A methodology for multi-objective cropping system design based on simulations. Application to weed management

Weeds are harmful for crop production but important for biodiversity. In order to design cropping systems that reconcile crop production and biodiversity, we need tools and methods to help farmers to deal with this issue. Here, we developed a novel method for multi-objective cropping system design aimed at scientists and technical institutes, combining a cropping system database, decision trees, the ``virtual field'' model FloaSys and indicators translating simulated weed floras into scores in terms of weed harmfulness (e.g. crop yield loss, weed-borne parasite risk, field infestation), weed-mediated biodiversity (e.g. food offer for bees) and herbicide use intensity. 255 existing cropping systems were simulated with FloaSys, individual indicator values were aggregated into a multi-performance score, and decision trees were built to identify combinations of management practices and probabilities for reaching performance goals. These trees are used to identify the characteristics of existing cropping systems that must be changed to achieve the chosen performance goals, depending on the user's risk strategy. Alternative systems are built and simulated with FLORSYS to evaluate their multi-criteria performance. The method was applied to an existing oilseed rape/wheat/barley rotation with yearly mouldboard ploughing from Burgundy which was improved to reconcile weed harmfulness control, reduced herbicide use and biodiversity promotion, based on a risk-minimizing strategy. The best alternative replaced a herbicide entering plants via shoot tips (during emergence) and roots after barley sowing by a spring herbicide entering via leaves, introduced crop residue shredding before cereals and rolled the soil at sowing, which reduced the risk of unacceptable performance from 90% to 40%. When attempting to reconcile harmfulness control and reduced herbicide use, the best alternative changed the rotation to oilseed rape/wheat/spring pea/wheat, replaced one herbicide in oilseed rape by mechanical weeding, delayed tillage before rape and applied the PRE herbicide before oilseed rape closer to sowing. This option reduced the risk of unacceptable performance to 30%. None of the initial or alternative cropping systems succeeded in optimal performance, indicating that more diverse cropping systems with innovative management techniques and innovative combinations of techniques are needed to build the decision trees. This approach can be used in workshops with extension services and farmers in order to design cropping systems. Compared to expert-based design, it has the advantage to go beyond well-known options (e.g. plough before risky crops) to identify unconventional options, with a particular focus on interactions between cultural techniques.