Edited by Mark H. Thiemens, University of California, San Diego, La Jolla, CA, and approved August 5, 2016 (received for review May 26, 2016)
Proterozoic climate dynamics, including both remarkable climate stability during the mid-Proterozoic and extreme low-latitude glaciation in the Neoproterozoic, must be understood in the framework of evolving oxidant reservoirs throughout the Precambrian. We present Earth system model simulations showing that recent constraints on atmospheric oxygen and oceanic sulfate during Proterozoic time have profound implications for marine methane cycling and the accumulation of methane in the atmosphere. Our model results challenge the paradigm of persistently elevated methane during the Precambrian, thus extending the relevance of the faint young Sun paradox throughout the Proterozoic. In light of the possibility of low methane during the mid-Proterozoic, we also suggest a conceptual model for the relationship between oxygenation, methane, and Neoproterozoic Snowball Earth events.
Pervasive anoxia in the subsurface ocean during the Proterozoic may have allowed large fluxes of biogenic CH4 to the atmosphere, enhancing the climatic significance of CH4 early in Earth’s history. Indeed, the assumption of elevated pCH4 during the Proterozoic underlies most models for both anomalous climatic stasis during the mid-Proterozoic and extreme climate perturbation during the Neoproterozoic; however, the geologic record cannot directly constrain atmospheric CH4 levels and attendant radiative forcing. Here, we revisit the role of CH4 in Earth’s climate system during Proterozoic time. We use an Earth system model to quantify CH4 fluxes from the marine biosphere and to examine the capacity of biogenic CH4 to compensate for the faint young Sun during the “boring billion” years before the emergence of metazoan life. Our calculations demonstrate that anaerobic oxidation of CH4 coupled to SO42− reduction is a highly effective obstacle to CH4 accumulation in the atmosphere, possibly limiting atmospheric pCH4 to less than 10 ppm by volume for the second half of Earth history regardless of atmospheric pO2. If recent pO2 constraints from Cr isotopes are correct, we predict that reduced UV shielding by O3 should further limit pCH4 to very low levels similar to those seen today. Thus, our model results likely limit the potential climate warming by CH4 for the majority of Earth history—possibly reviving the faint young Sun paradox during Proterozoic time and challenging existing models for the initiation of low-latitude glaciation that depend on the oxidative collapse of a steady-state CH4 greenhouse.
A dearth of glacial deposits during much of the Proterozoic is traditionally interpreted as evidence that Earth’s surface was as warm as, or warmer than, modern Earth throughout the mid-Proterozoic, 1.8 to 0.8 billion years ago (Ga) (1). This relatively warm climate state has conventionally been considered to require enhanced greenhouse warming to compensate for reduced solar luminosity early in Earth’s history (2, 3). Although the composition of the ancient greenhouse is still debated, the temporal proximity of extensive, low-latitude glaciation and oxygenation in the Paleoproterozoic (2.5 Ga to 1.6 Ga) supports speculation that a reduced greenhouse gas, such as biogenic CH4, played an important role in the Precambrian climate system (1, 4). This relationship has been envisaged in two ways for the Paleoproterozoic Snowball glaciations: (i) the Paleoproterozoic Great Oxidation Event (GOE) rapidly eroded a potent CH4 greenhouse, and oxidative collapse of the CH4 reservoir allowed the Earth system to plunge into snowball glaciation (5); or (ii) declining biogenic CH4 fluxes allowed the collapse of a CH4 greenhouse, extreme climate perturbation, and the initial rise of atmospheric O2 by reducing the atmospheric sink for O2 (6, 7). Several authors have similarly suggested that the oxidative collapse of a CH4 greenhouse could have also triggered climate destabilization during the Neoproterozoic (1.0 Ga to 0.54 Ga) if biogenic CH4 fluxes were elevated throughout the Proterozoic Eon, as assumed before the GOE (8⇓⇓–11).
Despite its conceptual convenience, the existence and/or effectiveness of a CH4 greenhouse at any time during the Proterozoic remains speculative. Although relatively small surface oxidant reservoirs (e.g., O2 and SO42−) during Proterozoic time may have allowed large CH4 fluxes from the marine biosphere (9), the geologic record does not provide quantitative constraints on either CH4 fluxes or the CH4 content of the atmosphere, and some models suggest only modest CH4 accumulation in the Proterozoic atmosphere compared with the Archean (12, 13). Characterization of the Proterozoic greenhouse is further complicated by (i) strong nonlinearity in biospheric CH4 fluxes as a function of marine SO42− concentrations, which arises because CH4 consumption and CH4 production are increased and decreased, respectively, with increasing [SO42−] (14, 15); and (ii) strong nonlinearity in the atmospheric lifetime of CH4 as a function of atmospheric pO2, which reflects the competing effects on CH4 stability from O2 content of the ocean−atmosphere system and correspondingly greater UV shielding by O3 in oxidizing atmospheres (16, 17). In other words, both the production and preservation of CH4 are disfavored by high levels of O2, but some low threshold level of O2 enhances CH4 preservation in the atmosphere because its photochemical destruction is muted when a protective O3 layer is well established (16, 17). These challenges for quantifying atmospheric CH4 (pCH4) are exacerbated by the nonlinear relationships between atmospheric pO2, crustal sulfide oxidation, and marine pyrite burial, muddling the relationship between atmospheric pO2 and oceanic [SO42−] (18, 19). Thus, the quantitative relationships between O2, SO42−, and CH4, on a global scale, are poorly understood for much of Earth’s history.
In the modern ocean, SO42−, not O2, is the primary oxidant for CH4 because of the coupling between SO4 and CH4 during anaerobic oxidation of methane (AOM) by microbes. Consequently, freshwater, terrestrial ecosystems (e.g., wetlands) are the most important source of CH4 to the modern atmosphere (excluding anthropogenic sources) because the vast quantity of CH4 produced deep within reducing marine sediments does not readily evade oxidation in overlying SO42−-rich pore waters (20). Quantitative isotopic constraints on seawater SO42− concentrations limit Proterozoic [SO42−] to only a few millimoles per liter (21), or possibly several hundred micromoles per liter (22)—a small fraction of the 28 mM characteristic of the modern ocean. Although less than ∼10% of the modern marine SO42− inventory is certainly low in relative terms, a seawater SO42− concentration of 1 mM still carries a far greater electron accepting capacity than the entire O2 reservoir of the generally well-oxygenated modern ocean. Thus, even at comparatively low concentrations, SO42− in pore waters and in an anoxic Proterozoic water column could have been a major obstacle preventing biogenic CH4 from escaping the ocean environment and entering the atmosphere.
Early arguments for elevated pCH4 during the Proterozoic were predicated on the notion that surface CH4 fluxes could have been >10× to 20× modern values due to enhanced methanogenesis and complete inhibition of methanotrophy under low [SO42−] conditions (9, 10), but this assumption has been challenged both experimentally and through the recognition of extensive anaerobic CH4 recycling in modern analog environments. For example, efficient microbial oxidation of CH4 coupled to SO42− reduction has been documented at SO42− concentrations as low as 100 μM (23), which is potentially much lower than the [SO42−] conditions that typified the Proterozoic ocean (21). Recent work has also shown that CH4 oxidation can be metabolically coupled to a range of alternate electron acceptors [e.g., Mn(III/IV) and Fe(III)] in freshwater lake systems (24) and in SO42−-deficient marine sediments (25). These metabolic pathways are theoretically more energetically favorable than SO42−-based AOM and are likely to have been important for Precambrian CH4 cycling given the apparent abundance of Fe and Mn in the Precambrian ocean (23, 26); however, there have been few attempts to include these metabolic sinks for CH4 in quantitative biogeochemical models, with attempts, to date, focusing on photochemical models that do not explicitly resolve redox cycling within the ocean (23, 27).
Somewhat counterintuitively, low pO2 in the Proterozoic may have also been an obstacle to CH4 accumulation in the atmosphere due to strongly diminished UV shielding by O3 at low pO2—which would act to greatly decrease the photochemical lifetime of atmospheric CH4. For this reason, atmospheres with low pO2 can be more oxidizing toward CH4 than higher pO2 atmospheres (16, 17). Thus, there is an optimization of the lifetime of atmospheric CH4 associated with having sufficient O2 to promote UV shielding by O3 but at O2 levels still low enough to minimize oxidative CH4 destruction. Previous models that have suggested the existence of a CH4 greenhouse during the Proterozoic have assumed that pO2 stabilized at, or above, ∼10% present atmospheric levels (PAL) following the GOE (9), but several diverse proxy records collectively indicate that pO2 may have eventually stabilized at much lower levels than those seen during the GOE and associated Lomagundi event (28). Recently, the absence of Cr isotope fractionation in mid-Proterozoic marine sediments suggests that atmospheric pO2 may have been as low as 0.1% PAL, an order of magnitude lower than envisioned by some authors (29). Consequently, it is possible that previous attempts to characterize the Proterozoic greenhouse have overestimated the lifetime for atmospheric CH4.
In response to recent revelations regarding the high efficiency of anaerobic CH4 oxidation under low [SO42−] conditions (23⇓–25) and refined constraints on atmospheric O2 levels during the mid-Proterozoic (29), we have revisited the role of CH4 in mid-Proterozoic climate stabilization and Neoproterozoic climate perturbation. Our calculations were performed using the grid-enabled integrated Earth system model (GENIE), an Earth system model of intermediate complexity. GENIE considers a 3D marine biosphere, including nutrient-limited export production, aerobic respiration, sulfate reduction, and methanogenesis. The marine biogeochemical module also includes aerobic methanotrophy and sulfide oxidation (see ref. 30 for a detailed description). The ocean is divided as a 36 × 36 equal-area grid with 16 depth layers, and the ocean system is coupled to a 2D energy and moisture balance model, a sea ice model, and a 2D model for the calculation of atmospheric chemistry and spatially resolved, bidirectional sea−air exchange fluxes (31).
For our Proterozoic simulations, we have expanded GENIE’s representation of the CH4 cycle compared with previously published versions of the model. In particular, we have (i) parameterized dynamic calculation of CH4 photooxidation (incorporated after ref. 17); (ii) refined the competition between methanogens and sulfate reducers (improved from ref. 32); and (iii) added AOM coupled to SO42− reduction, the rate law for which we derived from radio-labeled CH4 oxidation rates and chemical profiles from the anoxic water column of the Black Sea (Eq. S1) (33). In combination, these upgrades add strong nonlinearity to both CH4 fluxes and CH4 accumulation as a function of oxidant availability, allowing for an oceanographically realistic calculation of CH4 cycling during oxidant-deficient intervals of Earth history.
We calculated steady-state biogenic CH4 fluxes and resulting atmospheric pCH4 for >75 model configurations, differing primarily in the prescribed sizes of their surface oxidant reservoirs (e.g., O2 and SO42−). We systematically quantified the influence of pO2 (from 10−5 PAL to 10−1 PAL) and oceanic [SO42−] (from 0 to 2.8 mM). We then performed a wide range of sensitivity analyses (Supporting Information), starting from our baseline oxidant levels (10−3 PAL pO2, 280 μM SO42−) and examining the effects of varying (i) prescribed CH4 fluxes from terrestrial environments (0 Tmol CH4⋅y−1 to 22 Tmol CH4⋅y−1), (ii) the oceanic PO43− inventory (i.e., export production, from 0.25× to 2× modern), (iii) whether or not N limitation is considered, (iv) the depth distribution of organic carbon (Corg) remineralization, (v) the rate constants for both aerobic and anaerobic microbial CH4 oxidation, and (vi) the half-saturation constant for SO42− reduction. We also explored model sensitivity to the parameterization of CH4 oxidation within the atmosphere.
Importantly, our modeling experiments were designed to conservatively err in favor of overestimating pCH4 by calculating generous CH4 production fluxes while minimizing CH4 destruction. In our baseline model, we assume complete remineralization of exclusively P-limited export production in a closed ocean−atmosphere system; because this base model stipulates a modern PO43− inventory while neglecting potential N stress and C removal through organic burial, these calculations tend to overestimate the amount of organic matter remineralized via methanogenesis. This potential for overestimating CH4 production is further amplified by our use of a relatively high half-saturation constant for SO42− reduction (500 μM SO42−), which effectively reduces the competitive advantage of SO42− reducers over methanogens. Furthermore, because we neglect anaerobic oxidation of CH4 coupled to electron acceptors other than SO42− (e.g., oxidized Fe and Mn), it is unlikely that we overestimate CH4 oxidation by the marine biosphere. Consequently, the sea-to-air CH4 fluxes and subsequent pCH4 calculations presented here should be considered conservative upper limits (see Supporting Information for further details).
We find that, during the Proterozoic, SO42− would have been the primary control on biogenic CH4 fluxes from the ocean, not unlike today. Methane production by methanogenesis rapidly declines as SO42− reduction, the more energetically favorable metabolism, becomes increasingly competitive at higher [SO42−] and remineralizes a correspondingly greater fraction of organic material exported from the surface ocean. Meanwhile, CH4 destruction via AOM increases with increasing [SO42−], until low CH4 availability ultimately limits AOM kinetics and slows CH4 oxidation. The combined result of these effects is that net biogenic CH4 fluxes to the atmosphere plummet as oceanic [SO42−] increases (Fig. 1A), and the extreme sea-to-air CH4 fluxes suggested by Pavlov et al. (9) are not achievable for any of our model configurations. Thus, relatively modest oceanic SO42−levels can, in principle, preclude significant warming by CH4 on early Earth, regardless of atmospheric pO2.
Oxidant controls on CH4 cycling. (A) CH4 production flux (methanogenesis; blue squares) and CH4 consumption flux (the sum of aerobic methanotrophy and AOM; red circles) as a function of marine SO42−. The difference between methanogenesis and methanotrophy is equal to the net biogenic CH4 flux (i.e., the flux of CH4 that enters the atmosphere through sea−air exchange). Note that the SO42− experiments shown here assume pO2 = 10−3 PAL (29). (B) Net biogenic CH4 flux to the atmosphere (red squares) and atmospheric pCH4 (blue circles) as a function of atmospheric pO2. Net biogenic CH4 production declines with increasing O2 due to enhanced methanotrophic oxidation of CH4, but CH4 accumulation in the atmosphere increases despite reduced CH4 supply due to the establishment of an increasingly effective O3 layer (see figure 7 from ref. 34). Note that pO2 experiments shown here assume [SO42−] = 280 μM.
At higher atmospheric O2 levels, net biogenic CH4 fluxes also decline as a consequence of higher rates of methanotrophy—both aerobic (i.e., through direct metabolic consumption of CH4 with O2) and anaerobic (i.e., through more effective regeneration of SO42− via S2− reoxidation). Globally, however, aerobic methanotrophy is quantitatively much less significant than AOM because, when pO2 is low, oxygenation and aerobic methanotrophy are spatially restricted to the photic zone (Fig. 2A), which represents less than a few percent of the ocean by volume, whereas AOM occurs throughout the ocean interior in closer association with CH4 production (Fig. 2C). Consequently, CH4 oxidation linked to SO42− reduction in a broadly anoxic Proterozoic ocean greatly exceeds CH4 oxidation by O2 globally. The influence of atmospheric pO2 on aqueous CH4 oxidation is further muted when atmospheric pO2 is low because the surface ocean—the site of O2 production—is widely supersaturated with O2 with respect to the atmosphere when pO2 is less than ∼2.5% PAL. Under these conditions, sea−air exchange of O2 is effectively unidirectional, which partially decouples atmospheric pO2 from dissolved O2 and aerobic metabolism in the surface ocean.
(A) Zonally and annually averaged dissolved O2 profile. The distribution of dissolved O2 for our standard Proterozoic ocean is strikingly similar to the O2 landscape of the Archean ocean (32). Although the photic zone is widely supersaturated with respect to O2, the subsurface ocean is pervasively anoxic under Proterozoic conditions, and aerobic methanotrophy is restricted to the surface ocean. (B) Zonally and annually averaged SO42− profile. SO42− is strongly heterogeneous in the subsurface ocean. Where organic fluxes are elevated and water masses are older, pronounced SO42− minimum zones are established. These environments are directly analogous to modern O2 minimum zones (Fig. S1). Note that, where [SO42−] is plotted elsewhere, globally uniform concentrations are not implied; instead, [SO42−], as in Fig. 1 and Fig. 3, represents the global average concentration of SO42−. (C) Zonally and annually averaged CH4 oxidation rates. Contours represent the combined rates of aerobic and anaerobic methanotrophy. Generally, CH4 oxidation rates mirror CH4 concentrations (Fig. S1). For reference, the modern rate of CH4 oxidation in the anoxic water column of the Black Sea is ∼600 nmol kg−1⋅y−1 (33). Like the Black Sea, CH4 oxidation is dominated by anaerobic metabolism.
Zonally and annually averaged CH4 profile. Biogenic CH4 production is greatest in the shallow subsurface where organic fluxes are high and SO42− concentrations are suppressed by extensive anaerobic respiration. CH4 accumulation in oxygenated surface waters is extremely limited, and CH4 is not effectively mixed into deep ocean. Unlike atmospheric CO2, the atmospheric CH4 reservoir is not buffered by the deep ocean (42).
Despite limited influence on net biogenic CH4 fluxes, atmospheric pO2 exerts strong control on the accumulation of CH4 in the atmosphere when [SO42−] is sufficiently low to allow significant CH4 escape to the atmosphere. For the range of atmospheric pO2 that is consistent with the absence of mass-independent fractionation of S isotopes and unfractionated Cr in Proterozoic marine sediments (10−5 to 10−3 PAL) (29, 34), the atmospheric lifetime of CH4 increases with increasing pO2 as the result of an increasingly effective O3 UV shield (Fig. S2). This effect ultimately dominates at steady state over enhanced methanotrophic oxidation of CH4 when pO2 is very low, with the result that atmospheric pCH4 is greater despite reduced production of CH4 in an increasingly oxygenated Earth system. For example, as pO2 increases from 0.1 to 1% PAL, net biogenic CH4 production declines more than 20% whereas atmospheric pCH4 increases more than 200% as a consequence of muted photochemistry and a correspondingly greater lifetime for atmospheric CH4 (Fig. 1B).
Atmospheric CH4 oxidation vs. pO2. For the range of Proterozoic pO2, highlighted in red, CH4 oxidation rates in the atmosphere decline with increasing pO2 as a consequence of enhanced UV shielding by O3; therefore, on Proterozoic Earth, higher pO2 allows higher pCH4 for the same net biogenic CH4 flux. Note that the y axis lacks labels because the actual CH4 oxidation rate also scales as pCH40.7 (after ref. 17).
Combining the opposing effects of [SO42−] and pO2, we find that a robust steady-state CH4 greenhouse is plausible for only a very narrow window of pO2−[SO42−] parameter space (Fig. 3A). Atmospheric pCH4 values greater than ∼10 ppm by volume (ppmv) do not occur for [SO42−] > 1 mM, whereas pCH4 values greater than ∼100 ppmv are restricted to pO2 between 1 and 10% PAL. Thus, substantial accumulation of CH4 in the atmosphere requires the unique combination of exceptionally low [SO42−](<<1 mM), which allows sufficiently large CH4 fluxes from the marine biosphere, and relatively high pO2 (1 to 10% PAL), which optimizes the atmospheric lifetime of CH4. Although uncertainties remain regarding the coevolution of pO2 and marine [SO42−], this requirement for modest pO2 and very low SO42− is not consistent with the most recent [SO42−] and pO2 proxy records and likely would have been difficult to maintain over geologic timescales considering the coupling between atmospheric pO2, crustal sulfide oxidation, and SO42− fluxes. These conclusions regarding oxidant controls on pCH4 are insensitive to uncertainty in other model parameters (Fig. 3 B–E and Table S1), indicating that our salient results are robust, and these constraints challenge the paradigm of persistently elevated pCH4 throughout the Precambrian.
(A) The pCH4 contours for possible Proterozoic oxidant conditions. For reference, 100 ppmv CH4 to 300 ppmv CH4 is required to reconcile the absence of glaciation with the faint young Sun (9), but this value is almost certainly an underestimate (37), unless warming by pCO2 was higher than that calculated by the current generation of 1D atmospheric models. A CH4 greenhouse is favored by pO2 ≈ 1 to 10% PAL, whereas pCH4 > 10 ppmv is precluded for [SO42−] > 1 mM. (B−E) The pCH4 sensitivity to select parameters. The red bar in each panel represents our standard model configuration, and the gray bars represent pCH4 results for simulations that were run under identical oxidant conditions and differ with respect to (B) the flux of CH4 from terrestrial environments, (C) the half-saturation constant for SO42− reduction, (D) the oceanic PO43− inventory, and (E) the depth distribution of organic matter remineralization. Model sensitivity to each of these parameters is small compared with the influence of surface oxidant conditions (A). Unless otherwise specified, the values on the x axis of each panel are presented as times standard value (red bar). See Table S1 for full sensitivity results.
Sensitivity experiments
Elevated pCH4 on the order of 100 ppmv could provide ∼6 K of greenhouse warming, which would substantially, although not completely, compensate for reduced solar luminosity if Proterozoic Earth was persistently warmer than today (10); thus we consider ∼6 K to be the minimum radiative deficit implied by our model if the Proterozoic [SO42−] approached 1 mM or if pO2 was lower than previously appreciated. This deficit is a substantial one that is difficult to reconcile with the absence of glaciation and the suggestion that Proterozoic pCO2 was <10× modern (35).
Other reduced greenhouse gases that are not explicitly considered by our model may have bolstered the mid-Proterozoic greenhouse, helping to alleviate the radiative deficit produced by low atmospheric pCH4. Most notably, atmospheric levels of nitrous oxide (N2O) during the Proterozoic may have been much higher if widespread euxinia limited Cu availability and inhibited effective conversion of N2O to N2 during the final stage of denitrification (36), potentially yielding ∼5 K of supplemental warming (10). Although SO42− would not severely throttle N2O fluxes, N2O—like CH4—is rapidly photolyzed at low pO2. Nitrous oxide, therefore, is only a viable greenhouse contributor for a partial range of the proposed pO2 conditions for the Proterozoic (10). Higher hydrocarbons, particularly ethane (C2H6), can be photochemically produced from CH4 and may have been important for regulating the climate of Archean Earth (37), but our model results generally preclude the elevated pCH4:pCO2 conditions required for substantial hydrocarbon polymerization. Thus, greenhouse contributions from other hydrocarbons are unlikely during Proterozoic time.
In combination with evidence for relatively low pCO2 during much of the Proterozoic (35), we conclude that the faint young Sun paradox may still be a problem for the Proterozoic if relatively high CH4 levels are precluded by current estimates of oceanic SO42− and/or atmospheric pO2. Continued exploration of the metabolic limitations of CH4 oxidizing consortia (11), revised pCO2 calculations, or more realistic consideration of heat transport in increasingly sophisticated 3D coupled climate models may eventually reconcile our suggestion of low pCH4 with clement conditions during the mid-Proterozoic, but, given current constraints on the warming potential of CO2 during Proterozoic time, the paradox remains unresolved.
Given the possibility that baseline Proterozoic pCH4 was much lower than previously envisioned, the role of CH4 in triggering Neoproterozoic climate collapse also requires revisiting. Several authors have previously suggested that the oxidative collapse of a CH4 greenhouse was involved in the initiation of Neoproterozoic snowball glaciations (1, 8, 9). Indeed, emerging trace metal proxies suggest atmospheric oxygenation in advance of the glaciation (29, 38), despite unclear evidence for oxygenation within the marine realm (39⇓–41). The climatic consequences of the oxygenation of a low pCH4 atmosphere, however, have not been previously considered.
We suggest that a Neoproterozoic oxygenation event might have triggered a CH4-based climate collapse, despite low steady-state pCH4, as a result of the vastly differing timescales over which photochemical, biological, and weathering processes modulate Earth’s greenhouse. For example, the immediate consequence of an atmospheric oxygenation event would be an increase in the photochemical lifetime of CH4 via UV shielding by O3, and thus an initial increase in atmospheric pCH4. Subsequent destabilization of CH4 hydrates as the result of a warming climate may amplify this initial accumulation of CH4 in the atmosphere (12, 42). The resulting greenhouse, however, would be inherently unstable because CH4-induced warming would promote the drawdown of CO2 on weathering timescales (12, 42); thus, the climate system would then be vulnerable to any disruption of the CH4 source to the atmosphere (42), including the gradual accumulation of oceanic SO42− that would accompany oxygenation (9). A CH4 greenhouse would also be sensitive to CH4 instability arising as the result of either further oxygenation or even deoxygenation.
Although conceptually similar to the “methane shotgun” scenario advanced by Schrag et al. (42), our snowball initiation scenario has the advantage of eliminating the need for a large, sustained external source of CH4 to the ocean−atmosphere system and does not rely on either enhanced biological CH4 production or high steady-state pCH4 before climate destabilization. Glacial initiation through the buildup and collapse of a CH4 greenhouse in response to an oxygenation event may also provide an explanation for the enigmatic C isotope excursions of the late Neoproterozoic (12, 42). Future geochemical and model analyses that clarify the relationship between oxygenation, glaciation, and isotopic perturbation in the Neoproterozoic will be required to validate our proposed snowball scenario and could provide implicit support for our conclusion that baseline pCH4 was very low during much of the Proterozoic.
We have shown that long-term stabilization of Earth’s climate system by CH4 is challenging for much of Earth history, despite the likelihood of greatly enhanced CH4 cycling within the broadly anoxic Proterozoic ocean. Even at very low SO42− concentrations, anaerobic oxidation of CH4 in the ocean is sufficient to severely throttle CH4 fluxes to the atmosphere, precluding the stability of an atmosphere with >100 ppmv CH4 as previously suggested for Proterozoic Earth. If recent pO2 constraints from Cr isotopes are correct, reduced UV shielding by O3 may exacerbate this issue, further limiting the potential accumulation of CH4 in the atmosphere—with the implication that mid-Proterozoic pCH4 may not have been markedly different from the low levels present today. In this scenario, our results suggest that elevated CH4 should not be invoked to reconcile existing constraints on the composition of the mid-Proterozoic greenhouse with the absence of glaciation in the mid-Proterozoic. Thus, the faint young Sun paradox remains unresolved.
Although our model results imply relatively low steady-state pCH4 (<10 PAL) during the second half of Earth history, they do not preclude a significant role for O2 and CH4 in destabilizing the Neoproterozoic climate system. Rather than occurring as the result of the demise of a steady-state CH4 greenhouse, we suggest that the extreme low-latitude glaciations during the Neoproterozoic are the consequence of non–steady-state CH4–oxidant dynamics produced by the interplay of the marine biosphere, photochemistry, weathering, and climate.
Our calculations were performed using GENIE, an Earth System model of intermediate complexity. GENIE considers a 3D marine biosphere, including phosphate-limited export production, aerobic respiration, sulfate reduction, and methanogenesis. The marine biogeochemical module also includes aerobic methanotrophy and sulfide oxidation (see ref. 30 for a detailed description). The ocean system is coupled to a 2D energy and moisture balance model, a sea ice model, and a 2D model for the calculation of atmospheric chemistry and bidirectional sea−air exchange fluxes (31). The model is well calibrated for simulations of the Cenozoic Earth system, and it has also been used successfully to explore older intervals of Earth history, including the Permian and the Precambrian (32, 43).
For our Proterozoic simulations, we have greatly expanded GENIE’s representation of the CH4 cycle compared with previously published versions of the model. In particular, we have parameterized dynamic calculation of CH4 photooxidation (incorporated after ref. 17), refined the competition between methanogens and sulfate reducers (improved from ref. 32), and added anaerobic oxidation of CH4 (AOM) coupled to SO42− reduction, which is modeled after water column AOM rates and chemical profiles from the modern Black Sea (33) In combination, these upgrades add strong nonlinearity to both CH4 fluxes and CH4 accumulation as a function of oxidant availability, and they allow realistic calculation of CH4 cycling during oxidant-deficient intervals of Earth history.
The previously published representation of water-column methanogenesis in GENIE assumed that methanogens did not initiate biological production of CH4 until the SO42− reservoir was completely depleted (32). Here, we revised the competition between SO42− reducers and methanogens to allow CH4 production under low SO42− conditions. In the model, export productivity from each of the 934 surface ocean cells is calculated based on phosphate availability during each time step, neglecting the possibility of metal or N limitation. Remineralization of exported organic matter occurs following a fixed e-folding depth, which specifies the fraction of exported organic matter that is remineralized in each of the 16 depth layers. That is, the oxidant demand—and, ultimately, potential CH4 production—at each depth is specified by the e-folding depth and the supply of phosphate to surface waters. Upon the exhaustion of O2, the competition between SO42− reduction and methanogenesis begins. The maximum rate of SO42− reduction is specified by the calculated pool of organic matter awaiting remineralization at the end of each time step in each subsurface cell. Then, the actual rate of SO42− reduction is calculated by scaling down this maximum rate using a Monod-type expression, invoking a half-saturation constant for SO42− reduction equal to 500 μM, which is higher than most published values [e.g., 100 μM to 300 μM (44)]. All remaining organic matter in each subsurface cell after SO42− reduction is completely remineralized by methanogens—we do not allow organic matter burial; this means that, at 500 μM SO42−, the rates of SO42− reduction and methanogenesis are equal within anoxic grid cells. For SO42− >> 500 μM, rates of methanogenesis abruptly decline; for [SO42−] << 500 μM, methanogenesis dominates over SO42− reduction. Combining our assumption of 100% remineralization of phosphate-limited export with our high value for the SO42− half-saturation constant, our biogenic CH4 fluxes are generous. Our inability to generate a CH4 greenhouse at low, Proterozoic [SO42−], therefore, cannot be attributed to unrealistically low CH4 production.
In our model, anaerobic oxidation of CH4 coupled to SO42− reduction is first order with respect to [CH4] and obeys Monod-type limitation for [SO42−] (23),
We assigned the value of half-saturation constant,
Assuming CH4 oxidation is first order with respect to CH4 (46), and that SO42− was the only electron acceptor used in the oxidation of CH4 in the anoxic, but SO42−-replete, subchemocline water column of the Black Sea, we calculated
Finally, it is important to note that, in all of the simulations presented here, AOM exclusively occurs when it is thermodynamically favorable; neither CH4 nor SO42− is completely depleted by AOM. Our inability to consistently generate a CH4 greenhouse cannot be attributed to excess CH4 oxidation under energetically limiting CH4−SO42− conditions; this is particularly true considering that anaerobic oxidation of CH4 coupled to alternate electron acceptors [e.g., Mn(III/IV), Fe(III); refs. 23⇓–25] occurs in environments where SO42− is limiting on modern Earth (e.g., ferruginous Lake Matano), but we have neglected these CH4 oxidation pathways in our simulations. For this reason, it is very unlikely that we overestimate aqueous CH4 oxidation rates or underestimate sea-to-air CH4 exchange fluxes and Proterozoic pCH4.
Bidirectional sea−air exchange of CH4 is calculated after Wanninkhof (47) for each surface ocean cell. Then, oxidation of CH4 in the atmosphere is calculated after Goldblatt et al. (17). CH4 oxidation rates are strongly nonlinear with respect to pO2 (Fig. S2), which reflects increasingly effective UV shielding of CH4 by O3 at higher pO2. Although our photochemistry is highly simplified, the parameterization that we have used here captures the salient O2−CH4 dynamics simulated by the more sophisticated photochemical model of Claire et al. (16). Although there are notable discrepancies between the two models at the very low pO2 levels that were characteristic of the Archean, modeled CH4 oxidation kinetics are comparable for the range of pO2 that we explored for our Proterozoic simulations (see supplementary discussion from ref. 13). Nonetheless, we ran simulations in which we dramatically reduced the rate of CH4 oxidation calculated using the parameterizations from Goldblatt et al. (17) to assess the possibility that we systematically overestimated CH4 oxidation rates in the atmosphere by a factor of ∼2 based on the model agreement reported by Daines and Lenton (13). Although atmospheric pCH4 modestly increases for lower oxidation rates, the sensitivity of modeled pCH4 with respect to atmospheric oxidation kinetic parameters is small compared with control exerted by atmospheric pO2 and oceanic [SO42−]. Our conclusion that elevated pCH4 is difficult to reconcile with geochemical proxy records for pO2 and [SO42−] is thus not undermined by our relatively simple parameterization of coupled O2−O3−CH4 photochemistry.
We thank David Catling, Tim Lenton, and an anonymous reviewer for constructive and insightful reviews. This work was supported by the National Aeronautics and Space Administration Astrobiology Institute (C.T.R. and T.W.L.), the National Science Foundation Earth-Life Transitions program (C.T.R. and T.W.L.), and the National Science Foundation Frontiers in Earth System Dynamics program (T.W.L.).
Author contributions: S.L.O. and C.T.R. designed research; S.L.O. performed research; S.L.O. and C.T.R. analyzed data; and S.L.O., C.T.R., and T.W.L. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1608549113/-/DCSupplemental.
