SEARCH

SEARCH BY CITATION

Keywords:

  • carbon balance;
  • drought;
  • elevated CO2;
  • leaf respiration;
  • photorespiration;
  • Rlight;
  • temperature

SUMMARY

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

We investigated whether the degree of light inhibition of leaf respiration (R) differs among large Eucalyptus saligna grown in whole-tree chambers and exposed to present and future atmospheric [CO2] and summer drought. Associated with month-to-month changes in temperature were concomitant changes in R in the light (Rlight) and darkness (Rdark), with both processes being more temperature dependent in well-watered trees than under drought. Overall rates of Rlight and Rdark were not significantly affected by [CO2]. By contrast, overall rates of Rdark (averaged across both [CO2]) were ca. 25% lower under drought than in well-watered trees. During summer, the degree of light inhibition of leaf R was greater in droughted (ca. 80% inhibition) than well-watered trees (ca. 50% inhibition). Notwithstanding these treatment differences, an overall positive relationship was observed between Rlight and Rdark when data from all months/treatments were combined (R2 = 0.8). Variations in Rlight were also positively correlated with rates of Rubisco activity and nitrogen concentration. Light inhibition resulted in a marked decrease in the proportion of light-saturated photosynthesis respired (i.e. reduced R/Asat). Collectively, these results highlight the need to account for light inhibition when assessing impacts of global change drivers on the carbon economy of tree canopies.


INTRODUCTION

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

There is currently intense effort focused on understanding the balance between photosynthesis and leaf respiration (R) in relation to different climatic and atmospheric conditions (Cavaleri, Oberbauer & Ryan 2008; Drake et al. 2008; Kirschbaum 2010; Mahecha et al. 2010). This balance is strongly affected by factors that influence photosynthetic metabolism, but respiratory CO2 efflux from plants is also a large fraction of the total gross CO2 uptake and proceeds in daylight as well as during night. Although initially it was thought that R was similar throughout the day and during the night (Graham 1980), it is now accepted that R is strongly suppressed by daylight (Brooks & Farquhar 1985; Krömer 1995; Atkin et al. 2000a; Shapiro et al. 2004; Tcherkez et al. 2005; Nunes-Nesi, Sweetlove & Fernie 2007; Tcherkez et al. 2008). As a result, most leaf-level studies show that non-photorespiratory mitochondrial CO2 release in light (Rlight, which is Rd in Farquhar, Von Caemmerer & Berry 1980) is lower than mitochondrial CO2 release in the dark (Rdark, which is Rn in Farquhar et al. 1980), with the degree of light inhibition varying from 0 to 100%, reflecting the fact that Rdark and Rlight can respond differently to environmental signals (Hurry et al. 2005; Tcherkez et al. 2010). Failure to account for environment-dependent variations in light inhibition is likely to lead to large errors in the predicted balance between leaf-level R and light-saturated photosynthesis (Asat) (Atkin, Scheurwater & Pons 2006; Ayub et al. 2011), and in ecosystem level rates of R and primary productivity (Reichstein et al. 2005; Wohlfahrt et al. 2005; Mahecha et al. 2010).

Although we know that leaf Rdark can vary in response to water availability (Flexas et al. 2005; Galmes et al. 2007; Atkin & Macherel 2009), growth temperature (Atkin & Tjoelker 2003; Ow et al. 2010; Crous et al. 2011) and long-term elevated [CO2] (González-Meler, Taneva & Trueman 2004), relatively few studies have quantified the degree of light inhibition of leaf R in plants grown under different environmental conditions. Nevertheless, the available data suggest that light inhibition can vary in response to changes in temperature, atmospheric [CO2] and water availability. For example, light inhibition of leaf R increases within increasing measurement temperature in several plant species (Harley, Thomas & Reynolds 1992; Villar, Held & Merino 1995; Atkin et al. 2000a, 2006; Zaragoza-Castells et al. 2007); by contrast, sustained changes in growth temperature have less effect on the degree of light inhibition, reflecting the fact that Rlight and Rdark both thermally acclimate, albeit not necessarily to the same extent (Atkin et al. 2006; Zaragoza-Castells et al. 2007). Studies on herbaceous species Xanthium strumarium reported that light inhibition is lower in plants grown under elevated atmospheric [CO2] (Wang et al. 2001; Shapiro et al. 2004). By contrast, growth under elevated [CO2] had little impact on light inhibition in 12-year-old Liquidambar styraciflua trees (Tissue et al. 2002). Finally, Ayub et al. (2011) reported little impact of elevated [CO2] and growth temperature on the degree of light inhibition of leaf R in a 30 d study using glasshouse-grown Eucalyptus saligna seedlings; however, drought was found to substantially increase the degree of light inhibition, reflecting greater drought-mediated reductions in rates of Rlight (∼60%) compared with Rdark (∼17%). Whether such responses are observed in large trees experiencing seasonal variations in growth temperature has not however been investigated.

In addition to the need for further work characterizing seasonal variations in light inhibition of leaf R, there is also a need to determine whether environment-dependent variations in light inhibition can be predicted using a range of associated leaf traits (each of which may be affected by the surrounding environment). For example, if drought and atmospheric [CO2] mediated changes in leaf chemistry (e.g. concentrations of leaf N and sugars) and/or structure (leaf mass per unit leaf area) are correlated with changes in light inhibition of leaf R, then predictive models might be able to account for variations in Rlight using these data on these associated leaf traits. Similarly, correlations between light inhibition and chloroplast functioning (e.g. oxygenation by Rubisco) might provide another vehicle via which variations in light inhibition can be modelled under field conditions. A link between photorespiration and light inhibition of leaf R has been proposed (Atkin, Evans & Siebke 1998; Hurry et al. 2005; Tcherkez et al. 2008), although the significance and direction of this link remains unclear. To date, no field-based study has established whether environment-dependent variations in light inhibition of leaf R (and/or rates of Rlightper se) in large trees can be predicted using associated leaf traits such as those listed above.

Our study assessed the effects of elevated atmospheric [CO2] and summer drought on leaf non-photorespiratory mitochondrial CO2 release in the light (Rlight) and the degree of light inhibition in field-grown E. saligna Sm. (an evergreen, fast-growing tree). The study is the first to quantify rates of Rlight in large trees experiencing seasonal variations in the abiotic environment and subjected to treatments that simulate two major global change factors (elevated [CO2] and summer drought, implemented in a full factorial design). The specific objectives of our study were to: (1) determine how Rlight responds to elevated growth [CO2] and summer drought conditions, and whether temporal variations in the degree of light inhibition (in summer) differ among these treatments. As part of this analysis, we explored the potential impact of limitations in internal conductance (gi) on estimates of Rlight and associated gas exchange parameters; (2) determine potential proxy variables of Rlight by exploring relationships between Rlight and other leaf traits (e.g. other metabolic variables and structural/chemical composition traits); and (3) explore the consequences of environmental/seasonal-dependent variations in light inhibition on the balance between leaf R and Asat.

MATERIALS AND METHODS

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

Site description, plant material and experimental design

The study took place at the Hawkesbury Forest Experiment (HFE) research site in Richmond, NSW, Australia (33°36′40″S, 150°44′26.5″E, elevation 30 m a.s.l.) in a humid temperate climate with a mean annual temperature of 17 °C and a mean annual precipitation of 800 mm. Meteorological data such as rainfall, temperature and vapour pressure deficit (VPD) were collected at the site since the beginning of the experiment (Barton et al. 2010) and are summarized here for the duration of our study (Table 1). Additional environmental data are presented in Crous et al. (2011). December, January and February experienced high VPDs and wide temperature ranges with low rainfall, whereas March was more moderate (Table 1).

Table 1.  Environmental conditions at the Hawkesbury Forest Experiment (HFE) for the period between October 2008 and March 2009
Campaign dateAir temperature range (°C)VPD range (kPa)Rainfall between campaigns (mm)Avg. PPFD range (mol m−2 d−1)
  1. VPD, vapour pressure deficit; PPFD, photosynthetic photon flux density.

14–16 October 200813.0–20.30.01–2.5916.9–19.3
1–3 December 200812.5–31.20.03–4.47106.6 (46 d)37.7–57.4
5–7 January 200914.0–38.20.03–7.2563.2 (32 d)60.6–77.7
5–7 February 200919.7–41.90.04–6.8318.2 (28 d)67.8–74.4
10–12 March 200915.1–26.50.01–1.55173.4 (32 d)24.9–39.1

E. saligna trees were planted in April 2007 and the trees were on average 5.1 ± 0.2 m tall in October 2008 and grew an additional 3–4 m in height over the subsequent summer of 2008–2009. Two water treatments (droughted and well watered) and two atmospheric CO2 concentration treatments (ambient and elevated [CO2]) were applied to the whole-tree chambers, with three replicates per atmospheric [CO2] and water treatment combination. Six out of 12 whole-tree chambers received elevated [CO2] (ambient + 240 µmol mol−1), since the trees were planted. For each [CO2] treatment, three whole-tree chambers received a well-watered treatment where the equivalent of 10 mm precipitation was added every third day, equivalent to an annual precipitation of about 1200 mm. While this represents a rainfall that is 400 mm greater than the average rainfall in Richmond, 1200 mm is representative of the annual rainfall experienced by E. saligna in its natural habitat (Boland et al. 1984).

The drought treatment was achieved by withholding water from the assigned trees starting in October 2008 (i.e. spring in Australia) and extending through February 2009 (i.e. summer in Australia). Significant physiological effects of drought on plants were present in photosynthesis, leaf Rdark and pre-dawn water potentials in January and February until re-establishment of watering in March 2009, with the expected physiological recovery of photosynthesis and water potential for the droughted trees (Crous et al. 2011). More details about the technical chamber operation of the whole-tree chambers have been described in Barton et al. (2010), while the drought treatment and environmental conditions were presented in Crous et al. (2011).

In situ gas exchange measurements

Gas exchange measurements for our study were conducted between December 2008 and March 2009 in four consecutive campaigns (Table 1). One fully expanded leaf per tree was selected during each campaign in the sun-facing (i.e. north) lower third of the canopy to measure net CO2 exchange. At the completion of gas exchange measurements, leaves were harvested for analysis of leaf structure and chemical composition.

Gas exchange measurements were conducted using portable infrared gas analyser systems and 6 cm2 chamber (Li-Cor 6400; Li-Cor Inc., NE, USA). Flow rate in the cuvettes was set at 300 µmol s−1 for measurements made when photosynthetic photon flux density (PPFD) was ≤100 µmol photons m−2 s−1. For those measurements made at high irradiances (1800 µmol m−2 s−1), the flow rate was set to 600 µmol s−1. Temperatures and CO2 levels inside the leaf cuvettes were set to the prevailing conditions in each whole-tree chamber before each measurement, based on constantly updated readings from the tree chamber gas analysers (PP-systems SBA-1, Amesbury, MA, USA) and thermometers (NCT thermistors). As a result, the temperatures at which measurements were made varied from 1 month to another (being lowest in December/March and highest in January/February).

Given the inhibition of respiration by light, respiration rates during the day (Rlight) were assessed via light response curves between 1000 and 1130 h and analysed for the Kok effect (Kok 1948; Sharp, Matthews & Boyer 1984; Villar, Held & Merino 1994; Shapiro et al. 2004). One measurement at saturating light (Asat) was taken at 1800 µmol m−2 s−1 (in February and March only for the leaves used for Kok effect measurements) before stepping down PPFD from 100 µmol m−2 s−1 to 0 at 10 µmol m−2 s−1 intervals (example shown in Supporting Information Fig. S1). Each light level was applied for at least 150 s. Another measurement of Rdark was taken 10 min later to ensure depletion of all post-illumination transients (Atkin et al. 1998). For measurements made in December and January, Asat was measured on a leaf that was adjacent to that used for the Kok effect measurements (i.e. intervals over 10−100 µmol m−2 s−1 PPFD range). The Kok effect refers to the break in the slope of the measured photosynthetic rate in a light response (A-I) curve when measured with high resolution, which occurs at irradiances near the light-compensation point. At irradiances below the break, Rdark was determined at 0 µmol m−2 s−1, whereas above the break the line is extrapolated from the linear part of the A-I curve (using data measured at irradiances usually over the range 40–100 µmol m−2 s−1) to the y-axis representing Rlight (Kok 1948; Shapiro et al. 2004). Our decision to use 40–100 µmol m−2 s−1 PPFD range to estimate Rlight was based on the fact that slopes of A-I plots were near linear over this range (Supporting Information Fig. S1). Moreover, past work has shown that: (1) light inhibition reaches a maximal value at or below 40 µmol m−2 s−1 and is often constant over the 40–100 µmol m−2 s−1 PPFD range (e.g. 25 °C) (Atkin et al. 1998, 2000a; Shapiro et al. 2004); and (2) light inhibition remains near constant at high/saturating irradiances when measured at moderate to high temperatures (Atkin et al. 2000a). Thus, estimates of Rlight obtained over the 40–100 µmol m−2 s−1 PPFD range are likely to be indicative of Rlight at saturating irradiance.

In our study, Rlight was corrected for minor changes in Ci (intercellular CO2 concentration) associated with changes in photon flux density (Kirschbaum & Farquhar 1987) where Rlight was iteratively adjusted, thereby minimizing the intercept of photosynthetic electron transport (J) and irradiance. J, oxygenation (Vo) and carboxylation (Vc) of Rubisco at any given irradiance (Farquhar & von Caemmerer 1982) were calculated as follows:

  • image(1)
  • image(2)
  • image(3)

where Ci is the intercellular CO2 concentration, Γ* is the CO2 compensation point in the absence of Rlight (von Caemmerer & Farquhar 1981), Rlight values are positive values and Anet is the rate of net CO2 exchange at any given irradiance. When using Eqn 1, we assumed a Ci-based Γ* at 25 °C (Γ*25) of 36.9 µL L−1 (von Caemmerer et al. 1994) and that Γ* at any given leaf temperature (Γ*LeafT) (Jordan & Ogren 1984; Brooks & Farquhar 1985) can be calculated according to:

  • image(4)

Rates of photosynthetic carboxylation (Vc) and photorespiration (Vo) were calculated at light saturation (1800 µmol m−2 s−1) and at an irradiance that was in the upper range of that used to estimate Rlight (100 µmol m−2 s−1). By calculating rates of Vc and Vo at 100 and 1800 µmol m−2 s−1, we sought to assess relationships between Rlight and associated rates of Rubisco activity at irradiances near where inhibition is likely to be maximal (100 µmol m−2 s−1) and under conditions where photosynthesis is light saturated (1800 µmol m−2 s−1).

To explore how variations in internal conductance (gi) impact on the concentration of CO2 at sites of carboxylation (Cc) and subsequent calculations of Rlight, we calculated gi using a relationship reported by Evans & Von Caemmerer (1996) linking rates of Asat with corresponding ‘estimated’gi values:

  • image(5)

We then calculated Cc at any given irradiance according to:

  • image(6)

Finally, Rlight under the ‘estimated’gi scenario was estimated via application of the Kirschbaum & Farquhar (1987) correction procedure after replacing Ci with Cc. Similarly, we also calculated rates of J, Vc and Vo after replacing Ci with Cc; where relevant, we assumed a Cc-based Γ* at 25 °C (Γ*25) of 38.6 µL L−1 (von Caemmerer et al. 1994).

Leaf analyses for structural and chemical properties

All leaves on which the Kok effect was measured were harvested each month to measure leaf area (using a LI-3100 Leaf Area Meter; Li-Cor Inc.), fresh mass and dry mass (oven dried at 70 °C) using the 6 cm2 segment contained within the Li-Cor 6400 cuvette to calculate the ratio of leaf dry mass per unit leaf area (LMA) of each replicate. Concentrations of nitrogen and phosphorus in the individual 6 cm2 leaf segments were then determined with a Technicon Auto-analyser II (Bran + Luebbe Pty. Ltd, Norderstedt, Germany) using Kjeldahl acid digests. The remaining portion of each leaf (i.e. that not contained within the Li-Cor 6400 cuvette) was used to analyse sugars, starch and total non-structural carbohydrates (TNC) as described previously (Loveys et al. 2003).

Statistical analyses

All statistical analyses were conducted in JMP v.5.0.1 (SAS Institute, Cary, NC, USA). The statistical approach used here analysed repeated measurements with a linear mixed-effects model framework (Jennrich & Schluchter 1986; Piepho, Buchse & Richter 2004). The analysis considered the ‘Month’ variable a random factor, assuming that the variance within subjects (tree chambers) is homogeneous. This approach is advantageous for dealing with missing values and a known covariance structure in the data.

Variables were transformed where necessary to meet the normality and equal variance assumptions (typically a log10 transformation). Normality was tested according to Shapiro & Wilk (1965). In cases where the Shapiro–Wilk test rejected the null hypothesis (i.e. that data were normally distributed), a transformation of the data was applied; thereafter, transformed data were re-tested to show that the data can be treated as approximately normal and hence consistent with the assumptions of both regression analyses and analysis of variance (anova). Using chamber as a true replicate, effects of drought occurring within each summer month were tested with a two-way anova for CO2 and H2O. Differences between means were considered significant at P < 0.05 using a Tukey test.

To test [CO2] treatment differences in slopes with regard to relationships between Rlight and other variables, a linear regression analysis with dummy variables was employed (Neter et al. 1996) in which a significant interaction between [CO2] treatment (i.e. the dummy variable) and the independent variable represented a difference in slopes between [CO2] treatments. To account for small differences in respiration due to monthly variation in temperature [i.e. temperatures were much lower in March (i.e. early autumn) compared with the summer months] leaf temperature was used as a covariate but the covariate was never significant and therefore not further mentioned.

RESULTS

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

Response of respiration to environmental conditions

Site conditions during the summer months of the study (December–February) were consistently hot and dry with diurnal air temperatures ranging between 12.5–18 °C (minimum) and 30–42 °C (maximum), high VPD and reduced summer rainfall (<200 mm) especially compared with the long-term rainfall average of 266 mm (Table 1) during these months. The combination of these environmental conditions created drought conditions in January and February 2009 (Crous et al. 2011; Duursma et al. 2011).

The responses of respiration to drought and/or elevated [CO2] treatments were different for area-based rates of respiration in the dark (Rdark) and in the light (Rlight) measured mid-morning each month. Rates of Rdark exhibited strong monthly differences (P = 0.048) with rates in March being much lower compared with the other months. Moreover, overall area-based rates of Rdark were significantly reduced under drought conditions (−27%, P = 0.027), with average rates tending to be higher in plants grown under elevated [CO2] (+42%, P = 0.08). Similarly, although area-based rates of Rlight did not exhibit statistically significant differences among the treatments and across the 4 month period (Table 2), rates of Rlight averaged across both [CO2] treatments were nevertheless lower (−52%) in drought-treated trees compared with their well-watered counterparts (Fig. 1).

Table 2.  Results from three-way mixed analysis of variance using month as a random factor while CO2 treatment (CO2) and H2O treatment (H2O) are fixed factors
Variabled.f.RlightRdarkRlight/RdarkVo100Vc100Rdark/Asat
Source of variationMSF-ratioMSF-ratioMSF-ratioMSF-ratioMSF-ratioMSF-ratio
  • a

    One less degrees of freedom (d.f.) for Rlight, Rlight/Rdark and Rdark/Asat due to outlier.

  • Mean squares (MS) and F-ratios are reported for the following variables: light respiration (Rlight), dark respiration during the day (Rdark), the inverse ratio of light inhibition (Rlight/Rdark), oxygenation rates of Rubisco at 100 µmol m−2 s−1 (Vo100), carboxylation rates of Rubisco at 100 µmol m−2 s−1 (Vc100) and ratio of dark respiration to light-saturated photosynthesis (Rdark/Asat). Significant F-ratios are indicated with + for P < 0.1, * for P ≤ 0.05, ** for P < 0.01, *** for P < 0.0001. Log transformations were used for Rlight, Rdark and Rdark /Asat variables whereas Rlight/Rdark needed a square root transformation to normalize distributions. Vo100 and Vc100 were not transformed because of their normal distributions. Sample size varied between 45 and 46 observations. There were no significant outcomes (P > 0.05) for Rlight /Asat and Narea (not shown).

Month (random factor)36.282.542.4118.95*0.0961.100.676.47*0.390.130.00947.07*
CO211.160.790.896.52+0.0090.212.3535.48**1.6310.40*0.00324.48
H2O15.574.310.7515.88*0.163.600.183.5510.143.490.00394.57
CO2 * month (random)31.475.170.142.460.059.10*0.0665.210.151.120.00073.32
H2O * month (random)31.294.550.0470.840.058.83*0.0513.972.9221.23*0.00083.89
CO2 * H2O10.080.280.0681.230.0091.630.0876.83+0.0930.660.00125.66+
CO2 * H2O * month (random)30.280.230.0550.280.0050.140.0130.600.140.220.00020.12
Errora301.232.02*6.023.30**1.071.460.6416.1***18.682.43*0.0511.46
Whole model R245 0.51 0.62 0.43 0.89 0.55 0.43
image

Figure 1. Means and standard error per month of non-photorespiratory mitochondrial respiration in illuminated leaves (Rlight; a and b), mitochondrial respiration in dark-adapted leaves (Rdark; middle and d), and their ratio Rlight/Rdark (e and f) in ambient (left panels) and elevated CO2 (right panels) for well-watered (filled symbols) and droughted (open symbols) plants of Eucalyptus saligna. Drought effects were significant across summer months (December–February) when drought occurred (see text).

Download figure to PowerPoint

One factor explaining the month-to-month variation in rates of Rdark and Rlight was temporal changes in prevailing leaf temperature (Fig. 2). Temperatures during the mid-morning measurement period were highest in January, followed by slightly lower leaf temperatures in February, which in turn were higher than those in December; lowest mid-morning leaf temperatures were found in March. Associated with the month-to-month change in temperatures were concomitant changes in both Rdark and Rlight, indicating that both processes were temperature dependent. Indeed, for most growth [CO2]/water treatment combinations, R2 values of linear regressions fitted to leaf Rlight-T and Rdark-T data were in the 0.77–0.99 range (Supporting Information Table S3), which highlights the strong relationship between month-to-month fluctuations in temperature and variations in leaf R. It appears that both Rdark and Rlight were more sensitive to month-to-month fluctuations in temperature in well-watered trees, compared with their drought-treated counterparts (Fig. 2), especially in elevated [CO2]-grown trees where significant differences were found in the slopes of regressions fitted to leaf Rlight-T versus Rdark-T data of well-watered (P = 0.027) and drought-treated trees (P = 0.024) (Supporting Information Table S3).

image

Figure 2. Monthly means and standard errors of rates of Rdark (a) and Rlight (b) plotted against leaf temperature for each treatment. Open symbols represent the droughted plants and closed symbols represent the well-watered plants. Circles are used for ambient CO2 and triangles for the elevated CO2 treatment.

Download figure to PowerPoint

The ratio of Rlight/Rdark of which the remainder from one (i.e. 1-Rlight/Rdark) represents the degree of light inhibition, did not exhibit significant differences between the two [CO2] treatments. However, Rlight/Rdark exhibited significant interactions for Month × [CO2] and Month × H2O (both P = 0.05, Table 2). The Rlight/Rdark ratio was higher in elevated [CO2] during January and February, but not for December and March. There was also consistently higher light inhibition (i.e. lower Rlight/Rdark ratio) in droughted plants (∼80%) compared with well-watered plants (∼50%) for all months except March, resulting in a significant Month × H2O difference between the summer months and March (Fig. 1, Table 2). Indeed, in summer months Rlight exhibited a 60% (P = 0.07) reduction in drought-treated plants compared with their well-watered counterparts (Fig. 1), but no significant [CO2] treatment differences. Thus, summer drought increased the degree of light inhibition of leaf R (P = 0.047).

Given the different O2/CO2 ratios in leaves of ambient and elevated [CO2] treatments, a [CO2] treatment effect was expected in the Vo and Vc (i.e. oxygenation and carboxylation catalysis by Rubisco). When calculated at an irradiance of 100 µmol m−2 s−1 PPFD (the low irradiance where light inhibition was likely to be near maximal), Vc100 was only slightly higher (6%) in elevated [CO2] compared with ambient [CO2] (P = 0.045), whereas Vo100 was significantly lower in elevated [CO2] (−34%, P = 0.009) and showed a significant month effect (P = 0.04) with March being lower compared with the other months (Table 2, Supporting Information Table S1). Under 1800 µmol m−2 s−1 PPFD, rates of Vc1800 were 7% higher under elevated [CO2] (20.88 ± 1.67 µmol m−2 s−1) than their ambient [CO2] counterparts (19.58 ± 1.02 µmol m−2 s−1) (when averaged across all months and water treatments); conversely, elevated [CO2] resulted in an 18% decrease in Vo1800 (4.71 ± 0.60 and 5.76 ± 0.46 µmol m−2 s−1 for ambient and elevated [CO2] grown plants, respectively) (Supporting Information Table S1).

When compared in terms of the absolute change in CO2 exchange rates, growth under elevated [CO2] resulted in an increase in overall CO2 uptake by Vc100 of 0.37 µmol CO2 m−2 s−1 (when averaged across all months and water treatments), and an overall decrease in photorespiratory CO2 release (i.e. 0.5Vo100) of 0.24 µmol CO2 m−2 s−1 (i.e. a net increase of 0.37 + 0.24 = 0.61 µmol CO2 m−2 s−1) (Supporting Information Table S1). Interestingly, overall rates of A100 only increased by an average of 0.21 µmol CO2 m−2 s−1. Similarly, the net result of [CO2]-mediated changes in Vc1800 and 0.5 Vo1800 was greater than the increase in Asat (Supporting Information Table S1). These discrepancies likely reflect higher rates (albeit not statistically different) of Rlight exhibited by elevated [CO2] grown trees (Fig. 1).

There was a weak interaction between [CO2] and H2O (P = 0.07) where the Vo100 in ambient [CO2] was reduced slightly more under drought conditions compared with well-watered plants than Vo100 in elevated [CO2] (Table 2, Supporting Information Table S1). Vc100 was slightly higher (8.1% ± 2.6%) in well-watered conditions compared with drought-treated plants in December, February and March, but January exhibited a 53% difference between a high Vc100 in well-watered conditions and a low Vc100 in drought conditions, resulting in a significant Month × H2O interaction (P = 0.016, Table 2).

Impact of limitations in internal conductance

When initially calculating RlightVc and Vo, we assumed that the concentration of CO2 at the site of carboxylation (Cc) was the same as that in the internal airspace (Ci); i.e. internal conductance (gi) was infinite. For leaves with low gi, substantial differences between Cc and Ci are likely with the relative difference between Cc and Ci declining with decreasing measuring irradiance. Supporting Information Fig. S2 shows Rlight for all plants measured across the December–March study period, irrespective of treatment and calculated assuming that gi = ∞ and gi = 0.012Asat. Averaged across all months and treatments, rates of Rlight were 0.56 ± 0.10 and 0.61 ± 0.10 µmol CO2 m−2 s−1 assuming infinite and ‘estimated’gi values, respectively. Application of the ‘estimated’gi (Evans & von Caemmerer 1996) therefore has little effect on calculated values of Rlight; similarly, it has little effect on differences in Rlight exhibited by well-watered and droughted trees (Supporting Information Table S2). When measured under light-saturating conditions, accounting for limitations in gi resulted in substantial increases in predicted rates of J1800 (Supporting Information Fig. S3); by contrast, limitations in gi had negligible effects on estimates of J100 (Supporting Information Fig. S3). Incorporation of these gi-limited J-values into Eqn 2 resulted in minimal changes in predicted rates of Vc at both irradiances (Supporting Information Fig. S4). In the case of Vo (Eqn 3), accounting for potential limitations in gi resulted in large increases Vo1800 but little change in estimates of Vo100. Collectively, these results suggest that limitations in gi are only of substantial importance when considering predicted rates of photorespiration under high irradiance, with predicted rates of Rlight and Vc being largely independent of likely restrictions in internal conductance.

Relationships of Rlight with other leaf traits

One of our objectives was to explore relationships between Rlight and various other leaf traits (e.g. Rdark, chloroplast activity, leaf chemistry and/or structure) to determine whether those leaf traits could act as a proxy variable for Rlight. Given the significant month by treatment interactions in Rlight/Rdark ratios (Table 2), we expected significant drought and/or [CO2]-induced differences in linear regressions fitted to Rlight versus Rdark plots. However, linear regression analysis revealed no significant differences (in slopes or intercepts) among treatment relationships (Fig. 3). When data from all months and treatments were combined, a strong overall correlation between Rdark and Rlight was observed, with variations in Rlight being positively associated with variations in Rdark (Fig. 3; Table 3), with the degree of light inhibition being ca. 30 or 50% (when not forced or forced through the origin, respectively). Fitting of a first order linear regression to the overall Rlight versus Rdark relationship (Fig. 3) resulted in a small negative y-axis intercept when not forced through the origin. While this may indicate that a non-linear relationship may be more suitable, fitting of higher order equations was not possible because of insufficient data points and variability in the measured rates of leaf R. Clearly, there is a need for additional seasonal data to establish whether scaling between Rlight and Rdark is constant throughout the year or varies as a function of season, growth [CO2] and/or water availability.

image

Figure 3. Rates of Rlight plotted against corresponding rates of Rdark. Open symbols represent the droughted plants and closed symbols represent the well-watered plants. Circles are used for ambient CO2 and triangles for the elevated CO2 treatment. Regression equations and associated statistics are summarized in Table 3 for scenarios where the regression was/was not forced through the origin. The dashed line shows the 1:1 relationship.

Download figure to PowerPoint

Table 3.  Results from linear regression analysis of relationships between area-based respiration and associated traits
RelationshipCO2 treatmentEquationR2P-valueSlope difference?n
  1. Rlight represents the non-photorespiratory respiration in illuminated leaves (assuming infinite internal conductance), and Rdark represents the mitochondrial respiration in dark-adapted leaves. Relationships with rates of oxygenation (Vo) and carboxylation (Vc) of Rubisco at two irradiances (100 and 1800 µmol m−2 s−1 PPFD) are shown. Only significant relationships are reported. Relationships that did not exhibit a significant difference in slope between CO2 treatments (‘Slope difference?’) are reported across CO2 treatments (‘Combined’). n represents the number of observations included in the regression.

  2. NS, not significant.

Rlight − RdarkCombinedRlight = −0.36 + 0.67 Rdark0.80<0.0001No45
(forced through origin: Rlight = 0.49 Rdark)0.72<0.0001
Rlight − Vc100CombinedRlight = −1.34 + 0.35 Vc1000.290.0001No44
Rlight − Vo100AmbientRlight = −0.78 + 0.88 Vo1000.410.001Yes, P = 0.0322
ElevatedRlight = −1.29 + 2.26 Vo1000.410.00122
Rlight − NareaCombinedRlight = −0.24 + 0.48 Narea0.150.007No45
Rdark − Vc100CombinedRdark = −0.69 + 0.38 Vc1000.180.0034No45
Rdark − Vo100AmbientRdark = −0.36 + 1.10 Vo1000.450.0004Yes, P = 0.05723
ElevatedRdark = −0.81 + 2.75 Vo1000.310.00722
Rlight − Vc1800CombinedRlight = −0.35 + 0.03 Vc18000.310.0091 21
Rlight − Vo1800CombinedRlight = −0.31 + 0.11 Vo18000.420.0014 21
Rdark − NareaCombinedRdark = 0.078 + 0.74 Narea0.210.001No46
Rlight/Rdark − Vc100CombinedRlight/Rdark = −0.30 + 0.11 Vc1000.280.0003No44
Rlight/Rdark − Vo100AmbientRlight/Rdark = −0.05 + 0.28 Vo1000.180.051Yes, P = 0.05022
ElevatedRlight/Rdark = −0.31 + 0.72 Vo1000.480.000322
Rlight/Rdark − Vc1800CombinedRlight/Rdark = 0.12 + 0.005 Vc18000.053NS 21
Rlight/Rdark − Vo1800CombinedRlight/Rdark = 0.13 + 0.018 Vo18000.075NS 21
Rlight/Rdark − Vo1800/Vc1800CombinedRlight/Rdark = 0.21 + 0.061 Vo1800/Vc18000.0013NS 21
Rlight/Rdark − Vo100/Vc100CombinedRlight/Rdark = 0.30 + 0.142 Vo100/Vc1000.002NS 44
Rlight − Vo1800/Vc1800CombinedRlight = 0.11 + 0.69 Vo1800/Vc18000.03NS 21
Rlight − Vo100/Vc100CombinedRlight = 0.67 − 0.47 Vo100/Vc1000.003NS 44

As part of the study, we also sought to assess relationships between Rlight and associated rates of Rubisco activity at irradiances where inhibition is likely to be maximal (100 µmol m−2 s−1 PPFD) and under conditions where photosynthesis is light saturated (1800 µmol m−2 s−1 PPFD). In the absence of same-leaf, high-light data from December and January, assessment of RlightVo1800-Vc1800 relationships was limited to datasets where all parameters were measured on the same leaves (i.e. February and March). When assuming infinite gi, positive relationships were found between Rlight and both Vo and Vc (irrespective of irradiance; Fig. 4 and Supporting Information Fig. S5; Table 3). Similarly, positive Rlight-Vo and Rlight-Vc relationships were observed when assuming gi = 0.012Asat (e.g. Supporting Information Fig. S6). Most drought-treated plants (open symbols in figures) exhibited lower Vo and Vc values and were therefore at the lower end of the scale on both the x- and y-axis. Irrespective of the irradiance, Vo was more strongly correlated (higher R2) to Rlight than Rlight was correlated with Vc (Table 3). No treatment-induced differences in Rlight-Vo1800-Vc1800 scaling relationships were evident; similarly, there were no clear treatment-induced differences in RlightVc100 relationships. By contrast, clear differences were found between ambient and elevated [CO2] grown trees in Rlight-Vo100 relationships, with the slopes differing significantly between ambient and elevated [CO2] grown trees (P = 0.02, Fig. 4b). Similarly, growth [CO2] also altered the Rdark-Vo100 relationship (P = 0.056, Table 3). Subsequently, higher respiration rates were observed for both Rdark and Rlight in elevated [CO2] at a given Vo100 compared with ambient [CO2].

image

Figure 4. Plots of rates of leaf respiration in the light (Rlight) [(a) and (b)] and the ratio of Rlight to that in darkness (Rlight/Rdark) [(c) and (d)] against corresponding rates of photorespiration (i.e. oxygenation rate of Rubisco) at 100 µmol m−2 s−1 PPFD (Vo100) [(a) and (c)], the carboxylation rate of Rubisco at 100 µmol m−2 s−1 PPFD (Vc100) [(b) and (d)] in ambient (circles) and elevated (triangles) CO2 treatments. Open symbols represent the droughted plants and closed symbols represent the well-watered plants. The regression equation and associated statistics are summarized in Table 3. Similar plots of Rlight and Rlight/Rdark ratios against rates of Vc and Vo under saturating irradiance (1800 µmol m−2 s−1 PPFD) are shown in Supporting Information Fig. S5.

Download figure to PowerPoint

Inspection of individual leaf data showed that the degree of light inhibition in individual plants across treatments ranged between 30% and close to 100% (i.e. Rlight/Rdark ratios varied between 0.7 and near 0) (Fig. 4). To explore whether variations in Rlight/Rdark were linked to concomitant variations in Rubisco activity, we plotted Rlight/Rdark against rates of Vo and Vc at both irradiances, and against Vo/Vc ratios, the latter being a measure of the balance between photorespiration and carboxylation. Rlight/Rdark exhibited a significant, positive relationship with both Vo100 and Vc100 (Fig. 4; Table 3). As was the case with Rdark and Rlight, [CO2] treatment had no significant effect on the Rlight/Rdark-Vc100 relationship; by contrast, there was a significant difference in the Rlight/Rdark-Vo100 relationship exhibited by trees grown under high and low [CO2] (P = 0.050, Fig. 4), reflecting the higher rates of Rlight at given Vo100 (and the lower rates of Vo100per se) in trees grown under elevated [CO2] compared with their ambient [CO2] grown counterparts (Table 2 and Supporting Information Table S1). For the smaller high-light dataset, no significant relationship was found between Rlight/Rdark and Vo1800. Finally, irrespective of irradiance, there was no significant relationship between Rlight/Rdark or Rlightper se and the Vo/Vc ratio (Table 3), suggesting that changes in the balance between photorespiration and carboxylation were not associated with predictable changes in light inhibition of respiration or rates of Rlightper se.

Similar patterns were observed when Vc100 and Vo100 were related to a nitrogen-based respiration, RN (data not shown). Area-based rates of leaf R were weakly related to Narea (R2 = ∼0.15). Both Rlight and Rdark exhibited a positive significant relationship with Narea (data not shown), but the degree of light inhibition was not significantly related to Narea. However, neither area-based nor mass-based respiration rates (both in illuminated and dark-adapted leaves) exhibited significant relationships with sugars, starch or TNC, likely because of a small variation in values on the x-axis. In addition, there was no relationship between area-based or mass-based rates of leaf R with phosphorus concentration per unit leaf area or leaf mass per area (LMA; Supporting Information Table S1).

Balance between leaf respiration and photosynthesis

Finally, given the tight coupling between photosynthesis and respiration, we investigated the impact of each treatment combination on the balance between leaf R and light-saturated photosynthesis (using the mid-morning measured data, see Supporting Information Table S1 in Supplementary Information). Averaged across the four summer months, there were no significant differences between [CO2] or H2O treatments for both Rdark/Asat and Rlight/Asat ratios in E. saligna, apart from a significant month effect on the Rdark/Asat ratio (P = 0.04) due to lower prevailing temperatures in March. Rdark/Asat ranged between 0.04 in March and 0.09–0.13 in the summer months (December–February), whereas Rlight/Asat ranged between 0.02 and 0.07 in summer months. The Rlight/Asat ratio was on average approximately one-third of the Rdark/Asat ratio (Fig. 5). This difference in Rdark/Asat and Rlight/Asat ratios emphasizes the potential error of using Rdark measurements rather than Rlight measurements when assessing leaf daytime carbon balance.

image

Figure 5. Ratios of respiration to light-saturated photosynthesis (R/Asat) to compare the differences between using Rdark (i.e. mitochondrial respiration in dark-adapted leaves) and Rlight (i.e. non-photorespiratory respiration in illuminated leaves values). Values shown are averages across treatments within each month (there were no significant differences in R/A between CO2 and drought treatments; Table 2). Asat measurements in December and January were conducted on the same day on adjacent leaves of the same branch as the Rlight measurements. Leaf temperatures varied across months and were 26.2 °C in December, 30.7 °C in January, 28.3 °C in February and 22.7 °C in March.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

Environmental impacts of elevated [CO2] and drought on Rlight

Our study evaluated how Rlight and Rdark responded to elevated [CO2] and drought conditions in large field-grown E. saligna trees that experienced natural diel and seasonal variations in temperature. To enable comparison of treatment effects, and to assess how rates of Rlight and Rdark varied through the growing season, we measured gas exchange at the prevailing mid-morning air temperature four times over the summer/early autumn period. Although this approach increased variability (e.g. due to variations in the abiotic environment over the 4 month period), it also provided a unique opportunity to assess how light inhibition of leaf R, and rates of leaf Rlightper se, respond to current and future climate scenarios under field conditions. Although Rlight exhibited similar patterns as Rdark, distinct differences were found in the response of Rlight and Rdark to environmental variations, even after taking into account potential limitations in internal conductance (gi).

Our results show that temporal variations in mid-morning temperature (ca. 22–32 °C) over the 4 month study period were strongly correlated with variations in rates of Rlight and Rdark, with the month-to-month temperature sensitivity differing between the two processes (Fig. 2). To understand why the temperature sensitivity differed between Rlight and Rdark, consideration needs to be given to the short-term temperature dependence of both processes (i.e. Q10– the proportional change in leaf R per 10 °C rise) and the extent to which Rlight and Rdark acclimated to sustained increases in growth temperature. Past studies have reported that the short-term Q10 of Rlight is often lower than Rdark (Atkin et al. 2000b, 2006; Zaragoza-Castells et al. 2007), with both Rlight and Rdark acclimating to sustained changes in growth temperature. Thus, it is possible that the lower month-to-month temperature sensitivity exhibited by Rlight (compared with Rdark; Fig. 2) might have been due to lower short-term Q10 values and/or greater seasonal acclimation of Rlight than Rdark. While data are available on Q10 values and acclimation of Rdark in E. saligna trees (Crous et al. 2011) and other species (Atkin & Tjoelker 2003), we lack similar comparative short- and long-term data for Rlight. Addressing this data gap needs to be a priority of future research.

In our current study, Rdark measured at prevailing temperatures in the mid-morning tended to be higher in trees grown under elevated [CO2] (Fig. 1; Table 2). This was consistent with previous studies, where Rdark was higher in elevated [CO2] compared with ambient [CO2] (Wang et al. 2001; Davey et al. 2004; Shapiro et al. 2004; Crous et al. 2011). Although there was a trend towards higher Rlight in trees grown under elevated [CO2] (Fig. 1), growth [CO2]per se did not have a statistically significant effect on Rlight (Table 2). By contrast, Shapiro et al. (2004) reported higher Rlight in an herbaceous species (X. strumarium) grown under elevated [CO2]. Although our finding of no significant [CO2]-mediated change in Rlight may have been masked by an analysis across months and different prevailing temperatures during each campaign, it is also possible that trees may respond differently to elevated [CO2] than herbaceous species. Indeed, neither glasshouse-grown young E. saligna saplings (Ayub et al. 2011) nor field-grown 12-year-old L. styraciflua (Tissue et al. 2002) exhibited higher Rlight under elevated [CO2].

Lower light inhibition of leaf R (and higher rates of Rlightper se) in plants grown under elevated [CO2] has been previously reported in several studies (Wang et al. 2001; Shapiro et al. 2004) and attributed to a greater demand for energy and C-skeletons, and increased availability of respiratory substrates in elevated [CO2] compared with ambient [CO2] (Dewar, Medlyn & Mcmurtrie 1999; Atkin et al. 2000a; Wang et al. 2001). Our study found that Rlight/Rdark ratios were indeed higher (i.e. lower light inhibition of leaf R) in trees grown under elevated [CO2] compared with ambient [CO2] trees, when compared in the hot months of January and February (Fig. 1). While these results indicate a potential effect of atmospheric [CO2] on light inhibition of leaf R, they also suggest that the effect of elevated [CO2] on light inhibition may vary seasonally, with the greatest differences between the ambient and elevated [CO2] occurring when air temperatures are high (i.e. summer). Past work using plants grown under ambient [CO2] has shown that exposure to heat results in an increase in the degree of light inhibition of leaf R (i.e. Rlight/Rdark ratios decrease with increasing temperatures), with little inhibition occurring in the cold (Atkin et al. 2000b; Atkin et al. 2006; Zaragoza-Castells et al. 2007). One possibility is, therefore, that while degree of light inhibition might increase with rising temperature in ambient [CO2] trees, increases in temperature might have less effect on light inhibition in trees grown under elevated [CO2]. If correct, this may explain why the greatest differences in light inhibition (between ambient and elevated [CO2]) were most evident at the high temperatures experienced in January and February.

In our study, high summer temperature coincided with drought, enabling us to assess the interactive effects of drought and elevated [CO2] on Rlight in large field-grown trees. Recently, Ayub et al. (2011) found that drought stress reduced Rlight, resulting in greater light inhibition of respiration in drought compared with well-watered plants in potted, sunlit glasshouse-grown E. saligna saplings. Our study showed similar results (Fig. 1) in summer months where drought-treated trees exhibited about 80% light inhibition (and a 60% reduction in rates of Rlightper se) compared with well-watered trees, where the degree of light inhibition was approximately 50% across [CO2] treatments. Similarly, in Ayub et al. (2011) light inhibited leaf R by 73 and 40% in drought-treated and well-watered saplings, respectively. Hence, the overall responses observed in young glasshouse-grown seedlings (Ayub et al. 2011) were observed in our large field-grown trees in the summer when temperatures were high.

Several mechanisms could contribute to variation in the degree of light inhibition of leaf R, including changes in enzyme activity and reductant levels in respiratory pathways during the day compared with the night. Several studies have shown illumination of leaves can result in deactivation of pyruvate dehydrogenase (Gemel & Randall 1992; Tovar-Mendez, Miernyk & Randall 2003), the inhibition of isocitrate dehydrogenase due to higher ATP levels during the day (Igamberdiev & Gardeström 2003; Kasimova et al. 2006) and inhibition of pyruvate kinase (Lin, Turpin & Plaxton 1989; Tcherkez et al. 2009). Moreover, studies have shown that the TCA cycle functions in a non-cyclical manner in daylight producing both glutamate and fumarate to provide organic acids and C-skeletons for nitrate assimilation while also responding to the feedback inhibition by NADH and ATP on the TCA cycle in illuminated leaves (Tcherkez et al. 2009). Changes in demand for other respiratory products in the light (compared with in darkness) may also contribute to variability in degree of light inhibition of leaf R (Flexas et al. 2005; Atkin & Macherel 2009).

Establishing proxy variables of Rlight

One of the aims of our study was to determine potential proxy variables of Rlight by exploring relationships between Rlight and other leaf traits (e.g. other metabolic variables and structural/chemical composition traits). We focused on assessing relationships between Rlight and associated rates of Rdark, chloroplast activity and leaf chemistry/structure. In our current study, variations in Rdark explained 80% of the overall variation in Rlight regardless of treatment and temperature differences, demonstrating a tight coupling between Rdark and Rlight in field-grown trees. In previous work using glasshouse-grown E. saligna grown saplings, variations in Rlight were positively correlated with variations in Rdark (with an R2 of 0.52) (Ayub et al. 2011). Factors contributing to differences in explained variation between Rlight-Rdark correlations in our current study and Ayub et al. (2011) may reflect the contrasting stage of plant development (trees versus seedlings), growth conditions (whole-tree chambers versus glasshouse), differences in the severity of drought (more severe in the glasshouse compared with moderate summer drought in the field) and/or differences in thermal environment. Thus, notwithstanding that changes in the abiotic environment can have unequal effects on fine-level variations Rlight and Rdark, the available data suggest that overall variations in Rdark act as a good proxy variable of Rlight.

We also assessed whether variations in Rlight were linked to concomitant variations in leaf nitrogen concentration. Past work has shown that a strong correlation between leaf Rdark and leaf N concentration exists in a broad range of species (Niinemets & Tenhunen 1997; Tjoelker et al. 2005; Reich et al. 2008; Ordonez et al. 2009), with rates of leaf Rdark varying with canopy position (Bolstad, Mitchell & Vose 1999; Turnbull et al. 2003) due to the amount of N invested in Rubisco (Evans 1989). Such Rdark–N relationships have been successfully adopted into models (Williams et al. 1997; Dewar et al. 1999). What has not been established, however, is the extent to which [N] can be used to predict rates of leaf Rlight. Our study found a significant (albeit, not particularly strong) relationship of both area-based Rlight (and Rdark) with Narea with a steeper slope for Rdark-Narea compared with Rlight-Narea, resulting in 50% higher Rdark for a given Narea of about 2.5 g m−2 compared with Rlight (using the relationships reported in Table 3). Thus, while not as effective as Rdark as a proxy of Rlight, variations in Narea may provide an additional variable for predicting variation in Rlight.

The final group of proxy variables we assessed were fluxes though the photosynthetic and photorespiratory pathways. While many studies have demonstrated a close coupling between rates of net photosynthesis (A) and Rdark (Krömer 1995; Hoefnagel, Atkin & Wiskich 1998), less attention has been given to the relationship between variations in the components of A (photosynthetic carboxylation and photorespiration) and Rlight. Although past studies have provided evidence linking variations in Rlight to photorespiration (Vo), the direction and significance of this relationship is currently unclear, with two hypotheses having been proposed. On one hand, increased Vo might reduce Rlight, reflecting photorespiratory-mediated inhibition of pyruvate dehydrogenase (Randall et al. 1990) and changes in cellular energy demand (Igamberdiev et al. 2001). Alternatively, a positive relationship would arise if increased Vo enhanced Rlight. This relationship has been attributed by Tcherkez et al. (2008) to the requirement of glutamate cycling to provide amino groups for glycine synthesis in the peroxisomes during the day. Moreover, a high demand for glycine may increase glutamate synthesis and hence Rlight to maintain nitrate assimilation underpinning the synthesis of glycine (Tcherkez et al. 2009). While our results appear to support the latter (i.e. variations in Vo were positively correlated with rates of Rlight and Rlight/Rdark ratios), positive relationships were also found between Rlight and Vc. As such, we suggest that variations in Rlight of field-grown E. saligna trees are linked, in part, to the prevailing rate of CO2 fixation and/or O2 uptake by Rubisco. Given this, and the linkages between Rlight, Rdark and leaf N concentration, we suggest that proxy variables of Rlight can be identified using leaf trait data commonly obtained in intra- and inter-species comparisons (i.e. N concentration and components of photosynthesis).

Impact of light inhibition of R on leaf-level carbon balance

The balance between leaf R and photosynthesis (A) is crucial in estimating how much C is gained by the plants to sustain growth and reproduction. Whereas responses to these factors have been examined extensively for A, the responses for R are less well understood. The ratio of R/A could decrease in elevated [CO2] if R is proportionally less responsive to elevated [CO2] than A (Ziska & Bunce 1998). Regarding drought conditions, most studies reported reduced respiration rates during drought (Atkin & Macherel 2009), potentially resulting in an increase in R/A ratios if photosynthesis decreased proportionally more in drought conditions (due to stomatal closure) compared with leaf respiration. Although an increase in R/A occurred in E. saligna during a 30 d sustained severe drought (Ayub et al. 2011), our study did not find significant differences between [CO2] or drought treatments for both Rdark/Asat and Rlight/Asat ratios in a moderate summer drought. The same was true when considering R expressed as a proportion of ‘gross’ photosynthesis (where Agross = Asat plus Rlight), albeit that R/Agross ratios were marginally lower than their R/Asat counterparts (Supporting Information Table S1). The lack of treatment effects on R/A may represent a longer-term response over time as trees could acclimate to a sustained R/A ratio throughout the year (Dewar et al. 1999; Enquist et al. 2007). However, as a consequence of consistently lower respiration rates in the light, the Rlight/A ratio was always lower compared with the Rdark/A ratio in E. saligna, both when considered on a R/Asat (Fig. 5, Ayub et al. 2011) and R/Agross basis (Supporting Information Table S1). If more widespread, this finding has important implications for ecosystem level models that currently assume a fixed R/A ratio based on measurements of Rdark alone. Moreover, because leaf R can account for up to 25% of daily photosynthesis (Poorter, Remkes & Lambers 1990; Ryan et al. 1994; Atkin, Scheurwater & Pons 2007), accounting for variations in light inhibition of leaf R will be important for studies seeking to predict ecosystem level gross primary productivity (Wohlfahrt et al. 2005; Davidson, Janssens & Luo 2006).

CONCLUSIONS

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

Increasing attention is being given to understanding how climate affects leaf R, but few studies have quantified the impact of global change drivers on both leaf Rlight and Rdark, or sought to establish criteria by which environment-dependent variations in light inhibition of leaf R can be predicted under field conditions. While several studies have investigated the effect of climate variables on Rlight under controlled environment conditions (Harley et al. 1992; Villar et al. 1995; Atkin et al. 2000a, 2006; Wang et al. 2001; Shapiro et al. 2004; Zaragoza-Castells et al. 2007; Ayub et al. 2011), little is known about the impact of global change drivers on light inhibition in large trees grown under seasonally variable field conditions (but see Tissue et al. 2002). Our results demonstrate that while growth [CO2] has little effect on Rlight or the degree of light inhibition of leaf R, exposure to drought in the summer months decreased both Rlight and Rdark. Moreover, exposure to summer drought resulted in a marked increase in the degree of light inhibition of leaf R in our large field-grown E. saligna trees. Importantly, we found that both Rlight and Rdark were temperature dependent, with Rdark more sensitive to month-to-month variations in temperature than Rlight. Moreover, our results highlight the utility of using metabolic variables, such as leaf Rdark and CO2 fixation and/or O2 uptake by Rubisco, as proxy variables to predict environment-dependent variations in Rlight and Rlight/Rdark ratios. In our initial exploration of how variation in light inhibition of respiration affected predictions of the carbon balance of individual leaves, we found that accounting for a moderate level of light inhibition of leaf R can result in substantial decreases in R/A ratios. Collectively, these results strongly suggest that light inhibition of leaf R per se, and drought-mediated variation in light inhibition of leaf R, will need to be included in models to accurately predict rates of forest carbon exchange at the canopy level.

ACKNOWLEDGMENTS

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

This work was funded by the Natural Environment Research Council (NERC) in the UK (NE/D01168X/1, O.K.A.) and the Australian Research Council (ARC FT0991448 and DP1093759, O.K.A.; DP0881221, D.S.E.; and DP0879531, D.T.T.). Core support for the overall experiment and facilities from the Australian Department of Climate Change and the University of Western Sydney made this work possible and the first author was supported by the Australian Department of Agriculture, Fisheries and Forestry during the preparation of this manuscript. The NSW government is also gratefully acknowledged (NSW DECC Grant T07/CAG/16, D.S.E. and D.T.T.). The expert technical assistance of Mr David Sherlock and Ms Stephanie McCaffery is gratefully acknowledged.

REFERENCES

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

Supporting Information

  1. Top of page
  2. SUMMARY
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Supporting Information

Figure S1. Representative plot of net CO2 assimilation (Anet, µmol CO2 m−2 s−1) versus irradiance (µmol photons m−2 s−1) to illustrate the Kok effect and related calculations. Data are from a well-watered, ambient 'CO2'-grown replicate plant measured in February 2009 of the experiment. Solid symbols show measured rates of Anet over the 0–100 µmol photons m−2 s−1 range, with rates of leaf respiration in darkness (Rdark = 1.17 µmol CO2 m−2 s−1) shown. The break from linearity at irradiances below 40 µmol photons m−2 s−1 (dotted line) is shown, with a linear regression fitted (r2 = 0.992 for this replicate) to values between 40 and 100 µmol photons m−2 s−1 to estimate apparent rates of leaf R in the light (Rlight ‘apparent’, ◊ = 0.21 µmol CO2 m−2 s−1) at the y-axis intercept. Actual rates of Rlight (□ = 0.22 µmol CO2 m−2 s−1) that take into account changes in internal CO2 concentration (ci) that occur as irradiance declined (Kirschbaum & Farquhar 1987) are also shown 'assuming infinite internal conductance (gi)'. For this replicate, measurements of Anet were also made under saturating irradiance (1800 µmol photons m−2 s−1), yielding rates of Asat of 21.96 µmol CO2 m−2 s−1, with the underlying rates of carboxylation (Vc) and oxygenation (Vo) being 26.32 and 8.28 µmol CO2 m−2 s−1, respectively (calculated using Eqns 2 and 3 in the main text). At an irradiance of 100 µmol photons m−2 s−1, the corresponding values were: A100 = 4.24 µmol CO2 m−2 s−1, Vc100 = 5.08 µmol CO2 m−2 s−1, and Vo100 = 1.22 µmol CO2 m−2 s−1. Finally, for this replicate, actual Rlight values when assuming gi = 0.012Asat were 0.26 µmol CO2 m−2 s−1 (Eqn 5 in main text). For parameters above, Γ* was assumed to be 36.9 and 38.6 µL L−1 when assuming Ci and Cc, respectively (Von Caemmerer et al. 1994).

Figure S2. Rates of leaf respiration in the light (Rlight, µmol CO2 m−2 s−1) calculated assuming that internal conductance (gi) is infinite plotted against the corresponding Rlight values assuming an ‘estimated’ gi (gi = 0.012Asat; Evans & Von Caemmerer 1996). For the latter, we calculated cc at any given irradiance according to: cc = ci − (Anet/gi), with Rlight then being estimated via application of the Kirschbaum & Farquhar (1987) correction procedure after replacing ci with cc. For these calculations, Γ* was assumed to be 36.9 and 38.6 µL L−1 when assuming Ci and Cc, respectively (Von Caemmerer et al. 1994). Data are shown for all growth 'CO2' and water availability treatments, using data collected over the 4 month experimental period (December–March). Values shown are individual replicates. The dashed line shows the 1:1 relationship. See Supporting Information Table S2 for treatment averages of the ‘estimated’ gi values and corresponding Rlight values.

Figure S3. Rates of photosynthetic electron transport (J, µmol m−2 s−1) calculated assuming that internal conductance (gi) is infinite plotted against the corresponding J-values assuming an ‘estimated’ gi (gi = 0.012Asat; Evans & Von Caemmerer 1996). For the latter, we calculated cc at any given irradiance according to: cc = ci − (Anet/gi), with J then being estimated after replacing ci with cc. For these calculations, Γ* was assumed to be 36.9 and 38.6 µL L−1 when assuming Ci and Cc, respectively (Von Caemmerer et al. 1994). Data are shown for all growth 'CO2' and water availability treatments, using data collected over the 4 month experimental period (December–March). Values are shown for individual replicates of J calculated at two irradiances (100 and 1800 µmol m−2 s−1 PFFD). The dashed line shows the 1:1 relationship.

Figure S4. Rates of carboxylation (Vc) and oxygenation (Vo) by Rubisco (µmol m−2 s−1) calculated assuming that internal conductance (gi) is infinite plotted against the corresponding J-values assuming an ‘estimated’ gi (gi = 0.012Asat; Evans & Von Caemmerer 1996). For the latter, we calculated cc at any given irradiance according to: cc = ci − (Anet/gi), with J then being estimated via application of the Kirschbaum & Farquhar (1987) correction procedure after replacing ci with cc. For these calculations, Γ* was assumed to be 36.9 and 38.6 µL L−1 when assuming Ci and Cc, respectively (Von Caemmerer et al. 1994). Data are shown for all growth 'CO2' and water availability treatments, using data collected over the 4 month experimental period (December–March). Values are shown for individual replicates calculated at two irradiances (100 and 1800 µmol m−2 s−1 PFFD). The lines show the 1:1 relationship.

Figure S5. Plots of rates of leaf respiration in the light (Rlight) '(a) and (b)' and the ratio of Rlight to that in darkness (Rlight/Rdark) '(c) and (d)' against corresponding rates of photorespiration (i.e. oxygenation rate by Rubisco) at 1800 µmol m−2 s−1 PPFD (Vo1800) '(a) and (c)', the carboxylation rate by Rubisco at 1800 µmol m−2 s−1 PPFD (Vc1800) '(b) and (d)' in ambient (circles) and elevated (triangles) CO2 treatments. Open symbols represent the droughted plants and closed symbols represent the well-watered plants. The regression equation and associated statistics are summarized in Table 3. Values shown were calculated assuming infinite internal conductance (gi). Similar plots of Rlight and Rlight/Rdark ratios against rates of Vc and Vo 100 µmol m−2 s−1 PPFD are shown in Fig. 4 (main text). Data shown are for the two months where Rlight and light-saturated photosynthesis were measured on the same leaf (i.e. February and March only). In December and January, data on light-saturated photosynthesis were measured using adjacent leaves to those used for Rlight measurements; consequently, only data from February and March are shown in this figure.

Figure S6. Comparison of linear relationships between Rlight and photorespiration (i.e. oxygenation rate of Rubisco) at 100 µmol m−2 s−1 (Vo100) for calculations made assuming that internal conductance (gi) is infinite (left-hand panel) and assuming an ‘estimated’ gi (gi = 0.012Asat; Evans & Von Caemmerer 1996). For the latter, we calculated Cc at any given irradiance according to: Cc = Ci − (Anet/gi), with Rlight then being estimated via application of the Kirschbaum & Farquhar (1987) correction procedure after replacing Ci with Cc. Similarly, Vo100 values were calculating using J-values that employed Ci or Cc values. For these calculations, Γ* was assumed to be 36.9 and 38.6 µL L−1 when assuming Ci and Cc, respectively (Von Caemmerer et al. 1994). Data are shown for the four growth 'CO2' and water availability treatments, using data collected over the 4 month experimental period (December–March). Values shown are individual replicates.

Table S1. Means ± standard error per month and per treatment (n = 3) of Eucalyptus saligna leaves for the following variables: area-based rates of leaf respiration the light (Rlight) and dark (Rdark), carboxylation rates of Rubisco at 100 and 1800 µmol m−2 s−1 PPFD (Vc100 and Vc1800), oxygenation rates of Rubisco at 100 and 1800 µmol m−2 s−1 PPFD (Vo100 and Vo1800), net photosynthesis at 100 and 1800 µmol m−2 s−1 PPFD (A100 and Asat), ratios of leaf R (in the light and in the dark) to ‘gross’ rates of light-saturated photosynthesis (where ‘gross’ A = Asat plus Rlight), total soluble sugars, mass-based leaf nitrogen concentration (N), mass-based leaf phosphorus concentrations (P) and leaf mass per surface area (LMA). Note: gas exchange rates are for leaves measured in the mid-late morning at the prevailing ambient air temperature of each month.

Table S2. Overview of the effect of assumed internal conductance (gi) on calculated rates of leaf R in the light (Rlight; µmol CO2 m−2 s−1) for trees grown under ambient/elevated 'CO2' and well-watered/droughted conditions. Rlight/Rlight ratios are shown in parentheses. Rlight values were calculated on the assumption of an infinite internal conductance (gi) (i.e. Ci = Cc) and ‘estimated’ gi (i.e. gi = 0.012Asat) (Evans & von Caemmerer, 1996). For the latter, we calculated Cc at any given irradiance according to: Cc = Ci − (Anet/gi), with Rlight then being estimated via application of the Kirschbaum & Farquhar (1987) correction procedure after replacing Ci with Cc. For these calculations, Γ* was assumed to be 36.9 and 38.6 µL L−1 when assuming Ci and Cc, respectively (Von Caemmerer et al. 1994). Data shown are averages over the 4 month experimental period (December–March).

Table S3. Results from linear regression analysis of relationships between area-based respiration and month-to-month variations in leaf temperature (Tleaf). Rlight represents the non-photorespiratory respiration in illuminated leaves (assuming infinite internal conductance) and Rdark represents the mitochondrial respiration in dark-adapted leaves. See Fig. 2 in main text for treatment/monthly average data.

FilenameFormatSizeDescription
PCE_2465_sm_FigS1-S6-TableS1-S3.pdf216KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.