Estimated reactive nitrogen losses for intensive maize production in China

• Both N2O emissions and N leaching increased exponentially with the N application rate or N surplus.
• NH3 volatilization increased linearly with the N application rate.
• Estimated Nr losses based on N-rate were a little higher than based on N surplus.
• An optimal N management reduced Nr loss intensity by 45%, compared to farmers’ practice.

A thorough understanding of reactive N (Nr) losses from N fertilization applications and the factors that influence it is necessary to better evaluate various Nr losses mitigation scenarios and improve N management practices. The objectives of this study were to develop empirical models to calculate Nr losses using meta-analysis and to evaluate trade-offs among grain yield, N recovery efficiency (REN), and Nr loss intensity for in-season N management for intensive summer maize (Zea mays L.) production in China. A meta-analysis with 55 studies and 170 observations suggested that both N2O emissions and N leaching increased exponentially with the N application rate or N surplus, while NH3 volatilization increased linearly with the N application rate. According to regression curves, models based on the N rate (N-R) and N surplus (N-S) were used to estimate Nr losses. Because the N-R did not account for large variations in REN or grain yield across farmers’ fields for difference competence in N management, estimated Nr losses were a little higher than those estimated by the N-S, especially for high-yield, high-REN systems. Across 162 on-farm experimental sites, an in-season root-zone N management strategy with a 39% lower N application rate and 6% higher grain yield increased the REN by 98% (from 16% to 31%) and reduced Nr loss intensity (based on the N-S) by 45% (from 13.6 to 7.5 kg N Mg−1) compared to farmers’ typical N practices. In conclusion, the reconciliation of food security with greater environmental protection for the future can be driven by improved agronomic management to increase grain yield as well as REN, rather than by solely focusing on optimizing the N application rate.