Reduced stomatal conductance in sweetgum (Liquidambar styraciflua) sustained over long-term CO2 enrichment

Authors

  • Jeffrey D. Herrick,

    Corresponding author
    1. National Risk Management Research Laboratory, US Environmental Protection Agency, Research Triangle Park, NC 27711, USA;
      Author for correspondence: Jeffrey D. Herrick Tel: +1 919 541 7745 Fax: +1 919 541 7885 Email: Herrick.Jeffrey@epa.gov
    Search for more papers by this author
  • Hafiz Maherali,

    1. Department of Botany, University of Guelph, Guelph, ON N1G 2W1, Canada;
    Search for more papers by this author
  • Richard B. Thomas

    1. Department of Biology, West Virginia University, Morgantown, WV 26506, USA
    Search for more papers by this author

Author for correspondence: Jeffrey D. Herrick Tel: +1 919 541 7745 Fax: +1 919 541 7885 Email: Herrick.Jeffrey@epa.gov

Summary

  • • Over 4 yr (1998–2001) we examined the effects of CO2 enrichment on stomatal conductance (gs) of sun and shade leaves of overstory sweetgum (Liquidambar styraciflua) grown at the Duke Forest Free Air Carbon CO2 Enrichment (FACE) experiment.
  • • Gas-exchange measurements were taken in June and September of each year and relationships between water stress and stomatal conductance were examined. Stomatal density was measured in June 2000. Relative stomatal limitation (lg) was calculated from gas-exchange measurements.
  • • We found a 28% reduction of gs in elevated CO2 that was sustained over the study period and was similar in sun and shade leaves. Elevated CO2 reduced lg by 26%. Stomatal density was not affected by CO2 enrichment. Elevated CO2 did not change the sensitivity of gs to soil moisture or vapor pressure deficit.
  • • The data illustrate that decreased gs of sweetgum leaves in CO2 enrichment is consistent over long periods of time and under varying environmental conditions.

Introduction

Numerous studies have shown that elevated atmospheric CO2 reduces stomatal conductance (gs) in many woody plants (Morison, 1987; Drake et al., 1997; Medlyn et al., 2001). At the leaf, whole-plant and ecosystem levels, this decrease could lead to reduced evopotranspiration and result in increased soil moisture and run-off production (Field et al., 1995; Sellers et al., 1996; Bounuoa et al., 1999). Stomatal responses to elevated CO2 are quite variable, with literature reviews indicating average reductions ranging from 11 to 40% (Morison, 1987; Drake et al., 1997; Curtis & Wang, 1998; Medlyn et al., 2001). In some species gs may not change, or may even increase with elevated CO2 (Dixon et al., 1995; Ellsworth, 1999).

Variation in stomatal response to elevated CO2 may be caused by several factors, indicating a need for measurements over extended periods. For example, plants may be less sensitive to CO2 in the short term (< 1 yr) relative to the long term (> 1 yr; Medlyn et al., 2001). Plant and leaf age may also be factors, as stomata of mature trees are typically less responsive to CO2 than are tree seedlings and saplings (Medlyn et al., 2001). In addition, environmental variations in light, temperature, humidity and soil moisture conditions can also interact with elevated CO2 to influence gs (Sage, 1994; Santrucek & Sage, 1996; Curtis & Wang, 1998; Wullschleger et al., 2002; Maherali et al., 2003).

The development of leaf structure and function is tightly coupled to variation in the abiotic environment. For example, leaves within plant canopies are often exposed to different light, humidity and temperature environments, leading to distinct sun and shade morphologies (Givnish, 1988). Sun leaves are typically thicker, and have more nitrogen per unit leaf area and greater photosynthetic capacity than shade leaves. In addition, sun leaves usually have a higher stomatal density than shade leaves. These observations suggest that structural differences in leaf tissue caused by the environment, in addition to temporal environmental variation, may lead to differential responses of gs to elevated CO2.

In this study we report the effects of elevated CO2 on gs of overstory sweetgum trees (Liquidambar styraciflua L.) over 4 yr at the Duke Forest Free Air CO2 Enrichment (FACE) experiment. Sweetgum is the most common deciduous tree species in the overstory in the Duke Forest FACE experiment, and differential light environments at the top and the bottom of the forest canopy cause distinct sun and shade leaf types. The long-term nature of this study and the physiological differences among sun and shade leaves allowed us to examine whether stomatal responses to CO2 could be modified by variation in the light environment, as well as year-to-year changes in atmospheric and soil moisture deficits. We predicted that responses of gs to CO2 enrichment would be greater in sun leaves than in shade leaves because photosynthesis responds to CO2 to a greater extent in sun leaves (Herrick & Thomas, 1999, 2001). We were specifically interested in the relative response of stomatal conductance to elevated CO2 in sun and shade leaves, and how changes in stomatal conductance resulting from growth under elevated CO2 limit photosynthesis.

Materials and Methods

Duke Forest Free-Air CO2 Enrichment (FACE) experiment

The Duke Forest FACE experiment is located in a Pinus taeda L. (loblolly pine) plantation in the Blackwood division of the Duke Forest (35°97′ N 79°09′ W). No management measures have been taken to prevent the growth of other tree species since the current plantation was established in 1983 after a clear-cut in 1979. As a result, the forest is dominated by loblolly pine (1733 stems ha−1), but there are significant numbers of sweetgum (620 stems ha−1) and yellow poplar trees (Liriodendron tulipifera L., 68 stems ha−1) as well as other hardwood species in the canopy and the understory. Sweetgum is common in the south-eastern USA and in piedmont North Carolina and it invades early succession broomsedge (Anthropogon virginicus L.) fields during the course of secondary succession (Oosting, 1942). The experimental forest occurs on a nutrient-poor, clay-rich loam soil that is typical of many upland areas in the south-eastern USA.

Within this forest, six 30 m diameter experimental circular plots (FACE rings) were established in 1995. Three of these FACE rings are replicate CO2 treatments and the remaining three plots are ambient experimental controls. Each FACE ring consists of 32 vertical pipes that extend from the forest floor through the forest canopy. In the elevated treatment FACE rings, these pipes deliver a controlled amount of CO2 throughout the entire forest volume with a target CO2 concentration of ambient plus 200 µl l−1. Three control rings receive the same volume of air to replicate any micrometeorological effects on the forest that occurs during the operation of the FACE facility. Beginning in August 1996, the CO2 treatment has been applied continuously 24 h d−1 except when the air temperature was < 5°C for more than an hour. During the first 5 yr of the experiment (1997–2001) the daytime average CO2 concentration was 572 µl l−1 in the elevated treatment rings and 376 µl l−1 in the ambient control rings. To control for topographic variation (≈ 5 m) and potential gradients in site fertility between rings, the three control and three elevated-CO2 rings were arranged in a complete block design (three pairs, two rings in each block).

Steady-state gas exchange  To determine gs, steady-state gas exchange was measured during 7 d twice a year around June 25 and September 1 1998, 1999, 2000 and 2001. June measurements were made ≈ 68 d after leaf initiation, and September measurements ≈ 50 d before leaves began to senesce. Minimum air temperatures during June and September sample periods ranged from 14 to 19°C and maximum temperatures from 30 to 34°C. A drought occurred from late July to early September 1998. Soil moisture averaged 24.4% in late June 1998 and 15.0% during early September 1998 (Schäfer et al., 2002). During the 1999, 2000 and 2001 sample periods, soil moisture averaged ≈ 21%, except in late June 2001 where soil moisture was nearly 30% (Schäfer et al., 2002; H. Kim, unpublished results). Volumetric soil moisture was calculated from measurement of the soil dieletric constant in the upper 30 cm of the soil profile using modified time-domain reflectometry techniques with waveguides (CS615, Campbell Scientific, Ogden UT, USA). There were four sensors within each ring and the signals were recorded every half hour. Average soil moisture for each measurement period was calculated by averaging soil moisture from all the rings for the week when gas exchange was measured (Schäfer et al., 2002; H. Kim, unpublished results).

Two overstory sweetgum trees (7–11 m high in 1997) were selected in each FACE ring, based on the proximity of trees to areas accessible from static towers and portable hydraulic lifts. All these trees had leaves exposed to full sunlight at the top of the crown and deep shade at the bottom of the crown. The same trees and leaf positions were used by Herrick & Thomas (1999, 2001). Sun leaf irradiance was typically saturating and varied between 1100 and 1400 µmol m−2 s−1 during midday on sunny days. Diffuse irradiance for the shade leaves was ≈ 50 µmol m−2 s−1 punctuated by intermittent sunflecks (Herrick & Thomas, 1999).

Steady-state leaf-gas exchange was measured in situ on one sun leaf and one shade leaf from each tree using an open flow infrared gas analyzer with an attached red–blue LED light source (LI-6400, Li-Cor, Lincoln, NE). Measurements were made with a constant irradiance of 1400 µmol m−2 s−1 photon flux density after the leaves were allowed to equilibrate for at least 10 min in the leaf cuvette. Preliminary trials indicated that stomatal conductance reached steady state within 7 min. Steady-state stomatal conductance (gs) was determined when the coefficient of variation of change in water vapor was < 0.05% as indicated by the LI-6400. Fully expanded leaves at least 2–3 wk old were measured between 1000 and 1500 EST on sunny days to minimize diurnal effects on gas exchange. Leaf temperatures were not controlled during measurements and did not differ between CO2 treatments. Leaf temperatures averaged 31.3 ± 1.2°C in June and 30.6 ± 0.9°C in September. Vapor pressure deficit averaged 1.7 ± 0.1 kPa and did not differ between CO2 treatments or canopy position. Trees in one blocked pair of rings were measured each day so that slight differences in daily weather conditions could be included in the block effect in the ANOVA.

The relative stomatal limitation of photosynthesis (lg) for each sample period was based on the response of net photosynthesis (Anet) to variation in calculated intercellular CO2 (AnetCi curves) measured concurrently with the steady-state gas-exchange measurements. Some of these data were originally taken to assess photosynthetic acclimation (Herrick & Thomas, 2001). AnetCi curves were measured over a range of 10 external CO2 partial pressures (Ca) from ≈ 50–1200 µl l−1. A steady-state measurement at growth CO2 was made first, then Ca was dropped down to 50 µl l−1 and increased in nine steps up to 1200 µl l−1 (Herrick & Thomas, 2001). Measurements were made with a constant saturating irradiance of 1400 µmol m−2 s−1 photon flux density. Preliminary trials indicated that photosynthetic rates reached steady state within 3 min following an increase in Ca. The curves were fitted using nonlinear least squares with the equation:

image(Eqn 1)

where Amax = Anet at CO2 saturation, α = y intercept and Γ = CO2 compensation point. This model provided a good fit to our data (r2 = 0.99) and has been used by Gunderson et al. (1993) for other tree species. Relative stomatal limitation was calculated using the differential method (V) of Jones (1985), which is specifically intended for nonlinear AnetCi curves:

lg = rg/(rg + r*)(Eqn 2)

where rg is the gas-phase resistance to CO2 uptake and r* is the slope of the AnetCi curve. We calculated rg as (Ca − Ci)/Anet at the operating Ci and r* as the first derivative of equation 1 at the operating Ci (Jones, 1985).

Stomatal density  Stomatal density (s) was measured by counting the number of stomata per unit area after measuring gas exchange in June 2000. Two sun and two shade leaves were collected from two trees in each ring. Clear nail polish was used to make impressions of areas on the leaf that were not heavily veined (Ceulemans et al., 1995). Sweetgum leaves are hypostomatous, therefore impressions of only the abaxial side of each leaf were made. Each impression was analyzed at 400 × with a light microscope interfaced with a solid-state TV camera (Model CCD-72-SX; DAGE-MTI Inc., Michigan City, IN) using NIH Image 1.58 (US National Institute of Health). Three fields of view per slide were sampled and averaged for each leaf. In addition we measured epidermal cell density (e, number of epidermal cells mm−2) and stomatal index. Stomatal index was calculated as [s/(e + s)] × 100 (Salisbury, 1927). Guard cells were included in the epidermal cell density.

Data analysis

Gas-exchange parameters (gs, lg, Ci/Ca) were analyzed using a repeated-measures analysis of variance (ANOVAR) model with CO2 treatment, leaf position, month, year and blocked ring pair as main effects (Data Desk, 1997). Measurements from each ring were averaged, and individual rings (n = 3) were considered replicates for the purposes of statistical analysis. Stomatal density, epidermal cell density and stomatal index were analyzed with an ANOVA with CO2 treatment, leaf position and blocked ring pair as main effects (Data Desk, 1997). Post hoc comparisons of parameter means were determined using Bonferroni corrected multiple-comparison tests (Data Desk, 1997). Least-squares linear regression was used to analyze the relationship of gs to soil moisture and leaf-to-air vapor pressure deficit. Analysis of covariance (ANCOVA) was used to determine if the CO2 treatment influenced these relationships. A significant interaction would indicate that the slopes were different between CO2 treatments. If the slopes were the same, then the y intercepts were compared by examining the main effect of CO2 treatment (Sokal & Rolf, 1995). Data were log-transformed, when appropriate, to meet the assumptions of parametric statistics. The probability level was set a priori to 0.1 because of the low number of replicate rings in this study (Ellsworth, 1999; Herrick & Thomas, 2001).

Results

Elevated CO2 reduced steady-state stomatal conductance (gs) by an average of 28% in sun and shade sweetgum leaves during 1998–2001 (P = 0.026). CO2 enrichment decreased gs by 31% in sun leaves and by 25% in shade leaves, but the relative response to CO2 was not affected by leaf position (P = 0.398; Fig. 1) or time of year (June and September, P = 0.166). Steady-state gs was more than twice (122%) as high for sun leaves than for shade leaves (P = 0.005; Fig. 1). Variation in gs was influenced by time of year (P = 0.050) and block (P = 0.001). Stomatal conductance decreased with decreasing soil moisture content (r2 = 0.53, P = 0.040), but the slopes of the relationships did not differ significantly between CO2 treatments or leaf position (Fig. 2). In general, gs decreased with increasing leaf-to-air vapor pressure deficit (VPD) (r2 = 0.49, P = 0.059; Fig. 3), but high VPD data influence this relationship with gs. Nonetheless, elevated CO2 did not affect the slope of the response of gs to VPD in the sun leaves and there was not a strong relationship between gs and VPD in shade leaves grown at elevated CO2 (r2 = 0.32, P = 0.144). Despite decreased gs of sweetgum leaves under CO2 enrichment, relative stomatal limitation of photosynthesis (lg) was lower at elevated than at ambient CO2 (P = 0.029; Fig. 4). The CO2 response was not modified by canopy position (P = 0.265) or time of year (P = 0.375). Relative stomatal limitation of photosynthesis was 24% greater in sun leaves than in shade leaves (P = 0.002). Relative stomatal limitation of photosynthesis also varied significantly between blocks (P = 0.001) and between years (P = 0.006).

Figure 1.

Steady-state stomatal conductance (gs) of sun and shade leaves of overstory Liquidambar styraciflua trees growing at the Duke Forest FACE experiment in ambient (closed symbols) and elevated (open symbols) CO2. Gas exchange was measured around 25 June and 1 September of each year. Measurements were made at saturating irradiance and were restricted to the hours between 10 : 00 and 15 : 00 h on sunny days. Each point is the mean of three rings (± SE) for each CO2 treatment. Soil moisture during the measurement period is added for reference.

Figure 2.

Relationship between steady-state stomatal conductance (gs) from overstory Liquidambar styraciflua trees and soil moisture at the Duke Forest FACE experiment. Sun leaves and shade leaves were grown and measured in ambient (closed symbols, solid line) and elevated (open symbols, dashed line) CO2.

Figure 3.

Relationship between steady-state stomatal conductance (gs) from overstory Liquidambar styraciflua trees and leaf-to-air vapor pressure deficit (VPD) at the Duke Forest FACE experiment. Sun leaves and shade leaves were grown and measured in ambient (closed symbols, solid line) and elevated (open symbols, dashed line) CO2.

Figure 4.

Relative stomatal limitation of photosynthesis (lg; Jones, 1985) calculated from ACi curves measured on sun and shade leaves of overstory Liquidambar styraciflua trees growing at the Duke Forest FACE experiment in ambient (closed symbols) and elevated (open symbols) CO2. Gas exchange was measured around 25 June and 1 September of each year. Measurements were made at saturating irradiance and were restricted to the hours between 10 : 00 and 15 : 00 h on sunny days. Each point is the mean of three rings (± SE) for each CO2 treatment. Soil moisture during the measurement period is added for reference.

Elevated CO2 reduced the Ci/Ca ratio by an average 5% in sun and shade sweetgum leaves over the 4 yr of this study (P = 0.064; Fig. 5). Ci/Ca ratio was not affected by leaf position or time of the year (June and September). However, Ci/Ca ratio was affected by year (P = 0.007) and block (P > 0.001).

Figure 5.

Steady-state Ci/Ca ratio of sun and shade leaves of overstory Liquidambar styraciflua trees growing at the Duke Forest FACE experiment in ambient (closed symbols) and elevated (open symbols) CO2. Gas exchange was measured around 25 June and 1 September of each year. Measurements were made at saturating irradiance and were restricted to the hours between 10 : 00 and 15 : 00 h on sunny days. Each point is the mean of three rings (± SE) for each CO2 treatment. Soil moisture during the measurement period is added for reference.

During the fourth year of CO2 treatment stomatal densities of sun and shade leaves were unaffected by elevated CO2 (Table 1). Sun leaves had 32% higher stomatal densities than shade leaves (P = 0.005). Epidermal cell density was not affected by CO2 enrichment, nor was it different between sun and shade leaves. Stomatal index was not affected by CO2 treatments, but was higher in sun leaves than in shade leaves (P = 0.034).

Table 1.  Stomatal density, epidermal cell density and stomatal index of sun and shade leaves from canopy Liquidambar styraciflua trees grown in a forest ecosystem in ambient or elevated CO2
ParameterSun leavesShade leaves
Elevated CO2Ambient CO2Elevated CO2Ambient CO2
  1. Leaves were collected after gas-exchange measurements in June 2000. Each value is the mean of three rings (± SE) from both CO2 treatments ignoring block effects. Within a measurement period values designated by the same letter are not different at the 0.1 level of significance.

Stomatal density (mm−2)385 ± 39a395 ± 19a288 ± 28b303 ± 21b
Epidermal cell density (mm−2)2795 ± 159a2665 ± 80a2406 ± 140a2513 ± 182a
Stomatal index12.00 ± 0.45a13.12 ± 0.56a10.66 ± 0.32b10.75 ± 0.06b

Discussion

Stomatal conductance (gs) of overstory sweetgum trees exposed to elevated CO2 was reduced by an average of 28% in sun and shade leaves throughout the 4 yr study period. The decrease in gs of sweetgum leaves was sustained over 4 yr of CO2 treatment, indicating that stomatal sensitivity to CO2 did not subside over time. This is consistent with the 21% reduction in gs for 13 long-term field-based studies of woody plants (Medlyn et al., 2001), but greatly different from the dominant species of the Duke FACE experiment, loblolly pine, which shows no effect of elevated CO2 on gs (Ellsworth, 1999). This underscores the importance of species composition when predicting CO2 effects on stand-level processes.

The elevated CO2-induced reduction in gs was similar for both sun (31%) and shade (25%) leaves. Wullschleger et al. (2002) observed a 23% decrease in gs with elevated CO2 for sun leaves of sweetgum trees at the Oak Ridge, TN FACE experiment, but no effect of CO2 on gs for leaves in the middle and lower portions of the canopy. The difference between our observations on shade leaves at the Duke Forest FACE experiment and those at the Oak Ridge FACE experiment may reflect differences in canopy light or temperature regimes. The Oak Ridge FACE experiment is a closed-canopy sweetgum plantation, whereas in our study sweetgum trees are growing among a more open loblolly pine plantation. Our shade leaf gs measurements were taken at the very bottom of the sweetgum canopy where the light regimes are relatively stable and diffuse (Herrick & Thomas, 1999), resulting in relatively uniform shade leaf morphology (Table 1; Herrick & Thomas, 2001). By contrast, shade leaves in the broadleaved canopy of sweetgum trees at the Oak Ridge FACE experiment are probably more deeply shaded and cooler than in our study.

Our leaf-level results were similar to the 25% reduction in sap-flux density (gH2O m−2 sapwood s−1) at elevated CO2 of the same sweetgum trees from 1999–2000 (Schäfer et al., 2002). There has been no evidence that leaf-area index (LAI) has changed between the CO2 treatments at the Duke FACE site (Schäfer et al., 2002), but LAI of sweetgum alone has not been directly measured. Nevertheless, the overall similarity of LAI across CO2 treatments suggests that our leaf-level results are representative of whole-tree responses in this system. Wullschleger et al. (2002) found only a 14% reduction in mean canopy conductance (mmol m−2 s−1) of sweetgum at the Oak Ridge FACE experiment, and concluded that weaker responses in the lower canopy moderated the overall influence of CO2 on canopy conductance. Results from the Oak Ridge FACE site also indicate no differences in LAI between the CO2 treatments (Norby et al., 2003). The differences between the Oak Ridge and Duke Forest FACE experiments among overstory sweetgum trees demonstrate the utility of experiments at multiple sites to quantify the effects of elevated CO2 on gs of a given species.

Stomatal acclimation to CO2 enrichment can enhance the reduction of gs by elevated CO2 (Bunce, 2001; Lodge et al., 2001; Maherali et al., 2002), but has rarely been examined in plants growing in forest ecosystems (Medlyn et al., 2001). Sage (1994) suggested the use of the Ci/Ca ratio as an index for stomatal acclimation because it reflects any changes in the relationship between stomatal conductance and photosynthetic capacity. Stomatal acclimation would occur if stomata close relative to photosynthetic activity and reduce Ci/Ca (Sage, 1994). We found a modest 5% (P = 0.064) reduction overall for sun and shade leaves during all measurement periods (Fig. 5). However, the overall effect of elevated CO2 on Ci/Ca varied based on month, year and canopy position. For example, in September 1998 and 1999, elevated CO2 reduced Ci/Ca by 12% for both sun and shade leaves, whereas there was no effect of CO2 on Ci/Ca in 2000. In 2001 elevated CO2 reduced Ci/Ca by 8% in sun leaves, but only by 2% in shade leaves. These results suggest that at some points, stomatal acclimation probably occurred for sweetgum. Our results also indicate that stomatal acclimation during these times happened independently from the regulation of photosynthesis, because no photosynthetic acclimation occurred in these same trees (Herrick & Thomas, 2001). Although Ci/Ca responses were variable, we note that the strongest effect of CO2 on Ci/Ca was in the driest year (1998), suggesting that water stress is associated with stomatal acclimation. This interpretation is consistent with Sage's (1994) observation that stomatal acclimation to elevated CO2 is more common in cotton during leaf water stress.

Several studies suggest that a reduction in gs in long-term CO2 enrichment may be related to a reduction in stomatal density or stomatal index (Woodward & Bazzaz, 1987; Kurschner, 1997; Beerling et al., 1998). We found no evidence to support this in sweetgum trees at the Duke Forest FACE experiment. Stomatal density, epidermal cell density and stomatal index were all unaffected by CO2 treatment in June 2000, despite a 33% reduction in gs during that period (Table 1). These results indicate that gs responses to CO2 need not correlate with the magnitude or direction of stomatal density responses to CO2.

To determine whether the reduction in gs at elevated CO2 could be limiting carbon assimilation, we calculated relative stomatal limitation of photosynthesis (lg; Jones, 1985) from A–Ci curves measured concurrently with steady-state gas-exchange measurements. Elevated CO2 decreased lg (26%, Fig. 4) despite reducing gs (Fig. 1). Similarly, Gunderson et al. (2002) found a reduction in relative stomatal limitation in elevated CO2 in sweetgum trees at the Oak Ridge FACE experiment. These results indicate that the enhancement of photosynthesis by elevated CO2 more than compensated for the diffusional limitation imposed by stomatal closure at elevated CO2 (Tissue et al., 1995).

There were effects of canopy position on stomatal characteristics that were independent of CO2 treatment. Steady-state stomatal conductance was more than twice as high (+122%) in sun leaves as in shade leaves (Fig. 1), which is a common observation in forest trees (Whitehead, 1998). This might be partially caused by a 32% higher stomatal density in sun leaves than in shade leaves (Table 1), or related to a greater photosynthetic capacity in sun leaves (Herrick & Thomas, 2001). Epidermal cell density was not different between sun and shade leaves, but stomatal index was lower in the shade, indicating there were effects of leaf position on the initiation of stomata. Relative stomatal limitation (lg) of photosynthesis was 24% greater in sun leaves than shade leaves, and was probably a result of lower photosynthetic capacity in the shade leaves (Herrick & Thomas, 2001). These results demonstrate that the complex nature of within-canopy light environments is important in considering larger-scale canopy water use.

The response of gs to water stress in CO2 enrichment is particularly important because global climate change may increase the potential for drought conditions in some areas (Rind et al., 1990). Stomatal conductance decreased with decreasing soil moisture content, but elevated CO2 did not change the drought sensitivity of gs in sun or shade leaves (Fig. 2). We also observed that gs generally decreased with increasing leaf-to-air VPD, yet elevated CO2 did not affect stomatal sensitivity to VPD (Fig. 3). This is consistent with results in sweetgum at the Oak Ridge FACE experiment (Gunderson et al., 2002). In other species, the relative reductions in gs at elevated CO2 have been reported to be larger in low light and at low vapor pressure deficits than in high light and high vapor pressure deficits (Heath & Kerstiens, 1997; Bunce, 2001). We note, however, that these relationships with sweetgum leaves were pooled from different measurement dates with different VPDs, and observed responses would probably change if VPD was experimentally manipulated over individual leaves (Maherali et al., 2003).

Medlyn et al. (2001) and Sage (1994) observed that elevated CO2 most strongly reduced gs in drought stress. Our results provide some evidence for this pattern, as elevated CO2 reduced gs by 40% during a moderate drought in September 1998, while gs was reduced by only an average of 26% during the other measurement periods in this study (Fig. 1). However, the low magnitude of gs during the drought caused the absolute reduction of gs to be less during the drought of September 1998 than during other measurement periods (Fig. 1). In a previous study we found that elevated CO2 had no effect on gs when measured during the late season (23 September−16 November 1998) after the 1998 drought, as sweetgum leaves senesced (Herrick & Thomas, 2003). In that study it was not possible to separate the effects of leaf age from drought. Wullschleger et al. (2002) also found that the CO2-induced reduction in gs disappeared after mid-September, and argued that any condition that reduces the absolute magnitude of gs will also reduce the effect that elevated CO2 has on these gas-exchange processes. Our results support this conclusion with respect to the effects of leaf age and the absolute difference under water stress.

In summary, we found that the reductions in stomatal conductance in elevated CO2 in our study were sustained over four growing seasons with varying soil moisture conditions, a result consistent with other long-term studies of trees (Medlyn et al., 2001). The observed decrease in leaf-level stomatal conductance at elevated CO2 was similar in magnitude to the reduction in sap flux density of sweetgum at the Duke Forest FACE experiment (Schäfer et al., 2002). We also found consistent CO2 effects between sun and shade leaves. Our results demonstrate the value of long-term studies on large trees under variable environmental conditions as a tool to understand the effects of rising CO2 on the biosphere. In particular, long-term studies in natural systems are essential to confirm or refute previous studies done on smaller trees under more artificial conditions. Our results confirm that decreases in leaf-level stomatal conductance caused by CO2 enrichment are relatively consistent across a variety of environmental conditions, and do not abate over time.

Acknowledgements

We acknowledge the Brookhaven/Duke Forest FACE site which is supported by the US Department of Energy. We thank Dr Chantal Reid for advice on stomatal density measurements. We also thank two anonymous reviewers for helpful comments. In addition, we appreciate the hospitality and office space provided by the Duke University Phytotron. The research discussed here was supported by the Biological and Environmental Research (BER) Program, US Department of Energy. It was not funded by the US Environmental Protection Agency, is not subject to EPA quality assurance, and does not necessarily reflect the views of the agency and no official endorsement should be inferred.

Ancillary