Author for correspondence: J. B. Shapiro Tel: +44 1845 3658395 Fax: +44 1845 3658150 Email: firstname.lastname@example.org
• The leaf-level response of respiration in the light (RL) is a vital component of a plant's energy and carbon balance.
• Xanthium strumarium (common cocklebur) plants were grown in various combinations of CO2, nitrogen and temperature, and RL was measured using the Kok effect method.
• RL was significantly lower than respiration in the dark (RD), with the lowest percentage inhibition in the elevated CO2, high-N treatment. In general RL increased in response to increased CO2 concentration and N availability across all temperature treatments. However, there was a significant interactive effect of growth CO2 concentration and measurement temperature on RL which indicated that RL responded more positively to temperature changes in elevated CO2 conditions. Additionally, across all CO2 and N treatments the percentage of respired C with respect to assimilated C increased as temperature increased.
• Collectively, these results improve our understanding of the magnitude and sensitivity to foreseen environmental changes of mitochondrial respiration during light hours.
Motivated by projections of extensive climate change over the next century, substantial research has been conducted to characterize and understand the photosynthetic response of terrestrial vegetation to elevated CO2 concentrations. Results from CO2 enrichment experiments conducted on plants grown in controlled-environment growth chambers, outdoor open-top chambers and large-scale FACE rings indicate that elevated concentrations of CO2 significantly increase photosynthetic rates (Norby et al., 1999; Smith et al., 2000; Herrick & Thomas, 2001). Although it has been well established that plants respire approximately half the carbon that is photosynthetically fixed (Ryan, 1991), the metabolic process of respiration has been less intensively studied with respect to its response to predicted environmental changes (Brooks & Farquhar, 1985; Thomas et al., 1993; Wang et al., 2001). Specifically, little is known about the extent to which elevated CO2 concentrations, nitrogen availability and temperature change affect mitochondrial respiration that occurs in the light (Sharp et al., 1984; Thomas et al., 1993; Wang et al., 2001; Tissue et al., 2002). Obtaining a better understanding of the mechanistic response of respiration in the light to the aforementioned climate change variables has the potential to aid in producing more precise estimates of global carbon cycle budgets.
Mitochondrial respiration in the light influences, and is influenced by, various plant processes that take place only during hours of illumination. ATP derived from oxidative phosphorylation is necessary for sucrose synthesis; otherwise glucose and fructose build up in the cytosol and feed back negatively on Calvin cycle activity (Kromer, 1995). It is also thought that the respiratory electron transport chain aids in modification of the redox state of the stroma during photosynthetic activity (Foyer & Noctor, 2000). In addition, respiratory pathways produce organic precursors that are necessary for metabolic synthesis in the light; notably, C skeletons derived from the tricarboxylic acid cycle are necessary for amino acid synthesis in the light (Kromer, 1995; Villar et al., 1995). These cumulative demands for energy and C skeletons can modulate the degree of mitochondrial respiration in the light (Wang et al., 2001), which has the potential to change in response to variations in atmospheric CO2 concentration, N availability and ambient temperature.
Despite the importance of respiration in the light to plant metabolism and C balance, and observed differences between dark respiration rates and respiration rates in the light, the effect of elevated CO2 concentration on respiration in the light (RL) has received little attention (Thomas et al., 1993; Wang et al., 2001; Tissue et al., 2002). Results from Wang et al. (2001) using the herbaceous species Xanthium strumarium L. indicate that plants grown in elevated CO2 conditions show a lower rate of RL compared with respiration in darkness (RD), and that both RL and RD are higher in plants grown in elevated CO2 compared with plants grown at ambient CO2. It is of particular significance that the percentage inhibition of RL is lower for plants grown at elevated CO2 (17–24%) than for those grown in ambient CO2 (29–35%). This difference in inhibition is attributed to the greater demand for energy and C skeletons, and the greater availability of respiratory substrate in elevated vs ambient CO2 conditions (Dewer et al., 1999; Atkin et al., 2000a; Wang et al., 2001).
Although not studied in conjunction with the effect of elevated CO2 concentration, data indicate a variable response of C loss through RL and RD to increases in temperature (Sharp et al., 1984; Villar et al., 1995; Atkin et al., 2000b). Atkin et al. (2000b) report that RL for the evergreen species Eucalyptus pauciflora was relatively unresponsive to increases in temperature in the range 6–30°C when measured at high irradiance. By contrast, Villar et al. (1995) find that RL and RD for deciduous species increase with increasing temperature in the range 10–30°C.
The response of respiration to climate change will be a major factor in determining whether forests are a net C source or a sink (Drake et al., 1999; Valentini et al., 2000). Given that respiration is clearly a fundamental factor in understanding and modeling C budgets at both leaf and ecosystem levels, consideration should be given to determining the most representative values for each individual respiratory flux, including that of mitochondrial respiration during light hours. It is therefore important that global C cycle models incorporate corrected values of respiration by using RL in place of RD when modeling C balance during the day, which would have the effect of increasing the total amount of C gained, and by using RL values that are indicative of the sensitivity to variation in both temperature and CO2 concentration. In doing so, the individual responses of RL and photosynthesis to such environmental changes, which may be dissimilar, would be more accurately represented.
The following research investigates and quantifies the individual and interactive effects of temperature, CO2 concentration and leaf N availability on RL and RD in the herbaceous species X. strumarium. Specifically, we hypothesized that individual increases in CO2 concentration, N treatment and measurement temperature would affect RL and RD positively. Furthermore, we examined the collective effects of these variables on RL and RD to understand the leaf-level response of vegetation to a dynamically changing environment. Additionally, although data from previous studies have suggested that RL and RD are not necessarily equivalent, a further goal of this research is to confirm whether RL is significantly lower than RD across the aforementioned combinations of environmental conditions. The possible effect of leaf N availability on the response of RL to CO2 concentration and measurement temperature is studied because, in most cases, respiration can be related either to leaf N concentration or to leaf carbohydrate content, and the relative proportion of the two can vary in response to changes in environmental conditions (Amthor, 1989; Ryan, 1991).
Materials and Methods
Seed of Xanthium strumarium L. (common cocklebur), an annual C3 herbaceous plant, used in this experiment were from a single population in Lubbock, TX, USA. On 1 July 2002 two seeds were planted in each of 40 6 l pots filled with sand; 10 pots were placed into each of four 1.4 m2 controlled-environment growth chambers (E-15, Conviron Inc., Winnipeg, Manitoba, Canada) at the Lamont–Doherty Earth Observatory, Palisades, New York, USA. Seedlings emerged c. 7 d after planting, and were thinned to one per pot 7 d after emergence. Two of the growth chambers were programmed to supply 365 µmol mol−1 CO2 (ambient CO2 treatment), and two were programmed to supply a CO2 concentration of 730 µmol mol−1 CO2 (elevated CO2 treatment). All growth chambers maintained a daytime/night-time temperature of 28/22°C and a relative humidity of c. 50%. Leaf-level photosynthetic photon flux density was maintained at c. 500 µmol m−2 s−1 in all the chambers throughout the 9 wk experiment. Plants were maintained in their vegetative growth stage throughout the experiment with a light/dark regime of 18/6 h.
To study the effect of N availability on RL and RD in elevated and ambient CO2, half the plants in each chamber were fertilized to saturation with a half-strength Hoagland's solution containing a high N concentration of 7.0 mm NH4NO3, and the other half were fertilized to saturation with a half-strength Hoagland's solution containing a limiting N concentration of 1.5 mm of NH4NO3. The concentration of the additional nutrients in both high- and low-N fertilizers were as follows: 3.0 mm K as KCl and KH2PO4 monobasic, 3.0 mm Ca as CaCl2 dihydrate, 1.5 mm Mg as MgCl2 hexahydrate, 1.0 mm P as KH2PO4 monobasic, 2.0 mm S as Na2SO4, 0.14 mm Fe as Fe-EDTA, 0.05 mm B as boric acid, 0.01 mm Mn as MnCl2 4-hydrate, 0.001 mm Zn as ZnO, 0.001 mm Cu as CuSO4 and 0.05 µm Mo as MoO3. Nutrient additions began 7 d after emergence and were repeated 3 d wk−1 for the remaining 7 wk of the experiment. Plants were watered until saturation with deionized water on days when they did not receive fertilizer solution.
Two methods commonly used to measure RL are the Laisk method and the Kok method. The Laisk method measures photosynthesis at varying irradiance and at low internal CO2 concentrations where C fixation and photorespiration are balanced; at that concentration, additional CO2 evolution is attributed to RL. As pointed out by Villar et al. (1994), the main drawback of using this method is that measurements must be made at a CO2 concentration that is far from a given growth CO2 concentration. Because it is not known whether there is a direct short-term effect of CO2 concentration on RL, we chose not to use this method in an effort to avoid unnecessary complications in interpreting the results. By contrast, the Kok method used in this study measures the response of photosynthetic rate over incrementally decreasing irradiance, and can be measured at a given growth CO2 concentration. As applied and referred to in this experiment, the Kok method produced measurements of the long-term effect of growth CO2 concentration on physiological function, not the short-term effect of measurement CO2 concentration on such functions. The Kok effect, initially observed by B. Kok in 1948, specifically refers to the break in the slope of the measured photosynthetic rate in a light response (AQ) curve when taken with high resolution over low irradiances. At very low irradiance the slope, or quantum yield of photosynthesis, is relatively steep; at the vicinity of the light compensation point a distinct break occurs and the slope decreases. The line at irradiances below the break extends to RD, where it is determined at 0 PAR, and the line at irradiance above the break extrapolates to RL (Kok, 1948; Sharp et al., 1984; Kromer, 1995; Padmavathi & Raghavendra, 2001; Fig. 1). At high irradiances photosynthetic rate saturates, so when calculating RL the only points included are those in the linear section of the AQ curve.
During the eighth week of vegetative growth AQ curves were measured on the youngest fully expanded leaf of three randomly selected plants per treatment per chamber, using an open-flow gas-exchange system (LI-6400, Li-Cor, Lincoln, NE, USA); however, in one of the ambient CO2 chambers only one plant was suitable for measurement in the low-N treatment. AQ curves were generated using a red : blue light source (LI-6400-02B) at 20 light levels, 15 of which were < 150 µmol m−2 s−1 in an effort to have high resolution of the photosynthetic and respiratory responses at low irradiances. For each individual plant AQ curves were measured at the CO2 partial pressure of the particular growth environment (either 365 or 730 µmol mol−1 CO2), and were measured at three temperatures (23, 28 and 33°C) by adjusting the temperature in the cuvette and allowing conditions in the cuvette to equilibrate for c. 20 min before the AQ curve was taken. AQ curves were analyzed using photosynthesis assistant software (version 1.1.2, 1998, Dundee Scientific, Dundee, UK) from which a value was obtained per AQ curve for the light-saturating rate of net photosynthesis (Asat). Values for Asat for each individual plant reflected the same growth and measurement conditions as those for the corresponding measurements of RL and RD. Leaf tissue samples for C : N analysis were taken from each plant after the AQ curve was completed. Samples were dried, ground and analyzed using a NCS 2500 Elemental Analyzer (Carlo Erba NCS 2500, Milan, Italy).
Vital corrections were applied to the extrapolated values of RL as well as the measured values of RD to compensate for suspected intrinsic limitations of employing the LI-6400 when measuring relatively low gas-exchange rates associated with leaf respiration (Pons & Welschen, 2002). In order to correct for potential overestimations of measured values of RL and RD when using a clamp-on leaf chamber, as opposed to a leaf chamber that accommodates an entire leaf, values for RL and RD were recalculated to account for the diffusion of respired CO2 from darkened leaf material under the chamber gasket that travels into the leaf chamber. Specifically, measured values of RD were corrected by recalculating measured values to include the inward gasket area that is in contact with the leaf during gas-exchange measurements. Pons & Welschen (2002) used a simple model that partitioned the gasket area in half, i.e. into an inward and outward gasket area, and assumed that the CO2 evolved under the inward gasket area diffused into the leaf chamber and contributed to the measured value of RD. Results from this model were in good agreement with measured values of RD when taken with a cuvette that enclosed an entire leaf (Pons & Welschen, 2002). In addition, Pons & Welschen (2002) reported that overestimations of RL using a leaf chamber that does not enclose an entire leaf are greater than those associated with measurements of RD using the same apparatus. They attribute this to the high concentration gradient between the leaf chamber and the room in which the measurements were taken. However, the Kok method for measuring RL used in this experiment does not rely on a difference in CO2 concentration between the chamber and the room in which the measurement was taken any more than measuring RD does, and it was possible to make corrections to RL in a similar manner to those for RD. The extrapolated value of RL was recalculated according to the simple equation:
recalculated value of RL = [(extrapolated value of RL × 6 cm2) − (recalculated value of RD × 3.55 cm2)]/6 cm2
where 6 cm2 is the area of the window of the Li-Cor 6400 and 3.55 cm2 is the area of the inward side of the gasket as reported by Pons & Welschen (2002). These corrections, once applied, reduced original values of RL and RD by an average of 1.14 (± 0.047) µmol m−2 s−1. All values of RD and RL reported here refer to corrected values. Values for Asat were not corrected because errors associated with diffusion through the gasket are not detectable when measuring photosynthetic rate at light saturation (Pons & Welschen, 2002).
Q10 values for respiration, which compare the rate of respiration at one temperature to the rate at a second temperature that is 10° C offset from the first, were calculated for RL and RD over the measured temperature. Interval from 23 to 33°C for each CO2 by N treatment using the following equation:
The variables R2 and R1 refer to the measured respiration rate at high and low temperatures, respectively, and T2 and T1 refer to the specific high and low temperatures (°C) at which the respiration measurement was taken. Q10 values are presented primarily to facilitate comparisons with existing literature values, but in general should be used with caution (Tjoelker et al., 2001; Turnbull et al., 2001; Atkin & Tjoelker, 2003).
Temperature sensitivity of RL and RD of each treatment was calculated using a modified Arrhenius equation of the form:
where R is the respiratory rate (either RL or RD) measured at a specific temperature Ta (K), Rg is the molar gas constant 8.314 J mol−1 K−1, To (K) is the temperature at which Ro was calculated, and Eo is a fitted parameter and represents a temperature coefficient related to the overall activation energy for the metabolic process of respiration (Lloyd & Taylor, 1994; Turnbull et al., 2001). The equation was solved using a nonlinear regression function (Data Desk, Data Description, Inc., Ithaca, NY, USA, 1996).
A two-way ANOVA, with CO2 concentration (365 vs 730 µmol mol−1 CO2) and N treatment (7.0 mm vs 1.5 mm NH4NO3) as main effects, and respiration measurement (RD vs RL) and measurement temperature (23, 28 and 33°C) as repeated measures, was used to test for significance between subjects as well as within subjects. Initially the statistical model design included the CO2 growth treatment as a nested effect within the variable designated for CO2 growth chamber. Because the statistical significance of the chamber effect was low (P = 0.82) the analyses were rerun without chamber as a variable. A second ANOVA with an identical configuration was performed on RL : Asat (%) and RD : Asat (%) values (Data Desk). A multivariate analysis of variance (MANOVA) with R28 (respiration normalized to growth temperature, 28°C) and Eo28 (temperature coefficient at 28°C) as the dependent variables, CO2 concentration and N treatment as the main effects, and respiration measurement (RL vs RD) as a repeated measure variable was used to test for significance on the output of the Arrhenius model. In all analyses, effects were considered significant at P ≤ 0.10.
A main objective of this study was to determine whether RL maintained a unique value lower than RD over several experimentally imposed environmental variables. Statistical analyses indicate that RL values were significantly and consistently lower than RD values across CO2 and N treatments and across consecutive increases in measured temperature (Table 1). Additional hypotheses addressed the individual treatment effects of CO2 concentration and N availability across measurement temperature and forms of respiration. Both main effects were found to be significant (Table 1) such that both RL and RD significantly increased with an increase in CO2 growth concentration or an increase in N availability. Furthermore, the repeated-measures effect of measurement temperature was found to be highly significant (Table 1), where an increase in measured temperature was positively associated with an increase in RL and RD across all CO2 and N treatments.
Table 1. Summary of degrees of freedom (df), F ratios and statistical significance (P values) for the repeated-measures analysis of variance (ANOVA) on corrected values for Xanthium strumarium for respiration, with CO2 concentration and nitrogen availability as the between-subject factor and respiration measurement (RL vs RD) and measurement temperature (23, 28 and 33°C) as the within-subject factors.
CO2 Concentration (CO2)
Nitrogen Treatment (N)
CO2 * N
Respiration Measurement (R)
Measurement Temperature (T)
CO2 * R
N * R
CO2 * N * R
CO2 * T
N * T
CO2 * N * T
R * T
CO2 * R * T
N * R * T
CO2 * N * R * T
Another objective of this study was to examine whether RL and RD differed in their relative responses to the combined effects of CO2 and N supply. Although RL was consistently lower than RD across treatments, and both RL and RD increased with increasing CO2 and N supply, RL and RD differed significantly in their relative responses to the combined effects of CO2 and N supply (Table 1; Fig. 2). For RL the combined effects of increasing both CO2 and N had a strong positive effect that was substantially greater than for the other three treatments. By contrast to this interactive effect of CO2 and N on RL, there was no significant interaction between CO2 and N on RD. This differential response of RL and RD to the elevated CO2, high-N treatment (CN) was similarly reflected in the percentage inhibition of RL relative to RD for each CO2 × N treatment combination. The percentage inhibition of RL relative to RD was 28% on average for the CN treatment, but 48% for the ambient CO2, high-N treatment (cN); 53% for the elevated CO2, low-N treatment (Cn); and 48% for the ambient CO2, low-N treatment (cn). The overall CO2 × N effect on respiration measurement was constant across the three temperatures measured (Table 1).
A second set of interactions of interest in this experiment were the interactive effects of CO2 concentration and measurement temperature on RL and RD. This interaction was statistically significant across RL and RD (Table 1), and indicated that the respiratory flux per CO2 treatment was positively and differentially affected by measurement temperature (Fig. 3). The response of RD and RL to the two CO2 treatments was not statistically different when measured at 23°C, but as measurement temperature increased from 23 to 28°C and from 28 to 33°C the trajectory of the response surface of the elevated and ambient CO2 treatments diverged, with the response to the elevated CO2 treatment becoming progressively greater.
Although N availability as a main effect had a significant positive effect on RL and RD across CO2 growth concentration and measurement temperature, the interactive effect of N and measurement temperature was not statistically significant (Table 1). The relationship between respiration rates and N availability was examined further by determining whether respiration rates were correlated to leaf N concentration (%). While the statistical significance appears marginal, results indicate that RL had a stronger positive correlation with leaf N concentration (%) in both ambient (R2 = 0.48) and elevated (R2 = 0.45) CO2 treatments than those between RD and leaf N concentration (%) (R2 = 0.42 and R2 = 0.15, respectively).
Q10 values for RL and RD over the measured temperature interval of 23 to 33°C were calculated (equation 1) for each of the four CO2 × N treatments (Table 2). For plants in the cN and cn treatments, mean (± SE) Q10 for RL was c. 2 (2.04 ± 0.44 and 1.89 ± 0.27, respectively). The Q10 values of RL for plants grown in the CN and Cn treatments were substantially higher at 3.33 ± 0.66 and 4.91 ± 0.59, respectively. These latter, high Q10 ratios may reflect the small absolute values of RL measured at low temperatures, and as such the relative differences in RL become magnified over the temperature range measured. Similar trends in Q10 values across all treatments were observed for calculated values of RD.
Table 2. Treatment-specific values for Xanthium strumarium: calculated Q10 for RL and RD over the measured temperature interval of 23 to 33°C; normalized values for RL at 28°C from the Arrhenius model; temperature coefficients (Eo) for RL at 28°C from the Arrhenius model; normalized values for RD at 28°C from the Arrhenius model; temperature coefficients (Eo) for RD at 28°C from the Arrhenius model; and light-saturating rates of net photosynthesis. Values are means ± SE of four to six replicates
Q10 for RL
Q10 for RD
RL28µmol m−2 s−1
J mol−1 K−1
RD28µmol m−2 s−1
J mol−1 K−1
Asatµmol m−2 s−1
Elevated CO2, high nitrogen (CN)
3.33 ± 0.66
3.37 ± 0.29
1.61 ± 0.24
75199 ± 19693
2.26 ± 0.22
71794 ± 22726
54.73 ± 4.10
Ambient CO2, high nitrogen (cN)
2.01 ± 0.44
2.00 ± 0.27
1.01 ± 0.15
53784 ± 13543
2.02 ± 0.10
37415 ± 7457
32.07 ± 3.22
Elevated CO2, low nitrogen (Cn)
4.91 ± 0.59
3.00 ± 0.69
0.855 ± 0.14
68356 ± 15669
1.76 ± 0.11
59041 ± 3975
27.97 ± 4.07
Ambient CO2, low nitrogen (cn)
1.89 ± 0.27
1.94 ± 0.12
0.707 ± 0.13
80143 ± 52602
1.47 ± 0.32
49857 ± 9033
20.25 ± 1.90
Using the modified Arrhenius equation previously described (equation 2), and the measured RL and RD values, normalized values of R28 and E028 for both RL and RD were calculated for each set of temperature measurements from each plant (Table 2). These values were used to generate temperature-sensitivity curves for each CO2 × N treatment combination over the expanded temperature range of 20 to 35°C (Fig. 4a,b). Both CO2 and N treatment were significant main treatment effects (P = 0.06 and P < 0.01, respectively) on R28 and E028. Elevated CO2 grown plants show a significantly more positive respiratory response (a steeper slope) as temperatures increase, with this effect compounded by an increase in N availability (Fig. 4) and distinct for each respiration measurement (RD or RL) (P = 0.08). This result expands on the trend observed in Fig. 3 and demonstrates the presence of treatment-specific temperature sensitivity, particularly in the elevated CO2, high-N treatment (Fig. 4a).
Values for RL : Asat (%) were consistently and significantly lower than those for RD : Asat (%) across CO2 and N treatments and short-term temperature measurements (P < 0.001; Fig. 5). Additionally, Fig. 5 shows that the average leaf RL and RD as a percentage of Asat in the N × temperature treatment range in values from 2.4–4.4 and 4.6–8.1%, respectively. The statistically significant trend (P = 0.06) for both RL : Asat (%) and RD : Asat (%) values was an increase in percentage respiration with respect to Asat over temperature increases across all four CO2 × N treatments.
Global and significant increases in mean annual temperature and CO2 concentration and regional increases in N deposition are predicted for the near future (Vitousek et al., 1997; IPCC, 2001). It is therefore imperative that studies of physiological processes of terrestrial vegetation take these environmental factors into consideration. In order to gain a complete picture of how a plant's total C balance can be affected by foreseen environmental changes, it is not sufficient to measure only the net photosynthetic response of vegetation to such climate change variables. Instead, the individual effects of both photosynthetic C gain and respiratory C loss must be taken into account if a mechanistic, predictive understanding is to be gained.
This experiment demonstrated that RL was consistently and significantly lower then RD across all CO2 and N treatments and over all measured temperatures. Consistent with our hypotheses, both CO2 concentration and N availability had significant positive effects on RL and RD, indicating that an increase in resource availability resulted in an increase in respiratory flux. Furthermore, there was a significant interactive effect between C and N that differentially affected RL and RD (Fig. 2), accounting for the broad range of values for the percentage of RD that was inhibited in the light (28–53%). In this experiment the percentage inhibition of respiration in the light in the cN and cn treatments was the same (48%), but the percentage inhibition in the Cn (53%) and CN (28%) treatments showed a nearly twofold difference. Relative inhibition of RL for the CN and cN treatments are in accordance with those measured by Wang et al. (2001), who proposed that the percentage inhibition of RL in the CN treatment is lower than that for the cN treatment, because elevated CO2-grown plants have a greater demand for respiratory products such as energy and C skeletons. In addition, the number of mitochondria, the organelles that supply energy to the cell in the form of ATP, has been reported to be significantly higher in plants grown in elevated vs ambient CO2 (Griffin et al., 2001; Griffin et al., 2004; Wang et al., 2004). This suggests that plants may modulate respiratory capacity to meet a variable demand for respiratory products, which can be affected by growth CO2 concentration.
In plants grown in the Cn treatment, RL was highly inhibited relative to the other treatments because N limitation had a greater negative effect on RL than the positive effect of elevated CO2 on RL. This general relationship between N availability and respiration rate is partially supported by the correlations between RL and tissue N concentration (%) for all CO2 × N treatments. However, the lack of strong statistical support precludes the use of tissue N concentration as a potential predictor of respiration rates in the light.
Although RL and RD responded in the same direction to the CO2 concentration × measurement temperature interaction, the effect of CO2 treatment on respiration varied as measurement temperature increased. The response surface for the elevated and ambient CO2-grown plants (Fig. 3) indicated that, at measurement temperatures > 23°C, RL and RD in plants grown in elevated CO2 responded more positively to increasing temperatures than did RL and RD in plants grown in ambient CO2.
In order to explore the response of RL and RD to temperature more fully, two descriptive models were applied to the data and used to extrapolate the relationship beyond the temperature range measured. The models used were a Q10 and an Arrhenius model, chosen to allow broad comparisons with other data sets. The effect of temperature on RL is an important factor to consider because respiration can be more responsive than photosynthesis to variations in temperature (Atkin et al., 2000a). In this experiment, Q10 values for RL calculated from the gas-exchange data over the measured temperature interval of 23–33°C for each of the four CO2 × N treatments fall into two groups (Table 2). For plants grown in the ambient CO2 concentration, the calculated Q10 value for RL is c. 2, as is often assumed for RD (Tjoelker et al., 2001). However, for plants grown in the elevated CO2 treatment in both high- and low-N treatments, the calculated Q10 value for RL for the CN (3.33 ± 0.66) and Cn (4.91 ± 0.59) treatments were considerably higher. Q10 values > 3.0 have also been reported in species from both boreal and temperate forest biomes (Atkin & Tjoelker, 2003). However, an inaccuracy can arise in using a set Q10 over consecutive temperature intervals, as they tend to decrease with increasing temperatures, therefore overestimating C loss at high temperatures and underestimating C loss at low temperatures (Kirschbaum & Farquhar, 1984; Tjoelker et al., 2001; Bruhn et al., 2002).
In place of a traditional Q10, a modified Arrhenius equation that uses a temperature coefficient can be used (Lloyd & Taylor, 1994). In theory, the use of the temperature coefficient should allow more robust predictions of the temperature response of respiration over a broader temperature range. Using the modified Arrhenius equation to model the temperature sensitivity of RL we observe that RL and RD continue to increase appreciably as temperature increases in elevated CO2 conditions, particularly in the high-N treatment (Fig. 4). In addition, the differences between RD and RL tend to get larger as temperatures increase. The marked increase in respiration with increasing temperature, especially in the elevated CO2 treatments, has important implications for considering the simultaneous balance between photosynthetic C uptake and respiratory C loss. However, before these important short-term temperature effects can be extended to predictions of long-term temperature responses, it is crucial to consider whether or not acclimation to a new growth temperature will occur, and if so to what extent (Atkin & Tjoelker, 2003).
In analyzing the relative effect that short-term temperature change has simultaneously on both net photosynthetic C gain and respiratory C losses in the light, two important trends emerged. The first indicated that using RL instead of RD increased the proportion of C gained to C lost for all CO2 × N treatments and at all measured temperatures (Fig. 5). The second, perhaps more consequential, trend suggests that in general as temperatures rise the proportion of C respired via RL increased with respect to C gained at Asat. This suggests that an increase in RL with an increase in temperature has the potential to have a significant impact on a plant's net C balance (Graham, 1980; Sharp et al., 1984).
In conclusion, the magnitude of RL and how it changes in a variable environment is a vital factor in determining the proportion of photosynthetically fixed C that is respired, and hence in refining our estimates of ecosystems as net sinks or sources of C (Drake et al., 1999; Atkin et al., 2000b). This is especially important in light of the conclusion by Valentini et al. (2000) that ecosystem respiration was the deciding factor in categorizing 15 European forests as net C sources or sinks. In this experiment, when measuring leaf respiration rates using the Kok method, RL was consistently and significantly lower than RD across CO2 and N treatments and over the experimentally measured temperature range. This difference in RL and RD underscores the importance of using RL when attempting to describe accurately respiration rates that occur during light hours, and its use would increase the accuracy of estimating the total amount of C gained. In practice it may be possible to use previously measured RD values in C models with an appropriate correction for factors such as light-induced inhibition, relative resource availability and temperature sensitivity. Such corrections could be further established with additional experimentation. If the findings in X. strumarium are present in other species it will be important to include these significant adjustments for respiration during hours of illumination into models that budget C balance for terrestrial ecosystems.
We thank Ardis Thompson for her technical assistance in executing this experiment and Natasja van Gestel for the C : N analyses. This work was funded by a National Science Foundation Grant (IBN-0130885) to K.L.G., J.D.L. and D.T.T.