Growth and physiological responses of isohydric and anisohydric poplars to drought

Highlight Isohydric poplars have high water-use efficiency, while anisohydric poplars show faster growth under a variable water supply, with implications for performance of the different genotypes for woody biomass production.


Introduction
Society's dependence on fossil fuels contributes to greenhouse gas emissions and environmental pollution, leading to a demand for renewable energy sources. Woody biomass production (Kenney et al., 1990), and trees are expected to account for 377 million dry tons of the 1.37 billion dry tons total biomass necessary for a 30% replacement of US petroleum consumption with biofuels by 2030 (Perlack et al., 2005). Thus, tree growth rate, which underlies dry biomass gain, is a fundamental characteristic that can be used to increase productivity in tree plantations. As hybrid poplars are among the fastest growing temperate trees in the world, they serve as a promising feedstock for biofuels and other value-added products (Sannigrahi et al., 2010). Significant efforts have therefore been invested in poplar research, including genome sequencing (Tuskan et al., 2006;Bolger et al., 2014), in an attempt to produce high-yield cultivars.
Of the environmental factors constraining tree growth rate, water is usually the most critical, and water stress restricts plant growth and yield (Bréda et al., 2006;McDowell et al., 2008). This is at least partly because water loss via transpiration (E) is an inevitable consequence of photosynthesis, via the link between CO 2 diffusion into, and water flux out of, stomata (von Caemmerer and Baker, 2007). Stomatal conductance (g s ) thereby acts as a key control on both tree water loss and carbon gain, while carbon gain is closely linked to biomass production. At the leaf level, the ratio between CO 2 uptake and E (i.e. leaf-level water-use efficiency, WUE l ) is ~3-40 µmol CO 2 mmol H 2 O -1 across different well-watered poplar genotypes (Liang et al., 2006;Soolanayakanahally et al., 2009;Larchevêque et al., 2011), implying a more than 10-fold difference in potential carbon assimilation under a variety of soil water and evaporative demand conditions. These variations in WUE l often result mainly from variations in g s and not differences in net CO 2 assimilation rate (A N ) or photosynthetic capacity (Blum, 2005); therefore, an increase in WUE l usually results in reduced photosynthesis and yield (Flexas et al., 2013). Despite decades of research on stomatal physiology, the complex mechanisms that adjust stomatal aperture and regulate g s are still poorly understood although they are vital for plant function, especially when water supply is limited (Tardieu and Simonneau, 1998). Nevertheless, there seems to be general agreement that stomata sense leaf water potential (Ψ leaf ) so that both g s and leaf hydraulic function decline when Ψ leaf decreases (Brodribb and Cochard, 2009;Domec et al., 2009).
Leaf and plant water transport capacity can be quantified in terms of leaf and whole plant hydraulic conductance (K leaf and K plant , respectively). Plants with high hydraulic conductance can supply water rapidly from their roots to the leaves, maximizing g s , A N , and, ultimately, productivity under wellwatered conditions (Nardini and Salleo, 2000). Indeed, differences in xylem traits, such as vessel diameter and sapwood area, as well as the water potential gradient from the soil/root to the leaf may generate variation in E and g s among tree species, as well as within a species (Kleiner et al., 1992;Vivin et al., 1993;Comas et al., 2002;Wikberg and Ögren, 2004;Cocozza et al., 2010). However, when evaporative demand exceeds the ability to supply water to the transpiration stream, g s declines to protect the plant hydraulic system from cavitation (Zimmermann, 1983;Oren et al., 1999;Sperry, 2000). Consistent with earlier work documenting the coordination of g s with K leaf (Meinzer et al., 1995), recent work has revealed that maximum g s is very sensitive to K leaf (Ewers et al., 2000;Brodribb and Holbrook, 2003;Woodruff et al., 2007;Domec et al., 2009). As K leaf declines, owing to cavitation or regulated changes in mesophyll conductance (Johnson et al., 2009), Ψ leaf will also decline, stomata will close, and yield will be negatively affected (Sack and Holbrook, 2006). Therefore, maintaining the integrity of the root to leaf water continuum while avoiding embolism during transpiration is essential for sustaining photosynthetic gas exchange and growth in plants (Meinzer and McCulloh, 2013).
Depending on their genetically dictated molecular and physiological attributes, plants budget their water in very different ways, along a continuum that ranges from the water-conserving behaviour displayed by isohydric plants to the 'risk-taking' behaviour displayed by anisohydric plants (Tardieu and Simonneau, 1998;Moshelion et al., 2014;Sade and Moshelion, 2014). In isohydric species, stomata conservatively regulate plant water status by controlling the rate of water loss to the atmosphere such that it matches the capacity of the soil-plant hydraulic system to supply water to leaves. In order to decrease the risk of hydraulic dysfunction and leaf dehydration, isohydric plants maintain a constant, or nearly constant, minimum daily Ψ leaf (thus reflecting a narrowing soil-leaf water potential gradient) and relative water content by reducing g s and E under water stress. Anisohydric plants, on the other hand, allow Ψ leaf to decrease under drought conditions relative to a well-watered environment, thus reaching a lower Ψ leaf and relative water content with rising evaporative demand and maintaining the driving force for water flow to leaves (reviewed by Moshelion et al., 2014). Yet, the physiological mechanism for the regulation of isohydric and anisohydric behaviours is not fully understood (Klein, 2014;Martínez-Vilalta et al., 2014).
These different stomatal behaviours have implications for selecting the appropriate tree species or genotype to maximize yield and biomass production for bioenergy. Because isohydric plants are expected to reduce g s as soil water becomes limiting, water loss and growth rates should also be reduced. Consequently, higher WUE l can be expected in isohydric plants than in anisohydric plants as soil dries, implying that isohydric trees would not maximize yield on a plantation, though they may be the most water-use efficient trees for producing biomass. By contrast, under prolonged severe drought, isohydric trees might be expected to survive, whereas anisohydric trees are expected to die, generating no yield at all (reviewed by Moshelion et al., 2014). Although hybrid poplars are generally considered to be relatively isohydric (Tardieu and Simonneau, 1998), they actually vary widely in their stomatal sensitivity to soil and air drying and susceptibility to xylem cavitation (Arango-Velez et al., 2011;Silim et al., 2009).
In this work, three poplar genotypes were assessed to determine how fundamental differences in stomatal behaviour, photosynthesis, leaf hydraulics, and leaf size affect growth, drought tolerance, water-use efficiency, and drought recovery rate with the goal of identifying stomatal strategies to maximize biomass productivity under variable water supplies. The genotypes used were Populus balsamifera L. (BS), a North American boreal species that maintains a constant Ψ leaf under mild water stress (Almeida-Rodriguez et al., 2010), and is considered isohydric; Populus simonii carr. (SI), a fastgrowing Asian species that has unknown stomatal regulation behaviour; and their cross, P. balsamifera x simonii (BSxSI), which has been reported to act in an anisohydric manner (Almeida-Rodriguez et al., 2010). Previous studies reported physiological differences in A N , g s , and WUE l , and molecular differences (e.g. aquaporin expression) between BS and BSxSI exposed to a range of water stress (Soolanayakanahally et al., 2009;Almeida-Rodriguez et al., 2010), despite their close genetic relationship. It was hypothesized that isohydric poplars would produce less biomass but have greater WUE l , and therefore would have greater survival and recovery, than anisohydric plants in response to prolonged water deprivation. By contrast, anisohydric poplars would accumulate more biomass under no water stress, as well as mild to moderate drought stress, owing to the maintenance of high g s and A N under low Ψ leaf , but would have low survival rates during prolonged drought. It was also hypothesized that the reduction of Ψ leaf in anisohydric plants would result from the coordinated maintenance of high K leaf and high g s as water stress progressed.

Methods
Complementary experiments were conducted in two locations, Duke University and the Hebrew University of Jerusalem, using a single set of cuttings from dormant stems of the three poplar genotypes (supplied by Agriculture and Agri-Food Canada, Ottawa, ON, Canada) that were split between the two locations.
Duke University, NC, USA Stem cuttings were put into 3.9 L pots filled with Fafard 52 mix potting soil (www.sungro.com; Agawam, MA, USA) and grown in the Duke University Phytotron greenhouses for ~4 months to root and establish leaves. Plants were then moved to fully controlled conditions [25/20°C day/night, 18/6 h light/dark, 70% relative humidity (RH), and 700 µmol photons m -2 s -1 ] in growth chambers (Environmental Growth Chambers, Chagrin Falls, OH, USA), and were watered as needed to maintain a moist growing medium and fertilized with half-strength Hoagland's solution once per week. Gravimetric soil water content (SWC g ) was calculated as the ratio of the mass of water in the soil sample to the mass of dry soil.
The growth rate was measured on well-watered trees grown in 0.3 L pots maintained in these growth chambers. Cuttings with one lateral bud were grown for 62-65 days (until they were 30-40 cm tall) and then cut at the soil surface and weighed for fresh weight. The mass of each shoot was normalized to its stem diameter (mean of three measurements taken at the soil surface level with digital callipers). Leaf size and total leaf area per plant were measured on the same trees using a leaf area meter (Li-3100; LI-COR, Lincoln, NE, USA). Stomatal density measurements were made on a subset of leaves using a rapid imprinting technique (Geisler and Sack, 2002), which allowed the reliable scoring of hundreds of stomata at the same time. In brief, light-bodied vinyl polysiloxane dental resin (Heraeus-Kulzer, http://heraeus-dental.com) was attached to the abaxial and adaxial leaf sides and then removed as soon as it had dried (1 min). The resin epidermal imprints were covered with transparent nail polish, which was removed once it had dried. The nail-polish imprints were put on microscope slides and photographed under a bright-field inverted microscope (Zeiss Axio Imager, http:// www.zeiss.com) with a QImaging MicroPublisher 5.0 MP colour camera (http://www.qimaging.com). Stomatal images were analysed using IMAGEJ (http://rsb.info.nih.gov/ij).
Photosynthetic capacity was assessed on five well-watered individuals per genotype with a portable photosynthesis system (Li-6400; LI-COR). Responses of A N to changes in intercellular CO 2 concentrations were made at a leaf temperature of 25°C, a vapor pressure deficit of ~1.6 kPa, and saturating light (1500 µmol m -2 s -1 photosynthetic photon flux density); ambient cuvette CO 2 concentrations were lowered stepwise from 400 to 50 µmol mol -1 , returned to 400 µmol mol -1 , and subsequently raised stepwise to 2000 µmol mol -1 . Both maximum Rubisco carboxylation rates (V cmax ) and maximum electron transport rates (J max ) values were calculated according to Farquhar et al. (1980), using Rubisco kinetic parameters from von Caemmerer et al. (1994).
When the plants were ~1 m tall, they were moved to a semi-controlled greenhouse (18/6 h light/dark, 50-60% RH, 25°C and natural irradiance). Measurements took place from June to August 2013. The plants were exposed to progressive reductions in SWC g encompassing three categories of water stress based on the lowest SWC g measured in most plants (~30% SWC g ) and on SWC g at field capacity (>70% SWC g ): 70-100% SWC g , 50-69% SWC g , and 30-49% SWC g . Note that the values in the high SWC g class were always >70% but remained mostly <85%. The SWC g was continuously measured with Theta Probes (model ML2x; Delta-T Devices Ltd, Cambridge, UK) connected to a CR10 data logger (Campbell Scientific, Inc., Logan, UT, USA). These SWC g values were then used to calculate Ψ soil based on a 'dynamic' water retention function, obtained by pairing the values of water content and a water pressure head placed in the pot at a given time (Klute, 2003).
Unless mentioned otherwise, all measurements during the drydown experiment were taken from randomly selected individuals from each genotype. The drought treatments at Duke lasted 5 days maximum for each plant, and involved 42 BS plants, 121 BSxSI plants, and 43 SI plants. For a selected plant, point measurements of g s , E, and A N were made with a portable photosynthesis system (Li-6400; LI-COR) between 09:00 and 12:00 hours, with the cuvette set to growth conditions in the greenhouse. Immediately after, Ψ leaf was assessed with a Scholander pressure chamber (PMS Instruments, Albany, OR, USA) and the stem water potential (Ψ stem ) was estimated using the bagged-sealed leaf technique (Begg and Turner, 1970). These values were used to estimate K leaf and K plant as described in Equations 1 and 2: Hebrew University of Jerusalem, Rehovot, Israel A similar set of cuttings from the same three poplar genotypes were grown in Israel to determine whole plant water use and responses to drought. Whole plant daily transpiration (DT) and growth were determined using an array of lysimeters placed in a greenhouse at the Faculty of Agriculture, Rehovot, Israel, as described in detail previously (Sade et al., 2009). Briefly, cuttings were planted in 3.9 L pots filled with potting soil and grown under semi-controlled conditions of 30/25°C day/night under natural day length and light in Rehovot, Israel, from April to May 2014, with ample water supply. Each pot was placed on a temperature-compensated load cell with a digital output (Vishay Tedea-Huntleigh, Netanya, Israel), and was sealed to prevent evaporation from the soil surface. The weight output of the load cells was monitored every 15 s and the average readings over 3 min periods were logged (Campbell Scientific CR1000 , and BSxSI (n = 311). Different letters above the columns indicate significant differences between the three poplar genotypes within an SWC g bin according to Tukey's HSD test, P < 0.05. Asterisks indicate significant differences in comparisons within a genotype to well-irrigated controls using Dunnett's method, P < 0.05.
Data Logger) for further analysis. DT was assessed as the difference in mass between 04:30 and 18:00 hours. A drainage hole at the base of the lysimeters maintained a constant water level following irrigation events, enabling the calculation of the weight gained by the plant (ΔPW k ) between two consecutive irrigation events as described in Equation 3: where W k and W k+1 are the total weight of the pot placed on the lysimeter on two consecutive days (days k and k+1). Therefore, the plant weight gain over the entire experiment period was the sum of the daily plant weight gains from the first to the last day. Agricultural water use efficiency (WUE a ) was calculated as the cumulative weight gain over cumulative transpiration during measurements days as describe in Equation 4: The plants were watered daily until the onset of drought treatment, where no watering was applied until the SWC g in each pot fell below 30%. Because the trees growing in Israel were smaller than those at Duke, the drought varied between 10 and 20 days in duration (depending on plant size); the number of individuals from each genotype varied based on initial availability, and survival during growth and drought (four from BS, 20 from BSxSI, 15 from SI). Once SWC g fell below 30%, watering was reinitiated. Recovery patterns from severe drought stress (SWC g <30%) were determined as the proportion of DT after the return of irrigation to each plant compared to maximal DT prior to drought treatment.

Statistical analyses
Shapiro-Wilk tests for normal distribution of the data were made prior to Tukey's honestly significant difference (HSD) tests used for comparisons of means; comparison to controls were made using the Dunnett's method. Both tests were considered to be significantly different at P < 0.05; all statistics were analysed with JMP 10 Pro (SAS Institute Inc., Cary, NC, USA).

Results
Under well-watered conditions (70-100% SWC g ), Ψ leaf of the SI was significantly less negative than the BS and BSxSI, but Ψ leaf differences between the three genotypes disappeared at 30-49% SWC g (Fig. 1A). Thus, only the BS presented isohydric behaviour, maintaining constant Ψ leaf with declining SWC g (Fig. 2A), and the stem water potential (Ψ stem ) showed the same tendency ( Supplementary Fig. 1A). As a consequence, the water potential difference between the stem and the leaf (∆Ψ leaf ) did not vary between the three genotypes: ∆Ψ leaf remained constant as SWC g decreased, generating a constant driving force (of around 0.3 MPa) for water flow from the stem to the leaf (Supplementary Fig. 1B). This behaviour was made possible by the fact that the BS sharply reduced E and g s in response to the declining SWC g , while BSxSI and SI kept higher E and g s as water depletion progressed, and were thus insensitive to the declining Ψ leaf (Figs 1D,E and 2A,B).
K leaf showed a similar pattern to E and g s ; while SI and BSxSI kept K leaf relatively constant as SWC g declined, BS K leaf decreased by ~50% under the same conditions (Fig. 1B). Thus, g s declined in concert with the decrease in K leaf in the isohydric BS genotype, but g s was less tightly correlated with decreases in K leaf in the two anisohydric poplars (Fig. 2C). Only the BSxSI maintained constant K plant as SWC g decreased (Fig. 1C).
There were no significant differences in the photosynthetic capacity between the genotypes, as measured by maximum V cmax J max (means ± SE: V cmax = 125.7 ± 5.2 µmol CO 2 m -2 s -1 ; J max = 167.8 ± 8.6 µmol CO 2 m -2 s -1 ). However, instantaneous measurements of A N in the greenhouse revealed that the SI had a higher A N at 70-100% and the BS had the lowest A N at 30-49% SWC g (Fig. 1F).
The different genotypes varied in leaf size and number (Fig. 3A,B). BS and BSxSI had bigger leaves than SI seedlings (Fig 3B,C), but SI plants had more leaves per plant, such that there was a larger total leaf area in SI plants than in the other two genotypes (Fig 3D,E). Owing to its lower total leaf area and stomatal density (Fig. 3F), BS plants had the lowest stomatal number per plant (i.e. the gas exchange capacity per plant), while SI seedlings had the highest. Under well-watered conditions, the BS genotype also had significantly lower growth rates than the two anisohydric genotypes ( Figure 4D).
To better understand how these morphological and physiological differences contribute to plant growth rates, water-use efficiency, and drought tolerance, whole-plant transpiration, growth rate, and WUE a were measured. The SI and BSxSI had higher cumulative transpiration and weight gain than the BS under well-watered conditions (Fig. 4A,B,D), which did not correspond with a higher leaf-level E and g s (Fig. 1D,E), but could be explained by considering the different canopy morphology and stomatal densities between the genotypes (Fig. 3). However, the isohydric BS gained more biomass for a given amount of water transpired, generating a higher WUE a compared to the anisohydric SI plants (Fig 4C). The recovery patterns from severe water stress (reaching SWC g <30%) showed that the BS and SI fully recovered within 3-4 days of irrigation, while the BSxSI did not recover to their initial transpiration rates even after 11 days of irrigation (Fig. 5).

Discussion
This study demonstrates that three genotypes of poplar, a key woody biomass species, have different strategies to cope with drought stress, with implications for their suitability for biomass production. The contrasting stomatal and leaf hydraulic behaviours between genotypes ranged from a rapidly responding isohydric behaviour (BS), which is hypothesized to increase survival under drought at the cost of low biomass production, to an anisohydric behaviour (SI and BSxSI) that is thought to allow carbon uptake and maintain high growth rates for a longer period during drought, but to expose the plant to greater risk of drought-induced mortality if the drought persists. Overall, the results indicated three strategies for how the closely related biomass genotypes deal with water stress: survival-isohydric, sensitive-anisohydric, and resilience-anisohydric (Fig. 6). By reducing hydraulic and stomatal conductance to maintain a constant Ψ leaf , the isohydric poplars minimized the exposure of their leaves to water stress, ; for (D) BS (n = 5), SI (n = 6), and BSxSI (n = 16). White, BS; grey, SI; black, BSxSI. Different letters above the columns indicate significant differences between the genotypes for each day, according to Tukey's HSD test, P < 0.05. but also decreased their ability to fix carbon for growth as the soil dried. By contrast, the anisohydric plants kept a high A N while Ψ leaf declined, which should enable higher productivity in the anisohydric poplars, but also made them more vulnerable to damage from prolonged drought stress (Figs 2,4,5,6). Given that recent work in 37 hybrid and pure species of poplar has shown that mean water potentials at which 50% of conductivity is lost range between −1.3 and −1.5 MPa, with a large number of hybrids losing 50% of conductivity at water potentials >−1 MPa (Fichot et al., 2015), the Ψ stem values of near −0.9 MPa shown here were likely sufficient to induce significant cavitation.
While the ability of the isohydric BS to maintain high Ψ leaf and Ψ stem in both well-irrigated and drought conditions compared to the anisohydric BSxSI (Fig. 1A, Supplementary  Fig. 1A) has already been documented (Almeida-Rodriguez et al., 2010), the stomatal strategy of the SI has not. Here, the paternal SI demonstrated the opposite water balance regulation strategy to the maternal BS used to generate the crosses. The conservative water balance regulation of BS comes at the cost of slower growth rates (Figs 1F and 4B,D), which should be exacerbated by a fast reduction in E, g s , K leaf , and A N as soil dries, as hypothesized. Yet, the BS plants benefitted from this behaviour through their ability to maintain a high Ψ leaf under dry soils and, if subjected to drought, BS plants recovered faster than the anisohydric BSxSI cross (Fig. 5), and will thus likely have higher survival in dry conditions. By contrast, the anisohydric behaviour of the SI and BSxSI plants enabled them to sustain faster growth rates (Fig. 4B,D) through longer periods of high E and g s as water availability declined ( Supplementary Fig. 1A,B, Fig. 4B,D). This, in turn, enabled longer periods of high A N as SWC g decreased (Fig. 1F), making these poplars more suited for high biomass productivity. In fact, this anisohydric behaviour was suggested to be an agronomic trait, because anisohydric plants may outperform isohydric plants in terms of growth and yield (McDowell et al., 2008;Sade et al., 2009;Sade et al., 2012). As hypothesized, the reduction in Ψ leaf of the SI and BSxSI plants was possible by maintaining high K leaf (Fig. 1B). Nevertheless, the risk of keeping high hydraulic conductance as well as high g s during deteriorating SWC g might be hydraulic failure. Thus, under short-term stress conditions, the cost of anisohydry should be slow drought recovery (as seen in the BSxSI), and if the stress is prolonged, possible tree mortality.
Interestingly, despite the fact that the SI and BSxSI presented very similar anisohydric stomatal regulation (and therefore similar growth patterns), SI plants showed much better recovery from drought compared with the BSxSI plants (Fig. 5). With all else being equal, leaf and wholeplant tolerance to low SWC g should be conferred by the ability to provide transport pathways from major veins to the sub-stomatal cavity, which is generally associated with small leaf size (McKown et al., 2010;Scoffoni et al., 2011), and big leaved plants are less adapted to dry habitats (Gibson, 1998;Ackerly, 2004). Therefore, the greater drought recovery ability of the SI may be partly due to its smaller leaves, although other parameters, such as its ability to maintain high K leaf while reducing its K plant during drought, might also serve as embolism defence mechanisms (Domec and Johnson, 2012). The fact that BSxSI poplars maintained high K leaf and K plant (but suffered from slow water-stress recovery), while both K leaf and K plant were reduced in BS poplars (Fig. 1B,C), supports this hypothesis. In addition, the ability of the SI to maintain relatively low E but relatively high A N for longer periods under drought, together with the larger SI leaf area per plant, provides additional advantages for growth under drought.
Days since drought-breaking irrigaƟon RelaƟve recovery (%) 0% 25% 50% 75% 100% 125% 150% 1 2 3 4 5 6 7 8 9 10 11 BS SI BSxSI Fig. 5. Recovery rate (relative to pre-treatment transpiration level) from water deprivation of ~30% SWC g . Data are shown as means ± SE: BS (n = 4), SI (n = 15), and the BSxSI (n = 19). Arrows (with colours matching the symbols for the respective genotypes) indicate when post-stress transpiration rates were not significantly different from the pre-treatment values according to a Student t-test, P < 0.05. Poplars are long-lived trees characterized by a dioecious breeding system, wind dispersal of pollen and seeds, clonality, and often continental-scale distribution; as a result, poplars potentially comprise interbreeding populations of immense size. These life history traits typify a plant expected to exhibit abundant genetic variation (Breen et al., 2009), and hybrid poplars are a promising feedstock for multiple industrial applications (Sannigrahi et al., 2010) owing to their large germplasm and their status as the model species for tree genomics (Tuskan et al., 2006). In the long developmental process of breeding new tree genotypes, understanding the physiological responses of the plants to their intended growth environment is a crucial step to developing appropriate cultivars for commercial plantations. Under high soil moisture, anisohydric poplars (such as the SI and BSxSI) had a clear advantage because of their faster growth and higher photosynthetic rates, which may facilitate higher biomass production. Yet increasing demand for food and energy, combined with rising pressure for land conversion, may affect productivity (Harvey and Pilgrim, 2011) as biomass crops are grown on increasingly marginal lands to prevent competition with food production (Murphy et al., 2011;Swinton et al., 2011). Under these conditions, planting the isohydric BS is preferable, because it has high wateruse efficiency and is able to grow and survive under poorer conditions, although its performance is limited in terms of growth.
While the implications of varied stomatal regulation strategies for growth, water-use efficiency, and survival under a variable environment should be tested in the field, the SI's dynamic resilience-anisohydric behaviour patterns (Fig. 6) might be the ultimate strategy for growing under mild to moderate drought conditions, and may provide a suitable role model for future development in woody biomass production. Figure S1: The effect of SWC g on (A) Ψ leaf and (B) the difference between Ψ stem and Ψ leaf of three poplar genotypes. Data is shown as means ± SE from at least 20 independent measuring days and 24 technical repetitions per day. Different letters above the columns indicate significant differences between treatments according to Tukey's HSD test, P < 0.05. Asterisks indicate significant differences within a genotype in comparisons to well-irrigated controls using Dunnett's method, P < 0.05.