The Archean Sulfur Cycle and the Early History of Atmospheric Oxygen

See allHide authors and affiliations

Science  28 Apr 2000:
Vol. 288, Issue 5466, pp. 658-661
DOI: 10.1126/science.288.5466.658


The isotope record of sedimentary sulfides can help resolve the history of oxygen accumulation into the atmosphere. We measured sulfur isotopic fractionation during microbial sulfate reduction up to 88°C and show how sulfate reduction rate influences the preservation of biological fractionations in sediments. The sedimentary sulfur isotope record suggests low concentrations of seawater sulfate and atmospheric oxygen in the early Archean (3.4 to 2.8 billion years ago). The accumulation of oxygen and sulfate began later, in the early Proterozoic (2.5 to 0.54 billion years ago).

Life has dramatically modified the surface chemistry of Earth. A most conspicuous expression of this is the accumulation of oxygen, a product of oxygenic photosynthesis by plants and cyanobacteria, into the atmosphere and oceans. Atmospheric oxygen promotes the oxidative weathering of rocks on land, forming oxidized species such as iron oxides and soluble sulfate (1). As a result, the accumulation of sulfate into the oceans (the concentration is presently 28 mM) and the formation of iron oxides during weathering on land are two substantial geochemical expressions of oxygen accumulation into the atmosphere (1,2). Still, considerable controversy and debate surround when atmospheric oxygen first began to accumulate. In one scenario, atmospheric oxygen reached present-day levels by the earliest Archean [3.8 billion years ago (Ga)] and has persisted in high concentrations ever since (3). In another scenario, atmospheric oxygen first began to accumulate much later, around 2.2 to 2.3 Ga in the early Proterozoic (2). Present-day levels may not have been reached until sometime in the Neoproterozoic, 0.54 to 1.0 Ga (4).

The history of seawater sulfate concentrations is germane to differentiating between these two models for atmospheric oxygen accumulation. Low concentrations of seawater sulfate into the early Archean would be consistent with, and provide evidence for, low early Archean concentrations of atmospheric oxygen (5). Inferences as to the history of seawater sulfate accumulation are based, primarily, on interpretations of the sulfur isotope record of ancient sedimentary sulfide minerals. This record shows sedimentary sulfides between 3.4 and 2.8 Ga with isotopic compositions of ±5 per mil (‰) around a contemporaneous seawater sulfate isotopic composition [δ34S (6)] of 2 to 3‰ (7). The principal feature of this record is the small isotope difference between seawater sulfate and sedimentary sulfides.

The interpretation of this record is based on our understanding of the factors controlling isotope fractionation during sulfate reduction by sulfate-reducing bacteria. These bacteria are responsible for most of the sulfide formed in modern marine sediments. There is a tendency for pure cultures of sulfate-reducing bacteria to fractionate less as specific rates (rate per cell) of sulfate reduction increase (8). Increasing temperature can lead to higher specific rates for individual species of sulfate reducers, and, therefore, lower fractionations might be expected at higher temperatures. Thus, minimally fractionated early Archean sedimentary sulfides may have formed at rapid rates of sulfate reduction in a warm, sulfate-rich (10 to 28 mM) ocean (3, 9), providing support for high early Archean atmospheric oxygen concentrations (3, 9).

This ocean model requires extensions of relations between specific rates of sulfate reduction, temperature, and isotope fractionation beyond current observations. Thus, for Desulfovibrio desulfuricans, the most studied sulfate-reducing bacterium, fractionations of 10 to 26‰ are observed at 40° to 45°C, the upper temperature limit for the survival of the organism, where specific rates of sulfate reduction are also the highest (8). These fractionations are too large to explain the small fractionations preserved in the early Archean isotope record (7). Therefore, we explored the relation between temperature and isotope fractionation for natural populations of sulfate-reducing bacteria metabolizing at temperatures >45°C. For these experiments, we used sediment collected by the deep-sea submersible Alvin from an active hydrothermal vent area in the Guaymas Basin, Gulf of California (10). At the sampling location, sediment temperatures increased from 2.8°C at the sediment surface to 100°C by 15- to 20-cm depth. Sediment was sectioned into depth intervals with measured in situ temperatures of 50° to 60°C and 70° to 90°C. Sediment sections were subsequently incubated, intact, in flow-through reactors with lactate, ethanol, or acetate as substrates for sulfate-reducing bacteria and with sulfate concentrations of 2, 5, and 28 mM (11). Incubations were conducted at temperatures within the in situ range for each section, with organic substrates supplied both in excess (nonlimiting) and in concentrations limiting sulfate reduction rate.

We report the highest temperatures to which isotope fractionation during sulfate reduction has been measured and see fractionations of 13 to 28‰, at temperatures up to 85°C (Fig. 1A) (12). These high fractionations were independent of sulfate concentration between 2 and 28 mM and were observed both when organic substrate was limiting in concentration, suppressing microbial activity, and when it was nonlimiting. Specific rates of sulfate reduction could not be measured in these experiments, although the highest specific rates would be expected when organic matter was nonlimiting. Indeed, we measured the highest volume-based rates under these conditions (Fig. 1B). Our natural population fractionation results provide no support for reduced fractionations at temperatures above 40° to 45°C (13).

Figure 1

(A) Isotope fractionation is shown during sulfate reduction by mixed populations of sulfate-reducing bacteria metabolizing on ethanol, lactate, and acetate, under both limiting and nonlimiting substrate conditions with artificial seawater containing 2, 5, and 28 mM sulfate. Under limiting substrate conditions, all, or nearly all, of the substrate was used within the flow-through reactor. Under nonlimiting conditions, only a portion of the substrate was used, and considerable excess substrate exited the reactor. Isotope fractionation was calculated relative to the isotopic composition of the input sulfate, with a small correction for sulfate depletion within the reactor with a Rayleigh distillation model. No systematic differences in fractionation were observed between the substrates used or with sulfate between 2 and 28 mM, and, therefore, these data have not been indicated separately [data are available at Science Online (29)]. (B) Rates of sulfate reduction within the flow-through reactors are shown. Rates were measured with both limiting and nonlimiting substrate and are shown separately.

Although we could not measure specific rates of sulfate reduction in our experiments, pure cultures of sulfate-reducing bacteria metabolizing between −1.7° and 80°C, including organisms with different temperature adaptations (14) (Fig. 2), show little systematic variation in specific rates of sulfate reduction with temperature. There appear to be upper limits on specific rates of sulfate reduction in nature, and these limits should constrain the extent to which high specific rates can limit isotope fractionation. Thus, although correlations between specific rate of sulfate reduction, temperature, and fractionation might hold for individual organisms (15), these correlations cannot be projected beyond the temperature range of the organism.

Figure 2

Specific rates of sulfate reduction (rate per cell) are compiled here for a wide range of sulfate-reducing bacteria with different temperature adaptations (14). Specific rates were binned into 5°C intervals. Vertical lines indicate the full range of measured rates, and the average rate is provided by the symbol. Experiments reporting specific rates were conducted under both optimal and suboptimal conditions for bacterial growth. Therefore, the range of specific rates encountered here may reasonably reflect the range of specific rates that might be encountered in nature. Data come from (8, 30, 31). The specific rates reported at 80°C were determined in chemostat experiments on the Archaeal sulfate reducer Archaeoglobus fulgidus as part of this study. Shown for comparison is the increase in specific rates predicted if organisms responded across different temperature adaptations with a Q 10 = 3, a value typical for sulfate-reducing bacteria (14).

High rates of sedimentary sulfate reduction (rate per volume of sediment) are a key feature of the warm, sulfate-rich, early Archean ocean model, resulting in complete sulfate depletion near the sediment-water interface and producing a sediment closed to sulfate exchange (3, 9). A closed system could explain the generally small 2 to 3‰ differences between the average isotopic composition of sulfides, and of contemporaneous seawater sulfate, in early Archean sediments (16). We used a diffusion-reaction diagenetic model to explore the relation between sulfate reduction rates and sulfate depletion in an hypothetical early Archean sediment. We assumed a flux of metabolizable organic carbon (0.5 mmol cm−2 year−1) and sediment deposition rate (0.1 cm year−1) comparable to active modern shelf sediments (17) and a sulfate concentration of 28 mM, the same as today. The model (18) was started with sulfate reduction rates comparable to active modern shelf sediments and produced a sulfate profile also similar to modern shelf sediments (Fig. 3A). With increasing rates of sulfate reduction, more rapid depletion of organic matter occurs near the sediment surface and, consequently, leads to more vigorous near-surface sulfate consumption (Fig. 3, B and C). The result is less active sulfate reduction and minimal sulfate depletion deeper in the sediment.

Figure 3

Modeled depth distributions of sulfate reduction rate and sulfate concentration are shown for a hypothetical early Archean sediment. In the first three cases (A toC), the same flux of metabolizable carbon (0.5 mmol cm−2 year−1) to the sediment surface is used, comparable to active modern shelf sediments (see text). This flux of organic carbon is equivalent to a reactive carbon concentration of 10 weight % (wt %) (about 25% organic matter) at the sediment deposition rate (0.1 cm year−1) used in our calculations. Rates of sulfate reduction are increased (A to C) in one order of magnitude steps by increasing the reactivity of the organic carbon decomposing [the value of the rate constant k i(18)]. For (A), k 1 = 0.1 year−1 and k 2 = 0.001 year−1; for (B), k 1 = 1.0 year−1 and k 2 = 0.01 year−1; and for (C), k 1 = 10 year−1 and k 2 = 0.1 year−1. In the final case (D), the flux of metabolizable carbon (and concentration of reactive carbon; 20 wt %) is doubled from the other three examples, whereas the same high rate constants used in case (C) are retained. The insets provide an expanded view of the upper portions of the sediment column.

The organic carbon flux was doubled, retaining the high intrinsic organic matter reactivity in Fig. 3C, and still, even with these higher sulfate reduction rates (Fig. 3D) (19), only limited sulfate depletion was observed. The highest rates of sulfate reduction explored here (30 mol liter−1 year−1; Fig. 3D) are higher than any modern measurements and are comparable to proposed early Archean rates (10 to 100 mol liter−1year−1) (3, 9). We show here, however, that such high rates are associated with rapid organic matter consumption and, therefore, attenuate quickly with sediment depth. Importantly, and counter to predictions, increasing rates of sulfate reduction do not result in closed-system behavior.

Overall, rapid rates of sulfate reduction, with abundant sulfate and at elevated temperatures up to 85°C, should produce sedimentary sulfides depleted in 34S by about 13 to 28‰ compared with seawater sulfate. We have previously shown that modern microbial mats, supporting very high rates of sulfate reduction (up to 15 mol liter−1 year−1), produce sulfide depleted in34S by 20 to 40‰ at temperatures from 10° to 30°C (20). These high fractionations are preserved as pyrite in the mats (21). Therefore, at high and low temperatures, high fractionations are expected during sulfate reduction with abundant sulfate. The minimally fractionated early Archean sedimentary sulfides are most consistent with either sulfate reduction at low sulfate concentrations of <1 mM (22), where isotope fractionation during sulfate reduction is greatly reduced (23), or a nonbiogenic source of sulfide, if sulfate-reducing bacteria had not yet evolved. As sulfate-reducing bacteria likely evolved before cyanobacteria (24), a low oxygen atmosphere is consistent with both of these scenarios.

By 2.75 Ga, sedimentary sulfides with δ34S values as low as −18‰ are found (25, 26). Although some of these sulfides are volcanogenic (25), in other instances, a bacteriagenic source is possible (26). By 2.75 Ga, therefore, sulfate reduction may have been well established, but only locally expressed [probably in restricted sulfate-rich environments (25)], as evidenced by the general lack of 34S-depleted sulfides in organic matter–containing shales of this age (25). By 2.2 to 2.3 Ga, 34S-depleted sulfides of certain biological origin and reflecting generally abundant seawater sulfate concentrations of 1 mM or greater become a continuous feature of the geologic record (7, 27). Independent lines of geochemical evidence point to the first accumulation of oxygen into the atmosphere around this time (2).

Organic biomarker (28) and organic carbon stable isotope evidence (7) support the evolution of oxygenic photosynthesis by 2.7 Ga. There existed on Earth, therefore, a protracted period of at least 400 to 500 million years, from >2.7 Ga to 2.2 to 2.3 Ga, where biological oxygen production resulted in little net oxidation of Earth's surface. It appears that the evolution of metabolic innovations such as oxygenic photosynthesis and sulfate reduction is separated in time from their geochemical expression. This separation in time complicates our attempts to pace the timing of metabolic evolution and underscores the importance of elucidating the poorly understood biogeochemical mechanisms regulating Earth's surface chemistry.

  • * To whom correspondence should be addressed. E-mail: dec{at}


View Abstract

Navigate This Article