N applications that increase plant N during panicle development are highly effective in increasing spikelet number in rice

Efficient use of nitrogen fertilizer is critical in improving yield stability in rice. The objective of this study was to determine the effect of nitrogen (N) top-dressing on the number of total spikelet (fertile plus sterile) production and evaluate the effect among rice cultivars. We analyzed 136 sets of experimental data on growth and spikelet production for three lowland cultivars, grown under various regimes of N over 10 seasons at Kyoto, Ibaraki and Kanagawa, Japan. In each season, one to three of the lowland cultivars, Nipponbare (japonica), Koshihikari (japonica) and Takanari (indica), were studied. In 1986, 1995 and 1999–2001, the N regimes included basal application only, light basal and heavy top-dressing from the panicle initiation stage onward, heavy basal and heavy top-dressing from the spikelet formation stage onward, and no applications. In 2002 and 2005–2008, we set up experimental plots with varied time of N top-dressing, with or without N basal application. Takanari had the largest spikelet number averaged over all plots and was considered better efficient in spikelet production per applied N than the other cultivars. Although the trend is not clear, the effect of time of top-dressing on spikelet number was generally the greatest when N was top-dressed from 35 to 30 days before heading. The variation of observed spikelet number was analyzed with a linear regression of plant N 14 days before heading and by a model that estimates spikelet production accounting for plant N 14 days before heading and crop growth rate (CGR) during the 14-day period preceding heading. For the variation of spikelet number within each cultivar, the linear function model expressed the observed spikelet number than the two function model with R² 0.43** versus 0.13*–0.28** for the former and later models, respectively. When the results of all cultivars were combined, the two function model was much better for estimation of spikelet number than the linear function model (R²=0.36** vs. 0.20*). This indicates that yearly and varietal variation of spikelet number was caused mainly by plant N status at the late spikelet differentiation stage. The varietal variation in spikelet production efficiency is explained by CGR during this 14-day period. We concluded that N applications that increase plant N 14 days before heading is highly effective in maximizing spikelet production among cultivars.