A new measurement technique reveals rapid post-illumination changes in the carbon isotope composition of leaf-respired CO2


M. M. Barbour. Fax: +64 3 325 2418; e-mail: barbourm@landcareresearch.co.nz


We describ an open leaf gas exchange system coupled to a tunable diode laser (TDL) spectroscopy system enabling measurement of the leaf respiratory CO2 flux and its associated carbon isotope composition (δ13CRl) every 3 min. The precision of δ13CRl measurement is comparable to that of traditional mass spectrometry techniques. δ13CRl from castor bean (Ricinus communis L.) leaves tended to be positively related to the ratio of CO2 produced to O2 consumed [respiratory quotient (RQ)] after 24–48 h of prolonged darkness, in support of existing models. Further, the apparent fractionation between respiratory substrates and respired CO2 within 1–8 h after the start of the dark period was similar to previous observations. In subsequent experiments, R. communis plants were grown under variable water availability to provide a range in δ13C of recently fixed carbohydrate. In leaves exposed to high light levels prior to the start of the dark period, CO2 respired by leaves was up to 11‰ more enriched than phloem sap sugars within the first 10–15 min after plants had been moved from the light into the dark. The 13C enrichment in respired CO2 then decreased rapidly to within 3–7‰ of phloem sap after 30–60 min in the dark. This strong enrichment was not observed if light levels were low prior to the start of the dark period. Measurements of RQ confirmed that carbohydrates were the likely respiratory substrate for plants (RQ > 0.8) within the first 60 min after illumination. The strong 13C enrichment that followed a high light-to-dark transition coincided with high respiration rates, suggesting that so-called light-enhanced dark respiration (LEDR) is fed by 13C-enriched metabolites.


The stable carbon isotope composition of leaf-respired CO2 (δ13CRl) has enormous potential to allow partitioning of ecosystem respiration into various components (Tu & Dawson 2005), to provide information on key physiological processes (Ghashghaie et al. 2003) and to trace carbon fluxes through plants and ecosystems (e.g. Barbour et al. 2005). However, difficulties in measuring and understanding variation in δ13CRl have limited its application. Recent experimental work has demonstrated significant variability in δ13CRl (Tcherkez et al. 2003; Ocheltree & Marshall 2004; Xu et al. 2004; Hymus et al. 2005), and theoretical work (Ghashghaie et al. 2003; Tcherkez et al. 2005) has provided a framework with which to understand some of this variability.

CO2 respired by isolated leaf protoplasts displays no apparent fractionation (denoted as ed in this paper to distinguish respiration in the dark from mitochondrial respiration in the light) compared to substrates (Lin & Ehleringer 1997). However, intact leaves commonly respire 13C-enriched CO2 compared to likely respiratory substrates. For example, Phaseolus vulgaris leaves respired CO2 in the dark that was 6‰ more enriched than sucrose (Duranceau et al. 1999), while in Nicotiana sylvestris and Helianthus annuus, values of ed were 4 and 3‰, respectively (Ghashghaie et al. 2001). Further, δ13CRl varies with time spent in the dark (Tcherkez et al. 2003; Nogués et al. 2004; Tu & Dawson 2005), with light level during growth (Ocheltree & Marshall 2004) and over a diurnal cycle in field-grown Quercus leaves (Hymus et al. 2005).

When carbohydrates are the respiratory substrate, 13C-enriched respiratory CO2 is thought to result from the non-statistical (i.e. non-random) carbon isotope distribution within carbohydrate molecules (Ghashghaie et al. 2001). Carbon atoms in positions C-3 and C-4 of glucose are 2 and 6‰ more enriched than the average for the whole molecule (Roßmann, Butzenlechner & Schmidt 1991). These atoms are released as CO2 during glycolysis by pyruvate dehydrogenase (PDH) and are thought to predominate in respired CO2, so that the latter is enriched in 13C (Tcherkez et al.2003). Tcherkez et al. (2004) have termed this effect ‘fragmentation fractionation’. The overall result is an enrichment in CO2 released and a depletion of 13C in acetyl-CoA and subsequent fatty acids. Further, PDH itself may fractionate against 13C during acetyl-CoA formation (DeNiro & Epstein 1977; Melzer & Schmidt 1987).

This picture is complicated by the dynamics of respiratory metabolism of leaf cells. When carbohydrates are limiting (i.e. leaf starch or sucrose pools are fully consumed), substrates other than carbohydrates may be used for respiration. In particular, Tcherkez et al. (2003) showed that if the citric acid cycle coupled to β-oxidation of fatty acids dominates over PDH in producing CO2, then the CO2 respired will be depleted in 13C. This pattern explains the positive relationship between the ratio of CO2 produced to O2 consumed [i.e. the respiratory quotient (RQ)] and δ13CRl, with the most 13C-enriched CO2 when PDH activity dominated and RQ was close to 1, and more 13C-depleted CO2 when RQ dropped below 0.6 as time in the dark increased (Tcherkez et al. 2003). RQ indicates the type of substrate used in respiration because more reduced substrates, like fatty acids, require more oxygen per CO2 molecule produced than less reduced substrates like carbohydrates.

Although previous work describes and explains the slow kinetics (half an hour to several days in continuous darkness) of the carbon isotope composition of leaf-respired CO2 in darkness, the rapid kinetics just after illumination is unknown. However, it is recognized that leaves experience an important modification of the respiratory regime during the transition from high light to dark: after the photorespiratory burst, the respiration rate often displays a peak within 10–20 min, termed light-enhanced dark respiration (LEDR) (Atkin et al. 2000). Organic acids, which are thought to accumulate under high light conditions (Cornic 1973) and may provide the respiratory substrates for LEDR, often differ in δ13C compared to carbohydrates (Ghashghaie et al. 2001), so measurements of δ13C of the CO2 evolved in darkness during LEDR may help to determine its metabolic origin.

A new instrument for measurement of the stable isotope composition of atmospheric CO2 provides the potential of high-frequency measurements of leaf-respired δ13C. Tunable diode laser (TDL) spectroscopy is a relatively new technique for online CO2 measurements at high frequency, and has proven valuable for ecosystem scale measurements (Bowling et al. 2003; Griffis et al. 2004). We adopted this technique for quantification of leaf respiration by plumbing it to a portable gas exchange system designed for leaf-level measurements, and tested the coupled system in this study using castor bean (Ricinus communis L., Euphorbiaceae).

We have three objectives in this paper. Firstly, we demonstrated the application of TDL absorption spectroscopy coupled to a portable gas exchange system for leaf-level measurements of δ13CRl, and compared values and measurement precision with published studies using traditional mass spectrometric techniques. The first experiment involved exposing leaves to a long-term period of darkness (52 h) for comparison with previously published experiments. Secondly, we quantified short-term temporal variability in δ13CRl during the high respiration rates associated with LEDR at the start of the dark period. This experiment was conducted in order to test the hypothesis that measuring δ13CRl during LEDR enhances understanding of the substrate for respiration during the LEDR peak. Thirdly, we quantified short-term variability in photosynthetic carbon isotope fractionation (Δ13CA) and δ13CRl during a switch from low light to dark in the absence of LEDR in order to determine the degree to which δ13C of very recently fixed carbon is reflected in δ13CRl. The experiments were conducted on castor bean plants grown under two levels of water availability to provide a range in δ13C of respiratory substrates.


Growth conditions

Castor bean (R. communis L.) plants were grown in 7 L pots with potting mix and a slow-release fertilizer in a glasshouse at Los Alamos National Laboratory, NM, USA, during April and May 2005 (for the first and second experiments) and during July and August 2006 (for the third experiment). Average half-hourly photosynthetically active radiation (PAR) was measured with a quantum sensor (LI190SA; Li-Cor, Inc., Lincoln, NE, USA), and average half-hourly air temperature and relative humidity were measured (CS500L; Campbell Scientific, Inc., Logan UT, USA) and recorded with a datalogger (CR23; Campbell Scientific, Inc.). A shade cloth stretched across the roof of the glasshouse reduced the direct light within the glasshouse by about half, giving a maximum midday irradiance of 1050 µmol m−2 s−1 PAR.

For the first and second experiments, average daytime air temperature within the glasshouse varied between 24.1 and 31.2 °C, and relative humidity between 19 and 35%. Average daytime air temperature over the experimental period was 27.1 ± 0.8 °C, and relative humidity 25 ± 2%. Average night temperature varied between 18.2 and 22.1 °C and relative humidity between 14 and 40%. Average night temperature over the experimental period was 20.6 ± 0.4 °C, and relative humidity 27 ± 2%. The plants were well watered every second day, except five plants for which water was withheld for 7 d. Droughted plants were measured on the sixth and seventh day after watering ceased, and were observed to be slightly wilted on the morning of the seventh day.

For the third experiment, average daytime air temperature over the experimental period was 28.3 ± 2.4 °C, and relative humidity 42 ± 9%. Average night temperature over the experimental period was 21.5 ± 2.5 °C, and relative humidity 56 ± 13%. The plants were well watered every second day, except three plants for which water was withheld for 7 d.

Experimental description

Three different experiments were conducted. The first experiment (hereafter referred to as the ‘long-term respiration’ experiment) aimed at demonstrating the measurement technique and comparing the technique and measurements to previously experiments. Hence, changes in respiratory CO2 production (Rc), respiratory O2 consumption (Ro) and the carbon isotope composition of leaf-respired CO2 (δ13CRl) were measured (at the 3 min temporal resolution available with the TDL) over an extended period of continuous darkness after 24 h of illumination. The goal of the 24 h illumination period was to maximize carbohydrate storage within the leaves, and the extended dark period to deplete these carbon reserves and so induce changes in respiratory substrates. Plants grown under high light and well-watered conditions were moved into a controlled-environment growth cabinet (E15; Conviron, Winnipeg, Manitoba, Canada) set at 30 °C and 70% relative humidity, and maintained under constant light (400 µmol m−2 s−1 at the level of the uppermost leaves) for 24 h. The cabinet lights were then turned off, and the plants kept in the dark for 52 h to assess the variation in δ13CRl over the extended dark period. The plants remained in the dark growth cabinet while not being measured, but were moved under the cardboard box during Rc measurement. The results were compared to an experiment of similar design described by Tcherkez et al. (2003).

The second experiment (hereafter referred to as the ‘short-term respiration’ experiment) aimed to quantify short-term variability in δ13CRl during the first 40 min of a dark period, after leaves had been exposed to differing levels of cumulative irradiance to provide varying levels of LEDR. To do this, changes in Rc, Ro and δ13CRl were measured immediately after the plants were placed in the dark. The plants were moved into the dark at different times of the day, so that the leaves were exposed to light periods of different intensity and length over the 2 d of measurement. Five well-watered and five droughted plants were measured. The cumulative intercepted irradiance on the day of measurement (in mol m−2) was calculated for each leaf using quantum sensors at the height of the uppermost leaves. Just prior to Rc measurements, the plants were placed in the dark under a large cardboard box. The air temperature within the darkened cardboard box was the same as ambient glasshouse air temperature, and the plants spent between 25 and 35 min in the dark.

The third experiment (hereafter referred to as the ‘short-term photosynthesis–respiration’ experiment) aimed to quantify short-term variability in photosynthetic discrimination and in δ13CRl during a switch from light to dark. Light levels prior to the start of the dark period were relatively low to avoid possible complications in interpretation of δ13CRl because of LEDR. Changes in the net CO2 exchange (photosynthesis and respiration) and either photosynthetic 13C discrimination (Δ13CA, when expressed relative to δ13C of substrate CO2, or δ13CA when expressed relative to the standard, as described further) or δ13CRl were measured for 30 min in the light and then for 60 min after each plant had been moved into the dark. Three well-watered and three droughted plants were measured, with the soil water content of each pot evaluated gravimetrically. Watering regimes and the degree of water stress were qualitatively similar between Plants 3 and 5 in the short-term respiration experiment and those in the short-term photosynthesis–respiration experiment.

Gas exchange measurements

Measurements of leaf photosynthesis or respiration rate (i.e. CO2 production rate, Rc) were made on the youngest fully expanded leaf (leaves 3–6, depending on the growth environment) using a portable photosynthesis system (LI6400; Li-Cor, Inc.) fitted with one of two custom-built leaf chambers. The dark respiration chamber, used for both the long- and short-term respiration experiments, was milled from stainless steel to allow enclosure of up to 80 cm2 leaf area, and sealed with a closed cell foam gasket of 1 cm width. A large, clear topped chamber was used for the short-term photosynthesis–respiration experiment, allowing enclosure of up to 130 cm2 leaf area, and constructed as described by Sharkey, Berry & Raschke (1985,and see Field, Ball & Berry 1989). The large leaf area of both chambers maximized the difference in concentration and isotopic composition between incoming and outgoing chamber air, reducing errors in calculated δ13CRl and photosynthetic discrimination.

For both long-term and short-term respiration experiments, a thermocouple was placed within the chambers to measure leaf temperature, and a 120 L buffer volume used to stabilize the CO2 concentration of air entering the leaf chamber. The flow rate of air through the leaf chamber was 250 or 300 µmol s−1. The relationship between leaf area and boundary layer conductance was measured for the chamber using wet filter paper and applied to determine boundary layer conductance for each leaf. Leaf area within the chamber was measured from digitized images of the leaf and imaging software (Scion Image; Scion Corp., Frederick, MD, USA), and gas exchange variables were recalculated with the corrected leaf area.

After measurement of Rc, a disc was cut from the portion of the leaf within the chamber for measurement of oxygen consumption rate (Ro) using an oxygen electrode (Hansatech, King's Lynn, UK). The precision of the oxygen electrode is ± 0.001 µmol mol−1 O2. However, with an empty chamber, Ro values were quite variable for the first 5 min after the chamber was sealed (perhaps as the air within the chamber mixed). Thereafter, the net O2 flux was 0.00 ± 0.01 µmol m−2 s−1. Average Ro for each leaf was calculated between 5 and 6 min after the chamber was sealed. The temperature within the oxygen electrode leaf chamber was controlled using a water bath, and was altered during each measurement day to track ambient leaf temperature (as measured by the LI6400). Leaf temperature varied between 22 and 31 °C, and the electrode was re-zeroed after each change in chamber temperature. The RQ was calculated from the ratio of CO2 produced to O2 consumed (RQ = Rc/Ro).

The LI6400 reference and sample infrared gas analyzers were matched with an empty leaf chamber prior to the start of measurements. Measurements for the short-term respiration experiment commenced when Rc stabilized (i.e. the coefficient of variation for the CO2 partial pressure differential between the sample and the reference analysers was below 1%), and usually within 15 min of the plant being moved to the dark. Rc was recorded at 1 min intervals and tended to be very high initially, then reduced with time in the dark. Rc for some leaves, particularly those in the dark for more than 30 min before measurements commenced, tended to stabilize over the 20 min measurement period, and the average Rc over the last four measurements was used to calculate RQ. For most leaves, the initial rapid drop in Rc reduced to a linear decrease with time, so that Rc was estimated at the time of Ro measurement (for RQ calculation) from a linear regression of Rc against time over the linear part of the decline curve (the last four to six measurements). The linear model fit the data well, with between 72 and 99% of variation in Rc explained by time since the start of the dark period for an individual leaf. Using extrapolated Rc for RQ estimation, rather than average Rc over the last four measurements, results in values for RQ being 0.05 lower (on average). The values of RQ presented in the Results section refer to those estimated by the extrapolation technique.

Rc and δ13CRl measurements were made as described earlier for four plants in the long-term respiration experiment between 45 min and 4.6 h after the start of the dark period. The plants were returned to the growth cabinet after measurements, and re-measured between 23 and 26 h and between 48 and 52 h after the start of the dark period. During these three time periods, two leaf discs were removed from each of the remaining four plants within the growth cabinet for Ro measurement. As the leaves were used repeatedly for Rc and δ13CRl measurements, Ro measurements were necessarily made on different leaves. This meant that only an average RQ (i.e. the ratio of average Rc to average Ro, with eight individual Ro measurements used to calculate average Ro) could be calculated for each time period.

For the short-term photosynthesis–respiration experiment, compressed air from a cylinder was used, rather than buffered glasshouse air, to reduce temporal variability in δ13C of air entering the leaf chamber. The clear topped chamber allowed measurements to be made at ambient light levels, which were very low (102–439 µmol m−2 s−1 PAR) during the experimental period. The CO2 partial pressure of the cylinder air was 425 µmol mol−1, and δ13C was −8.3‰. The cylinder air had extremely low water vapor content, meaning that any water vapour in the leaf chamber was derived from leaf transpiration, and that air saturation deficits were generally low. The boundary layer conductance for the chamber was measured using wet filter paper, as described earlier. The flow rate into the leaf chamber was controlled by a mass flow controller at between 400 and 2070 mL min−1 (212–1104 µmol s−1), depending on the net CO2 flux, and gas exchange variables were recalculated from LI6400 measurements recorded every 20 s using corrected leaf areas, boundary layer conductance and flow rates. The time taken after the decrease in flow rate (coinciding with the switch from light to dark) for air within the large chamber to turnover and CO2 fluxes to stabilize (i.e. the coefficient of variation for the CO2 partial pressure differential between the sample and the reference analysers was below 1%) was between 9 and 22 min. Further details about the gas exchange measurement technique will be outlined in a subsequent paper.

Isotopic measurements

The ratio of 13C to 12C of CO2 entering and leaving the leaf chamber was determined using TDL absorption spectroscopy (TGA100A; Campbell Scientific, Inc.). This instrument measures the mixing ratio of the individual isotopologues (Bowling et al. 2003), but we recalculated the values as isotope ratios in the familiar delta notation, that is, relative to the Vienna Pee Dee belemnite (VPDB) standard. The carbon isotope composition of CO2 (δ13C) is then


where Rs and RVPDB are the 13C/12C ratios of the sample and the VPDB standard. δ13C is reported in parts per thousand (‰). Two primary calibration cylinders with total CO2 mixing ratios of 352.02 and 566.65 µmol mol−1, and δ13C of −8.44 and −17.06‰, respectively, (measured by National Oceanic and Atmospheric Administration-Climate Monitoring and Diagnostic Laboratory) were used to calibrate two working standards with CO2 concentrations of 336.65 and 553.74 µmol mol−1, and δ13C of −30.48 and −30.41‰, respectively. Mole mixing ratios of 12CO2 and 13CO2 for the high and low CO2 working standards were calculated as 334.62, 3.63 and 550.39, 5.97 µmol mol−1, respectively. Calibration of the TDL requires that the working standards span the range in expected mixing ratios of each isotopologue, not the isotope ratio per se (Bowling et al. 2003). Our calibration tanks span the expected range of isotope mixing ratios for the glasshouse air and leaf-respired CO2.

Mixing ratios of total CO2 ([CO2]t) were calculated from the mixing ratios of individual isotopologues by (Griffis et al. 2004)


where fother is the fraction of CO2 containing all isotopologues other than 12C16O16O and 13C16O16O, and assumed to be 0.00474 (Griffis et al. 2004).

The TDL measures at 10 Hz, with an SD of approximately 1.2‰. However, for the current application, a manifold was used to switch between each of the two working standards and the leaf chamber inlet (reference) and outlet (sample) lines to allow a mean mixing ratio to be measured over 15 s at the end of a 30 s period (for each working standard) or a 1 min period (for each of the reference and sample air streams). This means that each measured value for air entering and leaving the leaf chamber is an average of 150 individual measurements. All air streams passed through a low-flow Nafion counterflow water trap (PD625 dual configurations, Campbell Scientific, Inc.), prior to entry into the instruments optical cell at a flow rate of approximately 200 mL min−1 (controlled by a critical flow orifice). The optical cell pressure was typically 22.10 ± 0.04 mbar during measurement of the standards, and 22.11 ± 0.04 during measurement of the sample and reference lines. Repeated measurements of each working standard produced mean SDs of 0.09 µmol mol−1 for [CO2]t and 0.09‰ for δ13C. Further details about the coupled LI6400-TDL measurement technique, including lag times between the LI6400 and the TDL and error analysis will be provided in a subsequent paper.

Calculating δ13CRl, Δ13CA and δ13CA

The stable carbon isotope ratio of CO2 respired by the leaves (δ13CRl) was calculated by mass balance from the δ13C and concentration of CO2 entering (δ13Ce and Ce, respectively) and leaving (δ13Co and Co, respectively) the leaf chamber (Evans et al. 1986):


where p = (Co − Ce)/Co. A single value of δ13CRl was calculated from δ13Ce, δ13Co, Co and Ce measured in dry air by the TDL. Similarly, photosynthetic discrimination (Δ13CA) and fractionation (δ13CA) were calculated (Evans et al. 1986):




where ξ = Ce/(Ce − Co).

Phloem sap sampling

Sucrose transported within the phloem in the stem in darkness (assumed to be isotopically the same as sucrose in the measured leaf; Badeck et al. 2005) was sampled for δ13C analysis as described by Barbour et al. (2000) & Cernusak, Wong & Farquhar (2003). Samples were taken within 15 min after the completion of δ13CRl measurements for all three experiments. Samples were dried in pre-weighed tin capsules at 70 °C for 48 h, then re-weighed prior to analysis on an isotope ratio mass spectrometer (Europa Scientific 20/20; Europa Scientific, Crewe, UK) interfaced to a Dumas elemental analyser (Europa Scientific ANCA-SL, Europa Scientific) at the Waikato Stable Isotope Unit, University of Waikato, New Zealand. The SD for the repeated analysis of an internal standard, commercial sugar, was ± 0.14‰. Calibration versus VPDB was achieved using a certified secondary standard from CSIRO, Canberra, ACT, Australia.


Precision of δ13CRl measurement

The large leaf chambers allowed a large difference in concentration between CO2 entering and leaving the leaf chamber. However, when buffered glasshouse air was used, high transpiration rates limited minimum flow rates that could be used without creating vapor pressure and condensation problems. This problem was overcome with the use of essentially dry cylinder air in the short-term photosynthesis–respiration experiment. A range in p (as defined in Eqn 3) between 0.034 and 0.129 was observed, with higher p values found at higher Rc. A Monte Carlo analysis (following Barbour, Andrews & Farquhar 2001), including 10 000 individual calculations using Eqn 3 and the SD of [CO2]t and δ13C (0.09 µmol mol−1 and 0.09‰, respectively) produced an SD of individual measurements of δ13CRl between 0.53 and 1.99‰ (for p = 0.129 and p = 0.034, respectively).

Data from the first 5 h of the long-term respiration experiment were used to assess the degree of noise associated with the system, and between replicate plants. Figure 1 shows that the large buffer volume effectively smoothed temporal variability in δ13Ce, so that δ13Ce ranged between −8.69 and −8.21‰ over 5 h. A Monte Carlo analysis (as previously described, but limited to 1000 individual calculations of δ13CRl) produced an SD of between 0.5 and 1.2‰, with an average of 0.9‰ over the 5 h of measurements. Over 9–15 min, the temporal variability in δ13CRl, expressed as an SD, was between 0.2 and 1.7‰, with an average of 0.8‰. Repeated measures of δ13CRl on the same leaf over 5 h produced an SD of between 0.8 and 2.0‰, with an average SD of 1.7‰. Including variability between four replicate plants grown in the same environment, δ13CRl was found to vary between −25.6 and −22.1, with an average and SD of −24.3 and 1.0‰, respectively. The advantage of the high temporal resolution of TDL measurements is revealed when the SE of the average δ13CRl is calculated; δ13CRl for four replicate plants over 5 h is −24.3 ± 0.1‰.

Figure 1.

Isotope compositions and respiratory CO2 fluxes for four replicate plants during the first 5 h of the long-term respiration experiment. Panel (a) shows the isotope composition of CO2 in the inlet air stream (δ13Ce; buffered glasshouse air); panel (b) shows the difference in CO2 mol fraction between the dry chamber inlet and outlet air streams (Co − Ce, as measured by the TDL); panel (c) shows the Monte Carlo-derived SD of 1000 estimates of the carbon isotope composition of leaf-respired CO2[using Eqn 3 and the precision of δ13C and [CO2] measurement]; panel (d) shows measured values of the carbon isotope composition of leaf-respired CO2 (δ13CRl), and panel (e) shows the rate of respiratory CO2 production (Rc, as measured by the LI6400).

Temporal variability in δ13Ce and Ce was reduced by using cylinder air in the short-term photosynthesis–respiration experiment. However, this did not result in an appreciable reduction in the Monte Carlo SD compared to the long-term respiration experiment. SDs (over 1000 individual calculations for each set of measured values in Eqn 3) varied from 0.5 to 1.3‰, with an overall average of 0.9‰.

Long-term respiratory response

After 24 h in the light, both Rc and Ro were high (> 1.5 and > 2 µmol m−2 s−1, respectively) but declined with time when the plants were moved into the dark (Table 1). Rc averaged 0.68 µmol m−2 s−1 after both 24 and 48 h in the dark, but Ro increased from 1.1 to 1.7 µmol m−2 s−1 from 24 to 48 h in the dark. RQ decreased from 0.99 after 3 h to 0.62 after 24 h to 0.52 after 48 h (Table 1). This suggests that carbohydrates were the main respiratory substrate for at least 5 h after plants had been placed in the dark, but then other, more reduced, substrates became important. δ13Csuc increased from −30.5 to −27.6‰ over the first 5 h of the dark period (data not shown), but then remained constant at about −27.5‰ to 51 h in the dark (Table 1). δ13CRl was fairly constant, or slightly increasing with time, over the first 5 h of the dark period (Fig. 1). After the plants had been in the dark for more than 20 h, δ13CRl was more depleted and stable (−28.5 and −29.3‰ at 24 and 48 h, respectively, Table 1). The apparent fractionation between sucrose and respired CO2 declined from 4.5‰ at 5 h to −1.1‰ at 26 h to −1.6‰ at 48 h (Table 1). δ13CRl tended to increase with increasing RQ in the long-term (Fig. 2), confirming relationships presented by Tcherkez et al. (2004).

Table 1.  Gas exchange and stable carbon isotope ratios of leaf-respired CO2 and phloem sap sucrose from castor bean plants grown under well-watered conditions for the long-term respiration experiment, after 2–50 h in the dark following 24 h of illumination
Variable2–5 h24 h48 h
  1. Values are averages ± SE from measurements on four replicate plants. Ro was measured on four replicate plants under the same environmental conditions, to allow repeated gas exchange measurements on Plants 1–4.

  2. D, leaf-to-air saturation deficit; Tl, leaf temperature; gs, stomatal conductance; Rc, respiratory CO2 production; Ro, respiratory O2 consumption; RQ, respiratory quotient Rc/Ro; δ13CRl, δ13C of leaf-respired CO2; δ13Csuc, δ13C of phloem sap; ed, apparent fractionation between assumed respiratory substrates and leaf-respired CO2.

D (kPa)0.9 ± 0.11.34 ± 0.261.09 ± 0.12
Tl (°C)28.5 ± 0.527.6 ± 0.729.4 ± 0.1
gs (mol m−2 s−1)0.14 ± 0.040.07 ± 0.020.12 ± 0.02
Rc (µmol m−2 s−1)2.01 ± 0.090.69 ± 0.050.68 ± 0.03
Ro (µmol m−2 s−1)2.03 ± 0.251.09 ± 0.211.30 ± 0.22
RQ0.99 ± 0.090.62 ± 0.030.52 ± 0.03
δ13CRl (‰)−24.2 ± 0.4−28.5 ± 0.5−29.3 ± 0.1
δ13Csuc−28.7 ± 0.3−27.4 ± 0.2−27.7 ± 0.1
ed4.5 ± 0.7−1.1 ± 0.6−1.6 ± 0.6
Figure 2.

The relationship between the respiratory quotient (RQ) and the isotope composition of leaf-respired CO2 (δ13CRl) for the long-term respiration experiment. Values are averages of four replicate plants. Also shown is the linear relationship between the two values fitted to data presented by Tcherkez et al. (2003) for comparison.

Short-term respiratory response

In the short-term respiration experiment, the rate of CO2 production, Rc, tended to decline rapidly in the first 20 min after the leaves had been placed in the dark, then either continued to decline more slowly with time or stabilized (Fig. 3). Table 2 shows values of Rc used for calculation of RQ. As RQ was > 0.80 for all plants, carbohydrates were assumed to be the main respiratory substrate for all plants in the short-term respiration experiment. Stomatal conductance (gs) was typically high for well-watered plants, and tended to decline with time in the dark. Table 2 presents average gs during the last 6 min of Rc measurements for each leaf, and was significantly lower for droughted compared with well-watered plants (P < 0.05). These values for gs, although high for leaves in the dark, are consistent with gs measured at low leaf-to-air saturation deficit for castor bean (Barbour & Buckley, in press).

Figure 3.

Rates of CO2 efflux (a & b) and stable carbon isotope composition of respired CO2 (c & d) from castor bean leaves for the short-term respiration experiment. Plants were grown under two water availabilities: well watered (a & c) and droughted (b & d). Conditions inside the leaf chamber during measurements are listed in Table 2. Rc, respiratory CO2 production; δ13CRl, δ13C of leaf-respired CO2.

Table 2.  Gas exchange and stable carbon isotope ratios of leaf-respired CO2 (δ13CRl) and phloem sap sucrose (δ13Csuc) from castor bean plants grown under well-watered and droughted conditions for the short-term respiration experiment, after 25–35 min in the dark
Growth env.Plant no.Cum. PAR (mol m−2)D (Kpa)Tl (°C)gs (mol m−2 s−1)Ro (µmol m−2 s−1)Rc (at Ro) (µmol m−2 s−1)RQδ13CRl (‰)Δδ13CRlt (‰ min−1)δ13Csuc (‰)ed (‰)
  • Also shown are cumulative photosynthetically active radiation (PAR) prior to the being moved into the dark (Cum. PAR) and the slope of the relationship between δ13CRl and time (Δδ13CRlt), with asterisks indicating the significance of the relationship.

  • *

    P < 0.10;

  • **

    P < 0.05.

  • NS, non-significant relationship. D, leaf-to-air saturation deficit; Tl, leaf temperature; gs, stomatal conductance; Rc, respiratory CO2 production; Ro, respiratory O2 consumption; RQ, respiratory quotient Rc/Ro; δ13CRl, δ13C of leaf-respired CO2; δ13Csuc, δ13C of phloem sap; ed, apparent fractionation between assumed respiratory substrates and leaf-respired CO2.

Well watered15.5611.4023.00.0150.830.740.89−23.2NS−29.96.7
Well watered25.6970.6721.70.0660.331.160.87−23.7NS−30.06.3
Well watered311.1990.5028.40.2731.991.840.92−21.5−0.31**−26.95.4
Well watered413.5011.7630.30.0491.481.731.17−22.8−0.62**−26.73.9
Well watered513.9930.8029.00.1321.321.431.08−21.6−0.41**−27.05.4

The carbon isotope ratio of phloem sap sucrose (δ13Csuc) varied between −29.9 and −23.7‰ (Table 2). Two of the five droughted plants had quite enriched δ13Csuc (−23.7 and −24.1‰). These plants also had very low gs in the dark (Table 2). Half of the plants (three well-watered plants and two droughted plants) had rapidly declining values of δ13CRl between 10 and 20 min after the start of the dark period (Fig. 3). This decline was as great as 8‰ in the case of Plant 4 of the well-watered treatment, but substantial variation was observed in the rate of decrease of δ13CRl with time in the dark. Linear regressions of δ13CRl on time revealed slopes between 0 and −0.62‰ min−1, with a tendency towards steeper slopes with increasing cumulative irradiance (Table 2). The missing point for Plant 2 relates to a problem with the manifold switching between air streams, which was corrected at the end of the 3 min cycle. Subsequent measurements were not affected by this problem.

δ13CRl measured 20–60 min after plants were placed in the dark, after values had stabilized, varied between −23.7 and −20.4‰. δ13CRl increased with increasing δ13Csuc, with 79% of variation in δ13CRl explained by variation in δ13Csuc when values from both short-term experiments were included (P < 0.001, Fig. 4a). δ13CRl was unrelated to RQ for either well-watered or droughted plants (Fig. 4b). The strong relationship between δ13Csuc and δ13CRl points to a rather constant apparent fractionation of δ13CRl over δ13Csuc across the treatments, of 5.1 ± 0.2‰.

Figure 4.

The relationships between δ13C of leaf-respired CO2 (at quasi steady state, between 20–60 min after the start of the dark period) and δ13C of phloem sucrose (δ13Csuc) or photosynthetic 13C fractionation immediately prior to the start of the dark period (δ13CA) (a) and the respiratory quotient (RQ) (b). Data from both the short-term experiments are included, with open symbols representing values for well-watered plants and closed symbols values from droughted plants. In both panels, square symbols represent data from the short-term respiration experiment, and in (a), triangular symbols represent δ13Csuc, and stars represent δ13CA from the short-term photosynthesis–respiration experiment. The 1:1 line is shown in (a), as well as the fitted regression between δ13Csuc and δ13CRl across all plants; δ13CRl = 0.58 δ13Csuc − 6.18, r2 = 0.79.

Interestingly, apparent fractionation after 20–60 min in the dark decreased with increasing leaf temperature (Fig. 5). When the data presented by Tcherkez et al. (2003) from French bean were recalculated and plotted with current data from castor bean, it was apparent that ed was higher for castor bean, although the slope of the relationship was similar.

Figure 5.

Change in apparent fractionation (i.e. difference between δ13C of sucrose and leaf-respired CO2) with changing leaf temperature. The French bean data are recalculated from fig. 1C of Tcherkez et al. (2003), over the first 2 h of a dark period following a light period of 8–10 h. The castor bean values are taken at quasi steady state (20–60 min after the start of a dark period). Data from both the short-term experiments are included, with open symbols representing values for well-watered plants and closed symbols values for droughted plants. Square symbols represent data from the short-term respiration experiment, and triangular symbols represent data from the short-term photosynthesis–respiration experiment. Two plants for which quasi steady state had not been reached (δ13CRl not stable) are marked *. The fitted regression between δ13Csuc and δ13CRl across all plants is shown; ed = 10.77 Tl − 0.21, r2 = 0.45, p = 0.004.

Short-term photosynthetic-respiratory response

Watering regime had a clear effect on the gas exchange of plants in the short-term photosynthesis–respiration experiment. Droughted plants, with an average soil water content of 19%, were found to have gs between 0.003 and 0.021 mol m−2 s−1 in the dark (about 11% of well-watered plants, Table 3). Values for gs were lower than for the short-term respiration experiment because of generally higher air saturation deficits (Tables 2 & 3). Photosynthetic rates tended to be slightly higher for well-watered plants compared with droughted plants (Fig. 6). Higher stomatal conductances and (generally) higher photosynthetic rates resulted in higher Δ13CA and more negative δ13Csuc, on average, for well-watered plants (Table 3). δ13CRl was more depleted for the well-watered plants compared with the droughted plants (P = 0.02). However, as in the previous experiment, δ13CRl was positively related to δ13Csuc, and ed was negatively related to Tl. Interestingly, δ13CRl was not significantly related to photosynthetic fractionation (δ13CA, expressed relative to the VPDB standard) immediately prior to the start of the dark period (Fig. 4a), probably reflecting the small amount of carbon fixed during this time because of low light and low gs.

Table 3.  Gas exchange and stable carbon isotope ratios of leaf-respired CO2(δ13CRl) and phloem sap sucrose (δ13Csuc) from castor bean plants grown under well-watered and droughted conditions for the short-term photosynthesis–respiration experiment, after 40–60 min in the dark
  1. Values are averages ± SEs of measurements from three replicate plants. Also listed are values for soil water content (SWC), average photosynthetically active radiation (PAR) inside the leaf chamber (PAR) and photosynthetic discrimination against 13C (Δ13CA) prior to the start of the dark period.

SWC (%)88 ± 519 ± 2
PAR (µmol m−2 s−1)186 ± 48299 ± 64
D (kPa)0.40 ± 0.212.27 ± 0.68
Tl (°C)24.4 ± 0.527.4 ± 0.2
gs (mol m−2 s−1)0.081 ± 0.0220.009 ± 0.006
Rc (µmol m−2 s−1)0.73 ± 0.030.80 ± 0.09
Δ13CA (‰)20.7 ± 2.216.5 ± 2.1
δ13Csuc (‰)−28.4 ± 0.5−26.4 ± 0.2
δ13CRl (‰)−23.3 ± 0.5−21.1 ± 0.2
ed (‰)5.2 ± 0.45.3 ± 0.2
Figure 6.

The rates of net CO2 exchange (panels a & d), photosynthetic CO2 discrimination (Δ13CA, panels b & e) and photosynthetic 13C fractionation or isotope composition of leaf-respired CO2 (δ13CA and δ13CRl, respectively; panels c & f) for three plants grown under well-watered (panels a–c) or droughted (panels d–f) conditions during short-term photosynthesis–respiration experiment. Photosynthetic data are presented as both Δ13CA and δ13CA to allow the influence of δ13C of recently fixed carbohydrates on δ13CRl to be assessed. Conditions inside the leaf chamber during measurements are listed in Table 2.


Precision of the combined LI6400-TDL system for Δ13CA and δ13CRl measurement

The SD of δ13C measurement with the TDL (0.09‰) is similar or lower than typically reported for traditional mass spectrometric analysis of CO2 in air (e.g. 0.1‰, Cernusak et al. 2004; 0.2‰, Ghashghaie et al. 2001; 0 25‰, Ocheltree & Marshall 2004). Calculation of δ13CRl using an open system necessarily includes the combined errors associated with measurement of both inlet and outlet δ13C. To compare the precision of the LI6400-TDL system with more traditional mass spectrometry systems, we ran a Monte Carlo analysis for three examples from recently published studies in which δ13CRl or Δ13CA were measured, and compared these values to the SD estimated from the same calculation using the system described in the current paper.

Tcherkez & Farquhar (2003, 2005) measured δ13CRl in a closed gas exchange system flushed with CO2-free air prior to sealing. In this system, the CO2 molar fraction was allowed to build up to 300 µL L−1, with all CO2 derived from leaf respiration (i.e. Eqn 3 does not apply). In this case, the precision of the system is simply the precision of mass spectrometer measurement, or 0.2‰.

Applying an off-line mass spectrometry system with a very large leaf chamber (400–800 cm2), Cernusak et al. (2004) report ξ varying between 1.5 and 3.0, Ce between 533 and 967 µmol mol−1, and δ13Ce of −33.1‰. Using these values, the reported SD of δ13C measurement (0.1‰), and estimating the SD of measured CO2 partial pressures to be 0.1 µmol mol−1, we made 10 000 Monte Carlo calculations of Δ13CA with Eqn 4. The SD of 10 000 Monte Carlo estimates of Δ13CA varied between 0.13 and 0.28‰.

Cousins, Badger & von Caemmerer (2006) applied an online mass spectrometry system to measure Δ13CA. Using their reported values for ξ (between 4.9 and 29.9, depending on light level) and Δ13CA (between 3.6 and 11.0‰), and assuming δ13Ce is −33.1‰, and the SDs of CO2 partial pressures and δ13C are the same as for the Cernusak et al. (2004) experiment, the SD of 10 000 Monte Carlo calculations using Eqn 4 varied between 0.44 and 2.80‰. Applying the same calculations to Δ13CA measurements made during the short-term photosynthesis–respiration experiment described here, we obtain SDs between 0.28 and 1.02‰.

Tcherkez et al. (2005) also employed an open gas exchange system for online measurement of Δ13CA. Using a large leaf chamber (typical leaf surface area of 100 cm2), values for ξ were typically 3.2, and with an SD of 0.05 µmol mol−1 and 0.2‰ for CO2 partial pressure and isotope composition, respectively, 10 000 Monte Carlo calculations of Eqn 4 produce an SD for Δ13CA of 0.6‰.

The SD of individual δ13CRl measurements calculated by Monte Carlo analysis using the coupled LI6400-TDL system described here varied between 0.5 and 2.0‰, depending on the difference in CO2 mixing ratios between inlet and outlet air streams. Therefore, we conclude that the LI6400-TDL system has a precision between the off-line system of Cernusak et al. (2004), the closed-chamber online system of Tcherkez et al. (2003) and the online system of Cousins et al. (2006). However, it is worth noting that the higher precision of the Tcherkez et al. (2003) system comes from the closed nature of the system, which necessarily involves exposing the leaf to CO2-free air, and that the higher precision of the Cernusak et al. (2004) system is due to the very large leaf area enclosed by the chamber. The disadvantage of using a very large leaf chamber is the slow turnover time for the chamber air, meaning that dynamic leaf responses at high temporal resolution cannot be measured.

We believe the combined LI6400-TDL system has a number of advantages over other systems including (1) high temporal resolution; (2) realistic O2 and CO2 concentrations within the leaf chamber [unlike the mass spectrometer system described by Cousins et al. (2006), the O2 partial pressure is the same as ambient air, and unlike the closed system described by Tcherkez et al. (2003), ambient CO2 was used]; (3) relative ease of control of environmental conditions inside the leaf chamber using the LI6400; (4) real-time visualization of isotopic fluxes that allows quality control methods such as assessment of isotopic stability and leak testing during the measurement period; and (5) no need to correct for potentially interfering masses such as N2O.

Variation in δ13Csuc and δ13CRl from the long-term dark experiment

δ13Csuc increased from −30.5 to −27.6‰ over the first 5 h of the dark period. This change may reflect a change from export of recently fixed carbon, with depleted δ13C resulting from high Ci/Ca in the high humidity and relatively low-light growth chamber, to export of stored carbohydrates, with a more enriched δ13C from the lower Ci/Ca under lower humidity in the glasshouse. The Ci/Ca for castor bean photosynthesizing in glasshouse conditions was likely to be about 0.77 (giving δ13Csuc of about −27‰), while Ci/Ca at 70% relative humidity and 400 µmol m−2 s−1 PAR would be closer to 0.90 (giving δ13Csuc of about −30‰; M.M. Barbour, unpublished data from controlled environment experiments). In support of this suggestion, δ13Csuc sampled after 5 h in the dark is the same as δ13Csuc sampled from well-watered plants grown in the low-humidity glasshouse. An alternative explanation is that over time, starch is progressively re-mobilized for export from the leaves, and that starch is more enriched in 13C than sucrose (Tcherkez et al. 2004).

The data from the long-term dark experiment confirm the decline in δ13CRl over a number of days in the dark, and the positive relationship between δ13CRl and RQ, observed for P. vulgaris plants (Tcherkez et al. 2003). This supports the suggestion (Ghashghaie et al. 2001; Tcherkez et al. 2003; Pataki 2005) that the enriched δ13CRl at quasi steady state and high RQ near the start of a dark period reflects release of enriched carbon from positions 3 and 4 of glucose via the action of PDH, while after 24 and 48 h in the dark, a switch to other respiratory substrates has occurred, because both RQ and δ13CRl are lower.

Variation in δ13CRl in the quasi steady state

In both the short-term respiration experiment and the short-term experiment during a switch from photosynthesis to respiration, CO2 fluxes and δ13CRl values stabilized after 20–35 min in the dark. In quasi steady state, δ13CRl was positively related to δ13C of phloem sap sucrose (r2 = 0.79), although δ13CRl was 5.1‰ more enriched than δ13Csuc. This result compares well with previous observations of apparent fractionation: ed determined between leaf carbohydrates and δ13CRl was 6‰ in P. vulgaris (Duranceau et al. 1999), 4 and 3‰ for N. sylvestris and H. annuus, respectively (Ghashghaie et al. 2001), and 3.9‰ on average for four tropical rain forest species (Xu et al. 2004). The similarity in ed between previous measurements and those in the current study suggest that published values of ed are not an artifact of leaf respiration in chambers with unrealistically low or high CO2 concentrations (Tu & Dawson 2005). δ13CRl was not significantly related to δ13C of carbon fixed immediately prior to the start of the dark period when plants were exposed to low light levels, but rather δ13CRl was related to the more long-term, integrative signal of phloem sap sucrose.

An interesting observation was that ed decreased as leaf temperature increased for both short-term experiments. This relationship was also observed for French bean by Tcherkez et al. (2003). Because RQ also decreased with increasing leaf temperature (from 10 to 32 °C) in the Tcherkez experiment, the relationship was interpreted as a switch from carbohydrates to other more reduced respiratory substrates. In the short-term respiration experiment, RQ was unrelated to leaf temperature (leaf temperature varied only 10 °C; between 21 and 31 °C), and RQ was relatively invariable, so that changes in respiratory substrates do not provide an explanation. However, Rc tended to be higher at higher leaf temperatures (as is typical), so that it seems possible that pools of intermediates were used up more quickly at high temperatures, leading to a more rapid decline in δ13CRl. This view is in accord with data interpretation presented by Ghashghaie et al. (2003). Further experiments with more careful control of leaf temperature and measurements of intermediate pool sizes and δ13C are required to test this hypothesis.

Variation in δ13CRl at the start of the dark period

CO2 respired after high levels of illumination (within 20 min) was strongly 13C enriched in the short-term respiration experiment, coinciding with higher respiration rates associated with LEDR (Atkin et al. 2000). LEDR and coincident enrichment in δ13CRl was not observed in the short-term photosynthesis–respiration experiment, and δ13CRl was not related to photosynthetic 13C fractionation prior to the start of the dark period. These effects are probably due to the low light levels prior to the start of the dark period.

While it has been shown that the height of the LEDR peak is correlated to the cumulative photosynthesis rate during the previous light period (Hoefnagel, Atkin & Wiskich 1998), the metabolic origin of such an effect is not well known (Hill & Bryce 1992; Atkin, Evans & Siebke 1998). It has been suggested that this effect is related to the inhibition of respiration by light causing an accumulation of organic acids like malate (Cornic 1973). A plausible explanation for the strong 13C enrichment in the first 20 min of the dark period is decarboxylation of the mitochondrial malate pool. Malate (and oxaloacetate) in the cytosol contains carbon from HCO3- initially fixed by phosphoenolpyruvate carboxylase (PEPC) during the light period (Fig. 7). Malate may be transported to the mitochondria and decarboxylated by the NAD+-dependent malic enzyme in the dark, generating NADH. PEPC is known to discriminate against H13CO3- by about 2.2‰ (Brugnoli & Farquhar 2000), while the hydration equilibrium of CO2 to HCO3- favours 13C (fractionation of 9‰). As a result, light-produced cytosolic malate is enriched in 13C compared to triose phosphates produced by the Calvin cycle. The contribution of malate decarboxylation to leaf respiratory CO2 efflux would thus strongly enrich leaf-respired CO2 and would be highest at the start of the dark period, then slowly decline as PEPC activity decreased and as mitochondrial respiration is up-regulated.

Figure 7.

The biochemical pathways involved in leaf respiration in the dark. The pathways that are inhibited in the light and likely somewhat inhibited at the start of the dark period are indicated (Tcherkez et al. 2005).

We nevertheless recognize that malic enzyme-catalysed decarboxylation is almost certainly associated with a kinetic isotope effect. Indeed, it has been shown that the chicken liver NADP-dependent malic enzyme fractionates by 34‰ (Edens, Urbauer & Cleland 1997) against 13C. Unless all the malate molecules are consumed after illumination, this may eliminate the 13C enrichment in C-4 of malate. Although very attractive, the malate decarboxylation hypothesis needs further evidence, namely, the rate of the metabolic fluxes associated with decarboxylation reactions just after illumination, and a compound-specific isotope analysis.


The TDL coupled to a gas exchange system provides a technique ideal for measurement of environmentally driven and temporally variable carbon isotope composition of leaf-respired CO2. The present results confirm earlier observations that leaf-respired CO2 is enriched in 13C compared with likely respiratory substrates, but also provide the first evidence of variable and highly 13C-enriched respired CO2 over the first 10 min of a dark period for leaves previously in high light. This enrichment correlates with higher respiration rates, suggesting that LEDR is associated with the decarboxylation of heavier metabolites, among which malate is a plausible candidate. To test this hypothesis, measurements of the carbon isotope composition of purified metabolites are required, and this will be addressed in subsequent work.


We thank J. Ghashghaie, L. Wingate, T.E. Dawson, K.P. Tu and K.L. Griffin for valuable discussion; H. Powers, C. Meyer and K. Brown for technical assistance; T. Sharkey for loaning us the large leaf gas exchange chamber, and two anonymous reviewers and the editor for suggestions that greatly improved the manuscript. This work was supported through funding from the Royal Society of New Zealand Marsden Fund (contract LCR201, M.M. Barbour), International Science and Technology Linkages Fund (contract ISAT04 7, M.M. Barbour and N.G. McDowell), the Institute of Geophysics and Planetary Physics (IGPP) at Los Alamos National Laboratory (project 095566-001-05, N.G. McDowell and D.T. Hanson), a Laboratory Directed Research and Development-Exploratory Research grant (N.G. McDowell) and the University of New Mexico Research Allocation Committee (grant 05-L-07, D.T. Hanson).