A management framework for phosphorus use on agricultural soils using sorption criteria and soil test P

The variation in sorption and desorption of phosphorus (P) among soil types is not captured in current agronomic advice for agri-environmentally sustainable use of P. Phosphorus use is typically based on soil test P (STP) and soils are assumed to have the same rate of response to added P, regardless of sorption properties. The development of P sorption categories, coupled with STP information could improve fertiliser decision making, by making it more site specific and soil type specific. A framework for P sorption specific advice is proposed here integrating soil P sorption dynamics with STP for agronomic and environmental management, at farm and catchment scale. Using a national population of agricultural soils, laboratory measurement of Langmuir sorption maximum (Smax₅₀, mg kg^-1) and binding energy (k₅₀, l mg^-1), were coupled with STP (Morgan P) to derive novel categories for P management advice, specifically addressing the build-up and draw-down phases of P in soils. In addition to measured values, Smax₅₀ and k₅₀ were predicted from MIR spectroscopy and pedotransfer functions and used to allocate soils into these new sorption categories. The allocation of soils into a P management category using predicted values indicated that pedotransfer functions offered greater reliability (90% allocation accuracy using an independent test set), however MIR spectroscopy is faster and less resource intensive (67% allocation accuracy using an independent test set). Phosphorus sorption dynamics should be interpreted alongside soil test P and P Index information so that water quality policy can consider the difference between high and very high STP soils based on sorption information. In the absence of laboratory data on P sorption, soils can be classified into P management classes using predicted values from spectroscopy (rapid and cheap) or pedotransfer functions (greater reliability). Further development of the MIR methodology is recommended along with field validation.