letters to nature J. Chem. Phys. 108, 5888±5897 (1998). 26. MacGorman, D. R. & Rust, W. D. The Electrical Nature of Storms 32 (Oxford Univ. Press, New York, a 200 1998). 0 27. Pruppacher, H. R. & Klett, J. D. Microphysics of Clouds and Precipitation 813±814 (Kluwer Academic, Dordrecht, 1996). –200 28. Di Palma, T. M., Latini, A., Satta, M. & Giardini Guidoni, A. Molecular beam studies of ammonia –400 clustered with metals produced by pulsed laser reactive ablation. Int. J. Mass Spectrom. 179/180, 319± /Ar vs HLA 326 (1998). 2 –600 O δ –800 Acknowledgements. We thank R. D. Levine for discussions, and H.-J. Schmidtke for contributions to the early stage of this work. b 5 Correspondence and requests for material should be addressed to H.S. (e-mail: [email protected]). 0 –5 –10 O vs HLA Triple-isotope composition 18 –15 δ of atmospheric oxygen –20 –25 as a tracer of c 250 biosphere productivity 200 150 Boaz Luz*, Eugeni Barkan*, Michael L. Bender², O vs HLA 100 Mark H. Thiemens³ & Kristie A. Boering§ 7 17 50 * The Institute of Earth Sciences, The Hebrew University of Jerusalem, ∆ Jerusalem 91904, Israel 0 ² Department of Geosciences, Princeton University, Princeton, New Jersey 08544, 0 50 100 150 200 USA Day ³ Department of Chemistry, University of California, San Diego, La Jolla, California 92093, USA Figure 1 Removal of the atmospheric 17O anomaly by biological cycling. Shown 18 17 § Departments of Chemistry and of Geology and Geophysics, are variations of dO2/Ar (a) d O(b) and D O(c) in the terrarium experiment. Data University of California, Berkeley, California 94720-1460, USA points: diamonds, terrarium PK; circles, terrarium PDS. Horizontal scale: days ......................................................................................................................... from the beginning of the experiment. Light: continuous illumination, days 1±42; Oxygen has three naturally occurring isotopes, of mass numbers room light, days 43±90; 10 h light,14 h dark, days 91±136 and days 142±198; dark, 16, 17 and 18. Their ratio in atmospheric O2 depends primarily on days 137±141. the isotopic composition of photosynthetically produced O2 from terrestrial and aquatic plants1±3, and on isotopic fractionation due respiration, thereby inducing a wide range of d18O and d17O values. 4 to respiration . These processes fractionate isotopes in a mass- After several turnovers, anomalous ambient air O2 had been 17 dependent way, such that O enrichment would be approximately removed by respiration and replaced with normally fractionated O2, half of the 18O enrichment relative to 16O. But some photochemi- and steady state was attained. Subsequent illumination changes cal reactions in the stratosphere give rise to a mass-independent signi®cantly affected O2 concentration (measured as dO2/Ar; see isotope fractionation, producing approximately equal 17O and 18O Methods) and d18O but not the 17O anomaly (Fig. 1). The d18O enrichments in stratospheric ozone5 and carbon dioxide6,7, and versus d17O trend for data points for which the anomaly was at 2 consequently driving an atmospheric O2 isotope anomaly. Here steady state plot on a nearly perfect straight line (R 0:99999) we present an experimentally based estimate of the size of the with a slope of 0.5211 (60.0005), as expected for mass-dependent 17 16 8 O/ O anomaly in tropospheric O2, and argue that it largely fractionation (data not shown). The intercept of the regression line re¯ects the in¯uences of biospheric cycling and stratospheric is 0:155 6 0:008½ or 155 6 8 in units of per meg (see Methods). photochemical processes. We propose that because the biosphere Based on this analysis, we de®ned the D17O anomaly as the 17 removes the isotopically anomalous stratosphere-derived O2 by deviation from normal mass-dependent fractionation (D O respiration, and replaces it with isotopically `normal' oxygen by d17O 2 0:521d18O); in the case of the terrarium experiment, D17Ois photosynthesis, the magnitude of the tropospheric 17O anomaly equal to the intercept value of the regression line. HLA was the can be used as a tracer of global biosphere production. We use preferred standard for high-precision measurements in our study. 17 measurements of the triple-isotope composition of O2 trapped in However, as a reference for D O it is admittedly confusing, because bubbles in polar ice to estimate global biosphere productivity at it has anomalous isotopic composition. An air sample with no 17 various times over the past 82,000 years. In a second application, photosynthetic O2 added will have a D O 0 with respect to HLA, we use the isotopic signature of oxygen dissolved in aquatic and a sample of biologically equilibrated O2 will have a systems to estimate gross primary production on broad time D17O 155 per meg with respect to HLA. In this treatment, and space scales. ocean and meteoric waters are de®ned as normal with D17O 155 The magnitude of the 17O anomaly in the present atmosphere can per meg. A comparison of the isotopic composition of ocean water 17 18 18 be estimated by comparing d O and d O of ambient air O2 (represented by V-SMOW) to air O2 (d O 2 22:960½ and 17 6 (represented by the HLA standard; see Methods) with O2 that was d O 2 11:778½) , supports our conclusions that air bears a not affected by stratospheric processes. To make the latter, we built mass-independent signature. The D17O of V-SMOW is calculated as 17 two airtight terrariums in which O2 was consumed and replaced 184 per meg D O 2 11:778 2 0:521 3 22:96 1;000. In this biologically. Ultraviolet radiation that could lead to mass-indepen- calculation, the derived anomaly is very sensitive to the slope term dent fractionation was absent. The terrariums contained (0.521), and thus the small difference between 184 and 155 per meg Philodendron plants, soil and natural water. In both terrariums, may not be signi®cant. O2 production and consumption occurred via the higher plants as Importantly, the terrarium experiment cannot represent all well as via bacteria and algae. Illumination was changed during the Earth-surface processes affecting D17O. For example, the range of experiment with the aim of varying the ratio of photosynthesis to isotopic variations in global meteoric waters is much greater than in NATURE | VOL 400 | 5 AUGUST 1999 | www.nature.com © 1999 Macmillan Magazines Ltd 547 letters to nature 9 the experiment, and a recent study indicates that in these waters the ozone and CO2 species could result in anomalous O2. In the 17 18 d O/d O slope is slightly different than in respiration. Further- stratosphere, the ozone recombination reaction, O O2 ! O3, more, the humidity in the terrariums was at saturation and causes O3 to be mass-independently fractionated. The anomalous important isotope fractionation due to evapo-transpiration from fractionation is well documented, although its cause is debated11±13. leaves2,3 was not re¯ected in our experiment. In addition, the Theory14 and laboratory experiments15 suggest that the anomalous 17 18 d O/d O slopes in the various processes consuming oxygen (dark ozone enrichment is transferred to CO2. Ultraviolet photolysis of respiration, cyanide resistant respiration, photorespiration and the ozone in the stratosphere generates an electronically excited oxygen Mehler reaction) may vary slightly from 0.521. Because the relative atom which can undergo isotope exchange with CO2: rates of these processes in our experiment are not expected to be the O hn ! O 1DO same as in the global biosphere, the intercept of the regression line in 3 2 natural systems may differ from the value we measured. Thus, while O 1DCO ! COp the 155 per meg estimate clearly demonstrates the anomalous 2 3 isotopic signature of atmospheric O , further study is needed in 2 COp ! CO O 3P order to better constrain the magnitude of the anomaly. 3 2 1 Our experimental determination of the mass-independent Thus, because the ultimate source of the oxygen in O3 and O( D) in anomaly in O2 can be compared with estimates of its stratospheric the stratosphere is the O2 reservoir, O2 becomes anomalously 10 production. Bender et al. ®rst suggested that photochemical mass- depleted as CO2 becomes anomalously enriched. We note that independent fractionation processes in the stratosphere involving there is no stratospheric loss term for the enrichment in CO2, and the net stratospheric enrichments are lost only at the Earth's surface by isotope exchange with liquid water in leaves and in the ocean3,14,16 17 (Fig. 2). In contrast, O2 does not exchange isotopes with water , and the depletion disappears only through the consumption of O2 by respiration and its replacement by photosynthesis. The respiratory and photosynthetic ¯uxes are relatively small (Fig. 2) compared with the stratospheric production, and thus the D17O anomaly accumulates to a measurable level of 155 per meg over the residence time of atmospheric O2 (,1,200 yr; ref. 10). To test the hypothesis that stratospheric processes can generate the 155 per meg anomaly in atmospheric O2, we estimate the production rate of anomalous O2 from stratospheric photochem- istry. We use observations of mass-independently fractionated CO2 and its correlation with N2O (a long-lived tracer that is photolysed in the stratosphere), coupled with calculations of the annual mass ¯ux of air from the stratosphere to the troposphere. We frame the calculations in terms of N2O for two reasons. First, observations of D17OofCO (D17O ) are extremely sparse, so their tight correla- 2 CO2 tion with N2OÐa species that has been extensively measured in the stratosphereÐserves as a proxy for the distribution of D17O in CO2 the stratosphere.