8.11 The Global Oxygen Cycle S. T. Petsch University of Massachusetts, Amherst, MA, USA

8.11.1 INTRODUCTION 515

8.11.2 DISTRIBUTION OF O2 AMONG EARTH SURFACE RESERVOIRS 516 8.11.2.1 The Atmosphere 516 8.11.2.2 The Oceans 516 8.11.2.3 Freshwater Environments 519 8.11.2.4 Soils and Groundwaters 520

8.11.3 MECHANISMS OF O2 PRODUCTION 520 8.11.3.1 Photosynthesis 520 8.11.3.2 Photolysis of Water 522

8.11.4 MECHANISMS OF O2 CONSUMPTION 522 8.11.4.1 Aerobic Cellular Respiration 522 8.11.4.2 Photorespiration 523 8.11.4.3 C1 Metabolism 523 8.11.4.4 Inorganic Metabolism 523 8.11.4.5 Macroscale Patterns of Aerobic Respiration 523 8.11.4.6 Volcanic Gases 524 8.11.4.7 Mineral Oxidation 524 8.11.4.8 Hydrothermal Vents 525 8.11.4.9 Iron and Sulfur Oxidation at the Oxic–Anoxic Transition 525 8.11.4.10 Abiotic Organic Matter Oxidation 526 8.11.5 GLOBAL OXYGEN BUDGETS AND THE GLOBAL OXYGEN CYCLE 526

8.11.6 ATMOSPHERIC O2 THROUGHOUT EARTH’S HISTORY 526 8.11.6.1 Early Models 526 8.11.6.2 The Archean 528 8.11.6.2.1 Constraints on the O2 content of the Archean atmosphere 528 8.11.6.2.2 The evolution of oxygenic photosynthesis 530 8.11.6.2.3 Carbon isotope effects associated with photosynthesis 531 8.11.6.2.4 Evidence for oxygenic photosynthesis in the Archean 533 8.11.6.3 The Proterozoic Atmosphere 533 8.11.6.3.1 Oxygenation of the Proterozoic atmosphere 533 8.11.6.3.2 Atmospheric O2 during the Mesoproterozoic 536 8.11.6.3.3 Neoproterozoic atmospheric O2 536 8.11.6.4 Phanerozoic Atmospheric O2 539 8.11.6.4.1 Constraints on Phanerozoic O2 variation 539 8.11.6.4.2 Evidence for variations in Phanerozoic O2 539 8.11.6.4.3 Numerical models of Phanerozoic oxygen concentration 542 8.11.7 CONCLUSIONS 550 REFERENCES 552

8.11.1 INTRODUCTION processes interacting on and beneath the Earth’s surface determine the concentration of O2 and One of the key defining features of Earth as a variations in O2 distribution, both temporal and planet that houses an active and diverse biology is spatial. In the present-day Earth system, the presence of free molecular oxygen (O2) in the the process that releases O2 to the atmosphere atmosphere. Biological, chemical, and physical (photosynthesis) and the processes that consume

515 516 The Global Oxygen Cycle

O2 (aerobic respiration, sulfide mineral oxidation, Approximately 21% O2 in the atmosphere oxidation of reduced volcanic gases) result in represents an average composition. In spite of large fluxes of O2 to and from the atmosphere. well-developed turbulent mixing in the lower Even relatively small changes in O2 production atmosphere, seasonal latitudinal variations in O2 and consumption have the potential to generate concentration of ^15 ppm have been recorded. large shifts in atmospheric O2 concentration These seasonal variations are most pronounced at within geologically short periods of time. Yet all high latitudes, where seasonal cycles of primary available evidence supports the conclusion that production and respiration are strongest (Keeling stasis in O2 variation is a significant feature of and Shertz, 1992). In the northern hemisphere, the the Earth’s atmosphere over wide spans of the seasonal variations are anticorrelated with atmos- geologic past. Study of the oxygen cycle is pheric pCO2; summers are dominated by high O2 therefore important because, while an equable (and high inferred net photosynthesis), while O2 atmosphere is central to life as we know it, our winters are dominated by lower O2. In addition, understanding of exactly why O2 concentrations there has been a measurable long-term decline in 14 21 remain nearly constant over large spans of atmospheric O2 concentration of ,10 mol yr , geologic time is very limited. attributed to oxidation of fossil fuels. This This chapter begins with a review of distri- decrease has been detected in both long-term bution of O2 among various reservoirs on Earth’s atmospheric monitoring stations (Keeling and surface: air, sea, and other natural waters. The key Shertz, 1992, Figure 1(a)) and in atmospheric factors that affect the concentration of O2 in the gases trapped in Antarctic firn ice bubbles atmosphere and surface waters are next con- (Figure 1(b)). The polar ice core records extend sidered, focusing on photosynthesis as the major the range of direct monitoring of atmospheric process generating free O2 and various biological composition to show that a decline in atmospheric and abiotic processes that consume O2.The O2 linked to oxidation of fossil fuels has been chapter ends with a synopsis of current models occurring since the Industrial Revolution (Bender on the evolution of an oxygenated atmosphere et al., 1994b; Battle et al., 1996). through 4.5 billion years of Earth’s history, including geochemical evidence constraining ancient O2 concentrations and numerical models of atmospheric evolution. 8.11.2.2 The Oceans

Air-saturated water has a dissolved O2 concentration dependent on temperature, the Henry’s law constant kH, and ionic strength. 8.11.2 DISTRIBUTION OF O2 AMONG In pure water at 0 8C, O2 saturation is 450 mM; EARTH SURFACE RESERVOIRS at 25 8C, saturation falls to 270 mM. Other 8.11.2.1 The Atmosphere solutes reduce O2 solubility, such that at normal seawater salinities, O2 saturation is reduced by The partial pressure of oxygen in the present- ,25%. Seawater is, of course, rarely at perfect day Earth’s atmosphere is ,0.21 bar, correspond- O2 saturation. Active photosynthesis may locally 18 ing to a total mass of ,34 £ 10 mol O2 (0.20946 increase O2 production rates, resulting in super- bar (force/area) multiplied by the surface area of saturation of O2 and degassing to the atmos- the Earth (5.1 £ 1014 m2), divided by average phere. Alternately, aerobic respiration below the 22 gravitational acceleration g (9.8 m s ) and the sea surface can consume dissolved O2 and lead 21 formula weight for O2 (32gmol )yields to severe O2-depletion or even anoxia. 18 ,34 £ 10 mol O2). There is a nearly uniform Lateral and vertical gradients in dissolved O2 mixture of the main atmospheric gases (N2,O2, concentration in seawater reflect balances Ar) from the Earth’s surface up to ,80 km altitude between O2 inputs from air–sea gas exchange, (including the troposphere, stratosphere, and biological processes of O2 production and con- mesosphere), because turbulent mixing dominates sumption, and advection of water masses. In over molecular diffusion at these altitudes. general terms, the concentration of O2 with depth Because atmospheric pressure (and thus gas in the open ocean follows the general structures molecule density) decreases exponentially with described in Figure 2. Seawater is saturated to altitude, the bulk of molecular oxygen in the supersaturated with O2 in the surface mixed layer atmosphere is concentrated within several kilo- (,0–60 m water depth). Air–sea gas exchange meters of Earth’s surface. Above this, in the and trapping of bubbles ensures constant dissol- thermosphere, gases become separated based on ution of atmospheric O2. Because gas solubility is their densities. Molecular oxygen is photodisso- temperature dependent, O2 concentrations are ciated by UV radiation to form atomic oxygen greater in colder high-latitude surface waters (O), which is the major form of oxygen above than in waters near the equator. Oxygen concen- ,120 km altitude. trations in surface waters also vary strongly with Distribution of O2 among Earth Surface Reservoirs 517

0 –20 La Jolla (32.9˚ N. 117.3˚W) –40 –60 )

–1 –80 –100 (meg 2

/N –120 2 O

d –140 –160 in situ data –180 Flask data –200 Scripps data Fitted curves –220 Jan Apr Jul Oct JanApr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan 1989 1990 1991 1992 (a) )

3,000 –1 2,000 (meg 1,000 2 /N 2

0 O d

0 20 40 60 80 100 120 140 Depth (mbs) (b)

Figure 1 (a) Interannual variability of atmospheric O2/N2 ratio, measured at La Jolla, California. 1 ppm O2 is 21 equivalent to ,4.8 meg (source Keeling and Shertz, 1992). (b) Variability in atmospheric O2/N2 ratio measured in firn ice at the South Pole, as a function of depth in meters below surface (mbs). The gentle rise in O2/N2 between 40 m and 100 m reflects a loss of atmospheric O2 during the last several centuries due to fossil fuel burning. Deeper than 100 m, selective effusion of O2 out of closing bubbles into firn air artificially boosts O2/N2 ratios (source Battle et al., 1996). season, especially in high productivity waters. and burial of organic matter in sediments may be Supersaturation is strongest in spring and summer enhanced. Below the oxygen minima zones in the (time of greatest productivity and strongest water open ocean, O2 concentrations gradually increase column stratification) when warming of surface again from 2000 m to the seafloor. This increase in layers creates a shallow density gradient that O2 results from the slow progress of global inhibits vertical mixing. Photosynthetic O2 pro- thermohaline circulation. Cold, air-saturated sea- duction exceeds consumption and exchange, and water sinks to the ocean depths at high-latitudes in supersaturation can develop. O2 concentrations the Atlantic, advecting in O2-rich waters below drop below the surface mixed layer to form O2 the O2 minimum there. Advection of O2-rich deep minimum zones (OMZs) in many ocean basins. water from the Atlantic through the Indian Ocean O2 minima form where biological consumption of into the Pacific is the source of O2 in deep Pacific O2 exceeds resupply through advection and waters. However, biological utilization of this diffusion. The depth and thickness of O2 minima deep-water O2 occurs along the entire path from vary among ocean basins. In the North Atlantic, the North Atlantic to the Pacific. For this reason, the OMZ extends several hundred meters. O2 O2 concentrations in deep Atlantic water are concentrations fall from an average of ,300 mM slightly greater (,200 mM) than in the deep in the surface mixed layer to ,160 mM at 800 m Pacific (,150 mM). depth. In the North Pacific, however, the O2 In some regions, dissolved O2 concentration minimum extends deeper, and O2 concentrations falls to zero. In these regions, restricted water fall to ,100 mM. Along the edges of ocean circulation and ample organic matter supply result basins, where OMZs impinge on the seafloor, in biological utilization of oxygen at a rate that aerobic respiration is restricted, sediments are exceeds O2 resupply through advection and anoxic at or near the sediment–water interface, diffusion. Many of these are temporary zones of 518 The Global Oxygen Cycle

Pacific Atlantic 0 0

50˚ N 53˚ N 1,000 1,000 12˚ N 8˚ N 45˚ S 2,000 2,000

45˚ S

Depth (m) 3,000 Depth (m) 3,000

4,000 4,000

5,000 5,000 0246810 0246810 Oxygen concentration (mL L–1) Oxygen concentration (mL L–1) (a) (b)

Figure 2 Depth profiles of oxygen concentration dissolved in seawater for several latitudes in the Pacific (a) and Atlantic (b). Broad trends of saturation or supersaturation at the surface, high dissolved oxygen demand at mid-depths, and replenishment of O2 through lateral advection of recharged deep water are revealed, although regional influences of productivity and intermediate and deep-water heterotropy are also seen (source Ingmanson and Wallace, 1989).

Figure 3 Map detailing locations of extensive and permanent oxygen-deficient intermediate and deep waters (Deuser, 1975) (reproduced by permission of Elsevier from Chemical Oceanography 1975, p.3).

anoxia that form in coastal regions during summer, and silled coastal fjords, larger restricted basins when warming facilitates greatest water column (e.g., the Black Sea, Cariaco Basin, and the stratification, and primary production and organic chain of basins along the southern California matter supply are high. Such O2-depletion is now Borderlands). Lastly, several regions of the open common in the Chesapeake Bay and Schelde ocean are also associated with strong O2-depletion. estuaries, off the mouth of the Mississippi River These regions (the equatorial Pacific along Central and other coastal settings. However, there are and South America and the Arabian Sea) are several regions of the world oceans where associated with deep-water upwelling, high rates of stratification and anoxia are more permanent surface water primary productivity, and high features (Figure 3). These include narrow, deep, dissolved oxygen demand in intermediate waters. Distribution of O2 among Earth Surface Reservoirs 519 Oxygen concentrations in the pore fluids of when and where the productivity is lowest sediments are controlled by a balance of entrain- (Najjar and Keeling, 2000). Low-latitude ocean ment of overlying fluids during sediment depo- surface waters show very little net air–sea O2 sition, diffusive exchange between the sediment exchange and minimal change in outgassing or and the water column, and biological utilization. In ingassing over an annual cycle. marine sediments, there is a good correlation between the rate of organic matter supply and the depth of O2 penetration in the sediment (Hartnett 8.11.2.3 Freshwater Environments et al., 1998). In coastal sediments and on the Oxygen concentrations in flowing freshwater continental shelf, burial of organic matter is environments closely match air-saturated values, sufficiently rapid to deplete the sediment of oxygen due to turbulent mixing and entrainment of air bub- within millimeters to centimeters of the sediment– bles. In static water bodies, however, O2-depletion water interface. In deeper abyssal sediments, can develop much like in the oceans. This is where organic matter delivery is greatly reduced, particularly apparent in some ice-covered lakes, O2 may penetrate several meters into the sediment where inhibited gas exchange and wintertime before being entirely consumed by respiration. respiration can result in O2-depletion and fish There is close coupling between surface water kills. High productivity during spring and summer and atmospheric O2 concentrations and air–sea in shallow turbid aquatic environments can result gas exchange fluxes (Figure 4). High rates of in extremely sharp gradients from strong O2 marine primary productivity result in net out- supersaturation at the surface to near O2-depletion gassing of O2 from the oceans to the atmosphere in within a few meters of the surface. The high spring and summer, and net ingassing of O2 during concentration of labile dissolved and particulate fall and winter. These patterns of air–sea O2 organic matter in many freshwater environments transfer relate to latitude and season: outgassing of leads to rapid O2-depletion where advective O2 during northern hemisphere high productivity resupply is limited. High rates of O2 consumption months (April through August) are accompanied have been measured in many temperature and by simultaneous ingassing in southern latitudes tropical rivers.

Sea-to-air oxygen flux (mol m–2 yr–1) zonal average of world ocean 90

0 0

–10 –20 60 –20 20 10 30 –10

0 0 0 Latitude

0 –30 –10 0

10 20 10 20

–60 –20 0

0 –90 JFMAMJJA SOND Month

Figure 4 Zonal average monthly sea-to-air oxygen flux for world oceans. Outgassing and ingassing are concentrated at mid- to high latitudes. Outgassing of oxygen is strongest when primary production rates are greatest; ingassing is at a maximum during net respiration. These patterns oscillate during an annual cycle from northern to southern hemisphere (source Najjar and Keeling, 2000). 520 The Global Oxygen Cycle 8.11.2.4 Soils and Groundwaters anion and ferric oxyhydroxides in the case of pyrite oxidation). In some instances, the volume In soil waters, oxygen concentrations depend of gaseous O2 consumed during mineral oxidation on gas diffusion through soil pore spaces, generates a mild negative pressure gradient, infiltration and advection of rainwater and drawing air into soils above sites of sulfide groundwater, air–gas exchange, and respiration mineral oxidation (Lovell, 2000). of soil organic matter (see review by Hinkle, 1994). In organic-matter-rich temperate soils, dissolved O2 concentrations are reduced, but not entirely depleted. Thus, many temperate shallow 8.11.3 MECHANISMS OF O2 PRODUCTION groundwaters contain some dissolved oxygen. 8.11.3.1 Photosynthesis Deeper groundwaters, and water-saturated soils and wetlands, generally contain little dissolved The major mechanism by which molecular O . High-latitude mineral soils and groundwaters oxygen is produced on Earth is through the 2 biological process of photosynthesis. Photosyn- contain more dissolved O2 (due to lower temperature and lesser amounts of soil organic thesis occurs in higher plants, the eukaryotic protists collectively called algae, and in two matter and biological O2 demand). Dry tropical soils are oxidized to great depths, with dissolved groups of prokaryotes: the cyanobacteria and the O concentration less than air saturated, but not prochlorophytes. In simplest terms, photosyn- 2 thesis is the harnessing of light energy to anoxic. Wet tropical forests, however, may chemically reduce carbon dioxide to simple experience significant O -depletion as rapid 2 organic compounds (e.g., glucose). The overall oxidation of leaf litter and humus occurs near reaction (Equation (1)) for photosynthesis shows the soil surface. Soil permeability also influences carbon dioxide and water reacting to produce O2 content, with more clay-rich soils exhibiting oxygen and carbohydrate: lower O2 concentrations. In certain environments, localized anomalously 6CO2 þ 6H2O ! 6O2 þ C6H12O6 ð1Þ low concentrations of soil O2 have been used by exploration geologists to indicate the presence of a Photosynthesis is actually a two-stage process, large body of chemically reduced metal sulfides in with each stage broken into a cascade of chemical the subsurface. Oxidation of sulfide minerals reactions (Figure 5). In the light reactions of during weathering and soil formation draws photosynthesis, light energy is converted to p down soil gas O2 below regional average. chemical energy that is used to dissociate water Oxidation of sulfide minerals generates solid and to yield oxygen and hydrogen and to form the aqueous-phase oxidation products (i.e., sulfate reductant NADPH from NADPþ.

Light

CO2

ATP Calvin ADP Light cycle reactions NADPH NADP

O + e– 2 H2O S0 H S Sugars + 2 H H 2– 2 SO4 S0 2– S2O3

Figure 5 The two stages of photosynthesis: light reaction and the Calvin cycle. During oxygenic photosynthesis, H2O is used as an electron source. Organisms capable of anoxygenic photosynthesis can use a variety of other 0 22 electron sources (H2S, H2,S,S2O3 ) during the light reactions, and do not liberate free O2. Energy in the form of ATP and reducing power in the form of NADPH are produced by the light reactions, and subsequently used in the Calvin cycle to deliver electrons to CO2 to produce sugars. Mechanisms of O2 Production 521 The next stage of photosynthesis, the Satellite-based measurements of seasonal and Calvin cycle, uses NADPH to reduce CO2 to yearly average chlorophyll abundance (for marine phosphoglyceraldehyde, the precursor for a variety systems) and vegetation greenness (for terrestrial of metabolic pathways, including glucose syn- ecosystems) can be applied to models that thesis. In higher plants and algae, the Calvin cycle estimate net primary productivity, CO2 fixation, operates in special organelles called chloroplasts. and O2 production. In the oceans, there are However, in bacteria the Calvin cycle occurs significant regional and seasonal variations in throughout the cytosol. The enzyme ribulose-1,5- photosynthesis that result from limitations by biphosphate carboxylase (rubisco) catalyzes light, nutrients, and temperature. Yearly averages reduction of CO2 to phosphoglycerate, which is of marine chlorophyll abundance show concen- carried through a chain of reactions that consume trated primary production at high-latitudes in ATP and NADPH and eventually yield phospho- the North Atlantic, North Pacific, and coastal glyceraldehyde. Most higher plants are termed C3 Antarctica, in regions of seawater upwelling off plants, because the first stable intermediate formed of the west coasts of Africa, South America, and during the carbon cycle is a three-carbon com- the Arabian Sea, and along the Southern Sub- pound. Several thousand species of plant, spread tropical Convergence. At mid- and high latitudes, among at least 17 families including the grasses, marine productivity is strongly seasonal, with precede the carbon cycle with a CO2-concentrating primary production concentrated in spring and mechanism which delivers a four-carbon com- summer. At low latitudes, marine primary pound to the site of the Calvin cycle and rubisco. production is lower and varies little with season These are the C4 plants. This four-carbon com- or region. On land, primary production also pound breaks down inside the chloroplasts, supply- exhibits strong regional and seasonal patterns. 22 21 ing CO2 for rubisco and the Calvin cycle. The Primary production rates (in g C m yr ) are C4-concentrating mechanisms is an advantage in greatest year-round in the tropics. Tropical hot and dry environments where leaf stomata are forests in South America, Africa, and Southeast partially closed, and internal leaf CO2 concen- Asia are the most productive ecosystems on trations are too low for rubisco to efficiently Earth. Mid-latitude temperate forests and high- capture CO2. Other plants, called CAM plants, latitude boreal forests are also highly productive, which have adapted to dry climates utilize another with a strong seasonal cycle of greatest pro- CO2-concentrating mechanisms by closing sto- duction in spring and summer. Deserts (concen- mata during the day and concentrating CO2 at trated at ,308 N and S) and polar regions are less night. All higher plants, however, produce O2 productive. These features of seasonal variability and NADPH from splitting water, and use the in primary production on land and in the oceans Calvin cycle to produce carbohydrates. Some are clearly seen in the seasonal variations in prokaryotes use mechanisms other than the Calvin atmospheric O2 (Figure 4). cycle to fix CO2 (i.e., the acetyl-CoA pathway or Transfer of O2 between the atmosphere and the reductive tricarboxylic acid pathway), but none surface seawater is controlled by air–sea gas of these organisms is involved in oxygenic exchange. Dissolved gas concentrations trend photosynthesis. towards thermodynamic equilibrium, but other Global net primary production estimates have factors may complicate dissolved O2 concen- been derived from variations in the abundance trations. Degassing of supersaturated waters can and isotopic composition of atmospheric O2. only occur at the very surface of the water. Thus, in These estimates range from 23 £ 1015 mol yr21 regions of high primary production, concen- 15 21 (Keeling and Shertz, 1992)to26£ 10 mol yr , trations of O2 can accumulate in excess of the 15 21 distributed between 14 £ 10 mol yr O2 rate of O2 degassing. In calm seas, where the production from terrestrial primary production air–sea interface is a smooth surface, gas exchange and 12 £ 1015 mol yr21 from marine primary is very limited. As seas become more rough, and production (Bender et al., 1994a). It is estimated especially during storms, gas exchange is greatly that ,50% of all photosynthetic fixation of CO2 enhanced. This is in part because of entrainment of occurs in marine surface waters. Collectively, bubbles dispersed in seawater and water droplets free-floating photosynthetic microorganisms are entrained in air, which provide much more surface called phytoplankton. These include the algal for dissolution or degassing. Also, gas exchange eukaryotes (dinoflagellates, diatoms, and the depends on diffusion across a boundary layer. red, green, brown, and golden algae), various According to Fick’s law, the diffusive flux depends species of cyanobacteria (Synechococcus and on both the concentration gradient (degree of Trichodesmium), and the common prochlorophyte super- or undersaturation) and the thickness of Prochlorococcus. Using the stoichiometry of the the boundary layer. Empirically it is observed that photosynthesis reaction, this equates to half of the boundary layer thickness decreases with all global photosynthetic oxygen production increasing wind speed, thus enhancing diffusion resulting from marine primary production. and gas exchange during high winds. 522 The Global Oxygen Cycle 8.11.3.2 Photolysis of Water Oxidative phosphorylation involves the transfer of electrons from NADH and FADH2 through a In the upper atmosphere today, a small amount cascade of electron carrying compounds to of O2 is produced through photolysis of water molecular oxygen. Compounds used in the vapor. This process is the sole source of O2 to the electron transport chain of oxidative phosphoryl- atmospheres on the icy moons of Jupiter ation include a variety of flavoproteins, quinones, (Ganymede and Europa), where trace concen- Fe–S proteins, and cytochromes. Transfer of trations of O have been detected (Vidal et al., 2 electrons from NADH to O releases considerable 1997). Water vapor photolysis may also have been 2 energy, which is used to generate a proton the source of O to the early Earth before the 2 gradient across the mitochondrial membrane evolution of oxygenic photosynthesis. However, and fuel significant ATP synthesis. In part, this the oxygen formed by photolysis would have been gradient is created by reduction of O to H O through reactions with methane and carbon 2 2 as the last step of oxidative phosphorylation. monoxide, preventing any accumulation in the atmosphere (Kasting et al., 2001). While eukaryotes and many prokaryotes use the Krebs cycle to oxidize pyruvate to CO2, there are other pathways as well. For example, prokar- yotes use the glyoxalate cycle to metabolize fatty acids. Several aerobic prokaryotes also can 8.11.4 MECHANISMS OF O2 CONSUMPTION use the Entner–Doudoroff pathway in place of 8.11.4.1 Aerobic Cellular Respiration normal glycolysis. This reaction still produces In simple terms, aerobic respiration is the pyruvate, but yields less energy in the form of oxidation of organic substrates with oxygen to ATP and NADH. yield chemical energy in the form of ATP and Glycolysis, the Krebs cycle, and oxidative NADH. In eukaryotic cells, the respiration path- phosphorylation are found in all eukaryotes way follows three steps (Figure 6). Glycolysis (animals, plants, and fungi) and many of the occurs throughout the cytosol, splitting glucose aerobic prokaryotes. The purpose of these reaction into pyruvic acid and yielding some ATP. pathways is to oxidize carbohydrates with O2, Glycolysis does not directly require free O2, and yielding CO2,H2O, and chemical energy in the thus occurs among aerobic and anaerobic organ- form of ATP. While macrofauna generally require isms. The Krebs citric acid (tricarboxylic acid) a minimum of ,0.05–0.1 bar (,10 mM) O2 to cycle and oxidative phosphorylation are localized survive, many prokaryotic microaerophilic organ- within the mitochondria of eukaryotes, and along isms can survive and thrive at much lower O2 the cell membranes of prokaryotes. The Krebs concentrations. Because most biologically cycle completes the oxidation of pyruvate to CO2 mediated oxidation processes occur through the from glycolysis, and together glycolysis and the activity of aerotolerant microorganisms, it is Krebs cycle provide chemical energy and reduc- unlikely that a strict coupling between limited tants (in the form of ATP, NADH, and FADH2) atmospheric O2 concentration and limited global for the third step—oxidative phosphorylation. respiration rates could exist.

Figure 6 The three components of aerobic respiration: glycolysis, the Krebs cycle, and oxidative phosphorylation. Sugars are used to generate energy in the form of ATP during glycolysis. The product of glycolyis, pyruvate, is converted to acetyl-CoA, and enters the Krebs cycle. CO2, stored energy as ATP, and stored reducing power as NADH and FADH2 are generated in the Krebs cycle. O2 is only directly consumed during oxidative phosphorylation to generate ATP as the final component of aerobic respiration. Mechanisms of O2 Consumption 523

8.11.4.2 Photorespiration autotrophs, meaning they reduce CO2 to generate cellular carbon in addition to oxidizing inorganic The active site of rubisco, the key enzyme compounds. In these organisms, CO2 fixation is involved in photosynthesis, can accept either CO2 not tied to O2 production by the stoichiometry of or O2. Thus, O2 is a competitive inhibitor of photosynthesis. Hydrogen-oxidizing bacteria photosynthesis. This process is known as photo- occur wherever both O2 and H2 are available. respiration, and involves addition of O2 to While some H2 produced by fermentation in ribulose-biphosphate. Products of this reaction anoxic environments (deep soils and sediments) enter a metabolic pathway that eventually may escape upward into aerobic environments, produces CO2. Unlike cellular respiration, photo- most H2 utilized by hydrogen-oxidizing bacteria respiration generates no ATP, but it does consume derives from nitrogen fixation associated with O2. In some plants, as much as 50% of the carbon nitrogen-fixing plants and cyanobacteria. Nitrify- fixed by the Calvin cycle is respired through ing bacteria are obligate autotrophs that oxidize photorespiration. Photorespiration is enhanced in ammonia. Ammonia is produced in many environ- hot, dry environments when plant cells close ments during fermentation of nitrogen compounds stomata to slow water loss, CO2 is depleted and O2 and by dissimilatory nitrate reduction. Nitrifying accumulates. Photorespiration does not occur in bacteria are common at oxic–anoxic interfaces in prokaryotes, because of the much lower relative soils, sediments, and the water column. Other concentration of O2 versus CO2 in water com- chemolithoautotrophic bacteria can oxidize pared with air. nitrite. Non-photosynthetic bacteria that can oxidize reduced sulfur compounds form a diverse group. Some are acidophiles, associated with sulfide mineral oxidation and tolerant of extre- 8.11.4.3 C1 Metabolism mely low pH. Others are neutrophilic and occur in Beyond metabolism of carbohydrates, there are many marine sediments. These organisms can several other biological processes common in utilize a wide range of sulfur compounds produced prokaryotes that consume oxygen. For example, in anaerobic environments, including H2S, thio- sulfate, polythionates, polysulfide, elemental sul- methylotrophic organisms can metabolize C1 compounds such as methane, methanol, formal- fur, and sulfite. Many of the sulfur-oxidizing dehyde, and formate, as in bacteria are also autotrophs. Some bacteria can live as chemoautotrophs through the oxidation of þ þ CH4 þNADHþH þO2 !CH3OHþNAD þH2O ferrous-iron. Some of these are acidophiles growing during mining and weathering of sulfide minerals. However, neutrophilic iron-oxidizing CH3OHþPQQ!CH2OþPQQH2 bacteria have also been detected associated with the metal sulfide plumes and precipitates at mid- CH OþNADþ þH O!HCOOHþNADHþHþ ocean ridges. Other redox-sensitive metals that 2 2 can provide a substrate for oxidation include manganese, copper, uranium, arsenic, and chro- þ þ HCOOHþNAD !CO2 þNADHþH mium. As a group, chemolithoautotrophic micro- organisms represent a substantial flux of O2 These compounds are common in soils and consumption and CO2 fixation in many common sediments as the products of anaerobic fermenta- marine and terrestrial environments. Because the tion reactions. Metabolism of these compounds net reaction of chemolithoautotrophy involves can directly consume O (through monooxygen- 2 both O2 and CO2 reduction, primary production ase enzymes) or indirectly, through formation of resulting from chemoautotrophs has a very NADH which is shuttled into oxidative phos- different O2/CO2 stoichiometry than does photo- phorylation and the electron transport chain. synthesis. Chemolithoautotrophs use the electron Oxidative metabolism of C compounds is an 1 flow from reduced substrates (metals, H2, and important microbial process in soils and sedi- reduced sulfur) to O2 to generate ATP and ments, consuming the methane produced by NAD(P)H, which in turn are use for CO2-fixation. methanogenesis. It is believed that most aerobic chemoautotrophs utilize the Calvin cycle for CO2-fixation.

8.11.4.4 Inorganic Metabolism 8.11.4.5 Macroscale Patterns of Aerobic Chemolithotrophic microorganisms are those Respiration that oxidize inorganic compounds rather than organic substrates as a source of energy and On a global scale, much biological O2 con- electrons. Many of the chemolithotrophs are also sumption is concentrated where O2 is abundant. 524 The Global Oxygen Cycle

This includes surficial terrestrial ecosystems and O2 from sites of deep-water formation, O2 may marine surface waters. A large fraction of penetrate uniformly 1 m or more into the terrestrial primary production is consumed by sediment. aerobic degradation mechanisms. Although most of this is through aerobic respiration, some fraction of aerobic degradation of organic 8.11.4.6 Volcanic Gases matter depends on anaerobic breakdown of larger biomolecules into smaller C1 compounds, Gases emitted from active volcanoes and which, if transported into aerobic zones of soil fumaroles are charged with reduced gases, or sediment, can be degraded by aerobic C1 including CO, H2,SO2,H2S, and CH4. During metabolizing microorganisms. Partially degraded explosive volcanic eruptions, these gases are terrestrial primary production can be incorporated ejected high into the atmosphere along with into soils, which slowly are degraded and eroded. H2O, CO2, and volcanic ash. Even the relatively Research has shown that soil organic matter (OM) gentle eruption of low-silicon, low-viscosity- can be preserved for up to several millennia, and shield volcanoes is associated with the release of riverine export of aged terrestrial OM may be a reduced volcanic gases. Similarly, reduced gases significant source of dissolved and particulate are released dissolved in waters associated with organic carbon to the oceans. hot springs and geysers. Oxidation of reduced Aside from select restricted basins and special- volcanic gases occurs in the atmosphere, in natural ized environments, most of the marine water waters, and on the surfaces of minerals. This is column is oxygenated. Thus, aerobic respiration predominantly an abiotic process, although many dominates in open water settings. Sediment trap chemolithoautotrophs have colonized the walls and particle flux studies have shown that sub- and channels of hot springs and fumaroles, stantial fractions of marine primary production are catalyzing the oxidation of reduced gases with completely degraded (remineralized) prior to O2. Much of biological diversity in hyperthermo- deposition at the sea floor, and thus by impli- philic environments consists of prokaryotes cation, aerobic respiration generates an O2 con- employing these unusual metabolic types. sumption demand nearly equal to the release of Recent estimates of global average volcanic gas oxygen associated with photosynthesis. The bulk emissions suggest that volcanic sulfur emissions of marine aerobic respiration occurs within the range, 0.1–1 £ 1012 mol S yr21, is nearly equally water column. This is because diffusion and distributed between SO2 and H2S(Halmer et al., mixing (and thus O2 resupply) are much greater 2002; Arthur, 2000). This range agrees well with in open water than through sediment pore fluids, the estimates used by both Holland (2002) and and because substrates for aerobic respiration (and Lasaga and Ohmoto (2002) for average volcanic S thus O2 demand) are much less concentrated in the emissions through geologic time. Other reduced water column than in the sediments. O2 consump- gas emissions (CO, CH4, and H2) are estimated to tion during aerobic respiration in the water be similar in magnitude (Mo¨rner and Etiope, column, coupled with movement of deep- 2002; Arthur, 2000; Delmelle and Stix, 2000). All water masses from O2-charged sites of deep- of these gases have very short residence times in water formation, generates the vertical and lateral the atmosphere, revealing that emission and profiles of dissolved O2 concentration observed in oxidative consumption of these gases are closely seawater. coupled, and that O2 consumption through volca- In most marine settings, O2 does not penetrate nic gas oxidation is very efficient. very far into the sediment. Under regions of high primary productivity and limited water column mixing, even if the water column is oxygenated, 8.11.4.7 Mineral Oxidation O2 may penetrate 1 mm or less, limiting the amount of aerobic respiration that occurs in the During uplift and erosion of the Earth’s sediment. In bioturbated coastal sediments, O2 continents, rocks containing chemically reduced penetration is facilitated by the recharging of pore minerals become exposed to the oxidizing fluids through organisms that pump overlying conditions of the atmosphere. Common rock- water into burrows, reaching several centimeters forming minerals susceptible to oxidation include into the sediment in places. However, patterns of olivine, Fe2þ-bearing pyroxenes and amphiboles, oxic and anoxic sediment exhibit a great degree metal sulfides, and graphite. Ferrous-iron oxi- of spatial and temporal complexity as a result of dation is a common feature of soil formation. 2þ spotty burrow distributions, radial diffusion of O2 Iron oxides derived from oxidation of Fe in the from burrows, and continual excavation and parent rock accumulate in the B horizon of infilling through time. Conversely, in deep-sea temperate soils, and extensive laterites consisting (pelagic) sediments, where organic matter deliv- of iron and aluminum oxides develop in tropical ery is minor and waters are cold and charged with soils to many meters of depth. Where erosion Mechanisms of O2 Consumption 525 rates are high, iron-bearing silicate minerals may containing oxidized iron-mineral can extend as be transported short distances in rivers; however, much as 500 m below the seafloor. Much more O2 iron oxidation is so efficient that very few is consumed during oxidation of black smoker and sediments show deposition of clastic ferrous- chimney sulfides. Chimney sulfide may be only iron minerals. Sulfide minerals are extremely partially oxidized prior to transport off-axis susceptible to oxidation, often being completely through spreading and burial by sediments. weathered from near-surface rocks. Oxidation of However, black smoker metal-rich fluids are fairly sulfide minerals generates appreciable acidity, rapidly oxidized in seawater, forming insoluble and in areas where mining has brought sulfide iron and manganese oxides, which slowly settle minerals in contact with the atmosphere or O2- out on the seafloor, generating metalliferous charged rainwater, low-pH discharge has become sediments. a serious environmental problem. Although Because of the diffuse nature of reduced ferrous silicates and sulfide minerals such as species in hydrothermal fluids, it is not known pyrite will oxidize under sterile conditions, a what role marine chemolithoautotrophic micro- growing body of evidence suggests that in many organisms may play in O2 consumption associ- natural environments, iron and sulfur oxidation is ated with fluid plumes. Certainly such organisms mediated by chemolithoautotrophic microorgan- live within the walls and rubble of cooler isms. Prokaryotes with chemolithoautotrophic chimneys and basalts undergoing seafloor weath- metabolic pathways have been isolated from ering. Because of the great length of mid-ocean many environments where iron and sulfur ridges throughout the oceans, and the abundance oxidation occurs. The amount of O2 consumed of pyrite and other metal sulfides associated with annually through oxidation of Fe2þ and sulfur- these ridges, colonization and oxidation of metal bearing minerals is not known, but based on sulfides by chemolithoautotrophic organisms may sulfur isotope mass balance constraints is on the form an unrecognized source of primary order of (0.1–1) £ 1012 mol yr21 each for iron production associated with consumption, not net and sulfur, similar in magnitude to the flux of release, of O2. reduced gases from volcanism.

8.11.4.9 Iron and Sulfur Oxidation at 8.11.4.8 Hydrothermal Vents the Oxic–Anoxic Transition The spreading of lithospheric plates along mid- In restricted marine basins underlying highly ocean ridges is associated with much undersea productive surface waters, where consumption of volcanic eruptions and release of chemically O2 through aerobic respiration near the surface reduced metal sulfides. Volcanic gases released allows anoxia to at least episodically extend by subaerial volcanoes are also generated by beyond the sediment into the water column, submarine eruptions, contributing dissolved oxidation of reduced sulfur and iron may occur reduced gases to seawater. Extrusive lava flows when O2 is present. Within the Black Sea and generate pillow basalts, which are composed in many anoxic coastal fjords, the transition from part of ferrous-iron silicates. Within concentrated oxic to anoxic environments occurs within the zones of hydrothermal fluid flow, fracture-filling water column. At other locations, such as the and massive sulfide minerals precipitate within modern Peru Shelf and coastal California basins, the pillow lavas, large chimneys grow from the the transition occurs right at the sediment–water seafloor by rapid precipitation of Fe–Cu–Zn interface. In these environments, small concen- sulfides, and metal sulfide-rich plumes of high- trations of sulfide and O2 may coexist within a temperature “black smoke” are released into very narrow band where sulfide oxidation occurs. seawater. In cooler zones, more gradual emana- In such environments, appreciable sulfur recy- tions of metal and sulfide rich fluid diffuse upward cling may occur, with processes of sulfate through the pillow basalts and slowly mix with reduction, sulfide oxidation, and sulfur dispropor- seawater. Convective cells develop, in which cold tionation acting within millimeters of each seawater is drawn down into pillow lavas off other. These reactions are highly mediated by the ridge axis to replace the water released at the microorganisms, as evidenced by extensive hydrothermal vents. Reaction of seawater with mats of chemolithoautotrophic sulfur-oxidizing basalt serves to alter the basalt. Some sulfide is Beggiatoa and Thioploca found where the oxic– liberated from the basalt, entrained seawater anoxic interface and seafloor coincide. sulfate precipitates as anhydrite or is reduced to Similary, in freshwater and non-sulfidic brack- 2þ sulfide, and Fe and other metals in basalt are ish environments with strong O2 demand, replaced with seawater-derived magnesium. Oxy- dissolved ferrous-iron may accumulate in gen dissolved in seawater is consumed during groundwaters and anoxic bottom waters. Signifi- alteration of seafloor basalts. Altered basalts cant iron oxidation will occur where the water 526 The Global Oxygen Cycle table outcrops with the land surface (i.e., ground- although smoldering fires with inefficient oxi- water outflow into a stream), or in lakes and dation can be maintained. At high pO2, even wet estuaries, at the oxic–anoxic transition within the wood can support flame, and fires are easily water column. Insoluble iron oxides precipitate initiated with a spark discharge such as lightning. and settle down to the sediment. Recycling of iron Although most methane on Earth is oxidized may occur if sufficient organic matter exists for during slow gradual transport upwards through ferric-iron reduction to ferrous-iron. soils and sediments, at select environments there The net effect of iron and sulfur recycling on is direct injection of methane into the atmosphere. atmospheric O2 is difficult to constrain. In most Methane reacts with O2 in the presence of light or cases, oxidation of sulfur or iron consumes O2 metal surface catalysts. This reaction is fast (there are some anaerobic chemolithoautotrophic enough, and the amount of atmospheric O2 large microorganisms that can oxidize reduced sub- enough, that significant concentrations of atmos- strates using nitrate or sulfate as the electron pheric CH4 are unlikely to accumulate. Cata- acceptor). Reduction of sulfate or ferric-iron are strophic calving of submarine, CH4-rich hydrates almost entirely biological processes, coupled with during the geologic past may have liberated the oxidation of organic matter; sulfate and large quantities of methane to the atmosphere. ferric-iron reduction individually have no effect However, isotopic evidence suggests that this on O2. However, the major source of organic methane was oxidized and consumed in a substrates for sulfate and ferric-iron reduction is geologically short span of time. ultimately biomass derived from photosynthetic organisms, which is associated with O2 gener- ation. The net change derived from summing 22 2þ 22 the three processes (S -orFe -oxidation, SO4 - 8.11.5 GLOBAL OXYGEN BUDGETS AND or Fe3þ- reduction, and photosynthesis), is net gain THE GLOBAL OXYGEN CYCLE of organic matter with no net production or One window into the global budget of oxygen is consumption of O2. the variation in O2 concentration normalized to N2. O2/N2 ratios reflect changes in atmospheric O2 abundance, because N2 concentration is assumed 8.11.4.10 Abiotic Organic Matter Oxidation to be invariant through time. Over an annual cycle, d(O2/N2) can vary by 100 per meg or more, Aerobic respiration is the main means by especially at high-latitude sites (Keeling and which O2 is consumed on the Earth. This Shertz, 1992). These variations reflect latitudinal pathway occurs throughout most Earth-surface variations in net O2 production and consumption environments: soils and aquatic systems, marine related to seasonal high productivity during surface waters supersaturated with O2, within the summer. Observations of O2/N2 variation have water column and the upper zones of sediments. been expanded beyond direct observation (limited However, reduced carbon materials are also to the past several decades) to records of atmos- reacted with O2 in a variety of environments pheric composition as trapped in Antarctic firn ice where biological activity has not been demon- and recent ice cores (Battle et al., 1996; Sowers strated. Among these are photo-oxidation of et al., 1989; Bender et al., 1985, 1994b). These dissolved and particulate organic matter and records reflect a slow gradual decrease in atmos- fossil fuels, fires from burning vegetation and pheric O2 abundance over historical times, attrib- fossil fuels, and atmospheric methane oxidation. uted to the release and oxidation of fossil fuels. Olefins (organic compounds containing double As yet, details of the fluxes involved in the bonds) are susceptible to oxidation in the processes that generate and consume molecular presence of transition metapls, ozone, UV light, oxygen are too poorly constrained to establish a or gamma radiation. Low-molecular-weight oxi- balanced O2 budget. A summary of the processes dized organic degradation products form from believed to dominate controls on atmospheric O2, oxidation reactions, which in turn may provide and reasonable best guesses for the magnitude of organic substrates for aerobic respiration or C1 these fluxes, if available, are shown in Figure 7 metabolism. Fires are of course high-temperature (from Keeling et al., 1993). combustion of organic materials with O2. Research on fires has shown that O2 concen- tration has a strong influence on initiation and maintenance of fires (Watson, 1978; Lenton and 8.11.6 ATMOSPHERIC O2 THROUGHOUT Watson, 2000). Although the exact relationship EARTH’S HISTORY between p and initiation of fire in real terrestrial O2 8.11.6.1 Early Models forest communities is debated (see Robinson, 1989, 1991), it is generally agreed that, at low Starting with Cloud (1976), two key geologic pO2, fires cannot be started even on dry wood, formations have been invoked to constrain the Atmospheric O2 throughout Earth’s History 527

Hydrogen escape ~10–5

Atmosphere 34,000

~140 ~140 Primary Respiration Primary production production 11 Surface waters 12 9.2 6 Autotroph respiration 4.6 Heterotroph 0.6 0.2 and soil respiration 4.6 Respiration 0.4 Intermediate and deep water 219

Weathering of organic matter and sulfide minerals, volcanic gas oxidation 0.01

Figure 7 Global budget for molecular oxygen, including gas and dissolved O2 reservoirs (sources Keeling et al., 1993; Bender et al., 1994a,b).

history of oxygenation of the atmosphere: banded Time (Ga) iron formations (BIFs) and red beds (Figure 8). BIFs are chemical sediments containing very 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 Proterozoic little detritus, and consist of silica laminae Archean Phanero- interbedded with layers of alternately high and Paleo- Meso- Neo- zoic low ratios of ferric- to ferrous-iron. As chemical sediments, BIFs imply the direct precipitation of Banded iron formations ferrous-iron from the water column. A ferrous- iron-rich ocean requires anoxia, which by impli- Red beds cation requires an O2-free atmosphere and an anaerobic world. However, the ferric-iron layers Figure 8 Archean distribution of banded iron for- in BIFs do reflect consumption of molecular O2 mations, with short reoccurrence associated with wide- (oxidation of ferrous-iron) at a rate much greater spread glaciation in the Neoproterozoic, and the Proterozoic and Phanerozoic distribution of sedimen- than supply of O2 through prebiotic H2O photolysis. Thus, BIFs may also record the tary rocks containing ferric-iron cements (red beds). The end of banded iron formation and beginning of red evolution of oxygenic photosynthesis and at bed deposition at ,2.2 Ga has been taken as evidence least localized elevated dissolved O2 concen- for a major oxygenation event in Earth’s atmosphere. trations (Walker, 1979). Red beds are sandy sedimentary rocks rich in coatings, cements, and particles of ferric-iron. Red beds form during and after sediment deposition, and thus require that agreed on for several decades, the details and both the atmosphere and groundwater are oxidiz- texture of oxygenation of the Earth’s atmosphere ing. The occurrence of the oldest red beds are still being debated. (,2.0 Ga) coincides nearly with the disappear- Geochemists and cosmochemists initially ance of BIFs (Walker, 1979), suggesting that looked to models of planetary formation and some threshold of atmospheric O2 concentration comparison with other terrestrial planets to was reached at this time. Although the general understand the earliest composition of Earth’s concept of a low-O2 atmosphere before ,2.0 Ga atmosphere. During planetary accretion and core and accumulated O2 since that time has been formation, volatile components were liberated 528 The Global Oxygen Cycle from a molten and slowly convecting mixture of H2 from Earth’s gravity. Because H2 is a of silicates, metals, and trapped gases. The strongly reducing gas, loss of H2 from the Earth gravitational field of Earth was sufficient to equates to loss of reducing power or net increase retain most of the gases released from the in whole Earth oxidation state (Walker, 1979). interior. These include CH4,H2O, N2,NH3, Today, gravitational escape of H2 and thus and H2S. Much H2 and He released from the increase in oxidation state is minor, because interior escaped Earth’s gravitational field into little if any H2 manages to reach the upper levels space; only massive planets such as Jupiter, of the atmosphere without oxidizing. Early in Saturn, Neptune, and Uranus have retained an Earth’s history, sources of H2 included volcanic H2 –He rich atmosphere. Photolysis of H2O, gases and water vapor photolysis. The small flux NH3, and H2S produced free O2,N2, and S, of O2 produced by photolysis was rapidly respectively. O2 was rapidly consumed by consumed by reaction with ferrous-iron and oxidation of CH4 and H2S to form CO2, CO, sulfide, contributing to oxidation of the crust. and SO2. High partial pressures of CO2 and CH4 Today, volcanic gases are fairly oxidized, maintained a strong greenhouse effect and warm consisting mainly of H2O and CO2, with smaller average Earth surface temperature (,90 8C), in amounts of H2, CO, and CH4. The oxidation state spite of much lower solar luminosity. Recogniz- of volcanic gases derives in part from the ing that the early Earth contained an atmos- oxidation state of the Earth’s mantle. Mantle pheric substantially richer in strong greenhouse oxygen fugacity today is at or near the quartz- f gases compared with the modern world provided fayalite-magnesite buffer (QFM), as is the O2 of a resolution to subfreezing average Earth surface eruptive volcanic gases. Using whole-rock and temperatures predicted for the early Earth due to spinel chromium abundance from volcanogenic reduced solar luminosity (Kasting et al., 2001, rocks through time, Delano (2001) has argued that 1983; Kiehl and Dickinson, 1987). the average oxidation state of the Earth’s mantle Liquid water on the early planet Earth allowed was set very early in Earth’s history (,3.6– f a hydrologic cycle and silicate mineral weath- 3.9 Ga) to O2 at or near the QFM buffer. Magmas ering to develop. Fairly quickly, much of the with this oxidation state release volcanic gases atmosphere’s CO2 was reacted with silicates to rich in H2O, CO2, and SO2, rather than more produce a bicarbonate-buffer ocean, while CH4 reducing gases. Thus, throughout much of Earth’s was rapidly consumed by oxygen produced history, volcanic gases contributing to the atmos- through photolysis of H2O. Early microorgan- phere have been fairly oxidized. More reduced isms (and many of the most primitive organisms magma compositions have been detected in in existence today) used inorganic substrates to diamond-bearing assemblages likely Hadean derive energy, and thrive at the high tempera- in age (.4.0 Ga) (Haggerty and Toft, 1985). tures expected to be widespread during Earth’s The increase in mantle oxidation within early history. These organisms include Archea several hundred million years of early Earth’s history reveals very rapid “mantle þ crust” over- that oxidize H2 using elemental sulfur. Once turn and mixing at this time, coupled with photolysis and CH4 oxidation generated sufficient subduction and reaction of the mantle with pCO , methanogenic Archea may have evolved. 2 hydrated and oxidized crustal minerals (gene- These organisms reduce CO2 to CH4 using H2. However, sustainable life on the planet is rated from reaction with O2 produced through unlikely to have developed during the first H2O photolysis). several hundred million years of Earth’s history, due to large and frequent bolide impacts that would have sterilized the entire Earth’s surface prior to ,3.8 Ga (Sleep et al., 1989; Sleep and 8.11.6.2 The Archean Zahnle, 1998; Sleep et al., 2001; Wilde et al., 8.11.6.2.1 Constraints on the O2 content 2001), although recent work by Valley et al. (2002) of the Archean atmosphere suggests a cool early Earth that continually supported liquid water as early as 4.4 Ga. Several lines of geochemical evidence support Much of present understanding of the earliest low to negligible concentrations of atmospheric evolution of Earth’s atmosphere can trace O2 during the Archean and earliest Proterozoic, descent from Walker (1979) and references when oxygenic photosynthesis may have evolved. therein. The prebiological atmosphere (before The presence of pyrite and uraninite in detrital the origin of life) was controlled principally by Archean sediments reveals that the atmosphere the composition of gases emitted from volca- in the earliest Archean contained no free O2 noes. Emission of H2 in volcanic gases has (Cloud, 1972). Although Archean-age detrital contributed to net oxidation of the planet pyrites from South Africa may be hydrothermal through time. This is achieved through several in origin, Australian sediments of the Pilbara mechanisms. Simplest is hydrodynamic escape craton (3.25–2.75 Ga) contain rounded grains of Atmospheric O2 throughout Earth’s History 529 pyrite, uraninite, and siderite (Rasmussen and detected in sulfide inclusions in diamonds also Buick, 1999), cited as evidence for an anoxic are believed to derive from photochemical SO2 atmosphere at this time. Although disputed oxidation in an O2-free atmosphere at 2.9 Ga (Ohmoto, 1999), it is difficult to explain detrital (Farquhar et al., 2002). Not only do these isotope minerals that are extremely susceptible to dissol- ratios require an O2-free atmosphere, but they ution and oxidation under oxidizing condition also imply significant contact between the unless the atmosphere of the Archean was mantle, crust, and atmosphere as recently as essentially devoid of O2. 2.9 Ga. Archean paleosols provide other geochemical Nitrogen and sulfur isotope ratios in Archean evidence suggesting formation under reducing sedimentary rocks also indicate limited or conditions. For example, the 2.75 Ga Mount Roe negligible atmospheric O2 concentrations #2 paleosol of Western Australia contains up to (Figure 9). Under an O2-free environment, 0.10% organic carbon with isotope ratios nitrogen could only exist as N2 and reduced between 233‰ and 255‰ (Rye and Holland, forms (NH3, etc.). Any nitrate or nitrite produced 2000). These isotope ratios suggest that metha- by photolysis would be quickly reduced, likely nogenesis and methanotrophy were important with Fe2þ. If nitrate is not available, then pathways of carbon cycling in these soils. For denitrification (reduction of nitrate to free N2) modern soils in which the bulk organic matter is cannot occur. Denitrification is associated with a strongly 13C-depleted (,240‰), the methane substantial nitrogen-isotope discrimination, gen- 15 fueling methanotrophy must be derived from erating N2 that is substantially depleted in N 2 somewhere outside the soil, because reasonable relative to the NO3 source. In the modern rates of fermentation and methanogenesis cannot system, this results in seawater nitrate (and 15 supply enough CH4. By extension, Rye and organic matter) that is N-enriched relative to Holland (2000) argue that these soils formed air. Nitrogen in kerogen from Archean sedimen- 15 under an atmosphere rich in CH4, with any O2 tary rocks is not enriched in N (as is found in consumed during aerobic methanotrophy having all kerogen nitrogen from Proterozoic age to the been supplied by localized limited populations of present), but instead is depleted relative to oxygenic photoautotrophs. Other paleosol studies modern atmospheric N2 by several ‰ (Beaumont have used lack of cerium oxidation during soil and Robert, 1999). This is consistent with an formation as an indicator of atmospheric anoxia Archean nitrogen cycle in which no nitrate and no in the Archean (Murakami et al., 2001). In a free O2 was available, and nitrogen cycling was broader survey of Archean and Proterozoic limited to N2-fixation, mineralization and ammo- paleosols, Rye and Holland (1998) observe that nia volatilization. Bacterial sulfate reduction is all examined paleosols older than 2.4 Ga indicate associated with a significant isotope discrimi- loss of iron during weathering and soil formation. nation, producing sulfide that is depleted in 34S This chemical feature is consistent with soil relative to substrate sulfate. The magnitude of development under an atmosphere containing sulfur isotope fractionation during sulfate 24 5 ,10 atm O2 (1 atm ¼ 1.01325 £ 10 Pa), reduction depends in part on available sulfate although some research has suggested that anoxic concentrations. Very limited differences in soil development in the Archean does not sulfur isotopic ratios among Archean sedimentary necessarily require an anoxic atmosphere sulfide and sulfate minerals (,2‰ d 34S) indicate (Ohmoto, 1996). only minor isotope fractionation during sulfate Other evidence for low Archean atmospheric reduction in the Archean oceans (Canfield et al., oxygen concentrations come from studies of 2000; Habicht et al., 2002), best explained by 22 mass-independent sulfur isotope fractionation. extremely low SO4 concentrations (,200 mMin Photochemical oxidation of volcanic sulfur contrast with modern concentrations of ,28 mM) species, in contrast with aqueous-phase oxidation (Habicht et al., 2002). The limited supply of sulfate and dissolution that characterizes the modern in the Archean ocean suggests that the major source sulfur cycle, may have been the major source of of sulfate to Archean seawater was volcanic gas, sulfate to seawater in the Archean (Farquhar because oxidative weathering of sulfide minerals et al., 2002; Farquhar et al., 2000). Distinct shifts could not occur under an O2-free atmosphere. in d 33S and d34S in sulfide and sulfate from The limited sulfate concentration would have Archean rocks occurred between 2.4–2.0 Ga, suppressed the activity of sulfate-reducing consistent with a shift from an O2-free early bacteria and facilitated methanogenesis. How- atmosphere in which SO2 photochemistry could ever, by ,2.7 Ga, sedimentary sulfides that are dominate among seawater sulfate sources to an 34S-depleted relative to sulfate are detected, O2-rich later atmosphere in which oxidative suggesting at least sulfate reduction, and by weathering of sulfide minerals predominated implication sources of sulfate to seawater through over photochemistry as the major source of sulfide oxidation, may have developed (Canfield seawater sulfate. Sulfur isotope heterogeneities et al., 2000). 530 The Global Oxygen Cycle

60

40

20

34 d S 0 (‰)

–20

–40

–60 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0

(a) Time (Ga)

Evolution of nitrogen isotopic composition compared to oxygen atmospheric level evolution 0.0 0.0 → + R-NH2 NH4 + → – 0.5 NH4 NO3 0.5 – → NO3 N2 1.0 1.0

1.5 1.5

+ Modern atmospheric value (Ga) NH4 + 2O2 2.0 2.0 – + NO3 + H2O + H Time before present Time 2.5 2.5 PAL

3.0 3.0 → + R-NH2 NH4 3.5 3.5 –10 0 10 10–12 10–5 10–4 10–3 10–2 10–1 1 Reactions involved in 15 O2 concentration / (PAL) d Nair ‰ nitrogen (b) cycle

Figure 9 (a) The isotopic composition of sedimentary sulfate and sulfides through geologic time. The two upper lines show the isotopic composition of seawater sulfate (5‰ offset indicates uncertainty). The lower line indicated d 34S of sulfate displaced by 255‰, to mimic average Phanerozoic maximum fractionation during bacterial sulfate reduction. Sulfide isotopic data (circles) indicated a much reduced fractionation between sulfate and sulfide in the Archean and Proterozoic (source Canfield, 1998). (b) Geologic evolution of the nitrogen isotopic composition of kerogen (left), estimated atmospheric O2 content, and representative reactions in the biogeochemical nitrogen cycle at those estimated O2 concentrations (sources Beaumont and Robert, 1999; Kasting, 1992).

8.11.6.2.2 The evolution of oxygenic generated a hydrocarbon-rich smog that could photosynthesis screen UV light and protect early life in the absence of O2 and ozone (Kasting et al., 2001). In the early Archean, methanogenesis (reac- Other means of protection from UV damage in tion of H2 þ CO2 to yield CH4) was likely a the O2-free Archean include biomineralization significant component of total primary pro- of cyanobacteria within UV-shielded iron–silica duction. O2 concentrations in the atmosphere sinters (Phoenix et al., 2001). were suppressed due to limited O2 production Biological evolution may have contributed and rapid consumption with iron, sulfur, and to early Archean oxidation of the Earth. Catling reduced gases, while CH4 concentrations were et al., (2001) have recognized that CO2 fixation likely very high (Kasting et al., 2001). The high associated with early photosynthesis may have methane abundance is calculated to have been coupled with active fermentation and Atmospheric O2 throughout Earth’s History 531 methanogenesis. Prior to any accumulation of oxidized crustal rocks. BIF formation in the 2þ atmospheric O2,CH4 may have been a large Archean may be related to dissolved Fe that component of Earth’s atmosphere. A high flux of was oxidized in shallow-water settings associated biogenic methane is supported by coupled eco- with local oxygenic photosynthesis or chemoau- system–climate models of the early Earth (Pavlov totrophy (Kasting and Siefert, 2002). After the et al., 2000; Kasting et al., 2001) as a supplement biological innovation of oxygenic photosynthesis to Earth’s greenhouse warming under reduced evolved, there was still a several hundred million solar luminosity in the Archean. CH4 in the upper year gap until O2 began to accumulate in the atmosphere is consumed by UV light to yield atmosphere, because O2 could only accumulate hydrogen (favored form of hydrogen in the upper once the supply of reduced gases (CO and CH4) atmosphere), which escapes to space. Because H2 and ferrous-iron fell below rates of photosynthetic escape leads to net oxidation, biological pro- O2 supply. ductivity and methanogenesis result in slow oxidation of the planet. Net oxidation may be in the form of direct O2 accumulation (if CO2 fixation is associated with oxygenic photosyn- 8.11.6.2.3 Carbon isotope effects associated with photosynthesis thesis) or indirectly (if CO2 fixation occurs via anoxygenic photosynthesis or anaerobic che- The main compound responsible for harvesting moautotrophy) through production of oxidized light energy to produce NADPH and splitting iron or sulfur minerals in the crust, which upon water to form O2 is chlorophyll. There are several subduction are mixed with other crustal rocks and different structural variants of chlorophyll, includ- increase the crustal oxidation state. One system ing chlorophyll a, chlorophyll b, and bacterio- that may represent a model of early Archean chlorophylls a–e and g. Each of these shows biological productivity consists of microbial mats optimum excitation at a different wavelength of found in hypersaline coastal ponds. In these mats, light. All oxygenic photosynthetic organisms cyanobacteria produce H2 and CO that can be used utilize chlorophyll a and/or chlorophyll b. Other as substrates by associated chemoautotrophs, and non-oxygenic photoautotrophic microorganisms significant CH4 fluxes out of the mats have been employ a diverse range of chlorophylls. measured (Hoehler et al., 2001). Thus, mats may The earliest evidence for evolution of oxygenic represent communities of oxygenic photosyn- photosynthesis comes from carbon isotopic signa- thesis, chemoautotrophy, and methanogenesis tures preserved in Archean rocks (Figure 10). CO2 occurring in close physical proximity. Such fixation through the Calvin cycle is associated with communities would have contributed to elevated a significant carbon isotope discrimination, such atmospheric CH4 and the irreversible escape of that organic matter produced through CO2 fixation hydrogen from the Archean atmosphere, contri- is depleted in 13C relative to 12C by several per mil. buting to oxidation of the early Earth. In a closed or semi-closed system (up to and The earliest photosynthetic communities may including the whole ocean–atmosphere system), have contributed to oxidation of the early Earth isotope discrimination during fixation of CO2 then through production of oxidized crustal minerals results in a slight enrichment in the 13C/12C ratio without requiring production of O2. Crustal rocks of CO2 not taken up during photosynthesis. Thus, a today are more oxidized than the mantle, with biosignature of CO2 fixation is an enrichment 13 12 compositions ranging between the QFM and in C/ C ratio in atmospheric CO2 and seawater f hematite-magnesite O2 buffers. This is best bicarbonate over a whole-Earth averaged isotope explained as an irreversible oxidation of the ratio. When carbonate minerals precipitate from crust associated with methanogenesis and hydro- seawater, they record the seawater isotope value. gen escape (Catling et al., 2001). It is unlikely that Enrichment of the 13C/12C ratio in early Archean O2 produced by early photosynthesis ever directly carbonate minerals, and by extension seawater entered the atmosphere. bicarbonate, is taken as early evidence for CO2 Maintaining low O2 concentrations in the fixation. Although anoxygenic photoautotrophs atmosphere while generating oxidized crustal and chemoautotrophs fix CO2 without generating rocks requires oxidation mechanisms that do not O2, these groups either do not employ the Calvin involve O2. Among these may be serpentinization cycle (many anoxygenic photoautotrophs use the of seafloor basalts (Kasting and Siefert, 2002). acetyl-CoA or reverse tricarboxylic acid pathways) During seafloor weathering, ferrous-iron is or require O2 (most chemolithoautotrophs). Thus, released to form ferric oxyhydroxides. Using the most likely group of organisms responsible for H2OorCO2 as the ferrous-iron oxidant, H2 and this isotope effect are oxygenic photoautotrophs CH4 are generated. This H2 gas (either produced such as cyanobacteria. Kerogen and graphite that is directly, or indirectly through UV decomposition isotopically depleted in 13C is a common and of CH4) can escape the atmosphere, resulting in continuous feature of the sedimentary record, net oxidation of the crust and accumulation of extending as far back as ,3.8 Ga to the Isua 532 The Global Oxygen Cycle

Lomagundi–Jatulian Terminal Neoproterozoic Late Permian event (2.0–2.3 Ga) event (~0.6 Ga) event (~0.26 Ga) +10

0

Ccarb –10

, PDB) –20 Approximate mean ‰

C ( –30 13 d A2 C Francevillian excursion org –40 Isua metasedi· (~2.0 Ga) mentary suite (max. 3.85 Ga) Fortescue excursion –50 A1 (~2.7 Ga) Archean Proterozoic Phanerozoic 4 3.5 3 2.5 2 1.5 1 0.5 0 Age (Ga = 109 yr)

Figure 10 Isotopic composition of carbonates and organic carbon in sedimentary and metasedimentary rocks through geologic time. The negative excursions in organic matter d13C at 2.7 Ga and 2.0 Ga may relate to extensive methanogenesis as a mechanism of carbon fixation (Schidlowski, 2001) (reproduced by permission of Elsevier from Precanb. Res. 2001, 106, 117–134).

Supracrustal Suite of Greenland (Schidlowski, carbonate has not varied throughout geologic time 1988, 2001; Nutman et al., 1997). Although some raises several intriguing issues. First, biogeochem- isotopically depleted graphite in the metasedi- ical cycling of carbon exhibits remarkable con- ments from the Isua Suite may derive from abiotic stancy across 4 Gyr of biological evolution, in hydrothermal processes (van Zuilen et al., 2002), spite of large-scale innovations in primary pro- rocks interpreted as metamorphosed turbidite duction and respiration (including anoxygenic and deposits retain 13C-depleted graphite believed to oxygenic photosynthesis, chemoautotrophy, sul- be biological in origin. Moreover, the isotopic fate reduction, methanotrophy, and aerobic res- distance between coeval carbonate and organic piration). Thus, with a few notable exceptions matter (in the form of kerogen) can be used to expressed in the carbonate isotope record, burial estimate biological productivity through time. flux ratios between organic matter and carbonate With a few exceptions, these isotope mass balance have remained constant, in spite of varying estimates reveal that, since Archean times, global dominance of different modes of carbon fixation scale partitioning between inorganic and organic and respiration through time, not to mention other carbon, and thus global productivity and carbon possible controls on organic matter burial and burial, have not varied greatly over nearly 4 Gyr of preservation commonly invoked for Phanerozoic Earth’s history (Schidlowski, 1988, 2001). systems (anoxia of bottom waters, sedimentation Approximately 25% of crustal carbon burial is in rate, selective preservation, or cumulative oxygen the form of organic carbon, and the remainder is exposure time). Second, the constancy of organic inorganic carbonate minerals. This estimate matter versus carbonate burial through time derives from the mass- and isotope-balance reveals that throughout geologic time, the relative equation: contributions of various sources and sinks of þ 13 13 carbon to the “ocean atmosphere” system have d Cavg ¼ f d Corganic matter þð1 2 f Þ remained constant. In other words, to maintain a 13 constant carbonate isotopic composition through d Ccarbonate ð2Þ time, not only must the relative proportion of 13 where d Cavg is the average isotopic composition organic matter versus carbonate burial have of crustal carbon entering the oceans from remained nearly constant, but the relative intensity continental weathering and primordial carbon of organic matter versus carbonate weathering 13 emitted from volcanoes, d Corganic matter is the also must have remained nearly constant. In the average isotopic composition of sedimentary earliest stages of Earth’s history, when continents 13 organic matter, d Ccarbonate is the average isotopic were small and sedimentary rocks were sparse, composition of carbonate sediments, and f is the inputs of carbon from continental weathering fraction of carbon buried as organic matter in may have been small relative to volcanic inputs. sediments. However, the several billion year sedimentary The observation that the proportion of carbon record of rocks rich in carbonate minerals buried in sediments as organic matter versus and organic matter suggests that continental Atmospheric O2 throughout Earth’s History 533 weathering must have formed a significant 8.11.6.3 The Proterozoic Atmosphere contribution to total oceanic carbon inputs fairly 8.11.6.3.1 Oxygenation of the Proterozoic early in the Archean. Intriguingly, this indicates that oxidative weathering of ancient sedimentary atmosphere rocks may have been active even in the Archean, Although there is evidence for the evolution of prior to accumulation of O2 in the atmosphere. oxygenic photosynthesis several hundred million This runs counter to traditional interpretations of years before the Huronian glaciation (e.g., Brocks geochemical carbon–oxygen cycling, in which et al., 1999) or earlier (Schidlowski, 1988, 2001), organic matter burial is equated to O2 production high fluxes of UV light reacting with O2 derived and organic matter weathering is equated to O2 from the earliest photosynthetic organisms would consumption. have created dangerous reactive oxygen species that severely suppressed widespread development of large populations of these oxygenic photoauto- 8.11.6.2.4 Evidence for oxygenic trophs. Cyanobacteria, as photoautotrophs, need photosynthesis in the Archean be exposed to visible light and have evolved Fossil evidence for photosynthetic organisms several defense mechanisms to protect cell con- from the same time period can be traced to the tents and repair damage. However, two key existence of stromatolites (Schopf, 1992, 1993; metabolic pathways (oxygenic photosynthesis Schopf et al., 2002; Hofmann et al., 1999), and nitrogen fixation) are very sensitive to UV although evidence for these early oxygenic damage. Indisputably cyanobacterial fossil occur photosynthetic communities is debated (Buick, in 1,000 Ma rocks, with putative fossils occurring 1990; Brasier et al., 2002). Stromatolites are 2,500 Ma and possibly older (Schopf, 1992). laminated sediments consistently of alternating Sediment mat-forming cyanobacteria and stroma- organic matter-rich and organic matter-lean tolites are least ambiguous and oldest. Terrestrial layers; the organic matter-rich layers are largely encrusting cyanobacteria are only known in the composed of filamentous cyanobacteria. Stroma- Phanerozoic. Planktonic forms are not known for tolites and similar mat-forming cyanobacterial the Archean and early Proterozoic. They may not crusts occur today in select restricted shallow have existed, or they may not be preserved. marine environments. Fossil evidence from many Molecular evolution, specifically coding for locales in Archean rocks (Greenland, Australia, proteins that build gas vesicles necessary for South Africa) and Proterozoic rocks (Canada, planktonic life, are homologous and conserved in Australia, South America) coupled with carbon all cyanobacteria. Thus, perhaps planktonic cya- isotope geochemistry provides indirect evidence nobacteria existed throughout Earth’s history for the evolution of oxygenic photosynthesis early since the late Archean. in Earth’s history (,3.5 Ga). Today, ozone forms in the stratosphere by Other evidence comes from molecular fossils. reaction of O2 with UV light. This effectively All cyanobacteria today are characterized by the screens much incoming UV radiation. Prior to the presence of 2a-methylhopanes in their cell accumulation of atmospheric O2, no ozone could membranes. Brocks et al., (1999) have demon- form, and thus the UV flux to the Earth’s surface strated the existence of this taxon-specific bio- would be much greater (with harmful effects on marker in 2.7 Ga Archean rocks of the Pilbara DNA and proteins, which adsorb and are altered Craton in NW Australia. Also in these rocks is by UV). A significant ozone shield could develop 22 found a homologous series of C27 to C30 steranes. at ,10 PAL O2 (Kasting, 1987). However, a Steranes today derive from sterols mainly pro- fainter young Sun would have emitted somewhat duced by organisms in the domain Eukarya. lower UV, mediating the lack of ozone. Although Eukaryotes are obligate aerobes that require seawater today adsorbs most UV light by 6–25 m molecular O2, and thus Brocks and colleagues (1% transmittance cutoff), seawater with abun- argue that the presence of these compounds dant dissolved Fe2þ may have provided an provides strong evidence for both oxygenic effective UV screen in the Archean (Olson and photosynthesis and at least localized utilization Pierson, 1986). Also, waters rich in humic of accumulated O2. Of course, coexistence of two materials, such as modern coastal oceans, are traits within a biological lineage today reveals nearly UV opaque. If Archean seawater contained nothing about which evolved first. It is uncertain DOM, this could adsorb some UV. Iron oxidation whether 2a-methylhopane lipid biosynthesis pre- and BIF at ,2.5–1.9 Ga would have removed ceded oxygenic photosynthesis among cyanobac- the UV screen in seawater. Thus, a significant teria. Sterol synthesis certainly occurs in modern UV stress may have developed at this time. prokaryotes (including some anaerobes), but no This would be mediated coincidentally by existing lineage produces the distribution of accumulation of atmospheric O2. steranes found in the Brocks et al., (1999) study In the early Archean before oxygenic photo- except eukaryotes. synthesis evolved, cyanobacteria were limited. 534 The Global Oxygen Cycle Planktonic forms were inhibited, and limited by Lomagundi event (,2.3–2.0 Ga) is recorded in dissolved iron content of water, and existence of the sediment record as a prolonged period of stratified, UV-screen refuges. Sedimentary mat- elevated seawater carbonate d 13C, with carbon- forming and stromatolite-forming commu- ate d 13C values reaching nearly 10‰ in several nities were much more abundant. Iron-oxide sections around the world (Schidlowski, 2001). precipitation and deposition of screen enzymes This represents an extended period of time may have created UV-free colonies under a shield (perhaps several hundred million years) during even in shallow waters. which removal of carbon from the “ocean þ Advent of oxygenic photosynthesis in the atmosphere” system as organic matter greatly Archean generated small oxygen oases (where exceeded supply. By implication, release of O2 dissolved O2 could accumulate in the water) through photosynthesis was greatly accelerated within an overall O2-free atmosphere waters during this time. containing oxygenic photoautotrophs might have A third mechanism for oxygenation of the reached 10% air saturation (Kasting, 1992). At atmosphere at ,2.3 Ga relies on the slow, gradual this time, both unscreened UV radiation and O2 oxidation of the Earth’s crust. Irreversible H2 may have coincided within the water column escape and basalt-seawater reactions led to a and sediments. This would lead to increased UV gradual increase in the amount of oxidized and stress for cyanobacteria. At the same time, hydrated minerals contained in the Earth’s crust precipitation of iron from the water would and subducted in subduction zones throughout the make the environment even more UV-transpar- Archean. Gradually this influenced the oxidation ent. To survive, cyanobacteria would need to state of volcanic gases derived in part from evolve and optimize defense and repair mech- subducted crustal rocks. Thus, although mantle anisms for UV damage (Garcia-Pichel, 1998). oxygen fugacity may not have changed since the Perhaps this explains the ,500 Myr gap between early Archean, crustal and volcanic gas oxygen origin of cyanobacteria and accumulation of O2 fugacity slowly increased as the abundance of in the atmosphere. For example, the synthesis of oxidized and hydrated crust increased (Holland, scytonemin (a compound found exclusively in 2002; Kasting et al., 1993; Kump et al., 2001). cyanobacteria) requires molecular oxygen Although slow to develop, Holland (2002) f (implying evolution in an oxic environment); it estimates that an increase in O2 of less than 1 log optimally screens UV-a, the form of UV unit is all that would have been required for radiation only abundant in an oxygenated transition from an anoxic to an oxic atmosphere, atmosphere. assuming rates of oxygenic photosynthesis con- Oxygenation of the atmosphere at ,2.3– sistent with modern systems and the sediment f 2.0 Ga may derive from at least three separate isotope record. Once a threshold volcanic gas O2 causes. First, discussed earlier, is the titration of had been reached, O2 began to accumulate. O2 with iron, sulfur, and reduced gases. The other Oxidative weathering of sulfides released large is global rates of photosynthesis and organic amounts of sulfate into seawater, facilitating carbon burial in sediments. If the rate of bacterial sulfate reduction. atmospheric O2 supply (oxygenic photosynthesis) There are several lines of geochemical evidence exceeds all mechanisms of O2 consumption that suggest a rise in oxygenation of the (respiration, chemoautotrophy, reduced mineral atmosphere ,2.3–2.0 Ga, beyond the coincident oxidation, etc.), then O2 can accumulate in the last occurrence of BIFs (with one late Proterozoic atmosphere. One means by which this can be exception) and first occurrence of red beds evaluated is through seawater carbonate recognized decades ago, and carbon isotopic 13 d C. Because biological CO2 fixation is associ- evidence suggesting ample burial of sedimentary ated with significant carbon isotope discrimi- organic matter (Karhu and Holland, 1996; Bekker nation, the magnitude of carbon fixation is et al., 2001; Buick et al., 1998). The Huronian indicated by the isotopic composition of seawater glaciation (,2.3 Ga) is the oldest known glacial carbonate. At times of more carbon fixation and episode recorded in the sedimentary record. One burial in sediments, relatively more 12Cis interpretation of this glaciation is that the cooler removed from the atmosphere þ ocean inorganic climate was a direct result of the rise of carbon pool than is supplied through respiration, photosynthetically derived O2. The rise of O2 organic matter oxidation, and carbonate mineral scavenged and reacted with the previously high dissolution. Because carbon fixation is dominated atmospheric concentration of CH4. Methane by oxygenic photosynthesis (at least since the late concentrations dropped, and the less-effective Archean), periods of greater carbon fixation and greenhouse gas product CO2 could not maintain burial of organic matter in sediments (observed equable surface temperatures. Kasting et al., 13 as elevated seawater carbonate d C) are equated (1983) estimate that a rise in pO above 24 2 to periods of elevated O2 production through ,10 atm resulted in loss of atmospheric CH4 oxygenic photosynthesis. The early Proterozoic and onset of glaciation. Atmospheric O2 throughout Earth’s History 535 Paleosols have also provided evidence for a achieve an increase in oxidation state once change in atmosphere oxygenation at some time seawater sulfate concentrations increased. between 2.3–2.0 Ga. Evidence from rare earth A strong model for oxygenation of the element enrichment patterns and U/Th fraction- atmosphere has developed based largely on the ation suggests a rise in O2 to ,0.005 bar by the sulfur isotope record and innovations in microbial time of formation of the Flin Flon paleosol of metabolism. The classical interpretation of the Manitoba, Canada, 1.85 Ga (Pan and Stauffer, disappearance of BIFs relates to the rise of 2000; Holland et al., 1989; Rye and Holland, atmospheric O2, oxygenation of the oceans, and 1998). Rye and Holland (1998) examined several removal of dissolved ferrous-iron by oxidation. early Proterozoic paleosols and observed that However, another interpretation has developed, negligible iron loss is a consistent feature from based largely on the evolving Proterozoic sulfur soils of Proterozoic age through the present. These isotope record. During the oxygenation of the atmosphere at ,2.3–2.0 Ga, the oceans may not authors estimate that a minimum pO2 of .0.03 atm is required to retain iron during soil have become oxidized, but instead remained anoxic and became strongly sulfidic as well formation, and thus atmospheric O2 concentration has been 0.03 or greater since the early Paleozoic. (Anbar and Knoll, 2002; Canfield, 1998 and However, re-evaluation of a paleosol crucial to the references therein). Prior to ,2.3 Ga, the oceans argument of iron depletion during soil formation were anoxic but not sulfidic. Ferrous-iron was under anoxia, the Hekpoort paleosol dated at abundant, as was manganese, because both are 2.2 Ga, has revealed that the iron depletion very soluble in anoxic, sulfide-free waters. The detected by previous researchers may in fact be high concentration of dissolved iron and manga- the lower zone of a normal oxidized lateritic soil. nese facilitated nitrogen fixation by early cyano- Upper sections of the paleosol that are not bacteria, such that available nitrogen was depleted in iron have been eroded away in the abundant, and phosphorus became the nutrient exposure examined by Rye and Holland (1998), limiting biological productivity (Anbar and but have been found in drill core sections. The Knoll, 2002). Oxygenation of the atmosphere at , depletion of iron and occurrence of ferrous-iron 2.3 Ga led to increased oxidative weathering of sulfide minerals on the continents and increased minerals in the lower sections of this paleosol sulfate concentration in seawater. Bacterial have been reinterpreted by Beukes et al., (2002) to sulfate reduction generated ample sulfide, and indicate an abundant soil surface biomass at the in spite of limited mixing, the deep oceans would time of deposition that decomposed to generate have remained anoxic and now also sulfidic as reducing conditions and iron mobilization during p , long as O2 remained below 0.07 atm (Canfield, the wet season, and precipitation of iron oxides 1998), assuming reasonable rates of primary during the dry season. production. Both iron and manganese form Sulfur isotope studies have also provided insoluble sulfides, and thus were effectively insights into the transition from Archean low scavenged from seawater once the oceans became p to higher values in the Proterozoic. In the O2 sulfidic. Thus, the Proterozoic oxygenation led to same studies that revealed extremely low significant changes in global oxygen balances, Archean ocean sulfate concentrations, it was with the atmosphere and ocean mixed layer found that by ,2.2 Ga, isotopic compositions of becoming mildly oxygenated (probably sedimentary sulfates and sulfides indicate bac- ,0.01 atm O2), and the deep oceans becoming terial sulfate reduction under more elevated strongly sulfidic in direct response to the rise of seawater sulfate concentrations compared with atmospheric O2. the sulfate-poor Archean (Habicht et al., 2002; In addition to increased oxygen fugacity of Canfield et al., 2000). As described above, volcanic gases, and innovations in biological nitrogen isotope ratios in sedimentary kerogens productivity to include oxygenic photosynthesis, show a large and permanent shift at ,2.0 Ga, the oxygenation of the Proterozoic atmosphere consistent with denitrification, significant sea- may be related to large-scale tectonic cycles. water nitrate concentrations, and thus available There are several periods of maximum deposition atmospheric O2. of sedimentary rocks rich in organic matter Prior to ,2.2 Ga, low seawater sulfate concen- through geologic time. These are ,2.7 Ga, trations would have limited precipitation and 2.2 Ga, 1.9 Ga, and 0.6 Ga (Condie et al., 2001). subduction of sulfate-bearing minerals. This The increased deposition of black shales at 2.7 Ga would maintain a lower oxidation state in volcanic and 1.9 Ga are associated with superplume events: gases derived in part from recycled crust (Holland, highly elevated rates of seafloor volcanism, 2002). Thus, even while the oxidation state of the oceanic crust formation. Superplumes lead to mantle has remained constant since ,4.0 Ga increased burial of both organic matter and (Delano, 2001), the crust and volcanic gases carbonates (through transgression, increased derived from subduction of the crust could only atmospheric CO2, accelerated weathering, and 536 The Global Oxygen Cycle nutrient fluxes to the oceans), with no net effect on decoupling is a natural result of a shift in the carbonate isotopic composition. Thus, periods of source of the biological limiting nutrient from increased absolute rates of organic matter burial phosphorus (derived from continental weather- and O2 production may be masked by a lack ing) to nitrogen (limited by N2 fixation). of carbon isotopic signature. Breakup of Furthermore, the isotopic composition of carbon- supercontinents may be related to the black shale ates throughout the Mesoproterozoic is 1–2‰ depositional events at 2.2 Ga and 0.6 Ga. Breakup depleted relative to average carbonates from the of supercontinents may lead to more sediment early and late Proterozoic and Phanerozoic. This accommodation space on continental shelves, as is consistent with a decrease in the relative well as accelerated continental weathering and proportion of carbon buried as organic matter delivery of nutrients to seawater, fertilizing versus carbonate during this time. It appears that primary production and increasing organic matter after initial oxygenation of the atmosphere from p burial. These supercontinent breakup events at completely anoxic to low O2 in the early 2.2 Ga and 0.6 Ga are clearly observed on the Proterozoic, further oxygenation was halted for carbonate isotopic record as increases in relative several hundred million years. burial of organic matter versus carbonate. Model- Global-scale reinvigoration of primary pro- ing efforts examining the evolution of the carbon, duction, organic matter burial and oxygenation sulfur, and strontium isotope records have shown of the atmosphere may be observed in the latest that gradual growth of the continents during Mesoproterozoic. Shifts in carbonate d 13Cofup the Archean and early Proterozoic may in fact to 4‰ are observed in sections around the globe at play a very large role in controlling the onset ,1.3 Ga (Bartley et al., 2001; Kah et al., 2001). of oxygenation of the atmosphere, and that These positive carbon isotope excursions are biological innovation may not be directly associated with the formation of the Rodinian coupled to atmospheric evolution (Godderı´s and supercontinent, which led to increased continental Veizer, 2000). margin length, orogenesis, and greater sedimen- tation (Bartley et al., 2001). The increased organic matter burial and atmospheric oxygenation associ- ated with this isotope excursion may be related to 8.11.6.3.2 Atmospheric O during 2 the first occurrence of laterally extensive CaSO4 the Mesoproterozoic evaporites (Kah et al., 2001). Although the rapid 34 The sulfur-isotope-based model of Canfield 10‰ increase in evaporate d S across these and colleagues and the implications for limiting sections is taken to indicate a much reduced nutrient distribution proposed by Anbar and seawater sulfate reservoir with much more rapid Knoll (2002) suggest that for nearly thousand turnover times than found in the modern ocean, million years (,2.2–1.2 Ga), oxygenation of the the isotope fractionation between evaporate sul- p , fates and sedimentary sulfides indicates that the atmosphere above O2 0.01 atm was held in stasis. Although oxygenic photosynthesis was oceans were not sulfate-limiting. Atmospheric O2 active, and an atmospheric ozone shield had was of sufficient concentration to supply ample developed to protect surface-dwelling organisms sulfate for bacterial sulfate reduction throughout from UV radiation, much of the deep ocean was the Mesoproterozoic. still anoxic and sulfidic. Removal of iron and manganese from seawater as sulfides generated severe nitrogen stress for marine communities, 8.11.6.3.3 Neoproterozoic atmospheric O suggesting that productivity may have been 2 limited throughout the entire Mesoproterozoic. Elevated carbonate isotopic compositions The sluggish but consistent primary productivity (,4‰) and strong isotope fractionation between and organic carbon burial through this time is sulfate and biogenic sulfide minerals through the seen in the carbonate isotope record. For several early Neoproterozoic indicates a period of hundred million years, carbonate isotopic com- several hundred million years of elevated bio- position varied by no more than ^2‰, revealing logical productivity and organic matter burial. very little change in the relative carbonate/ This may relate in part to the oxygenation of the organic matter burial in marine sediments. atmosphere, ammonia oxidation, and increased Anbar and Knoll (2002) suggest that this seawater nitrate availability. Furthermore, oxi- indicates a decoupling of the link between dative weathering of molybdenum-bearing sul- tectonic events and primary production, because fide minerals, and greater oxygenation of the although variations in tectonic activity (and oceans increased availability of molybdenum associated changes in sedimentation, generation necessary for cyanobacterial pathways of of restricted basins, and continental weathering) N2-fixation (Anbar and Knoll, 2002). Much of occurred during the Mesoproterozoic, these are the beginning of the Neoproterozoic, like the not observed in the carbonate isotope record. This late Mesoproterozoic, saw gradual increases in Atmospheric O2 throughout Earth’s History 537 oxygenation of the atmosphere, and possibly of d 13C (‰) the surface oceans. Limits on the oxygenation of the atmosphere are provided by sulfur isotopic m –50510 and molecular evidence for the evolution of 40 sulfide oxidation and sulfur disproportionation. The increase in sulfate–sulfide isotope fraction- 30 ation in the Neoproterozoic (,1.0–0.6Ga) reflects a shift in sulfur cycling from simple 20 7 one-step reduction of sulfate to sulfide, to a system in which sulfide was oxidized to sulfur intermediates such as thiosulfate or elemental 10 sulfur, which in turn were disproportionated into sulfate and sulfide (Canfield and Teske, 1996). 0 Ghaub Glaciation Sulfur disproportionation is associated with 0 significant isotope effects, generating sulfide that is substantially more 34S-depleted than can 6 be achieved through one-step sulfate reduction. –10 The sulfur isotope record reveals that an increase in sulfate–sulfide isotope fractionation occurred –20 between 1.0 Ga and 0.6 Ga, consistent with evolution of sulfide-oxidation at this time as –30 derived from molecular clock based divergence of non-photosynthetic sulfide-oxidizing bacteria Ombaat JIE FM Maieberg CAP (Canfield and Teske, 1996). These authors –40 estimated that innovation of bacterial sulfide oxidation occurred when much of the coastal –50 shelf sediment (,200 m water depth) was –5 0510 exposed to water with 13–46 mMO2, which p Figure 11 Composite section of carbonate isotope corresponds to 0.01–0.03 atm O2. Thus, after the initial oxygenation of the atmosphere in the early stratigraphy before and after the Ghaub glaciation from , , p Namibia. Enriched carbonates prior to glaciation Proterozoic ( 2.3–2.0 Ga) to 0.01 atm, O2 was constrained to this level until a second oxygen- indicate active primary production and organic matter burial. Successive isotopic depletion indicates a shut ation in the late Proterozoic (,1.0–0.6 Ga) to down in primary production as the world became 0.03 atm or greater. covered in ice (source Hoffman et al., 1998). During the last few hundred million years of the Proterozoic (,0.7–0.5 Ga), fragmentation of the Rodinian supercontinent was associated with at (Hoffman and Schrag, 2002), possibly accelerated least two widespread glacial episodes (Hoffman through high rates of OM burial, and reduced et al., 1998; Hoffman and Schrag, 2002). Neo- meridional Hadley cell heat transport, because proterozoic glacial deposits are found in Canada, tropical air masses were drier due to increased Namibia, Australia, and other locations worldwide continentality (Hoffman and Schrag, 2002). (Evans, 2000). In some of these, paleomagnetic Growth of initially polar ice caps would have evidence suggests that glaciation extended com- created a positive ice–albedo feedback such that pletely from pole to equator (Sumner, 1997; see greater than half of Earth’s surface became ice also Evans (2000) and references therein). The covered, global-scale growth of sea ice became “Snowball Earth”, as these events have come to be inevitable (Pollard and Kasting, 2001; Baum and known, is associated with extreme fluctuations in Crowley, 2001). Kirschvink (1992) recognized carbonate isotopic stratigraphy, reoccurrence of that escape from the Snowball Earth becomes BIFs after an ,1.0 Ga hiatus, and precipitation of possible because during extreme glaciation, the enigmatic, massive cap carbonate sediments continental hydrologic cycle would be shut down. immediately overlying the glacial deposits He proposed that sinks for atmospheric CO2, (Figure 11). It has been proposed that the particular namely, photosynthesis and silicated weathering, configuration of continents in the latest Neopro- would have been eliminated during the glaciation. terozoic, with land masses localized within the Because volcanic degassing continued during the middle–low latitudes, lead to cooling of climate glaciation, CO2 in the atmosphere could rise to through several mechanisms. These include very high concentrations. It was estimated that an p higheralbedointhesubtropics(Kirschvink, increase in CO2 to 0.12 bar would be sufficient to 1992), increased silicate weathering as the bulk induce a strong enough greenhouse effect to begin of continents were located in the warm tropics, warming the planet and melting the ice (Caldeira resulting in a drawdown of atmospheric CO2 and Kasting, 1992). Once meltback began, it 538 The Global Oxygen Cycle would subsequently accelerate through the eukaryotes and other organisms dependent on positive ice–albedo feedback. Intensely warm aerobic respiration. Obviously, oxygenated refu- temperatures would follow quickly after the gia must have existed, because the rise of many glaciation, until accelerated silicate weathering eukaryotes predates the Neoproterozoic Snowball under a reinvigorated hydrologic cycle could events (Butterfield and Rainbird, 1998; Porter and p consume excess CO2 and restore CO2 to more Knoll, 2000). Nonetheless, carbon isotopic evi- equable values. Carbonate mineral precipitation dence leading up to and through the intense glacial was inhibited during the global glaciation, so intervals suggests that overall biological pro- seawater became enriched in hydrothermally ductivity was severely repressed during Snowball derived cations. At the end of glaciation, mixing events. Carbonate isotope compositions fall from of cation-rich seawater with high alkalinity extremely enriched values to depletions of surface runoff under warming climates led to the ,(26‰) at glacial climax (Hoffman et al., precipitation of massive cap carbonates. 1998; Kaufman et al., 1991; Kaufman and Geochemical signatures before and after the Knoll, 1995). These carbonate isotope values Snowball episodes reveal rapid and short-lived reflect primordial (volcanogenic) carbon inputs, changes in global biogeochemical cycles that and indicate that effectively zero organic matter impacted O2 concentrations in the atmosphere and was buried in sediments at this time. Sustained oceans. Prior to the Snowball events, the Neopro- lack of organic matter burial and limited if any terozoic ocean experienced strong primary pro- oxygenic photosynthesis under the glacial ice for duction and organic matter burial, as seen in the several million years would have maintained an carbon isotopic composition of Neoproterozoic anoxic ocean, depleted in sulfate, with low sulfide seawater in which positive isotope excursions concentrations due to iron scavenging, and reached 10‰ in some sections (Kaufman and extremely high alkalinity and hydrothermally- Knoll, 1995). These excursions are roughly derived ion concentrations. Above the ice, the coincident with increased oxygenation of the initially mildly oxygenated atmosphere would atmosphere in the Neoproterozoic (Canfield and slowly lose its O2. Although the hydrologic cycles Teske, 1996; Des Marais et al., 1992). Although were suppressed, and thus oxidative weathering of sulfur isotopic evidence suggests that much of the sulfide minerals and organic matter exposed on ocean may have become oxygenated during the the continents was inhibited, the emission of Neoproterozoic, it is very likely that this was a reduced volcanic gases would have been enough temporary phenomenon. The global-scale glacia- to fully consume 0.01 bar O within several tions of the late Neoproterozoic would have 2 hundred thousand years. Thus, it is hypothesized driven the oceans to complete anoxia through that during Snowball events, the Earth’s atmos- several mechanisms (Kirschvink, 1992). Ice cover inhibited air–sea gas exchange, and thus surface phere returned to pre-Proterozoic anoxic con- waters and sites of deep-water formation were cut ditions for at least several million years. Gradual increases in carbonate d 13C after glaciation and off from atmospheric O2 supplies. Intense oxi- dation of organic matter in the water column and deposition of cap carbonates suggests slow sediment would have quickly consumed any restored increases in primary productivity and organic carbon burial. available dissolved O2. Extreme positive sulfur- isotope excursions associated with snowball- Carbon isotopic evidence suggests that reox- succession deposits reflect nearly quantitative, ygenation, at least of seawater, was extremely closed-system sulfate reduction in the oceans gradual throughout the remainder of the Neopro- during glaciation (Hurtgen et al., 2002). As terozoic. Organic matter through the terminal seawater sulfate concentrations were reduced, Proterozoic derives largely from bacterial hydrothermal inputs of ferrous-iron exceeded heterotrophs, particularly sulfate reducing bac- sulfide supply, allowing BIFs to form (Hoffman teria, as opposed to primary producers (Logan and Schrag, 2002). Closed-system sulfate et al., 1995). These authors suggested that reduction and iron formations require ocean throughout the terminal Neoproterozoic, anaero- anoxia throughout the water column, although bic heterotrophy dominated by sulfate reduction restricted oxygenic photosynthesis beneath thin was active throughout the water column, and O2 tropical sea ice may be responsible for ferrous- penetration from surface waters into the deep iron oxidation and precipitation (Hoffman and ocean was inhibited. Shallow-water oxygen- Schrag, 2002). deficient environments became widespread at Once global ice cover was achieved, it is the Precambrian–Cambrian boundary (Kimura estimated to have lasted several million years, and Watanabe, 2001), corresponding to negative based on the amount of time required to carbonate d 13C excursions and significant bio- accumulate 0.12 bar CO2 at modern rates of logical evolution from Ediacaran-type metazoans volcanic CO2 emission (Caldeira and Kasting, to emergence of modern metazoan phyla in the 1992). This must have placed extreme stress on Cambrian. Atmospheric O2 throughout Earth’s History 539

8.11.6.4 Phanerozoic Atmospheric O2 species to species, and is probably impossible to reconstruct for extinct lineages, but modern 8.11.6.4.1 Constraints on Phanerozoic infaunal and epifaunal metazoans can accommo- O2 variation date dissolved O2 dropping to the tens of Oxygenation of the atmosphere during the micromolar in concentration. If these concen- latest Neoproterozoic led to a fairly stable and trations are extrapolated to equilibrium of the well-oxygenated atmosphere that has persisted atmosphere with well-mixed cold surface waters, , p through the present day. Although direct they correspond to 0.05 atm O2. Although local measurement or a quantifiable proxy for Phaner- and regional anoxia occurred in the oceans at p particular episodes through Phanerozoic time, the ozoic paleo- O2 concentrations have not been reported, multiple lines of evidence point to continued presence of aerobic metazoans suggests that widespread or total ocean anoxia did not upper and lower limits on the concentration of O2 occur during the past ,600 Myr, and that p has in the atmosphere during the past several hundred O2 million years. Often cited is the nearly continuous been maintained at or above ,5% O2 since the record of charcoal in sedimentary rocks since the early Paleozoic. evolution of terrestrial plants some 350 Ma. The presence of charcoal indicates forest fires 8.11.6.4.2 Evidence for variations throughout much of the Phanerozoic, which are in Phanerozoic O2 unlikely to have occurred below pO2 ¼ 0.17 atm (Cope and Chaloner, 1985). Combustion and Several researchers have explored possible p sustained fire are difficult to achieve at lower O2. links between the final oxygenation of the atmos- The existence of terrestrial plants themselves also phere and explosion of metazoan diversity at the provides a crude upper bound on pO2 for two Precambrian–Cambrian boundary (McMenamin reasons. Above the compensation point of and McMenamin, 1990; Gilbert, 1996).While the pO2/pCO2 ratios, photorespiration outcompetes Cambrian explosion may not record the origin of photosynthesis, and plants experience zero or these phyla in time, this boundary does record the negative net growth (Tolbert et al., 1995). development of large size and hard skeletons Although plants have developed various adaptive required for fossilization (Thomas, 1997). Indeed, strategies to accommodate low atmospheric pCO2 molecular clocks for diverse metazoan lineages or aridity (i.e. C4 and CAM plants), net terrestrial trace the origin of these phyla to ,400 Myr before photosynthesis and growth of terrestrial ecosys- the Cambrian explosion (Doolittle et al., 1996; tems are effectively inhibited if pO2 rises too high. Wray et al., 1996). The ability of lineages to This upper bound is difficult to exactly constrain, develop fossilizable hardparts may be linked with but is likely to be ,0.3–0.35 atm. Woody tissue increasing oxygenation during the latest Precam- is also extremely susceptible to combustion at p brian. Large body size requires elevated O2 and high pO2, even if the tissue is wet. Thus, terrestrial dissolved O2 concentrations, so the diffusion can ecosystems would be unlikely above some upper supply O2 to internal tissues. Large body size limit of pO2 because very frequent reoccurrence provides several advantages, and may have of wildfires would effectively wipe out terrestrial evolved rather quickly during the earliest Cam- plant communities, and there is no evidence of brian (Gould, 1995), but large size also requires this occurring during the Phanerozoic on a global greater structural support. The synthesis of scale. What this upper pO2 limit is, however, collagen (a ubiquitous structural protein among remains disputed. Early experiments used combus- all metazoans and possible precursor to inorganic tion of paper under varying humidity and pO2 structural components such as carbonate or (Watson, 1978) but paper may not be the most phosphate biominerals) requires elevated O2. appropriate analog for inception of fires in woody The threshold for collagen biosynthesis, and tissue with greater moisture content and thermal associated skeletonization and development of thickness (Robinson, 1989; Berner et al., 2002). large body size, could not occur until pO2 reached Nonetheless, the persistence of terrestrial plant some critical threshold some time in the latest communities from the middle Paleozoic through Precambrian (Thomas, 1997). p the present does imply that O2 concentrations There is evidence to suggest that the late have not risen too high during the past 350 Myr. p Paleozoic was a time of very elevated O2,to Prior to evolution of land plants, even circum- concentrations substantially greater than observed p stantial constraints on early Paleozoic O2 are in the modern atmosphere. Coals from the difficult to obtain. Invertebrate metazoans, which Carboniferous and Permian contain a greater have a continous fossil record from the Cambrian abundance of fusain, a product of woody tissue on, require some minimum amount of dissolved combustion and charring, than observed for any oxygen to support their aerobic metabolism. The period of the subsequent geologic past (Robinson, absolute minimum dissolved O2 concentration 1989, 1991; Glasspool, 2000), suggesting more able to support aerobic metazoans varies from abundant forest fires and by implication possibly 540 The Global Oxygen Cycle

higher pO2 at this time. Less ambiguous is the Devonian black shales of the Illinois, Michigan, biological innovation of gigantism at this time and Appalachian Basins (USA). Widespread among diverse arthropod lineages (Graham et al., burial of organic matter in the late Devonian has 1995). All arthropods rely on tracheal networks been linked to increased fertilization of the surface for diffusion of O2 to support their metabolism; waters through accelerated continental weathering active pumping of O2 through a vascular system due to the rise of terrestrial plant communities as found in vertebrates does not occur. (Algeo and Scheckler, 1998). High rates of This sets upper limits on body size for a given photosynthesis with relatively small rates of p O2 concentration. In comparison with arthropod global respiration led to accumulation of organic communities of the Carboniferous and Permian, matter in marine sediments, and the beginnings of modern terrestrial arthropods are rather small. a pulse of atmospheric hyperoxia that extended Dragonflies at the time reached 70 cm wingspan, through the Carboniferous into the Permian. mayflies reached 45 cm wingpans, millipedes Coalescence of continental fragments to form reached 1 m in length. Even amphibians, which the Pangean supercontinent at this time led to depend in part on diffusion of O2 through their widespread circulation-restricted basins that skin for aerobic respiration, reached gigantic facilitated organic matter burial and net oxygen size at this time. These large body sizes could release, and later, to extensive infilling to generate not be supported by today’s 21% O2 atmosphere, near-shore swamps containing terrestrial veg- p and instead require elevated O2 between 350 Ma etation which was often buried to form coal and 250 Ma. Insect taxa that were giants in the deposits during the rapidly fluctuating sea levels Carboniferous do not survive past the Permian, of the Carboniferous and Permian. suggesting that declining pO2 concentrations in The largest Phanerozoic extinction occurred at the Permian and Mesozoic led to extinction the end of the Permian (,250 Ma). A noticeable (Graham et al., 1995; Dudley, 1998). decrease in the burial of organic matter in marine Increases in atmospheric O2 affect organisms in sediments across the Permian–Triassic boundary several ways. Increased O2 concentration facili- may be associated with a global decline in primary tates aerobic respiration, while elevated pO2 productivity, and thus, with atmospheric pO2.The p against a constant N2 increases total atmospheric gigantic terrestrial insect lineages, thought to p pressure, with associated changes in atmospheric require elevated O2, do not survive across this gas density and viscosity (Dudley, 1998, 2000). In boundary, further suggesting a global drop in pO2, tandem these effects may have played a strong role and the sedimentary and sulfur isotope records in the innovation of insect flight in the Carbon- indicate an overall increase in sulfate reduction and iferous (Dudley, 1998, 2000), with secondary burial of pyritic shales (Berner, 2002; Beerling and peaks in the evolution of flight among birds, bats Berner, 2000). Although a long-duration deep-sea and other insect lineages corresponding to times anoxic event has been proposed as a cause for the of high O2 in the late Mesozoic (Dudley, 1998). Permian mass extinction, there are competing

In spite of elevated pO2 in the late Paleozoic, models to explain exactly how this might have leading up to this time was an extended period of occurred. Hotinski et al., (2001) has shown that water column anoxia and enhanced burial of while stagnation of the water column to generate organic matter in the Devonian. Widespread deep-water anoxia might at first seem attractive, deposition of black shales, fine-grained laminated global thermohaline stagnation would starve the sedimentary rocks rich in organic matter, during oceans of nutrients, extremely limiting primary the Devonian indicate at least partial stratification productivity, and thus shutting down dissolved O2 of several ocean basins around the globe, with demand in the deep oceans. Large negative oxygen deficiency throughout the water column. excursions in carbon, sulfur, and strontium iso- One example of this is from the Holy Cross topes during the late Permian may indicate Mountains of Poland, from which particular stagnation and reduced ventilation of seawater for molecular markers for green sulfur bacteria have extended periods, coupled with large-scale over- been isolated (Joachimski et al., 2001). These turn of anoxic waters. Furthermore, sluggish organisms are obligately anaerobic chemopho- thermohaline circulation at this time could derive toautotrophs, and indicate that in the Devonian from a warmer global climate and warmer water at basin of central Europe, anoxia extended upwards the sites of high-latitude deep-water formation through the water column well into the photic (Hotinski et al., 2001). The late Permian paleogeo- zone. Other black shales that indicate at least graphy of one supercontinent (Pangea) and one episodic anoxia and enhanced organic matter superocean (Panthalassa) was very different from burial during the Devonian are found at several the arrangement of continents and oceans on the sites around the world, including the Exshaw modern Earth. Coupled with elevated pCO2 at the Shale of Alberta (Canada), the Bakken Shale of time (Berner, 1994; Berner and Kothavala, 2001), the Williston Basin (Canada/USA), the Woodford GCM models predict warmer climate, weaker wind Shale in Oklahoma (USA), and the many stress, and low equator to pole temperature Atmospheric O2 throughout Earth’s History 541 gradients. Although polar deep-water formation (OAEs) are recognized from the Cretaceous in all still occurred, bringing O2 from the atmosphere to major ocean basins, suggesting possible global the deep oceans, anoxia was likely to develop at deep-ocean anoxia. Molecular markers of green mid-ocean depths (Zhang et al., 2001), and sulfur bacteria, indicating photic zone anoxia, thermohaline circulation oscillations between have been detected from Cenomanian–Turonian thermally versus salinity driven modes of circula- boundary section OAE sediments from the North tion were likely to develop. During salinity driven Atlantic (Sinninghe Damste´ and Ko¨ster, 1998). modes, enhanced bottom-water formation in Thepresenceofthesemarkers(namely, warm, salty low-latitude regions would limit isorenieratene, a diaromatic carotenoid accessory oxygenation of the deep ocean. Thus, although pigment used during anoxygenic photosynthesis) sustained periods of anoxia are unlikely to have indicates that the North Atlantic was anoxic and developed during the late Permian, reduced euxinic from the base of the photic zone (,50– oxygenation of deep water through sluggish 150 m) down to the sediment. High concentrations thermohaline circulation, coupled with episodic of trace metals scavenged by sulfide and an anoxia driven by low-latitude warm salty bottom- absence of bioturbation further confirm anoxia water formation, may have led to reoccurring throughout the water column. Because mid- episodes of extensive ocean anoxia over period of Cretaceous oceans were not highly productive, several million years. accelerated dissolved O2 demand from high rates Other researchers have invoked extraterrestrial of respiration and primary production cannot be causes for the End-Permian extinction and anoxia. the prime cause of these OAEs. Most likely, the Fullerenes (cage-like hydrocarbons that effec- warm climate of the Cretaceous led to low O2 tively trap gases during formation and heating) bottom waters generated at warm, high salinity have been detected in late Permian sediments regions of low-latitude oceans. External forcing, from southern China. The noble gas complement perhaps through Milankovic-related precession- in these fullerenes indicates an extraterrestrial driven changes in monsoon intensity and strength, origin, which has been interpreted by Kaiho et al., influenced the rate of salinity-driven deep-water (2001) to indicate an unrecognized bolide impact formation, ocean basin oxygenation, and OAE at the Permian–Triassic boundary. The abrupt formation (Wortmann et al., 1999). Sapropels are decrease in d 34S across this boundary (from 20‰ organic matter-rich layers common to late Cen- to 5‰) implies an enormous and rapid release of ozoic sediments of the eastern Mediterranean. 34S-depleted sulfur into the ocean–atmosphere They are formed through a combination of system. These authors propose that volatilized increased primary production in surface waters, bolide- and mantle-derived sulfur (,0‰) oxi- and increased organic matter preservation in the dized in air, consumed atmospheric and dissolved sediment likely to be associated with changes in O2, and generated severe oxygen and acid stress in ventilation and oxygenation of the deep eastern the oceans. Isotope mass balance estimates require Mediterranean basins (Stratford et al., 2000). The ,1019 mol sulfur to be released, consuming a well-developed OMZ located off the coast of 19 similar mass of oxygen. 10 mol O2 represents Southern California today may have been more some 10–40% of the total available inventory of extensive in the past. Variations in climate atmospheric and dissolved O2 at this time, affecting intensity of upwelling and primary removal of which led to immediate anoxia, as production, coupled with tectonic activity altering these authors propose. the depth of basins and height of sills along the Other episodes of deep ocean anoxia and California coast, have generated a series of anoxia- extensive burial of organic matter are known facies organic-matter-rich sediments along the from the Jurassic, Cretaceous, Miocene, and west coast of North America, beginning with Pleistocene, although these have not been linked the Monterey Shale and continuing through to the to changes in atmospheric O2 and instead serve as modern sediments deposited in the Santa Barbara examples of the decoupling between atmospheric and Santa Monica basins. and deep-ocean O2 concentrations through much As shown by the modeling efforts of Hotinski of the geologic past. Widespread Jurassic black et al., (2001) and Zhang et al., (2001), extensive shale facies in northern Europe (Posidonien deep ocean anoxia is difficult to achieve for Schiefer, Jet Rock, and Kimmeridge Clay) were extended periods of geologic time during the p deposited in a restricted basin on a shallow Phanerozoic when O2 were at or near modern continental margin. Strong monsoonal circulation levels. Thus, while localized anoxic basins are led to extensive freshwater discharge and a low- common, special conditions are required to salinity cap on basin waters during summer, and generate widespread, whole ocean anoxia such intense evaporation and antiestuarine circulation as observed in the Cretaceous. Deep-water during winter (Ro¨hl et al., 2001), both of which formation in highly saline low-latitude waters contributed to water column anoxia and black was likely in the geologic past when climates were shale deposition. Several oceanic anoxic events warmer and equator to pole heat gradients were 542 The Global Oxygen Cycle reduced. Low-latitude deep-water formation has a buried as organic matter. However, the simplified significant effect on deep-water oxygenation, not stoichiometry of (5) is applied for most geochem- entirely due to the lower O2 solubility in warmer ical models of C–S–O cycling. waters, but also the increased efficiency of nutrient If burial of organic matter equates to O2 release use and recycling in low-latitude surface waters to the atmosphere over long timescales, then the (Herbert and Sarmiento, 1991). If phytoplankton oxidative weathering of ancient organic matter in were 100% efficient at using and recycling sedimentary rocks equates to O2 consumption. nutrients, even with modern high-latitude modes This process has been called “georespiration” by of cold deep-water formation, the deep oceans some authors (Keller and Bacon, 1998). It can be would likely become anoxic. represented by Equation (6), the reverse of (5): Weathering of organic matter from rocks:

8.11.6.4.3 Numerical models of Phanerozoic O2 þ “CH2O” ! CO2 þ H2O ð6Þ oxygen concentration Both Equations (3) and (4) contain terms for Although photosynthesis is the ultimate source addition and removal of O2 from the atmosphere. Thus, if we can reconstruct the rates of burial and of O2 to the atmosphere, in reality photosynthesis and aerobic respiration rates are very closely weathering of OM into/out of sedimentary rocks coupled. If they were not, major imbalances in through time, we can begin to quantify sources and sinks for atmospheric O . The physical mani- atmospheric CO2,O2, and carbon isotopes would 2 result. Only a small fraction of primary production festation of this equation is the reaction of organic (from photosynthesis) escapes respiration in the matter with O2 during the weathering and erosion water column or sediment to become buried in of sedimentary rocks. This is most clearly seen in deep sediments and ultimately sedimentary rocks. the investigation of the changes in OM abundance This flux of buried organic matter is in effect and composition in weathering profiles developed “net photosynthesis”, or total photosynthesis on black shales (Petsch et al., 2000, 2001). minus respiration. Thus, while over timescales of In addition to the C–O system, the coupled C–S–O system has a strong impact on atmos- days to months, dissolved and atmospheric O2 may respond to relative rates of photosynthesis or res- pheric oxygen (Garrels and Lerman, 1984; Kump piration, on longer timescales it is burial of organic and Garrels, 1986; Holland, 1978, 1984). This is matter in sediments (the “net photosynthesis”) that through the bacterial reduction of sulfate to sulfide matters. Averaged over hundreds of years or using organic carbon substrates as electron longer, burial of organic matter equates to release donors. During bacterial sulfate reduction (BSR), OM is oxidized and sulfide is produced from of O2 into the “atmosphere þ ocean” system: sulfate. Thus, BSR provides a means of resupply- Photosynthesis: ing oxidized carbon to the “ocean þ atmosphere” 6CO2 þ 6H2O ! C6H12O6 þ 6O2 ð3Þ system without consuming O2. The net reaction for BSR shows that for every 15 mol of OM Respiration: consumed, 8 mol sulfate and 4 mol ferric-iron are C6H12O6 þ 6O2 þ 6CO2 þ 6H2O ð4Þ also reduced to form 4 mol of pyrite (FeS2): If for every 1,000 rounds of photosynthesis there 22 4FeðOHÞ3 þ 8SO4 þ 15 CH2O ! 4 FeS2 are 999 rounds of respiration, the net result is one þ 2 þ þð Þ2 ð Þ round of organic matter produced by photosyn- 15 HCO3 13 H2O OH 7 thesis that is not consumed by aerobic respiration. Oxidation of sulfate using organic substrates as The burial flux of organic matter in sediments electron donors provides a means of restoring represents this lack of respiration, and as such is a inorganic carbon to the ocean þ atmosphere net flux of O to the atmosphere. In geochemists’ 2 system without consuming free O2. Every 4 mol shorthand, we represent this by the reaction of pyrite derived from BSR buried in sediments represents 15 mol of O2 produced by photosyn- Burial of organic matter in sediments: thesis to generate organic matter that will not be CO2 þ H2O ! O2 þ “CH2O” ð5Þ consumed through aerobic respiration. In effect, pyrite burial equates to net release of O2 to the where “CH2O” is not formaldehyde, or even any atmosphere, as shown by (8), obtained by the specific carbohydrate, but instead represents addition of Equation (7) to (5): sedimentary organic matter. Given the elemental 22 composition of most organic matter in sediments 4FeðOHÞ3 þ 8SO4 þ 15CH2O and sedimentary rocks, a more reduced organic 2 2 ! 4FeS þ 15HCO þ 13H O þðOHÞ matter composition might be more appropriate, 2 3 2 i.e., C10H12O, which would imply release of þ15CO þ 15H O ! 15O þ 15“CH O” 12.5 mol O2 for every mole of CO eventually 2 2 2 2 Atmospheric O2 throughout Earth’s History 543

2 global-scale coupling C–S–O geochemistry. 4FeðOHÞ þ 8SO2 þ 15CO þ 2H O 3 4 2 2 First, the broad-scale features of Phanerozoic ! þ 2 þð Þ2 þ ð Þ 4FeS2 15HCO3 OH 15O2 8 O2 evolution were established. O2 concentrations in the atmosphere were low in the early Oxidative weathering of sedimentary sulfide Paleozoic, rising to some elevated pO levels minerals during exposure and erosion on the con- 2 during the Carboniferous and Permian (probably tinents results in consumption of O (Equation (9): 2 to a concentration substantially greater than 4FeS þ 15O þ 14H O ! 4FeðOHÞ today’s 0.21 bar), and then falling through the 2 2 2 3 Mesozoic and Cenozoic to more modern values. 22 þ þ 8SO4 þ 16H ð9Þ This model confirmed the suspicion that in contrast with the Precambrian, Phanerozoic O2 Just as for the C–O geochemical system, if we evolution was a story of relative stability can reconstruct the rates of burial and oxidative p through time, with no great excursions in O2. weathering of sedimentary sulfide minerals Second, by linking C–S–O cycles with sedi- through geologic time, we can use these to ments and specifically sedimentation rates, this estimate additional sources and sinks for atmos- model helped fortify the idea that a strong pheric O2 beyond organic matter burial and control on organic matter burial rates globally, weathering. and thus ultimately on release of O2 to the In total, then, the general approach taken in atmosphere, may be rates of sedimentation in modeling efforts of understanding Phanerozoic O2 near-shore environments. These authors variability is to catalog the total sources and sinks extended this idea to propose that the close for atmospheric O2, render these in the form of a linkage between sedimentation and erosion (i.e., rate of change equation in a box model, and the fact that global rates of sedimentation are integrate the changing O2 mass through time matched nearly exactly to global rates of implied by changes in sources and sinks: X sediment production—in other words, erosion) = ¼ may in fact be a stabilizing influence on dMO2 dt FO2 into the atmosphere X atmospheric O2 fluctuations. If higher sedimen-

2 FO2 out of the atmosphere tation rates result in greater burial of organic matter and pyrite, and greater release of O to ¼ F 2 burial of organic matter the atmosphere, at the same time there will be = þð15 8ÞFburial of pyrite greater rates of erosion on the continents, some

2 Fweathering of organic matter of which will involve oxidative weathering of ancient organic matter and/or pyrite. 2 ð15=8ÞF weathering of pyrite ð10Þ The other principal approach towards modeling One approach to estimate burial and weath- the Phanerozoic evolution of atmospheric O2 rests ering fluxes of organic matter and sedimentary on the isotope systematics of the carbon and sulfides through time uses changes in the relative sulfur geochemical cycles. The significant isotope abundance of various sedimentary rock types discriminations associated with biological fix- estimated over Phanerozoic time. Some sedi- ation of CO2 to generate biomass and with mentary rocks are typically rich in both organic bacterial reduction of sulfate to sulfide have matter and pyrite. These are typically marine been mentioned several times previously in this shales. In contrast, coal basin sediments contain chapter. Given a set of simplifications of the much organic carbon, but very low amounts of exogenic cycles of carbon, sulfur and oxygen, sedimentary sulfides. Non-marine coarse-grained these isotopic discriminations and the isotopic clastic sediments contain very little of either composition of seawater through time (d13C, organic matter or sedimentary sulfides. Berner d 34S) can be used to estimate global rates of and Canfield (1989) simplified global sedimen- burial and weathering of organic matter, sedi- tation through time into one of three categories: mentary carbonates, pyrite, and evaporative marine shales þ sandstone, coal basin sediments, sulfates (Figure 13). and non-marine clastic sediments. Using rock (i) The first required simplification is that the abundance estimates derived from the data of total mass of exogenic carbon is constant through Ronov and others (Budyko et al., 1987; Ronov, time (carbon in the oceans, atmosphere, and 1976), these authors estimated burial rates for sedimentary rocks). This of course neglects inputs organic matter and pyrite as a function of time of carbon and sulfur from volcanic activity and for the past ,600 Myr (Figure 12). Weathering metamorphic degassing, and outputs into the rates for sedimentary organic matter and pyrite mantle at subduction zones. However, if these were calculated as first order dependents on the fluxes into and out of the exogenic cycle are small total mass of sedimentary organic matter or enough (or have no effect on bulk crustal carbon pyrite, respectively. Although highly simplified, or sulfur isotopic composition), then this simpli- this model provided several new insights into fication may be acceptable. 544 The Global Oxygen Cycle

10

Org C Burial )

1 8 – mol Myr 18

– 6

4 Org C Burial (10

2

0 –600 –500 –400 –300 –200 –100 0 (a) Time (Myr) 2

Pyr S Burial

1.5 ) 1 – mol Myr 18 – 1

Pyr S Burial (10 0.5

0 –600 –500 –400 –300 –200 –100 0 (b) Time (Myr) 100 40

Oxygen 80 35

60 30 mol) 18 – 25 O2 (%) 40 20 Oxygen (10 15 20 10

5 OSD C PTR J K T 0 0 –600 –500 –400 –300 –200 –100 0 (c) Time (Myr)

Figure 12 Estimated organic carbon burial (a), pyrite sulfur burial (b), and atmospheric oxygen concentrations (c) through Phanerozoic time, derived from estimates of rock abundance and their relative organic carbon and sulfide content (source Berner and Canfield, 1989). Atmospheric O2 throughout Earth’s History 545 (ii) The second simplification is that the total strongly buffered by carbonate mineral precipi- mass of carbon and sulfur dissolved in seawater tation and dissolution, and thus it is unlikely that plus the small reservoir of atmospheric carbon extensive regions of the world ocean could have and sulfur gases remains constant through time. become significantly enriched or depleted in For carbon, this may be a realistic simplification. inorganic carbon during the geologic past. For Dissolved inorganic carbon in seawater is sulfur, this assumption may not be completely accurate. Much of the interpretation of sulfur isotope records (with implications for atmos- 5 pheric O2 evolution) in the Precambrian depend 13 on varying, but generally low dissolved sulfate 4 d C (‰) concentrations. Unlike carbonate, there is no 3 great buffering reaction maintaining stable sulfate 2 concentrations in seawater. And while the 1 sources of sulfate to seawater have likely varied 0 only minimally, with changes in the sulfate flux to seawater increasing or decreasing smoothly –1 through time as the result of broad-scale tectonic –2 activity and changes in bulk continental weath- ering rates, removal of sulfate through excessive 34S 30 d BSR or rapid evaporate formation may be much (‰) more episodic through time, possibly resulting in 25 fairly extensive shifts in seawater sulfate con- 30 centration, even during the Phanerozoic. None- theless, using this simplification allows us to 15 establish that for C–S–O geochemical models, 10 total weathering fluxes for carbon and sulfur must 5 equal total burial fluxes for carbon and sulfur, respectively. 0 Referring to Figure 14, we can see that these 500 400 300 200 100 0 simplifications allow us to say that Time (Ma) = dMoc dt ¼ Fwg þ Fwc 2 Fbg 2 Fbc ¼ 0 ð11Þ Figure 13 Globally averaged isotopic composition of carbonates (d13C) and sulfates (d34S) through Phaner- = ozoic time (source Lindh, 1983). dMos dt ¼ Fws þ Fwp 2 Fbs 2 Fbp ¼ 0 ð12Þ

Weathering Weathering OC Ocean + atmosphere Burial Burial ∑ CO2

C G Carbonate Organic matter rocks OS in rocks Dissolved seawater sulfate

wwb b

E P Evaporite Sedimentary sulfates sulfides

Figure 14 The simplified geochemical cycles of carbon and sulfur, including burial and weathering of sedimentary carbonates, organic matter, evaporites, and sulfides. The relative fluxes of burial and weathering of organic matter and sulfide minerals plays a strong role in controlling the concentration of atmospheric O2. 546 The Global Oxygen Cycle The full rate equations for each reservoir mass cell growth rate, geometry and nutrient avail- in Figure 14 are as follows: ability (Rau et al., 1989, 1992), CO2 availability species-specific effects, modes of CO2 sequestra- = dMc dt ¼ Fbc 2 Fwc ð13Þ tion (i.e., C3,C4, and CAM plants). For sulfur, these can include species–specific effects, the dM =dt ¼ F 2 F ð14Þ degree of closed-system behavior and sulfate g bg wg concentration, sulfur oxidation and disproportio- = nation. However, as a simplification in geochem- dMs dt ¼ Fbs 2 Fws ð15Þ ical modeling of the C–S–O cycles, variability in carbon and sulfur isotopic fractionation is limited. = dMp dt ¼ Fbp 2 Fwp ð16Þ In the simplest case, fractionations are constant through all time for all environments. As a result, for example, the isotopic composition of organic dM =dt ¼ F 2 F þ 15=8ðF 2 F Þð17Þ O2 bg wg bp wp matter buried at a given time is set at a constant This system of equations has four unknowns: 25‰ depletion relative to seawater dissolved two burial fluxes and two weathering fluxes. carbonate at that time, and pyrite isotopic (iii) At this point, a third simplification of the composition is set at a constant 35‰ depletion carbon and sulfur systems is often applied to the relative to seawater sulfate. Of course, in reality, weathering fluxes of sedimentary rocks. As a first ac and as (the isotopic discriminations assigned approach, it is not unreasonable to guess that the between inorganic–organic carbon and sulfate– rate of weathering of a given type of rock relates in sulfide, respectively) vary greatly in both time and some sense to the total mass of that rock type space. Regardless of how ac and as are set, available on Earth’s surface. If that relation is however, once fractionations have been defined, assumed to be first order with respect to rock the mass balance equations given in (13)–(17) mass, an artificial weathering rate constant for above can be supplemented with isotope mass each rock reservoir can be derived. Such constants balances as well. have been derived by assuming that the weath- Based on the first simplification listed above, : the exogenic cycles of carbon and sulfur are ering rate equation has the form Fwi ¼ ki Mi If we can establish the mass of a sedimentary rock regarded as closed systems. As such, the bulk reservoir i, and also the average global river flux isotopic composition of average exogenic carbon to the oceans due to weathering of reservoir i, and sulfur do not vary through time. We can write a rate equation for the rate of change in (mass £ then ki can easily be calculated. For example, if the total global mass of carbonate in sedimentary isotopic composition) of each reservoir in rock is 5000 £ 1018 mol C, and annually there Figure 14 to reflect this isotope mass balance. are 20 £ 1012 mol C discharged from rivers to the For example: oceans from carbonate rock weathering, then 12 21 kcarbonate becomes (20 £ 10 mol C yr )/ dðd M Þ d dM M dd (5,000 £ 1018 mol C), i.e., 4 £ 1023 Myr21. oc oc ; oc oc þ oc oc These simple first-order weathering rate constants dt dt dt have been calculated for each sedimentary rock re- ¼ dcFwc þ dcFwg 2 docFbg ð Þ servoir in the C–S–O cycle, derived entirely from 2 ðdoc 2 acÞFbg 18 estimated preanthropogenic carbon and sulfur fluxes from continental weathering. Lack of a = ¼ ; true phenomenological relationship relating mic- Using the simplification that dMoc dt 0 roscale and outcrop-scale rock weathering reac- Equation (18) reduces to an equation relating the tions to regional- and global-scale carbon and organic matter burial flux Fbg in terms of other sulfur fluxes remains one of the primary weak- known entities: nesses limiting the accuracy of numerical models of the coupled C–S–O geochemical cycles.  h   1 ddoc If weathering fluxes from eroding sedimentary Fbg ¼ Moc þ Fwcðdoc 2 dcÞ rocks are independently established, such as ac dt i through use of mass-dependent weathering þ Fwgðdoc 2 dgÞ ð19Þ fluxes, then all of the weathering fluxes in our system of equations above become effectively Using (11) above, known. This leaves only the burial fluxes as un- knowns in solving the evolution of the C–S–O F ¼ F þ F 2 F ð20Þ system. bc wg wc bg The exact isotope discrimination that occurs during photosynthesis and during BSR is depen- A similar pair of equations can be written for dent on many factors. For carbon, these include the sulfur system: Atmospheric O2 throughout Earth’s History 547 relationships provide a useful guide for evaluating 1 dd changes in organic matter and pyrite burial fluxes ¼ os þ ð Þ Fbp Mos Fws dos 2 ds and the impact these have on atmospheric O , as dt 2 simply by examining the isotopic records of þ Fwpðdos 2 dpÞ ð21Þ marine carbonate and sulfate. Early efforts to model the coupled C–S–O cycles yielded important information. The work of Fbs ¼ Fwp þ Fws 2 Fbp ð22Þ Garrels and Lerman (1984) showed that the exogenic C and S cycles can be treated as closed Full rate equations for the rate of change in systems over at least Phanerozoic time, without sedimentary rock reservoir isotopic composition exchange between sedimentary rocks and the deep can be written as “crust þ mantle.” Furthermore, over timescales of millions of years, the carbon and sulfur cycles were ddc Fbcðdoc 2 dcÞ ¼ ð23Þ seen to be closely coupled, with increase in dt M c sedimentary organic carbon mass matched by loss of sedimentary pyrite (and vice versa). Other dd F ðd 2 a 2 d Þ g ¼ bg oc c g ð24Þ models explored the dynamics of the C–S–O dt Mg system. One important advance was promoted by Kump and Garrels (1986). In these authors’ model, dds Fbwðdos 2 dsÞ a steady-state C–S–O system was generated and ¼ ð25Þ perturbed by artificially increasing rates of organic dt Ms matter burial. These authors tracked the shifts in seawater carbon and sulfur isotopic composition ddp Fbpðdos 2 as 2 dpÞ ¼ ð26Þ that resulted, and compared these results with the dt M p true sedimentary record. Importantly, these authors as well as the rate of change in the mass of O . The recognized that although there is a general inverse 2 34 isotopic composition of seawater carbonate and relationship between seawater d 13C and d S, the sulfate through time comes directly from the exact path along an isotope–isotope plot through sedimentary rock record. However, it is uncertain time is not a straight line. Instead, because of the how to average globally as well as over substantial vastly different residence times of sulfate versus period of geologic time. The rates of change in carbonate in seawater, any changes in the C–S–O seawater carbonate and sulfate isotopic compo- system are first expressed through shifts in carbon sitions are fairly small terms, and have been left isotopes, then sulfur isotopes (Figure 15). out of many modeling efforts directed at the The authors also pointed out large-scale divisions geochemical C–S–O system; however, for com- in C–S isotope coupling through Phanerozoic pleteness sake these terms are included here. time. In the Paleozoic, organic matter and pyrite One implication of the isotope-driven modeling burial were closely coupled (largely because the approach to understanding the C–S–O coupled same types of depositional environment favor cycle is that burial fluxes of organic matter and burial of marine organic matter and pyrite). During pyrite (which are the sources of atmospheric O2 in this time, seawater carbonate and sulfur isotopes these models) are nearly proportional to seawater co-varied positively, indicating concomitant isotopic composition of inorganic carbon and increases (or decreases) in burial of organic matter sulfate. Recalling the equation for organic matter and pyrite. During the late Paleozoic, Mesozoic, burial above, it is noted that because sedimentary and Cenozoic, terrestrial depositional environ- carbonate and organic matter mass are nearly ments became important settings for burial of constant through time, weathering fluxes do not organic matter. Because pyrite formation and vary greatly through time. Likewise, the average burial in terrestrial environments is extremely isotopic composition of carbonates and organic limited, organic matter and pyrite burial became matter are also nearly constant, as is the mass of decoupled at this time, and seawater carbonate and dissolved carbonate and the rate of change in sulfur isotopes co-varied negatively, indicating dissolved inorganic carbon isotopic composition. close matching of increased sedimentary organic Assuming that Fwc, Fwg, Moc,ddoc/dt, dc, and dg are matter with decreased sedimentary pyrite (and vice constant or nearly constant, Fbg becomes linearly versa), perhaps suggesting a net balance in O2 proportional to the isotopic composition of sea- production and consumption, and maintenance of p water dissolved inorganic carbon. The same nearly constant, equable O2 throughout much rationale applies for the sulfur system, with pyrite of the latter half of the Phanerozoic. The model burial becoming linearly proportional to the of Berner (1987) introduced the concept of rapid isotopic composition of seawater sulfate. Although recycling: the effort to numerically represent these relationships are not strictly true, even within the observation that younger sedimentary rocks the constraints of the model simplifications, these are more likely to be eroded and weathered than are 548 The Global Oxygen Cycle

14 13 16 3 15 20 Myr averages 1 0 – 20 Myr 6 2 20 – 40 Myr ... . 5 .. 2 9 33 640 – 660 Myr 3 4 34 660 – 680 Myr 12 8 d 13C 2 10 1 (‰) 7 17 18 22 1 23 11 34 33 19 20 32 31 0 21 24 26 30 27 28 29 –1 25 10 20 30 d 34S (‰)

Figure 15 Twenty-million-year average values of seawater d 34S plotted against concomitant carbonate d 13C for the last 700 Myr (source Kump and Garrels, 1986). older sedimentary rocks. Because young sedimen- Phosphorus is a key nutrient limiting primary tary rocks are likely to be isotopically distinct from productivity in many marine environments. older rocks (because they are recording any recent If phosphorus supply is increased, primary shifts in seawater carbon or sulfur isotopic production and perhaps organic matter burial composition), restoring that isotopically distinct will also increase. Degradation and remineraliza- carbon or sulfur more quickly back into seawater tion of OM during transit from surface waters into provides a type of negative feedback, dampening sediments liberates phosphorus, but most of this is excessively large or small burial fluxes required for quickly scavenged by adsorption onto the surfaces isotope mass balance. This negative feedback of iron oxyhydroxides. However, work by Ingall serves to reduce calculated fluctuations in organic and Jahnke (1994) and Van Cappellen and Ingall matter and pyrite burial rates, which in turn reduce (1996) has shown that phosphorus recycling and fluctuations in release of O2 to the atmosphere. release into seawater is enhanced under low O2 or Results from this study also predict large increases anoxic conditions. This relationship provides a in OM burial fluxes during the Permocarboniferous strong negative feedback between primary pro- (,300 Ma) above values present earlier in the duction, bottom water anoxia, and atmospheric Paleozoic. This increase, likely associated with O2. As atmospheric O2 rises, phosphorus scaven- production and burial of refractory terrigenous ging on ferric-iron is enhanced, phosphorus organic matter (less easily degraded than OM recycling back into surface waters is reduced, produced by marine organisms), led to elevated primary production rates are reduced, and O2 concentrations of O2 ,300 Ma. declines. If O2 concentrations were to fall, One flaw with efforts to model the evolution of phosphorus scavenging onto ferric-iron would be Phanerozoic O2 using the carbon and sulfur inhibited, phosphorus recycling back into isotope records is that unreasonably large fluctu- surface waters would be accelerated, fueling ations in organic matter and pyrite burial fluxes increased primary production and O2 release to (with coincident fluctuations in O2 production the atmosphere. Van Cappellen and Ingall (1996) rates) would result. Attempts to model the whole applied these ideas to a mathematical model of the Phanerozoic generated unreasonably low and high C–P–Fe–O cycle to show how O2 concentrations O2 concentrations for several times in the could be stabilized by phosphorus recycling rates. Phanerozoic (Lasaga, 1989), and applications of Petsch and Berner (1998) expanded the model what seemed to be a realistic feedback based on of Van Cappellen and Ingall (1996) to include the reality (weathering rates of sedimentary organic sulfur system, as well as carbon and sulfur isotope matter and pyrite dependent on the concentration effects. This study examined the response of the of O2) were shown to actually become positive C–S–O–Fe–P system, and in particular carbon feedbacks in the isotope-driven C–S–O models and sulfur isotope ratios, to perturbation in global (Berner, 1987; Lasaga, 1989). ocean overturn rates, changes in continental Atmospheric O2 throughout Earth’s History 549 weathering, and shifts in the locus of organic of photorespiration. Photorespiration is a net matter burial from marine to terrestrial depocen- consumptive process for plants; previously fixed ters. Confirming the idea promoted by Kump and carbon is consumed with O2 to produce CO2 and Garrels (1986), these authors showed that pertur- energy. CO2 produced through photorespiration bations of the exogenic C–S–O cycle result in may diffuse out of the cell, but it is also likely to be shifts in seawater carbon and sulfur isotopic taken up (again) for photosynthesis. Thus, in cells composition similar in amplitude and duration to undergoing fairly high rates of photorespiration in observed isotope excursions in the sedimentary addition to photosynthesis, a significant fraction of record. total CO2 available for photosynthesis derives from Other proposed feedbacks stabilizing the con- oxidized, previously fixed organic carbon. The centration of atmospheric O2 over Phanerozoic effect of this on cellular carbon isotopic compo- time include a fire-regulated PO4 feedback sition is that because each round of photosynthesis (Kump, 1988, 1989). Terrestrial primary pro- results in 13C-depletion in cellular carbon relative duction requires much less phosphate per mole to CO2, cells with high rates of photorespiration 13 CO2 fixed during photosynthesis than marine will contain more C-depleted CO2 and thus will primary production. Thus, for a given supply of produce more 13C-depleted organic matter. PO4, much more CO2 can be fixed as biomass and In controlled-growth experiments using both O2 released from photosynthesis on land versus in higher plants and single-celled marine photosyn- the oceans. If terrestrial production proceeds too thetic algae, a relationship between ambient O2 rapidly, however, p levels may rise slightly and O2 concentration and net isotope discrimination has lead to increased forest fires. Highly weatherable, been observed (Figure 16)(Berner et al., 2000; PO4-rich ash would then be delivered through Beerling et al., 2002). The functional form of this weathering and erosion to the oceans. Primary relationship has been expressed in several ways. production in the oceans would lead to less CO2 The simplest is to allow isotope discrimination to fixed and O2 released per mole of PO4. vary linearly with changing atmospheric O2 mass: Hydrothermal reactions between seawater and ¼ £ ð = Þ: ac 25 MO2 38 More complicated relation- young oceanic crust have been proposed as ships have also been derived, based on curve- an influence on atmospheric O2 (Walker, 1986; fitting the available experimental data on isotopic Carpenter and Lohmann, 1999; Hansen and fractionation as a function of [O2]. O2-dependent Wallmann, 2002). While specific periods of isotopic fractionation during photosynthesis has oceanic anoxia may be associated with acceler- provided the first mathematically robust isotope- ated hydrothermal release of mantle sulfide (i.e., driven model of the C–S–O cycle consistent with the Mid-Cretaceous, see Sinninghe-Damste´ and geologic observations (Berner et al., 2000). Ko¨ster, 1998), long-term sulfur and carbon Results of this model show that allowing isotope isotope mass balance precludes substantial inputs fractionation to respond to changes in ambient O of mantle sulfur to the Earth’s surface of a 2 different net oxidation state and mass flux than what is subducted at convergent margins (Petsch, 6 1999; Holland, 2002). 4 One recent advance in the study of isotope- 2 driven models of the coupled C–S–O cycles is 0 re-evaluation ofisotope fractionations. Hayes etal., C) (per mil)

(1999) published a compilation of the isotopic 13 –2 ∆ ( +

composition of inorganic and organic carbon for ∆ –4 the past 800 Myr. One feature of this dual record is –6 a distinct shift in the net isotopic distance between 0 200 400 600 800 1,000 1,200 1,400 carbonate and organic carbon, occurring during the Growth atmospheric O /CO ratio past ,100 million years. When carbon isotope 2 2 distance is compared to estimates of Cenozoic and Figure 16 Relationship between change in D(D13C) of p Mesozoic O2, it becomes apparent that there may vascular land plants determined experimentally in be some relationship between isotopic fraction- response to growth under different O2/CO2 atmospheric ation associated with organic matter production mixing ratios. D(D13C) is the change in carbon isotope fractionation relative to fractionation for the controls at and burial and the concentration of O2 in the atmosphere. The physiological underpinning present day conditions (21% O2, 0.036% CO2). The solid line shows the nonlinear curve fitted to the data, behind this proposed relationship rests on compe- 13 given by D(D C) ¼ 219.94 þ 3.195 £ ln(O2/CO2); tition between photosynthesis and photorespiration (†) Phaseolus vulgaris;(V) Sinapis alba;(A) from in the cells of photosynthetic organisms. Because Berner et al. (2000);(þ) from Berry et al. (1972) O2 is a competitive inhibitor of CO2 for attachment (Beerling et al., 2002) (reproduced by permission of to the active site of Rubisco, as ambient O2 Elsevier from Geochem. Cosmochim. Acta 2002, concentrations rise relative to CO2, so will rates 66, 3757–3767). 550 The Global Oxygen Cycle

70

60

50 2 40 mol O

18 30 10

20

10

0 500 400 300 200 100 0 Time (Ma)

Figure 17 Evolution of the mass of atmospheric O2 through Phanerozoic time, estimated using an isotope mass balance described in Equations (11)–(26). The model employs the isotope date of Figure 13, and includes new advances in understanding regarding dependence of carbon isotope discrimination during photosynthesis and sulfur isotope discrimination during sulfur disproportionation and organic sulfur formation. The system of coupled differential equations were integrated using an implicit fourth-order Kaps–Rentrop numerical integration algorithm appropriate for this stiff set of equations (sources Petsch, 2000; Berner et al., 2000). provides a strong negative feedback dampening extreme sulfur isotope depletion due to several excessive increases or decreases in organic matter cycles of BSR, sulfide oxidation and burial rates. Rates of organic matter burial in this sulfur disproportionation, organic matter is sulfur- model are no longer simply dependent on seawater ized within the sediments. Closed, or nearly 13 carbonate d C, but now also vary with 1/ac.As closed, system behavior of BSR in the sediments fractionation becomes greater (through elevated results in late-stage sulfide (the source of sulfur O2), less of an increase in organic matter burial in sedimentary organic matter) to be more rates is required to achieve the observed increase in enriched compared with pyrite in the same 13 seawater d C than if ac were constant. sediments. It is known that burial of sulfide as The same mathematical argument can be organic sulfur is facilitated in low O2 or anoxic applied to sulfate–sulfide isotope fractionation waters. If lower atmospheric O2 in the past encouraged development of more extensive during BSR. As O2 concentrations increase, so does sulfur isotope fractionation, resulting in a anoxic basins and increased burial of sulfide as strong negative feedback on pyrite burial rates. organic sulfur instead of pyrite, the 10‰ offset This is consistent with the broad-scale changes in between pyrite and organic sulfur would become sulfur isotope dynamics across the Proterozoic, effectively a change in net sulfur fractionation in reflecting a large increase in D34S (between sulfate response to O2. Applying these newly recognized modifications and sulfide) when atmospheric O concentrations 2 of carbon and sulfur isotope discrimination in were great enough to facilitate bacterial sulfide response to changing O availability has allowed oxidation and sulfur disproportionation. Perhaps 2 development of new numerical models of the during the Phanerozoic, when O concentrations 2 evolution of the coupled C–S–O systems were greater, sulfur recycling (sulfate to sulfide and variability of Phanerozoic atmospheric O2 through BSR, sulfur oxidation, and sulfur dis- concentration (Figure 17). proportionation) was increased, resulting in greater net isotopic distance between sulfate and sulfide. Another means of changing net sulfur isotope discrimination in response to O2 may be the distribution of reduced sulfur between sulfide 8.11.7 CONCLUSIONS minerals and organic matter-associated sulfides. Molecular oxygen is generated and consumed Work by Werne et al., 2000, 2003) has shown that by a wide range of processes. The net cycling of organic sulfur is consistently ,10 ‰ enriched in O2 is influenced by physical, chemical, and 34S relative to associated pyrite. This is believed to most importantly, biological processes acting on result from different times and locations of and beneath the Earth’s surface. The exact organic sulfur versus pyrite formation. While distribution of O2 concentrations depends on the pyrite may form in shallow sediments or even specific interplay of these processes in time and anoxic portions of the water column, reflecting space. Large inroads have been made towards Conclusions 551

100 H escape from Evolution of Huronian glaciation, Rodinian Phanerozoic –1 CH photolysis photosynthesis oxidation of crust supercontinent 10 4 oxygenation of atmosphere 10–2

10–3

10–4 [O2] (bar) 10–5 Snowball 10–6 anoxia?

10–7

10–8

10–9

10–10

4,5004,000 3,500 3,000 2,500 2,000 1,500 1,000 500 0 Time (Ma)

Figure 18 Composite estimate of the evolution of atmospheric oxygen through 4.5 Gyr of Earth’s history. Irreversible oxidation of the Earth resulted from CH4 photolysis and hydrogen escape during early Earth’s history. Evolution of oxygenic photosynthesis preceded substantial oxygenation of the atmosphere by several hundred million p years. Relative stasis in atmospheric O2 typified much of the Proterozoic, with a possible pulse of oxygenation associated with formation of the Rodinian supercontinent in the Late Mesoproterozoic, and possible return to anoxia associated with snowball glaciation in the Neoproterozoic (sources Catling et al., 2001; Kasting, 1992; Rye and Holland, 1998; Petsch, 2000; Berner et al., 2000). understanding the processes that control the con- times when biological productivity and ocean centration of atmospheric O2, especially regarding circulation facilitate anoxic conditions, but in the O2 as a component of coupled biogeochemical atmosphere, O2 concentrations have remained , , cycles of many elements, including carbon, sulfur, within 0.05–0.35 bar pO2 for the past 600 Myr. nitrogen, phosphorus, iron, and others. Several outstanding unresolved gaps in our Earth’s modern oxygenated atmosphere is the understanding remain, in spite of a well- product of over four billion years of its history developed understanding of the general features (Figure 18). The early anoxic atmosphere was of the evolution of atmospheric O2 through time. slowly oxidized (although not oxygenated) as the These gaps represent potentially meaningful result of slow H2 escape. Evolution of oxygenic directions for future research, including: photosynthesis accelerated the oxidation of Earth’s (i) assessing the global importance of mineral crust and atmosphere, such that by ,2.2 Ga a small oxidation as a mechanism of O2 consumption; but significant concentration of O2 was likely (ii) the flux of reduced gases from volcanoes, present in Earth’s atmosphere. Limited primary metamorphism, and diffuse mantle/lithosphere production and oxygen production compared with degassing; the flux of reduced volcanic gases maintained this (iii) the true dependence of organic matter p low O2 atmosphere for over one billion years until oxidation on availability of O2, in light of the great the Neoproterozoic. Rapid oscillations in Earth’s abundance of microaerophilic and anaerobic carbon and sulfur cycles associated with global microorganisms utilizing carbon respiration as a Snowball glaciation may also have expression in a metabolic pathway, carbon isotopic evidence return to atmospheric anoxia at this time, but suggesting continual and essentially constant subsequent to the late Proterozoic isotope excur- organic matter oxidation as part of the sedimen- sions, oxygenation of the atmosphere to near- tary rock cycle during the entire past four billion modern concentrations developed such that by the years, and the inefficiency of organic matter Precambrian–Cambrian boundary, O2 concen- oxidation during continental weathering; trations were high enough to support widespread (iv) stasis in the oxygenation of the atmosphere skeletonized metazoans. Phanerozoic seawater during the Proterozoic; and atmospheric O2 concentrations have fluctuated (v) contrasting biochemical, fossil, and mol- in response to tectonic forcings, generating ecular evidence for the antiquity of the innovation regional-scale anoxia in ocean basins at certain of oxygenic photosynthesis; and 552 The Global Oxygen Cycle (vi) evaluating the relative strength of biologi- Berner R. A. (2002) Examination of hypotheses for the cal productivity versus chemical evolution of the Permo–Triassic boundary extinction by carbon cycle modeling. Proc. Natl. Acad. Sci. 99, 4172–4177. Earth’s crust and mantle in controlling the early Berner R. A. and Canfield D. E. (1989) A new model for stages of oxygenation of the atmosphere. atmospheric oxygen over Phanerozoic time. Am. J. Sci. 289, Thus, study of the global biogeochemical cycle 333–361. of oxygen, the component of our atmosphere Berner R. A. and Kothavala Z. (2001) GEOCARB III: a revised integral and crucial for life as we know it, remains model of atmospheric CO2 over Phanerozoic time. Am. J. Sci. 301, 182–204. a fruitful direction for Earth science research. Berner R. A., Petsch S. T., Lake J. A., Beerling D. J., Popp B. N., Lane R. S., Laws E. A., Westley M. B., Cassar N., Woodward F. I., and Quick W. P. (2000) Isotope fractionation and atmospheric oxygen: Implications for Phanerozoic O2 evolution. Science 287, 1630–1633. REFERENCES Berry J. A., Troughton J. H., and Bjo¨rkman O. (1972) Effect of oxygen concentration during growth on carbon isotope Algeo T. J. and Scheckler S. E. (1998) Terrestrial-marine discrimination in C3 and C4 species of Atriplex. Carnegie teleconnections in the Devonian: links between the evolution Inst. Yearbook 71, 158–161. of land plants, weathering processes, and marine anoxic Beukes N. J., Dorland H., Gutzmer J., Nedachi M., and Ohmoto events. Phil. Trans. Roy. Soc. London B 353, 113–128. H. (2002) Tropical laterites, life on land, and the history of Anbar A. D. and Knoll A. H. (2002) Proterozoic ocean atmospheric oxygen in the Paleoproterozoic. Geology 30, chemistry and evolution: a bioinorganic bridge? Science 491–494. 297, 1137–1142. Brasier M. D., Green O. R., Jephcoat A. P., Kleppe A. K., van Arthur M. A. (2000) Volcanic contributions to the carbon and Kranendonk M. J., Lindsay J. F., Steele A., and Grassineau sulfur geochemical cycles and global change. In Encyclo- N. V. (2002) Questioning the evidence for Earth’s earliest pedia of Volcanoes (eds. H. Sigurdsson, B. F. Houghton, fossils. Nature 416, 76–81. S. R. McNutt, H. Rymer, and J. Stix). Academic Press, San Brocks J. J., Logan G. A., Buick R., and Summons R. E. (1999) Diego, pp. 1045–1056. Archean molecular fossils and the early rise of eukaryotes. Bartley J. K., Semikhatov M. A., Kaufman A. J., Knoll A. H., Science 285, 1033–1036. Pope M. C., and Jacobsen S. B. (2001) Global events across Budyko M. I., Ronov A. B., and Yanshin A. L. (1987) History the Mesoproterozoic–Neoproterozoic boundary: C and Sr of the Earth’s Atmosphere isotopic evidence from Siberia. Precamb. Res. 111, . Springer, Berlin. 165–202. Buick R. (1990) Microfossil recognition in Archean rocks: an Battle M., Bender M., Sowers R., Tans P. P., Butler J. H., appraisal of spheroids and filaments from a 3500 M. Y. old Elkins J. W., Ellis J. T., Conway T., Zhang N., Pang P., and chert-barite unit at North Pole, Western Australia. Palaios 5, Clarke A. D. (1996) Atmospheric gas concentrations over 441–459. Buick I. S., Uken R., Gibson R. L., and Wallmach T. (1998) the past century measured in air from firn at the South Pole. 13 Nature 383, 231–235. High-d C Paleoproterozoic carbonates from the Transvaal Baum S. K. and Crowley T. J. (2001) GCM response to late Supergroup, South Africa. Geology 26, 875–878. Precambrian (,590 Ma) ice-covered continents. Geophys. Butterfield N. J. and Rainbird R. H. (1998) Diverse organic- Res. Lett. 28, 583–586. walled fossils, including possible dinoflagellates, from early Beaumont V. and Robert F. (1999) Nitrogen isotope ratios of Neoproterozoic of arctic Canada. Geology 26, 963–966. kerogens in Precambrian cherts: a record of the evolution of Caldeira K., and Kasting J. F. (1992) Susceptibility of the early atmospheric chemistry? Precamb. Res. 96, 63–82. Earth to irreversible glaciation caused by carbon dioxide Beerling D. J. and Berner R. A. (2000) Impact of a Permo– clouds. Nature 359, 226–228. Canfield D. E. (1998) A new model for Proterozoic ocean Carboniferous high O2 event on the terrestrial carbon cycle. Proc. Natl. Acad. Sci. 97, 12428–12432. chemistry. Nature 396, 450–453. Beerling D. J., Lake J. A., Berner R. A., Hickey L. J., Taylor Canfield D. E. and Teske A. (1996) Late Proterozoic rise in D. W. and Royer D. L. (2002) Carbon isotope evidence atmospheric oxygen concentration inferred from phylo- genetic and sulphur-isotope studies. Nature 382, implying high O2/CO2 ratios in the Permo–Carboniferous atmosphere. Geochim. Cosmochim. Acta 66, 3757–3767. 127–132. Bekker A., Kaufman A. J., Karhu J. A., Beukes N. J., Quinten Canfield D. E., Habicht K. S., and Thamdrup B. (2000) The S. D., Coetzee L. L., and Kenneth A. E. (2001) Chemos- Archean sulfur cycle and the early history of atmospheric tratigraphy of the Paleoproterozoic Duitschland Formation, oxygen. Science 288, 658–661. South Africa. Implications for coupled climate change and Carpenter S. J. and Lohmann K. C. (1999) Carbon isotope carbon cycling. Am. J. Sci. 301, 261–285. ratios of Phanerozoic marine cements: re-evaluating global Bender M., Labeyrie L. D., Raynaud D., and Lorius C. (1985) carbon and sulfur systems. Geochim. Cosmochim. Acta 61, Isotopic composition of atmospheric O2 in ice linked with 4831–4846. deglaciation and global primary productivity. Nature 318, Catling D. C., Zahnle K. J., and McKay C. P. (2001) Biogenic 349–352. methane, hydrogen escape, and the irreversible oxidation of Bender M., Sowers T., and Labeyrie L. (1994a) The Dole effect the early Earth. Science 293, 839–843. and its variations during the last 130,000 years as measured in Cloud P. (1972) A working model of the primitive Earth. Am. the Vostok ice core. Global Biogeochem. Cycle 8, 363–376. J. Sci. 272, 537–548. Bender M. L., Sowers T., Barnola J.-M., and Chappellaz J. Cloud P. E. (1976) Beginnings of biospheric evolution and (1994b) Changes in the O2/N2 ratio of the atmosphere their biogeochemical consequences. Paleobiol. 2, during recent decades reflected in the composition of air 351–387. in the firn at Vostok Station, Antarctica. Geophys. Res. Condie K. C., Des Marais D. J., and Abbott D. (2001) Lett. 21, 189–192. Precambrian superplumes and supercontinents: a record in Berner R. A. (1987) Models for carbon and sulfur cycles and black shales, carbon isotopes, and paleoclimates? Precamb. atmospheric oxygen: application to Paleozoic geologic Res. 106, 239–260. history. Am. J. Sci. 287, 177–196. Cope M. J. and Chaloner W. G. (1985) Wildfire: an interaction Berner R. A. (1994) GEOCARB II: a revised model of of biological and physical processes. In Geological Factors atmospheric CO2 over Phanerozoic time. Am. J. Sci. 294, and the Evolution of Plants (ed. B. H. Tiffney). Yale 56–91. University Press, New Haven, CT, pp. 257–277. References 553

Delano J. W. (2001) Redox history of the Earth’s interior since of carbon during the past 800 Myr. Chem. Geol. 161, ,3900 Ma: implications for prebiotic molecules. Origins 103–125. Life Evol. Biosphere 31, 311–341. Herbert T. D. and Sarmiento J. L. (1991) Ocean nutrient Delmelle P. and Stix J. (2000) Volcanic gases. In Encyclopedia distribution and oxygenation: limits on the formation of of Volcanoes (eds. H. Sigurdsson, B. F. Houghton, S. R. warm saline bottom water over the past 91 m.y. Geology 19, McNutt, H. Rymer, and J. Stix). Academic Press, San 702–705. Diego, pp. 803–816. Hinkle M. E. (1994) Environmental conditions affecting the Des Marais D. J., Strauss H., Summons R. E., and Hayes J. M. concentrations of He, CO2,O2 and N2 in soil gases. Appl. (1992) Carbon isotope evidence for stepwise oxidation of the Geochem. 9, 53–63. Proterozoic environment. Nature 359, 605–609. Hoehler T. M., Bebout B. M., and Des Marais D. J. (2001) The Deuser W. G. (1975) Chemical Oceanography. Academic role of microbial mats in the production of reduced gases on Press, Orlando, FL, p. 3. the early Earth. Nature 412, 324–327. Doolittle R. F., Feng D. F., Tsang S., Cho G., and Little E. Hoffman P. F. and Schrag D. P. (2002) The snowball Earth (1996) Determining divergence times of the major kingdoms hypothesis: testing the limits of global change. Terra Nova of living organisms with a protein clock. Science 271, 14, 129–155. 470–477. Hoffman P. F., Kaufman A. J., Halverson G. P., and Schrag Dudley R. (1998) Atmospheric oxygen, giant Paleozoic insects D. P. (1998) A Neoproterozoic snowball Earth. Science 281, and the evolution of aerial locomotor performance. J. Exp. 1342–1346. Biol. 201, 1043–1050. Hofmann H. J., Gery K., Hickman A. H., and Thorpe R. I. Dudley R. (2000) The evolutionary physiology of animal flight: (1999) Origin of 2.45 Ga coniform stromatolites in Warra- Paleobiological and present perspectives. Ann. Rev. Phys. woona Group, Western Australia. Geol. Soc. Am. Bull. 111, 62, 135–155. 1256–1262. Evans D. A. D. (2000) Stratigraphic, geochronological and Holland H. D. (1978) The Chemistry of the Atmosphere and paleomagnetic constraints upon the Neoproterozoic climatic Oceans. Wiley, New York. paradox. Am. J. Sci. 300, 347–433. Holland H. D. (1984) The Chemical Evolution of the Farquhar J., Bao H., and Thiemens M. (2000) Atmospheric Atmosphere and Oceans. Princeton University Press, influence of Earth’s earliest sulfur cycle. Science 289, Princeton, NJ. 756–758. Holland H. D. (2002) Volcanic gases, black smokers and the Farquhar J., Wing B. A., McKeegan K. D., Harris J. W., great oxidation event. Geochim. Cosmochim. Acta 66, Cartigny P., and Thiemens M. H. (2002) Mass-independent 3811–3826. sulfur of inclusions in diamond and sulfur recycling on early Holland H. D., Feakes C. R., and Zbinden E. A. (1989) The Flin Earth. Science 297, 2369–2372. Flon paleosol and the composition of the atmosphere 1.8 Garcia-Pichel F. (1998) Solar ultraviolet and the evolutionary BYBP. Am. J. Sci. 289, 362–389. history of cyanobacteria. Origins Life Evol. Biosphere 28, Hotinski R. M., Bice K. L., Kump L. R., Najjar R. G., and 321–347. Arthur M. A. (2001) Ocean stagnation and End-Permian Garrels R. M. and Lerman A. (1984) Coupling of the anoxia. Geology 29, 7–10. sedimentary sulfur and carbon cycles—an improved Hurtgen M. T., Arthur M. A., Suits N. S., and Kaufman A. J. model. Am. J. Sci. 284, 989–1007. (2002) The sulfur isotopic composition of Neoproterozoic Gilbert D. L. (1996) Evolutionary aspects of atmospheric seawater sulfate: implications for a snowball Earth? Earth oxygen and organisms. In Environmental Physiology: 2 (eds. Planet. Sci. Lett. 203, 413–429. M. J. Fregly, and C. M. Blatteis). Oxford University Press, Ingall E. and Jahnke R. (1994) Evidence for enhanced Oxford, UK, pp. 1059–1094. phosphorus regeneration from marine sediments overlain Glasspool I. (2000) A major fire event recorded in the by oxygen depleted waters. Geochim. Cosmochim. Acta 58, mesofossils and petrology of the late Permian, Lower 2571–2575. Whybrow coal seam, Sydney Basin, Australia. Palaeogeogr. Ingmanson D. E. and Wallace W. J. (1989) Oceanography: An Palaeoclimat. Palaeoecol. 164, 373–396. Introduction. Wadworth, Belmont, CA, pp. 99. Godderı´s Y. and Veizer J. (2000) Tectonic control of chemical Joachimski M. M., Ostertag-Henning C., Pancost R. D., and isotopic composition of ancient oceans: the impact of Strauss H., Freeman K. H., Littke R., Sinninghe-Damste´ continental growth. Am. J. Sci. 300, 434–461. J. D., and Racki G. (2001) Water column anoxia, Gould S. J. (1995) Of it and not above it. Nature 377, 681–682. enhanced productivity and concomitant changes in d 13C Graham J. B., Dudley R., Anguilar N., and Gans C. (1995) and d 34S across the Frasnian–Famennian boundary Implications of the late Paleozoic oxygen pulse for (Kowala—Holy Cross Mountains/Poland). Chem. Geol. physiology and evolution. Nature 375, 117–120. 175, 109–131. Habicht K. S., Gade M., Thamdrup B., Berg P., and Canfield Kah L. C., Lyons T. W., and Chesley J. T. (2001) Geochemistry D. E. (2002) Calibration of sulfate levels in the Archean of a 1.2 Ga carbonate-evaporite succession, northern Baffin ocean. Science 298, 2372–2374. and Bylot Islands: implications for Mesoproterozoic marine Haggerty S. E. and Toft P. B. (1985) Native iron in the evolution. Precamb. Res. 111, 203–234. continental lower crust: petrological and geophysical Kaiho K., Kajiwara Y., Nakano T., Miura Y., Kawahata H., implications. Science 229, 647–649. Tazaki K., Ueshima M., Chen Z., and Shi G. (2001) Halmer M. M., Schmincke H.-U., and Graf H.-F. (2002) The End-Permian catastrophe by a bolide impact: evidence of annual volcanic gas input into the atmosphere, in particular a gigantic release of sulfur from the mantle. Geology 29, into the stratosphere: a global data set for the past 100 815–818. years. J. Volcanol. Geotherm. Res. 115, 511–528. Karhu J. and Holland H. D. (1996) Carbon isotopes and the Hansen K. W. and Wallmann K. (2003) Cretaceous and rise of atmospheric oxygen. Geology 24, 867–870. Cenozoic evolution of seawater composition, atmospheric Kasting J. F. (1987) Theoretical constraints on oxygen and O2 and CO2: a model perspective. Am. J. Sci. 303, 94–148. carbon dioxide concentrations in the Precambrian atmos- Hartnett H. E., Keil R. G., Hedges J. I., and Devol A. H. (1998) phere. Precamb. Res. 34, 205–229. Influence of oxygen exposure time on organic carbon Kasting J. F. (1992) Models relating to Proterozoic Atmos- preservation in continental margine sediments. Nature 391, pheric and Oceanic Chemistry. In The Proterozoic Bio- 572–574. sphere (eds. J. W. Schopf, and C. Klein). Cambridge Hayes J. M., Strauss H., and Kaufman A. J. (1999) The University Press, Cambridge, pp. 1185–1187. abundance of 13C in marine organic matter and Kasting J. F. and Siefert J. L. (2002) Life and the evolution of isotopic fractionation in the global biogeochemical cycle Earth’s atmosphere. Science 296, 1066–1068. 554 The Global Oxygen Cycle

Kasting J. F., Zahnle K. J., and Walker J. C. G. (1983) Mo¨rner N.-A. and Etiope G. (2002) Carbon degassing from the Photochemistry of methane in the Earth’s early atmosphere. lithosphere. Global. Planet. Change 33, 185–203. Precamb. Res. 20, 121–148. Murakami T., Utsunomiya S., Imazu Y., and Prasad N. (2001) Kasting J. F., Eggler D. H., and Raeburn S. P. (1993) Mantle Direct evidence of late Archean to early Proterozoic anoxic redox evolution and the oxidation state of the Archean atmosphere from a product of 2.5 Ga old weathering. Earth atmosphere. J. Geol. 101, 245–257. Planet. Sci. Lett. 184, 523–528. Kasting J. F., Pavlov A. A., and Siefert J. L. (2001) A coupled Najjar R. G. and Keeling R. F. (2000) Mean annual cycle of the ecosystem-climate model for predicting the methane con- air–sea oxygen flux: a global view. Global Biogeochem. centration in the Archean atmosphere. Orig. Life Evol. Cycles 14, 573–584. Biosph. 31, 271–285. Nutman A. P., Mojzsis S. J., and Friend C. R. L. (1997) Kaufman A. J. and Knoll A. H. (1995) Neoproterozoic Recognition of ^3850 Ma water-lain sediments in variations in the C-isotopic composition of seawater: West Greenland and their significance for the early Archean stratigraphic and biogeochemical implications. Precamb. Earth. Geochim. Cosmochim. Acta 61, 2475–2484. Res. 73, 27–49. Ohmoto H. (1996) Evidence in pre-2.2 Ga paleosols for the Kaufman A. J., Hayes J. M., Knoll A. H., and Germs G. J. B. early evolution of atmospheric oxygen and terrestrial biota. (1991) Isotopic composition of carbonates and organic Geology 24, 1135–1138. carbon from upper Proterozoic successions in Namibia: Ohmoto H. (1999) Redox state of the Archean atmosphere: stratigraphic variation and the effects of diagenesis and evidence from detrital heavy minerals in ca. 3250– metamorphism. Precamb. Res. 49, 301–327. 2750 Ma sandstones from the Pilbara Craton, Australia: Keeling R. F. and Shertz S. R. (1992) Seasonal and interannual comment and reply. Geology 27, 1151–1152. variations in atmospheric oxygen and implications for the Olson J. M. and Pierson B. K. (1986) Photosynthesis 3.5 carbon cycle. Nature 358, 723–727. thousand million years ago. Photosynth. Res. 9, 251–259. Keeling R. F., Najjar R. P., Bender M. L., and Tans P. P. (1993) Pan Y. and Stauffer M. R. (2000) Cerium anomaly and Th/U What atmospheric oxygen measurements can tell us about fractionation in the 1.85 Ga Flin Flon Paleosol: clues from the global carbon cycle. Global Biogeochem. Cycles 7, REE- and U-rich accessory minerals and implications for 37–67. paleoatmospheric reconstruction. Am. Mineral. 85, 898–911. Keller C. K. and Bacon D. H. (1998) Soil respiration and Pavlov A. A., Kasting J. F., Brown L. L., Rages K. A., and georespiration distinguished by transport analyses of vadose Freedman R. (2000) Greenhouse warming by CH4 in the 13 14 CO2, CO2, and CO2. Global Biogeochem. Cycles 12, atmosphere of early Earth. J. Geophys. Res. 105, 361–372. 11981–11990. Kiehl J. T. and Dickinson R. E. (1987) A study of the Petsch S. T. (1999) Comment on Carpenter and Lohmann radiative effects of enhanced atmospheric CO2 and CH4 on (1999). Geochim Cosmochim Acta 63, 307–310. early Earth surface temperatures. J. Geophys. Res. 92, Petsch S. T. (2000) A study on the weathering of organic matter 2991–2998. in black shales and implications for the geochemical cycles Kimura H. and Watanabe Y. (2001) Oceanic anoxia at the of carbon and oxygen. PhD Dissertation. Yale University. Precambrian–Cambrian boundary. Geology 29, 995–998. Petsch S. T. and Berner R. A. (1998) Coupling the geochemical Kirschvink J. L. (1992) Late Proterozoic low-latitude global cycles of C, P, Fe, and S: the effect on atmospheric O2 and glaciation: the snowball earth. In The Proterozoic Biosphere the isotopic records of carbon and sulfur. Am. J. Sci. 298, (eds. J. W. Schopf, and C. Klein). Cambridge University 246–262. Press, Cambridge, pp. 51–52. Petsch S. T., Berner R. A., and Eglinton T. I. (2000) A field Kump L. R. (1988) Terrestrial feedback in atmospheric oxygen study of the chemical weathering of ancient sedimentary regulation by fire and phosphorus. Nature 335, 152–154. organic matter. Org. Geochem. 31, 475–487. Kump L. R. (1989) Chemical stability of the atmosphere Petsch S. T., Smernik R. J., Eglinton T. I., and Oades J. M. and ocean. Palaeogeogr. Palaeoclimat. Paleoecol. 75, (2001) A solid state 13C-NMR study of kerogen degradation 123–136. during black shale weathering. Geochim. Cosmochim. Acta Kump L. R. and Garrels R. M. (1986) Modeling atmospheric 65, 1867–1882. O2 in the global sedimentary redox cycle. Am. J. Sci. 286, Phoenix V. R., Konhauser K. O., Adams D. G., and Bottrell S. H. 337–360. (2001) Role of biomineralization as an ultraviolet shield: Kump L. R., Kasting J. F., and Barley M. E. (2001) Rise of implications for Archean life. Geology 29, 823–826. atmospheric oxygen and the upside down Archean mantle. Pollard D. and Kasting J. K. (2001) Coupled GCM-ice sheet Geochem. Geophys. Geosys. 2 No. 2000 GC 000114. simulations of Sturtian (750–720 Ma) glaciation: when in Lasaga A. C. (1989) A new approach to isotopic modeling of the snowball-earth cycle can tropical glaciation occur? EOS the variation in atmospheric oxygen through the Phaner- 82, S8. ozoic. Am. J. Sci. 289, 411–435. Porter S. M. and Knoll A. H. (2000) Testate amoebae in the Lasaga A. C. and Ohmoto H. (2002) The oxygen geochemical Neoproterozoic Era: evidence from vase-shaped microfos- cycles: dynamics and stability. Geochim. Cosmochim. Acta sils in the Chuar Group, Grand Canyon. Paleobiology 26, 66, 361–381. 360–385. Lenton T. M. and Watson A. J. (2000) Redfield revisited 2. Rasmussen B. and Buick R. (1999) Redox state of the Archean What regulates the oxygen content of the atmosphere? atmosphere: evidence from detrital heavy minerals in ca. Global Biogeochem. Cycles 14, 249–268. 3250–2750 Ma sandstones from the Pilbara Craton, Aus- Lindh T. B. (1983) Temporal variations in 13C, 34S and global tralia. Geology 27, 115–118. sedimentation during the Phanerozoic. MS Thesis, Univer- Rau G. H., Takahashi T., and Des Marais D. J. (1989) Latitudinal 13 sity of Miami. variations in plankton d C: implications for CO2 and Logan G. A., Hayes J. M., Hleshima G. B., and Summons R. E. productivity in past oceans. Nature 341, 516–518. (1995) Terminal Proterozoic reorganization of biogeochem- Rau G. H., Takahashi T., Des Marais D. J., Repeta D. J., ical cycles. Nature 376, 53–56. and Martin J. H. (1992) The relationship between d13C Lovell J. S. (2000) Oxygen and carbon dioxide in soil air. of organic matter and [CO2(aq)] in ocean surface water: In Handbook of Exploration Geochemistry, Geochemical Data from a JGOFS site in the northeast Atlantic Remote Sensing of the Subsurface (ed. M. Hale). Elsevier, Ocean and a model. Geochim. Cosmochim. Acta 56, Amsterdam, vol. 7, pp. 451–469. 1413–1419. McMenamin M. A. S. and McMenamin D. L. S. (1990) The Robinson J. M. (1989) Phanerozoic O2 variation, fire, and Emergence of Animals—The Cambrian Breakthrough. terrestrial ecology. Palaeogeogr. Palaeoclimat. Palaeoecol. Columbia University Press, New York. (Global Planet Change) 75, 223–240. References 555

Robinson J. M. (1991) Phanerozoic atmospheric reconstruc- facies and stratigraphy of the Gamohaan and Frisco tions: a terrestrial perspective. Palaeogeogr. Palaeoclimat. Formations, Transvaal Supergroup, South Africa. Am. J. Palaeoecol. 97, 51–62. Sci. 297, 455–487. Ro¨hl H.-J., Schmid-Ro¨hl A., Oschmann W., Frimmel A., and Thomas A. L. R. (1997) The breath of life—did increased Schwark L. (2001) The Posidonia Shale (Lower Toarcian) of oxygen levels trigger the Cambrian Explosion? Trends Ecol. SW-Germany: an oxygen-depleted ecosystem controlled by Evol. 12, 44–45. sea level and paleoclimate. Palaeogeogr. Palaeoclimat. Tolbert N. E., Benker C., and Beck E. (1995) The oxygen and Palaeocol. 165, 27–52. carbon dioxide compensation points of C3 plants: possible Ronov A. B. (1976) Global carbon geochemistry, volcanism, role in regulating atmospheric oxygen. Proc. Natl. Acad. Sci. carbonate accumulation, and life. Geochem. Int. 13, 92, 11230–11233. 172–195. Valley J. W., Peck W. H., King E. M., and Wilde S. A. (2002) Rye R. and Holland H. D. (1998) Paleosols and the evolution of A cool early Earth. Geology 30, 351–354. atmospheric oxygen: a critical review. Am. J. Sci. 298, Van Cappellen P. and Ingall E. D. (1996) Redox stabilization 621–672. of the atmosphere and oceans by phosphorus-limited marine Rye R. and Holland H. D. (2000) Life associated with a productivity. Science 271, 493–496. 2.76 Ga ephemeral pond? Evidence from Mount Roe #2 van Zuilen M., Lepland A., and Arrhenius G. (2002) paleosol. Geology 28, 483–486. Reassessing the evidence for the earliest traces of life. Schidlowski M. (1988) A 3,800-million year isotopic record of Nature 418, 627–630. life from carbon in sedimentary rocks. Nature 333, 313–318. Vidal R. A., Bahr D., Baragiola R. A., and Peters M. (1997) Schidlowski M. (2001) Carbon isotopes as bio- Oxygen on Ganymede: laboratory studies. Science 276, geochemical recorders of life over 3.8 Ga of Earth history: 1839–1842. evolution of a concept. Precamb. Res. 106, 117–134. Walker J. C. G. (1979) The early history of oxygen and ozone Schopf J. W. (1992) Paleobiology of the Archean. In The in the atmosphere. Pure Appl. Geophys. 117, 498–512. Proterozoic Biosphere (eds. J. W. Schopf, and C. Klein). Walker J. C. G. (1986) Global geochemical cycles of carbon, Cambridge University Press, pp. 25–39. sulfur, and oxygen. Mar. Geol. 70, 159–174. Schopf J. W. (1993) Microfossils of the early Archean Apex Watson A. J. (1978) Consequences for the Biosphere of Chert: new evidence of the antiquity of life. Science 260, Grassland and Forest Fires. PhD Dissertation. Reading 640–646. University, UK. Schopf J. W., Kudryavtvev A. B., Agresti D. G., Wdowiak T. J., Werne J. P., Hollander D. J., Behrens A., Schaeffer P., Albrecht and Czaja A. D. (2002) Laser-Raman imagery of Earth’s P., and Sinninghe-Damste´ J. S. (2000) Timing of early earliest fossils. Nature 416, 73–76. diagenetic sulfurization of organic matter: a precursor- Sinninghe-Damste´ J. S. and Ko¨ster J. (1998) A euxinic southern product relationship in Holocene sediments of the anoxic North Atlantic Ocean during the Cenomanian/Turonian Cariaco Basin, Venezuela. Geochim. Cosmochim. Acta 64, oceanic anoxic event. Earth Planet. Sci. Lett. 158, 165–173. 1741–1751. Sleep N. H. and Zahnle K. (1998) Refugia from asteroid Werne J. P., Lyons T. W., Hollander D. J., Formolo M., and impacts on early Mars and the early Earth. J. Geophys. Res. Sinninghe-Damste´ J. S. (2003) Reduced sulfur in euxinic E—Planets. 103, 28529–28544. sediments of the Cariaco Basin: sulfur isotope constraints on Sleep N. H., Zahnle K. J., Kasting J. F., and Morowitz H. J. organic sulfur formation. Chem. Geol. 195, 159–179. (1989) Annihilation of ecosystems by large asteroid impacts Wilde S. A., Valley J. W., Peck W. H., and Graham C. M. on the early earth. Nature 342, 139–142. (2001) Evidence from detrital zircons for the existence of Sleep N. H., Zahnle K. J., and Neuhoff P. S. (2001) Initiation of continental crust and oceans on the Earth 4.4 Gyr ago. clement surface conditions on the earliest Earth. Proc. Natl. Nature 409, 175–178. Acad. Sci. 98, 3666–3672. Wortmann U. G., Hesse R., and Zacher W. (1999) Major- Sowers T., Bender M., and Raynaud D. (1989) Elemental and element analysis of cyclic black shales: paleoceanographic isotopic composition of occluded O2 and N2 in polar ice. implications for the early Cretaceous deep western Tethys. J. Geophys. Res. 94, 5137–5150. Paleoceanography 14, 525–541. Stratford K., Williams R. G., and Myers P. G. (2000) Impact Wray G. A., Levinton J. S., and Shapiro L. H. (1996) Molecular of the circulation on sapropel formation in the evidence for deep Precambrian divergences among eastern Mediterranean. Global Biogeochem. Cycles 14, Metazoan phyla. Science 274, 568–573. 683–695. Zhang R., Follows M. J., Grotzinger J. P., and Marshall J. Sumner D. Y. (1997) Carbonate precipitation and oxygen (2001) Could the late Permian deep ocean have been anoxic? stratification in late Archean seawater as deduced from Paleoceanography 16, 317–329.

q 2003, Elsevier Ltd. All rights reserved Treatise on Geochemistry No part of this publication may be reproduced, stored in a retrieval system or ISBN (set): 0-08-043751-6 transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Volume 8; (ISBN: 0-08-044343-5); pp. 515–555