Process-based mass-balance modeling of soil phosphorus availability: Testing different scenarios in a long-term maize monoculture

• A process-based modeling predicted solution P concentration in maize monoculture.
• Transfer periods = 1 year improved model simulations for balanced P application.
• Transfer periods > 1 year improved model simulations for non-limiting P application.
• P uptake below the plow layer improved simulations for limiting P fertilization.

Unraveling phosphorus cycling is an important issue for managing agricultural ecosystem sustainably. We built a process-based mass balance model that equilibrates annual P budget with plant-available soil P, assessed as the sum of phosphate ions (Pi) in solution and the time-dependent diffusive Pi (Pr) that replenishes Pi in solution. We evaluate the predictive ability of this model considering different scenarios of (a) periods of Pi diffusion at the solid-to-solution interface, and (b) contributions of the plow layer to plant nutrition. On average, the P applications were: 0 (P0), 27 (P27), and 79 (P79) kg P ha− 1 applied every year and 52 (P52/2) kg P ha− 1 applied every 2 years as triple superphosphate over 17 years (1975–1992). Climatic data, grain yields, grain P contents and annual P budgets were determined. Batch experiments were conducted in laboratory to analyze the concentration of Pi (Cp) and Pr kinetics by isotopic dilution for short periods (< 400 min). All experimental Pr values closely fitted to the following Freundlich kinetic equation: Pr = 5.72Cp0.69t0.24 (144 observations, R2 = 0.95, P < 0.001). Assuming that slow reactions lasted for one year and that the P removed by harvest was entirely derived from the plow layer, Cp simulations over 17 years accurately reflected the long-term effect of balanced P fertilization, P27 and P52/2 treatments, but not of unbalanced P fertilization, P0 and P79 treatments; root mean square deviation (RMSD) was 0.4343. The Cp simulations based on the assumption that slow reactions lasted more than a year were significantly improved for unbalanced P fertilization, P0 and P79 treatments (RMSD = 0.2976 for 3 years and RMSD = 0.2329 for 5 years). In addition, the Cp simulations based on the assumption that for P0 treatment, part of grain P was taken up from below the plow layer were significantly improved (RMSD = 0.0613 for 80% and RMSD = 0.0670 for 60% P uptake from the plow layer). The proposed model accounted for extended periods of Pi equilibration and the contribution of the plow layer to plant nutrition in this maize monoculture.