Returning organic nutrient sources (for example, straw and manure) to rice fields is inevitable for coupling crop–livestock production. However, an accurate estimate of net carbon (C) emissions and strategies to mitigate the abundant methane (CH₄) emission from rice fields supplied with organic sources remain unclear. Here, using machine learning and a global dataset, we scaled the field findings up to worldwide rice fields to reconcile rice yields and net C emissions. An optimal organic nitrogen (N) management was developed considering total N input, type of organic N source and organic N proportion. A combination of optimal organic N management with intermittent flooding achieved a 21% reduction in net global warming potential and a 9% rise in global rice production compared with the business-as-usual scenario. Our study provides a solution for recycling organic N sources towards a more productive, carbon-neutral and sustainable rice–livestock production system on a global scale.

将有机养分源（如秸秆、粪肥）返还稻田是农牧业耦合生产的必然选择。然而，对净碳 (C) 排放量的准确估计以及减少有机源稻田大量甲烷 (CH _{4 ) 排放的策略仍不清楚。}在这里，我们利用机器学习和全球数据集，将实地调查结果扩展到全球稻田，以协调稻米产量和净碳排放量。考虑总氮输入、有机氮源类型和有机氮比例，制定了最佳有机氮（N）管理。与正常情况相比，最佳有机氮管理与间歇性洪水相结合，使全球净变暖潜力降低了 21%，全球稻米产量提高了 9%。我们的研究提供了一种回收有机氮源的解决方案，以在全球范围内建立更具生产力、碳中和和可持续的稻畜生产系统。