Pesticides are widely used in agriculture to manage pests, weeds and diseases with consequences on ecosystem and human health. Reducing pesticide reliance appears as a necessary step for transitioning to sustainable farming systems. Many non-chemical techniques to control pests, weeds and diseases are individually well documented, but quantitative assessment of how much the reliance on pesticides could be decreased through a holistic approach of cropping system re-design combining several individual techniques was lacking. A diachronic approach was conducted on 913 French commercial farms from the DEPHY network, spanning a diversity of contexts and management strategies. Regression Trees and Random Forest models were run to identify combinations of changes in management strategies associated with the evolution of pesticides reliance over time, in a range of production contexts. The consequences of these changes on pesticide-related risks to human health and the environment were also investigated. Farm reliance on pesticides and pesticide-related risks to human health and the environment were reduced in 65.7% and 60.3% of the farms, respectively. The greatest reductions in pesticide reliance were observed in farms with the highest initial levels of reliance on pesticides. Crop diversification and reduced nitrogen fertilization were often associated with reduced pesticide reliance. This study highlights the value of real farm networks for identifying and promoting combinations of techniques that help reduce pesticide dependency and move towards more sustainable farming systems.

中文翻译：

---

作物管理策略的重新设计可以减少对农药的依赖：基于法国商业农场多样性的历时方法

农药在农业中广泛用于防治害虫、杂草和疾病，对生态系统和人类健康造成影响。减少农药依赖似乎是向可持续农业系统过渡的必要步骤。许多控制害虫、杂草和疾病的非化学技术都有详细的记录，但缺乏对通过结合几种单独技术的种植系统重新设计的整体方法可以减少对农药的依赖程度的定量评估。对 DEPHY 网络中的 913 个法国商业农场进行了历时分析，涵盖了不同的背景和管理策略。运行回归树和随机森林模型来确定在一系列生产环境中与农药依赖随时间演变相关的管理策略变化的组合。还调查了这些变化对人类健康和环境与农药相关风险的影响。65.7%和60.3%的农场对农药的依赖以及农药对人类健康和环境造成的相关风险降低。在最初对农药依赖程度最高的农场中，农药依赖程度下降幅度最大。作物多样化和氮肥施用量的减少往往与农药依赖的减少有关。这项研究强调了真实农场网络对于识别和推广有助于减少农药依赖并转向更可持续农业系统的技术组合的价值。