A trait-based approach to explain weed species response to agricultural practices in a simulation study with a cropping system model

• RLQ and fourth-corner analyses linked weed species traits to selection pressures.
• Mouldboard ploughing selected thick-coated seeds.
• Long rotations and early sowing selected flattened/elongated seeds.
• Pre-sowing glyphosate selected seeds with a low area/mass ratio.
• Trait selection was explained by impacts on seed bank and emergence processes.

The objective of the present study was to use a cropping system model to evaluate the potential impacts of modified agricultural practices ex ante on weeds and to adapt ecological analytical methods to analyse simulation output to identify species traits selected by cropping techniques. Cropping systems typical of three French regions were determined from farm surveys and a French data base monitoring agricultural practices and weeds (“Biovigilance-Flore”). Current scenarios as well as recent and probable future changes were identified in surveys and by expert opinion were simulated in each region, using the FlorSys model and a weed flora consisting of sixteen major annual weed species. Each scenario was simulated over 24 years and repeated 10 times with randomly chosen regional weather series. The simulated weed flora was analysed with RLQ-analyses and fourth-corner analyses. These showed that longer rotations selected weed species with elongated/flattened seeds; spring crops and late-sown autumn crops favoured round seeds and species with short emergence periods; frequent mouldboard ploughing selected species with thick-coated seeds; frequent tillage favoured monocotyledons and species with a late and/or short emergence period; glyphosate before sowing in direct-drilled fields selected seeds with low area/mass ratios. The identified selection pressures could be explained by effects on life-cycle processes (e.g. thick-coated seeds survive better in the soil after burial, low area/mass ratios delay germination because of slower imbibition). The trait × practice relationships were then used to predict the behaviour of additional weed species in cropping systems.