QTL Mapping of Agronomic and Physiological Traits at the Seedling and Maturity Stages under Different Nitrogen Treatments in Barley

Nitrogen (N) stress seriously constrains barley (Hordeum vulgare L.) production globally by influencing its growth and development. In this study, we used a recombinant inbred line (RIL) population of 121 crosses between the variety Baudin and the wild barley accession CN4027 to detect QTL for 27 traits at the seedling stage in hydroponic culture trials and 12 traits at the maturity stage in field trials both under two N treatments, aiming to uncover favorable alleles for N tolerance in wild barley. In total, eight stable QTL and seven QTL clusters were detected. Among them, the stable QTL Qtgw.sau-2H located in a 0.46 cM interval on the chromosome arm 2HL was a novel QTL specific for low N. Notably, Clusters C4 and C7 contained QTL for traits at both the seedling and maturity stages. In addition, four stable QTLs in Cluster C4 were identified. Furthermore, a gene (HORVU2Hr1G080990.1) related to grain protein in the interval of Qtgw.sau-2H was predicted. Correlation analysis and QTL mapping showed that different N treatments significantly affected agronomic and physiological traits at the seedling and maturity stages. These results provide valuable information for understanding N tolerance as well as breeding and utilizing the loci of interest in barley.


Introduction
Nitrogen (N) is one of the indispensable nutrient elements for plant growth and development [1]. The increase in crop yield depends mainly on the application of N fertilizer [2]. Over the past 40 years, the doubling of global grain production has been closely related to the surge in N application rates [3]. Excessive application of N fertilizer will decrease N use efficiency (NUE) and cause a series of environmental problems [4], such as global acidification, the destruction of stratospheric ozone and the eutrophication of water bodies [5]. As global soil waterlogging is expected to increase under the future climate, this will further lead to more N losses [6]. With the decrease in global planting area and the increase in food demand, cultivating crops with a high NUE plays a very important role in reducing the amount of N fertilizer applied and increasing grain yield (GY) [7].
Low N (LN) stress is a compound trait with a complex genetic basis. Many agronomic and physiological traits are involved in the response to LN stress in crops [8,9]. N deficiency will result in numerous adverse effects on the growth and development of crops. For example, it can phenotypically decrease plant height (PH), dry matter biomass and leaf area [10] and severely affect the root system's architecture and yield-related traits [11]. It can also physiologically reduce the N content and a series of enzyme activities in the leaves [12]. However, some studies have shown that there are great differences in LN tolerance among different crop genotypes [13,14], which may provide a theoretical basis for breeding LN-tolerant cultivars.
QTL for agronomic and physiological traits have been mapped in several crops under N stress, including wheat [15][16][17][18][19], barley [20][21][22][23][24][25][26], maize [27][28][29] and rice [30,31]. For example, Saal et al. [20] detected 82 QTL for yield-related traits in two N treatments and six environments. Hoffmann et al. [22] located 58 QTL and identified beneficial wild barley alleles that improved the performance of several traits in response to different N treatments. Schnaithmann and Pillen [26] detected 65 QTL for 15 traits under different N treatments on seven chromosomes. Kindu et al. [25] detected 41 QTL on seven chromosomes for 18 traits in different treatments. Although some QTL for agronomic and physiological traits were identified, few of them were stably expressed, meaning that they were hard to utilize in breeding. More QTL for N tolerance under realistic barren soil conditions should be detected for marker-assisted selection (MAS) breeding.
Compared with other crops, barley is one of the earliest domesticated plants. It can grow under a wider range of conditions, especially under various stresses [32]. With the development of modern breeding and intensive farming, the genetic diversity of cultivated barley varieties has been reduced and many excellent alleles have been lost [33]. As a result, many cultivated barley varieties are sensitive to a variety of abiotic and biotic stress environments. Wild barley can maintain normal growth and productivity in soils with poor fertility. It has a rich genetic diversity and can provide excellent genes for breeding barley, thereby breaking the bottleneck of genetic improvement [34]. For example, Schmalenbach et al. [35] reported that 1000-grain weight (TGW) increased when wild barley alleles (Hsp) were introgressed into spring barley. Wild barley is generally better adapted to poor soil fertility. Therefore, we can assume that wild barley may contain some unique N tolerance genes. In this study, 39 agronomic and physiological traits were measured in a population of 121 recombinant inbred lines derived from the variety Baudin and the wild barley accession CN4027 in a hydroponic culture trial with low N (LN) and normal N (NN) treatments in a single timeframe, and in field trials with N deficiency (LN) and N sufficiency (NN) over two growing seasons, with the aim of uncovering favorable alleles for N stress tolerance in wild barley.
Six growth-related traits, including shoot dry matter weight (SDW, g/plant) and root dry matter weight (RDW, g/plant), were measured using a 1/10,000 electronic balance. The dry root-shoot ratio (RSA, RDW/SDW) and total dry matter weight (TDW, SDW + RDW) were calculated. PH (cm/plant) was measured with a ruler. Leaf number (Ln) was recorded by counting the number of leaves longer than 2 cm 30 days after planting. The average values from three plants under LN and NN for each line were used for the following analyses.
For T1, the values for all phenotype traits except Ln were significantly influenced by different N treatments (Table 1). The SDW, RDW, TDW, PH, TRV, ARL and ARSA of CN4027 were significantly higher (p < 0.01) than those of Baudin at LN, whereas ARTN and ARTNP were significantly lower (p < 0.01). The TRL and ARTN of Baudin were significantly higher than CN4027 under NN (p < 0.01). There were no significant differences in RSA, Ln, TRS, ARD, LRL, LRSA, LRV, ARV, LRTN, LRLP, ARLP, LRSAP, ARSAP, LRVP, ARVP and LRTNP under the two N treatments. The coefficient of variation for the growth-related traits and root-related traits between the two N treatments ranged from 0.45 to 51.83% (Table 1). Among the RILs, transgressive segregation was observed for all traits (Table 1), indicating that they are typical quantitative traits.   (Table 2). CN4027 had a significantly higher PH, AL, LDR, StDW, and ADW than Baudin under the two different N treatments, whereas SL, GP, GN, TGW, GY and HI were significantly lower (p < 0.01). The SN of CN4027 was significantly higher than that of Baudin under LN (p < 0.01). The coefficient of variation ranged from 1.77 to 56.85% for all traits. Among the RILs, transgressive segregation was also observed for all traits ( Table 2).

Correlation Analysis between Investigated Traits
For T1, the correlations between growth-and root-related phenotypic traits under the two N treatments are presented in Table 3. Spearman's correlations for growth-and root-related traits ranged from −0.17 to 0.98. Growth-related traits were significantly and positively correlated with all root-related traits under the NN treatment except for ARD, TRV and ARSA (p < 0.05). Under the LN treatment, significant and positive correlations were detected between root-related traits and both SDW and Ln (p < 0.01). TDW and PH were partly correlated with root-related traits (p < 0.05). However, RDW and RSR were not significantly correlated with root-related traits. For T2 and T3, correlations between phenotypic agronomic traits and yield-related traits were listed in Table 4. Spearman's correlation coefficient ranged from −0.81 to 0.92 (Table 4). Significant correlations were detected among each of the agronomic traits and most of the yield-related traits under both N conditions (p < 0.05). Under the LN treatment, agronomic traits, including SL, AL and LDR, were significantly and positively correlated with all the yield-related traits except StDW (Table 4). Under the NN treatment, SN, GN and TGW had significant positive correlations with SL, AL, LDR and GP (p < 0.05). Significant and positive correlations were detected for GY with PH, SL, AL and LDR and for StDW with PH, AL, LDR and GP (p < 0.05). ADW was significantly and positively correlated with all agronomic traits (p < 0.05). Consistently, significant and positive correlations were detected between HI and SL and between AL and GP (p < 0.01). Significant and negative correlations were detected between HI and both PH and LDR (p < 0.01) under the two N treatments at T2 and T3.

QTL for Seedling Traits
Twenty-one QTL for 27 traits, individually explaining 14.20-34.60% of the phenotypic variation rate (PVE) with a LOD value of 2.72-7.37, were detected. These were located on four chromosomes: 3H (9QTL), 5H (1), 6H (10) and 7H (1) (Figure 1 and Table 5). Positive alleles of 12 QTL were from CN4027, as well as 9 from Baudin. Interestingly, only Qph.sau-3H for PH, explaining 14.5-34.6% of the PVE, was detected under both LN and NN conditions, and its positive allele was derived from CN4027. Other QTL were detected under either the LN or NN treatment.
In total, 16 QTL for yield-related traits were detected, and the PVE explained by them ranged from 11.70% to 57.30% (Figure 1 and Table 5). The positive alleles of nine traits were contributed by Baudin, and those of the remaining seven traits were from CN4027. Of these, Qgn.sau-3H was detected on chromosome 3H, which explained 22.20% to 57.30% of the PVE under both N treatments in both T2 and T3. Qtgw.sau-2H was detected under LN in both T2 and T3, suggesting that it belongs to an LN-specific QTL. It explained up to 19.40% of the PVE, with a LOD value of 5.29. The positive allele of this trait was from Figure 1. Quantitative trait loci (QTL) for agronomic and physiological traits identified at the seedling and maturity stages in the RIL population. Blue represents QTL conferring growth-related traits; green represents QTL conferring root-related traits; red represents QTL conferring yield-related traits; purple represents QTL conferring agronomic traits. 1H, chromosome 1H; 2H, chromosome 2H; 3H, chromosome 3H; 4H, chromosome 4H; 5H, chromosome 5H; 6H, chromosome 6H; 7H, chromosome 7H.
In total, 16 QTL for yield-related traits were detected, and the PVE explained by them ranged from 11.70% to 57.30% (Figure 1 and Table 5). The positive alleles of nine traits were contributed by Baudin, and those of the remaining seven traits were from CN4027. Of these, Qgn.sau-3H was detected on chromosome 3H, which explained 22 For agronomic traits, 16 QTL were detected. The PVE explained by them was between 11.50% and 62.20% (Figure 1 and Table 5). The positive alleles of six traits were from Baudin, and 10 were from CN4027. Qph.sau-3H was detected under both N treatments in T2 and T3 and had the largest PVE (62.20%), with a LOD value of 24.45. Its positive allele was from CN4027. Qal.sau-3H was mapped to chromosome 3H under both N treatments in both trials conducted, explaining 12.3% to 30.10% of the PVE. Qldr.sau-3H was detected under four treatments and explained up to 35.80% of the PVE, with a LOD value of 11.27. Qgp.sau-3H was detected under the two different N treatments in T3 but only under the NN treatment in T2. with a PVE of 24.70% and a LOD value of 7.21. Its positive allele was derived from Baudin.

QTL Clusters
The QTL for different traits cluster together in a given interval, which is usually pleiotropic and important [37]. In this study, seven QTL clusters (C1-C7) were mapped to three chromosomes (2H, 3H and 6H). More than three QTL for each cluster were co-located (Table S3). These clusters could be classified into three types: those detected only at the seedling stage (Type I, including C2, C5 and C6) or the maturity stage (Type II, including C1 and C3), and those detected simultaneously at the seedling and maturity stages (Type III, including C4 and C7). Among them, Type I contained QTL only for root-related traits at the seedling stage. Notably, four LN-specific QTL for LPLP, LRSAP, ARSAP and ARLP were located in the region of 15.21-34.83 cM on chromosome 6H in C6. Type II contained QTL for both agronomic traits and yield-related traits at maturity. Four stable QTL (Qgn.sau-3H, Qgy.sau-3H, Qhi.sau-3H and Qal.sau-3H) in C3 were detected under four treatments and had higher PVE values (more than 12.30%). Type III contained QTL for traits at both seedling and maturity. Three of the 10 QTL in C4 were stable, namely Qph.sau-3H, Qldr.sau-3H and Qgp.sau-3H. They could be detected under all treatments and individually explained 12.30-62.20% of the PVE. In C7, QTL for LRLP and ARLP at the seedling stage and for TGW at maturity were co-located in the region of 58.84-66.07 cM on chromosome 6H.

Discussion
It has been shown that a limited N supply constrains crop production globally by influencing the crops' growth and development [20,38]. For example, N deficiency can decrease barley yield, shorten the growth period and also decrease total biomass, the N content of organs and a series of enzyme activities [39][40][41]. Due to the complexity of physiological phenotypic characteristics, there is no standard evaluation index for a crop's tolerance to LN stress. In this study, 39 traits under two N treatments at the seedling and maturity stages were measured to further evaluate the indexes of N tolerance in barley. Correlation analysis showed that different N treatments strongly affected the correlations among different types of traits at the seedling and maturity stages. In addition, the seven QTL clusters identified in this study provided strong evidence that the genetic mechanisms of growth-related traits, root-related traits, agronomic traits and yield-related traits in barley were closely related under different N treatments. In conclusion, the correlation analysis and QTL mapping showed that there was a close relationship between the four N tolerance evaluation indexes in barley under the different N treatments.
Under LN stress, plants may adopt different strategies to cope with the change in N [42]. In comparison with cultivated barley, wild barley has wider genetic diversity and is better adapted to N deficiency [43]. Identification of the resources of wild barley with higher tolerance to LN will provide excellent genetic material for improving LN tolerance in barley and other cereal crops [42]. QTL analysis suggested that the major Qtgw.sau-2H was stably expressed under only the LN treatment and the positive allele was from wild barley (CN4027). We speculate that it is an LN-sensitive QTL and has the potential for breeding to develop barley varieties with LN tolerance.
Roots, as the most important organ for N absorption, can adapt to changes in the N supply by changing their morphology over time [44]. In this study, Baudin had a larger LRLP and LRSAP, while CN4027 had a larger ARLP and ARSAP under LN. Thus, the characterization and aggregation of LN-specific QTL in different varieties may be helpful for breeding high-NUE varieties [45]. Remarkably, there was a co-localized region for rootrelated traits under LN. Qlrlp.sau-6H.1, Qarlp.sau-6H.1, Qlrsap.sau-6H and Qarsap.sau-6H were co-located between bpb-3665654 and bpb-5256842 on chromosome arm 6HS, indicating that the favorable alleles for this region may increase N absorption capacity by changing root systems and root phenotypes. Eight stably expressed QTL for traits identified at maturity under more than three treatments were obtained in this study. To confirm whether the newly identified QTL were novel loci, we compared their physical intervals with previously reported QTL through anchoring their flanking markers on barley using BLASTn analyses against the Morex genome database. Qgy.sau-3H, Qhi.sau-3H, Qgn.sau-3H and Qal.sau-3H were mapped in the interval between 16.54 and 435.82 Mb on chromosome 3H. This interval overlapped with QTL for GY on 3H (345.20 Mbp) [46], with QTL for AL [47] and a QTL hotspot for yield-related traits [48]. In addition, Qph.sau-3H, Qldr.sau-3H and Qgp.sau-3H were identified in the interval between 473.84 and 528.17 Mb on chromosome arm 3HL, which overlapped with the sdw1/denso genes for PH [49,50]. These results suggested that they may not be novel loci.
An LN-specific Qtgw.sau-2H for TGW was located in a 0.46 cM interval and physically mapped between 524.23 and 524.63 Mbp on chromosome arm 2HL. Few QTL-controlling TGW have been reported on chromosome 2H. For example, Wang et al. [48] identified some QTL for grain length, grain width, grain length-width ratio, TGW, grain area, grain perimeter, grain diameter, grain roundness and factor form density between 647.83 and 653.98 Mbp on chromosome arm 2HL. Wang et al. [51] detected qTgw2-1 and qTgw2-2 for TGW on 2HL, which were close to the morphological marker Vrs1 (690.31 Mbp). These previously reported QTL were unlikely to be allelic to the Qtgw.sau-2H locus identified in this study, suggesting that Qtgw.sau-2H might be a novel locus.
In the present study, Qtgw.sau-2H was located between the markers bpb3927777 and bpb3396835. Only one candidate gene was predicted in this region (Table S4). The colocated intervals for Qgy.sau-3H, Qhi.sau-3H, Qgn.sau-3H and Qal.sau-3H were located between the markers bpb6282426 and bpb3264570; there were eight candidate genes in this region. The intervals for Qph.sau-3H, Qldr.sau-3H and Qgp.sau-3H were co-located between markers bpb3433483 and bpb3257096; there were 12 candidate genes in this interval (Table S4). Among these were candidate genes in each interval for the QTL. For example, HORVU2Hr1G080990.1 encodes a seed storage 2S albumin superfamily protein.
A large amount of Seed storage 2S albumin superfamily protein is synthesized at the late stage of seed development [52,53]. The protein content and the degradation rate of stored protein are closely related to seed vigor [54]. HORVU3Hr1G091720.2 encodes a NAC domain-containing protein, and it plays a regulatory role in the growth and development of millet [55]. HORVU3Hr1G091460.10 encodes an SET domain-containing protein that has been reported to play a role in nutrient regulation under abiotic stresses [56]. HORVU3Hr1G061710.1 and HORVU3Hr1G089840.1 encode an L-asparaginase that has been reported to affect NUE and GY in Arabidopsis [57,58]. HORVU3Hr1G063180.1 encodes glutamate dehydrogenase 2, which has been reported to improve NUE traits [23]. HORVU3Hr1G066050.1 and HORVU3Hr1G093310.2 encode the transcription factor basic helix-loop-helix 62 (bHLH62), which has been reported to participate in the transport of NH4 + [59].
N stress in the soil is becoming a huge problem for crop production around the world. It is urgent to develop crop cultivars with strong LN tolerance for coping with the problem. A wild barley allele introgression of S42IL-119 was associated with a 20% increase in TKW under LN conditions [26]. Quan et al. [60] showed that soluble acid invertase, MYB and NAC transcription factors, ethylene biosynthesis and oxygen species play an important role in the LN tolerance of barley. Meanwhile, agronomic and physiological traits are often used as indicators of plants' tolerance to LN stress [61][62][63]. In this study, eight stable QTL were identified that were related to agronomic and physiological traits, which provide clues for developing barley with LN tolerance.
QTL clusters containing stable QTL are particularly important [64]. In this study, two QTL clusters (C3 and C4) contained stable QTL. C3 on chromosome 3H contained four stable QTL (Qgn.sau-3H, Qgy.sau-3H, Qhi.sau-3H and Qal.sau-3H). The most important cluster, C4 on chromosome 3H, included 3 and 7 QTL for seedling and maturity traits, respectively. Interestingly, Qph.sau-3H and Qldr.sau-3H were co-localized between the markers bpb3433483 and bpb3257096, which showed the existence of a close genetic relationship between PH and LDR [65]. The markers in these two QTL clusters should be helpful for marker-assisted selection in barley breeding programs under LN stress. The QTL clusters could provide important information for selecting genotypes with better early vigor and N uptake efficiency.

Plant Materials
An F 8 RIL population, including 121 lines, was constructed using the single-seed descent method from the cross between the barley cultivar Baudin and the wild accession CN4027 [66]. Baudin is a high-yielding variety that can adapt to longer seasons and regions with moderate rainfall; CN4027 is a wild barley genotype.

Hydroponic Trial
A hydroponic trial (T1) was adopted at the seedling stage with two N concentrations provided using Ca(NO 3 ) 2 ·4H 2 O at 2 mmol·L −1 (NN) and 0.5 mmol·L −1 (LN) [22]. Twenty seeds with a similar size from each line in the BC population were selected and disinfected with 10% H 2 O 2 for 5 min, then cleaned with distilled water and germinated at 25 • C and 60% humidity. Three seedlings from each line with consistent growth were transferred to a plastic tank (60 × 40 × 15 cm 3 ) containing Hoagland nutrient solution (Table S1) at the two-leaf stage. The nutrient solution was changed every 5 days, and ventilation was conducted every 2 h. The pH of the nutrient solution was adjusted to 5.5-6.0 with 0.1 mol·L −1 HCl or NaOH. Each treatment was repeated three times. T1 was conducted in the rainproof greenhouse facilities of the Chengdu Campus of Sichuan Agricultural University from September to November in 2019. After transplantation, the samples were collected after being treated with different N concentrations for 30 days. The samples were washed with tap water, moistened with distilled water and then dried with absorbent paper. Epson Expression 10,000 XL was used for collecting images of the roots. Win RHIZO-Pro V2007d (Regent Instruments Inc., Quebec, QC, Canada) was used to analyze the root-related traits. The Ln was calculated for leaves measuring more than 0.2 cm at harvest. PH was measured from the base of the plant to the top. The shoot and root tissues were packed in bags and placed in an oven for deactivation at 105 • C for 30 min and dried at 75 • C until they had a constant weight. The RDW (g/plant) and SDW (g/plant) were measured using a 1/10,000 electronic balance. The dry root-shoot ratio (DRS, RDW/SDW) and TDW (RDW + SDW) were calculated.

Field Trials
The field trials were conducted using a split-plot design with the N treatments (normal N (NN) and no N (LN)) as the main plot and the genotypes as the subplots. The two main plots were separated by a trench 60 cm deep. Pure N at 150 kg·hm −2 , with the N source being urea (N, 46%), was adopted as the normal N treatment. In both trials, pure phosphorus (P 2 O 5 75 kg·hm −2 ), superphosphate (P 2 O 5 , 12%), pure potassium (K 2 O 75 kg hm −2 ) and potassium chloride (K 2 O, 60%) were applied. The trials were conducted at Shifang and Deyang in Sichuan Province during the growing seasons of 2020-2021 (T2) and 2021-2022 (T3). These two experiment fields belong to the mid-latitude of the subtropical humid climate zone. The type of soil in the experimental field is paddy soil, and the basic physical and chemical properties of soil before planting are shown in Table S2. Each line was planted as a single seed in two rows 0.8 m in length, with 6 cm between plants within a row and 20 cm between the rows. Each row was treated as a replicate, and each treatment was repeated three times. The fields were managed by conventional practices for barley production. We collected the aerial parts of five plants with similar growth conditions for each of the three repetitions at the mature stage, then numbered and bagged them. We then rinsed the samples with distilled water and finally dried them with absorbent paper. The straw dry weight per plant and spike weight was measured using an electronic balance after the straw and spike tissues had been deactivated at 105 • C for half an hour and then dried to a constant weight at 75 • C.

QTL Mapping and Statistical Analysis
The genetic linkage map reported in an earlier study [57] was used for mapping the QTL. Spearman's correlation coefficients among traits at seedling and maturity stages were calculated using SPSS 22.0 (SPSS Inc., Chicago, IL, USA). Excel (2016) was used for drawing the charts. Interval mapping in MapQTL5.0 was used for mapping the QTL with LOD > 2.5 at the seedling stage and LOD > 3.0 at maturity according to the standards of previous studies [19,58], and MapChart2.0 software was used for map construction. QTL were named following the recommendations of the International Rules of Genetic Nomenclature (http://wheat.pw.usda.gov/ggpages/wgc/98/Intro.htm (accessed on 15 April 2022)).

Conclusions
In our study, we identified eight stable QTL for agronomic and physiological traits, including GN, TGW, GY, HI, PH, AL, LDR and GP under different N treatments. Thirtyfive QTL formed seven QTL clusters on chromosomes 2H, 3H and 6H. Additionally, we evaluated the relationships among growth-related traits and root-related traits, agronomic traits and yield-related traits. A novel stable LN-specific QTL for TGW was identified on chromosome 2H under the LN treatments. A predicted gene (HORVU2Hr1G080990.1) in the Qtgw.sau-2H interval may be valuable for subsequent fine mapping of candidate genes. These results provide valuable information for understanding N tolerance as well as breeding and utilizing the loci of interest in barley. Funding: This study was supported by the National Natural Science Foundation of China (31971944) and Sichuan Province's Science Foundation for Distinguished Young Scholars (2022JDJQ0006). We thank the anonymous referees for critically reading and revising this manuscript. Data Availability Statement: All data generated or analyzed during this study are included in this published article and its supplementary information files.

Conflicts of Interest:
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.