Recent Progress in Germplasm Evaluation and Gene Mapping to Enable Breeding of Drought-Tolerant Wheat

There is a need to increase wheat productivity to meet the food demands of the ever-growing human population. However, accelerated development of high yielding varieties is hindered by drought, which is worsening due to climate change. In this context, germplasm diversity is central to the development of drought-tolerant wheat. Extensive collections of these genetic resources are conserved in national and international genebanks. In addition to phenotypic assessments, the use of advanced molecular techniques (e.g., genotype by sequencing) to identify quantitative trait loci (QTLs) for drought tolerance related traits is useful for genome- and marker-assisted selection based approaches. Therefore, to assist wheat breeders at a critical time, we searched the recent peer-reviewed literature (2011-current), first, to identify wheat germplasm observed to be useful genetic sources for drought tolerance, and second, to report QTLs associated with drought tolerance. Though many breeders limit the parents used in breeding programs to a familiar core collection, the results of this review show that larger germplasm collections have been sources of useful genes for drought tolerance in wheat. The review also demonstrates that QTLs for drought tolerance in wheat are associated with diverse physio-morphological traits, at different growth stages. Here, we also briefly discuss the potential of genome engineering/editing to improve drought tolerance in wheat. The use of CRISPR-Cas9 and other gene-editing technologies can be used to fine-tune the expression of genes controlling drought adaptive traits, while high throughput phenotyping (HTP) techniques can potentially accelerate the selection process. These efforts are empowered by wheat researcher consortia.


INTRODUCTION
Bread wheat (Triticum aestivum L.) is one of the world's major cereal crops with global production of 756.7 million tons in 2017 (FAO, 2018). The world's population is expected to exceed 9 billion by 2050, requiring at least a 60% increase in wheat yield (United Nations, 2019). An increase in wheat yield from the current level of 1% per year to at least 1.6% is deemed necessary to address this challenge (GCARD, 2012). This is further challenged by projected unpredictable rainfall patterns associated with climate change which are expected to lead to more drought events (IPCC, 2013). However, the degree of impact on final yield depends on the growth stage and the intensity and duration of stress events (Daryanto et al., 2016;Sarto et al., 2017).
In general, breeding can be accelerated by exploiting the diversity of genetic resources as sources of alleles that enhance desirable traits. More than 800,000 wheat accessions, including local landraces and synthetics, are conserved in genebanks globally (FAO, 1998). A large proportion of these accessions are accessible to wheat breeders. Recent advancements in genomics (e.g., genotype by sequencing, see below) have enabled exploration of this genetic diversity, leading to the discovery of markers and associated quantitative trait loci (QTLs) that can be utilized in marker assisted selection and genomic selection to accelerate variety development (Huang and Han, 2014). Wheat has a large and complex genome (Marcussen et al., 2014;Shi and Ling, 2017;Uauy, 2017;Abrouk et al., 2018). Very recently, however, an annotated reference whole genome sequence for bread wheat was released, which described 107,891 high-confidence level genes (IWGSC, 2018). This release has provided a significant opportunity to use genetic resources for exploring the wheat genome, and selecting alleles that encode desirable physio-morphological traits associated with drought tolerance.
There have been excellent recent reviews on the progress in breeding wheat and other cereals for tolerance to abiotic stress including drought (Mohammadi, 2018;Choudhary et al., 2019;Gupta et al., 2020). However, a unique feature of this current paper is that it reviews the different genetic resources that have been or can be exploited to accelerate the breeding of wheat for drought tolerance. The paper comprehensively updates recent progress (2011-2020) on the discovery of QTLs that promote drought tolerance in wheat, expands the list of associated physiomorphological traits and provides helpful details to breeders (e.g., QTL source populations). The QTL section is now the most up-to-date review on the topic. Furthermore, this paper also presents very recent advances in genetic engineering including the CRISPR/Cas9 genome editing system that has been used to target genes conferring drought tolerance.

UTILIZATION OF GENETIC VARIATION FROM DIVERSE SOURCES
Use of diverse germplasm is key to the development of droughttolerant wheat varieties. As noted in Table 1, T. aestivum landraces are one of the major groups of genetic resources valuable for breeding drought-tolerant wheat (Mwadzingeni et al., 2017). They have complex morphological diversity and are mostly grown in low input environments (Padulosi et al., 2012) which make them more adapted to stress (Padulosi et al., 2012;Lopes et al., 2015a). For example, a group of Creole wheat landraces (the landraces introduced to Mexico from Europe) showed better adaptation to different abiotic stresses, including drought, due to the presence of rare but beneficial alleles (Vikram et al., 2016). Similarly, the Japanese landrace "Aka Komugi" is one of the sources of the dwarfing Rht8c allele (Lopes et al., 2015a;Grover et al., 2018) that contributes to breeding drought tolerant wheat. This is because Rht8c promotes a higher-yielding semi-dwarf phenotype, but does not reduce the length of the coleoptile and thus permits deep sowing (Lopes et al., 2015a). Furthermore, in a study that evaluated 21 genotypes, nine wheat landraces exhibited drought tolerance based on a stress susceptibility index (SSI) (Sareen et al., 2014).
The rye (Secale cereale L.) derived wheat genotypes, also known as the 1B/1R chromosome translocation lines, have been utilized by different wheat breeding programs (Kumar et al., 2003;Hoffmann, 2008;Ren et al., 2017). These 1B/1R chromosome translocation genotypes consist of the long arm of wheat chromosome 1B including its centromere, and the short arm or a portion of rye chromosome 1R (Cai and Liu, 1989;Heslop-Harrison et al., 1990). These genotypes are also important sources of abiotic stress tolerance including drought-tolerance (Kumar et al., 2003;Rajaram, 2005). Hoffmann (2008) demonstrated that the 1B/1R translocation wheat genotypes possess greater drought tolerance. However, this may further depend on the genetic background of the recipient wheat genotype (Monneveux et al., 2003;Singh et al., 1998) as shown by Tahmasebi et al. (2015) where 1B/1R translocation genotypes were not suitable sources of diversity for improving drought tolerance.
The challenges in utilizing these landraces and wild relatives are often associated with epistatic and pleiotropic effects of some genes, leading to linkage drag (Sehgal et al., 2015). However, The wheat genotypes representing major wheat growing zones in India were tested for two seasons under different water regimes using the three marked water stress indices.

Modern cultivars 15
Normal (5 irrigations) and restricted irrigation treatments were applied to assess drought tolerance.
Three genotypes, NI-5439, WH-1021, and HD-2733, were identified as the most droughttolerant cultivars. (Meena et al., 2015) 6 University of Reading, United Kingdom A set of wheat genotypes were assessed for drought tolerance in an ambient glasshouse environment using different drought indices.
Modern cultivars and a local check 6 Three watering regimes (100%, 35%, and 25% capacity) were used to test for drought tolerance.
Hashim-8 was identified as the superior variety for drought tolerance. (Khakwani et al., 2011) 7 Mansoura University, Mansoura, Egypt The wheat genotypes were evaluated for the effect of two levels of osmotic stress at seedling stage.
Modern cultivars (released varieties from 1999 to 2011)

10
The treatments were normal irrigation until 45 days, and irrigation withheld for 21 days in the stress treatment.
The results showed a negative effect of the stress on morphological seedling traits. However, variation was observed among the genotypes. Sids 13, one of the tested genotypes, was the most drought-tolerant, while Shandawel 1 was observed to be the most sensitive genotype.  Landraces 7 Five levels of water stress were applied on seedlings (0%, 6%, 12%, 18%, and 24%) using PEG-6000.
Two landraces, Leweucei and Mateteleki, were found to be more drought-tolerant based on different drought tolerance related parameters such as higher relative water content (RWC), tolerance index (TI) and activities associated with a and b-amylases. (Abido and Zsombik, 2018) 9 Nanjing Agricultural University, PR China One drought-tolerant (Luhan-7) and one drought sensitive (Yangmai-16) variety were evaluated to assess improved tolerance to water stress during post-anthesis growth phase as a result of pre-drought priming at different stages during the vegetative growth phase.
Modern winter wheat cultivars 2 Drought priming was done at tillering and jointing stages with moderate stress (55-60% of field capacity). Severe stress was applied at 35-40% field capacity, seven days after anthesis.
Results showed positive effect of drought priming on both varieties when they were exposed to postanthesis drought. However, the drought-tolerant genotype showed greater response to priming, while the growth stage of priming also contributed to drought tolerance to some extent. (Abid et al., 2016) 10 Cereal Research Non-Profit Ltd., Hungary Snowy River Seeds Pty Ltd., Australia Wheat genotypes were phenotyped for different root and shoot traits under drought in a glasshouse along with one drought-tolerant and one drought susceptible check.

Modern cultivars (released varieties) 29
The well-watered pots were irrigated to 60% soil water capacity, while the drought stress pots were irrigated to 20% soil water capacity.
Based on final grain yield, three varieties were identified as drought-tolerant. (Nagy et al., 2018)   The wheat alien chromosome addition lines derived using Chinese Spring as the common parent, were evaluated using 10 important agronomic traits under irrigated and water-limited conditions.
Landrace derived during the advanced lines 82 The control and drought stress treatments were provided with a total of 210 mm and 120 mm irrigation, respectively, in a two year evaluation using rainout shelters.
The result showed that 26 out of 82 lines possessed high levels of drought tolerance.  15 Seedlings of six modern semi-dwarf and six old tall genotypes were evaluated for drought stress tolerance.

Modern cultivars and landraces 12
The seedlings were subjected to stress by withholding water for six days starting 14 days after planting.
The results indicated that the modern varieties were more drought-tolerant, as the water balance maintenance was better compared to the old varieties. (Petrov et al., 2018)  these unfavorable genetic effects can be minimized through recurrent backcrossing with elite genetic materials and selection. In general, the development of modern varieties has reduced the genetic diversity of bread wheat, and this reduction in diversity shows both spatial and temporal trends (Rauf et al., 2010). Studies have shown that there is extensive reduction of nucleotide diversity in the A and B genome in modern wheat, including elite lines, compared to their progenitors (Haudry et al., 2007;Dreisigacker et al., 2008). Nonetheless, elite germplasm is a convenient source of genetic variation and has been considerably utilized in breeding for drought-tolerant wheat ( Table 1). Most of the breeding programs around the world work on elite materials to develop drought-tolerant cultivars despite the advantages that landraces and wild relatives potentially carry. The major reason behind this is that in elite materials, there is less linkage drag associated with coinheritance of undesirable and defective genes and rare alleles. As a result, elite materials have fewer challenges related to pleiotropic and epistatic gene effects in breeding populations (Sehgal et al., 2015;Mwadzingeni et al., 2017). The advantages offered by modern elite genetic materials compared to landraces and wild relatives make enhancement of genetic gain and selection response more rapid.
As a way to move forward, synthetic hexaploid wheats (SHWs) have been developed to widen the genetic variation in modern wheat. SHWs have deployed ancestral genomes using interspecific hybridization techniques (Reynolds et al., 2007;Khan et al., 2016). The SHW lines (AABBD′D′) have been developed by artificial hybridization of tetraploid durum wheat (T. turgidum, AABB) and diploid wild goat grass (Aegilops tauschiii, D′D′) (Reynolds et al., 2007;Jafarzadeh et al., 2016;Khan et al., 2016;Rosyara et al., 2019). A study involving CIMMYT SHWs has demonstrated the importance of the D′ genome of Ae. tauschii in maintaining genetic diversity as well as improving genetic gain (Rosyara et al., 2019). The diversity in the D′ genome may be contributing to increased stress tolerance in SHW lines (Khan et al., 2016). In one study, six SHW lines were compared to four winter wheat cultivars grown in the U.S. Great Plains under drought stress . One of the SHW lines was shown to be superior under drought conditions for root morphological traits including deep root biomass and length of the longest root. In another study, an evaluation of 33 Ae. tauschii accessions and corresponding SHW lines revealed a wide range of variability for drought tolerance (Sohail et al., 2011). CIMMYT's efforts to develop SHW lines based on 600 genebank Ae. tauschii accessions has been successful, as indicated by the encouraging adoption of SHW cultivars in India and South-West China; of these, at least 30% have been reported to be drought-tolerant (Aberkane et al., 2019). Similarly, Song et al. (2017) found that six out of 34 SHW lines were drought-tolerant, and these SHW lines showed high antioxidant activities (superoxidase and peroxidase activities) that minimize drought-associated oxidative cellular damage, and thus improve drought tolerance (Zhang and Kirkham, 1994;Laxa et al., 2019). In this study, the SHWs also demonstrated superior drought tolerance over the ( Arshad et al., 2016) parental lines (Song et al., 2017). Similarly, Lopes and Reynolds (2011) observed that under drought conditions, four synthetic hexaploid derived lines out-yielded the parental lines by an average of 26%. Combined, these studies suggest that there are opportunities to exploit broad genetic variation in SHWs to improve different agronomic traits including drought tolerance in bread wheat. Furthermore, mutation breeding, including using elite germplasm, can be another avenue to create variants suitable for drought adaptation. Among the cereals, wheat is one of the major crops in which mutation breeding has been employed for cultivar development. For example, "Sharbati Sonora," an early maturing wheat cultivar developed by gamma radiation of a Mexican cultivar, made a major contribution to wheat production in India (Raina et al., 2016). Using gamma ray radiation, 11 drought-tolerant wheat mutant lines were identified in a recent study (Sen et al., 2017). Although mutation breeding is less common compared to other breeding methods, it has potential to generate novel stress tolerance alleles.
The major sources of the above mentioned genetic resources are national and international genebanks that maintain extensive collections of wheat landraces, wild relatives, breeding populations, obsolete varieties, and modern elite varieties. The diversity of wheat germplasm deposited in these genebanks appears to be under-utilized in breeding wheat for drought tolerance, though there are growing initiatives to evaluate and identify accessions that are suitable for wheat breeding programs. These initiatives include: The Wheat Pre-breeding Project (https://www.cwrdiversity.org/partnership/wheat-pre-breedingproject/), The Wheat Improvement Strategic Programme (WISP) Consortium (http://www.wheatisp.org/Consortium/ WISP.php) and Seeds of Discovery (SeeD) (https:// seedsofdiscovery.org/) . Genesys (https:// www.genesys-pgr.org/welcome) is an online portal that hosts information about plant genetic resources deposited in genebanks. Out of almost 320,000 wheat accessions documented in Genesys, 23% are landraces, and 7% are wild relatives. The International Maize and Wheat Improvement Centre (CIMMYT) (http://www.cimmyt.org/seed-request/) hosts the most extensive collection of wheat germplasm (102,375 accessions). USDA-ARS genebanks hold 67,615 wheat accessions: most of them are maintained at the National Small Grains Collection (https://www.ars.usda.gov/pacific-west-area/ aberdeen-id/small-grains-and-potato-germplasm-research/docs/ national-small-grains-collection/) in Aberdeen, ID. Wild species are held at the Wheat Genetic Resource Center (WGRC) (https:// www.k-state.edu/wgrc/), and additional materials at the Germplasm Resources Information Network (GRIN) (https:// www.ars-grin.gov/). The Australian Grains Genebank (http:// www.seedpartnership.org.au/associates/agg) has 42,624 accessions, and the International Center for Agriculture Research in the Dry Areas (ICARDA) (http://www.icarda.org/) has 41,471 accessions. ICARDA and CIMMYT together are managing 1,570 accessions of Ae. tauschii (Aberkane et al., 2019). Other organizations maintaining wheat germplasm are: Navadanya in India (http://www.navdanya.org/), the Svalbard Global Seed Vault in Norway (https://www.croptrust.org/our-work/svalbard-global-seed-vault/) and the National Bureau of Plant Genetic Resources (NBPGR) in India (http://www.nbpgr. ernet.in/). However, a limitation associated with the Svalbard, Navadanya, and NBPGR collections is their restricted access.
Once genetic diversity has been identified for crossing, a double haploid (DH) strategy can be employed. The DH lines possess identical copies of chromosomes in the genome which are complete homozygotes: this allows quick fixation of desired alleles. Wheat DH populations (Fleury et al., 2010;Yan et al., 2017) show potential in improving drought tolerance, as they enhance selection efficiency by improved additive genetic variance for complex quantitative traits (Dashti et al., 2007;Mwadzingeni et al., 2017). The DH lines are considered important for QTL x environment interactions, as trait means are estimated more efficiently, resulting in precise selection due to complete homozygosity (Shamasbi et al., 2017). DH breeding, augmented by advanced molecular tools, has potential to contribute to breeding drought-tolerant wheat. Different studies in wheat have shown the potential of DH lines as sources of stress tolerance. For example, Fatima et al. (2018) identified five drought-tolerant DH lines out of 84, which were tested under control and drought stress conditions. However, DH lines are associated with negative effects from colchicine treatment, and also gametoclonal and somaclonal variation that affect plant performance. Furthermore, a high level of homozygosity may make DH lines inferior to conventionally developed inbred lines (Niemirowicz-Szczytt, 1997). Other major challenges associated with DH in breeding for drought tolerance are financial expense and unique technical expertise involving plant cell and tissue culture techniques.

Identification of QTLs for Drought Tolerance Related Traits
By employing the above genetic diversity, conventional linkage mapping, using biparental populations and double haploid lines, has commonly been used to locate QTLs/genes associated with target traits, including those associated with drought (Sallam et al., 2016;Zhao et al, 2018;Li L. et al., 2019). The objective of these studies is to facilitate marker/genome assisted selection. As the mapping populations used in traditional linkage mapping are derived by hybridization between two parents (most cases), and have limited genetic variation, only low natural allelic diversity can be captured which results in low resolution QTLs (Sallam et al., 2016;Zhao et al., 2018). Furthermore, the large genome size of wheat, which has different epistatic interactions among the QTLs, in addition to a large number of genes influencing a trait, have reduced the identification of useful QTLs (Ashraf, 2010). However, more recently, genome-wide association studies (GWAS) studies have been used to identify marker-trait associations for the trait(s) of interest (Corvin et al., 2014;Scherer and Christensen, 2016). GWAS represents a powerful alternate to linkage mapping. GWAS facilitates the exploration of the genetic variation of complex quantitative traits such as drought, controlled by several genes and their interactions (Kooke et al., 2016). GWAS exploits linkage disequilibrium (LD) resulting from variants at a locus caused by different factors such as historical mutations, natural and artificial selection, and other forces Huang and Han, 2014;Visscher et al., 2017). The method accommodates natural populations with diverse genetic backgrounds (Huang and Han, 2014) and hence takes advantage of the diversity available in the genebanks described above. The efficiency of GWAS depends upon individual factors such as the number of loci for a trait that segregates in the population, genetic architecture, and size of the study population (Visscher et al., 2017). For more precise detection of QTLs/genes, some studies have combined linkage mapping and GWAS (Sallam et al., 2016;Shi W. et al., 2017;Liu et al., 2018;Zhao et al., 2018. GWAS and linkage mapping of drought tolerance QTLs have been enabled by the availability of a high-quality reference genome for common wheat along with next-generation sequencing (NGS) technologies (Kilian and Graner, 2012;Ray and Satya, 2014;Ramirez-Gonzalez et al., 2015;IWGSC, 2018). Genotype by sequencing (GBS), a DNA sequencing method that follows the NGS protocol, does not require prior genome sequence information and has enormous potential to genotype complex genomes such as in wheat (Poland et al., 2012;Mwadzingeni et al., 2016a;Chung et al., 2017).
Using these methodologies, important QTLs for wheat physio-morphological traits associated with drought tolerance have been reported in various studies. Many earlier studies have identified genomic regions associated with different physiological traits such as carbon isotope discrimination (DC) on chromosomes 1BL, 2BS, 3BS, 4AS, 4BS, 5AS, 7AS, and 7BS (Rebetzke et al., 2008); flag leaf senescence on chromosomes 2B and 2D (Verma et al., 2004); coleoptile length on chromosomes 4B and 6A (Rebetzke et al., 2001); seedling vigor on chromosome 6A (Spielmeyer et al., 2007); and canopy temperature on 1B, 2B, 3B, 4A, and 5A (Pinto et al., 2010). Building upon these earlier discoveries, the last few years have shown significant advances. For example, Tura et al. (2020) mapped yield QTLs under drought in a double haploid population on chromosomes 4A, 5B, and 7A. In a comprehensive GWAS study involving a panel of 210 European elite wheat lines, Touzy et al. (2019) discovered 24, 31, and 28 QTLs associated with drought tolerance, respectively, under low, medium and high water stress conditions. Bhatta et al. (2018) identified 90 marker-trait associations (MTAs) related to yield and associated traits under limited water conditions in a GWAS study with GBS markers using 123 synthetic hexaploid wheat lines. Similarly, QTLs for drought-related traits such as the drought susceptibility index (DSI), normalized difference vegetative index (NDVI), and leaf traits (including green leaf area, leaf senescence, and flag leaf phenotypes) were detected on chromosomes 1B, 4A, 6B, 5B, 7A, and 7B (Edae et al., 2014) in a GWAS involving a spring wheat panel. Table 2 summarizes QTLs reported for drought tolerance related traits on different chromosomes of the wheat genome, published in the peer reviewed literature from 2011 onwards.
A number of the QTLs reported in Table 2 that are associated with different physio-morphological traits that contribute to wheat grain yield under drought, appear to be restricted to specific genetic backgrounds and environments (Griffiths et al., 2012;Acuna-Galindo et al., 2015). To determine which QTLs are relevant across genetic backgrounds and environments, some studies have used a meta-analysis approach to identify common QTLs discovered in different studies. For example, in a QTL meta-analysis recently conducted by Soriano and Alvaro (2019), 634 QTLs retrieved from earlier studies were projected on a consensus map having 7352 markers, which resulted in 94 consensus QTL regions, of which 35 were associated with root architecture and response to moisture stress. Similarly, in another QTL meta-analysis, Acuna-Galindo et al. (2015) reported 43 meta-QTLs related to drought and heat stress of which 20 were specific to drought stress tolerance. These results suggest that QTL meta-analysis, along with better estimations of QTL effects, may maximize the benefits of past drought QTL studies (Acuna-Galindo et al., 2015;Gupta et al., 2017;Soriano and Alvaro (2019).
Once candidate genes or loci have been identified, another approach is to use allele-specific markers to identify favorable alleles (haplotypes) that promote drought tolerance in a defined population. For example, Khalid et al. (2019) in a study comprising 213 advanced lines derived from synthetic hexaploid wheat and elite cultivars (CIMMYT and Pakistan), identified favorable alleles at five candidate genes associated with drought adaptation, including those that encode a cell wall invertase gene (TaCwi-A1) that converts sucrose into glucose and fructose, a dehydration responsive element binding protein (Dreb1) and COMT-3B that promotes lignin under water stress. The cell wall invertase gene discovery was consistent with an earlier study involving 348 modern Chinese cultivars which identified a favorable allele (haplotype) at another invertase gene (TaCWI-5D) on chromosome 5D in terms of promoting drought tolerance in wheat .
Combined, these advances in genome sequencing, QTL and favorable allele identification, have provided wheat breeders with numerous targets for introgression and selection of drought tolerance promoting alleles that are potentially stable across environments and genetic backgrounds.

Genome Engineering Techniques
Genome engineering techniques, including gene pyramiding, gene stacking, and transgenics, broaden the methods available to breed wheat for tolerance to abiotic stress including drought (Budak et al., 2015;Mwadzingeni et al., 2016a). Clustered Regulatory Interspaced Short Palindromic Repeats (CRISPR) is a bacterium-derived genome editing system; in this system, a DNA fragment encoding a non-coding RNA sequence is designed to target and then cleave a chromosomal DNA target of interest, within the Cas9 protein complex (Garneau et al., 2010;Budak et al., 2015;Zhang et al., 2018;Kumar et al., 2019;Okada et al., 2019;El-Mounadi et al., 2020;Shinwari et al., 2020). Recently, Wang et al. (2018) used CRISPR/Cas9 to increase grain size in wheat. Kim et al. (2018) also successfully established a   Zhang et al., 2013 (Continued) CRISPR/Cas9 genome editing system in wheat protoplasts to edit the stress-responsive factor genes TaDREB2 and TaERF3. Since some essential regulatory genes that control the biosynthesis of metabolites associated with drought tolerance have been identified (Yang et al., 2010), the CRISPR/Cas9 gene editing technology could potentially be used to target such genes and traits in the future (Singh B. et al., 2018). In addition, CRISPR/Cas9 mediated base editing (Zong et al., 2017;Li C. et al., 2018) and prime editing (Anzalone et al., 2019) techniques could be potentially utilized in breeding for drought tolerant wheat in the future as alternatives to the standard CRISPR/Cas9 genome editing system. Gene editing using CRISPR technologies, which depends on double strand breaks induced by the Cas9 protein (Garneau et al., 2010), has limitations associated with the delivery of donor DNA to the targeted cell types due to the low frequency of homologous recombination in plants ( Molla and Yang, 2019;Hassan et al., 2020). The advantage of base editing technology over the standard CRISPR technology is: it does not require double stranded breaks and can simply perform substitution mutations allowing four kinds of modifications (C/G-to-T/A and A/T-to-G/C) (Komor et al., 2016;Gaudelli et al., 2017;Li C. et al., 2018;Molla and Yang, 2019). However, as base editing does not create indels (Komor et al., 2016;Gaudelli et al., 2017), an alternative prime editing technology which also does not require double stranded DNA cleavage, has been engineered to introduce indels; it also allows all 12 base to base conversions (Anzalone et al., 2019). As the prime editing method overcomes the abovestated limitations and enables different mutations (point mutations and indels) in wheat protoplasts (Lin et al., 2020), it may be a game changer technology in the future in terms of engineering drought tolerant wheat (Hassan et al., 2020;Marzec and Hensel, 2020).
In addition to gene editing, there is evidence suggesting that wheat genetic improvement is also possible through genetic engineering. For instance, transfer into wheat of the AISAP gene encoding a stress-associated protein from the halophyte grass/ Mediterranean saltgrass (Aeluropus littoralis) has been shown to enhance the germination rate, biomass, and grain yield of wheat under osmotic-and salinity stress (Ben-Saad et al., 2012). A recent study (Ayadi et al., 2019) demonstrated that overexpression of the wheat aquaporin gene TdPIP2 in transgenic wheat improved drought tolerance. Similarly, the overexpression of the Escherichia coli CspA gene (modified as SeCspA) in transgenic wheat lowered the rate of water loss and maintained higher chlorophyll and proline under drought compared to nontransgenic plants (Yu et al., 2017). Additionally, improved drought tolerance (enhanced growth, delayed senescence, higher relative water content, higher photosynthetic rate, and higher total chlorophyll content) was observed in transgenic wheat in which an Arabidopsis thaliana cysteine protease (OVERLY TOLERANT TO SALT-1, OTS1) was over-expressed (using a ubiquitin promoter) compared to non-transformed wheat plants (le Roux et al., 2019). It has also been reported that overexpression of a fructan exohydrolase encoding gene (1-FEH w3) promotes higher fructan remobilization under drought which contributes positively to grain yield Hou et al., 2018). Several other studies have shown that overexpression of different drought responsive genes confers greater drought tolerance in transgenic wheat including the genes AtHDG11 , TaCIPK23 (Cui et al., 2018), TaBZR2 (Cui et al., 2019), TaWRKY2 , TaWRKY1, and TaWRKY33 (He et al., 2016), and TaPYL4 (Mega et al., 2019). Interestingly, TaSnRK2.9, which is a sucrose non-fermenting 1-related protein kinase gene cloned from bread wheat, elevated ABA content in tobacco (Nicotiana tabacum) resulting in greater drought tolerance (Feng et al., 2019). Therefore, both gene editing and genetic engineering show promise in terms of achieving drought tolerance in wheat. However, time, effort, skill, financial resources, regulatory issues, and public acceptance are major constraints to wider adoption of these techniques in wheat research programs.

Future Perspectives
The world is currently facing the prospect of food insecurity due to increasing drought. An assessment of genetic gain over fifty years of wheat breeding at CIMMYT showed an increase in grain yield by~18 kg/ha per year under drought (Mondal et al., 2020) but this rate will be insufficient to meet demand. As this review has shown, promising approaches to improve drought tolerance in wheat include exploration of wheat diversity in genebanks, identification of genome-wide QTLs, validation of putative QTLs, and introgression of markers, genomic segments or candidate genes. However, undertaking this pipeline de novo is a major challenge in terms of time, skill, and resources. Therefore, the QTL meta-analysis approach (Acuna-Galindo et al., 2015;Gupta et al., 2017;Soriano and Alvaro, 2019) could be an immediate option that utilizes the results of already-existing QTLs associated with drought tolerance in wheat. Another promising near-term technique is genome editing using CRISPR/Cas9, given the potential it has shown to improve traits such as male sterility, grain weight, protein content, and powdery mildew resistance Wang et al., 2018;Zhang et al., 2018;Okada et al., 2019). Furthermore, the pioneering CRISPR-based technologies including base editing and prime editing may have potential future. Though transgenic wheat offers potential, public acceptance may limit its adoption. By contrast, synthetic hexaploid wheats (SHWs) have shown promise in terms of drought tolerance and simplify the use of existing wild genetic resources (Lopes and Reynolds, 2011;Sohail et al., 2011;Becker et al., 2016;Aberkane et al., 2019;Laxa et al., 2019).
Apart from the use of wider germplasm, the growth environment in which selection and breeding is undertaken is important as well as the phenotyping strategies employed. The tradition of breeding for drought tolerance in wheat has been to compare grain yield in a water deficit environment compared to an optimal environment (Khakwani et al., 2011). However, the complex nature of drought (e.g., drought on sandy soil is more severe than on clay soil) complicates these efforts in terms of their actual impact on yield stability (Hoover et al., 2018). Furthermore, the large and complex nature of the wheat genome makes breeding for drought-tolerant wheat challenging. Therefore, adoption of a comprehensive strategy to develop drought-tolerant wheat varieties is needed. Such a strategy requires precise phenotyping in a water deficit environment (Fleury et al., 2010) as the efficiency of genomics also rests on advances in phenomics (Araus and Cairns, 2014;Fahlgren et al., 2015;Bai et al., 2016;Fernandez et al., 2017;Zhang X. et al., 2017). Many ground-based and aerial phenotyping platforms have been developed recently, with advances made in remote sensing, aeronautics, and computing (White et al., 2012;Araus and Cairns, 2014). In addition, new bioinformatics platforms have shown potential to overcome challenges associated with the management of the high volume of data generated by precision phenotyping tools (Shi et al., 2016;Singh et al., 2016;Coppens et al., 2017;Araus et al., 2018;Yang et al., 2020). These include Minimum Information About a Plant Phenotyping Experiment (MIAPPE), a platform that enables harmonization of phenotyping experiments (Bolger et al., 2019); and Crop Ontology, a platform that promotes proper use of genotypic and phenotypic data through data annotation (Shrestha et al., 2012).
Global agricultural research institutions such as CIMMYT and ICARDA have already initiated precision phenotyping coupled with NGS protocols to improve the utilization of wheat genebank materials and their development into prebreeding materials for stress tolerance breeding programs (Crossa et al., 2016;Singh et al., 2019). However, other national and international institutions holding wheat genetic materials may need to prioritize similar efforts in order to deploy potential germplasm into wheat breeding programs. Traditional efforts may be revived to generate new allelic diversity for drought tolerance traits through mutagenesis involving chemical mutagens, ultraviolet light, and high energy radiation (Chen et al., 2014). Similarly, to discover the genetic basis of complex traits, transcriptomics, proteomics, and metabolomics offer promise (Reddy et al., 2014). For example, the information generated concerning different metabolic pathways have potential to add value in terms of developing targeted strategies that modulate the expression of genes associated with stress tolerance (Yang et al., 2010).
Further enhancement and strengthening of current global research collaborations is another way forward. CIMMYT is using the concept of 12 mega environments (ME) (http:// wheatatlas.org/megaenvironments), defined as broad geographic regions with similar biotic and abiotic stresses, as well as agronomic practices and consumer preferences, to develop and promote wheat cultivars for wider adoption. One immediate activity of this collaboration could be to support the initiation of genomic characterization of genebank materials in developing countries. In the same vein, a very promising global initiative is the "The Heat and Drought Wheat Improvement Consortium (HeDWIC)" (https://www.hedwic.org), which is a worldwide platform of wheat scientists from more than 90 countries working on drought and heat tolerance. Similarly, "The International Wheat Yield Partnership (IWYP)" (https://iwyp.org) is a recent global consortium of partners including public and private research organizations, that aims to improve the genetic potential of wheat grain yield by 50% throught collaborative efforts in two decades. It is hoped that such global scientific cooperation in wheat breeding has the potential to generate new drought-tolerant varieties to combat climate change.

AUTHOR CONTRIBUTIONS
KK and AN conceptualized the manuscript, while KK undertook the literature review, analysis, and wrote the manuscript, and AN and MR edited the manuscript.

FUNDING
The manuscript is a part of PhD thesis (KK) supported by a grant to MR from the Canadian International Food Security Research Fund (CIFSRF), jointly funded by Global Affairs Canada and the International Development Research Centre (IDRC, Ottawa).
The PhD program is also partially supported by grants to AN by the Agricultural Adaptation Council of Canada; Grain Farmers of Ontario, Canada; and SeCan, Canada.