Use of handheld mid-infrared spectroscopy and partial least-squares regression for the prediction of the phosphorus buffering index in Australian soils

The development of techniques for the rapid, inexpensive and accurate determination of the phosphorus (P) buffer index (PBI) in soils is important in terms of increasing the efficiency of P application for optimum crop requirements and preventing environmental pollution due to excessive use of P fertilisers. This paper describes the successful implementation of partial least-squares regression (PLSR) from spectra obtained with bench-top and handheld mid-infrared (MIR) spectrometers for the prediction of PBI on 601 representative Australian agricultural soils. By contrast, poor predictions were obtained for available (Colwell) P. Regression models were successfully derived for PBI ranges of 0–800 and 0–150, the latter range resulting in the optimum model considering the dominance of low PBI soils in the sample set. Concentrations of some major soil minerals (mainly kaolinite and gibbsite content for high PBI, and smectites or illites for low PBI), quartz (representative of low surface area of soils) and, to a lesser extent, carbonate and soil organic matter were identified as the main drivers of the PBI models. Models developed with soils sieved to <2 mm presented an accuracy similar to those developed using fine-ground material. The accuracy of the PLSR for the prediction of PBI by using bench-top and handheld instruments was also similar. Our results confirm the possibility of using MIR spectroscopy for the onsite prediction of PBI.