests (Bray,, Bray,, alkaline fluoride, Mehlich, Truog, lactate, Olsen and Colwell) were evaluated and calibrated in terms of responsiveness (â) and response curvature (C) parameters derived from the Mitscherlich equation. The results showed that, regardless of how well correlated a soil test is with yield responsiveness, it cannot give a satisfactory estimate of fertilizer requirement unless yield response curvature is also taken into account. The tendency of soil test values, especially of the Colwell test, to be negatively related to response curvature, and hence inversely related to fertilizer effectiveness, compounded the problem of directly relating soil test values to fertilizer requirement. The best test (lactate) accounted for only 28% of the variance in fertilizer requirement, compared with 50% of the variance in responsiveness, and the worst test (Colwell) was completely unrelated to fertilizer requirements. When fertilizer requirement was estimated from the lactate test value and the actual response curvature for each experiment, 68% of the variance (from the actual fertilizer requirement) was accounted for. Thirteen experiments were subject to drier conditions than the others, and these were less responsive and had lower fertilizer requirements relative to soil test values. In relation to yield responsiveness, the Colwell test was most sensitive (P < 0.001) to dry conditions, while the two best tests (lactate and Bray,) were the least sensitive (P > 0.05). The results demonstrated the superiority of acidic anionic extractants over alkaline bicarbonate extractants on moderately acid to alkaline wheat-growing soils.