Predicting phosphorus sorption isotherm parameters in soil using data of routine laboratory tests

Knowledge of phosphorus (P) sorption dynamics across different soil types could direct agronomic and environmental management of P. The objective of this study was to predict P isotherm parameters for a national soil population using data of routine laboratory tests. Langmuir and Freundlich sorption parameters were calculated from two different ranges (0–25 and 0–50 mg P L^?1) using an archive of representative agricultural soil types from Ireland. Multiple linear regression (MLR) identified labile forms of aluminium (Al) and iron (Fe), organic matter (OM), cation exchange capacity (CEC), and clay as significant drivers. Langmuir and Freundlich sorption capacities, Freundlich affinity constant, and Langmuir buffer capacity were predicted reliably, with R² of independent validation > 0.9. Sorption isotherm parameters were predicted from P sorbed at a single concentration of 50 mg P L^?1 (S₅₀). An MLR prediction of P sorption maximum in the 0–50 mg P L^?1 range was achieved, to an accurate standard, using S₅₀, OM, and Mehlich-3 Fe (R² of independent calibration and validation being 0.91 and 0.95, respectively). Using Giles’ four shapes of isotherms (C, L, H, and S), L non-strict- and C-shaped isotherm curves accounted for 64% and 27% of the soils, respectively. Hierarchical clustering identified a separation of isotherm curves influenced by two ranges of Mehlich-3 Al. Soils with a low range of Mehlich-3 Al (2.5–698 mg kg^?1) had no incidence of rapid sorption (C shape). Single point indices, Al, or available soil data make the regression approach a feasible way of predicting Langmuir parameters that could be included with standard agronomic soil P testing.