Advancing the EPIC model for glyphosate and AMPA fate prediction across different agroecosystems

Biogeochemical models are valuable tools for simulating agroecosystems, yet reproducing the dynamics of plant protection products (PPPs) under varying environmental conditions remains challenging. This study sought to improve predictions of the environmental fate of glyphosate (GLY) and its metabolite aminomethylphosphonic acid (AMPA) with a newly implemented version of the EPIC model, compared with the original, using data from lysimeter and field-scale experiments conducted at three sites in the Veneto region, northeastern Italy. Key model modifications incorporated degradation and wash-off rates based on weather conditions, a simplified preferential flow representation, the Freundlich coefficient, and AMPA-derived fate from GLY dynamics. Experimental observations were reproduced in terms of magnitude of pore-water concentrations and their temporal dynamics in soil profile. Results indicated that the implementation enhanced the prediction potential of the fate of GLY and AMPA, with dissipation dynamics that were well aligned with experimental data across sites (RMSE = 0.12–0.98 mg kg-1 for GLY and 0.11–1.23 mg kg-1 for AMPA) despite the model calibration being minimal. The new preferential flow component provided reliable prediction of the rapid GLY movement to deep layers, overcoming the limitation of traditional biogeochemical models in describing the vertical movement of highly adsorptive compounds. Notably, GLY levels in groundwater remained below 1% of applied mass, consistent with observed data. Despite the simplified representation of PPP fate, results support model transferability across agroecosystems with minimal parameter requirements and provide a valuable tool for farmers and decision makers to apply targeted and evidence-based mitigation strategies.