Interactive effects of elevated CO2, warming, and drought on photosynthesis of Deschampsia flexuosa in a temperate heath ecosystem.

Global change factors affect plant carbon uptake in concert. In order to investigate the response directions and potential interactive effects, and to understand the underlying mechanisms, multifactor experiments are needed. The focus of this study was on the photosynthetic response to elevated CO(2) [CO2; free air CO(2) enrichment (FACE)], drought (D; water-excluding curtains), and night-time warming (T; infrared-reflective curtains) in a temperate heath. A/C(i) curves were measured, allowing analysis of light-saturated net photosynthesis (P(n)), light- and CO(2)-saturated net photosynthesis (P(max)), stomatal conductance (g(s)), the maximal rate of Rubisco carboxylation (V(cmax)), and the maximal rate of ribulose bisphosphate (RuBP) regeneration (J(max)) along with leaf δ(13)C, and carbon and nitrogen concentration on a monthly basis in the grass Deschampsia flexuosa. Seasonal drought reduced P(n) via g(s), but severe (experimental) drought decreased P(n) via a reduction in photosynthetic capacity (P(max), J(max), and V(cmax)). The effects were completely reversed by rewetting and stimulated P(n) via photosynthetic capacity stimulation. Warming increased early and late season P(n) via higher P(max) and J(max). Elevated CO(2) did not decrease g(s), but stimulated P(n) via increased C(i). The T×CO2 synergistically increased plant carbon uptake via photosynthetic capacity up-regulation in early season and by better access to water after rewetting. The effects of the combination of drought and elevated CO(2) depended on soil water availability, with additive effects when the soil water content was low and D×CO2 synergistic stimulation of P(n) after rewetting. The photosynthetic responses appeared to be highly influenced by growth pattern. The grass has opportunistic water consumption, and a biphasic growth pattern allowing for leaf dieback at low soil water availability followed by rapid re-growth of active leaves when rewetted and possibly a large resource allocation capability mediated by the rhizome. This growth characteristic allowed for the photosynthetic capacity up-regulations that mediated the T×CO2 and D×CO2 synergistic effects on photosynthesis. These are clearly advantageous characteristics when exposed to climate changes. In conclusion, after 1 year of experimentation, the limitations by low soil water availability and stimulation in early and late season by warming clearly structure and interact with the photosynthetic response to elevated CO(2) in this grassland species.

Introduction

Environmental changes caused by increased emissions of greenhouse gasses have been predicted to influence the stability of ecosystems worldwide (IPCC, 2007). For terrestrial plants and ecosystems, increases in the atmospheric n class="Chemical">CO2 concentration and air temperature as well as changes in precipitation patterns are expected to have strong impacts on the carbon balance. Regional climate models predict enhanced temperature and a change in the pattern of precipitation in Denmark, resulting in longer summer drought periods and the occasional incidences of high amounts of precipitation (Christensen and Christensen, 2003). Effects of environmental changes such as elevated CO2, temperature, and precipitation on plants and ecosystems have primarily been investigated as effects of a single factor or two factors in combination (e.g. Beier ; for an overview see Rustad, 2008). Since all environmental changes occur concurrently, in order to investigate potential interactions between factors, it is important to study the effects in response to the factors alone and when combined.

Warming generally increases net primary production (Rustad ; Shaw ; Dukes ; Penuelas et al., 2007), with the strongest responses on sites with low aridity (Penuelas et al., 2007). On the leaf level, increased daytime temperature may increase light-saturated net photosynthesis (Pn), espen class="Chemical">cially during cold periods (Sage and Kubien, 2007). Plant respiration in response to warming probably affects the plant carbon uptake. During the day, respiration increases with temperature and decreases the plant net carbon uptake (Atkin and Tjoelker, 2003), whereas during the night-time increased plant respiration can stimulate carbon sink strength and daytime Pn (Turnbull , 2004). Indirect effects of warming can increase growing season length (Menzel and Fabian, 1999; Walther ; Cleland ), change phenology (Harte and Shaw, 1995; Wan ; Menzel ), increase soil nitrogen mineralization and availability (Rustad ), reduce the soil water content (SWC) (Volk ; Morgan ; Wan ), and affect species composition and community structure (Shaver ; Wan ). The indirect effects have been argued to be more important than the direct effects (Körner, 2000; Shaver ; Morgan ; Lou et al., 2007).

In response to mild to moderate drought, stomatal conductance (gs) reduction decreases transpiration (Tr) an class="Disease">nd Pn. More intensive drought induces the down-regulation of light- and CO2-saturated net photosynthesis (Pmax), the maximal rate of Rubisco carboxylation (Vcmax), and the maximal rate of ribulose bisphosphate (RuBP) regeneration (Jmax); a reduced content of Rubisco decreases Pn together with gs, and mesophyll conductance reduction dominates during severe drought (Flexas and Medrano, 2002; Penuelas et al., 2007).

Elevated CO2 predominantly decreases g, stimulates Pn, an class="Disease">nd increases plant water use efficiency (WUE) (Curtis, 1996; Curtis and Wang, 1998; Ainsworth and Long, 2005; Ainsworth and Rogers, 2007). At elevated CO2, Vcmax is often reduced through reduction of Rubisco content and thereby nitrogen investment (Drake ; Parry ). This response leads to a down-regulation response with decreased leaf nitrogen, increased leaf C/N ratio, and reduced Jmax and Pmax found in many studies (Ainsworth and Long, 2005; Ainsworth and Rogers, 2007). The magnitude of the response of Pn to elevated CO2 is modulated by nitrogen supply and the constraints of the capacity to utilize photosynthates (Ainsworth and Rogers, 2007). The importance of improved water relations via reduced water consumption in elevated CO2, which to some degree sustain Pn in dry periods, has been stressed in particular in grasslands and forest exposed to elevated CO2 (Körner, 2000; Volk ; Medlyn ; Morgan ; Leuzinger and Körner, 2008). During dry periods, the combined responses of leaf area index (LAI) and leaf gs determine the plant water consumption. Therefore, water status may strongly co-determine plant biomass responses to CO2 enrichment (Acock and Allen, 1985; Gifford, 1992; Koch and Mooney, 1996; Fredeen ; Field ; Niklaus ).

This study is part of the CLIMAITE experiment (Mikkelsen ), where treatments of passive night-time warming, elevated CO2, an class="Disease">nd periods of drought are applied alone and in all combinations. To investigate the photosynthetic responses of the dominant grass Deschampsia flexuosa, monthly campaigns of leaf gas exchange measurements (A/Ci curves) were performed, and leaf characteristics of carbon, nitrogen, and 13C natural abundance (δ13C) were analysed. The values of δ13C represent an integrated measure of WUE over time (Farquhar ). It was assumed that the measured δ13C values were the product of gs reductions over time but that they would also be influenced by a higher leaf Ci/Ca ratio, mediated by reduction in photosynthetic capacity measures (Jmax, Vcmax, and Pmax). The measurements made allowed for analysis of the accumulated responses of carbon uptake, water status, and nitrogen balance in the experimental system. The impacts of the three environmental factors alone and in combination were evaluated to elucidate possible shifts in importance of the treatments during the growing season.

It was hypothesized that:

(i) Elevated CO2 (CO2) would increase Pn via increased Ci despite acclimation with gs reduction and photosynthetic capacity (probed with Pmax, Jmax, and Vcmax) down-regulation.

(ii) Elevated CO2 would sustain Pn in dry periods via soil water savings.

(iii) Passive night-time warming (T) would increase Pn and decrease δn class="Chemical">13C in spring via earlier maturation of photosynthetic capacity and in autumn via delayed senescence maintaining high photosynthetic capacity.

(iv) Summer drought (D) would decrease Pn via gs reduction and only decrease Pn via photosynthetic capan class="Chemical">city down-regulation under severe water shortage.

(v) Responses to combinations of T, D, and CO2 would mainly be additive, but with potential for interactive effects.

Materials and methods

Site and experimental set-up

The experimental site is a dry heathland ecosystem in North Zealan class="Disease">nd, Denmark, dominated by the C3 grass D. flexuosa (L.) Trin and the evergreen dwarf shrub Calluna vulgaris (L.) Hull. The experimental treatments were elevated CO2 (CO2), passive night-time warming (T), summer drought (D), all combinations, TD, TCO2, DCO2, and TDCO2, and unmanipulated control (A), replicated in six blocks in a split-plot design. Each block consisted of two 6.8 m diameter octagons, each divided into four plots. In one octagon the CO2 concentration was elevated to 510 ppm during the daytime with the FACE technique (free air CO2 enrichment; Miglietta ) and in the other octagon the CO2 level was ambient. In each octagon, automated curtains covered one half (two plots) during the night, preserving a proportion of the daily incoming radiation energy, which increased the night air temperature by up to 4 °C, on average 1–2 °C. Perpendicular to the infrared (IR)-reflective curtains was a rain-excluding curtain which automatically was activated by rain during intended drought periods. In each experimental plot the soil temperature was measured continuously at 2 cm and 10 cm depth, and the SWC at 0–20 cm and 0–60 cm depth. Two climatic stations recorded precipitation, air temperature, photosynthetically active radiation, and wind speed and direction. The CO2 and warming treatments were initiated on 3 October 2005. A drought period was initiated on 3 July 2006 and continued until 4 August when the SWC reached ∼5% in the top 20 cm of the soil. For further description of the site and experimental set-up, see Mikkelsen .

Leaf gas exchange

CO2 an class="Disease">nd H2O leaf gas exchange measurements were conducted in situ by using two LI-6400s (LI-COR Biosciences, Lincoln, NE, USA) connected to standard 2×3 cm chambers with light-emitting diode (LED) light sources (6400-02B). They were carried out in the periods 11–18 May, 12–16 June, 10–13 July, 14–21August, 11–22 September, and 9–20 October, 2006. During each campaign, measurements were conducted on one leaf sample in each experimental plot (48 plots) and the resulting treatment replications were n=6. Only fully expanded, healthy leaves from the top of the vegetation were selected. Pilot studies resulted in the following methodology securing highly reproducible measurements: 10–20 D. flexuosa leaves were positioned side by side and gently fixed by hairgrips during each of the monthly periods of measurements. The leaf cuvettes were fixed to the samples during measurements using flexible arms (Magic Arm 143, Manfrotto, Italy). During the following harvest, care was taken to ensure that area and weight were determined on exactly the same material as inserted in the leaf cuvette.

Samples were acclimated for 2–4 min at ambient CO2 (380 ppm), until net photosynthesis an class="Disease">nd stomatal conductance were stabilized [±1 coefficient of variation (CV) over 30 s]. The CO2 response curves were measured by stepping CO2 down from 380 ppm to 300, 200, 100, and 50 ppm and then back to 380 ppm for 4 min re-acclimation, until the initial state was reached again. Then CO2 was then stepped up to saturation from 380 ppm to 450, 510, 650, 800, 1000, 1200, and 1400 ppm at the saturating light level of 1500 μmol photons m−2 s−1. The LI-COR Auto program ‘A/Ci-curve’ was used (settings: minimum 45 s and maximum 60 s, reference CO2 stable in 10 s with CV <1%, Ci stable in 10 s with CV <1 %, IRGA matching performed at each step). Block temperature was held constant at 25 °C and sample relative humidity was stabilized at 45–55% during measurements. All measurements were area corrected. Leak corrections was applied with the empty chamber approach (Manter and Kerrigan, 2004). Gas exchange parameters, such as Pn, Tr, WUE (=Pn/Tr), gs, and the Ci, were extracted from the CO2 response curves at the reference CO2 level (380 ppm CO2 in non-FACE plots and 510 ppm CO2 in FACE plots). This allowed calculation of the Ci/Ca with Ca set to 380 ppm in ambient CO2 and 510 ppm in elevated CO2. Pmax was extracted at 1200 ppm CO2. Calculation of Vcmax and Jmax followed the approach of Dubois after Farquhar .

Leaf weight, area, nitrogen, carbon, and δ13C

Immediately after harvest, digital pictures were taken of the leaves flattened by transparent acrylic plates beside a quadrant of known area. The photographs were converted to black and white an class="Disease">nd loaded into a pixel-counting program (Bitmap, S. Danbæk, Department of Biology, University of Copenhagen), by which the leaf areas were determined. Then the fresh weight was determined. The dry weight was determined after oven drying at 80 °C for 48 h. The plant material was analysed for C and N concentration and δ13C with an elemental analyser (EuroVector, Milano, Italy) coupled to an IsoPrime isotope ratio mass spectrometer (GV Instruments, Manchester, UK). During analysis, the internal reference gas was calibrated against certified standards from the International Atomic Energy Agency, and plant material calibrated against certified standards was used as a working standard. From the dry weight and leaf area, the specific leaf area (SLA) was calculated. Leaf water content was calculated from fresh and dry weights.

Statistics

Analyses of variance were performed on linear mixed effects models with the following fixed effects: month, T, D, and n class="Chemical">CO2, and their interactions T×D, T×CO2, D×CO2, and T×D×CO2, using proc mixed in SAS (Statistical Analysis Systems Institute, 2009). The split-plot design of the experiment was described by including octagon (O), O×T, and O×D as random factors. Pre-treatment data were initially included as covariates as fixed factors. The full model was reduced by backward selection until factors with P <0.1 remained. The denominator degrees of freedom were corrected according to the Satherwaite procedure. Significance levels are reported in the figures and tables as a tendency with † when P <0.10 and as significant with * when P <0.05, ** when P<0.001, and *** when P <0.0001. All data were tested for and confirmed to be normally distributed; some response data were transformed to fulfil the assumption of homogeneity of variance. To interpret the direction of the significant responses, the group least square means were compared.

Results

Environmental conditions

In 2006, the last snow melted in late March and from early April the mean daily temperature (at 2 m height) gradually increased from 0 °C to 25 °C in late July; 2006 was the warmest year ever recorded in Denmark, 9.4 °C (www.DMI.dk), an class="Disease">nd mean daily air temperatures were >11 °C in late November. The warming treatment increased the 24 h mean daily temperature by 1–2 °C and there was a 33% higher number of accumulated growing degree days (GDDs) from 1 April to 15 May and an annual 7% increase in GDDs in warmed plots compared with controls (Mikkelsen ). The warming treatment increased the maximum temperatures in the late night/early morning in both the air and the soil, but the effect was gradually reduced during the day. In the air, this effect of the warming treatment occasionally lasted 3–5 h after sunrise, while in the soil the warming effect was less dynamic and some warming was sustained throughout the day. (Mikkelsen ). However, the average air temperature increased in response to warming only during the night-time, and there were no effects on average daytime temperature (Fig. 1). The concentrations of CO2 in the FACE octagons were close to 510 ppm on an houly basis during the daytime (Mikkelsen ), but there were a large variation on a smaller time scale just as experienced in all FACE studies. The control plots were not affected by the CO2 in the FACE octagons.

Fig. 1.

Temperature. Daytime air temperature at 2 m height (hourly means) at the experimental site is depicted as a black line and the 0 °C reference as a dotted line. (b) Daytime temperature difference at 20 cm height for the warming minus control treatments. No significant effects on daytime temperature appeared; thus, temperature was not directly affected when leaf gas exchange and fluorescence measurements were conducted. (c) Night-time temperature difference at 20 cm height for warming minus control treatments. Night-time corresponds to the period where the passive night-time warming treatment was active.

Temperature. Daytime air temperature at 2 m height (hourly means) at the experimental site is depicted as a black line and the 0 °C reference as a dotted line. (b) Daytime temperature difference at 20 cm height for the warming minus control treatments. No significant effects on daytime temperature appeared; thus, temperature was not directly affected when leaf gas exchange an class="Disease">nd fluorescence measurements were conducted. (c) Night-time temperature difference at 20 cm height for warming minus control treatments. Night-time corresponds to the period where the passive night-time warming treatment was active.

SWC fluctuated with incoming precipitation, ranging from ∼5% to 25% at 0–20 cm an class="Disease">nd 0–60 cm depth, and the SWC at 0–20 cm decreased from 17.2% in April to 8.8% in July in the ambient plots (Figs 2, 3). This natural low SWC was extended by the experimental drought period (D) in which SWC decreased even further towards August. In this period the non-drought plots benefitted from some incidents of precipitations. Rewetting increased the SWC at 0–20 cm to 18% in August, 13.6% in September, and 16.5% in October in the controls. SWCs were significantly decreased in the warming treatments in May, June, August, and September and in response to the experimental drought in July (Figs 2, 3). There was a significant interaction effect of D×CO2 on SWC at 0–60 cm in the experimental drought period (Fig. 2) and on SWC at 0–20 cm during the leaf gas exchange campaign in July (Fig. 3). A D×CO2 interaction revealed that the combination of elevated CO2 and experimental drought compensated for the negative effects of drought by keeping soil moisture levels marginally higher in TDCO2 and DCO2 compared with the TD and D treatments (Figs 2, 3).

Fig. 2.

Precipitation and soil water content. Shown are: (a) the daily averages of soil water content in control plots at 0–20 cm and 0–60 cm depth and the daily accumulated precipitation. The panel is divided into pre-drought, experimental drought, and post-drought by vertical black lines. The vertical dashed line indicates a lag phase after the end of the experimental drought period due to sparse precipitation. (b) Change in percentage of the soil water content at 0–20 cm with treatment compared with control. The dotted line is the zero reference line for the unmanipulated control. The treatments are: elevated CO2 (CO2); experimental drought (D); passive night-time warming (T); and the combination of DCO2. Warming and elevated CO2 treatments were active in all periods, whereas the drought was only active in the experimental drought period. All treatments were replicated six times, in total 48 plots. In each period the significant effects of the experimental factors T, D, CO2, and their interactions are indicated with *** for P <0.001, ** for P <0.01, and * for 0.01

Fig. 3.

Soil water content at 0–20 cm depth during each gas exchange campaign. Shown are the mean ±SE (n=6). Treatments are: A, unmanipulated control; T, passive night-time warming; D, experimental drought, active in July to early August; TD, warming and drought combined; CO2, elevated CO2; TCO2, warming and elevated CO2 combined; DCO2, drought and elevated CO2 combined; and TDCO2, warming, drought, and elevated CO2 combined. Statistically significant effects of single factors (T, D, and CO2) and interactions (T×CO2 and D×CO2) are shown with treatment letters (*** if P <0.0001, ** if P <0.001, * if P <0.05 and † if P <0.10).

Precipitation an class="Disease">nd soil water content. Shown are: (a) the daily averages of soil water content in control plots at 0–20 cm and 0–60 cm depth and the daily accumulated precipitation. The panel is divided into pre-drought, experimental drought, and post-drought by vertical black lines. The vertical dashed line indicates a lag phase after the end of the experimental drought period due to sparse precipitation. (b) Change in percentage of the soil water content at 0–20 cm with treatment compared with control. The dotted line is the zero reference line for the unmanipulated control. The treatments are: elevated CO2 (CO2); experimental drought (D); passive night-time warming (T); and the combination of DCO2. Warming and elevated CO2 treatments were active in all periods, whereas the drought was only active in the experimental drought period. All treatments were replicated six times, in total 48 plots. In each period the significant effects of the experimental factors T, D, CO2, and their interactions are indicated with *** for P <0.001, ** for P <0.01, and * for 0.01ndicated with triangles in a horizontal line at –12.5%. (c) Change in percentage of the soil water content at 0–60 cm with treatment compared with control.

Soil water content at 0–20 cm depth during each gas exchange campaign. Shown are the mean ±SE (n=6). Treatments are: A, unmanipulated control; T, passive night-time warming; D, experimental drought, active in July to early August; TD, warming an class="Disease">nd drought combined; CO2, elevated CO2; TCO2, warming and elevated CO2 combined; DCO2, drought and elevated CO2 combined; and TDCO2, warming, drought, and elevated CO2 combined. Statistically significant effects of single factors (T, D, and CO2) and interactions (T×CO2 and D×CO2) are shown with treatment letters (*** if P <0.0001, ** if P <0.001, * if P <0.05 and † if P <0.10).

Seasonal changes in leaf gas exchange

Pn generally showed high levels across treatments in May (range 8–16 μmol CO2 m−2 s−1), lower levels during June, July, August, an class="Disease">nd September (range 4–13 μmol CO2 m−2 s−1), and again high levels in October (range from 6 μmol CO2 m−2 s−1 to 11.5 μmol CO2 m−2 s−1) (Fig. 4). gs levels in the control plots decreased during May, June, and to July from 0.116 to 0.077 and 0.066 mmol H2O m−2 s−1. In August the gs in the controls was again high (0.195 mmol H2O m−2 s−1). It dropped in September to 0.090 mmol H2O m−2 s−1 and was intermediate in October, 0.132 mmol H2O m−2 s−1 (Fig. 5). The WUE across season varied from ∼3 μmol CO2 mmol H2O−1 to 9 μmol CO2 mmol H2O−1, with a seasonal low in August and the highest values in May and October (Table 1). When including the clear effect of month (P <0.0001), elevated CO2 significantly increased Pn, gs, and WUE (all P <0.0001) across seasons.

Fig. 4.

Light-saturated net photosynthesis at treatment CO2, Pn. Shown are the mean ±SE. Treatment abbreviations and statistics are as in Fig. 3.

Fig. 5.

Leaf stomatal conductance at treatment CO2, gs. Shown are the mean ±SE. Treatments abbreviation and statistics are as in Fig. 3.

Table 1.

For each month the F-values and significance levels (*P <0.05; ** P <0.01; *** P <0.0001) for the main effects night-time warming (T), drought (D), elevated CO2 (CO2), and their interactions on light-saturated net photosyntheis (Pn), stomatal conductance (gs), water use efficiency (WUE), maximal light- and CO2-saturated net photosynthesis (Pmax), maximal RuBP regeneration rate (Jmax), maximal rate of Rubisco carboxylation (Vcmax), leaf carbon to nitrogen ratio (C/N), and leaf δ13C by a linear mixed model ANOVA

Deschampsia flexuosa		T			D			CO2			T×D			T×CO2			D×CO2			T×D×CO2
Num df		1			1			1			1			1			1			1
Denom df		10			10			5			10			10			10			10
Pn	May	1.27			0.14			0.24			0.03			5.83	*	↑↑	0.05			1.37
	June	0.09			0.00			12.54	**	↑	0.01			10.81	**	↑↑	0.47			0.94
	July	0.31			5.15	*	↓	8.65	**	↑	0.98			0.56			0.01			3.80
	August	2.62			4.82	*	↓	2.57			0.09			5.02	*	↑↑	0.80			1.66
	September	0.18			8.47	**	↑	6.24	*	↑	2.63			0.01			13.41	**		1.08
	October	2.06			0.83			32.84	***	↑	0.29			0.72			0.96			0.01
gs	May	0.00			0.03			0.04			0.03			0.07			0.08			2.37
	June	0.03			0.25			19.03	**	↑	0.13			8.85	*	↑↑	0.21			0.79
	July	0.45			5.97	*	↓	0.98			1.50			0.66			0.05			1.78
	August	0.79			9.79	**	↓	0.01			0.26			0.06			1.21			0.04
	September	2.26			9.52	**	↑	0.14			0.01			2.19			0.01			4.83
	October	0.04			0.07			18.24	**	↑	0.03			2.47			0.01			0.10
WUE	May	5.46	*	↑	0.45			20.05	**	↑	0.31			0.53			2.69			1.29
	June	0.58			0.12			2.85			1.16			0.99			3.16			1.60
	July	3.65			0.83			10.99	**	↑	2.02			0.21			0.05			2.98
	August	1.83			1.48			1.34			1.23			3.90	†	↑↑	0.01			0.75
	September	2.89			1.34			14.44	**	↑	0.52			0.01			1.83			0.79
	October	1.48			0.23			30.94	***	↑	2.84			1.59			1.12			0.46
Pmax	May	7.31	*	↑	0.37			0.86			1.59			0.12			0.38			2.78
	June	0.03			1.00			10.13	**	↑	1.51			9.85	*	↑↑	3.13			2.02
	July	0.02			4.88	*	↓	1.81			0.11			4.64	*	↓↓	0.13			2.36
	August	1.69			5.75	*	↓	11.09	**	↓	0.30			0.85			1.04			0.19
	September	3.91			8.69	*	↑	0.68			0.64			0.44			6.87	*	↑↑	1.36
	October	8.59	*	↑	0.31			0.43			1.45			0.35			3.52	†	↑↑	0.13
Jmax	May	ND		ND	ND			ND			ND			ND			ND			ND
	June	0.45			0.01			5.56	*	↑	1.16			7.00	**	↑↑	2.54			0.78
	July	0.36			6.96	*	↓	2.10			1.00			7.89	*	↓↓	0.32			0.25
	August	0.33			6.07	*	↓	9.85	*	↓	0.17			0.94			0.26			2.84
	September	0.15			6.19	*	↑	7.13	**	↓	2.73			0.00			7.02	*	↑↑	0.05
	October	12.00	*	↑	0.63			0.10			0.27			0.23			4.15	†	↑↑	0.11
Vcmax	May	ND			ND			ND			ND			ND			ND			ND
	June	0.24			1.38			4.86	*	↓	2.43			0.86			0.34			0.00
	July	0.67			1.66			6.04	*	↓	2.32			7.60	*	↓↓	2.15			3.32
	August	0.01			9.73	*	↓	4.76	†	↓	3.07			1.12			0.95			1.10
	September	0.00			9.02	*	↑	9.47	*	↓	0.11			0.71			0.03			0.03
	October	0.64			0.96			0.39			2.53			1.49			0.76			0.04
C/N	May	1.18			1.26			5.82	*	↑	0.04			0.72			0.03			0.50
	June	1.79			0.68			6.30	*	↑	0.04			0.39			1.07			0.03
	July	0.62			2.31			8.72	*	↑	0.53			0.44			0.54			2.48
	August	0.45			0.26			0.71			0.04			0.19			0.17			0.02
	September	0.72			17.93	***	↓	15.69	***	↑	1.03			0.89			1.23			0.06
	October	0.17			1.88			2.75			0.00			0.23			3.50	†	↓↓	0.00
δ13C	May	0.48			1.42			74.49	***	↓	0.01			7.17	**	↓↓	0.80			0.04
	June	1.31			0.01			152.62	***	↓	0.12			12.52	**	↓↓	4.39			0.82
	July	1.34			0.23			364.90	***	↓	0.79			1.34			9.45	**	↑↑	0.18
	August	6.41	*	↓	0.09			428.33	***	↓	1.00			5.59	**	↓↓	3.08			0.27
	September	6.08	*	↓	0.03			361.69	***	↓	1.96			0.63			2.49			0.41
	October	15.38	**	↓	0.03			332.44	***	↓	3.25			0.71			1.34			1.47

Degrees of freedom (df), numerator (Num), denominator (Denom), not detected (ND) increase (↑), decrease (↓), synergistic increase (↑↑), synergistic decrease (↓↓).

For each month the F-values and significance levels (*P <0.05; ** P <0.01; *** P <0.0001) for the main effects night-time warming (T), drought (D), elevated n class="Chemical">CO2 (CO2), and their interactions on light-saturated net photosyntheis (Pn), stomatal conductance (gs), water use efficiency (WUE), maximal light- and CO2-saturated net photosynthesis (Pmax), maximal RuBP regeneration rate (Jmax), maximal rate of Rubisco carboxylation (Vcmax), leaf carbon to nitrogen ratio (C/N), and leaf δ13C by a linear mixed model ANOVA

Leaf stomatal conductance at treatment n class="Chemical">CO2, gs. Shown are the mean ±SE. Treatments abbreviation and statistics are as in Fig. 3.

Treatment responses in leaf gas exchange

Pn was higher in the TCO2 an class="Disease">nd TDCO2 treatments, as compared with T or CO2 alone, leading to a significant T×CO2 interaction in May, June, and August (Fig. 4). Elevated CO2 increased Pn in June, July, and October. Drought decreased Pn in July and August, but after rewetting the Pn was increased in the previously drought-treated plots in September. In September the D and CO2 in combination synergistically increased Pn (Fig. 4).

gs was increased in response to elevated CO2 in June an class="Disease">nd October (Fig. 5). In the TCO2 treatments, a T×CO2 interaction synergistically increased gs in June. Drought decreased gs in July and August, but after rewetting, gs increased in the previously drought-treated plots in September. WUE increased in elevated CO2 in May, July, September, and October (Table 1). Warming increased WUE in May and a T×CO2 interaction caused a synergistic increase in WUE in August. Intercellular CO2 was increased in elevated CO2 in all months, increased by warming in May and August, and a T×CO2 effect synergistically increased Ci in TCO2 and TDCO2 treatments in August and October (Table 1).

Pmax was significantly increased in response to warming in May an class="Disease">nd October (Fig. 6). Elevated CO2 decreased Pmax in June (Fig. 6). In the TCO2 and TDCO2 treatments a T×CO2 effect synergistically increased Pmax in June, but this was reversed in July where T×CO2 synergistically decreased Pmax. Drought decreased Pmax in the experimental drought period in July and beyond it into August, but after rewetting in September and October the Pmax increased in the experimental drought plots. In DCO2 and TDCO2, a D×CO2 interaction synergistically increased Pmax in September and October (Fig. 6).

Fig. 6.

Light- and CO2-saturated net photosynthesis, Pmax. Shown are the mean ±SE. Treatment abbreviations and statistics are as in Fig. 3.

Jmax demonstrated similar effects to Pmax, except that elevated n class="Chemical">CO2 was not observed to reduce Jmax in September (Table 1). The Vcmax was lowered in elevated CO2 in most months (June, July, August, and September). Experimental drought reduced Vcmax (August), but rewetting increased Vcmax in the former drought plots (September). In TCO2 and TDCO2, a T×CO2 interaction synergistically decreased Vcmax in July (Table 1).

Leaf characteristics

In all treatments the leaf characteristics of Deschampsia were related to the biphasic growth pattern, with two peaks of vegetative green biomass closely linked to the SWC. The first phase (May, June, an class="Disease">nd July) was characterized by biomass increments in May and June during ample SWC and a large dieback in July along with declining water availability. The second leaf phase (August, September, and October) was initiated by rewetting and was characterized by the leaves having a high nitrogen content and a balanced C/N ratio in contrast to the nitrogen dilution during first phase. The leaf C/N ratio increased in elevated CO2 in May, June, July, and September (Fig. 7). Drought decreased the C/N ratio in September, while the interaction detected (D×CO2) in October meant that the increased C/N ratio in elevated CO2 did not occur when combined with drought. In general, the changes in the C/N ratios were governed by differences in nitrogen concentrations (Fig. 7).

Fig. 7.

Leaf carbon to nitrogen ratio, C/N. Shown are the mean ± SE. Treatment abbreviations and statistics are as in Fig. 3.

Leaf water content increased in elevated n class="Chemical">CO2 in May, June, and September (P <0.006, P <0.009, and P <0.002, data not shown) and decreased in drought in July and August (P <0.04 and P <0.0001, data not shown). There were no significant effects on SLA. Leaf δ13C was decreased, in all months, under elevated CO2, as a result of uptake of the lower δ13C signature of the industrial CO2 gas (average air δ13C during the primary and secondary leaf phase in the elevated CO2 treatment was –10.6‰ and –13.8‰, respectively, as opposed to –8‰ in ambient air) added to the FACE plots (Fig. 8). Warming decreased the δ13C in August, September, and October, and a T×CO2 interaction was significant in May, June, and August due to a decrease in δ13C only when warming and elevated CO2 were combined. The interaction D×CO2 was significant in July due to an increase in δ13C in response to drought only when combined with elevated CO2 (Fig. 8).

Fig. 8.

Natural abundance of leaf 13C, δ13C. Shown are the mean ±SE. Treatment abbreviations and statistics as in Fig. 3. Note that the CO2 gas dosed in the FACE plots had a lower δ13C signature of –10.6‰ δ13C (May, June, and July) and of –13.8‰ δ13C (August, September, and October) as opposed to –8.0‰ in ambient air. This caused the significantly lower δ13C values in leaves in the elevated CO2 treatment.

Natural abundance of leaf n class="Chemical">13C, δ13C. Shown are the mean ±SE. Treatment abbreviations and statistics as in Fig. 3. Note that the CO2 gas dosed in the FACE plots had a lower δ13C signature of –10.6‰ δ13C (May, June, and July) and of –13.8‰ δ13C (August, September, and October) as opposed to –8.0‰ in ambient air. This caused the significantly lower δ13C values in leaves in the elevated CO2 treatment.

The slopes of regressions between leaf δ13C an class="Disease">nd WUE were negative in both elevated CO2 and ambient CO2 (P <0.05, Fig. 9a). Correcting δ13C for the difference in air δ13C signatures between phases in elevated CO2 resulted in similar linear regression slopes (α) for the δ13C versus WUE relationship in elevated and ambient CO2 (α= –0.29). Lower soil water conditions occurred during the primary phase compared with the secondary leaf phase, and the leaf δ13C did not vary significantly with SWC at 0–20 cm or 0–60 cm across treatments (Fig. 9b, d). The relationships between leaf δ13C and Ci/Ca were positive, with higher α in elevated compared with ambient CO2 (Fig. 9c).

Fig. 9.

Regressions with leaf carbon isotope ratio (δ13C). (a) The leaf δ13C versus water use efficiency. (b and d) the leaf δ13C versus the soil water content during each of the leaf gas exchange campaigns at 0–20 cm and 0–60 cm. (c) The leaf δ13C versus the ratio between the intercellular CO2 concentration and the CO2 concentration outside the leaf. Points are the averages from each treatment per month. Leaves from the primary leaf phase (May, June, and July) are indicated by filled symbols and leaves from the secondary leaf phase (August, September, and October) by open symbols. Circles are elevated CO2 (FACE) and triangles ambient CO2 (non-FACE). Note the average air δ13C during the primary and secondary leaf phase under elevated CO2 was –10.6‰ and –13.8‰, respectively. In ambient CO2 the average air δ13C was –8‰. This caused the offsets in leaf δ13C between the leaf phases and ambient versus elevated CO2. Only significant regression lines with slope (α) different from zero are depicted (P <0.05).

Regressions with leaf carbon isotope ratio (δn class="Chemical">13C). (a) The leaf δ13C versus water use efficiency. (b and d) the leaf δ13C versus the soil water content during each of the leaf gas exchange campaigns at 0–20 cm and 0–60 cm. (c) The leaf δ13C versus the ratio between the intercellular CO2 concentration and the CO2 concentration outside the leaf. Points are the averages from each treatment per month. Leaves from the primary leaf phase (May, June, and July) are indicated by filled symbols and leaves from the secondary leaf phase (August, September, and October) by open symbols. Circles are elevated CO2 (FACE) and triangles ambient CO2 (non-FACE). Note the average air δ13C during the primary and secondary leaf phase under elevated CO2 was –10.6‰ and –13.8‰, respectively. In ambient CO2 the average air δ13C was –8‰. This caused the offsets in leaf δ13C between the leaf phases and ambient versus elevated CO2. Only significant regression lines with slope (α) different from zero are depicted (P <0.05).

Discussion

Responses to elevated CO2

When water availability was declining (May–July), the C/N ratio was higher in the elevated n class="Chemical">CO2 treatment, indicating that some of the additional assimilated carbon was allocated to leaf growth without a corresponding allocation of nitrogen. As some level of soil moisture is necessary for the mineralization processes, this probably influenced the C/N ratios. In the longer term, nitrogen limitation very probably constrains the potential for increased productivity in response to CO2 (Finzi ; Lou ; Hungate ; Reich ), but in this dry ecosystem it may be difficult to separate this effect from the effect of the ample water conditions.

The Pn was higher in elevated CO2, mainly driven by higher n class="Chemical">Ci during most of the growing season in accordance with the first hypothesis proposed here. Photosynthetic capacity regulations were dynamic, but down-regulation took place with Vcmax reduction in most months, whereas Pmax and Jmax responses were more stable and only declined in the second leaf phase after rewetting. These responses are consistent with most findings on photosynthetic responses to elevated CO2 (Ainsworth and Long, 2005; Ainsworth and Rogers, 2007; Leaky ). The WUE was higher in elevated CO2, but not via gs reduction even in the driest month, July, and therefore the higher WUE in this study was caused by a higher Pn which is in contrast to several other studies with other species (Ainsworth and Long, 2005; Ainsworth and Rogers, 2007).

In contrast to the first hypothesis proposed here, no general reduction in gs was seen under elevated n class="Chemical">CO2. Complex gs responses have been reported, and are purported to depend on the degree of drought and the species-specific water demand; the gs net response can be either positive, negative, or zero, owing to the antagonistic nature of direct (reducing) and indirect (enhancing) CO2 effects on gs (Knapp ; Volk ). In the present study the observed gs values were actually higher in the elevated CO2 treatment, along with Pmax and Jmax up-regulation, which in combination increased Pn in June. This demonstrates the dynamic capability of the photosynthetic capacity in Deschampsia, which may be related to the peak of the first phase of leaf growth and in combination with the relatively high June SWC level (10–15%). Further, δ13C, an integrating measure of the response of WUE, did not vary significantly with SWC, but instead with Ci/Ca. This indicates both opportunistic water consumption in Deschampsia and also that the up-regulation of photosynthetic capacity was the primary cause of the reduction of Ci/Ca.

It was hypothesized that elevated CO2 would sustain Pn in dry periods via soil n class="Chemical">water savings; this is supported by the data obtained. Other studies report improved water relations in elevated CO2 to originate mainly from gs reductions in combination with LAI reductions (Körner, 2000; Volk ; Morgan ). In the present study no general gs reductions were observed; other factors such as LAI reduction due to leaf wilting or the influence of co-occuring species could be proposed to explain the water savings that developed in the late part of the dry period in elevated CO2 plots. The gs of the co-occurring C. vulgaris decreased in response to dry conditions, but not as a general response to elevated CO2, even when the soil water savings developed, partly excluding this option to explain the observed phenomenon (Albert ).

As reported in other climate change experiments, drought was the main factor determining grass biomass (Beier ; Penuelas , 2007; Damgaard ; Prieto ). In elevated CO2, spen class="Chemical">cies-specific responses have been shown; some species increase biomass whereas other species remain unchanged (Ainsworth and Long, 2005; Maestre and Reynolds, 2007; Leakey ). This indicates that biomass reductions could have taken place in response to drought and that biomass could have been higher in elevated CO2. Therefore, the opportunistic water consumption alone, or in combination with high biomass, probably led to depletion of soil water to a threshold near the wilting point in elevated CO2 in late July. Thereafter, in response to the very low July soil water availability, gs was at its seasonal low and above-ground grass biomass probably died back. Consequently, this leaf wilting left the gs in the remaining leaves unaffected, but Pn was stimulated. This complex balance between the opportunistic water consumption and drought-induced leaf wilting of Deschampsia caused the modest soil water savings observed under elevated CO2. Such response patterns provide further support for the concept that limitations on photosynthetic improvement in elevated CO2 are governed by water relations, as shown by several other FACE experiments (Körner, 2000; Volk ; Morgan ; Leuzinger and Körner, 2007), but the opportunistic water consumption and variable grass biomass complicate the picture and need further investigation.

Responses to passive night-time warming and elevated CO2 in combination

In accordance with the hypothesis proposed herein, warming improved photosynthetic capacity (n class="Chemical">Pmax and Jmax) in the early and late season, although, contrary to this hypotheses, this did not affect Pn. This could be due to higher leaf respiration, since leaf respiration increases with temperature (Atkin and Tjoelker, 2003). However, it does clearly indicate an earlier maturation of photosynthetic capacity in response to warming, which is in line with other studies (Penuelas and Fiellala, 2001; Cleland ; Menzel ; Prieto ) showing an advancement of the growing season. Warming also appeared to shift the timing of the second phase leaves as δ13C was reduced in warmed plots in August, September, and October. In this period, water supply was sufficient; there were no significant effects of warming on Pn or gs. This indicates that the effect of warming on δ13C was caused by an earlier leaf emergence allowing for a longer period of carbon uptake and thus discrimination against the 13C.

The effects of warming and elevated n class="Chemical">CO2 were not additive as hypothesized; T×CO2 synergistically increased Pn in parallel with T×CO2 synergistically decreasing δ13C (May and June). Further, T×CO2 synergistically increased June Pmax and Jmax, but in July this reversed. These responses clearly demonstrate that T×CO2 increases plant carbon uptake early in the growing season and that this is closely related to the regulation of photosynthetic capacity. This shows that earlier maturation of photosynthetic capacity in response to warming is a prerequisite for the interactive stimulation of Pn when combined with elevated CO2, early in the growing season.

After rewetting, the T×CO2 interaction synergistically increased August Pn an class="Disease">nd decreased δ13C, but this increased plant carbon uptake was not governed by photosynthetic capacity regulations. While the warming treatment resulted in an extended period of carbon uptake (see above), this cannot explain the profound short-term stimulation of Pn. Therefore, it is suggested that the T×CO2 stimulation of Pn could be linked to better access to soil water mediated by extended root growth, facilitated by the previous month's high carbon uptake in the plots with both warming and elevated CO2. Alternatively, it may be the result of a complex mechanism facilitating enhanced export of photoassimilates to sink tissues via increased night-time respiration in response to warming (Turnbull , 2004) or direct stimulation of leaf respiration by elevated CO2 (Leaky ). In a parallel study at the experimental site the grass biomass was synergistically increased in response to the combination of warming and elevated CO2 (Andresen ). These responses demonstrate that T×CO2 synergistically increased plant carbon uptake to be strongly influenced by photosynthetic capacity up-regulation early in the season. After rewetting, the processes are less clear, but mechanisms stimulating the carbon sink strength in the combination of elevated CO2 and warming could be involved.

Responses of drought and elevated CO2 in combination

In accordance with the hypothesis proposed herein, the experimental drought decreased Pn via gs reduction (July), but also via photosynthetic capacity (n class="Chemical">Pmax, Jmax, and Vcmax) down-regulation under severe water shortage (August). Gradually drier conditions in general reduce above-ground Deschampsia biomass, and the experimental drought treatment also reduced the vegetation coverage (J. Kongstad et al., personal communication). These responses demonstrate the intensity of the experimental drought to push the response of Deschampsia beyond the threshold where dry conditions could be handled solely by acclimation, and leaf wilting occurred.

During dry conditions in the combination of elevated n class="Chemical">CO2 and drought, there were no indications of SWC being higher than in the control. Although a D×CO2 interaction did increase the SWC above what was expected from the single factors, the resulting SWC was relatively lower in the DCO2 plots compared with controls, but above the level in the D plots (Figs 2, 3). The absence of beneficial water savings was reflected in the D×CO2 interaction inducing the higher δ13C in DCO2 and TDCO2 plots. In other grassland studies, elevated CO2 has resulted in soil water savings (Bremer ; Niklaus ; Morgan ), mainly brought about by gs reduction along with higher potential to sustain Pn under dry conditions (Ainsworth and Long, 2005; Lauber and Körner, 1997; Robredo ). As discussed above, such gs reductions were not seen in elevated CO2 to counteract the negative influence of the drought. Furthermore, the experimental drought caused very strong negative effects on photosynthetic performance, explaining why the experimental drought responses seem to dominate the physiological responses in the DCO2 and TDCO2 treatments.

Rewetting more than reversed all negative eco-physiological effects induced by the drought. After the experimental drought treatment, Pn, gs, n class="Chemical">Pmax, and Jmax increased and the C/N ratio decreased. In addition, the D×CO2 interaction synergistically increased the photosynthetic capacity (Pmax and Jmax) mediating parallel stimulatory effects on Pn in September and October. These responses to rewetting indicate that better nitrogen availability, perhaps in combination with pronounced re-allocation from rhizomes, does control the magnitude of Pn beyond the drought period. Increased nitrogen availability could also explain why the C/N ratio did not increase in elevated CO2 in the autumn even though Pn was increased. These variable responses show that Pn, in the combination of drought and elevated CO2, depends on soil water availability, with additive effects on Pn when SWCs are low and synergistic D×CO2 effects on Pn after rewetting.

Influence of T×CO2 and D×CO2 interactions on carbon uptake is controlled by photosynthetic capacity and plant available water

The heathland ecosystem investigated can be considered to be fully developed, with the LAI an class="Disease">nd fine root renewal at steady state, and can as such be compared with mature forest ecosystems as conceptualized by Körner (2006). In such a system, the stimulation of biomass production may to a large degree result from CO2-induced improved water relations and to a lesser degree directly from elevated CO2. The Deschampsia responses in the present study add complexity to this concept as the per leaf area water consumption was opportunistic and the biphasic growth pattern allowed for leaf dieback at low soil water availability and rapid re-growth of new active leaves when rewetted. This growth characteristic with an opportunistic resource exploration allowed for photosynthetic capacity regulations that mediated the synergistic T×CO2 and D×CO2 interaction effects on photosynthesis. These characteristics are advantageous when exposed to climatic changes, and the carbon uptake of this species is likely to be strongly influenced by changes in precipitation patterns, such as event frequency, intensity, magnitude, and interannual variability. More long-term experimentation including years with different variability in precipitation and the warming impact at the shoulders of the growing season is needed to elaborate these findings. In conclusion, after 1 year of experimentation the limitations imposed by low soil water availability and stimulation in the early and late season by warming clearly structure and interact with the photosynthetic response to elevated CO2 in this grassland species.