Aegilops tauschii (2n = 2× = 14, DD) is the diploid progenitor of the D subgenome of hexaploid wheat (Triticum aestivum L., 2n = 6× = 42, AABBDD) and a vital genetic resource for the improvement of wheat quality and yield (Dvořák et al., 1998; Ogbonnaya et al., 2013). The A. tauschii has a rich genetic diversity and multiple biological and abiotic resistances, including excellent genetic resources such as stress resistance (Qin et al., 2016), disease resistance (Zhang et al., 2019), and improved yield, which are uncommon in ordinary hexaploid wheat.

Hexaploid wheat arose through natural hybridization and chromosome doubling between a cultivated allotetraploid (2n = 4× = 28, AABB) and A. tauschii (Dvořák et al., 1998; Matsuoka, 2011). However, common wheat descends from a small number of spontaneous interspecific hybrids (Cox, 1997). Thus, there is scope for A. tauschii to improve wheat and increase wheat yield by artificially synthesizing hexaploids. After several collaborative long-time research efforts, the International Center for Maize and Wheat Improvement (CIMMYT) have synthesized hexaploid wheat lines by crossing elite tetraploid durum with A. tauschii (Ogbonnaya et al., 2013). The A. tauschii has several yield traits or components that may be transferred to synthetic hexaploid wheat when used as a paternal parent. Furthermore, it has previously been used to introgress yield traits into wheat, such as the large-kernel wheat Chuanmai 42 (Zhang et al., 2004) and heavy panicle Shumai 830 (Hao et al., 2019). With the rapid development of scientific research technology, identification of the linked markers in the genetic background of A. tauschii can enable targeted introgressions, thus making it economical.

Single nucleotide polymorphism (SNP) is a third-generation genetic marker technology. SNPs are abundant and have traits such as high frequency and good genetic stability. Currently, SNP genetic studies are widely used for kernel yield, disease resistance, and stress resistance. Genome-wide association study (GWAS) based on linkage disequilibrium (LD) has been widely adopted to identify loci significantly associated with important and complex morphological traits in several species, including A. tauschii (Liu et al., 2015a,b), rice (Chen et al., 2014), wheat (Lin Y. et al., 2017, 2019, 2020a; Liu Y. et al., 2017), and maize (Lu et al., 2010, 2011 Yang et al., 2014). Moreover, only a few GWAS have reported kernel size traits in A. tauschii. For example, using 193 A. tauschii accessions worldwide, 58 SSR were identified in three environments for seven grain traits (Zhao et al., 2015). Using 5,249 SNPs, a GWAS was performed for 114 A. tauschii germplasm, and a total of 17 SNPs associated with grain size traits distributed over all the seven chromosomes (Arora et al., 2017). However, this study aimed to investigate marker-trait associations for kernel size traits using SNPs in a core collection of 223 A. tauschii of diverse origin. Moreover, our objective was to scan candidate gene responses to kernel size traits. These identified genes and SNPs will provide an important research framework for cloning kernel trait genes in A. tauschii.

The improvement of common wheat has gone through the cross between landraces and main popularized varieties to the cross between elite varieties now. However, wheat has been affected by domestication and selection of long-term backbone parents, and genetic “evolutionary bottlenecks” have appeared, which leads to a decrease in yield. Wheat kernel traits are the most important factor affecting yield, and excellent kernel traits greatly increase yield. However, A. tauschii is one of the ancestral species of hexaploid wheat and the donor species of D subgenome. It has a lot of valuable genes and a rich genetic diversity, and a vital genetic resource for the improvement of wheat quality and yield (Dvořák et al., 1998; Ogbonnaya et al., 2013). The purpose of this research is to dig out the excellent genes that regulate kernel size in A. tauschii, and lay a foundation for the transfer of common wheat and the broadening of genetic diversity. There have been many successful cases of introducing A. tauschii genes into wheat through hybridization. Such as Chuanmai 42, Shumai 969, and Shumai 830 etc. thereinto, Chuanmai 42 is a large and heavy spike cultivar with a large kernel (Zhang et al., 2004; Duan et al., 2006), and Shumai 830, is also a heavy spike cultivar (Hao et al., 2019). These successful wheat varieties suggest the considerable potential of A. tauschii for wheat improvement, especially for breeding cultivars with large and heavy spikes. However, the aim of this study was to discover genes in A. tauschii that regulate kernel size, introducing the significant target markers that affect kernel traits directly into wheat would accelerate the breeding of target varieties and save time.

In this study, 223 A. tauschii were divided into T-group and S-group subgroups through population STRUCTURE. T-group and S-group representing A. tauschii ssp. tauschii and A. tauschii ssp. strangulata, respectively. The results of this study are completely consistent with previous studies (Arora et al., 2017). Population structure is one of the most important factors affecting LD (Flint-Garcia et al., 2003). Thus, the LD analyses was conducted separately for the T-group and S-group, respectively. Only one previous study reported LD decay distance in A. tauschii. The LD decay distance was reported at 9.8 and 2.7 cM for T-group and S-group, respectively (Arora et al., 2017). This study firstly reported the LD decay distance based on physical distance. The LD decay distance was highest in the T-group (approximately 110 kb) and S-group (approximately 290 kb), the average LD decay distance was approximately 200 kb. In wheat, the LD decay distance was 250 kb for D subgenome, consistent with our results (Long et al., 2019).

The present study aimed to identify significant markers for kernel size trait in A. tauschii, the D subgenome donor of hexaploid wheat. Significant (p < 0.001) differences were noted among genotypes and environments; A. tauschii showed high diversity, indicating high research and utilization value, and it could make a major contribution to broadening the genetic diversity of the wheat D subgenome. In the present study, heritabilities of kernel traits were medium to high. In previous studies, moderate or high heritabilities were also observed in A. tauschii (Arora et al., 2017), consistent with our study. In some reports in wheat, moderate to high heritability were also observed (Kuchel et al., 2007; Liu Y. et al., 2017). These results indicated that the kernel size-related traits was more controlled by genetic factors. We used correlation and linear regression analyses revealed a significant positive correlation between KL, KW, KV, and HKW, and HKW increased with increasing KL, KW, and KV. Previous studies have found that, in the tetraploid and hexaploid wheat, TaGW2-A1 mutations could increase KW and KL, thereby increasing yield by increasing TKW (Simmonds et al., 2016). The correlations between these trait points to a causal relationship between kernel size and weight because longer and wider kernels can accumulate more starch and, therefore, have greater kernel weight (Duan and Sun, 2005). Our study showed the consistent results. Thus, in the process of breeding, breeders could pay attention to discovering varieties with long or wide kernels. The identified A. tauschii accessions with long or wide kernels could be used in further breeding through SHW to broaden the genetic diversity of wheat.

In the present study, cluster analysis results were highly consistent (83%) with the population structure results. Similar results have been reported in A. tauschii and wheat (Liu et al., 2015b; Liu Y. et al., 2017). A previous cluster analysis performed using 29 morphological traits in 322 A. tauschii showed 72% consistency with the population structure results (Liu et al., 2015b). This result may be caused by an intermediate type between A. tauschii ssp. tauschii and A. tauschii ssp. strangulata subspecies (Zhao et al., 2018). Indeed, Kihara et al. (1965) reported that intermediate forms and hybrids existed between the two subspecies. Due to the presence of intermediate type or hybrids, there may be uncertainty in morphological identification, resulting in differences between morphology- and genotype-based identification. Besides, morphological traits may also be more easily affected by the ecological environment. Plant growth will be affected to varying degrees in different ecological environments, affecting plant growth and, ultimately, kernel size (Wang et al., 2009; Liu et al., 2015b). Thus, the morphological traits clustering is roughly correct, but there will be some classification errors. Compared with genotype data classification results, cluster analysis could correctly classify most A. tauschii.

For GWAS results, the threshold is set to Bonferroni correction method a = 0.01 or 0.05. Because MLM is too strict and can lead to over-correction. The Bonferroni correction method a = 1 to reduce negative errors caused by overcorrection (Yang et al., 2014). The p-value was 1.49 × 10^–4 for the 6,723 SNPs, with a corresponding −log₁₀^(p) = 3.80. However, only 22 significant SNPs were identified. This is due to overcorrection caused by MLM. In addition, −log₁₀^(p) = 3.00 is also commonly set as a threshold (Liu J. et al., 2017; Ye et al., 2019; Fu et al., 2020) to reduce the negative false rate. Meanwhile, it has also been successfully applied in GWAS in wheat (Liu J. et al., 2017; Ye et al., 2019; Fu et al., 2020). This indicates that this threshold is a frequently used empirical value and the results are reliable. In this study, the threshold setting as −log₁₀^(p) = 3.00, 66 significant loci were identified for kernel traits using 6,723 SNPs markers by GWAS. To the best of our knowledge, to date, only two studies have reported QTL for kernel characteristic traits in A. tauschii based on GWAS (Zhao et al., 2015; Arora et al., 2017). Owing to the lack of publicly available marker sequence information, our results were compared to those of previous studies based on chromosomes. In previous studies for KL, significant loci were identified on chromosomes 1D, 2D, 5D, and 6D (Zhao et al., 2015; Arora et al., 2017), while in this study, 11 significant markers were identified on chromosome 7D; these may represent novel loci. For KW, we detected 28 significant markers on chromosomes 2D, 3D, 4D, 5D, and 7D. Significant markers on these chromosomes were also found in previous studies (Zhao et al., 2015; Arora et al., 2017). For KSA, 14 significant markers were identified on chromosomes 2D, 4D, 5D, and 7D, and loci on chromosome 4D may be novel according to a comparison with previous research (Zhao et al., 2015). For HKW, loci were identified on all chromosomes in previous studies, except for 7D (Zhao et al., 2015; Arora et al., 2017). In this study, significant markers were identified on chromosome 7D, as well as on chromosomes 3D, 4D, 5D, and 6D. Besides, 21 significant markers for KWL were identified on all seven chromosomes, and 22 for KV were identified on chromosomes 2D, 4D, 5D, and 7D. GWAS is an important and effective approach for wheat breeding by helping in the design of hybrid crosses. The identified significant markers/variants can be designed for use as molecular markers in wheat breeding directly.

To further serve the breeding of target varieties with desirable kernel traits, six candidate genes were identified based on homologous functions in arabidopsis and rice. AET2Gv20774800 was flanking the SNP marker GBF1XID01D2CAC_283. This gene was homologous to the MST1 gene in arabidopsis; they are also called sulfurtransferases 1 (STR1). Of note, a mutation of STR1 alone resulted in a shrunken seed phenotype. The shrunken seed phenotype was associated with delayed/arrested embryo development (Mao et al., 2011). In addition, the MST1 family gene OsMST5 plays an important role in early seed development in rice (Takeda et al., 2001). AET5Gv20084100 was flanking SNP marker F5XZDLF01AU4HH_125 for KW on chromosome 5D. It was homologous to the ZAR1 gene in arabidopsis, which belongs to the RLK/Pelle kinase family (Yu et al., 2016). Maize ARGOS1 (ZAR1) transgenic alleles increase hybrid maize yield, as ZAR1 increased plant and organ size primarily through increasing cell numbers (Guo et al., 2013). AET5Gv20005900 was flanking SNP marker be405667Contig1ATwsnp1 for KV and KSA on chromosome 5D, and it was homologous to ETR1 in arabidopsis, respectively. In Rice, a reduction of ETR2 expression could increase the thousand-seed weight (Wuriyanghan et al., 2009). However, enhanced ethylene response also may be related to a larger or heavier kernel (Yin et al., 2017). The starch and protein of rice kernel determine factors of seed dry weight and size (Duan and Sun, 2005; Yin et al., 2017). Thus, AET5Gv20005900 may presumably affect starch accumulation in A. tauschii, finally affecting kernel size. The AET4Gv20799000 gene related to KW and KWL was predicted on the 4D chromosome. It was homologous to the MAC3B gene in arabidopsis, which belongs to the U-box family. U-box is a ubiquitin ligase activity-related protein domain in plants (Li and Li, 2014; Li S. et al., 2018). Ubiquitin ligases have been identified as key factors of seed size control in plants (Li N. et al., 2018). For example, grain width gene GW2 encodes a RING-type E3 ubiquitin ligase and controls kernel width and weight in rice (Matsuoka and Ashikari, 2007). Finally, AET7Gv20644900 was homologous to NAC047 in arabidopsis and belonged to the NAC gene family. It was reported that NAC2 regulates embryogenesis, affecting seed shapes in arabidopsis (Kunieda et al., 2008; Mathew et al., 2016). As NAC2 and NAC047 belong to the same family, we speculate that these candidate genes may affect embryo development, kernel size, and yield. The genes identified by GWAS can be speeds up selective breeding using CRISPR-Cas9 system, which is a powerful tool for rapid and effective genetic improvement and allows several QTL/genes to be edited precisely and simultaneously or even novel alleles to be created.

This in-depth study of A. tauschii provides new insight into its potential role in wheat improvement. Six kernel traits, including KL, KW, KWL, KSA, KV, and HKW, were evaluated among 223 A. tauschii over 3 years. H’ was in the range 0.80-0.89, showing that A. tauschii had high diversity. Kernel traits showed medium to high heritability (0.74-0.87), and correlation and linear regressions analyses showed that HKW increased with increasing KL, KW, and KV. Kernel size traits affected kernel weight and, subsequently, yield. Our research results revealed that there are favorable varieties with longer and wider kernels in both subspecies of A. tauschii. Based on BLUP values, a total of 66 significant SNPs was identified using GWAS, and six candidate genes were identified as potential genetic drivers of these yield-related traits. The identified SNPs/genes will speed up the wheat breeding by MAS and genome-editing technology. It is expected that the excellent target gene from the D subgenome can be successfully introduced into wheat, so as to increase the yield of wheat and broaden the genomic resources of wheat.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

QW and NY drafted and revised the manuscript and contributed to data analysis. HC, HH, YL, HS, KZ, XJ, and SY performed the phenotypic evaluation and helped with data analysis. SL helped to perform linkage disequilibrium analysis and revise the manuscript. CL, GC, and ZY helped to draft the manuscript. YL designed and coordinated the study and revised the manuscript. All authors have read and approved the final manuscript for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.