‘Two‐floret spikelet’ as a novel resource has the potential to increase rice yield

Yield in rice (Oryza sativa) is determined by three major components: panicle number per plant, grain weight and grain/ spikelet number per panicle (Zhou et al., 2015). Grain number per panicle is one of the main targets and mainly results from the number of spikelets. Traditionally, rice breeders have focused on the improvement of spikelet number per panicle and rarely focused on the number of florets because a normal rice spikelet has one fertile floret and produces one seed. In grass, the spikelet comprises one to 40 florets depending on the species and shows determinacy or indeterminacy. In rice (O. sativa) with a determinate spikelet, the spikelet meristems produced the fixed floral meristems, resulting in the formation of one floret. In wheat (Triticum aestivum) with an indeterminate spikelet, the spikelet meristems produced the variable floral meristems, resulting in the formation of more florets. How to further increase rice yield? If the number of florets in a spikelet could be increased, it may be a new method for rice high production. In our study, we characterized two allelic mutants with two florets within a single spikelet, double floret1-1 (df1-1) and df1-2. We next focused on the df1-1 mutant to investigate the regulation of floret number in rice, and this provided a new perspective for increasing grain number per panicle and yield. The wild-type rice spikelet has one fertile floret that is flanked by one pair of glumes, which are generated from the spikelet meristem, and one floret per spikelet is strictly regulated in the Oryza genus (Figure 1a–d). The floret comprises the lemma, palea, lodicule, stamen and pistil (Figure 1a–d). The 15%–20% spikelets developed two florets inside one pair of sterile lemma which randomly distributed in the df1-1 mutant (Figure 1g–h). Rarely, three florets were also observed (Figure 1c). In the df1-1 mutant, we observed four whorls of floral organs within each single floret (Figure 1g–j). Each lemma and palea had a similar histological texture and vascular bundles as the wild type, respectively (Figure 1c,i). To confirm the identity of the organs in each single floret, we investigated the expressions of OsMADS1, OsMADS14, OsMADS15, OsMADS6 and DL responsible for the lemma and/or palea identity. No differences in gene expression were found within each single spikelet between the wild type and df1-1 mutant. We next investigated the expression of the OSH1 gene, which is essential for floral meristem activity. OSH1 showed a much higher expression in the young panicles of the df1 mutant. These results suggested that each floret in the df1-1 mutant has a normal lemma and palea, and two florets form within a single spikelet. At maturity, the seed setting rate, grain size and weight of the normal and original florets in the df1-1 mutant were comparable with those in the wild type, respectively. The seed setting rate of the secondary florets was lower, and the grain size and weight were variable in the df1-1 mutant. These results suggested that DF1 has the potential for increasing the grain number per panicle and yield. We next investigate early spikelet development. Compared with the wild type, we found two pairs of the lemma and palea in a single df1-1 spikelet at the Sp4 stage, which were generated from different floral meristems (Figure 1m,q). At the Sp5 and Sp6 stages, two florets within a single spikelet developed normal stamen primordia which exhibited a similar developmental process with that of the wild-type floret (Figure 1c,i,n,r). At the Sp7 stage, with formation of the pistil, the stamen primordia of