Global Biogeochemical Cycles

Effects of photorespiration, the cytochrome pathway, and the alternative pathway on the triple isotopic composition of atmospheric O2



[1] The triple isotopic composition of atmospheric O2 is a new tracer used to estimate changes in global productivity. To estimate such changes, knowledge of the relationship between the discrimination against 17O and the discrimination against 18O is needed. This relationship is presented as θ = ln(17α)/ln(18α). Here, the value of theta was evaluated for the most important processes that affect the isotopic composition of oxygen. Similar values were found for dark respiration through the cytochrome pathway (0.516 ± 0.001) and the alternative pathway (0.514 ± 0.001), and slightly higher value was found for diffusion in air (0.521 ± 0.001). The combined effect of diffusion and respiration on the atmosphere was shown to be close to that of dark respiration. The value we found for photorespiration (0.506 ± 0.005) is considerably lower than that of dark respiration. Our results clearly show that the triple isotopic composition of the atmosphere is affected by the relative rates of photorespiration and dark respiration. Also, we show that closing the current global isotopic balance will enable the estimation of the current global rate of photorespiration. Using the Last Glacial Maximum as a case study, we show that variations in global rate of photorespiration affected the triple isotopic composition in the past. Strong fractionations measured in illuminated plants indicated that the alternative pathway is activated in the same conditions that favor high rate of photorespiration. This activation suggests that the global rate of the alternative pathway is higher than believed thus far, and may help to close the gap between the calculated and measured Dole Effect.

1. Introduction

[2] The isotopic composition of atmospheric O2 fluctuated over glacial-interglacial timescales. Variations in the ratios of all three stable O2 isotopes were applied to estimate changes in global gross production [Luz et al., 1999]. The difference between the 18O/16O ratio of atmospheric O2 and seawater H2O is known as the Dole Effect. Variations in the Dole Effect, are used to estimate changes in the ratio of marine to terrestrial productivity in the past [Bender et al., 1994]. To improve interpretation of past atmospheric changes, it is necessary to understand the basic processes affecting atmospheric isotopic composition.

[3] Atmospheric O2 is produced by photosynthesis without isotopic fractionation from the substrate water [Guy et al., 1993]. Thus, marine photosynthesis results in production of O2 with identical isotopic composition to seawater. The substrate water in leaves is enriched in 17O and 18O through evapotranspiration [Dongmann, 1974]. As a result, terrestrial photosynthesis produces O2 that, on average, is enriched in these heavy isotopes. In both marine and terrestrial systems, the major cause of the heavy isotope enrichment is preferential removal of 16O (over 17O and 18O) by biological uptake mechanisms [Lane and Dole, 1956]. This preferential removal is mass-dependent and the relative increase in 17O/16O is about half (∼0.52) of the relative increase in 18O/16O (Figure 1). In addition to evapotranspiration and biological uptake, atmospheric O2 is also affected by photochemical reactions in the stratosphere [Luz et al., 1999]. These reactions preferentially transfer 17O and 18O from O2 to CO2 in a mass-independent way. Hence, in these reactions the ratio of the increase in 17O/16O over 18O/16O in CO2 is equal to 1 or larger [Lämmerzahl et al., 2002; Thiemens, 1999]. As a result, O2 becomes mass-independently depleted in the heavy isotopes [Bender et al., 1994]. At the surface of the Earth, the 17O depletion of atmospheric O2 is removed by exchange through photosynthesis and uptake. Therefore, the net result is depletion in 17O of atmospheric O2 in comparison to O2 affected by biological uptake alone (Figure 1). The magnitude of this depletion depends on the ratio of production of 17O depleted O2 in the stratosphere, and its destruction by biological cycling. Thus, past variations in the 17O depletion can be used to infer changes in global biological productivity.

Figure 1.

Schematic plot (not to scale) of 17O and 18O relative to 16O variations. Oxygen is produced by photosynthesis and it is fractionated by biological uptake in a mass-dependent way, and by stratospheric photochemistry in a mass-independent way. The balance between these two types of fractionation control the triple isotopic composition of atmospheric O2.

[4] In order to apply this approach, it is necessary that the relationship between 17O/16O and 18O/16O to be known with great accuracy. However, the triple isotope relationships in different physical-chemical mass-dependent processes, varies slightly about 0.52 [Luz and Barkan, 2000; Matsuhisa et al., 1978; Young et al., 2002]. If this is also the case for O2 consumption by different biological mechanisms, then a change in the ratio of uptake by the different pathways will result in variations in the magnitude of the 17O depletion. As a consequence, past variations in this parameter will be misinterpreted.

[5] In interpretations of variations in the Dole Effect it is usually assumed that the major driving force is changes in the terrestrial to marine production rate (terrestrial produced O2 is heavier than marine produced, because of the enrichment of leaf water in evapotranspiration). An implied assumption is that the enrichment due to biological uptake remains constant. In the present paper, we explore the possibility that time variations in Dole Effect were also affected by changes in the relative rates of O2 uptake by different biological mechanisms, that have different fractionations associated with them. Consumption (or uptake) of O2 in plants occurs via four mechanisms: ordinary respiration through the cytochrome oxidase pathway (COX), respiration by the alternative oxidase pathway (AOX), Mehler reaction, and photorespiration. The first two processes take place in light as well as in dark conditions, while the latter two occur only under illumination. In addition, O2 diffusion in air precedes biological uptake in some cases. The discrimination against 18O associated with the AOX (∼28‰) is stronger than the discrimination associated with other mechanisms (COX ∼ 18‰, Mehler ∼ 15‰, photorespiration ∼ 22‰). Hence, an increase in the proportion of O2 uptake by AOX will significantly intensify the Dole Effect.

[6] To improve the interpretation of past variations in the triple O2 isotopes and the Dole Effect, we have investigated the effects of biological mechanisms in controlled laboratory experiments. In one set of experiments, we studied triple isotope variations due to uptake alone by COX and AOX, and due to removal by diffusion. In a second set of experiments, conducted in an airtight terrarium, we analyzed the combined effects of production and consumption by both dark and light reactions in steady state between photosynthesis and uptake.

2. Triple Isotope Systematics

2.1. Terminology

[7] Atmospheric O2 is depleted in 17O due to stratospheric reactions. Hence, O2 that is produced by photosynthesis will have 17O excess with respect to atmospheric O2. This 17O excess was formulated in previous papers [Luz and Barkan, 2000; Luz et al., 1999] as Δ17O = δ17O − 0.521 × δ18O. However, this definition is somewhat problematic since it contains a linear approximation to the mass-dependent fractionation, and because the parameter 0.521 (that represents mass-dependent processes) depends on the choice of reference material. Here, we follow the suggestion of Miller [2002] for a new definition of the 17O excess that is free of these problems:

equation image

[8] The new definition of the 17O excess will be noted as 17Δ to distinguish it from the old definition (Δ17O). 17R stands for the isotope ratio 17O16O/16O2, 18R stands for the isotope ratio 18O16O/16O2, and the subscript “ref” stands for the reference. It should be noted that 17Δ is a calculated value that depends both on the standard used and on the value of the parameter C. This parameter represents the triple isotope relationships in mass-dependent processes. However, as was mentioned in the introduction, these relationships vary slightly about 0.52 for different processes. As a result, The chosen C value can only correspond to the relationship associated with one of the processes, preferably the main one. In this study we choose to use a value of C = 0.516, which corresponds to ordinary dark respiration (COX), the most wide spread global O2 uptake mechanism.

[9] In the old definition of the 17O excess (Δ17O) the variations in the isotopic composition were described on a δ17O versus δ18O plot. To describe the variation in the isotopic composition according to the new definition, two new terms that will be used for the axes of such plot are suggested,

equation image
equation image

[10] Another advantage of 17Δ, Ln17O and Ln18O is that they are dimensionless. Hence, the value of a sample versus the reference exactly equals to the value of the reference versus the sample with the opposite sign (this is not true for Δ17O, δ17O and δ18O). For convenience, Ln17O and Ln18O are expressed as permil (‰) and 17Δ as per meg deviations from the standard.

[11] The slope of a line passing between two points (A and B) on a Ln17O versus Ln18O plot, will be defined as λ, as in previous work [e.g., Miller, 2002],

equation image

[12] Following Mook [2000], the relationship between the discrimination against 17O and 18O (relative to 16O) is presented as

equation image

where 17α and 18α are the fractionation factors xRp/xRs. The subscripts “p” and “s” stand for “product” and “substrate,” respectively (x can be 17 or 18).

[13] Blunier et al. [2002] used an additional relationship between the discriminations,

equation image

where xε is xα − 1.

[14] The terms θ and γ describe the inherent relationship between the discriminations, but cannot be measured directly. What can be measured is the slope (λ) that is controlled by these relationships and the processes that take place in the system. In the special case of a system at steady state, in which production equals uptake, the slope (λ) is equal to θ (see section 2.2). In the special case of a system where only uptake takes place, the slope (λ) is equal to γ (see section 2.3). Hence, by conducting experiments in such systems, θ or γ could be calculated from the measured λ. In other systems (for example, in ones where mixing takes place) the slope can be different from both θ and γ. The difference between the value of θ and γ is small (∼0.003), but since the 17O excess variations in O2 are minute such a difference cannot be neglected.

[15] This paper deals with the effect of the biosphere on the triple isotopic composition of the atmosphere. Since the biosphere-atmosphere system is close to production-uptake steady state, we chose to use θ to report the relationship between discriminations.

2.2. Production-Uptake Steady State

[16] In this study, we used a terrarium similar to that of Luz et al. [1999] to estimate the θ value associated with photorespiration. To understand how the θ value of a process can be evaluated from such experiment, we will discuss the triple isotope systematics of a closed system in biological production-uptake steady state. Such a system can also be used as a model for the Earth atmosphere and biosphere without the effects of stratospheric photochemistry and the hydrological cycle.

[17] The starting point for such a system is oxygen that is produced by photosynthesis in photosystem 2 from water (noted as “W”) without isotopic fractionation. Uptake by respiration causes enrichment in the heavy isotopes. Since the fractionation in biological uptake is mass dependent, the enrichment in 17O will be about half of the enrichment in 18O. In steady state, depending on the value of θ, the system will reach either the point that is marked as “BSS1” (Biological Steady State 1) or “BSS2” in Figure 2. The Ln18O value of the O2 in steady state relative to the substrate water is illustrated by the horizontal distance between “BSS1” (or “BSS2”) and “W,” and is the system's equivalent of the “Dole Effect.”

Figure 2.

Schematic plot (not to scale) of Ln17O versus Ln18O for a closed system in production-uptake steady state. Point “W” represents oxygen that is produced by photosynthesis in photosystem 2 from water, and “BSS1” and “BSS2” represent biological steady states. The slope of the line that connects “BSS1” and “W” is equal to θ1 and the slope of the line connecting “BSS2” and “W” is equal to θ2(θ = ln(17α)/ln(18α)). When θ of the system equals θ1 the system reaches the steady state condition indicated by “BSS1” with an 17O excess of 17ΔBSS1. Since C equals to θ1, 17ΔBSS1 equals to the 17ΔW. When θ of the system equals θ2 the system reaches steady state indicated by “BSS2” with 17O excess of 17ΔBSS2 that is lower than 17ΔBSS1. The horizontal distance between “BSS” and “W” is the system's equivalent of the global Dole Effect (in Ln18O terms). The difference between 17ΔBSS1 and 17ΔBSS2 is given by the system's “Dole Effect” times the difference between θ1 and θ2.

[18] A system in production-uptake steady state can be analyzed graphically as in Figure 2, or, more rigorously, by a model that deals with the fluxes of the three isotopic species. Such a one-box model of production-uptake steady state can be formulated as follows:

equation image

where yP is production rate of yO16O (y can be 16, 17 or 18), and yU is uptake rate of yO16O.

[19] Equation (7) can be rewritten as follows:

equation image

where x can be 17 or 18, the subscript “W” stands for oxygen produced by photosynthesis, and “BSS” stands for the system air in biological steady state.

[20] Rearranging equation (8) and substituting it into equation (4) gives

equation image

[21] Hence, the slope of the line connecting the O2 of a biological system in production-uptake steady state, and the O2 produced from the substrate water, is equal to the value of θ. However, as we will show in section 2.3 below, the value of λ is not always equal to the value of θ.

[22] The 17Δ in production-uptake steady state can be found by substituting equations (7) and (8) into equation (1),

equation image

[23] According to equation (1), the 17Δ of the photosynthetically produced oxygen is

equation image

[24] Subtracting equation (11) from equation (10), and using equation (5) yields

equation image

[25] When θ equals C, the 17Δ value of the air in steady state equals to that of the oxygen that is produced from the substrate water. However, if θ is different from C, then 17ΔBSS is controlled both by θ and 18α.

[26] Currently, the value of 17ΔW cannot be measured directly with sufficient accuracy. However, this value can be estimated by conducting a terrarium experiment in which the θ value of the uptake process is known (this known θ will be noted as θ1). In such an experiment, the value of 17Δ in steady state (17ΔBSS1) will be identical to that of 17ΔW, if we choose C = θ1. By conducting an additional experiment in which the θ value of the uptake process is unknown (this unknown θ will be noted as θ2), we can calculate the value of θ2 from the 17O excess in steady state of this experiment (17ΔBSS2), by rearranging equation (12),

equation image

[27] The value of ln(18α) for the system can be found by rearranging equation (8),

equation image

[28] The right-hand side of equation (14) is the Ln18O value of “BSS” versus “W,” which is equivalent to the terrarium Dole Effect in Ln18O terms (the value of the Dole Effect in δ18O terms is greater by 0.3‰ if SMOW is the reference and lower by 0.3‰ if atmospheric O2 is the reference). Hence, the calculation of θ2 from equation (13) is identical to the solution presented graphically in Figure 2.

[29] To estimate θ values by the method describe above, the θ value of at least one process (θ1) must be known independently. This value can be found by conducting experiments in systems in which there is O2 uptake but no production.

2.3. O2 Removal Only

[30] In our dark respiration and binary diffusion experiments, O2 is only removed from the system. The change in isotopic composition in such experiments follows the Raleigh distillation equation,

equation image

where xR0 is the initial isotope ratio xO16O/16O2 and f is the remaining 16O2 fraction. Substituting equation (15) into equation (4) yields

equation image

where the subscript R stands for Raleigh distillation. Equation (16) shows that in a system where there is only uptake of oxygen, the value of λ is different from the value of θ. The value of θ in “removal only” experiments can be calculated by substituting equation (16) into equation (5),

equation image

[31] The range of 18ε in biological uptake is −15 to −32‰ and the value of θ in mass-dependent fractionation is about 0.5. Hence, λR will be larger by 0.002–0.004 than θ in most systems. For O2 samples close in composition to the standard O2, using slightly different C values will not significantly affect the calculated 17Δ. However, for the purpose of the present study where comparisons between steady state O2 and its substrate water are necessary, a small difference in parameter C becomes significant. To illustrate the importance this point, consider a water sample having Ln18O = −23‰ and Ln17O = −11.7‰ with respect to air O2. Using equation (1) and C = 0.52, 17Δ of this sample is calculated as 260 per meg. Similar calculation with C = 0.524 yields 17Δ of 352 per meg.

3. Experimental Methods

3.1. Dark Respiration Experiments

[32] Axenic cultures of Lemna gibba, a small water plant, were grown in a nutrient solution. Illumination from mixed fluorescent/incandescent lamps was at an intensity of 200 ± 20 μE m−2 s−1 photosynthetic photon flux (PPF), at the level of the fronds, for 12 h d−1. Air temperature was 24°C and 19 ± 0.5°C in the light and dark periods, respectively.

[33] The seeds of Orobanche aegyptica, an obligatory root parasite, were placed in empty tea bags and suspended for sterilization in 1% solution of NaOCl for 5 min. The seeds were then rinsed thoroughly for 20 min in sterile distilled H2O. After the rinse, the seeds were activated by 3 days imbibition in 5 ml of H2O in 9-cm-diameter Petri dishes with one layer of Whatman paper, to which a mixture of antibiotics was added. The mixture contained 50 μg each of streptomycin and penicillin G and 25 μg of chloramphenicol, in order to prevent the development of bacterial infections during the imbibition period.

[34] To determine the θ value associated with the COX, we inhibited the AOX in the plant samples with 2 mM Salicylhydroxamic acid (SHAM). To determine the value associated with the AOX, we inhibited the COX with 1 mM NaCN. The inhibitors were added to a closed aerated chamber containing the plant samples and water for at least 90 min before measurements were made. Additional control experiments without inhibitors were conducted with Lemna and natural soil (the resulting θ value was expected to lie between that of the AOX and the COX).

[35] For the incubation, about 1 g of plant sample was inserted into 6 cm3 blood collection tubes (Vacutainers®) that were closed with rubber septa. The tubes were immersed in water during incubation in order to prevent air leaks. After an incubation period of 1 to 10 hours, the air in the tubes was sampled, via a needle, directly to the vacuum preparation line. In other experiments, plant samples were incubated in water and the changes in the dissolved oxygen were monitored. At the end of these experiments, 120 cm3 water were sampled in 250 cm3 pre-evacuated flasks closed with a Lowers Happert® valve. Sampling and extraction of the dissolved gases were done according to Luz et al. [2002].

3.2. Terrarium Experiments

[36] We used an airtight terrarium as described by Luz et al. [1999]. The terrarium contained Philodendron plants, soil and natural water from Lake Kinneret, Israel (δ18O = 0.5‰ versus SMOW). It was illuminated by a fluorescent lamp (100 μE m−2 s−1) for 24 h d−1, 10 h d−1, 4 h d−1 and 2 h d−1 in different stages of the experiment. Different light conditions were used to manipulate the CO2 concentration in the terrarium, which, in turn, affects the relative rate of photorespiration [Badger, 1985]. The terrarium was also covered and darkened for dark incubations. The CO2 concentration in the terrarium air was determined by sampling 6 cm3 air in pre-evacuated blood collection tubes (Vacutainers®), and measuring the air sample with an infrared gas analyzer (LI-COR-6252®) by a method similar to Davidson and Trumbore [1995]. The relative accuracy for CO2 concentration measurement was ±5%.

3.3. O2-N2 Diffusion Experiments

[37] In the diffusion of O2 in N2 experiments, a 4-cm3 flask was filled with pure oxygen. The neck of the flask was filled with a diffusive medium (plastic sponge) that prevented advective mixing, and the oxygen that diffused out of the tube was immediately removed by a flow of N2. After 1–2 hours, the O2 concentration in the flasks was considerably lowered and the flasks were closed and transferred for isotopic analysis.

3.4. Sample Preparation and Mass Spectrometry

[38] Sampling, sample preparation, and mass spectrometry were according to Angert and Luz [2001], Luz et al. [2002], and Luz and Barkan [2000]. The preparation of the sample included cryogenic removal of water vapor and CO2, and chromatographic separation of N2, which is needed for accurate 17Δ measurements. The samples were frozen at ∼4°K into stainless steel tubes and transferred for analysis in a Finnigan Delta-Plus mass-spectrometer. Corrections were applied in order to account for the sensitivity of ionization efficiencies of the three isotopic species to variations in the O2/Ar ratio. The analytical error (absolute difference from the average) for δ18O, δO2/Ar, and Δ17O was 0.02‰, 0.5% and 5 per meg, respectively. All values are reported with respect to the atmospheric O2 (HLA standard).

4. Results

[39] The results of the dark incubation and the diffusion of O2 in N2 experiments, are illustrated in Figure 3, and summarized in Table 1. The isotopic discrimination (18ε) was calculated by equation (15) from the results of the dark incubation experiments. As expected, higher fractionation for each plant species was measured when the plants were inhibited with NaCN, and lower fractionation when the plants were inhibited with SHAM. Considerably lower uptake rates were measured when the plant was inhibited with NaCN (data not shown). The value of λR in these experiments was determined from the Ln17O versus Ln18O according to equation (4). The λR values that were found are 0.518 ± 0.001 for the COX, 0.5179 ± 0.0003 for the AOX. There was no significant difference between the λR values in the incubations of Lemna gibba and Orobanche aegyptica, nor between different experiments with the same plants. The λR value determined for incubation without inhibitors of Lemna was 0.517 ± 0.001, and a value of 0.517 ± 0.002 was found for the natural soil. In the two diffusion experiments we found λR value of 0.5228 ± 0.0002, with no significant difference between the experiments.

Figure 3.

Relative changes in the triple isotopic composition of O2 (plotted as Ln17O versus Ln18O in per mils) as result of (a) uptake through the COX, (b) uptake through the AOX, and (c) removal by diffusion into N2. The slope of the fitted regression line is λ, and the reported precision accuracy is standard error. The values of θ are calculated according to equation (17).

Table 1. Summary of Dark Incubation Experiments and Diffusion Experimentsa
  • a

    Abbreviations: n, number of data points; SE, standard error.

Orobanche aegypticaSHAM30.5170.003−14.50.9
Orobanche aegypticaSHAM40.5200.006−14.20.7
Lemna gibbaSHAM70.5180.003−20.70.1
Lemna gibbaSHAM70.5180.002−19.60.2
Lemna (in water)SHAM80.5180.001−14.10.4
COX 290.5180.001  
Orobanche aegypticaNaCN40.5160.002−18.50.9
Orobanche aegypticaNaCN50.5170.002−21.51.4
Lemna gibbaNaCN60.51810.0003−27.90.1
Lemna gibbaNaCN60.5180.001−27.10.2
AOX 210.51790.0003  
Lemna gibba-40.5170.001−20.10.3
Diffusion in N2-60.52260.0004−14.90.1
Diffusion in N2-60.52310.0003--
Diffusion in N2 120.52280.0002  

[40] The 17Δ, δO2/Ar, Ln18O and [CO2] of the terrarium air in the course of the experiment are presented in Table 2 and shown in Figure 4. In days 1–58, in which there was a 24-hour illumination per day, the CO2 concentrations in the terrarium were about 150 ppm. In days 127–178, when the illumination was of 10 h d−1, CO2 concentration decreased from 4500 ppm in the beginning of the light period to 150 ppm, after about 3 hours. In days 184–235 and 330–365, in which there were 2–4 hours illumination, the CO2 concentration were 10,000–40,000 ppm, and in days 249–252, in which there were 4 hours of illumination, CO2 decreased to about 150 ppm after 2–3 hours of illumination. The isotopic discrimination (18ε) that was calculated from the dark periods (days 183–184 and 283–285) is −16.7‰.

Figure 4.

Summary of the terrarium experiment results. Shown in the markers are the variations in (a) 17Δ versus the atmosphere in per megs (with C = 0.516), (b) Ln18O in per mils versus the terrarium water, and (c) δO2/Ar in per mils versus the atmosphere. The solid line shows the hours per day in which photorespiration was engaged due to low CO2 concentration (<500 ppm) and high illumination. Encircled are results of three time intervals in which 17Δ was close to steady state. These three intervals correspond to constant behavior of the CO2 concentration.

Table 2. Summary of the Terrarium Experiment
DayΔO2/Ar Versus Atm,aΔ18O versus Atm,aLn18O versus Waterb17Δ versus Atm,aLight, h d−1CO2, ppmCO2,<500 ppm,c h d−1
  • a

    Atmospheric standard HLA.

  • b

    Water in the terrarium.

  • c

    Number of hours per day that CO2 was less than 500 ppm.

3620.323.19124 24
3610.423.19524 24
4681.424.1 2413024
161220−1.521.214510 7
177246−1.321.513610 7
183247−0.921.91650 0
183198−0.222.51520 0
1841680.223.01510 0
235−157−6.616.12074 0
249−161−4.318.51822 2
252−230−2.620.21862 3
283−133−2.720.1 0 0
283−132−2.720.01850 0
284−191−1.621.11880 0
284−191−1.521.21770 0
285−247−0.422.41840 0
285−248−0.422.41940 0
330−439−1.920.92264 0
345−309−4.518.32244 0
355−197−6.016.81944 0
355−197−6.016.71994 0
365−87−5.817.02134 0

5. Discussion

5.1. Dark Incubation Experiments and Diffusion in Liquid Phase

[41] The value of θ was calculated from the measured λR values by equation (17). The fractionations (17ε and 18ε) are strongly controlled by the effects of slow diffusion to the consumption site [Angert and Luz, 2001; Guy et al., 1989]. However, the value of λR in our experiments can be shown to be independent of such effects. The overall fractionation of a system in which the oxygen supply is limited by diffusion is given by

equation image

taken from Farquhar et al. [1982], where xε is the overall fractionation, xεcon is the fractionation in consumption, xεdiff is the fractionation in diffusion, and Ca,Ci are the substrate concentrations in the ambient air and in the reaction site, respectively. The diffusion into the organic tissues of our plant samples and the water film around them is in liquid phase, hence, the fractionation associated with it (xεdiff) is very small [Farquhar and Lloyd, 1993] and could be taken as zero. In Raleigh distillation the value of λ is given by 17ε/18ε (equation (16)). Substituting xε obtained from equation (18) into equation (16), shows that λ is independent of the diffusion limitation (Ci/Ca) when the fractionation in diffusion is zero.

[42] To calculate the value of θ by equation (17) the value of 18ε is needed. Since the value of 18ε we measured in our experiments (Table 1) is affected by limiting diffusion, we used published values from experiments in which special care was taken to avoid this effect. We used a 18ε value of −18‰ for the COX [Guy et al., 1989] and −30‰ for the AOX [Ribas-Carbo et al., 2000]. The θ values that were calculated are 0.516 ± 0.001 for COX and 0.514 ± 0.001 for AOX.

[43] The θ values measured for each of the dark respiration pathways were identical in the two plant species. The θ values of the COX and the AOX were found to be very close. An agreement exists between the values found for the COX and the AOX, the values found in experiment without inhibitors for Lemna and for natural soil, and the value found for plankton in lake water (B. Luz and E. Barkan, unpublished results, 2001). This agreement indicates that the θ values reported here represent the θ of dark respiration pathways in general.

5.2. Diffusion in Air and the Effect of Soil Respiration

[44] The θ value for diffusion in N2 was calculated from the measured λR value by equation (17), using a 18ε value of −14‰ that was calculated from the theoretical equation for binary diffusion [Mason and Marrero, 1970]. This θ value was found to be 0.521. This value is significantly different from the value of 0.512 that can be calculated from the 18α and 17α resulting from the equation of binary diffusion. This difference may have originated from assumptions of the binary diffusion equation that are not realistic for our experiment. However, the θ value we found should be relevant for diffusion through porous media, such as soils.

[45] The measured θ value is 0.005 greater than the value we found for the COX (0.516). Thus, if O2 is consumed by COX from a reservoir that is not well mixed with the atmosphere, and to which oxygen enters by diffusion in air (which is very similar to diffusion in N2), the effective θ will be an intermediate value between the θ of COX and that of diffusion. Such a situation occurs in soils in general (and also the soil in our terrarium experiments), in which fractionation of entering O2 by diffusion in air has been shown to have strong effects [Angert et al., 2001]. Since soil respiration is an important component of the oxygen cycle, such an effect should be considered in any attempt to estimate the terrarium or global θ.

[46] The effective fractionation of a system in which the oxygen supply is limited by diffusion is given by equation (18). We can estimate the effective θ for different Ci/Ca ratios by calculating the 17ε and 18ε using this equation. Oxygen concentration in soils vary widely with soil depth, soil type, soil moisture, and rate of respiration in the soil. However, most of the soil respiration takes place at the top of the soil profile. Hence, it will be justified to assume that the Ci/Ca ratio for weighted average soil respiration is not smaller than 0.9. This value corresponds to about 20,000 ppm of CO2, which is in the upper range of the concentrations in soils. Even for this high value, the effective θ is 0.5164, only 0.0004 larger than that of COX and within the margin of error of its assessment. This effect might be compensated if some of the consumption in soil respiration is through the AOX, in which θ is lower than in the COX. Thus, the effect of diffusion on the effective θ in soils is very small. Confirmation for that comes from the λ value (0.517 ± 0.001) that was measured in natural soil, in which diffusion limitation was indicated by relatively weak fractionation (18ε = −16‰). In summary, effects of diffusion in air through soil profiles can be neglected in the discussion of the global triple isotope balance as well as the isotopic balance in the terrarium. Additional diffusion limitations in soils occur in liquid phase, but, as shown in the previous section, the effect on the triple isotopic composition is negligible.

5.3. The 17Δ and θ in the Terrarium Experiment

[47] Since the values of θ we found for the COX and AOX are very close, and because COX is the most widespread global uptake mechanism, we choose to use the θ value of the COX - 0.516 for the parameter C (which is used to calculate 17Δ). In this respect, we must emphasize that inappropriate choice of parameter C may lead to misinterpretation of terrarium experimental results. For example, choosing C = 0.525 representing Ln18O-Ln17O relationship in rocks, led Young et al. [2002] to suggest that the major origin of the 17O deficiency in the atmosphere is not mass independent fractionation in the stratosphere. Clearly, air-rock oxygen exchange is negligible in comparison to the fluxes in photosynthesis and respiration. Therefore, the relationship between the latter two processes should be used in the interpretation of the triple isotope composition of atmospheric O2.

[48] Three time intervals in which the 17Δ of the terrarium air was in steady state can be identified in Figure 4. These three intervals correspond to constant behavior of the CO2 concentration in the terrarium air. In the first interval (marked as “High” in Figure 4,) CO2 concentration was very high (>5000 ppm) throughout the day and the average 17Δ was 215 ± 14 per meg. In the second interval (“Low”), the terrarium was illuminated for 24 h d−1, CO2 concentration was low (∼150 ppm), and the average 17Δ was 93 ± 6 per meg. In the third interval (“Variable”), the terrarium was illuminated for 10 h d−1 and the CO2 concentration changed with the illumination. The plant was exposed to CO2 levels below 500 ppm, while illuminated, for about 7 hours per day. The average 17Δ in this interval was 145 ± 6 per meg. Similar dependence of the Δ17 O values in illumination can be seen in the terrarium experiments of Luz et al. [1999]. Although there is no CO2 concentration data for these experiments, the relationship between illumination and CO2 were probably similar to the relationships in the current experiment.

[49] The different Δ17 O values in steady state indicate changes in the average θ of the terrarium (Figure 2). These changes indicate variations in the relative rates of different processes in the terrarium, each with its characteristic θ. The difference between the θ of COX and AOX we report here is only 0.002. Thus, even if the entire uptake in the terrarium was shifted from COX to AOX, which is not very likely, it will only cause a lowering of 17Δ of about 50 per meg (according to equation (12)). Therefore, the change of 122 per meg in 17Δ between intervals “High” and “Low” cannot be explained by changes in the ratio between COX and AOX. It can be explained by an increase in the importance of a process (or processes) in which θ is considerably smaller than in dark respiration. Because lowering of θ occurs when CO2 concentration is low, photorespiration is a likely mechanism. Another process that might be activated in the light is the Mehler reaction. Whether or not this is an important mechanism is debatable, but preliminary experiments in our laboratory indicate that the θ value in Mehler reaction is relatively high (Y. Helman, unpublished results, 2002), and hence this reaction could not be responsible for the lowering of the terrarium θ in the light. In addition, in a recent review, Badger et al. [2000] argue that O2 uptake by Mehler reaction is of small importance in higher plants.

[50] When CO2 concentration in the terrarium was high throughout the day (interval “High”), photorespiration was inhibited and the average θ was controlled by COX and AOX alone. Since the θ values of COX and AOX are close (0.516 and 0.514, respectively), and usually most of the respiration takes place through the COX, we assume that the average θ in interval “High” (“θ1”) was 0.516. Since θ1 is equal to C, the 17ΔW of the terrarium water is equal to 17ΔBSS in interval “High” (see equation (12)), 215 per meg on average. In interval “Variable,” the CO2 concentration was low enough to allow photorespiration in part of the day and Δ17 O dropped from 215 to 145 per meg, and it went down even further to 93 per meg, in interval “Low,” when low CO2 concentration and illumination were present for 24 hours per day (Figure 5).

Figure 5.

Analysis of the terrarium experiments (not to scale). In interval “High,” photorespiration was inhibited as a result of high CO2 concentrations and the relationship between the fractionations was controlled by dark respiration (θ1 = 0.516). In Interval “Low,” both dark respiration and photorespiration (PR) were engaged (θ2 = 0.511). To account for such an effect, the θ of photorespiration must be lower than 0.511, and it can be calculated from the relative rates of dark respiration and photorespiration as 0.506.

[51] By using the value of 215 for 17ΔW and 18ε of −24‰ (see section 5.5) in equation (13), we calculated an average θ for interval “Low” (“θ2”) of 0.511. This value of θ2 represents a weighted average for both dark respiration and photorespiration. This clearly indicates that the θ value of photorespiration must be significantly less than 0.516. A quantitative estimate of the θ value for photorespiration will be derived in section 5.6.

5.4. Deviations From Steady State

[52] While the 17Δ of the terrarium air was in steady state in intervals “Low,” “Variable” and “High,” the δO2/Ar values were constant only through interval “Low.” In intervals “Variable” and “High,” the O2 concentration increased with time, indicating a net production in the terrarium. To study the effect of this net production on the 17Δ, we used a simple numerical model. This model deals with the fluxes of production and uptake of the different isotopic species without the steady state assumption. The temporal concentration of isotopic species is given by equation (19) where Δt stands for time interval.

equation image

[53] As expected, running the model with 16U=16P and θ = C gave identical result to the analytical model (equation (12)); hence, when the value of 17Δ reached a plateau (equivalent to 17ΔBSS in the analytical model), it was equal to that of 17ΔW. However, running the model with production (16P) that is higher then uptake (16U) resulted in 17Δ values in a plateau slightly higher then 17ΔW. For example, running the model with 16U/16P ratio of 0.5 and 17ΔW of 211 per meg resulted in 17Δ in a plateau of 221 per meg. To estimate uptake in interval “High,” we assumed that it was identical to the uptake rate in the dark periods (the oxygen depletion rate). This assumption is probably true since the high CO2 levels in interval “High” inhibited photorespiration. From this uptake rate and the observed net production (rate of increase in O2 concentration), we found the ratio of uptake to production was 0.9 for interval “High.” Using this value in the numerical model, we found that this deviation from steady state will cause an increase of 4 per meg in the value of 17Δ, when it reached a plateau. Thus, 17ΔW is 211 per meg instead of the 215 per meg calculated based on the steady state assumption. This correction for 17ΔW is small with respect to other uncertainties and does not significantly change the value calculated for θ2 (0.511).

[54] Another nonsteady state effect results from the diurnal illumination cycle in the terrarium. Photosynthesis took place in the few hours of illumination, while the uptake through dark respiration continued all day, and as a result the O2 concentration fluctuated. By modeling this condition in the numerical model, we found that it will cause the 17Δ value to fluctuate around the value 17ΔBSS. The amplitude of the fluctuations in 17Δ depends on the amplitude of the fluctuations in O2 concentration. For interval “High,” we can calculate from the dark periods that dark respiration consumed about 9% of the O2 reservoir of the terrarium per day. Using this value and the value of 0.9 for uptake to production ratio, the calculated magnitude of the 17Δ fluctuations resulting from the light-dark cycle is ±2 per meg. Again, this value is considerably smaller than the analytical uncertainty, and therefore this effect can be also neglected.

5.5. Dependence of 18ε on Illumination and [CO2]: Implication for the Dole Effect

[55] The weighted-average 18ε of all the processes in the terrarium can be estimated from the terrarium equivalent of the global Dole Effect (equation (14)), which is the value of Ln18O of “BSS” versus “W.” The Ln18O values of the terrarium air versus the value of the substrate water are presented in Figure 4b. In the three intervals in which the 17Δ of the terrarium was constant, the Ln18O was almost constant.

[56] In interval “High,” the fractionation (ε) in the terrarium according to equation (14) is −18.4 ± 1.8‰, in interval “Low” −23.9 ± 0.5‰, and in interval “Variable” −21.7 ± 0.2‰. The fractionation in interval “High” is in agreement with the known fractionation for the COX, −18‰ [Guy et al., 1989], and with the fractionation that was calculated for the dark periods −16.7‰. This agreement indicates that, as was assumed in section 5.3., the uptake in the terrarium was dominated by COX in interval “High” in which the CO2 concentration was high.

[57] The fractionation by COX is −18‰, in photorespiration it is −21.7‰, and that of AOX is about −30‰ [Ribas-Carbo et al., 2000]. Thus, photorespiration and COX alone cannot explain the high Ln18O values measured in interval “Low” and interval “Variable.” These high values seem to indicate that a considerable portion of the uptake was through the AOX. Since in interval “Variable” there was net production that introduced oxygen with light isotopic composition, the fractionation must have been even stronger than that calculated above for the same interval according to the steady state assumption (−21.7‰). The fractionation in interval “Low” was extremely strong. The relative rate of uptake through the AOX in this interval was estimated from the observed Ln18O values in section 5.6. as 41–31% of gross production.

[58] Some enrichment of the terrarium leaf water by evapotranspiration might have contributed to the high Ln18O. However, since the relative humidity in the terrarium was 100% this effect was probably very small. In fact, no enrichment was found when we compared the δ18O of the terrarium free water and the terrarium leaf water (data not shown). However, this result might originate from measuring total leaf water, which includes depleted vein water. Even if we assume that the enrichment at the site of photosynthesis was as high as 1‰, our main conclusions will remain the same. The relative rate of the AOX in interval “Low” will be 19–29%, still a very high figure, and the correction to θP (see section 5.6) will be much smaller than the other uncertainties.

[59] The strong measured fractionation indicates that the AOX was activated in the same conditions that favor high rate of photorespiration-illumination and low CO2. This finding is in agreement with the indication for high AOX rates in the light inferred from in situ measurements in a lake [Luz et al., 2002]. The CO2 concentration in the terrarium were very low (150 ppm), much lower than in most natural environments. However, since the relative humidity in terrarium was 100% stomatal conductance must have been high. Consequently, the internal CO2 concentration in the leaves was similar to that of midday in many natural environments. Since strong fractionation occurred not only with the 24 h d−1 illumination but also with the 10 h d−1 illumination, which is closer to the natural cycle, we conclude that the engagement of the AOX in the light is likely also in many natural systems.

[60] In previous models of the Dole Effect, the global rate of the AOX was assumed to be very low and was neglected [Bender et al., 1994; Malaize et al., 1999]. This low rate is based on measurements of the AOX activity in the dark. However, if the AOX activation is enhanced in illuminated leaves in natural systems, then its global rate should be considerably higher. This higher rate may help to close the gap between the calculated value of the Dole Effect (20.8‰ [Bender et al., 1994]) and the measured one (23.5‰ [Kroopnick and Craig, 1972]), and compensate for the weak fractionation recently reported for soil respiration [Angert et al., 2001]. The connection between photorespiration and AOX might also explain past changes in the Dole Effect. Increased rate of photorespiration will be coupled with an increased rate of AOX. Drier and hotter climate is expected to cause an increased rate of photorespiration, as well as more evapotranspiration that will result in 18O enriched leaf water. Thus, such climate will cause an increased Dole Effect by both heavier composition of leaf water and increased rate of photorespiration and the AOX, two processes that have high fractionation relative to that of COX.

5.6. Estimating the θ of Photorespiration

[61] The average θ in the terrarium in interval “Low” was 0.511. This value and the values found for the AOX and COX, can be used to estimate the θ associated with photorespiration by assuming production uptake steady state, and a simple weighted average equation,

equation image
equation image

where GP is gross production in the terrarium, U stands for uptake, and the subscript P stands for photorespiration. In order to solve equation (20) for θP the rates of the different uptake pathways in the terrarium relative to gross production must be estimated.

[62] The gross oxygen production in the terrarium, at interval “Low,” can be estimated from previous experiments in the same terrarium [Luz et al., 1999]. In one of these experiments the δ18O of the water was −6.6‰ (on SMOW scale). As a result, the δ18O value in steady state was very different from the initial atmospheric value. At the beginning of the experiment (days 1–37, illumination of 24 h d−1, as at interval “Low”), changes in the δ18O and δO2/Ar of the terrarium were observed, when it approached steady state. The changes in the δ18O and δO2/Ar were modeled by the numerical model described in section 5.4. The best fit to the observed data was reached with uptake to gross production ratio of 0.99 and gross production rate (GP) that is 19.2% of the terrarium O2 reservoir per day.

[63] The uptake rate measured in the dark periods was 45% of the GP estimated above, and was probably mostly through the COX. When the terrarium was illuminated, considerable portion of the total dark respiration must have been through the AOX (section 5.5.); hence, the rate of COX in interval “Low” was lower than 45% of GP. If we assume a COX of 10–30% of GP, the rate of photorespiration (UP) can be calculated from the known fractionations of the different pathways and the terrarium-fractionation (18ε) we evaluated for interval “Low” (−24‰) by equation (20) and a weighted average equation similar to equation (21),

equation image

[64] Using fractionation values of −30‰ for AOX [Ribas-Carbo et al., 2000], −18‰ for COX [Guy et al., 1989], and −21.8‰ for photorespiration UP was calculated as 29–59% of GP. The corresponding rate of AOX is 41–31% of GP. According to these rates and the θ values of the COX and the AOX, the θ associated with photorespiration is calculated as 0.506 ± 0.003. The error margin is based only on the uncertainty in the rate of the COX, and neglects the uncertainty in the fractionations of the different pathways. The uncertainty in the fractionation in the COX and photorespiration is small [Guy et al., 1989, 1993]; however the uncertainty in the fractionation of the AOX is larger, about ±2‰ for green tissue [Ribas-Carbo et al., 2000; Robinson et al., 1992]. Including this uncertainty the θ associated with photorespiration is 0.506 ± 0.005.

[65] In photorespiration, O2 is consumed by two enzymes, Rubisco and glycolate-oxidase. Hence, the θ value of photorespiration represents the weighted average of the two consumption processes. Further study is necessary in order to derive the θ values associated with each of the two enzymes.

[66] The θ value of 0.506 ± 0.005 we report here for photorespiration is considerably lower than that of both dark respiration pathways. In the next section, we will use this value to estimate the effect of photorespiration on the triple isotopic composition of the atmosphere.

5.7. Implication for the Triple Isotopic Composition of Atmospheric O2

[67] In order to estimate how a change in the global average θ of biological processes will affect the triple-isotope composition of the atmosphere, we can analyze it in a graphical approach presented in Figure 6, or with a rigorous analytical model. In this model, the troposphere (which is well mixed relatively to the lifetime of oxygen in it) is represented by one box, and O2 is exchanged between the biosphere, the troposphere, and the stratosphere. In steady state we can write the following equation:

equation image

where yS is the stratosphere to troposphere flux and yT represent the flux from the troposphere to the stratosphere of any isotopic species (yO16O, y = 16,17,18). Flux yP represents global photosynthetic production and yU is global uptake of any isotopic species. In the model, the transfer of O2 to the stratosphere involves no fractionation and O2 reentering the troposphere is fractionated by xαs (as in the work of Blunier et al. [2002]). The relationship between 18αs and 17αs is given by θs. The parameter xαs describes the sum of stratospheric effects on O2 isotopes, and does not represent any single physical process. The global average isotopic ratio of oxygen produced from water by photosynthesis will be noted as xRAW. The isotopic composition of leaf water (which is the substrate for photosynthesis on land) is controlled by the hydrological cycle and evapotranspiration. However, since information on the different θ's in the hydrological cycle is scarce [Meijer and Li, 1998; Miller, 2002] and since there is no information about the θ of evapotranspiration, we will assume an average isotopic ratio for all the oxygen produced by photosynthesis and will not deal directly with differences in the substrate water. This average isotopic ratio is assumed to be constant in time. The average global fractionation in biological uptake is given by xαb and θb. We also assume 16T = 16S and 16P = 16U. Including these formulations in equation (23) gives

equation image

where x can be 17 or 18, and the subscript “T” stands for tropospheric oxygen.

Figure 6.

Analysis of triple isotopic composition (not to scale) of the Last Glacial Maximum (LGM). The point that represents the LGM atmosphere (after correction for lower rate of stratospheric photochemistry) lies 12 per meg lower than the present atmosphere. A change in the global average θ (θb) will affect the triple isotopic composition of the atmosphere in a way that is similar to the way a change in θ affected the terrarium air (Figure 5). Hence, a change in θb will cause a change in the atmospheric 17Δ with a magnitude of the Ln18O difference between atmospheric O2 and photosynthetic oxygen times the difference in θb. Thus, a change of 0.001 in θb will result in a change of about 19 per meg in the atmospheric 17Δ. The higher global rates of photorespiration (PR) in the LGM (27% of global production instead of 24% today) caused lowering of θb that can explain 5 per meg of the 12 per meg difference. The rest of this difference can be attributed to lower global gross production.

[68] According to equation (1), the 17Δ of tropospheric O2 relative to oxygen produced by photosynthesis is

equation image

[69] The value of 17ΔT-AW is always negative, since tropospheric O2 is depleted in 17O. Equations (24) and (25) give

equation image

[70] The average global θ associated the biological uptake will be noted as θb. The difference in the tropospheric 17Δ resulting from a change of the global average θb from θb1 to θb2 is given by substituting equation (26) with equation (5),

equation image

5.7.1. Implications for the 17Δ in Ice Cores

[71] The new θ values reported here may change the interpretation of 17Δ measured in ice cores. To demonstrate this, we will use the Last Glacial Maximum (LGM) as a case study. According to Blunier et al. [2002], the ratio of photorespiration to terrestrial production was 43% in the LGM relative to 38% at the modern biosphere. This difference is due to relatively low atmospheric CO2 concentration in the LGM (since the residence time of O2 in the atmosphere is ∼1200 years, the preindustrial CO2 levels are used for present), which was partially compensated by an increase in the relative rate of photosynthesis in C4 plants. We can apply equation (27) or Figure 6 to estimate the effect of this different photorespiration rate on the LGM's 17ΔT-AW. By assuming that the ratio of terrestrial production to marine production is 1.7 [Blunier et al., 2002] and that no photorespiration takes place in the oceans, we estimate that the ratio of photorespiration to global production is 24% at present and was 27% in the LGM. Using the θ values for dark respiration and photorespiration we are reporting here, and assuming that 10% of dark respiration is through the AOX, we calculated that the θb in the LGM was lower by about 0.00025 from the present value. Assuming the AW is 40% seawater and 60% leaf water and using an average Ln18O value of −23‰ versus the atmosphere for the former and +6‰ versus SMOW for the later [Gillon and Yakir, 2001], we calculated the average global fractionation (18εb) as −19.4 (0.4 × [−23] + 0.6 × [−23 + 6] = −19.4). In addition, we used a 16P/16S ratio of 0.0097 [Luz et al., 1999].

[72] Using equation (27), with the above parameters we calculated that the effect of higher photorespiration rates in the LGM is expected to cause the 17Δ of the LGM troposphere to be lower by 5 ± 2 per meg relative to the present troposphere (the ±2 per meg error margin is derived from the uncertainty in the θ values). This result agrees with the simple graphical solution presented at Figure 6 (0.00025 × 19.4 = 5 per meg). This 5 per meg estimate is insensitive to most of the assumptions in the calculation, except the assumption on the variations in the global rate of photorespiration. Hence, this calculation gives the order of magnitude of the variation in 17Δ that are caused by variations in global photorespiration. To estimate whether or not this 5 per meg signal is significant, it should be compared to the 17Δ signal generated by changes in global productivity.

[73] The Δ17O of the LGM troposphere relative to the present troposphere was measured in air from ice cores and was found to be +38 per meg [Blunier et al., 2002; Luz et al., 1999]. This corresponds to a 17Δ value of +43 per meg (with C of 0.516). This value is affected by a lower rate of mass-independent stratospheric processes that resulted from lower CO2 concentration in the LGM atmosphere. This lower rate resulted in 17Δ of the LGM troposphere higher by 55 per meg than the present value (under the assumptions of linear dependence of the stratospheric processes rate on CO2 concentration and constant ozone concentration [Luz et al., 1999]. There is a −12 per meg difference between the +55 per meg expected from the lower rate of stratospheric processes and the +43 per meg actually measured. This difference was considered in previous papers [Blunier et al., 2002; Luz et al., 1999] to indicate lower global productivity in the LGM. However, as shown above, about half of this −12 per meg change can be explained as the result of higher photorespiration rate in the LGM. In conclusion, small changes in the global rate of photorespiration can have an important impact on the atmospheric 17Δ. Consequently, any interpretation of triple-isotope ice core data must take this impact into account.

5.7.2. Global 17Δ Budget in the Present Atmosphere

[74] To gain some insight on the global 17Δ budget, we estimate the parameters of equation (26) that represents the effects of the biosphere and the stratosphere on the triple isotopic composition. The current value of θb was calculated, as in the previous section, from the derived θ values and the estimated global relative rates of photorespiration (24%), COX (68%) and AOX (8%), as 0.513 ± 0.002 (0.24 × 0.506 + 0.68 × 0.516 + 0.08 × 0.514 = 0.513, the error margin is derived from the uncertainty in the θ values). According to Luz et al. [1999] the stratospheric mass-independent processes result in lowering of the current atmospheric Δ17O in 117 ± 13 per meg. In the terms of equation (26) (with 16P/16S ratio of 0.0097 [Luz et al., 1999] and θs = 1) this 117 ± 13 per meg lowering corresponds to xαs of 0.99999765 ± 0.0000001 assuming 17Δ ≅ Δ17O and 18εb that was calculated in section 5.7.1 as 19.4‰. Using these parameters, we calculated by equation (26) that 17ΔT-AW (the troposphere versus AW) is −175 ± 50 per meg (the error margin is based on the error in θb and xαs). Correspondingly, 17ΔAW (the average substrate water for photosynthesis versus the troposphere) is 175 ± 50 per meg. This calculation can also be preformed graphically on Figure 7. The calculated 17ΔAW value is smaller than either 215 per meg obtained for Lake Kinneret water (section 5.3) or 249 per meg for seawater [Luz and Barkan, 2000]. The difference between the calculated 17ΔAW and value obtained for seawater can be solved if higher global rates of photorespiration will be assumed in the calculation. However, this will not explain the difference between the values obtained for Kinneret water and seawater. Both this difference, and the difference between seawater and the calculated 17ΔAW, can be explained by various isotope effects in the hydrological cycle.

Figure 7.

Schematic illustration (not to scale) of suggested triple isotope budget in the present global atmosphere. The global composition of oxygen produced by photosynthesis (AW) is a weighted average of about 60% O2 produced from leaf water and 40% from seawater. Average leaf water is formed from seawater that is depleted in 17O and 18O during meteoric precipitation (θ ∼ 0.525) and enriched by evapotranspiration (θ < 0.525). AW oxygen is enriched due to biological uptake (θb = 0.513) to form BSS oxygen. BSS signifies the hypothetical composition of the atmosphere that would have been produced in the absence of stratospheric photochemistry. Further mass independent depletion (θ ≥ 1) by stratospheric photochemistry causes 117 per meg lowering of atmospheric 17Δ.

[75] Reported λ values for meteoric precipitation range from 0.528 [Meijer and Li, 1998] to 0.525 [Miller, 2002]. Since θ is only slightly lower than λ (∼0.002), it is clear that the corresponding θ values are higher than the ones we determined for biological uptake (0.506–0.516). The source of Lake Kinneret water is depleted meteoric precipitation that later becomes enriched by evaporation. Hence, a relatively low θ associated with evaporation can explain why the 17ΔW of lake Kinneret is lower than that of seawater [Luz and Barkan, 2000]. Similar explanation can show how 17ΔAW becomes 74 per meg smaller than 17ΔW of seawater. If the θ of evapotranspiration is also low, then the depletion in the heavy isotopes of meteoric precipitation and the enrichment in evapotranspiration will yield a low 17Δ of average leaf water. Such low 17Δ of leaf water will contribute to a low 17ΔAW (175 per meg, Figure 7).

[76] The discussion above demonstrates that rigorous closure of the triple isotopic balance requires direct measurements of the θ values associated with the hydrological cycle. Such estimation will enable independent estimate of the isotopic composition of average water (17ΔAW). The average global biological θ (θb) could be calculated from 17ΔAW and the measured triple isotopic composition of atmospheric O2. Our study shows that the relative global rates of dark respiration and photorespiration control θb. Hence, the rates of these globally important processes could be evaluated from triple isotope studies.

6. Conclusions

[77] The θ values for the cytochrome pathway, the alternative pathway, photorespiration, and diffusion in air are 0.516 ± 0.001, 0.514 ± 0.001, 0.506 ± 0.005, and 0.521 ± 0.001, respectively. The combined effect of diffusion and respiration on the atmosphere was shown to be close to that of dark respiration. Since the value associated with photorespiration was found to be considerably lower than that of dark respiration, the triple isotopic composition of atmospheric O2 is strongly controlled by the relative global rates of these two processes. This control makes the triple isotopic composition a tracer of the global rate of photorespiration in the past, as well as in the present.

[78] The engagement of the alternative oxidase pathway (AOX) in the light is larger in low CO2 than in high CO2 concentrations. Thus, environmental conditions that lower the internal CO2 concentration in leaves are expected to cause an increase in photorespiration rate as well as in the AOX rate. Since the fractionation in both processes is strong, such environmental conditions will cause an increased Dole Effect.


[79] We thank Michael Bender, Thomas Blunier, Dan Yakir, and Joe Berry for valuable discussion, and Y. Helman for providing unpublished data. The comments of two anonymous reviewers significantly improved the manuscript. We greatly appreciate support from the Israel Science Foundation, the USA-Israel Binational Science Foundation and Moshe-Shilo Minerva Center.