Prediction of phosphorus sorption indices and isotherm parameters in agricultural soils using mid-infrared spectroscopy

•Phosphorus sorption parameters predicted using MIR to rough screening standard.•P sorption indices were better predicted than isotherm derived parameters using MIR.•Langmuir isotherm parameters predicted to rough screening standard for soil.•Raw spectra and reference data outlier removal was sufficient for model optimisation. Phosphorus is a macro nutrient essential for optimum crop growth and animal health. The soil’s ability to supply P in an available form is influenced by sorption capacity and P binding energies in soil. These properties are usually derived from sorption isotherms that are time consuming and difficult for routine analysis. Mid-infrared diffuse reflectance Fourier transform (MIR DRIFT) spectroscopy is a rapid analysis technique that can potentially replace extractive and digestive techniques traditionally used in soil analysis. This study explored the application of MIR DRIFT in combination with chemometrics, to predict indicators of soil fertility and quality, specifically, P sorption properties. Using an archive of 11 great soil groups a P sorption reference library was generated using five different sorption models; (1) single point sorption index; (2) Langmuir sorption isotherm in the 0–25 mg l−1 P range; (3) Langmuir sorption isotherm in the 0–50 mg l−1 P range; (4) Freundlich sorption isotherm in the 0–25 mg l−1 P range; and (5) Freundlich sorption isotherm in the 0–50 mg l−1 P range. Thirteen reference values were generated for each sample for calibration and validation of P sorption models developed from MIR DRIFT spectra. Validation of the single point P sorption index and Langmuir parameters were satisfactory for rough screening with the exception of the Langmuir binding energy in the 0–50 mg l−1 P range (k50 [8]). The P sorption capacity remaining was the best predicted parameter and the Freundlich sorption isotherm predictions were poor. The results indicated that there is potential for benchtop MIR to describe P sorption properties in agricultural soil to improve management decisions and that soil specific models could be developed to further enhance prediction performance.