Cropping system diversification does not always beget weed diversity

International audienceCropping system (CS) diversification appears as a promising solution to increase CS sustainability. However, weed community response to different options of CS diversification remains poorly documented. Moreover, these effects are expected to be more pronounced in experimental than commercial farms because experimental farms explore more diverse combinations of farming practices. We hypothesized that (i) CS diversification would increase weed diversity at multiple spatio-temporal scales but that (ii) different options of CS diversification would select different weed communities and that (iii) responses could differ between experimental and commercial farms. Hence, weed density per species was measured over a 6-year time period in a CS experiment and in a farmers’ network (both resorting to diverse CSs that were numerically summarized to allow their comparison, i.e. different positions along gradients of tillage intensity, herbicide use, crop rotation length etc.). Weed density measures were used to compute weed diversity indices (taxonomic and functional, at annual and plurennial scales) and community weighted means on key response traits for each CS. All experimented alternative CSs (diversified crop sequences with coherent but different combinations of weed management tools) showed that diverse combinations of agronomic tools are available to increase weed diversity, as highlighted by a 3 and 2-fold increase in species richness at the annual and plurennial scales, respectively. In contrast, only one farmer CS (3-year rotation, low tillage intensity, intermediate herbicide reliance) showed significantly higher levels of weed diversity, possibly because the reduced tillage intensity was not compensated by other agronomic levers (e.g. increase of herbicide use and/or crop rotation diversity). Such outcomes were attributed to (i) reduced CS complexity in commercial compared to experimental farms and (ii) high herbicide reliance in commercial farms, irrespectively of CS complexity. Across both experimental and commercial farms, tillage, weed management and crop type appeared as the main factors structuring weed communities. Systems with reduced tillage were associated with a higher percentage of grasses and perennials. Systems with spring/summer crops and/or mechanical weeding were associated with a higher proportion of spring/summer and perennial species. These results suggest that solutions are readily available for farmers to implement sustainable weed management, but supports are required to address the factors hindering the adoption of these experimented CS in commercial conditions