Changes in atmospheric carbon dioxide concentration ([CO2]) affect plant carbon/water tradeoffs, with implications for drought tolerance. Leaf-level studies often indicate that drought tolerance may increase with rising [CO2], but integrated leaf and xylem responses are not well understood in this respect. In addition, the influence of the low [CO2] of the last glacial period on drought tolerance and xylem properties is not well understood.
We investigated the interactive effects of a broad range of [CO2] and plant water potentials on leaf function, xylem structure and function and the integration of leaf and xylem function in Phaseolus vulgaris.
Elevated [CO2] decreased vessel implosion strength, reduced conduit-specific hydraulic conductance, and compromised leaf-specific xylem hydraulic conductance under moderate drought. By contrast, at glacial [CO2], transpiration was maintained under moderate drought via greater conduit-specific and leaf-specific hydraulic conductance in association with increased vessel implosion strength.
Our study involving the integration of leaf and xylem responses suggests that increasing [CO2] does not improve drought tolerance. We show that, under glacial conditions, changes in leaf and xylem properties could increase drought tolerance, while under future conditions, greater productivity may only occur when higher water use can be accommodated.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Changes in atmospheric carbon dioxide concentration ([CO2]) are known to alter tradeoffs between plant carbon gain and water loss, with implications for drought tolerance and productivity. Studies focused on leaf-level responses to changes in [CO2] often suggest that drought tolerance may improve as [CO2] rises. However, changes in [CO2] can alter a wide variety of traits related to drought tolerance beyond leaf traits. Specifically, the effects of [CO2] and drought on xylem structure and function and the integration of xylem with leaf-level responses is not well understood, hindering our ability to accurately predict whole-plant responses to changing [CO2] and drought. Furthermore, until now, there have been few studies incorporating the effects of low [CO2] that occurred during the last glacial period (20 000 yr ago). In this paper we investigate the effects of a broad continuum of [CO2] and drought on a suite of plant traits, including leaf-level responses, xylem structure and function, as well as patterns of allocation to leaves vs xylem and roots. The results of this work offer unique insights into the mechanisms that drive the fundamental tradeoffs between plant water use and carbon gain over geologic and contemporary timescales.
A simultaneous change in both [CO2] and water availability has occurred since the last glacial period, and rapid changes are expected to continue into the future. During the last glacial period, [CO2] was among the lowest concentrations that occurred during the evolution of vascular plants (as low as 180 ppm; Berner, 2006). With the onset of anthropogenic activities during the current interglacial period, [CO2] has risen from 270 to 390 ppm, and is expected to reach 700–1000 ppm by the end of this century (Meehl et al., 2007). Concurrent changes in soil water availability have also occurred, with many regions experiencing more extreme drought during glacial periods than in modern times (Jansen et al., 2007). Further changes in water availability are also anticipated for the near future, with the length and severity of droughts expected to worsen (Christensen et al., 2007; Sheffield & Wood, 2008). Thus, analysis of the interactive effects of changes in [CO2] and water availability is important for understanding the impact of both past and future climates on plant productivity. Actual plant specimens dating to the last glacial period show strong support for leaf-level carbon limitation, as evidenced by reduced CO2 availability in the leaf intercellular space (Ward et al., 2005; Gerhart & Ward, 2010; Gerhart et al., 2011). In addition, leaf-level studies of modern plants often find that increasing [CO2] from glacial to elevated future concentrations enhances carbon gain for C3 plants, while at the same time allowing for decreased stomatal conductance (Gerhart & Ward, 2010; Vanaja et al., 2011; Franks et al., 2013). Furthermore, the relative increase in photosynthetic rates at elevated [CO2] is often greatest under drought conditions (Franks et al., 2013), such that the negative impact of drought may be ameliorated as [CO2] increases from glacial to future predicted concentrations by improved carbon gain, even when stomatal closure occurs in response to drought.
Studies examining how changes in [CO2] affect water use at the leaf level do not provide a complete assessment of how drought tolerance may change as [CO2] rises, however, mainly because [CO2] affects many other plant components that dictate water use. In fact, studies that have examined whole-plant responses to altered [CO2] often find no positive effect of elevated [CO2] on drought tolerance (Vaz et al., 2012; Perry et al., 2013). One reason for this may be that plants must also tolerate the damaging effects of drought on xylem water transport, but little is known about the interactive effects of [CO2] and water availability on this aspect of plant function. Hydraulic conductance can be increased when plants are grown under elevated vs current (Tognetti et al., 1999, 2001) or current vs glacial [CO2] (Quirk et al., 2013). This may result because lower transpiration rates at elevated [CO2] allow plants to maintain water potential farther from the point where xylem embolism begins. In some cases, though, hydraulic conductance decreases at elevated [CO2] (Tognetti et al., 2005), which may become particularly problematic for plants with poor stomatal control or for those that retain active physiologically functioning during drought. This may be attributable, at least in part, to increased rates of xylem embolism, as suggested by the fact that an increase in xylem conduit size is often observed at elevated compared with current [CO2], resulting in a reduced ratio of conduit wall thickness to diameter (Eguchi et al., 2008; Kostiainen et al., 2009). Furthermore, [CO2] effects on water transport may be strongly linked to concurrent changes in leaf area (Gebauer & BassiriRad, 2011), as an increase in total leaf area under elevated [CO2] could increase total canopy water demand (Ward et al., 1999; Wullschleger et al., 2002).
If canopy water demand changes with increasing [CO2], then greater allocation to water transport may be necessary for plants to maintain similar water supply on a leaf area basis across a broad [CO2] range. At elevated [CO2], increased allocation of carbon to shoots and/or roots could improve leaf-specific hydraulic conductance during drought, and thereby decrease the risk of hydraulic failure (Maseda & Fernández, 2006). Indeed, plants usually allocate fewer resources to above-ground structures as [CO2] increases in both drought and nondrought conditions (Ward et al., 2005; Inauen et al., 2012). Under well-watered conditions, hydraulic conductance on a leaf area basis can be conserved as [CO2] increases from preindustrial to current (Phillips et al., 2011), and from current to elevated concentrations (Maherali & DeLucia, 2000; Eguchi et al., 2008; Phillips et al., 2011). In other cases, however, leaf-specific hydraulic conductance can also be reduced at elevated compared with current [CO2] (Tognetti et al., 2005; Eguchi et al., 2008). Thus, the effects of changing [CO2] on the interaction of leaf and xylem traits may provide a mechanistic explanation as to why, in some cases, the positive effect of increasing [CO2] on leaf-level traits does not translate into higher growth rates (Ward et al., 1999; Ghannoum et al., 2010). Although such whole-plant effects ultimately determine drought performance and survival, data examining the combined effects of [CO2] and drought on the integration of leaf and xylem traits are generally lacking from the literature.
We addressed this knowledge gap by determining the effects of a range of [CO2] and plant water potentials on leaves, xylem and whole-plant function. Specifically, we hypothesized that increasing [CO2] from glacial to future predicted concentrations would reduce the impact of drought on productivity and increase the capacity for xylem water transport to meet leaf water demand. We investigated these hypotheses using the annual Phaseolus vulgaris, which has been previously developed as a model system in [CO2] studies (Sage & Reid, 1992; Cowling & Sage, 1998). Although drought does not typically limit P. vulgaris growth in irrigated cropping systems, varieties of P. vulgaris are commonly cultivated using dryland farming techniques in regions where extended droughts frequently limit their yield (Subbarao et al., 1995). In addition, the effect of changing [CO2] and drought on dryland agriculture is expected to intensify in the coming century, such that a more thorough understanding of the interactive effects of [CO2] and drought will be crucial for determining future management practices in dryland agricultural regions (Jat et al., 2012). We chose a drought-tolerant variety, Bolita, providing a more conservative test of the effects of changing [CO2] on drought tolerance. We grew plants under [CO2] representing glacial to elevated concentrations predicted for the future and drought conditions ranging from none to severe. We measured biomass accumulation and allocation, leaf gas exchange, xylem hydraulic conductance, as well as xylem anatomical features related to embolism vulnerability and water transport capacity. In addition, we applied our data to a water transport model developed by Sperry et al. (1998) that quantifies the risk of runaway embolism during drought.
Materials and Methods
Plant material and growth conditions
Phaseolus vulgaris L. var. Bolita (hereafter P. vulgaris; seeds from Plants of the Southwest, Santa Fe, NM, USA) were grown for 28–31 d in a full factorial design with three [CO2] treatments representing glacial (180 ppm), current (380 ppm), or elevated (700 ppm) conditions in combination with four water regimes, resulting in a range of water potentials representing the full operating range of this species (c. −0.2 to −1.8 MPa, Holste et al., 2006). Each [CO2] treatment was replicated in two different growth chambers (Conviron, BDR16, Winnipeg, MB, Canada), and all reported measurements were made on the day of harvest. Plants in all treatments were grown in 3.8 l pots in a sandy clay loam soil (49% sand, 18% silt, 33% clay) for 29–31 d. Growth chamber temperature and relative humidity (RH) were controlled at 25 : 18°C and 50 : 65% (day : night), respectively. Light was provided from 08:00 to 22:00 h at 1048 (± 66) μmol m−2 s−1. Within each growth chamber, pots containing sown seeds were randomly assigned to one of four water availability regimes ranging from high to low (see Supporting Information, Table S1, for a detailed description of water regimes). For each plant, the degree of drought was determined on the day of harvest as predawn plant water potential (ΨPD) measured using a pressure chamber (PMS Instrument Co., Model #1000; Corvalis, OR, USA). We also measured the midday plant water potential (ΨMD) for use in the water transport model described later.
As P. vulgaris forms a symbiotic relationship with nitrogen fixing bacteria, all soil was autoclaved before planting to eliminate nodulation and ensure equal access to nitrogen at all [CO2]. Nitrogen demand is known to increase at elevated [CO2] in P. vulgaris (Jifon & Wolfe, 2002), but high nutrient concentrations have a negative impact on hydraulic function and reduce transpiration (Ewers et al., 2000). Therefore, we used preliminary experiments to identify a nutrient regime that maximized transpiration for plants grown at the control [CO2] of 380 ppm (see Table S1 for description of nutrient regime).
Growth and allocation
Total plant dry biomass (leaves, stems and roots) on the day of harvest was determined after drying at 70°C for 72 h. Biomass allocation was determined as the ratio of absorbing root area to transpiring leaf area (AR : AL). We measured fresh leaf area using a leaf area meter (LI-3100; Li-Cor Biosciences, Lincoln, NE, USA). We estimated fresh absorbing root area for each [CO2] × water availability combination using a regression coefficient relating fresh root surface area to dry mass (detailed methods and coefficient for each treatment combination are presented in Table S2).
Leaf gas exchange
Gas exchange measurements were made at midday with a LI 6400 (Li-Cor Biosciences) on the day of harvest, using the uppermost leaf that was large enough to fill the cuvette. For all plants in all treatments, this leaf had developed after the start of drought treatments. Cuvette conditions were controlled to be similar to growth conditions (Tblock = 25°C, Tleaf = 26 ± 1.3°C, RHsample = 48 ± 4%, vapor pressure deficit (VPD) = 1.8 ± 0.3 kPa, 1000 μmol m−2 s−1 light intensity, and [CO2] representing experimental treatments). Leaves were allowed to acclimate to cuvette conditions for a minimum of 10 and up to 20 min, or until stomatal conductance, photosynthesis and transpiration were fully stable for at least 1 min. Upon reaching steady state, we made five measurements over a 1 min period and report the average of these five measurements. Whole-plant water use per unit of time was calculated by multiplying transpiration rate measured using the LI 6400 by total leaf area. Instantaneous water-use efficiency (WUE) was calculated from photosynthetic and transpiration rates measured by the LI 6400.
Xylem function and anatomy
Immediately following gas exchange measurements on the day of harvest, we measured total stem hydraulic conductance (Kh) using an Ultra-Low Flow Meter (apparatus described in Tyree et al., 2002). The precise methods used here are described in detail in Medeiros & Pockman (2011), except that in this study we removed each leaf at the base of the petiole, cut the stem underwater c. 1 cm above the root/shoot junction and then measured Kh for the entire above-ground portion of the plant stem. We calculated leaf-specific hydraulic conductance (Kl) by dividing Kh by leaf area measured as described earlier. Measurements of xylem anatomy were made using the same plants measured for Kh. Stem samples were cut from c. 1 cm above the root/shoot junction, embedded in resin (catalog #14300; Electron Microscopy Sciences, Hatfield, PA, USA), sectioned, and then imaged using a digital camera (CoolSNAP ES; Photometrics, Tuscon, AZ, USA) connected to a microscope (Eclipse 80i; Nikon Instruments, Melville, NY, USA). We measured vessel diameter for all vessels in a single cross-section. From this, we calculated the hydraulically weighted diameter (Dh = Σ D5/Σ D4), and the conduit area-specific hydraulic conductance (Kc = Kh/total xylem conduit area), which both provide estimates of total water transport capacity. Based on the method described in Hacke et al. (2001), we also determined vessel implosion strength (t/b), or the ratio of the thickness (t) to the span (b) of the double wall between two adjacent vessels, which is strongly correlated with xylem vulnerability to drought embolism (Hacke et al., 2001). Plants from only one chamber were used in this case for these exceptionally time-intensive measures.
Water transport model
Using the water transport model of Sperry et al. (1998), we estimated the critical transpiration rate at which runaway embolism would begin, and the corresponding critical plant water potential (Ψcrit). A brief description of our implementation is provided here, but complete methods are described in the supporting information for this article (Methods S1). The model uses a standard finite difference approach to solve Darcy's law for the main portion of the hydraulic pathway from the bulk soil to the atmosphere. The model was parameterized for each [CO2] treatment using maximum measured values of transpiration (E), Kl, ΨPD and ΨMD. Measurements of E, ΨPD, ΨMD and AR: AL for individual plants were used for model inputs, resulting in an output critical transpiration rate and Ψcrit for each plant. The safety margin from runaway embolism was calculated as the difference between estimated critical transpiration rate and measured E. These model results provide a quantitative estimate of the relative risk of runaway embolism across [CO2] treatments based on measured leaf and xylem traits, but should be interpreted with some caution as transpiration is typically overestimated in a cuvette because of elimination of the boundary layer (LI-COR, 2012). We performed t-tests and regression analysis to determine the fit of the model. The fitted regression lines are presented here (Fig. 7a,b) and the results of the model fit analysis are presented in the supporting information for this article.
All data were analyzed using SAS (version 9.2; SAS Institute Inc., Cary, NC, USA). We performed two-way ANOVA to test for differences across treatments using PROC GLM with [CO2], ΨPD, and [CO2] × ΨPD as main effects with interactions (growth chamber was treated as a random variable). Because ΨPD was not continuously distributed (Fig. 1), it was necessary to treat it as a categorical variable in all analyses. Plants were assigned to one of four drought categories according to measured ΨPD (Fig. 1): none, mild, moderate or severe drought, with each [CO2] × drought combination including four to 11 plants. To account for unequal sample sizes, type III sums of squares were used to determine significance for all analyses. Overall, growth chamber effects were negligible, exhibiting a significant effect in only a small subset of measured characters (Table 1) and no significant interactions with [CO2] or degree of drought. Therefore, interactions with growth chamber were removed from all analyses to conserve model degrees of freedom. Gas exchange parameters were analyzed using MANOVA as a result of the high degrees of autocorrelation typical of these measures. Planned pairwise comparisons were made using the LSMEANS option. Because t/b and AR : AL are ratios, these data were transformed for analysis using an arcsin square root.
Table 1. F-statistics from ANOVAs testing for the effects of [CO2], drought intensity and growth chamber on Phaseolus vulgaris var. Bolita physiological and anatomical characteristics, including biomass and allocation (AR: AL, ratio of absorbing root area to transpiring leaf area), whole-plant water use and water-use efficiency (WUE), xylem hydraulic conductance (Kh (kg s−1 MPa−1), total xylem hydraulic conductance; Kl (kg s−1 MPa−1 m−2), leaf specific xylem hydraulic conductance; and Kc (kg s−1 MPa−1 mm−2), conduit area specific xylem hydraulic conductance), and xylem anatomy (Dh (μm), hydraulic diameter; t (μm), width of the double wall; t/b, implosion strength)
Drought df = 3
[CO2] df = 2
[CO2] × drought df = 6
Growth chamber df = 1
Bold text indicates a significant treatment effect. *, P <0.05; **, P <0.01; ***, P <0.0001.
For all [CO2] treatments, leaf area was significantly reduced in response to increased drought intensity (Fig. 2a; Table 1) and the response to drought was similar across [CO2] (no significant [CO2] × drought interaction). For plants grown at 180 ppm, leaf area was reduced most dramatically in response to drought and was always significantly lower than for plants grown at 380 and 700 ppm. By contrast, plants grown at 380 ppm had leaf areas similar to, or larger than, those grown at 700 ppm at all drought intensities.
Plants grown at 180 ppm exhibited the lowest root areas across all drought conditions (Fig. 2b), and there was a significant effect of [CO2] on this character (Table 1). For plants grown at the control value of 380 ppm, drought did not significantly affect root area, but for plants grown at 180 and 700 ppm, root area declined in response to greater drought intensity (significant [CO2] × drought interaction).
For plants grown at the control value of 380 ppm, more intense drought resulted in a significant increase in AR : AL (Fig. 2c), while for plants grown at 180 and 700 ppm AR : AL was more similar across all drought intensities (significant [CO2] × drought interaction; Table 1). Under moderate and severe drought, AR : AL did not differ between plants grown at 380 and those grown at 700 ppm. AR : AL was significantly lower for plants grown at 180 ppm than for those grown at 700 ppm, regardless of drought.
Total plant biomass was significantly reduced by the most severe drought treatment (Fig. 2d; Table 1), and the relative patterns across [CO2] treatments were similar (no significant [CO2] × drought interaction). Total plant biomass was always significantly lower for plants grown at 180 ppm than for those grown at 380 and 700 ppm regardless of drought. By contrast, significant differences between plants grown at 700 and 380 ppm were only observed in the absence of drought.
Leaf gas exchange
In the absence of drought, A (net assimilation; Fig. 3a) increased with increasing [CO2], while E (transpiration; Fig. 3b) and gs (stomatal conductance; Fig. 3c) declined. MANOVA indicated a significant effect of [CO2] on gas exchange parameters (F126 = 33.44, df = 6, P <0.0001). For plants grown at 700 ppm, mild drought did not affect A. However A, E and gs were reduced significantly under moderate and severe drought for all plants (drought effect, F153.48 = 41.45, df = 9, P <0.0001). The drop in net photosynthetic rates under drought conditions was least dramatic for plants grown at 180 ppm, compared with those grown at 380 and 700 ppm. Thus, we did not detect a significant difference among the three [CO2] treatments under moderate or severe drought, resulting in a significant [CO2] × drought interaction (F178.68 = 8.27, df = 18, P <0.0001). Similarly, E and gs remained significantly higher for plants grown at 180 ppm than for those grown at 380 and 700 ppm under all but the most severe drought conditions. There was not a significant effect of growth chamber on gas exchange (F63 = 1.86, df = 3, P = 0.1461).
Total water use reduced significantly as drought intensity increased (Fig. 4a; Table 1), and the relative pattern across [CO2] treatments was similar (no significant [CO2] × drought interaction). Across all drought intensities, plants grown at 180 ppm exhibited the highest rates of total water use, resulting in a significant effect of [CO2] on this character, although differences across [CO2] were not significant under the most severe drought. Although, in the absence of drought, total water use was higher for plants grown at 380 ppm than for those grown at 700 ppm, there was not a significant difference among these plants under any degree of drought.
Plants grown at 180 ppm exhibited the lowest WUE of all drought conditions (Fig. 4b), and there was a significant effect of [CO2] on this measure (Table 1). WUE did not increase significantly at higher drought intensities for plants grown at 180 ppm. By contrast, WUE increased under moderate and severe drought for plants grown at 380 and 700 ppm, and this increase was greatest for plants grown at 700 ppm (significant [CO2] × drought interaction).
Xylem hydraulic function
In the absence of drought, Kh was significantly higher for plants grown at 180 and 380 than for those grown at 700 ppm (Fig. 5a). Moderate and severe drought significantly reduced Kh for all [CO2] treatments (significant drought effect, Table 1), although Kh was reduced more dramatically for plants grown at 700 ppm than for those grown at 180 and 380 ppm (significant [CO2] × drought interaction). Similarly, in the absence of drought, Kl was significantly higher for plants grown at 180 and 380 ppm than for those grown at 700 ppm (Fig. 5b). There were also striking differences among the [CO2] treatments in the response of Kl to moderate and severe drought. For plants grown at 180 ppm, Kl increased under moderate and severe drought, while it remained similar for plants grown at 380 ppm, and decreased for plants grown at 700 ppm (significant [CO2] × drought interaction). We also found a significant effect of [CO2] on Kc, with plants grown at 180 and 380 ppm maintaining higher Kc than plants grown at 700 ppm regardless of drought intensity (Fig. 5c). Although there was a trend for Kc to decline with increasing drought for all [CO2], we did not detect a significant effect of drought or [CO2] × drought interaction on this character.
There was no significant effect of drought or significant [CO2] × drought interaction for any of the measured xylem anatomical traits (Table 1). Therefore, for simplicity, only the [CO2] effect is shown in Fig. 6. Increasing [CO2] resulted in a significant increase in mean xylem vessel diameter (Fig. 6a). Vessel density decreased significantly as [CO2] increased from 180 to 700 ppm (Fig. 6b). There was also a significant increase in Dh as [CO2] rose from 180 to 380 ppm, although plants grown at 380 and 700 ppm did not differ in this measure (Fig. 6c). In addition, t/b (Fig. 6d) decreased significantly as [CO2] increased from 180 to 700 ppm. There were no significant differences across [CO2] in xylem cell wall thickness (t; Table 1, mean t: at 180 ppm, 4.24 ± 0.43 μm; at 380 ppm, 4.45 ± 0.41 μm; and at 700 ppm, 4.26 ± 0.23 μm).
Water transport model
Based on the measurements of AR: AL, plant water potential, gas exchange, and xylem hydraulic conductance made in the current study, the water transport model predicted lower Ψcrit for plants grown at 180 and 380 ppm than for those grown at 700 ppm, regardless of ΨPD (Fig. 7a). At higher ΨPD, the safety margin from runaway embolism was predicted to be lower for plants grown at 180 ppm than for plants grown at 380 and 700 ppm (Fig. 7b). Under severe drought, however, plants grown at 180 ppm were predicted to have the largest safety margin relative to plants grown at 380 and 700 ppm, respectively. By contrast, a larger safety margin was always predicted for plants grown at 380 than for those grown at 700 ppm.
A number of leaf-level studies have now indicated that increasing [CO2] from glacial to future predicted concentrations may improve drought tolerance by enhancing photosynthesis while at the same time allowing for reduced transpiration (Vanaja et al., 2011; Franks et al., 2013). Yet [CO2] can affect a number of additional components relating to drought tolerance, including xylem function and allocation to leaves, roots and/or xylem. Until now, however, the impact of glacial and elevated [CO2] on the coordination of leaf and xylem function has not been adequately addressed. We hypothesized that increasing [CO2] from glacial to future predicted concentrations would reduce the negative effects of drought on productivity and increase the capacity for xylem water transport to meet leaf water demand. We investigated these hypotheses using P. vulgaris grown across a range of [CO2] from glacial to elevated concentrations under drought conditions ranging from none to severe. Surprisingly, we found that the hydraulic function of plants grown at glacial [CO2] was least affected by drought. This was associated with reduced leaf area and changes in xylem structure in response to growth at low [CO2], which together allowed higher transpiration rates under moderate drought. By contrast, plants grown at elevated [CO2] exhibited a reduced ability of xylem water transport capacity to meet leaf water demand and stronger drought-induced limitations on transpiration.
We found evidence that the steeper drought-induced declines in net photosynthesis under increasing [CO2] observed in this study (Fig. 3a) and in others (André & Du Cloux, 1993; Engel et al., 2004) could be driven by negative effects of elevated [CO2] on water transport. In our study, Kh was reduced during drought for plants grown at current [CO2] (Fig. 5a), but simultaneous declines in leaf area (Fig. 2a) resulted in similar Kl as drought progressed (Fig. 5b). By contrast, at elevated [CO2], significantly steeper reductions in Kh combined with similar reductions in leaf area produced a mismatch between water supply and demand under moderate and severe drought (reduced Kl). In conjunction with this, plants grown at elevated [CO2] exhibited no difference in biomass accumulation during drought compared with plants grown in the current [CO2] treatment (Fig. 2d). In addition, we saw that plants grown at glacial [CO2] exhibited a significantly smaller decline in Kh in response to more intense drought compared with plants grown at elevated [CO2], and had the greatest relative reduction in leaf area. Thus, Kl went up during drought for glacial [CO2]-grown plants. Furthermore, this was associated with higher transpiration rates and a smaller decrease in net photosynthetic rates under moderate drought compared with plants grown at current and elevated [CO2] (Fig. 3).
It is worth noting here that simultaneous and opposing effects of [CO2] on leaf area (increased with [CO2]) and Kh (decreased at elevated [CO2]) resulted in an overall effect that was different than would be predicted from measurement of leaf function, allocation to leaf area, or hydraulic function alone. Consequently, the integration of all factors was critical for a clear understanding of the impact of [CO2] on plant drought tolerance. Using this approach we found that the effects of [CO2] under well-watered conditions probably have a negative impact on the response of plants to drought: Kl was significantly and negatively related to [CO2]. Published data for a wide range of plant taxa ranging from trees and crops to lianas suggest that Kl is often reduced at elevated [CO2] under well-watered conditions (Fig. 8a). Interestingly, though, xeric shrubs appear to maintain Kl at elevated [CO2] (Fig. 8b), while fast-growing tree species like Eucalyptus and Quercus sp. show a relative increase in Kl (Fig. 8c) as [CO2] rises. Thus, the ability of plants to maintain coordinated leaf and xylem function under changing [CO2] may vary according to traits such as drought tolerance or life-history strategy. Furthermore, our work extends our understanding of the effects of [CO2] on whole-plant hydraulic function by providing evidence that Kl can become increasingly compromised during drought for plants grown at elevated [CO2]. The commonality of this remains to be seen, however, as other studies concerning the combined effects of drought and [CO2] on Kl are lacking in the current literature.
One possible explanation for the drought responses we observed is that reduced Kl at elevated [CO2] under nondrought conditions caused plants to experience an increase in drought embolism. We found several lines of evidence in support of this. First, we saw that the differences in Kh and Kl between nondrought and drought plants were greatest for plants grown under elevated [CO2]. Secondly, we found that, for a given conduit size, relative investment in cell walls declined significantly as [CO2] increased (lower t/b, Fig. 6d). Across a wide variety of woody plants, reduced t/b is correlated with increased embolism vulnerability (Hacke et al., 2001), although the applicability of this relationship to herbaceous plants has not been experimentally justified. However, our results with glacial [CO2] do extend and support observations that t/b is reduced at elevated [CO2] in both herbaceous (Davey et al., 2004) and tree species (Kilpeläinen et al., 2007). Thirdly, an increased risk of embolism with increasing [CO2] is further suggested by the results of the water transport model, which predicted the highest Ψcrit for plants grown at elevated [CO2] (Fig. 7a). The model also predicted a smaller safety margin from runaway embolism for plants grown at elevated vs current [CO2], regardless of water availability (Fig. 7b). A global data set examining 191 forest species (Choat et al., 2012) indicates that, in comparison with most other angiosperms, P. vulgaris var. Bolita has a similar or lower Ψcrit (at current [CO2], Ψcrit in our study was below −2.5 MPa), suggesting that our results using the drought-tolerant Bolita variety of P. vulgaris are possibly applicable to a wide range of species.
We also found evidence, however, that the negative effect of increasing [CO2] on Kl resulted, at least in part, from a change in saturated water transport capacity. In support of this, we saw that Kc was c. 50% lower for plants grown at elevated [CO2] than for those grown at glacial and current [CO2], regardless of water availability (Fig. 5c). It is possible that Kc was compromised at elevated [CO2] as a result of greater embolism even under nondrought conditions, but the differences in Kc across [CO2] did not increase as drought progressed, so this possibility cannot be discerned under the current study design. We also saw that the significant positive effect of elevated [CO2] on xylem water transport capacity estimated from xylem conduit diameter (Dh; Fig. 6c) was not associated with an increase in either Kh or total canopy water use at elevated [CO2] (Fig. 4a). Taken together, these results raise the intriguing possibility that the hydraulic and subsequent drought effects of [CO2] we observed were driven not only by changes in embolism rate, but also by changes in xylem anatomical traits related to saturated water transport capacity, such as vessel length. Recent evidence for substantial within-species variation in vessel length across a broad taxonomic range of woody plants (Jacobsen et al., 2012) suggests that this character may display plasticity in response to environmental conditions, but the impact of [CO2] on vessel length has not been previously determined. Although our study cannot provide direct support for an impact of [CO2] on vessel length, our whole-plant approach did allow us to define more narrowly the impact of [CO2] on plant hydraulic function and drought tolerance, and highlights the need for more studies of this nature.
Despite the negative effect of increasing [CO2] on hydraulic function, we observed several characteristics of plants grown at current and elevated [CO2] that could increase their drought tolerance compared with those grown at glacial [CO2]. First, we saw that plants grown at current and elevated [CO2] always accumulated more biomass than glacial [CO2]-grown plants, regardless of drought (Fig. 2d), which could improve drought survival and recovery (Ward et al., 1999; Wullschleger et al., 2002). Secondly, we saw that plants grown at glacial [CO2] had the highest total water use in most cases (Fig. 4a) and did not significantly increase WUE in response to drought (Fig. 4b). By contrast, WUE increased significantly during drought for those grown at current [CO2], and even more so for those grown at elevated [CO2]. Still, it is important to note here that our data concerning xylem hydraulic function (Fig. 5), xylem structure (Fig. 6) as well as Ψcrit and safety margin (Fig. 7) suggest that lower WUE in glacial plants was accommodated without increasing the risk of hydraulic failure, an important component of drought tolerance.
We also found some evidence that growth at elevated [CO2] could improve drought tolerance compared with current [CO2], but in this case the benefits did not extend beyond mild drought. First, although total water use was significantly lower for plants grown at elevated than for those grown at current [CO2] in the absence of drought (Fig. 4a), we did not find a significant difference between current and elevated [CO2] under drought conditions. Secondly, we observed, for example, significantly greater root area (Fig. 2b) and higher AR: AL (Fig. 2c) with increasing [CO2] under well-watered and mild drought conditions. This pattern has been observed in a large number of previous studies (Inauen et al., 2012) and could improve access to soil water resources. But, again, under moderate and severe drought, the advantage of elevated vs current [CO2] was diminished. At this level of drought, plants grown under elevated [CO2] had significantly lower root area than those grown under current conditions, and we saw no significant difference between plants grown at current and elevated for a wide variety of plants [CO2] in AR: AL. Taken together, these data indicate that elevated [CO2] may provide some protection from the onset of drought conditions and/or from the negative effects of mild drought, but the opposite may be true under moderate and severe drought.
Our observations ultimately have broad implications for productivity under both future and past [CO2]. Reduced Kl observed in our study (Fig. 5b) and others (Fig. 8a) suggests that elevated [CO2] could compromise productivity under moderate and severe drought by limiting leaf water supply, and thereby limiting stomatal opening. In addition, the results of the water transport model suggest that cessation of growth may occur at higher plant water potentials as [CO2] rises in the future. In addition, our investigation of xylem structure and function indicates that productivity may be reduced at glacial [CO2], not only as a result of low substrate availability and increased photorespiration (Ward, 2005), but also because of the need for relatively greater investment in xylem cell walls (higher t/b). In our study this came at a cost of reduced water transport capacity (Fig. 6c), which in turn can reduce productivity (Pittermann et al., 2006). Although this could lower the risk of xylem embolism, it also increases carbon inputs into construction of xylem (Hacke et al., 2001). Based on this, one might predict that carbon limitation at glacial [CO2] would result in reduced investment in cell walls; nevertheless, optimal carbon-use efficiency may be achieved in some cases when conduits remain functional (Holtta et al., 2011).
In conclusion, our study involving whole-plant responses highlights the fact that a change in [CO2] elicits simultaneous changes in xylem structure and function, allocation to leaves and roots as well as leaf function, and it is the collective impact of these changes that governs plant responses to drought. We provide evidence that growth under low-[CO2] conditions could have increased the drought tolerance of some C3 plants during glacial times compared with the present. Our work also provides mechanistic support for studies indicating that water availability will remain an important factor limiting the productivity of C3 plants, even as [CO2] increases (Linares et al., 2009; Vaz et al., 2012; Perry et al., 2013). Furthermore, we show that increased productivity at elevated [CO2] may only occur under conditions where higher water use can be accommodated.
The authors would like to thank J. Sperry for generously providing the hydraulic limits model described in Sperry et al. (1998), as well as T. Leibbrandt, E. Duffy, C. Bone and M. Walker for help with plant care. Also, thanks to authors listed in Fig. 8 for providing data on leaf-specific hydraulic conductance and anonymous reviewers for helpful comments on the manuscript. Funding for this research was provided by an NIH IRACDA postdoctoral training grant to J.S.M. and an NSF CAREER award to J.K.W.