<<

REVIEWS OF GEOPHYSICS, SUPPLEMENT, PAGES 1253-1262, JULY 1995 U.S. NATIONAL REPORT TO INTERNATIONAL UNION OF GEODESY AND GEOPHYSICS 1991-1994

The atmospheric cycle: The oxygen of atmospheric C02 and 02 and the 02/N2 ratio

Ralph F. Keeling Scripps Institution of Oceanography, University of California, San Diego, La Jolla

Introduction to come to isotopic equilibrium with liquid is the same as the time scale for hydration, i.e., around 30 sec­ 18 16 Oxygen is the most abundant element in the earth's onds [Mills and Urey, 1940]. The 0/ 0 ratio of C02 , it accounts for 89% of the mass of the , and in equilibrium with water at 25° C is 1.041 times higher 18 16 it is the most abundant element in the earth's than the 0/ 0 ratio of the water. This equilibrium atmosphere. Much work on the oxygen cycle has fo­ fractionation factor varies slightly with temperature. Isotopic ratios are generally reported according to cused on the question of the origin of atmospheric 02 and its variations over geologic time [see Kump et al, 18 16 18 16 1991, and references therein). This review focuses on S = ( 0/ 0)sampie/( 0/ 0),Wflrd " 1 (1) several other aspects of the oxygen cycle including the short-term controls on the oxygen isotopic abundance where the S value is customarily multiplied by 1000 and expressed in per mille (%<>)• Bottinga and Craig [1969] of atmospheric C02 and 02, and the short-term vari- abilitiy in the 0 /N ratio. suggested using a standard based on CO2 in equilib­ 2 2 rium with the Standard Mean Ocean Water (SMOW) at These aspects of the oxygen cycle depend mainly on 25°C [Craig and Gordon, 1965]. Most recent measure­ material exchanges between the atmosphere and living ments have been reported relative to the 180/160 ra­ organisms at the earth's surface or in the ocean. Like tio of C0 derived from the Pee-Dee Belemnite (PDB) several other atmospheric variables which have received 2 carbonate standard. This standard has an 180/160 ra­ much attention recently, e.g., the abundances of C0 , 2 tio that is 0.22%o higher than the Bottinga and Craig CH4, and N 0, the oxygen isotopic content of C0 and 2 2 standard [Friedman and O'Neil, 1977]. 02 and the 02/N2 ratio have atmospheric lifetimes that 18 16 are long relative to the time scale of atmospheric mix­ The 0/ 0 ratio of atmospheric CO2 is primarily ing and thus reflect an integration of material exchanges determined by exchanges with leaf water, soil water, over the globe. Recently, our knowledge of these vari­ and surface sea water [Francey and Tans, 1987; Far- ables has expanded through laboratory experiments ex­ quhar et al., 1993]. Oxygen exchange with leaf water occurs because a significant fraction of the C0 ploring the exchange pathways, and through measure­ 2 ments on contemporary air samples or in ancient air which diffuses into the chloroplasts of leaf cells is not samples extracted from polar ice cores. This review assimilated but diffuses back into the air, and this frac­ summarizes recent literature on these subjects, and also tion will have equilibrated isotopically with chloroplast emphasizes how these aspects of the global oxygen cycle water. Equilibration occurs in spite of the short (< 1 second) residence time of C0 in leaves because of the can provide new information on the material exchanges 2 between the atmosphere and biota integrated over large presence of the enzyme carbonic anhydrase, which is areas. concentrated in the chloroplasts of leaf cells and which dramatically speeds up the hydration reaction. Oxygen atom exchange with soil water occurs primarily through The 180/160 Ratio of Atmospheric C0 2 C02 which is released into the soil by below-ground res­ piration and which subsequently diffuses into the atmo­

The oxygen isotopic content of atmospheric C02 is sphere. Oxygen atom exchange with occurs mainly determined by interactions between C02 and through the exchange of C02 across the air- the global reservoirs of liquid water. This follows be­ sea interface. cause direct gas phase interactions of C02 with 02 and The oxygen isotopic composition of soil water and

H20 vapor do not result in O atom exchange [Francey leaf water vary considerably. Soil water isotopic com­ and Tans, 1987]. When C02 dissolves in water, oxy­ position tends to follow the composition of precipitation gen are exchanged through a mechanism that which is progressively depleted in 180 relative to seawa­ involves the hydration of dissolved C02 to form car­ ter towards high latitudes and towards the-interior of bonic acid (H2C03). The time scale for dissolved C02 continents. Chloroplast water, in turn, tends to be en­ riched in 180 relative to soil water by evaporation from leaves because H2160 evaporates preferentially relative Copyright 1995 by the American Geophysical Union. to H2180. This enrichment of chloroplast water is sensi­ tive to relative humidity and temperature, which can be Paper number 95RG00438. highly variable [Dongmann et al., 1974; Forstel, 1978; 8755-1209/95/95RG-00438$15.00 Zundel et al, 1978].

1253 1254 KEELING: ATMOSPHERIC OXYGEN CYCLE

A global steady-state budget for 6lsO of atmospheric is where does chloroplast water fall in this range. Far­ CO2 is shown in Figure 1. This budget uses figures quhar et al. [1993] present results based on ex­ from Farquhar et al [1993] for fluxes and isotopic ex­ change experiments with several varieties of fruit trees changes of atmospheric CO2 with leaf, soil, and sea wa­ that suggest the isotopic composition of chloroplast wa­ ter. One significant source of uncertainty here is the ter is virtually identical to that of water at the evap­ global average isotopic composition of chloroplast wa­ orating surfaces in leaves. The budget in Figure 1 is ter. Logically, the 6lsO of chloroplast water should be based on this assumption, taking into account the vari­ intermediate between that of soil water and water at ability of leaf water over the surface of the earth. In the evaporating surface in the leaves where the max­ contrast, Yakir and coworkers have conducted isotope imum isotopic enrichment occurs. A critical question exchange experiments on sunflowers that indicate that

Figure 1. The global "pre-anthropogenic" steady-state budget for the oxygen isotopes of atmo­ spheric CO2 based on Farquhar et al. [1993] showing annual fluxes of CO2 in units of 1015 moles of

and showing the isotopic composition of C02 in equilibrium with dominant exchangeable water reservoirs [see also Keeling, 1993]. CO2 exchange with soil water involves uptake of CO2

by leaves, respiration within the soil, and diffusion of the respiratory C02 out through the soil. The budget shown here assumes that the kinetic isotope fractionation that results from diffusion through stomata and through the soil cancel each other out (see also Table 2, Eq. (F)). According to this budget, the bulk composition of atmospheric CO2 can be explained by assuming that 45% of the oxygen atoms come from chloroplast water at an average isotopic composition of -f5%o> 34% come from soil water at an average of — 7%o, and 21% come from sea water at an average of l%o- This combination yields atmospheric CO2 at approximately 0%o- All numbers here are relative to the PDB standard. KEELING: ATMOSPHERIC OXYGEN CYCLE 1255 chloroplast water is typically 6 to 10%o depleted in 180 Table 1. Summarizing Two Alternative Formulations for compared to water at the evaporating surface [Yakir et Describing Exchanges of Oxygen Isotopes of C02 with al, 1993; Yakir et al, 1994]. The difference in 6180 be­ Terrestrial Ecosystems tween chloroplasts and evaporation sites probably varies NOTATION: significantly from species to species [Yakir et al, 1993]. C0 Atmospheric CO2 partial pressure (PCO2)

A global model describing oxygen atom exchanges of Cc PCO2 in chloroplast

18 18 ie CO2 with terrestrial ecosystems has been developed by Ca Atmospheric C O O partial pressure Farquhar et al [1993] (see Table 1, Equation G). This 18 16 Ra 0/ 0 ratio of atmospheric C02 model is based on a formulation in which the oxygen- 18 16 Rc 0/ 0 ratio of C02 in equilibrium with chloro­ atom exchanges with leaf water are described using an plast water effective fractionation factor A a (see Table 1) against 18 16 R0 0/ 0 ratio of C02 in equilibrium with soil 180 on net uptake of CO2. The isotopic exchange flux water between the atmosphere and leaves is thus obtained by Rpdb 180/160 ratio of CO2 derived from the carbonate multiplying Aa by net flux of CO2 into the leaves (basi­ standard Pee-Dee Belemnite [see Friedman and cally equal to gross , GPP). The fac­ O'Neil, 1977] tor Aa is not a true fractionation factor because it de­ 6a (Ra-RPDB)/RPDB pends on the isotopic composition of atmospheric CO2. 6c (Rc—Rpdb)/Rpdb

Aa is nevertheless useful because it can be measured 63 (R«—Rpdb)/Rpdb

in controlled experiments as well as modeled over the 6r 68 + e80u surface of the earth [Farquhar et al, 1993]. Fin Gross flux of CO2 into stomata The latitudinal distribution of Aa as estimated by Fout Gross flux of CO2 out of stomata Farquhar et al [1993], is shown in Figure 2. Also shown R Flux of CO2 out of soil from root and soil respiration is the latitudinal variation of C02 in equilibrium with surface seawater (6°), the isotopic composition of CO2 ttsto Fractionation factor of diffusion through stomata (same both directions) returned to the atmosphere through soils (6r), the sum

1S or8oii Fractionation factor for diffusion out of soil 6r + AA, and the annual mean surface values of S 0 —5 er Csto Qfsto ~ 1 [ P Farquhar et al., 1993] of atmospheric CO2. Exchanges with leaves and soils tend to drive the local S180 of atmospheric CO2 tends Csoil Of soil — 1

A Gross primary production (A = Fin— Fout) towards the sum 6r + AA. This sum tends to decrease 18 16 towards high latitudes in the northern hemisphere, fol­ A a Effective discrimination against C 0 0 relative lowing the depletion of 180/160 of precipitation. The to net CO2 assimilation [Farquhar et al, 1993];

AA = -e.to + [Ce/(C«- CC)](6C - 6a) latitudinal gradient in 6r-h Aa can account qualitatively for the latitudinal gradient in 180/160 of CO2 that was M Number of moles of air in the atmosphere observed by Francey and Tans [1987], although the ac­ N Net ecosystem production tual profile in the air is smoothed by atmospheric mix­ (N = Fi„- Fout- R) ing. EQUATIONS: In addition to exhibiting a gradient with latitude, the Mass balance of atmospheric CO2: 18 16 M dCa/dt = Fout~ Fin 4" R (A) 0/ 0 ratio of C02 is known to undergo a seasonal cycle in the northern hemisphere [Keeling, 1961; Friedli Mass balance of atmospheric C180160:

et al, 1987] with a maximum in early summer and a M d(RaCa)/dt = <*8to Rc Fout - Qfsto Ro

minimum in early winter. This seasonal cycle prob­ Fin + c*80ii R, R (B) ably results mostly from the seasonality of exchanges Flux/gradient proportionality:

with terrestrial ecosystems. These exchanges will tend Fout/(Fin- Fout) = Ce/(C- Cc) (C) to cause a decrease in 180/160 of CO2 during the Combining (A) and (B): warmer months when atmospheric CO2 exchanges most M Ca dRa/dt = a«to(Rc- Ra)Fout- Ro rapidly with leaf and soil reservoirs which are depleted (t*8to " l)(Fin- Fout) + ("soil Ra) R (D) 18 16 in 0/ 0 at middle and high northern latitudes. The Rearranging (D): ratio will tend to increase during other seasons as a M Ca dRa/dt = a8to(Rc- Ro) Fout + ("soil result of transport of higher 180/160 ratios from more R, - C* to Ro)R + RoKto - 1)(R - Fin southern latitudes. Other factors, such as seasonal vari­ 8 + Fout) (E) ations in soil water 180/160 ratios, in leaf water isotopic FORMULATION 1 (results from converting (D) enrichment, and in the ratio of CO2 exchange rate with 1 to 6-notation): leaves and soils probably also play a role. Modeling C M C d6 /dt = (6 - $a)Fout + (*• + ">il this seasonal cycle remains an important area of future a a C - e .) R -e to N (F) work. (t 8 FORMULATION 2 (combining (C) and (D), and 18 16 What can we learn from measurements of 0/ 0 converting to 6-notation):1 ratios of CO2? Farquhar et al [1993] propose using the M Ca d6a/dt = AAA + («r-6a) R (G) measurements to distinguish between CO 2 exchanges with different biomes and between terrestrial ecosys­ 1 See Tans et al, 1993, for example of approximations tems and the . This application is suggested be- used. 1256 KEELING: ATMOSPHERIC OXYGEN CYCLE

information on rates of gross primary production and stomatal conductance, as these exchanges can produce large variations in <$180 without producing variations atmospheric CO2 concentration or carbon isotopes of

C02. At present, several research programs are engaged in measuring the 180/160 ratio of atmospheric CO2. These measurements will be useful for validating global- scale numerical models including physiologically based exchanges of H2O and CO2 with leaves and soils. The -10 180/160 measurements can be expected to provide in­ formation on rates of gross primary production and stomatal conductance integrated over large spatial scales -20 L and in variations in these quantities in response to climate change, increasing atmospheric CO2, or other global variables. -30 -80 -60 -40 -20 0 20

Latitude (°) 18 16 The 0/ 0 Ratio of Atmospheric 02: Figure 2. Latitudinal averages of the effective dis­ The Dole Effect crimination factor on uptake of CO2 by leaves Aa, the ls 18 6 O of C0 emitted by soils (6 ), the <5 0 of C0 in 18 16 2 r 2 The 0/ 0 ratio of atmospheric O2 is higher than equilibrium with surface seawater 6q , and the observed 18 that of average seawater H2O by 23.5%o [Kroopnick atmospheric 6 0 of atmospheric CO2 (circles), and the and Craig, 1972]. This observation was first made in­ sum (6r + Aa), from Farquhar et al [1993]. Here 6r was computed from the isotopic composition of precip­ dependently by Dole [1935] and Morita [1935] and has become known as the Dole effect. It is caused mainly by itation minus 7.6%0- Latitudinal averages for Aa and 18

6r are weighted by GPP. 6r and So are expressed relative discrimination against O during respiration, as was re­ to the PDB standard. alized in early investigations [Lane and Dole, 1956]. By this reasoning, produces O2 from H2O with the same 180/160 ratio as the H2O, while res­ 16 cause of the large variability in the effective discrimina­ piration preferentially removes O2 from the air. A 18 16 tion factor A A between different biomes. steady-state balance is achieved when the 0/ 0 ra­ Further insight into what can be learned from mea­ tio of atmospheric O2 is enriched relative to photosyn­ surements of oxygen isotopes of CO2 can be obtained by thetic O2 by the discrimination factor associated with the formulation shown in Table 1 (Equation F) which respiration. divides the isotopic exchanges with terrestrial ecosys­ It is now recognized that additional processes also tems into three separate terms. The first term is pro­ contribute to the Dole effect. While careful investiga­ portional to the gross flux of C02 out of leaves which tions have confirmed that photosynthesis produces O2 is related to stomatal conductance. The second is pro­ without fractionation [Stevens et al, 1975; Guy et al, portional to the flux of CO2 out of soils which, for an 1993], the isotopic composition of photosynthetic wa­ ecosystem which is neither gaining or losing carbon, ter can vary, and these variations will be passed on to is closely related to ecosystem gross primary produc­ the 02 produced by photosynthesis. On average, this tion (GPP). The third is proportional to the net flux to an increased Dole effect because of evaporative of CO2 into the ecosystem, i.e., to net ecosystem pro­ enrichment of 180 in leaf water [Dongmann, 1974]. Ad­ duction (NEP). The first two terms depend on gross ditional processes which influence the Dole effect are the (i.e., two-way) exchanges of CO2 while the third term equilibrium fractionation of 180/160 between dissolved depends on the net (one-way) CO2 exchange. The third and gaseous O2, which is relevant because O2 consumed term is present because net CO2 uptake by the ecosys­ by respiration is derived from dissolved O2, and photo­ tem results in discrimination against 180 in CO2 as the chemical processes in the which to a CO2 diffuses into stomata. Under some circumstances, slight decrease in 618O of O2 through exchanges of oxy­ e.g., over a diurnal cycle, the instantaneous value of this gen atoms between O2 and CO2 [Bender et al, 1994a]. third term may be comparable in magnitude to the first Estimating the effective average fractionation factor two terms. On a time averaged basis, however, this term for global respiration is complicated because O2 con­ will be relatively unimportant because NEP is generally sumption can occur via several distinct biochemical at least an order of magnitude smaller than the flux of pathways [Guy et al, 1989; Guy et al, 1993; Bender

C02 out of soils or leaves. (Figure 1 was drawn assum­ et al, 1994a] including the light reactions, such as the ing this term is zero). What this means is that 6180 Mehler reactions and photorespiration reactions, and of CO2 is mainly sensitive to gross rather than net ex­ the dark reactions, such as the cytochrome pathway and changes [see also Yakir et al, 1993]. The unique value of the alternative cyanide-resistant pathway. To compute the SlsO measurements therefore lies in providing new the global respiratory contribution to the Dole effect it KEELING: ATMOSPHERIC OXYGEN CYCLE 1257 is necessary to know the fractionation factors and the A recent budget of the global contributions to the relative O2 consumption for each pathway at the global Dole effect by Bender et al [1994a] is presented in Fig­ scale [Berry, 1992; Bender et al, 1994a]. ure 3. This budget adopts the value of 4.4%o [Farquhar Respiration in the deep sea requires special consider­ et al, 1993] for the average enrichment of terrestrial ation because here respiratory O2 utilization depletes a chloroplast water relative to SMOW. The budget takes significant fraction of the O2 originally present in the account of respiratory fractionation using fractionation water. If total depletion occurred, then the effective factors from Guy et al [1989], Guy et al [1993], Kiddon fractionation for respiration in the deep sea would be et al [1993], and Bender [1990], and using estimates of zero because the 180/160 ratio of the removed O2 would the global O2 uptake on land and in the ocean from be equal to the 180/160 ratio of the O2 originally dis­ Farquhar et al [1980], Guy et al [1993], and Keeling solved in the water. In the case where O2 is only par­ and Shertz [1992]. tially depleted, the effective respiratory fractionation Interestingly, this budget yields an estimate for the factor can be calculated based on the percentage O2 global Dole effect of 20.8%o which is significantly smaller depletion that actually occurs [Bender et al, 1994a]. than the observed value of 23.5%o. The difference may

THE DOLE EFFECT STRATOSPHERE

8 = 8, |=8ATM-0.018%o c... TROPOSPHERE AVERAGE CHLOROPLAST ATMOSPHERIC 0 (3.7 x 1019 MOLES) WATER 2 18 18 8 0 4.4%o 8 0 OF ATMOSPHERIC 02 (8ATM)

8ATM = 23.5%o (OBSERVED) 20.8%o (CALCULATED)

ROSS MARINE PHOTOSYNTHESIS 818O = 0.0%o

MASS BALANCE FOR 8180 OF O2: ° ' =*?

20(4.4) + (12)(0.0) + (660)(8ATM - 0.018) =

20(8ATM - 18.8) + 11.4(8ATM + 0.7 - 20) + 0.6(8ATM + 0.7 -12.0) + 6608ATM CALCULATED GLOBAL DOLE EFFECT: 20.8%o

MASS BALANCE CONSIDERING ONLY TERRESTRIAL EXCHANGES:

20(4.4)= 20(8ATM-18.8) CALCULATED TERRESTRIAL DOLE EFFECT: 22.4%o

MASS BALANCE CONSIDERING ONLY OCEANIC EXCHANGES:

12(0.0) = 11.4(8ATM + 0.7 - 20) + 0.6(8ATM + 0.7 - 12.0) CALCULATED OCEANIC DOLE EFFECT: 18.9%o

Figure 3. The global steady-state budget for the oxygen isotopes of atmospheric O2 per Bender et al [1994]. Fluxes are in units of 1015 moles O2 yr"1. The <5180 values represent estimates of global averages of spatially and temporally variable quantities. Photorespiration and pho­

tooxidation reactions are grouped here as part of total terrestrial respiration. The 02 flux from leaves thus exceeds the net O2 production by leaves, i.e., the O2 production associated with gross primary production, by the amount required by balance photorespiration and photooxidation reactions. 1258 KEELING: ATMOSPHERIC OXYGEN CYCLE

either reflect errors in the values adopted or unknown core O2 and sediment records has made it possible to additional processes. A possible problem is the value establish more firmly the age of the air extracted from

of 4.4%0 adopted from Farquhar et al. [1993] for av­ ice cores relative to the sediment chronologies [Sowers erage chloroplast water. A value of 8.7%o would bring et al, 1991].

18 the budget into balance, and Bender et al. [1994a], ar­ Variations in 6 0 of atmospheric 02 must lag behind gue that a higher value is plausible given the uncertain­ variations of 6180 in surface seawater by the turnover ties involved. In this case, however, the Farquhar et al. time of atmospheric O2 with respect to gross photo­ 18 16 [1993] budget for 0/ 0 of C02 would be out of bal­ synthesis and respiration. If the sediment and ice core ance. One possible way of reconciling both the CO2 and chronologies were improved sufficiently, this turnover O2 isotope budgets might be by increasing the 6180 of time, currently estimated at 1500 years, could be di­ chloroplast water and decreasing the 618O of CO2 leav­ rectly determined [Bender et al., 1985; Bender et al., ing soils relative to the Farquhar et al. [1993] budget 1994a]. (M. Bender, personal communication). Some additional How variable is 6180 of atmospheric O2 on shorter flexibility may be provided by the fact that the O2 and time scales? Temporal and spatial surveys [Dole et al, CO2 budgets depend on different weighted averages of 1954; Kroopnick and Craig, 1972] showed that 6180 in chloroplast water. For the O2 budget, the average needs the present atmosphere is constant to at least 0.25%o- to be weighted by GPP plus photorespiration, while for More recent tropospheric measurements indicate that CO2 the average needs to be weighted by the flux of 6180 is constant to least 0.03%o (M. Theimens, per­

C02 out of stomata, which is equal to the gross flux of sonal communication). Known sources of variability are CO2 into stomata minus GPP. In any case more work expected produce changes about an a order of magni­ is needed to construct mutually consistent budgets for tude smaller than this. For example, we can expect 6*180 in both atmospheric O2 and CO2. 8180 to be lower in summer than in winter in both Bender et al. [1994a] estimate that the Dole effect northern and southern hemispheres by about 0.002%o- which would result from exchanges with the oceans This estimate is based on assuming that the 0.01% sea­ alone is around 2 to 3%o lower than that which would sonal increase in the atmospheric O2/N2 ratio (see next result from terrestrial exchanges alone (see Figure 3). section) is driven by the input of photosynthetic O2 This difference would be even larger if a 6180 value that is 20%o lower in 180 than atmospheric O2. Sea­ higher than 4.4%o is adopted for globally averaged chlo­ sonal variability might also be caused by seasonal phase roplast water. Either way, the overall magnitude of differences in gross photosynthesis in the oceans or on the Dole effect is sensitive to the ratio of gross primary land, or by seasonality in leaf water 6180. Detecting production on land to gross primary production in the such small changes may eventually be feasible with very oceans. This suggests that measurements of the Dole ef­ precise mass spectrometric measurements. fect and its variation over time may be used to constrain In summary, our knowledge of variations in

18 16 atmospheric O2 has closely followed the 0/ 0 ratio The atmospheric reservoirs of 02 and C02 are linked of surface seawater as established from sediment records by processes that involve the formation and destruction [Shackleton and Pisias, 1985], which in turn has varied of organic matter such as photosynthesis, respiration, due to the expansion and contraction of the continental and combustion. On times scales shorter than many ice sheets. The Dole effect, i.e., difference in 180/160 thousands of years, these organic oxidation-reduction ratio between atmospheric O2 and surface seawater, has reactions are the main source of variability in atmo­ been constant to around ±0.5%o over this period, with spheric 02 abundance. These reactions also produce possible small cyclic variations with a period of 23 thou­ and destroy C02, but the chemistry of atmospheric C02 sand years corresponding to the precession period of the is further complicated by reactions with seawater. In earth's orbital axis [Bender et al., 1994a]. The high seawater, CO2 dissolves to form carbonic acid which can degree of constancy can probably be explained only if react to form basic compounds like carbonate and bi­ some of the factors controlling the Dole effect changed carbonate . These acid-base reactions have no effect in ways that compensated for each other. This could on oxygen abundance so that atmospheric oxygen varia­ occur, for example, if reductions in terrestrial produc­ tions essentially reveal how atmospheric carbon dioxide tivity during glacial conditions were accompanied by re­ would behave if the acid-base reactions did not occur. ductions in marine productivity [Bender et al, 1994a]. The difference between the geochemistry of atmo­ The similarity in the patterns of 6180 variations in ice spheric O2 and CO2 can be quantified in terms of the KEELING: ATMOSPHERIC OXYGEN CYCLE 1259

1 relative fluxes of 02 and C02 expected from certain 1 part-per-million by volume (ppmV) because O2 com­ types of processes, as summarized in Table 2. One prises 20.95% of air by volume [Machta and Hughes, important difference between CO2 and O2 is that the 1970]. uptake of fossil-fuel CO2 by the ocean essentially pro­ Measurements on air samples collected at three sea- ceeds through reaction of dissolved CO2 with carbonate level sites using the interferometric technique were re­ ions and therefore involves no O2. Another difference is ported by Keeling and Sheriz [1992], and are shown here that marine photosynthesis and respiration can produce in Figure 4. Significant seasonal variations in 6XO2/N2) much larger changes in atmospheric O2 than CO2, espe­ are evident at all three sites. An interannual decrease

cially on short time scales. Here the difference depends in 02/N2 is clearly evident in the La Jolla data. Con­ mainly on the fact that CO2 exchange between the at­ current CO2 data are also shown. mosphere and oceans proceeds much more slowly than One process leading to seasonal variations in 02/N2 O2 exchange. CO2 is exchanged slowly because most is the seasonal uptake and release of O2 due to pho­ of carbon in the oceans is in the form of carbonate and tosynthesis and respiration of terrestrial ecosystems. bicarbonate ions which are not exchanged across the These exchanges of O2 are closely tied to exchanges in air-^ea interface. CO2 with an exchange ratio of approximately —1.05:1 Two techniques are now available for measuring (02:C02). The seasonal variations in CO2 in the north­ changes in atmospheric oxygen, one involving interfer- ern hemisphere are almost entirely caused by these ter­ ometry [Keeling, 1988; Keeling and Sheriz, 1992], the restrial exchanges, and they can be used to correct for other [Bender et al, 1993]. Both the effects of terrestrial exchange on the O2/N2 varia­ methods determine changes in atmospheric oxygen tions [Keeling and Sheriz, 1992]. The residual variations through changes in the O2/N2 ratio of air. Changes in O2/N2 must be oceanic in origin. The oceanic com­ in the O2/N2 ratio are mainly caused by changes in O2 ponent is especially pronounced in the southern hemi­ because N2 is constant to a very high level. Like iso­ sphere where the seasonal O2/N2 variations are accom­ topic ratios, the O2/N2 ratio is expressed as deviations panied by only very small variations in C02. from a reference Oxygen is released to the atmosphere by the oceans at middle and high latitudes in the spring and summer

*(02/N2) = " l) (2) when the net rate of photosynthesis in surface \ (L>2/preference / exceeds the rate of respiration. Oxygen is removed from The resulting deviations are multiplied by 106 and the the atmosphere in the fall and winter when marine pho­ result is expressed in a new unit called a "per meg." tosynthesis rates are lower and when deeper water, un- In these units 1/0.2095 = 4.8 per meg is equivalent to dersaturated in oxygen, mixes upwards to the surface. These seasonal air-sea O2 fluxes are linked to the rate at Table 2. which organic material is produced and exported from the euphotic zone [Jenkins and Goldman, 1985; Keeling et al, 1993] and they are linked to changes in dissolved Ratio of inorganic carbon (DIC) in the water. Seasonal heating 02 Flux to and cooling of the upper ocean also contributes to sea­ Process C0 Flux 2 sonal variations in atmospheric O2/N2 because of the Photosynthesis and Respiration on Land —1.05a solubility temperature dependence of O2 and N2 [Keel­

C02 -h H20 ^ CH20 + 02 ing and Sheriz, 1992]. b Burning Fossil Fuel -1.42 Measurements of seasonal variations in O2/N2 will be

CH„ + (1 + J) Oa ^ J H20 + C02 useful constraining estimates of the annual net photo­ Oceanic Uptake of Excess CO2 0 synthetic production of organic carbon in the euphotic zone. To succeed this application also requires taking H20 + C02 +CO? — 2HCOJ- Ocean Photosynthesis and Respiration —2 to — 8C account of transport within the atmosphere and trans­

106CO2 + 16NOJ" + H2POr + 17H+ port of O2 between the euphotic zone and deeper wa­ ^ Cioe H263 OHO Ni6 P + 13802 ters. Atmospheric oxygen data may be especially help­ ful in determining productivities over large regions be­ a On average, the ratio for terrestrial organic matter is cause the air mixes so rapidly. slightly more reduced than for carbohydrate, which yields a Measurements of O2/N2 ratios will also be useful for 02:C02 ratio slightly higher than 1.0 [Keeling, 1988]. determining the mechanisms by which excess carbon b This is the global average ratio estimated for the year dioxide produced from fossil-fuel burning is being re­ 1989 based on fuel production data [Marland and Boden, moved from the atmosphere. Over the long-term, we 1991], and using 02:C02 ratios of different fuel types [Keel­ ing, 1988]. can represent the global budget for atmospheric CO2 c Ocean photosynthesis adds O2 to seawater and removes according to CO2 from seawater in proportions of approximately —1.3:1 AC0 = F + C-0 + £ (3) as determined by the composition of marine organic matter 2 [Redfield et al, 1963]. The relative fluxes across the air- sea interface also depend on the relative efficiencies of gas where AC02 is the annual averaged change in atmo­ exchange and can vary depending on the time scale involved spheric CO2, F is the source of CO2 from burning fossil [Keeling and Severinghaus, 1993]. fuels, C (virtually negligible) is the CO2 source from ce- 1260 KEELING: ATMOSPHERIC OXYGEN CYCLE

340 340 Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan j ...... —___ Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan 1989 1990 1991 1992 1989 1990 1991 1992

b 0 -20 -40 -60 E -80 1 -100 CM -120 z -140

OCO -160 -180 -200 in situ dm -220 Flask data Scripps data Figure 4. Measurements of 8(02/N2) and CO2 368 Fitted curves 364 fraction at (a) Alert, (b) La Jolla, and (c) Cape Grim as 360 reported previously by Keeling and Sheriz [1992]. The E axes are scaled (5 per meg » 1 ppm) so that changes in cL 356 8(02/^2) and CO2 are directly comparable on a mole ci 352 O2 to mole CO2 basis. Supplemental CO2 data from J 348 the Climate Monitoring and Diagnostics Laboratory of O 344 the National Oceanic and Atmospheric Administration 340 Jon Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan and from the Scripps Institution of Oceanography are 1989 1990 1991 1992 also shown.

O 2 B ment manufacturing, is the oceanic CO sink, and O = -(A02 + H) - (AC02 - C) - (aB - 1)B (5) is the net source of CO2 from terrestrial ecosystems (B can be positive or negative), all in units of moles yr"1. The last term on the right-hand side of Eq. (5) can be Likewise, we can represent the budget for atmospheric evaluated by solving Eq. (4) for B, although this term oxygen according to is virtually negligible since a# » 1. Solving Eq. (4) for B yields

A02 = -F-H- aBB (4) B = -(l/a )(A0 + F + H) (6) where AO2 is the change in atmospheric oxygen, H is B 2 the O2 sink owing to the oxidation of elements other These equations show how observations of the change in than carbon (predominately ) in fossil fuels, atmospheric oxygen combined with estimates of fossil- and aj3 represents the 02^C exchange ratio for terres­ fuel CO2 production and O2 consumption can be used trial biomatter. to directly calculate the net exchange of CO2 with the Adding Eqs. (3) and (4), and solving for O yields oceans and with the land . KEELING: ATMOSPHERIC OXYGEN CYCLE 1261

Using preliminary estimates of the 02 trend based ing GPP and stomatal conductances. This work has been on the data shown in Figure 4, Keeling and Sheriz supported by the Environmental Protection Agency Global [1992] derive an oceanic uptake of 3.0 ± 2.0 gT C/yr Change Research Program, IAG#DW49935603-01-2, and (1 gT = 1015g) and a net terrestrial of the National Science Foundation under grant ATM-9309765. 0.2 ± 2.0 gT C/yr for the 1989-1991 period. This estimate is clearly preliminary, and the uncertainties References are too large to make these results very useful in con­ Bender, M., The delta 18O of dissolved O2 in seawater: a straining CO2 sinks. The primary source of uncertainty unique tracer of circulation and respiration in the deep comes from uncertainty in the O2 trend, and this uncer­ sea, /. Geophys. Res., 95, 22243-22252, 1990. tainty should decrease as longer records are obtained. Bender, M., L. D. Labeyrie, D. Raynaud, and C. Lorius, Bender et al. [1994b] have extended our knowledge Isotopic composition of atmospheric O2 in ice linked with of variations in atmospheric O2/N2 ratio back over the deglaciation and global primary production, Nature, 318, 349-352, 1985. past decade by measurements on air samples extracted Bender, M. L., P. P. Tans, J. T. Ellis, J. Orchardo, and K. from glacial firn at Vostok Station, Antarctica. The de­ Habfast, High precision isotope ratio mass spectrometry tected O2/N2 variations imply that the terrestrial bio­ method for measuring the O2/N2 ratio of air, Geochemica, sphere was neither a large source nor sink of CO2 over 1993. this longer period, agreeing with Keeling and Shertz Bender, M., T. Sowers, and L. Labeyrie, The Dole effect and [1992], although the uncertainties in this preliminary its variations during the last 130,000 years as measured work are again quite large. Attempts to extend the in the Vostok ice core, Global Biogeochemical Cycles, 8, 363-376, 1994a. records even further into the past from air extracted Bender, M. L., T. Sowers, J.-M. Barnola, and J. Chappel- from bubbles in the glacial ice have so far been frus­ laz, Changes in the O2/N2 ratio of the atmosphere during trated by processes which fractionate O2 relative to recent decades reflected in the composition of air in the

N2 in the ice bubbles or during the extraction process firn at Vostok Station, Antarctica, Geophys. Res. Lett., [Craig et al., 1988; Sowers et al., 1989; Bender et al, 21, 189-192, 1994b. 1995]. Bender, M., T. Sowers, M.-L. Dickson, J. Orchardo, P. Although uptake of anthropogenic CO2 by the oceans Grootes, P. A. Mayewski, D. A. Meese, Climate correla­ tions between Greenland and Antarctica during the past has no effect on atmospheric O2, it is possible that nat­ 100,000 years, Nature, 372, 663-666, 1994c. ural variability in the oceans could lead to net O2 ex­ Bender, M., T. Sowers, V. Lipenkov, On the major element change with the oceans on interannual time scales. This composition of trapped gases in ice cores, J. Geophys. possibility, which was neglected in Eq. (4), would com­ Res., in press, 1995. plicate the use of O2/N2 data for discriminating be­ Berry, J. A., Biosphere, atmosphere, ocean interactions: tween terrestrial and oceanic sinks for CO2. Such air- a plant physiologists's perspective, in , edited by P. G. sea exchanges are especially likely on the 3 to 6 year Falkowski and A.D. Woodhead, pp. 441-453, Plenum Press, New York, 1992. time scale of the El Nino phenomenon [Keeling and Sev- Bottinga, Y., and H. Craig, Oxygen isotope fractionation eringhaus, 1994] which means that the O2/N2 records between CO2 and water, and the isotopic composition of will probably need to span several El Nino events before marine atmospheric CO2, Earth and Planetary Science the data can be used to place firm constraints on the Letters, 5, 285-295, 1969. sources and sinks of anthropogenic CO2. Craig, H. and L. Gordon, and oxygen-18 varia­ tion in the ocean and marine atmosphere, in Proc. Con. Stable Isotopes in Oceanography Studies of Paleotemper- Conclusions ature, Laboratory of and Nuclear Science, Pisa, 9-130, 1965. This review has discussed the controls of the oxygen Craig, H.,Y. Horibe, and T. Sowers, Gravitational separa­ tion of gases and isotopes in polar ice caps, Science, 242, isotope ratios in atmospheric CO2 and O2 and in the 1675-1678, 1988. O2/N2 ratio and how measurements of these quantities Dole, M., The relative atomic weight of oxygen in water and can be used to study the material exchanges with biota air, /. Am. Chem. Soc, 57, 2731, 1935. over large areas. Measurements of the 180/160 ratio Dole, M., G. A. Lane, D. P. Rudd, and D. A. Zaukelies, of atmospheric C02 can provide information on stom­ Isotopic composition of atmospheric oxygen and , atal conductance and gross primary production of ter­ Geochimica et Cosmochimica Acta, 6, 65-78, 1954. Dongmann, G., The contribution of land photosynthesis to restrial ecosystems. Measurements of variations in the 18 18 16 the stationary enrichment of O in the atmosphere, Rod. 0/ 0 ratio of atmospheric O2 can provide informa­ Environm. Biophys., 11, 219-225, 1974. tion on variations in the isotopic composition of globally Dongmann, G., H. W. Niirnberg, H. Forstel, and K. Wa- averaged metabolic water, which is linked to the ratio ls gener, On the enrichment of H2 O in the leaves of tran­ of gross primary production on land versus the ocean. spiring plants, Rad. Environm. Biophys., 11, 41-52, 1974. Measurements of the O2/N2 ratio can provide informa­ Farquhar, G. D., J. Lloyd, J. A. Taylor, L. B. Flanagan, tion on net rates of carbon storage or release from the J. P. Syvertsen, K. T. Hubick, S. C. Wong, and J. R. Ehleringer, Vegetation effects on the isotope composition terrestrial biosphere and on rates of in of oxygen in atmospheric CO2, Nature, 363, 439-443, the ocean. 1993. Farquhar, G. D., S. von Caemmerer, and J. A. Berry, A Acknowledgments. I thank Joe Berry for first articu­ biochemical model of photosynthetic CO2 assimilation in lating to me the idea of using 6lsO in CO2 for constrain­ leaves of C3 species, Planta, 149, 78-90, 1980. 1262 KEELING: ATMOSPHERIC OXYGEN CYCLE

Forstel, H., The enrichment of 180 in leaf water under nat­ Marland, G. and T. Boden, CO2 emissions-modern record, ural conditions, Rad. Environm. Biophys., 15, 323-344, in Trends 91: A compendium of data on global change, 1978. edited by T. A. Boden, R. J. Sepanski et al., pp. 386- Francey, R. J. and P. P. Tans, Latitudinal variation in ,389, Carbon Dioxide Information Analysis Center, Oak oxygen-18 of atmospheric CO2, Nature, 327, 495-497, Ridge, Tenn., 1991. 1987. Mills, G. A., and H. C. Urey, The kinetics of isotope ex­ Friedli, H., U. Siegenthaler, D. Rauber, and H. Oeschger, change between carbon dioxide, bicarbonate , carbon­ Measurements of concentration, 13C/ C and 180/160 ra­ ate ion and water, J. Am. Chem. Soc., 62, 1019-1026, tios of tropospheric carbon dioxide over Switzerland, Tel­ 1940. lus, 39B, 80-88, 1987. Morita, N., 1935, The increased density of air oxygen rela­ Friedman, I. and O'Neil, J. R., U.S. Geological Survey Pro­ tive to water oxygen, J. Chem. Soc. Japan, 56, 1291. fessional Paper 440-KK, Stable Isotope Fractionation Fac­ Redfield, A. B., B. H. Ketchum, and F. A. Richards, The tors of Geochemical Interest, U. S. Printing Office, Wash­ influence of organisms on the composition of seawater, in ington, D.C, 1977. The Sea, Vol. 2, edited by M. N. Hill, pp. 26-77, Wiley Guy, R. D., J. A. Berry, M. L. Fogel, and T. C. Hoering, Interscience, New York, 1963. Differential fractionation of oxygen isotopes by cyanide- Shackle ton, N. J., and N. G. Pisias, Atmospheric carbon resistant and cyanide-sensitive respiration in plants, Planta, dioxide, orbital forcing, and climate, in The Carbon Cy­ 177, 483-491, 1989. cle and Atmospheric CO2: Natural Variations Archean to Guy, R. D., M. L. Fogel, and J. A. Berry, Photosynthetic Present, edited by E. T. Sundquist and W. S. Broecker, fractionation of the stable isotopes of oxygen and carbon, pp. 303-317, American Geophysical Union, Washington, Plant Physiol, 101, 37-47, 1993. 1985. Jenkins, W. J., and J. C. Goldman, Seasonal oxygen cycling Sowers, T., M. Bender, and D. Raynaud, Elemental and and primary production in the Sargasso Sea, J. Mar. Res., isotopic composition of occluded O2 and N2 in polar ice, 43, 465-491, 1985. J. Geophys. Res., 94, 5137-5150, 1989. Keeling, C. D., The concentration and isotopic abundance Sowers, T., M. Bender, D. Raynaud, Y. S. Korotkevich, and of carbon dioxide in rural and marine air, Geochimica et J. Orchardo, The delta 18O of atmospheric O2 from air Cosmochimica Acta, 24, 277-298, 1961. inclusions in the Vostok ice core: timing of CO2 and ice Keeling, R. F., Measuring correlations in atmospheric O2 volume changes during the penultimate deglaciation, Pa­ and CO2 mole fractions: a preliminary study in urban leoceanography, 6, 679-696, 1991. air, J. Atm. Chem., 7, 153-176, 1988. Stevens, C. L. R., D. Shultz, C. Van Baalen, and P. L. Keeling, R. F., Development of an interferometric oxygen Parker, Oxygen isotope fractionation during photosynthe­ analyzer for precise measurement of the atmospheric O2 sis in a blue-green and an green alga, Plant Physiol, 56, mole fraction, Doctoral Thesis, Harvard University, Cam­ 126-129, 1975. bridge, Massachusetts, 1988. Tans, P. P., J. A. Berry, and R. F. Keeling, Oceanic 13C/12C Keeling, R. F., Heavy Carbon Dioxide, Nature, 363, 399- observations: a new window on oceanic CO2 uptake, 400, 1993. Global Biogeochemical Cycles, 7, 353-368, 1993. Keeling, R. F., R. G. Najjar, M. L. Bender, and P. P. Tans, Yakir, D., J. A. Berry, L. Giles, and C. B. Osmond, , in Sta­ What atmospheric oxygen measurements can tell us about ble Isotopes and Plant Carbon Water Relationships, edited the global , Global Biogeochemical Cycles, 7, by J. R. Ehleringer, A. E. Hall et al., Academic Press, San 37-67, 1993. Diego, 1994. Keeling, R. F. and J. P. Severinghaus, Atmospheric oxy­ Yakir, D., J. A. Berry, L. Giles, C. B. Osmond, and R. gen measurements and the carbon cycle, in Proceedings Thomas, Applications of stable isotopes to scaling bio­ of the 1993 Global Change Institute, The Carbon Cycle, spheric photosynthetic activities, in Scaling Physiological Cambridge University Press, 1993. Processes: Leaf to Globe, edited by J. R. Ehleringer and Keeling, R. F. and S. R. Shertz, Seasonal and interannual C. B. Field, pp. 323-338, Academic Press, San Diego, variations in atmospheric oxygen and implications for the 1993. global carbon cycle, Nature, 358, 723-727, 1992. Zundel, G., W. Miekeley, B. M. Grisi, and H. Forstel, The Kiddon, J., M. Bender, J. Orchardo, J. Goldman, D. Caron, H2180 enrichment in the leaf water of tropic trees: com­ and M. Dennett, Isotopic fractionation of oxygen by respir­ parison of species from the tropical rain forest and the ing marine organisms, Global Biogeochemical Cycles, 7, semi-arid region in Brazil, Rad. Environm. Biophys., 15, 679-694, 1993. 203-212, 1978. Kroopnick, P. and H. Craig, Atmospheric oxygen: isotopic composition and solubility fractionation, Science, 175, 54- 55, 1972. Kump, L. R., J. F. K as ting, and J. M. Robinson, Atmo­ spheric oxygen variation through geologic time-introduction, R. F. Keeling, Scripps Institution of Oceanography, Uni­ Pelaeogeography, Palaeoclimatology, Palaeoecology, 97, 1- versity of California, San Diego, La Jolla, CA 92093-0236. 3, 1991. (e-mail: [email protected]) Lane, G. A. and M. Dole, Fractionation of oxygen isotopes during respiration, Science, 123, 574-576, 1956. Machta, L. and E. Hughes, Atmospheric oxygen in 1976 and 1970, Science, 168, 1582-1584, 1970. (Received July 12, 1994; accepted November 22, 1994.)