Understanding how water-use regulation relates to biomass accumulation is imperative for improving crop production in water-limited environments. Here, we examine how the vulnerability of xylem to water stress-induced cavitation and the coordination between water transport capacity and assimilation (A) influences diurnal water-use efficiency (WUE) and dry-matter production in Lolium perenne L. – a commercial forage grass. Plants were exposed to a range of water stresses, causing up to 90% leaf death, by withholding water and then rewatering to observe the recovery process. Leaf hydraulic conductance (Kleaf) declined to 50% of maximum at a leaf water potential (ψleaf) of −1 MPa, whereas complete stomatal closure occurred well after this point, at −2.35 MPa, providing no protection against hydraulic dysfunction. Instantaneous A remained maximal until >70% of hydraulic conductivity had been lost. Post-stress rewatering showed that 95% loss of Kleaf could be incurred before the recovery of gas exchange exceeded 1 d, with a rapid transition to leaf death after this point. Plants exposed to sustained soil water deficits through restricted nightly watering regimes did not suffer cumulative losses in Kleaf; instead, ψleaf and gas exchange recovered diurnally. The effect was improved WUE during the day and optimal ψleaf during the night for the maintenance of growth.
A fundamental limitation to the ability of plants to survive and grow under water-limited conditions is the capacity of the water transport system to maintain a hydraulic connection between leaves and the soil. As the soil dries, xylem vessels come under increasing physical tension, which when great enough, causes the water column to break and become air-filled, resulting in cavitation and loss of hydraulic conductivity (Tyree & Sperry 1989). Reduced hydraulic conductivity has the potential to impact plant growth rate (Mencuccini 2003; Tyree 2003; Poorter et al. 2010; Russo et al. 2010) because of the close relationship between water transport efficiency and photosynthetic capacity (Hubbard et al. 2001; Brodribb, Feild & Jordan 2007). Furthermore, drought-induced xylem failure can lead to plant death without efficient control of leaf hydration, or the ability to repair dysfunction in the hydraulic system (Sperry 2000; Sperry et al. 2002). Despite the risks to productivity and survival, it is acknowledged that some plants operate close to the minimum leaf water potential at which incipient loss in hydraulic conductivity occurs (Tyree & Sperry 1988). Therefore, water stress-induced hydraulic dysfunction presents a problem not only for long-term survival during severe drought events, but also for daily productivity during short-term fluctuations in soil water availability.
Closure of stomata to regulate the decline in leaf water potential (ψleaf) under water stress is an important mechanism in slowing the loss of hydraulic conductivity. Stomatal closure generally pre-empts significant loss of hydraulic conductivity, and the gap between the ψleaf at 95% stomatal closure and the ψleaf at 50% loss of hydraulic conductivity provides an indication of the hydraulic safety margin (Brodribb & Holbrook 2004; Sack & Holbrook 2006). In species with a wide safety margin, the risk of reaching leaf water potentials low enough to induce xylem failure is minimal because stomata close before hydraulic conductivity substantially declines. However, in other species with a narrow or even negative safety margin, stomata tend to exhibit a threshold response to ψleaf which triggers after the point of incipient loss in hydraulic conductivity (Brodribb et al. 2003).
The importance of stomatal response to desiccation as an indicator of water-stress resilience requires an understanding of the timing and mechanisms associated with hydraulic dysfunction (Zwieniecki & Holbrook 2009). For example, a rapidly reversible decline in hydraulic conductivity has been associated with turgor loss of bundle sheath cells (Brodribb & Holbrook 2006; Kim & Steudle 2007) and xylem collapse (Cochard et al. 2004; Brodribb & Holbrook 2005; Blackman, Brodribb & Jordan 2010), whereas slow recovery of hydraulic conductivity is more likely to be where cavitation of xylem has occurred, requiring positive pressure for embolism repair. In conifers displaying resistant xylem and conservative stomatal regulation (i.e. operate with a large safety margin), the transition from recoverable to non-recoverable water stress has been shown to be rapid once 50% hydraulic loss has been transgressed (Brodribb & Cochard 2009). Comparatively in woody angiosperms displaying a narrow safety margin (Brodribb & Holbrook 2004), it has been recently demonstrated that near-complete loss of leaf hydraulic conductivity can be incurred before plant death ensues (Blackman, Brodribb & Jordan 2009).
The relationship between xylem dysfunction and plant recovery rate still remains to be specifically quantified; however, in the case of cavitation-induced hydraulic dysfunction, it is likely to relate to both physical constraints on the generation of root pressure under soil-saturated conditions and limitations on the rate at which replacement xylem can be grown (Brodribb et al. 2010). In herbaceous plants, root pressure is readily observed alongside observations of diurnal cavitation and refilling of xylem. In both rice (Stiller, Lafitte & Sperry 2003) and sugarcane (Neufeld et al. 1992), for example, over 60% of hydraulic conductivity may be lost during peak transpiration under well-watered conditions but can be restored overnight by root pressure. Diurnal declines and recovery of hydraulic conductivity have also been demonstrated in the petioles (Bucci et al. 2003) and leaves (Brodribb et al. 2003; Johnson et al. 2009) of some woody species. It is, however, unlikely that once cavitation is widespread throughout the plant, woody plants would be able to generate enough root pressure to traverse the long distances required (Fisher et al. 1997).
The comparative analysis of hydraulic efficiency and vulnerability in crop plants remains scant. Furthermore, it has only been in recent years that the direct use of hydraulic traits in breeding efforts for improved water use and production has been considered (Cochard et al. 2008; Sinclair, Zwieniecki & Holbrook 2008). This is in part due to the mixed interpretations of what may be regulating stomatal closure – namely the situation where ψleaf and hydraulic conductivity appear to remain stable, but stomata close due to increases in the hormone abscisic acid (ABA) in the transpirational stream (Comstock 2002). However, from the work undertaken on woody plants, the application of hydraulic information to crop plants seems paramount to the characterization of absolute drought tolerance and water use under soil water deficit conditions (Sperry, Stiller & Hacke 2003; Monclus et al. 2005; Fichot et al. 2009). Knowing why stomata regulate with certain safety margins and the consequences of hydraulic failure will be important questions when it comes to designing irrigation practices with limited water resources and for selecting plant attributes for production in arid environments.
In this study, we investigate how hydraulic conductivity, xylem vulnerability and patterns of diurnal gas exchange influence production and water-use efficiency (WUE) in a commercial pasture grass. From a theoretical perspective, a temperate forage grass was selected specifically to contrast the extensive work on woody plants. Two watering methodologies were employed in order to answer two related sets of questions. The first method involved withholding water from pots to five levels of increasing drought stress followed by rewatering to field capacity to: (1) analyse the relations between hydraulic dysfunction and water-loss regulation, and therefore production potential with declining leaf water potential; (2) quantify the hydraulic safety margin and how this relates to xylem vulnerability; and (3) identify the point at which plants suffer irreversible desiccation damage. Using a different cohort of plants, the second method examined whether plants were able to recover from daily losses in hydraulic conductance when maintained at five narrow soil water levels using a nightly watering regime. The interaction between acclimation, dynamic recovery of hydraulic conductance and gas exchange were examined in the context of growth rate over a ryegrass regrowth cycle.
MATERIALS AND METHODS
Growing conditions and establishment
The experiment was conducted in a controlled glasshouse environment (14 h days at 20/10C day/night), with unfiltered natural light supplemented with sodium vapour lamps to ensure a minimum of 300–500 µmol quanta m−2 s−1 at the leaf surface throughout the day period. Relative humidity was maintained between 30 and 50% by a dehumidifier coupled to a humidity probe. Seeds of Lolium perenne L., cv. ‘Impact’ were germinated in seedling trays in a combination of potting mix and pea gravel, before being transplanted to 3.7 L pots, packed to an average density of 1.04 g cm3−1 with sandy-loam soil. To transplant, the roots were cut to 3 cm and washed to remove any remaining potting mix, and the leaf material above a standing height of 5 cm was cut. Seedlings with 7–10 tillers were selected to ensure that plants were of a similar size and maturity. Triple superphosphate and potassium phosphate fertilizers at rates of 200 kg P/ha, 45 kg K/ha and 184 kg S/ha were mixed into the sandy-loam soil before sieving and potting occurred to ensure that soil fertility would not limit plant growth during the experiment. Subsequent soil testing showed that at the commencement of treatments, the soil nutrient status was 80.3 mg kg−1 P (Olsen extraction), 272 mg kg−1 K (Colwell extraction) and 42 mg kg−1 S [KCl extraction, with a pH of 6.3 (H2O)]. During the establishment phase, on the completion of the third leaf expanding (one ‘regrowth’ period), plants were cut to 5 cm and urea was applied at a rate of 46 kg N ha−1. Soil water content during the establishment phase was maintained at field capacity, determined to be 43.5% (volumetric) using the Haynes apparatus.
Experimental watering strategies
Two watering strategies were employed, each using a different cohort of plants – (1) dry-down with rewatering from different drought intensities to investigate recoverability and limitations of plant hydraulics on leaf gas exchange; and (2) sustained watering levels by pot weight maintenance to examine daily water use and growth.
Dry-down with recovery
Plants were arranged in a completely randomized design, with three plants maintained at field capacity throughout the duration of the experiment to act as controls, and the remaining 15 plants (three plants per treatment) allowed to dry-down by withholding water to different drought intensities determined according to visual signs of leaf necrosis once a pre-dawn leaf water potential (ψPD) of 1.5 MPa (turgor loss point of non-drought adapted plants) had been reached. Leaf damage was calculated as the percentage length of leaf that was dead, totalled across all leaves per tiller and expressed as an average of five tillers. The resulting five plant stress levels at which plants were rewatered to field capacity were 10, 20, 40, 80 and 90% tiller death.
Corresponding minimum leaf water potentials (ψleaf) and soil relative water contents (RWC%) were also defined at each of the five treatment levels. ψleaf was measured on excised leaves, placed immediately in a humidified bag, using a Scholander pressure bomb (Model 615, PMS Instrument Company, Albany, OR, USA). Soil RWC% was calculated for each individual plant according to the weight of the soil at the time of rewatering (initial weight) and the weight of the soil after being dried at 105C for 24 h (final weight), according to the equation:
Once leaves had reached approximately −3 MPa, the live portion of the leaf was quite minimal and easily damaged during measurement, which is a common problem in ryegrass (Thomas 1987, 1991; Clark, Newton & Barker 1999). As a result, minimum ψleaf of the two driest treatments were estimated according to the relationship between leaf RWC % and 1/ψleaf where leaf turgid weight was calculated from the relationship of leaf dry weight to turgid weight (data not shown).
At the target tiller damage plants were rewatered to field capacity by placing the pot in a shallow tray and left to draw water from the bottom of the pot until field capacity had been attained. During the subsequent recovery phase, pots were weighed every second day and watered from the top to maintain field capacity. During dry-down and recovery, plants were not cut at the usual completion of three leaves fully expanded in order for there to be enough leaf material to perform the measurements. As a result of ryegrass only maintaining 3–4 live leaves, leaf turnover was relatively high, particularly in the controls.
Sustained drought levels
To assess the impact of water stress on WUE and growth potential, three plants per five soil water deficit levels were maintained according to pre-defined pot weights to achieve a range of ψMD and corresponding gas exchange capacities. The average pot weight targets included soil volumetric water contents (VWC %) of 28, 21, 19, 16.5 and 15%, which corresponded to average daily watering applications of 119, 78, 64, 44 and 27 mL day−1. Foil was placed on top of the soil surface and around the base of the tillers, and the pot double-bagged so that water loss could be attributed to transpiration. After 6 pm each night, pots were individually weighed and watered back up to target weight. Water was applied down a 2.5 cm wide plastic pipe that was inserted 2 cm into the soil. This method was used to avoid preferential flow down the side of the pots as well as to ensure that water uptake was only via the roots and not through direct contact with the leaf or tiller bases. The plastic pipe was capped to prevent water loss via soil evaporation. There were three plants in each treatment, arranged in a randomized design. To ensure time for plant adjustment, plants were pre-treated at their prospective target weights for one regrowth before being cut and the measurement period beginning.
Leaf vulnerability (Kleaf)
The response of leaf hydraulic conductivity (Kleaf) to ψleaf was assessed according to the method of Brodribb & Holbrook (2003) where Kleaf is calculated from the kinetics of ψleaf relaxation in leaves rehydrated through the tiller base. Five plants at various initial hydration states were used to construct the curve. Individual tillers were removed roots intact, and allowed to further dry on the lab bench at 25 °C for different periods, before being placed in a sealed plastic bag for approximately 1 h to minimize variation in water potential between leaves. Initial ψleaf was then determined by measuring leaves subtending the youngest fully expanded leaf (YFEL), or where equilibrium ψleaf between leaves could not be obtained in severely desiccated tillers; half of the YFEL was sampled prior to rehydration. The tiller base was then cut under water, taking care not to wet the leaf itself, and allowed to rehydrate between 30 s and 4 min depending on the initial ψleaf. The remaining leaf was then cut at the leaf sheath junction and immediately placed in a humidified bag for measurement of final ψleaf. Kleaf was calculated from the ratio of initial to final ψleaf and the leaf capacitance, and corrected for the viscosity of water at 20 °C.
where ψo = initial water potential (MPa), ψf = final water potential (MPa), t = duration of rehydration and Cleaf = leaf capacitance (mmol m−2 MPa−1).
The relationship between ψleaf and Kleaf was plotted and a sigmoidal regression of the form Kleaf = α / [1 + e(β-γϕleaf)] was fitted to the data to describe the leaf vulnerability. To assess the degree of xylem protection against cavitation afforded by stomatal closure, the hydraulic safety margin between stomatal closure and the depression in Kleaf was calculated. This was defined as the difference between the ψleaf at 50% loss of hydraulic conductivity determined from the vulnerability curve, and the ψleaf where stomatal conductance first reached a minimum and did not appear to substantially decline further with ψleaf, determined during the withholding of water from pots. This definition was preferred to a set stomatal closure point of 95%, due to epidermal conductance potentially being a large portion of conductance measured at the tail-end of the observed stomatal range.
For determination of leaf turgor loss, osmotic adjustment (OA) and leaf capacitance, pressure–volume (PV) curves were constructed for each treatment replicate using three leaves (Tyree & Hammel 1972). Leaves were cut at the leaf-sheath intersection before being bagged and recut at the distal end under water to leave only a small amount of leaf area in direct contact with water. Leaves were sampled at dusk and allowed to rehydrate overnight before PV determination. Leaf weight and water potential were measured periodically during slow desiccation of sample leaves in the laboratory. The turgor loss point was located at the inflection point of the 1/ψleaf versus RWC curve. Using this value in the relationship of ψleaf versus RWC, leaf capacitance was determined by fitting linear regressions either side of the inflection point. The capacitance slopes pre- and post-turgor loss were averaged across the treatments to take into account the use of both well-watered and droughted plants used in the construction of the vulnerability curve. Construction of the vulnerability curve requires leaf capacitance in the calculation of Kleaf, be expressed in absolute terms and normalized by leaf area (Brodribb & Holbrook 2003). This was determined by multiplying the capacitance calculated from the PV curve by the ratios of (leaf dry weight:leaf area) and (saturated mass of water:leaf dry weight), which were averaged across all plants:
where DW is leaf dry weight (g), LA is leaf area (m2), WW is mass of leaf water at 100% RWC (g) and M is molar mass of water (18 g).
Limitations of Kplant on leaf gas exchange
Gas exchange parameters including assimilation (A), stomatal conductance (gs) and transpiration (E) were measured using a portable gas analyser (Li-6400; Licor, Lincoln, NE, USA), with a light intensity of 1000 µmol quanta m−1 s−1 at a target air temperature of 20 °C and vapour pressure deficit (VPD) of 1.5 KPa. Three YFELs were sampled intact, per pot, and the results corrected for leaf area. Measurements were made between 1100 and 1400 h, including ψleaf, which was taken directly after gas analysis. For the determination of ψleaf, two leaves off each plant were removed and placed in a humidified bag for immediate measurement. Kplant was calculated as:
where ψPD = pre-dawn leaf water potential, ψMD = midday leaf water potential and E = transpiration according to leaf gas exchange measurements.
Gas exchange and ψleaf measurements were made almost daily in plants subjected to the one drought cycle, but were restricted to five times during the growth of plants under sustained soil water deficit conditions to limit the amount of leaf material removed during ψleaf measurements so that dry-matter (DM) production could be estimated.
Recovery from water stress was assessed according to the number of days taken for plants to return to 50% of the pre-drought maximum rate of Kplant and A following rewatering (t1/2). The inverse of half-time (t1/2−1) was plotted as a function of the treatment ψleaf minimum (Brodribb & Cochard 2009), where t1/2−1 decreases as the time taken to recover increases – zero indicating plant death. This was compared with the proportion of leaf dieback associated with the treatment levels to provide a measure of the net loss in A capacity compared with the individual leaf potential.
Diurnal patterns in leaf extension, water use and WUE
Leaf extension was monitored over a 6 d period at the emergence of the third leaf on two tagged tillers per plant under the sustained soil water deficit conditions. The length of the elongating leaf was measured from the ligule to the tip with a ruler at 8 am (beginning of light hours) and at 6 pm (when plants were watered), and the rate of extension (cm h−1) during the night (6 pm–8 am) expressed as a percentage of daily growth.
The day before plants were cut for DM yield determination, a diurnal series of whole-plant water loss was performed by weighing pots every hour between 600 and 2100 h, to an accuracy of ±0.01 g (Mettler-Toledo PG5002-S). Plant E was calculated by the loss of weight of each plant between measurements divided by the total leaf area. Leaf area was determined for each plant by taking a mixed subsample of leaves after being cut, and the relationship between DM and leaf area used to calculate leaf area of the whole plant.
Leaf tissue (above 5 cm) was harvested from plants and leaf DM yield/plant determined after drying samples for at least 24 h at 60 °C in a forced draught oven. WUE was calculated according to leaf DM per plant (g)/cumulated water applied (mL). Instantaneous WUE (A/gs) calculated from gas exchange information and averaged across the experimental period per treatment was used as a comparative measure.
Nonlinear and linear regressions were analysed using PROC NLMIXED procedures from the SAS statistical package, version 9.1 (SAS Institute Inc., Cary, NC, USA) and graphed using Sigmaplot (SPSS Inc., Chicago, IL, USA). A log likelihood ratio chi-square test was used to compare nested models, for the purpose of selecting the most parsimonious model. The P-value for model selection was derived from the difference in the −2 log likelihood statistics between two models, assuming chi-square distribution with degrees of freedom given by the difference in the number of parameters between models. Where there was no significant difference, the reduced model was selected. Akaike's information criterion (AIC) (Akaike 1974) and biological reasonableness was used to distinguish between non-nested models.
To determine whether plants under sustained soil moisture conditions had adjusted their water-use patterns, the progression of gs with ψleaf was compared with progressively droughted plants. A sigmoid model that fitted each dataset independently by solving for two curves with the form was compared to the single parameterization of the model applied to the combined dataset using a chi-square test as described earlier. If non-significant, the simpler model was retained. For visualizing the main differences in the shapes of the two response curves, gs data was condensed by pooling across −0.2 MPa increments and standard error bars calculated. The coefficient of determination (R2) was used to indicate the level of variance explained by the model and was expressed as where SSerror = residual sum of squares and SStotal = total sum of squares. Treatment effects of the five different sustained watering levels on DM production were analysed using analysis of variance (anova) of a complete randomized block design using Genstat version 9.1 (Lawes Agricultural Trust, Rothamsted Experimental Station, UK). Unless otherwise stated, a significance level of 5% was used throughout.
Leaf hydraulic vulnerability and relationship to stomatal closure
Hydraulic conductivity of excised leaves decreased in response to declining ψleaf. Below a threshold ψleaf of −1 MPa, Kleaf fell sharply over a narrow range of leaf water potentials before reaching a minimum at −2.35 MPa (Fig. 1). Stomatal closure in droughted plants followed a similar trajectory (Fig. 2) indicating a close link between liquid and vapour conductance across the entire range of functional leaf water potentials. A negative safety margin of 1.35 MPa was observed, indicating that stomata closed after 50% loss of Kleaf. In plants subjected to sustained soil moisture deficits, OA occurred in the leaves, with the turgor loss point increasing as a function of the average ψMD (Fig. 2). Consequently, stomatal closure of plants under sustained water deficits followed a shallower sigmoid curve despite having a similar gs minimum as non-acclimated plants (Fig. 2).
Relationship between plant hydraulic conductivity and assimilation rate
Assimilation rate and plant hydraulic conductivity, estimated from gas exchange and leaf water potentials, were sampled during the dry-down and recovery (after rewatering) of plants spanning a leaf water potential range of −0.2 to −2.4 MPa. Although A and Kplant were strongly correlated, the relationship was non-linear of the form , with A initially insensitive to declining Kplant, then dropping in parallel only once Kplant fell below 5 mmol m−2 s−1 MPa−1 (Fig. 3).
Drought-induced leaf dieback and gas exchange recovery
In plants subjected to water stress, there was a clear threshold of ψPD after which leaves began to become visibly damaged progressively from the leaf tip. This threshold for leaf damage occurred at a ψPD of −1.5 which coincided with ∼80% loss of functionality of both hydraulic conductivity and stomatal opening (Fig. 4). Within a soil water content range of 12.6–9.8%, the amount of leaf dieback increased exponentially from 10 to 90% of the total length of leaves. Although the transition from non-damaged to highly damaged occurred over a very small change in both the water status of the leaves and soil (Fig. 4), this represented an extended period of time, around 18 d, due to the fact that stomata were closed and evaporation from the soil by this stage was minimal.
Upon rewatering, recovery of the remaining live portion of leaves was rapid, requiring only half a day to regain 50% of gas exchange capacity in leaves which had lost leaf turgor, and up to 2 d in plants dehydrated to −3.1 MPa. Although in the latter case we observed hydraulic recovery in the living portions of these plants, leaf loss by this stage was already as great as 60% (Fig. 4 inset). The gradient between hydraulically recoverable and non-recoverable (in terms of existing leaf material) paralleled the steep progression of leaf dieback. Accordingly, 100% leaf damage was predicted to occur at −4 MPa from the fitted curve of leaf dieback, and unrecoverable A and Kplant at a similar average ψleaf of −3.6 MPa based on where t1/2−1 falls to zero (Fig. 4). Overall net recovery of leaf biomass was limited to a greater degree by leaf dieback and therefore the time to grow new leaves, than gas exchange and plant hydraulic conductivity. Complete plant death, i.e. the point where new leaves were unable to be initiated, was not reached in this experiment.
Diurnal water use and DM production
Nightly applications of water ranging from 27 to 119 mLs resulted in average ψMD of −0.5, −0.8, −0.9, −1.1 and −1.9 MPa for each of the soil water content targets of 28, 21, 19, 16.5 and 15%, respectively. This was in contrast to ψPD which was similar across all watering levels at an average value of 0.15 MPa (Fig. 5), despite the large range of water applied to the different treatments (27 to 119 mL day−1). The diurnal course of transpiration showed strong variation between treatments, with drier treatments producing depressed maximum E (Fig. 6). When compared to the well-watered controls, the average diurnal E rates declined by 21, 30, 53 and 71% for each of the soil water content targets of 21, 19, 16.5 and 15%, respectively.
Plants maintained at drier soil water targets were found to change leaf elongation patterns such that a greater proportion of daily growth occurred during the night compared with plants maintained at higher soil water contents (Fig. 7). Cumulatively, the restrictions to gas exchange imposed by the soil moisture deficits were insufficient to significantly (P ≤ 0.05) affect DM production; the largest yield penalty being 30% as observed in the driest treatment (Fig. 8). Furthermore, WUE increased by twofold in the most droughted treatment according to instantaneous gas exchange measures, and threefold based on DM production and total water applied (Fig. 9).
One of the constraints in plant water-stress research has been the lack of a cohesive framework by which to assess how genetic manipulations relate to functional changes in plant water use and survival, along with a quantifiable scale by which to compare levels of plant water stress and hence relative drought tolerance. Due to the fundamental link between water transport capacity and gas phase conductance, plant hydraulics provide a useful framework by which to evaluate the capacity of plants to lose water and the ultimate limitations to carbon gain. Using a hydraulics approach in this study, we identified that ryegrass has the capacity to transport a lot more water than required to maintain maximal A rates. Furthermore, water use could be reduced without permanent hydraulic damage, i.e. leaves remained responsive to water inputs for the majority of the leaf functional water potential range. With knowledge of the hydraulic limits of ryegrass, it was demonstrated that manipulations in diurnal soil water availability can achieve major increases in WUE without affecting biomass production. The implications of the results and the application of the hydraulics framework are discussed with relevance to both natural and agricultural systems.
Water-spending hydraulic strategy
Perennial ryegrass was found to be extremely vulnerable to hydraulic dysfunction under soil water deficit, with 50% loss in hydraulic conductivity occurring at −1 MPa and 95% loss of function at −2.2 MPa (Fig. 1). While it cannot be deduced using the pressure relaxation technique whether cavitation was the cause of declining hydraulic function with leaf water potential, Brodribb & Holbrook (2006) suggest that low levels of conductance, particularly 95% loss of functionality, are most likely the result of embolism formation.
Closure of stomata during drying followed a similar trajectory to the loss in hydraulic conductivity, but importantly, it was found that stomatal closure did not occur until after the loss of 50% Kleaf (Fig. 2). Although a similar degree of sensitivity of xylem to water stress-induced hydraulic dysfunction has been reported for both herbaceous (Neufeld et al. 1992; Maize, Cochard 2002; Rice, Stiller et al. 2003) and woody plants (Vogt 2001; Brodribb & Holbrook 2004; Hukin et al. 2005), most angiosperms tend to produce parallel declines in Kleaf and gs resulting in safety margins close to zero. Ryegrass is unusual in this context, producing a significantly negative safety margin of 1.35 MPa.
This risky stomatal behaviour depicts the extreme scenario of the water-spending strategy described by Jones & Sutherland (1991). Based on hydraulic models of maximum permissible water extraction (Sperry et al. 2002), the stomatal response of ryegrass indicates that this species operates at the extraction capacity of the xylem, therefore maximizing the utilization of xylem investment. This is in contrast to plants with higher resistance to xylem cavitation which tend to produce a concomitant increase in the hydraulic safety margin (Meinzer et al. 2009). In these plants, not only does the relative sensitivity of stomata increase such that plants utilize xylem investment less efficiently, but the costs associated with construction of resistant xylem are also much greater. As such, these intrinsically drought tolerant plants tend to be limited in distribution to arid environments as they are unable to compete in mesic or shaded environments where growth rate and carbon economy are at a premium (Givnish 1988).
The water extraction capacity of ryegrass plants was enhanced through OA in the leaves, with acclimated plants observing higher rates of gs under water stress (ψMD range 1.3–1.7 MPa) compared with plants exposed to only the one drought cycle (Fig. 2). It is interesting to note, however, that OA shifted the leaf turgor loss point beyond the point of 95% loss of Kleaf in three of the driest treatments (Fig. 2 inset). This would appear maladaptive, unless there was a coordinated shift in the vulnerability curve (Awad et al. 2010). The fact that stomatal closure occurred at a similar ψleaf between adapted and non-adapted plants suggests that hydraulic acclimation did not occur. This may help to explain the inconsistent benefits reported of OA in increasing crop yield under water-deficit conditions (Serraj & Sinclair 2002). A large research effort with rice, for example, has failed to produce evidence of a benefit of OA on crop yield (Jongdee, Fukai & Cooper 2002). Considering that rice (which displays similar physiology to ryegrass) already incurs losses of Kleaf during normal diurnal cycles in plants under well-watered conditions (Stiller et al. 2003), there seems little benefit, in terms of gas exchange at least, in further increasing the osmotic concentration in the leaf and stomata beyond the transport capacity of the leaves. In some situations, therefore, reduced OA and a resultant protection from hydraulic failure may be more beneficial where water supply is transient.
Recovery and hydraulic limits
Ryegrass plants rewatered after exposure to water stress conformed to a hydraulic-stomatal limitation model (Brodribb & Cochard 2009), whereby gas exchange was strongly mediated by the recovery of leaf hydraulic conductance (Fig. 3). However, due to the saturating relationship between A and Kplant, the recovery of A to 50% of the pre-drought maximum (t1/2) was more rapid than for Kplant (Fig. 4). While the mechanism of hydraulic repair was not investigated in this study, the fact that ryegrass operates so close to its hydraulic limits suggests an inherent ability to repair dysfunctional xylem on a daily basis. One possible explanation for this ability to rapidly repair xylem after water stress is that dysfunction is caused by cavitation, and repair is achieved by the generation of root pressure to refill emboli at night.
Root pressure is readily observed in the herbaceous crops including grasses (Tyree et al. 1986; Neufeld et al. 1992; Cochard, Ewers & Tyree 1994; Hughes & Brimblecombe 1994; McCully, Huang & Ling 1998; Tang & Boyer 2008), and may have similarly provided for a rapid recovery of ryegrass. This is supported by the observation that ψPD remained close to zero in all plants (Fig. 5). For root pressure to have been active, however, the majority of the root system would have to be hydraulically disconnected from the soil. Otherwise, the majority of the water applied to the soil surface would have been predicted to have redistributed throughout the pot under the passive process of equilibration of water tension in the soil–plant-atmosphere–continuum (Richards & Caldwell 1987), resulting in a much lower ψPD than observed.
Alternatively, cavitation may not be responsible for loss of hydraulic function and instead turgor loss or cell collapse may be driving diurnal declines in hydraulic conductance, both of which do not require positive pressure to refill emboli in order to restore hydraulic conductivity (Brodribb & Holbrook 2005; Blackman et al. 2010). This is quite probable as plants sustained at even at the most severe soil water deficit still maintained ψMD above the TLP. Furthermore, beyond a ψPD of −2.5 MPa, there was a steep transition to leaf death, suggesting that the transition between a readily reversible process and non-recoverable hydraulic dysfunction is very close to the functional limits of the leaves. To a degree, this transition was buffered by leaf dieback which increased exponentially with declining ψPD (Fig. 4). Leaf dieback or plant segmentation theory suggests that the sacrificial death of more disposable plant regions would help improve the hydraulic conductance and water status to remaining foliage, acting as a last resort mechanism to avoid whole-plant mortality and aid long-term survival by protecting vital plant parts such as stems and nodal regions (Cochard, Ridolfi & Dreyer 1996; Davis et al. 2002; Vilagrosa et al. 2003; Lo Gullo et al. 2004). For ryegrass, this is a reasonable proposition because regrowing three live leaves (maximum number that can be supported per tiller) under temperate summer conditions would require only ∼21 d. This compares to woody species described in Brodribb & Cochard (2009) which in some cases required over 100 d to replace leaves damaged during drought.
Understanding the critical nature of hydraulic dysfunction for maximizing water productivity
Daily water use was determined by the nightly watering volume, resulting in a decline in maximum E with decreasing soil moisture availability, which was sustained over a longer period of the day (Fig. 6). There was an expectation that water would be used preferentially during the morning in the drier treatments when ψleaf was optimal, with steady declines subsequently during the day. This was not observed which suggests hysteresis in the ψleaf versus gs relationship during drought. The plant hormone ABA has been mooted as a candidate for such transpirational control (Lovisolo, Hartung & Schubert 2002), and in a recent study by Lovisolo et al. (2008), ABA was suggested to help promote the gradual hydraulic repair of xylem by limiting the rate of stomatal reopening. The water supply limitations during the daylight hours were further evidenced by a linear increase in the proportion of leaf elongation growth conducted during the night with decreasing soil VWC % (Fig. 7). The summation of responses to restrictions in water uptake was therefore a water stress-induced increase in WUE during the daylight hours, with optimal leaf water status overnight when plants were watered, ensuring carbon fixed could be utilized in growth.
The fact that DM accumulation was minimally affected by exposure to ψleaf capable of reducing gs and Kleaf in three of the five treatments (Fig. 8) is a significant finding, as to date, the majority of plant hydraulic enquiry has been based on the idea that prevention of hydraulic dysfunction is the a priori position, and any hydraulic conductivity loss incurred is likely to have deleterious effects on gas exchange and long-term survivability from drought (e.g. Tyree et al. 2003). Such conclusions have been drawn from the extensive research conducted on woody plants where loss of hydraulic conductivity is slow to recover (Blackman et al. 2009; Brodribb & Cochard 2009). The assumption that hydraulic dysfunction needs to be prevented is challenged here by the fact that, firstly, recovery was rapid (Fig. 4), and, secondly, A in ryegrass remained maximal until >70% of Kleaf had been lost (Fig. 3). Saturation of the A versus Kleaf relationship across species has been alluded to (Brodribb & Holbrook 2006, 2007; Brodribb et al. 2007) but few studies have shown this in an individual species (Stiller et al. 2003). Hydraulic redundancy provides a greater buffering capacity of ψleaf and hence gas exchange to VPD (Maherali & DeLucia 2000, 2001; Brodribb et al. 2005), and from this perspective may be considered the equivalent of the stomatal safety margin displayed in drought-tolerant plants. However, saturating A versus Kleaf does incur the loss of more water for a similar carbon gain. In the context of this study, increases in VPD may therefore have resulted in larger differences in DM production obtained under the restricted watering regimes.
In agricultural production systems and natural systems alike, predicting both maximal production under optimal soil moisture conditions and production risks under water-limited conditions are important to both the design of irrigation strategies and the distribution of species across soil moisture gradients. The plant hydraulics approach utilized in this work reflects the basic need to conserve the hydraulic connection between soil and plant in order to maintain gas exchange and prevent desiccation. Ryegrass was identified here as being intrinsically drought-sensitive with xylem highly susceptible to water stress-induced hydraulic dysfunction, along with stomatal behaviour characteristic of a water-spending hydraulic strategy. In this regard, stomata maximized short-term carbon gain by allowing hydraulic conductivity to decline. However, unlike previous observations in woody species, hydraulic dysfunction was neither detrimental to A up to a point, nor restrictive to the recovery of hydraulic conductivity on rewatering. By exploiting these two characteristics, diurnal soil water availability could be manipulated in order to achieve increases in WUE without affecting biomass production. Further characterization of crop plants in terms of their hydraulic traits is recommended if our understanding and prediction of plant water use is to improve, and will be imperative to the achievement of yield maximization in water-limited cropping systems.
The authors would like to gratefully acknowledge the statistical advice provided by Ross Corkrey and glasshouse maintenance by Ian Cummings. This research has been funded by an Australian Research Fellowship (TJB), a Discovery Grant from the Australian Research Council and Dairy Australia Research and Development Scholarship (MHP). Ms Holloway-Phillip's stipend was provided by an Australian Postgraduate Award and additional funds by the Tasmanian ICT Centre, CSIRO.