The DH population consisting of 123 individuals was developed through in vitro anther culture (De Buyser and Henry, 1980) of the F₁ hybrids from a cross between ND3753 and LX99. The DH population and two parents were used for genome-wide identification of QTLs related to SSV.

For mapping of the possible QTL, ND3753 that carried the positive allele of the QTL in the confidence interval was crossed with LX99. F₁ plants were backcrossed with LX99 for three generations with insertion-deletion (InDel) marker-assisted selection to generate a BC₃F₁ population containing 418 plants. Subsequently, 126 heterozygotes lines at the QTL-anchored region were self-pollinated to BC₃F₂ containing 1,081 plants. This population with LX99 background is presented as BC₃F₂-L in the present paper.

In addition, in order to evaluate the effects of two QTL, LX99 was crossed with recurrent parent ND3753 and 64 BC₃F₁ heterozygotes lines at the QTL-anchored region were self-pollinated to construct another BC₃F₂ population containing 387 plants with marker-assisted selection. This population with the background of ND3753 is presented as BC₃F₂-N.

The DH population and two parents were planted in seven environments during the wheat-growing seasons of 2016, 2017, and 2018 in Beijing (BJ) (40°08′N, 116°10′E), Linfen (LF) (36°04′N, 111°31′ E), Xi’an (XA) (34°16′N, 108°55′E), and Cangzhou (CZ) (38°18′N, 116°49′E), China. The seven environments, namely 2016BJ, 2017BJ, 2017LF, 2017XA, 2018BJ, 2018LF, and 2018CZ were presented in this study as E1, E2, E3, E4, E5, E6, and E7, respectively. The field trials were conducted following a complete random block design with three biological replicates. However, only one biological duplication was harvested in E7 due to an accident of field management. Then 60 seeds for each of the lines and two parents were planted in two rows of 1.5 long and the row space was 20 cm.

The BC₃F₁, BC₃F₂-L, and BC₃F₂-N populations were all planted in Beijing. The BC₃F₁ population was planted in the wheat-growing seasons of 2018, while the BC₃F₂-L and BC₃F₂-N populations were planted in 2019. All of these backcross populations were sown in rows of 1.5 m long and 30 cm row space with a sowing density of 20 seeds per row. The BC populations were all planted in one trial and designed as a single replicate. During the whole growing season, the local standard field management methodologies were adopted for plant cultivation.

For DHs, 80 plants were harvested in each line of the seven environments and wheat flour of each line planted in E1 was obtained with a CD1 Quadrumat Junior laboratory mill (Chopin Technology, Paris, France), while the whole wheat flour of each line in the other six environments was produced by an XF-98B experimental mill (Zhenxing Electromechanical Instrument Factory, Cangzhou, China). SSV was determined according to a modified protocol of Axford et al. (1979) and Preston et al. (1982) using 2 g of samples. The specific procedure of the SSV test was similar to that described by Li et al. (2009). In particular, SSVs of each DH line with only one biological duplication harvesting from E7 were measured. SSV of the BC₃F₁, BC₃F₂-L, and BC₃F₂-N populations was measured with whole wheat flour from a single plant.

In addition, 300 g of grains of 30 randomly selected DH lines based on the minimum sample required for the Pearson’s correlation coefficient calculated according to the formula provided by Mangard et al. (2007) and Chen et al. (2011) in E5 were ground into flour with a flour yield of approximately 60% in all samples. Their farinograph parameters (GB/ICC) were recorded by a Farinograph (DongFu JiuHeng, Beijing, China) to evaluate the correlation with SSV (Chicago, IL, United States) (ICC, 1996; Luo et al., 2018).

Deoxyribonucleic acid (DNA) was extracted from fresh leaves of individual DH lines and two parents using the hexadecyltrimethy ammonium bromide (CTAB) method (Allen et al., 2006). The 15 K Axiom^® Wheat Breeder single-nucleotide polymorphism (SNP) Genotyping Array (China Golden Marker Co., Beijing, China) containing 13,947 SNP markers was used to genotype the DH population and parents. SNP markers with a missing data rate > 20% were removed, and the remaining polymorphic markers were used to construct a wheat genetic map based on the inclusive composite interval mapping (ICIM) method using IciMapping v4.1 (Chinese Academy of Agricultural Sciences, China) and MapChart v2.32 (Plant Research International, P.O. Box 16, 6700 AA Wageningen) (Voorrips, 2002). The physical locations of unique SNP markers were obtained from the International Wheat Genome Sequencing Consortium (Appels et al., 2018).

The average value of SSV in each environment and the BLUP were employed for QTL analysis using inclusive composite interval mapping (ICIM¹) method in software IciMapping v4.1 (Meng et al., 2015). A QTL with LOD ≥ 2.5 was defined as a significant QTL. The confidence intervals (±2 LOD away from the peaks of likelihood ratios) of several QTLs were coincident, which were preliminarily considered as the same QTL. In this study, the QTL that can be detected in three or more environments is defined as an environmentally stable QTL.

High-quality genomic DNA of ND3753 and LX99 was extracted to construct paired-end-sequencing libraries. According to the procedures described by Li et al. (2020), the parents were re-sequenced with an average sequencing depth of 6 × and paired-end reads of length 150 bp for two parents using the Illumina HiSeq X Ten platform (Illumina, California, United States), and the re-sequencing data were processed. The InDels were identified using the HaplotypeCaller module of the Genome Analysis Toolkit (GATK). The InDel markers were developed based on the sequence difference between the parents around the target region. Primer3 version 0.4.0² was used to design the sequences of InDel primers.

Deoxyribonucleic acid (DNA) amplification was programmed for an initial 5 min at 94°C, then followed by 35 cycles of 30 s at 94°C, 30 s at 56°C, and 30 s at 72°C, and finally 5 min at 72°C. A 10 μL PCR reaction system was used, containing 5 μL of 2 × Taq PCR StarMix (GenStar, Beijing, China) (for PAGE), 1.5 μL of DNA template (about 50–100 ng), 1.5 μL of each InDel primer, and double-distilled H₂O. The PCR products were analyzed on 8% non-denaturing polyacrylamide gels with silver staining.

Insertion-deletion (InDel) markers tightly linked to QSsv.cau-1A.1.1 and QSsv.cau-1A.1.2 were used to genotype DH, BC₃F₁, BC₃F₂-L, and BC₃F₂-N populations. The putatively annotated high confidence (HC) genes located between the flanking markers of QSsv.cau-1A.1.1 were acquired based on the reference genome of T. aestivum cv. Chinese Spring³.

To conduct phenotypic statistical analysis and calculate correlation coefficients between SSV and farinograph parameters, IBM SPSS Statistics 21.0 (SPSS, Chicago, United States) was used. R software v3.6.2⁴ was used to perform the Shapiro-Wilk test across seven environments and the best linear unbiased prediction (BLUP), as well as to estimate the broad-sense heritability (h2B) following the formula: hB2=σg2/(σg2+σg⁢e2/n+σ2/n⁢r), as described by Liu et al. (2014).

Descriptive statistics for SSV of two parents and the DH population in the seven environments are shown in Table 1. SSV of ND3753 was significantly higher than that of LX99 in all environments. SSV showed bi-directional transgressive segregation, suggesting that both parents have increasing alleles for SSV. The broad-sense heritability values in all environments were greater than 0.8, indicating that SSV was mainly controlled by genetic factors. The result of the Shapiro-Wilk test displayed that SSV exhibited normal distribution under six environments and BLUP value, indicating SSV was determined by many genes (Figure 1). Pearson’s correlation coefficient analysis was carried out between farinograph parameters and SSVs of 30 DH lines planted in E5 which indicated a significant positive correlation between the two (Table 2).

A total of 2,523 SNP markers showed polymorphisms between the two parents. Finally, 2,474 SNP markers participated in the map construction and were mapped to 21 linkage groups, covering the 21 chromosomes of common wheat (Supplementary Table 1). The total length of the map was 7,349.01 cm, and the average interval distance between two adjacent markers was 7.24 cm (Supplementary Table 1). The A genome contained the most SNP markers (963), followed by the B genome (902), while the D genome had the least (609) (Supplementary Table 1). The total length of chromosome 7D was the largest (619.02 cm), while that of chromosome 4B was the smallest (180.86 cm) (Supplementary Table 1). Chromosome 5A harbored the most SNP markers (236), while chromosome 6D contained the least (42) (Supplementary Table 1).

A total of 15 QTLs were detected on 11 chromosomes (1A, 1B, 1D, 2A, 2D, 4B, 4D, 5A, 5D, 6B, and 6D) in the seven environments (Table 3 and Supplementary Table 2). Four environmentally stable QTLs (QSsv.cau-1A.1, QSsv.cau-1A.2, QSsv.cau-2D and QSsv.cau-5D.1) were identified on chromosomes 1A, 1A, 2D, and 5D, respectively (Table 3). The favorable allele of QSsv.cau-2D came from LX99, while the superior alleles of the other three QTLs were contributed by ND3753. The major QTL QSsv.cau-1A.1 was repeatedly detected in five environments and the BLUP data, explaining 39.67% of the phenotypic variation in the BLUP analysis (Table 3). QSsv.cau-2D and QSsv.cau-5D.1 contributed 3.17 and 4.82% of the phenotypic variation in the BLUP analysis, respectively. QSsv.cau-1A.2 explained 8.17–18.62% of the phenotypic variation. The remaining 11 were putative QTLs (Supplementary Table 2).

Considering its stability and the genetic effect, QSsv.cau-1A.1 was chosen to saturate the positioning interval. Then 15 InDel markers near this region were developed according to the re-sequencing results of two parents (Table 3 and Supplementary Table 3) and a new genetic linkage map of 1A long arm was constructed. QTL remapping detected an additional QTL located next to the original interval. These two QTLs were named QSsv.cau-1A.1.1 and QSsv.cau-1A.1.2 (Figure 2).

QSsv.cau-1A.1.1 was detected in six environments and BLUP, explaining 17.21–26.47% of the phenotypic variation, and the favorable allele was from ND3753. The confidence interval was between the markers 1A248 and 1A332 corresponding CS physical position of 386,222,224–463,923,653 bp (Appels et al., 2018; Figure 2 and Supplementary Table 5). QSsv.cau-1A.1.2 was repeatedly detected in three environments as well as BLUP data, contributing 7.02–12.13% of the phenotypic variation and LX99 contributed the favorable allele. The physical position of QSsv.cau-1A.1.2 located on 462,634,655–492,004,197 bp by comparing flanking markers 1A36 and 1A215 to CS RefSeqv1.0 (Appels et al., 2018; Figure 2 and Supplementary Table 5).

The flanking markers 1A248, 1A332 and 1A36, 1A215 delimiting confidence intervals of QSsv.cau-1A.1.1 and QSsv.cau-1A.1.2 (Figure 2) separately were used to detect genotypes in the DH and BC₃F₂-N populations. AA and aa represented genotypes with homozygous favorable and unfavorable alleles of QSsv.cau-1A.1.1 from ND3753, respectively, whereas BB and bb were symbols of that of QSsv.cau-1A.1.2 from LX99.

DH lines could be grouped into three genotypes which included two parental genotypes AAbb, aaBB, and one recombined genotype aabb, and each contained 51, 59, and 13 lines. The reason for the absence of genotype AABB remained unknown. The average SSV values in BLUP of genotype AAbb, aaBB, and aabb were 21.2, 19.9, and 17.8 mL, respectively, of which AAbb was significantly higher than aaBB and both were significantly higher than that of aabb (Figure 3A). This suggested that QSsv.cau-1A.1.1 had a stronger effect on SSV than QSsv.cau-1A.1.2, which was consistent with their contribution rates of phenotypic variation and additive effects in QTL analysis.

In BC₃F₂-N population four allele combinations AaBB, AAbb, aaBB, and aabb were identified, which had SSV average values of 20.4, 17.5, 17.9, and 15.8 mL from 2, 23, 20, and 19 plants, respectively. Similarly, the SSV values of genotypes AaBB, AAbb, and aaBB were significantly higher than that of aabb (Figure 3B). However, although genotype AaBB had the distinct highest value, three genotypes with favorable alleles had no significant difference. This was possibly due to the too-small sample number of genotypes AaBB. Nevertheless, all results above could still prove that the favorable allele has positive effects.

This study did not find the combination type of AABB but found the type AaBB in the BC₃F₂-N population. In summary, the combination of two favorable loci had the high SSV, whereas recombinants with unfavorable loci had the lowest. In future research, the homozygous lines of the BC₃F₂-N population will be extracted to develop the corresponding NIL pairs to further verify their effects and examine their genetic effect on some wheat qualities, such as gluten content, extensograph, farinograph, alveograph, and gluten index.

In order to further verify and narrow down the confidence interval of QSsv.cau-1A.1.1, the BC₃F₁ population in the background of LX99 was genotyped using six InDel markers, and four recombinant types were obtained (Figure 4B). The SSV values of types 3 and 4 were similar and were significantly lower than that of types 1 and 2, indicating that QSsv.cau-1A.1.1 was delimited to the interval between markers 1A1 and 1A366.

The BC₃F₁ individuals with heterozygous genotypes between markers 1A1 and 1A39 were selected to generate the BC₃F₂-L population. The six markers between 1A1 and 1A39 were used to genotype the BC₃F₂-L population, and six recombinant types were obtained (Figure 4C). The SSV value of type 6 was similar to that of types 4 and 5 but was significantly lower than that of the other types (Figure 4D). These results once again proved the effectiveness of QSsv.cau-1A.1.1 and further confirmed that its location was between markers 1A1 and 1A366 (Figure 4C). There was no significant phenotypic difference between the heterozygous genotype and the ND3753 genotype, which implied that the ND3753 genotype was dominant. The corresponding physical interval of CS between these two markers was 14.9 Mb, with 89 putative high-confidence annotated genes (Supplementary Table 6). In addition, further fine mapping of QSsv-cau-1A.1.1 is under research.

Sodium dodecyl sulfate (SDS)-sedimentation volume (SSV) is a comprehensive indicator for indirectly testing wheat quality and one of the important tests to evaluate the gluten strength of flour and is closely related to the processing and baking quality of flour (Axford et al., 1978; Peña-Bautista, 2002; He et al., 2004). SSV is well correlated with other quality traits, such as grain protein content, gluten index, wet gluten content, bread volume, and farinograph parameters (Cubadda et al., 1992).

Our results showed that SSV was significantly positively correlated with stability time, formation time, and water absorption of dough, which is consistent with previous studies (Table 2; Cubadda et al., 1992). This confirms that SSV can be used as a suitable substitute for farinograph indicators that require a great number of samples to evaluate the rheological properties of wheat dough, thereby indirectly measuring the baking and processing quality of wheat flour.

We compared the physical locations of QTLs for SSV reported in previous studies and those revealed in this study (Li et al., 2009; Reif et al., 2011; Deng et al., 2015; Würschum et al., 2016; Yang et al., 2020). The physical position of QSsv.cau-1A.2 was agreed with that of a previously reported QTL controlling SSV (Yang et al., 2020). SSV was found to be affected by allelic variations at Glu-A1 (508,726,618–508,725,448 bp, RefSeqv1.0) and Glu-A3 (4,203,001–4,202,275 bp, RefSeqv1.0) loci in several previous studies (Li et al., 2009; Reif et al., 2011; Deng et al., 2015; Würschum et al., 2016). Some QTLs associated with SSV was reported on chromosome 1A. For instance, Yang et al. (2020) identified a QTL (540,660,000–544,610,000 bp, RefSeqv1.0) for SSV that is located on chromosome 1A by genome-wide association study (GWAS). However, the physical position of QSsv.cau-1A.1.1 (371,573,909–386,426,688 bp, RefSeqv1.0) and QSsv.cau-1A.1.2 (419,490,584–492,004,197 bp, RefSeqv1.0) did not overlap with those of the above-mentioned QTLs/genes, suggesting these two QTLs may be novel. Li et al. (2009) identified a QTL (470,230,000–570,420,000 bp, RefSeqv1.0) for SSV on chromosome 2D using a recombinant inbred line population. Four QTLs (16,340,000, 59,102,000, 615,470,000, 646,600,000 bp, RefSeqv1.0) controlling SSV were reported to be located on chromosome 2D by multi-locus GWAS (Yang et al., 2020). However, the physical position of these QTLs and that of QSsv.cau-2D (140,759,212–467,689,413 bp, RefSeqv1.0) were not consistent, indicating that QSsv.cau-2D may also be a novel QTL. Li et al. (2009) identified a QTL for SSV on chromosome 5DS. SSV was found to be affected by the allelic variation at the Pinb-D1 locus on chromosome 5DS in some previous studies (Li et al., 2009; Reif et al., 2011; Deng et al., 2015; Würschum et al., 2016). However, the physical locations of these QTLs/genes and QSsv.cau-5D.1 does not match, implying that QSsv.cau-5D.1 on chromosome 5DL may be a novel QTL.

Sodium dodecyl sulfate (SDS)-sedimentation volume (SSV) is a quantitative trait affected by both environmental and genetic factors; thus, some QTLs can only be detected in specific environments (Supplementary Table 2). We found 11 such QTLs located on chromosomes 1B, 1D, 2A, 4B, 4D, 5A, 5D, 6B, and 6D, which is consistent with previous results (Li et al., 2009; Kerfal et al., 2010; Reif et al., 2011; Deng et al., 2015; Würschum et al., 2016; Liu et al., 2017a; Mir Drikvand et al., 2018; Goel et al., 2019; Guo et al., 2020; Yang et al., 2020). The Glu-D1 gene, which is located in the interval of QSsv.cau-1D, may be a candidate gene for QSsv.cau-1D. QSsv.cau-4B.1 and QSsv.cau-4D was located next to the dwarf genes Rht-B1 and Rht-D1, respectively. Previous studies have also revealed QTL-enrichment areas near Rht-B1 and Rht-D1, which are associated with kernel size, kernel hardness, kernel protein, pasting properties, and mixing properties (Shanhong et al., 2001; Li et al., 2006; Wang et al., 2012, 2017; Patil et al., 2013; Zhang et al., 2013; Jin et al., 2016; Liu et al., 2017b).

However, there is not enough evidence to support the correlation between Rht-B1/D1 and SSV, and the gene that controls SSV near Rht-B1 and Rht-D1 has not been cloned. We hypothesized: (1) there may be other genes affecting quality traits near Rht-B1 and Rht-D1; (2) allelic variations between Rht-B1 and Rht-D1 may also regulate certain quality traits, such as SSV. However, these hypotheses need to be further tested. The co-localization of dwarf genes and QTLs related to quality traits may also remind breeders to consider the selection of plant height and grain quality in the wheat breeding program.

In recent years, a large number of QTLs for SSV have been identified and characterized through GWAS and linkage analysis, and some of these QTLs are related to allelic variants of Glu-1, Glu-A3, Glu-B3, Gli-B1, and Pina-D1 (Ahn et al., 2014; Deng et al., 2015; Guo et al., 2020). However, most of the other QTLs have not been further verified or fine mapped. SSV is a typical quantitative trait with a complex genetic mechanism. The lack of information on the authenticity and genetic effects of these QTLs for SSV not only hinders the exploration of their genetic and molecular mechanisms but also fails to provide breeders with sufficient new high-quality genetic resources for wheat quality improvement. In this study, we verified the effects of QSsv.cau-1A.1.1 and QSsv.cau-1A.1.2 on SSV and the interaction between the two QTLs in the DH and BC3F2-N populations. Therefore, the InDel markers are closely linked to QSsv.cau-1A.1.1 and QSsv.cau-1A.1.2 developed in this study can be used by breeders to aggregate high-quality genes for wheat quality improvement.

In particular, QSsv.cau-1A.1.1 was delimited to an approximate 14.9 Mb between markers 1A1 and 1A366 (Figure 4). We conduct an orthologous analysis for the candidate region to predict HC genes in QSsv.cau-1A.1.1, but no ones are associated with SSV in Oryza sativa and Arabidopsis thaliana (Supplementary Table 6). This may be because Oryza sativa and Arabidopsis thaliana do not have gluten, and SSV is related to gluten strength in the common wheat.