Objective The comprehensive effects of intelligent fertilizer recommendation on potato production in northern China have yet to be systematically quantified based on multi-site field experimentation. This study aimed to systematically evaluate its agronomic, environmental, and economic performance based on multi-year, multi-site field validation trials and county-level household fertilizer survey data.Furthermore, the key drivers underlying effect heterogeneity and the spatial distribution of carbon and nitrogen mitigation potential were investigated.

Methods A dataset comprising 256 field validation trials of intelligent fertilizer recommendation conducted across the northern potato monocropping region from 2017 to 2024, coupled with household fertilizer survey data spanning 122 counties was compiled. Potato tuber yield, nutrient use efficiency, and economic benefits were evaluated. Greenhouse gas emissions (GHG), reactive nitrogen losses (Nr), carbon footprint, and nitrogen footprint (NF) were quantified via life cycle assessment. Random forest modeling was applied to identify the dominant factors driving observed variation, and machine learning-based spatial prediction was employed to delineate the spatial pattern of cropland mitigation effects.

Results Relative to conventional farmer fertilization practice, intelligent fertilizer recommendation enhanced nitrogen, phosphorus, and potassium use efficiency by 12.8, 2.78, and 8.31 percentage points, respectively, with no significant effect on tuber yield. In terms of environmental outcomes, intelligent fertilizer recommendation reduced Nr, GHG, and NF of potato cropland by 22.5%, 9.23%, and 18.5%, respectively. Economic assessment revealed that net economic benefit and net ecological economic benefit under intelligent fertilizer recommendation reached 33533.8 and 29359.7 CNY/hm², representing increases of 1.70% and 6.05%, respectively, compared with conventional farmer practice. Random forest analysis identified that N and K₂O application rates were the primary drivers influencing variations in potato yield, reactive nitrogen losses, and greenhouse gas emissions. Spatial prediction results showed that areas with with high synergistic carbon and nitrogen mitigation potential were mainly concentrated in central Inner Mongolia, central-eastern Gansu, northwestern Shaanxi, northern Hebei, and western Liaoning, exhibiting significant spatial aggregation.

Conclusion Intelligent fertilizer recommendation can synergistically achieve fertilizer reduction, efficiency gains, and carbon-nitrogen mitigation without compromising yield, underscoring its considerable potential for widespread adoption across the northern potato production region. These findings provide a quantitative scientific basis and technological support for optimizing zonal nutrient management strategies, advancing precision fertilization, and facilitating the green and low-carbon transition of potato production in northern China.