Molybdenum Isotope Evidence for Widespread Anoxia in Mid-Proterozoic Oceans

See allHide authors and affiliations

Science  02 Apr 2004:
Vol. 304, Issue 5667, pp. 87-90
DOI: 10.1126/science.1091785


How much dissolved oxygen was present in the mid-Proterozoic oceans between 1.8 and 1.0 billion years ago is debated vigorously. One model argues for oxygenation of the oceans soon after the initial rise of atmospheric oxygen ∼2.3 billion years ago. Recent evidence for H2S in some mid-Proterozoic marine basins suggests, however, that the deep ocean remained anoxic until much later. New molybdenum isotope data from modern and ancient sediments indicate expanded anoxia during the mid-Proterozoic compared to the present-day ocean. Consequently, oxygenation of the deep oceans may have lagged that of the atmosphere by over a billion years.

Understanding how and when the Earth's early atmosphere and oceans became rich in oxygen is essential to our understanding of early life on Earth and to the search for life beyond the solar system (13). A large body of geochemical and sedimentological evidence points to a substantial rise in atmospheric oxygen ∼2.3 billion years ago (Ga) (46). The history of ocean oxygenation is less clear. The occurrence of banded iron formations (BIFs) until ∼1.8 Ga suggests deep oceans with high concentrations of ferrous iron until this time and therefore, because of the low solubilities of ferric oxides, little dissolved O2 (4). The subsequent disappearance of BIFs may indicate the beginning of persistent deep-sea oxygenation, which continued with only brief interruptions to the present day (4). This model of ocean redox evolution is challenged by the recent proposal that euxinic (anoxic and sulfidic) conditions were common in the deep oceans from ∼1.8 Ga to at least 1.0 Ga, after which atmospheric and deep-sea oxygen levels approached modern values (79). Euxinic conditions could also account for the disappearance of BIFs because of the low solubilities of ferrous sulfides (8). This debate has important implications for evolution linked to the availability of redox-sensitive, bioessential trace metals (2) and for climate, because the flux of biogenic methane from oceans to atmosphere is sensitive to ocean oxygenation (10, 11).

Sulfur isotopes and sedimentary iron chemistry indicate that some mid-Proterozoic sedimentary basins were oxic at the surface and euxinic at depth (9, 12, 13). Sulfur isotope data further suggest that sulfate concentrations in mid-Proterozoic oceans were <1/10th the modern value (9, 1214). However, low sulfate does not uniquely indicate anoxic or sulfidic oceans, and redox-stratified basins exist even today although >99% of the oceans are oxygenated to the seafloor. Hence, such data cannot directly address the extent of ocean anoxia.

Molybdenum stable isotopes may clarify this issue. Molybdenum enters the oceans largely via rivers. In oxygenated waters, Mo speciation is dominated by the molybdate anion MoO42–, which is so unreactive that Mo is the most abundant transition metal in modern oceans. However, under euxinic conditions, such as in the modern Black Sea where H2S is >100 μM in bottom waters, Mo is readily removed from solution (1518). As a result, ocean sediments accumulating today beneath euxinic waters account for 10 to 50% of the annual removal flux of Mo from the oceans even though such waters cover an extremely small percentage of the modern seafloor (<0.5%) (15, 18, 19). The remainder of marine Mo is primarily removed in association with sedimentary Mn oxides in oxygenated environments (18, 19). It is not known if sediments accumulating under bottom waters with <5 μM O2 but lacking sulfide (“suboxic” sediments) constitute a net source or sink of Mo from the oceans (18, 20), but they probably account for <10% of Mo cycling today (18). The ocean Mo budget is therefore highly sensitive to the extent of euxinic bottom waters (15).

The marine Mo isotope budget also appears to be redox-sensitive. In previous studies (table S1; Fig. 1A), values for δ97/95Mo in granite, basalt, and continental molybdenite ores were found to be – 0.01 ± 0.27‰ (2124). Molybdenum in Holocene euxinic Black Sea sediments is significantly heavier: Values for δ97/95Mo are 1.52 ± 0.23‰ in the surface-most microlaminated sediments (Unit I) and 1.14 ± 0.08‰ in the underlying sapropel (Unit II) (22). In contrast, δ97/95Mo in Mn oxide sediments, represented by Atlantic and Pacific ferromanganese nodules (22) and ferromanganese crusts from the Atlantic, Pacific, and Indian oceans (23), is lighter: δ97/95Mo = –0.47 ± 0.18‰.

Fig. 1.

(A) Molybdenum isotope compositions of sources, sinks, and reservoirs important to the Mo ocean budget, including new measurements and previous data (22, 23). δ97/95Mo ∼ δ98/95Mo × 2/3. Open circles represent the mean of replicate measurements for individual samples. Large vertical hash marks represent the mean of all samples for each sample group. Error bars on these means are ±2σ. Uncertainties of measurements reported here are given in table S2. The mass balance model source is defined by Eq. 1 using δ97/95Moeux = -0.47‰, δ97/95Moox = 1.28‰, and fox = 50 to 90% (18, 19). Experimentally determined fractionation during Mo adsorption to Mn oxides is also indicated (29), relative to mean δ97/95Mosw. The hatched region represents the ±0.5‰ uncertainty associated with these experiments. (B) Mo isotope compositions of mid-Proterozoic black shales. Open diamond is the mean for all Proterozoic samples with ±2σ uncertainty, as indicated. Reproducibility of individual samples ranged from ±0.05 to 0.24‰ (2σ), as reported in table S2.

New δ97/95Mo data are presented here for another sample from Black Sea Unit I and a suite of samples from the Cariaco Basin, Venezuela, which is second only to the Black Sea in its extent of modern anoxic-sulfidic water-column conditions (table S2) (25). Our δ97/95Mo data from Cariaco Basin sediments are similar to earlier measurements of euxinic sediments (tables S1 and S2), and new data for Black Sea Unit I are identical within error to previous measurements (tables S1 and S2). Collectively, δ97/95Mo averages 1.28 ± 0.42‰ in Holocene euxinic sediments, nearly 2‰ heavier than in Mn oxide sediments. The offset between the average values of these redox end members encompasses almost the full range of δ97/95Mo variation measured to date.

Molybdenum in both euxinic and Mn oxide sediments is largely hydrogenous. The isotopic offset between these sediment types suggests that Mo entering the oceans from continental sources is isotopically fractionated during removal into one or both of these major sedimentary sinks (22, 23). As a result, the two sediment types have complementary isotopic compositions—one heavy and the other light, relative to the input (22, 23, 26, 27) (Fig. 2A).

Fig. 2.

Schematic depiction of Mo ocean isotope budget (A) today and (B) in the mid-Proterozoic, including average δ97/95Mo values for reservoirs, as indicated. Widths of arrows depict proportions of Mo removed to each sediment type. Offset between oceans and Mn oxides (Δsw-ox ≈ 2) is consistent with observations and laboratory experiments. Because of the possible importance of suboxic sediments to the Mo isotope budget in the mid-Proterozoic oceans, the proportion of Mo removed to Mn oxides may have been smaller than indicated in (B).

If the isotopic compositions presented here are representative of the major Mo sources and sinks (28), then to a first approximation the global ocean Mo isotope budget can be expressed in terms of an input of ∼0‰ and outputs to oxic and euxinic sinks with isotopic compositions of ∼ –0.5‰ and 1.3‰, respectively (Fig. 2A). To test this hypothesis, consider the following simple steady-state mass balance model: Math Math(1) where the subscripts input, ox, and eux denote the riverine input and oxic and euxinic sedimentary sinks, respectively, and fox and feux represent the fraction of total Mo removed to each sediment type. Using the mean values for δ97/95Moox and δ97/95Moeux and the range of literature estimates of fox and feux in the modern oceans (15, 18, 19), we can estimate the value of δ97/95Moinput needed to balance removal to oxic and euxinic sediments. Our predicted range for modern δ97/95Moinput matches well with measurements of Mo in crustal materials, used here as a proxy for riverine input (table S1; Fig. 1A) (22, 23). This agreement implies that these data provide a first-order representation of Mo sources to, and sinks from, the oceans (22, 23).

Mechanisms of Mo isotope fractionation in the ocean are best viewed in light of the δ97/95Mo value of seawater. The Mo isotope composition of seawater (δ97/95Mosw) is uniform with location and depth, with a presentday mean value of +1.56 ± 0.13‰ (table S1; Fig. 1A) (22, 23). Such uniformity is expected from the conservative behavior and ∼800,000-year residence time of Mo in oxygenated oceans (18). The δ97/95Mo of seawater is ∼2‰ heavier than that of hydrogenous Mo associated with sedimentary Mn oxides. Laboratory experiments document a fractionation factor, α, of 1.0018 ± 0.0005 between dissolved Mo and Mo adsorbed to Mn oxides (Δdissolved-adsorbed = δ97/95Modissolved – δ97/95Moadsorbed = 1.8 ± 0.5‰ ≈ 103·ln(α) ≈ 1.8‰) (29). The similarity between experiments and observed δ97/95Mosw – δ97/95Mooxsw-ox) and the importance of Mn oxides as a Mo sink strongly suggest that preferential adsorption of light Mo isotopes to authigenic Mn oxides is the primary driver of Mo isotope fractionation in the modern ocean.

In contrast to the fractionation associated with Mn oxides, δ97/95Mosw is within 0.3‰ of δ97/95Moeux, indicating relatively little net fractionation when Mo is removed from the water column and deposited with sediments in euxinic basins. This is most likely a consequence of mass balance: Mo removal from aqueous systems when [H2S] > 100 μM is nearly quantitative, and a large fraction is removed even when [H2S] is lower (16, 17, 30). Hence, the δ97/95Mo value of sediments accumulating under euxinic conditions should be similar to δ97/95Mo of the Mo source to the immediately overlying waters, even if there are large instantaneous isotope effects during Mo removal to these sediments. Small (<±0.4‰) variations in δ97/95Mo of Holocene euxinic sediments may result from local effects (22, 23, 25). However, even relatively restricted basins like the modern Black Sea, where the mean residence time of the deep water is ∼387 years, are well mixed with the open ocean because of the long residence time of oceanic Mo (31). Consequently, the similarity of δ97/95Mosw and δ97/95Moeux today is not surprising, and removal of Mo to euxinic sediments is unlikely to drive global-scale Mo isotope fractionation. Instead, δ97/95Moeux should track the Mo isotope evolution of seawater and thus the balance between oxic and euxinic marine deposition over geologic time.

The preceding systematics provide a basis for using Mo isotopes in ancient euxinic sediments (i.e., black shales) to examine changing ocean redox conditions. If Mo isotope fractionation in the oceans results from an equilibrium isotope effect between dissolved and Mn oxide–associated Mo (22, 23, 29), then Δsw-ox = Δdissolved-adsorbed ≈ 2‰. Additionally, because of the near-quantitative removal of Mo in euxinic settings, it seems reasonable that δ97/95Moeux approximates δ97/95Mosw, at least in settings similar to the Black Sea and Cariaco Basin. Using these relationships and assuming fox + feux = 1, Eq. 1 can be modified to obtain Math(2) Hence, changes in ocean redox, which affect the relative proportions of feux and fox, should result in changes in the isotopic composition of hydrogenous Mo in euxinic sediments. Although highly idealized, this model illustrates that expanded euxinic or contracted oxic conditions relative to the modern oceans should shift δ97/95Moeux toward δ97/95Moinput.

To examine the extent of mid-Proterozoic ocean redox, we measured δ97/95Mo in black shales from the Roper Group (Velkerri Formation; ∼1.4 Ga; [Mo] = 11 to 60 ppm) and the Tawallah Group (Wollogorang Formation; ∼1.7 Ga; [Mo] = 42 to 52 ppm) in northern Australia. Both sample suites were obtained from drill cores and are exquisitely preserved. Specifically, they have experienced no oxidative weathering and little metamorphism or deformation (25, 32, 33). Therefore, it is expected that the primary δ97/95Mo of these sediments should be well preserved.

Detailed chemostratigraphic studies of sulfur isotopes and sedimentary iron chemistry have been conducted for both formations (12, 13). Based on their ratios of highly reactive Fe to total Fe and the degrees to which reactive Fe has been pyritized, our samples come from those parts of the depositional basins showing euxinic character similar to that of the modern Black Sea (12, 13, 25, 34). By analogy to modern euxinic settings, Mo sequestration into these sediments was likely quantitative, or nearly so. Sulfur stable isotopes and sedimentary features strongly suggest that during accumulation of these sediments, neither basin was isolated from the global ocean (12, 13). Hence, we can be confident that our δ97/95Mo data are representative of the overlying Proterozoic water column and that this water column was in contact with the open ocean.

We find that the Mo isotope compositions of these mid-Proterozoic black shales have significantly lighter δ97/95Mo values than modern equivalents (table S2; Fig. 1B). Specifically, the mean mid-Proterozoic value is 1.1‰ lighter than that of modern Black Sea sediments (Unit I), which are the most obvious modern analog, and are 0.8‰ lighter than the average of all Holocene euxinic sediments measured to date (Fig. 1A).

It is likely that δ97/95Mosw was well mixed in the mid-Proterozoic ocean and that these data are globally representative. In support of this assumption, the concentration of Mo in mid-Proterozoic euxinic sediments, although low, is within the range of Phanerozoic equivalents (which vary over a factor of ∼10). This comparison suggests a global ocean in which the Mo concentration was probably not more than an order of magnitude lower than it is today (35). Unless ocean mixing times were substantially slower than they are today, or Mo concentrations decreased by more than a factor of ∼50, δ97/95Mosw was likely well mixed.

It follows that the mid-Proterozoic global ocean Mo isotope budget was very different compared to that of the present day. Equation 2 indicates that fox accounted for ∼30% of Mo removal during the mid-Proterozoic, compared to ∼75% today (36, 37) (Fig. 2). This value probably underestimates the extent to which mid-Proterozoic ocean redox differed from that of the modern oceans because Eq. 2 ignores suboxic sedimentary environments. Considerable variation in Mo isotopes is observed in such sediments today (0.35‰ < δ97/95Mo < 1.08‰) (23) (table S1). In a pore-water profile beneath suboxic waters, δ97/95Mo in the interstitial waters is as much as 0.8‰ heavier than in seawater (20). If these data are broadly representative, this relationship indicates preferential removal of light Mo isotopes to sediments in such settings. The effect of such sediments on the Mo isotope budget is semiquantitatively similar to that of Mn oxide sediments (38). Although suboxic settings are of secondary importance in the modern Mo budget, this may not have been the case in the mid-Proterozoic. It can be reasonably hypothesized that expanded suboxic conditions would have gone hand-inhand with contraction of oxic seafloor and expanded euxinic deposition. Because oxic and suboxic removal fractionate Mo isotopes similarly, mass balance requires that an increase in the suboxic sink could only occur at the expense of the oxic sink. Thus, it is probable that fox was <30%.

It is difficult to account for the Proterozoic data without invoking ocean redox arguments. If the balance between oxic and euxinic sinks were unchanged, then Eq. 2 requires that δ97/95Moinput was ∼ –0.8‰ during the mid-Proterozoic. Such a value for δ97/95Moinput is substantially lighter than those for any known crustal materials (22, 23). Regional or local influences on δ97/95Mo in the studied sediments, as might result from localized hydrothermal or riverine influences, are also unlikely but less easily dismissed (25).

Molybdenum isotope data therefore point to redox conditions in mid-Proterozoic oceans that differed from those observed today. If we assume a linear scaling of fox and feux to the areal extent of oxic and euxinic seafloor, then the ratio fox/feux can be used to assess changes in the relative extents of such environments (39). The Mo isotope data indicate that fox/feux changed from <0.4 in the mid-Proterozoic to ∼3 today. Hence, the area of oxic sedimentation in the mid-Proterozoic oceans was nearly a factor of 10 smaller than the modern value, or the area of euxinic sedimentation approached a factor of 10 larger than the modern value, or, most likely, an intermediate situation occurred involving both substantial contraction of the extent of oxic deposition and expansion of the euxinic deep ocean (40).

Such conditions may have persisted until after 1 Ga, when S isotopes suggest an increase in ocean-atmosphere oxygenation (7, 8). Subsequently, ocean anoxia was generally confined to small areas of the seafloor. Although episodic anoxic events occurred during the Phanerozoic, they probably did not approach the duration or areal extent of mid-Proterozoic oxygen deficiency. For example, intervals of expanded euxinic deposition were common during the Devonian, as documented by extensive deposition of black shales on multiple continents (41). However, anoxic conditions were likely confined to shallow epicontinental seas, whereas the vast majority of the seafloor remained oxic (41). Consistent with this notion, δ97/95Mo in Devonian black shale is not as strongly shifted from modern seawater as are mid-Proterozoic shales (22) (table S1).

Further research is needed to relate δ97/95Mo variations quantitatively to ocean redox conditions. However, based on current understanding of Mo isotope systematics, seafloor anoxia during the mid-Proterozoic was expanded compared to today during at least two intervals separated by ∼300 million years. The δ97/95Mo data are consistent with globally extensive seafloor anoxia persisting after oxygenation of the atmosphere ∼2.3 Ga. The identification of mid-Proterozoic Mn oxides with a δ97/95Mo value ∼2‰ lighter than in coeval euxinic sediments would strengthen these arguments. Nevertheless, the full impact of nearly a billion years of ocean anoxia on Proterozoic ecosystems and biogeochemical cycles must be seriously considered.

Supporting Online Material

Materials and Methods

SOM Text

Fig. S1

Tables S1 and S2

References and Notes

References and Notes

  1. Data are reported using the δ notation, relative to a Mo standard (Johnson Matthey Chemical, Specpure lot #702499I) where Math
View Abstract

Stay Connected to Science

Navigate This Article