Rise of atmospheric oxygen and the ``upside-down ... - CiteSeerX

Geochemistry Geophysics Geosystems

3

Research Letter

G

Volume 2 January 23, 2001 Paper number 2000GC000114

AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES Published by AGU and the Geochemical Society

ISSN: 1525-2027

Rise of atmospheric oxygen and the `ùpside-down'' Archean mantle Lee R. Kump and James F. Kasting

Department of Geosciences and Astrobiology Research Center, Pennsylvania State University, 503 Deike Building, University Park, Pennsylvania 16802 ([email protected]; [email protected])

Mark E. Barley

Key Centre for Strategic Mineral Deposits, University of Western Australia, Nedlands WA 6907, Australia ([email protected])

[1] Abstract: The establishment of an oxygen-rich atmosphere dramatically altered the evolution of life on Earth. Most of the recent discussion of the topic has been focused on the timing of the event rather than on its mechanism. Here we draw upon recent developments in the understanding of Earth's interior to propose that the rise of oxygen followed a geologically abrupt period of mantle overturn and/ or intense plume activity near the Archean-Proterozoic transition, 2470 ± 2450 million years ago. The magmatic event has already been linked to the widespread deposition of oxide-facies banded iron formation, and the rise of oxygen has been implicated as the trigger for iron deposition and Earth's first major glaciation. We argue that these events are all related to a change in redox state of volcanic gases brought about by deep-seated Late Archean and earliest Paleoproterozoic magmatism.

Keywords: Mantle redox; atmospheric oxygen; Late Archean; Paleoproterozoic; mantle plume. Index terms: Evolution of the atmosphere; chemical evolution; composition of the mantle; geochemical cycles. Received October 16, 2000; Revised December 8, 2000; Accepted December 11, 2000; Published January 23, 2001. Kump, L. R., J. F. Kasting, and M. E. Barley, 2001. Rise of atmospheric oxygen and the `ùpside-down'' Archean mantle, Geochem. Geophys. Geosyst., vol. 2, Paper number 2000GC000114 [5162 words, 2 figures]. Published January 23, 2001.

1. Introduction Most researchers agree that the initial rise in atmospheric O2 levels occurred between 2400 and 1800 Ma [Holland, 1984, 1994; Kasting, 1993; Karhu and Holland, 1996; Collerson and Kamber, 1999]. However, the exact cause of the rise of O2 remains uncertain. Cyanobacteria, the first oxygenic photosynthesizers, have been identified from organic biomarkers in sediments dated at 2700 Ma [Brocks et al., [2]

Copyright 2001 by the American Geophysical Union

1999] and have been associated morphologically with microfossils dated at 3500 Ma [Schopf, 1993]. Carbon isotopes, in particular, the roughly 30% average spread in d13C between carbonates and sedimentary organic carbon, are also consistent with an early origin of oxygenic photosynthesis [Schidlowski et al., 1983; Des Marais, 1997]. Indeed, the fact that the d13C values of carbonates have remained near 0% throughout most of geologic time has been interpreted as evidence that the rate of


3

G

kump et al.: research letter

organic carbon burial and by implication that of O2 production as well has remained approximately constant [Schidlowski et al., 1983]. This interpretation is consistent with the observed approximately constant organic carbon content of shales [Holland, 1984]. It must be recognized, however, that the carbon isotopic constraints on oxygen production are not particularly severe. Nonoxygenic autotrophs could have generated a considerable fraction of Archean sedimentary organic matter without significantly modifying its isotopic composition. Moreover, significant fluctuations or substantial secular trends in organic carbon burial through the Archean cannot be ruled out based on existing data. A system burying no organic carbon would produce limestones with a d13C near the lower limit of the data envelope, and a system burying organic carbon at twice the modern rate would be near the upper limit. The constraints on the isotopic composition of organic matter are even less restrictive. Nevertheless, there is no compelling evidence to support the suggestion that a permanent increase in oxygen production caused the rise of atmospheric O2. Why, then, did it take half a billion to a billion years after the evolution of cyanobacteria before significant amounts of O2 first appeared in the atmosphere? Theories regarding the cause for the rise in O2 fall into two categories: either the source of O2 was lower in the past (i.e., nonoxygenic autotrophs provided the organic matter and isotopic discrimination for Archean sediments) or the sinks for O2 must have been larger. In this paper, we pursue the second option, namely, that the O2 sinks were larger in the distant past. The first option, which remains a valid possibility, will be explored elsewhere. [3]

2. Oxygen Sources and Sinks Before embarking on a discussion of how the sinks for O2 may have changed, let us

[4]

2000GC000114

briefly review how the O2 budget is balanced today. According to Holland [1978], the net O2 production (organic carbon burial) rate is 1 1013 mol/yr. (This number represents 0.1% of gross photosynthetic production of O2. Respiration and/or decay recycle the other 99.9% of O2.) O2 is consumed by oxidation of reduced minerals in rocks and reduced volcanic gases. We include reduced fluids emanating from midocean ridge hydrothermal vents as part of the volcanic gas sink. Among the reduced gases emanating from surface volcanoes, H2, CO, and SO2 appear to be the most important [Holland, 1978, pp. 291±292]. Their total contribution to O2 loss is 14 1011 mol/ yr, or 14% of the net O2 budget. Fe(II) and S(II-) are the predominant reduced species in hydrothermal vent fluids. Holland [1984, p. 385] estimates a total O 2 demand of 20 mmol/kg for fluids emanating from the hot, axial, hydrothermal vent systems. Estimates for the global rate of seawater flow through the axial vent systems of range from (2.4±12) 1013 kg/yr [Morton and Sleep, 1985; Palmer and Edmond, 1989; Derry and Jacobsen, 1990]. Adopting a median value of 5 1013 kg/yr yields a global O2 consumption rate of 1 1012 mol/yr, or 10% of the rate of O2 production. Thus the total volcanic sink for O2 is 25% of the rate of O2 generation by organic C burial. This number is uncertain by at least 50%. The other 75% of the O2 generated is consumed by weathering of reduced compounds in rocks, predominantly organic C, pyritic S, and ferrous iron in shales. Although oxidative weathering is the major O2 sink today, this loss process would not have been important on the Archean Earth. The presence of detrital uraninite and pyrite in Archean sediments, coupled with the absence of red beds, indicates that surface oxidation was minimal at that time. Hence O2 must have been consumed mainly by reaction with reduced volcanic gases and with reduced

[5]


3

G


hydrothermal vent fluids. Ferrous iron leached from the continents during weathering may have been oxidized in the ocean either photochemically [Braterman et al., 1983] or biologically; however, this is not thought to constitute a quantitatively important O2 sink [Kump and Holland, 1992]. Hence we are confronted with a dilemma: How could volcanic outgassing, which accounts for only 25% of the O2 sink today, have consumed nearly all the biologically produced O2 at that time?

2.1. Archean Oxygen Consumption The rate of heat loss from Earth's interior was higher early in its history. A likely consequence of this was greater volcanic activity and a larger volcanic sink for O 2 in the Archean. However, by the same reasoning, the amount of CO2 released by volcanoes was also higher then. This, together with the approximately constant d13C composition of carbonates, reflecting a roughly constant proportion of organic carbon burial, requires that the burial rate of organic carbon was proportionately larger. If oxygenic photosynthesis dominated Archean productivity, then the O2 source was elevated in proportion to the volcanic sink. Thus a change in volcanic outgassing rate cannot by itself account for the delay in the rise of atmospheric O2. [6]

This objection does not apply if the composition of volcanic gases has changed with time Kasting et al., 1993]. Consider first the fluids emanating from midocean ridge hydrothermal vents. A significant change in the composition of vent fluids does appear to be needed to explain the widespread occurrence of banded iron formations (BIFs) prior to 1850 Ma [Beukes and Klein, 1992] and the enrichment of Fe in Precambrian shales [Kump and Holland, 1992]. Evidence from rare earth elements and Nd isotopes indicates that most of the iron was derived from hydrothermal

[7]

2000GC000114

sources rather than from continental weathering [see, e.g., Derry and Jacobsen, 1990]. In seafloor hydrothermal systems, much of the ferrous iron is removed through precipitation with sulfide. However, higher equilibration temperatures within the vent systems could increase the Fe/S ratio dramatically [Seyfried and Janecky, 1985], and this may indeed be the factor that made BIF formation possible prior to 1850 Ma. However, such a change is unlikely to have significantly increased the total O2 demand of the vent fluids. Iron and sulfide are inversely correlated in high-temperature fluids [Seyfried and Janecky, 1985], and each S(II-) atom has a potential O2 demand that is 8 times higher than that of Fe(II). So iron-rich vent fluids on the early Earth may actually have consumed less O2 than they do today. However, because heat flow was higher in the Archean, the rate of hydrothermal emanation likely was higher, so the overall sink was probably somewhat greater than today. Nevertheless, it could not have been this change that kept atmospheric O2 concentrations low prior to 2400 Ma.

2.2. Change in Oxidation State of Volcanic Gases [8] A change in the composition of surface volcanic gases, on the other hand, could have produced the desired result. Modern magmas have oxygen fugacities ( fO2) that are close to equilibrium with the fayalite-magnetite-quartz (FMQ) buffer, for which fO2 10ÿ8 at P = 5 atm and T = 1500 K [Holland, 1984], but there are good arguments that Earth's upper mantle was originally more reduced [Kasting et al., 1993]. If the upper mantle was more reduced in the past, then volcanic gases would have been more reduced as well. For surface outgassing at equilibrium with the iron-wustite (IW, Fe-FeO) buffer (a reasonable buffer just after the formation of Earth's core), the relevant value of fO2 is 10ÿ12 atm [Holland, 1984, p. 50]. This

3

G


would make H2 and CO the dominant components of surface volcanic emissions. The potential O2 sink from surface volcanism could therefore have been higher than today by a factor of 40 or more, allowing it to easily overwhelm the photosynthetic O2 source. The excess hydrogen could have escaped to space, so the oxygen budget would have been balanced in a manner totally different from today. In order for this hypothesis to be valid, the mantle must have become progressively oxidized over time. This process could have proceeded by subduction of hydrated, carbonated seafloor, followed by dissociation of H2 and CO and their outgassing in arc volcanoes [Kasting et al., 1993] or by subduction of ferric iron [Arculus, 1985; Lecuyer and Ricard, 1999]. The rate at which the mantle was oxidized depended on efficacy of the subduction process and on the oxygen buffering capacity of the mantle, neither of which is readily determined. One can attempt to constrain this process by measuring the redox state of ancient volcanic rocks [Delano, 1993; Canil, 1997], and this indeed constitutes the greatest challenge to this hypothesis (see below). For the moment, however, let us bypass this objection and simply estimate how much of a change in fO2 would be needed to allow volcanic outgassing to overwhelm the photosynthetic O2 source (Figure 1). [10] Assuming that surface volcanic gases are in thermodynamic equilibrium with the magmas from which they are released [Holland, 1978], the ratios of H2/H2O and CO/CO2 are determined from

H2 1=2O2 $ H2 O CO 1=2O2 $ CO2

and are therefore proportional to fO 20.5. SO2 would not vary in this manner. However,

Total volcanic sink

Surface H2 +CO 1013 Modern Borg 1012

Submarine H2S Submarine CH4

1011 IW

QFM

1010 0

[9]

2000GC000114

1014

O2 Sink (mol/yr)


1

2

3

4

∆ log fO

2

Figure 1. Magnitude of the volcanic O2 sink as a function of the oxygen fugacity of mantle source regions. Borg is the organic carbon burial rate and the presumed rate of oxygen production if most of the organic matter came from oxygenic photoautotrophs. logfO2 is relative to the iron-wustite (IW) buffer. Calculations were performed at T = 1500 K and P = 1 bar for subaerial volcanic gases and T = 900 K and P = 400 bar for subaqueous volcanoes, using the thermodynamic data from JANAF [Chase et al., 1986].

Holland [1978] points out that sulfur is overrepresented in the volcanic gases he has studied; hence we assume that the reducing potential observed in SO2 would be released globally as H2 and/or CO. The reducing potential of vent fluids is primarily in the form of sulfide, which is already fully reduced, so we assume that this O2 sink would remain constant at lower mantle fO2 values. An additional O2 sink, not discussed above, that becomes important at low fO2 values is hydrothermal CH4. The methane content of vent fluids is governed by CO2 2H2 O $ CH4 2O2 ;

so the CH4/CO2 ratio is proportional to fO 22. CH4 constitutes only 1% of the carbon in vent fluids today [Welhan, 1988], but it would become the dominant carbon-bearing species if fO2 were more than 1 log unit lower than


3

G


FMQ. At this point, it becomes a significant sink for O2. [11] Inspection of Figure 1 indicates that the total volcanic O2 sink rises above the photosynthetic O2 source at fO2 = 2 log units below FMQ, or roughly halfway (on a log scale) between the FMQ and IW buffers. This is the point at which one would expect the atmosphere to make a transition from reducing to oxidizing conditions. At first glance, it appears discouraging for our model because it implies that much of the oxidation of the upper mantle must have taken place subsequent to 2300 Ma. Once the atmosphere had become oxidizing, hydrogen escape to space would have slowed to its present, essentially negligible rate, so any further oxidation of the mantle as a whole must have come at the expense of reduction of the crust. A growth of the organic carbon reservoir in sedimentary rocks could contribute to this deficit; Des Marais et al. [1992] estimate an increase in this inventory of 4 1020 moles subsequent to 2200 Ma. This is only 6% of the 7 1021 moles of O2 needed to oxidize the upper mantle from IW to FMQ [Kasting et al., 1993], so it seems unlikely that this could account for a 2 log unit shift in fO2. [12] The other serious stumbling block for the mantle redox evolution theory of the rise of atmospheric O2 is evidence from trace element partitioning that mantle oxygen fugacities were near FMQ in the Archean [Delano, 1993]. The partitioning of vanadium between komatiite liquid and olivine is sensitive to fO2. Canil [1997] utilized this dependence to estimate the fO2 of six Archean komatiites, ranging in age from 2700 to 3500 Ma. He concluded that the mantle source region for komatiites was unlikely to have been significantly more reduced than today. However, he recognized that sulfides in Archean diamonds indicate that their source region (albeit perhaps

2000GC000114

volumetrically small) was significantly more reduced (4±5 log units lower fO2), implying considerable spatial and/or temporal heterogeneity of mantle redox state. Thus, although his work does indicate that some regions of the Archean upper mantle were near FMQ, it does not necessarily disprove the hypothesis that the upper mantle as a whole has evolved in its oxidation state since 3500 Ma [cf. Lecuyer and Ricard, 1999]. In contrast, previous considerations of mantle redox evolution have been based on the tacit assumption that the upper mantle source region for volcanic gases evolved homogeneously, i.e., that its mixing time was short relative to the rate of input of oxidized crustal material [Kasting et al., 1993]. However, geochemical and geophysical data argue for considerable compositional heterogeneity of the present-day mantle, supporting the new model of mantle structure proposed by Albarede and van der Hilst [1999] [see also Tackley, 2000]. Mantle tomography reveals distinct regions where subducted lithospheric slabs have penetrated the lower mantle but appear to have retained their integrity, as revealed by anomalously fast seismic wave velocities. Source regions for modern mid-ocean ridge basalts (MORB) are degassed and depleted in rare earth elements, K, U, Th, Rb, and Ba (the ``lithophile'' elements), whereas oceanic island basalts (OIB) are less degassed (i.e., they are relatively enriched in 3He). This implies a deep mantle source, but OIB show clear isotopic signatures of recycled oceanic crust. Albarede and van der Hilst [1999] use these and other observations to argue for the existence of a vast region of undepleted, undegassed (primordial) mantle situated above the core-mantle boundary (CMB) (Figure 2). The depressions in this interface are envisioned as the repositories (``graveyards'') for subducted lithospheric materials and the source regions [13]


3

G


for oceanic island basalts as plumes detach and rise to the surface. In contrast, MORB is sourced by a more rapidly mixed, depleted, and outgassed asthenosphere. The asthenosphere is separated from the primordial mantle region by a transition zone (from 400 to 1000 km), whose viscosity contrast to the asthenosphere can deflect subducting plates, and a lower mantle (1000±2000 km) that is mixed more slowly.

3. Archean Mantle Dynamics and Redox Evolution: A New Hypothesis [14] This view of mantle structure provides a nice solution to the problems noted above with the mantle redox evolution model of Kasting et al. [1993], along with a consistent explanation for the timing of the apparent rise of atmospheric O2. During the Archean, volcanic gases were derived from mantle source regions that were more reduced than today. The associated oxygen sink exceeded the substantial oxygen source associated with cyanobacterial productivity and organic carbon burial, maintaining oxygen concentrations at vanishingly small values. Ferrous iron in basalts that erupted on the ocean floor or land surface reacted with water to generate hydrogen gas and ferric iron. The hydrogen produced in this way was utilized by the biota or escaped to space, while the oxidized oceanic lithosphere was subducted and began accumulating in the lithospheric graveyard near the core-mantle boundary. This transfer of oxidized material to the base of the mantle created an `ùpside-down Earth,'' with oxidized material at depth isolated from reduced upper mantle source regions for volcanic gases (Figure 2). The sink for atmospheric oxygen was not diminished because there was little net oxidation of the upper mantle. [15] The accumulation of oxidized mantle near the CMB requires penetration of oceanic slabs

2000GC000114

through the mantle and fairly efficient heat transfer from the lower to upper mantle to prevent strong thermal stratification and thermally driven overturns of the mantle. Models of Archean mantle dynamics indicate that the ability of lithospheric slabs to penetrate the mantle depends on the magnitude of the Clapeyron slope of the pressure-induced phase transition of mantle minerals and the stiffness of the slab. Lower efficiency of heat transfer could have facilitated mantle layering during the Archean, punctuated by episodic whole mantle overturning once a critical geothermal gradient across the phase-change barrier was exceeded, a model preferred by Davies [1995]. In Davies' models, even with high thermal efficiency and stiff slabs, the mantle remains thermally stratified through much of the Archean. Slabs first penetrate into the lower mantle at 2700 Ma, although the timing is sensitive to model parameters. If his preferred model is correct, then the upper mantle may have become more rapidly oxidized than it does in our model, and there could have been more than one Archean ``rise'' of oxygen, each terminated by a mantle overturning event [cf. Barley et al., 1998]. The geologic record indicates two periods of intense magmatic plume activity in the Late Archean and earliest Paleoproterozoic [Barley et al., 1997; Condie, 1997; Heaman, 1997]. The first event (2740±2660 Ma [Barley et al., 1998]) generated komatiites from oxidized deep-mantle sources, perhaps explaining the vanadium partitioning pattern of these rocks. The second event, dated at approximately the Archean-Proterozoic boundary (2470 ± 2450 Ma), was a widespread plume event that culminated in the emplacement of a large igneous province (LIP) recognized on several continents [Barley et al., 1997; Heaman, 1997]. This event brought even more oxidized Archean mantle to the surface, ultimately resulting in a reduction in the volcanic O2 [16]


3

G


2000GC000114

oxidized reduced primordial

Figure 2. Model of Late Archean mantle structure and dynamics, depicting heterogeneity of mantle redox states. Basalts erupted at Earth's surface become oxidized. The oxidized slabs of oceanic lithosphere are subducted, penetrating into the lower mantle and accumulate at the core-mantle boundary. Plumes carry oxidized mantle back to the surface. Source regions for mid-ocean ridge basalts remain reduced through the Archean. Interior structure (but not redox characteristics) after Albarede and van der Hilst [1999].

sink below the O2 source. On the short term, however, oxygen levels remained low because of the high oxygen demand from increased overall rates of volcanism. As the plume episode waned, the demand on oxygen from this less reduced source of volcanic gas fell, ultimately dropping below its preevent level. As soon as it fell below the oxygen production

rate, atmospheric O2 rose essentially instantaneously to a level that again ensured that the oxygen production and consumption rates (now dominated by organic carbon burial and oxidation during weathering, respectively) were in balance. Although this event would have occurred when the mantle source regions averaged two log units below FMQ, the upper


3

G


mantle would have continued to increase in fO2 as more and more oxidized material was exhumed from the Archean lithospheric graveyard and recycled into the upper mantle. In a sense, the rise of oxygen was inevitable. Adopting the ferric iron subduction rate of Lecuyer and Ricard [1999] of 1.2 1013 mol Fe(III)/yr, the accumulation of ferric iron necessary to transform the upper mantle from IW to FMQ [Kasting et al., 1993] would require 2.3 billion years. The scenario we propose requires somewhat faster rates of transfer of oxidized crust to the deep mantle, depending on when slab penetration first became possible. Increased plume activity in the Late Archean simply accelerated the oxidation of the upper mantle in two step changes, the latter one taking the system through the crossover point to an oxic atmosphere. [17]

[18] The interval between the plume event at 2470±2450 Ma and the rise of oxygen (sometime before 2200 Ma) was a time of major environmental change. It began with extensive deposition of oxide-facies BIF, best represented by the Hamersley Province BIF of Western Australia, which was deposited contemporaneously with the LIP [Barley et al., 1997]. BIF deposition on shelf environments likely was associated with substantial seafloor magmatic and hydrothermal activity. Atmospheric oxygen levels could have been quite low at this time. As in the earlier Archean, cyanobacterial photosynthesis may have created oxygenated surface waters that existed dynamically in the face of supply of reductant from both above and below [Kasting, 1987; Kasting et al., 1992]. The period ended with a very deep (possibly global) glaciation [Kirschvink et al., 2000], which may have been triggered by the loss of methane from the atmosphere as it became oxygen-rich [Pavlov et al., 2000].

2000GC000114

4. Summary With Extensions The hypothesis requires that low fO 2 source regions dominated Archean volcanic emissions, and this is certainly testable. How important were plumes and arc volcanoes (potentially at FMQ) compared to mid-ocean ridge volcanoes, which would have been derived from passive upwelling of shallow, and according to our hypothesis, more reduced upper mantle? Also, were these source areas indeed more reduced? Clearly needed are the sort of data Canil [1997] has obtained on the fO2 of komatiites for other types of Archean volcanic source rocks, especially MORB. A major uncertainty is the nature of Archean tectonics and the timescale of deep mantle mixing. The scenario described here requires slabs of oxidized mantle to penetrate the mantle transition zone and accumulate near the coremantle boundary for hundreds of millions of years. These timescales are not unreasonable given our knowledge and its uncertainty concerning Archean mantle dynamics [Davies, 1995; Tackley, 2000]. [19]

[20] However, alternative scenarios are possible. One such scenario, described above, involves mantle overturnings preceding the time when slabs gain the ability to penetrate into the lower mantle (2700 Ma), the favored model of Davies [1995]. Prior to this time, the upper mantle would have become oxidized and oxygen levels could have risen. In this model the upper mantle would evolve to higher redox states between episodes of whole mantle overturn, which would replace the upper mantle with reduced material, setting back the redox evolution. The ``final'' rise of oxygen would occur only after such vigorous overturnings subsided. The appeal of our preferred hypothesis is that it provides a causative link between remarkable features of the late Archean and early Paleoproterozoic: widespread plume activity, iron formation deposi-


3

G


tion, the rise of atmospheric oxygen, and global glaciation.

Acknowledgments L.R.K. and J.F.K. acknowledge support NASA Astrobiology Institute Cooperative Agreement (NCC21057). M.E.B. has been supported by the Australian Research council, the Australian Minerals Industry Research Association, and the Minerals and Energy Research Institute of Western Australia. He also thanks Brian Krapez for his many contributions to the BIF and greenstone belt work and for review of the manuscript. Richard Arculus and two anonymous reviewers provided considerable guidance in revising this manuscript. We are also indebted to H. D. Holland for many stimulating discussions on the timing and causes of the rise of oxygen.

[21]

References Albarede, F., and R. D. van der Hilst, New mantle convection model may reconcile conflicting evidence, Eos Trans. AGU, 80(45), 535 ± 537, 1999. Arculus, R. J., Oxidation status of the mantle; past and present, Annu. Rev. Earth Planet. Sci., 13, 75 ± 95, 1985. Barley, M. E., A. L. Pickard, and P. J. Sylvester, Emplacement of a large igneous province as a possible cause of banded iron formation 2.45 billion years ago, Nature, 385, 55 ± 58, 1997. Barley, M. E., B. Krapez, D. I. Groves, and R. Kerrich, The Late Archaean bonanza: Metallogenic and environmental consequences of the interaction between mantle plumes, lithospheric tectonics and global cyclicity, Precambrian Res., 91, 65 ± 90, 1998. Beukes, N. J., and C. Klein, Models for iron-formation deposition, in The Proterozoic Biosphere: A Multidisciplinary Study, edited by J. W. Schopf, pp. 147 ± 151,Cambridge Univ. Press, New York, 1992. Braterman, P. S., A. G. Cairns-Smith, and R. W. Sloper, Photo-oxidation of hydrated Fe 2+ ; significance for banded iron formations, Nature, 303, 163 ± 164, 1983. Brocks, J. J., G. A. Logan, R. Buick, and R. E. Summons, Archean molecular fossils and the early rise of eukaryotes, Science, 285, 1033 ± 1036, 1999. Canil, D., Vanadium partitioning and the oxidation state of Archaean komatiite magmas, Nature, 389, 842 ± 845, 1997. Chase, M. W., Jr., C. A. Davies, J. R. Downey Jr., D. J. Frurip, R. A. McDonald, and A. N. Syverud, JANAF Thermochemical Tables, 3rd ed., J. Phys. Chem. Ref. Data, 14, 1 ± 1856, 1986.

2000GC000114

Cloud, P. E., A working model of the primitive Earth, Am. J. Sci., 272, 537 ± 548, 1972. Collerson, K. D., and B. S. Kamber, Evolution of the continents and the atmosphere inferred from Th-U-Nb systematics of the depleted mantle, Science, 283, 1519 ± 1522, 1999. Condie, K. C., Episodic growth of juvenile crust and catastrophic events in the mantle, Mem. Natl. Inst. Polar Res., Spec. Issue, 53, 1 ± 7, 1997. Davies, G. F., Punctuated tectonic evolution of the earth, Earth Planet. Sci. Lett., 136, 363 ± 379, 1995. Delano, J. W., Oxidation state of the Earth's upper mantle during the last 3800 million years; implications for the origin of life, Lunar Planet. Sci., 24, 395 ± 396, 1993. Derry, L. A., and S. B. Jacobsen, The chemical evolution of Precambrian seawater; evidence from REEs in banded iron formations, Geochim. Cosmochim. Acta, 54, 2965 ± 2977, 1990. Des Marais, D. J., Isotopic evolution of the biogeochemical carbon cycle during the Proterozoic Eon, Org. Geochem., 27, 185 ± 193, 1997. Des Marais, D. J., H. Strauss, R. E. Summons, and J. M. Hayes, Carbon isotope evidence for the stepwise oxidation of the Proterozoic environment, Nature, 359, 605 ± 609, 1992. Heaman, L. M., Global mafic magmatism at 2.45 Ga: Remnants of a large igneous province?, Geology, 25, 299 ± 302, 1997. Holland, H. D., The Chemistry of the Atmosphere and Oceans, 351 pp., John Wiley, New York, 1978. Holland, H. D., The Chemical Evolution of the Atmosphere and Oceans, 582 pp., Princeton Univ. Press, Princeton, N. J., 1984. Holland, H. D., Early Proterozoic atmospheric change, in Early Life on Earth, edited by S. Bengtson, pp. 237 ± 244, Columbia Univ. Press, New York, 1994. Karhu, J. A., and H. D. Holland, Carbon isotopes and the rise of atmospheric oxygen, Geology, 24, 867 ± 870, 1996. Kasting, J. F., Theoretical constraints on oxygen and carbon dioxide concentrations in the Precambrian atmosphere, Precambrian Res., 24, 205 ± 229, 1987. Kasting, J. F., Earth's early atmosphere, Science, 259, 920 ± 926, 1993. Kasting, J. F., J. F. Kump, and H. D. Holland, Atmospheric evolution; the rise of oxygen, in The Proterozoic Biosphere: A Multidisciplinary Study, edited by J. W. Schopf, pp. 159 ± 163, Cambridge Univ. Press, New York, 1992. Kasting, J. F., D. H. Eggler, and S. P. Raeburn, Mantle redox evolution and the oxidation state of the Archean atmosphere, J. Geol., 101, 245 ± 257, 1993.


3

G


Kirschvink, J. L., E. J. Gaidos, E. Bertani, N. J. Beukes, J. Gutzmer, L. N. Maepa, and R. E. Steinberger, Paleoproterozoic snowball Earth: Extreme climatic and geochemical global change and its biological consequences, Proc. Natl. Acad. Sci., 97, 1400 ± 1405, 2000. Kump, L. R., and H. D. Holland, Iron in Precambrian rocks: Implications for the global oxygen budget of the ancient Earth, Geochim. Cosmochim. Acta, 56, 3217 ± 3223, 1992. Lecuyer, C., and Y. Ricard, Long-term fluxes and budget of ferric iron: Implication for the redox states of the Earth's mantle and atmosphere, Earth Planet. Sci. Lett., 165, 197 ± 211, 1999. Morton, J. L., and N. M. Sleep, A mid-ocean ridge thermal model; constraints on the volume of axial hydrothermal heat flux, J. Geophys. Res., 90, 11,345 ± 11,353, 1985. Ohmoto, H., Evidence in pre-2.2 Ga paleosols for the early evolution of atmospheric oxygen and terrestrial biota, Geology, 24, 1135 ± 1138, 1996. Palmer, M. R., and J. M. Edmond, The strontium isotope budget of the modern ocean, Earth Planet. Sci. Lett., 92, 11 ± 26, 1989. Pavlov, A. A., J. F. Kasting, L. L. Brown, K. A. Rages, and R. Freedman, Greenhouse warming by CH4 in the atmosphere of early Earth, J. Geophys. Res., 105, 11,981 ± 11,990, 2000.

2000GC000114

Schidlowski, M., J. M. Hayes, and I. R. Kaplan, Isotopic inferences of ancient biochemistries: Carbon, sulfur, hydrogen and nitrogen, in Earth's Earliest Biosphere: Its Origin and Evolution, edited by J. W. Schopf, pp. 149 ± 186, Princeton Univ. Press, Princeton, N. J., 1983. Schopf, J. W., Microfossils of the early Archean Apex Chert; new evidence of the antiquity of life, Science, 260, 640 ± 646, 1993. Seyfried, W. L., and D. R. Janecky, Heavy metal and sulfur transport during subcritical and supercritical hydrothermal alteration of basalt; influence of fluid pressure and basalt composition and crystallinity, Geochim. Cosmochim. Acta, 49, 2545 ± 2560, 1985. Tackley, P. J., Mantle convection and plate tectonics: Toward an integrated physical and chemical theory, Science, 78, 2002 ± 2007, 2000. Walker, J. C. G., C. Klein, M. Schidlowski, J. W. Schopf, D. J. Stevenson, and M. R. Walter, Environmental evolution of the Archean-Early Proterozoic Earth, in Earth's Earliest Biosphere: Its Origin and Evolution, edited by J. W. Schopf, pp. 260 ± 290, Princeton Univ. Press, Princeton, N. J., 1983. Welhan, J. A., Origins of methane in hydrothermal systems, Chem. Geol., 71, 183 ± 198, 1988.