Late Archean Biospheric Oxygenation and Atmospheric Evolution

See allHide authors and affiliations

Science  28 Sep 2007:
Vol. 317, Issue 5846, pp. 1900-1903
DOI: 10.1126/science.1138700


High-resolution geochemical analyses of organic-rich shale and carbonate through the 2500 million-year-old Mount McRae Shale in the Hamersley Basin of northwestern Australia record changes in both the oxidation state of the surface ocean and the atmospheric composition. The Mount McRae record of sulfur isotopes captures the widespread and possibly permanent activation of the oxidative sulfur cycle for perhaps the first time in Earth's history. The correlation of the time-series sulfur isotope signals in northwestern Australia with equivalent strata from South Africa suggests that changes in the exogenic sulfur cycle recorded in marine sediments were global in scope and were linked to atmospheric evolution. The data suggest that oxygenation of the surface ocean preceded pervasive and persistent atmospheric oxygenation by 50 million years or more.

The history of Earth-surface oxygenation is written in the geological record of redox-sensitive elements preserved in ancient sediments. The discovery of large non–mass-dependent (NMD) S isotope anomalies in Archean and the earliest Paleoproterozoic sediments are believed to record changes in atmospheric O2 levels, as these result from photochemical reactions in a low O2 atmosphere (14). To document temporal changes in the magnitude of these isotope excursions, we focused on high stratigraphic resolution analyses of organic-rich shale and carbonate from a recently drilled scientific core (5) through the ∼2500 million-year-old Mount McRae Shale of northwestern Australia. Previous studies of the Mount McRae Shale identified abundant 2-α methyl hopanoids (6), produced by cyanobacteria that most likely generated O2, as well as eukaryotic sterols (7), which are biomolecules that require O2 for their synthesis (8). We investigated the time-series history of elemental and isotope variations through the succession and interpreted the upper half of the formation as capturing the oxygenation of the terminal Archean surface ocean and biosphere, a result further supported by a companion trace-metals study (9).

In the present study, a modified online combustion method (5, 10) for rapid analysis of whole-rock S and a high-precision fluorination technique for analysis of chemically extracted sulfide S were applied to samples from the core. We used unprecedented, high-resolution records (δ34S, Δ33S, and Δ36S) in concert with stratigraphic variations in elemental abundances [weight percent (wt %) C and S] and 13C compositions of carbonate and organic matter to address the cause(s) of fluctuation in NMD effects preserved in these ancient sediments. These results were compared to a previous study of the Mount McRae Shale (11) and to new S isotope data from the stratigraphically equivalent Gamohaan and Kuruman Iron formations in South Africa (1214) to evaluate the spatial extent of the interpreted events.

The Mount McRae Shale core intersects laminated and well-preserved sediments that accumulated in a marine environment below the wave base. A regional sequence analysis (15) indicates the presence of two depositional cycles; each sequence starts in carbonate or siliciclastic turbidite or breccia and deepens upwards to either pelagic shale or banded iron-formation (Fig. 1). The succession has experienced only mild regional metamorphism (prehnite-pumpellyite facies to <300°C) and minimal deformation (gentle folding to dips <5°) (16). Radiometric age constraints place the Mount McRae Shale very near the Archean/Proterozoic boundary (∼2500 million years old) (9) and just before the disappearance of large NMD effects that are inferred to mark the rise in atmospheric O2 (13, 17).

Fig. 1.

Lithologic and time-series elemental (C and S) and isotopic (δ13C, δ18O, δ34S, and Δ33S) trends in the ∼2500 million-year-old Mount McRae Shale. Sequence subdivisions are based on (15). Trends in Δ33S in the lower Mount McRae Shale are correlated with equivalents in a separate core drilled some 300 km away from the core in this study (11). VPDB, Vienna Pee Dee belemnite; TOC, total organic C; VCDT, Vienna Canyon Diablo Triolite.

Geochemical data from the Mount McRae Shale (Fig. 1) suggest a tight coupling between environmental and biological signals, with a substantial transition recorded at ∼153 m in the core. Acid leaching of extractable iron from Mount McRae samples indicates that siderite dominates in the lower half of the formation, which is consistent with the absence of O2 in deeper depositional environments. On the other hand, calcite is a primary carbonate phase in the upper Mount McRae Shale, indicating the general absence of soluble iron in the shallow water column at this time. Carbonate and total organic C δ13C values increase progressively up the core. Total organic C and total S values are high throughout the Mount McRae Shale but are notably enriched above the mineralogic transition in the interval between 135 and 153 m, where up to 16 wt % C and S are observed. In the interval between 130 and 135 m, visual evidence of pyrite nodules, laminations, and graded beds suggests some degree of sulfide remobilization, which may help to explain the sharp drop in total S abundance in the homogeneous-shale host rock.

The high-resolution S isotope record reveals considerable stratigraphic variation in δ34S and Δ33S (18), including substantial bed-to-bed oscillations. In evaluating the time-series S isotope data, the Mount McRae Shale can be divided (19) into a lower unit (>153 m), where δ34S and Δ33S show positive correlation, and an upper unit (<153 m), where δ34S values become increasingly negative. Of particular interest is the interval above 130 m, where positive Δ33S values are coupled with negative δ34Svalues; this S isotope relation may record an important environmental and biological event near the Archean/Proterozoic boundary.

The lower half of the Mount McRae Shale from the Archean Biosphere Drilling Project (ABDP)–9 core is interpreted to have accumulated in a deep, anoxic environment insofar as sediments are dominated by sideritic shale and banded iron-formation (13, 14). The positive correlation between δ34S and Δ33S in the lower Mount McRae Shale has been interpreted as either a primary atmospheric array (11) or mixing between atmospherically derived NMD S with mass-dependent terrestrial inputs. This mixing may well explain the long-term and bed-to-bed variability in S isotope compositions (Figs. 1 and 2B). It is difficult to independently assess the quantitative contribution of terrestrial inputs [with δ34S and Δ33S ∼ 0 per mil(‰)] relative to NMD inputs. However, nonzero values of Δ33S (either positive or negative) must ultimately be linked to fluxes of sulfate and elemental S from the atmosphere. We interpret the dominance of positive Δ33S through this stratigraphic interval as indicating the preferential incorporation of reduced NMD S (atmospheric elemental S) into marine sediments, probably facilitated by microbial elemental S reduction. This microbial process is capable of transferring the Δ33S to pyrite, while imparting little to no additional isotopic fractionation in δ34S. Ono et al. (11) previously explained variations in NMD signatures within the Mount McRae strata by physical and biological mixing of these atmospheric sources. The isotopic similarity (in δ34S and Δ33S) between their core and ours, presently 300 km apart, points to a basin-scale phenomenon linked through atmospheric inputs (20). However, rapid bed-to-bed (or even within bed) variability and small differences in the magnitude of the positive Δ33S excursion probably reflect local controls related to variable mixing of S from distinct surface reservoirs, including the deep and shallow ocean as well as terrestrial environments.

Fig. 2.

Triple isotope plots [δ34S versus Δ33Sin(A) and (B) and Δ33S versus Δ36S in(C) and (D)] for the Mount McRae Shale in Western Australia (W. A.) and the equivalent Gamohaan and Kuruman formations in South Africa (S. A.). Data from both cores are divided into upper [(A) and (C)] and lower [(B) and (D)] intervals. All data outside the target stratigraphic interval is shown in light gray for comparison. The anomalous S isotope compositions recorded in (A) are interpreted as reflecting oxidizing conditions. Mass-dependent fractionations of 33S and 36S resulted in an array with aslope of ∼6.85 (30, 32) [labeled as MD in (C) and (D)], whereas data from this study fit the general Archean slope of ∼–1 (1). The Δ33S versus Δ36S relation between the correlated upper and lower intervals is statistically different, pointing to the evolution of atmospheric composition in the late Archean Eon. I. F., iron formation.

As noted above, the upper half of the Mount McRae Shale, which is dominated by turbidites of carbonate and shale that accumulated below the storm wave base, is characterized by sulfides with negative δ34S values coupled with positive Δ33S values (Figs. 1 and 2A). The δ34S and Δ33S values of these sulfides imply microbial sulfate reduction with larger isotopic fractionations, which may reflect sulfate reduction in the water column (21), possibly coupled with rising sulfate concentrations (22). This interpretation is consistent with the high organic C contents in sediments above 153 m in the core, which are plausibly linked to high rates of primary productivity that released oxidants into the shallow marine environment. On the other hand, the positive Δ33S values reflect incorporation of reduced photolytic S. To account for these two features, we propose that the S isotope signatures in the upper Mount McRae Shale reflect the establishment of a widespread and possibly permanent oxidative S cycle, perhaps for the first time in Earth's history, in a water column that was stratified with respect to oxygen (23).

In the late Archean oceans, O2 would accumulate in highly productive regions along continental margins and perhaps to a lesser degree in distal settings, where nutrient levels were high enough to stimulate oxygenic photosynthesis. Possible explanations to account for the isotopic observations above 153 m include elemental S reducers capable of producing large 34S depletions (an unlikely scenario given the small redox change associated with this metabolic pathway) or the activation of microbial disproportionation reactions (24). Because the former are currently unknown and the latter are not clearly evident until the mid-Proterozoic (25), we suggest an alternative solution related to increases in O2 initiated during the productivity event recorded in the core above 153 m. Inorganic S oxidation generally requires high levels of dissolved O2, whereas microbial S oxidation, which is thought to be ancient in origin, would proceed at lower (or absent) O2 concentrations but still would require an electron acceptor to drive phototrophic oxidation. In either case, the magnitude of isotopic fractionation associated with oxidation is small (26) and unlikely to account for the negative δ34S values of sulfides in the upper Mount McRae sediments. Thus, we propose that the sulfate formed through oxidation (with positive δ34S and Δ33S) was re-reduced by microbial sulfate reduction to form sulfides depleted in δ34S but retaining positive NMD Δ33S values (Figs. 1 and 2A).

The organic C and S spike between 153 and 135 m corresponds to an interval where both δ34S and Δ33S values are typically negative (19). Although broadly binned with the upper Mount McRae Shale interval, these sediments provide important environmental constraints on a source of S (with a negative Δ33S composition) and the mechanism for its sink into sediments. Low Archean atmospheric O2 levels would generally limit oxidative weathering, the principal source of sulfate to the modern oceans. With rising atmospheric O2 levels, however, some metals and associated S from terrestrial sources may have been released to the shallow marine environment (9), but contributions from juvenile S(Δ33S = 0) or preexisting sedimentary sources (Δ33S > 0) cannot account for the negative Δ33S value of the S from this interval. Thus, a major source of sulfate to the Archean ocean at this time would have been atmospheric in origin and would have carried a negative Δ33S signature (1, 4). In keeping with previous arguments for Archean seawater sulfate (27), we interpret deep- and open-ocean seawater sulfate as having a negative Δ33S composition and acting as the major source of S in the pyritic interval above 153 m. Sulfate in the anoxic deep ocean was nonetheless likely to have been low (potentially < 200 μM) (22) and possibly even lower on the continental shelves. However, we suggest that enhanced microbial sulfate reduction, stimulated by high rates of organic C burial in the presence of abundant reactive iron, would serve as an effective long-term sulfate sink and conceivably result in the concentration of open-ocean S (with negative Δ33S) into the sediments (28).

Further insight into the Archean S cycle, specifically atmospheric evolution, is gained through the evaluation of the Δ36S/Δ33S relation. Whereas mass-dependent processes fractionate 33S and 36S in systematic ways(Δ36S/Δ33S ∼–7) (29, 30), NMD photochemical experiments (2, 31) suggest a wavelength dependence to the Δ36S/Δ33S relation, and with a few exceptions, observations from the Archean record generally follow Δ36S/Δ33S ∼–1. This value is broadly consistent with measurements of sulfides from a wide range of Archean sediments (1, 32). Additionally, the Δ36S/Δ33S relation may characterize NMD contributions to surface environments even when the absolute magnitude of Δ33S is small (33, 34). The high-resolution 36S analysis of the Mount McRae Shale reveals measurable differences for Δ36S/Δ33S within the succession (Fig. 2, C and D), further supporting the stratigraphic distinctions outlined above. The resolvable difference between the Δ36S/Δ33S relation in the upper and lower Mount McRae Shale indicates a change in atmospheric composition, because according to current knowledge, such shifts can only be caused by changing photochemical reactions involving S-bearing gases (35).

If the change in atmospheric composition suggested above is real, we expect the signal to be widespread in nature. To test this prediction, we have undertaken S isotope analyses of samples from the broadly equivalent Transvaal Basin in South Africa. The studied South African core intersects the Gamohaan and Kuruman Iron Formations (5, 12, 13, 36), which record similar lithologic transitions to those observed in northwestern Australia. Although it is possible that these two successions (now over 8000 km apart) accumulated along the margins of a contiguous ocean basin, palinspastic reconstruction (37) of the two subbasins on the basis of existing outcrop area suggests that the core locations were at least 1000 km apart when the sediments accumulated.

The similarity in S isotope records between the South African and Australian sediments is pronounced (Fig. 2). The correlation between these widely separated basins strongly supports the spatially pervasive character of Δ33S (and Δ36S) production, implying a degree of lateral atmospheric homogeneity. The Δ36S/Δ33S of the lower portion of the South African core matches that of the lower Mount McRae Shale, whereas the Δ36S/Δ33S from the upper portion is quite similar to that of the upper Mount McRae sediments. The consistency of the δ34S versus Δ33S and Δ33S versus Δ36S relations between the Australian and South African cores indicates that the S isotope variations reflect widespread and probably global variations in the Archean S cycle. The origin of the profound δ34S and Δ33S anomaly at ∼170 m in the Mount McRae core and its equivalent in South Africa is unknown, but it is probably related to a pulsed flux of atmospheric inputs to surface environments that was captured over long distances in similar depositional settings. Whereas the transition captured at ∼153 m might reflect changes in the atmospheric O2 budget, it is also possible that changes in the abundance of other atmospheric species (CO2 and CH4) may be responsible for differences in the Δ36S/Δ33S relations. However, the independent trace-metal evidence (9) and lower stability of methane under oxidizing conditions point to an increasingly important role for O2 in surface environments.

We interpret our data from Western Australia and South Africa to suggest a progressive oxygenation of the Archean biosphere. This conclusion is in accord with the trace-metal data (9), which similarly suggest the onset of oxidative processes. Combined, these time-series records of mineralogic, elemental, and S isotopic change provide clues to coupled changes in the redox state of the shallow ocean (largely before the atmosphere became oxygenated) in relation to biological innovation before the Archean/Proterozoic boundary, including the oldest evidence for an active and globally distributed oxidative S cycle.

Supporting Online Material

SOM Text

Figs. S1 to S4

Tables S1 and S2


References and Notes

Stay Connected to Science

Navigate This Article