Linking Measurable Phosphorus Pools With Simulations of Soil P Dynamics: Results for the Long-Term Experiment ‘Rostock’

Abstract

Phosphorus (P) is removed from agroecosystems through harvesting, and sustainable management must include P fertilization as P availability affects crop performance. However, accurate assessment of plant-available P is challenging. In this study, two promising approaches are combined to assess the plant-available P of a 22-year long-term experiment (LTE) near Rostock, Germany. We hypothesize agreement between a modern P test method and process-based model estimates of plant-available P. The diffusive gradients in thin films (DGT) technique offers an accurate P test method because it mimics the diffusion and desorption of soil P in the presence of root uptake. This was applied in a synergetic combination with a state-of-the-art agroecosystem model that was extended with a P cycling module. The simulations and yearly DGT-P analyses comprise 4 treatments: no P fertilization, mineral P fertilization with triple-superphosphate, organic P fertilization with compost, and mineral plus organic P fertilization. Soils at 0–30 cm depth were sampled in four replicates on a yearly basis between 1999 and 2021. In addition, a P fractionation was applied for 2015 using the Hedley approach, which made it possible to link non-plant-available, steady P fractions with the respective model pools. The comparison between DGT-P determined plant-available P up to a depth of 30 cm and that estimated from the pools of the agroecosystem model AgroC showed agreement with respect to the differences between the treatments and with respect to the temporal evolution (R² between 0.65 and 0.7). Less agreement was detected for DGT-P and the respective model pools in deeper soil. A closer match over soil depth was found between grouped Hedley P fractions and AgroC model pools. Both, model and DGT-P analyses indicate that a new plant-available P equilibrium will be established under the new P management after about 12 years for the Rostock site, which points to the resilience of P cycling in agroecosystems. We conclude that the combined application of DGT-P analysis and agroecosystem modeling offers a robust and accurate quantification of plant-available P in the plough layer and can be used to create an agricultural digital twin with respect to soil P availability and its impact on crop yield.