Pleiotropic Effect of GNP1 Underlying Grain Number per Panicle on Sink, Source and Flow in Rice

Rice yield potential is largely determined by the balance among source capacity, sink strength, and flow fluency. Our previous study indicated that the gene GNP1 encoding gibberellin biosynthesis gene GA20ox1 affects grain number per panicle (GNP) in rice, thus resulting in increase of grain yield. To clarify GNP1 effect on sink, source and flow in regulating rice grain yield, we compared Lemont, a japonica (geng) cultivar, with its near-isogenic line (NIL-GNP1 ^TQ) in Lemont background with introgression of the allele at GNP1 from Teqing, a high-yielding indica (xian) cultivar. NIL-GNP1 ^TQ exhibited averagely 32.8% more GNP than Lemont with the compensation by reduced seed setting rate, panicle number and single-grain weight. However, NIL-GNP1 ^TQ still produced averagely 7.2% higher grain yield than Lemont in two years, mainly attributed to significantly more filled grain number per panicle, and greater vascular system contributing to photoassimilates transport to spikelets. The significantly decreased grain weight of superior spikelets (SS) in NIL-GNP1 ^TQ was ascribed to a significant decrease of grain size while the significantly decreased grain weight of inferior spikelets (IS) ascribed to both grain size and poor grain-filling as compared with Lemont. The low activities of key enzymes of carbon metabolism might account for the poor grain-filling in IS, which resulted in more unfilled grains or small grain bulk density in NIL-GNP1 ^TQ. In addition, low seed setting rate and grain weight of IS in NIL-GNP1 ^TQ might be partially resulted from significantly lower carbohydrate accumulation in culms and leaf sheath before heading compared with Lemont. Our results indicated that significantly increased GNP from introgression of GNP1 ^TQ into Lemont did not highly significantly improve grain yield of NIL-GNP1 ^TQ as expected, due primarily to significant low sink activities in IS and possible insufficient source supply which didn't fully meet the increased sink capacity. The results provided useful information for improving rice yield potential through reasonably introgressing or pyramiding the favorable alleles underlying source-related or panicle number traits by marker-assisted selection.