Box models for the evolution of atmospheric oxygen: an update

Palaeogeography, Palaeoclimatology, Palaeoecology (Global and Planetary Change Section), 97 (1991): 125-131

125

Elsevier Science Publishers B.V., Amsterdam

Box models for the evolution of atmospheric oxygen: an update James F. Kasting Department of Geosciences, The Pennsylvania State Un&ersi~, University Park, PA I6802, USA (Received December 3, 1990; accepted February 1, 1991)

ABSTRACT Kasting, J.F., 1991. Box models for the evolution of atmospheric oxygen: an update. Palaeogeogr., Palacoclimatol., Palacoecol. (Global Planet. Change Sect.), 97: 125-131. A simple 3-box model of the atmosphere/ocean system is used to describe the various stages in the evolution of atmospheric oxygen. In Stage I, which probably lasted until redbeds began to form about 2.0 Ga ago, the Earth's surface environment was generally devoid of free O2, except possibly in localized regions of high productivity in the surface ocean. In Stage II, which may have lasted for less than 150 Ma, the atmosphere and surface ocean were oxidizing, while the deep ocean remained anoxic. In Stage III, which commenced with the disappearance of banded iron formations around 1.85 Ga ago and has lasted until the present, all three surface reservoirs contained appreciable amounts of free 0 2. Recent and not-so-recent controversies regarding the abundance of oxygen in the Archean atmosphere are identified and discussed. The rate of 0 2 increase during the Middle and Late Proterozoic is identified as another outstanding question.

Introduction A convenient way of trying to understand the complex behavior of the atmosphere-ocean system is by using box models. Elsewhere (Kasting, 1987; Kasting et al., in press), I have argued that a 3-box model, consisting of the atmosphere, the surface ocean, and the deep ocean, can be used to constrain possible variations in atmospheric oxygen concentration through geologic time. Here, I update the discussion to bring it into accord with developments over the past two years. Before doing so, however, I wish to point out that the 3-box model has no predictive capability by itself. The factors controlling atmospheric oxygen are too complicated to be reliably simulated by such a crude model. Indeed, even more sophisticated models are presently incapable of predicting atmospheric O 2 levels from first principles. Models can, however, be used to aid in the interpretation of evidence obtained from the geologic record. That is the strategy followed below. In what follows, I hope to convince the reader that

thinking about atmospheric oxygen in the context of a simple model does help to clarify what is possible and what is not.

Stages of atmospheric evolution Within the framework of the 3-box model, the rise of oxygen can be divided into three stages (Fig. 1). In the 'reducing' Stage 1, all three reservoirs are essentially devoid of free 0 2. The actual atmospheric oxygen concentration can be estimated by treating O 2 as a trace constituent and calculating its abundance with a photochemical model (Kasting, 1979; Kasting et al., 1984). The predicted ground-level O 2 mixing ratio is of the order of 10 -1° PAL (times the present atmospheric level) or below. Earth's atmosphere should have remained in Stage 1 until the amount of 0 2 produced by photosynthesis followed by organic carbon burial was sufficient to overwhelm the flux of reduced gases from volcanos (Walker, 1977; Holland, 1978) and from photooxidation of iron in the surface ocean (Braterman et al., 1983;

0921-8181/91/$03.50 © 1991 - Elsevier Science Publishers B.V. All rights reserved

126

J .F. KASTING

somewhat weakened. If the Earth's atmosphere did pass through such a stage, the 3-box model can be used to estimate an upper limit on pO 2 during this time (Kasting, 1987)

ATMOSPHERE R

0

0

OCEAN 0

OCEAN 0

R

0

Borg SURFACE O

P02 <- ka

DEEP OCEAN R

Fig. 1. Three-stage box model for the evolution of atmospheric oxygen.An 0 indicates the presenceof free Oz; an R indicates reducing conditions. The shaded area in Stage I represents a localized 'oxygen oasis' in an otherwise anoxic surface ocean.

Kasting et al., 1984). Certain regions of the surface ocean where photosynthetic activity was high could have formed localized 'oxygen oases'in which these generally reducing conditions did not apply (Kasting et al., in press). This is indicated pictorially by the small, shaded region in Fig. 1. The dissolved oxygen partial pressure in these oases can be estimated by balancing the photosynthetic production rate of oxygen with its loss by diffusion to the atmosphere. Values as high as 10% percent of the present dissolved 0 2 concentration are theoretically possible. So, an anoxic atmosphere does not necessarily imply a uniformly reducing surface ocean. In the 'oxidizing' Stage II, the atmosphere and surface ocean contained appreciable concentrations of free 0 2 but the deep ocean remained anoxic. The perceived need for this stage comes from the simultaneous occurrence of redbeds and banded iron formations (BIF's) during the early Proterozoic (Kasting, 1987). The redbeds indicate that free O 2 was present in the atmosphere at a level of ~ 10 - 7 PAL or more (Kasting and Walker, 1981); the BIF's imply that the deep ocean was anoxic (Holland, 1973). As discussed below, the time period during which these conditions are both met may be short, or even nonexistent, so the theoretical motivation for this stage is

where Borg ( = 1013 moles C yr -1 at present) is the burial rate of organic carbon, k ( = 1.3 × 1021 liters/103 yr) is the turnover rate of the deep ocean, and a ( = 1.28 × 10 -3 M a t m -I at 15°C) is the Henry's Law coefficient for 0 2. This limit is derived by balancing the production rate of oxygen with its rate of consumption by reaction with ferrous iron in the deep ocean. If one assumes that Bor~ and k have not changed with time, then the predicted upper limit o n p O 2 is 6 × 10 -3 atm, or 0.03 P A L The 'aerobic' Stage III atmosphere corresponds to the modern system in which the atmosphere, surface ocean, and deep ocean all contain appreciable quantities of free 0 2. The 3-box model can be used to set a lower limit on the atmospheric oxygen concentration during this stage (Kasting et al., in press) SMoR P02 > ka where SMoR ( = 4 × 101~ moles 0 2 yr -I at present) is the oxygen demand of hydrothermal fluids emanating from the midocean ridges. This limit comes from the requirement that the 0 2 content of downwelled surface water be sufficient to oxidize the iron and sulfide in the vent fluids. For modern values of StubR and k, the calculated lower limit on 0 2 is 4 × 10 -4 atm, or 0,002 PAL. It is considered likely that vent activity was more vigorous in the past as a consequence of higher internal heat flow from the young Earth. Thus, this lower limit might be adjusted upward by a factor of perhaps two or three, depending on when Stage III is considered to have been reached. More stringent lower limits on pO 2 during Stage III can be set by considering the metabolic requirements of fossil organisms during the Midto Late Proterozoic, when Stage III is thought to have been reached (Runnegar, this issue; see also

127

BOX M O D E L S F O R T H E E V O L U T I O N O F A T M O S P H E R I C O X Y G E N

Fig. 2). These metabolic constraints are, of course, predicated on the assumption that the regions where the organisms lived were representative of the surface ocean as a whole. This assumption is not wholly secure, given the possible existence of oxygen oases. Figure 2 has, however, been drawn as if this assumption is valid. Evidence that the fossil organisms in question (Grypania, Dickinsorim) were widely distributed would help to relieve any lingering doubts on this matter. An additional, quasi-biological constraint that applies to more recent Earth history is that pO 2 must have been at least 0.6 PAL (13% by volume) since the Devonian to account for the continuity of the fossil charcoal record (Jones and Chaloner, this issue). More speculative is their upper limit of 1.7 PAL (35% by volume) based on the idea that the forests would have burned down entirely had pO 2 exceeded this value.

Timing of the various stages Most of the controversy surrounding the application of the 3-box model to Earth history concerns the timing of the transitions between the various stages. Elsewhere (Kasting, 1987), 1 proposed that Stage II began around 2.4 Ga before present and that Stage III commenced at approximately 1.7 Ga. This interpretation was based on the data presented in figs. 11-12 of Walker et at. (1983), specifically the first appearance of redbeds at 2.4 Ga and the last appearance of BIF's at 1.7 Ga (excluding the Late Proterozoic Rapitan and Urukum formations, which are thought to be different from earlier deposits.) Both of these dates have since been revised. According to Lowe (in press), those redbeds dated prior to 2.0 Ga are all suspect. At the same time, the date of the most recent BIF has been pushed back to at least

10'

10 2

<

~O

10.2

10 ~ 1~ 13..

O

.lo ~

10~

o: 0

E 0 0

10S

10-12 10 '~

3.5

2.5

1.5

0,5

Time before present (Ga) Fig. 2. Estimated change in atmospheric oxygen over geologic time. The shaded area shows the range of 0 2 partial pressures permitted by various interpretations of the data. The solid vertical lines indicate the most likely dates for the transitions between stages. The dashed vertical line at 2.4 Ga indicates a possible earlier date for the beginning of Stage II that is consistent with the paleosol data and with some disputed redbed data (Kasting et al., in press). The dashed horizontal lines show the theoretical limits on pO 2 during Stages I1 and III derived from the 3-box model. The solid vertical bar represents data from Early Proterozoic paleosols (Kasting et al., in press). Points labelled 'Eukaryotes', 'Grypania' and 'Dickinsonia" are from Runnegar (this issue). Arrows labelled 'uraninite survival' are discussed by Kasting et al. (in press); the arrow labelled 'abundant phytoplankton' is from Kasting (1987). Other constraints are discussed in the text.

128

1.85 Ga (Beukes and Klein, in press). If these new dates are correct, Stage II shrinks from 700 m.y. to only 150 m.y. in duration. Were the dates on the most recent BIF to change by a small amount, Stage II would disappear entirely and the model would collapse to the two-stage model of Cloud (1972): the atmosphere would then be either reducing or oxidizing. Not everyone agrees that the atmosphere switched from reducing to oxidizing near 2.0 Ga. Holland (1984) and Grandstaff et al. (1986) favor low, hut finite (10 -4 to 10 -2 PAL), 0 2 levels during the Archean and Early Proterozoic to explain the retention of iron in paleosols derived from weathering of granitic rocks. Though small by comparison to today, such 0 2 levels would have been sufficient to prevent the buildup of reduced gases such as CH 4 or H2, which could have reached concentrations of tens to hundreds of parts per million in a Stage I atmosphere (Kasting et al., 1983, 1984). Towe (1981, 1983, 1990) suggests that an oxidizing Archean atmosphere is required to provide an effective UV screen, to explain the detailed structure of BIF's, and to effectively recycle the organic matter produced by photosynthesis. In my view, none of these arguments in favor of an oxidizing Archean atmosphere is persuasive. Only three Archean paleosols have been identified and their interpretation is not at all straightforward (Grandstaff et al., 1986; Holland et al., 1989). It is difficult to see why these isolated analyses should outweigh the redbed data (see discussion below). An Archean UV screen may be unnecessary (Margulis et al., 1976; Rambler and Margulis, 1980), or it might have been provided by hydrocarbon smog produced from methane photolysis (Lovelock, 1988) or by S8 vapor produced from photolysis of SO 2 (Kasting et al., 1989). BIF's could have been formed photochemically in the complete absence of 02 (Braterman et al., 1983), or they could have formed from upwelling of Fe2+-rich deep water into the oxygen oases suggested above. The argument that 0 2 was required to recycle organic matter (Towe, 1990) is spurious because it assumes that a fixed fraction (20%) of the primary organic matter is buried (see discussion below).

J.F. KASTI NG

In short, there is little reason to overturn the suggestion by Cloud (1972) that the Archean atmosphere was oxygen-free. Interpretation of the redbed record Two of the above items merit further discussion because there has been some misunderstanding about them in the literature and elsewhere. The first concerns the significance of redbeds. To my mind the biggest question is not "When did they first appear?" but, rather, What does the absence of redbeds imply about atmospheric oxygen levels?" One interpretation (Holland, 1984) is that the appearance of redbeds simply signifies a rise in pO 2 above some poorly defined critical level (10 3 to 10- ~ PAL?). In this view the atmosphere could have been oxidizing throughout the entire geologic record. This is certainly possible if oxygenic photosynthesis evolved very early, as suggested by some authors (Schopf et al., 1983). A second interpretation (Kasting and Walker, 1981) is that the appearance of redbeds indicates a switch from a weakly reducing, Stage I atmosphere to an oxidizing, Stage II atmosphere. Photochemical models predict that this transition should occur at or below 10 -7 PAL. The transition is marked by a change in the ground-level abundances of various free radicals. In the Kasting and Walker (1981) model, atomic H is the most abundant free radical in Stage I, whereas OH, HO2, and H 2 0 z predominate in Stage II. The argument, then, is that iron on the surface of sediment grains would have been either oxidized or reduced by reactions with these free radicals, depending on which ones were most abundant. Free 0 2 concentrations of 10 -4 to 10 -2 PAL, sufficient to explain the paleosol data, should have resulted in a clear preponderance of oxidizing free radicals and the consequent formation of redbeds. So, according to this view, the Archean/ Early Proterozoic atmosphere must have been reducing if Lowe's evaluation of the redbed record is correct. The question of how redbeds form deserves further investigation. Is the iron oxidized by direct contact with the atmosphere or by oxidants dissolved in groundwater? Is 0 2 the dominant

BOX MODELS FOR THE EVOLUTION OF ATMOSPHERIC OXYGEN

129

oxidant during Stage II or are photochemicallyproduced trace species important (Kasting et al., 1985)? Would HCO and HzCO have been the dominant reductants in a weakly reducing, highCO 2 atmosphere (Kasting et al., 1984) and how would this have affected the transition from Stage I to Stage II? To what extent would oxidizing and reducing free radicals have reacted with each other in rainwater and in groundwater? An integrated atmospheric/hydrologic/geologic approach is needed to answer these questions.

by Veizer and Jansen (1985) and the approximately constant observed weight percent of organic carbon in sedimentary rocks of all ages (Holland, 1984; Schidlowski, 1988). Any global values derived from these numbers must be highly uncertain, given the small percentage of Archean sediments that has been preserved until the present. The burial fraction is based on values observed in high-productivity upwelling regions in the modern ocean (Muller and Suess, 1979). There is no good reason to believe that these same values apply to the Archean. Even if Towe's figures were correct, his conclusion that they require an O2-rich atmosphere is not. In the first place, aerobic respiration might have proceeded in local oxygen oases even if the atmosphere itself was anoxic (see above). This would reduce the amount of carbon that must have been recycled anaerobically. Second, other anaerobic decay processes such as sulfate reduction could have operated in conjunction with methanogenesis. Sulfate reduction would also have consumed 02, since the sulfate would presumably have been derived from photochemical oxidation of SO 2 and H2S (Walker and Brimblecombe, 1985). Finally, if anaerobic decay processes were unable to recycle the required 2 × 1013 moles (C) yr -1, the burial fraction would have necessarily increased, implying that primary productivity must have been lower than Towe assumes. After all, the observational constraint is on the carbon burial rate, not the rate of primary productivity. More to the point is Towe's argument that even 5 × 1012 moles yr-1 of photosynthetic 0 2 would have overwhelmed the available oxygen sinks. Towe estimates that perhaps 2.2× 1012 moles 02 yr -1 could have been consumed by oxidation of ferrous iron. This number is based on Veizer and Jansen's sedimentation rate and an estimated 5% iron content in average sediments (Holland, 1984). The remaining 2.8 × 1012 moles 02 yr -1 would have to be consumed by reaction with reduced volcanic gases, primarily H2, CO, and SO 2. The estimated present release rates of these gases are 11 × 10 it moles yr -1 for H 2 (Holland, 1978, p. 292), 3 x 1011 moles yr-l for CO (ibid., p. 291), and 2 x 1011 moles yr -1 for

Aerobic versus anaerobic recycling of organic matter

A second point of contention concerns the recent assertion (Towe, 1990) that the Archean atmosphere must have contained enough free oxygen (pO 2 > 0.01 PAL) to allow recycling of organic carbon by aerobic respiration. Otherwise, says Towe, the flux of oxygen produced by photosynthesis CO 2 +

H 2 0 --+ CH20

+ 0 2

would have swamped the available 02 sinks, resulting in a logical inconsistency. To arrive at this conclusion, Towe assumes an Archean organic carbon burial rate of 5 × 1012 moles yr-t and a burial fraction of 20%. This requires a net primary production rate of 2.5 x 1013 moles (C) yr -1, 80% of which must be recycled to CO 2. Towe argues that anaerobic processes, specifically methanogenesis 2 CH20 ~ CH 4 +

CO 2

followed by methane oxidation CH 4 + 2 02 -~ CO2 + 2H20 would have been unable to recycle this much organic carbon. Even if they could, he says, the remaining 5 × 1012 moles 02 yr-1 corresponding to the buried carbon exceeds the amount that could have been consumed by reaction with ferrous iron and reduced volcanic gases. There are several flaws in this argument, the first of which is the assumption that we really know the organic carbon burial rate and burial fraction in the Archean. The burial rate is derived from a global sedimentation rate estimated

130 SO 2 (Berresheim and Jaeschke, 1983). (Holland's estimate for SO 2 emissions is probably too high.) Since one mole of each of these gases consumes 0.5 moles of 0 5, the present 0 2 consumption rate is about 8 × l0 II moles yr -t. The Archean volcanic outgassing rate would have to have been 3.5 times higher than this to neutralize the remaining photosynthetic 0 2 in Towe's model. Given the higher internal heat flow at that time, such an increase in volcanic activity does not seem unreasonable. An anoxic atmosphere would also result if Towe's organic carbon burial rate were reduced by a factor of two or three. Given the large uncertainties in the carbon burial rate, the ferrous iron flux, and the volcanic outgassing rate, it is impossible to predict whether the Archean atmosphere should have been oxidizing or reducing. We must rely on the rock record to make this decision.

Atmospheric 0 5 levels during Stage III A final point of discussion concerns the rate at which atmospheric oxygen levels increased during the early parts of Stage III, that is, during the Middle to Late Proterozoic. From a theoretical standpoint it is attractive to speculate that the atmospheric 0 2 concentration rose rather slowly during the Proterozoic and that biological evolution essentially tracked this rise; that is, higher organisms evolved soon after it became metabolically possible for them to do so (Nursall, 1959; Berkner and Marshall, 1965). If this idea is correct, atmospheric 0 2 levels ought to have been near the lower end of the range shown in Fig. 2, i.e. pO 2 could have been well below 0.l PAL until the early Phanerozoic. Certain indirect arguments support this hypothesis: Runnegar (1982) suggested that the wide, flat body shape of the Ediacaran worm Dickinsonia was an evolutionary adaptation to low oxygen conditions; Knoll et al. (1986) theorized that pO 2 must have increased dramatically at the end of the Precambrian to account for the prevalence of isotopically heavy carbonates during Vendian time. (Heavy carbonates imply high rates of organic carbon burial, which in turn implies high rates of 0 2 production.) Counterbalancing these arguments is a re-

J.F. KASTING cent study by Holland and Beukes (1990) of a weathering horizon developed on the Kuruman iron formation in South Africa. Their analysis suggests that pO 2 was already > 0.15 PAL by 2.2-1.9 Ga before present. This conclusion is inconsistent with the upper limit on pO 2 deduced above for Stage II; however, it might be reconciled with the box model if the BIF's actually disappeared earlier than 1.85 Ga ago or if the rates of organic carbon burial or ocean circulation were significantly different than today.

Summary The rise of atmospheric oxygen remains poorly understood because of the inherent shortcomings of purely theoretical models and the incompleteness and occasional ambiguity of the rock record. Two specific questions on which no consensus has been reached concern the oxidation state of the Archean atmosphere and the rate of rise in atmospheric O z during the Middle to Late Proterozoic. The first question may eventually be resolved by improved models for redbed formation and by additional data from paleosols; the second question can be addressed by detailed examination of the carbon isotope record. Simple box models of the atmosphere-ocean system can aid in evaluating the geological evidence and incorporating it into a self-consistent theoretical framework.

Acknowledgements This work was supported by the National Science Foundation under grant no. ATM 8901775. The author acknowledges helpful discussions with J.C.G. Walker, H.D. Holland, and B. Runnegar.

References Berkner, L.V. and Marshall, L.C., 1965. On the origin and rise of oxygenconcentration in the Earth's atmosphere. J. Atmos. Sci., 22: 225-261. Berresheim, H. and Jaeschke, W., 1983. The contribution of volcanoes to the global atmosphericsulfur budget. J. Geophys. Res., 88: 3732-3740. Beukes, N.J. and Klein, C., in press. Models for iron-formation deposition. In: J.W. Schopf and C. Klein (Editors),

BOX M{)DEt.S FOR THE EVOLUTION OF ATMOSPHERIC OXYGEN

131

The Proterozoic Biosphere. A Multidisciplinary Study, Cambridge University Press, New York. Braterman. P.S., Cairns-Smith, A.G. and Sloper, R.W., 1983. Photooxidation of hydrated Fe+2--significance for banded iron formations. Nature, 303: 163-164. Cloud, P.E., 1972. A working model of the primitive Earth. Am. J. Sci., 272: 537-548. Grandstaff, D.E., Edelman M.J., Foster, R.W., Zbinden, E. and Kimberley, M.M., 1986. Chemistry and mineralogy of Precambrian paleosols at the base of the Dominion and Pongola Groups (Transw~al, South Africa). Precamb. Res., 32: 97-131. Holland, H.D., 1973. The oceans: A possible source of iron in iron-formations. Econ. Geol., 68:1169-1172. tlolland, H,D.. 1978. The Chemistry of the Atmosphere and Oceans. Wiley, New York. 351 pp. Holland, H.D., 1984. The Chemical Evolution of the Atmosphere and Oceans. Princeton University Press, Princeton, N.J., 582 pp. Holland, H.D. and Beukes, N.J., 1990. A paleoweatbering profile from Griqualand West, South Africa: Evidence for a dramatic rise in atmospheric oxygen between 2.2 and 1.9 B.Y.B.P. Am. J. Sci., 29(/-A: 1-34. Holland, HD., Feakes, C.R. and Zbinden, E.A., 1989. The Flin Flon paleosol and the composition of the atmosphere 1.8 BYBP. Am. J. Sci., 289: 362-389. Jones, T.P. and Chaloner. W.G., 1991. Fossil charcoal, its recognition and paleoatmospheric significance. Global Planet Change, this issue. Kasting, J.F., 197% Evolution of oxygen and ozone in the Earth's atmosphere. University of Michigan, Ann Arbor, Mich. Kasting, J.F., 1987. Theoretical constraints on oxygen and carbon dioxide concentrations in the Precambrian atmosphere. Precamb. Res., 34: 205-229. Kasting, J.F. and Walker, J.C.G., 1981. Limits on oxygen concentration in the prebiological atmosphere and the rate of abiotic fixation of nitrogen. J. Geophys. Res., 86: 1147 1158. Kasting, J.F., Zahnle, K.J. and Walker, J.C.G., 1983. Photochemistry of methane in the Earth's early atmosphere. Precamb. Res., 2/): 121-148. Kasting, J.F., Pollack, J.B. and Crisp, D., 1984. Effects of high CO, levels on surface temperature and atmospheric oxidation state of the early earth. J. Atmos. Chem., 1: 403428. Kasting, J.F., Holland, H.D. and Pinto, J.P., 1985. Oxidant abundances in rainwater and the evolution of atmospheric oxygen. J. Geophys. Res., 90: 10,497-10,510. Kasting, J.F., Zahnle, K.J., Pinto, J.P. and Young, A.T., 1989. Sulfur, ultraviolet radiation and the early evolution of life. Origins Life, 19: 95-108. Kasting, J.F.. Holland, H.D. and Kump, L.R., in press. Atmospheric evolution: The rise of oxygen. In: J.W. Schopf and C. Klein (Editors), The Proterozoic Biosphere: A Multidisciplinary Study, Cambridge University Press, New York. Knoll. A.H., Hayes, J.M., Kaufman, A.L, Swett, K. and Lain-

bert, l.B., 1986. Secular variation in carbon isotope ratios from Upper Proterozoic successions of Svalbard and East Greenland. Nature, 321: 832-838. Lovelock, J., 1988. The Ages of Gaia. Norton, New York, 252 PP. Lowe, D., in press. Other geological indicators. In: J.W. Schopf and C. Klein (Editors[, The Proterozoic Biosphere: A Multidisciplinary Study, Cambridge University Press, New York. Margulis, L., Walker, J.C.G. and Rambler. M.B., 1976. Reassessment of roles of oxygen and ultraviolet light in Precambrian evolution. Nature, 264: 62(/-624. Muller, P.J. arid Suess, E., 1979. Productivity, sedimentation rate and sedimentary organic matter in the oceans--l. Organic carbon preservation. Deep-Sea Res., 26A: 13471362. Nursall, J.R., 1959. Oxygen as a prerequisite to the origin of the metazoa. Nature, 183: 1170-1172. Rambler, M. and Margulis, L., 1980. Bacterial resistance to ultraviolet irradiation under anaerobiosis: Implications for pre-Phanerozoic evolution. Science, 210: 638-640. Runnegar, B., 1982. Oxygen requirements, biology and phylogenetic significance of the late Precambrian worm Dickinsonia and the evolution of the burrowing habit. Alcheringa, 6: 223-239. Runnegar, B., 1991. Precambrian oxygen levels estimated from the biochemistry and physiology of early eukaryotes. Global Planet. Change, this issue. Schidlowski, M., 1988. A 3,800-million-year isotopic record of life from carbon in sedimentary rocks. Nature, 333: 313318. Schopf, J.W., Hayes, JM. and Walter, M.R., 1983. Evolution of Earth's earliest ecosystems: recent progress and unsolved problems. In: J.W. Schopf (Editor), Earth's Earliest Biosphere: Its Origin and Evolution, Princeton University Press, Princeton, NJ, pp. 361-384. Towe, K.M., 1981. Environmental conditions surrounding the origin and early Archean evolution of life: a hypothesis. Precamb. Res., 16: 1-10. Towe, K.M., 1983. Precambrian atmospheric oxygen and banded iron formations: a delayed ocean model. Precamb. Res., 20: 161-170. Towe, K.M., 1990. Aerobic respiration ira the Arcbaean? Nature, 348: 54-56. Veizer, J. and Jansen, S.L., 1985. Basement and sedimentary recycling-2: Time dimension to global tectonics. J. Geol., 93: 625-643. Walker, J.C.G., 1977. Evolution of the Atmosphere. Macmillan, New York. Walker, J.C.G. and Brimblecombe, P., 1985. lron and sulfur in the pre-biologic ocean. Precamb. Res.. 28:205 222. Walker, J.C.G., Klein, C,, Schidlowski, M., Schopf, J.W., Stevenson. D.J. and Walter, M.R., 1983. Environmental evolution of the Archean-Early Proterozoic Earth. In: J.W. Schopf (Editor), Earth's Earliest Biosphere: hs Origin and Evolution, Princeton University Press, Princeton, N.J., pp. 26(/-290.