A population of 103 RILs developed from a cross between highly polymorphic parents Synthetic W7984 (T. turgidum cv. Altar 84/Aegilops tauschii Coss. line WPI 219) and Mexican spring wheat (Opata M85) accessed through the International Triticeae Mapping Initiative (Salem et al., 2007; Sorrells et al., 2011) was used for the genetic mapping study. In addition, F₂ populations of Synthetic W7984 × Chara and Opata M85 × Cascade with different genetic backgrounds to RILs were developed to validate the phenotypic effect of two identified QTL for RD and RM, respectively.

Synthetic W7984, Opata M85, and the 103 RILs were grown in a semi-hydroponic system (Chen et al., 2011b) for 42 days in a randomized block design with four replicates for each genotype. The experimental conditions and trait measurements were the same as described by Halder et al. (2021). All plants were assessed at tiller onset [Zadoks 2.4; (Zadoks et al., 1974)], i.e. 42 days after transplanting. Briefly, the experiment was conducted in a temperature controlled (10–24°C) glasshouse at The University of Western Australia (UWA), Perth, from mid-June to late-August 2019. Wheat seedlings (4–5 cm long roots) grown in washed river sand were transplanted into bins for a semi-hydroponic system containing 35 L nutrient solution.

At harvest, the maximum depth of a plant root (RD) was measured with a ruler from its crown, and the number of nodal roots per plant (NNR) was counted manually. After capturing photographs of the root system using a portable photographing system, the root system were separated from the shoot. Root sections (≤ 20 cm) were scanned at 400 dpi using a desktop scanner (Epson Perfection V800/850) to determine other root traits—RL (sum of all root length types), Rdia, RSA and root diameter length (RDCL) of fine roots (root diameter< 0.25 mm) and coarse roots (root diameter > 0.25 mm)—were measured using WinRHIZO Pro software (v2009, Regent Instruments Inc., Montreal, QC, Canada). Specific root length (SRL) was calculated as the RL per unit of RM, and root length intensity (RLI) was the RL per unit of RD. Root growth rate is the RD per day. RM is the weight of the whole root system after air-forced oven drying (65°C for 72 h). Further, using the phenotypic data, broad-sense heritability (H²) of the root traits was calculated as:

H² = σ G 2 / [ σ G 2 +   σ E 2 n ] , where σ G 2   is genotypic variance (mean sum of squares of a trait) and σ E 2 is environmental variance (residual mean sum of square) from the analysis of variance, and n is replication number per genotype (4) (Wu et al., 2013; Ben Sadok et al., 2015).

Molecular marker data and the linkage map of the Synthetic W7984 × Opata 85 RIL mapping population were accessed from the GrainGenes database (https://wheat.pw.usda.gov/cgi-bin/GG3/report.cgi?class=mapdata&name=Wheat%2C%20Synthetic%20x%20Opata%2C%20BARC). The linkage map comprised 1,476 simple-sequence repeats (SSR) and restriction fragment length polymorphism (RFLP) markers distributed across 21 linkage groups. Among the available markers, 1,018 with known chromosomal locations were used for QTL mapping of the target root traits. The genetic map spanned a length of about 500 cM with an average marker density of 1 cM after filtering the 20% missing values from the dataset.

The composite interval mapping method in Windows QTL Cartographer V2.5_011 was used to identify root traits QTL; the logarithm of odds (LOD) threshold value was set to ≥ 2.5 based on 500 and 1,000 permutations at the 5% significance level. LOD > 2.5 indicate the presence of significant QTL in a particular genomic region for an individual trait. The square of the partial correlation coefficient (R²) estimates the phenotypic variance of a single QTL (Balakrishnan et al., 2020). The sequences of the SSR and RFLP flanking markers (left-and right-hand sides closest to the QTL regions) were identified from GrainGenes database (https://wheat.pw.usda.gov/cgi-bin/GG3/browse.cgi?class=marker; accessed on 05 October 2022) and/or NCBI database (https://www.ncbi.nlm.nih.gov/; 05 October 2022), respectively. Further the sequences were blasted in JBrowse (https://urgi.versailles.inra.fr/blast/?dbgroup=wheat_iwgsc_refseq_v1_chromosomes&program=blastn) with the wheat reference genome RefSeq v1.0 to identify the physical position of the markers. The graphical representation of the QTL was drawn using MapChart 2.32 software.

One-third of an individual seed (excluding the embryo) was used to extract the genomic DNA of Synthetic W7984, Opata M85, Chara, Cascade, and the F₂ of Synthetic W7984 × Chara and Opata M85 × Cascade. The remaining seed with embryo was preserved in the cold room for seed germination to validate the phenotypic effect of the targeted QTL. The one-third seed part was crushed manually using a small hammer, and then crushed further with a SPEX^® SamplePrep 2010 GenoGrinder at 1,400 rpm for 2 minutes for DNA extractions following the cetyl trimethyl ammonium bromide (CTAB) method. The extracted DNA was suspended in 0.1× TE buffer (pH 8.0) for storage. DNA concentrations were measured by NanoDrop (NanoDrop-1000 spectrophotometer) and Qubit 2.0 fluorometer using the Qubit dsDNA BR (Broad-Range) Assay Kit. The primers (forward and reverse) for SSR marker Xbarc50 were obtained from Sigma-Aldrich (Sigma-Aldrich Pty Ltd., NSW, Australia). DNA primers of Xgwm334 (forward (5´) dyed with fluorescent PET) were obtained from Alpha ADN (225 Bridge CP 4023, Montreal, Quebec H3C 0J7, Canada: http://www.alphaadn.com/contact.html).

An EmeraldAmp^®MAX HS PCR Master Mix reaction mixture (15 µL) containing 20 ng template DNA of Synthetic W7984 and Synthetic W7984 × Chara populations, 0.2 µM of each forward and reverse primers was amplified in a thermocycler (Eppendorf Mastercycler EP Gradient S) to validate the Q.rd.uwa.7BL with a flanking marker Xbarc50. The annealing temperature of the marker (53°C) was found in GrainGenes (https://wheat.pw.usda.gov/cgi-bin/GG3/report.cgi?class=marker&name=&id=86860). The PCR conditions were 98°C for 1 min, 35 cycles of denaturation at 98°C for 10 sec, annealing at 53°C for 30 sec, elongation at 72°C for 1 min kb^–1 and final extension (Taq polymerase) at 72°C for 5 min. The PCR products were run on a 2.5% agarose gel electrophoresis using GelRed™ (1:10 ratio) at 120 V for 1 h 20 min. The experiment was conducted at the genetics and molecular genetics laboratories at the UWA School of Agriculture and Environment.

A DNA fragment analysis was undertaken using the Applied Biosystems Genetic Analyzer at Biodiversity Conservation Centre, Kings Park, WA, to validate Q.rm.uwa.6AS with a flanking marker Xgwm334. The annealing temperature of the marker was determined by a gradient PCR using RT-PCR. The master mix (1rxn) for gradient PCR was 5 µL SYBR Green, 1.5 µL of each forward and reverse primer and 2 µL template DNA of Opata M85 and Cascade. Using PCR conditions at 98°C for 2 min, 40 cycles denaturation at 98°C for 10 sec, a range of annealing at 52–67°C for 45 sec, elongation at 72°C for 30 sec kb^–1 and final extension at 72°C for 5 min, the best annealing temperature for the marker was set at 58°C. Singleplex PCR of template DNAs (20 ng) from the Opata M85 × Cascade populations and both parents was done in the wheat genetics laboratory at UWA. The master mix (1×) for a singleplex PCR was 3.52 µL PCR grade water, 2 µL 5× buffer, 0.8 µL MgCl₂ (25mM), 0.08 µL Taq polymerase (0.04 u µL^-1), 0.8 µL of each fluorescent forward primer, and reverse primer, and 2 µL of template DNA (≥ 2 ng µL^-1). Singleplex PCR conditions were 94°C for 5 min, 40 cycles denaturation at 94°C for 30 sec, annealing at 58°C for 1min, elongation at 72°C for 45 sec kb^–1, final extension at 72°C for 7 min and hold at 10°C. A multiplex PCR was done using 1 µL PCR product mixed with 9 µL highly deionized (Hi-Di) formamide with LIZ Size Standard for fragment analysis in an ABI sequencer.

Further, 1 µL PCR product was mixed with 9 µL Hi-Di with LIZ Size Standard for fragment analysis using capillary electrophoresis in an Applied Biosystems 3500 series Genetic Analyzer at Biodiversity Conservation Centre, Kings Park, WA. The DNA fragment size were identified by analyzing the electrogram from SeqPartitioner, Geneious plugin.

Homozygous (AA from Synthetic W7984 or Opata M85, and BB from Cascade or Chara alleles) individuals were identified by comparing differences in band size in the agarose gel and the DNA fragment size of their respective parents. Selected individuals were grown in a semi-hydroponic system in a controlled environment (day/night 24°C, 14°C) as described above. The average RD and RM of the genotypes of two allelic combinations (AA and BB) were compared using a student’s t-test at 0.05% significance level.

Phenotypic data were analyzed using GenStat statistical software 19th edition, with the frequency analysis done in SPSS Version 28.0.0 (142) (https://www.ibm.com/support/pages/node/6525830).

Potential high confidence candidate genes for root traits were identified in the two QTL considered for validation. The physical position of the flanking markers of the QTL was found in the GrainGenes wheat database (https://wheat.pw.usda.gov/cgi-bin/GG3/browse.cgi?class=marker), and blasted in the JBrowse (http://www.wheatgenome.org/Tools-and-Resources/Sequences, accessed on 01 June 2022) wheat genome browser with RefSeq v1.0 to identify the candidate genes on the QTL region.

Further, the gene functions were identified in the International Wheat Genome Sequencing Consortium (IWGSC) RefSeq v1.0 website (https://wheaturgi.versailles.inra.fr/Seq-Repository/Annotation, accessed on 20 June 2022) (Appels et al., 2018). Genes involved in root growth and development from other studies were considered putative candidate genes. The biological functions of the individual genes were obtained from Uniprot (https://www.uniprot.org/?-id+2fYRW1ChXSa+-fun+Pagelibinfo+-info+TREMBL). WheatExp revealed expression of the candidate genes on root tissues in other wheat cultivars (Pearce et al., 2015) (http://www.wheat-expression.com/; accessed on 01 December 2022). Gene expression levels for the candidate genes in different wheat tissues, including roots were downloaded from WheatExp. Further, gene expression in root tissue was filtered, with the highest expression level considered for this study.

Root traits of the Synthetic W7984 × Opata M85 RILs and their parents varied considerably ( Table 1 ). Opata M85 had higher RL, RD, RM, root surface area of fine roots (root diameter < 0.25 mm, cm²; RSA2), and total length of coarse roots (root diameter < 0.25 mm, cm; RDCL2) than Synthetic W7984, while Synthetic W7984 had higher root surface area of fine roots (root diameter > 0.25 mm, cm²; RSA1), SRL, and total length of coarse roots (root diameter > 0.25 mm, cm; RDCL1) than Opata M85 ( Table 1 ). For the RILs, RL, RD, RM, Rdia, RSA1, RSA2, and SRL ranged from 173.10–12,783 cm, 8.00–158 cm, 0.03–0.44 g, 0.21–0.66 mm, 10.07–216.20 cm², 1.16–55.17 cm², and 2,150–74,013 cm g^–1, respectively ( Table 1 ). Transgressive segregation with approximately normal distribution for various root traits (RL, RD, RM, Rdia, RSA1, RSA2, and SRL) between the RILs and the parents was detected ( Figure 1 ). H² was high (84.5–92.1%) for all root traits except RLI (data not shown).

The permutation tests identified 14 and nine QTL for eight root traits, with LOD scores ≥ 2.5 in CIM at 500 and 1,000 permutations, respectively. However, considering that root system architecture is complex and governed by many genes of small effect (Sharma et al., 2011), the study considered the QTL identified at 500 permutations. Most of the QTL were distributed on chromosome groups 5, 6 and 7, except 5B and 7D chromosomes. The QTL for RSA1 were found on chromosome 2A and 3B ( Table 2 ). Opata M85 contributed alleles to all the QTL for RL, RM, RSA1 and NNR and a QTL for RD (Q.rd.uwa.5DL), and Synthetic W7984 contributed all other alleles.

Seven of the QTL on chromosome groups 5, 6 and 7 explained more than 10% phenotypic variance (R²). Three QTL for RD (R² = 11.03–16.52%; LOD 3.14–4.14) were identified and distributed on the long arms of chromosomes 5A, 5D and 7B ( Figure 2 ). Synthetic W7984 contributed alleles to Q.rd.uwa.5AL and Q.rd.uwa.7BL, and Opata M85 contributed alleles to Q.rd.uwa.5DL. Among the 11 QTL, Q.rd.uwa.5DL had the highest LOD (4.14) and R² (16.52) values. For the two QTL for RM (both contributed by Opata M85), one was distributed on the long arm of chromosome 7A (R² = 12.86%; LOD = 2.87) and the other on the short arm of chromosome 6A (R² = 9.49%; LOD = 2.80) ( Figure 2C ); the LOD and R² were same for both 500 and 1,000 permutations. QTL for RL and NNR occurred on the long arm of chromosome 5A. Two QTL for RSA2 were detected on chromosomes 5DL ( Figure 2B ) and 6BS, and QTL for SRL was identified on chromosome 6DS.

Co-location of Q.rd.uwa.5DL and Q.rsa2.5DL were identified at 120.50 cM ( Figure 2B ). The flanking marker interval was 112.2–126.7 cM for Q.rd.uwa.5DL and 112.2–134.1 cM for Q.rsa2.5DL.

Xbarc50, the closely linked marker of Q.rd.uwa.7BL, showed polymorphism between Synthetic W7984 and Chara in agarose gel electrophoresis. Xgwm334, the closely linked marker of Q.rm.uwa.6AS, showed polymorphism between Opata M85 and Cascade in DNA fragment analysis. The fragment size of the parents ( Table 3 ) was used to score the randomly selected F₂ populations of the two validation population lines—19 lines for Synthetic W7984 × Chara, and 13 lines for Opata M85 × Cascade. The Synthetic W7984 × Chara hybrids (F₂) were divided into two groups (only homozygous lines). The group containing the negative allele from Synthetic W7984 (Q.rd.uwa.7BL) had a significantly (P < 0.01) shorter (52%) RD than Chara ( Table 4 ). Similarly, in the Opata M85 × Cascade hybrids, the group containing the positive allele from Opata M85 (Q.rm.uwa.6AS) had a significantly (P < 0.01) higher (31.58%) RM than Cascade ( Table 4 ).

The 329.79–700.63 Mb mapping interval of Q.rd.uwa.7BL contained 2,323 genes, with the functions of 215 genes associated with the wheat root system (Halder et al., 2022) ( Supplementary Table S1 ). Twenty-one genes were putative candidate genes for root traits and abiotic and biotic stress tolerance in wheat, reported earlier in other crops and Arabidopsis ( Table 5 ). The 8.00–22.02 Mb mapping interval of Q.rm.uwa.6AS contained 387 genes, with the functions of 34 genes reported in the wheat root system (Halder et al., 2022) ( Supplementary Table S1 ) and 13 were putative candidate genes for root traits and abiotic and biotic stress tolerance in wheat ( Table 5 ). The in-silico expression study identified, high expression levels of the candidate genes in wheat cultivars ‘Chinese Spring,’ ‘Nulliterea Chinese Spring,’ ‘Azhurnaya,’ and ‘N1DT1A’ ( Supplementary Table S2 ). Among the putative candidate genes from both QTL, TraesCS7B01G374800, had the highest expression level (log2 of transcripts per million: 360.65) in the roots of ‘Chinese Spring’. In the roots of ‘Azhurnaya,’ TraesCS7B01G404000, TraesCS7B01G368400, and TraesCS6A01G026500 had high-expression levels.

The semi-hydroponic system used in this study offered an excellent opportunity to acquire reliable root trait data with high accuracy and repeatability (Chen et al., 2020). In our recent study, root trait variability of 184 bread wheat genotypes originating from 37 countries was characterized in the same semi-hydroponic phenotyping system (Chen et al., 2020), followed by validation of genotypes with contrasting root systems in soil-filled rhizoboxes (Figueroa-Bustos et al., 2018). The consistent ranking of genotypes for some important root traits in the semi-hydroponic system and soil conditions indicates the reliability of the phenotyping study for root studies, as confirmed in other crop species, such as narrow-leafed lupin (Chen et al., 2011a; Chen et al., 2014), barley (Wang et al., 2021), and soybean (Liu et al., 2021; Salim et al., 2022). The wheat lines used in this study will be examined further under the field conditions. Significant phenotypic variation for all measured root traits in the biparental population indicates the successful identification of the polygenic trait ( Table 1 ). Further, the continuous distribution of different root traits such as RL, RD, RM, Rdia, RSA1, RSA2 and SRL ( Figure 1 ) indicates that the genetic architecture of individual trait has many genes responsible for the variation. Similarly, high broad-sense heritability (> 80%) for all root traits except RLI (data not shown) indicates the potential for selecting these traits in future wheat breeding. Earlier studies demonstrated significant phenotypic variation in root traits under drought stress (Ayalew et al., 2017), heat stress (Lu et al., 2022), and waterlogging stress (Yu and Chen, 2013) of the same population suggesting that the population in the current study is suitable for genetic mapping of root traits.

Understanding root trait variation is essential for manipulating the traits according to the soil and environmental conditions to improve stress tolerance and yield in wheat. For example, a large wheat root system (in terms of RL and RM) was beneficial for higher grain yield, capturing water and nutrient from sandy soil under well-watered conditions (Palta and Watt, 2009; Palta et al., 2011). However, large root systems reduced yield at terminal drought due to lower (59%) water use efficiency than shallow root systems (Figueroa-Bustos et al., 2020). RM is another important root trait positively correlated with grain yield under drought stress (Ehdaie et al., 2012). SRL (ratio of RL and RM) is an indicator of utilization in nutrient uptake. Wheat genotypes with large SRL and more fine roots take up nutrient and water from the subsoil and contribute to high yield (Chen et al., 2020). Similarly, Rdia significantly correlates with wheat yield and P acquisition (Atta et al., 2013; Nahar et al., 2022).

Several studies have reported the genetics of RL (Ibrahim et al., 2012; Atkinson et al., 2015; Danakumara et al., 2021; Yang et al., 2021a; Li et al., 2022), RD (Liu et al., 2013; Xu et al., 2013; Ayalew et al., 2017; Ren et al., 2017; Salarpour et al., 2020; Danakumara et al., 2021) and RM (Yu and Chen, 2013; Acuna et al., 2014; Meng-jiao et al., 2020; Danakumara et al., 2021) under different environmental conditions. In our study, we discovered 11 QTL on wheat chromosome groups 5, 6 and 7 responsible for root traits contributed by the two parents, Synthetic W7984 and Opata M85. Both parents could contribute favorable alleles for root traits (Bhoite et al., 2018) ( Table 2 ).

Identification of putative QTL alone is insufficient for trait improvement using MAS. Therefore, validating QTL—testing the allelic effect in populations other than the original population—is essential to eliminate statistical error (Langridge et al., 2001). In this study, we validated two flanking markers for RD and RM QTL in populations different than the original population used for QTL identification.

Mapping QTL can also identify the relationship between the traits through the co-localization of QTL (Colombo et al., 2022) which is important for plant performance improvement. Q.rd.uwa.7BL (Xgwm611–Xbarc20) for RD had a LOD of 4.00 and phenotypic variation of 13.14% ( Table 2 ; Figure 2D ) for both 500 and 1,000 permutations. Q.rd.uwa.7BL were co-located with previously identified QTL including qMRL.CK-7B (Xbarc257.2–Xgwm46) under controlled conditions. Q.rd.uwa.7BL also co-located with qMRL.LP-7B and qMRL-7B1 (Xbarc1181–Xbarc1116) under low P (Ren et al., 2017) and well-watered conditions (Ren et al., 2012), respectively. Two other QTL under well-watered conditions—qLR-7B (Wms400–Wms573) (Ehdaie et al., 2016) and Q.RL-7BL (AX-94528392) (Danakumara et al., 2021) were co-located with Q.rd.uwa.7BL. Two drought-stress specific QTL for RD found from the RILs derived in Synthetic W7984 and Opata M85 (Ayalew et al., 2017) were co-located with Q.rd.uwa.7BL, suggesting that Q.rd.uwa.7BL may contribute to drought stress tolerance. Importantly, Q.rd.uwa.7BL was co-located with grain yield QTL (Xm43p78.14–Xm86p65.0) (Quarrie et al., 2005), kernel number per spike QTL, QKNPS-DH-7B-2.1 (Xbarc276.1–Xwmc396), and thousand-grain weight QTL, QTKW-DH-7B (Xgwm333–Xwmc10) (Zhang et al., 2016) on chromosome 7B. These co-location evidence strongly suggest that Q.rd.uwa.7BL contributes to improving RD and wheat yield. However, for the first time Q.rd.uwa.7BL was successfully validated for RD in other populations with different genetic backgrounds. The QTL may also contribute to biotic stress tolerance: Xgwm344, a closely linked marker of Q.rd.uwa.7BL previously validated for leaf rust resistance in wheat (Zhang et al., 2020). Comparing of Q.rd.uwa.7BL with other previously reported co-located QTL revealed that Q.rd.uwa.7BL is physically located in a larger interval (329.79–700.63 Mb) than the above-mentioned QTL, except qLR-7B ( Figure 3 ).

A number of QTL for RD have been reported on chromosome 7B ( Figure 3 ) (Ren et al., 2012; Liu et al., 2013; Xu et al., 2013; Ehdaie et al., 2016; Ren et al., 2017; Ayalew et al., 2018; Li et al., 2020; Danakumara et al., 2021) under different stress conditions, suggesting that chromosome 7B harbors important genes for RD to improve stress tolerance in wheat. However, none of the QTL has been validated in other populations. In this study, Q.rd.uwa.7BL with a peak marker of Xbarc50 was used to screen the validation population. Synthetic W7984 contributed to Q.rd.uwa.7BL for shallow RD; therefore, Xbarc50 was validated in Synthetic W7984 × Chara hybrids (F₂). A significant reduction in RD in the validation population confirmed the reliability of the marker performance. However, the marker could be further tested in other genotypes for application in wheat breeding. Previous studies have reported the significance of shallow RD in high yield of wheat. Under well-watered conditions, genotypes with shallow RD in durum wheat (Bellario and Jabal2), contributed to high yield. Under irrigated conditions, the shallow-rooted genotypes contributed to 20–40% higher yield than deep-rooted genotypes (El Hassouni et al., 2018). The short root length gene, TaSRL1 (on chromosome 4A: 3.37 Mb) improved thousand grain weight as a pleiotropic effect (Zhuang et al., 2021). Importantly, it was revealed in an earlier study that despite taking up 20% less water under drought conditions, Synthetic W7984 (donor parent of Q.rd.uwa.7BL in our study) had higher grain numbers per spike than Opata M85 (Onyemaobi et al., 2018). Therefore, the validated marker for shallow RD from this study could be used to improve the grain yield and stress tolerance in wheat.

Q.rm.uwa.6AS (Xbcd21–Xcmwg652) identified in this study, was co-located with a previously identified grain yield QTL, QGY.cgb-6A (Xgwm334–WMC297) under both well-watered and water-stressed conditions (Liu et al., 2013) indicating its potential for drought tolerance and grain yield improvement. A meta-QTL for RM, Root_MQTL_67 (Soriano and Alvaro, 2019), overlapped with Q.rm.uwa.6AS (Xgwm334), but was not validated. In this study, Opata M85 contributed positive alleles to Q.rm.uwa.6AS for increased RM. Testing the allelic performance of Xgwm334, in Opata M85 × Cascade hybrids (F₂), Opata M85 had significantly higher RM than Cascade confirming the functionality of the identified QTL. Therefore, using of Q.rm.uwa.6AS in future MAS or other advanced genetic approaches may help improve RM, yield, and stress tolerance in wheat. The QTL could be tested in wider populations for wheat breeding. On the other hand, Q.rm.uwa.7AL for RM identified in this study was co-located with previously identified QTrl.D84-7A (Xbarc275) for RL under both well-watered and drought-stress conditions in a back cross population of Devon × Syn084 (Ibrahim et al., 2012). Another grain yield QTL (Xpsp3094.1–Xm68p78.6) on chromosome 7A was co-located with Q.rm.uwa.7AL (Quarrie et al., 2005). As no similar QTL for root traits on chromosome 6AS and 7AL were reported, Q.rm.uwa.6AS and Q.rm.uwa.7AL were novel discoveries in this study. Validation of Q.rm.uwa.7AL is recommended in the future.

In this study, QTL for RD (Q.rd.uwa.5DL) and QTL for RSA2 (Q.rsa2.5DL) overlapped and shared the same marker, Xmwg900 ( Figure 2B ). These QTL also overlapped with a previously reported QTL for root volume (Ibrahim et al., 2012). Furthermore, the marker interval was found within a recently identified QTL (IWB61072–IWB49479) for grain yield (Li et al., 2018) suggesting the importance of Q.rd.uwa.5DL and Q.rsa2.5DL over other root traits and grain yield. However, there was no overlapping evidence for Q.rl.uwa.5AL, Q.rd.uwa.5AL, Q.rdia.uwa.6AL, Q.rm.uwa.7AL, or Q.srl.uwa.6DS discovered in this study with any previous root trait QTL ( Figure 3 ) which suggesting that they are novel QTL for controlling root traits.

We identified 2,323 genes in the Q.rd.uwa.7BL regions and 387 genes in the Q.rm.uwa.6AS regions. Proteins encoded by 215 genes of Q.rd.uwa.7BL and 34 genes of Q.rm.uwa.6AS were associated with the wheat root traits ( Supplementary Table S1 ). In a recent review article, Halder et al. (2022) listed the number of proteins associated with the wheat root system. However, among the identified genes, proteins encoded by 21 and 13 genes in Q.rd.uwa.7BL (329.79–700.63 Mb) and in Q.rm.uwa.6AS (8.00–22.02 Mb), respectively, had roles in controlling root traits in different crops such as rice, barley, maize and soybean, and Arabidopsis ( Table 5 ).

Phytohormones such as cytokinin play important role in root development (Aloni et al., 2006), with histidine-containing phosphotransfer (HK) and glutaredoxin proteins regulating cytokinin signaling. Transgenic plants with reduced cytokinin had greater root growth and more lateral roots than those plants with high cytokinin (Nishimura et al., 2004). HK proteins regulate phosphorylation to control the root growth of Arabidopsis (Hutchison and Kieber, 2007) and barley (under limited/resupply of P) (Ma et al., 2021). We found TraesCS6A01G020400 encoded HK proteins which located very close (0.23 Mb) to the validated marker, Xgwm334, and TraesCS7B01G404000 encoded glutaredoxin protein which located on the QTL for shallow RD. In Arabidopsis, a genotypes AtGRXS3/4/5/8 with silenced glutaredoxin proteins had large primary roots (Patterson et al., 2016) indicating the negative role of glutaredoxin protein in RD. Glutaredoxin also play important role in stress tolerance through redox state of cell, redox dependent pathway regulation, and also improve nutrient uptake. In rice root, glutaredoxin improved arsenic (Verma et al., 2020) and salinity (Verma et al., 2021) stress tolerance. In Arabidopsis root, glutaredoxin improved nitrogen uptake and ammonium stress tolerance (Patterson et al., 2016).

Another protein, 3-ketoacyl-CoA synthase (KCS) condenses very-long-chain fatty acids essential for cuticular waxes and suberin production in roots (Lee et al., 2009; Kim et al., 2021). Suberin is critical role in drought and salinity stress tolerance in root (de Silva et al., 2021). For example, reduced suberin restricted root growth in Arabidopsis (Lee et al., 2009). A KCS6 barley mutant with reduced cuticular waxes had reduced seminal root length but increased lateral root length (Weidenbach et al., 2015). Therefore, future exploration of the KCS encoding genes (TraesCS6A01G024400 and TraesCS7B01G254900) identified in this study could be useful for improving stress tolerance in wheat through improved RM, and RD improvement. TraesCS6A01G026500, identified in our study, encoded lysine-specific demethylase (LSD). LSD belongs to histone demethylase (amine oxidase superfamily) which contributes to root elongation and abiotic stress tolerance through histone modification (Sun et al., 2019; Liu et al., 2022). In maize, the LSD encoding hub gene [Zm00001d002266: genes with top 10% correlation within a module (Liu et al., 2019)], controlled seminal root length under drought and controlled conditions (Guo et al., 2020). Transgenic Arabidopsis with overexpressed LSD encoding Glyma.17G022500 improved salinity stress tolerance (Sun et al., 2019).

Few candidate genes for RD and RM encode proteins with similar functions indicating their importance for future wheat breeding. Germin-like protein (GLP) was first identified in germinating wheat grains (Bernier and Berna, 2001). However, the role of GLP genes in wheat root trait control is unclear, except for an association between GLP and cell wall modification for improved aluminum stress tolerance in wheat (Delisle et al., 2001; Houde and Diallo, 2008). GLP genes Gs1 and Gs2 are highly expressed in barley roots and expressed salinity stress tolerance (Hurkman et al., 1991). GLP contributed to multiple stress (e.g. as drought, heat, cold, and oxidative stress) tolerance in Arabidopsis and rice (Li et al., 2016). Therefore, it would be interesting to explore the genes encoding GLP for RD and RM ( Table 5 ). We also found six candidate genes encoding mitochondrial transcription termination factor (mTERF) in Q.rm.uwa.6AS, and thaumatin-like protein encoding genes—TraesCS6A01G021000 in Q.rm.uwa.6AS and TraesCS7B01 G446200 and TraesCS7B01G25590 in Q.rd.uwa.7BL. mTERF and thaumatin-like protein expressed in roots of different crops and Arabidopsis and play important role in abiotic and biotic stress tolerance. For example, mTERF encoding genes shot1 and mterf6-5 in an Arabidopsis mutant expressed heat tolerance (Kim et al., 2012) and salt tolerance (Robles et al., 2018), respectively. Thaumatin-like protein highly expressed in roots in barley (Iqbal et al., 2020) and Arabidopsis, and contributed to abiotic (e.g. drought and salt) and biotic (e.g. fungus) stress tolerance in Arabidopsis (Misra et al., 2016; de Jesús-Pires et al., 2020). In-silico studies of the putative candidate genes suggested that the genes expressed at different levels (0.02–360.65) in root tissues of different wheat cultivars. Important genes, TraesCS7B01G404000, TraesCS7B01G254900 and TraesCS7B01G446200, from Q.rd.uwa.7BL showed high (124.32) to low (0.89) gene expression while important genes from Q.rm.uwa.6AS, TraesCS6A01G024400, TraesCS6A01G021000 and TraesCS6A01G020400 showed low (0.20–0.11) gene expression in roots of previously studied wheat cultivars ( Supplementary Table S2 ). However, the study conditions and the studied root traits of the previous studies may cause variation in gene expression, which could be confirmed by future gene expression approaches. Moreover, after further functional validation, the identified putative candidate genes may be useful for wheat breeding programs for root improvement.