Reduced herbicide use does not increase crop yield loss if it is compensated by alternative preventive and curative measures

Herbicide use must be reduced because of environmental and health issues. This raises the question of whether weeds and the resulting crop yield loss will increase. Previous studies analysing relationships between herbicide use intensity, weeds and yield loss suffer from methodological shortcomings in terms of weed flora and farm diversity as well as temporal scales. Here, we collected data on 272 arable cropping systems from one Spanish and six French regions, from farm surveys, the Biovigilance-Flore network, expert opinion, cropping system trials, crop advisors and scientists. Each system was simulated over 27 years and with 10 weather repetitions consisting of 28 randomly chosen weather years, using the virtual-field model FloxSys. This process-based model simulates multi-species weed floras and crop canopies as a function of cropping systems and pedoclimate at a daily time-step over the years. Four series of simulations were run, 1) starting with a typical regional weed flora, 2) eliminating all herbicides without any other change in management practices. The two series were run again, but without an initial weed seed bank. Comparing series 1 and 2 to respectively 3 and 4 led to calculating a crop yield loss due to weeds in series 1 and 2. Comparing series 1 and 2 quantified the herbicide impact on weeds, crop production and yield loss. The simulations showed that (1) crop yield loss increased with increasing weed biomass, and that the weed/crop biomass ratio at crop flowering was the best indicator of the year's yield loss, (2) herbicide use intensity was not correlated to either weed variables or yield loss, because herbicide use intensity greatly depended on other management practices; e.g., it decreased with increasing frequency and interannual variation of mechanical weeding and superficial tillage, (3) weed biomass and yield loss increased when herbicides were eliminated without any other change in management practices, (4) effects were more visible at the multi-annual than the annual scales. The systems the most sensitive to herbicide suppression were characterized by monotonous rotations with short crop cover, high herbicide use, no plough or winter ploughing and frequent rolling operations. Finally, a decision tree predicting yield loss as a function of management practices was proposed to support farmers and crop advisors when designing innovative cropping systems reconciling low herbicide use and low yield loss.