Joint linkage and association analysis identifies genomic regions and candidate genes for yield-related traits in wheat

Abstract

Key message

Twenty-six QTLs associated with yield-related traits in wheat were identified through joint linkage and association analysis, with TraesCS5A03G0002500 being selected as a candidate gene for QGl.caas-5A.1.

Abstract

As a major staple crop worldwide, continuously increasing wheat yield is crucial for ensuring food security. Wheat yield is influenced by multiple traits, and elucidating the genetic basis of yield-related traits lays a foundation for future gene cloning and molecular mechanism studies. In this study, a recombinant inbred line (RIL) population derived from 292 lines of Hengguan 35/Zhoumai 18 was genotyped with the Affymetrix wheat 660 K SNP array. Combined with the phenotype of the RIL population in 13 environments, linkage analysis of six yield-related traits including plant height, grain number per spike, thousand-grain weight, grain length, grain width, and grain thickness was conducted. A total of 262 quantitative trait locus (QTLs) (logarithm of odds [LOD] > 3) were identified across 21 chromosomes, in which 50 QTLs were repeatedly detected in more than three environments. Numerous QTLs harbored cloned genes and overlapped with those reported in previous studies. Subsequently, joint analysis of genome-wide association study (GWAS) data from the advanced backcross-nested association mapping plus inter-crossed (AB-NAMIC) population and QTLs identified in the RIL population revealed 26 overlapping genomic regions. Notably, the QGl.caas-5A.1 associated with grain length on chromosome 5A was detected in both the RIL and AB-NAMIC populations, and TraesCS5A03G0002500 was selected as a candidate gene. A kompetitive allele-specific PCR (KASP) marker based on a variant [A/G] in TraesCS5A03G0002500 was developed and validated in a natural population containing 350 accessions. Taken together, these results provide valuable information for fine mapping and cloning of yield-related wheat genes in the future.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00122-025-04900-4.

Introduction

Common wheat (Triticum aestivum L.) is an important staple crop that feeds approximately 40% of the world's population (Xiao et al. 2022). China is the largest wheat producer and consumer in the world (Gupta et al. 2020). Given the increasing world population and decreasing cultivated land area, increasing wheat yield per unit area is an important goal of current breeding programs. Grain yield is a comprehensive trait that is affected by multiple factors. Generally, thousand-grain weight, grain number per spike, and spike number per unit area are usually defined as the three components of wheat yield. In addition, important agronomic traits, such as heading date and plant height, have different degrees of correlation with the three wheat yield components, indirectly affecting wheat yield (Simmonds et al. 2014; Lopes et al. 2012). Yield-related traits are quantitative traits influenced by multiple genes with minor effects and are susceptible to environmental factors. Linkage analysis and association mapping based on forward genetics are two effective approaches for discovering quantitative trait locus (QTL) (Cadic et al. 2013; Khan et al. 2021).

Numerous QTLs and genes associated with grain yield have been identified in wheat. Specifically, 25 genes influencing plant height have been identified (McIntosh et al. 2020). These genes play pivotal roles in modulating the plant architecture of wheat, which in turn affects grain yield by influencing factors such as lodging resistance and the harvest index (Gale and Youssefian 1985). Notably, RHB-B1, RHT-D1, and RHT8 have been successfully cloned and are the most widely utilized wheat dwarfing genes worldwide (Chai et al. 2022). Efforts to improve grain yield have been constrained by the trade-off between grain weight/size and grain number (Quintero et al. 2018). The number of grains per spike depends on floret fertility and the number of spikelets per spike. Many flowering time-related genes (FT1, Ppd-1, VRN1, and FUL2) affect spikelet and grain numbers per spike by regulating early spike development (Finnegan et al. 2018). Additionally, GNI1 (Sakuma et al. 2019), TaSUT1 (Al-Sheikh Ahmed et al. 2018), TaPIN1 (Wang et al. 2017a, b), and TaSPL14 (Cao et al. 2021), among others, regulate the number of grains per spike in wheat. Many genes controlling grain size have been cloned in rice, such as those involved in the ubiquitin–proteasome pathway like GW2 (Song et al. 2007), WTG1 (Huang et al. 2017), and OsOTUB1 (Wang et al. 2017a, b); G-protein signaling such as GS3 (Sun et al. 2018) and DEP1 (Huang et al. 2009); and phytohormone signaling and homeostasis including GS5 (Li et al. 2011), GSK2 (Tong et al. 2012), and GL3.1 (Qi et al. 2012). Recently, numerous genes related to grain weight/size in wheat, e.g., TaGW2 (Hong et al. 2014), TaDEP1 (Li et al. 2022), and Tasg-D1 (Cheng et al. 2020), have been cloned via reverse genetics strategies. However, few yield-related genes have been cloned through forward genetics, and further research is needed.

Currently, a series of single-nucleotide polymorphism (SNP) arrays have been designed for genomic research and molecular breeding using marker-assisted selection (MAS) in various crop species. For wheat, 9 K and 90 K chips based on Illumina Infinium BeadChip technology, as well as 660 K, 820 K, and 35 K arrays based on Affymetrix Axiom technology, have been successively developed (Rasheed et al. 2017). The 660 K array contains 630,517 SNP loci, with approximately 75.8% of the markers located in intergenic regions and 24.2% in genic regions. This array addresses the disadvantage of limited markers in the D subgenome and has a relatively uniform distribution of markers across the genome. The 660 K array has tremendous potential in the genotyping and molecular marker selection of polyploid wheat (Sun et al. 2020) and has been widely applied in various genetic studies, including comparative genomics (Zhou et al. 2018), linkage map construction and QTL mapping (Cui et al. 2017), and genome-wide association analysis (Yang et al. 2019; Niaz et al. 2023).

Wheat is an allopolyploid with a genome size of ~ 16 Gb and 80% DNA repeats (Xiao et al. 2022). The number of genes cloned from wheat is much lower than those cloned from rice, maize, and other crops. A large number of studies have preliminarily dissected the genetic basis of wheat yield-related traits via linkage analysis or genome-wide association studies (GWAS) and identified numerous QTLs. However, linkage analysis typically requires the construction of a specific genetic population, a process that is time-consuming and constrained by parental mating types. Limited recombination events result in broad mapping intervals, making it difficult to fine-map target genes (Liang et al. 2021). On the other hand, GWAS, although conducted in natural populations, is susceptible to population stratification, which may lead to false positive results if not properly controlled, particularly in the detection of low-frequency and rare variants, where the effectiveness of this method is weaker (OTT et al. 2015; Sazonovs and Barrett 2018). Therefore, a joint linkage analysis with a GWAS, important genomic regions, and critical genes associated with yield-related traits can be rapidly discovered (Bergelson and Roux 2010), laying the foundation for subsequent gene cloning and elucidation of the underlying molecular mechanism.

In our previous studies, a GWAS was performed on 11 yield-related traits using an advanced backcross-nested association mapping plus inter-crossed (AB-NAMIC) population comprising 590 lines. Specifically, the AB-NAMIC population was enriched in beneficial traits, and a large number of substantial GWAS signals were detected (Jiao et al. 2023). In the present study, the wheat 660 K array was used to genotype the recombinant inbred line (RIL) population derived from Hengguan 35/Zhoumai 18, and linkage analysis of yield-related traits was carried out in combination with phenotypic data. The candidate regions for yield-related traits were mapped through comparative analysis of the RIL and AB-NAMIC populations. The objectives of this study were to: (1) construct a high-density genetic map suitable for QTL mapping based on the Affymetrix wheat 660 K SNP array, (2) integrate linkage analysis with GWAS to perform a joint analysis of yield-related traits and identify important genomic regions, (3) predict candidate gene for targeted QTL interval according to reference genome annotation information and expression patterns, and (4) develop practicable kompetitive allele-specific PCR (KASP) marker of the candidate gene and verify the marker in a natural population consisted of 350 accessions for MAS in high-grain length wheat breeding. Overall, this study aimed to provide valuable information for wheat yield molecular breeding by co-anchoring yield-related genomic regions in different genetic populations.

Materials and methods

Plant materials

A mapping population composed of 292 RILs derived from Hengguan 35/Zhoumai 18 was used as the study material. The RIL population was constructed by the single-seed descent method, and a stable F₁₄ family was formed via continuous self-crossing from F₂. Hengguan 35 is a winter wheat cultivar bred from Heng84-Guan749 and Heng87-4263 in Hebei, China; this cultivar has short stalks, large spikes, early maturity, high yield, water savings, and multiresistance. Zhoumai 18, developed by crossing Neixiang 185 with Zhoumai 9 in Henan, China, exhibits characteristics such as a high tillering-to-panicle ratio, medium maturity, high stable yield potential, and wide adaptability. The RIL population generally exhibited moderate resistance to powdery mildew and stripe rust under field conditions. In addition, a natural population of 350 accessions, comprising 113 Chinese landraces, 226 modern Chinese cultivars, and 11 introduced cultivars (Table S1), was used for KASP marker screening.

Field trials and phenotypic evaluation

Field experiments were conducted at the Zhengzhou Experimental Station, Henan (113.7°E, 34.8°N), in 2014; Chinese Academy of Agricultural Sciences (CAAS)-Xinxiang Experimental Station, Henan (113.5°E, 35.2°N), from 2015–2020; and CAAS-Shunyi Experimental Station, Beijing (116.3°E, 40.0°N), from 2015–2020. Both parents and the 292 RILs were planted in those 13 environments. In addition, a natural population consisting of 350 accessions was planted at Xinxiang Experimental Station from 2016 to 2019. Hereafter, the 13 phenotypic identification environments were designated 2014-Zhengzhou, 2015-Xinxiang, 2015-Shunyi, 2016-Xinxiang, 2016-Shunyi, 2017-Xinxiang, 2017-Shunyi, 2018-Xinxiang, 2018-Shunyi, 2019-Xinxiang, 2019-Shunyi, 2020-Xinxiang, and 2020-Shunyi, respectively. Field experiments were grown in a randomized block design with three replications. Each accession was planted in a 2-m four-row plot with 25 cm between rows and 20 seeds per row. Field management followed local practices. The six agronomic traits were investigated in the RIL population according to the Descriptors and Data Standard for wheat (Li and Li 2006), including plant height (PH, cm), grain number per spike (GN), thousand-grain weight (TGW, g), grain length (GL, mm), grain width (GW, mm), and grain thickness (GT, mm). In the natural population, four traits, namely TGW, GL, GW, and GT, were measured. Specifically, at physiological maturity, ten plants from the middle rows in each plot were selected to determine PH and GN. PH was measured as the length from the soil surface to the top of the spike (excluding awns); GN was the total number of grains per spike. After physiological maturity, spikes in the middle two rows were hand-harvested to determine TGW, GL, GW, and GT. Harvested wheat grains were air-dried to a 12% water content and then weighed. Grain-related traits were measured semiautomatically by an SC-G automatic grain analyzer (Hangzhou Wanshen Testing Technology Co., Ltd.).

Phenotypic statistical analysis

The best linear unbiased prediction (BLUP) values for each trait from different environments were calculated using the "lme4" package in R (Bates et al. 2015). The formula of model for BLUP was as follows: Y = μ + Line + Loc + (Line × Loc) + Rep (Loc) + ε, where Y represents the phenotype; μ represents the intercept; Line and Loc denote the respective random effects of the genotype and the environment; Line × Loc is the interaction effect between genotype and environment; Rep (Loc) is the random effect of replication nested in the environment; and ε represents the random residual error effect. Basic descriptive statistics and analysis of variance (ANOVA) were conducted using SPSS v25.0 (https://www.ibm.com/spss). The formula ($H^2$=${\mathrm{\sigma }}_g^2/({\mathrm{\sigma }}_g^2+\frac{{\mathrm{\sigma }}_{\mathit{ge}}^2}{n}+\frac{{\mathrm{\sigma }}_e^2}{\mathit{nr}})$) was used to calculate the broad-sense heritability ($H^2$), where ${\mathrm{\sigma }}_g^2$, ${\mathrm{\sigma }}_{\mathit{ge}}^2$ and ${\sigma }_e^2$ represent the variance components of genotype, interaction between genotype and environment, and error, respectively, r is the number of replicates, and n denotes the number of environments for phenotypic assessment. Pearson’s correlations between traits were calculated using the BLUP values with the "PerformanceAnalytics" package in R (https://github.com/braverock/PerformanceAnalytics). The analysis of normal distribution was conducted with Origin software (https://www.originlab.com/). Additionally, visualization was performed with the "ggplot2" package in R (Villanueva and Chen 2019).

Genotyping and construction of the genetic map

All the lines and both parents of the RIL population were planted in the greenhouse of the Institute of Crop Sciences, Chinese Academy of Agricultural Sciences (116.3°E, 40.0°N). When the seedlings reached the three-leaf stage, partial leaf samples were collected for whole-genome DNA extraction. The wheat genomic DNA was extracted using a fast plant genomic DNA extraction kit (Tiangen Biochemical Technology Co., Ltd., DP321) according to the manufacturer’s instructions. After preliminary testing for DNA quality and purity using a 1% agarose gel, the qualified samples were sent to Beijing CapitalBio Technology Co., Ltd. (http://www.capitalbiotech.com/), and the Affymetrix wheat 660 K SNP array was used to genotype 294 samples. On the basis of the genotyping results, three categories of poor-quality genotypes were removed: mono high resolution (clear clustering, but lacking one homozygous marker), call rate below threshold (low response, poor genotyping), and other (poor genotyping and unmapped to chromosomes). Furthermore, the genotype data of Hengguan 35 and Zhoumai 18 were compared, and SNP markers with homozygous genotypes and with different allelic variations ("aa × bb" type) between the parents were extracted. The genotypes of the RIL population were extracted for the polymorphic SNP markers identified between the parents, and the heterozygous markers in the progeny were converted to missing types. Then, SNP marker quality control was performed using the VCFtools software (Danecek et al. 2011) with parameter thresholds set as follows: ‌minor allele frequency (MAF) ≥ 0.05‌ and ‌missing rate ≤ 0.1‌.

A redundancy analysis of high-quality polymorphic SNP loci based on the separation pattern of the RIL population was conducted using the BIN module of IciMapping 4.2 software (Meng et al. 2015). The "missing rate" and "P value for segregation distortion test" parameters were set to 10% and 0.001, respectively. One marker was randomly selected from each bin for genetic map construction. The bin markers were grouped according to the chromosome information from the Affymetrix wheat 660 K SNP array, and a genetic map was constructed using the maximum likelihood (ML) method in JoinMap 4.0 software (Stam 1993). The conversion between the recombination rate and genetic distance was performed via the Kosambi function. The genetic map was plotted using the "LinkageMapView" package in R (Ouellette et al. 2017) and MapChart (Voorrips 2002. https://doi.org/10.1093/jhered/93.1.77). The quality of the genetic map was assessed by collinearity analysis between the genetic map and the physical map. The correlation coefficients between the genetic positions and the physical positions of the markers were calculated by Excel. The collinearity between the genetic and physical maps was visualized using the "Circlize" package in R (Gu et al. 2014).

QTL mapping

QTL mapping for yield-related traits was performed using the inclusive composite interval mapping (ICIM) method in the BIP module of IciMapping 4.2 software (Li et al. 2008; Meng et al. 2015). The parameters were set as follows: step size = 0.1 cM, probability in stepwise regression (PIN) = 0.001, and threshold of logarithm of odds (LOD) value ≥ 3. The QTLs associated with the same trait, located within a genetic distance of approximately 10 cM on the same chromosome, were merged and considered as a single QTL. Additionally, QTLs detected in more than three environments were considered stable major QTLs. The QTL was named following the method proposed by Liu et al. (2018a, b), which was formatted as "Q + trait abbreviation + research facilities abbreviation + chromosome name + order on the chromosome." In addition, according to breeding selection objectives and previous studies (Pang et al. 2020), the variations that increased GN, TGW, GL, GW, and GT, while moderately decreased PH were defined as favorable alleles in this study. QTL cluster analysis was then performed on stable QTLs associated with six yield-related traits. When two or more stable QTLs were located within a specific chromosomal region (≤ 20 cM), they were grouped into a single QTL cluster.

Overlapping genomic regions between the RIL and AB-NAMIC populations

The flanking sequences of the QTLs identified by linkage mapping were aligned to the Chinese Spring reference genome (IWGSC RefSeq v2.1) to determine their physical position. Based on the GWAS signals from our previous study of Jiao et al. (2023), the merge command of Bedtools was used to merge overlapping QTLs identified for each trait into a new QTL. The QTLs identified by linkage analysis in the RIL population were subsequently compared with those detected via GWAS in the AB-NAMIC population, and the overlapping QTLs between the two populations were obtained when the physical positions of their intervals partially overlapped with a minimum distance of ≥ 0.5 Mb.

Candidate gene prediction and KASP marker development

The IntervalTool tool on the WheatOmics 1.0 website (http://wheatomics.sdau.edu.cn/) was utilized to obtain candidate genes within high-confidence genomic regions. The expression patterns of the candidate genes were analyzed using the Wheat Expression Browser database (http://www.wheat-expression.com/). The SNP tightly linked to the QGl.caas-5A.1, and the variant of [A/G] on the second intron of the candidate gene was converted into a KASP marker (Table S2). The natural population was genotyped using the developed KASP markers to preliminarily validate gene function. KASP reactions were carried out on QuantStudio™ 7 Flex (Applied Biosystems by Life Technologies, San Diego, CA, USA) using the KASP V4.0 2 × Master mix (LGC Genomics, Teddington, UK) as previously described (Zhao et al. 2019).

Results

Phenotypic variation and correlation analysis

The basic descriptive statistics for PH, GN, TGW, GL, GW, and GT indicated significant phenotypic differences in GN and GT between the parents, whereas the differences in other traits were not significant (Fig. 1, Table 1). Compared with Hengguan 35, Zhoumai 18 had a greater GT but a smaller GN. However, the range of coefficient of variation (CV) of the six yield-related traits in the RIL population was quite wide, ranging from 2.76% to 23.19%. The BLUP values exhibited a continuous distribution, ranging from 37.88 cm to 116.58 cm for PH, 39.87 to 64.82 for GN, 35.07 g to 58.17 g for TGW, 6.12 mm to 7.12 mm for GL, 3.11 mm to 3.83 mm for GW, and 2.82 mm to 3.36 mm for GT. The broad-sense heritability (H²) was high, greater than 80%, for all six traits, indicating that these traits were less affected by the environment and could be inherited stably in the RIL population.

Fig. 1.

Phenotypic characterization of Hengguan 35 and Zhoumai 18. a Morphology of the whole plants of the parental lines Hengguan 35 and Zhoumai 18. b Spike morphology of the parental lines Hengguan 35 and Zhoumai 18. c Grain size of the parental lines Hengguan 35 and Zhoumai 18. The left figure shows the 150-grain sizes of Hengguan 35 and Zhoumai 18, and the figure on the right shows the grain length, width, and thickness of Hengguan 35 and Zhoumai 18, respectively

Table 1.

Basic descriptive statistics of the RIL parents and lines based on BLUP values

Traita	Parents		RIL lines
	Hengguan 35	Zhoumai 18	Mean ± S.E.b	Range	CV (%)c	Kurtosis	Skewness	H2 (%)d
PH	69.82	68.41	73.82 ± 1.00	37.88–116.58	23.19	-0.33	0.32	98.95
GN	57.97	53.84 *	52.58 ± 0.31	39.87–64.82	10.12	-0.72	0.11	84.46
TGW	45.75	49.29	48.22 ± 0.26	35.07–58.17	9.17	-0.30	-0.38	89.73
GL	6.58	6.62	6.55 ± 0.01	6.12–7.12	2.76	0.20	0.44	90.06
GW	3.44	3.57	3.52 ± 0.01	3.11–3.83	3.72	-0.29	-0.32	88.55
GT	3.01	3.17 *	3.11 ± 0.01	2.82–3.36	3.65	-0.65	-0.07	83.83

^aPH, plant height (cm); GN, grain number per spike (number); TGW, thousand-grain weight (g); GL, grain length (mm); GW, grain width (mm); GT, grain thickness (mm). ^bS.E., standard error; ^cCV, coefficient of variation; ^dH², broad-sense heritability. *indicates significant differences at P < 0.05

Correlation analysis showed that in the RIL population, PH had a significant negative correlation with GN, whereas there was a significant positive correlation between PH and grain weight-related traits, i.e., TGW, GL, GW, and GT. There was a significant negative correlation between GN and grain weight-related traits, and TGW had a significant positive correlation with GL, GW, and GT (Fig. 2a). Additionally, all traits exhibited a continuous distribution in the RIL population, with absolute skewness and kurtosis values of less than 1.0 for all the traits, indicating that these traits largely conformed to a normal distribution (Fig. 2b, Table 1).

Fig. 2.

Pearson correlation and normal distribution analysis of six yield-related traits in the RIL populations based on the BLUP values. a Pearson correlation analysis of yield-related traits. **, *** represent significance at P < 0.01 and P < 0.001, respectively. b Normal distribution analysis of traits in the RIL population

Bin-based genetic map construction and assessment

The Affymetrix wheat 660 K SNP array was used to genotype 292 RILs. On the basis of the SNP marker flanking sequence alignment to the Chinese Spring reference genome (IWGSC RefSeq v2.1), the specific physical positions of 66,063 (10.48%) high-quality polymorphic SNP markers between the parents were obtained (Table S3). Figure 3a shows the distribution of those SNP markers on the physical map across 21 wheat chromosomes. Generally, the segregation ratio of the two alleles for one SNP marker in the RIL population is 1:1. In this study, after removing SNPs with P < 0.001 via a partial segregation test, the remaining 51,536 (8.17%) SNP markers were used for genetic map construction, and these markers were divided into 2,890 bins. A genetic map containing 21 wheat chromosomes with a length of 4,243.31 cM was successfully constructed (Fig. 3b). More precisely, the lengths of the A, B, and D subgenomes were 1,494.81 cM, 1422.33 cM, and 1,326.17 cM, respectively. The genetic length of different chromosomes ranged from 80.20 cM (1B) to 277.41 cM (5B), with an average of 202.06 cM. The collinearity analysis of the genetic map and the physical map revealed that the SNP order in the genetic map had good collinearity with that in the assembly of the wheat reference genome (Fig. 3c), indicating that the genetic map constructed in this study was of high quality.

Fig. 3.

Bin-based genetic map construction and quality assessment. a Distribution of 66,063 high-quality polymorphic SNP markers on the physical map across 21 chromosomes. b Genetic map of 21 chromosomes in the RIL population. c Collinearity analysis between the genetic map and the physical map

QTL detection for yield-related traits

A total of 262 putative QTLs were detected for PH, GN, TGW, GL, GW, and GT distributed across 21 chromosomes (Table S4). Among them, 50 QTLs could be repeatedly detected in more than three environments; thus, these QTLs were considered environmentally stable (Fig. 4, Table 2). The greatest number of QTLs controlling yield-related traits was detected on chromosomes 7A and 4A, whereas only two QTLs were identified on chromosome 1D.

Fig. 4.

Genetic position of QTL regions associated with plant height (PH), grain number per spike (GN), thousand-grain weight (TGW), grain length (GL), grain width (GW), and grain thickness (GT). Named QTLs represent stable QTLs identified in three or more environments

Table 2.

Stable quantitative trait locus (QTL) for six traits identified across multiple environments in the RIL population

Traita	QTL	Environmentb	Marker interval (cM)	Left marker of peak	Right marker of peak	LODc	PVE (%)d	Adde	Donor of favorable alleles	Previous reference reportsf
PH	QPh.caas-3A.1	E2/E3/E10/E12/E14	94.1–96.9	AX-108732873	AX-110378636	7.33	2.98	-2.69	Hengguan 35	Zhang et al. (2017)
	QPh.caas-3A.4	E5/E7/E9/E11/E13	193.1–200.1	AX-109343446	AX-111463342	5.92	2.18	-2.96	Hengguan 35	Guan et al. (2018)
	QPh.caas-4A.3	E3/E5/E7/E6/E9/E10/E11/E12/E14	138.3–149.2	AX-109580567	AX-94978517	5.56	2.18	2.58	Zhoumai 18	Li et al. (2015)
	QPh.caas-6A.1	E2/E3/E4/E5/E6/E8/E9/E10/E11/E12/E13/E14	62.6–70.4	AX-110975340	AX-109974361	12.89	5.76	-4.22	Hengguan 35	Zhai et al. (2016)
	QPh.caas-7A.1	E5/E9/E11/E13/E14	52.8–63.9	AX-111761655	AX-110395009	7.04	2.49	-3.22	Hengguan 35	NA
	QPh.caas-7A.2	E2/E5/E8/E9/E10/E11/E12/E13/E14	88.2–91.9	AX-111156184	AX-109848559	11.03	4.10	3.80	Zhoumai 18	Hu et al. (2020)
	QPh.caas-3B.1	E3/E6/E8	1.0–2.2	AX-110468825	AX-110735793	5.62	2.63	-2.55	Hengguan 35	Cui et al. (2022)
	QPh.caas-5B.2	E2/E5/E6/E7/E9/E10/E12/E13	14.0–14.9	AX-108871901	AX-109866963	12.45	4.89	-3.94	Hengguan 35	NA
	QPh.caas-5B.3	E4/E8/E12/E14	46.9–48.0	AX-109913757	AX-109845322	6.09	2.48	-2.69	Hengguan 35	NA
	QPh.caas-5B.5	E5/E7/E9/E13/	72.2–72.4	AX-111656355	AX-109492223	7.58	2.78	-3.21	Hengguan 35	NA
	QPh.caas-5B.6	E4/E6/E8	160.6–160.8	AX-111001298	AX-89376056	3.36	1.56	-2.01	Hengguan 35	NA
	QPh.caas-7B.2	E3/E4/E6/E7/E8/E9/E10/E11/E12/E14	10.5–18.7	AX-94780393	AX-109486292	8.56	3.69	3.15	Zhoumai 18	Yan et al. (2006)
	QPh.caas-2D.2	E2/E3/E4/E6/E7/E8/E9/E10/E14	9.2–12.0	AX-111021196	AX-111618622	6.45	3.21	2.85	Zhoumai 18	Chai et al. (2022)
	QPh.caas-4D.1	E2/E3/E4/E5/E6/E7/E8/E9/E10/E11/E12/E13/E14	58.0–58.1	AX-89703298	AX-110572006	46.09	40.10	11.19	Zhoumai 18	Peng et al. (1999)
	QPh.caas-5D.1	E5/E7/E9/E10/E14	131.6–133.0	AX-111175690	AX-110958036	6.58	2.41	-2.74	Hengguan 35	NA
	QPh.caas-5D.2	E2/E3/E6/E13	173.9–174.0	AX-109534644	AX-110295931	6.54	2.81	-2.62	Hengguan 35	Zhao et al. (2018)
GN	QGn.caas-2A.2	E3/E7/E10/E11/E14	59.7–60.3	AX-111466604	AX-110620573	4.64	3.29	1.21	Hengguan 35	NA
	QGn.caas-4A.4	E3/E7/E13	162.7–163.0	AX-95220664	AX-94503747	8.23	5.18	1.98	Hengguan 35	Gao et al. (2015)
	QGn.caas-5A.4	E7/E8/E14	82.1–82.8	AX-109343369	AX-108832618	5.37	2.88	1.09	Hengguan 35	NA
	QGn.caas-7A.2	E3/E4/E8/E12/E13/E14	170.9–173.6	AX-109279553	AX-110721033	6.14	4.76	-1.63	Zhoumai 18	Kuzay et al. (2022)
	QGn.caas-3B.1	E5/E6/E9/E14	1.0–1.2	AX-110468825	AX-111556751	6.45	4.13	1.57	Hengguan 35	NA
	QGn.caas-5B.1	E3/E9/E11	13.1–14.9	AX-110416577	AX-109866963	7.23	4.39	1.67	Hengguan 35	NA
	QGn.caas-4D.1	E2/E3/E4/E5/E6/E7/E8/E9/E10/E11/E12/E13/E14	56.7–62.7	AX-108944764	AX-111354775	25.29	21.90	-3.59	Zhoumai 18	Jia et al. (2013)
TGW	QTgw.caas-3A.2	E5/E11/E14	218.5–219.0	AX-111079045	AX-109872128	7.51	4.28	-1.31	Zhoumai 18	NA
	QTgw.caas-4A.5	E2/E5/E9/E13/E14	142.0–143.0	AX-109857541	AX-108966947	7.81	6.03	-1.32	Zhoumai 18	Gao et al. (2015)
	QTgw.caas-4A.6	E1/E3/E6/E7/E8	162.9–171.2	AX-95220664	AX-111600193	8.00	7.09	-1.49	Zhoumai 18	Guan et al. (2018)
	QTgw.caas-7A.1	E3/E7/E14	51.9–55.5	AX-111761655	AX-109959178	8.14	4.98	-1.24	Zhoumai 18	NA
	QTgw.caas-7A.3	E5/E9/E11/E14	165.7–170.4	AX-111583862	AX-108784724	7.66	4.78	1.37	Hengguan 35	Keeble-Gagnère et al. (2018)
	QTgw.caas-7A.6	E7/E8/E12	221.6–222.3	AX-110554268	AX-109934866	4.77	4.51	1.20	Hengguan 35	Kumar et al. (2016)
	QTgw.caas-4D.1	E1/E2/E3/E5/E6/E7/E8/E9/E10/E11/E12/E13/E14	57.0–57.8	AX-108944764	AX-89703298	28.00	27.65	3.00	Hengguan 35	Mohler et al. (2016)
GL	QGl.caas-1A.1	E1/E9/E11	7.7–7.9	AX-111486000	AX-86178568	5.85	5.06	0.06	Hengguan 35	NA
	QGl.caas-4A.1	E5/E8/E9/E13	141.9–146.9	AX-111148488	AX-110543806	6.63	5.88	-0.07	Zhoumai 18	NA
	QGl.caas-4A.4	E1/E2/E3/E4/E5/E7/E9/E10/E11/E14	242.2–242.5	AX-109548008	AX-110952283	6.60	4.47	-0.06	Zhoumai 18	NA
	QGl.caas-5A.1	E4/E5/E8/E10/E12/E14	1.0–1.9	AX-110680739	AX-111128087	10.15	5.64	0.07	Hengguan 35	Mohler et al. (2016)
	QGl.caas-5A.4	E4/E11/E12/E14	197.6–201.1	AX-109423091	AX-109581017	13.32	6.42	-0.07	Zhoumai 18	NA
	QGl.caas-6A.4	E10/E11/E12/E14	137.8–139.3	AX-94815082	AX-89481603	9.31	4.93	-0.06	Zhoumai 18	NA
	QGl.caas-7A.2	E2/E3/E5/E12/E13	97.2–101.5	AX-110478459	AX-111628682	3.85	3.56	0.05	Hengguan 35	Cui et al. (2016)
	QGl.caas-7A.4	E4/E7/E8/E9/E10/E12/E13/E14	173.6–179.5	AX-109908479	AX-111601461	8.77	5.90	0.07	Hengguan 35	Su et al. (2016)
	QGl.caas-3B.1	E8/E10/E11/E12	137.2–140.2	AX-108894034	AX-110081374	5.11	3.37	-0.05	Zhoumai 18	NA
	QGl.caas-4D.1	E3/E7/E9/E11	55.2–57.9	AX-89654830	AX-110572006	5.54	4.67	0.06	Hengguan 35	Peng et al. (1999)
	QGl.caas-4D.2	E2/E8/E10/E13/E14	71.9–75.8	AX-110696146	AX-108726608	12.25	7.27	0.07	Hengguan 35	NA
	QGl.caas-5D.3	E3/E4/E8/E12	135.1–139.8	AX-110958036	AX-109809706	6.52	4.60	-0.06	Zhoumai 18	Li et al. (2019)
	QGl.caas-6D.1	E2/E3/E5/E8/E9/E10/E13	0.0–3.3	AX-94797722	AX-109871624	4.98	4.37	-0.06	Zhoumai 18	NA
GW	QGw.caas-4A.1	E5/E11/E13	119.8–121.8	AX-111612772	AX-109558533	5.05	3.00	-0.05	Zhoumai 18	NA
	QGw.caas-4A.3	E4/E8/E9/E10/E14	141.2–142.0	AX-109844107	AX-110540586	12.37	8.42	-0.05	Zhoumai 18	Guan et al. (2018)
	QGw.caas-5A.1	E8/E10/E14	123.9–124.1	AX-109516381	AX-109306311	3.87	2.96	0.03	Hengguan 35	NA
	QGw.caas-4D.1	E1/E2/E3/E4/E5/E7/E8/E9/E10/E11/E12/E13/E14	56.8–59.3	AX-108944764	AX-110572006	21.90	22.55	0.09	Hengguan 35	Peng et al. (1999)
GT	QGt.caas-7A.4	E3/E5/E7/E11/E14	221.5–221.9	AX-110362833	AX-111046608	4.67	4.03	0.03	Hengguan 35	NA
	QGt.caas-2B.1	E3/E9/E10	64.0–67.1	AX-108769797	AX-109364692	4.05	3.34	0.04	Hengguan 35	NA
	QGt.caas-4D.1	E1/E2/E3/E4/E5/E7/E8/E9/E10/E11/E12/E13/E14	51.3–57.9	AX-108735064	AX-110572006	21.84	22.76	0.08	Hengguan 35	Peng et al. (1999)

^aThe abbreviated names of all the phenotypic traits are shown in Table 1. ^bE1, E2, E3, E4, E5, E6, E7, E8, E9, E10, E11, E12, E13, and E14 indicate 2014-Zhengzhou, 2015-Shunyi, 2015-Xinxiang, 2016-Shunyi, 2016-Xinxiang, 2017-Shunyi, 2017-Xinxiang, 2018-Shunyi, 2018-Xinxiang, 2019-Shunyi, 2019-Xinxiang, 2020-Shunyi, 2020-Xinxiang, and BLUP, respectively. ^cLOD, likelihood of odds; ^dPVE, proportion of phenotypic variation of the corresponding QTL; ^eADD, additive effect; ^fNA: not available

A total of 57 QTLs for PH were identified, 16 of which were stable QTLs detected in at least three environments (Table 2, Fig. S1). The percentage of phenotypic variance explained (PVE) ranged from 1.56% to 40.10%, with LOD values ranging from 3.36 to 46.09 for the 16 stable QTLs. Among the alleles for decreased PH, 11 QTLs were contributed by Hengguan 35 and five were contributed by Zhoumai 18. The PVEs of QPh.caas-6A.1 (5.76%) and QPh.caas-4D.1 (40.10%) both exceeded 5%, and these QTLs were detected in 12 environments and 13 environments, respectively.

A total of 50 QTLs for GN were identified, seven of which were stable QTLs detected in at least three environments (Table 2, Fig. S2). The PVE ranged from 2.88% to 21.90%, with LOD values ranging from 4.64 to 25.29 for the seven stable QTLs. Among the alleles for increased GN, five QTLs were contributed by Hengguan 35 and two were contributed by Zhoumai 18. The PVEs of QGn.caas-4A.4 (5.18%) and QGn.caas-4D.1 (21.90%) both exceeded 5%, and these QTLs detected in three environments and 13 environments, respectively.

A total of 34 QTLs for TGW were identified, seven of which were stable QTLs detected in at least three environments (Table 2, Fig. S3). The PVE ranged from 4.28% to 27.65%, with LOD values ranging from 4.77 to 28.00 for the seven stable QTLs. Among the alleles for increased TGW, three QTLs were contributed by Hengguan 35 and four were contributed by Zhoumai 18. The PVEs of QTgw.caas-4A.5 (6.03%), QTgw.caas-4A.6 (7.09%), and QTgw.caas-4D.1 (27.65%) both exceeded 5%, and these QTLs were detected in five, five, and 13 environments, respectively.

A total of 58, 27, and 36 QTLs for GL, GW, and GT, respectively, were identified, of which 13, four, and three, respectively, were stable QTLs detected in at least three environments (Table 2, Figs. S4, S5 and S6). The PVE of GL ranged from 3.37% to 7.27%, with LOD values ranging from 3.85 to 13.32 for the 13 stable QTLs. The PVE of GW ranged from 2.96% to 22.55%, with LOD values ranging from 3.87 to 21.90 for the four stable QTLs. The PVE of GT ranged from 3.34% to 22.76%, with LOD values ranging from 4.05 to 21.84 for the three stable QTLs. GL, GW, and GT had six, two, and one stable QTL, respectively, with PVE values exceeding 5%.

Ten pleiotropic QTL clusters, comprising 27 stable QTLs, were identified across chromosomes 3B, 4A, 4D, 5B, 5D, and 7A (Table S5), with each cluster containing two to six QTLs. Among these, the cluster QGt.caas-4D.1/QGl.caas-4D.1/QGn.caas-4D.1/QGw.caas-4D.1/QTgw.caas-4D.1/QPh.caas-4D.1 on chromosome 4D (spanning 51.3 cM to 62.7 cM) contained the most QTLs and exhibited positive additive effects for TGW, GL, GW, GT, and PH, while showing a negative additive effect for GN. The cluster QPh.caas-4A.3/QGw.caas-4A.3/QGl.caas-4A.1/QTgw.caas-4A.5 on chromosome 4A (spanning 138.3 cM to149.2 cM) had a positive additive effect for PH, but negative additive effects for TGW, GL and GW. The cluster QTgw.caas-7A.3/QGn.caas-7A.2/QGl.caas-7A.4 on chromosome 7A (spanning 165.7 cM to 179.5 cM) exhibited a positive additive effect for TGW and a negative additive effect for GN.

Additive effect of identified QTLs in the RIL population

PH in the RIL population exhibited significant and extensive phenotypic variation, ranging from 37.9 cm to 116.6 cm, with a high CV of 23.2% (Fig. 5a, Table 1). Sixteen stable QTLs controlling PH were identified through linkage analysis in the RIL population, with 11 QTLs containing the allele for reduced PH from Hengguan 35 and the other five containing the allele from Zhoumai 18 (Figs. S1 and 5b). The phenotypic effect was evaluated by the linear regression between the number of reduced PH alleles and PH. The results showed that as the number of reduced PH alleles increased, PH obviously trended to decrease, and a strong negative correlation (r² = 0.40) was found, indicating that the accumulation of more alleles for reduced PH led to lower PH (Fig. 5c and d). Hence, when the RILs carried only one allele for reduced PH, the PH was 115.9 cm, whereas RILs with 13 alleles for reduced PH had the lowest average PH (55.9 cm). Similarly, there were strong positive correlations between the number of alleles for increased GN, GW, TGW, GT, and GL and the corresponding phenotypic values (Fig. S7). These results imply the reliability of QTL mapping and the significant additive effect of alleles for yield-related traits.

Fig. 5.

Plant height variation with the accumulation of favorable alleles in the RIL population. a Plant height of parents and representative lines of the RIL population. HG35, Hengguan 35; ZM18, Zhoumai 18. b Sixteen stable QTLs associated with plant height were identified through linkage analysis in the RIL population. RHT8, RHT5, RHT-D1, RHT23 represent the presence of cloned genes in the corresponding QTL region. c Plant heights of the accessions harboring different numbers of favorable alleles. d Correlations between the number of favorable alleles and plant height

Co-localized genome regions discovered by the RIL and AB-NAMIC populations

On the basis of our previous study (Jiao et al. 2023), a total of 365 QTLs controlling PH, GN, TGW, GL, GW, and GT were identified in the AB-NAMIC population via association mapping (Table S6). Thus, joint linkage analysis and association mapping in the current study revealed co-localized genomic regions for all six yield-related traits in the RIL and AB-NAMIC populations (Figs. 6 and S8). Altogether, there were 26 overlapping QTL regions for PH, GN, TGW, and GL on chromosomes 1A, 2A, 2D, 3A, 3B, 4A, 5A, 5B, 5D, 6A, and 7A between these two populations (Table S7). Among them, QPh.caas-2D.2, QPh.caas-3B.1, QTgw.caas-4A.6, QTgw.caas-7A.6, QGl.caas-4A.1, QGl.caas-5A.1, QGl.caas-5D.3, and QGl.caas-7A.4 were detected in the RIL population in at least three environments, indicating an important candidate genomic region for yield-related traits to be fine-mapped in subsequent research.

Fig. 6.

QTL overlapping regions on chromosomes of the RIL and AB-NAMIC populations. Chromosomes of different colors represent the RIL population (left) and the AB-NAMIC population (right), and different shapes represent different traits. PH, GN, TGW, GL, GW, and GT represent plant height, grain number per spike, thousand-grain weight, grain length, grain width, and grain thickness, respectively. The bottom shading on the chromosomes in both populations indicates segments co-anchored by GWAS and linkage analysis. The black box beside the chromosome represents the QTL segment size

Candidate genes within the GL‑associated genomic region on chromosome 5A

The analysis of overlapping genome regions between the RIL and AB-NAMIC populations revealed a QTL that was simultaneously detected through both linkage analysis and GWAS, namely QGl.caas-5A.1, located on chromosome 5A and delimited by the flanking markers AX-110680739 and AX-111128087, with a LOD value > 10 (Fig. 7a). Further analysis indicated that the physical interval of QGl.caas-5A.1 was between 1.68 Mb and 2.73 Mb, which contained 11 high-confidence genes. The gene expression data from the Wheat Expression Browser database revealed that only three of these genes were expressed in the spike or grain, namely TreasCS5A03G0002000, TreasCS5A03G0002100, and TreasCS5A03G0002500 (Fig. 7b). The functional annotation of TraesCS5A03G0002500 was carotenoid cleavage dioxygenase, which was preliminarily considered the candidate gene for QGl.caas-5A.1 (Table S8). TraesCS5A03G0002500 has 14 exons and 13 introns, and the 660 K genotyping results in the RIL population revealed four SNP variations in the intron region (Fig. 7c). KASP primers were designed based on the [A/G] variation in the second intron region in TraesCS5A03G0002500 (Table S2). Subsequently, a natural population consisting of 350 accessions was screened using the KASP marker, and the results showed that effective genotyping of both allelic and heterozygous variants (Fig. 7d, Table S1). Furthermore, a two-tailed t-test was conducted on four grain-related traits in the natural population carrying two types of alleles (Fig. 7e). Compared with the accessions carrying the A allele, the accessions carrying the G allele presented greater TGW, GL, GW, and GT; therefore, the G allele was the favorable allele of TraesCS5A03G0002500 for yield-related traits.

Fig. 7.

Candidate gene prediction and preliminary functional validation of QGl.caas-5A.1. a QGl.caas-5A.1 detected in the RIL and AB-NAMIC populations. b Transcriptome data for four tissues from Chinese Spring in the physical interval of QGl.caas-5A.1. c Structure of TraesCS5A03G0002500. d Allelic segregation of the KASP marker for TraesCS5A03G0002500 alleles. e Phenotypic differences between the two alleles of TraesCS5A03G0002500 in thousand-grain weight (TGW), grain length (GL), grain width (GW), and grain thickness (GT) across four environments

Discussion

The rich diversity of phenotypic variations lays the foundation for QTL mapping

The phenomenon of transgressive segregation has been widely applied in agricultural production to increase crop yield and improve disease resistance and stress tolerance (Gu et al. 2023). In this study, the basic descriptive statistics of the six traits indicated that there was no significant phenotypic difference between the two parents, Hengguan 35 and Zhoumai 18, in terms of PH, TGW, and other traits, but the RILs showed extremely high phenotypic variation, especially in PH (Table 1). Cui et al. (2011) reported that there was no significant difference in PH between parents, and their offspring showed transgressive segregation, which was consistent with the results of the present study. Taking PH as an example, among the 16 identified QTLs associated with PH, 11 QTLs for reduced PH were derived from Hengguan 35, each exhibiting a relatively small additive effect within the absolute range of 2.0–4.2 cm. Meanwhile, the other five identified QTLs responsible for reducing PH were traced back to Zhoumai 18, with QPh.caas-4D.1 exhibiting a high additive effect of 11.2 cm (Fig. 5). These results suggest that Hengguan 35 carries multiple minor-effect QTLs for reducing PH, whereas Zhoumai 18 carries a major-effect QTL for reducing PH, thus resulting in no significant difference in plant height between the parents. Furthermore, the RIL population carried varying numbers of QTLs for reduced PH, and with an increase in the number of favorable alleles, there was an obvious decrease in PH, which was consistent with the findings of Pang et al. (2020). The significant differences in PH among the offspring can be attributed to the recombination of chromosomes for the QTLs with reducing PH during the hybridization and multi-generation self-cross of the RIL population. Simply speaking, the rich phenotypic variation in the RIL population lays the foundation for gene discovery.

QTL comparison with previous studies

The PH of wheat is a composite trait, comprising spike length and internode length below the spike. The "Green Revolution" in the 1960s introduced semidwarf genes that reduce PH, increase lodging resistance, and significantly improve grain yield (Gale and Youssefian 1985). Currently, 25 Rht genes have been identified on chromosomes 2A, 2B, 2D, 3B, 4A, 4B, 4D, 5A, 5D, 6A, 7A, and 7B (Mo et al. 2018). In this study, 16 stable QTLs controlling PH were identified in the RIL population. BLAST analysis revealed that QPh.caas-2D.2, QPh.caas-4D.1, QPh.caas-5D.2, and QPh.caas-3B.1 contain the known genes RHT8, RHT-D1, RHT23, and RHT5, respectively (Chai et al. 2022; Peng et al. 1999; Zhao et al. 2018; Cui et al. 2022). QPh.caas-4D.1 simultaneously affects TGW, GL, GW, GT, as well as GN. Furthermore, QPh.caas-3A.1, QPh.caas-3A.4, QPh.caas-4A.3, QPh.caas-6A.1, QPh.caas-7A.2, and QPh.caas-7B.2 overlapped with the loci for PH reported in previous studies (Zhang et al. 2017; Guan et al. 2018; Li et al. 2015; Zhai et al. 2016; Hu et al. 2020; Yan et al. 2006). However, six new loci controlling PH were identified in the current study, four of which were located on chromosome 5B. Generally, the GN depends on the fertility of the florets and the number of spikelet per spike, which is determined in the early stages of spikelet development (Xiao et al. 2022). In the present study, QGn.caas-7A.2 contained the cloned WAPO1 gene (Kuzay et al. 2022). QGn.caas-4D.1 and QGn.caas-4A.4 identified through linkage analysis were consistent with those reported by Jia et al. (2013) and Gao et al. (2015). Other stable QTLs, which represent new loci controlling spikelet numbers and grain numbers, have not been previously reported. Increasing yield is significantly correlated with an increase in TGW, with seed size being a crucial factor in determining grain weight. In this study, seven stable QTLs related to grain weight were identified in the RIL population, of which QTgw.caas-4A.5, QTgw.caas-4A.6, QTgw.caas-7A.3, QTgw.caas-7A.6, and QTgw.caas-4D.1 were consistent with previous studies (Gao et al. 2015; Guan et al. 2018; Keeble-Gagnère et al. 2018; Kumar et al 2016; Mohler et al. 2016), which indicates that these QTLs are stable in different genetic backgrounds. Additionally, only two TGW-related QTLs were most likely novel loci. Starch synthesis plays a significant role in regulating seed size, with starch synthesis-related genes TaSus1 and Tasus2 being associated with seed size (Jiang et al. 2011; Hou et al. 2014). However, the genetic loci underlying wheat seed size are poorly understood. This study identified several QTLs associated with GL, GW, and GT, some of which overlapped with previously reported QTLs (Mohler et al. 2016; Cui et al. 2016; Su et al. 2016; Li et al. 2019; Guan et al. 2018). Additionally, several QTLs located on chromosomes 1A, 4A, 5A, 6A, 7A, 2B, 3B, 4D, and 6D are novel loci associated with grain size. Taken together, these data are largely consistent with previous findings and provide novel targets for the discovery of genes underlying yield-related traits in wheat.

QTL clusters for yield‑related traits

Pleiotropy is a critical factor influencing the mutation and evolution of biological traits. This phenomenon plays a key role in the inheritance, evolution, and development of crops, making it especially valuable for strategies such as pyramiding breeding and MAS (Deng et al. 2017). QTL cluster analysis provides essential insights into deciphering the genetic regulatory networks of complex traits and the mechanisms underlying trait-trait correlations‌. Previous studies identified several QTL clusters associated with PH, TGW, GL, and GW (Ren et al. 2021), and a major QTL cluster spanning chromosome 3AS (between wsnp_Ku_c50833_56310208 and wsnp_Ra_c19079_28210937) governing spike length, grain number per spike, spikelet numbers per spike, and total spikelet number per spike (Liu et al. 2018a, b). This study demonstrated that multiple QTL clusters regulated PH, GN, and grain weight-related traits (TGW, GL, GW, GT) through synergistic or antagonistic effects, revealing the genetic relatedness among these traits‌. Among the ten pleiotropic QTL clusters identified in this study, the QGt.caas-4D.1/QGl.caas-4D.1/QGn.caas-4D.1/QGw.caas-4D.1/QTgw.caas-4D.1/QPh.caas-4D.1 cluster negatively regulates PH/GN and positively regulates PH/TGW, explaining the significant negative correlation between PH and GN, as well as the positive correlation between PH and TGW (Fig. 2, Table S5). The additive effect directions of QTL clusters further help optimize breeding strategy. For instance, the antagonistic additive effects of QTgw.caas-7A.3 (+ 1.37 g) and QGn.caas-7A.2 (-1.63) in the QTgw.caas-7A.3/QGn.caas-7A.2/QGl.caas-7A.4 cluster directly validate the TGW/GN trade-off, highlighting the importance of designing allele pyramiding strategies targeting specific QTL clusters in MAS‌. In summary, the study and application of pleiotropic QTLs offer novel insights and approaches for crop genetic improvement, particularly in the selection and optimization of complex traits.

Joint linkage analysis and association mapping provide a new perspective for discovering yield-related genes

Currently, research focused on the genetic basis dissection of crop quantitative traits by using artificial genetic populations is increasing. Generally, the construction of parental populations is relatively simple and has high QTL detection efficacy. The advanced backcross-nested association mapping (AB-NAM) population combines the characteristics of the advanced backcross (AB) and nested association mapping (NAM) populations and has become an effective population for mining new loci of exotic germplasm resources (Nice et al. 2016). In the previous study, an AB-NAMIC population was developed by crossing three elite Chinese wheat cultivars with approximately 20 accessions from a mini core collection of Chinese wheat genetic resources, and in this population, rare alleles were strongly detected (Jiao et al. 2023). Given these findings for the AB-NAMIC population, a combination of the mapping results of the RIL and AB-NAMIC populations was attempted to explore QTLs that can be consistently identified across different genetic backgrounds. By joint linkage analysis and association mapping, 26 overlapping genomic regions were identified (Table S7, Fig. 6), which should be the focus of further research. However, the yield-related traits of wheat are controlled by multiple genes and are susceptible to environmental influences, and the distance of genomic linkage disequilibrium (≥ 2 Mb) of wheat is much greater than that of rice, maize, and other crops (Li et al. 2022; Jiao et al. 2023). Thus, it is still challenging to identify genetic loci related to yield traits in candidate regions. Therefore, for QTLs with large genomic intervals identified across both panels, such as QGl.caas-4A.2, located in the 641.8–688.5 Mb region on chromosome 4A, it is essential to employ a combination of methodologies such as bulked segregant analysis (BSA), bulk segregant RNA sequencing (BSR-seq), and RNA sequencing (RNA-seq) for further fine mapping of potential candidate genes (Zhao et al. 2023).

In the current study, one stable QTL, QGl.caas-5A.1, was associated with GL, and contained relatively small candidate regions for further exploration of the candidate gene (Fig. 7). Candidate genes were predicted based on gene expression data in different tissues and functional annotations of the homologous gene. Those results lay a solid foundation for the subsequent cloning of yield-related genes. In the future, functional validation of target genes will be conducted via overexpression and clustered regularly interspaced palindromic repeats (CRISPR)/Cas9 technologies, aiming to elucidate the molecular mechanisms underlying GL and other yield-related traits.

Conclusions

In this study, a high-quality genetic map spanning 4,243.31 cM was constructed using data from the Affymetrix wheat 660 K SNP array. An RIL population was used to identify QTLs for yield-related traits, i.e., PH, GN, TGW, GL, GW, and GT in multiple environments. A total of 50 stable QTLs were identified in more than three environments. The pyramiding of favorable alleles detected in this study indicated highly positive phenotypic effects on all the traits. Joint linkage analysis and association mapping discovered 26 overlapping QTLs. Notably, TraesCS5A03G0002500 was selected as the candidate gene of QGl.caas-5A.1 for grain length. Furthermore, a KASP marker was developed based on the [A/G] variation of TraesCS5A03G0002500 and verified in a natural population. The G allele was found to be favorable and was associated with higher TGW, GL, GW, and GT. These findings provide new insights into dissecting the genetic basis of yield-related traits and molecular-assisted breeding and lay the foundation for cloning and molecular mechanism analysis of yield-related genes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution statement

CH designed the experiments and edited the manuscript; LZ performed the experiments and wrote the manuscript; LD constructed the RIL population; HL participated in the experiments and data analyses; HL, HL, YZ, YL, JH, and TL participated in phenotypic investigation; DY participated in data analyses and revising the manuscript; XZ participated in field trials and revising the manuscript. All authors contributed to the article and approved the final manuscript.

Funding

This work was financially supported by grants from the STI 2030-Major Projects (2023ZD0406802), the Biological Breeding-National Science and Technology Major Project (2022ZD04017), the National Key Research and Development Program of China (2023YFD1200404), and the Innovation Program of Chinese Academy of Agricultural Sciences (CAAS‐CSCB‐202401).

Data availability

Data generated in this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare that there is no conflict of interest.