Modelling herbicide distribution through soil profiles under conventional tillage and non-tillage

International audience

English. Conservation agricultural practices based on mulching and herbicides use are becoming more widespread. However, they cannot be simulated with pesticide fate models used for risk assessment before placing pesticides on the European market. Thus, this work sought to assess and compare the performance of PRZM and MACRO models to replicate the moisture content and vertical distribution of S-metolachlor, foramsulfuron, and thiencarbazone-methyl herbicides in two agricultural soils under conventional tillage (CT) and non-tillage (NT) over a two-year field experiment. A new approach was developed to consider the impact of the mulch on soil moisture and on the amount of herbicides reaching the NT topsoil. This involved removing the presence of the mulch layer while maintaining its impact on potential evapotranspiration and herbicide adsorption and degradation through the meteorological file and the herbicide application inputs, respectively. In addition, the Kf of S-metolachlor was calibrated for an acceptable agreement between predictions and observations. MACRO outperformed PRZM in simulating soil water content and herbicide mobility, and both models performed better under CT compared to NT. Model performances decreased the second year, especially in NT soils, because herbicide DT50 were parameterized with values measured the first year, which were less representative of the second year. This new modelling approach is appropriate for simulating moisture and herbicide transport in NT soils with MACRO on a bi-annual scale. However, to assess the fate of herbicides in NT, models require a module allowing change in DT50 according to the soil surface covered annually by the mulch.