Keywords

Introduction

Among all abiotic stresses most affected by climate change, drought is the major one curtailing global wheat production. Although conventional breeding has steadily increased crop productivity under drought conditions and across a broad range of environmental constraints, the present rate of increase in wheat productivity is insufficient to meet food security globally. Genomics-assisted crop improvement provides novel opportunities to enhance the yearly rate of increase in wheat yield while advancing our understanding of the genetic and functional basis of the adaptive response to water-limited conditions (Habash et al. 2009; Fleury et al. 2010; Able and Atienza 2014; Tuberosa et al. 2014). This chapter illustrates how genomics approaches have been applied to genetically dissect durum wheat performance under water-scarce conditions and, more in general, how this information might help to mitigate the negative effects of drought on wheat productivity. In view of the genetic and functional similarity shared by durum and bread wheat, a number of relevant examples for the latter have also been considered.

Dissecting the Genetic Basis of Drought Resistance in Durum Wheat

The genetic basis of grain yield and the morpho-physiological traits that determine durum wheat performance under drought involves a myriad of quantitative trait loci (QTLs) of widely different effects, mostly too small to be detected experimentally. One of the reasons accounting for the modest impact of genomics-assisted breeding on the release of drought-tolerant cultivars is that screening conditions adopted under controlled conditions (e.g. growth chamber) usually provide a rather poor surrogate of the dynamics of the drought episodes that crops are exposed in the field (Passioura 2007; Tuberosa 2012). Additionally, the high context-dependency of QTL effects according to the genetic background and environment (Maphosa et al. 2014) further limits the effectiveness of marker-assisted selection (MAS) for improving field performance under drought conditions.

Until the introduction of association mapping, QTL identification has been pursued via linkage mapping based on the evaluation of biparental populations of recombinant inbred lines (RILs). Notably, the availability of maps obtained with different crosses and sharing common polymorphisms allows for the construction of a consensus map (Maccaferri et al. 2014) that in turn enables an even more accurate comparative analysis (e.g. meta-analysis) of QTL positions.

Association mapping (AM) based on sets of unrelated accessions provides additional opportunities to identify the loci (genes and/or QTLs) for target traits. In durum wheat, the evaluation of a panel of elite accessions characterized by high LD (>1 cM) has allowed for a genome-wide search using a limited number of markers (Maccaferri et al. 2005, 2011). Although AM has mostly targeted traits with a genetic basis less complex than drought tolerance (e.g. resistance to biotic stress), some studies have targeted drought-adaptative traits and grain yield in durum wheat grown under varying water regimes (Sanguineti et al. 2007; Maccaferri et al. 2011; Canè et al. 2014; Graziani et al. 2014).

Notwithstanding the clear advantages of AM as compared to biparental linkage mapping, a major limitation of the former is the high rate of false positives (i.e. Type-I error rate) due to the presence of hidden population structure. Additionally, for highly integrative and functionally complex traits such as yield, particularly under drought conditions, the effectiveness of AM is reduced by the fact that different genotypes may show similar phenotypes due to trait compensation (e.g. yield components), which inevitably undermines the identification of significant marker-trait association. Although the number of studies is clearly too limited to draw more certain conclusions on the validity of AM in identifying QTLs for yield, the results reported by Maccaferri et al. (2008, 2011) in durum wheat suggest that the identification of yield QTLs under different water regimes should also be pursued via biparental mapping. This is particularly true whenever the investigated trait (e.g. yield under drought conditions) is strongly influenced by flowering time, in which case the overwhelming effects on yield of this phenological covariate will overshadow the effects due to the action of yield per se QTLs, hence reducing the possibility of identifying them.

QTLs for Drought-Adaptive Traits

On the discovery side, the past two decades have witnessed remarkable progress in several areas as shown also by the manuscripts in this special issue. The pivotal role of phenotyping in drought-related research is now universally recognized and receiving renewed attention (Tuberosa 2012; Araus and Cairns 2014). This revival in phenotyping has been sparked by the recent availability of new phenotyping technologies and highly automated platforms coupled with a better appreciation of the role of phenotyping in accelerating the response to selection for drought resistance, either through conventional or non-conventional approaches.

Dehydration avoidance and dehydration tolerance are the main mechanisms that contribute to maintain yielding ability under water-limited conditions (Blum 1988). Deep rooting and osmotic adjustment – classified under dehydration avoidance – enable the plant to maintain better hydration while other biochemical and physiological features (e.g. accumulation of molecular protectants, remobilization of stem water-soluble carbohydrates, etc.) classified under dehydration tolerance enable the plant to sustain metabolism even under severely dehydrated conditions. Notably, most genes induced under extreme dehydration have been shown to belong to metabolic pathways with doubtful functional significance under the water-limited field conditions encountered by wheat (Passioura 2007). Conversely, exploitation of naturally occurring variation for yield and/or drought-adaptive traits has allowed for slow albeit unequivocal progress in wheat performance under drought conditions (Reynolds and Tuberosa 2008).

Given the quantitative nature of abiotic stress tolerance, QTLs have been the main target of studies attempting to identify the loci regulating the adaptive response of crops to environmentally constrained conditions. In very few cases, major QTLs affecting yield and other drought-adaptive traits across a broad range of soil moisture conditions have been identified (Quarrie et al. 2005; Maccaferri et al. 2008).

QTLs for Root Architecture and Size

Among the traits that affect the water balance of the plant, roots play a key role in conditions of limited soil moisture (Richards 2008). Roots show a high level of morphological and developmental plasticity, a peculiarity that allows plants to adapt to moisture-limited conditions (de Dorlodot et al. 2007; Den Herder et al. 2010). An example is provided by root aerenchyma and root angle, root features that are receiving increasing attention for their effects on the response to drought and other abiotic stresses (Christopher et al. 2013). Other root features remain much more challenging to investigate, particularly under field conditions, such as in the case of root depth, a trait that has repeatedly shown a key role in crop adaptation to drought conditions when residual moisture at maturity is mainly available in deeper soil layers (Blum 2009, 2011; Watt et al. 2013). In bread wheat, soil coring down to 2 m depth revealed a broad range of genetic variation in deep root traits and showed that root features of high-performing genotypes were superior to those of low-performing genotypes or commercial varieties (Wasson et al. 2014). Since direct measuring of root depth remains an unresolved challenge, large-scale phenotyping for this trait can only be addressed through the use of proxies (e.g. canopy temperature depression) that through aerial remote sensing allow for monitoring the water status of a large number of genotypes in the field (Lopes and Reynolds 2010; Lopes et al. 2014).

In wheat, root metaxylem diameter is another feature that has shown an association with yield under drought conditions (Schoppach et al. 2014). Notably, selection for higher water-use efficiency (WUE) has shown merits in Australian environments where the crop prevalently grows on moisture stored in the soil prior to planting. Under these conditions, a wheat plant using water conservatively is able to complete grain filling with greater amount of water available in the soil. The adoption of this conservative strategy led to the release of two cultivars (‘Drysdale’ and ‘Rees’; Condon et al. 2004) characterized by yield increases of up to 23 % when compared with control cultivars. The final effects of root architecture and size on yield will depend on the distribution of soil moisture and the level of competition for water resources within the plant community.

A most challenging aspect is to define the most desirable root ideotype able to optimize yield according to the prevailing dynamics of soil moisture profile but also accounting for the concurrent presence of gradients in the soil profile for other abiotic factors (e.g. salinity, toxic elements, high pH, etc.) that may impair plant growth. Therefore, each root ideotype should be established based upon the prevailing soil features in the target environment, a good understanding of the root architectural features that limit water uptake, and the metabolic cost required to develop and functionally sustain the root system. Along this line, loci that affect root growth under particular abiotic (e.g. boron toxicity) and biotic (e.g. nematode resistance) constraints are interesting targets for MAS aimed at improving drought resistance through a more vigorous root system of wheat grown in problematic soils.

QTLs for Carbohydrate Accumulation and Relocation

In wheat, the accumulation of carbohydrates and their relocation to the ear are key factors for optimizing yield under adverse environmental conditions (Blum 1998; Reynolds et al. 2009). In bread wheat, QTLs for stem reserve, water-soluble carbohydrates (WSC) remobilization and leaf senescence have been reported across well-watered and water-stressed conditions (Snape et al. 2007; Rebetzke et al. 2008; Bennett et al. 2012; Zhang et al. 2015). Although these studies showed an important role for WSC in assuring stable yield and grain size, Rebetzke et al. (2008) concluded that the small effects of many independent WSC QTLs may limit their direct use for MAS. A combined QTL analysis for yield of several wheat populations evaluated across different environments and seasons enabled Snape et al. (2007) to identify QTLs showing stable and differential expression across irrigated and non-irrigated conditions. Variation for stem water-soluble carbohydrate reserves was associated with the chr. 1RS arm of the 1BL/1RS translocated (from rye to wheat) chromosome, and was positively associated with yield under both irrigated and rainfed conditions, thus contributing to general adaptability (Snape et al. 2007). The beneficial role of this translocation on wheat performance under drought-stressed conditions has already been reported (Ehdaie et al. 2003).

QTLs for Other Traits of Interest for the Control of Water Balance

Measurement of traits such as stomatal conductance, canopy temperature and leaf rolling provides indications of water extraction patterns and the water status of the plant. Therefore, measuring these traits together with soil moisture may help in selecting deep-rooted germplasm in environments where water is available at depth (Blum 1988; Reynolds et al. 2009). Stomatal conductance integrates important environmental and metabolic cues and allows the plant to modulate and optimize its transpiration and WUE (Brennan et al. 2007). A study conducted on a series of successful bread wheat cultivars released from 1962 to 1988 showed a strong and positive correlation between stomatal conductance and grain yield (r = 0.94; Fischer et al. 1998), suggesting that the more modern cultivars extract more water from the soil. These results indicate the possibility of raising the yield potential using stomatal conductance as proxie and suggest the value of identifying the relevant QTLs. Canopy temperature is an integrative trait that reports on the water balance at the leaf and whole-plant level, thus providing a proxie of the capacity of the plant to extract soil moisture (Blum 1988, 2009; Reynolds and Tuberosa 2008). Canopy temperature depression (CTD) is mainly useful in hot and dry environments, with measurements preferably made on recently irrigated crops in cloudless and windless days at high vapour pressure deficits (Blum 1988; Reynolds et al. 2009). Under these circumstances, CTD can be a good predictor of grain yield in bread wheat (r varying from 0.6 to 0.8; Reynolds et al. 2009), where yield progress has been associated with cooler canopies, hence higher transpiration (Fischer et al. 1998). Genetic gains in yield have also been reported in response to direct selection for CTD (Reynolds et al. 2009).

QTLs for Yield Under Different Water Regimes

As global climate change intensifies, the identification of loci with consistent per se effects on yield (i.e. not loci for flowering time) across a broad range of soil moisture regimes becomes increasingly important to raise yield potential (Maccaferri et al. 2008; Pinto et al. 2010; Reynolds et al. 2011; Turner et al. 2014). Major QTLs for grain yield and its components across a broad range of soil moisture regimes have all been reported in bread wheat (Quarrie et al. 2005; Kirigwi et al. 2007; Snape et al. 2007) with only one notable exception in durum wheat where Maccaferri et al. (2008) searched for QTLs for grain yield in RILs evaluated in 16 environments with a broad range in grain yield values (from 0.56 to 5.88 t ha−1), mainly consequent to different soil moisture availability. Two major QTLs on chr. 2BL and 3BS (QYld.idw-2B and QYld.idw-3B, respectively) showed highly significant and consistent effects in eight and seven environments, respectively. In both cases, an extensive overlap was observed between the LOD profiles for grain yield and plant height, but not with those for heading date, thus indicating that the effects of these two QTLs on yield were not due to escape from drought, a well-known factor in determining yield under terminal drought stress conditions that typically characterize Mediterranean environments (Araus et al. 2008). Accordingly, this population was originally chosen because it had shown limited variability in flowering time. For plant height and grain yield, a strong epistasis between QYld.idw-2B and QYld.idw-3B was detected across several environments, with the parental combinations providing the higher performance. These two QTLs evidenced significant additive and epistatic effects also on ear peduncle length and kernel weight (Graziani et al. 2014). As a prerequisie to positional cloning, progeny derived from the cross of isogenic lines have been evaluated for fine mapping of both QTLs (Maccaferri et al. unpublished).

Improving Drought Resistance via Marker-Assisted Selection

Several factors limit the possibility of obtaining reliable QTL data and, most importantly, their deployment in breeding programs through MAS (Tuberosa et al. 2007). Among such factors, the environment dependence of QTL expression is of utmost importance in order to obtain reproducible data and effectively assess the value of a particular QTL. This aspect is particularly relevant for stress tolerance traits since the effect of the same QTL can markedly differ according to the prevailing environmental conditions (Collins et al. 2008). Although many studies have described QTLs that influence tolerance to drought, MAS has so far contributed marginally to the release of drought-resistant cultivars. Improving crop performance under water-limiting conditions via MAS may also require considering QTLs for tolerance to abiotic (e.g. high boron) and biotic (e.g. nematodes) factors that impair root growth and functions. A common feature of cereal responses to drought near flowering and during early stages of seed growth is a reduction of reproductive fertility due to partial sterility and/or early abortion. This loss of fertility has been attributed to different factors acting alone and more likely on reproductive fertility. The QTL approach attempts to dissect out the genetic and physiological components affecting source-sink relationships under abiotic stress and to what extent these may influence yield (Miralles and Slafer 2007). Major QTLs for seed weight and grain yield at different moisture conditions have been identified in durum wheat (Maccaferri et al. 2008) and are being introgressed in different genetic backgrounds. In bread wheat, Fleury et al. (2010) have implemented a strategy where a specific environment is targeted and appropriate germplasm adapted to the chosen environment is selected, based on extensive definition of the morpho-physiological and molecular mechanisms of tolerance of the parents. This information was then used to create structured populations and develop models for QTL analysis, MAS and positional cloning.

Future Perspectives

Increasing attention has been devoted to the use of crop modeling for elucidating the genetic basis of genotype × management × environment (G × M × E) interaction at the level of the entire genotype and, more recently, at the level of single loci (Ludwig and Asseng 2010; Richards et al. 2010; Tardieu and Tuberosa 2010; van Eeuwijk et al. 2010; Parent and Tardieu 2014). The objective is to predict, via modeling, yield differences among genotypes grown under different environmental conditions (Cooper et al. 2009; Tardieu and Tuberosa 2010). The benefits accrued by modeling studies are expected to increase as the complexity of the genetic control of traits increases provided it is possible to account for the effects of genetic interactions for predicting trait variation (Cooper et al. 2009). Ultimately, modeling aims to predict the best combinations of QTL alleles able to optimize yield. The main underlying assumption of the modelling approach is that yield and other functionally complex traits can be analyzed and improved by dissecting it into simpler processes, and then by re-assembling such processes to reconstruct via modelling higher order of plant functionality and ultimately yield itself. Models have been used to generate an index of the climatic environment (e.g. of drought stress) for breeding program trials. In wheat grown in northern Australia, this has shown that mid-season drought generates large genotype by environment interaction (Chapman 2008).

With only a few exceptions as listed above, the vast majority of loci that affect crop yield per se have a rather small effect, particularly under drought conditions. Therefore combining the favorable alleles by MAS to achieve a significant improvement quickly becomes impractical and would excessively constrain the potential for achieving yield gain due to the action of other loci. In this case, MAS for mapped QTLs (Randhawa et al. 2013) can be replaced by genome-wide selection (Bernardo 2010; Storlie and Charmet 2013). Nowadays, genome selection is facilitated by the availability of large numbers of markers, particularly Single Nucleotide Polymorphysms (SNPs; Wang et al. 2014) that are amenable to high-throughput profiling at very low cost.

Conclusions

The release of cultivars better adapted to a broader range of environmental conditions will become an increasingly important goal of breeding projects worldwide. Compared to conventional breeding practices, the contribution in this direction of molecular breeding has somehow fallen short of expectations (Blum 2014; Tuberosa et al. 2014). Nonetheless, genomics approaches and sequence-based breeding will expedite the dissection of the genetic basis of abiotic stress tolerance while providing unprecedented opportunities to tap into wild relatives of wheat. To what extent this will actually impact the release of improved cultivars will largely depend on a more complete and comprehensive understanding of the adaptive response of crops to abiotic stress and our capacity to integrate this information into breeding programs via modeling or other approaches such as genomic selection. In view of the complexity of yield, particularly under drought, we foresee that genomic selection will provide the most effective way to raise the yield potential to the levels required to keep up with the fast-increasing demand in food worldwide. However, MAS will remain a valid option for major loci (genes and/or QTLs) as long as their effects will be sufficiently predictable and economically viable (Tuberosa and Pozniak 2014). Additionally, QTL cloning will become a more routine activity thanks to a more widespread utilization of high-throughput, accurate phenotyping (Tuberosa 2012), sequencing and the identification of suitable candidate genes via ‘omics’ profiling. Ultimately, reducing wheat vulnerability to drought will require a multidisciplinary and integrated approach that will eventually allow breeders to more effectively select drought-resistant cultivars.