Intensive agriculture with high reliance on pesticides and fertilizers constitutes a major strategy for ‘feeding the world’. However, such conventional intensification is linked to diminishing returns and can result in ‘intensification traps’—production declines triggered by the negative feedback of biodiversity loss at high input levels. Here we developed a novel framework that accounts for biodiversity feedback on crop yields to evaluate the risk and magnitude of intensification traps. Simulations grounded in systematic literature reviews showed that intensification traps emerge in most landscape types, but to a lesser extent in major cereal production systems. Furthermore, small reductions in maximal production (5–10%) could be frequently transmitted into substantial biodiversity gains, resulting in small-loss large-gain trade-offs prevailing across landscape types. However, sensitivity analyses revealed a strong context dependence of trap emergence, inducing substantial uncertainty in the identification of optimal management at the field scale. Hence, we recommend the development of case-specific safety margins for intensification preventing double losses in biodiversity and food security associated with intensification traps.

高度依赖农药和化肥的集约化农业是“养活世界”的重大战略。然而，这种传统的集约化与收益递减有关，并可能导致“集约化陷阱”——高投入水平下生物多样性丧失的负反馈引发产量下降。在这里，我们开发了一个新颖的框架，该框架考虑了生物多样性对作物产量的反馈，以评估集约化陷阱的风险和程度。基于系统文献综述的模拟表明，集约化陷阱出现在大多数景观类型中，但在主要谷物生产系统中出现的程度较小。此外，最大产量的小幅下降（5-10%）经常会转化为生物多样性的巨大收益，导致跨景观类型普遍存在小损失大收益的权衡。然而，敏感性分析揭示了陷阱出现的强烈背景依赖性，导致在现场规模的最佳管理识别中存在很大的不确定性。因此，我们建议制定针对具体情况的集约化安全边际，防止与集约化陷阱相关的生物多样性和粮食安全双重损失。