Carbon isotope discrimination by Sorghum bicolor under CO2 enrichment and drought


Author for correspondence: D. G. Williams Tel: +1 520 621 7259 Fax: +1 520 621 8801


  •  Sorghum bicolor was exposed to free-air CO2 enrichment (FACE) and drought at the Maricopa Agricultural Center, AZ, USA, in summer 1998. We predicted that bundle sheath leakiness (Φ) would be insensitive to FACE under well-irrigated (wet) conditions, but would be lower in FACE compared with control-CO2 treatments when irrigation was withheld (dry).
  •  Leaf and air δ13C values and leaf pi/pa from gas exchange were measured to estimate carbon isotope discrimination (Δ) and F. Midday leaf water potential (Ψ) and photosynthetic rate were simultaneously measured to evaluate the influence of plant water status on Φ and the association between Φ and carbon gain.
  •  Irrigation treatments affected Ψ, pi/pa, Δ and Φ in control CO2 and FACE rings. Differences in leaf Δ between wet- and dry-treatment plots resulted from changes in Φ and to stomatal influences on pi/pa. FACE had very little effect on Ψ, Δ and Φ in wet-treatment plots. However, Φ and Δ in dry plots were higher in control than in FACE rings.
  •  FACE ameliorated the effects of drought on bundle sheath leakiness and Δ by reducing transpiration, prolonging soil water availability and enhancing plant water status. Direct effects of CO2 enrichment on C4 photosynthetic metabolism in Sorghum apparently are minimal and indirect effects depend on soil water supply.


The CO2 concentrating mechanism in C4 photosynthesis involves coordinated functioning between mesophyll and bundle sheath cells and two carboxylation pathways. High CO2 concentrations in bundle sheath cells greatly reduce the rate of oxygenation by ribulose bisphosphate carboxylase/oxygenase (Rubisco) and resultant photorespiration. The leakiness of the bundle sheath to CO2 (Φ), defined as the fraction of CO2 originally fixed by phosphoenol pyruvate carboxylase (PEPC) in the mesophyll that subsequently leaks out of the bundle sheath cells quantifies the efficiency of the CO2 concentrating mechanism (Farquhar, 1983). Bundle sheath leakage reduces C4 pathway efficiency and represents an added energy cost to the plant because ATP is required for regeneration of phosphoenol pyruvate (PEP) (Hatch & Osmond, 1976). Estimates of Φ for a variety of C4 plants range from 0.2 to 0.6 (Farquhar, 1983; Henderson et al., 1992; Sandquist & Ehleringer, 1995; Saliendra et al., 1996; Meinzer & Zhu, 1998). Some studies suggest that PEPC and Rubisco activities are regulated such that Φ is kept fairly constant over a range of environmental conditions (Krall & Edwards, 1990; Henderson et al., 1992). However, substantial environmental and genetic variation in Φ has been observed in some species (Saliendra et al., 1996; Meinzer & Zhu, 1998).

It is not clear how rising CO2 levels in the atmosphere will alter Φ in C4 plants. C4 grasses evolved during the Miocene and Pleistocene when CO2 partial pressures in the atmosphere were much lower than present values (Ehleringer et al., 1991). PEPC is generally saturated at relatively low CO2 partial pressure so increases in mesophyll CO2 concentrations should not have much impact on photosynthetic rates, quantum yield and Φ (Ehleringer & Bjorkman, 1977; Tissue et al., 1995). Thus, direct effects of rising CO2 concentrations above current ambient levels should be negligible for C4 plants. However, indirect effects of rising CO2 levels and their interaction with other environmental factors, such as drought, may have significant impacts on these photosynthetic parameters. Moderate increases in A and biomass production have been observed in C4 grasses exposed to elevated CO2 concentrations in field experiments, but these increases are generally apparent only when water is limiting (Knapp et al., 1993; Owensby et al., 1993; Hamerlynck et al., 1997). It is not known how changes in Φ factor into these responses.

Carbon isotope discrimination (Δ) in C4 plants can be used to estimate Φ. Δ in C4 plants is related to fractionations associated with diffusion of CO2 into the leaf (a = 4.4‰), carboxylation by Rubisco (b3 = 29‰), dissolution of CO2 to HCO3 and fixation by PEPC (b4c. −5.7‰), the ratio of leaf internal to ambient CO2 concentration (pi/pa) and Φ (Farquhar, 1983) in the following manner:

image( Eqn 1)

s, the fractionation associated with diffusion of CO2 out of the bundle sheath cells (1.8‰), has been added to Farquhar’s original model as suggested by Peisker & Henderson (1992). Equation 1 can be rearranged to yield an expression for Φ:

image( Eqn 2)

Since a, b4, b3, and s are relatively constant, then Φ can be estimated from measurements of Δ and pi/pa.

Here we present data on Δ and Φ from a free-air CO2 enrichment (FACE) experiment with the C4 grass Sorghum bicolor (L.) Möench. We predicted that leakiness would be insensitive to changes in CO2 concentration under well-irrigated conditions, but would be lower at enriched CO2 levels compared with current ambient CO2 concentration when water was limiting. We hypothesized that lower stomatal conductance and transpiration under FACE would reduce the magnitude of water-stress-induced changes in Φ.

Materials and Methods

Free-air CO2 enrichment (FACE) experiments were conducted at the University of Arizona Maricopa Agricultural Center (MAC), Maricopa, AZ, USA, in 1998 and 1999 to determine interactive effects of elevated CO2 and limited soil-water supply on grain sorghum [Sorghum bicolor (L.) Möench cv. Dekalb 54]. This paper describes carbon isotope discrimination by Sorghum in the 1998 FACE experiment. The experimental design, plant culture and treatments are summarized below for this experiment. See Ottman et al. (2001)– pp. 000–000 in this issue, for a more comprehensive description of Sorghum FACE experiments at MAC.

CO2 treatments

The free-air CO2 enrichment (FACE) technique was used to enrich the air in circular plots within a 12-ha Sorghum field similar to prior cotton and wheat experiments at the Maricopa Agricultural Center (MAC) (Hendrey et al., 1993; Wall & Kimball, 1993; Mauney et al., 1994; Kimball et al., 1995; Wechsung et al., 1995). Eight 25-m-diameter toroidal plenum rings were placed in the field shortly after planting. These were arranged into 4 blocks with each block containing one FACE and one control ring. Air enriched with CO2 was blown into 4 of the rings through tri-directional jets in vertical pipes at elevations near the top of the sorghum canopy. A computer-control system used wind speed and CO2 concentration measurements taken within the rings to adjust CO2 flow rates and maintain desired CO2 concentrations. Air blowers were installed in four nonCO2-enriched ambient ‘control’ plots to provide air movement similar to that of the FACE plots.

FACE plots were enriched to a target 200 µmol mol−1 above ambient. The FACE treatment was applied continuously from the date when 50% of the plants emerged until grain maturity. The average daytime CO2 concentrations in the FACE and control plots were 556 and 363 µmol mol−1, respectively, while the night time values were 603 and 428 µmol mol−1.

Irrigation treatments

Each of the main circular FACE and control plots was split in semicircular halves, with each half receiving either an ample (wet) or a water-limited (dry) irrigation regime. The water was applied using flood irrigation. Only two irrigations were applied to the dry treatments compared with seven to the wet treatments. The criterion used to decide when to irrigate the wet plots was after 30% of the available water in the rooted zone was depleted (Kimball et al., 1999); plots were irrigated with an amount calculated to replace 100% of the potential evapotranspiration since the last irrigation, adjusted for rainfall (Fox et al., 1992). The dry plots were planned to receive one-third the number of irrigations and total water amount as the wet plots, so as to stress the Sorghum plants in the dry plots. This was achieved by applying water only twice, once at the start and once again near mid-season. Irrigation plus rain applied during 1998 was 1218 mm and 474 mm to the wet and dry plots, respectively.

Crop culture

The FACE Sorghum experiments were conducted in the east end of Field 28 of the University of Arizona, Maricopa Agricultural Center. The soil was classified as Trix clay loam (Post et al., 1988). A Sorghum crop had been grown on the land in the summer-fall of 1997, and a barley crop had been grown in the winter-spring and harvested before maturity for hay at the beginning of April 1998. The description of planting (below) refers only to the 1998 experiment.

The field was laser leveled and disked in two directions prior to 8 April 1998. Fertilizer (93 kg N ha−1 and 41 kg P ha−1 as urea (46-0-0) and monoammonium phosphate (11-52-0)) and pre-emergent herbicide (Dual (metolachlor; formulation 8E) at 2.7 kg ha−1) were applied by air on 10 and 11 June 1998. Certified grain Sorghumbicolor (L.) Moench seed (Dekalb DK54) was planted 15–16 July, 1998 into relatively dry soil in north–south rows spaced 0.76 m (30 inches) apart at a rate of 328 000 seeds ha−1 (10.915 kg ha−1; or 1 seed per 4 cm of row). Erection of the FACE and control apparatus commenced immediately after planting and was completed by 27 July 1998 (days after planting (DAP) 11) when the first irrigation was applied to all plots. The FACE treatment started on 31 July (DAP 15), when slightly less than 50% emergence had occured. Stand counts shortly after emergence revealed a plant population of 223 100 plants ha−1.

There was only one mid-season irrigation for the dry plots, so a second application of fertilizer was applied to the dry plots on this date (11 September 1998, DAP 57) at a rate of 186 kg N ha−1 to give a total N application of 279 kg N ha−1 for the season. The wet plots were irrigated and fertilized on this date as well, receiving 124 kg N ha−1. Because the dry plots were more cracked, they were given supplemental water for coverage, and they also received more N. To compensate, the wets were given an additional 62 kg N ha−1 on their next scheduled irrigation day (25 September 1998, DAP 71) so that both dry and wet plots received the same total N for the season. The fertilizer was applied in irrigation water as Uran-32 (urea ammonium nitrate (32-0-0)).

Physiological measurements

Leaf-level physiological measurements were made during the vegetative growth period and into the initial reproductive phase of the Sorghum crop (DAP 36–77). Measurements were made only from replicate blocks 2, 3 and 4. Block 1 was excluded due to time constraints for diurnal gas exchange measurements.

Leaf water potential

Leaf water potential (Ψ) was measured weekly at midday (1000–1300 h) with a pressure chamber on the youngest, fully expanded blade collected from one Sorghum plant in each of the wet and dry sides of the FACE and control plots. The terminal portion of the blade was excised and immediately placed in a plastic bag. Water potential measurements were made within 2 or 3 min after collection from each plant.

δ13C and discrimination

Leaf tissue was collected from two plants adjacent to each of those used for leaf Ψ measurements in blocks 2, 3 and 4 on DAP 36, DAP 49, DAP 61 and DAP 77. Leaf tissue (c. 20 cm2) was collected from the mid-portion of the youngest, fully expanded leaf, approximately 20 cm from the apex of the blade. Immediately following collection, the duplicate leaf samples were sealed in screw-cap glass vials and placed on dry ice. Samples were kept frozen during transport to the lab at the University of Arizona in Tucson and then freeze dried. Dried samples were stored in a dessicator and later ground to a fine powder using a ball mill after removal of mid ribs.

Sub-samples (c. 2 mg) of the ground leaf tissue were placed in Vycor tubes containing cupric oxide and silver foil, sealed under vacuum, and then combusted at 850°C for 4 h. CO2 from these combustions was purified cryogenically (Buchanan & Corcoran, 1959) and then analyzed on a Finnigan delta-S isotope ratio mass spectrometer (Finnigan MAT, San Jose, CA, USA) at the University of Arizona Geosciences Stable Isotope Facility. Carbon isotope values (δ13C, ‰) are expressed relative to the Pee Dee Belemite standard. Duplicate samples were analyzed separately, but averaged for each replicate ring 2–4 and treatment combination (n = 3). A laboratory internal standard (spinach) was processed and analyzed with each batch of Sorghum samples. Standard deviation for δ13C values of the spinach standard during these runs was 0.14‰.

Carbon isotope discrimination (Δ) was calculated according to Farquhar et al. (1989) from plant δ13C values (δp) and the corresponding δ13C values of air (δa) measured in FACE and control rings:

image( Eqn 3)

The δ13C values of tank CO2 and air in the FACE and control rings were collected several times between 7 August (DAP 22) and 23 September 1998 (DAP 69). The δ13C value of tank CO2 was −4.36‰, which was c. 4‰ more positive than δ13C of ambient CO2. As a result, the δ13C values of CO2 in FACE rings (−7.49 ± 0.13‰, n = 9) was higher than in control rings (−8.42 ± 0.10‰, n = 4).

Bundle sheath leakiness

Gas exchange measurements in the Sorghum FACE study were made by G. Wall (unpublished) and were used here with Δ values to calculate Φ. Diurnal gas exchange measurements were made on four dates (DAP 36, DAP 49, DAP 61, and DAP 77) with a LI-6200 portable gas exchange system on plants adjacent to those used for δ13C measurements and on comparable leaves of each plant (Wall et al., 2001). The mid-morning estimates of pi/pa from these gas exchange measurements were used in the calculations of Φ to correspond with the time period of maximum daily rates of photosynthesis. There is an inherent temporal discrepancy in this approach for calculating Φ. We combine instantaneous measures of pi/pa with time-integrated measures of Δ from bulk leaf δ13C values. However, since Sorghum in this study produced leaves at a very high rate (every 3–4 d), and pi/pa was relatively stable and shifted slowly over measurement dates (G. Wall, unpublished) we felt that our approach gave valid results.

Statistical analysis

The experiment had a split-plot design with main plots (FACE and control rings) in complete blocks. Each main plot was split into semicircular halves and each half was randomly assigned to either wet or dry irrigation treatments. Because we collected data only from three blocks, we did not have adequate error degrees of freedom (df = 5) to test interactions between irrigation and CO2 treatments and between the treatments and block. To test the interactive effect between CO2 and irrigation treatments, we calculated the difference in physiological response between dry and wet sides of the main plots and analyzed the effect of CO2 treatment on this difference. This is conceptually similar to analyzing the interactive effect between CO2 and irrigation treatments. A significant CO2 treatment effect on the difference between dry and wet subplots then indicates that plants exposed to different levels of water availability responded differently to FACE conditions, which is what we predicted. We also performed independent analyses of CO2 effects on responses within dry and wet treatments. To evaluate the effect of irrigation alone, we performed independent analyzes on data within control and FACE treatments, respectively. In each case, multivariate ANOVA for repeated measures (MANOVA) was performed on physiological responses [Ψ, Δ, Φ, pi/pa and A (net assimilation rate)] with measurement date as the repeated factor. Measurement dates fell before and after the second irrigation for the dry treatment subplots. Block was left in the models as a main effect, but we could not test the CO2 × block or irrigation × block interactions because of the limited degrees of freedom. The JMP statistical software for Macintosh (SAS Institute, 1995) was used for all statistical analyzes.


Irrigation treatments had a prominent effect on Sorghum midday leaf water potential (Ψ), carbon isotope discrimination (Δ), bundle sheath leakiness (Φ), photosynthetic rate (A) and the ratio of internal and ambient CO2 partial pressure (pi/pa) in this study (Table 1). Irrigation significantly affected the response of leaf Ψ in control and FACE plots, when these were analyzed independently. Midday Ψ-values of Sorghum in the dry half of control and FACE plots were initially high (near −1.5 MPa), but dropped rapidly between DAP 36 and DAP 49 prior to the second dry-treatment irrigation on DAP 57. Ψ recovered to −1.5 MPa by DAP 70 after irrigation in these plots (Fig. 1). Midday Ψ values in the wet half of control and FACE plots averaged about −1.5 MPa and changed little through the vegetative growth period of Sorghum in this study (Fig. 1). Only in the control CO2 plots did the effect of irrigation depend on sampling date, suggesting that the impact of the drought cycle was lessened by FACE. Indeed, the average difference in leaf Ψ between wet and dry treatments was 0.9 MPa in control CO2 plots, but only 0.4 MPa in FACE plots at peak drought stress on DAP 49. However, because of the high variance across the dy subplots at this time the analysis of the difference in Ψ between wet and dry subplots did not show a significant effect of CO2 treatment (Table 2). Furthermore, the CO2 treatments did not significantly influence midday leaf Ψ within an irrigation treatment (Table 3).

Table 1.  Repeated measures (MANOVA) for irrigation effects on Sorghum analyzed independently in control CO2 and FACE rings. Significance levels for Wilks’ Lambda test are reported
SourceYDFApi : pa
  • *


  • **


  • ***

    , and ns for P ≤ 0.100, P ≤ 0.050, P ≤ 0.010, and not significant, respectively. The actual P-values are reported when P > 0.100 and P < 0.200.

Control CO2     
Time × block0.156nsnsnsns
Time × irrigation**nsns*ns
Time × blocknsnsnsns0.178
Time × irrigationnsnsnsnsns
Figure 1.

Midday leaf water potential (Ψ) of Sorghum bicolor in wet- and dry-irrigation treatments in FACE and control rings in 1998. Measurements were made weekly between 10.00 and 13.00 h during the vegetative growth period of the Sorghum crop. Vertical bars represent one standard error of the mean. FACE wet, open squares; FACE dry, closed squares; Control wet, open triangles; Control dry, closed trianlges.

Table 2.  Repeated measures analysis of variance (MANOVA) for CO2 effects on the difference between dry and wet irrigation treatments. Significance levels for Wilks’ Lambda test are reported
Source Ψ Δ Φ Api : pa
  • *


  • **


  • ***

    , and ns for P ≤ 0.100, P ≤ 0.050, P ≤ 0.010, and not significant, respectively. The actual P-values are reported when P > 0.100 and P < 0.200.

Time × Blocknsns*nsns
Time × CO2nsns**nsns
Table 3.  Repeated measures analysis of variance (MANOVA) for CO2 effects on Sorghum analyzed independently in wet and dry treatment subplots. Signifcance levels for Wilks’ Lambda test are reported
Source Ψ Δ Φ Api : pa
  • *


  • **


  • ***

    , and ns for P ≤ 0.100, P ≤ 0.050, P ≤ 0.010, and not significant, respectively. The actual P-values are reported when P > 0.100 and P < 0.200.

Time × Blocknsnsnsnsns
Time × CO2nsnsnsnsns
Time × Blockns***nsns0.125
Time × CO2ns**nsns0.110

Carbon isotope discrimination (Δ) during the early vegetative growth period (between DAP 36 and DAP 49) was high, ranging from 3.5 to 4.5‰, in the wet- and dry-treatment subplots, respectively, but then declined after DAP 49 to c. 3.0‰ (Fig. 2). However, the influence of the irrigation treatments on Δ did not depend on sampling date as it did for leaf Ψ (Table 1). CO2 treatments had an influence on the response of Δ to the irrigations. There was a significantly greater difference in Δ-values between dry and wet halves in control compared with FACE plots (Fig. 2; Table 2). When wet- and dry-treatment subplots were analyzed separately, only in the dry subplots did CO2 influence leaf Δ (Table 3).

Figure 2.

Carbon isotope discrimination (Δ) of Sorghum bicolor in wet- and dry-irrigation treatments in FACE and control rings in 1998. The Δ-values were calculated from δ13C values of CO2 in air collected within FACE and control rings and bulk leaf material. Vertical bars represent 1 SE of the mean. FACE wet, open squares; FACE dry, closed squares; Control wet, open triangles; Control dry, closed trianlges.

Bundle sheath leakiness (Φ) in our FACE experiment estimated from Δ and pi : pa values (Eqn 2) were similar to those reported previously for other C4 grasses (Bowman et al., 1989; Saliendra et al., 1996; Meinzer & Zhu, 1998). Bundle sheath leakiness was stable over the vegetative growth period of Sorghum in wet subplots between DAP 36 and DAP 77 (Fig. 3). However, as with Δ, Φ was more strongly affected by CO2 treatments in dry than in wet subplots. When dry and wet subplots were analyzed separately, CO2 significantly affected Φ-values only in the dry-irrigation plants (Table 3). Bundle sheath leakiness was c. 25% higher in the control-dry than in the other treatment combinations on DAP 49 (Fig. 3). However, by DAP 77 water stress had diminished (Fig. 1) and Φ in all treatments had converged to a value of c. 0.3 (Fig. 3). The difference in Φ-values between dry and wet halves of each plot were strongly affected by CO2 treatment and by the interaction between CO2 treatment and sampling date (Table 2). Irrigation moderately affected Φ (P = 0.103) in control CO2 plots, but had no effect on leakiness in FACE plots when CO2 treatments were analyzed separately (Table 1).

Figure 3.

Bundle sheath leakiness to CO2 (Φ) of Sorghum bicolor in wet- and dry-irrigation treatments in FACE and control rings in 1998. The Φ-values were calculated from spot pi : pa measurements and Δ-values of bulk organic carbon of leaves. Vertical bars represent 1 SE of the mean. FACE wet, open squares; FACE dry, closed squares; Control wet, open triangles; Control dry, closed trianlges.

The greatest differences in bundle sheath leakiness (Φ) among treatments occured on DAP 49 at peak drought stress just before the second dry-treatment irrigation (Figs 1 and 3). We performed a series linear regression analyzes on mean treatment responses (n = 4) to evaluate relationships between Φ, carbon isotope discrimination (Δ) and ancillary leaf physiological responses on DAP 49. Midday leaf water potential (Ψ) was a strong determinant of Φ (r2 = 0.66; P = 0.181) and Δ (r2 = 0.91; P = 0.048) on this date; these parameters were both negatively correlated with Ψ across treatments (Fig. 4). Carbon isotope discrimination is codetermined by pi : pa and F in C4 plants, so we analyzed the response of D as a function of F and pi : pa in separate analyzes (Fig. 5). pi : pa varied as expected across a range of water-stress conditions. Well-irrigated Sorghum plants on DAP 49 had fairly high pi : pa values (near 0.55), whereas with drought stress, pi : pa values were lower (near 0.30). D was positively correlated with F (r2 = 0.87; P = 0.068) and negatively correlated with pi : pa (r2 = 0.85; P = 0.077) on DAP 49 in our study (Fig. 5). Carbon isotope discrimination and F also were related to net assimilation rate (A) on DAP 49 (Fig. 6). Net assimilation rate was negatively correlated to both Δ (r2 = 0.96; P = 0.018) and Φ (r2 = 0.76; P = 0.129) on this date.

Figure 4.

Relationship between midday leaf water potential and (a) carbon isotope discrimination (Δ) and (b) bundle sheath leakiness to CO2 (Φ) of Sorghum bicolor 49 d after planting in the 1998 FACE experiment. Vertical and horizontal bars represent ± 1 SE of the mean. FACE wet, open squares; FACE dry, closed squares; Control wet, open triangles; Control dry, closed trianlges.

Figure 5.

Relationship between carbon isotope discrimination (Δ) and (a) the ratio of leaf internal to ambient CO2 partial pressure (pi : pa) and (b) bundle sheath leakiness to CO2 (Φ) in Sorghum bicolor 49 d after planting in the 1998 FACE experiment. Vertical and horizontal bars represent ± 1 SE of the mean. FACE wet, open squares; FACE dry, closed squares; Control wet, open triangles; Control dry, closed trianlges.

Figure 6.

Relationship between net photosynthetic rate at midday (A) and (a) carbon isotope discrimination (Δ) and (b) bundle sheath leakiness to CO2 (Φ) in Sorghum bicolor 49 d after planting in the 1998 FACE experiment. Vertical and horizontal bars represent ± 1 SE of the mean. FACE wet, open squares; FACE dry, closed squares; Control wet, open triangles; Control dry, closed trianlges.


Stomatal conductance and transpiration are often lower, and midday leaf water potentials (Ψ) and soil water availability are generally greater in C4 grasses grown under elevated CO2 in field settings (Knapp et al., 1993; Knapp et al., 1996; Hamerlynck et al., 1997). This effect is particularly apparent in dry years. Elevated CO2 under drought conditions may have an indirect effect on Φ if soil water depletion and/or leaf water deficits are ameliorated by reduced transpiration resulting from stomatal closure. We predicted that FACE would reduce drought stress and lessen its impact on photosynthetic metabolism in Sorghum. We detected a moderately significant effect atmospheric CO2 concentration on the difference between dry and wet subplot Δ and Φ-values in this experiment. Volumetric soil water contents were higher and evapotranspiration was lower in FACE compared to control rings in dry plots in the 1998 Sorghum experiment (Conley et al., unpublished). FACE-induced reductions in stomatal conductance were responsible for this difference (Wall et al., 2001). These data support the findings of other field-based CO2 enrichment studies (Knapp et al., 1993; Knapp et al., 1996; Hamerlynck et al., 1997) suggesting that productivity could potentially be enhanced in drought susceptible C4 grasslands under future atmospheric CO2 conditions.

Bundle sheath leakiness (Φ) is stable in Sorghum bicolor (Henderson et al., 1992; Henderson et al., 1998) and in other NADP-me subtype C4 grasses (Buchmann et al., 1996) with changes in some environmental conditions. Under well-irrigated conditions in the present study, Φ of Sorghumbicolor was very stable ranging only between 0.27 and 0.34. However, when irrigation was applied only twice on our field-grown Sorghum crop (once during planting and once midway through the vegetative growth stage) to promote drought stress, Φ rose to as high as 0.42. Close to half of the CO2 fixed by PEPC and transported to bundle sheath cells subsequently leaked back to the mesophyll under these water-stressed conditions. Similarly, Bowman et al. (1989) observed Φ-values of up to 0.55 in water-stressed Zea mays and Andropogon glomeratus, two other NADP-me subtype grasses. Drought stress also promoted high leakiness values in sugarcane (Saccharum spp.) (Saliendra et al., 1996). Apparently the CO2 concentrating mechanism in these C4 grasses is particularly sensitive to drought stress. Henderson et al. (1998) demonstrated the usefulness of carbon isotope discrimination as an indicator of water-use efficiency in Sorghum under certain environmental conditions. However, estimates of water-use efficiency in Sorghum and other C4 grasses from carbon isotope discrimination values may be problematic under conditions of drought stress since Φ apparently is not stable under these conditions.

Carbon isotope discrimination at peak drought stress on DAP 49 in this study was correlated with Φ and pi : pa. Saliendra et al. (1996) attributed all of the variation in D in sugarcane to changes in F; pi : pa was stable across irrigation treatments in their study. Other studies suggest that environmental variation in D is attributable to stomatal influences on pi : pa only (Madhavan et al., 1991; Henderson et al., 1992). Our study suggests that Φ and pi : pa both vary and together determine Δ in Sorghum. However, shifts in Φ appear to be more prominent at greater water stress levels.

Variation in Φ in C4 plants has been attributed to physical changes in the membrane, cell wall and suberized lamella of the bundle sheath, and to decreased C3 (Rubisco) to C4 cycle (PEPC) activity (Hattersley & Browning, 1981; Ehleringer & Pearcy, 1983; Farquhar, 1983). Disproportional suppression of C3-cycle activity by drought would cause CO2 concentrations in the bundle sheath to rise, causing greater CO2 leakage and diminished efficiency of the C4 concentrating mechanism. Water limitation reduced the ratio of Rubisco to PEPC activity and increased Φ in sugarcane (Saliendra et al., 1996). This mechanism was invoked also to explain the high leakiness in N-deficient and salt-stressed sugarcane in parallel studies (Meinzer et al., 1994; Meinzer & Zhu, 1998). There were no appreciable effects of drought on bundle sheath conductance to CO2 in mature leaves of Sorghum in our FACE experiment (Cousins et al., 2001). However, differences in the relative activities of PEPC and Rubisco developed among the plants exposed to wet and dry treatments (Adam et al., unpublished). The ratio of Rubisco to PEPC activities on DAP 49 was c. 14% lower in control-dry plots than in FACE-dry plots in our FACE experiment in 1998. Thus, variation in Φ in this study apparently was related to changes in C4 overcycling rather than to changes in bundle sheath conductance.

Leakage of previously fixed CO2 from the bundle sheath reduces quantum yield by increasing the ATP requirement for photosynthate production. Our data support earlier findings (Bowman et al., 1989; Saliendra et al., 1996; Meinzer & Zhu, 1998) that variation in Φ in C4 plants leads to variation in Δ and photosynthetic performance. Net photosynthetic rates from gas exchange measurements in Sorghum were negatively correlated with Φ in this study. However, since variation in Δ was attributed to changes in pi : pa and Φ in this study, it may be difficult to use Δ as an indicator of the underlying changes in biochemistry that accompanies reduced A under environmental stress in Sorghum.

In conclusion, there appears to be very limited direct effect of elevated CO2 on leaf Δ and Φ in Sorghumbicolor; these parameters were unchanged by FACE in well-irrigated plants in our 1998 study at Maricopa, AZ, USA. However, under conditions of limited water supply, variation in plant water potentials develops as a result of changes in leaf stomatal conductance and transpiration rate under FACE and control CO2 conditions. FACE ameliorates drought stress in Sorghum. Variation in water stress levels in Sorghum, like in other NADP-me subtype C4 grasses, was responsible for changes in the proportion of CO2 leaked from bundle sheath cells, and potentially for changes in photosynthetic rate, and to some degree the productivity of this C4 grass ecosystem.


We thank Dan Keopke, John King and Sean Schaeffer for assistance in the field and Dr Darren Sandquist for providing helpful comments on an earlier draft of this manuscript. Dr Bob Steidl is gratefully acknowledged for providing valuable advice. The research was supported by Interagency Agreement No. DE-AI03–97ER62461 between the Department of Energy, Office of Biological and Environmental Research, Environmental Sciences Division and the USDA, Agricultural Research Service (Bruce A. Kimball, PI); by Grant no. 97–35109–5065 from the USDA, Competitive Grants Program to the University of Arizona (S. W. Leavitt, M. Ottman, A. D. Mattias, T. L. Thompson, D. G. Williams, and R. L. Roth, PIs); and by the USDA, Agricultural Research Service. It is part of the DOE/NSF/NASA/USDA/EPA Joint Program on Terrestrial Ecology and Global Change (TECO III). This work contributes to the Global Change Terrestrial Ecosystem (GCTE) Core Research Programme, which is part of the International Geosphere-Biosphere Programme (IGBP). We also acknowledge the helpful cooperation of Dr Robert Roth and his staff at the Maricopa Agricultural Center. Portions of the FACE apparatus were furnished by Brookhaven National Laboratory, and we are grateful to Mr Keith Lewin, Dr John Nagy and Dr George Hendrey for assisting in its installation and consulting about its use.