Sustained rice yields and decreased N runoff in a rice-wheat cropping system by replacing wheat with Chinese milk vetch and sharply reducing fertilizer use

Pollution from the paddy fields has posed a threat to surface water quality, and the reactive N in runoff has been recognized as the dominant contributor. In the rice-wheat systems of eastern China, replacing wheat (Triticum aestivum) with Chinese milk vetch (CMV) (Astragalus sinicus) is known to reduce total fertilizer N use and associated N losses during winter; however, the function of the rice-CMV system in controlling the N runoff loss was overlooked during the summer rice-growing season. Over 6 years, we monitored soil mineral N, plant N accumulation, rice grain yield, N agronomic efficiency (AEN), and N runoff in rice-CMV fertilizer N rate-response experiments and made comparisons with the conventional N inputs in rice-wheat rotation. Aboveground CMV residues added 65–116 kg N ha−1 yr−1; therefore, by adjusting the fertilizer time, the rice in this system required 44–56% less N fertilizer to produce rice yields equivalent to the 270 kg N ha−1 (district average, C270) used in the rice-wheat system. In all fertilizer N application treatments, 120 kg ha−1 seemed to be the threshold that ensured the soil N supply, the N accumulation at rice critical stages, and consequently, the current level rice yield. The corresponding runoff N averaged 9.3 kg ha−1 season−1, which was 51.8% less than that in C270 (19.3 kg ha−1 season−1). Cumulative N runoff (total N and NH4+-N) correlated strongly with fertilizer N input for any single year (sample size = 108, P