Nitrogen addition enhances soil carbon and nutrient dynamics in Chinese croplands: a machine learning and nationwide synthesis

Nitrogen (N) addition is a critical driver of soil organic carbon (SOC) sequestration and nutrient cycling in croplands. However, its spatial variability and long-term effects under diverse environmental conditions remain poorly understood. We synthesised data from 479 cropland sites across China and apply machine learning models to evaluate the impacts of N addition on SOC and key soil nutrient indicators, including total nitrogen (TN), nitrate (NO₃⁻-N), ammonium (NH₄⁺-N), the carbon-to-nitrogen ratio (C/N), and available phosphorus (AP). We further evaluated the moderating roles of climate zones, fertiliser types, and fertilisation duration. Our findings demonstrate that N addition significantly increased SOC, TN, NO₃⁻-N, NH₄⁺-N, and AP contents, whereas the C/N ratio remains unaffected. SOC sequestration was greater in arid regions, whereas nutrient accumulation was more pronounced in humid zones. Organic and integrated (organic-inorganic) fertilisers outperformed chemical ones in enhancing SOC and nutrient cycling. Long-term N input (> 10 years) markedly intensified SOC storage and nutrient accumulation. We further developed the high-resolution (5 km) national-scale dataset that predicts the spatial responses of SOC and nutrient dynamics to nitrogen addition across China. This AI-derived dataset enables automated mapping of soil carbon and nutrient functions, capturing substantial spatial heterogeneity under varying environmental conditions. These results provide critical insights for optimising nitrogen management strategies, enhancing soil carbon sink functions, and informing precision agriculture policies in China.