Run-off transport of herbicides during natural and simulated rainfall and its reduction by vegetated filter strips

The efficiency of filter strips in protecting watercourses against herbicides in run‐off was evaluated in field experiments in western Germany. Surface run‐off caused by natural rainfall and related transport of metolachlor, terbuthylazine and pendimethalin were measured on plots of 40 m length without filter strips (F0), and after passing over three types of herbicide‐untreated field margin: 12 m conservation headland (CH12), 6 m (GF6) and 12 m grass strips (GF12). Run‐off was also measured after simulated rainfall on 7 m long plots without (F0) and with 3 m grass strips (GF3). All three herbicides were transported both in dissolved and in adsorbed forms; the partitioning depended on their water solubility with metolachlor and terbuthylazine mainly translocated in dissolved form (F0: highest mean concentrations for a natural run‐off event 721 and 220 μg L^?1, respectively). Pendimethalin was predominantly transported in adsorbed form (maximum mean concentration 11.2 μg L^?1). In the sediment, the highest mean herbicide contents in a single natural event (F0) accounted for 2294 μg kg^?1 (metolachlor), 1317 μg kg^?1 (terbuthylazine) and 5648 μg kg^?1 (pendimethalin). The proportions of applied herbicide translocated were 0.3% (metolachlor), 0.2% (terbuthylazine) and 0.06% (pendimethalin; F0, natural rainfall). The extent of herbicide transport decreased with time but within this trend soil sealing, soil moisture and amount and intensity of rainfall increased losses. Compared with the F0 plots, the reduction of herbicide translocation after natural rainfall reached 80–83% (CH12), 80–88% (GF6) and >99% (GF12) over the 3‐year period. The 12 m grass strips allowed only one extreme run‐off event to pass through, thus providing a highly effective watercourse protection against herbicide pollution.