Climate models predict a rise in CO2 levels from around 400 ppm now to 400–1000 ppm by 2100, correlated with more extreme weather events such as droughts, storms and flooding (Intergovernmental Panel on Climate Change, 2021). Although increased CO2 levels have negative effects on ambient temperature, they have a positive impact on plant growth by increasing the photosynthesis rate, known as the CO2 fertilization effect (Chen et al., 2022). However, the potential of adapting crop plants to higher CO2 levels and thereby increase their production has not yet been fully exploited by breeding programs. Toshiyuki Takai and Yusaku Uga, authors of the highlighted publication, met at the International Rice Research Institute (IRRI) during their PhD studies, where they started to discuss the future of rice (Oryza sativa) breeding. Already back then, Takai suggested that breeders could combine larger panicles with more tillers as a next ideotype to increase the rice yield potential. Increasing the sink capacity, which is a product of both spikelet number and seed size, could be a key factor to enhance yield under elevated atmospheric CO2 levels (Ainsworth & Long, 2021). However, high-yielding rice varieties already produce large panicles, so further increases in panicle size are not considered feasible. Instead, Takai et al. addressed panicle number in an attempt to increase sink capacity (Takai et al., 2023). They focused on the quantitative trait locus (QTL) MORE PANICLES 3 (MP3). The Koshihikari allele of MP3 in a near-isogenic line (NIL) in the Takanari genetic background produced 19% more panicles and 12–20% more spikelets (Takai et al., 2014). In the highlighted study, the authors showed that NIL-MP3 produced more tillers than the parental cultivars until the heading stage and had more elongated axillary buds at the axil of the second leaf, suggesting that the increased tiller number was caused by a different elongation rate of axillary buds (Figure 1a). MP3 affects tillering in rice and grain yield under elevated CO2 conditions. (a) Takanari, NIL-MP3 and NIL-fc1 plants grown in pots. (b) Takanari, NIL-MP3 and NIL-fc1 plants grown in a paddy field in the late grain-filling stage. (c) FACE plot with the CO2 emission tubes in an octagonal arrangement. (d) Grain yield of Takanari and NIL-MP3 under ambient and elevated CO2 (FACE) conditions; figure modified from Takai et al. (2023). To clone the causal gene underlying the QTL, the authors used high-resolution mapping on segregating F2 plants and subsequent recombinant F3,4 lines. They found only one predicted open reading frame in the candidate region, TEOSINTE BRANCHED1/FINE CULM 1 (OsTB1/FC1). OsTB1/FC1 is a TCP transcription factor that is a negative regulator of tiller bud outgrowth (Takeda et al., 2003). The authors analyzed the expression pattern of OsTB1/FC1 and detected expression in the axillary buds, but saw no difference in expression level between Takanari and NIL-MP3. However, by collecting transcriptome data of the shoot apical meristem and axillary meristem at various time points and analyzing the expression trajectories, they found that NIL-MP3 reached a more advanced growth stage on the trajectory than Takanari, suggesting that MP3 exerts positive effects on axillary bud growth. MP3 had three polymorphisms in the 5′ UTR, coding sequence (CDS), and 3′ UTR, so the authors tested their importance by identifying different combinations in chromosome segment substitution lines with the Koshihikari background. They found that all three polymorphisms were necessary for the difference in panicle number. Perhaps these polymorphisms lead to an alteration in translational efficiency and protein structure, and thereby cause a moderate increase in panicle number without decreasing the number of spikelets per panicle. Analysis of haplotypes in a rice single-nucleotide polymorphism database showed that the Koshihikari allele accounted for 74% of accessions in temperate japonica subgroups, while the Takanari allele accounted for 50–59% of accessions in indica subgroups. This prompted the authors to introduce the Koshihikari MP3 allele as NILs into two high-yielding indica cultivars, Hokuriku 193 and IR64, which are categorized as Takanari MP3 haplotypes. Phenotyping under field conditions confirmed that the NILs developed a greater number of tillers and a similar number of spikelets per panicle as the parental cultivars. They previously found that the OsTB1/FC1 allele was derived from a loss-of-function allele in the fc1 mutant, which had a 1-bp deletion in the CDS, resulting in a premature stop codon. When the authors compared a NIL with this allele in the Takanari background with the Takanari wild type and the mild NIL-MP3 allele, they found that NIL-fc1 produced more panicles, and overall more spikelets per m2 than Takanari. Nevertheless, the grain yield in Takanari-fc1 was much lower than that in the Takanari wild type because the plants lodged during the grain-filling period, i.e., the stems bent near the ground level, because they were thinner and lighter than wild-type stems (Figure 1a,b). Even though the number of panicles and the number of spikelets per m2 were increased in NIL-MP3, the overall yield was not. The authors hypothesized that this was caused by limited photosynthetic capacity that was not sufficient to fill the spikelets. Therefore, they compared the yield under normal air and free-air CO2 enrichment (FACE) treatments in open paddy fields, in which CO2 is supplied from tubes installed around the experimental plots above the rice canopy (Figure 1c). NIL-MP3 had a higher sink capacity than Takanari in both CO2 conditions, and under higher CO2 conditions, the yield of NIL-MP3 was significantly higher than the yield of Takanari (Figure 1d). In conclusion, the authors propose that mild alleles such as MP3, which moderately increase panicle number, combined with large-sized panicle varieties can be used to create high-yielding varieties adapted to rising CO2 levels. At the National Agriculture and Food Research Organization in Tsukuba, Japan, Uga focuses additionally on the genetic improvement of the root system. His group is studying the effects of pyramiding some QTLs for root traits such as root length, volume, and thickness, and in the future he will analyze the potential of pyramiding above- and belowground QTLs affecting yield and drought resistance.