Phosphorus dynamics in P-depleted sub-surface of cropping soils in northeast Australia: Evaluating the potential of APSIM for simulation and the influence of soil properties

• Degree of P saturation identified as critical factor for fertiliser P recovery. • An exponential decay model represented accurately the decrease in P extractability. • Fertiliser P type did not affect long-term P extractability. • APSIM labile-P simulation possible with a specific loss available rate coefficient. • Loss available rate coefficient related to the P buffering index of the soils. Sub-surface (10–30 cm depths) phosphorus (P) depletion in Vertisols of northern Australia necessitates deeper placement of phosphate fertilisers to sustain rain-fed crop productivity. However, predicting crop P availability in these layers is challenging due to the variability in soil properties and seasonal rainfall. This study had two main objectives: (1) to determine which soil properties influence the decrease in extractable-P after the addition of different types of fertilisers and, (2) to evaluate the performance of the Agricultural Production Systems sIMulator (APSIM) model in simulating soil extractable-P dynamics within the labile-P pool (i.e., plant-available P pool in APSIM) in P-depleted sub-surface soils following mineral P fertiliser application. In an incubation study, monoammonium phosphate and diammonium phosphate were applied at a rate of 50 mg P kg −1 to nine Vertisols and three contrasting soil types. Extractable-P was measured using Colwell-P (0.5 M NaHCO 3 , pH 8.5) and BSES-P (0.005 M H 2 SO 4 ) at intervals of 10, 30, and 90 days, and extrapolated beyond 365 days of incubation. Extractable-P levels decreased rapidly, reaching equilibrium at 30 days, following an exponential decay model. The degree of P saturation, derived from the P Buffering Index (PBI) and highly correlated with the concentration of extractable amorphous aluminium and iron oxides, was identified as the most significant factor in determining fertiliser P availability. Using Colwell-P to initialise APSIM, the simulated labile-P pool reasonably captured the actual decline of Colwell-P over time by employing a specific loss rate coefficient “r” for each soil, which was related to the soil PBI. The study indicates that the degree of P saturation is crucial in determining the reduction in extractable-P following fertiliser application in these soils. Furthermore, the P module in the APSIM model shows potential for predicting long-term (>1 year) labile-P dynamics, although further testing on diverse soils, varying fertiliser application rates, and under field conditions is necessary to optimise the model’s parameterisation and user confidence.