Changing agricultural practices modifies the species and trait composition of the weed flora. A simulation study using a cropping system model

EA EcolDur CT1International audienceCropping systems change over time to adapt to socio-economical and environmental constraints and to profit from technological innovations. These changes can result in unexpected side-effects which are difficult to determine in fields. The objective of the present study was to use a cropping system model to evaluate the impacts of modified agricultural practices ex ante on weeds. The FLORSYS model quantifies the effect of crop succession, management techniques and climate on multi-specific weed dynamics over the years; it was parameterized with functional relationships predicting model parameters (e.g. pre-emergent seedling mortality) from species traits (e.g. seed mass). Cropping systems typical of three French regions were determined from farm surveys and the Biovigilance data base. These control scenarios as well as various management modifications were simulated in each region, using a weed flora consisting of sixteen major weed species. Each scenario was simulated over 27 years and repeated 10 times, by randomly choosing each year annual climate series measured in the tested region. The control and prospective scenarios were analysed for their total weed density and their species compositions, showing for instance that the weed flora was more diverse in the longer Burgundy (oilseed rape / winter wheat / winter barley, with mouldboard ploughing every three years) and Poitou-Charentes rotations (OSR / winter wheat / sunflower / winter wheat, with ploughing three years out of four) than in the Aquitaine maize monoculture. Modifying management practices modified both weed density and composition. For instance, simplifying or abandoning tillage greatly increased weed infestation. If it was moreover accompanied by earlier sowing, infestations increased even more, but only when autumnal weed species dominated. Introducing temporary crops before spring crops (as required by recent French regulations) reduced densities, particularly in autumnal weeds. Simplifying the three-year Burgundy rotation to a two-year oilseed rape/wheat increased weed densities, particularly when mouldboard ploughing was also abandoned. Conversely, diversifying rotations decreased densities in Burgundy (addition of a spring pea) and, particularly, in Aquitaine (maize monoculture changed to a four-year rotation). Abandoning ploughing though still increased densities. Some cultural changes (e.g. no-till) greatly changed weed flora composition. To understand these changes, the relationships between management practices and weed species traits were studied with RLQ and fourth-corner analyses. Longer rotations were thus shown to select weed species with elongated/flattened seeds (which present little dormancy) and long emergence periods. These species are able to emergence in crops irrespective of their sowing date. Conversely, rotations with late-sown crops (i.e. spring crops or late-sown autumn crops) favoured species with round seeds which are more dormant, thus reducing germination prior to crop sowing. Frequent mouldboard ploughing selected species with thick-coated seeds which are also those least prone to mortality. Ploughing also favoured species with low-lipidic seeds which germinate later and more slowly. Both consequences, low mortality and reduced germination, improve seed survival after burial by ploughing until a further ploughing moves seeds again closer to soil surface where emergence is possible. More generally, frequent tillage favoured monocotyledonous species and species with a late and short emergence period, thus limiting fatal germination triggered by early tillage. Glyphosate before sowing in direct-drilled fields selected seeds with low area/mass ratios which germinate later and can thus avoid the glyphosate application