Relations between phosphorus retention parameters and some other properties of benchmark soils in Lesotho

Abstract

The benchmark soils of Lesotho are characterized by high P retention and this aspect should be accommodated in fertilization programmes. This together with long procedures for determination of P retention parameters and a lack of information on the relations between them and other soil properties motivated the study. The objective was to establish which properties influence P retention by the eleven benchmark soils of Lesotho and subsequently develop regression models for predicting P retention parameters from those properties that are determined in routine laboratory analyses. In an incubation study two P retention parameters were quantified with the extraction procedures of Bray and Olsen: (i) phosphorus retention index (PRI), which was the slope of retained P plotted against applied P and (ii) phosphorus retention capacity at an application level of 400 kg P ha–1 (PRC400). Simple linear and non–linear regression models showed regardless of extraction procedure significant correlations (r2>0.60) between either PRI or PRC400 and the following soil properties: sample density, CEC, sand, clay, organic C, crystalline Fe oxides and amorphous Fe and Al oxides. When the data were fitted into a stepwise multiple regression analysis, the effect of organic C and CEC on P retention became insignificant while that of available P and exchangeable K became significant. The r2 values for the multiple regression models were in the order of 0.93–0.94 for Bray or 0.96 for Olsen. Regression models excluding Fe and Al were also established to cater for situations where it may not be possible to determine those oxides and their r2 values were in the order of 0.92–0.93 for Bray or 0.92 for Olsen. Hence any of these multiple regression models can be used to predict P retention in the benchmark soils of Lesotho.