Generating Plants with Improved Water Use Efﬁciency

: To improve sustainability of agriculture, high yielding crop varieties with improved water use efﬁciency (WUE) are needed. Despite the feasibility of assessing WUE using different measurement techniques, breeding for WUE and high yield is a major challenge. Factors inﬂuencing the trait under ﬁeld conditions are complex, including different scenarios of water availability. Plants with C 3 photosynthesis are able to moderately increase WUE by restricting transpiration, resulting in higher intrinsic WUE ( i WUE) at the leaf level. However, reduced CO 2 uptake negatively inﬂuences photosynthesis and possibly growth and yield as well. The negative correlation of growth and WUE could be partly disconnected in model plant species with implications for crops. In this paper, we discuss recent insights obtained for Arabidopsis thaliana (L.) and the potential to translate the ﬁndings to C 3 and C 4 crops. Our data on Zea mays (L.) lines subjected to progressive drought show that there is potential for improvements in WUE of the maize line B73 at the whole plant level (WUE plant ). However, changes in i WUE of B73 and Arabidopsis reduced the assimilation rate relatively more in maize. The trade-off observed in the C 4 crop possibly limits the effectiveness of approaches aimed at improving i WUE but not necessarily efforts to improve WUE plant .


Introduction
Green Revolution technologies and significant expansion in the use of land, water, and other natural resources for agricultural purposes have led to a tripling in agricultural production between 1960 and 2015 [1]. Despite this success, the high costs to the natural environment that accompany elevated productivity and changes in the food supply chain threaten the sustainability of food production [1]. Global food security is further challenged by climate change, with a predicted increase in frequency of droughts [2,3]. Globally, agriculture accounts for at least 70% of withdrawals from freshwater resources, with large effects on ecosystems [4,5]. Despite this high water deployment, major yield losses due to water deficits are experienced in crops [6]. At the same time, global population growth increases the demand for food, feed, and fuel, which intensifies the pressure to improve water use efficiency (WUE) of crops [7,8]. While better crop and water management practices provide an immediate opportunity to increase crop water productivity, breeding for superior varieties can achieve a medium-and long-term increase [9,10].

Disconnecting Improved WUE and Growth Trade-Offs
Being a ratio, iWUE can be improved by reducing g s per amount of CO 2 assimilated or by enhancing the assimilation rate at a given g s . Both cases result in lowered intercellular CO 2 concentration (C i ) and consequently in an increased stomatal CO 2 gradient (C a − C i , with external CO 2 concentration C a ), which is directly proportional to the ratio of A n to g s according to Fick's law applied to carbon assimilation in leaves, A n = g s (C a − C i ) [16,38,39]. Increased iWUE has been observed in several C 3 species under water deficit conditions when plants reduce g s [11,15,[40][41][42], although a decrease in g s caused by drought was found to be overridden by heat stress [43]. However, closing stomatal pores to reduce transpiration often results in a reduction of A n [41,44]. Lowering g s impinges on C i and unless this change in C i is counteracted by an elevated mesophyll conductance Agronomy 2018, 8,194 3 of 13 (g m ), the CO 2 concentration at the site of Rubisco-dependent carboxylation (C c , CO 2 concentration in chloroplasts) will be reduced [39,45,46]. A reduction of C c affects the carboxylation efficiency of Rubisco and favors photorespiration [47,48]. Sustaining net photosynthesis under these conditions might require a higher electron transfer rate (ETR) and/or reduced nonphotochemical quenching to support enhanced carboxylation by Rubisco for compensation of enhanced photorespiration [49][50][51][52]. There are reports that water deficit results in increased g m ; however, in most analyses, no change or a reduced g m was observed under drought [44,[53][54][55][56][57][58].
Gains in WUE are often associated with growth trade-offs [59,60]. As pointed out by Blum [61], crops with high CO 2 assimilation and high biomass accumulation per unit land area require high stomatal conductance. This is supported by the observation of a constant WUE on the field level over a broad range of yields [8]. Nevertheless, there might be ways to achieve elevated WUE and high photosynthesis, namely by exploring CO 2 concentrating mechanisms, increased g m , and increased CO 2 specificity of Rubisco [10].
Interestingly, several reports of C 3 plants have shown enhanced iWUE without the expected negative impact on A n or growth [11,15,[62][63][64], as postulated by plant physiologists [4]. In these studies with transgenic tomato and Arabidopsis plants, g s was moderately reduced by enhancing the biosynthesis or the responsiveness to the phytohormone abscisic acid (ABA) or by reducing the size and density of leaf stomata [15,[63][64][65]. Plants overexpressing distinct ABA receptors-termed ABA-Binding Regulatory Component (RCAR)/Pyrabactin Resistance 1-(like) (PYR1/PYL)-caused increases of 40% in iWUE, integrated WUE based on ∆ 13 C of biomass and cellulose fractions, and WUE plant [15]. Growth rates and biomass accumulation were not significantly different from wild type [15]. Hence, the ABA receptor lines revealed higher water productivity, i.e., WUE per time, both under well-watered growth conditions and under water deficit. Net carbon assimilation was comparable to the wild type, however, at lowered C i levels and without detectable changes in g m . This report and other studies show that improving WUE is possible without growth trade-offs. The underlying physiological mechanisms are largely unknown and might involve the root system, as grafting experiments have suggested [15], and enzymes of the C 4 metabolism, such as PEP carboxylase and its regulatory protein kinase PEPC kinase, which are both upregulated in C 3 plants at low CO 2 availability [65]. C 4 and C 3 plants differ in WUE [66][67][68]. At a given g s , C 4 plants show higher net carbon assimilation rates and higher WUE [66]. The CO 2 concentrating mechanism involving PEP carboxylase results in saturation of C 4 photosynthesis at relatively low C i [69,70]; therefore, lower g s and a steeper CO 2 gradient (C a − C i ) are realized in C 4 plants compared to C 3 plants [66,70]. C 3 plants have C i values in the range of 300 ± 60 µmol mol −1 , while the C i of the C 4 plants is around 150 ± 40 µmol mol −1 [71][72][73] at ambient CO 2 of 370-400 µmol mol −1 in well-watered conditions. Under optimal growth conditions, maize and sorghum with C 4 metabolism therefore have higher yields per water transpired than the C 3 crop wheat [9].

Comparative Analysis of Maize and Arabidopsis
The question arises as to whether it is possible to transfer the finding of improved iWUE without having the negative impact on growth to crops. The data on g s -modified tomato plants suggests that it might work for C 3 plants [63], but the lower C i level of C 4 plants could preclude such an accomplishment in maize.
To explore the relevance of these findings of uncoupling WUE improvement and yield decreases for the C 4 crop maize (Zea mays L.), we analyzed gas exchange data obtained from the maize inbred line B73 and compared them to findings in Arabidopsis. In addition, we analyzed the WUE plant of maize lines subjected to drought. B73 is an inbred line that is commonly used in breeding programs, but is known to be drought-sensitive [74]. B73 was included in a progressive drought stress experiment adapted from Yang et al. [15] in which biomass production with a given amount of water was analyzed and WUE plant was determined. In this experiment, B73 showed the lowest WUE plant (Figure 1a) compared to the maize inbred Mo17 and lines derived from an introgression library described by Gresset et al. [75]. In Figure 1, data are shown of the recurrent parent (RP) of the introgression library as well as two introgression lines differing from RP by reduced (IL-05) or elevated kernel ∆ 13 C (IL-81) [75]. A significantly reduced WUE plant compared with the recurrent parent for IL-05 shows the potential of genetic improvement for this trait. However, the largest difference in WUE plant was observed between B73 and Mo17, with an increase of~27% ( Figure 1a). Our data is in accordance with a previous drought stress experiment conducted on seedlings of maize inbred lines, where Mo17 ranked top in yield per plant [76]. The results indicate genetic variation in the efficiencies of water use among maize lines and a potential for genetic improvement of the WUE plant for B73. parent for IL-05 shows the potential of genetic improvement for this trait. However, the largest difference in WUEplant was observed between B73 and Mo17, with an increase of ~27% ( Figure 1a). Our data is in accordance with a previous drought stress experiment conducted on seedlings of maize inbred lines, where Mo17 ranked top in yield per plant [76]. The results indicate genetic variation in the efficiencies of water use among maize lines and a potential for genetic improvement of the WUEplant for B73. Maize lines showed a difference in water consumption over the five weeks of the experiment (ANOVA, p < 0.001, Figure 1b). However, differences in water consumption cannot explain the differences observed in WUEplant and towards the end of the progressive drought, all genotypes included in the experiment had consumed an equal amount of water (5.8 kg ± 0.02 kg, mean ± SE).  [15]. In the greenhouse, 23 maize genotypes were grown in a randomized complete block design, including the two maize inbred lines Mo17 and B73 and introgression lines described by Gresset et al. [75].  The influence of initial biomass on the biomass at the end of the experiment (28.74 g ± 2.22 g and 25.5 g ± 2.39 g, respectively) was approximately 2%. The plants were transplanted into 10 L pots containing 8 L water-saturated soil (85% v/v soil water content; CL ED73, Einheitserdewerke Patzer, Germany, particle diameter <15 mm). A cover of polyethylene foil was used to prevent evaporation, and the progressive drought experiment was initiated by no further watering. The experiment was conducted in the greenhouse (Gewächshauslaborzentrum Dürnast in Freising, Germany) in Oct-Nov 2017 at full sunlight plus supplemental light at 25-33 • C, 19-20 • C day/night, 400 µmol m −2 s −1 PAR, 40% RH. Soil water content declined progressively during the course of the experiment until the plants used all available water. The water consumed was determined gravimetrically (means ± SE of n ≥ 4 biological replicates). WUE plant was calculated as final aboveground biomass per water consumed (means ± SE of n ≥ 4 biological replicates). The increase in WUE in Mo17 compared to B73 is indicated with an arrow. Student's paired t-tests of the maize lines were adjusted for multiple comparisons with the Bonferroni method and lines, which did not differ significantly (p < 0.01), and are marked with common letters.
Maize lines showed a difference in water consumption over the five weeks of the experiment (ANOVA, p < 0.001, Figure 1b). However, differences in water consumption cannot explain the differences observed in WUE plant and towards the end of the progressive drought, all genotypes included in the experiment had consumed an equal amount of water (5.8 kg ± 0.02 kg, mean ± SE).
The way in which the change in soil water content (SWC) during the progressive drought experiment affected photosynthesis and iWUE was analyzed by gas exchange measurements. The A n of leaves was fairly constant for maize B73 plants exposed to high SWC levels up to 40%, then the A n dropped steadily approaching zero at approximately 20% SWC (Figure 2a). In parallel, g s changed moderately between 70% and 40% SWC and declined to zero at 20% SWC (Figure 2b). The C i values were in the range of 80-100 µmol CO 2 per mol between 40-60% SWC. They were somewhat higher in plants from water-saturated soil and were lowered to a minimum of approximately 40 µmol mol −1 at 25% SWC (Figure 2c). Further reduction of the water content in the soil resulted in the steep rise of C i values, indicating collapsing photosynthesis at very low g s of plants experiencing severe drought stress. As the ambient CO 2 concentration (C a ) surrounding the leaf was maintained at 400 µmol mol −1 , the CO 2 gradient (C a -C i ) at the stomatal pores increased from approximately 250 µmol mol −1 (C i of 150 µmol mol −1 ) at soil water saturation to approximately 360 µmol mol −1 (C i of 40 µmol mol −1 ) at the brink of terminal drought. The SWC also influenced iWUE (Figure 2d). Values increased from well-watered conditions to a maximum at 25% SWC, with a plateau around 170 µmol CO 2 per mol H 2 O between 40-60% SWC. Under mild water deficit between 40-60% SWC, there was little variation in A n , and g s and, consequently, the iWUE values.
The results for maize B73 differed from data gained by similar analyses of Arabidopsis plants (Figure 3a-d; Reference [15]). The A n remained constant between 30-70% SWC, which might be caused by light-limited, but not water-limited, photosynthesis. However, g s and C i steadily decreased with decreasing SWC and, concomitantly, the iWUE increased by twofold from approximately 35 to 70 µmol mol −1 at 30% SWC. The CO 2 gradient at stomata increased more than twofold from approximately 80 µmol mol −1 at soil water saturation to approximately 170 µmol mol −1 at 30%. The data were obtained at light conditions that did not saturate photosynthesis, but analysis at saturating light confirmed the capacity of Arabidopsis to lower C i and maintain photosynthetic rates unchanged [15]. The improvement in iWUE by limiting g s without major trade-offs in A n (Figure 3) was observed for the C 3 plants Arabidopsis [15,24] and tomato [63]. A twofold enhancement in iWUE has been reported in different C 3 species under drought [11,15,[77][78][79]. Besides, considerable differences in WUE in the absence of drought stress have been observed among natural variants [80,81].
plants experiencing severe drought stress. As the ambient CO2 concentration (Ca) surrounding the leaf was maintained at 400 µmol mol −1 , the CO2 gradient (Ca -Ci) at the stomatal pores increased from approximately 250 µmol mol −1 (Ci of 150 µmol mol −1 ) at soil water saturation to approximately 360 µmol mol −1 (Ci of 40 µmol mol −1 ) at the brink of terminal drought. The SWC also influenced iWUE (Figure 2d). Values increased from well-watered conditions to a maximum at 25% SWC, with a plateau around 170 µmol CO2 per mol H2O between 40-60% SWC. Under mild water deficit between 40-60% SWC, there was little variation in An, and gs and, consequently, the iWUE values. Figure 2. Changes in intrinsic water use efficiency of maize exposed to a progressive depletion of soil water content. (a) Net carbon assimilation rate (An), (b) stomatal conductance (gs), (c) intercellular CO2 concentration (Ci), and (d) intrinsic WUE (iWUE; defined as the ratio of An to gs) of B73 plants at different soil water content. Gas exchange measurements using the GFS-3000 gas exchange system (Heinz, Walz GmbH, Effeltrich, Germany) were conducted at a photon flux density of 1000 µmol m −2 Figure 2. Changes in intrinsic water use efficiency of maize exposed to a progressive depletion of soil water content. (a) Net carbon assimilation rate (A n ), (b) stomatal conductance (g s ), (c) intercellular CO 2 concentration (C i ), and (d) intrinsic WUE (iWUE; defined as the ratio of A n to g s ) of B73 plants at different soil water content. Gas exchange measurements using the GFS-3000 gas exchange system (Heinz, Walz GmbH, Effeltrich, Germany) were conducted at a photon flux density of 1000 µmol m −2 s −1 , an external CO 2 (C a ) of 400 µmol mol −1 CO 2 , and vapor pressure deficit (VPD) of 26 Pa kPa −1 ± 2 Pa kPa −1 . The first fully expanded leaf counting from the top of the plants was clamped into an 8 cm 2 cuvette for measurements, and plants were subjected to progressive drought as detailed in Figure 1. Plants were grown in soil (Classic Profi Substrate Einheitserde Werkverband) as described in Reference [15]. The experiment was conducted in a greenhouse in the Department of Botany in Freising, Germany from June to August. The maize plants were exposed to full sunlight, at an average temperature of 27 • C, and an average relative humidity of 55% in the experimental period. (a-d) five biological replicates and each data point represents single measurements with five technical replicates.
at saturating light confirmed the capacity of Arabidopsis to lower Ci and maintain photosynthetic rates unchanged [15]. The improvement in iWUE by limiting gs without major trade-offs in An (Figure 3) was observed for the C3 plants Arabidopsis [15,24] and tomato [63]. A twofold enhancement in iWUE has been reported in different C3 species under drought [11,15,[77][78][79]. Besides, considerable differences in WUE in the absence of drought stress have been observed among natural variants [80,81]. Comparing the results from the analysis of maize and Arabidopsis, the potential for increasing iWUE in maize was more limited relative to Arabidopsis. Between 30-60% SWC-reflecting mainly mild water deficit [15]-Arabidopsis responded to the mounting water deficit by a 70% increase in iWUE, while maize showed an increase of less than 20%. Based on the iWUE values obtained at water-saturated soil, the iWUE increase in Arabidopsis and maize was approximately 100% and 40%, respectively, but water logging might be an issue at these high SWC levels. Between 30% and 60% SWC, the stomatal CO 2 gradient of maize increased from 300 to 350 µmol mol −1 , which corresponded to a 17% increase, whereas in Arabidopsis, the gradient was enhanced by 70%, from 100 to 170 µmol mol −1 .
To sum up, mild drought stress (30-60% SWC) had a minor effect on A n , g s , and iWUE in maize. However, in Arabidopsis, g s and iWUE changed dynamically, while A n was little affected. A reduction in SWC from 35% to 25% led to a rapid decline of A n in maize.
These results are in accordance with data from C 3 and C 4 grass species [77,78] but only partly meet the behavior expected for C 3 and C 4 plants based on a meta study [82] where decreases in g s and A n under mild drought stress were more pronounced in C 3 relative to C 4 species. Comparing the A n and g s curves of maize B73 and Arabidopsis Col-0 (Figure 2a,b and Figure 3a,b), a reduction in stomatal conductance led to an immediate reduction in assimilation rate A n for maize but not for Arabidopsis. However, not all C 3 plants show the same flat A n /g s curve as Arabidopsis [83,84] and therefore the data cannot be translated to C 3 crops in general. A previous study on maize lines found differences in g s without trade-offs in A n at well-watered conditions [85], and the author noted that C i values did not become low enough to limit A n as it might in high VPD conditions [85] or as it was observed here under drought.
Still, the limitations in improving WUE in maize B73 could be unique to this inbred line, and comparable data from other maize lines is needed before implications can be expanded to maize in general. Additionally, results from maize cannot represent C 4 photosynthesis in general because this pathway evolved independently in 19 angiosperm families [67]. However, other C 4 grass species have also shown a slight increase in iWUE with progressive depletion of water followed by a steep decline under very severe drought conditions driven by a pronounced decline in A n [77,78]. The loss of A n in the C 4 grass species under drought could partly be attributed to stomatal limitations, while other limitations dominate, including photoinhibition, limitations of CO 2 fixation due to desiccation, and decreases in g m [73,77].
It has also been shown that for subspecies of Alloteropsis semialata (R.Br.) Hitchc., A n is massively reduced in C 4 subspecies under drought in such a way that C 4 photosynthesis totally loses its advantage over photosynthesis of the C 3 subspecies [73]. This conclusion cannot be drawn from our experiments. However, the observation that C 3 plants become more water use efficient during mild and moderate drought while C 4 plants show more stable WUE [86] is in accordance with our data on iWUE.
The less potent improvement in WUE observed in maize under drought might be attributable to a limitation in increasing the CO 2 gradient (C a − C i ) further. Maize, like other C 4 species, possesses a CO 2 concentrating mechanism utilizing precarboxylation of CO 2 by PEP carboxylase, which results in C i values approximately half compared to C 3 species [87]. Such a mechanism results in advantages in A n and WUE under non-or mild-water-deficit conditions [9,66,69,73,77,78]. However, these advantages cannot be maintained when the drought gets severe, especially for maize and C 4 grass species [66,69,73,77,78,86], which is in agreement with our observation for maize B73 at SWC below 27%. The differences can also translate to the field level, where maize has been found to be more sensitive to drought than wheat (C 3 ), with yield reductions of 39.3% compared to 20.6%, respectively, at approximately 40% water reduction [88]. Maize and sorghum are equally or even more sensitive to water stress than many C 3 plants [69,73].
Our results show a potential to increase WUE in maize. The inbred lines displayed a broad variation in WUE plant under progressive drought, and iWUE-measured under the same conditions for the least efficient line at whole plant level-still showed a moderate increase in iWUE with declining SWC. However, the potential of WUE improvement is limited in this C 4 plant compared to Arabidopsis. This limitation is caused by a very high iWUE and low C i under well-watered conditions, which provides a minor degree of freedom for further lowering the C i . The C 3 plant is more responsive concerning increases in iWUE under mild water deficit compared to maize. Hence, screening C 3 plants for enhanced iWUE in combination with efficient growth is a suitable approach to identify crops with improved WUE plant . This approach is less promising for C 4 plants. Establishing higher C a − C i gradients in C 3 crops at a given soil water potential, e.g., by biotechnical engineering using ABA receptors, has the potential to increase iWUE at the cost of minor reductions in A n . Moderate reductions in A n do not necessarily influence yield. In barley, improvements in iWUE and WUE plant have been associated with trade-offs in carbon assimilation but without deleterious effects on plant growth or seed yield [89].