Estimating regional N application rates for rice in China based on target yield, indigenous N supply, and N loss.

Decision-making related to nitrogen (N) applications based solely on historic experience is still widespread in China, the country with the largest rice production and N fertilizer use. By connecting N application rates with target N uptake, indigenous N supply, and N loss estimates collected from 1078 on-farm experiments, we determined regional N application rates for five rice-based agroecosystems, including a quantification of the reduction potential of application rates when using low-loss N sources, such as organic N and slow-release N. Based on our results, the moderate regional N application rates were 165, 180, 160, 153, and 173 kg N ha for single, middle-CE (Central and Eastern China), middle-SW (Southwestern China), early, and late rice, respectively; lower (99-148 kg N ha) and upper (195-217 kg N ha) limits of N application rates were developed for situations with sufficient and insufficient indigenous N supplies, respectively. The depletion of soil N mineralization was quantified as 46.8-67.3 kg ha, and straw return is determined to be a robust measure to maintain soil N balance. Substituting manure or slow-release N for conventional N fertilizer significantly decreased N losses via NH volatilization, leaching, runoff, and NO emissions. Overall, we observed 7.2-11.3 percent point reductions of N loss rate for low-loss N sources when compared to conventional N applications. On average, total N application rates could be theoretically reduced by 27 kg N ha by using a slow-release N fertilizer, or by 30 kg N ha when using manure due to their effectiveness at decreasing system N losses. Greater productivity, sustainable soil fertility, and a lower risk of N pollution would result from the ideal N application rate coupled with appropriate management practices. Widespread adoption of using low-loss N sources could become a key solution for future reduction in environmental N pollution and agricultural N inputs.