Identification of QTL underlying the leaf length and area of different leaves in barley

Leaf is the main organ of photosynthesis, which significantly impacts crop yield. A high-density linkage map containing 1894 single nucleotide polymorphism (SNP) and 68 simple sequence repeats (SSR) markers was used to identify quantitative trait locus (QTL) for flag leaf length (FLL), second leaf length (SLL), third leaf length (TLL), fourth leaf length (FOLL), flag leaf area (FLA), second leaf area (SLA), third leaf area (TLA) and fourth leaf area (FOLA). In total, 57 QTLs underlying the top four leaf length and area traits were identified and mapped on chromosome 2H, 3H, 4H and 7H. Individual QTL accounted for 5.17% to 37.11% of the phenotypic variation in 2015 and 2016. A major stable QTL qFLL2-2 close to the marker 2HL_25536047 was identified on the long arm of chromosome 2H. The most important QTL clustered region at M_256210_824 - 2HL_23335246 on chromosome 2H was associated with FLL, SLL, FLA and SLA and explained high phenotypic variation. These findings provide genetic basis for improving the leaf morphology of barley. In addition, our results suggested that the top four leaves were significantly positively correlated with plant height and some yield-related traits.

Barley (Hordeum vulgare L.), as one of the earliest domesticated and most important cereal crops for humanity, is widely used for food, feed and malting. Leaf is the main organ of photosynthesis and plays a crucial role in determining yield of crops 1 . In cereals, the upmost three leaves, particularly the flag leaf, assimilate the most light energy and are main source of generation for carbohydrate production [2][3][4] . The top two leaves produce over 80% photosynthesis of the whole plant during grain filling 4 . Moreover, leaf size and shape are regarded as important traits determining photosynthesis capability and grain yield [5][6][7] . Therefore, it is necessary to disclose the genetic mechanism of leaf characteristics.
QTL underlying leaf morphological traits has been identified in some crops [8][9][10][11] , such as QTL controlling flag leaf length in rice 8 , QTL for flag leaf size and morphology in wheat 10 .
To date, the morphological traits of flag leaf have been widely studied in barley and were determined by polygenes with vulnerability to influence from environment [12][13][14][15][16][17][18] . Digel et al. 19 identified PHOTOPERIOD-H1 (Ppd-H1) as a candidate gene controlling leaf size in barley. Xue et al. 17 reported QTLs for FLL and FLW on chromosome 5H and 7H using a comprehensive DArT and SSR genetic map. Liu et al. 18 identified two pleiotropic genomic regions on chromosome 2H and 7H controlling FLL, FLW, FLA and some physiological traits. The limitation of previous studies was the limited number of markers used. High-density map can improve the accuracy of QTL and precise location of important traits 20 .
Our group has constructed a high density of linkage map containing 1962 SNP and SSR markers using a barley DH population 21 , which has previously been used for identifying QTLs underlying yield-related traits 22 , physiological traits of flag leaf including net photosynthesis rate 18 and mapping a new dwarf gene btwd1 23 . The objectives of this study were to use this high density of linkage map to identify QTL associated with the topmost four leaf length and area traits; and to reveal the relationships among the topmost four leaves, plant height and yield-related traits. The detected QTLs and their closely linked markers can be used for marker-assisted selection (MAS) in barley breeding. (2019) 9:4431 | https://doi.org/10.1038/s41598-019-40703-6 www.nature.com/scientificreports www.nature.com/scientificreports/ Methods Experimental materials and field trial. The 122 doubled haploid (DH) lines from a cross between dwarf-barley cultivar Huaai11 and feed-barley cultivar Huadamai6 were obtained 21 . Huaai11 is a six-rowed, dwarf-barley cultivar derived from barley landrace, Daofu Baiqingke. Huadamai6 is a two-rowed, feed-barley cultivar. In addition to the differences in plant height between two parents, they also showed distinct difference in leaf length and area traits. The DH lines and parents were sown in a plot with line length of 1.5 m and line interval of 0.2 m at the experimental farm of Huazhong Agricultural University, Wuhan, China in 2015 and 2016. The field trial followed a completely randomized block design, with 3 replications each year. Each DH and parental line was grown in two rows with eight seedlings in each row.
Trait measurement. At the stages of pre-filling and with fully unfolded flag leaves on the main stem, 6 plants in the middle of rows from each replication were randomly chosen to measure 8 morphological traits including flag leaf length (FLL, cm), second leaf length (SLL, cm), third leaf length (TLL, cm) and fourth leaf length (FOLL, cm), as well as flag leaf area (FLA, cm 2 ), second leaf area (SLA, cm 2 ), third leaf area (TLA, cm 2 ) and fourth leaf area (FOLA, cm 2 ). The leaf length was examined from leaf base to tip. Leaf area was calculated using width × length × 0.75 24 , in which the length and width were measured on the widest and longest part of the blade, respectively. The measurement of plant height (PH) and yield-related traits of all lines including main spike length (MSL), grain number per plant (GP), spikelet number per plant (SLP), grain weight per spike (GWS), grain weight per plant (GWP) and thousand grain weight (TGW) was described by Wang et al. 22 .
Phenotypic data analysis. The mean values of each year were used for statistics, correlation analysis and QTL mapping. Descriptive statistics of all phenotypic data were performed using IBM SPSS Statistics 22 software. Correlation analysis between traits and their significance levels were calculated using the Pearson's correlation coefficients analysis method and two-tailed T test. The broad-sense heritability value was calculated using /rn), where the genotype variance σ g 2 = (MS g − MS ge )/rn, genotype and environment interaction variance σ ge 2 = (MS ge − MS error )/r, error variance σ e 2 = MS error . MS g , MS ge and MS error represent genotype mean square, genotype and environment interaction mean square and error mean square, respectively, r and n represent the number of replicates of each genotype and environments, respectively. The variance of the components was estimated using a general linear model (GLM). QTL analysis. The genetic map containing 1894 SNP and 68 SSR markers 21-23 was used to identify QTL using WinQTLcart v2.5 with composite interval mapping 25 . The window size and walking speed were set to 10 cM, and 1 cM, respectively. After performing 1000 permutations, a LOD (Logarithm of the odds) threshold was obtained 26 . The QTL with LOD score greater than LOD threshold was considered to be an important QTL using the significance level (P = 0.05) as the LOD threshold. QTL was considered as minor or major, depending on the phenotypic variance in a primary genetic analysis, if the phenotypic variance was more than 15% which was considered as major QTL 27 . QTL nomenclature followed the principle of McCouch 28 . The QTL location on the map was drawn using MapChart ver. 2.2 software 29 .

Results
Phenotypic data evaluation. Phenotypes Table 1. The values of all traits in Huadamai 6 were higher than those in Huaai 11, and the two parents showed significant differences in all traits (p < 0.01). The frequency distributions of those traits in two years were consecutive ( Supplementary Fig. S1), and the values of skewness and kurtosis for all traits were among −1 and 1, indicating all traits controlled by multiple genes (Table 1). ANOVA showed that genotypic effects of 122 DH lines and parents were significant difference for all traits (p < 0.01). Year effects were significant (p < 0.05) for all traits except SLL, FOLL and SLA. Genotype × year interactions were also significant (p < 0.05) for all traits except SLA. ( Table 2). The variable coefficients varied from 13.57% to 33.94% in 2015, and 11.12% to 30.89% in 2016. The broad-sense heritability (h B 2 ) of all traits ranged from 88.54% to 91.11%, indicating that genetic factors played an important role in determining these traits.  Table S1). Among the 57 QTLs, 26 QTLs (46%) were detected in the both years.
Third-leaf area. Six QTLs for TLA trait were detected on 2H (2 QTLs), 4H (2 QTLs) and 7H (2 QTLs) with phenotypic variation varying from 6.23% to 31.24% and all positive alleles originated from Huadamai6 (Supplementary Table S1). Only qTLA4-2 was identified in both years and explained 7.05% of phenotypic variation in 2015 and 12.98% in 2016. The remaining QTLs were detected in one year and accounted for 6.23-31.24% of phenotypic variations.
Fourth-leaf area. Six QTLs for FOLA trait were identified on 2H (2 QTLs), 4H (2 QTLs) and 7H (2 QTLs) with phenotypic variation ranging from 7.64% to 35.24%, all carried positive alleles from Huadamai6 (Supplementary Table S1). The QTL qFOLA4-1 was identified in both years and explained 7.64% of phenotypic variation in 2015 and 14.09% in 2016. The rest of QTLs were found in one year, and qFOLA2-3 accounted for the highest phenotypic variation up to 35.24% in 2016.

QTL clusters for leaf length and leaf area traits.
In total, 57 QTLs were detected in this population, five QTL clusters associated with top four leaf length and area traits were found on chromosome 2H (two clusters), 3H (one cluster), 4H (one cluster) and 7H (one cluster) (Fig. 1, Table 4). These QTLs increased top four leaf length and area were contributed by Huadamai6 alleles in these QTL cluster regions. Six QTLs were co-localized in C1 on chromosome 2H underlying TLL, FOLL, SLA, TLA and FOLA. The major QTL qFLL2-2 in C2 was identified in two years and co-localized with qSLL2-2 and qFLA2-2, the positive alleles from Huadamai6 increased the FLL, SLL and FLA. A major QTL for FOLL (qFOLL3-1) in C3 clustered with two stable QTL qSLL3-1 and qTLL3-1, and the favorite alleles from Huadamai6 increased these traits simultaneously. The C4 on chromosome 4H contained stable QTLs (qTLL4-2, qTLA4-2 and qFOLA4-1) controlling TLL, TLA and FOLA. Nine QTLs were co-localized in C5 on chromosome 7H underlying FLA, SLL, SLA, TLL, FOLL, TLA and FOLA. This QTL cluster contained two stable QTLs (qSLL7-1 and qSLA7-2) controlling SLL, SLA.

Discussion
Leaf is the main organ of photosynthesis, and the top two leaves produce over 80% of the net photosynthesis product during grain filling, particularly the flag leaf, which assimilates most of the light energy and converts it to 41-43% of the carbohydrates for kernel filling 3,30 . Thus, leaf morphological traits, such as length, width and area,  www.nature.com/scientificreports www.nature.com/scientificreports/ are considered as important components in determining grain yield potential. Liu et al. 18 have reported the correlation and QTL of physiological and morphological traits of flag leaf in barley. However, QTL analysis related to the uppermost four leaf length and area traits has not been reported.
In our study, we detected 57 QTLs for eight leaf morphological traits: FLL, SLL, TLL, FOLL, FLA, SLA, TLA and FOLA. Six QTLs for FLL in two years were located on chromosome 2H and 7H (Fig. 1, Supplementary  Table S1). A major QTL qFLL2-2 that was stable and not influenced by environment was close to the SNP marker 2HL_25536047 on the chromosome 2H long arm. QTL for flag-leaf length was previously reported on chromosome 2H, 3H, 5H and 7H [14][15][16][17][18] . The QTL qFLL2-2 was different from those QTL reported by Elberse et al. 14 on 2HS, and is likely a new QTL. The QTL qFLL2-4 on 2H and nearby the SNP marker 2HL_13648618 was associated with row number (Vrs1). The QTL qFLL7-2 on 7H is different from the QTL on 7HS reported by Xue et al. 17 .
As expected, three coincident genetic regions were detected between FLA and FLL. For example, qFLA2-2, qFLA2-4 and qFLA7-2 were located at the same regions as qFLL2-2, qFLL2-5 and qFLL7-2. In this study, we detected 10 QTLs associated with FLA on 2H, 3H, 4H and 7H over two years (Fig. 1, Supplementary Table S1). A stable QTL qFLA2-2 was detected on 2H. The QTL qFLA2-2 was close to the SNP marker 2HL_25536047 and SSR marker GBM1218. Li et al. 15 detected Qla2.1 for FLA close to the marker HVHOTR1 on 2H in a BC 3 F 2 population. The genetic-linkage map of Varshney et al. 31 illustrated the marker HVHOTR1 was near to the marker GBM1218, indicating that qFLA2-2 was likely the same locus to the QTL Qla2.1.
QTL for leaf morphological traits from the second to the fourth leaf of barley has not been studied in detail. In our study, 41 QTLs associated with SLL, TLL, FOLL, SLA, TLA and FOLA were detected over two years with phenotypic variation ranging from 5.28% to 35.64%. The six traits were controlled by genetic regions on 2H, 3H, 4H and 7H (Fig. 1, Supplementary Table S1). Wherein, ten QTLs were detected in both years and included three major QTLs. These major QTLs can be used in MAS to improve the leaf morphological traits from the second leaf to the fourth leaf.
In the present study, we detected 57 QTLs associated with top four leaf length and area traits, including five QTL clusters on 2H, 3H, 4H and 7H. Our previously reported QTL for FLL and FLA in C2 on chromosome 2H also influenced those traits in other populations 15,32 . The most important QTL cluster region at the M_256210_824 -2HL_23335246 was close to the marker GMS3 on 2H, associated with QTLs for FLL, SLL, FLA and SLA (Fig. 1, Table 4, Supplementary Table S1). QTLs affecting the chlorophyll, thousand grain weight and plant height were previously located on 2H close to the SSR marker GMS3 inferred from GrainGenes (http:// wheat.pw.usda.gov/GG3/) [33][34][35][36] , which was in the same region detected here. This region also has QTL associated with grain number per spike and net photosynthetic rate, stomatal conductance and chlorophyll content of the flag leaf physiological traits previously detected in the same population 18,22,37 , indicating this QTL cluster region not only affected the morphology of leaf, but also had positive effect on the physiological traits of flag leaf.
The second noticeable clustered QTL region was Bmag13 -3_504106156 on 3H contained qSLL3-1, qTLL3-1 and qFOLL3-1 underlying SLL, TLL and FOLL (Fig. 1, Table 4). Ren et al. 37 detected QTL for heading date on 3H, which was linked with the marker Bmag13 in the same population, and near to this QTL cluster region. The QTLs underlying rachis internode length, plant height, grain yield, number of tillers and days until heading were near to the marker Bmag13 in this clustered QTL region 35,38,39 .
The third QTL cluster was in the region 4_42378533 -4_8073993 on 4H determined SLL, TLL, FOLL, SLA, TLA and FOLA (Fig. 1, Table 4). According to GrainGenes (http://wheat.pw.usda.gov/GG3/), this region affecting ears per m 2 , heading date, thousand kernel weight, plant height and yield were located on 4H, and near to the SSR marker HVM40 in this clustered QTL region 33,35,39 . The fourth QTL cluster in the region 7_542104217 -7HL_3360534 was close to the SSR marker GBM1102 on 7H, which was associated with FLA, SLL, SLA, TLL, FOLL, TLA and FOLA (Fig. 1, Table 4). Search against GrainGenes (http://wheat.pw.usda.gov/GG3/) found that QTLs affecting plant height, heading date and single plant yield were located on 7H adjacent to the marker GBM1102, which was similar to this region detected here 39,40 . Considerable to note was the above-mentioned four QTL cluster regions, each containing QTL-controlling plant height and heading date. Previous studies reported that plant-height and heading-date traits had significant effect on grain yield [41][42][43] . The co-localization of these QTLs likely resulted from closely linked QTL or pleiotropic QTL.
The leaf is the main organ of photosynthesis, in which the top three leaves, especially the flag leaf, absorb most of the plant's light energy and is the main source of carbohydrate production [2][3][4] . In this study, the top four leaf traits were significantly positively correlated with yield-related traits, such as MSL and TGW, with the highest correlation coefficient between flag leaf and TGW. Significant positive correlation was also detected between flag leaf and GWS and GWP (Table 3). Previous studies have also reported the importance of flag leaf traits such as leaf length, leaf angle and leaf area on yield 6,44-46 . We detected co-localized QTL between the top four leaf traits and yield-related traits, such as MSL, SLP, GWP and TGW, on 2H, 4H and 7H 22 . Therefore, QTLs for different leaf traits co-localized with yield-related traits can be effectively utilized in MAS. In addition, significant positive correlations between the top four leaves and PH, and QTL for FLL (qFLL7-2) near the dwarf gene btwd1 on chromosome 7H 23 suggested that leaf size might be affected by plant height.
A major QTL was detected for FLL, SLL, FLA and SLA on 2HL. Blast SNP-tag sequence searched against http:// floresta.eead.csic.es/barleymap using the Barleymap program 47 to anchor the QTL marker (2HL_25536047) in the Barke × Morex POPSEQ population 48 . The QTL annotation by Barleymap program found three potentially relevant functional genes, FHY3/FAR1, gibberellin-regulated gene and SAUR gene. Tang et al. 49 reported FHY3 (FAR-RED ELONGATED HYPOCOTYL3) and FAR1 (FAR-RED IMPAIRED RESPONSE1) were transcriptional factors derived from ancient transposases in evolutionary process. They were involved in chlorophyll biosynthesis via the activation of HEMB1 gene expression in Arabidopsis thaliana, and may have a wide range of functions in plant growth and development. Aubert et al. 50 described that gibberellin-regulated, gene-encoded and gibberellin-regulated family proteins have some role in plant development. Gil and Green 51 found that SAUR gene expressed in growing hypocotyls or other extended tissues had a certain function in regulating cell elongation.
www.nature.com/scientificreports www.nature.com/scientificreports/ The above-mentioned genes might be considered as candidates in determining these traits. Of course, we cannot rule out the role of other genes in this QTL region. At present, the function of these genes on barley is unknown and will be characterized in future studies.
In this research, a total of 57 QTLs associated with top four leaf length and area traits were identified with individual QTL explaining 5.17% and 37.11% of the phenotypic variation. Five clustered QTL regions were detected on chromosome 2H, 3H, 4H and 7H. Ten QTLs were co-localized in the C2 cluster, such as qFLL2-2, qSLL2-2, qFLA2-2 and qSLA2-3. Two major QTLs qFLL2-2, qSLL2-2 and one stable qFLA2-2 explained high phenotypic variation in this QTL cluster. Two stable QTLs qSLL3-1, qTLL3-1 and one major qFOLL3-1 on 3H detected in two years were associated with SLL, TLL and FOLL and co-localized in the C3. The C4 on 4H contained stable QTLs (qTLL4-2, qTLA4-2 and qFOLA4-1) controlling TLL, TLA and FOLA. Five QTLs were co-localized in the C5 on 7H associated with TLL, FOLL, FLA, TLA and FOLA. These QTL clusters could be used as target regions for improving leaf morphology of barley.