Soil and Plant Water Relations Determine Photosynthetic Responses of C₃ and C₄ Grasses in a Semi‐arid Ecosystem under Elevated CO₂

Abstract

To model the effect of increasing atmospheric CO₂ on semi‐arid grasslands, the gas exchange responses of leaves to seasonal changes in soil water, and how they are modified by CO₂, must be understood for C₃ and C₄ species that grow in the same area. In this study, open‐top chambers were used to investigate the photosynthetic and stomatal responses of Pascopyrum smithii (C₃) and Bouteloua gracilis (C₄) grown at 360 (ambient CO₂) and 720 µmol mol^–1 CO₂ (elevated CO₂) in a semi‐arid shortgrass steppe. Assimilation rate (A) and stomatal conductance (g_s) at the treatment CO₂ concentrations and at a range of intercellular CO₂ concentrations and leaf water potentials (ψ_leaf) were measured over 4 years with variable soil water content caused by season and CO₂ treatment. Carboxylation efficiency of ribulose bisphosphate carboxylase/oxygenase (V_c,max), and ribulose bisphosphate regeneration capacity (J_max) were reduced in P. smithii grown in elevated CO₂, to the degree that A was similar in elevated and ambient CO₂ (when soil moisture was adequate). Photosynthetic capacity was not reduced in B. gracilis under elevated CO₂, but A was nearly saturated at ambient CO₂. There were no stomatal adaptations independent of photosynthetic acclimation. Although photosynthetic capacity was reduced in P. smithii growing in elevated CO₂, reduced g_s and transpiration improved soil water content and ψ_leaf in the elevated CO₂ chambers, thereby improving A of both species during dry periods. These results suggest that photosynthetic responses of C₃ and C₄ grasses in this semi‐arid ecosystem will be driven primarily by the effect of elevated CO₂ on plant and soil water relations.

INTRODUCTION

Arid and semi‐arid ecosystems occupy 40 % of the Earth’s land surface (Dregne, 1991), so an understanding of the responses of these biomes to increasing atmospheric CO₂ is critical. Water‐limited ecosystems are predicted to have relatively large responses to increasing CO₂ because elevated CO₂ decreases stomatal conductance and leaf transpiration, thereby improving soil water content (Volk et al., 2000). A complicating factor in predicting the responses of grasslands to increasing CO₂ is that many grasslands have co‐occuring C₃ and C₄ photosynthesis species (Wand et al., 1999). Ehleringer et al. (1997) proposed that the global distribution of C₃ and C₄ plants was related to inherent differences in the quantum yield of photosynthesis. In theory, quantum yield is improved in C₃ plants as CO₂ increases, potentially expanding the global distribution of C₃ species. But the influence of elevated CO₂ on photosynthesis may be less important in arid ecosystems where the effects of water may dominate. An understanding of how seasonal and CO₂‐induced variation in soil water affects photosynthetic and stomatal responses of competing C₃ and C₄ species is needed to predict the responses of arid grasslands to increasing CO₂.

Although carbon assimilation is stimulated relatively more in C₃ than in C₄ species when exposed to short‐term increases in CO₂, this increase is often not maintained in plants exposed to elevated CO₂ for weeks or months (Sage, 1994; Moore et al., 1999; Stitt and Krapp, 1999). This phenomenon has been termed ‘photosynthetic acclimation’. The physiological basis for photosynthetic acclimation is not completely understood, but is often related to reduced leaf nitrogen and photosynthetic enzyme activity (Moore et al., 1999; Stitt and Krapp, 1999). Little research has been done on possible stomatal acclimation to elevated CO₂, though reduced stomatal conductance and leaf transpiration may significantly affect ecosystem processes (Drake et al., 1997; Jarvis et al., 1999; Lee et al., 2001).

Field studies are necessary to ascertain the responses of plants to elevated CO₂. There are few field studies in which the effects of elevated CO₂ on grasslands have been investigated, and even fewer studies in which C₃ and C₄ species occur in the same ecosystem (Wand et al., 1999). There was no evidence of photosynthetic acclimation in C₃ or C₄ grasses in a tallgrass prairie (Nie et al., 1992; Knapp et al., 1993), a Texas grassland (Anderson et al., 2001) or in a C₃ annual grassland (Jackson et al., 1995). Conversely, photosynthetic acclimation was reported in both C₃ and C₄ species in a Minnesota grassland, and in C₃ grasses, but not a C₄ grass, in New Zealand (Lee et al., 2001; von Caemmerer et al., 2001). Therefore, field studies suggest that the photosynthetic responses to elevated CO₂ in grasslands are site‐ and species‐dependent. A deficit of resources other than CO₂ (soil N, soil water, rooting volume) can limit the production of new sink tissues when assimilation rates are improved (Drake et al., 1997; Rogers et al., 1998; Lee et al., 2001). This source/sink imbalance causes an accumulation of carbohydrates, which is thought to feed back on photosynthetic processes (Sage, 1994; Stitt and Krapp, 1999).

Growth chamber studies have been used to investigate the photosynthetic and growth responses of a C₃ [Western wheatgrass; Pascopyrum smithii (Rydb.) A. Love] and a C₄ grass [Blue grama; Bouteoloua gracilis (H.B.K.) Lag.] to elevated CO₂ (Morgan et al., 1994a, b; Hunt et al., 1996; Read et al., 1997; LeCain and Morgan, 1998; Morgan et al., 1998). These are major species of the semi‐arid shortgrass steppe (SGS) of the North American Great Plains (Milchunas et al., 1989). Photosynthetic acclimation to elevated CO₂ was seen in both C₃ and C₄ species (Morgan et al., 1994a; Read et al., 1997; LeCain and Morgan, 1998). Despite the relative reduction, photosynthesis rates were slightly higher under elevated vs. ambient CO₂. Growth improvement in both species was attributed to higher rates of photosynthesis and improved soil and plant water status (Morgan et al., 1994,a, b; Read and Morgan, 1996; Morgan et al., 1998).

To increase understanding of the responses of C₃ and C₄ grasses to elevated CO₂ in a water‐limited ecosystem, a field study was performed on the native SGS in north‐eastern Colorado, USA. We hypothesized that photosynthesis of both P. smithii and B. gracilis would acclimate under elevated CO₂, but that there would still be a direct photosynthetic stimulation, especially in P. smithii. Leaf nitrogen, carbohydrates and assimilation at varying intercellular [CO₂] were measured to investigate the physiology of photosynthetic adaptations. We also hypothesized that stomatal conductance and transpiration of leaves would be reduced under elevated CO₂, and that stomatal acclimation would not occur. Measurements were conducted over four seasons (1997–2000) to document the responses to seasonal and CO₂‐induced soil moisture variation.

MATERIALS AND METHODS

The study site is the USDA‐ARS Central Plains Experimental Range, 40°50′N, 104°47′W, in the SGS region of north‐eastern Colorado, USA. Mean annual precipitation (55 years) averaged 320 mm, and air temperatures averaged 15·6 °C in summer and 0·6 °C in winter, with maximum July temperatures of 30·6 °C (Milchunas et al., 1989). The soil at the experimental site is a Remmit fine sandy loam (Ustollic camborthids) to greater than 1 m. This is considered to be a nitrogen‐poor soil (Hunt et al., 1988), with a total nitrogen content of 0·085 % in the upper 30 cm (A. Mosier, pers. comm.).

This experiment was established in spring 1997 on a 6 ha field of native grassland. Although much of the SGS is dominated by B. gracilis (Milchunas et al., 1989), this site was chosen for its nearly equal mix of C₃ and C₄ grasses (Morgan et al., 2001). The field was divided into three blocks, based on vegetation, and three plots per block were chosen at random. From mid‐March until late October in 1997, 1998, 1999 and 2000, large (60·5 m³) open‐top chambers were placed on two plots in each of the three blocks. One chamber received ambient air (approx. 360 µmol mol^–1 CO₂), the other received air with elevated CO₂ (720 µmol mol^–1 CO₂). Each block also included an ‘unchambered control’ plot of equal ground area, which was used to monitor the effect of the chamber.

The chambers were 3·8 m high and 4·5 m in diameter, with plastic walls and top (Lexan; Regal Plastics, Littleton, CO, USA), with a 0·75‐m opening. An aluminium flange was buried to 0·75 m around the chamber wall. The chambers were aspirated at an exchange rate of 1·5 volumes min^–1. The CO₂ concentration was monitored using an infrared gas analyser (LI6262; LI‐COR, Lincoln, NE, USA) and was elevated with pure CO₂ in three of the chambers (720 ± 20 µmol mol^–1). Air temperature (15 cm above the soil) and soil temperature (10 cm depth), measured using thermocouples, light intensity above the plant canopy (LI‐190SA; LI‐COR) and relative humidity at 2 m height (Vaisala HMP45C; Campbell Scientific, Logan, UT, USA) were recorded hourly. Since the chamber top kept out much of the precipitation, rain was captured and returned to the plots using an automatic system. For a more detailed description of the chambers see Morgan et al. (2001). Soil water content (SWC) was measured weekly using a neutron probe (model 4301; Troxler Electronics Lab., Research Triangle Park, NC, USA).

Exchange of CO₂ and H₂O were measured on recently expanded leaves of B. gracilis and P. smithii using a portable gas analysis system [CIRAS‐1with PLC (N) leaf chamber; PP systems, Hitchin, UK]. Steady‐state measurements of CO₂ assimilation rate (A), stomatal conductance to water vapour (g_s), transpiration (E) and intercellular CO₂ concentration (C_i) at cuvette concentrations (C_a) of 50, 100, 200, 360, 500, 720, 900 (and 1200 in P. smithii) µmol mol^–1 CO₂ were made on leaves during mid‐May, mid‐July and late August (early, mid‐ and late season) in 1997, 1998 and 1999 in the ambient and elevated CO₂ plots. Drought prevented reliable A : C_i measurement in 2000. Cuvette temperature and vapour pressure were set to ambient conditions and kept constant when varying C_a. Measurements were performed in the unchambered plots only at ambient CO₂.

Since responses to seasonal and CO₂‐induced variation in soil water were of interest, A, g_s, E and the leaf water‐use efficiency (WUE; CO₂ gain/H₂O loss) were measured at the two growth CO₂ concentrations every 3 to 4 weeks during 1997, 1998, 1999 and 2000 on all plots. Leaf water potential (ψ_leaf) was measured on these dates using a pressure chamber (Plant Measurement Systems, Corvalis, OR, USA). The light unit of the gas exchange system provided 1300 µmol m^–2 s^–1 photosynthetic photon flux (PPF). Photosynthesis was saturated at this light intensity in both species (data not shown). In addition, during early, mid‐ and late season, the photosynthetic response to varying light intensity was measured at the growth CO₂ concentration at 0, 50, 100, 160, 290, 590, 850 and 1300 µmol m^–2 s^–1 PPF.

Leaves similar to those in the cuvette were collected at midday, frozen in liquid nitrogen, lyophylized and ground to 0·5 mm. The content of starch, sucrose, hexose, fructans and total non‐structural carbohydrates (TNC) was determined using a modified method of Hendrix (1993) (Morgan et al., 1998). Nitrogen concentration was measured on a combustion N analyser (PDZ Europa Ltd, Northwich, UK). Leaf carbohydrates and N were calculated on a structural dry mass basis by subtracting TNC from total leaf dry mass.

The mechanistic analysis of A : C_i and A : PPF curves for P. smithii (C₃) were made using ‘Photosynthesis Assistant’ software (Dundee Scientific, Dundee, UK). The maximum rate of carboxylation by ribulose‐1,5 bisphosphate carboxylase/oxygenase (Rubisco) (V_c,max), the PPF‐saturated rate of electron transport (J_max), and the apparent quantum efficiency (Q) were calculated. The light‐ and CO₂‐saturated assimilation rate (A_sat) was also determined from A : C_i curves. The mechanistic interpretation of A : C_i curves is not well worked out for C₄ plants. Therefore, to investigate photosynthetic adaptations in B. gracilis, the carboxylation efficiency (CE), which is an indicator of efficiency of the phosphoenol‐pyruvate carboxylase CO₂ pump, was calculated from the initial slope of the A : C_i curves (Long et al., 1993; Sage, 1994). A_sat and Q were also calculated for B. gracilis.

Data presented in the figures are means of three replications ± s.e., with statistical significance tested using the SAS ANOVA procedure (SAS Institute Inc., Cary, NC, USA). Data in the tables were analysed using the SAS PROC MIXED procedure, with year as a repeated measures variable. Correlations between parameters were tested using the SAS PROC REG or PROC NLIN procedures, depending on the best fit of the data.

RESULTS

Chamber conditions, precipitation and soil water content

Air and soil temperatures in the chambers were, on average, 2·6 and 1·3 °C higher than in the unchambered plots, but there was no difference in vapour pressure. The plastic covering blocked approx. 5 % of the PPF, while shading from the chamber framework reduced daily PPF by 28 %.

Total growing season precipitation was 480, 302, 523 and 247 mm in 1997, 1998, 1999 and 2000 (long term average = 280 mm). Soil water content was higher in the elevated vs. ambient CO₂ chambers on many dates (Fig. 1). When averaged over the 4 years of study, SWC in the upper metre was 129·4 mm in elevated CO₂ chambers compared with 113·6 mm in ambient CO₂ chambers, an improvement of 14 % (P = 0·02). Ambient CO₂ chambers often had lower SWC than the unchambered plots owing to higher air temperatures.

Fig. 1. Soil moisture (to 1 m) and growing season precipitation for 4 years in open‐top chambers with ambient (360 µmol mol^–1) and elevated (720 µmol mol^–1) CO₂ and in plots without chambers on the Colorado shortgrass steppe. Data are means of three replications ± s.e.

Photosynthetic adaptations to elevated CO₂

Table 1 shows photosynthetic parameters for P. smithii. Long‐term growth at elevated CO₂ reduced V_c,max, J_max, and A_sat by an average of 36, 39, and 36 % respectively (all P < 0·001). This indicates strong and consistent photosynthetic acclimation in this C₃ grass. Although there was a significant seasonal effect on these parameters, there was no interaction between CO₂ treatment and season or year.

Table 1.

The maximum rate of carboxylation by Rubisco (V_c,max), the PPF‐saturated rate of electron transport (J_max), assimilation at saturating light and CO₂ (A_sat), and the apparent quantum efficiency (Q) in Pascopyrum smithii (C₃) at three periods during the growing season

	Vc,max (µmol CO2 m–2 s–1)	Jmax (µmol e– m–2 s–1)	Asat (µmol CO2 m–2 s–1)	Q (CO2/photon)
Early season
Ambient CO2	63·3 ± 4	206 ± 20	32·8 ± 2·4	0·066 ± 0·01
Elevated CO2	40·9 ± 4	121 ± 16	19·9 ± 2·4	0·076 ± 0·007
Mid‐season
Ambient CO2	43·4 ± 9	97 ± 21	15·5 ± 3·7	0·051 ± 0·009
Elevated CO2	28·2 ± 3	57 ± 7	10·1 ± 1·3	0·053 ± 0·008
Late season
Ambient CO2	62·9 ± 6	155 ± 19	24·5 ± 3·3	0·053 ± 0·013
Elevated CO2	39·2 ± 5	99 ± 15	16·2 ± 3·4	0·053 ± 0·004
P
CO2trt	0·0001	0·0003	0·0006	0·4968
Year	0·8744	0·3395	0·2531	0·6905
Season	0·0132	0·0001	0·0001	0·1527
CO2trt × year	0·7552	0·5467	0·6649	0·8450
CO2trt × season	0·7006	0·4207	0·4028	0·9547

Plants were grown in open‐top chambers on native shortgrass steppe of Colorado, USA, with ambient or elevated (360 and 720 µmol mol^–1) CO₂. Data are averaged across three years and three replications ± s.e.

The CO₂ effect on CE and A_sat was not significant in B. gracilis (Table 2). Again, there was a significant effect of season, probably due to seasonal differences in soil water (see below). However, the interaction between season and CO₂ treatment was again not significant.

Table 2.

The initial slope of the A : C_i curve (carboxylation efficiency, CE), assimilation at saturating light and CO₂ (A_sat), and the apparent quantum efficiency (Q) in Bouteloua gracilis (C₄) at three periods during the growing season

	CE (CO2 m–2 s–1/mol–1 CO2)	Asat (µmol CO2 m–2 s–1)	Q (CO2/photon)
Early season
Ambient CO2	0·1267 ± 0·017	25·8 ± 3·7	0·047 ± 0·007
Elevated CO2	0·1107 ± 0·11	22·8 ± 2·2	0·064 ± 0·002
Mid‐season
Ambient CO2	0·0697 ± 0·07	12·7 ± 3·2	0·046 ± 0·007
Elevated CO2	0·0726 ± 0·013	16·0 ± 2·6	0·052 ± 0·009
Late season
Ambient CO2	0·1325 ± 0·024	23·9 ± 3·9	0·052 ± 0·006
Elevated CO2	0·1380 ± 0·021	25·3 ± 3·4	0·066 ± 0·007
P
CO2trt	0·8560	0·8948	0·2573
Year	0·2877	0·1075	0·6451
Season	0·0048	0·0029	0·1622
CO2trt × year	0·9026	0·7447	0·6857
CO2trt × season	0·8240	0·5861	0·5628

Plants were grown in open‐top chambers on native shortgrass steppe of Colorado, USA, with ambient or elevated (360 and 720 µmol mol^–1) CO₂. Data are averaged across three years and three replications ± s.e.

There was no consistent effect of CO₂ on the light response curves of photosynthesis (measured at the growth CO₂ concentration) at any time or in either species (curves not shown). The apparent quantum efficiency (Q) is summarized in Tables 1 and 2, and shows no significant variable effect in either P. smithii or B. gracilis.

Since there was no CO₂ effect, further presentation of A : C_i data from B. gracilis are not shown. The A : C_i curves of P. smithii from 1999 are given (Fig. 2) as this year was representative of the treatment responses. Early in the season, when SWC was high, A was reduced in elevated vs. ambient CO₂ leaves at most C_i (Fig. 2A). During mid‐season, when plants were water‐stressed, A was about 50 % of that measured in May and there was a (non‐significant) trend for lower A in leaves grown at elevated CO₂ (Fig. 2B). Precipitation in late July improved A, and a significant reduction in the A : C_i curve in leaves at elevated CO₂ was again seen late in the season (Fig. 2C). The reduction in photosynthetic capacity is such that A measured at the growth CO₂ concentration was similar in grasses grown at ambient and elevated CO₂ (Fig. 2, arrows). The assimilation rate in grasses in unchambered plots was similar to that in grasses in ambient CO₂ chambers.

Fig. 2. Assimilation vs. intercellular CO₂ concentration of Pascopyrum smithii grown in open‐top chambers with ambient (360 µmol mol^–1) and elevated (720 µmol mol^–1) CO₂, and in plots without chambers on the Colorado shortgrass steppe over the course of the 1999 growing season. Data are means of three replications ± s.e. An asterisk indicates a significant effect of CO₂ treatment at P < 0·05. Arrows indicate data at the treatment [CO₂].

Assimilation rates under seasonal and CO₂‐induced soil moisture variation

Variability in the patterns of A measured at the growth CO₂ concentration was primarily a function of variation in SWC induced by season and CO₂. Here we contrast results from a wet (1999) and a dry (2000) year. Data for 1997 and 1998 have been presented previously (Morgan et al., 2001).

Early in the season of 1999 [day of the year (DOY) 125], P. smithii plants grown at elevated CO₂ had higher A than plants grown at ambient CO₂ or without chambers (Fig. 3A). By DOY 153, this advantage had disappeared. However, as seasonal water stress developed, there were periods (DOY 188 and 210) when the assimilation rate of P. smithii grown at elevated CO₂ tended to be higher, probably due to better SWC in the elevated CO₂ plots (Fig. 1). For B. gracili, there was little difference in A early in the season of 1999 (Fig. 3B). However, during drought, and for the later third of the season, A of the elevated CO₂ plants was higher than that of ambient CO₂ plants (DOY 188, 210 and >225). Again, this was associated with improved SWC under elevated CO₂ (Fig. 1).

Fig. 3. Assimilation of Pascopyrum smithii and Boutleoua gracilis grown in open‐top chambers with ambient (360 µmol mol^–1) and elevated (720 µmol mol^–1) CO₂, and in plots without chambers on the Colorado shortgrass steppe during the growing seasons of 1999 and 2000. Measurements were made at the chamber CO₂ concentrations. Data are means of three replications ± s.e. An asterisk indicates an effect of CO₂ treatment significant at P < 0·05; t, P < 0·10.

During 2000, elevated CO₂ plots had higher SWC for much of the season (Fig. 1). Assimilation rates of P. smithii reflect these soil moisture differences, with the elevated CO₂ plants having higher A than ambient plants on most dates (Fig. 3C). This effect was less evident in B. gracilis (Fig. 3D). Assimilation in the unchambered plots was similar to that in the ambient plots in both species, even though soil moisture was similar to that in elevated CO₂ plots (Fig. 1). This was associated with improved ψ_leaf in the elevated CO₂ plants.

Leaf water potential under seasonal and CO₂‐induced soil moisture variation

Higher SWC and improved conditions for A under elevated CO₂ were also reflected in improved ψ_leaf. Leaf water potential in both species was higher in plants grown at elevated CO₂ on most measurements dates in 1999 and 2000 (Fig. 4).

Fig. 4. Leaf water potential of Pascopyrum smithii and Boutleoua gracilis grown in open‐top chambers with ambient (360 µmol mol^–1) and elevated (720 µmol mol^–1) CO₂, and in plots without chambers on the Colorado shortgrass steppe during the growing seasons of 1999 and 2000. Data are means of three replications ± s.e. An asterisk indicates a significant effect of CO₂ treatment at P < 0·05.

Stomatal responses to elevated CO₂

Responses of g_s to varying C_i early in the season of 1999 were representative of responses when interactions with drought were not occurring (see below) and thus are discussed here (Fig. 5). Stomatal conductance of P. smithii decreased from 0·5 to 0·28 mol m^–2 s^–1 over a range of C_i from 50 to 670 µmol mol^–1; this variation was less than that seen in B. gracilis which varied from 0·8 to 0·18 over similar values of C_i. In P. smithii there was a trend for lower g_s in leaves grown in elevated CO₂ (Fig. 5A). This trend was observed on most measurement dates. There was no indication of stomatal adaptation to elevated CO₂ in B. gracilis (Fig. 5B).

Fig. 5. Stomatal conductance vs. intercellular CO₂ of Pascopyrum smithii and Boutleoua gracilis grown in open‐top chambers with ambient (360 µmol mol^–1) and elevated (720 µmol mol^–1) CO₂ and in plots without chambers on the Colorado shortgrass steppe. Measurements made during May 1999. Data are means of three replications ± s.e. The CO₂ treatment effect was not significant.

During much of the growing season of 1999, stomatal conductance of P. smithii (measured at the growth CO₂ concentration) was lower in plants grown at elevated CO₂ (Fig. 6A). Stomatal conductance of B. gracilis in elevated CO₂ chambers also tended to be lower, although the difference was significant on fewer dates (Fig. 6B). During the dry spring of 2000, g_s was sometimes higher in P. smithii plants from elevated CO₂, and there were no differences in g_s in B. gracilis (Fig. 6C and D). Leaf water‐use efficiency of both species was always higher when measured at elevated CO₂ (data not shown).

Fig. 6. Stomatal conductance of Pascopyrum smithii and Boutleoua gracilis grown in open‐top chambers with ambient (360 µmol mol^–1) and elevated (720 µmol mol^–1) CO₂, and in plots without chambers on the Colorado shortgrass steppe during the growing seasons of 1999 and 2000. Measurements were made at the chamber CO₂ concentrations. Data are means of three replications ± s.e. An asterisk indicates an effect of CO₂ treatment significant at P < 0·05; t, P < 0·10.

Leaf nitrogen and carbohydrates responses to elevated CO₂

There was a significant CO₂ × season interaction for leaf N and TNC. Therefore, statistics are shown for each measurement period (Table 3). Early in the season, there was no CO₂ effect on the leaf N of P. smithii, but N was reduced under elevated CO₂ in mid‐ and late season. Leaf N was reduced in B. gracilis in elevated CO₂ in all three periods.

Table 3.

Percentage nitrogen and total non‐structural carbohydrates (TNC) on a structural dry mass basis of leaves of Pascopyrum smithii (C₃) and Bouteloua gracilis (C₄) grown in open‐top chambers on native shortgrass steppe of Colorado, USA, with ambient or elevated (360 and 720 µmol mol^–1) CO₂, or on plots with no chamber

	Nitrogen (%)		TNC (mg g–1)
	P. smithii	B. gracilis	P. smithii	B. gracilis
Early season
No chamber	1·59 ± 0·05	1·89 ± 0·03	111·6 ± 18·9	121·0 ± 10·9
Ambient CO2	1·42 ± 0·03	1·65 ± 0·02	119·6 ± 6·6	110·9 ± 2·7
Elevated CO2	1·33 ± 0·04	1·40 ± 0·05	193·1 ± 14·2	140·4 ± 11·3
P
CO2trt	0·1391	0·0001	0·0003	0·0103
Year	0·1106	0·0372	0·0001	0·0007
CO2trt × year	0·1909	0·0723	0·1845	0·5604
Mid‐season
No chamber	1·33 ± 0·08	1·24 ± 0·06	111·0 ± 16·3	100·5 ± 1·2
Ambient CO2	1·31 ± 0·05	1·15 ± 0·04	97·4 ± 7·9	106·7 ± 5·6
Elevated CO2	1·10 ± 0·02	1·02 ± 0·02	176·5 ± 25·3	117·0 ± 5·8
P
CO2trt	0·0029	0·0013	0·0007	0·0868
Year	0·6238	0·0014	0·0310	0·0001
CO2trt × year	0·9010	0·2490	0·1593	0·3753
Late season
No chamber	1·05 ± 0·03	1·26 ± 0·04	72·9 ± 10·1	76·5 ± 3·0
Ambient CO2	0·96 ± 0·06	1·20 ± 0·03	56·3 ± 7·1	74·7 ± 2·2
Elevated CO2	0·77 ± 0·06	1·06 ± 0·02	83·5 ± 9·3	73·1 ± 6·2
P
CO2trt	0·0096	0·0250	0·0670	0·8580
Year	0·0001	0·0001	0·0001	0·0001
CO2trt × year	0·3399	0·0029	0·0691	0·8497

Data are averaged over 4 years and three replications ± s.e. The season × CO₂ interaction was significant.

CO₂trt is the No chamber, ambient CO₂ and elevated CO₂ variables.

The content of non‐structural carbohydrates was higher in P. smithii leaves grown in elevated vs. ambient CO₂ in all three periods (Table 3). An increase in all of the individual sugar fractions was found, but fructans and sucrose were the dominant carbohydrates in P. smithii (data not shown). In contrast, B. gracilis grown in elevated CO₂ had significantly more carbohydrates only early in the season (primarily starch). In both species, unchambered plants had TNC and N contents similar to those of plants in the ambient CO₂ chambers.

DISCUSSION

Reports of photosynthetic acclimation under elevated CO₂ have been considered the exception in field studies (Sage, 1994; Drake et al., 1997). Contrary to this, our hypothesis that P. smithii would develop photosynthetic acclimation was confirmed in this study. Although there were periods, in early spring, that reflected a direct photosynthetic stimulation in elevated CO₂, 3 years of A : C_i measurements showed consistent photosynthetic acclimation in P. smithii. These findings agree with previous results of growth chamber studies (Morgan et al., 1994a, 1998; Read et al., 1997), but the photosynthetic acclimation was greater in the field.

Previous field studies of CO₂ enrichment in grassland ecosystems show contrasting evidence for photosynthetic acclimation. In the tall grass prairie and a Texas grassland there was no photosynthetic acclimation in either C₃ or C₄ species (Nie et al., 1992; Hamerlynck et al., 1997; Anderson et al., 2001). However, photosynthetic acclimation was reported in C₃ species in a New Zealand pasture (von Caemmerer et al., 2001) and in several diverse grassland species in the Minnesota prairie (Lee et al., 2001).

From these reports it is difficult to generalize about the basis of photosynthetic acclimation under elevated CO₂, as species, soils, climate and experimental conditions vary. However, it seems clear that the reduction in photosynthetic capacity helps to restore the balance between carbon assimilation, other resources (water, nutrients, rooting volume) and carbon demand (Sage, 1994; Midgley et al., 1999; Moore, 1999). In C₃ plants, this typically includes a reduction in Rubisco amount or activity, and reduced leaf N (Sage, 1994; Midgley et al., 1999; Moore et al., 1999; Stitt and Krapp, 1999).

Consistent with these studies, V_c,max and leaf N concentration of P. smithii were significantly reduced under elevated CO₂. We suspect that improved growth of SGS species under elevated CO₂ (Morgan et al., 2001) has further limited the available soil N in this N‐limited ecosystem (Hunt et al., 1988). The resultant inability of P. smithii plants to produce new sinks causes an imbalance in ‘supply and demand’, which feeds back on photosynthetic processes. Our leaf TNC data support this, as strong accumulation (average 61 %) of carbohydrates occurred. These results agree with those of Rogers et al. (1998) who concluded that photosynthetic acclimation results primarily from sink limitation in low N soils.

The A : C_i curves obtained also show a reduction in the regeneration capacity for ribulose‐1,5 bisphosphate (RuBP) in P. smithii, as indicated by reduced J_max (Table 1) (Long et al., 1993). The light response data also support this (Table 1). The apparent quantum efficiency (Q) of photosynthesis is related to the RuBP regeneration capacity of C₃ plants (Long et al., 1993). An increase in C_a should increase Q, owing to less oxygenation of RuBP. The lack of improvement in Q when measured at elevated CO₂ provides more evidence for a reduction in RuBP regeneration capacity (Table 1). These results agree with those of Midgley et al. (1999), who reported an integrated reduction in both carboxylation and RuBP regeneration capacity under elevated CO₂. However, these findings contradict the theory that Q will be improved in C₃ plants under increasing atmospheric CO₂ (Ehleringer et al., 1997).

Contrary to our hypothesis, B. gracilis did not display photosynthetic acclimation under elevated CO₂, although we have previously reported acclimation of well‐watered and fertilized B. gracilis (Read et al., 1997; LeCain and Morgan, 1998). The A : C_i curves in this field study show the short‐term, ‘potential’ photosynthetic improvement at elevated CO₂ is only approx. 10 % in B. gracilis (data not shown), compared with approx. 65 % in P. smithii. Therefore, B. gracilis is less likely to experience source/sink imbalances. Compared with growth chamber studies, the low soil N availability in the field may have limited the potential photosynthetic response of B. gracilis, which precluded photosynthetic acclimation to elevated CO₂. Our results agree with those of von Caemmerer et al. (2001) who reported photosynthetic acclimation in C₃ species, but not C₄ species, under elevated CO₂. Due to acclimation, there was little improvement in A of P. smithii grown at elevated CO₂ during periods of adequate soil moisture. However, during dry periods, the improved soil moisture at elevated CO₂ resulted in improved A in both species. Averaged over the study, A was 30 and 21 % higher in P. smithii and B. gracilis, respectively, grown at elevated compared with ambient CO₂ (Table 4).

Table 4.

Means of gas exchange data of Pascopyrum smithii and Bouteloua gracilis measured at the growth CO₂ concentration, and averaged over all of the data (4 years, 36 dates)

	Assimilation (µmol     CO2 m–2 s–1)	Conductance (mol     H2O m–2 s–1)	Transpiration (mmol     H2O m–2 s–1)	Water‐use efficiency (mmol CO2/mol H2O)
P. smithii
Ambient CO2	12·5	0·29	5·66	2·07
Elevated CO2	16·2	0·21	4·32	3·76
P
CO2trt	0·0002	0·0038	0·0058	0·0001
Year	0·0330	0·0068	0·8393	0·0051
CO2trt × year	0·4216	0·0681	0·0998	0·2757
B. gracilis
Ambient CO2	15·9	0·25	6·33	2·79
Elevated CO2	19·3	0·16	5·16	4·89
P
CO2trt	0·0097	0·0001	0·0001	0·0001
Year	0·0008	0·0096	0·0035	0·2667
CO2trt × year	0·9997	0·5184	0·9716	0·7657

Plants were grown in open‐top chambers on native shortgrass steppe of Colorado, USA, with ambient or elevated (360 and 720 µmol mol^–1) CO₂.

Figure 7 shows data for SWC and A in plants grown at elevated relative to ambient CO₂, plotted against the ambient CO₂ SWC, for 1999 and 2000. These plots demonstrate that the improvement in SWC under elevated CO₂ occurred primarily as soils dried out. Also, the improvement in A under elevated CO₂ was expressed primarily at low SWC. These results conflict with the views of Wand et al. (1999) that water stress reduces the relative photosynthetic enhancement of grasses grown at elevated CO₂. But our findings agree with those of Owensby et al. (1993) and Hamerlynck et al. (1997) for a tallgrass prairie, where responses were greater in drought seasons. Huxman et al. (1998), conducting a field study in the desert, also reported photosynthetic acclimation in Larrea tridentata Cav. when soil water was available, but assimilation was higher under elevated CO₂ during drought periods.

Fig. 7. 1999 and 2000 assimilation (A) and soil water content (SWC) data in elevated (E: 720 µmol mol^–1) relative to ambient (A: 360 µmol mol^–1) CO₂ chambers, plotted against the ambient CO₂ SWC. Lines are a linear regression for SWC and a second order polynomial for assimilation. Both correlations are significant (P < 0·05).

A reduction in g_s is an almost universal response to elevated CO₂ (Drake et al., 1997; Jarvis et al., 1999; Wand et al., 1999). Consistent with our hypothesis, stomatal conductance was lower under elevated CO₂ compared with ambient CO₂ when soil moisture was adequate. Averaged over the study, stomatal conductance was reduced by 27 and 36 %, and transpiration was reduced 24 and 20 % in leaves of P. smithii and B. gracilis (Table 4). Reduced transpiration is probably the reason for improved SWC under elevated CO₂ (Field et al., 1995).

An unexpected finding was that during periods in which the soil was very dry, g_s was sometimes higher in P. smithii grown under elevated CO₂ than that grown at ambient CO₂. The best example of this occurred in 2000, where higher SWC in elevated CO₂ plots was carried over from the previous year (Fig. 1). The lack of spring precipitation led to such extended drying of the soil that a point was reached where very low SWC under ambient CO₂ conditions actually reduced g_s below that of plants grown at elevated CO₂. Knapp et al. (1996) also reported an interaction between SWC and g_s during a dry year.

A consistent trend for lower g_s at an equivalent C_i was seen in P. smithii grown at elevated CO₂ (Fig. 5A). There is evidence that stomatal functioning is coupled with photosynthetic activity to maintain a relatively stable C_i : C_a ratio (Jarvis et al., 1999). If stomata adapt to elevated CO₂ independently of photosynthetic adaptations, the C_i : C_a ratio will differ (Ball and Berry, 1982). We found no effect of CO₂ treatment on C_i : C_a in either species (data not shown). Our results agree with those of Drake et al. (1997) and Jarvis et al. (1999) that apparent stomatal acclimation is a result of stomatal dependence on photosynthetic processes.

Leaf water potential was higher in elevated CO₂ plots on most dates. This is probably a result of improved SWC and g_s during much of the growing season (Tyree and Alexander, 1993; Hamerlynck et al., 1997). However, the regression analysis showed poor correlations between the improvement in ψ_leaf and improved A and g_s (data not shown). This is because ψ_leaf was improved on nearly all dates, whereas A was improved primarily during periods of low SWC, and g_s interacted with SWC. Factors such as osmotic adjustment and rooting dynamics will complicate the relationship between ψ_leaf and A and g_s (Tyree and Alexander, 1993). Still, we believe that the strikingly consistent improvement in ψ_leaf is evidence that plant water status was more favourable for assimilation in the elevated CO₂ plots. The present results agree with those of Hamerlynck et al. (1997) who reported that elevated CO₂ improved ψ_leaf and A in both C₃ and C₄ species during drought years.

Implications for grassland ecosystems

In this semi‐arid, N‐limited ecosystem, the C₃ grass was unable to produce adequate new sinks to maintain higher A under elevated CO₂. Despite the lack of direct photosynthetic stimulation in both the C₃ and C₄ species, the present results provide support for a positive growth response to elevated CO₂ in this ecosystem. Averaging the gas exchange data over the study, A was improved by 30 and 21 %, g_s was reduced by 27 and 36 %, E was reduced by 24 and 20 %, and WUE was increased by 80 and 75 % in P. smithii and B. gracilis under elevated compared with ambient CO₂ (Table 4). A 26–47 % increase in above‐ground biomass was found in the elevated CO₂ plots in 1997 and 1998 (Morgan et al., 2001). In 1999, a year with nearly twice the average seasonal precipitation, biomass was improved by only 17 % in elevated compared with ambient CO₂ (J. A. Morgan, pers. comm.). But during the drought year of 2000, biomass was increased by 95 % in elevated CO₂ plots. This large improvement in biomass was possible because of higher SWC and improved A and ψ_leaf in elevated CO₂ plots.

The present study suggests that improvements in WUE, soil water conservation and plant water relations will primarily affect responses to elevated CO₂ in this semi‐arid ecosystem, rather than a direct effect on assimilation. Although the improvement in SWC (relative average 14 %) may seem small, the effect is large in this water‐limited ecosystem. These results agree with those of Volk et al. (2000) that the effects of elevated CO₂ on soil moisture account for most of the variability in production. Improved soil moisture should extend the favourable growing season in arid and semi‐arid ecosystems (Knapp et al., 1996).

This study provides no evidence for a competitive advantage of the C₃ over the C₄ grass under elevated CO₂. Earlier notions that C₃ species would have a large advantage as global levels of CO₂ rise are not being supported by field studies (Owensby et al., 1993; Wand et al., 1999; Anderson et al., 2001). The comparative response of C₃ vs. C₄ species will depend on phenology, morphology, root distribution, reproduction strategies and nutrient‐use efficiency, as well as photosynthetic and stomatal responses.

ACKNOWLEDGEMENTS

The authors thank Scott Andre, Barry Weaver, Dean Ackerman and Dennis Mueller for designing and building the chambers, and for technical assistance, and Clenton Owensby, Jay Ham, Lisa Auen and Fred Caldwell of Kansas State University for guidance on chamber design. We also thank Alan Knapp and Lewis Ziska for critically reviewing the manuscript. This research was supported, in part, by funding from the NSF‐TECO awards IBN‐9524068 and USDA/NRICGP‐98‐134, NSF award DEB‐9708596 and from the Shortgrass Steppe LTER project DEB 9350273. Mention of a trademark or manufacturer by the USDA does not imply its approval to the exclusion of other products or manufacturers that may also be suitable.

Received: 2 December 2002; Returned for revision: 28 January 2003; Accepted: 12 March 2003    Published electronically: 9 May 2003