This review considers stomatal conductance as an indicator of genotypic differences in the growth response to water stress. The benefits of using stomatal conductance are compared with photosynthetic rate and other indicators of genetic variation in water stress tolerance, along with the use of modern phenomics technologies. Various treatments for screening for genetic diversity in response to water deficit in controlled environments are considered. There is no perfect medium: there are pitfalls in using soil in pots, and in using hydroponics with ionic and non-ionic osmotica. Use of mixed salts or NaCl is recommended over non-ionic osmotica. Developments in infrared thermography provide new and feasible screening methods for detecting genetic variation in the stomatal response to water deficit in controlled environments and in the field.
The drought tolerance of a crop is essentially linked to its ability to access soil water and to use it most productively (Richards et al., 2010). Strategies for water use that confer drought tolerance can be quite different for annual and perennial species, and for dryland versus irrigated agriculture. For annual crops such as wheat and barley in semi-arid environments, with mild winters and hot summers, one successful strategy is a fast rate of development, and a short time to flowering and grain maturity, allowing the available water to be used by the plant before it is lost from the soil as the temperature increases. Another is to choose a slow-developing cultivar and sow early (Passioura and Angus, 2010). Perennial species can employ a conservative strategy, minimizing the use the use of water to avoid the risk of leaf dehydration, and resuming a fast growth rate when the rainy season returns.
Over the time scale of days to months, water loss is determined by leaf area and stomatal conductance. Over the time scale of minutes to hours, it is regulated by stomatal behaviour. Stomatal conductance responds rapidly and sustainably to changes in soil water potential, and provides the main limitation to photosynthesis and growth.
A sensitive growth response to decreases in soil water potential would be beneficial for crops growing without irrigation on stored water and likely to endure a terminal drought. Root signals providing a feed-forward regulation of leaf area expansion and stomatal conductance can adapt the plant to a drying soil (Davies et al., 2005). A less sensitive response would be useful for crops growing with irrigation, as productivity could be maintained for the periods between the applications of irrigation water. Selecting genotypes with diverse responses to a decrease in soil water potential would provide an option to growers in different environments. This review discusses possible screening techniques to select for genetic variation in the growth response to soil water stress.
Stomatal conductance versus leaf area or photosynthesis as a screen for genotypic differences in tolerance of soil water deficit
Traditional methods of measuring growth have been time-consuming and have often involved the destructive harvest of plants. Recent imaging technologies to estimate biomass and growth parameters have gone some way to alleviate these issues, but lose accuracy as plants become larger and produce multiple shoots. It was found that when the total leaf area of a wheat plant exceeded 100 cm2, imaging analyses become a less reliable indicator of leaf area, the error being over ±10% (Rajendran et al., 2009). As an indicator of growth rate, measurement of stomatal conductance has the advantage over measurement of leaf area expansion rate or relative growth rate, because the latter two indices require two measurements. In addition, stomatal conductance is quicker to measure than leaf area or biomass, allowing for greater numbers of individual plants to be screened.
Measuring stomatal conductance has an advantage over measuring photosynthesis, as the former is often more sensitive to water deficit on a per unit area basis (James et al., 2002). Using salinity as a water stress, a stress-induced reduction in stomatal conductance of 50% was shown in the emerging leaf, but photosynthetic rate was affected by only 10% (James et al., 2002). The small effect on photosynthetic rate (per unit leaf area) was due to the changed leaf and cell morphology. Leaves were narrower, thicker, and visibly greener, due to smaller cells with different dimensions, giving a higher chloroplast density per unit leaf area. They were thus able to compensate for the lower stomatal conductance and maintain their photosynthetic rate (on a unit leaf area basis, but not a leaf basis). This increase in transpiration efficiency had been previously found for wheat (Rivelli et al., 2002) and barley (Rawson et al., 1988) with increasing salinity.
The disadvantage of stomatal conductance is that it can be highly variable (Jones, 1987). It is sensitive not only to light, but also to vapour pressure deficit and CO2 concentration, which need to be controlled. This is particularly problematic when measuring stomatal conductance with a hand-held porometer in a small space, such as a controlled environment chamber, as respired CO2 causes concentrations in the chamber to rise quickly above ambient and cause stomatal closure (James et al., 2008). The impact on conductance of some of these factors can be minimized when using infrared thermography to estimate stomatal conductance in the laboratory (Sirault et al., 2009) and field (Jones et al., 2009).
High throughput measurements of photosynthesis are challenging because traditional CO2 exchange requires the equilibration of leaves in specialized small chambers (cuvettes) on single leaves. Options for high throughput estimations of photosynthesis or for assays of the health of the photosynthetic apparatus under drought stress have been limited to measuring oxygen evolution at super saturating CO2 in a closed cuvette (often using leaf pieces: Jones, 1979; Delieu and Walker, 1983) or more recently, by estimating electron transport rate or photodamage from chlorophyll fluorescence quenching parameters (Baker and Rosenqvist, 2004; Baker, 2008; Jansen et al., 2009).
The most easily measured and hence the most commonly used fluorescence parameter in stress studies is dark adapted Fv/Fm (a measure of the intrinsic photochemical efficiency of light harvesting in photosystem II; Baker, 2008). As this measurement is now possible on whole plants using fluorescence imaging technologies (Baker, 2008), it is feasible to obtain a whole plant average measurement or target leaves at the same developmental stage. This is particularly applicable to high-throughput studies of stress response in species which grow predominantly in the horizontal plane in the seedling stage (such as Arabidopsis and tobacco) and thus can be grown in large trays or, for very small seedlings, in microtitre plates or similar vessels. Fluorescence imaging also allows the determination of projected leaf area and hence growth rate if measurements are made regularly over time (Barbagallo et al., 2003). Fv/Fm has recently been measured in two drought studies with Arabidopsis using systems that are scalable to high-throughput screening (Woo et al., 2008, Jansen et al., 2009). Figure 1 (adapted from Jansen et al., 2009) illustrates the time-course of relative growth rate (RGR) and Fv/Fm in the Arabidopsis ecotype C24 after withholding water. Clearly, as also shown in many other studies (e.g. Massacci and Jones, 1990) , Fv/Fm is relatively insensitive to drought stress whereas RGR falls off rapidly after watering is ceased. While stomatal conductance or photosynthetic rate were not assayed in these experiments, it is likely that both these parameters would have responded rapidly and similarly in extent to RGR. With salinity stress, the same result was found: Fv/Fm being no more sensitive than a measure of chlorophyll concentration with a SPAD meter (James et al., 2002).
An alternative to measuring Fv/Fm to detect drought-induced damage to the light harvesting system is to quantify the fast transients in chlorophyll fluorescence occurring in the first second or so following illumination, known as the OIJP transient (see Oukarroum et al., 2009, and references therein). This measurement has been shown to be a sensitive method for ranking drought tolerance in the early vegetative growth of barley cultivars (Oukarroum et al., 2009). As this measurement is generally made on single leaves, throughput is lower than with imaging methods and as with measurements of Fv/Fm, it is not clear whether adaptive or only survival traits can be screened for.
The experiments of Jansen et al. (2008) also show that Fv/Fm is useful only as a drought ‘survival’ screen, which is of limited utility for most annual crops. So far, chlorophyll fluorescence has not been widely used to estimate the response of photosynthetic electron transport rate to drought stress, although this can be calculated from chlorophyll fluorescence quenching if the incident light intensity and the absorption properties of the leaf are known (Baker, 2008).
Measurements of stomatal conductance or photosynthesis on a single leaf at a single developmental stage are not necessarily indicators of the average of all leaves. For this purpose, new imaging technologies for estimating conductance or photosynthesis, such as thermal infrared imaging for conductance (Jones et al., 2009), and reflectance spectroscopy for photosynthetic response (Siebke and Ball, 2009), have potential value as they can be used at the whole plant or canopy level.
Alternative treatment methods for imposing a plant water deficit: soils in pots versus ionic and non-ionic osmotica in hydroponic culture
The search for the best medium for growing plants to impose a controlled water deficit has been going on for decades, without a clear resolution. There is no perfect medium, all have limitations including pots containing real soil, that is, soil imported from the field.
Soil drying is hard to control especially when comparing genotypes of different vigour or rates of development. Even a drying soil, as well as being very difficult to maintain at a uniform and constant water potential through the whole soil profile, may exert specific effects; for example, transmission of nutrients through the soil will be reduced at low soil water potentials (Nye and Tinker, 1977). Another problem with soil in pots that are not deep is that they easily become saturated at the bottom (Passioura, 2006), so much so that often plants in the ‘control’ treatment do not grow as well as those in soil with a moderate water deficit. Pots should be tall to enhance drainage. Saturation can be avoided by the use of ‘inorganic soils’ such as fritted or calcined clay. Use of artificial material with large particles and little root contact may be problematic. Vermiculite caused a greater reduction in root growth at low water potential than did liquid media at the same water potential, at least up to –1.6 MPa (Verslues et al., 1998).
Hydroponics is often used as a way of overcoming the problems of heterogeneity, drainage, and inconstant water potential. A variety of non-ionic osmotica have previously been used to mimic a decrease in soil water potential, including mannitol, melibiose, and sorbitol. However, all these small molecules eventually enter roots and move in the xylem to the shoots, because membranes are permeable to neutral solutes of this size. Mannitol was rapidly taken up by cells, especially at low water potential (Hohl and Schopfer, 1991). Carbohydrates also support bacterial growth as it is impossible to create an aseptic root environment.
High-molecular-weight polyethylene glycol (PEG) has been examined in many early studies that attempted to impose a controlled water deficit. Its main problem is that it limits O2 diffusion to roots (Mexal et al., 1975). PEG is very viscous which reduces stirring of solution and decreases O2 movement to roots so that the roots become O2 deficient (Mexal et al., 1975; Verslues et al., 1998). This probably explains why it can interfere with ion transport (Yeo and Flowers, 1984). When PEG solutions were bubbled with O2 rather than air, the rate of root growth increased, and the partial pressure of O2 at the root tip increased (Verslues et al., 1998). These authors warned that vigorous bubbling with air is not sufficient to overcome the O2 problem; a supply of O2 with normal gentle aeration is needed as vigorous bubbling causes roots to break.
Not only bubbling causes damage to roots: laterals break when the solution is changed and the roots are unsupported (Miller, 1987), and solution enters through the damaged junction between a lateral root and the seminal axis. This is possibly the mode of entry of PEG described by Lawlor (1970). This problem can be overcome by using solid media such as quartz gravel which supports the roots when the solution is removed or aerated (Munns and James, 2003).
Although Verslues et al. (1998) were able to carry out experiments using PEG with an additional O2 supply, these were limited to 3 d and with seedlings that were not transpiring (maize seedlings grown in the dark). If plants are transpiring, it is likely that significant amounts of PEG will be taken up (Lawlor, 1970).
NaCl can be used as a surrogate for water stress. Both drought and salinity cause an inhibition of plant growth through the reduced soil water potential. Salt in the soil generates an osmotic stress which constitutes the major limitation to growth of wheat and barley in saline soil, causing reduced leaf expansion rates and tiller formation.
It has advantages over alternative media, because it is not toxic to most species when used at concentrations up to 100 mM (Munns, 2002). In the short term it is also unlikely that Na+ or C1– concentrations in the leaves are limiting or determining growth: leaf expansion recovered immediately after NaCl was removed from around the roots (Munns et al., 1982; Rawson and Munns, 1984). Na+ and Cl– concentrations are low in rapidly growing cells (Munns et al., 1988) and may indeed limit their growth because of this (Greenway and Munns, 1983).
Mixed salts, for example, high concentrations of the macronutrients in Hoagland's solution (a protocol described by Munns, 2010,a, in PrometheusWiki) are a better osmotica than any of the above, as their rate of uptake is tightly regulated by transporters that have evolved to deal with variable concentrations in soils, and they do not support bacterial growth. Concentrated macronutrients were shown by Termaat and Munns (1986), Rengasamy (2010), and Tavakkoli et al. (2010) to be a good surrogate for water deficit caused by osmotic stress.
Successful use of NaCl salinity and stomatal conductance to screen for genotypic variation in growth rate in response in response to water stress
Stomatal conductance measured with a cycling porometer was used to screen durum wheat for genotypic variation in response to water deficit, using NaCl as an osmotic stress (James et al., 2008; Rahnama et al., 2010). The stomatal closure was due to the water deficit rather than to a salt-specific effect, as shown by the immediacy of the response (Fig. 2), and by the fact that KCl caused exactly the same degree of stomatal closure (Rahnama et al., 2010). Significant stomatal closure occurred with 50 mM NaCl (0.25 MPa), but a concentration of 150 mM NaCl (0.75 MPa) was needed to reveal genotypic differences (Fig. 3) (Rahnama et al., 2010).
Higher stomatal conductance under stress was linked to higher growth rates, tiller numbers, and shoot biomass (James et al., 2008; Rahnama et al., 2010). Similarly, Jiang et al. (2006) concluded that stomatal conductance measured using a porometer could be useful to select for yield potential in barley under salt-stressed conditions. Further, a significant relationship was found between higher stomatal conductance and higher yield in field-grown wheat and Pima cotton (Lu et al., 1998). In that study, the authors concluded that a cooler canopy from more open stomata and higher transpiration rates was likely to account for improved yields as there was no apparent change in photosynthesis. Other studies have highlighted the link between higher stomatal conductance, cooler canopies, and increased photosynthesis leading to higher yields in wheat in non-water-limiting environments (Fischer et al., 1998).
As leaf temperature is an indicator of stomatal conductance, automating the analysis of thermal images acquired with long-wave infrared (IR) sensors (Sirault et al., 2009) have great potential for the development of high throughput screens, in particular, if the image capture can also be automated. In that study, leaf temperature differences between barley grown at 200 mM NaCl and 0 mM NaCl reached 1.6 °C (Fig. 4). The genotypic differences in temperature increased at the high salt concentrations (R Munns, unpublished data). Genotypes known to vary for osmotic stress tolerance were grown in control (no salt) and 150 mM NaCl treatments. The ranking of the 18 genotypes based on both a growth study and the thermal IR measurements was consistent with previous reports in the literature for these genotypes (James et al., 2008), thus supporting the potential of IR thermal imaging for the screening of large numbers of genotypes varying for stomatal traits.
Developments in methodology with infrared thermography: scaling up from controlled environment chamber to the field
Although thermography has been used successfully for many years for genetic screening in controlled environments (Raskin and Ladyman, 1988; Merlot et al., 2002) it is difficult to scale up the technology to the field environment (Jones et al., 2009). The main problems relate to (i) separation of the plant canopy of interest from the background soil (which is often hotter than the leaves), (ii) the wide range of leaf temperatures resulting from both variation in stomatal conductance through the canopy and variation in absorbed short-wave solar radiation depending on leaf and solar orientation and shading, and (iii) the temporal variation resulting from non-constant illumination (e.g. as a result of clouds) and the rapid variation in wind speed and boundary layer conductance. There has been substantial recent progress in all these areas, with the separation of the leaf from the background with both thermal thresholds (Giuliani and Flore, 2000; Jones et al., 2002) and image analysis techniques being useful (Leinonen and Jones, 2004). There is also the possibility of using linear un-mixing to extract the temperatures of canopy and soil components where there is a predominance of mixed pixels—as will occur with cereal canopies in the field (McCabe et al., 2008) (Fig. 5). The temperature variation from leaf-to-leaf, far from necessarily being a problem, provides the basis of one approach to the detection of stomatal closure (Fuchs, 1990), with stressed canopies theoretically showing a greater temperature variance than well-watered canopies (Bryant and Moran, 1999; Jones et al., 2002). Temperature variance is more of a problem when one attempts to estimate an absolute stomatal conductance for each genotype than where one simply aims to rank different genotypes (Jones et al., 2009).
Another factor confounding the detection of differences in stomatal response is that observed canopy temperature may depend to some extent on plant height (Giunta et al., 2008). The use of multiple view angles has been shown to be of value for the estimation of the true aerodynamic canopy temperature of some crops for the detection of water stress (Luquet et al., 2003), although such an approach may be too complex for routine screening applications.
Notwithstanding all the problems there have been a number of successful attempts to identify stress tolerance using canopy temperature data, both among rice genotypes (Jones et al., 2009) and wheat genotypes (Amani et al., 1996; Reynolds et al., 1998; Ayeneh et al., 2002), even when using infrared temperature probes, rather than the more powerful imagers.
In relation to stomatal screening, it is necessary to consider whether the species or cultivar displays isohydric or anisohydric behaviour, and to what extent it is influenced by vapour pressure deficit (Tardieu and Simonneau, 1998; Collins et al., 2010). Significant genetic variability of transpiration response to vapour pressure deficit was found amongst soybean cultivars (Sadok and Sinclair, 2009). Isohydric species maintain a nearly constant leaf water potential during the day, by stomatal closure, whereas in anisotrophic species the water potential decreases with evaporative demand (Tardieu and Simonneau, 1998). However, some plants can change from anisohydric to isohydric behaviour with increasing soil water deficit (Collins et al., 2010). Wheat and barley show anisohydric behaviour with increasing water deficit. Leaf water potential of wheat and barley decreased as the root water potential decreased, and osmotic adjustment resulted in turgor being maintained (shown in Fig. 2 in Boyer et al., 2008). In all cases, extra information on the effect of the stomatal behaviour on plant adaptation and growth rate would be gained by accompanying measurements of leaf water status, preferably turgor.
Measurements of water status, however, tend to be quite difficult to apply under high throughput screening conditions. Measurement of relative water content (RWC) is a moderately high-throughput measurement, which involves hydrating detached leaves for 3–4 h in distilled water, and measuring the increase in water content relative to the dry weight (protocol described by Munns, 2010,b, in PrometheusWiki). RWC measures the dehydration of a leaf, and is useful for assessing changes in leaf water status when osmotic adjustment has not occurred (Boyer et al., 2008). However, the traditional method is not valid if osmotic adjustment has occurred, because leaves with a higher concentration of solutes will take up more water than their unstressed counterparts, and give a falsely low indication of their water status (Boyer et al., 2008). The traditional method therefore must be modified to assess the water status of drought-adapted plants, so that leaves are rehydrated by eliminating transpiration (e.g. by placing the plants in the dark) while the plants remain intact with roots still in the soil of low water potential (Boyer et al., 2008).
There is increasing evidence that water content can be estimated remotely using hyperspectral sensing based on the water absorption bands in the mid-infrared. Many spectral indices have been proposed to estimate remotely the water content of tissues as a measure of water deficit stress (Chen et al., 2005; Seelig et al., 2008,a, b; Pimstein et al., 2009); these could, in principle, be adopted in a screening programme. These are based on spectral bands in the near- to mid-infrared where there are strong water-absorbing features (970 nm, 1200 nm, 1450 nm, 1930 nm, and 2500 nm). Use of the band at 1450 nm is generally found to represent a good trade-off between sensor accuracy and depth of the absorption band (Seelig et al., 2008b), but only for laboratory measurements. Normalizing the results against values at nearby wavelengths where water absorbs more weakly, if at all, can improve accuracy; for the 1450 nm band a suitable reference is 1300 nm. In the laboratory one can also quantify water in leaves by the use of transmittance measurements with the same ratios used as for reflectance measurements (Seelig et al., 2008a).
All measurements, but especially transmission, primarily detect the amount of water per unit leaf area, which can be expressed as an equivalent leaf water thickness (EWT, m) by dividing by the density of water; i.e. EWT=(fresh mass–dry mass)/ρw.A, where ρw is the density of water. These measures of the absolute water content of leaves may not be particularly useful in the absence of other information, such as a comparison of stressed and unstressed crops.
Drought stress leads to a wide range of physiological responses in addition to the obvious reductions in photosynthesis, stomatal conductance, and leaf growth. These include changes in spectral reflectance and fluorescence related to altered light harvesting and biochemistry caused by oxidative stress. A combination of spectral reflectance, fluorescence, and thermal sensing can provide a powerful tool both for diagnosing particular stresses (including drought responses) and for quantifying the responses in different genotypes (Chaerle et al., 2009).
Stomatal conductance can be a reliable indicator of growth rate response to stress, and thermal imaging is a possible screening method for both the laboratory and field. Genetic variation in response to stress can be exploited in annual crops if irrigation water is available. Tolerant (small stomatal response) lines could be useful for irrigation in arid conditions. Sensitive (large stomatal response) lines could be useful for long-term drought.
It is concluded that new high-throughput imaging technologies in controlled environments can provide a useful way of screening large numbers of genotypes and identifying small differences in growth rate or expression of certain traits. A variety of non-destructive spectroscopic and imaging techniques are now available to assess photosynthetic performance, plant function, and chemical composition, which are potentially scalable from the leaf to the canopy level (Furbank, 2009). However, the selection of parents for breeding should be made only after the results in controlled environments are validated in real soil and real environments.