Evaluation of EDTA-extractable Phosphorus as a Soil Test Procedure

Diagnostic techniques involving complex systems such as soils and plants are rarely infallable. Thus, a constant research effort is underway to improve decision making by adding information to presently used methods and/or developing new techniques. Predicting crop response to applied P requires a suitable extractant and a model that satisfactorily describes the yield response. A number of extractants and yield response models have been proposed and evaluated over the last 50 years. The objectives of the research reported here were to (i) develop a satisfactory procedure for extraction of P using one of these extractants (EDTA), (ii) evaluate this procedure using current mathematical and statistical models, and (iii) compare this procedure (by use of the models) with current procedures in order to determine if decision-making could be improved. Soil samples of four soil series from 23 location-years of irrigated fertilizer response trials with grain sorghum (Sorghum bicolor L. Moench) were extracted with ammonium acetate, sodium bicarbonate, and EDTA. Extraction time for EDTA was established utilizing soils from two locations of known P response. The extraction procedures were evaluated by the following models: linear, quadratic, logarithmic, Mitscherlich, and the Cate-Nelson analysis of variance method. For the soils used, an extraction time for EDTA of 1.0 hour was established as a reasonable compromise between ability to distinguish with a high degree of certainty between soils on a P needs basis and an extraction time suitable for a routine analysis procedure. The Cate-Nelson ANOVM (Analysis Variance Method) for partitioning soils for expected P response was found to be superior to the other models tested for all three extractants. For five of six models tested, the EDTA-extraction procedure resulted in better predictability of P fertilizer needs than either the ammonium acetate or sodium bicarbonate extraction procedures. The combination of the EDTA-extraction procedure and the Cate-Nelson three-class model resulted in the highest R² value (0.71) of all combinations of models and extractants evaluated.