Concilier la productivité des cultures et le maintien de la biodiversité est l'un des principaux défis de l'agriculture mondiale. Nous avons testé les hypothèses selon lesquelles (i) toutes les communautés adventices ne génèrent pas de pertes de rendement et (ii) que des communautés adventices plus diversifiées peuvent atténuer les pertes de rendement. L'étude est basée sur trois années d'observations des densités et biomasses adventices à quatre stades critiques de croissance des céréales d'hiver dans 54 zones (36 sans désherbage et 18 avec), sur un dispositif expérimental présentant un gradient de densité et de diversité d'espèces adventices. Dans ce contexte, quand la flore adventice est désherbée, les pertes de rendement ne sont pas significatives. Quand la flore n'est pas désherbée, quatre des six communautés identifiées ont généré des pertes de rendement significatives, variant de 19 à 56 %. Dans les zones non désherbées, le nombre d'épis par plante et le nombre de grains par épi ont été systématiquement affectés, quand des pertes de rendement sont détectées. Quand l'équitabilité des communautés augmente (répartition équitable des densités ou biomasse entre espèces), la biomasse totale des adventices baisse et la productivité de la culture augmente.

No abstract

There is a growing interest in the use of functional approaches for the study of weed assemblages, to disentangle underlying processes determining their composition and dynamics. Functional approaches are based on the assumption that weed community composition and dynamics can be best explained by a set of species traits expressing their response to agricultural disturbance. This knowledge should help develop more sustainable, ecologically based weed management systems. Trait-based data required for this kind of analysis are available from various sources, but most of them either cover mainly non-weedy species or, in the case of weed-focussed trait databases, they cover a limited number of species. In this work, we present a trait database for 240 weed species common throughout Europe, including not only response traits but also effect traits, that is linked to selected agroecosystem services and disservices. A case study is presented where our weed trait database is used in conjunction with appropriate statistical analysis to highlight the distribution of weed functional groups in soyabean crop communities from an experiment including different tillage and weed management systems. Finally, we discuss the strengths, weaknesses, opportunities and threats of this functional approach. By highlighting the links between weed species and agroecosystem (dis)services, this approach could be a useful resource for scientists, farm managers and policymakers.

Integrated weed management encourages long-term planning and targeted use of cultural strategies coherently combined at the cropping system scale. The transition towards such systems is challenged by a belief of lower productivity and higher weed pressure. Here, we hypothesize that diversifying the crop sequence and its associated weed management tools allow long-term agronomic sustainability (low herbicide use, efficient weed control, and high productivity). Four 6year rotations with different constraints (S2: transition from reduced tillage to no-till, chemical weeding; S3: chemical weeding; S4: typical integrated weed management system; S5: mechanical weeding) were compared to a reference (S1: 3-year rotation, systematic ploughing, chemical weeding) in terms of herbicide use, weed management, and productivity over the 2000-2017 period. Weed density was measured before and after weeding. Crop and weed biomass were sampled at crop flowering. Compared to S1, herbicide use was reduced by 46, 65, and 99% in S3, S4, and S5 respectively. Herbicide use in S2 was maintained at the same level as S1 (− 9%), due to increased weed pressure and dependence to glyphosate for weed control during the fallow period of the no-till phase. Weed biomass was low across all cropping systems (0 to 5 g of dry matter m −2) but weed dynamics were stable over the 17 years in S1 and S4 only. Compared to S1, productivity at the cropping system scale was reduced by 22% in S2 and by 33% in S3. These differences were mainly attributed to a higher proportion of crops with low intrinsic productivity in S2 and S3. Through S4's multiperformance, we show for the first time that low herbicide use, long-term weed management, and high crop productivity can be reconciled in grain-based cropping systems provided that a diversified crop rotation integrating a diverse suite of tactics (herbicides included) is implemented.

No abstract