Influential sources of uncertainty in glyphosate biochemical degradation in soil

Reactive transport models are important numerical tools to support decision making in many fields, such as herbicide use regulation. Though, models may be affected by multiple sources of uncertainty. Therefore, uncertainty and sensitivity analyses should become the practice to assess the confidence of such models. Here, the uncertainty in steady-state concentrations of glyphosate (GLP) and its metabolite aminomethylphosphonic acid (AMPA) was assessed using a reaction network that accounts for GLP and AMPA biotic and abiotic degradation pathways in soil including biological oxidation or hydrolysis in aerobic conditions via metabolic or cometabolic reactions. The mathematical framework is based on Michealis–Menten–Monod kinetic equations, which allow to account for microbial strategies to biodegrade contaminants. Chemical oxidation is assumed to occur independently from environmental conditions and resulted in a reduction of GLP concentration up to 15% when it was accounted for. The wide spectrum of interconnected catabolic reactions, each occurring at a different rate, as well as uncertainties in kinetic parameters estimation, suggest variability in modelling outcomes, which were addressed by means of a sensitivity analysis. In particular, the tested reaction network was mainly driven by GLP oxidation to AMPA; increasing the corresponding rate constant or decreasing the half-saturation constant resulted in a substantial decrease of GLP concentration but to an increase in AMPA concentration. Identification of the conditions responsible for GLP degradation to non-toxic metabolites, as well as for AMPA production and degradation, can allow to forecast unexpected consequences of GLP use and to design optimal land management and bioremediation plans.