•The response of nocturnal stomatal conductance (gs,n) to rising atmospheric CO2 concentration ([CO2]) is currently unknown, and may differ from responses of daytime stomatal conductance (gs,d). Because night-time water fluxes can have a significant impact on landscape water budgets, an understanding of the effects of [CO2] and temperature on gs,n is crucial for predicting water fluxes under future climates.
•Here, we examined the effects of [CO2] (280, 400 and 640 μmol mol−1), temperature (ambient and ambient + 4°C) and drought on gs,n, and gs,d in Eucalyptus sideroxylon saplings.
•gs,n was substantially higher than zero, averaging 34% of gs,d. Before the onset of drought, gs,n increased by 85% when [CO2] increased from 280 to 640 μmol mol−1, averaged across both temperature treatments. gs,n declined with drought, but an increase in [CO2] slowed this decline. Consequently, the soil water potential at which gs,n was zero (Ψ0) was significantly more negative in elevated [CO2] and temperature treatments. gs,d showed inconsistent responses to [CO2] and temperature.
•gs,n may be higher in future climates, potentially increasing nocturnal water loss and susceptibility to drought, but cannot be predicted easily from gs,d. Therefore, predictive models using stomatal conductance must account for both gs,n and gs,d when estimating ecosystem water fluxes.
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.
Another key factor that may regulate gs,n is [CO2]; yet, to our knowledge, the effect of rising [CO2] on gs,n has not been quantified. In a recent study, Zeppel et al. (2011) found that the nocturnal sap flow of well-watered Eucalyptus saligna increased under elevated [CO2], in contrast with the reductions seen in daytime sap flow dynamics with rising [CO2]. These whole-tree sap flow results suggest that leaf-level gs,n and gs,d may exhibit different responses to rising [CO2]. However, the nocturnal sap flow response to elevated [CO2] occurred during the night-time stem recharge period, and therefore may have reflected changes in stem recharge, rather than transpirational fluxes to the atmosphere.
In this study, the effect of rising [CO2] on gs,n was quantified. Eucalyptus sideroxylon trees were grown at three levels of [CO2] (280 μmol mol−1, pre-industrial; 400 μmol mol−1, ambient; and 640 μmol mol−1, elevated). We included a pre-industrial [CO2] treatment to provide a baseline for the assessment of the effects of rising [CO2] since the beginning of the industrial age, which may also help us to predict future responses to [CO2] (Sage & Coleman, 2001; Gill et al., 2002; Gerhart & Ward, 2010; Lewis et al., 2010), and allows for an extended response surface to more adequately understand correlations among functional traits. We tested whether gs,n decreased in response to rising [CO2], as might be expected from the effects of rising [CO2] on gs,d, or whether gs,n increased, as suggested by the study of nocturnal sap flow in Zeppel et al. (2011). In addition to comparing responses across a [CO2] gradient, this study also investigated the potential interactive effects of temperature (ambient and ambient + 4°C) and [CO2] on gs,n over the course of a protracted drought. Across our temperature treatments, relative humidity was approximately equal, meaning that D was higher at higher temperature; therefore, the elevated temperature treatment represented a combined elevated temperature and elevated D environment when compared with the ambient conditions. Based on a study showing that gs,n increased with D (Dawson et al., 2007), we anticipated that gs,n would increase with temperature. We also anticipated that gs,n would decrease with reduced soil water content, as has been shown in numerous other studies (Bucci et al., 2004; Barbour & Buckley, 2007; Cavender-Bares et al., 2007; Dawson et al., 2007; Howard & Donovan, 2007; Zeppel et al., 2010). It is not known whether these responses depend on growth [CO2].
Our main objectives were therefore to determine the following: whether gs,n would decrease in response to rising [CO2], as might be expected from the effects of [CO2] on gs,d; whether gs,n would increase with growth temperature, and an associated increase in D; whether gs,n would decrease as soil water content decreased; and whether rising [CO2] would modify these responses to temperature, D and soil water content. Overall, we sought to identify how nocturnal water use was affected by growth [CO2] to improve our understanding of the complex effects of climate change on vegetation water use.
Materials and Methods
Soil was collected from the A horizon (top 50 cm) of the Hawkesbury Forest Experiment site, University of Western Sydney, Richmond, NSW, Australia (Barton et al., 2010). The soil is a loamy-sand with low organic matter content, low water-holding capacity and low fertility (Ghannoum et al., 2010a). The soil was air dried and c. 95 kg was added to each of 36 pots (volume, 75 l). Six pots were placed in each of six adjacent, naturally lit glasshouse compartments. Temperature and [CO2] conditions in the glasshouse compartments were maintained as described previously by Ghannoum et al. (2010a). In summary, three glasshouse compartments were set to simulate ambient temperature, defined as the temperature of a 30-yr average local (Richmond, NSW, Australia) day for the months of November to May (hereafter ambient temperature treatment). Three glasshouse compartments were maintained at ambient + 4°C (i.e. elevated temperature treatment), which is within the range of temperatures predicted with rising [CO2] in the next century (IPCC, 2007). Over the course of every 24 h, temperatures were changed five times to simulate natural temperature variation. Maximum temperatures during the middle of the day and middle of the night for the ambient and elevated temperature treatments were 26 : 18 and 30 : 22°C (day : night), respectively. The vapour pressure deficit (D) varied between 0.4 and 2.5 kPa in the ambient temperature treatment and between 1.0 and 3.8 kPa in the elevated temperature treatment. Within each temperature treatment, compartments were automatically regulated to maintain pre-industrial (280 μmol mol−1, target), ambient (400 μmol mol−1, target) and elevated (640 μmol mol−1, target) [CO2]. Actual average daytime [CO2] values during the study period for the pre-industrial, ambient and elevated treatments were 290, 400 and 650 μmol mol−1, respectively.
Temperature and relative humidity were measured in 5-s intervals using Vaisala thermistors, one per glasshouse room (Vaisala Inc., Boston, MA, USA), and logged every 15 min on a Campbell CR 1000 data logger (Campbell Scientific Inc., Logan, UT, USA). Relative humidity was not regulated and averaged 57% over the growing season. The vapour pressure deficit (D) was estimated from the temperature and relative humidity (Pearcy et al., 1989). Soil moisture was measured using one Time Domain Reflectometry (TDR) probe to a depth of 0.30 m within each of the 36 pots. Data were measured every 5 min and logged hourly on a Campbell CR 800 data logger (Campbell Scientific Inc.). The volumetric water content was converted to the soil water potential (Ψs) using a previously determined calibration curve (Phillips et al., 2010).
Six E. sideroxylon (A. Cunn. ex Woolls) trees (one per 75-l pot) were grown in each of the six glasshouse compartments. Seeds of E. sideroxylon were obtained from Ensis (Australian Tree Seed Centre, ACT, Australia) and sown (mid-September 2009) in seedling tubes filled with seed raising mix (Plugger Custom; Debco Pty Ltd, Berkshire Park, NSW, Australia). Two months later, seedlings were transplanted into the 75-l pots and grown for an additional 7 months in the temperature and [CO2] treatments. The drought treatments commenced at this point. Within each [CO2]–temperature treatment, three pots were randomly assigned to each of the well-watered and drought treatments. Pots in the well-watered treatment were watered to field capacity (c. 2–3 l) every 3 d, and pots in the drought treatment received no water after the start of the drought treatment. Pots contained small drainage holes to prevent excessive soil waterlogging.
Tree size measurements
Each tree was destructively harvested at the completion of the experiment, 60 d after the onset of the drought treatment, when the trees were 11 months old. Total leaf area was quantified using a leaf area meter (LI-3100A; Li-Cor, Lincoln, NE, USA).
A Decagon porometer (Decagon Devices, Pullman, WA, USA), which minimally disturbs the leaf boundary layer and therefore more accurately reflects ambient atmospheric conditions, was used to measure nocturnal (gs,n) and daytime (gs,d) stomatal conductance under ambient growth conditions. Nocturnal measurements were conducted weekly during the drought treatment. Daytime measurements were conducted at the start of the drought period. Measurements on two to five mature, fully expanded leaves per tree were made during the daytime between 11:00 and 13:00 h. Nocturnal measurements were made between 19:30 and 22:00 h (sunset occurred at approximately 17:30 h during the study period). For all measurements, the porometer was equilibrated to room temperature for 2 h before initiating the measurements of gs,n and gs,d.
All statistical analyses, except partial correlations, were conducted using R versions 2.11.1 and 2.13.0 (R Development Core Team, 2010). In all tests, [CO2] was treated as a continuous variable and temperature was treated as a categorical variable. Results were considered to be significant if P ≤ 0.05. Normal probability plots and plots of residuals vs predicted values were used to assess whether data violated the assumptions of normality and homogeneity of variances. These assumptions were met for all variables.
Two-way analysis of variance (ANOVA) was conducted on all data (well-watered and drought plants) on the first day to examine the effects of [CO2] and temperature on gs,n before the onset of drought. To test whether variation in D could account for effects of [CO2] and temperature on gs,n, a partial correlation analysis (SPSS v16.0 for Windows) was conducted on all plants (six plants per treatment for six treatments) whilst holding D constant. A partial correlation analysis explains the measure of variance in the dependent variable that is explained by an independent variable (predictor), over and above the effects of other independent variables in the model (Murray & Hose, 2005). The use of this technique allowed an examination of the unique relationships between gs,n, [CO2] and temperature, whilst holding constant potentially confounding effects of D.
To test the interactive effects of [CO2] and temperature on the response of gs,n to D, we conducted a homogeneity of slopes test using a linear model on gs,n and D data collected from well-watered trees during 10 measurement periods that spanned the drought period. Further, a homogeneity of slopes test was conducted on data from drought plants to test whether the slopes of the Ψs vs gs,n relationship were different among [CO2] and temperature treatments. Finally, for each drought plant, the Ψs at which gs,n reached zero (Ψ0) was estimated as Ψ0 = gs,nmax/m, where m is the slope of the relationship between gs,n and Ψs, and gs,nmax is the value of gs,n when Ψs is zero. Subsequently, we used an ANOVA to test the effects of [CO2] and temperature on Ψ0.
Environmental conditions –D and soil water potential
Mean D increased with growth temperature, both during the day and at night. There were some differences in D among [CO2] treatments. At ambient temperature, mean (± SE) daytime (10:00–16:00 h) D values across the study period were 1.4 (± 0.01), 1.4 (± 0.01) and 1.3 (± 0.01) kPa in the 280, 400 and 640 μmol mol−1 treatments, respectively, whereas the mean nocturnal (18:00–06:00 h) D value was 0.5 kPa (± 0.01) in all of these treatments. At elevated temperature, mean daytime (10:00–16:00 h) D values were 1.9, 2.4 and 2.3 kPa in the 280, 400 and 640 μmol mol−1 treatments, respectively, whereas the mean nocturnal (18:00–06:00 h) D values were 0.9, 1.0 and 1.1 kPa in these treatments, respectively (± 0.01 kPa in all treatments). Statistical analyses of our data were designed to correct for these small differences in D among treatments.
The soil water potential (Ψs) remained high in well-watered plants throughout the experiment (Fig. 1). In drought plants, Ψs declined during the dry-down period, as expected. Rising [CO2] increased the rate of depletion of soil water, such that the most rapid decline occurred in the elevated [CO2] treatment. Elevated temperature also increased the rate of depletion of soil water. The effect of rising [CO2] on Ψs was probably a result of a larger tree size (Fig. 2). Leaf area increased linearly with increasing [CO2] (P = 0.07). However, elevated temperature had no effect on leaf area (Fig. 2). Accordingly, the effect of elevated temperature on Ψs was probably a result of increased evapotranspiration associated with increased D.
Effects of [CO2] and temperature on gs,n and gs,d before drought onset
At the beginning of the experiment, when all treatments were well watered, gs,n increased significantly with rising [CO2] in both temperature treatments (Fig. 3a, P = 0.02). Averaged across temperature treatments, gs,n increased by 85% as [CO2] increased from 280 to 640 μmol mol−1. The partial correlation analysis showed that there was a [CO2] effect, even after taking into account differences in D among chambers (P = 0.014). Under ambient temperature, there was a reduction in gs,d with rising [CO2], as has commonly been found. Under elevated temperature, there was no clear trend (Fig. 3b). Thus, the response of gs,n to rising [CO2] did not follow a similar pattern to that of gs,d within the same environmental context at the onset of this experiment. The mean value of gs,n as a percentage of gs,d across all treatments was 34% (range, 16–50%).
Effects of [CO2] and temperature on the relationship between gs,n and D
Overall, gs,n increased with rising D (P < 0.001; Fig. 4). There was a significant effect of growth temperature on the relationship between gs,n and D (P = 0.01; Table 1). At ambient growth temperatures, the response to rising D was relatively flat, whereas gs,n tended to increase more strongly with rising D at elevated temperature (Fig. 4). There was also a significant effect of [CO2] on the gs,n–D relationship (P = 0.04; Table 1). The slope of the gs,n–D relationship increased with rising [CO2] from 280 to 640 μmol mol−1 (Fig. 4). The effect of increasing [CO2] was marginally higher (P = 0.07) in the elevated relative to the ambient temperature treatment. In addition, there were significant effects of [CO2] and temperature (P = 0.01) on gs,n, even after taking into account differences in D (P = 0.015; Table 1). This result indicates that gs,n increased as temperature and [CO2] increased from 280 to 640 μmol mol−1 across the entire study period.
Table 1. Results of a test of homogeneity of slopes to examine the effects of [CO2] and temperature on the relationship between nocturnal stomatal conductance (gs,n) and vapour pressure deficit (D) in well-watered Eucalyptus sideroxylon plants across the entire sampling period
SS, sum of squares.
Effects of [CO2] and temperature on the relationship between gs,n and Ψs
Rising [CO2] modified the response of gs,n to decreasing Ψs. As the drought progressed, gs,n declined in the drought treatment (Fig. 5). However, there was a significant interaction between [CO2] and Ψs on gs,n (P = 0.02; Table 2). As soil dried, gs,n declined more slowly with lower Ψs in rising [CO2] (Fig. 5). The relative sensitivity of gs,n to elevated [CO2] increased as soil became drier, such that in dry soils, gs,n was highest under elevated [CO2] and lowest under pre-industrial [CO2].
Table 2. Results of a test of homogeneity of slopes to examine interactions among [CO2], temperature and soil water potential (Ψs) on nocturnal stomatal conductance (gs,n) of drought-treated Eucalyptus sideroxylon plants across the drought period
There was an interactive effect of [CO2] and temperature treatment on Ψ0 (the soil water potential at which gs,n reached zero). Ψ0 increased significantly with rising [CO2] in the elevated temperature treatment (P < 0.05), indicating that complete stomatal closure occurred at progressively more negative Ψs with rising [CO2], and in the elevated temperature treatment (Fig. 6).
We had four main objectives. We tested whether gs,n decreased in response to rising [CO2]; we found that an increase in [CO2] from 280 to 640 μmol mol−1increased gs,n by 85% in well-watered plants, when averaged across temperature treatments. We asked whether gs,n would increase with growth temperature, and an associated increase in D; we found that gs,n increased with rising D in the ambient + 4°C temperature treatment, and that the increase was steeper than in the ambient temperature treatment; however, this effect varied with rising [CO2]. We tested whether gs,n would decrease as soil water potential decreased; we found that gs,n declined with drought and lower soil water potential. Finally, we asked whether rising [CO2] would modify these responses to temperature, D and soil water content; we found that rising [CO2] and elevated temperature slowed the decline in gs,n with drought, and that the soil water potential at which gs,n was zero was significantly more negative with rising [CO2]. In summary, [CO2] modified the response of gs,n to temperature, D and soil water potential. Therefore, we conclude that there were synergistic effects of rising [CO2] and temperature on gs,n and on the responsiveness of gs,n to soil water potential.
The effect of temperature treatment on gs,n could partially be accounted for by changes in D, as gs,n increased with increasing D over a wide range of D in elevated temperatures. Interestingly, these results suggest that gs,n is not functionally linked to gs,d as a simple fractional formulation. Specifically, the increases in gs,n with rising [CO2] and D in elevated temperature were opposite to the commonly observed reductions in other studies of gs,d with rising [CO2] and D. However, our results confirmed those of Dawson et al. (2007), who found increasing gs,n with rising D in a field study. These results generally contradict the hypothesis that gs,n follows the patterns of gs,d, suggesting that gs,n is not readily predictable from our current understanding of gs,d responses to [CO2], temperature and drought. Given the growing importance of night-time water loss to plant and landscape water budgets, our lack of a mechanistic understanding constrains our ability to fully consider the consequences of a changing environment.
Increasing [CO2] increased gs,n
Rising [CO2] significantly increased gs,n at the leaf scale. The 85% increase in gs,n from pre-industrial to elevated [CO2], averaged across temperature treatments, was in contrast with the response of gs,d, which did not show a consistent pattern of response to rising [CO2] or temperature, and to previous studies on gs,d, which generally found a reduction in gs,d with rising [CO2] (Medlyn et al., 2001; Lewis et al., 2002a; Ainsworth & Long, 2005; Ghannoum et al., 2010b). However, the increase in gs,n with rising [CO2] in wet soils was consistent with a recent whole-tree study of nocturnal sap flux density, which found that, when soil was wet, nocturnal sap flux in E. saligna was higher under elevated [CO2] than under ambient [CO2] (Zeppel et al., 2011). What causes these changes in gs,n in E. sideroxylon with rising [CO2] and temperature conditions?
Several factors, including treatment differences in D (Dawson et al., 2007), higher water fluxes in immature leaves (Phillips et al., 2010) and nutrient transport (Caird et al., 2007a), have been proposed to account for changes in nocturnal fluxes. In our experiment, when D was held constant, [CO2] still had a significant influence on gs,n, ruling out D as a potential confounding factor. Differences in gs,n among treatments may also arise if leaves of different ages are sampled among treatments. Rising [CO2] may alter leaf phenology (Lewis et al., 2002b; Warren et al., 2011), and it is possible that the proportion of newly flushed leaves may increase with rising [CO2]. However, we selected mature, fully expanded leaves for all of our measurements, so that differences in leaf age cannot account for such findings. Increased nutrient demand with rising [CO2], as a result of an increased leaf area of trees, could potentially explain the increase in gs,n. We did not measure leaf nutrients for this study, and so we cannot discount this mechanism. Results on the effects of nutrient demand on nocturnal fluxes to date are inconclusive (Howard & Donovan, 2007, 2010; Scholz et al., 2007).
Alternatively, perhaps the increase in [CO2] generated more favourable leaf and stomatal water status at night. Eucalypts grown in higher [CO2] may exhibit higher branch conductance (Atwell et al., 2007), thereby potentially allowing faster refilling of roots, stems and branches. Subsequently, this may lead to more turgid guard cells and less water-stressed leaves at night, thereby facilitating higher stomatal conductance. However, a meta-analysis has shown that plant hydraulic efficiency (capacity to supply water per unit of leaf area) tends to decrease under elevated [CO2] (Mencuccini, 2003). Clearly, more detailed studies are required to examine the potential roles of nutrient transport (Howard & Donovan, 2007, 2010; Scholz et al., 2007), branch conductance and removal of embolisms in the regulation of nocturnal water fluxes (Dawson et al., 2007).
Increasing temperature and D increased gs,n
Although it has been observed in field-grown trees that gs,n increases with rising D (Dawson et al., 2007), these data may be confounded by concomitant increases in temperature with rising D. For example, when night temperature was held constant (c. 18°C) and D was allowed to vary through the night, gs,n in Ricinuscommunis declined in both well-watered and drought conditions with increasing D (Barbour & Buckley, 2007). In addition, in a study on six tree species –Pinus ponderosa, Pinus radiata, Dacrydium cupressinum, Weinmannia racemosa, Quintiniaacutifolia and Quercus rubra– in the field under varying night-time temperature conditions, significant gs,n was observed, but there was no significant relationship between gs,n and D (Barbour et al., 2005). In our study, gs,n generally increased with rising D. At ambient temperatures, night-time D ranged from 0.2 to 0.8 kPa and the relationship with gs,n was relatively flat. At elevated temperature, night-time D ranged from 0.6 to 1.6 kPa. gs,n was increased relative to the ambient temperature treatment, but also increased significantly with D. Thus, there were increases in gs,n with rising D when D was confounded with higher temperature, but also when temperature was held constant and differences in D were driven by the variation in absolute humidity from night to night.
The effects of temperature and D on gs,n in E. sideroxylon contrast starkly with studies on the response of gs,d to D, which generally show that gs,d decreases at higher D and temperatures (Berryman et al., 1994; Cunningham, 2004; Eamus et al., 2008). Importantly, different responses of gs,n and gs,d to D further demonstrate that gs,n responses to climatic variables cannot be predicted from gs,d responses. Although there have been many experiments on the impact of temperature on stomatal conductance, few have examined the effect of changes in D which typically occur with changing temperature (Lewis et al., 2002a; Way & Oren, 2010). Separating the effects of D from the effects of temperature is critical because stomatal processes are strongly influenced by D as well as by temperature. Further, because future climate projections predict both rising temperatures and D, in conjunction with increasing frequency and severity of drought in many regions (Allen et al., 2010), it is essential to account for D when studying the interactive effects of these variables on nocturnal water fluxes.
Rising [CO2] influenced the response of gs,n to Ψs
Rising temperature and [CO2] both increased gs,n, such that the highest values of gs,n were observed under elevated [CO2] and elevated temperature conditions. These results suggest that the combined effect of elevated [CO2] and elevated temperature may lead to increased water loss at the leaf level as a result of increased gs,n, and may partially account for the more rapid soil dry-down observed in the elevated [CO2] and elevated temperature treatment. The rapid dry-down is also partly explained by the higher leaf area in elevated [CO2] trees, which is commonly observed (Ainsworth & Long, 2005). Higher leaf area under elevated [CO2], combined with higher water loss at night under elevated [CO2] and temperature, suggests that E. sideroxylon may experience greater water stress in future climates. Decreasing gs,d with rising [CO2] has been suggested to lead to ‘water savings’ in the soil, and reduced drought stress (Wullschleger et al., 2002). However, this effect has not always been observed in trees (Schäfer et al., 2002; Wullschleger et al., 2002; Duursma et al., 2011; Warren et al., 2011). The effects of reduced gs,d were offset by a larger tree size and higher water fluxes at night under elevated [CO2]. Therefore, our results clearly indicate that reduced gs,d did not lead to sufficient ‘water savings’ to reduce drought stress. In fact, tree size had a greater influence than stomatal closure on soil water content under elevated [CO2], as plants with the lowest gs,d were also those that experienced the most rapid soil dry-down.
Drought has been observed to reduce both gs,n (Barbour & Buckley, 2007; Cavender-Bares et al., 2007) and nocturnal sap flow (Dawson et al., 2007; Zeppel et al., 2010). Consistent with these patterns, soil dry-down was associated with a decrease in gs,n in the present study. However, the response of gs,n to drying soil differed among [CO2] treatments. As soils dried, gs,n declined more slowly under elevated [CO2] relative to pre-industrial and ambient [CO2]. Further, the soil water potential at which gs,n was zero (Ψ0) was significantly more negative in elevated [CO2] and elevated temperature. Accordingly, across soil moisture levels, gs,n was higher under elevated temperature and elevated [CO2] rather than pre-industrial [CO2]. One key implication of these results is that stomata may remain open at night under drier soils in elevated [CO2] in conditions similar to the present experiment. This potential increase in nocturnal water loss may lead to reduced hydraulic redistribution (Howard et al., 2009), which, in conjunction with the higher leaf area under elevated [CO2], in turn may lead to greater susceptibility to drought stress and mortality.
It is important to obtain an idea of the magnitude of the [CO2] effect on gs,n on the whole-plant water balance (Meinzer et al., 2010). To achieve this, we carried out a simple estimation of [CO2] effects on plant water fluxes using measured values of gs and D during the daytime and night-time. We estimated E as gs × D. We assumed that D averaged 2.5 and 1 kPa during the day and night, respectively, that gs at ambient [CO2] averaged 125 and 30 mmol m−2 s−1 during the day and night, respectively, and that gs,d at elevated [CO2] was 20% lower than in ambient [CO2], averaging 100 mmol m−2 s−1. If both day and night gs decrease by 20% at elevated [CO2], then, all else being equal, we would expect total 24-h E to decrease by 20% at elevated [CO2]. If, however, daytime gs decreases by 20%, but night-time gs increases by 85%, as observed in this study, total 24-h E would still decrease, but the reduction will be 12%. Thus, our results suggest that elevated [CO2] could still lead to water savings, but these savings are reduced when rising gs,n is included (12% compared with 20%).
These estimates, however, assume that the leaf area remains unchanged. If the leaf area were to increase with rising [CO2], such that whole-plant daytime E remained unchanged, total 24-h E would increase by c. 9%. These ‘back-of-the-envelope’ calculations provide an idea of the importance of these changes in night-time gs for the overall water budget. Increased leaf gs,n, in conjunction with increased leaf area under elevated [CO2], may cause increases in whole-tree water stress, despite reductions in gs,d. These findings have implications for the water availability of forests, which, in conjunction with rising [CO2], may also experience extreme heatwaves and more severe and prolonged droughts under future climates (De Boeck et al., 2011).
Nocturnal stomatal conductance (gs,n) was substantial in our study, averaging 34% of daytime stomatal conductance (gs,d). Under high soil water content, rising [CO2] (from 280 to 640 μmol mol−1) increased gs,n by 85% when averaged across temperature treatments. During drought, gs,n declined more slowly in elevated [CO2] and elevated temperature. Further, the soil water potential at which gs,n was zero (Ψ0) was significantly lower (more negative) in elevated [CO2] and elevated temperature. As a result, gs,n was highest in elevated [CO2] and elevated temperature across the entire drought period. Taken together, these results indicate that there are synergistic effects of rising [CO2] and temperature on gs,n and on the responsiveness of gs,n to soil water potential. Critically, gs,n may be higher in future climates, potentially increasing nocturnal water loss and increasing the susceptibility of plants to drought, but this cannot be predicted from gs,d. Consequently, predictive models utilizing stomatal conductance must account for both gs,n and gs,d when estimating ecosystem water fluxes.
We thank three anonymous reviewers for insightful comments, Kaushal Tewari for technical assistance, and Markus Lowe and Honglang Duan for assistance with data collection. This research was funded by an Australian Research Council Discovery grant DP0879531 (D.T.T.), a University of Western Sydney International Science Research Schemes Initiative (71846; J.D.L.), Fordham University (J.D.L.), the Philecology Foundation of Fort Worth, Texas through a gift to the University of Arizona (T.E.H.) and a Macquarie University Research Fellowship (M.J.B.Z.).