Keywords

1 Learning Objectives

  • Drought and heat stress are common constraints across most wheat growing regions.

  • New phenotyping and genetic technologies and knowledge can be efficiently integrated in current wheat breeding strategies.

  • Physiological trait breeding is effective in improving wheat adaptation to stress.

  • Accurate breeding targets, relevant genetic diversity, efficient population screening methods and innovative whole wheat breeding program strategies are essential for sustained success.

2 Introduction

Abiotic stresses significantly limit wheat production globally and the extent and intensity of yield losses are increasing with climate change. Rainfall is declining and the distribution changing in many environments and the impacts will be more acute in rainfed production systems. Current yield losses in wheat are primarily a consequence of abiotic rather than biotic factors [1]; this was not always the case, but a consequence of the steady improvement of disease resistance over the past 100 years [2]. However, wheat breeders have also incrementally improved crop adaptation to stress. This was largely achieved by targeted use of diversity and extensive testing in the environment under prevailing stresses. Thus, empirical selection has improved adaptation to abiotic stresses across the world’s wheat growing areas, despite the genetic complexity and low heritability of these traits compared to disease resistances.

Climate modeling indicates that instability will increase in the major wheat producing areas of the world [3]. However, some regions will suffer more than others, including Australia, North Africa and large parts of North and South America. Expected losses in wheat production due to drought and heat stress, exacerbated by climate change, for key wheat growing regions are outlined below.

2.1 Australia

Australian wheat productivity will be limited by climate change. The Agricultural Production Systems Modulator (APSIM) was used to estimate changes in wheat productivity and response to high temperature for the period 1985–2017 [4]. The production environment had become more variable over the period, and heat stress was found to reduce grain weight more than grain number. Of the yield losses estimated, 26% were associated with heat and the remainder with drought stress. Wheat breeders need to target both stresses as a priority.

2.2 North America

The impact of climate change on North America will be mixed. Climatic changes between 1981 and 2015 have led to higher rainfall and longer growth periods and this was positively associated with grain yield [5]. New winter wheat cultivars with higher yield potential and improved disease resistance are required to meet this shift. However, spring wheat was subjected to increased temperature stress in the critical June period, thus requiring some heat tolerance at anthesis. Wheat breeders need to target higher yield potential and improved heat tolerance at anthesis.

2.3 Europe

Increasing temperatures are projected to reduce wheat yield in Europe. Semenov and Shewry [6] simulated various climate scenarios and predicted that high temperatures, particularly at flowering, would limit wheat yield more than drought. They reasoned that lower summer rainfall would be offset by earlier maturation thus crops would escape the impact of drought. They concluded that wheat breeders should target the improvement of heat tolerance at anthesis as a priority.

2.4 Russia and Ukraine

Like North America, the impacts of climate change will be mixed. The most productive zones of Russia are likely to experience yield losses from reduced precipitation and heat waves during vegetative development [7]. However, milder and drier winters and warmer spring periods in northern production zones are likely to see increases in productivity. In Ukraine, modeling suggests moderate climate change will have little impact on wheat yield. Nevertheless, under high emissions scenarios and higher levels of warming, yield is expected to decrease by more than 11% [8]. Wheat breeders should target improved heat tolerance at all stages of development.

2.5 India

Climate change and increasing temperature will and have already reduced wheat yield in India [9]. Wheat yield is estimated to be 13% higher than it would have been without irrigation trends since 1970 [10]. Irrigation dampens the effect of high temperature and irrigated wheat has just 25% of the sensitivity of rainfed wheat. However, yield gains have slowed due to warming. These authors found that irrigation will have little impact on future warming as opportunity to expand the system is limited. Wheat breeders need to target both high temperature tolerance and better water use efficiency as a priority.

2.6 China

Climate change will limit the productivity of wheat in China. Under the most severe climate change scenarios, wheat yield in China is projected to decline by 9.4% by 2050, which represents the largest yield reduction of all Chinese crops [11]. This 50-year study of Chinese climate data concluded that terminal heat stress was more severe in cooler regions. They concluded that the vegetative period had changed little in these cooler areas, but temperatures post heading had increased significantly thus reducing yield. Development of cultivars with improved terminal heat tolerance should be a priority as much of the wheat production in these regions is irrigated, thus negating the impacts of drought stress.

While climate change has already impacted wheat production in many environments, empirical selection has mitigated the impact of climate change on yield. Thus, rates of genetic gain have plateaued, rather than declined, in many regions. However, rates of genetic gain are not constant over time and fluctuate depending on access to new technologies, such as the introduction of dwarfing genes in the 1960s and 1970s, climatic changes or biophysical yield limitations which have limited recent gains in various regions [12]. However, wheat breeders have access to better technology than ever before, and it can be expected that optimized use of technology will further lessen the impacts of drought and high temperatures. Historically, we can already document changes in wheat morphology and physiology [13]. Yield improvements were associated with shorter vegetative and longer grain filling periods, more grain per unit area, shorter plant stature, wider leaves and higher harvest indices. Modern varieties tend to be earlier maturing, more N use efficient and translocate more assimilate to the developing grain. These changes have come about by coupling empirical selection in the target environment, with access to new diversity, improved understanding of physiological limitations and more recently, better understanding of the genetic control of traits. Nevertheless, optimal integration of technologies remains a significant challenge to wheat breeding and this is discussed further in Chaps. 5 and 6.

3 Breeding for Improved Adaptation to Water-Limited and Heat Stressed Environments

Wheat breeders have many tools available and technology has advanced rapidly in recent years. Molecular markers for high value traits are routinely used in most programs, genomic selection forms part of many strategies, proximal and remote sensing have extended beyond the physiologist’s experiments and is routinely used by some programs and many are considering ways to exploit gene editing effectively. However, no technology ensures high value varieties are delivered to farmers and strategy and the choices that breeders make are vital to success. Wheat breeding is needs driven. Breeding targets must be well defined and relevant to both producers and marketers. The breeder’s choice of technology will reflect the available diversity, heritability of phenotypic screens and availability of markers for high value traits and other genomic strategies that improve rates of genetic gain.

This section will follow the breeder’s decision-making process in the context of improving rates of genetic gain for heat and drought tolerance.

3.1 Relevant Breeding Targets

Most farmers are forthcoming in describing varietal limitations to wheat breeders. In fact, the farmer’s wish list can sometimes be extensive and bear little relationship to the available genetic diversity. Nevertheless, most breeders are aware of production constraints and wheat market requirements. Many production constraints can be solved agronomically and the influence of genetics is so limited that they should not be selection targets. The effectiveness of rotation in controlling take-all (Gaeumannomyces graminis var. tritici) is one such example. Other traits, such as crown rot resistance in wheat, are managed by the interaction of genetics with management practices, like non-host rotation and interrow sowing [14] (see Chap. 9). Other traits with high heritability, such as rust resistance (see Chap. 8), are clear targets for genetic selection. However, the picture is less clear regarding abiotic stress tolerances as a combination of optimized management and targeted traits for specific environments is almost always the goal. Thus, a genetic ideotype that assumes optimized agronomic management for specific environmental conditions, and reflects the most probable or frequent environment type, would help the breeding process [15]. Definition of wheat breeding target environments is discussed in more detail in Chap. 3. Ideotypes for drought stress [16] and heat stress [17] have been developed. These ideotypes are general in nature and the traits identified may not be effective in all environments. For example, under Australian conditions, soluble stem carbohydrates improve drought response in northern Australia but not in southern areas [18]. Knowledge of how these traits interact with the environment is crucial. To explore this further, a national, field-based managed environment facility (MEF) was established at three locations representing the key wheat growing regions of Australia and the effectiveness of traits assessed [19]. The network, where all confounding effects, such as soil heterogeneity and moisture, were minimized, was effective in assigning trait values by region. Similarly, if conservation agriculture is used by farmers to reduce the loss of soil moisture, then more vigorous genotypes that emerge from depth and carry resistance to stubble borne pathogens would be required. Thus, the ideotype must reflect the most frequent genotype x management practice x environment interaction. Building this ideotype will then depend on available genetic diversity; there is no point in including traits for which no diversity exists, and trait heritability, which reflects the accuracy of phenotypic screens and/or availability of molecular markers and their degree of linkage. An example is an ideotype constructed for the specific conditions of northwestern NSW is reproduced in Fig. 10.1 [15].

Fig. 10.1
figure 1

An ideotype for drought stress tolerance in northwestern NSW, Australia. (Reprinted with permission from Ref. [15])

Full size image

3.2 Meaningful Genetic Diversity

Genetic diversity should always be assessed and accessed from the adapted or primary wheat gene pool first. Species in the primary gene pool have completely homologous genomes with common wheat (AA, DD, AABB or AABBDD). Within the primary gene pool, the pathway to market is much shorter if the diversity is available in already adapted materials (AABB, AABBDD). The decision to access such diversity is a function of crossability with adapted wheat, either hexaploid or tetraploid, and the value of the trait. Much has been written about the value and use of synthetic wheat to improve the stress tolerance of wheat [20]. Primary synthetics, generated by crossing tetraploid wheat (such as Triticum turgidum cv durum, T. dicoccum or T. dicoccoides) with Aegilops tauschii, the donor of the D genome with subsequent embryo rescue and chromosome doubling, have been crossed to adapted wheat and cultivars released to farmers [21]. The D genome contributed by Ae. tauschii is more diverse than that in common wheat and this diversity has provided new alleles linked to stress adaptation. Wheat A and B genome diversity can also be introduced through direct crossing of wild tetraploid and adapted hexaploid wheat [17]. The wild tetraploid diversity, once introduced to hexaploid wheat, has been linked to improved drought [22] and high temperature [17] adaptation.

However, sometimes the diversity required for high value traits is not available in the primary gene pool. Only then is an exploration of the secondary gene pool; those materials with partial homology to the common wheat genome, warranted. Such materials include species such as T. timopheevi (AAGG) and Aegilops speltoides (BB) with one genome common to hexaploid wheat. One such example is the translocation of a segment of Aegilops speltoides in wheat linked to a more profuse root system and enhanced drought tolerance [23]. As a last resort, diversity for high value traits can be sourced from the tertiary gene pool where no homology with the common wheat genome exists. Historically, this diversity took a long time to introduce and was often associated with yield penalties caused by linkage drag. Examples include rye (RR) and Thinopyrum elongatum (EE). However, new genomic tools have made it easier to target and exploit tertiary diversity in wheat, and it is expected that the tertiary gene pool will be increasingly exploited to improve both the biotic and abiotic stress tolerance of tetraploid and hexaploid wheat. Chapters 16, 17 and 18 detail the conservation, characterization and use of genetic resources.

3.3 To Phenotype or Not?

Phenotyping is expensive and often comprises the greatest cost in any breeding program. Historically, parents are selected and crossed based on genetic and phenotypic information and availability of high value traits with high heritability that are amenable to high throughput screening. These have traditionally included traits such as disease resistance, plant height and phenology. However, marker assisted selection has broadened the suite of traits assessed in the early generations in recent years and grain quality, disease, phenology and even some major QTL linked to abiotic stress response have been used to truncate populations. If robust and tightly linked markers for high value traits exist, then it is not necessary to phenotype beyond parents and their fixed line progeny, thus reducing costs. It is assumed of course that the phenotypes used to identify the marker-trait linkages are accurate, repeatable and relevant. The same applies to the calculation of genomic estimated breeding values (GEBVs); it is assumed that the training population size is optimized, the phenotype accurate and the relationship between the training population and the breeding materials relatively close.

Nevertheless, some high value traits are difficult to phenotype and the available genetic information insufficient to justify a genomic approach alone. These include drought and heat tolerance as most observed QTL are of small effect and the influence of environment on QTL expression significant. Remote and proximal sensing are becoming increasingly valuable sources of information on genotype physiological responses to stress. Thermal infrared sun-induced fluorescence combined with solar-reflective hyperspectral remote sensing are considered state-of-the-art applications for assessing plant stress responses but require satellite access. However, proximal sensing using unmanned aerial vehicles (UAVs) or ground based phenomobiles, is now widely used by some breeding programs to capture real time thermal and spectral reflectance data on large numbers of genotypes. These data can be collected over time and responses with the highest heritability used to drive phenotypic selection and inform genomic prediction models. Kyratzis et al. [24] used an UAV to assess drought stress response in durum wheat and concluded that green NDVI (Normalized Difference Vegetation Index) was effective in discriminating genotypes. Nevertheless, the greatest limitation for the plant breeder is often the data processing to produce plot means and standard deviations that can be used for timely selection. The challenges of high-throughput phenotyping are discussed in Chap. 27.

Regardless of the technology used to capture field-based data, the information will have little value if field screening is confounded by heterogeneity or the season is not representative of the most common environment type. Pot-based screening in glasshouses or phenomics facilities can be effective in controlling environmental fluctuations for traits with high heritability. However, when applied to stresses such as heat and drought, the results rarely correlate with field responses thus limiting the utility of such data to the plant breeder [25]. While the field environment is subject to uncontrollable variation, the impact of confounding factors such as soil heterogeneity or season rainfall can be minimized.

As mentioned earlier, one such example is the MEF in Australia [19]. Materials are sown in a carefully managed crop sequences designed to limit wheat root diseases through rotation with non-host alternative crops, and soil heterogeneity is carefully assessed before sowing using an EM38 to detect differences in soil moisture and texture. The most homogenous areas are then selected for drought evaluation using an irrigation treatment split. Rainfall and soil moisture are assessed so that an environment type can be estimated, and this informs genotype responses. Once high value materials are identified, they are subsequently evaluated using rainshelters to control seasonal moisture and confirm drought responses in the field.

A slightly different approach can be used to evaluate genotype response to high temperature [26]. Here dates of sowing are used to screen thousands of genotypes for high temperature response at anthesis and grain filling. However, abnormal biomass development from a truncated vegetative period in late sown materials could influence estimations of grain number and seed weight under stress. To counter this, materials selected from delayed sowing are subsequently sown at an optimal time and portable field-based heat chambers used to apply a heat shock at anthesis for several days (Fig. 10.2). Those materials that maintain seed number and weight are then selected for final confirmation under controlled greenhouse conditions. This three-tiered phenotyping system thus overcomes the lack of relationship between glasshouse and field screening by inverting the process to initially screen in the field, followed by increasing levels of phenotyping precision on smaller numbers of lines. A more detailed discussion of heat stress methods and traits can be found in Chap. 22.

Fig. 10.2
figure 2

Heat chambers with attached air conditioning units deployed in the field at Narrabri, NSW, Australia

Full size image

3.4 Physiological Wheat Breeding

The term ‘physiological breeding’ was first coined by Reynolds et al. [27] and refers to crossing parents carrying complementary traits with subsequent progeny screening under stress. Parental materials are selected based on the suite of traits required in a breeding program, including yield potential, yield under stress, seed weight, grain quality and disease resistance. However, all materials are subsequently assessed for the physiological traits deemed effective in the target environment. For example, these might comprise the traits in the ideotype in Fig. 10.1, if the target environment is northwestern NSW. However, some traits are more easily assessed than others and their amenability for high-throughput field-based phenotyping will determine if they are assessed on parents only or used to truncate segregating materials during progeny selection. Trethowan [15] categorized many physiological traits into those associated with emergence and establishment, early growth, pre-flowering and post-flowering. Traits such as osmotic adjustment can be assessed in the pre and post-flowering periods; but is difficult and time consuming to measure. This trait would therefore only be assessed on parents and again on fixed line progeny expressing drought tolerance in multi-environment testing. In contrast, canopy temperature depression; a trait assessable at the same developmental stages, is easily measured and can be used to select segregating materials, either using handheld sensors or remote or proximal sensing.

The concept of physiological breeding has been successfully applied in wheat breeding [27]. Materials developed using physiological crossing for drought tolerance at the International Maize and Wheat Improvement Centre (CIMMYT), were subsequently deployed in south Asia and found to be tolerant to drought [28]. Lines developed by crossing complementary physiological traits had on average, higher yield, superior grain weights and cooler canopies. New drought tolerant wheat cultivars were subsequently released to farmers in Pakistan from materials developed at CIMMYT using physiological breeding, including Barani-2017 and Kohat-2017. Targeted physiological trait introgression was successfully used to develop the Australian wheat cultivar Drysdale [29]. Here carbon isotope discrimination, which is negatively correlated with transpiration efficiency, was backcrossed into an elite background and the transpiration efficient cultivar, Drysdale, was released in southern Australia in 2002 and the cultivar Rees for northern regions the following year. The Drysdale and Rees examples show that a single targeted physiological trait can have tangible benefits in a dry environment. This is discussed further in Chap. 23.

Wheat breeding and the enhancement of farmer profitability is more than just targeting stress adaptive traits. Most farmers in marginal environments tend to make most of their income in the better years. Hence, physiological traits that do not limit yield potential and have a higher heritability than yield alone would have high value in wheat improvement. For example, Pozo et al. [30] found that chlorophyll content was positively associated with yield under optimal and drought conditions, whereas carbon isotope discrimination was associated under optimal conditions only. In their study water soluble stem carbohydrates assessed at anthesis were not associated with yield at any level of moisture. Such findings help tailor trait selection for given production conditions.

3.5 Integration of Genomic Technologies in a Broader Physiological Breeding Strategy

Molecular markers linked to physiological traits that are deemed effective in the target environment and do not limit yield in the better years, significantly improve the effectiveness and cost of physiological breeding. Many studies have reported QTL linked to physiological traits with varying degrees of accuracy [31]. Unfortunately, many physiological traits are difficult to measure and have relatively low heritability, including stomatal conductance and photosynthetic rate, making marker development difficult and offering marginal value to the plant breeder [32]. Many QTLs are also cross specific and their expression fades when transferred to different backgrounds. Chen et al. [33] found that morphological traits such as plant height and peduncle length had high heritability while most physiological traits, including photosynthetic and transpiration rates, intercellular CO2 concentration and stomatal conductance were low. Nevertheless, they concluded that five traits, including yield per plant, plant height, peduncle length, spike length and transpiration rate explained more than 90% of the variation in genotype response to drought.

Genomic selection, once the realm of the animal breeder/geneticist, is now integrated into many wheat breeding programs. Nevertheless, success depends on the accuracy, depth and relevance of the training population phenotype as much as the relatedness to the breeding population. Genomic estimated breeding values can be calculated for stress response based on a weighted index of associated traits. If makers linked to specific physiological traits are known, then they can be targeted in progeny selection following genotyping so that both GEBV and known QTL or gene profiles are optimized. A more detailed analysis of genomic selection can be found in Chaps. 6 and 32.

4 Examples of Integrating Physiological Breeding in Wheat Improvement Programs

To examine the practicalities of integrating physiological trait breeding in the wheat breeding process, three examples that compass commonly adopted breeding methods are presented. These methods include modified pedigree, selected bulk and a genomic strategy (Fig. 10.3). Pedigree breeding was not considered as so few programs use a strict pedigree breeding scheme due to the significant resources required. However, before crossing begins, it is necessary to accurately define the target environment (see Chap. 3 for more detail), the most likely probability of stress occurrence based on historical evidence and the suite of traits to be targeted [15]. Northwestern NSW in Australia will be used as an example; however, the principle can be applied to any environment. Details of genomic selection, including the available models and their applications, are provided in Chap. 6 and the general principle only, in the context of a wider physiological breeding strategy, will be discussed.

Fig. 10.3
figure 3

Three strategies that integrate physiological breeding and selection

Full size image

4.1 Defining the Environment in Northwestern NSW

The environment in northwestern NSW is characterized by summer dominant rainfall and extensive vertosol soils with high water holding capacity. The region lies between 26–30° latitude south. In season drought and heat stress are common, particularly from anthesis onwards [15]. Heat stress, defined as temperatures in excess of 35 °C for short periods of time, is common [26]. Stem, leaf and stripe rust, crown rot and root lesion nematode are major biotic constraints. The region produces high quality, high protein wheat that attracts a premium price. The optimum sowing time, based on simulation modeling, to minimize the risk of temperature extremes lies between 6 – 20th May [34].

4.2 Establishing an Ideotype for Northwestern NSW

Phenology is a primary driver of yield and matching phenology to the environment is critical to minimizing the impacts of stress. These responses are controlled by three loci each, PpdD1, Ppd2 and Ppd3 and VrnA1, VrnB1 and VrnD1 for photoperiod and vernalization responses, respectively. Daylength and vernalization insensitivity are controlled by dominant alleles at these loci and at this latitude, dominant alleles at PpD1 and VrnA1 with recessive alleles at the remaining vrn loci optimize the flowering window [15]. Plants should be semi-dwarf in stature to avoid lodging and if possible, height should be controlled by gibberellic acid sensitive dwarfing genes that do not significantly reduce coleoptile length and hence emergence and establishment [35]. Rapid early growth and ground cover will assist crop establishment in standing stubble as conservation agriculture is widely practiced in the region. In the region, stay-green, associated with deeper roots that extract soil moisture from depth, is an important character [36]. Genotypes with high water-soluble stem carbohydrates are also high yielding [18]. Genotypes with cooler canopies use soil moisture more effectively and continue to photosynthesize as temperature and moisture stress increases [16]. However, it is not clear whether high levels of transpiration efficiency will be beneficial given the trade off with high yield under less-limiting conditions in better years. Pollen fertility under heat stress and maintenance of grain weight under both heat and drought are important characteristics. Resistances to the rust diseases, root lesion nematodes and crown rot are required.

4.3 Breeding Method – Modified Pedigree

Once the ideotype has been developed and the trait selection determined, it is necessary to screen breeding materials, introductions and new diversity for the suite of traits required. Assessment may be phenotypic and/or genetic if tightly linked or perfect trait markers are available. At this stage, managed field environments, augmented by controlled environment testing can be used to establish trait profiles. Materials are then combined in backcrosses (when the source or non-recurrent parent is unadapted or carries deleterious characteristics), two or three-way crosses to combine physiological traits and other important diversity. A large F2 population is sown and single plants selected based on highly heritable and economically important traits such as rust resistance, plant height and maturity. If co-dominant physiological trait markers are available, they can also be used to drive F2 single plant selection.

Under the modified pedigree scheme, each selected plant becomes an F2:3 plot. Simple to measure physiological tools, such as canopy temperature depression or NDVI can be assessed manually or using proximal/remote sensing. A bulk of spikes taken from selected plants from each plot is then advanced to an F2:4 plot and the process repeated until individual plants are retained from the F5 generation to form the new fined line (F5:6). Marker assisted selection can augment this process as required. The near homozygous materials are then multiplied and evaluated across the target environment and physiological trait combinations confirmed in the best performing materials.

4.4 Breeding Method – Selected Bulk

The process of parental selection and crossing is identical to the modified pedigree or pedigree system up to the F2:3 generation. Individual plants selected from the F2 generation are bulked and not maintained as individual plots. Marker assisted selection for linked physiological traits on F2 plants can still be performed if required before bulking. Individual plant selections are bulked each generation and once the required level of homozygosity is reached, usually by F5, individual plants are selected, and these F5:6 selections become the new fixed lines. These enter multi-environment testing and confirmation of physiological trait expression as per the modified pedigree method. There are many fewer but much larger plots in the selected bulk method between F3 and F5 compared to modified pedigree. Nevertheless, these populations can still be assessed using easy to measure traits, such as canopy temperature depression or NDVI. Higher numbers of plant can be selected from plots with better agronomic type and superior physiological trait values, thus favorably skewing gene frequency.

4.5 Breeding Method – Genomic Selection

This breeding approach varies from the previous two strategies once the target ideotype has been determined. This section will focus on integrating physiological traits in a broader genomic breeding scheme. A training population representing the diversity required to assemble the ideotype, preferably in adapted backgrounds, of more than 2000 individuals is assembled. If physiological traits are found in unadapted materials, it is better to first derive lines carrying the trait in better agronomic backgrounds using backcrossing, otherwise the training population phenotype will be compromised by morphological and phenological extremes. In general, if plant height and phenology fall within a relatively narrow range, then the population phenotype is deemed comparable [27]. The training population phenotype must be largely field-based and should extend over time and space. The more accurate the phenotype, including physiological traits, the better the GEBVs upon which crossing and selection decisions will be based. Traits that correlate with yield under stress can be integrated using a weighted index based on heritability and GEBVs subsequently calculated using all the available information.

Crosses would then be made among genetically distant lines with high GEBVs that include, where possible, known marker-physiological trait associations. To optimize linkage disequilibrium, materials should be genotyped and recombined in crosses by the F4. At this stage a reasonable degree of homozygosity has been reached and the materials can be advanced from F2 – F3 rapidly using single seed decent or other methods of population advance that maintain gene frequency. Single plant selections taken at F3 would be genotyped and the F3:4 grown in plots. F4:5 progeny selections would be retained for multi-environment testing and trait validation from those F4 plots with high GEBVs. The F3:4 materials with highest GEBVs and greatest genetic distance would then be recombined in crosses and the process begun again without phenotyping. Following at least two breeding cycles based on genotype alone, the derived materials would be evaluated in multi-environment trials and phenotyped to confirm combined physiological traits. Superior materials would then cycle back to the training population along with a continuous flow of new alleles.

5 Key Concepts and Conclusions

Drought and heat stress tolerance will become increasingly important wheat breeding objectives in most wheat growing regions, increasing investment in these stresses and the opportunities for collaboration within and across regions. Most wheat breeding programs use a handful of breeding methodologies or modifications of these methods to derive new cultivars for farmers. The integration of new wheat breeding tools and knowledge does not entail a complete restructuring of breeding programs as such changes do have significant economic consequences. Instead, new technologies and knowledge can be integrated effectively with current commonly used breeding methods. These technologies are simply efficiencies that advance the overall goal of delivering better cultivars faster.

Physiological breeding is one such strategy that is easily integrated and entails better characterization of parents for physiological traits relevant to the target environment, implementation of an appropriate selection strategy that may entail high-throughput phenotyping, molecular markers or empirical selection under stress, followed by extensive evaluation of fixed lines under the stress and across multiple environments within the target region.

However, physiological breeding should become an obsolete term, as these traits are simply part of the suite of traits accessible to the plant breeder interested in improving crop adaptation to stress. Decisions to use these traits in crossing and selection will depend, as always, on heritability, ease of assessment and importance to farmers and industry. However, as Reynolds and others have shown [28, 37], at a minimum, they can be incorporated at crossing and confirmed following empirical selection in the target environment. Proximal and remote sensing are also changing the method (and scale) of assessment of physiological traits and these data can be used to truncate populations and favorably skew gene frequency. Physiological characterization can be easily incorporated into genomic selection strategies including the calculation of GEBVs based on weighted trait values.