 © 2004 American Society of Plant Biologists
Abstract
We measured the oxygen isotope composition (δ^{18}O) of CO_{2} respired by Ricinus communis leaves in the dark. Experiments were conducted at low CO_{2} partial pressure and at normal atmospheric CO_{2} partial pressure. Across both experiments, the δ^{18}O of darkrespired CO_{2} (δ_{R}) ranged from 44‰ to 324‰ (Vienna Standard Mean Ocean Water scale). This seemingly implausible range of values reflects the large flux of CO_{2} that diffuses into leaves, equilibrates with leaf water via the catalytic activity of carbonic anhydrase, then diffuses out of the leaf, leaving the net CO_{2} efflux rate unaltered. The impact of this process on δ_{R} is modulated by the δ^{18}O difference between CO_{2} inside the leaf and in the air, and by variation in the CO_{2} partial pressure inside the leaf relative to that in the air. We developed theoretical equations to calculate δ^{18}O of CO_{2} in leaf chloroplasts (δ_{c}), the assumed location of carbonic anhydrase activity, during dark respiration. Their application led to sensible estimates of δ_{c}, suggesting that the theory adequately accounted for the labeling of CO_{2} by leaf water in excess of that expected from the net CO_{2} efflux. The δ_{c} values were strongly correlated with δ^{18}O of water at the evaporative sites within leaves. We estimated that approximately 80% of CO_{2} in chloroplasts had completely exchanged oxygen atoms with chloroplast water during dark respiration, whereas approximately 100% had exchanged during photosynthesis. Incorporation of the δ^{18}O of leaf dark respiration into ecosystem and global scale models of C^{18}OO dynamics could affect model outputs and their interpretation.
Variations in the oxygen isotope composition (δ^{18}O) of CO_{2} in the atmosphere have the potential to reveal vital information about the global carbon cycle (Francey and Tans, 1987; Farquhar et al., 1993; Ciais et al., 1997). Furthermore, measurements of oxygen isotope composition of CO_{2} in canopy air may allow differentiation of CO_{2} fluxes into photosynthetic and respiratory components (Yakir and Wang, 1996). It was also recently suggested that nighttime measurements of δ^{18}O in canopy air could be used to partition nocturnal ecosystem respiration between leaves and soil (Bowling et al., 2003a, 2003b). Leaf dark respiration is an important component of carbon cycling between vegetation and the atmosphere. An understanding of the factors controlling the δ^{18}O of CO_{2} respired by leaves in the dark could therefore be important for interpreting the δ^{18}O of atmospheric CO_{2} at local, regional, and global scales.
The net rate of CO_{2} efflux from a leaf in the dark can be thought of as the difference between two oneway diffusional fluxes, one from the atmosphere to the leaf and the other from the leaf to the atmosphere. For example, if the net respiratory CO_{2} efflux (ℜ_{n}) is defined as ℜ_{n} = g_{c}(c_{i} − c_{a}), where g_{c} is the leaf conductance to CO_{2}, and c_{i} and c_{a} are CO_{2} mole fractions in the intercellular air spaces and atmosphere, respectively, the oneway flux from leaf to atmosphere becomes g_{c}c_{i} and that from atmosphere to leaf becomes g_{c}c_{a}. The difference between ℜ_{n} and g_{c}c_{i} will depend on the magnitude of the CO_{2} concentration difference between c_{i} and c_{a}; this difference will in turn depend on the leaf conductance to CO_{2} and the CO_{2} production rate inside the leaf. If the CO_{2} concentration difference between c_{i} and c_{a} is very large, then the magnitude of the net CO_{2} efflux will approach that of the oneway CO_{2} efflux from leaf to atmosphere. However, if the CO_{2} concentration inside the leaf is only a little larger than that in the atmosphere, the net CO_{2} efflux from the leaf will be much smaller than the oneway CO_{2} efflux from the leaf.
It has previously been recognized that one of the primary controls over the δ^{18}O of CO_{2} diffusing out of leaves in the dark should be the δ^{18}O of leaf water (Flanagan et al., 1997, 1999). This is because gaseous CO_{2} exchanges oxygen atoms with water during interconversion between CO_{2} and bicarbonate. In plant tissues, this interconversion is catalyzed by the enzyme carbonic anhydrase. The rate constant for carbonic anhydrase is very fast, such that CO_{2} diffusing out of leaves is expected to reflect nearly complete oxygen isotope exchange with leaf water. There is an equilibrium fractionation that takes place during the exchange reaction, such that at 25°C, the δ^{18}O of CO_{2} will be enriched by approximately 41‰ compared to the δ^{18}O of water with which it has equilibrated.
In this article, we present measurements of the δ^{18}O of CO_{2} respired by Ricinus communis leaves in the dark. We theorized that it should be the oneway flux of CO_{2} out of a respiring leaf that is labeled with the leaf water δ^{18}O signal, rather than the net CO_{2} efflux. This led us to hypothesize that the effect of a respiring leaf on the δ^{18}O of CO_{2} in air passing over the leaf could be much greater than predicted by considering the net CO_{2} efflux alone.
THEORY
Interpretation of the Oxygen Isotope Composition of DarkRespired CO_{2}
Natural abundance oxygen isotope ratios are commonly expressed relative to the value of a standard:(1)where δ_{x} represents the proportional deviation of R_{X}, the ^{18}O/^{16}O of material X, from R_{Std}, the ^{18}O/^{16}O of a standard. Using δ notation, we present the following equation for the δ^{18}O of CO_{2} respired by leaves in the dark (δ_{R}):(2)where θ is the proportion of CO_{2} in the chloroplast that has completely exchanged oxygen atoms with chloroplast water, δ_{e} is the oxygen isotope composition of water at the evaporating sites within the leaf, ε_{w} is the equilibrium fractionation between water and CO_{2}, δ_{c0} is the oxygen isotope composition of CO_{2} in the chloroplast that has not exchanged oxygen atoms with chloroplast water, C_{a} is the ambient carbon dioxide partial pressure, C_{c} is the chloroplastic CO_{2} partial pressure, δ_{a} is the oxygen isotope composition of ambient CO_{2}, and is the weighted mean isotopic discrimination against C^{18}OO during diffusion from the chloroplast to the atmosphere. A summary of all symbols used in the text is given in Table I. A derivation of Equation 2 is presented (see “Derivation 1” in text). As described for photosynthesizing leaves by Gillon and Yakir (2000b), we make the assumption that CO_{2} inside the leaf comprises a mixture of CO_{2} completely equilibrated with leaf water (of proportion θ) and CO_{2} that has undergone no equilibration with leaf water (of proportion 1 − θ). We further assume that chloroplasts are appressed against intercellular air spaces in the mesophyll cells (Evans and von Caemmerer, 1996), such that CO_{2} evolved from mitochondria interacts with chloroplasts during diffusion out of the cells. Because carbonic anhydrase resides primarily in chloroplasts in C_{3} leaves (Everson, 1970; Jacobson et al., 1975; Tsuzuki et al., 1985), the chloroplastic CO_{2} concentration becomes the relevant parameter for modeling δ_{R}.
The diffusional discrimination, can be calculated as (Farquhar and Lloyd, 1993)(3)where C_{i} is the CO_{2} partial pressure in the intercellular air spaces, and C_{s} is that at the leaf surface. The term a_{w} describes the summed discrimination against C^{18}OO during liquidphase diffusion and dissolution (0.8‰); a is the discrimination during diffusion through the stomata (8.8‰); and a_{b} is the discrimination during diffusion through the leaf boundary layer (5.8‰). We note that Equation 3 is precisely the same as the equation given for by Farquhar and Lloyd (1993); we have simply multiplied both their numerator and denominator by −1. The equilibrium fractionation between water and CO_{2} can be calculated as (Brenninkmeijer et al., 1983)(4)where T is leaf temperature in K.
The oxygen isotope composition of CO_{2} in the chloroplast of a respiring leaf (δ_{c}) can be calculated from the following equation:(5)a derivation of equation 5 is presented (see “Derivation 1” in text). Equations 23 and 24 can be combined, and, after dividing through by R_{Std}, give(6)
For a series of measurements made at different values of δ_{e}, δ_{c} can be calculated from Equation 5 and plotted against δ_{e}. According to Equation 6, the slope of the relationship between δ_{c} and δ_{e} (m) is then equal to θ(1 + ε_{w}), such that θ can be calculated as θ = m/(1 + ε_{w}). The intercept of the relationship, I, is equal to θε_{w} + δ_{c0}(1 − θ), such that δ_{c0} can be calculated as δ_{c0} = (I − θε_{w})/(1 − θ). We note that such an analysis assumes that only δ_{e} varies across the series of measurements; thus, θ, ε_{w}, and δ_{c0} are assumed invariant.
The oxygen isotope enrichment at the evaporative sites in leaves (Δ_{e}) can be calculated as (Craig and Gordon, 1965; Dongmann et al., 1974; Farquhar and Lloyd, 1993)(7)where ε^{+} is the equilibrium fractionation that occurs during the phase change from liquid to vapor, ε_{k} is the kinetic fractionation that occurs during diffusion of vapor from the leaf intercellular air space to the atmosphere, Δ_{v} is the isotopic enrichment of vapor in the atmosphere, and e_{a}/e_{i} is the ratio of ambient to intercellular vapor pressures. The Δ_{e} and Δ_{v} are defined with respect to source water, such that Δ_{e} = R_{e}/R_{s} − 1 and Δ_{v} = R_{v}/R_{s} − 1, where R_{e} is ^{18}O/^{16}O of water at the evaporating sites, R_{s} is ^{18}O/^{16}O of source water, and R_{v} is ^{18}O/^{16}O of vapor in the atmosphere. The term δ_{e} can be calculated from Δ_{e} as(8)where δ_{s} is the oxygen isotope composition of source water relative to a standard. The parameter Δ_{v} in Equation 6 can be calculated from measurements of the oxygen isotope composition of vapor in the atmosphere (δ_{v}) and source water as Δ_{v} = (δ_{v} − δ_{s})/(1 + δ_{s}). The equilibrium fractionation between liquid and vapor, ε^{+}, can be calculated as (Bottinga and Craig, 1969)(9)where T is leaf temperature in K. The kinetic fractionation, ε_{k}, can be calculated as (Farquhar et al., 1989)(10)where r_{s} and r_{b} are the stomatal and boundary layer resistances to water vapor diffusion (m^{2} s mol^{−1}), and 32 and 21 are associated fractionation factors scaled to per mil. These fractionation factors have been revised up from values of 28 and 19, respectively, based on recent measurements showing the isotope effect for diffusion of H_{2}^{18}O in air to be 1.032 (Cappa et al., 2003), rather than 1.028 (Merlivat, 1978).
Measurement of the Oxygen Isotope Composition of DarkRespired CO_{2}
For our first dark respiration experiment, in which air entering the leaf chamber was free of CO_{2}, we calculated the oxygen isotope composition of respired CO_{2}, δ_{R}, simply as the oxygen isotope composition of CO_{2} exiting the chamber, δ_{a}. In our second dark respiration experiment, where air entering the leaf chamber had a CO_{2} concentration sufficient to bring that inside the chamber close to that normally found in the atmosphere, we calculated δ_{R} with a modified form of the equation presented previously by Evans et al. (1986):(11)where C_{a} is the CO_{2} partial pressure (μbar) of air within the chamber when dried, δ_{a} is δ^{18}O of CO_{2} within the chamber, C_{in} is the CO_{2} partial pressure (μbar) of dry air entering the chamber, and δ_{in} is the δ^{18}O of CO_{2} entering the chamber. A derivation of Equation 11 is provided (see “Derivation 2” in text). The terms C_{a} and δ_{a} are measured in gas exiting the leaf chamber, due to effective stirring of air within the chamber.
Calculation of Photosynthetic Discrimination against ^{13}C and ^{18}O
For measurements in the light, we calculated carbon and oxygen isotope discrimination during photosynthesis as described by Evans et al. (1986):(12)where R_{a} is ^{13}C/^{12}C or ^{18}O/^{16}O of CO_{2} within the leaf chamber, R_{A} is ^{13}C/^{12}C or ^{18}O/^{16}O of CO_{2} removed from the chamber by photosynthesis, δ_{a} is δ^{13}C or δ^{18}O of CO_{2} within the leaf chamber, δ_{in} is δ^{13}C or δ^{18}O of CO_{2} entering the chamber, and ξ is defined as C_{in}/(C_{in} − C_{a}), where C_{in} and C_{a} refer to CO_{2} partial pressures in dry air. We calculated the oxygen isotope composition of chloroplast CO_{2} during photosynthesis by rearranging the C^{18}OO discrimination equation presented by Farquhar and Lloyd (1993):(13)where Δ_{A} is discrimination against C^{18}OO during photosynthesis, as defined above, and Δ_{ca} is defined as (R_{c}/R_{a}) − 1, where R_{c} is ^{18}O/^{16}O of chloroplast CO_{2}. We then calculated δ_{c} as δ_{c} = Δ_{ca}(1 + δ_{a}) + δ_{a}.
For the photosynthesis measurements that comprised our third experiment, we compared the regression approach to calculating θ, as described above in the theory relating to dark respiration, to the method suggested by Gillon and Yakir (2000b), whereby θ can be calculated separately for each individual photosynthesis measurement:(14)where Δ_{ea} is the value of Δ_{ca} expected if chloroplastic CO_{2} were in full oxygen isotope equilibrium with δ_{e}. The Δ_{ea} was calculated as(15)
Equation 14 incorporates an assumption that is not applied in the regression approach to calculating θ that we described above for dark respiration. The assumption is that the δ^{18}O of CO_{2} in the chloroplast that has not equilibrated with leaf water can be calculated from the equation R_{c0} = R_{a}[1 − ā(1 − C_{c}/C_{a})] (Gillon and Yakir, 2000b), which can be replaced, to a close approximation, by Defining δ_{c0} in this way assumes no discrimination against C^{18}OO by Rubisco; it also ignores any possible effect of photorespiration or day respiration on δ_{c0}.
Calculation of the Conductance from C_{i} to C_{c}
The CO_{2} conductance from leaf intercellular air spaces to the sites of carboxylation in chloroplasts (g_{i}) was calculated from ^{13}C discrimination measurements during photosynthesis using the method of Evans et al. (1986):(16)where Δ_{obs} is the observed ^{13}C discrimination, b is the discrimination against ^{13}CO_{2} during carboxylation (taken as 29‰), is the sum of discriminations against ^{13}CO_{2} during dissolution of CO_{2} and liquid phase diffusion (1.8‰), A is the net photosynthetic rate (μmol CO_{2} m^{−2} s^{−1}), C_{a} is the ambient CO_{2} partial pressure (μbar), ℜ_{d} is day respiration (μmol CO_{2} m^{−2} s^{−1}), e is the associated discrimination against ^{13}CO_{2}, k is the carboxylation efficiency (mol m^{−2} s^{−1} bar^{−1}), Γ_{*} is the CO_{2} compensation point in the absence of ℜ_{d} (μbar), and f is the discrimination against ^{13}CO_{2} associated with photorespiration. The term Δ_{i} represents the discrimination that would occur if g_{i} were infinite, and if photorespiration and day respiration did not discriminate (Farquhar et al., 1982):(17)where is the discrimination against ^{13}CO_{2} during diffusion through the boundary layer (2.8‰), C_{s} is the CO_{2} partial pressure at the leaf surface, and a^{13} is the discrimination against ^{13}CO_{2} during diffusion through the stomata (4.4‰). The term ( was calculated from the slope, m^{13}, of a plot of Δ_{i} − Δ_{obs} against A/C_{a}. The term g_{i} was then calculated as ( The value of m^{13} is independent of values assigned to f and e in Equation 16 because varying these parameters affects the intercept of the regression rather than the slope. Therefore, there is no need to assign values to f and e for calculation of g_{i}.
Calculation of the Oxygen Isotope Composition of Average Lamina Leaf Water
We estimated the average lamina leaf water ^{18}O enrichment (Δ_{L}) of leaves during CO_{2} collections from a model relating Δ_{L} to Δ_{e} (Farquhar and Lloyd, 1993):(18)where Δ_{e} is as calculated in Equation 7, and ℘ is a lamina radial Péclet number (Farquhar and Gan, 2003). The term ℘ is defined as EL/(CD), where E is transpiration rate (mol m^{−2} s^{−1}), L is a scaled effective path length (m), C is the molar concentration of water (5.55 × 10^{4} mol m^{−3}), and D is the diffusivity of H_{2}^{18}O in water (2.66 × 10^{−9} m^{−2} s^{−1}). In a previous experiment, we found that the scaled effective path length for R. communis, grown and measured under the same conditions as in the present experiment, was 15.0 ± 3.5 mm (mean ± 1 sd; n = 5; Cernusak et al., 2003). This mean value was used to calculate Δ_{L}. The term δ_{L} was calculated as δ_{L} = Δ_{L}(1 + δ_{s}) + δ_{s}. Cernusak et al. (2003) also found that the ethanoldry ice traps on the bypass drying loop of the gas exchange system were not quite efficient enough to remove all of the water vapor from the air cycling back to the chamber. Due to fractionation during condensation of the vapor in the traps, vapor in the air returning to the chamber was slightly enriched compared to that retained in the traps. As a result, Δ_{v} for the air exiting the chamber was found to be 1.2 ± 0.5‰ (mean ± 1 se; n = 5). This mean value was used in calculations of Δ_{e}.
Derivation 1: Equation for Predicting the δ^{18}O of DarkRespired CO_{2}
We begin by writing an equation for the total CO_{2} flux from the leaf interior to the atmosphere in the dark in the steady state:(19)where ℜ_{n} is the net CO_{2} efflux (μmol m^{−2} s^{−1}); g_{tc} is the total conductance to CO_{2} from chloroplast to atmosphere (mol m^{−2} s^{−1}); C_{c} and C_{a} are the CO_{2} partial pressures in the chloroplast and atmosphere, respectively (μbar); and P is atmospheric pressure (bar). We make the assumption that, in C_{3} plants, carbonic anhydrase resides primarily in the chloroplast (Everson, 1970; Jacobson et al., 1975; Tsuzuki et al., 1985) and that it is therefore the chloroplastic CO_{2} concentration that should be considered when calculating the C^{18}OO efflux from the leaf. We further assume that the chloroplasts in C_{3} plants are appressed against the intercellular air spaces in the leaf and that CO_{2} evolved in mitochondria interacts with chloroplasts during diffusion out of the leaf. These assumptions may need to be reassessed for application of the model to C_{4} plants. Equation 19 can be written for C^{18}OO as(20)where R_{R} is the ^{18}O/^{16}O of darkrespired CO_{2}, ā is the weighted mean diffusional fractionation from chloroplast to atmosphere (calculated as described in Equation 3 above), R_{c} is ^{18}O/^{16}O of chloroplastic CO_{2}, and R_{a} is ^{18}O/^{16}O of ambient CO_{2}. Equations 19 and 20 can be combined to give(21)Dividing Equation 21 by the ^{18}O/^{16}O of a standard, R_{Std}, and applying the relationship R_{X}/R_{Std} = δ_{X} + 1 leads to(22)Solving Equation 22 for δ_{c} leads to Equation 5 above:
To write an expression for predicting δ_{R}, we apply an assumption proposed by Gillon and Yakir (2000b), under which the CO_{2} within the chloroplast can be divided into two pools: one pool, of proportion θ, has completely exchanged oxygen atoms with chloroplast water and therefore has an ^{18}O/^{16}O composition of R_{ce}; the other pool, of proportion 1 − θ, has not exchanged oxygen atoms with chloroplast water and retains its initial ^{18}O/^{16}O composition of R_{c0}. We note that the term R_{c0} could describe a mixture of mitochondrial CO_{2} and CO_{2} that has diffused into the leaf from the ambient air. Therefore, we do not define R_{c0} solely as a function of CO_{2} diffusing into the leaf from the atmosphere, as was done previously for photosynthesis (Gillon and Yakir, 2000b). The term R_{c} is then written as(23)The term R_{ce} can be calculated from the equilibrium fractionation between CO_{2} and water:(24)where R_{e} is ^{18}O/^{16}O of chloroplast water, which we assume to be equal to ^{18}O/^{16}O of water at the evaporative sites. Combining Equations 21, 23, and 24 leads to(25)Dividing through by R_{Std}, and substituting 1 + ε_{w} for α_{w}, gives(26)Solving Equation 26 for δ_{R} leads to Equation 2 above, which is
Derivation 2: Calculating δ_{R} From Online GasExchange Measurements
Under steadystate conditions, the increase in CO_{2} concentration in air flowing through a gasexchange cuvette containing a respiring leaf can be described as(27)where u is the flow rate through the cuvette (mol s^{−1}), Λ is the area of the leaf in the cuvette (m^{2}), C_{a} and C_{in} are CO_{2} partial pressures of dry air exiting and entering the cuvette (μbar), P is atmospheric pressure (bar), and ℜ_{n} is the respiration rate of the leaf (μmol CO_{2} m^{−2} s^{−1}). The corresponding mass balance for C^{18}OO can be written as(28)Combining Equations 27 and 28 gives(29)Dividing through by the isotope ratio of a standard, R_{Std}, and substituting from the relationship R_{X}/R_{Std} = δ_{X} + 1 gives(30)Canceling common terms leads to Equation 11 above, which is
We note that the equations derived in this and the previous section can also be applied in the light. Thus, for photosynthesis, the term δ_{R} in Equations 2, 5, and 11 above can simply be replaced with the term δ_{A}. The term δ_{A} relates to Δ_{A} by the relationship Δ_{A} = (δ_{a} − δ_{A})/(1 + δ_{A}).
RESULTS
Dark Respiration with CO_{2} Free Air Entering the Leaf Chamber
In the first dark respiration experiment, air entering the leaf chamber was free of CO_{2}, and air exiting the leaf chamber had a mean CO_{2} partial pressure of 47 μbar. The CO_{2} exiting the leaf chamber was collected and analyzed for its isotopic composition. A summary of gas exchange parameters measured just prior to each CO_{2} collection is presented in Table II. The dark respiration rates of the leaves ranged from 0.8 to 2.0 μmol CO_{2} m^{−2} s^{−1} on a projected leaf area basis, with a mean value of 1.5. The C_{a}/C_{i} values ranged from 0.46 to 0.93, with a mean value of 0.81.
Isotopic parameters derived by combining the results of the gas exchange measurements with results of analyses of the isotopic composition of CO_{2} exiting the leaf chamber, and of irrigation water fed to the plants, are given in Table II; these parameters are δ_{e}, δ_{L}, and δ_{c}. Results for δ_{a}, the δ^{18}O of CO_{2} exiting the leaf chamber, are also given in Table II. The observed δ_{R} values, which are equal to δ_{a} in the first experiment, ranged from 43.8‰ to 59.0‰, with a mean value of 51.6‰. All δ^{18}O values in this paper are reported relative to Vienna Standard Mean Ocean Water (VSMOW). The δ_{c} values were significantly, positively correlated with corresponding values of δ_{e} (Fig. 1 ); the Pearson correlation coefficient (r) between the two was 0.96 (P < 0.0001, n = 11). The δ_{c} values were also significantly correlated with values of δ_{L} (r = 0.90, P = 0.0001, n = 11), but the relationship was not as strong as that between δ_{c} and δ_{e}. The slope of the regression relating δ_{c} to δ_{e} was 0.82, yielding an estimate for θ of 0.79. Thus, we estimated, by applying Equation 6, that 79% of the CO_{2} in the chloroplasts had equilibrated with chloroplast water during dark respiration in the first experiment. The intercept of the regression relating δ_{c} to δ_{e} was 39.4‰; this intercept yields an estimate for δ_{c0} of 36.2‰. This is the mean δ^{18}O estimated for CO_{2} not equilibrated with chloroplast water.
By applying the mean value of g_{i} derived from carbon isotope discrimination measurements during photosynthesis (see results below), we generated estimates of C_{c} and These values are detailed in Table II. Estimates of C_{a}/C_{c} ranged from 0.45 to 0.88, with a mean value 0.78. When these values for C_{a}/C_{c} and were inserted into Equation 2, along with the values of θ and δ_{c0} described above, a mean modeled δ_{R} of 51.9‰ was predicted, in good agreement with the mean observed δ_{R} of 51.6‰. The range of modeled δ_{R} can be compared with the range of observed δ_{R} in Table II.
Dark Respiration at Atmospheric CO_{2} Concentration
In the second dark respiration experiment, the partial pressure of CO_{2} in the air entering the leaf chamber was adjusted such that the air exiting the chamber had a partial pressure of approximately 350 μbar. Under these conditions, leaf dark respiration rates were similar to those observed in the first experiment, ranging from 1.0 to 2.0 μmol CO_{2} m^{−2} s^{−1}, with a mean value of 1.4. Stomatal conductance was lower than in the first experiment, having a mean value less than half that observed in the first experiment (Table II). This presumably reflects a response to the increased CO_{2} partial pressure within the leaf chamber. Although stomatal conductance was lower, C_{a}/C_{i} values were higher than in the first experiment due to the increase in C_{a}; values ranged from 0.91 to 0.99, with a mean of 0.97. The δ^{18}O of CO_{2} in air entering the leaf chamber was 19.1 ± 0.1‰ (mean ± 1 se; n = 5). The mean δ^{18}O of CO_{2} exiting the chamber was 43.5‰.
The most striking difference between the first and second dark respiration experiments was the difference in observed δ_{R}. The mean observed δ_{R} in the second experiment was 277‰, which can be compared with 52‰ for the first experiment (Table II). Mean values for δ_{e}, δ_{L}, and δ_{c} were similar between the two experiments (Table II). Differences between δ_{e} and δ_{L} in the second experiment were slightly less than in the first experiment, reflecting the lower transpiration rates (Table II). As in the first experiment, variation in δ_{c} was significantly correlated with variation in δ_{e} (Fig. 2 ), showing an r value of 0.95 (P < 0.0001, n = 10). It was also correlated with δ_{L}, with a slightly lower correlation coefficient (r = 0.94, P < 0.0001, n = 10). The regression slope of the relationship between δ_{c} and δ_{e} was 0.82, resulting in an estimate for θ of 0.79, suggesting that 79% of the CO_{2} in chloroplasts had equilibrated with chloroplast water during dark respiration in the second experiment. This θ value is the same as the value of 0.79 estimated in the first experiment. The value of the intercept of the regression of δ_{c} on δ_{e} was 34.6‰, yielding an estimate for δ_{c0} of 14.3‰; this value is lower than the δ_{c0} of 36.2‰ estimated in the first experiment.
Values of C_{a}/C_{c} in the second experiment did not differ from values for C_{a}/C_{i} when calculated to two decimal places; the range was from 0.91 to 0.99, with a mean of 0.97. This mean of 0.97 is considerably higher than the mean C_{a}/C_{c} of 0.78 observed in the first experiment. Mean estimates for were similar between the two experiments (Table II). When the empirically determined coefficients for θ and δ_{c0} for the second experiment were inserted into Equation 2, along with the other relevant parameters, the mean value of modeled δ_{R} was 291‰, which compares reasonably well with the mean observed δ_{R} of 277‰. The relatively small difference between the two presumably reflects variation around the regression line in Figure 2, which was used to estimate θ and δ_{c0}.
A comparison of modeled δ_{R} values across both experiments with observed δ_{R} showed that modeled δ_{R} accounted for 80% of variation in observed δ_{R}. The regression line relating the two was δ_{R}(observed) = 0.72δ_{R}(modeled) + 39.5 (R^{2} = 0.80, P < 0.0001, n = 21).
Carbon and Oxygen Isotope Discrimination during Photosynthesis
In the third experiment, R. communis leaves were placed in the leaf chamber in the light, and gas exchange and isotopic analyses were conducted. Photosynthesis rates ranged from 8.5 to 30.9 μmol CO_{2} m^{−2} s^{−1}, with a mean value of 20.4. The CO_{2} partial pressure of air exiting the chamber ranged from 328 to 395 μbar, whereas the CO_{2} partial pressure of incoming air ranged from 533 to 967 μbar; this gave rise to ξ values ranging from 1.5 to 3.0. Stomatal conductance was approximately 4fold larger in the light than in the dark at similar CO_{2} partial pressure (Table II). The C_{i}/C_{a} ranged from 0.66 to 0.90, with a mean of 0.79. The δ^{18}O of CO_{2} entering the leaf chamber was 19.1 ± 0.1‰ (mean ± 1 se; n = 5); the δ^{13}C of CO_{2} entering the leaf chamber was −33.1 ± 0.2‰ (mean ± 1 se; n = 5). The δ^{18}O of CO_{2} exiting the leaf chamber ranged from 40.1‰ to 45.4‰; the δ^{13}C of CO_{2} exiting the chamber ranged from −25.3‰ to −19.3‰.
The mean observed oxygen isotope discrimination during photosynthesis (Δ_{A}) was 44.6‰; the range is given in Table II. The δ_{c} values for the photosynthesis experiment were somewhat higher than for the dark respiration experiment at similar CO_{2} concentration, presumably reflecting a higher proportion of chloroplast CO_{2} equilibrated with chloroplast water (i.e. higher θ). Differences between δ_{e} and δ_{L} were larger in the photosynthesis experiment than in the dark respiration experiments, reflecting the higher transpiration rates (Table II). Variation in δ_{c} was significantly correlated with variation in δ_{e} (r = 0.97, P < 0.0001, n = 8), as shown in Figure 3 . The δ_{c} was also correlated with δ_{L} (r = 0.91, P = 0.001, n = 8), but the correlation was not as strong as with δ_{e}. The slope of the relationship between δ_{c} and δ_{e} was 1.31; using Equation 6, this indicates a value for θ of 1.25. However, this slope estimate was strongly influenced by one outlying data point; this datum is identified by an arrow in Figure 3. If this outlying datum is excluded from the analysis, the slope of the relationship between δ_{c} and δ_{e} becomes 1.11, yielding an estimate for θ of 1.06. The individual θ values calculated according to the method of Gillon and Yakir (2000b) ranged from 0.93 to 1.24, with a mean value of 1.02. If the outlying data point identified with the arrow in Figure 3 is excluded, these individual θ estimates ranged from 0.93 to 1.06, with a mean of 0.99. Because the θ values were very close to 1.0, we did not estimate a δ_{c0} value for the photosynthesis experiment.
Observed carbon isotope discrimination values, Δ_{obs}, ranged from 19.4‰ to 25.2‰, whereas values predicted for infinite g_{i} and no discrimination by photorespiration or day respiration, Δ_{i}, ranged from 20.6‰ to 26.5‰. The slope of the relationship between Δ_{i} − Δ_{obs} and A/C_{a} was 47.8 ± 14.6 (slope ± 1 se), yielding a mean g_{i} estimate of 0.57 mol m^{−2} s^{−1} bar^{−1}.
DISCUSSION
The most important result of this study is that we have shown that it is the oneway CO_{2} efflux from a respiring leaf that is labeled with the leaf water δ^{18}O signal, rather than the net CO_{2} efflux. The oneway efflux can be calculated as g_{tc}C_{c}/P, where g_{tc} is the total conductance to CO_{2} from chloroplast to atmosphere (mol m^{−2} s^{−1}), and P is atmospheric pressure (bar). In our second dark respiration experiment, where C_{a} averaged 347 μbar, values for g_{tc}C_{c}/P ranged from 7.4 to 50.1 μmol CO_{2} m^{−2} s^{−1}, whereas the net respiratory efflux, ℜ_{n}, ranged from 1.0 to 2.0 μmol CO_{2} m^{−2} s^{−1}; the ratio of g_{tc}C_{c}/P to ℜ_{n} averaged 16.9. Thus, in cases where the CO_{2} diffusing out of a respiring leaf has a δ^{18}O different from CO_{2} in canopy air, the effect of δ_{R} on δ_{a} could be significantly underestimated if one assumes that only the net CO_{2} efflux is influenced by the isotopic composition of leaf water. The analogous requirement for considering oneway CO_{2} fluxes when calculating the effect of photosynthesizing leaves on δ^{18}O of atmospheric CO_{2} was discerned by Farquhar et al. (1993).
Previous attempts to model the effect of leaf dark respiration on the δ^{18}O of CO_{2} in canopy air have considered only the net respiratory CO_{2} efflux. We will refer to this method as the net flux model. In the net flux model, δ_{R} is calculated as δ_{R} = δ_{e} + ε_{w} − a, where a is usually taken as 8.8‰. The C^{18}OO isoflux is then calculated as the product of ℜ_{n} and δ_{R}. For the purposes of this discussion, we define an isoflux as the product of a net CO_{2} flux and its δ^{18}O. The net flux model has been used to interpret nighttime measurements of δ^{18}O in canopy CO_{2} (Flanagan et al., 1997, 1999; Mortazavi and Chanton, 2002; Bowling et al., 2003a, 2003b) and in global simulations of δ^{18}O dynamics in atmospheric CO_{2} (Cuntz et al., 2003a, 2003b). Earlier global studies did not differentiate leaf respiration from soil respiration, and thus did not define δ_{R} for leaves (Farquhar et al., 1993; Ciais et al., 1997). A slightly different version of the net flux model, with a modified term for diffusional fractionation, has also been applied at the leaf level (Yakir et al., 1994; Yakir, 1998). If we apply the net flux model to data from our second experiment, where C_{a} was near that found in the atmosphere, predicted values for δ_{R} range from 42‰ to 61‰. These values can be compared to observed δ_{R} values ranging from 233‰ to 324‰. Thus, in the second dark respiration experiment, the net flux model underestimated the observed δ_{R} by 180‰ to 266‰. Note that these observed δ_{R} values are effective values that result when one treats the modification of δ^{18}O of CO_{2} in air passing over the leaf as if it resulted from the net CO_{2} efflux alone. Thus, using these observed δ_{R} values, the C^{18}OO isoflux is still calculated as ℜ_{n}δ_{R}, and the large difference between ℜ_{n} and g_{tc}C_{c}/P becomes manifested in the δ_{R} term.
If we apply the net flux model to our first experiment, where air entering the leaf chamber was free of CO_{2}, it predicts δ_{R} values ranging from 36‰ to 60‰. Observed δ_{R} values in this experiment ranged from 44‰ to 59‰, in good agreement with predictions from the net flux model. The difference in the performance of the net flux model between the first and second dark respiration experiments can be effectively understood by examining an alternative formulation of Equation 2. If the definition of δ_{c} from Equation 6 is substituted into Equation 2, and the term () in the denominator of Equation 2 is assumed equal to unity, Equation 2 can be rewritten as
Equation 31 is informative in that terms I and II on the right side are analogous to the net flux model; the difference is that in Equation 31 δ_{c} is defined as in Equation 6, whereas in the net flux model δ_{c} is defined as δ_{e} + ε_{w}. Term III on the right side of Equation 31 reflects the proportion of CO_{2} that diffuses into the leaf and equilibrates with leaf water, then diffuses out of the leaf, thereby altering the isotopic composition of CO_{2} in the leaf chamber while leaving the net CO_{2} efflux rate unaltered. This process is analogous to the invasion effect that has been described for soil respiration (Tans, 1998; Miller et al., 1999; Stern et al., 2001). In the first dark respiration experiment, where air entering the leaf chamber was free of CO_{2}, this process was also occurring, but had a much smaller impact on δ_{R} than in the second experiment. This is because (δ_{c} − δ_{a}) was small in the first experiment, having a mean value of 1.8‰; in contrast, (δ_{c} − δ_{a}) in the second experiment had a mean value of 7.9‰. Additionally, [C_{a}/(C_{c} − C_{a})] was much smaller in the first experiment than in the second, having a mean value of 4.8 in the former versus 47.7 in the latter. As a result, the mean value for term III in Equation 31, which can be thought of as the invasion term, was 5.2‰ for the first dark respiration experiment, and 234‰ for the second dark respiration experiment.
Equation 31 can be used to highlight the conditions under which large departures in δ_{R} from values predicted by the net flux model can be expected at the ecosystem level under natural conditions. For example, if δ_{c} is very similar to δ_{a}, term III will be small. Additionally, if stomata are tightly closed, [C_{a}/(C_{c} − C_{a})] will be small, and term III will also be small. Thus, the largest departures in δ_{R} from the predictions of the net flux model should occur when there is a relatively large difference between δ_{c} and δ_{a}, and when stomata are relatively open, such that [C_{a}/(C_{c} − C_{a})] is large. The approximation in Equation 31 that () equals unity introduces a very small bias into calculations with this equation; however, this bias is less than 1% and is therefore negligible. Thus, Equation 31, in combination with Equation 6, can be used in place of Equation 2, if so desired.
Photosynthesis enriches the atmosphere in C^{18}OO due to exchange of CO_{2} with evaporatively enriched leaf water in the chloroplast, whereas soil respiration is generally thought of as depleting the atmosphere in C^{18}OO, because soil CO_{2} exchanges with water in soil that has generally not been enriched by evaporation (Flanagan and Ehleringer, 1998). In this study, we have observed that leaf dark respiration is capable of enriching air passing over a leaf in C^{18}OO to as great an extent as photosynthesis. The mean δ^{18}O value of CO_{2} exiting the leaf chamber in the respiration measurements at atmospheric CO_{2} partial pressure was 43.5‰; the mean value for photosynthesis measurements at similar C_{a} was 42.3‰. The δ^{18}O of incoming CO_{2} in both experiments was 19.1‰, and flow rates through the chamber were similar between the two experiments. Thus, dark respiration had as marked an effect as photosynthesis on the δ^{18}O of CO_{2} passing over the leaves, even though the net exchange of CO_{2} between the leaf and ambient air is roughly an order of magnitude less, and in the opposite direction, during dark respiration.
The effect of both photosynthesis and respiration on δ^{18}O of CO_{2} in canopy air is partly controlled by the isotopic composition of leaf water. In natural systems, nighttime leaf water δ^{18}O is typically intermediate between daytime leaf water δ^{18}O and the δ^{18}O of source water (Dongmann et al., 1974; Förstel, 1978; Zundel et al., 1978; Förstel and Hützen, 1983; Flanagan and Ehleringer, 1991; Flanagan et al., 1993, 1999; Cernusak et al., 2002; Mortazavi and Chanton, 2002). We therefore expect nighttime leaf respiration to impart a C^{18}OO signal on the atmosphere that is intermediate between the soil respiration signal and the photosynthesis signal.
Accurate prediction of the oxygen isotope composition of leaf water is important for interpreting vegetation effects on δ^{18}O of atmospheric CO_{2}. Equation 7 can be used to calculate δ_{e} under steady state conditions. However, leaf water δ^{18}O is unlikely to be at steady state at night (Flanagan and Ehleringer, 1991; Harwood et al., 1998; Cernusak et al., 2002). Cernusak et al. (2002) applied a nonsteady state equation for δ^{18}O in leaf water, derived by G.D. Farquhar and L.A. Cernusak (unpublished theory), and found good agreement between predicted and observed nighttime values. The combination of the nonsteady state leaf water equation and the model that we have provided here for δ_{R} should allow reasonable predictions to be made of the impact of leaf dark respiration on δ^{18}O of atmospheric CO_{2}.
Stomatal conductance will be an important parameter in the prediction of both δ_{e} and δ_{R} during the night. However, little attention has been paid historically to nighttime stomatal conductance. Snyder et al. (2003) recently observed nighttime stomatal conductance to water vapor ranging from 10 to 150 mmol m^{−2} s^{−1} for 17 plant species in the western United States. However, a mechanistic framework for interpreting such variation does not currently exist. Further investigation into the patterns and processes controlling nighttime stomatal conductance will lead to more accurate prediction of nighttime δ_{e} and δ_{R}. We note that the mean stomatal conductance that we observed in the dark for R. communis at normal atmospheric CO_{2} concentration was 130 mmol m^{−2} s^{−1} (Table II), near the high end of values observed by Snyder et al. (2003) at night in the field. Our measurements were made during the day, and it is likely that stomatal conductance was influenced by circadian rhythms, causing it to be higher than it would be in the dark at night.
The mean value of θ for the photosynthesis experiment calculated by the method described by Gillon and Yakir (2000b) was very close to 1.0. If the outlying data point, indicated by an arrow in Figure 3, was excluded from the analysis, the regression method resulted in a similar estimate of 1.06. Thus, both calculations suggested θ values close to unity for photosynthesizing R. communis leaves. A quick examination of Figure 3 shows that observed δ_{c} estimates lie very close to those expected for full equilibrium, with the exception of the one outlier, which is several per mil above the value expected for full equilibrium. We are unable to find a satisfactory explanation for why this particular datum should differ so markedly from the others. Results have been reported for a number of other C_{3} species in which the CO_{2} diffusing out of photosynthesizing leaves appeared to be very close to full equilibrium with δ_{e} (Farquhar et al., 1993; Gillon and Yakir, 2001). Interestingly, the θ values that we observed during dark respiration in R. communis were lower than those observed during photosynthesis, having values close to 0.80. Further research is necessary to determine the cause of this apparent discrepancy between θ in the light and in the dark.
Gillon and Yakir (2000b) suggested that during photosynthesis δ_{c0}, the δ^{18}O of CO_{2} in the chloroplast not equilibrated with chloroplast water, can be calculated, to a close approximation, as δ_{c0} = δ_{a} − ā(1 − C_{c}/C_{a}). This definition assumes no discrimination against C^{18}OO by Rubisco during photosynthesis, and neglects any influence of photorespiration or day respiration on δ_{c0}. The latter statement is tantamount to saying that CO_{2} evolved from the mitochondria in the light has the same oxygen isotope composition as CO_{2} in the chloroplast. In that case, any addition of mitochondrial CO_{2} will have no impact upon the δ^{18}O of chloroplast CO_{2}. The photosynthesis data set that we collected for R. communis did not allow us to test these assumptions because θ was very close to 1.0; thus, the δ_{c0} signal was completely washed out by the activity of carbonic anhydrase.
However, this was not the case for dark respiration, during which θ was approximately 0.80. The method of Gillon and Yakir (2000b) leads to mean δ_{c0} values for the first and second dark respiration experiments of 55.2 and 43.7‰, respectively. These values can be compared to the mean δ_{c0} values generated by the regression method of 30.8 and 14.3‰, respectively. Although the regression method makes no a priori assumptions about the controls on δ_{c0}, we caution against overinterpretation of these latter values for the following reason: the regression analysis, as summarized in Equation 6, assumes no variation in θ and δ_{c0} among individual measurements in each experiment. The δ_{a} values varied among measurements according to how the leaf was modifying the δ^{18}O of CO_{2} in the leaf chamber. Therefore, to the extent that δ_{c0} is controlled by δ_{a}, δ_{c0} could also have varied among individual measurements.
Nonetheless, the large variation between δ_{c0} calculated as suggested by Gillon and Yakir (2000b) and the apparent δ_{c0} values observed in the dark respiration experiments warrants some discussion. There are three possible sources for the oxygen in CO_{2} evolved in mitochondria during either dark respiration or photosynthesis: atmospheric O_{2}, organic oxygen from respiratory substrates, and oxygen from leaf water. Atmospheric O_{2} has a δ^{18}O near 23.5‰ (VSMOW scale), and discrimination against ^{18}OO during respiration in plant tissues ranges from about 17‰ to 26‰ (Guy et al., 1992). We would therefore expect the δ^{18}O of respiratory CO_{2} deriving its oxygen atoms from O_{2} to be in the range of 0‰ to 5‰. Assuming the O_{2} tank used in our experiments had a δ^{18}O similar to atmospheric O_{2}, this range of values would apply. Organic oxygen in phloem sap sugars of the R. communis plants that we studied had a mean δ^{18}O of 27.5 ± 0.6‰ (mean ± 1 sd; n = 10). Generally, this oxygen pool is expected to have a δ^{18}O enriched by 27‰ compared to δ_{L} at the time of photosynthesis (Cernusak et al., 2003). Oxygen atoms derived from water during respiratory reactions would also be expected to be enriched by 27‰ compared to the δ^{18}O of the water source. The difference between the δ^{18}O of CO_{2} derived from any of these three sources and that of CO_{2} diffusing into the leaf from the atmosphere, prior to equilibration with leaf water, would depend on δ_{a} and, in the case of organic oxygen and oxygen from water, δ_{L}. However, it seems likely that under most circumstances the effect of incomplete equilibration between CO_{2} evolved from mitochondria and leaf water would be to decrease δ_{c0} below the value predicted by the formulation given by Gillon and Yakir (2000b). More experiments like those conducted by Yakir et al. (1994) would be helpful for resolving this issue.
Farquhar and Lloyd (1993) discussed the departure of δ_{c} from that predicted for equilibrium with δ_{e} during photosynthesis in terms of the ratio of the rate of carboxylation by Rubisco to the rate of CO_{2} hydration by carbonic anhydrase. This ratio was termed ρ. A simplified nonequilibrium equation for discrimination against C^{18}OO during photosynthesis, neglecting the possible effects of photorespiration and day respiration, was presented as (Farquhar and Lloyd, 1993)(32)where b^{18} is discrimination against C^{18}OO by Rubisco. Using this equation, and assuming b^{18} = 0, we calculated a mean ρ value for our photosynthesis measurements of −0.002 ± 0.009 (mean ± 1 sd; n = 8); if the outlier in Figure 3 is excluded, the mean ρ value becomes 0.001 ± 0.006 (mean ± 1 sd; n = 7). These values can be compared to a mean ρ value calculated for Phaseolus vulgaris of 0.025 (Flanagan et al., 1994). Thus, the ρ values that we observed for R. communis were somewhat smaller than those observed previously for P. vulgaris. These values can be compared to a theoretical prediction for ρ of approximately 0.05 (Cowan, 1986).
In our calculations we have assumed that the δ^{18}O of chloroplast water is equivalent to δ_{e}. One might expect chloroplast water to be slightly less enriched than δ_{e} due to the Péclet effect (Farquhar and Lloyd, 1993), which describes the interplay between advection of water toward the evaporative sites and diffusion of heavy isotopes away from the evaporative sites. We found that correlations between δ_{c} and δ_{e} were generally stronger than between δ_{c} and δ_{L}. This agrees with previous results (Flanagan et al., 1994), and suggests that δ_{e} is a more relevant parameter for predicting δ^{18}O of CO_{2} diffusing out of leaves than δ_{L}.
Gillon and Yakir (2000a) suggested that the CO_{2} partial pressure at the chloroplast surface (C_{cs}) is a more appropriate parameter for predicting discrimination against C^{18}OO during photosynthesis than that at the sites of carboxylation by Rubisco (C_{c}). They reconstructed C_{cs} by combining measurements of C^{18}OO discrimination and carbonic anhydrase activity. We did not measure carbonic anhydrase activity directly, and so could not modify our calculations to take into account C_{cs}. In cases where the total resistance from the chloroplast to the atmosphere in the dark is dominated by the stomatal resistance, use of C_{cs} in place of C_{c} will likely not alter predictions of δ_{R} to a very large extent. However, if stomata are relatively open and (δ_{c} − δ_{a}) is large, such that the invasion term in Equation 31 is large, a variation between C_{c} and C_{cs} of as little as 2 μbar could have a significant effect on predicted δ_{R}. In such cases it may prove helpful to use C_{cs} in place of C_{c}, if possible.
Farquhar et al. (1993) found that a globally averaged leaf water δ^{18}O of 4.4‰ satisfactorily balanced the global budget for δ^{18}O of atmospheric CO_{2}. In the most recent study of the global budget for δ^{18}O of atmospheric O_{2}, a globally averaged leaf water δ^{18}O of between 6.1 and 6.8‰ was estimated (Hoffmann et al., 2004). Gillon and Yakir (2001) suggested that the globally averaged leaf water δ^{18}O could be as much as 3‰ more than the estimate of Farquhar et al. (1993), in agreement with the requirement for balancing the Dole effect (global ^{18}OO budget); the global C^{18}OO budget could then be maintained by incomplete equilibration of chloroplast CO_{2} with chloroplast water (i.e. θ < 1). They estimated a globally averaged θ of 0.80. The results presented in this study provide an additional reason that the apparent leaf water signals required to balance the global C^{18}OO and ^{18}OO budgets should not be expected to resolve into a single value. The apparent leaf water signal relevant to the global δ^{18}O budget for O_{2} is the average daytime leaf water δ^{18}O, weighted by diurnal (daytime) variation in photosynthetic oxygen evolution rates. In contrast, the apparent leaf water signal relevant to the global δ^{18}O budget for CO_{2} is the 24h average leaf water δ^{18}O, weighted by diel (day and night) variation in g_{tc}C_{c}/P. Thus, the apparent leaf water δ^{18}O signals relevant to the global C^{18}OO and ^{18}OO budgets are fundamentally different.
CONCLUSION
We observed a very large variation in the δ^{18}O of CO_{2} respired by leaves in the dark, with observed values ranging from 44‰ to as high as 324‰. We have shown that this large range of δ_{R} values can be satisfactorily explained by taking into account the flux of CO_{2} that enters the leaf, equilibrates with leaf water, and diffuses out of the leaf without affecting the net CO_{2} efflux. Incorporation of the correct expression for δ^{18}O of leaf dark respiration into ecosystem and global scales models of C^{18}OO dynamics could affect model outputs and their interpretation.
MATERIALS AND METHODS
Plant Material and Gas Exchange Measurements
Ricinus communis plants were grown from seeds in 10L pots for 8 to 12 weeks in a temperature and humidity controlled glasshouse. Growth conditions were essentially the same as those described by Cernusak et al. (2003). Daytime temperature and humidity were 27°C ± 2°C and 40% ± 10%, respectively. Nighttime temperature was 20°C, with the same humidity as during the day. Measurements were made on fully expanded leaves of plants that were approximately 1 m tall. Projected areas of measured leaves ranged from approximately 400 to 800 cm^{2}. The configuration of the gas exchange system was recently described (Cernusak et al., 2003). The throughflow rate of air in the leaf chamber was approximately 3 L min^{−1}. Chamber air cycled continuously through a bypass drying loop to remove water vapor. The flow rate through the bypass drying loop was varied between 5 and 45 L min^{−1} to achieve different vapor pressures within the chamber, and therefore different values of e_{a}/e_{i}, and consequently of δ_{e}. Air entering the leaf chamber was generated by mixing 79% dry nitrogen with 21% dry oxygen using two mass flow controllers. Carbon dioxide was added to this air stream from a cylinder of 10% CO_{2} in air. Leaf temperature was measured with eight thermocouples arrayed across the underside of the leaf, and the average of these measurements used in gasexchange and isotopic calculations. Gasexchange calculations were performed according to the equations of Caemmerer and Farquhar (1981).
After gas exchange conditions in the leaf chamber stabilized for a time period judged long enough for leaf water to reach isotopic steady state, CO_{2} was cryogenically trapped from air exiting the chamber, as described previously (Evans et al., 1986; Caemmerer and Evans, 1991). Trapping continued until approximately 50 μmol of CO_{2} was obtained. The time period sufficient for leaf water to reach isotopic steady state was assumed to be three times the residence time of lamina leaf water (Förstel, 1978). The residence time of lamina leaf water was calculated as W/g_{t}w_{i}, where W is the lamina water concentration (mol m^{−2}), g_{t} is the total conductance of boundary layer plus stomata to water vapor (mol m^{−2} s^{−1}), and w_{i} is the mole fraction of water vapor in the leaf intercellular air spaces (mol mol^{−1}). The term W was determined to be 6.3 ± 0.4 mol m^{−2} (mean ± 1 sd) from measurements of the difference between fresh weight and dry weight for one leaf from each of five plants. This mean value of W was assumed for all leaves in the experiment; g_{t} and w_{i} were calculated continuously for each leaf being measured. Time periods calculated in this way for leaf water to reach isotopic steady state after a step change in humidity ranged from approximately 0.5 to 3.5 h.
Three experiments were conducted, two in the dark and one in the light. In the first dark experiment, air entering the leaf chamber was free of CO_{2}. All CO_{2} in the air exiting the chamber was therefore derived from the leaf. Measurements were conducted on one leaf from each of five plants. Each leaf was subject to two or three different chamber vapor pressures, and CO_{2} collected after gas exchange had stabilized for the requisite amount of time at each vapor pressure. Chamber air temperature was maintained at approximately 30°C. The second dark experiment was similar to the first, but differed in that CO_{2} was added to the air entering the chamber, such that the partial pressure within the chamber was approximately 350 μbar. The third experiment was in the light. Irradiance varied between 300 and 800 μmol PAR m^{−2} s^{−1}, and chamber air temperature varied between 25°C and 30°C. The CO_{2} partial pressure within the chamber was approximately 350 μbar.
Isotope Measurements
The carbon and oxygen isotope composition of CO_{2} exiting the leaf chamber was determined on an Isoprime mass spectrometer (Micromass, Manchester, UK) operating in dual inlet mode. Repeated analyses of the same gas sample generally showed a precision of better than 0.1‰ (1 sd, n = 10) for δ^{13}C and δ^{18}O. The carbon and oxygen isotopic composition of the gas used as a reference for the dual inlet measurements was calibrated against standard gases supplied by the International Atomic Energy Agency (Vienna). Oxygen isotope ratios in this paper are presented relative to VSMOW; carbon isotope ratios are presented relative to the Vienna Pee Dee Belemnite standard (VPDB). The oxygen isotope composition of irrigation water fed to the plants was determined with an Isochrom mass spectrometer (Micromass) operating in continuous flow mode (Farquhar et al., 1997). The water samples were pyrolyzed in a custombuilt furnace at 1,300°C prior to entering the mass spectrometer. Precision of analyses, based on repeated measurements of a laboratory standard water sample, was 0.3‰ (1 sd, n = 10). The δ^{18}O of the irrigation water was found to be −7.2 ± 0.2‰ (mean ± 1 se; n = 6).
We assumed that the only source of N_{2}O in the leaf chamber was the compressed air that the CO_{2} was mixed into, and that the concentration of N_{2}O in this air was 300 nmol mol^{−1}. The CO_{2} concentration was 10%, giving a ratio of N_{2}O to CO_{2} of 3 × 10^{−6}. This ratio could have been doubled during photosynthesis measurements, when the CO_{2} concentration exiting the chamber was as little as onehalf that entering it, giving a ratio of 6 × 10^{−6}. Using the empirical equations of Mook and van der Hoek (1983), this ratio of N_{2}O to CO_{2} would result in measurement biases of 0.002‰ for both δ^{13}C and δ^{18}O. This bias was considered negligible, and no attempt was made to account for contamination of CO_{2} samples by N_{2}O.
Acknowledgments
We thank Matthias Cuntz, Roger Gifford, Ricardo Marenco, and Francesco Ripullone for helpful comments on earlier versions of the manuscript.
Footnotes

Article, publication date, and citation information can be found at www.plantphysiol.org/cgi/doi/10.1104/pp.104.040758.
 Received February 9, 2004.
 Revised April 29, 2004.
 Accepted May 3, 2004.
 Published September 17, 2004.