Statistical Aspects of Soil and Plant Test Measurement and Calibration in Australasia

Abstract

This article outlines the contemporary analytical performance of soil and plant analyses in Australasia. These derive from interlaboratory proficiency programs conducted by the Australasian Soil and Plant Analysis Council (ASPAC) in 2001/2002 and earlier. Contemporaneously, the best‐performed soil tests were pH (mean robust CVs around 2.2%), whereas soil methods with robust median CVs in excess of 20% included exchangeable Na, Bray‐P, water‐soluble Cl, hot CaCl2‐extractable B and KCl‐exchangeable Al. The plant tests with the lowest CVs were total C, N, and P; the worst were total Na, Al, Co, Mo, Cd, Pb, Si, and Se. For both soil and plant tests, improvements in analytical performance are less than expected, which may be due in part to some of ASPAC's interlaboratory proficiency program policies. There is brief comment on strategies that may improve analytical performance, including discussion of social perspectives that include the use of “rewards.” Analytical performance, however, is only one factor influencing outcomes from testing. Others relate to the derivation of critical values or ranges, and relationships between fertilizer rates and yields. This article draws attention to the influence statistics has on these. Relevant attributes of nonlinear models and other alternatives are discussed, with preference given to choices favoring low nutrient inputs to lessen the likelihood of off‐farm movement of nutrients to water. The consistently poor analytical performance of Bray‐P calls into question its diagnostic value, whereas there is a need in Australia to enhance the measurement quality of chemical indicators of soil salinity, particularly water‐soluble Cl by potentiometric titration.