Molecular markers, population structure, and linkage disequilibrium (LD)
The core collection of barley accessions belongs to ten geographical regions (Figure 1). There are two main clusters among the accessions with biological status and row-type whereas most of the accessions were 2-rowed and landraces were 6-rowed (Figure 2a and b). The accessions of the collection were originally from different geographical origins and could see a trend among the accessions that most of Northern Europe, Northern America, and South America were clustered around the center while the accessions from other origins were distributed (Figure 2c). These results showed a wide diversity among the landraces coming from central Asia, the Near and the Far East, North Africa, and Ethiopia compared to accessions mostly coming from Europe and the Americas (Figure 2).
All of the barley accessions were genotyped using GBS, which yielded 19,276 SNPs as described by Milner, et al. 23. Information on the number of markers for each chromosome, with the map length and marker density for each chromosome, are presented in Figure S1. The largest number of markers were detected on chromosome 2H by 3,373 SNPs, while the highest marker density was detected on chromosome 7H. Only 2,193 markers were noted on chromosome 1H with the lowest marker density among chromosomes. For the whole barley genome, the overall average values of r2 decreased as the distance between SNPs increased as well as the average LD decay distance was approximately 2.4 Mbp (Figure S2).
Natural variation of agronomic trait performance
Barley accessions showed excessive significant phenotypic variation for all morphological and yield-related traits with normal distribution under both treatment conditions (Figure S3). The minimum, maximum, and mean of each trait for all genotypes are presented in Table S2. All yield traits (SL, NSS, NGS, WGS, and TKW) had a lower mean performance under salinity stress than under control conditions as expected (Table S2 and Figure S8). On average, all yield traits had a reduction due to salt stress ranging from 15% (NGS) to 48% (WGS). Based on salt-tolerant indices (STI), genotypic mean values for all yield traits ranged from 52% (WGS) to 85% (NGS) (Table S2). The broad-sense heritability (H2) values were ranged from 99.38 for NGS_C to 96.32 for NSS_C. In response to salt stress, the highest H2 was detected for NGS (99.3) followed by SL (97.91) and WGS (97.06). Moreover, H2 was calculated and ranged from 53.64 (TKW_STI) to 81.26 (SL_STI) (Table S2).
Natural variation of enzymatic antioxidants
Significant natural variation was also detected for all enzymatic antioxidant activities with normal distribution in response to salinity conditions (Figure S4). Under control conditions, the genotypic mean values for SOD, CAT, APX, and GR were ranged from 2.29 (SOD) to 1.10 (APX). The mean values were ranged from 5.60 (SOD) to 2.37 (APX) under salt treatment. A significant increment in SOD, CAT, APX, and GR was observed under salt stress (58%, 48%, 50%, and 65%), respectively, compared to the control condition (Table S4 and Figure S9). For STI, genotypic mean values ranged from 252.18. (APX) to 1986.54 (CAT). The highest value was detected for CAT_STI by 53666.67. The highest values of H2 were observed for CAT_S by 99.74 and for GR _C by 97.93. H2 values ranged from 77.96 (SOD_STI) to 99.74 (CAT_STI) as shown in Table S3.
Natural variation of proline contents, soluble protein, and non-enzymatic antioxidants
Natural variation was detected for all non-enzymatic antioxidants and compatible solutes which expressed normal distribution under salt stress (Figure S5 and Figure S6). Proline content (PC) accumulation remarkably increased by 47% under salt treatment (Figure S10a), compared to the control condition (Table S4). Soluble protein (SP) significantly decreased under salt stress by 57% when compared to the control (Figure S10b). The mean salt tolerance indices for SPC and PC were 43 and 190, respectively. A significant increase in AsA, GSH, TPC, TFC, and TAC was detected under salt treatment by 65%, 49%, 32%, 66%, and 72%, respectively, in comparison to the control across over genotypes (Table S4). Genotypic mean values for such traits STI ranged from 154.72 (TPC) to 542 (AsA) (Table S4). H2 values were ranged from 97.94 for AsA_C to 61.78 for TAC_C. Under salt treatment, the highest H2 was detected for AsA (95.27) followed by TFC (89.13) and SP (76.3). H2 values for STI were low for PC_STI (34.70) and reached 95.27 in AsA_STI
Natural variation of ionic Na+ and K+
Na+ content showed a significant increment by 87% under salt treatment compared to the control treatment (Table S5). In contrast, K+ accumulation was decreased by 67% under salt treatment in comparison to the control across over accessions (Table S5). Moreover, the K+/ Na+ ratio was significantly increased under salt stress by 27% when compared to the control treatment (Table S5). For STI, mean values were 768.51, 32.64, and 4.25 for Na+, K+, and the K+/ Na+ ratio, respectively. Additionally, high H2 values were detected which ranged from 84.06 (K+/ Na+) to 90.87 (Na+) (Table S5).
Pearson's correlations among traits based on the mean of all accessions under each treatment condition reveal a significant correlation was observed amongst all studied traits under both treatments. SOD showed positive significant correlation at p-value 0.0001 **** with CAT, APX, and GR (r = 0.55***, 0.99***, 0.86***), respectively. Similarly, a positive significant correlation was noted between SOD and all agronomic traits except NGS and WGS. In contrast, there was a negative correlation between SOD activity and non-enzymatic antioxidants such as AsA, TPC, TFC, TAC, and compatible solutes such as SP and PC. Moreover, APX showed positive correlations with SOD, CAT, GR, and TKW (r = 0.99***, 0.54***, 0.85***, 0.26), respectively (Figure 3a).
Under salinity conditions, SOD activity detected positive associations with all physiological and antioxidant systems where the highest significant associations were detected for APX and GR (0.87***, 0.75***), respectively. Contrarily, SOD was negatively correlated with all agronomic traits except for NGS and WGS. Interestingly, high significant positive correlations were detected between APX activity with SOD, CAT, GR, AsA and GSH (r = 0.87***, 0.75***, 0.86***, 0.70***, 0.70***), respectively. Like SOD, APX was negatively correlated with all agronomic traits except for NGS and WGS (Figure 3b).
Genetic associations and candidate genes underlying the studied traits
GWAS identified a total of 299 SNPs as significantly associated with 19 traits and passing thresholds of (−Log10 (p) ≥ 4.0) and with R2 ≥ 10% under both control and salinity conditions (Table S6). Out of the 299 SNPs, we identified 163 SNPs (-log10 p-value ≥ 4) that had highly significant associations with enzymatic and non-enzymatic antioxidants under both stress and non-stress conditions (Table S6). For example, we observed sixteen, fifteen, fourteen, eleven, and ten SNPs associated respectively with APX_STI, CAT_STI, AsA_C, CAT_S, and SOD_S, whereas less than ten SNPs were associated with each of the remaining traits. Moreover, thirty-seven SNP markers were significantly associated with Na+, K+, and K+/Na+ ratio where the highest number of was detected for K+/Na+_S (12 SNPs), followed by K+/Na+_C (9 SNPs), whereas less than nine SNPs were associated with each of the remaining traits under both control and salinity treatments. Compatible solutes such as soluble protein (SP) and proline content (PC) showed 13 significant SNP markers present on chromosomes 1H, 2H, and 3H. For instance, we detected eight, four, and one SNPs highly associated, respectively, with SP_STI, SP_C, and PC_STI with (-log10 p-value ≥ 4).
For all agronomic traits, 86 SNPs (-log10 p-value ≥ 4) were significantly detected on all chromosomes (Table S7). The highest number of markers was detected for SL_S (23 SNPs), followed by NGS_C and SL_C (11 SNPs each), whereas ten SNPs were associated with NGS_S and eight SNPs for TKW_C.
Interestingly, highly significant associations were discovered on chromosomes which were found to be highly associated with the antioxidant system, K+/Na+ ratio, and some agronomic traits (Table S8 and Table S9). In this investigation, we identified highly significant SNPs associated with enzymatic and non-enzymatic antioxidants present on all chromosomes. For instance, we observed five, seven, three, and one SNPs highly associated respectively with CAT_S, CAT_STI, APX_STI, and TFC_STI with a -log10 (p-value) > 6 (i.e. p-value <10–6). For K+/Na+_C, four highly significant associations were detected on chromosomes 2H, 5H, and 7H. Moreover, we identified highly significant SNPs associated with some agronomic traits distributed on all over chromosomes. For instance, we observed four, five, three, and four SNPs highly associated with NGS_S, NGS_C, TKW_S, and TKW_C, respectively.
This study revealed five genomic regions based on marker-trait associations (MTAs), which harbors 193 potential candidate genes which are distributed on chromosomes 1H, 2H, 4H, 6H, and 7H (Table S10). Of these, 23 potential candidate genes were found to control all enzymatic and non-enzymatic antioxidant components under salt stress (Table 1).
On chromosome 1H, two candidate genes namely Pentatricopeptide repeat (PPR) superfamily protein that control the variation of Na+_C, K+_C, Na+_STI, TFC_C, and TFC_STI, at position (669,733,8-669,734,9 bp) and FBD-associated F-box protein that control the variation of APX_STI and SOD_STI at position (500,0276,4-512,105,27 bp). Exclusively, chromosome 2H harbors several candidate genes that are found to be associated with most enzymatic and non-enzymatic antioxidant components such as SOD, APX, GR, AsA, and GSH under salt stress (Figure 4). The most prominent candidate gene is HORVU.MOREX.r3.2HG0181470.1 at position (557,725,076-557,730,772 bp) annotated as Beta-glucosidase, putative. The third genomic region is located on chromosome 4H,chr4H:13826536:A:G SNP (138,265,36 bp) and chr4H:569747682:G:A SNP (569,747,682 bp) inside the HORVU.MOREX.r3.4HG0336230.1 (138,232,77-138,249,82 bp) annotated as Glycosyltransferase for SL_STI, and gene HORVU.MOREX.r3.4HG0405320.1 (569,744,191-569,748,338 bp) annotated 3-ketoacyl-CoA synthase that control the variation of K+/Na+_S, K+/Na+_STI, and K+_STI (Figure 5, Table S8 and S9). Interestingly, the fourth genomic region was detected on 6H, chr6H:451682318:A:T SNP near the gene HORVU.MOREX.r3.6HG0605480.1 (451,676,227- 451,677,811 bp) annotated as PLATZ transcription factor, regulating the variation of CAT_S, APX_S, GR_S, AsA_S, and GSH_S related to antioxidant components (Figure 4). Ultimately, the the genomic region at 7H was found to be associated with SOD_S and APX_S at 153,773,211 bp chr7H:153773211:A:C SNP inside the gene HORVU.MOREX.r3.7HG0676830.1 (153,772,300- 153,774,057 bp) that encodes L-gulonolactone oxidase, regulating the variation of SOD_S and APX_S related to enzymatic antioxidant components. The allelic variation at chr7H:153773211:A:C SNP inside the L-gulonolactone oxidase gene demonstrates a negative selection of accessions carrying C allele. This allele appears in a cultivar with lower activity of enzymatic antioxidants e.g. superoxide dismutase and ascorbate peroxidase under salinity conditions. (Figure 6).