PERSPECTIVE

The snowball : A disaster triggered by the evolution of oxygenic

Robert E. Kopp*, Joseph L. Kirschvink, Isaac A. Hilburn, and Cody Z. Nash Division of Geological and Planetary Sciences, California Institute of Technology 170-25, Pasadena, CA 91125

Communicated by Paul F. Hoffman, Harvard University, Cambridge, MA, June 14, 2005 (received for review April 8, 2004)

Although biomarker, trace element, and isotopic evidence have been used to claim that oxygenic photosynthesis evolved by 2.8 giga-annum before present (Ga) and perhaps as early as 3.7 Ga, a skeptical examination raises considerable doubt about the pres- ence of producers at these times. Geological features suggestive of oxygen, such as red beds, lateritic paleosols, and the re- turn of sedimentary sulfate deposits after a Ϸ900-million year hiatus, occur shortly before the Ϸ2.3–2.2 Ga Makganyene ‘‘snowball Earth’’ (global glaciation). The massive deposition of Mn, which has a high redox potential, practically requires the presence of envi- ronmental oxygen after the snowball. New age constraints from the Transvaal Supergroup of South Africa suggest that all three gla- ciations in the Huronian Supergroup of Canada predate the Snowball event. A simple cyanobacterial growth model incorporating the range of C, Fe, and P fluxes expected during a partial glaciation in an anoxic world with high-Fe oceans indicates that oxygenic pho- tosynthesis could have destroyed a greenhouse and triggered a snowball event on timescales as short as 1 million years. As the geological evidence requiring oxygen does not appear during the Pongola glaciation at 2.9 Ga or during the Huronian glacia- tions, we argue that oxygenic evolved and radiated shortly before the Makganyene snowball.

oxygen ͉ Makganyene glaciation ͉ Huronian glaciations ͉ cyanobacteria

fter the rise of itself, the and some abiotic change, such as the the Griqualand West region. The Makg- most radical transformation of long-term escape of hydrogen to space anyene interfingers with the Earth’s biogeochemical cycles (9), was the direct cause of planetary oxy- overlying Ongeluk flood basalts, which occurred in the transition from genation. We suggest, however, that the are correlative to the Hekpoort volca- Aan anoxic to an oxic world. This trans- data are also consistent with scenarios nics in the eastern domain and have a formation took place in three phases. without oxygenic photosynthesis in the paleolatitude of 11° Ϯ 5° (14). In its up- First, oxygenic photosynthesis evolved Archean. Herein we discuss an alternate per few meters, the Makganyene diamic- and brought into the world locally oxic hypothesis, one in which the evolution of tite also contains basaltic andesite clasts, environments. Second, oxygen became a cyanobacteria destroyed a methane green- interpreted as being clasts of the Onge- major component of the ; house and thereby directly and rapidly luk volcanics. The low paleolatitude of some authors (1, 2) have suggested that triggered a planetary-scale glaciation, the the Ongeluk volcanics implies that the this period was a protracted phase dur- Ϸ2.3–2.2 Ga Makganeyene ‘‘snowball glaciation recorded in the Makganyene ing which the ocean became euxinic. Earth.’’ and Boshoek Fms. was planetary in ex- Finally, the whole ocean–atmosphere tent: a snowball Earth event (15). Con- system took on its modern oxygen- Geological Setting sistent with earlier whole- Pb–Pb dominated cast. The earliest evidence for glaciation comes measurements of the Ongeluk Fm. (16), Although the timing of and relation- from the late Archean and early Protero- the Hekpoort Fm. contains detrital zir- ship between the three stages have been zoic, which suggests Earth at this time cons as young as 2,225 Ϯ 3 My ago (17), topics of active research for many de- experienced global temperatures not much an age nearly identical to that of the cades, there is still a wide divergence of different from those today. The oldest Nipissing diabase in the Huronian Su- opinion. Evidence from organic biomar- known midlatitude glaciation, recorded in pergroup. As the Makganyene glaciation kers (3–5) and arguments concerning the Pongola Supergroup diamictite, oc- begins some time after 2.32 Ga and trace element mobility (6) and biological curred at 2.9 Ga (10). The period from ends at 2.22 Ga, the three Huronian productivity (7) have convinced many 2.45 Ga until some point before 2.22 Ga glaciations predate the Makganyene that O2-generating cyanobacteria and saw a series of three glaciations recorded snowball. aerobic evolved no later than in the Huronian Supergroup of Canada In contrast to the Makganyene Fm., Ϸ2.78 giga-annum before present (Ga) (11) (Fig. 1). The final glaciation in the the three Huronian are un- and perhaps as long ago as 3.7 Ga. Huronian, the Gowganda, is overlain by constrained in . Poles from the Meanwhile, the developing record of several kilometers of sediments in the Matachewan dyke swarm, at the base of mass-independent fractionation (MIF) Lorrain, Gordon Lake, and Bar River the Huronian sequence, do indicate a Ϸ Ϸ of sulfur in the sedimentary formations (Fms.). The entire sequence is latitude of 5.5° (18), but 2kmof record, as well as some other geochemical penetrated by the 2.22 Ga Nipissing dia- sedimentary deposits separate the base tracers, has been interpreted as supporting base (12); the Gowganda Fm. is therefore of the Huronian from the first glacial a protracted atmospheric oxygenation significantly older than 2.22 Ga. unit (19), which makes it difficult to over the period of Ϸ2.5–2.2 Ga (8). In its eastern domain, the Transvaal draw conclusions about the latitude of An early origin for oxygenic photosyn- Supergroup of South Africa contains two the glacial units based on these poles. thesis demands an explanation of how glacial diamictites, in the Duitschland and surface oxidation was muted for perhaps Boshoek Fms. The base of the Timeball Abbreviations: Ga, giga-annum before present; My, million as long as 1,500 million years (My), until Hill Fm., which underlies the Boshoek years; BIF, banded formation; Fm., formation; MIF, cyanobacteria finally surmounted some Fm., has a Re-Os date of 2,316 Ϯ 7My mass-independent fractionation. geochemical, environmental, or ecological ago (13). The Boshoek Fm. correlates *To whom correspondence should be addressed. E-mail: barrier and successfully oxidized the with the Makganyene diamictite in the [email protected]. planet. Perhaps this scenario is correct, western domain of the Transvaal Basin, © 2005 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0504878102 PNAS ͉ August 9, 2005 ͉ vol. 102 ͉ no. 32 ͉ 11131–11136 Downloaded by guest on September 30, 2021 part of the marine S budget, whereas sharper thermal gradients would have driven homogenization of the S pool. Both effects would have decreased sedi- mentary MIF. If this interpretation is correct, only the final decrease of MIF after Ϸ2.09 Ga may reflect the rise of atmospheric O2, and the oxygenation event may have been more rapid than the length of Farquhar and Wing’s (8) MIF Stage II (2.45–2.09 Ga) indicates.

Appearance of Local O2 Both the Huronian and the Transvaal Supergroups contain features suggestive of, but not demanding, the first appear- ance of O2 in the period between the Gowganda and Makganyene glaciations. As noted in ref. 8, MIF in the Huronian rocks examined so far is at diminished values compatible with either increased atmospheric O2 or enhanced glacial Fig. 1. Proposed correlation of the Huronian Supergroup and the upper Transvaal Supergroup. The and ocean͞atmosphere mix- three Huronian glacial units, penetrated and capped by the Nipissing diabase, predate the Makganyene ing. The Lorrain Fm. contains red beds, diamictite in the Transvaal. The uppermost Huronian glacial unit, the Gowganda Fm., is overlain by as well as a hematitic paleosol at Ville- hematitic units, perhaps reflecting a rise in O2. The basal Timeball Hill Fm. contains pyrite with minimal MIF Marie, Quebec, and the Bar River Fm. (26), whereas the upper Timeball Hill Fm., which we suggest is correlative to the Lorrain or Bar River Fms., contains pseudomorphs after gypsum, contains red beds. The Makganyene diamictite records a low-latitude, snowball glaciation (29), perhaps consistent with an increase in sulfate triggered by the destruction of a CH4 greenhouse. It is overlain by the Kalahari Mn Field in the Hotazel Fm., levels from oxidative weathering of sul- the deposition of which requires free O2. Transvaal is based on ref. 17; Huronian stratigraphy is based on ref. 19. fide minerals on land. In the Transvaal Supergroup, the upper Timeball Hill Fm., like the Lorrain Fm., contains red Low latitude poles in the Lorrain Fm. and the McKim Fm. in the Huronian beds. (20, 21), which conformably overlies the Supergroup,† sedimentary sulfides often Stronger evidence for O2 appears af- Gowganda diamictite, are postdeposi- display MIF (8, 26, 27) over an order of ter the Makganyene glaciation. The On- tional overprints (22). magnitude larger than the largest values geluk Fm. is overlain by the Hotazel observed in modern sulfate aerosols (25, Fm., which consists predominantly of MIF of Sulfur 28). One plausible interpretation of the (BIF). The basal The recent discovery of MIF of S iso- diminished MIF observed in the Roo- half meter of the Hotazel Fm. contains topes in Archean and early ihogite, Timeball Hill, and McKim Fms. dropstones (29) (see Supporting Text, which is published as supporting infor- rocks has provided a major constraint is a rise in atmospheric O2. At present, mation on the PNAS web site), which on atmospheric O2. Large MIF of S (up only record the MIF sometimes to several hundred permil) is produced present in sulfate aerosols; in the marine suggests it was deposited toward the end of the . The Hotazel Fm. by photolysis of SO2 to S by light of environment, with riverine input compos- wavelengths of Ͻ200 nm, which would ing Ϸ90% of sulfate input and aerosols hosts a massive Mn member (29): a blanket of Mn deposition unmatched by have been unable to penetrate to the composing Ϸ10%, MIF is not preserved any other known in the world, Ϸ50 m lower atmosphere had O levels been (25). The small range of MIF could re- 2 thick and with an erosional remnant, the above a few percent of the present at- flect an environment in which atmo- Kalahari Mn Field, measuring Ϸ11 ϫ 50 mospheric level (23). To preserve MIF, spheric chemistry began to approach modern conditions, decreasing the mag- km in extent (30). multiple atmospheric S species must be The Kalahari Mn Field indicates the maintained as partially isolated reser- nitude of MIF, but in which aerosols formed the major component of marine release of large quantities of O2 into the voirs rather than being homogenized by ocean and therefore suggests a highly sulfate input, allowing its preservation. oxidation of reduced species like hydro- active postglacial aerobic biosphere. In gen sulfide and polysulfur, as would oc- The small range of MIF also permits Ϫ at circumneutral pH, only cur at greater than Ϸ10 5 present atmo- an opposite interpretation. Rather than Ϫ NO3 and O2 can chemically oxidize a decrease in atmospheric fractionation, ϩ spheric level O2 (24). Mixing and soluble Mn2 to produce insoluble the diminution could be a product of dilution of the atmospherically fraction- Mn(IV) (31) (Table 1). Although a increased continental input and ocean͞ ϩ ated component, both in the atmosphere Mn2 oxidizing phototroph also could atmosphere mixing driven by glacial and in the oceans, presumably yields the produce Mn(IV), no such organism has observed Archean MIF values of up to conditions. The period of 2.5–2.2 Ga ever been identified. Given the high re- Ϸ10‰ (25). An active oceanic S cycle, was a time of glaciations. Enhanced gla- dox potential of the Mn(IV)͞Mn(II) cial weathering could have made unfrac- as would exist at moderately high O2 couplet, it is likely that no such organ- tionated S from igneous sources a larger levels, also would likely prevent the ism exists. The unexcited P870 reaction preservation of MIF. center in purple bacteria has a redox Before the deposition of the Ϸ2.32 potential too low to accept electrons †Wing, B. A., Brabson, E., Farquhar, J., Kaufman, A. J., Ga Rooihogite and Timeball Hill Fms. Rumble, D., III, & Bekker, A. (2002) Geochim. Cosmochim. from Mn(II) (Table 1), but no known in the Transvaal Supergroup (13, 26) Acta 68, A840 (abstr.). photosystem reaction center aside from

11132 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0504878102 Kopp et al. Downloaded by guest on September 30, 2021 Table 1. Midpoint potentials of relevant Timeball Hall Fm. shows only slight is Ϸ1͞64 (1). Rare earth element pat- redox couplets MIF of S (26), consistent with the terns of BIFs (40) and the stability of E, mV initiation of planetary oxygenation or terrestrial ferrous sulfides in an anoxic enhanced glacial activity. atmosphere suggest that the main source Redox couplet pH 7 pH 8 3. The low-latitude glaciation that of reactive Fe in the Archean was hy- ͞ ͞ † Ϫ Ϫ formed the Makganyene and drothermal, so the P Fe ratio may have CO2 CH4* 230 289 ͞ ͞ 2Ϫ͞ Ϫ‡ Ϫ Ϫ Boshoek diamictites (14) was initi- been closer to 1 2 than 1 64. SO4 HS 217 284 ͞ 2ϩ‡ ated when the production of O2 trig- When the P Fe ratio falls below the Fe(OH)3 (ferrihydrite)͞Fe Ϫ5 Ϫ183 2Ϫ͞ Ϫ Ϫ gered the collapse of a methane critical value, essentially all O2 produced UO2(CO3)2 UO2 (uraninite) 18 137 2ϩ ϩ § greenhouse. Although Pavlov et al. is captured by Fe . Above the critical CO2* ͞ Ϫ͞ ϩ‡ ϩ ϩ (35) proposed a similar trigger, they P Fe value, there is a net release of O2 NO3 NH4 366 292 Ϸ P ͞Pϩ (purple bacteria) Ϸϩ450 assumed a delay of at least 400 My (Fig. 2a). Other dissolved O2 sinks, such 870 870 2ϩ 2ϩ‡ between the onset of oxygenic photo- as Mn ,H2, and CH4, will consume MnO2 (pyrolusite)͞Mn ϩ490 ϩ372 Ϫ͞ ‡ ϩ ϩ synthesis and its surficial expression. some O2, but hydrothermal fluxes of NO3 N2 717 646 O ͞H O ϩ815 ϩ756 We suggest a more immediate linkage. these reductants are up to an order of 2 2 2ϩ ͞ ϩ Ϸϩ 4. The Nipissing diabase intruded into magnitude less than the flux of Fe P680 P680 1,100 (Photosystem II) the Huronian sequence (36), and (41). At today’s levels, these are capable 10 the Ongeluk and Hekpoort volcanics of consuming Ϸ7 ϫ 10 mol of O2 per All reactants not specifically noted are at stan- were deposited (16). y, equivalent to 4 ϫ 10Ϫ10 bar͞y. Al- dard state. Thermodynamic constants are from 5. The Hotazel Fm., which includes BIF though the flux of H2S from vent fluids refs. 76 and 77. Photosystem potentials are from and Mn members, was deposited in in today’s high sulfate oceans is compa- ref. 78. 2ϩ ϭ oxygenated waters in the aftermath rable to that of Fe , in an ocean where *Calculated for pCO2 100 mbar. 2ϩ † of the snowball. Fe is the main electron donor, H2S Calculated for pCH4 ϭ 1 mbar. ‡Calculated with all aqueous reactants at 1 mM. 6. The upper siderite facies of the Ho- levels would be lower. A formaldehyde 12 §Calculated with dissolved U species at 20 nM. tazel Fm. and perhaps also part of rainout flux of Ϸ10 mol͞y (24) would the overlying Mooidrai were be the largest O2 sink, capable of ad- Ϫ9 deposited as cap in the sorbing Ϸ6 ϫ 10 bar of O2 per y, but the P670 reaction center of photosystem process of removing CO2 from a would be nullified by H2O2 rainout once II has a higher redox potential. Photo- thick postsnowball greenhouse (29). even a small amount of O2 accumulated synthetic reduction of Mn(IV) therefore in the atmosphere. O2 production in ex- would likely require a two-part photo- Timescale for Methane Greenhouse cess of Ϸ6 ϫ 10Ϫ9 bar͞y(1barϭ 100 system akin to that involved in oxygenic Collapse kPa) would therefore be sufficient to photosynthesis. For cyanobacteria to be directly responsi- initiate CH4 oxidation, and, once begun, One plausible interpretation of the ble for triggering a planetary glaciation, net O2 production in excess of Ϸ4 ϫ sequence of events leading up to the they must have been able to produce 10Ϫ10 bar͞y would suffice to continue it. Paleoproterozoic snowball Earth is enough O2 to destroy the methane green- At the rates predicted by our model, shown in Fig. 1 and is as follows. house before the – destruction of a 1-mbar methane green- weathering cycle could compensate. On house, if it occurred at all, would likely 1. Three glaciogenic units were depos- sufficiently long timescales, global cooling occur within a few My, a timescale com- ited in the Huronian. The extent of would slow silicate weathering and car- parable with the carbonate–silicate the glaciations is not constrained, but bonate precipitation, thereby allowing weathering cycle. Therefore, either cya- they generally lack the lithographic CO2 to build up in the atmosphere (37). nobacteria did not evolve until shortly features associated with snowball The response time of the carbonate– before the oxygenation event, or the , such as a sharp transition silicate weathering cycle is generally esti- nutrient flux did not reach sufficiently Ϸ from a diamictite to a cap carbonate mated at 1 My (38), although the time high levels at any point after the evolu- (32). Although the Espanola carbon- to replace the greenhouse capacity lost in tion of cyanobacteria until then. ate (33) could be a cap for the Bruce a methane greenhouse collapse may be If cyanobacteria were present during glaciation, without paleomagnetic longer. To estimate the timescale for de- the Huronian glaciations, the increased data or additional lithographic fea- struction of the methane greenhouse, we P flux into the oceans generated by gla- tures, its presence alone is insuffi- constructed a steady-state ocean biogeo- cial weathering (42) should have trig- cient to conclude that the Bruce chemistry model based on the assump- gered the oxygenation event. Instead, glaciation was a snowball event. tions that biological productivity was con- the oxygenation event seems to corre- 2. Some of the earliest continental red trolled by P and N and that Fe(II) late with the later Makganyene glacia- beds were deposited in the Firstbrook oxidation was the main inorganic O2 sink tion; this finding suggests the evolution member of the Gowganda Fm. and in (see Supporting Text). ͞ of cyanobacteria occurred in the interval the Lorrain and Bar River Fms. in The critical P Fe flux ratio for net between the Huronian glaciations and Canada, as well as in the upper oxidation of the surface ocean increases the Makganyene glaciation. Timeball Hill Fm. in South Africa. with the burial rate of C (see Fig. 2a, Whether N limitation could have de- The basal Timeball Hill Fm. has re- which is published as supporting infor- layed the destruction of a methane green- cently been dated at 2,316 Ϯ 7My mation on the PNAS web site). For a P ‡ Ϸ ϫ house depends on the Fe demand of the ago (13). In our proposed correla- flux similar to today’s value of 8 N fixers and the ability of cyanobacteria 10 ͞ 2 tion, all of the red bed-bearing units 10 mol y (39) and a burial fraction of to capture Fe before it reacted with O ϫ Ϫ2 2 were deposited after the last Huro- 2 10 , similar to modern anoxic en- and sank beneath the photic zone as a nian glaciation and before the Makg- vironments, the model results indicate ferric precipitate. With anoxic deep waters anyene glaciation. The formation of that the critical value is Ϸ1͞50. The cur- the red beds could involve local O2, rent ratio of P flux to hydrothermal Fe Ϸ ͞ although it does not demand it (34). flux is 1 2, whereas the ratio of P flux ‡Falkowski, P. G., Follows, M. & Fennel, K. (2003) EOS Trans. Syngenetic pyrite from the basal to hydrothermal and terrigenous Fe flux Am. Geophys. Union 84, Suppl., abstr. U52C-01.

Kopp et al. PNAS ͉ August 9, 2005 ͉ vol. 102 ͉ no. 32 ͉ 11133 Downloaded by guest on September 30, 2021 providing a large source of Fe to the pho- without the Ϸ400 My delay assumed by cur only in aerobic organisms. Fischer tic zone, life for an early Proterozoic dia- others (35). et al. (79) recently demonstrated, how- zotroph would have been easier than for Phosphate flux into the oceans corre- ever, that Geobacter sulfurreducens can modern diazotrophs that must subsist off lates with increased continental weather- synthesize diverse hopanols, although of Fe transported from the continents. ing during glacial intervals (42), so in- not 2-methyl-hopanols, when grown un- C͞Fe ratios in N2-fixing populations of creased continental weathering produced der strictly anaerobic conditions. Thus, Trichodesmium range from Ͻ2,000 to by the glaciations may have increased nu- fossil hopanes do not necessarily imply 50,000 (43). If early cyanobacteria were trient availability above the threshold for the presence of O2-producing organisms, inefficient at both capturing and using Fe net O2 release into the atmosphere. Be- and nothing about 2-methyl-hopanols (e.g., C͞Fe ϭ 2,000; capture efficiency ϭ cause of the relatively low global tempera- suggests that they are any different in 1%), then N limitation could have pro- ture, it would have taken only a trace of this respect. Archean 2-methyl-hopanes tected the methane greenhouse, but under OH radicals from the oxygenic bloom also might have been produced by an- more optimistic assumptions (either a produced when cyanobacteria appeared in cestral cyanobacteria that predated oxy- higher capture efficiency or a higher C͞Fe a high-P ocean to damage the CH4 green- genic photosynthesis. ratio), it would not have done so (Fig. 2b). house enough to bring average global The geologic record, not computational temperatures to Ͻ0°C. During the at least Steranes. Produced by eukaryotes for use models, must ultimately decide. Ϸ35 My (29, 47) it took to build up a suf- in the cell membrane, sterols are pre- ficient CO2 greenhouse to escape from served in the rock record as steranes. Implications of the Possible Late the snowball, hydrothermal fluxes would Brocks et al. (4, 5) recovered steranes, Evolution of Cyanobacteria have built up a rich supply of nutrients in along with 2-methyl-bacteriohopane- The interpretation presented here sug- the oceans. When the planet finally polyol, from 2.78 Ga Pilbara Craton gests that planetary oxygenation began warmed, the oceans were ripe for a cya- shales. Because there is no known an- in the interval between the end of the nobacterial bloom. The O2 produced by aerobic sterol synthesis pathway, they Huronian glaciations and the onset of the bloom cleared out tens of My worth used their discovery to argue for the the Makganyene glaciation and that the of accumulated reductants and thus pro- presence of O2. Paleoproterozoic snowball Earth was the duced the Kalahari Mn field (29). Cholesterol biosynthesis in modern direct result of a radical change in the organisms is a long biochemical path- biosphere. In the Archean and earliest Biomarker Counterevidence for way that employs the following four Proterozoic oceans, life may have been Archean O2 O2-dependent enzymes: (i) squalene fueled predominantly by Fe, with Fe(II) Despite the parsimony of cyanobacterial epoxidase, (ii) lanosterol 14-␣-methyl used as the electron donor for photosyn- evolution occurring within a few My be- demethylase cytochrome P450, (iii) thesis and Fe(III) as the main electron fore the onset of the Paleoproterozoic sterol-4-␣-methyl oxidase, and (iv) latho- acceptor for respiration. The sediments snowball Earth, some organic biomarker sterol oxidase (49). These O2-dependent therefore would be moderately oxidizing evidence and indirect sedimentological enzymes perform reactions that, al- and the surface waters reduced (34). and geochemical arguments have been though not currently known to occur Because the redox potential for Mn2ϩ used to suggest that the origin of cya- biochemically in anaerobic organisms, oxidation is much higher than that of nobacteria dates to far earlier times: at could feasibly occur without O2. More- 2ϩ Fe , Mn would have to be removed least 2.78 Ga and maybe as long ago as over, the substitution of an O2-depen- from the oceans in reduced form. The 3.7 Ga. dent enzyme for an anaerobic step in a carbonates precipitated at this time con- The critical piece of evidence placing biosynthetic pathway appears to be a tain up to Ϸ2% Mn (30), indicating that the origin of cyanobacteria and locally common evolutionary occurrence. Ray- Mn2ϩ reached shallow waters and copre- oxic environments in the Archean is the mond and Blankenship (50) found that, ϩ2 cipitated with Ca ; oxidized Mn is ex- discovery in bitumens from rocks as old of the 473 O2-dependent enzymatically tremely rare. The atmosphere was likely as 2.78 Ga of organic biomarkers appar- catalyzed reactions in the BioCyc data- reducing. Astrophysical models predict ently derived from lipids used by cya- base (www.biocyc.org), 20 have at least the was substantially dimmer than nobacteria and eukaryotes in their cell one O2-independent counterpart that today, but a CH4 greenhouse (44) pro- membranes. Although Brocks et al. (48) performs the same reaction. For in- duced by methanogens living in a re- concluded that the bitumens were likely stance, AcsF catalyzes an O2-dependent duced upper ocean would have kept the syngenetic, they could not exclude the cyclization step in the synthesis of chlo- planet warm enough to allow for the possibility that they were postdepositional rophyll and bacteriochlorophyll, a path- presence of liquid water without leading contaminants. Even if the biomarkers are way that must have existed before the to massive siderite precipitation (45). as old as their host rocks, however, the evolution of oxygenic photosynthesis. This world would have been over- uniformitarian extrapolation of modern The O2-independent enzyme BchE per- thrown when cyanobacteria capable of lipid distributions to the Archean should forms the same reaction as AcsF but oxygenic photosynthesis evolved, which be viewed cautiously. uses vitamin B12 in place of O2 (50). molecular phylogenies indicate occurred The assumption that sterol synthesis is later than the main bacterial radiation Hopanes. Among modern organisms, always O2-dependent and always has (46). The surface waters became oxidiz- 2-methyl-bacteriohopanepolyol is pro- been therefore merits close inspection. ing and more productive. Reduced C duced predominantly by cyanobacteria Indeed, the Hamersley bitumens in- accumulated in the sediments; methano- and in trace quantities by methylotrophs clude their own cautionary message genesis moved from the surface ocean like Methylobacterium organophilum about the application of uniformitarian to the deep ocean, where it was isolated (3). Hopanes derived from 2-methyl- assumptions to fossil lipids. Dinosterane from the atmosphere. Methanotrophy in bacteriohopanepolyol can be preserved is generally accepted to be characteristic the ocean and photochemistry in the in sedimentary rocks, where they have of dinoflagellates and is interpreted as a atmosphere used O2 to transform atmo- been used as tracers for cyanobacteria dinoflagellate biomarker in Phanerozoic spheric CH4 to CO2, a less effective (3). Hopanol synthesis has traditionally rocks (51). Yet even though an Archean . The rise of O2 thus been assumed, based on the understood origin for dinoflagellates seems implau- might have triggered a glacial interval modern distribution of hopanols, to oc- sible, because it would indicate the Ar-

11134 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0504878102 Kopp et al. Downloaded by guest on September 30, 2021 chean origin of modern eukaryotic ronment, either by photooxidation (66) or None of these indirect lines of evi- group not known in the fossil record by Fe(II)-oxidizing phototrophic bacteria dence necessitate oxygenic photosynthe- until at least the Paleozoic (52), Brocks (67, 68). sis. The case for cyanobacteria and lo- et al. (5) found dinosterane. They inter- cally oxic environments existing before preted the molecule as being produced Isotopic Evidence. Rosing and Frei (6) the disappearance of MIF of S isotopes by eukaryotes of unknown affinities, argued based on isotopic evidence that and the massive deposition of Mn in the although an alternative explanation is Ͼ3.7 Ga metashale from West Green- Kalahari Mn Field rests, at the moment, that these modern-style putative Ar- land preserves signs of an aerobic eco- solely on steranes. chean biomarkers are contaminants. system. They found organic C with ␦13C values of Ϫ25.6‰ in the same sedi- Future Directions Indirect Counterevidence for Archean O 2 ments as Pb with isotopic ratios indicat- Our model for a Paleoproterozoic origin Although organic biomarkers may be ing that the samples were originally of cyanobacteria is testable by several difficult to interpret, they are a signifi- enriched in U with respect to Th and methods. It suggests that sterols in Ar- cant improvement on several other geo- interpreted this finding as reflecting an chean rocks, if they are original, were logical tracers that have been used to environment in which U could be cycled synthesized by anaerobic reactions. It argue for the presence of cyanobacteria between its reduced, insoluble U(IV) therefore should be possible to find or and environmental O , including micro- form and its oxidized, soluble U(VI) 2 engineer enzymes capable of synthesizing fossils, , BIFs, and assorted form. They concluded the light C was cholesterol under anaerobic, biochemi- isotopic fractionations. produced by oxygenic phototrophs and cally plausible conditions. In addition, a that biogenic O had permitted the re- 2 continuous biomarker record that Microfossils. Before 2 Ga, when diversi- dox cycling of U. However, all biological fied assemblages with affinities to major C fixation pathways and some abiotic stretches back from time periods when groups of cyanobacteria first appear in mechanisms can produce light C, and, O2 is definitely present into the Archean the fossil record, the microfossil record just like the Fe(III) in BIFs, U(VI) can might reveal transitions in community composition. Current work with samples is murky (53). Some have interpreted form in the absence of O2. At circum- filamentous forms in earlier rocks as neutral pH values, the midpoint poten- from recent drilling programs targeting cyanobacterial remains (54–56), but tial of the U(VI)͞U(IV) couplet is Ϸ0 the late Archean and the Paleoprot- Brasier et al. (57) recently questioned V, similar to the Fe(III)͞Fe(II) couplet erozic has begun this task. An intensive the biogenic of these objects. and considerably less oxidizing than search for biomarkers with definite rela- Ϫ Moreover, cyanobacteria cannot be Mn(IV)͞Mn(II), NO3 ͞N2,orO2͞H2O tionships to metabolism, such as those identified solely by a filamentous form. (see Table 1). derived from the pigment molecules of Many nonoxygenic bacteria are also fila- A strong negative ␦13C excursion in phototrophic bacteria, also would pro- mentous, including some mat-forming organic C at Ϸ2.7 Ga has been inter- duce a more convincing record. A green nonsulfur and purple sulfur bacte- preted as evidence for the evolution of search for cyanobacterial or eukaryotic ria (58, 59) and a methanogenic archeon aerobic methanotrophy (69). Such light fossils that predate 2.0 Ga yet have af- (60). The wide variety of filamentous C suggests repeated fractionation: first finities to modern groups would comple- prokaryotes highlights a problem in in the production of CH4 subsequently ment the geochemical approaches. identifying fossil microbes lacking clear oxidized to CO2, then in re-reduction by With respect to the record of MIF of evidence of cell differentiation based on a primary producer; similar fraction- S, the timeline we propose for a rapid morphology: Any given form has proba- ations are observed today in environ- oxygenation scenario suggests that de- bly arisen many times in Earth history, ments with methanotrophs. But al- creased fractionation during the interval both in extant and extinct organisms. though CH4 oxidation was once thought of the Huronian glaciations may be a to require O2, geochemical measure- byproduct of enhanced glacial weather- Stromatolites and BIFs. Two types of late ments (70) and recent microbiological ing and ocean͞atmosphere mixing. If Archean rock Fms. have often been in- work (71, 72) have demonstrated that this scenario is correct, a similar phe- terpreted as indicating cyanobacterial CH4 oxidation also can occur with alter- nomenon should have occurred in asso- activity: stromatolites and BIFs. Des native electron acceptors, so O2 is not ciation with the Pongola glaciation at Marais (7) argued that large needed to explain the isotopic excur- Ϸ2.9 Ga. Investigation of the Huronian reefs indicate the presence of cyanobacte- sions. Although the anaerobic methane deposits where low-MIF S is found ria and therefore a locally aerobic envi- oxidizing bacteria studied today rely on should reveal these deposits to be sedi- ronment, but large reefs also can form sulfate, which is unlikely to have been mentologically immature, reflecting gla- under anaerobic conditions. Populations available in a high-Fe Archean ocean, cial input. Additionally, the syngenicity of anaerobic methane oxidizers, for thermodynamics permits a variety of instance, have built massive reefs at meth- electron acceptors, including Fe(III), to of the MIF values should be tested through techniques such as the paleo- ane seeps in the (61). In addi- be used for CH4 oxidation. tion, the Archean and Paleoproterozoic S data indicate local spikes in magnetic search for significantly post- oceans were likely more supersaturated sulfate concentration starting Ϸ3.45 Ga depositional formation of sulfides. with respect to calcite and aragonite than (73, 74). Canfield et al. (75) argued that Finally, our model predicts that the Makganyene snowball Earth was a singu- the modern oceans (62), which would these spikes were produced in high-O2 have facilitated the precipitation of large environments. But given the low redox lar event. Convincing paleomagnetic evi- reefs even without biological participation. potential of the sulfide͞sulfate couplet, dence (including positive syn-sedimentary Indeed, abiotic processes may have played local sulfate spikes can be explained by field tests) that demonstrated the Huro- a major role in the formation of many scenarios that do not involve O2. More- nian glaciations were low-latitude would stromatolites (63). Moreover, over, sedimentary sulfate deposits, which contradict our model and instead support although the deposition of ferric iron in disappear in the rock record after Ϸ3.2 a protracted episode of planetary oxygen- BIFs has traditionally been taken to imply Ga (75), do not reappear until after the ation with multiple snowball events not the presence of free O2 (40, 64, 65), BIFs Huronian glaciations, which suggests that directly triggered by a singular event, the also could have formed in a O2-free envi- high sulfate conditions were rare. evolution of cyanobacteria.

Kopp et al. PNAS ͉ August 9, 2005 ͉ vol. 102 ͉ no. 32 ͉ 11135 Downloaded by guest on September 30, 2021 Summary photosynthesis, triggered one of the Sessions, D. Sumner, T. Raub, B. Weiss, and Because of the importance of the evolu- world’s worst climate disasters, the Pa- three anonymous reviewers for advice and leoproterozoic snowball Earth. Intensive discussion; R. Tada for help with fieldwork in tion of cyanobacteria and the planetary the Huronian; A. Pretorius and Assmang oxygenation event in Earth history, it is investigation of the time period of the Limited for access to Nchwaning Mine; and particularly useful to consider multiple Paleoproterozoic glaciations may reveal P. Hoffman for communicating this manu- working hypotheses about these events. whether a novel biological trait is capa- script. This work was supported in part by We propose a model that takes a skepti- ble of radically altering the world and the Agouron Institute and by a National cal attitude toward the evidence for Ar- nearly bringing an end to life on Earth. Aeronautics and Space Administration Astro- biology Institute cooperative agreement with chean cyanobacteria and a protracted the University of Washington. R.E.K. was early Proterozoic planetary oxygenation. We thank R. Adler, N. Beukes, R. Blanken- supported by a National Science Foundation In our alternative scenario, an evolu- ship, J. Brocks, H. Dorland, A. Kappler, J. Graduate Research Fellowship and a Moore tionary accident, the genesis of oxygenic Kasting, A. Maloof, D. Newman, S. Ono, A. Foundation Fellowship.

1. Canfield, D. E. (1998) Nature 396, 450–453. 28. Savarino, J., Romero, A., Cole-Dai, J., Bekki, S. & 55. Schopf, J. W. & Packer, B. M. (1987) Science 237, 2. Anbar, A. D. & Knoll, A. H. (2002) Science 297, Thiemens, M. H. (2003) Geophys. Res. Lett. 30, 70–73. 1137–1142. 2131. 56. Schopf, J. W. (1993) Science 260, 640–646. 3. Summons, R. E., Jahnke, L. L., Hope, J. M. & 29. Kirschvink, J. L., Gaidos, E. J., Bertani, L. E., 57. Brasier, M. D., Green, O. R., Jephcoat, A. P., Logan, G. A. (1999) Nature 400, 554–557. Beukes, N. J., Gutzmer, J., Maepa, L. N. & Kleppe, A. K., Van Kranendonk, M. J., Lindsay, 4. Brocks, J. J., Logan, G. A., Buick, R. & Summons, Steinberger, R. E. (2000) Proc. Natl. Acad. Sci. J. F., Steele, A. & Grassineau, N. V. (2002) Nature R. E. (1999) Science 285, 1033–1036. USA 97, 1400–1405. 416, 76–81. 5. Brocks, J. J., Buick, R., Summons, R. E. & Logan, 30. Cairncross, B., Beukes, N. & Gutzmer, J. (1997) 58. Pierson, B. K. (1994) in Early Life on Earth, Nobel G. A. (2003) Geochim. Cosmochim. Acta 67, 4321– The Manganese Adventure: The South African Man- Symposium, ed. Bengtson, S. (Columbia Univ. 4335. ganese Fields (Associated Ore & Metal Corp. Ltd., Press, New York), No. 84, pp. 161–180. 6. Rosing, M. T. & Frei, R. (2004) Earth Planet. Sci. Johannesburg). 59. D’Amelio, E. D., Cohen, Y. & Des Marais, D. J. Lett. 217, 237–244. 31. Gaidos, E. J., Nealson, K. H. & Kirschvink, J. L. (1987) Arch. Microbiol. 147, 213–220. 7. Des Marais, D. J. (2000) Science 289, 1703–1705. (1999) Science 284, 1631–1633. 60. Joulian, C., Patel, B. K. C., Ollivier, B., Garcia, 8. Farquhar, J. & Wing, B. A. (2003) Earth Planet. 32. Hoffman, P. F. & Schrag, D. P. (2002) Terra Nova J. L. & Roger, P. A. (2000) Int. J. Syst. Evol. Sci. Lett. 213, 1–13. 14, 129–155. Microbiol. 50, 525–528. 9. Catling, D. C., Zahnle, K. J. & McKay, C. P. (2001) 33. Bernstein, L. & Young, G. M. (1990) Can. J. Earth 61. Michaelis, W., Seifert, R., Nauhaus, K., Treude, Science 293, 839–843. Sci. 27, 539–551. T., Thiel, V., Blumenberg, M., Knittel, K., 10. Nhelko, N. (2004) D.Phil. thesis (Rand Afrikaans 34. Walker, J. C. G. (1987) Nature 329, 710–712. Gieseke, A., Peterknecht, K., Pape, T., et al. (2002) University, Johannesburg). 35. Pavlov, A. A., Brown, L. L. & Kasting, J. F. (2001) Science 297, 1013–1015. 11. Young, G. M., von Brunn, V., Gold, D. J. C. & J. Geophys. Res. Planets 106, 23267–23287. 62. Grotzinger, J. P. & Kasting, J. F. (1993) J. Geol. Minter, W. E. L. (1998) J. Geol. 106, 523–538. 36. Noble, S. R. & Lightfoot, P. C. (1992) Can. J. Earth 101, 235–243. 12. Corfu, F. & Andrews, A. J. (1986) Can. J. Earth Sci. 29, 1424–1429. 63. Grotzinger, J. P. & Knoll, A. H. (1999) Annu. Rev. Sci. 23, 107–109. 37. Walker, J. C. G., Hays, P. B. & Kasting, J. F. (1981) Earth Planet. Sci. 27, 313–358. 86, 13. Hannah, J. L., Bekker, A., Stein, H. J., Markey, J. Geophys. Res. Oceans Atmos. 9776–9782. 64. Cloud, P. (1973) Econ. Geol. 68, 1135–1143. R. J. & Holland, H. D. (2004) Earth Planet. Sci. 38. Berner, R. A., Lasaga, A. C. & Garrels, R. M. 65. Holland, H. D. (1973) Econ. Geol. 68, 1169–1172. Lett. 225, 43–52. (1983) Am. J. Sci. 283, 641–683. 66. Cairns-Smith, A. G. (1978) Nature 276, 807–808. 14. Evans, D. A., Beukes, N. J. & Kirschvink, J. L. 39. Bjerrum, C. J. & Canfield, D. E. (2002) Nature 67. Widdel, F., Schnell, S., Heising, S., Ehrenreich, A., (1997) Nature 386, 262–266. 417, 159–162. Assmus, B. & Schink, B. (1993) Nature 362, 834– 15. Tajika, E. (2003) Earth Planet. Sci. Lett. 214, 40. Klein, C. & Beukes, N. J. (1992) in The Proterozoic 836. 443–453. Biosphere, ed. Klein, C. (Cambridge Univ. Press, 68. Kappler, A. & Newman, D. K. (2004) Geochim. 16. Cornell, D. H., Schutte, S. S. & Eglington, B. L. Cambridge, U.K.), pp. 147–152. Cosmochim. Acta 68, 1217–1226. (1996) Precambrian Res. 79, 101–123. 41. Elderfield, H. & Schultz, A. (1996) Annu. Rev. 69. Hayes, J. M. (1994) in Early Life on Earth, ed. 17. Dorland, H. (2004) D.Phil. thesis (Rand Afrikaans Earth Planet. Sci. 24, 191–224. Bengtson, S. (Columbia Univ. Press, New York), University, Johannesburg). 42. Fo¨llmi, K. B. (1995) 23, 503–506. Vol. 84, pp. 220–236. 18. Bates, M. P. & Halls, H. C. (1990) Can. J. Earth 43. Kustka, A., Carpenter, E. J. & Sanudo-Wilhelmy, 70. Barnes, R. O. & Goldberg, E. D. (1976) Geology Sci. 27, 200–211. S. A. (2002) Res. Microbiol. 153, 255–262. 4, 19. Young, G. M., Long, D. G. F., Fedo, C. M. & 44. Pavlov, A. A., Kasting, J. F., Brown, L. L., Rages, 297–300. Nesbitt, H. W. (2001) Sediment. Geol. 141, 233– K. A. & Freedman, R. (2000) J. Geophys. Res. 71. Boetius, A., Ravenschlag, K., Schubert, C. J., 254. Planets 105, 11981–11990. Rickert, D., Widdel, F., Gieseke, A., Amann, R., 20. Williams, G. E. & Schmidt, P. W. (1997) Earth 45. Rye, R., Kuo, P. H. & Holland, H. D. (1995) Jorgensen, B. B., Witte, U. & Pfannkuche, O. Planet. Sci. Lett. 153, 157–169. Nature 378, 603–605. (2000) Nature 407, 623–626. 21. Schmidt, P. & Williams, G. (1999) Earth Planet. 46. Olsen, G. J., Woese, C. R. & Overbeek, R. (1994) 72. Orphan, V. J., House, C. H., Hinrichs, K. U., Sci. Lett. 172, 273–285. J. Bacteriol. 176, 1–6. McKeegan, K. D. & DeLong, E. F. (2002) Proc. 22. Hilburn, I. A., Kirschvink, J. L., Tajika, E., Tada, 47. Pierrehumbert, R. T. (2004) Nature 429, 646–649. Natl. Acad. Sci. USA 99, 7663–7668. R., Hamano, Y. & Yamamoto, S. (2005) Earth 48. Brocks, J. J., Buick, R., Logan, G. A. & Summons, 73. Canfield, D. E., Habicht, K. S. & Thamdrup, B. Planet. Sci. Lett. 232, 315–332. R. E. (2003) Geochim. Cosmochim. Acta 67, 4289– (2000) Science 288, 658–661. 23. Farquhar, J., Savarino, J., Airieau, S. & Thiemens, 4319. 74. Shen, Y. A., Buick, R. & Canfield, D. E. (2001) M. H. (2001) J. Geophys. Res. Planets 106, 32829– 49. Risley, J. M. (2002) J. Chem. Ed. 79, 377–384. Nature 410, 77–81. 32839. 50. Raymond, J. & Blankenship, R. E. (2004) Geobi- 75. Huston, D. L. & Logan, G. A. (2004) Earth Planet. 24. Pavlov, A. A. & Kasting, J. F. (2002) Astrobiology ology 2, 199–203. Sci. Lett. 220, 41–55. 2, 27–41. 51. Moldowan, J. M. & Jacobson, S. R. (2000) Int. 76. Drever, J. I. (1997) The Geochemistry of Natural 25. Ono, S., Eigenbrode, J. L., Pavlov, A. A., Khare- Geol. Rev. 42, 805–812. Waters: Surface and Groundwater Environments cha, P., Rumble, D., Kasting, J. F. & Freeman, 52. Falkowski, P. G., Katz, M. E., Knoll, A. H., Quigg, (Prentice–Hall, Englewood Cliffs, NJ). K. H. (2003) Earth Planet. Sci. Lett. 213, 15–30. A., Raven, J. A., Schofield, O. & Taylor, F. J. R. 77. Langmuir, D. (1978) Geochim. Cosmochim. Acta 26. Bekker, A., Holland, H. D., Wang, P.-L., Rumble, (2004) Science 305, 354–360. 42, 547–569. D., III, Stein, H. J., Hannah, J. L., Coetzee, L. L. 53. Amard, B. & Bertrand-Sarfati, J. (1997) Precam- 78. White, D. (2000) The Physiology and Biochemistry & Beukes, N. J. (2004) Nature 427, 117–120. brian. Res. 81, 197–221. of Prokaryotes (Oxford Univ. Press, New York). 27. Farquhar, J., Bao, H. & Thiemens, M. (2000) 54. Awramik, S. M., Schopf, J. W. & Walter, M. R. 79. Fischer, W. W., Summons, R. E. & Pearson, A. Science 289, 756–758. (1983) Precambrian. Res. 20, 357–374. (2005) 3, 33–40.

11136 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0504878102 Kopp et al. Downloaded by guest on September 30, 2021