Elevated atmospheric CO2 improved Sorghum plant water status by ameliorating the adverse effects of drought

Authors


  • 1

    Mention of this or any other proprietary product does not imply an endorsement or recommendation by the authors or their Institution over products not mentioned.

Author for correspondence: Gerard W. Wall Tel: +1 602 437 1702 ext. 279 Fax: +1 602 437 5291 Email: gwall@uswcl.ars.ag.gov

Summary

  • • The interactive effects of atmospheric CO2 concentration and soil-water content on grain sorghum (Sorghum bicolor) are reported here.
  • • Sorghum plants were exposed to ambient (control) and free-air CO2 enrichment (FACE; ambient + 200 µmol mol−1), under ample (wet, 100% replacement of evapotranspiration) and reduced (dry, postplanting and mid-season irrigations) water supply over two growing seasons.
  • • FACE reduced seasonal average stomatal conductance (gs) by 0.17 mol (H2O) m−2 s−1 (32% and 37% for dry and wet, respectively) compared with control; this was similar to the difference between dry and wet treatments. FACE increased net assimilation rate (A) by 4.77 µmol (CO2) m−2 s−1 (23% and 9% for dry and wet, respectively), whereas dry decreased A by 10.50 µmol (CO2) m−2 s−1 (26%) compared with wet. Total plant water potential (ψw) was 0.16 MPa (9%) and 0.04 MPa (3%) less negative in FACE than in the control treatment for dry and wet, respectively. Under dry, FACE stimulated final shoot biomass by 15%.
  • • By ameliorating the adverse effects of drought, elevated atmospheric CO2 improved plant water status, which indirectly caused an increase in carbon gain.
Abbreviations
 

A, instantaneous leaf net assimilation rate (µmol (CO2) m−2 s−1)

Introduction

The Intergovernmental Panel on Climate Change (IPPC, 1996) projects that if anthropogenic-based emissions of CO2 are maintained at 1994 levels, a rise in atmospheric CO2 from present-day concentrations of c. 370 µmol mol−1 to c. 500 µmol mol−1 will occur by the 21st century. Climate modelers predict that rising levels of atmospheric CO2 will increase air temperature and alter precipitation patterns. Undoubtably, alterations in future climatic conditions will affect soil-water content and production of sorghum (Sorghum bicolor (L.) Möench) in semiarid regions where it predominates.

Sorghum is a major grain crop that is especially important in developing countries (Food and Agriculture Organization of the United Nations (FAO, 1996)) and with concerns about food security, sorghum has been identified as high priority for global change research (IPCC, 1996) by the Global Change and Terrestrial Ecosystems – International Geosphere – Biosphere Programme (GCTE-IGBP) (Steffen et al., 1992). Sorghum is a C4 grass representative of species prominent in warm prairies and savannas, which cover significant areas of the Earth and which have a large pool of soil carbon. The cooler pole-ward expanses of grassland are dominated by species with the C3 photosynthetic pathway, and global change could potentially cause large shifts in the transition area between C4 and C3-dominated ecosystems (Mayeux et al., 1991). Thus, for food security, ecosystem, and global carbon budget concerns, there is merit in obtaining a better understanding of effects of elevated CO2 and changes in soil-water supply on gas exchange (CO2 and H2O) processes and plant water relations that affect net primary production of sorghum.

Because of a CO2-based regulation of stomatal conductance (Morison, 1998), large reductions in transpiration rate have been reported for sorghum grown at elevated CO2 with ample water (Pallas, 1965; van Bavel, 1974). However, the response of stomata to elevated CO2 may differ depending on the interactive effect of CO2 level and soil-water content. The net effect of elevated CO2 on soil-water content ultimately depends on its effect on leaf area and stomatal response such that soil-water content can either be unaffected, increased, or decreased by an increase in CO2 (Samarakoon & Gifford, 1995). Typically however, water-stressed plants will utilize all the water that they can obtain, so that season-long cumulative water usage is minimally affected by elevated CO2 (Morison & Gifford, 1984a,b). Under drought conditions, however, any reduction in water usage will reduce soil-water depletion, which may feedback and improve physiological processes such as photosynthesis, water relations and growth (Kirkham, 1990; Kirkham et al., 1991; Nie et al., 1992).

In general plants with the C3 photosynthetic pathway are more responsive to elevated CO2 than C4 plants (Ziska et al., 1990; Lawlor & Mitchell, 1991; Poorter, 1993; Poorter et al., 1996; Wand et al., 1999). This occurs primarily because within the bundle sheath cells of C4 plants CO2 levels are at near saturation and photorespiration is suppressed (Nobel, 1991; Hatch, 1992; von Caemmerer, 2000). Under current ambient CO2 levels, therefore, gas exchange measurements have demonstrated that C4 photosynthesis is at near saturation (Ludlow & Wilson, 1971; von Caemmerer et al., 1977), but elevated CO2 has been reported to increase net assimilation rate for well-watered C4 plants (Wand et al., 1999; Ziska et al., 1999). Ghannoum et al. (2000) reported that any improvement in net assimilation rate and growth in C4 plants because of elevated CO2 will most likely occur because of an improvement in water relations (stomatal limitation) rather than any alteration in biochemical-based factors (nonstomatal limitation).

In communities of C4 salt march grasses, where water is nonlimiting, the photosynthetic response of C4 plants to elevated CO2 is much smaller than C3 (Drake & Leadley, 1991; Arp et al., 1993). A similar response has been observed for other well-watered C4 crop species (Poorter, 1993; Poorter et al., 1996; Rogers & Dahlman, 1993). However, competitiveness of C3 over C4 plants at elevated CO2 is less under drought conditions. The relative enhancement in carbon gain (Garcia et al., 1998; Osborne et al., 1998; Wall et al., 2000; Wechsung et al., 2000) and subsequent growth (Morison & Gifford, 1984a,b; Kimball et al., 1995; Pinter et al., 1996) in a C3 cool-season annual grass, wheat (Triticum aestivum L.) increased in response to elevated CO2 more under drought than under well-watered conditions. In rainfed tallgrass prairies containing C3 Kentucky bluegrass (Poa pratensis L.) and C4 big bluestem (Andropogon gerardii) grasses (Kirkham et al., 1991; Nie et al., 1992; Knapp et al., 1993; Owensby et al., 1993), and in a native shortgrass steppe containing C3 (Pascopyrum smithii) and C4 (Bouteloua gracilis) perennial grasses (Hunt et al., 1996), a significant increase in carbon gain and net primary productivity has been reported in response to elevated CO2. Nevertheless, across a wide range of environments (salt march, tallgrass prairies, native shortgrass steppe, agronomic cropping systems), species (domestic, wild-types), and photosynthetic pathways (C3 and C4), the beneficial effect of elevated CO2 on net primary production has been shown to be dependent on water status.

An increase in photosynthesis at elevated CO2 will increase the pool of total nonstructural carbohydrates (Hendrix et al., 1994; Drake et al., 1997; Estiarte et al., 1999) that can be used to develop a more robust root system (Rogers et al., 1992; Rogers & Runion, 1994; Wechsung et al., 1995, 1999). In sorghum, roots reach a maximum rooting depth of between 1.6 and 2.0 meters (Mayaki et al., 1976; Kiagama et al., 1977), with a typical soil-water extraction front velocity (rate of root zone penetration) of 27.2 mm day−1 (Robertson et al., 1993a,b). Chaudhuri et al. (1986a) demonstrated that elevated CO2 increased root mass at every growth stage of sorghum, and that roots reached the bottom of a 1.6-m mini-rhizotron faster under elevated than ambient CO2. A corresponding increase in root length density was also observed, which suggest alterations in morphological characteristics (increased branching and number of fine roots) in sorghum grown in elevated CO2, as observed in other species (Rogers et al., 1992).

An increase in stomatal resistance, that reduces water use, and greater capacity to extract available soil-water and nutrients by roots, would tend to enable plants to maintain greater relative water content and less negative total leaf water potential (Szeicz et al., 1973; Denmead & Millar, 1976; Ackerson et al., 1977), and elevated CO2 has been shown to cause further improvements in both drought avoidance and tolerance mechanisms (Kirkham et al., 1991; Wall, 2001). Less negative total leaf water potential in plants exposed to elevated CO2 cause a positive feedback between net assimilation rate (Graan & Boyer, 1990) and optimal growth and development (Boyer, 1968; Hsiao & Acevedo, 1974; Hsiao & Jing, 1987). The net result is an increase in net primary productivity in C3 plants such as wheat (Kimball et al., 1995; Pinter et al., 1996; Wall et al., 2000). However, the effect of elevated CO2 on growth for systems dominated by C4 species is still not well defined (Tyree & Alexander, 1993; Poorter, 1993; Poorter et al., 1996).

Our objective was to elucidate the interdependency of three edaphic (volumetric soil-water content (θs), evapotranspiration rate (ET), and cumulative evapotranspiration (ETc)), eight gas exchange (stomatal conductance (gs), net assimilation rate (A), intrinsic water use efficiency (IWUE: A/gs), transpiration rate (T), leaf temperature (Tl), leaf minus air (Ta) temperature (ΔT = Tl − Ta), internal CO2 concentration (Ci), ratio of Ci to atmospheric CO2 concentration (Ca) (Ci : Ca)), two water relations (relative water content (RWC) and total plant water potential (ψW)), and eight plant growth (average leaf area index (LAIa), average accumulated shoot biomass (Ba), specific leaf weight (SLW), peak leaf area index (LAIp), final shoot biomass (B), grain yield (YLD), harvest index (HI: YLD/B), and maximum depth of root penetration (Dr)) parameter responses of sorghum grown under ample and reduced soil-water content and ambient and elevated atmospheric CO2 levels. Specifically, we hypothesized the following: (1) elevated CO2 should directly reduce gs, but this response should be greater under wet than dry conditions; (2) a reduction in gs should cause a reduction in T resulting in reductions in ET and ETc, thereby lowering depletion of available soil-water content, but this response should be greater under wet than dry conditions; (3) reduction in gs should decrease transpiration cooling, thereby increase Tl; (4) although the photosynthetic response to elevated CO2 will be less than that observed in C3 plants, on a relative basis it should be proportionately greater under dry compared with wet plots; (5) the additive effect of elevated CO2 in decreasing consumptive water use and increasing water absorption capacity of roots should improve water relations of sorghum under drought conditions; and (6) improved water relations and a net increase in carbon gain should enhance both above- and below-ground growth, but any elevated CO2-based enhancement in growth will be proportionately greater under dry than wet conditions. To test the validity of these six hypotheses we quantified and characterized the effects of tissue dessication (during soil dehydration) and subsequent recovery (after soil rehydration) for a sorghum crop grown in an open field in ambient air and air enriched with CO2 and under two soil dehydration/rehydration cycles over a 2-yr study.

Materials and Methods

Two experiments were conducted during the 1998 and 1999 growing seasons to determine the interactive effects of atmospheric CO2 concentration and soil-water content on grain sorghum. The experiments were conducted at the University of Arizona Maricopa Agricultural Center (MAC), Maricopa, Arizona, USA, located 50 km south of Phoenix, Arizona (33.1°N, 112.0°W).

Carbon dioxide and irrigation treatments

The free-air CO2 enrichment (FACE) technique was used to enrich the air in circular plots within a 12-ha sorghum field (Hendrey, 1993; Wall & Kimball, 1993; Kimball et al., 1999; Ottman et al., 2001). Plants were exposed to an elevated CO2 treatment of c. 200 µmol mol−1 above ambient (c. 370 µmol mol−1) for 24 h per day.

Each of the main circular FACE and control plots was split into semicircular halves. Level-basin flood irrigation was utilized to supply each half with either an ample (wet) or a water-stress (dry) irrigation regime. All equipment and hard ware associated with the FACE and control apparatus was elevated to reduce restrictions to surface water distribution. Wet plots were irrigated after 30% of available water in the rooting zone was depleted. They were irrigated with an amount calculated to replace 100% of potential evapotranspiration since the last irrigation, adjusted for rainfall. Only two irrigations were applied to dry treatments (postplanting and mid-season) each season compared to 7 (1998) (Fig. 1b) or 5 (1999) (Fig. 1d) to the wet treatments.

Figure 1.

Mean volumetric soil-water content (θs) to a depth of 1.8 m of a Trix clay loam for day after planting (DAP) and calendar dates given for the 1998 (a) and 1999 (c) sorghum growing seasons (Conley et al., 2001). Vertical dashed lines separate the first from the second soil dehydration cycle. (a) and (c) Control-dry (CD), closed triangles; FACE-dry (FD), closed sqaures; control-wet (CW), open triangles; FACE-wet (FW), open squares. Flood irrigation and rainfall amounts and application dates given in (b) during 1998 and (d) during 1999. (b) and (d) hatched columns, dry; open columns, wet; closed columns, rain. Amount of fertilizer applied in the irrigation water is given in parenthesis above irrigation histograms for 1998 (b) and 1999 (d). +, * Fertilizer application by air during 1998 (b) and 1999 (d), respectively. Cumulative (Cum.) amount of water (flood irrigation + rainfall) given by y-intercept for dry and wet curves. Onset of the differential irrigation denoted by separation in Cum. curves on 14 Aug. 1998 (DAP 029) (b) and 23 July 1999 (DAP 038) (d). Up-pointing arrows denote DAP when physiological parameters were measured, whereas parenthesis below arrows denote number of days between measurements dates and last irrigated for (dry : wet) treatments. †During the 1999 season, all measurements were discontinued after hail damage on 16 Sept. (DAP 093).

Because no physical barrier existed between the dry and wet plots, some lateral flow of water occurred from the wet to dry plots. To minimize this effect on all edaphic, physiological, and growth parameters measured, the row of sorghum in the dry plots adjacent to the walkway between dry : wet plots was treated as an additional border row (Ottman et al., 2001).

Crop culture

Certified grain sorghum seeds (DeKalb Hybrid DK 54) were sown with a pneumatic planter (Model PNU 96, Monosem, Largeasse, France) containing a 72 hole planting plate 1(Model 6072.5a and 6073.5, Monosem, ATI, Inc., Lenexa, Kansas, USA) into flat beds in a Trix clay loam soil (fine-loamy, mixed (calcareous) hyperthermic Typic Torrifluvent) in north–south rows 0.76 m apart on 15–16 July 1998 and 14–15 June 1999. Seeding rates were 10.9 kg ha−1 (40 mm apart for 33 seeds m−2; plant density of 22 plants m−2) during 1998 and 9.97 kg ha−1 (41 mm apart for 32 seeds m−2; plant density of 21 plants m−2) during 1999. A final plant population of 223, 100 plants ha−1 was obtained during 1998 and 259, 500 plants ha−1 during 1999. To derive plant growth parameters treatment means were pooled over subsamples for each sample date for four replications (n = 4) as described by Ottman et al. (2001). Phenological development (average numerical code across all treatments) was used to group results by distinctive growth stages (Vanderlip, 1993).

Fertilizer was applied before emergence by air at a rate of 93 kg N ha−1 and 41 kg P ha−1 during both years. Fertilizer subsequently was applied in irrigation water as Uran-32 (urea ammonium nitrate, 0.32 kg N kg fertilizer−1) to both the wet and dry plots. Because dry plots were irrigated only once during mid-season, a second application of fertilizer was applied on 11 Sept. 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 (Fig. 1b). The wet plots were also irrigated and fertilized on this date, receiving 124 kg N ha−1.

Because the soil in the dry plots was more cracked than wet, they had to be given more water for coverage, consequently, they also received more N. To compensate, the wets were given an additional 62 kg N ha−1 on their next scheduled irrigation day (DAP 71). Nevertheless, both dry and wet plots received the same total N for the 1998 season, even though two applications occurred for dry and three occurred for wet. During 1999, the mid-season irrigation for the dry plots was on 6 August (DAP 52), and all plots were again fertilized at that time with 172 kg N ha−1 in the irrigation water to give a total of 265 kg N ha−1 for the season (Fig. 1d).

Volumetric soil-water content, evapotranspiration rate, and depth of root penetration

Measurements of θs were obtained with neutron scattering equipment (Hydroprobe Model 503 DR, Campbell Pacific Co., Martinez, CA, USA). Access tubes (50-mm-diameter and 3.0-m-long) were installed in the no-traffic section (Ottman et al., 2001) in each of the 16 plots before planting during 1998. Measurements of θs were taken at 0.3 m intervals to either 1.8 m or 3.0 m depths during the 1998 and 1999 seasons, respectively. In the 1998 season, the uppermost measurement was taken at 0.46 m, whereas in 1999 it was at 0.23 m. A multiple regression model, as described by Conley et al. (2001), was used to estimate θs for the 0.23 m data for 1998.

The active root depth was determined by estimating the expected water extraction front from 0 to 1.76 m (Robertson et al., 1993a). After expected root penetration had occurred to a given depth, a stair-step model incorporated values of θs at consecutively deeper depths in 0.3 m intervals. By assuming negligible deep percolation, ET was essentially set equal to the temporal changes in θs for the active root zone. Whenever irrigation or rain occurred, however, ET was estimated by interpolation by using the average ET before and after the wetting event for the active root zone. Hence, θs and the total amount of irrigation plus rain were used to calculate ET based on the soil-water balance method (Jensen et al., 1990) as implemented by Conley et al. (2001).

An estimate of Dr was determined by determining the soil water extraction front velocity according to Robertson et al. (1993a,b). During 1999, θs by depth was measured at 25.4 mm increments to a depth of 1.6-m when roots had reached their peak growth at inflorescence emergence (DAP 69–78) and again during grain filling (DAP 124–129). The soil depth at which values of θs intersected between inflorescence emergence and grain filling, when root growth had reached its maximum depth, was used to estimate Dr (Robertson et al., 1993a,b). Treatment means for θs and parameters derived from θs (ET, ETc, and Dr) were obtained from four replication (n = 4).

Measurements of midday leaf net assimilation rate, stomatal conductance and transpiration rate

On the first sampling date, midday (solar noon) CO2 and H2O gas exchange rates were measured on randomly selected leaves with three portable closed-exchange (transient) systems with a 0.25-l transparent plexiglass cuvette (Model LI-6200, Li-Cor, Inc., Lincoln, Nebraska, USA). Because leaf sizes were large, subsequent measurements were made with a 1.0-l cuvette. Each infrared gas analyser was calibrated against a gravimetrically prepared mixture of CO2 in air (± 1% Primary Standard, Matheson Gas Products, Inc., Cucamonga, California, USA) and cuvette humidity sensors were calibrated with a dew-point generator (LI-610, Li-Cor, Inc.) immediately before use. Before each measurement the cuvette was allowed to come into equilibrium with prevailing air temperature and relative humidity. Mid-day measurements of A (µmol (CO2) m−2 s−1), gs (mol (H2O) m−2 s−1) and T (mmol (H2O) m−2 s−1), on the central portion of fully expanded upper-canopy sunlit leaves, begun at a cuvette CO2 concentration of 370 ± 40 or 560 ± 40 µmol mol−1 for control and FACE, respectively. The leaf cuvette was held in the horizontal position and caution was used not to shade any portion of the leaf. One leaf per row from each of three different rows was randomly subsampled in the area of each subplot (Ottman et al., 2001) designated for gas exchange studies. Three separate portable closed-exchange systems were used, which enabled all three replications to be measured simultaneously, thereby minimizing variation in gas exchange measurements due to diurnal changes in meteorological conditions, particularly incident photon flux density (PPFD: µmole (photons) m−2 s−1), Ta and water vapor pressure deficit (e*). Individual runs were completed within 45 min of solar noon (1.5 h sample interval). Three observations were recorded at 10-s intervals (total measurement time, therefore, was approximately 55 s, which minimized leaf cuvette effects on gas exchange rate measurements). The first 10-s measurement interval was discarded from the statistical analysis because Ca in the assimilation chamber was unstable during that period. Consequently, each mean datum was derived from 3 leaves × 2 observation × 3 replications (i.e., means based on n = 18, standard error of replication mean based on n = 3).

Each gas exchange system made direct measurements of Ca, atmospheric water vapor pressure (ea), Tl, Ta and incident PPFD normal to the leaf surface. We calculated leaf A, gs, T, saturation water vapor pressure (es), water vapor pressure deficit (i.e., e* = es – ea) and Ci as suggested by Li-Cor (1990) following equations of von Caemmerer & Farquhar (1981). Because equations used to calculate Ci are more applicable to C3 than C4 plants, we assumed that calculated values of Ci were more representative of substomatal CO2 concentration in mesophyll tissue (Cm) rather that at the site of carboxylation by Rubisco in bundle sheath cells (von Caemmerer, 2000). Consequently, hereafter Ci will be referred to as Cm. A value for IWUE was derived as the ratio of A to gs, whereas ΔT was derived by taking the difference between Tl and Ta.

Measurements of total plant water potential and relative water content

Blades of uppermost fully expanded sunlit leaves were excised approx. 5 mm apical to the leaf collar at midday (solar noon) at weekly intervals and sealed in a plastic bag containing a damp paper towel. Plastic bags containing excised leaves were then stored in an insulated container containing ice. No direct contact occurred between the ice and leaf samples. Before inserting the leaf into a pressure chamber (Scholander et al., 1965), leaf blade tissue was removed from each side of the midrib. The ψW was measured on the midrib with a pressure chamber (Model 3000, Soil Moisture Equipment Corp, Santa Barbara, CA, USA). This protocol has given reliable results when precautions are taken to minimize error due to rapid water loss (Turner & Long, 1980).

Leaf disks (c. 15 mm diameter) were excised with a cork borer (Cork Borer Set – 12 Piece, Model No. 9664, Humboldt Manufacturing, Chicago, IL, USA) from the leaf material removed from each side of the midrib. Six leaf disks were placed in a test tube and weighed to obtain the fresh weight (f. wt). Afterwards, 100 ml of distilled water was placed in the test tubes and the leaf disks were rehydrated to obtain the 100% hydrated weight (f. wt(100)). The leaf disks were than dried in an oven at about 65°C for 2–4 d and weighed to obtain the dry weight (d. wt). The RWC was calculated as follows: RWC = ((f. wt − d. wt)/(f. wt(100) − d. wt)) (Barrs & Weatherley, 1962). Treatment means for ψW and RWC were derived from three to four subsamples for four replication (i.e., means based on n = 12 or n = 16, standard error of replication mean based on n = 3 or n = 4, respectively).

Statistical analysis

Data were analysed as a strip-split-plot design using the SAS ‘Mixed’ Procedure (Littell et al., 1996) for the ANOVAs with CO2 (C) as the main plot and irrigation (I) as the strip-split plot (Ottman et al., 2001). A third factor in the ANOVA was soil dehydration (D) cycles. Two soil dehydration cycles occurred for the dry treatments, whereas the wet plots remained hydrated throughout the experiment. Nevertheless, both dry and wet plots were classified as the second dehydration cycle following the mid-season irrigation on DAP 57 during 1998 and DAP 52 during 1999. Another effect in the ANOVA was year (Y). Growth stage (G) was treated as a repeated measure specifying first order, autoregression correlation for the covariance structure. Because growth stages were different between soil dehydration cycles each year, an ANOVA to determine the effect of G on C, I and C×I interaction effects was performed separately for each year and soil dehydration cycle. In order to avoid pseudo-replication all ANOVAs were performed on replication means.

Results

Overall, C, I and C×I interaction effects were consistent across G during the first soil dehydration cycle during both years. This probably occurred because of the narrow range in growth stage, between seedling growth through growing point differentiation (growth stages 1–3 during 1998, growth stages 2–3 during 1999; Vanderlip, 1993), and because only mild water stress occurred during this period. Even though a much broader range in growth stages, from final leaf in the whorl through soft dough (growth stages 4–7 during 1998 and 1999; Vanderlip, 1993), and greater soil water deficits occurred during the second soil dehydration cycle, the effect of G on C, I and C×I interaction effects were also consistent. Over the 2-yr study, the greatest effect of G was observed in the latter stages of growth in the second soil dehydration cycle during 1999 when soil-water content became the most depleted (Fig. 1c). During this period the affect of G on C×I interaction effects were more prevalent; particularly, for gs (Fig. 2d) and ψW (Fig. 4b).

Figure 2.

Mean midday (solar noon) net assimilation rate (A) (a,b), stomatal conductance (gs) (c,d), and intrinsic water use efficiency (IWUE: A/gs) (e,f) for fully expanded sunlit sorghum leaves during 1998 (a,c,e) and 1999 (b,d,f). Vertical dashed lines separate the first from the second soil dehydration cycle. (a) through (f) Control-dry (CD), closed triangles; FACE-dry (FD), closed sqaures; control-wet (CW), open triangles; FACE-wet (FW), open squares. Growth stage during measurements are described as follows: (0) emergence; (1) collar of third leaf visible; (2) collar of fifth leaf visible; (3) growing point differentiation; (4) final leaf in whorl; (5) boot; (6) half-bloom; (7) soft dough; (8) hard dough; (9) physiological maturity (Vanderlip, 1993). Because drought stress delayed phenological development (Ottman et al., 2001), growth stages are given for dry : wet treatments; particularly, during the second compared with the first soil dehydration cycle and in 1999 compared with 1998. Source of variance in ANOVA include carbon dioxide effect (C: control (ambient at 370 µmol mol−1) compared with FACE (c. 200 µmol mol−1 above ambient)) for main plot and irrigation effect (I: dry (postplant and mid-season irrigations) compared with wet (100% replacement of evapotranspiration)) for split-plot, and C×I interaction effect, in a strip split-plot experimental design. Each mean datum was derived from 3 leaves × 2 observations × 3 replications (n = 18). Vertical bar is one standard error from replication mean (n = 3). Results of ANOVA performed for each sampling date given above each datum as three asterisks, two asterisks, one asterisk, and ns for P≤ 0.01, P≤ 0.05, P≤ 0.10, and not significant, respectively.

Because there was no evidence that G was a significant factor, a four-way ANOVA across Y, C, I and D was performed. All results from this ANOVA (Y×C×I×D) given below follow a similar order. Higher order interactions will be discussed prior to lower order. Across all parameters the I effect predominated over the C effect, but of most interest was the C effect under dry compared with that under wet.

Atmospheric conditions

Midday PPFD, Ta and e* were obtained from the Arizona Meteorological Network station at Maricopa. During this 2-yr study this semiarid dessert region had predominantly clear skies, but because sorghum was planted 31 d later in 1998 than 1999 trends in PPFD, Ta and e* were different between years. At planting during 1998, PPFD had peaked at 2000 (µmol (photons) m−2 s−1), Ta had peaked at 39°C and e* had peaked at 4.2 kPa. At physiological maturity PPFD had diminished to 1350 (µmol (photons) m−2 s−1), Ta to 20°C and e* to 1.7 kPa. By contrast, during 1999, the earlier planting date resulted in a much narrower range in PPFD (2000–1600 µmol (photons) m−2 s−1), Ta (28–39°C) and e* (3.0–4.7 kPa) from planting until physiological maturity. Consequently, temporal trends in accumulated thermal time (Vanderlip, 1993) and subsequent phenological events differed between years (Ottman et al., 2001).

Soil-water content and evapotranspiration

The θs was maintained near field capacity in the wet plot, but became progressively lower in the dry plots with depletion of soil-water (Fig. 1a,c). The irrigation strategy, therefore, provided large enough differences in θs between dry and wet treatments for a comparative study. During 1998, before physiological maturity, two distinct soil dehydration cycles occurred for the dry plots (Fig. 1a,c). The first one began after the initial irrigation on 27 July and ended after the mid-season irrigation on 11 Sept. (DAP 11–56). The second soil dehydration cycle began after the mid-season irrigation and ended at physiological maturity during 1998 (DAP 57–166). During 1999, the first soil dehydration cycle began on 28 June and ended on 6 Aug. (DAP 13–51), whereas the second soil dehydration cycle began after the mid-season irrigation and ended at physiological maturity (DAP 52–82). Rainfall was negligible during 1998 (< 20 mm), but as much as 153 mm of rainfall occurred during 1999 (Fig. 1b,d), particularly between DAP 91–101 (Fig. 1d). Because rainfall significantly rehydrated the soil, a third soil dehydration cycle was observed after DAP 101 during 1999. Nevertheless, hail damage to sorghum leaves during these rainfall events precluded any further measurements of gas exchange or water relation parameters (Fig. 1d).

A significant Y×D×I interaction effect occurred (Table 2) because dry reduced θs more than wet in the second compared with the first dehydration cycle during 1999 (severe water stress) compared with 1998 (moderate water stress). A significant I effect occurred because θs was significantly lower by 0.05 m3 m−3 (23%) in dry compared with wet (Fig. 1a,c; Tables 1 and 2). Dry significantly reduced ET by 2.93 mm d−1 (44%) compared with wet (Tables 1 and 2), but FACE caused an insignificant reduction in ET by 0.43 mm d−1. Dry significantly reduced ETc by 226 mm (39%) compared with wet. An insignificant decrease in ETc by 36 mm occurred in FACE compared with control.

Table 2.  ANOVA for carbon dioxide (C: control (370 µmol mol−1); FACE (ambient + 200 µmol mol−1)), irrigation (I: dry (postplant and midseason irrigations); wet (100% replacement of evapotranspiration)), Year (Y: 1998, 1999), and soil dehydration cycle (D: first (seedling growth through growing point differentiation; growth stages 1–3 during 1998, growth stages 2–3 during 1999), second (final leaf in the whorl through soft dough; growth stages 4–7 during 1998 and 1999, Vanderlip, 1993)) effects on three edaphic, eight gas exchange, two water relations, and eight growth parameters for grain sorghum (Sorghum bicolor (L.) Möench cv. Dekalb Hybrid DK54)
SourcedfEdaphicGas exchangeWater relationsGrowth
θsETETcgsAIWUETTlΔT CmC: CaRWCψWLAIaBaSLWLAIpBYLDHIDr
  1. θs, volumetric soil-water content; ET, evapotranspiration rate; ETc, cumulative evapotranspiration; gs, stomatal conductance; A, net assimilation rate; IWUE, intrinsic water use efficiency (A/gs); T, transpiration rate; Tl , leaf temperature; ΔT, leaf minus air temperature; Cm, internal CO2 concentration of mesophyll tissue; Cm : Ca, ratio of Cm to atmospheric CO2 concentration (Ca); RWC, relative water content; ψW , total plant water potential; LAIa , average leaf area index; Ba, average accumulated shoot biomass; SLW, specific leaf weight; LAIp, peak leaf area index; B, final shoot biomass; YLD, yield; HI, harvest index (YLD/B); Dr, maximum depth of root penetration. ANOVA on gs, IWUE, Cm : Ca, SLW, and HI performed on Log(10) transformation. (***, P ≤ 0.01; **, P ≤ 0.05; *, P ≤ 0.10; ns, not significant, respectively. Actual probability of a greater F-value by chance reported if P ≥ 0.10 and P ≤ 0.25). Source of variance in ANOVA (error term) include C, I, D, and Y effects: C (R*C), I (R*I), C*I (R*C*I), Y (R×Y), Y×C (R×Y×C), Y×I (R×Y×I), Y×C×I (R×Y×C×I), D (R×D), D×C (R×D×C), D×I (R×D×I), DxCxI (R×D×C×I), Y×D (R×Y×D), Y×D×C (R×Y×D×C), Y×D×I (R×Y×D×I), Y×D×C×I (R×Y×D×C×I).

C1nsns0.18******0.18*****nsns***nsnsnsns*nsnsns
I1******************************ns************************0.22
C×I10.25ns0.230.25nsnsnsnsnsnsnsns*nsns0.20nsns0.12*ns
Y1****************0.17*********************ns*************** 
Y×C 1nsnsnsnsnsnsnsnsns0.13nsnsnsnsnsnsnsns*ns 
Y×I1nsnsnsnsns**nsnsns******nsns0.180.12*********** 
Y×C×I 1nsnsnsnsnsnsnsnsns0.18nsnsnsnsnsnsns**nsns 
D1***ns***0.14***ns*******nsns*************     
D×C 1nsnsns0.25ns0.13nsnsns0.180.11nsnsnsnsns     
D×I1*ns***0.200.23***ns****ns0.22*****ns     
D×C×I 1nsnsnsnsnsnsnsnsnsnsnsnsnsnsnsns     
Y×D 1ns0.18***ns*****nsnsnsns0.25*****     
Y×D×C 1nsnsnsnsnsnsnsnsnsnsnsnsnsnsnsns     
Y×D×I1***nsns***ns*ns*nsnsns***0.21**ns     
Y×D×C×I 1nsnsnsnsnsnsnsnsnsnsnsnsnsnsnsns     
Table 1.  Means ± standard error (SE) across both years and soil dehydration cycles for sorghum (Sorghum bicolor (L.) Möench cv. Dekalb Hybrid DK54) grown under two carbon dioxide (control: C (370 µmol mol−1), FACE: F (ambient + 200 µmol mol−1)), and two irrigation (dry: postplant and midseason irrigations, wet: 100% replacement of evapotranspiration) levels during 1998 and 1999. Edaphic parameters include volumetric soil-water content (θs), evapotranspiration rate (ET), and cumulative evapotranspiration (ETc) obtained from Conley et al. (2001). Fully expanded upper canopy sunlit leaf gas exchange paramaters include stomatal conductance (gs), net assimilation rate (A), intrinsic water use efficiency (IWUE: A/gs), transpiration rate (T), leaf temperature (Tl), leaf minus air temperature (ΔT), internal CO2 concentration in mesophyll tissue (Cm), ratio of Cm to atmospheric CO2 concentration (Ca) (C: Ca), relative water content (RWC), and total plant water potential (ψW). Growth parameters include average leaf area index (LAIa), average accumulated shoot biomass (Ba), specific leaf weight (SLW), peak leaf area index (LAIp), accumulated shoot biomass (B), yield (YLD), and harvest index (HI: YLD/B) obtained from Ottman et al. (2001). The maximum depth of root penetration (Dr) was estimated from θs by depth obtained from Conley et al. (2001). Also shown are the ratios of the FACE : control CO2 and dry : wet irrigation treatments.
ParameterIrrigationControlFACERatio F : C
  • Leaf (Tl) and air (Ta) temperatures measured in cuvette. Edaphic (θs, ET, and ETc) parameters treatment mean and standard error based on replication mean (n = 4). Gas exchange (gs, A, IWUE, T, Tl , ΔT, Cm, and Cm/Ca) parameters treatment mean and standard error based on replication mean (n = 3). Water relations (RWC and ψW) parameters treatment mean and standard error based on replication mean (n = 4). Growth (LAIa, Ba, SLW, LAIp, B, YLD, HI, and Dr) parameters treatment mean and standard error based on replication mean (n = 4).

θs (m3 m−3)Dry0.15 ± 0.010.15 ± 0.010.000
 Wet0.20 ± 0.010.19 ± 0.010.950
 Dry : Wet0.7500.789 
ET (mm d−1)Dry3.84 ± 0.413.65 ± 0.410.951
 Wet7.01 ± 0.416.34 ± 0.410.904
 Dry : Wet0.5480.576 
ETc (mm)Dry361 ± 26348 ± 260.964
 Wet611 ± 26552 ± 260.903
 Dry : Wet0.5910.631 
gs mol (H2O) m−2 s−1Dry0.37 ± 0.050.25 ± 0.050.676
 Wet0.59 ± 0.050.37 ± 0.050.627
 Dry : Wet0.6270.676 
A µmol (CO2) m−2 s−1Dry26.80 ± 2.5533.04 ± 2.551.23
 Wet38.79 ± 2.5142.09 ± 2.511.09
 Dry : Wet0.6210.785 
IWUE µmol (CO2) mol (H2O)−1Dry101 ± 7159 ± 71.59
 Wet84 ± 6132 ± 61.57
 Dry : Wet1201.21 
T mmol (H2O) m−2 s−1Dry12.68 ± 1.5411.07 ± 1.540.873
 Wet18.29 ± 1.5115.08 ± 1.510.824
 Dry : Wet0.6930.734 
Tl (°C)Dry38.52 ± 0.5739.48 ± 0.58 
 Wet37.13 ± 0.5638.58 ± 0.56 
 Dry : Wet   
ΔT (°C)Dry−0.28 ± 0.240.00 ± 0.23 
 Wet −1.48 ± 0.24−0.88 ± 0.23 
 Dry : Wet   
Cm (ppm)Dry168 ± 9271 ± 91.61
 Wet178 ± 9297 ± 91.67
 Dry : Wet0.9430.912 
Cm:CaDry0.49 ± 0.020.48 ± 0.020.980
 Wet0.53 ± 0.020.53 ± 0.021.00
 Dry : Wet0.9250.906 
RWC (%)Dry82.13 ± 1.0482.22 ± 1.021.01
 Wet82.37 ± 1.0384.32 ± 1.041.02
 Dry : Wet0.9970.987 
ΨW (MPa)Dry −1.82 ± 0.04 −1.66 ± 0.040.912
 Wet −1.43 ± 0.04 −1.39 ± 0.040.972
 Dry : Wet1.281.19 
LAIaDry2.39 ± 0.052.70 ± 0.111.130
 Wet4.07 ± 0.183.91 ± 0.120.961
 Dry : Wet0.5870.691 
Ba (g m−2)Dry 604 ± 17653 ± 261.081
 Wet977 ± 26963 ± 130.986
 Dry : Wet0.6180.678 
SLW (g m−2)Dry66 ± 6.156 ± 1.90.862
 Wet51 ± 1.554 ± 2.01.059
 Dry : Wet1.2941.037 
LAIpDry5.94 ± 0.625.92 ± 0.720.997
 Wet 6.95 ± 0.736.91 ± 0.430.994
 Dry : Wet0.8550.857 
B (g m−2)Dry999 ± 631151 ± 711.152
 Wet1560 ± 101604 ± 431.028
 Dry : Wet0.6400.718 
YLD (g m−2)Dry289 ± 35348 ± 291.200
 Wet573 ± 12 550 ± 160.960
 Dry : Wet0.5040.633 
HIDry0.262 ± 0.050.280 ± 0.051.069
 Wet0.368 ± 0.030.341 ± 0.030.927
 Dry : Wet0.7110.821 
Dr (m)Dry1.78 ± 0.171.93 ± 0.171.084
 Wet 2.03 ± 0.172.15 ± 0.171.059
 Dry : Wet0.8770.90 

Stomatal conductance, net assimilation rate and intrinsic water use efficiency

A significant Y×D×I interaction effect occurred because dry reduced gs more than wet in the second compared with the first dehydration cycle during 1999 compared with 1998 (Fig. 2c,d; Table 2). Although the overall C×I interaction effect on gs was minor (P = 0.25; Table 2), significant C×I interactions occurred on individual days, especially during the second dehydration cycle during 1999 (Fig. 2d) compared with 1998 (Fig. 2c). Compared with control, FACE reduced seasonal average gs by 0.12 (32%) and 0.22 (37%) mol (H2O) m−2 s−1 in dry and wet, respectively (Table 1; Fig. 2c,d). Dry reduced gs by a similar amount of 0.17 mol (H2O) m−2 s−1 (35%) (Fig. 2c,d; Table 1).

By contrast to trends in gs, a significant Y×D×I interaction effect occurred because A was proportionately greater in dry than wet in the second compared with the first dehydration cycle in 1999 compared with 1998 (Fig. 2a,b; Table 2). In dry, A was 10.50 µmol (CO2) m−2 s−1 (26%) lower than wet. Compared with control, FACE increased A by 4.77 µmol (CO2) m−2 s−1 (Table 1). Dry significantly increased IWUE by 22 µmol (CO2) mol (H2O) (20%) compared with wet (Tables 1 and 2). Regardless of irrigation level, FACE increased the seasonal average IWUE by 53 µmol (CO2) mol (H2O) (57%) compared with control (Fig. 2e,f; Table 1).

Transpiration rate and leaf temperature

Sorghum was planted 31 d later in 1998 than 1999. Consequently, meteorological conditions were different between growing seasons, and trends in Tl and ΔT (Fig. 3) reflect these differences between years. Nevertheless, the effects of I and C on T, Tl and ΔT were relatively consistent between years.

Figure 3.

Mean midday (solar noon) leaf temperature (Tl) (a,b) inside cuvette, and difference (ΔT) between leaf and air temperature (Ta) inside cuvette, for fully expanded sunlit sorghum leaves during 1998 (a,c) and 1999 (b,d). Vertical dashed lines separate the first from the second soil dehydration cycle. (a) through (d) Control-dry (CD), closed triangles; FACE-dry (FD), closed sqaures; control-wet (CW), open triangles; FACE-wet (FW), open squares. Growth stage during measurements are described as follows: (0) emergence; (1) collar of third leaf visible; (2) collar of fifth leaf visible; (3) growing point differentiation; (4) final leaf in whorl; (5) boot; (6) half-bloom; (7) soft dough; (8) hard dough; (9) physiological maturity (Vanderlip, 1993). Because drought stress delayed phenological development (Ottman et al., 2001), growth stages are given for dry : wet treatments; particularly, during the second compared with the first soil dehydration cycle and in 1999 compared with 1998. Source of variance in ANOVA include carbon dioxide effect (C: control (ambient at 370 µmol mol−1) compared with FACE (c. 200 µmol mol−1 above ambient)) for main plot and irrigation effect (I: dry (postplant and mid-season irrigations) compared with wet (100% replacement of evapotranspiration)) for split-plot, and C×I interaction effect, in a strip split-plot experimental design. Each mean datum was derived from 3 leaves × 2 observations × 3 replications (n = 18). Vertical bar is one standard error from replication mean (n = 3). Results of ANOVA performed for each sampling date given above each datum as three asterisks, two asterisks, one asterisk, and ns for P≤ 0.01, P≤ 0.05, P≤ 0.10, and not significant, respectively.

Because gs and T are highly correlated, seasonal trends in T and statistical results (data not shown) were similar to those reported for gs (Fig. 2c,d). A significant Y×D×I interaction effect occurred because T was lower in dry than wet during the second dehydration cycle compared with the first during 1999 compared with 1998 (Table 2). Dry reduced T by 4.81 mmol (H2O) m−2 s−1 (29%) compared with wet (Fig. 2e,f; Table 1). A reduction in T of 2.42 mmol (H2O) m−2 s−1 (16%) occurred in FACE compared with control (Table 2; P = 0.18). A significant Y×D×I interaction effect occurred because transpiration cooling was decreased more in the second compared with the first dehydration cycle, which increased ΔT (Table 2; Fig. 3c,d) more in dry than wet during 1999 compared with 1998. Although insignificant, leaf temperature was increased by 1.15°C in dry compared with wet, which was similar to the 1.20°C increase in Tl in FACE compared with control (Fig. 3a,b; Table 1). Compared with wet, ΔT was 0.14°C greater in dry (Fig. 3c,d; Table 1), whereas ΔT was greater by 0.44°C in FACE compared with control.

Relative water content and leaf water potential

Of all the parameters evaluated ψW was the most responsive to C and I treatments (Fig. 4a,b; Tables 1 and 2), whereas RWC was the least responsive (Fig. 4c,d; Tables 1 and 2). A significant Y×D×I interaction effect occurred because ψW was proportionately greater in dry than wet in the second compared with the first dehydration cycle in 1999 compared with 1998 (Fig. 4a,b; Table 2). The ψW was more negative by 0.33 (23%) MPa in dry compared with wet, but a significant C×I interaction effect occurred because ψW was 0.16 MPa (9%) and 0.04 MPa (3%) less negative in FACE than control for dry and wet, respectively. These C×I interaction effects for ψW were most prevalent when water stress was the most severe at the end of the second soil dehydration cycle during 1999 (Fig. 4b).

Figure 4.

Mean midday (solar noon) total water potential (ψW) (a,b) and relative water content (RWC) (c,d) for fully expanded sunlit sorghum leaves during 1998 (a,c) and 1999 (b,d). Vertical dashed lines separate the first from the second soil dehydration cycle. (a) through (d) Control-dry (CD), closed triangles; FACE-dry (FD), closed sqaures; control-wet (CW), open triangles; FACE-wet (FW), open squares. Growth stage during measurements are described as follows: (0) emergence; (1) collar of third leaf visible; (2) collar of fifth leaf visible; (3) growing point differentiation; (4) final leaf in whorl; (5) boot; (6) half-bloom; (7) soft dough; (8) hard dough; (9) physiological maturity (Vanderlip, 1993). Because drought stress delayed phenological development (Ottman et al., 2001), growth stages are given for dry : wet treatments; particularly, during the second compared with the first soil dehydration cycle and in 1999 compared with 1998. Source of variance in ANOVA include carbon dioxide effect (C: control (ambient at 370 µmol mol−1) compared with FACE (c. 200 µmol mol−1 above ambient)) for main plot and irrigation effect (I: dry (postplant and mid-season irrigations) compared with wet (100% replacement of evapotranspiration)) for split-plot, and C×I interaction effect, in a strip split-plot experimental design. Each mean datum was derived from 4 subsamples and 4 replication (n = 16). Vertical bar is one standard error from replication mean (n = 4). Results of ANOVA performed for each sampling date given above each datum as three asterisks, two asterisks, one asterisk, and ns for P≤ 0.01, P≤ 0.05, P≤ 0.10, and not significant, respectively.

Growth response

Compared with wet, dry significantly reduced LAIa by 1.45 (36%), LAIp by 1.0 (14%), but increased SLW by 8.2 g m−2 (16%) (Tables 1 and 2) (Ottman et al., 2001). But, FACE did not significantly affect LAIa, LAIp or SLW (Ottman et al., 2001). Dry significantly reduced Ba by 342 g m−2 (35%) compared with wet, but FACE had no significant affect on Ba compared with control (Ottman et al., 2001). Final shoot biomass was reduced by 507 g m−2 (33%) in dry compared with wet. A significant increase in B by 98 g m−2 occurred in FACE compared with control (Ottman et al., 2001). Dry significantly reduced YLD by 243 g m−2 (43%) compared with wet, but compared with control, FACE did not significantly increase YLD (Ottman et al., 2001). A significant reduction in HI of 0.083 (23%) was observed in dry compared with wet, but compared with control, FACE had no significant effect on HI. Neither I nor C effects on Dr were detected (Tables 1 and 2).

Discussion

One very significant direct effect of an approx. 200 µmol mol−1 increase in atmospheric CO2 concentration was a reduction in gs by 0.17 mol (H2O) m−2 s−1 (36%), which was similar to the difference observed between sorghum grown under water stress compared with ample soil-water supply. On a relative basis, however elevated CO2 caused a greater reduction in gs under ample compared with reduced soil-water supply (32% for dry, 37% for wet) (accept Hypothesis 1).

Seasonal trends in θs were different between sorghum grown with adequate and reduced water supply under ambient and elevated CO2. In the wet plots, part of this difference resulted from a systematic 1% lower θs in elevated CO2 compared with ambient. This was presumed to occur mostly because of differences in soil texture (greater clay content and less sand in FW than CW; data not shown). No soil textual characteristic effects on θs were presumed to occur in the dry plots. Because soil substrate properties such as texture and water holding capacity were inherently variable, differences in whole-canopy ET between 5 and 90 mm (1–20%) derived from θs were difficult to discern statistically (Conley et al., 2001). Nevertheless, despite a lack of statistical significance for C×I interaction in ET, and ETc, relative differences can be used to make inferences about treatment effects on ET and ETc. Based on a root depth adjusted step-wise ET model, elevated CO2 reduced calculated values of ETc by 36 mm (7%) – FACE reduced ETc by 59 mm (10%) when plants were given ample water and by 13 mm (4%) under drought conditions (Conley et al., 2001). These results were consistent with those from individual leaves where elevated CO2 reduced transpiration rate by 2.42 mmol (H2O) m−2 s−1 (16%). A CO2-based reduction in gs, therefore, caused enough of a reduction in both individual leaf and whole-canopy water use to reduce season-long consumption for a well-watered sorghum crop. But, reductions in gs because of elevated CO2 had less of an effect on season-long consumptive water use for water-stressed sorghum. Presumably, a reduction in consumptive water use because of elevated CO2 resulted in higher levels of available soil-water for a longer period into the soil dehydration cycle (support Hypothesis 2). These results parallel those of Hunsaker et al. (1996, 2000) for wheat grown in an open field under FACE and for a tallgrass prairie grass, big bluestem, grown in elevated CO2 in open-topped chambers (Kirkham et al., 1991).

A decrease in transpiration cooling because of elevated CO2 resulted in an increase in leaf temperatures (accept Hypothesis 3). An increase in individual leaf temperature contributes to an increase in whole-canopy temperature in sorghum (Chaudhuri et al., 1986b). Although drought stress has been known to decrease phenological development in sorghum (Chaudhuri et al., 1986a; Bremner et al., 1986; Donatelli et al., 1992; Crawford et al., 1993;1998), an increase in leaf temperature would tend to accelerate phenology. However, increased leaf temperatures, because of elevated CO2, had only a nominal effect on phenology. During vegetative growth stages, elevated CO2 slowed growth of the dry, but accelerated it for the wet plots. By contrast, during grain filling the inverse occurred (Ottman et al., 2001). Despite the lack of a consistent response in phenology, any increase in leaf temperature should affect leaf energy balance (Kimball et al., 1999; Allen, 1999). Because an increase in leaf temperature will increase the leaf-to-air vapor pressure gradient, a feedback should occur between transpiration rate and leaf temperature that would affect the direct effect of elevated CO2 on stomatal conductance. Perhaps, this could explain why an c. 200 µmol mol−1 increase in CO2 caused a 35% reduction in stomatal conductance (P < 0.05, Table 2), but only a 15% (P = 0.18; Table 2) reduction in transpiration. Morison & Gifford (1994a,b) also reported that because of leaf tissue warming, and greater leaf area in wheat grown under elevated CO2, the reduction in transpiration rate was proportionately less than reductions in stomatal conductance. Furthermore, Long (1991) reported that even a modest increment in leaf temperature can increase photosynthesis of C4 plants. Consequently, the observed increase in leaf temperature could have contributed to the season-long increase in carbon gain.

As observed in other C4 species, elevated CO2 stimulated photosynthetic activity in sorghum (Sionit & Patterson, 1984; Kirkham et al., 1991; Knapp et al., 1993). But, environmental factors that have been shown to alter the photosynthetic response of sorghum and the expression of the C4 phenotype were also observed in this study. Adam et al. (1999) reported that elevated CO2 caused a strong reduction in the initial slope of the A/Ci curve. However, this reduction in carboxylation efficiency in sorghum was limited to younger leaves. They concluded that any alteration in carboxylation in sorghum was dependent on leaf age and growth stage. Young leaves are more responsive to elevated CO2 because they have an immature C4 pathway that are C3-like, thereby more responsive to elevated CO2 (Sionit & Patterson, 1984; Poorter et al., 1996). In a companion study, Cousins et al. (2001) demonstrated that elevated CO2 stimulated C4 photosynthesis in young leaves by suppressing photorespiration and enhancing energy-use efficiency. They postulated that elevated CO2 may have decreased both CO2 leakage from bundle sheath cells and overcycling of the C4 pump. In yet another companion study, Williams et al. (2001) demonstrated that 49 d-old sorghum plants grown in elevated CO2 and ample water supply had no apparent change in bundle cell leakiness, but that drought stress increased leakiness. They also reported that the ameliorating effect of elevated CO2 in alleviating drought (Wall, 2001), minimized the deleterious affects of tissue desiccation on bundle sheath leakiness and δ13C discrimination. This suggest that the direct effect of CO2 enrichment on metabolism of C4 photosynthesis in sorghum is nominal, whereas the indirect effect of improved plant water status by elevated CO2 predominates. For well-watered 45- to 50-d-old sorghum plants grown in plant growth chambers maintained at 350 and 700 µmol (CO2) mol−1, and at 800 µmol (photons) m−2 s−1 at plant height (less than half full sunlight), Watling et al. (2000) demonstrated that elevated CO2 caused modifications in the C4 pathway at both anatomical and metabolic levels. Metabolic changes, because of elevated CO2, included a decrease in carboxylation efficiency and the CO2 saturated rate of photosynthesis, a 50% reduction in the leaf concentration of phospoenolpyruvate carboxylase, but no change in Rubisco content or quantum yield. Anatomical changes, because of elevated CO2, included a twofold decrease in bundle sheath cell wall thickness, which increased bundle sheath leakiness by 9%. Ghannoum et al. (2000), however, reported that any improvement in net assimilation rate and growth in C4 plants because of elevated CO2 will most likely occur because of improved water relations (stomatal limitation) rather than any significant modification in anatomical (bundle sheath leakiness) or biochemical (enzyme kinetics, CO2 saturation, reduced photorespiration, acclimation) responses (nonstomatal limitation).

Our results demonstrated that the stimulatory effect of elevated CO2 on carbon gain was also greatest for younger tissue early in the season. It also indicated that on a relative basis the photosynthetic response to elevated CO2 was proportionately greater in dry than wet (accept Hypothesis 4). Nevertheless, elevated CO2 also caused a modest increase in net assimilation rate in older tissue later in the season. Hence, elevated CO2 caused an increase in season-long carbon gain for sorghum, particularly under drought conditions.

Because elevated CO2 increased drought avoidance by conserving soil-water, less internal water deficits occurred as the soil-water content became depleted. Hence, elevated CO2 improved water relations. This was evidenced by a slight increase in leaf RWC, but more so by less negative ψW for sorghum grown in elevated CO2. Furthermore, improved water relations because of elevated CO2 became more pronounced as soil-water content was depleted (accept Hypothesis 5). Improved water relations would tend to decrease the deleterious effects of water stress on physiological process such as photosynthesis and enhance growth (Hsiao & Jing, 1987).

Under drought conditions xeromorphic adaptations such as thicker cuticles, mesophyll, and epidermal cells, and higher trichome density results in higher SLW (Kramer, 1983). In this field study, an 8.2-g m−2 (16%) increase in SLW occurred in dry compared with wet plots. Such an increase in SLW suggests that drought adaptations occurred in water-stressed sorghum leaves, which increased thickness and/or the concentration of total nonstructural carbohydrates. In a FACE wheat study, Estiarte et al. (1999) showed that elevated CO2 increased total non-structural carbohydrates in leaves at tillering, stem-elongation, anthesis, and soft dough. Consequently, we believe that an increase in carbohydrate supply because of elevated CO2 enhanced adaptation of leaf tissue, which minimized water loss (accept Hypothesis 5) (Turner & Kramer, 1980; Kramer, 1983).

An increase in carbon gain because of elevated CO2 has been correlated with enhanced root growth in sorghum (Chaudhuri et al., 1986a), wheat (Wechsung et al., 1995, 1999), and other grass species (Mo et al., 1992). Regardless of irrigation level, however, only a 1% (12–15 mm) increase in the depth of root penetration was observed because of elevated CO2 (reject Hypothesis 6 for root depth). Reduced water supply, however, decreased root depth by 12% (24 mm). Nonetheless, Chaudhuri et al. (1986a) reported that elevated CO2 enhanced morphological characteristics of sorghum roots that increased root mass, length density, surface area, and root linear density suggesting that changes in root structure and composition occurred. Changes in root lineal density are related to more compact or denser tissue and alteration in carbohydrate storage, cell number, cell size, and other structural modifications such as increased stele diameter, cortex width, and root diameter in the root hair zone (Rogers et al., 1992). Soil cores taken at the peak of root growth to a 1.2-m depth provided visual evidence of a more robust root system for sorghum grown in elevated compared with ambient CO2 (G. W. Wall, unpublished). Consequently, we believe that the root systems of sorghum grown under elevated CO2 had an increased capacity to extract available soil-water content (accept Hypothesis 6 for root morphology).

Morison & Gifford (1984a,b) postulated that during soil dehydration both stomatal aperture and leaf area expansion would continuously adjust in relation to soil-water content. But, Samarakoon & Gifford (1995) demonstrated that variation occurred between the counterbalancing effect of stomatal conductance and leaf area with soil-water content. In sorghum, a 35% reduction in stomatal aperture, because of elevated CO2, reduced the season-long cumulative evapotranspiration by 13 (4%) and 59 (11%) mm for dry and wet, respectively. Despite the effect of elevated CO2 on water use, however, wet plots increased season-long cumulative evapotranspiration by 227 mm (40%) compared with dry. Samarakoon & Gifford (1995) reported that net assimilation rate, derived from biomass accumulation, of corn (Zea Mays L.) suggested that a strong stimulation by elevated CO2 increased biomass production and leaf area in dry, but not in wet plots. In agreement, our results demonstrated that elevated CO2 and water stress increased average leaf area index by 13%, final shoot biomass by 15%, and marketable yield by 20% in sorghum (Ottman et al., 2001). Elevated CO2, however, had virtually no effect on growth under wet conditions (accept Hypothesis 6 for shoot).

In conclusion, simultaneous examination of 21 (three edaphic (θs, ET, and ETc); eight gas exchange (gs, A, IWUE, T, Tl, ΔT, Cm and Cm : Ca); two water relations (RWC and ψW); eight growth (LAIa, Ba, SLW, LAIp, B, YLD, HI, and Dr)) parameters demonstrated that improved water relations for a herbaceous, warm-season, annual, C4 grain crop, sorghum, are anticipated in a future high-CO2 world. This is presumed to occur because elevated CO2 decreased the rate of water loss by transpiration and increased the capacity for water absorption by roots. Because the stimulatory affect of elevated CO2 was greatest under drought conditions, our results also demonstrated the inherent complexity in unraveling the interaction between global change and net primary productivity of sorghum. Hence, the most beneficial effect of a rise in atmospheric CO2 concentration may be a yield enhancement during drought years. Nevertheless, a rise in atmospheric CO2 will minimize the deleterious effects of drought on physiological function and growth, and thereby, expand the range where a viable sorghum crop can be grown, especially in more marginal production areas.

Acknowledgements

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); and, by Grant No. 97–35109–5065 between the USDA, Competitive Grants Program to the University of Arizona (Steven W. Leavitt, PI) as part of the DOE/NSF/NASA/USDA/EPA Joint Program on Terrestrial Ecology and Global Change (TECO III). The research was also supported by Interagency Agreement No. IBN-9652614 between the National Science Foundation and the USDA, Agricultural Research Service (Gerard W. Wall, PI) as part of the NSF/DOE/NASA/USDA Joint Program on Terrestrial Ecology and Global Change (TECO II); and, by the USDA, Agricultural Research Service. 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. Technical assistance of Ms. Laura Olivieri and Mr. Matthew Conley, who assisted in collecting and summarizing results and preparation of the figures, is appreciated.

Ancillary