A Whiff of Oxygen Before the Great Oxidation Event?

See allHide authors and affiliations

Science  28 Sep 2007:
Vol. 317, Issue 5846, pp. 1903-1906
DOI: 10.1126/science.1140325


High-resolution chemostratigraphy reveals an episode of enrichment of the redox-sensitive transition metals molybdenum and rhenium in the late Archean Mount McRae Shale in Western Australia. Correlations with organic carbon indicate that these metals were derived from contemporaneous seawater. Rhenium/osmium geochronology demonstrates that the enrichment is a primary sedimentary feature dating to 2501 ± 8 million years ago (Ma). Molybdenum and rhenium were probably supplied to Archean oceans by oxidative weathering of crustal sulfide minerals. These findings point to the presence of small amounts of O2 in the environment more than 50 million years before the start of the Great Oxidation Event.

Many lines of evidence point to a rapid rise in the partial pressure of atmospheric O2 (PO2) from <10–5 times the present atmospheric level (PAL) between 2.45 and 2.22 billion years ago (Ga) (1, 2), a transition often referred to as the Great Oxidation Event (GOE). The GOE could have been an immediate consequence of the evolution of oxygenic photosynthesis (3). Alternatively, O2 biogenesis may be ancient (4). If so, the GOE was a consequence of an abiotic shift in the balance of oxidants and reductants at Earth's surface (58). This debate can be addressed by looking for evidence of localized or short-lived concentrations of O2 before 2.45 Ga.

The abundances of some transition elements in sedimentary rocks are sensitive to the availability of O2 (9). In particular, in the modern oxygenated environment, molybdenum (Mo) exists in rivers and oceans primarily as the unreactive molybdate ion (MoO42–). Oxidative weathering of Mo-bearing sulfide minerals in crustal rocks leads to the accumulation of Mo in the oceans, where it is the most abundant transition element (at a concentration of ∼105 nM) (10, 11). The abundance of Mo in the oceans is reflected in pyritic marine sediments deposited under oxygen-deficient conditions, where Mo is removed from solution in association with organic carbon (12, 13), probably after reacting with H2Sto form oxythiomolybdates (MoO4-xSx2–) (14). In such sediments deposited today and through much of the Phanerozoic, Mo contents are typically >100 ppm versus ∼1 ppm in average crust (10, 12, 13, 15, 16).

By comparison, on an anoxic Earth, Mo would be largely retained in unoxidized crustal sulfide minerals during weathering. Therefore, Mo concentrations in the oceans would be low, and organic-rich sediments would show little authigenic Mo enrichment as compared to modern equivalents. Similar logic applies to sulfur (S). In fact, studies of Mo and S concentrations and stable isotopes in black shales reveal systematic shifts in ocean budgets from the Archean through the Phanerozoic that are broadly consistent with the GOE and with another rise in PO2 later in the Proterozoic (2, 1719) (table S1). Rhenium (Re) and uranium (U) are also promising indicators because their aqueous geochemistry is similar to that of Mo.

Here we report Mo, Re, U, and S measurements, as well as other geochemical data obtained at high stratigraphic resolution in the Mount McRae Shale, deposited ∼2.5 Ga in the Hamersley Basin, Western Australia (20, 21). Approximately 100 samples were analyzed from a freshly recovered continuous drill core obtained for this study (22) (Fig. 1, fig. S1, and table S2). These samples were also analyzed for S isotope variations as part of a companion study (23).

Fig. 1.

Stratigraphy and geochemistry of the Mount McRae Shale, including percent of carbonate, TOC, S, Fe, Mn, Mo, Re, and U and EFs (24) for Mo, Re, and U (23). The intervals S1 and S2 span 125.5 to 153.3 m and 173.0 to 189.7 m, respectively. For comparison, dashed lines denote mean Mo concentrations and EFs in Archean and Proterozoic pyritic black shales, as indicated in the legend at bottom (18, 22) (tables S1 and S2).

The core intersected two intervals containing pervasive pyritic carbonaceous shale, which we refer to as S1 (from 125.5 to 153.3 m) and S2 (from 173.0 to 189.65 m). Shales in both intervals contain several weight % (wt %) S and typically >3% total organic carbon (TOC), which is consistent with anoxic (and potentially sulfidic) bottom waters and the presence of H2S in pore waters during these depositional intervals.

The most prominent feature of the data is the excursion in Mo content within S1 (Fig. 1). Mo concentrations below this layer are typically <5 parts per million (ppm), which is near the crustal value and is typical of Archean carbonaceous shales. Concentrations increase gradually up the section from the base of S1 to a peak value of ∼40 ppm at 143 m and then decrease to <10 ppm by ∼125 m. These variations and the Mo peak at ∼143 m are more pronounced when plotted as aluminum (Al)–normalized enrichment factors (24). Viewed this way, Mo content increases up the section by ∼50 times before falling sharply over an interval of ∼2 m. The Mo enrichment correlates with enrichments in TOC and Re and broadly coincides with variations in carbonate and S contents. However, U contents vary little through the section.

The coherent behavior of Mo, Re, and U makes it possible to use Re/Os geochronometry to verify that metal abundances were unaffected by remobilization. Postdepositional addition or loss of Re (or Os) would result in significant isochron scatter. We find that samples taken from 128 to 149 m define an isochron with mean square weighted deviation = 1.1 (22) (Fig. 2 and table S3) and an age of 2501.1 ± 8.2 million years, which is consistent with previous ages for the Mount McRae shale (20, 21). The element enrichments are therefore primary sedimentary features (as is the lack of associated U variation) deposited at the boundary of the Archean and Proterozoic eons, at least 50 million years before the beginning of the GOE.

Fig. 2.

Re/Os isochron from the Mount McRae Shale, based on data from two subintervals within S1 (128.71 to 129.85 m and 145.22 to 148.32 m). MSWD, mean square weighted deviation. The Re/Os age, 2501.1 ± 8.2 Ma (initial 187Os/188Os = 0.04 ± 0.06), falls between prior ages of 2479 ± 3 Ma for the overlying Dales Gorge Member of the Brockman Iron Formation (20) and 2561 ± 8 Ma for the underlying Bee Gorge Member of the Wittenoom Formation (21). Details are discussed in (22).

The Mo excursion cannot be explained by variable carbonate dilution, as documented by extreme enrichment factors (Fig. 1). Instead, the correlations of Mo with TOC are strong evidence of authigenic enrichment (Fig. 3). Such trends are common in sediments from modern anoxic basins where the concentration of H2S exceeds ∼10 μM. In such sediments, Mo/TOC scales with Mo concentrations in deep waters (12). We recognize two trends within S1, corresponding to the zone of increasing Mo enrichment (∼143 to 153 m) and the overlying zone in which Mo falls but remains elevated above average crustal values (∼125 to 143 m). Mo/TOC slopes in these zones are ∼3.4 ± 0.5 (±1σ) ppm Mo/wt % TOC and ∼1.8 ± 0.2 (±1σ) ppm Mo/wt % TOC, respectively (25). By comparison, Mo/TOC is 4.5 to 25 ppm Mo/wt % TOC in pyritic sediments from modern anoxic basins (12) and averages ∼26 ppm Mo/wt % TOC in Phanerozoic pyritic black shales (18) (table S1). Hence, small but substantial concentrations of dissolved Mo were present during S1 deposition. Similar reasoning can be applied to Re (fig. S2A).

Fig. 3.

Relationship between Mo and TOC in organic carbon-rich pyritic intervals in the Mount McRae Shale. Circles are from interval S1 (125.5 to 153.3 m). The metal-enriched zone of S1 below 143 m (open circles) is differentiated from the upper zone (solid circles). Triangles are from interval S2 (173.0 to 189.7 m). For comparison, the shaded region indicates the range of Mo/TOC slopes (forced through the origin) observed in modern sulfide-rich anoxic basins.

These observations are not easily explained by hydrothermal inputs to the oceans. Enhanced hydrothermal input should result, first and foremost, in enrichments of iron (Fe) and manganese (Mn), yet the S1 unit is depleted in these elements relative to S2. In any event, high-temperature mid-ocean ridge–type systems should be sinks, not sources, for Mo and Re because of the low solubilities of Mo and Re sulfides. A small amount of Mo enters seawater today as result of low-temperature hydrothermal seafloor weathering (13), but this Mo is probably derived from modern Mo-rich seafloor sediments.

Instead, these observations can be straightforwardly interpreted as evidence of oxidative weathering during S1 deposition. We hypothesize that O2 in the shallow oceans and possibly in the atmosphere enhanced the rate of dissolution of submarine and subaerial sulfide minerals, such as molybdenite (MoS2), that are important for the budgets of Mo and Re in igneous and metamorphic crustal rocks. Mo and Re released in this way would ultimately have produced authigenic enrichments in ocean sediments.

Sulfide minerals weather rapidly in the presence of O2, so PO2 need not have been high. For example, even if PO2 is only ∼10–5 PAL, a pyrite crystal of 100 μm3 volume will dissolve completely in <∼20,000 years (26, 27). This is a short time compared to the likely duration of S1 (28). Consistent with such low PO2, Mo/TOC values in S1 do not exceed those of sediments accumulating in the modern Black Sea, which implies that the concentration of Mo in contemporaneous seawater was of similar magnitude as that in the deep waters of the Black Sea, or <1% that of fully oxygenated modern oceans.

The same process could have contributed to the excursion in S content and δ34S in S1 (23). The long-term δ34S record of sedimentary sulfides exhibits a negative shift between 2.4 and 2.3 Ga that is thought to indicate an increase in ocean sulfate concentrations. This increase is ascribed to an increased rate of oxidative weathering of pyrites in crustal rocks during and after the GOE (2). The negative shift in sedimentary δ34S beginning at ∼153 m in the Mount McRae Shale may record the effects of less extreme oxygenation at 2.5 Ga.

Our hypothesis of mild oxygenation is supported by the absence of U enrichment coincident with Mo and Re enrichments (Fig. 1) and the lack of correlation between U and TOC in S1 (fig. S2B), observations indicating that dissolved U concentrations were very low. U in the crust is primarily hosted by feldspars, zircon, apatite, and sphene, but not sulfides. Therefore the rate of release of U from rocks is only weakly affected by oxygenation, unlike that of Mo and Re; experimental studies suggest that the rate of pyrite oxidation exceeds that of feldspar minerals when PO2 > 10–6 PAL (29, 30). U may also be less mobile than Mo and Re when O2 is low (31). Hence, enhancements of Mo and Re influx without U enhancement are expected in the presence of small amounts of O2.

Our interpretation is also consistent with the extremely nonradiogenic initial 187Os/188Os in the Mount McRae Shale (Fig. 2). Such low values, also seen in shales ∼200 million years younger (32), indicate that the ocean Os budget was dominated by hydrothermal sources rather than by radiogenic Os derived from the weathering of high-Re/Os crustal rocks. As with U, oxidative weathering of sulfide minerals in igneous or metamorphic rocks might have had little effect on the balance between hydrothermal and crustal sources of Os to the oceans, because the Os content of crustal sulfide minerals, particularly molybdenite, can be low.

The low levels of O2 that can account for our data are similar to the upper limit of 10–5 PAL for typical Archean PO2 derived from the observation of nonzero Δ33S in Archean sediments (33, 34), possibly explaining the juxtaposition of Mo and Re enrichments with the small nonzero Δ33S signals seen throughout S1 (23). Alternatively, PO2 above this threshold could have been present ephemerally within geographically restricted areas such as biologically productive regions of the oceans.

In contrast to the observations in S1, Mo concentrations and enrichment factors are very low below ∼153 m, including in the organic carbon–rich and pyritic S2, where Mo is essentially invariant with TOC [Mo/TOC = 0.15 ± 0.35 (±1σ) ppm Mo/wt % TOC] (Fig. 3). Weak correlations appear between Mo and Al in S2, suggesting that at these depths the Mo budget was influenced by detrital components; such correlations are absent from S1. These observations point to much less authigenic Mo enrichment in S2 than in S1.

The difference in Mo enrichment suggests that the Mo inventory in overlying waters was much larger during S1 deposition than during S2 deposition, as would follow from an increase in environmental oxygenation up the section beginning at ∼153 m. This interpretation is complicated by the fact that the Mo concentration differences between the units were also affected by differences in local depositional conditions that increased the efficiency with which Mo was transferred from water to sediments during S1 time (35). However, a difference in the dissolved Mo inventory and in ocean oxygenation provides a compelling explanation for the sharp difference in Mo/TOC between the units (36).

Other data from the core also point to greater surface ocean oxygenation above ∼153 m, including changes in δ34S-Δ33S systematics that may record the onset of an oxidative sulfur cycle (23). A redox shift can also explain differences in Fe and Mn concentrations above and below ∼160 m (Fig. 1). Below this depth, most of the Fe is present as siderite (FeCO3), and both elements are much lower in S1 than in S2. Fe and Mn would have been easily mobilized during anoxic weathering, enriched in anoxic S-poor Archean oceans and hence available for incorporation into sediments. Oxygenation of surface environments would have reduced the availability of both elements. At the same time, Re concentrations are slightly elevated above crustal average values throughout the core, and there is a positive correlation of Re with TOC in S2 as well as S1 (fig. S2B). Re can be more mobile than Mo during oxidative sulfide weathering (37), so this persistent Re enrichment suggests that some small degree of oxidative weathering occurred throughout.

The decrease in Mo content and Mo/TOC above 143 m may record a drop in the dissolved Mo inventory after its initial rise, even though the surface environment apparently remained persistently, if mildly, oxygenated (23). Re and S also decrease. Diagenetic complications notwithstanding (23), it is tempting to speculate that these decreases mirror a drop in atmosphere or ocean redox potential (38), as a result of biological or nonbiological feedbacks (39). However, declining trace metal abundances could simply reflect the exhaustion of exposed crustal sulfide sources, or areal expansion of sulfidic basins in the oceans in response to rising sulfate reduction, drawing down seawater Mo, Re, and S inventories.

The onset of oxidative weathering at 2.5 Ga was probably widespread. A recent examination of contemporaneous sediments from the Ghaap Group in South Africa found that authigenic Mo and Re increased between 2.64 and 2.5 Ga (40), although that study did not have the stratigraphic resolution to capture the scale of variations reported here. Changes in S isotope systematics like those in the Mount McRae Shale also appear in time-correlative units from South Africa (23). Theoretical models show that a shift toward more oxidizing conditions can occur before the rise of an oxygenated atmosphere (5). Hence, the whiff of oxygen in the Mount McRae Shale may presage the global and irreversible transition to an oxygenated world.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Tables S1 to S3

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article