Contribution of sink and source sizes to yield variation among rice cultivars

In order to identify the key factors that arrest yield improvement in rice, we observed fifteen divergent cultivars in a field at Kyoto, Japan in 1995 and 2001 under various nitrogen (N) regimes. The contribution that sink size (spikelet number X single fully ripened grain mass), source size (total available carbohydrate), and source components, non-structural carbohydrate (pre-reserved) at full heading (NSCh) and dry matter production during grain Filling (DMP) had to the variation in yield among cultivars was examined. The dry weight of rough brown rice (Y) ranged from 310 to 743 g m(-2) throughout two years and under all N regimes examined. Although Y correlated with both sink and ource sizes, it tended to correlate more closely with source size than with sink size. In many cultivars, source size was smaller than sink size at all conditions examined except for the low nitrogen regime. The contribution of source components to Y was analyzed with the equation: Y=C-n NSCh+C-d DMP, where C-n and C-d are coefficients of NSC utilization and of DMP utilization for grain filling. Y correlated with DMP more closely than with NSCh. DeltaNSC (NSCh - NSCm), where NSCm is NSC at maturity and C-n, vaguely correlated with the difference between sink size and DMP, showing that NSC is used to ompensate for the shortage of DMP to rill grains. At the same time, there were cultivar differences in NSCh and C-n. The highest yielding cultivar Takanari always had the greatest DMP, relatively high NSCh and stably high values of C-n. In conclusion, yield variation among rice cultivars correlated with source size more closely than with sink size, and DMP rather than NSC, primarily contributed to Y. While NSCh tended to be utilized complementarily to DMP, the contribution of NSCh seemed to depend on the ability of rice cultivars to utilize NSC.