Abstract
Cropland fertilization is the largest anthropogenic source of nitrous oxide (N 2 O‐N) emissions in China and could be mitigated through manure substitution. However, accurate quantification of N 2 O‐N losses under different combinations of manure and chemical fertilizer applications remains challenging due to limited national data, oversimplified methods of estimation, and neglect of the nitrogen (N) source (chemical‐N and manure‐N) interaction effects under mixed fertilization on N 2 O‐N production processes. Using a dataset consisting of 2186 observations across China, we developed N 2 O‐N models for fertilized (chemical fertilizer and manures) croplands incorporating climate, soil, fertilization and cropping variables, with particular emphasis on the manure component, through machine‐learning approaches. Model simulations estimated a median N 2 O‐N flux of 1.1 kg ha −1 and a total national N 2 O‐N loss of 243.7 Gg in 2019, with a hotspot in Northcentral China. During 2000–2019, N 2 O‐N losses were driven mainly by annual total N input. By 2050, merely pursuing 100% manure recycling without considering improved manure‐N use efficiency could lead to an 18% increase in N 2 O‐N losses compared with 2019. Hence, this study provides a high‐resolution modeling framework to predict how developments in manure substitution will affect the magnitude and spatial distribution of N 2 O‐N losses. We also demonstrate that future manure substitution policies will need to promote improved manure nutrient management at the same time as greater manure recycling rates.