Introduction

Rice is a major food crop globally, and its consumption is projected to rise significantly by 2025, from 503.5 million tonnes to 800–900 million tonnes (Resilience 2017). The introduction of hybrid rice and semi-dwarf cultivars has already led to a substantial increase in rice production. However, despite reaching desired production levels in recent decades, rice yield growth has plateaued (Xu et al. 2015; Zhu et al. 2017). Therefore, it is necessary to intensify efforts to boost the yield potential of rice as the world’s rapid population growth continues to pose a serious threat to food security. One of the key objectives of rice breeding remains the development of high-yielding improved cultivars. Grain yield is a complex quantitative trait, influenced by three major traits—panicle number, grain number per panicle, and grain weight—as well as genotype-environment (GE) interactions (Oladosu et al. 2017; Wang et al. 2020). Grain number per panicle is a major trait linked to grain yield, and understanding its genetic basis is useful for the development of higher-yielding rice cultivars. The genetic basis of grain number per panicle has been extensively studied in rice, and several major quantitative trait loci (QTL) or genes associated with grain numbers per panicle have been identified, including GN1a (Ashikari et al. 2005), NOG1 (Huo et al. 2017), LOG (Kurakawa et al. 2007), LP/EP3 (Li et al. 2011a), GNP1 (Wu et al. 2016), qGN4-1 (Singh et al. 2018), GNP 4 (Zhang et al. 2011), NAL1 (Fujita et al. 2013), GSN1 (Guo et al. 2018), APO 1 (Ikeda-Kawakatsu et al. 2009), GHD 7 (Xue et al. 2008), PAY1 (Zhao et al. 2015), OsSPL14 (Miura et al. 2010), DEP 1 (Huang et al. 2009), TAW1 (Yoshida et al. 2013), SP 1 (Li et al. 2009) and RCN 1 (Nakagawa et al. 2002) in across all 12 chromosomes.

Despite extensive research, the mechanism underlying grain number trait formation remains unclear. Traditional QTL mapping studies require a large breeding population and numerous markers, which can be time-consuming and labor-intensive (Wang et al. 2019; Weng et al. 2021). In contrast, bulked segregant analysis (BSA) (Michelmore et al. 1991) offers a simple and cost-efficient approach to rapidly identify the polymorphic markers associated with traits of interest. The availability of the whole-genome sequence of rice, combined with advances in second and third-generation sequencing, enables the introduction of innovative genomics-driven breeding strategies (Yano et al. 2016). Furthermore, QTL-seq has emerged as a useful method for detecting QTL when it is integrated with traditional BSA and sequencing technological advances (Takagi et al. 2013). Numerous studies have demonstrated that QTL-seq, which involves tracing QTL from whole-genome resequencing of two DNA bulks of progeny with extreme phenotypes, offers a faster and more cost-effective approach compared to traditional QTL mapping (Jia et al. 2023; Pujol et al. 2019b; Vogel et al. 2021; Yang et al. 2021; Yuan et al. 2015). QTL-seq has been extensively used for the detection of QTL for many traits, including grain elongation (Arikit et al. 2019), plant height (Zhang et al. 2021a), panicle grain number (Ma et al. 2022), brown plant hopper resistance (Wang et al. 2022), and salt tolerance (Gao et al. 2023) in rice.

In our previous study, we investigated the genetic variation of grain number per panicle and its related traits using a diverse set of rice genotypes and identified high and low grain number per panicle genotypes of CB12132 (423.00 ± 12.56) and IET 28835 (108.00 ± 11.45) (Gunasekaran et al. 2023). To further elucidate the genetic and molecular basis of grain number per panicle, this study employed QTL-seq, a combination of BSA and whole genome resequencing. Our objective is to identify genomic regions associated with grain number per panicle and pinpoint potential candidate genes that can useful for rice breeding programs aimed at improving grain number per panicle in rice.

Materials and methods

Plant materials and generation of segregating population

In the present study, we used an F2 segregating population composed of 256 individuals, derived from a cross between two Oryza sativa L. ssp. indica genotypes: CB12132 (High grain number) and IET28835 (Low grain number). To prepare the F2 segregating population, IET28835 plants were crossed with CB12132 in the summer of 2020, and F1 plants were obtained. After confirming the true F1s, the plants were selfed to produce the F2 population in the summer of 2021. All field experiments were conducted at the Paddy Breeding Station, Centre for Plant Breeding and Genetics, Tamil Nadu Agricultural University, Coimbatore, India. Each plant had a plant-plant distance of 20 cm × 20 cm, respectively. The 2012 edition of the TNAU-Crop Production Guide was followed for agronomic practices and plant protection techniques to ensure healthy crop growth.

Trait evaluation and construction of bulks

The parents and the F2 segregating population were subjected to trait evaluation and used to construct bulks. Traits such as flag leaf area (FLA), panicle length (PL), number of spikelets per panicle (NOSPP), number of grains per panicle (NOGPP), number of primary branches (NOPB), number of secondary branches (NOSB), number of secondary branches per primary branch (NOSBPB), number of spikelets in primary branches (NOSIPB), and number of spikelets in secondary branches (NOSISB) were measured agreeing to IRRI’s standard evaluation system (SES) (IRRI, 2014). In F2 plants, the first-formed primary panicle was tagged in each plant, and at physiological maturity, it was collected and used to count the number of grains per panicle. From a total of 256 F2 plants of CB12132 × IET28835 cross, 15 plants each with a high grain number per panicle and low grain number per panicle were selected to construct two bulks, referred to as the high grain number per panicle-bulk (HGNPP-B) and low grain number per panicle-bulk (LGNPP-B), respectively. F2 individuals from the two groups were selected, and DNA was isolated by the CTAB method (Murray and Thompson 1980). The extracted DNA of the 15 individual plants in each group was mixed at equal molar concentrations and pooled together to construct the bulks.

Library preparation and whole-genome resequencing

We prepared four genomic libraries, including two parents (CB12132 and IET28835) and two extreme bulks (HGNPP-B and LGNPP-B) using Nextera XT DNA library preparation kit (Illumina, Inc., San Diego, CA, USA). The libraries had a size of 350b and were sequenced on the Illumina NovaSeq 6000 (Illumina, Inc., San Diego, CA, USA) with a pair-end read length of 150 bp. The raw sequence data has been deposited in the NCBI Sequence Read Archive database (https://www.ncbi.nlm.nih.gov/sra) under the following Bio project and SRA accession numbers; PRJNA1094905, and SRX24115668- SRX24115671.

Data analysis via the QTL-seq pipeline

We performed quality control and preprocessing on the raw reads from four samples (CB12132, IET28835, HGNPP-B, and LGNPP-B) using Trimmomatic v0.39 (Bolger et al. 2014). The clean reads were then mapped to the reference genome, Nipponbare (Oryza_sativa_IRGSP-1.0, accessed on 10.02.2024) using Burrows-Wheeler Aligner (BWA) software (Version 0.7.17) (Li and Durbin 2009). To ensure that only uniquely mapped reads retained. We converted the SAM file to BAM format, sorted and indexed it using samtools (version 1.17), and removed duplicates using Picard Tools (version 3.00) (https://broadinstitute.github.io/picard/). The processed BAM files were used as input for QTLseq analysis using the pipeline reported by Sugihara et al. (2022) and Takagi et al. (2013). We calculated the SNP index for each SNP position in the HGNPP-B and LGNPP-B samples based on the count of reads harboring SNPs similar or dissimilar to the reads of CB12132. If all reads in the HGNPP-B was identical to CB12132, the SNP index was 0, and if all reads were different, the SNP index was 1. The ΔSNP index of each SNP was estimated according to the formula: ΔSNP index = SNP index of LGNPP-B – SNP index of HGNPP-B. The SNP index distribution among the 12 chromosomes was analysed using the sliding window method by setting 1 Mb of window size and 50 kb of increment steps. Genomic regions with an average ΔSNP index value greater than the surrounding region with a 95% or 99% confidence level were considered effective. Additionally, Genome Analysis Toolkit (GATK), Haplotype Caller was used to predict the SNP and InDel variation (McKenna et al. 2010). SnpSift and SnpEff (version 5.1) were used for the filtering (parameters: QUAL >  = 30 and MQ >  = 30 and DP >  = 10) and annotation of SNPs and InDels (Cingolani et al. 2012).

Search for candidate genes

We employed a three-stage approach to identify the possible candidate genes. First, we collected genes within the specific genomic region. Next, we pinpointed gene positions, sorted genes with nonsynonymous SNPs and InDels, and annotated them using the Rice Annotation Project Database (https://ricexpro.dna.affrc.go.jp accessed on 09.03.2024). We excluded genes labeled as ‘(retro) transposon’, ‘hypothetical’, or ‘unknown’ from further analysis. Eventually, expression pattern of selected genes in flag leaf and inflorescence developmental stages were analysed using the RiceXpro database (https://ricexpro.dna.affrc.go.jp accessed on 09.03.2024). By integrating these analyses with literature knowledge, we prioritized the possible genes associated with grain numbers per panicle in each candidate genomic region.

Quantitative real-time PCR (qRT-PCR) analysis

Fresh, young inflorescence (0.5 cm length at the branch primordium stage) were collected from various plants of CB12132 and IET28835. A 0.5 g sample was used for total RNA isolation, which was performed using the TRIzol® reagent kit (Invitrogen, Carlsbad, CA, USA) following to the user guidelines. We used the iScript cDNA synthesis kit (Bio-Rad Laboratories, Inc., Hercules, CA, USA) to convert DNA-free RNA from high and low grain number parents into high-quality cDNA, following the product guidelines. High-quality cDNA was used as the template for qRT-PCR analysis, and reactions were performed on the Bio-Rad CFX 96 (Bio-Rad Laboratories, Inc., Hercules, CA, USA). Each 20 μL PCR reaction contained 1 μL of cDNA, 1 μL of each forward and reverse primers, 10 μL AccuPower ® 2X GreenstarTM SYBR Green Master mix, and 6 μL dd H2O. We used a 2−ΔΔCt method to analyse the relative expression of the target genes, using OsActin1 (Han et al. 2024) as the internal control. In addition, all the samples were amplified at three biological replications, with three technical replications at each biological replication. The primer sequences were listed in supplementary Table 1.

Results

Phenotypic variation in F2 population and construction of bulks

The high and low grain number per panicle parents, CB12132 and IET28835 were crossed and advanced to F1 and F2 generation. The evaluation of traits such as, FLA, PL, NOSPP, NOGPP, NOPB, NOSBPB, NOSIPB, and NOSISB showed significant differences between parents and its F1 generation (Table 1). In the F1 generation, the grain number per panicle had an intermediate value of 187.50 ± 12.56 compared to CB12132 (423.00 ± 12.56) and IET 28835 (108.00 ± 11.45). In the F2 generation, 225 individual plants were evaluated, and the number of grains per panicle ranged from 89 to 420. The majority of plants had grain number per panicle ranging between 150 and 200. A total of 15 F2 plants with grain number per panicle ranging from 355 to 420 were selected to generate the HGNPP-B, while 15 F2 plants with grain number per panicle ranging from 89 to 125 were used to produce the LGNPP-B. The two parents and high and low-grain numbers F2 individuals are presented in Fig. 1.

Table 1 The phenotypic observation on grain number per panicle and its related traits
Full size table
Fig. 1
figure 1

Grain number per panicle distribution in the F2 generation derived from the cross between CB12132 and IET28835. Parents and F2 individuals (A), F2 individuals with a high grain number per panicle (B), and F2 individuals with a low grain number per panicle (C). Note: HGNPP-B stands for high grain number per panicle-bulk, and LGNPP-B stands for low grain number per panicle-bulk

Full size image

Whole-genome resequencing of parental lines and HGNPP and LGNPP bulks

Sequencing of CB12132, IET28835, HGNPP-B and LGNPP-B yielded 60.45, 56.40, 60.00 and 48.39 million reads, respectively (Table 2). After filtering, we obtained the following number of high-quality clean reads for each sample: 60.07 million for CB12132, 55.98 million for IET28835, 59.40 million for HGNPP-B, and 48.08 million for LGNPP-B. Alignment with the reference genome produced 8.98 Gb and 8.16 Gb of clean data (22.74X and 21.42X coverage) for CB12132 and IET28835, and 8.87 Gb, and 7.19 Gb clean data (21.53X and 18.36X coverage) for HGNPP-B and LGNPP-B. The number of genome-wide SNPs and InDels identified in HGNPP-B and LGNPP-B after mapping cleaned reads onto the reference genome were 877,063 and 846,524, respectively. Using a criterion of read depth exceeding 15 in both HGNPP-B and LGNPP-B and a SNP index surpassing 0.30 in at least one, a total of 204,923 SNPs and 66,459 InDels commonly detected in both HGNPP-B and LGNPP-B were used for QTL seq analysis (Fig. 2).

Table 2 The overview of whole genome resequencing data of parents (CB12132 and IET28835), and HGNPP-B and LGNPP-B derived from F2 individuals
Full size table
Fig. 2
figure 2

A summary of SNP and InDel variants identified in the high and low bulks mapped to the reference genome (A), and the common SNPs and InDels between the high and low grain number per panicle bulks (B). Note: HGNPP-B stands for high grain number per panicle-bulk, and LGNPP-B stands for low grain number per panicle-bulk

Full size image

Tracing candidate genomic region associated with grain number per panicle using QTLseq

QTL-seq analysis was done to trace the candidate genomic region associated with grain number per panicle. This analysis calculated the ∆ (SNP index) values across the genome in 1 Mb window with a 50 kb increment. It revealed five candidate QTL regions associated with grain number per panicle on chromosomes 1, 4, and 5 (Table 3). Of these, three QTL regions on chromosome 1 designated as qGNPP1.1, qGNPP1.2, and qGNPP1.3 spanned between 10.40 to 12.76 Mb (2.36 Mb), 24.61 to 25.33 Mb (0.72 Mb) and 26.57 to 27.26 Mb (0.68 Mb) with Δ SNP index values of 0.451, 0.459, and 0.233. The QTL on chromosome 4, qGNPP4.1 (3.64 Mb) spanned between 27.70 to 31.34 Mb with Δ SNP index values of 0.471. Another QTL, qGNPP5.1 (3.38 Mb) on chromosome 5, was identified in the region of 2.12 to 5.50 Mb with a Δ SNP index values of 0.452. The SNP index plot and candidate genomic regions across the rice genome are presented in Fig. 3 and 4.

Table 3 Details of candidate genomic regions associated with the grain number per panicle
Full size table
Fig. 3
figure 3

SNP index plots for HGNPP-B (green dotted) (A), LGNPP-B (orange dotted) (B), and Δ SNP index plots for the two bulks (C). Sliding window plots of the average SNP index, using a 1-Mb window size and 50-kb steps, are shown as red lines. The green and orange lines in the Δ SNP index plots represent the 95% and 99% confidence intervals, respectively. Note: HGNPP-B refers to the high grain number per panicle bulk, and LGNPP-B refers to the low grain number per panicle bulk

Full size image
Fig. 4
figure 4

Sliding window plots depict the SNP index for HGNPP-B and LGNPP-B, providing a comparison of the SNP index between the two bulks. The plots include: Pseudomolecules of the Nipponbare reference genome according to IRGSP 1.0 (A), upper probability threshold at 99% confidence level (P < 0.01) (B) and 95% confidence level (P < 0.05) (C), sliding window analysis of the ΔSNP index with a window size of 1 Mb and steps of 50 kb (D), lower probability threshold at 95% confidence level (P < 0.05) (E) and 99% confidence level (P < 0.01) (F), sliding window plots of average SNP index values in the HGNPP-B (G) and LGNPP-B (H) with a 1 Mb window and 50 kb steps. The identified candidate genomic regions for grain number per panicle are highlighted in the circos plot. Note: HGNPP-B stands for high grain number per panicle-bulk, and LGNPP-B stands for low grain number per panicle-bulk

Full size image

Briefing the putative candidate genes associated with grain number per panicle

A total of 199, 87, 73, 508, and 359 genes were identified in the targeted candidate genomic regions qGNPP1.1, qGNPP1.2, qGNPP1.3, qGNPP4.1 and qGNPP5.1, respectively. Further, we identified 23 candidate genes associated with grain number per panicle in five candidate genomic regions based on nonsynonymous SNPs/InDels, expression in inflorescence, and annotation details (Table 4 and Fig. 5). On chromosome 1, the qGNPP1.1 region includes gene such as squamosa-promoter binding-like protein 2 (Os01g0292900), similar to kinase interactor 1 (Os01g0310800), R2R3-MYB transcription factor 6 (Os01g0298400) and polygalacturonase (Os01g0296200). The qGNPP1.2 region contains genes like ATPase, AAA-type, core domain-containing protein (Os01g0623500), myosin tail 2 domain-containing protein (Os01g0621700), and glycosyltransferase (Os01g0622000), while the qGNPP1.3 region contains the TRITHORAX-like protein (Os01g0655300). On chromosome 4, the qGNPP4.1 region encompasses eight possible candidate genes, including the previously reported serine protease (NAL1; Os04g0615000) gene. Meanwhile chromosome 5’s qGNPP5.1 region includes six possible candidate genes such as serine carboxypeptidase (Os05g0158500), 4Fe-4S ferredoxin (Os05g0157300), S-Domain receptor-like kinase-36 (Os05g0166600), ethylene receptor-like (Os05g0155200), KIP1-like domain-containing protein (Os05g0168800) and similar to P-glycoprotein ABCB5 (Os05g0137200). These genes are potential contributors to grain number per panicle and useful for further investigation.

Table 4 Summary of possible candidate genes identified in the genomic region associated with grain number per panicle
Full size table
Fig. 5
figure 5

The expression patterns of 23 possible candidate genes across various reproductive stages and tissues, as obtained from the RiceXPro database (accessed on 09.03.2024)

Full size image

Expression profiling of candidate genes

A total of six candidate genes such as Os01g0292900 (SPL1), Os01g0622000 (OsCUGT1), Os01g0655300 (SDG705), Os04g0615000 (NAL1), Os04g0559800 (SMG2) and Os05g0155200 (ERS2) were selected from the 23 identified genes and their expression pattern were analysed in the young inflorescence of the two parents, CB12132 and IET28835 using qRT-PCR analysis. The results showed that all genes, except for Os04g0559800, were expressed at significantly higher levels in CB12132 compare to IET28835. Figure 6 shows the expression levels of these genes in CB12132 and IET28835.

Fig. 6
figure 6

The expression patterns of six candidate genes were analyzed in high and low grain number per panicle parents (CB12132 and IET28835) using qRT-PCR. Data are presented as the means of three biological replicates, with error bars indicating the standard deviation (SD)

Full size image

Discussion

The grain number per panicle is closely associated with grain yield and is an ideal trait targeted by rice breeders to improve the rice yield. In this study, we used a segregating F2 population derived from the cross between CB12132 and IET28835, to investigate the genetic mechanism of this trait. The F2 population showed a wide distribution for grain number per panicle, ranging from 89 to 420, indicating that the trait governed by additive gene action. This finding is consistent with the reports of Priyanka et al. (2019), Sekhar et al. (2021) and Wang et al. (2024). Furthermore, our study used whole-genome resequencing-based QTL-seq approach integrated with BSA to identify the genomic region regulating grain number per panicle in rice. We used a total of 15 individuals to construct the HGNPP-B and LGNPP-B, which is adequate for detecting the major loci associated with trait of interest, consistent with previous reports (Bommisetty et al. 2023; Huang et al. 2022; Kaur et al. 2022; Singh et al. 2022; Zhang et al. 2021b). The genome coverage for the HGNPP-B and LGNPP-B was 94.17% and 94.18%, respectively, which is within the ranges reported in previous studies: 96.72–96.95%, 93.70–94.39%, 93.51–94.05% and 93.37–95.53% (Gao et al. 2023; Luo et al. 2018; Ma et al. 2022). The average depth was above 20X obtained in both bulks, similar to findings reported in other studies (Nubankoh et al. 2020; Singh et al. 2022 and Takagi et al. 2013). We identified five candidate QTL regions associated with grain number per panicle using QTL-seq. These regions, qGNPP1.1-qGNPP1.3, qGNPP4.1, and qGNPP5.1, were located on chromosomes 1, 4, and 5. Notably, several of these QTLs coincide with or near previously reported QTLs for grain number per panicle. For instance, qGNPP1.2 was positioned 1.5 Mb downstream from LOG1 (Kurakawa et al. 2007), while qGNPP1.3 was near RGN1a (Zhang et al. 2022b), with a distance of 1 Mb. Additionally, qGNPP4.1 overlaps with NAL1 (Fujita et al. 2013) and qGN4.1, and qGNPP5.1 is linked to qSPP5 (Luo et al. 2013). Interestingly, the qGNPP1.1 region appears to be novel, with no reported genes related to grain number in this area.

According to the Nipponbare reference genome sequence, the five genomic regions (qGNPP1.1, qGNPP1.2, qGNPP1.3, qGNPP4.1 and qGNPP5.1) contain a total of 1226 genes. To identify possible candidate genes associated with grain number per panicle, we filtered these genes based on nonsynonymous SNPs/InDels, expression in inflorescence, and annotation. We first narrowed down the list to 294 genes with nonsynonymous SNPs/InDels and then selected 23 possible genes across the five regions using the expression and annotation data. These 23 genes show promise for further investigation into their role in regulating grain number per panicle. In the qGNPP1.1 genomic region (Chr 1; 10.40 Mb to 12.76 Mb), we identified four potential genes: Os01g0292900 (SPL1), Os01g0310800 (WAK4), Os01g0298400 (2R_MYB6), and Os01g0296200 (PSL). Notably, SPL1 belongs to a plant transcription factor family that regulates growth, development, and grain yield. SPL genes increase grain number per panicle by promoting panicle branching and enhance grain length and plant yield by regulating cell size (Jiao et al. 2010; Miura et al. 2010). We found three missense SNPs in SPL1, which may contribute to higher grain number per panicle. Additionally, WAK4, involved in cell expansion (Delteil et al. 2016), may also play a role in regulating panicle development, as loss of WAK function has been linked to decreased florets in rice panicles (Zhang et al. 2022b). In the qGNPP1.2 genomic region (Chr 1; 24.61 Mb to 25.33 Mb), we identified three potential candidate genes: Os01g0623500, Os01g0621700, and Os01g0622000 (OsCUGT1). Notably, OsCUGT1 is a Glycosyltransferase involved in regulating physiological and stress responses in rice. It was previously identified as a cold stress-responsive gene, and its mutant showed a sterile inflorescence phenotype with altered cytokinin metabolism and phenylpropanoid biosynthesis (Zhao et al. 2023). Os01g0655300 (SDG705) is the most possible candidate gene identified in the qGNPP1.3 genomic region (Chr 1; 26.57 Mb to 27.26 Mb). SDG705 belongs to the SET domain family, which is involved in histone modification and various biological processes, including flowering regulation. Enhanced expression of SDG genes promotes florigen genes and Ehd3, critical promoters of rice flowering. Interestingly, overexpression of a similar gene, GhD7, delayed flowering and increased panicle size, leading to enhanced grain yield in rice (Xue et al. 2008).

In the qGNPP4.1 genomic region (Chr 4; 27.70 Mb to 31.34 Mb), we identified nine potential candidate genes, including Os04g0615700 (OsAGO2), Os04g0563900 (OsRLCK157), Os04g0597400 (OsNPF7.6), Os04g0559800 (SMG2), Os04g0598300 (RFL), Os04g0560600 (CDPK12), Os04g0615000 (NAL1), Os04g0555000 (OsHAM3) and Os04g0591900 (OsFbox231). Notably, NAL1 (Os04g0615000) encodes a serine protease, previously reported to regulate grain number per panicle, with two missense SNPs. NAL1 has been extensively studied for its pleiotropic effects on various traits, including grain number per panicle, leaf width, and photosynthetic efficiency (Fujita et al. 2013; Takai et al. 2013; Xu et al. 2015; Yano et al. 2016; Zhang et al. 2014). Other notable candidate genes in this region include: OsNPF7.6 (Os04g0597400), involved in nitrate transportation and uptake, with potential for improving nitrogen utilization efficiency and yield (Zhang et al. 2022a). SMG2 (Os04g0559800), interacting with signaling cascades to modulate grain size (Xu et al. 2018). OsAGO2 (Os04g0615700), regulating anther development by controlling ROS levels and tapetal PCD through DNA methylation-mediated gene expression (Zheng et al. 2019). In the qGNPP5.1 genomic region (Chr 5; 2.12 Mb to 5.50 Mb), we identified six potential candidate genes: Os05g0158500 (GS5), Os05g0157300, Os05g0166600 (SDRLK-36), Os05g0155200 (ERS2), Os05g0168800, and Os05g0137200 (MDR3). Notably, ERS2 (Os05g0155200) encodes an ethylene receptor, previously reported to regulate grain number per panicle. Ethylene plays a crucial role in grain filling, with higher levels at anthesis associated with poorer grain filling (Panigrahi et al. 2023). Another notable candidate gene is GS5 (Os05g0158500), encoding serine carboxypeptidase, which positively regulates grain size and grain yield. These genes may play important roles in regulating grain number per panicle and grain yield (Li et al. 2011b).

In summary, by adopting QTL-seq approach, we identified five candidate genomic regions associated with grain number per panicle using an F2 population derived from CB12132 × IET28835 cross. Additionally, we identified 23 possible candidate genes through analysis of gene sequences, expression, and annotation data sets. Of these, six genes were validated by qRT-PCR analysis. We recommend further investigation of these candidate genes using functional genomics approaches. The genomic regions and genes identified in this study could be valuable for enhancing grain yield in rice breeding programs.

Consent for publication

All authors have read and agreed to the published version of the manuscript. All authors read and approved the final manuscript.

Ethics approval

This article does not contain any studies with human or animal subjects.