The application of spectroscopic techniques for the prediction of phosphorus dynamics in agricultural soils

Sustainable management of soil phosphorus (P) is important, because fertiliser supply is limited to the mining of finite phosphate rock and over-application of P causes harm to the environment. Currently in Ireland, only one parameter (Morgan's P (mg l-1)) is used to describe the supply of available P for crop uptake. Knowledge about P dynamics would allow an advance on current agronomic advice that relies on a quantification of current state to define what action a farmer should take. Dynamic P parameters (e.g. Langmuir sorption maximum, binding energy and maximum buffer capacity) are usually derived from sorption isotherms that are too time-consuming for routine analysis. Infrared (IR) spectroscopy offers a rapid analysis technique that can potentially replace some extractive and digestive techniques. The aim of this study was to develop a framework for incorporating soil P dynamics, quantified using rapid, low cost methods, into Irish agronomic advice to optimise productivity while supporting water quality policy. This was addressed by considering the accuracy of IR spectroscopic predictions of parameters that describe P sorption in soil, the accuracy of pedotransfer functions to predict P isotherm parameters, by identifying different mechanisms driving P isotherm properties and by defining a conceptual framework of categorisation for soils based on their P sorption parameters coupled with STP data. First horizon subsamples (n = 225) were taken from the archive of the Soil Information System (SIS) Ireland (Creamer et al., 2016) at Johnstown Castle, Wexford. Phosphorus sorption Index (PSI, R2c = 0.63, R2v = 0.65), phosphorus sorption capacity remaining (PSCr, R2c = 0.77, R2v = 0.67), phosphorus sorption capacity total (PSCt, R2c = 0.75, R2v = 0.60), Langmuir isotherm parameters in the 0 – 25 mg P l-1 range of added P (Smax25, R2c = 0.70, R2v = 0.60; k25, R2c = 0.66, R2v = 0.62; MBC25, R2c = 0.66, R2v = 0.60) and Langmuir isotherm parameters in the 0 – 50 mg P l-1 range of added P (Smax50, R2c = 0.75, R2v = 0.67; k50, R2c = 0.79, R2v = 0.47; MBC50, R2c = 0.74, R2v = 0.53) were predicted using MIR to a standard suitable for rough screening of agricultural soils. Langmuir sorption maximum, Smax50, was reliably predicted using multiple linear regression (MLR) with S50, OM and Mehlich-3 Fe (R2c = 0.91 and R2v = 0.95). The chemical data were also used to better understand the different isotherm shapes identified in the Irish soil population, where 64 % had Gile's non-strict L shape isotherm mechanism and 27 % were classified as having C shape isotherms. A conceptual framework based on P sorption parameters that incorporated STP was defined for agricultural soils. It was recommended that current Index classes be re- evaluated to consider having a target value (as Index 3) and to split those requiring reduction, or drawn-down, into a class near the target and a class that is far from the target. This split would contribute to a tool for critical source area (CSA) identification for P management. It was found that soils could be reliably classified into the new framework using values predicted from both MIR spectroscopy and pedotransfer functions. Optimisation of the framework for P sorption specific management of agricultural soils will require two additional steps: (1) further development of the MIR methodology for field use and (2) field validation of the thresholds and interpretation proposed.