The genomic landscape of recombination in rice. Recombination serves as the central mechanism for generating genetic variation, assembling advantageous traits, and disrupting deleterious genetic linkages in crop breeding. While rice breeding over the past century has successfully leveraged this process to achieve remarkable gains in yield and resistance, the underlying genomic patterns of recombination remain largely uncharted. This gap in fundamental knowledge limits our ability to fully harness recombination for future breeding breakthroughs. To globally map recombination landscapes in rice, we analyzed recombination events within a large nested association mapping (NAM) population (Wei et al., 2024, Supporting Information Methods S1). This population, derived from 16 genetically diverse parents spanning major subpopulations (indica, temperate japonica, tropical japonica, aus, basmati and intermediate), comprised 16 019 homozygous lines (Fig. 1a; Table S1). Huanghuazhan, a widely cultivated indica variety from southern China, was used as the common parent. A high-density genotype map revealed abundant, well-dispersed recombination events across all chromosomes (Fig. S1). In total, we identified 508 221 recombination events. The average number per line ranged from 17.7 to 40.3 across populations, with indica-japonica crosses exhibiting the highest recombination frequency (Fig. S2). The 15 populations were categorized into six groups corresponding to their parental ecotypes. Despite this genetic diversity, analysis of recombination across all 12 chromosomes revealed a conserved pattern among the groups (Fig. 1b). For example, specific recombination events, such as the regions at 19.2–19.5 Mb on Chromosome 4 and 0.9–1.2 Mb on Chromosome 11, were common to all six groups. A genome-wide statistical analysis using 300-kb bins identified distinct ‘hot spots’ and ‘cold spots’ of recombination – genomic regions with consistently higher or lower recombination frequency, respectively (Methods S1). We identified 115 hot spots and 151 cold spots in total (Fig. 1c; Table S2). Hotspots included the two regions where crossovers commonly occur (Chr4: 19.2–19.5 Mb, Chr11: 0.9–1.2 Mb). Hot spots were often distributed near chromosome ends, whereas cold spots were strongly concentrated around centromeres and on telomeres. While hot spots appeared as discrete, scattered sites, cold spots formed continuous domains, particularly in pericentromeric regions. The presence of cold spots at all 12 centromeres indicates systemic suppression of recombination in these regions. To assess the conservation of recombination hot spots and cold spots across diverse populations, we generated four additional BC1F1 mapping populations. Analysis revealed that 53.04% of hot spots and 72.86% of cold spots were detectable in these populations (Fig. S3). These results indicate that such genomic features are largely conserved in crosses involving diverse rice varieties. We evaluated the relationship between recombination rate and several genomic features: open chromatin regions, repetitive sequences, transposable element density, and H3K4me3 histone modification (Fig. 1d). The recombination rate showed a significant positive correlation with H3K4me3 histone modification (Pearson's correlation coefficient r = 0.46, P = 5.02 × 10−66) but was negatively correlated with transposable element density (r = −0.58, P = 1.10 × 10−104). Furthermore, we observed a nearly complete (90.73%) association of cold spots with repetitive sequence regions (r = −0.56, P = 1.15 × 10−111), while hot spots predominantly (66.96%) coincided with open chromatin (r = 0.39, P = 6.12 × 10−46). These results were consistent with findings in Arabidopsis, soybean, and maize (Fu et al., 2002; Wijnker et al., 2013; Rodgers-Melnick et al., 2016; Ma et al., 2023). We analyzed the correlation between structural variation (SV) length and the weighted average of recombination events per line within a 300-kb flanking region (c. 2 cM) and found almost no correlation (Pearson's correlation coefficient r = 0.002, P = 0.875) between SV length and recombination events (Fig. S4a). Further analysis revealed that two large inversions (13.1–17.6 Mb on Chromosome 6 and 10.1–13.3 Mb on Chromosome 9) and one retrotransposon insertion (1.96–1.98 Mb on Chromosome 9) suppress recombination frequency in rice. In subpopulations carrying these three SVs, the genetic distances (cM) of bins in the flanking regions decreased rapidly compared to subpopulations without these SVs (Fig. S4b). These results are consistent with previous studies on inversions (Salomé et al., 2012) and retrotransposon insertions (Dooner Table S4). Genetic effects of the QTL ranged from −3.96 to 4.65, and these QTL collectively explained 46.4% of the variation. Considering that the NAM population was cultivated in an open-field environment, where meiotic processes are subject to environmental factors including temperature (Lloyd et al., 2018), the estimated genetic contribution of QTLs is likely to be underestimated. The presence of these QTL was further validated through genetic linkage mapping (Table S4). For instance, a major-effect QTL on the short arm of Chromosome 3 was consistently detected in three independent populations with logarithm of the odd (LOD) scores exceeding 10 (Fig. 1f). We employed our previously developed gene discovery platform, RiceG2G, to identify candidate genes. This led to the identification of four strong candidates: HOMOLOGOUS PAIRING ABERRATION IN RICE MEIOSIS1 (PAIR1), RecQ-mediated Genome Instability 1 (RMI1), FASCIATA1 (FAS1), and Multicopy Suppressor of iral (MSI1). PAIR1 encodes a nuclear protein with coiled-coil motifs and basic regions at both termini. PAIR1 protein plays an essential role in the establishment of homologous chromosome pairing in rice meiosis (Nonomura et al., 2004). FAS1 and MSI1 encode subunits of the Chromatin Assembly Factor-1 (CAF-1) complex. Mutations of either subunit lead to a c. 40-fold increase in the frequency of somatic homologous recombination in Arabidopsis (Endo et al., 2006). RMI1 is essential for resolution of meiotic recombination intermediates in Arabidopsis (Hartung et al., 2008). Harnessing these recombination-modulating QTL in breeding programs holds significant potential for enhancing genetic gain by enabling more predictable and efficient reshuffling of allelic combinations (Epstein et al., 2023). Analysis of nucleotide variation across 404 rice accessions identified seven haplotypes in the PAIR1 promoter region (2-kb upstream of the ATG) and its coding sequence (Figs 1g, S6a). Among these, the 16 parental lines clustered into two distinct haplotypes: Hap1 and Hap3. Recombination frequency was significantly lower in accessions carrying Hap1 compared to Hap3 (Fig. 1h), suggesting that recombination efficiency could potentially be repressed in Hap1 varieties, which are predominantly of the indica subspecies. Notably, no nonsynonymous single-nucleotide polymorphisms were detected in the coding region, and nucleotide diversity was higher in indica than in japonica (Fig. S6b). Furthermore, expression analysis across 10 rice tissues revealed that PAIR1 transcript levels were highest in young panicles at the meiosis stage (Fig. S7). Additionally, expression analysis in young panicles at the meiosis stage from the 16 parental lines showed significantly higher PAIR1 transcript levels in Hap1 varieties (Fig. 1i). Together, these results indicate that natural variation in the PAIR1 promoter region likely influences recombination frequency by modulating the gene's expression level. In summary, this study provides a systematic genomic dissection of recombination in rice. We mapped the landscape of recombination, revealed patterns of recombination hot and cold spots, and identified QTL controlling recombination variation. These insights into genetic recombination will accelerate the development of molecular design breeding in rice. The approaches used in this study, NAM linkage mapping and GWAS, can be directly extended to identify genes controlling recombination distribution in rice. Manipulating such genes via genome editing may enable breeders to reshape the hot spot and cold spot landscape, thereby breaking undesirable linkage drags and accelerating the generation of favorable allelic combinations. This work was supported by the National Natural Science Foundation of China (32341030 and 32222064), the Shuguang Program of Shanghai Education Development Foundation and Shanghai Municipal Education Commission (24SG44), and the Shanghai Agricultural Science and Technology Innovation Program (T2023204). None declared. XW, JQ and XH conceived the project. XW, XH and MC designed the experiments and drafted the manuscript. HH, CY and HL contributed to the population development. MC, XW, JQ and ZW analyzed the data. MC and HH contributed equally to this work. The bin map and recombination events of NAM population are available from Figshare repository with doi: 10.6084/m9.figshare.31881931. Fig. S1 Genome-wide distribution of recombination events across all 15 NAM subpopulations. Fig. S2 Recombination rate variation in the rice NAM population. Fig. S3 Genome-wide distribution of recombination events across four BC1F1 populations. Fig. S4 Correlation analysis of recombination frequency and SVs in rice NAM population. Fig. S5 Allele frequencies of the four QTGs around hot and cold spots in 404 rice accessions. Fig. S6 Genetic variation and haplotype diversity of PAIR1. Fig. S7 Expression of PAIR1 in 10 rice tissues. Methods S1 Materials and methods. Table S1 Summary information of RILs in the NAM population. Table S2 Hot spots and cold spots of recombination in rice genome. Table S3 Quantitative trait genes around hot and cold spots in rice. Table S4 QTL of recombination in the rice NAM population. Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article. The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.
Chen et al. (Wed,) studied this question.