Comparison of soil test extractants used in mid-Atlantic United States

Abstract

A study was conducted to develop conversion equations between the five soil test extractants commonly used by public and private soil testing laboratories serving clients in the Mid-Atlantic United States. Three hundred soil samples were selected from among the samples submitted to the University of Delaware Soil Testing Program for routine fertility analysis. The samples were extracted with Mehlich 1, Mehlich 3, Bray P1, 1 N pH 7.0 ammonium acetate and 0.1 N HCl and analyzed for phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), manganese (Mn), zinc (Zn), copper (Cu), and iron (Fe), as appropriate to the method. Significant linear relationships suitable for use as conversion equations were obtained between all extractants across a wide range of soil concentrations. Analysis of the 95% confidence bounds around the predictive equations showed large confidence intervals for some nutrient–extractant combinations (e.g., P, Mn, and Fe). Separation of the data into sub-populations based upon nutrient concentration (e.g., Mehlich 3 P<or>200 mg P dm− 3) and subsequent statistical analysis reduced the width of the confidence interval for some, but not all, relationships. For nutrient–extractant combinations where multiple relationships have been produced for specific concentration ranges, selection of the appropriate equation should be made based upon the intended application of the resulting data and the relevant concentration range (e.g., development of agronomic nutrient recommendations would require accuracy at “low P” concentrations versus limiting P application might require accuracy at “high” P concentrations). Statistical analysis of the data set using regression analysis with the “no intercept” option produced simple conversion factors that could also be used to rapidly and easily convert the results obtained with one soil test to those of another extractant. These “simple conversion factors” had slopes and confidence intervals that were comparable to those produced using standard regression analysis. The absence of an intercept component in the resulting relationship, however, would make these conversion factors more simplistic and less cumbersome for clients to use when comparing soil test results.