Identifying the genetic control of salinity tolerance in the bread wheat landrace Mocho de Espiga Branca

Keywords: chloride, genotyping-by-sequencing, landrace, Na, phenotyping, plant growth, quantitative trait loci, QTL, salt tolerance, sodium, wheat.

Bread wheat is one of the main sources of calories in the human diet for many people (Shewry 2009; Wrigley 2009), and wheat yields of >10 t/ha can be achieved with sufficient water and nutrient supply (Shewry 2009). However, the global average wheat yield is currently 3.5 t/ha due to many factors including biotic and abiotic stresses (Shewry 2009; Food and Agriculture Organization 2019). Soils with an electrical conductivity (EC_e) of 4 dS/m (or 40 mM of NaCl) are generally considered saline, reducing plant growth and yield of cereal crops including bread wheat (Flowers 2004; Colmer et al. 2005; Munns and Tester 2008). Improving the tolerance of bread wheat to salinity therefore would help to minimise the gap between wheat yield potential and actual yield.

The abilities to exclude Na⁺ and to maintain a high cytosolic K⁺:Na⁺ in shoot tissue are considered key salinity tolerance mechanisms (Munns et al. 2006). In bread wheat, the Kna1 locus associated with Na⁺ exclusion and enhanced K⁺:Na⁺ in the shoot was detected on chromosome 4D, with the gene encoding a Na⁺ transporter, TaHKT1;5-D, identified as the most likley candidate underlying the locus (Dvořak et al. 1994; Byrt et al. 2007, 2014). The importance of Na⁺ exclusion in salinity tolerance has been reported in many crops, including durum wheat (Munns et al. 2003, 2012), rice (Ren et al. 2005), tomato (Martinez-Rodriguez et al. 2008) and maize (Fortmeier and Schubert 1995). Although studies have shown that both greater Na⁺ exclusion and K⁺:Na⁺ discrimination correlated with higher biomass and/or grain yield in wheat under salinity (Chhipa and Lal 1995; Ashraf and O’leary 1996; Ashraf and Khanum 1997), this relationship does not always exist across all genotypes (Ashraf and McNeilly 1988; El-Hendawy et al. 2005; Genc et al. 2013, 2019), indicating the importance of other subtraits linked to the salinity tolerance of bread wheat. These subtraits could include tolerance to high leaf Na⁺ (and/or Cl⁻) by compartmentalising toxic ions into vacuoles and, osmotic adjustment by synthesising organic solutes to prevent ion toxicity damaging cellular metabolism in the cytoplasm (Colmer et al. 2005; Munns and Tester 2008; Munns et al. 2016). Retention of Na⁺ and Cl⁻ in older leaves to maintain growth in the younger leaves is also reported as a key trait associated with salinity tolerance (Boursier et al. 1987; Boursier and Läuchli 1989; Colmer et al. 1995, 2005). Studies investigating salinity tolerance in cereals indicate that crops use more than one tolerance mechanism to maintain growth (Rajendran et al. 2009; Tilbrook et al. 2017; Asif et al. 2018; Munns et al. 2020). Many salt tolerant barley lines, particularly landraces, are known to utilise both shoot ion-independent tolerance and Na⁺ exclusion from the shoot (Tilbrook et al. 2017).

A concern with identifying new genetic variation for salinity tolerance subtraits, is a potential lack of variation in current elite germplasm. Due to a focus on yield and quality, elite germplasm may not have the best alleles of the genes required for improving salinity tolerance. Landraces have in the past proven to be a good source of genetic variation to identify novel genes and alleles for salinity tolerance. Two major Na⁺ exclusion loci, Nax1 and Nax2 on chromosome 2A and 5A, were identified in a durum wheat landrace, Line 149 (Munns et al. 2003). Both loci originated from the wheat relative Triticum monococcum L. (Munns et al. 2003; Lindsay et al. 2004; James et al. 2006). Introducing the Nax2 locus (TmHKT1;5-A) into a commercial durum wheat cultivar (Triticum turgidum ssp. durum var. Tamaroi) significantly reduced leaf Na⁺ accumulation and the lines with the Nax2 locus had a 25% improvement in grain yield in the field compared to near isogenic lines without this locus (James et al. 2011, 2012; Munns et al. 2012). Therefore, it is likely that landraces provide a valuable source of genetic variation in salinity tolerance, which could be used to enhance the overall salinity tolerance of current elite cereal cultivars.

Previously, a Portuguese bread wheat landrace, Mocho de Espiga Branca, was found to accumulate up to six-fold greater leaf and sheath Na⁺ concentrations whilst maintaining similar shoot growth as Gladius, an Australian commercial cultivar, under 150 mM NaCl. Borjigin et al. (2020) identified that a naturally occurring SNP in TaHKT1;5-D prevents Mocho de Espiga Branca from retrieving Na⁺ from the root xylem, resulting in a greater flux of Na⁺ to the shoot and higher accumulation of Na⁺ in the leaves. However, despite accumulating considerably higher leaf Na⁺ than Gladius, Mocho de Espiga Branca had the same salt tolerance, suggesting it had other mechanisms of tolerance separate to the HKT pathway. The genetic control of the salinity tolerance (including Na⁺ tissue tolerance) in Mocho de Espiga Branca has not yet been established, particulalry, how this line maintains shoot health despite the high levels of Na⁺. The aim of this study was to detect quantitative trait loci (QTL) associated with salinity tolerance subtraits in an F₂ population of Mocho de Espiga Branca and Gladius to determine which genetic regions may be involved in Na⁺ tissue tolerance.

An F₂ mapping population was derived from a bi-parental cross between a Portuguese bread wheat landrace Mocho de Espiga Branca and an Australian commercial cultivar Gladius (doubled haploid (DH)-derived Rac875/Karichauff//Excalibur/Kukri/3/Rec875/Krichauff/4/-Rac875//Excalibur/Kukri [3794]). A total of 412 F₂ progeny from a single cross along with 20 replicates of Mocho de Espiga Branca and seven replicates of Gladius were phenotyped in a fully automated conveyor system within a temperature controlled Smarthouse at the Australian Plant Phenomics Facility (The Plant Accelerator®, University of Adelaide, Australia; latitude: −34.971366°, longitude: 138.639758°) between June and August 2016. The pot experiment was conducted under natural daylight with a day temperature of 22°C and 15°C at night. Seeds were imbibed at room temperature for 4 h and placed in the dark for 3 days at 4°C prior to sowing. Three uniform seeds were sown in each 2.5 L pot (15 × 20 cm) containing 2.3 kg of substrate 50% (v/v) University of California mix, 35% (v/v) peat mix and 15% (v/v) clay loam. Pots were arranged in a greenhouse according to an augmented row and column design which was used to assign plants to a grid of 20 × 22 positions. At the emergence of the second leaf, seedlings were thinned to a single seedling per pot. At the emergence of the third leaf (16 days after planting (DAP)) plants were loaded onto an individual cart in the Smarthouse, where the pots were weighed daily and automatically watered to maintain a gravimetric water content of 17% (g/g). At emergence of the fourth leaf (23 DAP), 213 mL of 0.26 M NaCl stock solution was applied to the saucer of each pot. Pots were allowed to dry down to 17% (g/g) at which point the final salinity concentration of 150 mM NaCl in the soil solution was reached and automated watering resumed to maintain this concentration.

Daily non-destructive shoot imaging was performed using two 8-megapixel red-green-blue (RGB) cameras, capturing one top view and two side view images with a 90° rotation using a LemnaTec Scanalyzer 3D (LemnaTec GmbH) (Rajendran et al. 2009; Golzarian et al. 2011; Takahashi et al. 2015; Ward et al. 2019). Imaging of plants began 4 days before salt treatment and continued for 12 days after the application of NaCl. The imaging data was prepared using the SET method of Brien et al. (2020) that employs the package growthPheno (Brien 2020b) for the R statistical computing environment (R Core Team 2020). The total shoot pixel area derived from the three RGB images was used to derive the projected shoot area (PSA) in kilo pixels, which has previously been shown to be correlated with shoot biomass (Rajendran et al. 2009; Golzarian et al. 2011; Takahashi et al. 2015). The smoothed PSA (sPSA) values were calculated by fitting a cubic smoothing spline to the PSA for each plant using the R function smooth splines with d.f. of 5. The smoothed absolute growth rate (sAGR) and the smoothed relative growth rate (sRGR) were calculated for each plant using sPSA by subtracting the consecutive smoothed PSA and ln(sPSA) values, respectively, and then dividing by the differences in time (Al-Tamimi et al. 2016). Based on the plots for sPSA, sAGR and sRGR, the sPSA on 19, 22, 24, 28, 31 and 35 days after planting (DAP), the sAGR and sRGR intervals of 1–4 days prior to salt treatment (sAGR_19–22 and sRGR1_19–22), and 1–5 days (sAGR_24–28 and sRGR_24–28), 1–8 days (sAGR_28–31 and sRGR_28–31) and 8–12 days (sAGR_31–35 and sRGR_31–35) after salt treatment were selected to represent the plant growth response in the experiment according to the growth patterns observed. A predicted osmotic tolerance (OST) was determined by dividing the growth rate of the plant immediately after salt application by the growth immediately prior to the salt application (sRGR_24–28/sRGR_19–22). To produce phenotypic means, a mixed-model analysis was performed for each trait using the R package ASReml-R (Butler et al. 2009) and asremlPlus (Brien 2020a) packages for the R statistical computing environment (R Core Team 2020).

The fourth leaf blade, which fully expanded during salt stress, was sampled 12 days after NaCl treatment. Fresh weight was measured and the leaf blade was oven dried at 65°C for 2 days. The dried leaf was weighed and digested in 10 mL of 1% HNO₃ at 85°C for 4 h on a SC154 HotBlock® (Environmental Express Inc.). Na⁺ and K⁺ concentrations were measured using a Flame Photometer (Model 420; Sherwood Scientific Ltd), and Cl⁻ concentration was measured using a Chloride Analyzer (Model 926; Sherwood Scientific Ltd) with the titration solution consisting of combined acid buffer (0.006% nitric acid, 90% water and 10% acetic acid; v/v) and gelatine solution (1.2%; w/v) at 4:1 ratio.

The third leaf blade was sampled for genomic DNA (gDNA) extraction using the phenol/chloroform/iso-amyl alcohol (25:24:1) extraction method (Rogowsky et al. 1991) and the extracted gDNA was quantified using the Quant-iT™ PicoGreen™ dsDNA Assay Kit (Invitrogen) following the manufacturer’s instructions.

Genotyping-by-sequencing (GBS) was used for SNP discovery as described by Poland et al. (2012) with some minor modifications. Briefly, 10 μL of DNA from each plant was normalised to 20 ng/μL and arranged into 96-well PCR plates. After the restriction digest and ligation, each sample was pooled into a single 2 mL tube and purified using the PureLink™ Quick PCR Purification Kit (Invitrogen) following the manufacturer’s instructions. The purified product was multiplexed in eight 25 μL PCR reactions consisting of 10 μL ligated sample, 5 μL Taq® 5× Master Mix (New Energy Biology), 0.65 μL each of 10 μM Illumina forward and reverse primers and 8.7 μL Milli-Q water using the following conditions: an initial denaturation at 95°C for 30 s followed by 16 cycles of amplification at 95°C for 30 s, 62°C for 20 s, 68°C for 1 min 20 s a final extension at 72°C for 5 min and held at 8°C. The final PCR product was pooled, purified as described previously and eluted into 30 μL Milli-Q water to form a final GBS library. The library was quantified using a High Sensitivity D1000 ScreenTape® at the Australian Genome Research Facility (AGRF, SA, Australia). An Illumina High-Seq 2500 machine (Illumina Inc.) at the South Australian Health and Medical Research Institute (SAHMRI, SA, Australia) was used to perform the next generation sequencing (NGS) of the GBS libraries to generate 150 base pair (bp) paired-end sequences.

The NGS data was preprocessed to remove 5′ barcodes and 3′ adapter readthrough. Reads were required to contain the expected restriction enzyme cut site following or preceeding the detected barcode or 3′ adapter respectively. Reads were demultiplexed and assigned to a sample based on the detected barcode. Reads were mapped to the IWGSC RefSeq v1.0 genome assembly (International Wheat Genome Sequencing Consortium (IWGSC) 2018) using BioKanga v4.3.6 in paired-end mode and allowing for 1% mismatches. Approximately 45% of the reads mapped unambigously to the reference genome as pairs. This equated to 109 million reads mapped for Mocho de Espiga Branca, 23.4 million for Gladius and an average of 5.9 million mapped reads for the F₂ lines. From the reads of the two parents, 5633 SNP positions were identified as being homozygous and polymorphic between the two parents with at least seven reads coverage of both alleles. Of these, 178 SNP positions were dropped as the genotyping calls in the F₂ were either: (a) missing in all F₂s (175) or (b) had alleles inconsistent with the parent alleles (3). The remaining 5455 SNPs were used for map construction of population. The markers with more than 90% missing values and/or with high segregation distortion (P ≤ 0.00001) were removed, leaving a total of 2343 SNP markers with an average marker ratio of 26.7:45.5:27.8 for homozygous Mocho de Espiga Branca genotype (AA), heterozygous (AB) and homozygous Gladius genotype (BB) for map construction using the R package ASMap (Taylor and Butler 2017). The markers were distributed across all 21 chromosomes and assigned to 21 linkage groups which spanned a total of 4808.7 cM with an average spacing of 2.1 cM between markers (Supplementary Table S1, available at the journal website). Two additional markers tsl2SALTY-4D and wMAS000033 were added to the genetic map on chromosome 4D and 5A, respectively (Supplementary Fig. S1). The marker tsl2SALTY-4D on chromosome 4D is a cleaved amplified polymorphic sequence (CAPS) marker which was previously designed to distinguish between Gladius and Mocho de Espiga Branca alleles of the Na⁺ exclusion gene TaHKT1;5-D (Fig. S2) (Borjigin et al. 2020). There was no genetic linkage between the two markers on chromosome 4D, this chromosome was established based on the physical position of the two markers for the purpose of QTL analysis. The wMAS000033 marker on chromosome 5A is a known Kompetitive allele specific PCR (KASP™) assay marker for vernalisation gene TaVrn-A1 (Grogan et al. 2016; Garcia et al. 2019). All genotype, marker and phenotype data for the population can be found at https://doi.org/10.25909/14159054.

QTL analysis was conducted for 15 growth traits and four ion concentration traits using WinQTL Cartographer 2.5 (Wang et al. 2012). Single marker analysis (SMA) was performed using a simple linear regression model to identify individual markers that are significantly (P ≤ 0.001) associated to the traits. Composite interval mapping (CIM) was performed using the default settings. The genome-wide significance threshold (or logarithm of odd, LOD) was estimated using 1000 permutations at a significance level of P ≤ 0.05 and a walk speed of 1 cM. The notation of QTL followed the format: Qphenotype.lab-chrmosome with the ‘asl’ indicating ‘Adelaide Salt Lab’, and for the QTL detected for growth traits the subscript ‘x–y’ after phenotype referred to the days after planting.

To determine the genes underlying the QTL detected using CIM and the genetic region containing the significant markers detected using SMA, the IWGSC RefSeq v1.0 full pseudomolecule ID of all the markers within the detected region that were up to two LOD drops from the maximum likelihood of each QTL or all the significant markers detected in SMA, were obtained using the coordinate converter tool in DAWN (Diversity Among Wheat Genome) at http://crobiad.agwine.adelaide.edu.au/dawn/ (Watson-Haigh et al. 2018). The IWGSC RefSeq v1.0 scaffolds containing the markers and the scaffolds bridging the neighbouring markers within the selected region were retrieved from https://urgi.versailles.inra.fr/jbrowseiwgsc/gmod_jbrowse/?data=%2FIWGSC_RefSeq_v1 prior to retrieval of all the annotated IWGSC Chromosome Survey Sequence (CSS) genes on the scaffolds. The predicted functional properties of those IWGSC CSS genes along with their corresponding Munich Information Center for Protein Sequences (MIPS) annotation hit ID and rice annotation hit ID were detected using POTAGE (PopSeq Ordered Triticum aestivum Gene Expression) at http://crobiad.agwine.adelaide.edu.au/potage/ (Suchecki et al. 2017).

After 12 days at 150 mM NaCl, Mocho de Espiga Branca accumulated 18-fold greater fourth leaf Na⁺ concentration than Gladius (Table 1; Fig. 1a; Fig. S3A). Of the 412 F₂ progeny, 298 were skewed towards the Gladius phenotype (Fig. 1a), while 114 had similar Na⁺ levels to Mocho de Espiga Branca. The fourth leaf K⁺ concentration in Mocho de Espiga Branca was approximately 1.6-fold less than Gladius (Table 1; Fig. 1b; Fig. S3B), with 290 of the F₂ lines exhibiting a Gladius-like phenotype (Fig. 1b). The fourth leaf K⁺:Na⁺ of Mocho de Espiga Branca was almost 30 times less than Gladius, with 321 of the F₂ population skewed to the Mocho de Espiga Branca phenotype (Fig. 1c). Mocho de Espiga Branca accumulated 1.2-fold higher Cl⁻ concentration in the fourth leaf compared to Gladius, while the F₂ population was normally distributed for the trait (Table 1; Fig. 1d; Fig. S3C).

Table 1.  The mean ± s.e.m of the parents Mocho de Espiga Branca and Gladius, and range of the F₂ population (min–max) for plant growth and fourth leaf ion concentrations.

Fig. 1.  Frequency distribution of the F₂ population for the salinity tolerance subtraits. (a) Fourth leaf Na⁺ (μmol/g DW). (b) Fourth leaf K⁺ (μmol/g DW). (c) Fourth leaf K⁺:Na⁺ (DW). (d) Fourth leaf Cl⁻ (μmol/g DW). (e) sAGR_31–35 after 12 days in 150 mM NaCl applied at the emergence of fourth leaf. Arrows indicate the trait mean for parents. M, Mocho de Espiga Branca; G, Gladius.

Mocho de Espiga Branca had significantly higher sPSA and sAGR than Gladius throughout the experiment, reflecting its larger biomass accumulation over the time periods measured (Table 1, Figs 1e and 2, Fig. S4A–I). No significant differences were observed between the two cultivars for sRGR or predicted OST (Table 1; Fig. S4J–N). For all growth traits including the predicted OST, the F₂ population was normally distributed (Fig. 1e and Fig. S4A–N).

Fig. 2.  Plant growth response before and after 150 mM NaCl treatment. (a) Smoothed projected shoot area (sPSA) of the parent Mocho de Espiga Branca (in blue) and Gladius (in red). (b) sPSA of the F₂ population (in orange). (c) Smoothed absolute growth rate (sAGR) calculated based on the PSA of the parents Mocho de Espiga Branca (in blue) and Gladius (in red). (d) sAGR calculated based on the PSA of the F₂ population (in orange). (e) Smoothed relative growth rate (sRGR) calculated based on the PSA of the parents Mocho de Espiga Branca (in blue) and Gladius (in red). (f) sRGR calculated based on the PSA of the F₂ population (in orange). The stress was applied 23 DAP shown in black arrow and the intervals selected for the growth analysis are shown in green with the corresponding post salt stress days shown. The thick blue and red lines indicate the mean value of the parents, Mocho de Espiga Branca and Gladius, respectively.

Seven QTL were detected at five unique loci for six salinity tolerance subtraits using CIM (Table 2). Each QTL had a range of phenotypic variation and additive effects. A single QTL for plant growth under salinity was detected on chromosome 1A for the subtrait sAGR_31–35 (QsAGR_31–35.asl-1A) with a LOD score of 4.1 (Table 2). It accounted for 4.4% of the phenotypic variation and the additive effect was −0.3, indicating that the Gladius allele at this position contributed to a greater sAGR 31–35 days after sowing (Fig. 2d). There was a single QTL detected for plant growth performance before salt treatment for sPSA₁₉ (QsPSA₁₉.asl-4A) on chromosome 4A, with a LOD score of 4.3 and 4.8% phenotypic variation, with an additive effect of −0.7 (Table 2).

Table 2.  QTL detected for plant growth and fourth leaf ion concentration in the Mocho de Espiga Branca × Gladius F₂ population using composite interval mapping with 1000 permutation at P ≤ 0.05.

In total, three QTL were detected on chromosome 4D in the location of the TaHKT1;5-D gene. A QTL for fourth leaf Na⁺ concentration (QNa.asl-4D), with a LOD score of 204 accounted for 87.2% of the phenotypic variation with an additive effect of +666.0 being inherited from Mocho de Espiga Branca (Table 2). A QTL for fourth leaf K⁺ concentration (QK.asl-4D) had a LOD score of 77.1 and a total phenotypic variation of 57.3% with an additive effect of −304.0 indicating those with the Gladius allele accumulated more K⁺ (Table 2). The third QTL on chromosome 4D was identified for fourth leaf K⁺:Na⁺ (QK:Na.asl-4D) with a LOD score of 24.7, phenotypic variation of 24.4% and a negative additive effect of −10.9 (Table 2). Two QTL for fourth leaf Cl⁻ concentration were detected, one on chromosome 2B (QCl.asl-2B) and the other on chromosome 5D (QCl.asl-5D) with LOD scores of 5.3 and 6.4, respectively (Table 2). These two QTL accounted for 6.4 and 10.8% of the phenotypic variation with an additive effect of −66.8 and 17.6, respectively (Table 2).

In addition to the QTL detected using CIM, significant markers (P ≤ 0.001) were identified using SMA for five salinity tolerance subtraits and have been summarised in Table S2 and Fig. S5. Five significant markers on chromosome 5A were detected for sAGR_31–35 and were located within the interval of 233.3–238.8 cM (Table S2). Five markers significantly associated with sRGR_19–22 were located within the interval 265.5–270.4 cM on chromosome 1B (Table S2). For the subtrait sRGR_28–31 47 significant markers were identified within the interval of 24.1–73.6 cM on chromosome 4B and for sRGR_31–35, there was a number of significant markers with one marker located at 79.2 cM on chromosome 1D, 13 markers within 39.9–46.2 cM on chromosome 4B and three markers at 297.7 cM on chromosome 5A (Table S2). A total of 29 significant markers associated with fourth leaf K⁺ were detected on the genetic map and were located in the interval 54.3–110.7 cM on chromosome 2A (Table S2).

A list of expressed genes (Supplementary File S1) were identified for the QTL (QsAGR_31–35.asl-1A, QsPSA₁₉.asl-4A, QNa.asl-4D, QK.asl-4D, QK:Na.asl-4D, QCl.asl-2B, QCl.asl-5D) detected using CIM and the genomic regions containing the significant markers detected using SMA (File S2).

For the three co-located ion QTL on chromosome 4D (QNa.asl-4D, QK.asl-4D and QK:Na.asl-4D), a single scaffold44094 containing 64 expressed genes was retrieved (File S1). The gene TaHKT1;5-D, encoding a cation transporter, was shortlisted as candidate based on its role in limiting root-to-shoot Na⁺ transport to enable the plant to maintain low concentrations of Na⁺ in shoot (Table S3) (Byrt et al. 2007, 2014). A single SNP between Mocho de Espiga Branca and Gladius has previously been shown to be responisble for high shoot Na⁺ accumulation (Borjigin et al. 2020).

For traits linked to fourth leaf Cl⁻ accumulation, 18 scaffolds with 288 expressed genes, and 61 scaffolds containing 1491 expressed genes, for the QTL QCl.asl-2B and QCl.asl-5D were identified, respectively (File S1). Within the QTL QCl.asl-2B region, potassium transporter 2, NHX2, CIPK3 and CIPK29 were identified as potential candidates, based on their previously described roles in plant salinity tolerance (Table S3) (Gierth and Mäser 2007; Luan 2009; Wei et al. 2011). Within the QCl.asl-5D loci candidate genes included Na⁺/H⁺ antiporter 2 (NHD2), NHX5, CHXs and CDPKs, based on their potential roles in maintaining ion homeostasis and stress tolerance (Table S3) (Luan et al. 2002; Bassil et al. 2011; Shi et al. 2018).

Marker assisted selection also identified regions of the wheat chromosomes significantly linked to salt tolerance traits. Regions linked to the growth traits sAGR_31–35, on chromosome 5A (two scaffolds), sRGR_28–31 on chromosome 4B (131 scaffolds) sRGR_31–35 on chromosome 1D (one scaffold), sRGR_31–35 on chromosome 4B (73 scaffolds), sRGR_31–35 on chromosome 5A (one scaffold) had 164, 2877, 98, 1331 and 89 expressed genes, respectively (File S2). Candidate genes for sAGR_31–35 on chromosome 5A include (but not limited too) those encoding a two pore potassium channel a (TPK), inhibitor of growth protein 4, gibberellin-regulated proteins and several protein kinases (Table S4). The sRGR_28–31 region on chromosome 4B included CBL3, CIPK3 and CIPK9, based on their role in regulating ion homeostasis and Ca⁺ signalling pathways under stress, a number of transcription factors which regulate growth, and a number of genes encoding proteins involved in ion transport (Table S4). A candidate gene within the sRGR_31–35 region on chromosme 1D was CIPK18 (Table S4) while within the sRGR_31–35 on chromosome 4B a number of candidates were shortlisted including genes CBL18, CDPKs, CHXs, potassium transporter two, aquaporin-like superfamily protein and Abscisic acid-insensitve five transcription factor, due to their roles in regulating ion homeostasis, water transport in plants and regulation of plant growth (Table S4). Finally, the sRGR_31–35 on chromosome 5A had one obvious candidate gene, a voltage-dependent anion-selective channel (VDAC), due to its physiological functions associated with plant developmental processes and abiotic stress responses (Table S4) (Desai et al. 2006; Takahashi and Tateda 2013).

A total of 65 scaffolds with 2021 expressed genes were present within the significant marker interval for the fourth leaf K⁺ concentration on chromosome 2A (File S2). Within this region, genes shortlisted as potential candidates include the potassium channels (AKT1 and SKOR), CIPK23 and chloride channel C (CLC-c) (Table S4) based on their role in maintaining cation homeostasis and plant salinity tolerance (Table S4) (Gaxiola et al. 1998; Hirsch et al. 1998; Cheong et al. 2007; Shabala and Cuin 2008; Ragel et al. 2015; Wang et al. 2016).

In this study, 412 Mocho de Espiga × Gladius F₂ progeny were phenotyped using both non-destructive and destructive tools for 19 salinity tolerance subtraits. The F₂ population was genotyped using GBS to construct a high density genetic linkage map and to perform QTL analysis. Genomic regions significantly associated with salinity tolerance were detected on chromosomes 1A, 1D, 4B and 5A for the subtraits of plant growth (PSA, AGR and RGR), and on chromosome 2A, 2B, 4D and 5D for leaf ion (Na⁺, K⁺ and Cl⁻) accumulation.

QTL for Na⁺ and K⁺ accumulation (QNa.asl-4D, QK.asl-4D and QK:Na.asl-4D) were identified on chromosome 4D. The QTL for leaf Na⁺ accumulation, QNa.asl-4D, was particularly strong explaining up to 87.2% of the phenotypic variation, with those plants with the Mocho de Espiga Branca allele accumulating significantly higher leaf Na⁺ than those with the Gladius allele. Within the QTL interval is the TaHKT1;5-D gene, known to be responsible for low shoot Na⁺ accumulation and enhanced leaf K⁺:Na⁺ discrimination in bread wheat (Dubcovsky et al. 1996; Byrt et al. 2007, 2014). Recently Borjigin et al. (2020), demonstrated that one SNP in the coding region of the Mocho de Espiga Branca TaHKT1;5-D gene resulted in a non functional protein leading to significantly higher Na⁺ in the transpiration stream and consequently in the leaves. The QTL identified in this study suggest that this single SNP is responsible for the majority of Mocho de Espiga Branca’s high Na⁺ leaf phenotype. However, as observed in Borjigin et al. (2020) the high leaf Na⁺ concentration in Mocho de Espiga Branca was not linked to a decrease in salinity tolerance suggesting that the landrace uses other salinity tolerance mechanisms to survive. It was hoped that a study of other phenotypic traits between Gladius and Mocho de Espiga Branca would identify salinity tolerance mechanisms independent of leaf Na⁺ accumulation.

As previously observed, Mocho de Espiga has greater biomass than Gladius both before and after salt application (Table 1 and Fig. 2a). Despite accumulating over 10 times as much fourth leaf Na⁺ (Fig. 1a), one third less K⁺ (Fig. 1b) and 1.5 times as much Cl⁻ (Fig. 1d) than Gladius, Mocho de Espiga Branca’s relative growth rate (sRGR) is compariable to that of Gladius, indicating (for at least the time points measured) high accumulation of Na⁺ and Cl⁻ are not affecting its ability to grow under salt. This suggests that either the ability to exclude large amounts of leaf Na⁺ is not linked to enhanced plant tolerance, or that Mocho de Espiga Branca has a unique set of tissue tolerance subtraits that can be introduced into modern bread wheat cultivars. Despite there being no differences in the sRGR between Mocho de Espiga Branca and Gladius, genetic loci linked to growth parameters were identified in the F₂ population. A QTL for the subtrait sAGR_31–35 (QsAGR_31–35.asl-1A) for plant growth under salinity was detected on chromosome 1A. At this growth stage, the plants had been exposed to salinity for 8–12 days and 114 F₂ lines had accumulated more than 400 μmol/g DW Na⁺ in the leaf at 35 DAP, suggesting that this QTL is associated with shoot ion-dependent tolerance. A QTL for shoot biomass has previously been detected in the similar region on chromosome 1A in a bread wheat recombinant inbred lines (RILs) mapping population (Masoudi et al. 2015). This suggests the gene(s) residing in this region may be important to maintaining plant growth under salinity (Masoudi et al. 2015). Among 792 expressed genes within the region of QsAGR_31–35.asl-1A, there are two predicted AKT genes (AKT1 and AKT2) that encode K⁺ channels. AKT1 is known to regulate root K⁺ uptake under low K⁺ conditions in Arabidopsis, while AKT2 mediates K⁺ transport in the phloem (Hirsch et al. 1998; Lacombe et al. 2000; Gajdanowicz et al. 2011). K⁺ is one of the most important inorganic cations in plants and involved in many metabolic processes such as protein synthesis and signalling mechanisms (Shabala and Cuin 2008; Shabala 2017). Other genes encode for a series of proteins involved in calcium signalling, known to be important for signalling the plant is under salinity stress between tissues and within cells (Luan 2009; Kudla et al. 2010), ABA responsive transcription factors and bZIP transcription factors which control a plant’s response to environmental stress (Choi et al. 2000; Sornaraj et al. 2016; de Melo et al. 2020; Li et al. 2020).

This study also detected loci on chromosome 5A for sAGR_31–35 and sRGR_31–35 and chromosomes 1D and 4B for sRGR_31–35 (Table S2). QTL associated with plant growth under salinity have previously been reported on chromosome 1D, 4B and 5A for seedling vigour (Oyiga et al. 2018), leaf growth rate (Amin and Diab 2013) and shoot biomass (Ma et al. 2007; Genc et al. 2010; Amin and Diab 2013; Ghaedrahmati et al. 2014; Masoudi et al. 2015; Oyiga et al. 2018). Within the physical interval on 4B in this study, many Ca⁺ signalling genes (CBLs, CDPKs and CIPKs) were identified and Ca⁺ signalling has previously been reported to play an important role in plant salinity tolerance (Batistič and Kudla 2009; Luan 2009; Boudsocq and Sheen 2013).

Salinity tolerance is a complex trait and plants use multiple tolerance mechanisms to maintain growth across different developmental stages (Ward et al. 2019). The genomic region detected on chromosome 1B (for sRGR_19–22) and 4A (for sPSA₁₉) for shoot growth prior to the salt stress was not detected for the shoot growth after salinity stress was applied. Although the same region on chromosome 4B was associated with sRGR at both 28–31 and 31–35 DAP, it appears that genes on chromosome 1D and 5A associated with shoot growth have been activated at 31–35 DAP. In rice, QTL were identified for transpiration use efficiency (TUE) during the early response phase following salinity treatment were no longer significant after 6 days of salinity treatment (Al-Tamimi et al. 2016). Similarly, Ward et al. (2019) found highly time-dependant salinity response QTL in barley. This, in addition to the findings of the current study, suggests that the genetic response of plants to salinity treatments varies across different stages of salinity stress.

Two QTL for leaf Cl⁻ concentration were detected on chromosome 2B (QCl.asl-2B) and 5D (QCl.asl-5D). Unlike the Na⁺ and K⁺ frequency distributions which were biomodel (most likely caused by the SNP in the TaHKT1;5-D gene) the Cl⁻ distribution was normally distributed, suggesting multiple genes control this trait. In both this study and Borjigin et al. (2020), Mocho de Espiga Branca had significantly higher leaf Cl⁻, in part due to high translocation of Cl⁻ from the root to the shoot in the xylem. While Cl⁻ is a plant nutrient and important for regulating turgor and leaf osmotic potential (Raven 2017; Wege et al. 2017; Geilfus 2018; Wu and Li 2019), at high concentrations in leaves Cl⁻ can also be toxic (Tavakkoli et al. 2011, 2012; Geilfus 2018, 2019; Wu and Li 2019). Similar to the stratergies to avoid Na⁺ toxicity, strategies to minimise Cl⁻ toxicity are hypothesised to involve Cl⁻ exclusion from the leaves or sequestration of Cl⁻ in the vacuole (Geilfus 2018, 2019; Wu and Li 2019). Although genes that encode Cl⁻ channels or transporters have been identified (see Colmenero-Flores et al. 2007; Barbier-Brygoo et al. 2011; De Angeli et al. 2013 Henderson et al. 2014; Herdean et al. 2016; Li et al. 2016, 2017; Geilfus 2018, 2019; Wu and Li 2019 for examples), the genes/proteins specifically involved in Cl⁻ salinity tolerance in cereals remain unknown. To date, few studies have reported QTL for plant Cl⁻ accumulation in bread wheat under salinity (Asif et al. 2019).

Two QTL have been identified in this study, suggesting a genetic difference in the mechanisms relating to Cl⁻ movement between Mocho de Espiga Branca and Gladius. Previously, a major QTL was detected for leaf Cl⁻ accumulation on chromosome 5A, while 13 minor QTL were detected on chromosomes 1D, 2A, 2B, 3A, 3B, 6D, 7A and 7D using a doubled haploid (DH) mapping population in hydroponics or field trials under salinity stress (Genc et al. 2014). The QTL detected previously for leaf Cl⁻ accumulation and the QCl.asl-2B detected in this study were both located on the long arm of chromosome 2B, which may indicate the importance of the genes within this region of the genome in the regulation of plant Cl⁻ transport in saline conditions. Within the physical interval underlying QCl.asl-5D and QCl.asl-2B in this study, genes encoding proteins potentially involved in ion transport and homeostasis in plants were detected including NHX5 and CHXs (Table S3). CHXs have been demonstrated to have potential roles in maintaining organelle pH and K⁺ homeostasis and cellular stress responses in salinity (Yokoi et al. 2002; Bassil et al. 2011; Chanroj et al. 2012). There were also genes such as CBLs and CIPKs that encode proteins involved in signalling and regulation of cell response to stress (Table S3). Previous reviews have reported a number of CBLs and CIPKs with their potential roles in tolerance to low-K⁺, positive regulation of ABA signalling and plant abiotic stress tolerance (Batistič and Kudla 2009; Dodd et al. 2010; Kudla et al. 2010; Thoday-Kennedy et al. 2015; Steinhorst and Kudla 2019).

In conclusion, a number of QTL for salinity tolerance subtraits were detected using an F₂ mapping population of Mocho de Esgiga Branca × Gladius under salt stress in a controlled environment experiment. Using the bread wheat genome, candidate genes that encode proteins associated with plant salinity tolerance subtraits were identified. These include genes encoding K⁺ channels, Na⁺/H⁺ antiporters, H⁺-ATPase, CBLs, CIPKs and CDPKs. This study provides a new insight into the genetic control of salinity tolerance in Mocho de Espiga Branca to assist with the future improvement of salinity tolerance in bread wheat cultivars. Future work should focus on developing advanced mapping populations derived from Mocho de Espiga Branca and Gladius to reduce the interval of the detected QTL regions in order to identify tightly linked candidate genes responsible for Na⁺ tissue tolerance in Mocho de Espiga Branca.

Supplementary material is available online.

The marker, genotype and phenotype data that support this study are available in Figshare at https://doi.org/10.25909/14159054.

The authors declare that they have no conflicts of interest.

This project was funded by the Grains Research and Development Corporation (GRDC): Projects UA00145, UA00151, ANU00027. The Australian Centre for Plant Functional Genomics (ACPFG) was jointly funded by the Australian Research Council (ARC) and GRDC. The Australian Plant Phenomics Facility, The Plant Accelerator® provided infrastructure and technical support and is funded under the National Collaborative Research Infrastructure Strategy (NCRIS). CBo thanks the China Scholarship Council and the University of Adelaide Joint Postgraduate Scholarships Program for her PhD stipend, and acknowledges the financial support from the Research Travel Scholarship and the Global Learning Travel Grant at the University of Adelaide, and the Plant Nutrition Trust to attend conferences.