Optimizing nitrogen supply increases rice yield and nitrogen use efficiency by regulating yield formation factors

Recently, many optimized nutrient management strategies have been applied to improve rice yield and nitrogen use efficiency (NUE) in China, most of which achieved higher yield and NUE than farmers’ usual fertilizer practices. The objective of this study was to investigate the aboveground biomass, nitrogen (N) accumulation and contribution of yield factors (i.e., number of effective panicles, number of spikelets per panicle, and grain weight) on rice yield regulated by optimized nitrogen application at different growth stages. Field experiments were conducted in 2009 and 2010 at 7 different sites, and three optimized N treatments (OPT) were compared with local farmers’ fertilizer practices (FFP). Rice yields of the optimized treatments increased 8.2–12.6% over the farmers’ fertilizer practices. Recovery efficiency of N (REN), agronomic N use efficiency (AEN) and partial factor productivity of applied N (PFPN) for OPT1 and OPT2 were all significantly higher than those obtained using the farmers’ fertilizer practices. The aboveground biomass and nitrogen accumulation of OPTs were lower than FFP at the vegetative and earlier reproductive growth stages, but the accumulation rate became faster than FFP at the later reproductive stage. In our result, grain yield was positively correlated with the panicles per unit area for OPTs, but not for FFP. Spikelets per panicles had negative correlation with panicles per unit area but the slop of OPTs was slower than the slope of FFP. We found that the OPTs treatments could mitigate the contradiction between yield formation factors. According to yield responses to the treatments, we defined two dominant yield-promotion stages: the primary nutrient contribution stage and the advanced nutrient contribution stage. Our results illustrated that at the primary nutrient contribution stage, the main promotion factor for yield was panicle number, which was mainly promoted by high N application at the early vegetative stage, and at the advanced nutrient contribution stage, the yield increase depended upon resolving the contradiction between number of effective panicles and spikelet number per panicle, which resulted from appropriate adjustments to the proportions of added N applied at different growth stages. In conclusion, our results indicated that adjusting the proportion of N application at different growth stages may reform the source–sink contradiction of yield, thus further increasing rice yield.