Different genomic footprint of small insertion-deletion and structural variants determines the genetic divergence of indica and japonica rice

Abstract

Background

As the two main variety groups in improved rice, IND (indica) and JAP (temperate japonica and tropical japonica), extensive studies have utilized (single-nucleotide polymorphisms) SNPs on 233 selected improved rice cultivars to analyze the genetic basis of agronomic traits during their improvement and to compare their similarities and differences. However, the roles of other types of variations, such as insertions and deletions (INDELs) and structural variations (SVs), remain relatively underexplored.

Results

Here, using resequencing data from 233 improved rice accessions (IND: 142, JAP: 91), we identified 811,646 INDELs and 16,231 SVs in IND, and 652,793 INDELs and 10,533 SVs in JAP. The abundance of INDELs and SVs decreased as their length increased in both variety groups. INDELs and SVs also showed an uneven distribution across the chromosomes of the two variety groups. In IND, 38.92% of INDELs and 44.1% of SVs were located within the 2 kb upstream and downstream of genes; in JAP, this proportion was 38.49% for INDELs and 43.22% for SVs. By performing genome wide association studies (GWAS) using phenotypic data of six agronomic traits (heading date, flag leaf length, flag leaf width, panicle number, plant height, and thousand grain weight) along with INDELs and SVs, we identified 3,222 significant IND-INDELs, 537 significant IND-SVs, 1,996 significant JAP-INDELs, and 286 significant JAP-SVs. Comparison of significant loci revealed that IND and JAP shared only one INDEL associated with flag leaf length and one SV associated with panicle number, suggesting distinct genetic architectures determined by INDELs and SVs for these traits in the two groups. Furthermore, haplotype analysis of candidate genes demonstrated that INDELs and SVs influence key functional genes, such as the gene TAD1 (IND-INDELs) in flag leaf length, RCN2 (IND-SVs) in heading date, PLS2 (JAP-INDELs) in plant height, and OsYLC2 (JAP-SVs) in leaf development.

Conclusions

This study analyzed the variation patterns of INDELs and SVs during the improvement of IND and JAP varieties, and identified INDELs and SVs associated with agronomic traits. These findings will provide valuable genetic and material resources for rice breeding.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12864-026-13053-7.

Background

Differences in DNA sequences and structures among individuals are defined as genetic variation which serves as a source of biodiversity and is an indispensable element for the exploration and development of modern genetics. With the progress in sequencing technologies and data analysis methods, our understanding of genetic variation continues to deepen [1]. Genetic variation manifests as variants of varying ranges and sizes, which are primarily classified into single nucleotide polymorphisms (SNPs), insertions and deletions (INDELs, 1–50 bp), and structural variations (SVs, > 50 bp) based on their base length [2]. Notably, some SVs can span up to millions of base pairs or even entire chromosomes. Additionally, SVs encompass multiple types, such as insertions, deletions, duplications, inversions, and translocations [3]. Existing research has shown that SVs alone affect the number of bases in the genome several times greater more than SNPs [2, 4]. When considering the additional impact of INDELs on base number, their combined influence becomes remarkably substantial. Therefore, compared to SNPs, INDELs and SVs exert broader effects in biological process. In recent years, with the continuous optimization of sequencing and computational approaches, various forms of INDELs and SVs have been identified at the population level [2, 5]. Extensive studies have demonstrated their prevalence in genomes and their role as key contributors to numerous evolutionary processes, such as phenotypic outcomes in human evolution and disease, phenotypic diversity in plants and animals, and local adaptation [6–8].

Growing evidence indicates that INDELs and SVs have contributed to crop domestication and improvement by influencing traits such as growth, development, yield and stress resistance. For example, a 1,133 bp structural variation located 16 kb upstream of GhPH1 which encodes gibberellin 2-oxidase 1A affects plant height by regulating the gene expression [9]. Through super-pangenome analysis, a large number of structural variations have been identified in tomatoes and rice [10, 11]. For example, a 244 bp deletion in the exon of a cytochrome P450 gene influences plant architecture and yield in tomatoes [10]. A 43.3 kb and 66.6 kb deletion covered OsWAK112d increased the rice blast resistance [11]. Xia et al. identified a gene, RHW1, controlling husk leaf width in maize using GWAS. A 18 bp INDEL in the 3’ untranslated region of RHW1 determines husk leaf width by altering its protein abundance [12]. He et al. screened a Q-interacting protein, TaLAX1, which encodes a basic helix-loop-helix transcription factor, using yeast two-hybrid assays. A natural variation (+/- TATA) in the promoter region of TaLAX1 is associated with differences in promoter activity between bread wheat and its ancestor Aegilops tauschii [13]. Furthermore, numerous INDELs and SVs have been identified in crops at the population level, and the roles of some variations in domestication and improvement have been validated [11, 14–16]. Nevertheless, studies analyzing the variation patterns of INDELs and SVs and their potential effects on genes during crop improvement at the whole-genome level remain limited.

Rice (Oryza sativa L.) is one of the world’s most important crops, serving as a staple food source for nearly half of the global population. It is also an ideal model system for biological research [17]. As one of the oldest domesticated cereal crops, the transition of rice from its wild ancestors to cultivated varieties primarily involved two processes: domestication (from wild species to landraces) and improvement (from landraces to modern breeding varieties) [18, 19]. Although rice improvement has taken less than 100 years, cultivated varieties have become better suited to modern agricultural production and consumption needs under artificial selection compared to landraces, in terms of agronomic traits, yield, quality, biotic and abiotic stress. Based on differences in morphological characteristics, physiological and biochemical responses, rice can be divided into two subspecies: Indica (O. sativa ssp. indica) and Japonica (O. sativa ssp. japonica) [20]. Indica is tolerant to humidity and heat and is mainly grown in the tropical and subtropical lowland regions of south and southeast Asia, while Japonica is tolerant to cold and low light, primarily distributed in the tropical highland areas of southeast Asia and the cold climates of northeast Asia. Previous studies have shown that there are six genetically distinct groups within cultivated rice, including indica (IND) and aus (AUS) belonging to the Indica subspecies, and temperate japonica (TEJ), tropical japonica (TRJ), aromatic (ARO), and rayada (RAY) belonging to the Japonica subspecies [21, 22]. Among these, IND and JAP (TEJ and TRJ) are the primary groups used in modern rice breeding. To date, numerous studies have explored the genetic basis of agronomic traits in rice using population genomic data [23, 24]. However, these studies have primarily relied on SNPs rather than INDELs and SVs, which may lead to the omission of key genetic information. Although some studies have utilized INDELs and SVs to investigate the genetic basis of phenotypes in rice, they have not distinguished between different groups, and reports on the patterns of INDELs and SVs in different populations remain limited [25]. Therefore, it is necessary to analyze the variation patterns of INDELs and SVs in the IND and JAP during rice improvement and uncover the genetic basis underlying agronomic traits.

In this study, we utilized resequencing data from 233 improved rice varieties (IND: 142, JAP: 91) to identify INDELs and SVs within the IND and JAP groups, respectively. We analyzed the genomic patterns of these two variations in IND and JAP, including the number, size, distribution, and proportion across different genomic regions. Furthermore, we conducted GWAS using both INDELs and SVs data along with six agronomic traits (heading date, flag leaf length, flag leaf width, plant height, panicle number, and thousand grain weight), and identified key genes in both groups and pinpointed potential functional mutations. Finally, we performed a comparative analysis of the patterns of INDELs and SVs and the genetic basis underlying the agronomic traits between IND and JAP. Our research highlights the importance of INDELs and SVs in the genetic divergence of the IND and JAP rice populations. Additionally, the identified candidate genes and potential mutations will provide valuable genetic resources for rice breeding.

Materials and methods

Materials and phenotypes

We selected a panel of 233 improved rice cultivars, comprising 142 IND and 91 JAP. These materials included 148 previously reported in earlier studies [26], along with 87 newly added materials (35 IND and 52 JAP), of which 2 were excluded for not conforming to the phylogenetic relationships of the IND and JAP groups. Among them, the information for 79 accessions was obtained from the Genesys platform (https://www.genesys-pgr.org/), where they were documented as advanced/improved cultivars. The remaining 154 accessions were derived from a previously published study [19], in which the authors performed a stratification referred to analysis on these accessions. 110 accessions were classified as typical improved varieties, and 44 accessions containing modern variety components were also regarded by us as improved cultivars. Phenotypic data for six agronomic traits (flowering time, flag leaf length, flag leaf width, panicle number, plant height, and thousand grain weight) were obtained from Rice Functional Genomics & Breeding (RFGB) database (https://rfgb.rmbreeding.cn/index). Five traits (excluding thousand grain weight) were collected for all accessions in Hainan, China in 2018; while thousand grain weight were obtained for 182 accessions from Hainan, China in 2016. Each phenotype was recorded with least three replicates per accession.

Sequencing data

The resequencing data of 233 rice varieties were retrieved from the publicly available 3,000 Rice Genomes Project [24]. We used Trimmomatic (v0.36) [27] to remove adapters and low-quality bases from raw reads. We generated alignment indices for the Nipponbare reference genome (IRGSP-1.0) using BWA (v0.7.10) [28] and SAMtools (v1.1) [29]. Clean reads were aligned to the Nipponbare reference genome (IRGSP-1.0) using the BWA-MEM algorithm (v0.7.17-r1188). We processed the alignment files through Genome Analysis Toolkit 4 (GATK4, v4.1) (https://gatk.broadinstitute.org/hc/en-us) by: (1) converting alignment file from SAM to BAM format, (2) sorting aligned reads, (3) marking PCR duplicates, and (4) creating BAM index. We calculated the number of mapping reads, and sequencing depths (×) by SAMtools (v1.1) [29] and GATK3 (v3.8), respectively.

INDELs calling

Variant calling for insertion and deletions (INDELs) was performed on all samples through the HaplotypeCaller in GATK using standard default settings. From the called variants, we isolated INDELs and applied quality filters using GATK’s VariantFiltration tool with the following thresholds: Quality by depth (QD) < 2; FisherStrand (FS) > 200; Read position rank sum (ReadPosRankSum) < -20. Additional filtering was conducted using VCFtools (v0.1.16) [30] to retain biallelic INDELs (--min-alleles 2 --max-alleles 2) with minor allele frequencies (MAFs) ≥ 0.01 and missing rate < 40% [26].

SVs calling

Structural variants calling was performed using Delly (v0.8.7) [31] and Manta (v1.6.0) [32] software with input BAM files generated from resequencing alignments. Per-accession SVs calling was executed with Delly, followed by consolidation of individual VCF outputs into a unified reference panel. Then, SVs calling was performed using the merged VCF as a reference panel. We retained only variants with PASS flags for downstream analysis, followed by multi-sample combination using BCFtools (v1.9) [33]. For manta, the recommended workflow was utilized to conduct SVs calling for each individual. All SVs calling files were merged using SURVIVOR (v1.0.7) [34]. Furthermore, Delly and Manta-derived SVs were merged via an in-house Perl script to generate a comprehensive SVs catalog. SVs across IND or JAP cultivars were filtered using VCFtools (v0.1.16) [30] with minor allele frequency (MAF) ≥ 0.01 and missing rate < 40% [35].

Distribution of INDELs and SVs

According to the locations of INDELs and SVs breakpoints relative to the gene, we categorized INDELs and SVs into six groups: (1) 2 kb upstream of gene, (2) 2 kb downstream of gene, (3) UTR region (5’ UTR and 3’ UTR), (4) coding region, (5) intron region, and (6) intergenic region (the remaining region outside the five specified groups). We statistically analyzed the percentage distribution across all categories.

Identification of INDELs and SVs hotspot regions

We computed the density of INDELs and SVs breakpoints in 1,000 kb windows with 500 kb step size across each chromosome. Then, according to the numbers of INDELs and SVs within the window, we sorted all 1,000 kb windows in descending order. The top 10% of windows were defined as INDELs or SVs hotspots. In addition, all the continuous hotspot windows were merged as a single hotspot region [11].

GWAS and genetic correlation analysis

We conducted genome-wide association studies (GWAS) for INDELs and SVs against six agronomic traits (days to heading, plant height, flag leaf length, flag leaf width, panicle number, and thousand grain weight) using the BLINK-R model in Genome Association and Prediction Integrated Tool (GAPIT) [36, 37]. The first three principal components were included as fixed effects to control for population structure. The percentage of variance explained by each principal component was calculated as its eigenvalue divided by the total variance. For each trait, we derived significance thresholds by conducting 1,000 conditional permutation tests [38]. The -log (p) value corresponding to the 95% confidence level was selected as the significance criterion for each trait [39]. Furthermore, we performed clumping analysis using PLINK (v1.9) [40] to detect the independent significant associated site. To evaluate haplotype-phenotype correlations, we defined gene haplotypes according to the presence or absence of variants and conducted differential phenotype analysis across haplotype groups via two-tailed Student’s t-test in R [41]. The shared significant associated INDELs and SVs (defined as those with exactly the same physical positions) were tallied both within and between IND and JAP group.

Results

Identification of INDELs and SVs

To investigate the roles of INDELs and SVs in the improvement of IND and JAP, we selected 233 typical improved varieties (IND: 142, JAP: 91) from the Rice 3 K database and downloaded their resequencing read data. Sequence alignment was performed using the Nipponbare genome (IRGSP1.0) as the reference. The sequencing depth and coverage of the 233 accessions are provided in Table S1. Overall, the average sequencing depth for IND was 14.88 × (6.77–33.33×), with an average coverage of 90.10% (83.2%–97.5%). For JAP, the average sequencing depth was 15.36 × (4.84–33.07×), with an average coverage of 95.28% (88.0%-98.7%). We used GATK to identify small deletion and insertion variants (INDELs, 2–50 bp). Ultimately, 811,646 INDELs were detected in IND, including 458,265 deletions and 353,381 insertions. In JAP, we identified 652,793 INDELs, including 364,656 deletions and 288,137 insertions (Fig. 1a). For structural variants (SVs, > 50 bp), we utilized Manta and Delly to identify 16,231 SVs in IND, including 13,270 deletions, 1,092 insertions, 1,085 duplications, 224 inversions, and 560 translocations. In JAP, we identified 10,533 SVs, including 8,798 deletions, 647 insertions, 643 duplications, 136 inversions, and 309 translocations (Fig. 1b). Due to the inability to accurately determine translocation breakpoints, translocations were excluded from subsequent analyses. We then analyzed the number of INDELs and SVs of different lengths in IND and JAP, and found that the number of variants gradually decreased as the length of INDELs and SVs increased in both groups (Fig. 1c, d). A similar pattern was observed for small deletions and insertions in INDELs, and deletions and insertions in SVs in both groups (Fig. S1). Notably, INDELs (≤ 10 bp) accounted for 89.18% of the total, while SVs (≤ 500 bp) accounted for 75.96% in IND (Fig. 1c, d). In JAP, INDELs (≤ 10 bp) accounted for 89.24% of the total, and SVs (≤ 500 bp) accounted for 72.88% (Fig. 1c, d). These results indicate that small INDELs (≤ 10 bp) and SVs (51–500 bp) may play important roles in the improvement of IND and JAP.

Fig. 1.

Length and Number of INDELs and SVs in IND and JAP. a–b Number of different types of INDELs (a) and SVs (b). For INDELs, DEL: deletion (≤ 50 bp), INS: insertion (≤ 50 bp). For SVs, DEL: deletion (> 50 bp), INS: insertion (> 50 bp), DUP: duplication, INV: inversion, TRA: translocation. c–d Number of INDELs (c) and SVs (d) of different lengths. The blue rectangles and solid lines represent the IND group (indica), and the red rectangles and solid lines indicate the JAP group (temperate japonica and tropical japonica)

Distribution of INDELs and SVs

To investigate the distribution patterns of INDELs and SVs during the improvement of IND and JAP, we first analyzed the number of these two types of variants on each chromosome in both groups. We found that IND contained an average of 67,637 INDELs per chromosome (Fig. 2a). Chromosome 1 had the highest number of INDELs (92,850), while chromosome 10 had the lowest (48,441), representing a nearly twofold difference between these two chromosome (Fig. 2a). For SVs, IND had an average of 1,305 per chromosome, with chromosome 1 showing the highest count (1,915). Chromosomes 9 and 10 had significantly fewer SVs, with only 962 and 980, respectively (Fig. 2b). An average of 54,399 INDELs were identified per chromosome in JAP (Fig. 2a). Chromosome 1 also had the highest number of INDELs (78,073), while chromosome 9 had the lowest (37,373) (Fig. 2a). For SVs, JAP averaged 852 per chromosome, with chromosome 11 containing the most (1,325) and chromosome 9 the fewest (554). In JAP, the chromosomes with the highest numbers of INDELs and SVs both had more than twice the number found on the chromosomes with the lowest counts (Fig. 2b). A similar pattern was observed for small insertions and deletions (INDELs) as well as for insertions and deletions (SVs) in both groups (Fig. S2). Subsequently, we used the chi-square test to assess whether the distribution patterns of INDELs and SVs across chromosomes were consistent between IND and JAP. The results showed significant differences in the distribution of INDELs numbers on all chromosomes except chromosome 6, and significant differences in SVs numbers on nine chromosomes between the two groups (Table S2). These findings indicate that there are significant differences in the distribution of INDELs and SVs between IND and JAP, suggesting these two groups exhibit distinct variant patterns during improvement.

Fig. 2.

Genomic distribution of INDELs and SVs in IND and JAP. a–b Number of INDELs (a) and SVs (b) on different chromosomes. c–d Distribution of INDELs (c) and SVs (d) hotspot regions on chromosomes. Gray areas indicate centromeric regions. e–f Proportion of INDELs (e) and SVs (f) in different genomic regions

To further explore whether INDELs and SVs were uniformly distributed across the chromosomes of IND and JAP, we analyzed their distribution patterns. The results revealed that both INDELs and SVs were non-uniformly distributed across the chromosomes in both groups (Fig. 2c, d). In IND, we identified a total of 36 INDELs hotspots and 39 SVs hotspots (Fig. 2c, d). Chromosome 4 contained six INDELs hotspots, while chromosomes 3, 5, 7, and 10 each had only one. The longest INDELs hotspots were detected on chromosomes 4 and 8, reaching 3.5 Mb. For SVs, chromosomes 2 and 12 each contained six hotspots, while no SVs hotspot was detected on chromosome 3. The longest SVs hotspots were located on chromosomes 8 and 11, reaching 3.0 Mb. In JAP, we identified 29 INDELs hotspots and 34 SVs hotspots (Fig. 2c, d). Chromosome 11 contained six INDELs hotspots, whereas chromosomes 3, 5, 6, and 9 had only one each. The longest INDELs hotspot, spanning 7.5 Mb, was located on chromosome 11. For SVs, chromosome 7 contained five hotspots, while chromosomes 3, 5, and 9 had one each. The longest SV hotspot, also 7.5 Mb in length, was found on chromosome 11. Notably, the longest INDELs hotspot and the longest SVs hotspot in JAP were both located on chromosome 11, with completely overlapped intervals (Fig. 2c, d), suggesting that this region is a highly enrichment zone for both INDELs and SVs. Analysis of the genes within this region, we identified 48 cloned genes based on China Rice Data Center (https://www.ricedata.cn/). Among these, 33 genes (~ 70%) are associated with plant responses to biotic and abiotic stresses (Table S3), indicating that this hotspot region plays a crucial role in the environmental adaptation of JAP during genetic improvement. When comparing the hotspots identified between IND and JAP, we found only 10 overlapping INDEL hotspots, accounting for 27.78% (IND) and 34.48% (JAP) of the total hotspots, respectively (Fig. S3a). Similarly, only 12 overlapping SVs hotspots were identified, representing 30.77% (IND) and 35.29% (JAP) of the total (Fig. S3b). These results indicate distinct distributions of hotspots between IND and JAP, further suggesting unique variant patterns during the improvement of IND and JAP.

Additionally, we analyzed the genomic distribution of INDELs and SVs in IND and JAP. There were 54.37% of INDELs located in genic regions (2 kb upstream and downstream regions of gene, 5’ and 3’ UTR, exons, and introns), while 45.63% were in intergenic regions in IND (Fig. 2e). Among the INDELs in genic regions, 71.59% were distributed within the 2 kb flanking regions. Similarly, 58.52% of SVs were located in genic regions, and 41.48% in intergenic regions (Fig. 2f). Of the SVs in genic regions, 75.36% were found within the 2 kb flanking regions. In JAP, 53.81% of INDELs were located in genic regions, and 46.19% in intergenic regions (Fig. 2e). Among the genic regions, 71.53% of INDELs were distributed within the 2 kb flanking regions. Similarly, 58.20% of SVs were located in genic regions, and 41.80% in intergenic regions (Fig. 2f). Of the SVs in genic regions, 74.26% were distributed within the 2 kb flanking regions. These results indicate that a substantial number of INDELs and SVs are enriched in the regulatory regions of both IND and JAP, suggesting that INDELs and SVs in these regulatory regions may play important roles in the genetic improvement of both groups.

The impact of INDELs and SVs on phenotypes in IND and JAP improvement

To investigate the impact of INDELs and SVs on phenotypes during the improvement of IND and JAP varieties, we selected six traits for analysis based on previous research [26], including heading date, plant height, panicle number, flag leaf length, flag leaf width, and thousand grain weight (Table S4). We summarized the phenotypic data for these six traits in IND and JAP. As shown in the figures, all traits exhibited continuous distribution in both groups, suggesting that each trait is controlled by multiple loci (Fig. 3a–f). Subsequently, we performed GWAS using INDELs and SVs genotypic data along with the each of phenotypic data. In IND, for the GWAS of INDELs, we identified 331 significant loci for heading date, 604 for flag leaf length, 565 for flag leaf width, 323 for panicle number, 931 for plant height, and 468 for thousand grain weight (Figs. 3g and 4a; Fig. S4). For SVs, we identified 119 significant loci for heading date, 56 for flag leaf length, 102 for flag leaf width, 120 for panicle number, 84 for plant height, and 56 for thousand grain weight (Figs. 3h and 4d; Fig. S5). In JAP, for the GWAS of INDELs, we identified 322 significant loci for heading date, 787 for flag leaf length, 77 for flag leaf width, 406 for panicle number, 288 for plant height, and 116 for thousand grain weight (Figs. 3i and 5a; Fig. S6). For SVs, we identified 37 significant loci for heading date, 31 for flag leaf length, 33 for flag leaf width, 63 for panicle number, 82 for plant height, and 40 for thousand grain weight (Figs. 3j and 5d; Fig. S7). Furthermore, clumping analysis revealed that 67.8% of INDELs and 94.4% of SVs in IND, as well as 52.9% of INDELs and 95.8% of SVs in JAP were independent significant associated loci (Table S5). We then compared the significant loci detected for the same traits between IND and JAP. Among the six traits, only one shared significant INDEL was detected for flag leaf length between the two groups. Similarly, only one shared significant SV was detected for panicle number. Additionally, we compared the significant loci identified for the six traits within the IND and JAP groups. In IND group, 25 shared significant INDELs were identified across at least two traits, accounting for 0.78% of the total (Fig. 3g). For SVs, 43 shared significant loci were detected, accounting for 8.72% of the total (Fig. 3h). In JAP group, 28 shared significant INDELs were identified across at least two traits, accounting for 1.42% of the total (Fig. 3i). For SVs, 32 shared significant loci were detected, accounting for 12.70% of the total (Fig. 3j). These findings further indicate that the genetic basis underlying the same traits in IND and JAP improvement may be determined by distinct INDELs and SVs and the genetic basis (INDELs and SVs) of different traits varies also significantly within the same group.

Fig. 3.

Phenotype and GWAS analysis. a–f Statistical analysis of six traits in IND and JAP, including heading date (a), flag leaf length (b), flag leaf width (c), plant height (d), panicle number (e), and thousand grain weight (f). g–h Number of significant loci identified for six traits based on INDELs (g) and SVs (h) in IND group. i–j Number of significant loci identified for six traits based on INDELs (i) and SVs (j) in JAP group. DH (heading date), FLL (flag leaf length), FLW (flag leaf width), PN (panicle number), PH (plant height) and TGW (thousand grain weight)

Fig. 4.

GWAS analysis for flag leaf width and heading date in IND. a Manhattan plot of INDELs and flag leaf width association in 142 IND varieties. Green dots indicate significant loci located in 2 kb upstream, 5’ UTR, and coding regions of genes. Red dots imply the subset of green dots where the function of involved genes have been previously reported. Red dashed lines represent the significant threshold. b Location of a 1 bp insertion in TAD1. c Comparison of flag leaf width between different haplotypes of TAD1 (with and without insertion). d Manhattan plot of SVs and heading date association in 142 IND varieties. The green dots, red dots and red dashed lines in panel (d) share the same meaning as those in panel (a). e Location of a 237 bp deletion in RCN2. f Comparison of heading date between different haplotypes of RCN2 (with and without deletion). Statistical significance was calculated by Student’s t-test; ^**p < 0.01, ^***p < 0.001

Fig. 5.

GWAS analysis for plant height and flag leaf length in JAP. a Manhattan plot of INDELs and plant height association in 91 JAP varieties. Green dots indicate significant loci located in 2 kb upstream, 5’ UTR, and coding regions of genes. Red dots imply the subset of green dots where the function of involved genes have been previously reported. Red dashed lines represent the significant threshold. b Location of a 6 bp deletion in PLS2. c Comparison of plant height between different haplotypes of PLS2 (with and without deletion). d Manhattan plot of SVs and flag leaf length association in 91 JAP varieties. The green dots, red dots and red dashed lines in panel (d) share the same meaning as those in panel (a). e Location of a 242 bp deletion in OsYLC2. f Comparison of flag leaf length between different haplotypes of OsYLC2 (with and without deletion). Statistical significance was calculated by Student’s t-test; ^*p < 0.05, ^***p < 0.001

Important genes associated with INDELs and SVs

To explore the potential genes influenced by INDELs and SVs in IND and JAP groups improvement, we further screened the significant loci obtained for the six traits. First, we retained significant loci located in the 2 kb upstream region, 5’ UTR, and coding regions of genes, as indicated by the green dots in the figures (Figs. 4 and 5; Figs. S4–7; Tables S6–9). Second, by integrating functional genomic annotation information (funRiceGenes, https://funricegenes.github.io/), significant loci located within cloned genes were selected for subsequent analysis. Finally, based on literature review, we focused on significant loci involved in functional genes related to the target traits. For the INDELs association results in IND, we identified two cloned genes for flag leaf width, seven for plant height, and two for thousand grain weight (Fig. 4a; Fig. S4d, e). For the SVs association results in IND, we found two cloned genes for heading date (Fig. 4d). For the INDELs association results in JAP, we identified one cloned gene for flag leaf width and two for plant height (Fig. 5a; Fig. S6c). For the SVs association results in JAP, we found one cloned gene for flag leaf length (Fig. 5d). No shared cloned genes were detected for the same traits between IND and JAP, suggesting heterogeneity in the genetic basis underlying these traits between these two groups.

Here, we selected one gene and its variant locus from the INDELs and SVs association results of IND and JAP for detailed description, respectively. Previous research showed that the TAD1 mutant exhibit narrower leaves [42]. In this study, we found a 1 bp insertion at 911 bp upstream of TAD1 in some IND materials compared to the reference genome (Fig. 4b). Based on the presence or absence of this insertion, we divided the materials into two haplotypes. Statistical analysis showed that materials with the 1 bp insertion (Hap2) had significantly smaller flag leaf width than those without the insertion (Hap1) (Fig. 4c), suggesting that the 1 bp insertion in the TAD1 promoter region may reduce flag leaf width in IND. For heading date, Nakagawa et al. found that overexpression of RCN2 delays flowering time in rice [43]. In this study, we identified a 237 bp deletion at 815 bp upstream of RCN2 in some IND materials (Fig. 4e). We divided the materials into two haplotypes based on the presence or absence of this deletion. Results showed that materials with the 237 bp deletion (Hap2) flowered significantly earlier than those without the deletion (Hap1) (Fig. 4f), suggesting that the 237 bp deletion in the RCN2 promoter region may promote earlier flowering. For plant height, previous studies indicated that PLS2 positively regulates plant height in rice [44]. In this study, we found a 6 bp deletion at 530 bp upstream of PLS2 in some JAP materials compared to the reference genome (Fig. 5b). The materials were divided into two haplotypes based on the presence or absence of this deletion. Statistical analysis showed that materials with the 6 bp deletion (Hap2) had significantly lower plant height than those without the deletion (Hap1) (Fig. 5c), suggesting that the 6 bp deletion in the PLS2 promoter region may reduce plant height. For flag leaf length, Li et al. found that OsYLC2 is most highly expressed in leaf and shoot apical meristems, and OsYLC2 mutants exhibit affected leaf development [45]. In this study, we identified a 242 bp deletion at 444 bp upstream of OsYLC2 in some JAP materials (Fig. 5e). The materials were divided into two haplotypes based on the presence or absence of this deletion. Results showed that materials with the 242 bp deletion (Hap2) had significantly longer flag leaves than those without the deletion (Hap1) (Fig. 5f), suggesting that the 242 bp deletion in the OsYLC2 promoter region may increase flag leaf length. Furthermore, we performed haplotype analysis on other genes with known functions and found that the variant sites in these functionally characterized genes contribute to the phenotype (Figs. S8–9). These results indicate that INDELs and SVs may play important roles in IND and JAP improvement by altering gene function and thereby influencing key agronomic traits.

Discussion

Genomic pattern of INDELs and SVs in IND and JAP

Different types of DNA mutations provide the raw material for phenotypic variation. As important sources of genetic variation, INDELs and SVs contain more abundant variation information compared to SNPs. With the advancement in sequencing technologies and the optimization of algorithms, they have received increasing attention in recent years. Previous studies have shown that INDELs and SVs play crucial roles in species differentiation, local adaptation, and phenotypic diversity. However, research analyzing variation patterns (number, size, location, etc.) at the whole genome level remains limited [11, 46–48]. In particular, the study about systematic analyses the genomic patterns of both INDELs and SVs are rarely. This study reports the genomic variation patterns of INDELs and SVs in improved cultivars of IND and JAP. We found that INDELs of ≤ 10 bp and SVs of 51–100 bp were the most abundant in both IND and JAP varieties (Fig. 1c, d), suggesting that small INDELs and SVs play a major role in genomic variation. Furthermore, we observed that INDELs and SVs are clustered on the chromosomes (Fig. 2c, d), which is consistent with previous studies [11, 49]. Notably, we identified a 7.5 Mb hotspot region on chromosome 11 and found that this region contains a large number of genes related to biotic and abiotic stress, implying that this hotspot may have facilitated environmental adaptation during rice improvement. Some studies have found that SVs are predominantly located in genic regions [11], while others have reported that SVs are mainly concentrated in intergenic regions [48]. Our study analyzed the genomic distribution of INDELs and SVs in IND and JAP cultivars indicating that both types of variations are predominantly located in genic regions in both groups (IND-INDELs: 54.37%; IND-SVs: 58.52%; JAP-INDELs: 53.81%; JAP-SVs: 58.20%) (Fig. 2d–f). Further analysis revealed that INDELs and SVs are primarily concentrated within the 2 kb upstream and downstream regions of genes, suggesting that variations in regulatory regions play a significant role in rice improvement.

Distinct INDELs and SVs determine the genetic basis of improved IND and JAP varieties

Rice comprises two subspecies, Indica and Japonica, which can be further divided into six variety groups. Among these, IND and JAP (TEJ and TRJ), as the most widely cultivated groups, are the primary components of improved varieties. These two groups exhibit significant differences in morphological characteristics, physiological traits, and geographical distribution [20]. Although previous studies have suggested substantial genetic differentiation between IND and JAP [19, 23, 24], most research has focused on SNPs rather than INDELs and SVs, which have a greater impact on genome [50, 51]. Therefore, comparing the genetic basis underlying the improvement of IND and JAP varieties using INDELs and SVs is of great importance. In this study, we performed GWAS using phenotypic data for six agronomic traits and identified 3,690 significant loci in IND group (INDELs: 3,197; SVs: 493) (Fig. 3g, h) and 2,221 significant loci in JAP group (INDELs: 1,968; SVs: 253) (Fig. 3i, j). When comparing shared significant loci between the two groups, only seven significant INDELs were shared between IND and JAP, accounting for merely 0.22% of the IND and 0.36% of the JAP INDELs, respectively (Table S10). Similarly, only four significant SVs were shared, representing just 0.81% of the IND and 1.58% of the JAP SVs, respectively (Table S10). Furthermore, we analyzed variations shared for each trait. One shared INDEL for flag leaf length and one shared SV was found for panicle number between IND and JAP groups; no shared SVs or INDELs were detected for other traits (Table S10). These results indicate that IND and JAP groups have independent genetic architectures for these six traits during the improvement, indirectly supporting the view of a high level of genetic differentiation between them. In addition, we compared the genetic basis among the six traits within each group. In IND group, we identified 25 significant INDELs shared among traits, primarily concentrated between flag leaf width and plant height (12) (Fig. 3g, Table S11). We also identified 43 significant SVs shared among traits, mainly between heading date and flag leaf width (22) (Fig. 3h, Table S12). In JAP group, we identified 28 significant INDELs shared among traits, primarily between heading date and flag leaf length (9), and between plant height and flag leaf length (9) (Fig. 3i, Table S13). 32 significant SVs shared among traits were identified, mainly between heading date and plant height (11) (Fig. 3j, Table S14). Notably, there were no overlap even among the trait pairs with the highest proportion of shared variations between the IND and JAP, further suggesting independent genetic basis for the two groups.

INDELs and SVs play an important role in IND and JAP improvement

Rice, as one of the most important food crops, has undergone rapid changes in agronomic traits due to strong artificial selection, despite the period of its improvement being much shorter than its domestication. To date, studies have reported the roles of INDELs and SVs by impacting some traits such as heading date, plant height, yield, and biotic and abiotic stress responses in rice improvement. For example, a 12 bp insertion in the promoter of NOG1 enhanced its expression level, thereby increasing grains number per panicle and yield in rice [52]. A 29 bp deletion in the OsTCP regulatory region determined the nitrogen response level among different rice germplasms [53]. A 520 bp deletion in the promoter of DNR1 reduced gene expression and improved nitrogen uptake efficiency in rice [54]. A 4 kb insertion at 53 bp upstream of spd6 disrupted gene function and reduced grain length in rice [49]. As the two major cultivar groups in improved rice (IND and JAP), identifying the genetic basis of INDELs and SVs underlying agronomic trait can provide more useful genetic information for breeding. In this study, we discovered potential new functional mutations in previously cloned genes using INDELs- and SVs-based GWAS. For instance, in IND group, a 1 bp insertion at upstream of TAD1, which controls flag leaf width (Fig. 4a–c), and a 237 bp deletion at upstream of RCN2, which regulates heading date (Fig. 4d–f). In JAP group, a 6 bp insertion at upstream of PLS2, which controls plant height (Fig. 5a–c), and a 242 bp deletion at upstream of OsYLC2, which regulates flag leaf length (Fig. 5d–f). Additionally, several genes with reported functions were identified, and the traits controlled by these genes were consistent with those mapped in this study. For example, based on INDELs-GWAS analysis, we identified seven genes related to plant height and two genes related to thousand grain weight in IND group (Fig. S4d, e), and one gene related to flag leaf width in JAP group (Fig. S6c). These results demonstrate that INDELs and SVs play important roles in the genetic improvement of IND and JAP varieties by altering gene function and influencing key agronomic traits. The novel mutations and functional genes identified in this study will provide valuable genetic resources for rice breeding.

Conclusions

This study utilized resequencing data from IND and JAP to identify INDELs and SVs across the whole genome. We then analyzed the genetic architecture of these variations, including the size, number, distribution, and relative positions to genes in the two groups. Exploiting phenotypic data from six agronomic traits, we identified a large number of significantly associated loci through GWAS in each group. By integrating existing functional reports, we pinpointed potential candidate genes and key mutation site. Furthermore, we systematically compared the significantly associated INDELs and SVs between IND and JAP. These results not only elucidate the variation patterns of INDELs and SVs in the improvement of IND and JAP, but also provide abundant genetic resources and candidate genes for rice breeding.

Authors’ contributions

Jilong Li, Jingcheng Li, and Lei Chen conceived and designed the research. Jilong Li, Hu Li, and Zhigan Xie performed the data analysis. Yinghua Pan, Yihao Xing, Renmin Liang, Zhangbo Liao, Hongping Jiang, and Qi Zhong participated in the material preparation. Jilong Li, Jingcheng Li, and Lei Chen wrote the paper. All authors read and approved the final manuscript.

Funding

This work was funded by the Natural Science Foundation of Guangxi Province of China (2025GXNSFAA069784 and 2025GXNSFAA069272), the Key Project of Guangxi Natural Science Foundation (Guike AA22068087-6), the National Natural Science Foundation of China (31900198), the Guangxi Key Research and Development Program (Guikenong AB2506910007), the Base Business Project of Guangxi Academy of Agricultural Sciences (Guinongke 2026YP106, Guinongke 2026YT076, Guinongke 2026YP020, and Guinongke2025YP033).

Data availability

All data supporting the findings of this study are included in the paper and its Supplementary Materials. Additional datasets generated and analyzed are available from the corresponding author upon reasonable request. The resequencing data for the 233 accessions were obtained from the European Nucleotide Archive under project PRJEB6180. The corresponding phenotypic data was sourced from the Rice Functional Genomics & Breeding database. A complete list of the accession numbers is provided in Table S1.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.