Fractionation between inorganic and organic carbon during the Lomagundi (2.22–2.1 Ga) carbon isotope excursion

Earth and Planetary Science Letters 271 (2008) 278–291

Contents lists available at ScienceDirect

Earth and Planetary Science Letters j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e p s l

Fractionation between inorganic and organic carbon during the Lomagundi (2.22–2.1 Ga) carbon isotope excursion A. Bekker a,⁎, C. Holmden b, N.J. Beukes c, F. Kenig d, B. Eglinton b, W.P. Patterson b a

Geophysical Laboratory, Carnegie Institution of Washington, 5251 Broad Branch Rd., N.W., Washington, DC 20015 USA Saskatchewan Isotope Laboratory, Department of Geological Sciences, University of Saskatchewan, 114 Science Place, Saskatoon, Saskatchewan, S7 N 5E2, Canada Department of Geology, University of Johannesburg, Auckland Park 2006, South Africa d Department of Earth and Environmental Sciences, University of Illinois at Chicago, Chicago, IL 60607, USA b c

A R T I C L E

I N F O

Article history: Received 25 November 2007 Received in revised form 7 April 2008 Accepted 10 April 2008 Available online 26 April 2008 Editor: H. Elderfield Keywords: Precambrian carbon cycle Lomagundi Event carbon isotope fractionation between organic and carbonate carbon ocean redox state

A B S T R A C T The Lomagundi (2.22–2.1 Ga) positive carbon isotope excursion in shallow-marine sedimentary carbonates has been associated with the rise in atmospheric oxygen, but subsequent studies have demonstrated that the carbon isotope excursion was preceded by the rise in atmospheric oxygen. The amount of oxygen released to the exosphere during the Lomagundi excursion is constrained by the average global fractionation between inorganic and organic carbon, which is poorly characterized. Because dissolved inorganic and organic carbon reservoirs were arguably larger in the Paleoproterozoic ocean, at a time of lower solar luminosity and lower ocean redox state, decoupling between these two variables might be expected. We determined carbon isotope values of carbonate and organic matter in carbonates and shales of the Silverton Formation, South Africa and in the correlative Sengoma Argillite Formation, near the border in Botswana. These units were deposited between 2.22 and 2.06 Ga along the margin of the Kaapvaal Craton in an open-marine deltaic setting and experienced lower greenschist facies metamorphism. The prodelta to offshore marine shales are overlain by a subtidal carbonate sequence. Carbonates exhibit elevated 13C values ranging from 8.3 to 11.2‰ vs. VPDB consistent with deposition during the Lomagundi positive excursion. The total organic carbon (TOC) contents range from 0.01 to 0.6% and δ13C values range from −24.8 to −13.9‰. Thus, the isotopic fractionation between organic and carbonate carbon was on average 30.3 ± 2.8‰ (n = 32) in the shallow-marine environment. The underlying Sengoma shales have highly variable TOC contents (0.14 to 21.94%) and δ13C values (−33.7 to −20.8‰) with an average of −27.0 ± 3.0‰ (n = 50). Considering that the shales were also deposited during the Lomagundi excursion, and taking δ13C values of the overlying carbonates as representative of the δ13C value of dissolved inorganic carbon during shale deposition, a carbon isotope fractionation as large as ~ 37‰ appears to characterize the production of bulk organic matter in the deeper part of the Pretoria Basin at that time. This enhanced fractionation relative to that observed in shallow-water environments likely reflects heterotrophic (secondary) and chemotrophic productivity at and below a pronounced redoxcline, consistent with the euxinic conditions inferred from independent evidence for the deeper part of the Pretoria Basin. Greater variability in organic carbon vs. carbonate carbon isotopic values on the shallow-marine carbonate platform suggests that the carbon cycling was dominated by a large dissolved inorganic carbon reservoir during the Lomagundi excursion. Our study suggests that in contrast to the Late Neoproterozoic and Phanerozoic, when carbon isotope fractionation between carbonate and organic carbon in the open ocean was mostly controlled by primary producers, in the Paleoproterozoic redox-stratified ocean heterotrophic and chemotrophic productivity overprinted a signal of primary productivity below the redoxcline. This strong imprint of heterotrophic and chemotrophic productivity on organic carbon isotope records complicates the reconstruction of spatial patterns and secular trends in the δ13C values of dissolved inorganic carbon in the Paleoproterozoic seawater. © 2008 Elsevier B.V. All rights reserved.

1. Introduction

⁎ Corresponding author. Present address: Department of Geological Sciences, University of Manitoba, Winnipeg, MB, R3T 2N2 Canada. Tel.: +1 204 474 7343; fax: +1 204 474 7623. E-mail address: [email protected] (A. Bekker). 0012-821X/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2008.04.021

Subsequent to its discovery within carbonates of the Paleoproterozoic Lomagundi Group, Zimbabwe (Schidlowski et al., 1975, 1976), the positive carbon isotope excursion in shallow-marine sedimentary carbonates, later named the Lomagundi Event, has been observed in a number of Paleoproterozoic basins worldwide (e.g. Baker and Fallick,

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

1989a,b; Karhu, 1993; Karhu and Holland, 1996; Melezhik and Fallick, 1996; Bekker et al., 2006a). Although the details of secular carbon isotope variations in Paleoproterozoic seawater are as yet poorly constrained, these chemostratigraphic studies clearly demonstrated that the Lomagundi carbon isotope excursion started before 2.22 Ga, shortly after the Paleoproterozoic glacial epoch, and ended between 2.11 and 2.06 Ga. Accordingly, carbon isotope values of seawater remained high for more than 100 Ma of Paleoproterozoic history (Karhu and Holland, 1996; Melezhik et al., 2007). Atmospheric oxygenation was linked genetically to the Lomagundi Event by Karhu and Holland (1996). More recently, however, Bekker et al. (2004) provided geochronological evidence that the rise in atmospheric oxygen occurred earlier. However, regardless of this genetic relationship, steady-state carbon isotope mass balance constraints predict that a large amount of oxygen (~12 and 22 times the present atmospheric level) was released to the atmosphere over the 100 Ma duration of the Lomagundi Event (Karhu and Holland, 1996). The wide range of values mostly reflects a poor knowledge of the global average carbon isotope fractionation between carbonates and organic carbon for this time interval. Unfortunately, carbonate successions of this age are generally very lean in organic matter, preventing determination of carbonate and organic carbon isotope values on the same samples. An additional complication is the finding of Karhu and Holland (1996) that the carbon isotope record of organic-rich shales deposited in the deeper-water settings appears not to record the Lomagundi Event. This led Hayes and Waldbauer (2006) to reinterpret the Lomagundi Event as a period of globally-enhanced methane production below the marine sediment–water interface during the progressive oxidation of diagenetic environments accompanying the rise in atmospheric oxygen; an explanation that strongly contrasts with the earlier model based on high relative burial rates of organic carbon (e.g. Karhu and Holland, 1996). Because the age of these shales is poorly constrained, we cannot be entirely sure that they are in fact stratigraphically equivalent to the carbonate deposits that record the Lomagundi Event in the shallower-water settings. The Precambrian ocean might have had large dissolved inorganic carbon (DIC) and dissolved organic carbon (DOC) reservoirs (Rothman et al., 2003; Bartley and Kah, 2004); the latter suggestion was used to explain decoupled organic and carbonate carbon records in the early Neoproterozoic (Fike et al., 2006). The DOC reservoir, following Fike et al. (2006), refers to a mass of dead organic carbon suspended in the Precambrian ocean that includes colloidal, particulate, and truly ‘dissolved’ organic matter. The concept of a high DOC reservoir in the Precambrian ocean was originally developed by Logan et al. (1995), who argued for a redox-stratified ocean in the Late Precambrian as a result of slow depositional rates of organic matter before the advent of metazoan grazers and fecal pellet production. Logan et al. (1995) inferred that suspended organic matter was processed by heterotrophs in a water column with a prominent redoxcline. Because the biomass of heterotrophs is 13C-enriched compared to primary producers, the contribution of heterotrophs to TOC may be recognized if specific biomarkers are present or if TOC displays elevated δ13C values. For example, the TOC of the Terminal Neoproterozoic deeper-water shales is 13C-enriched relative to sediments deposited in shallower-water settings, consistent with a larger contribution of heterotrophs to TOC in the deeper-water settings where degradation of primary biomass would have been more advanced (Logan et al., 1999). The concept of a large DIC reservoir in the early Precambrian seawater is based on the inferred lower solar luminosity at that time requiring higher atmospheric CO2 levels to maintain greenhouse conditions (Grotzinger and Kasting, 1993) and finds supporting evidence in carbonate textures such as inorganic precipitates (tufa), synsedimentary cement crusts, and micrites (e.g. Grotzinger, 1994; Kah and Knoll, 1996). The required atmospheric CO2 level at ca. 2.3 Ga was estimated at 0.03 to 0.3 bars, which is 100 to 1000 times larger than the present atmospheric CO2 level (Grotzinger and Kasting, 1993).

279

Another related issue is the relatively common assumption that the isotopic fractionation between organic and carbonate carbon (Δδ) during the Precambrian was constant (e.g. Eichmann and Schidlowski, 1975; Schidlowski and Aharon, 1992). This assumed constant fractionation is used as a test for post-depositional alteration (e.g. Knoll et al., 1986; Kaufman and Knoll, 1995) and for the reconstruction of carbonate carbon isotopic values from the organic carbon isotope record (e.g. Calver, 2000; Karlstrom et al., 2000; Walter et al., 2000; Kaufman and Xiao, 2003). In both cases, fractionation between 25 and 30‰ is commonly used as a canonical range of values despite the observation of temporal variations in this parameter in the Neoproterozoic and Phanerozic rock record (Hayes et al., 1999). Large DIC and DOC reservoirs as well as redox stratification in the Precambrian ocean should have had a dramatic effect on isotopic fractionation between carbonate and organic carbon, possibly leading to the decoupling of their records both locally and globally. To address these issues, we present new δ13C data on both the carbonate and organic fractions in subtidal carbonates of the 2.22– 2.06 Ga Silverton Formation (Pretoria Group, South Africa) and the correlative Sengoma Argillite Formation, Segwagwa Group across the border in Botswana (Figs. 1 and 2). These data are compared to the δ13Corg values in the underlying organic-rich shales of the Sengoma Argillite Formation. Buick et al. (1998) found a positive δ13Ccarb excursion in carbonates of the lower part of the Silverton Formation from the northeastern part of the Transvaal Basin (Button, 1973; Fig. 2) within the contact-metamorphic aureole of the 2.06 Ga Bushveld Complex, which they correlated to the 2.22–2.1 Ga Lomagundi carbon isotope excursion. Because both the Silverton–Sengoma carbonates and Sengoma shales were deposited during the Lomagundi Event, δ13Ccarb and δ13Corg data will be used to test whether the carbonate and organic carbon records are coupled, and whether enhanced methane cycling alone can account for the massive and prolonged shift in δ13Ccarb values recorded in shallow-marine carbonates between 2.22 and 2.1 Ga. 1.1. Regional geology and stratigraphy The Silverton Formation belongs to the Paleoproterozoic Pretoria Group (Fig. 1) deposited in the open-marine epicontinental Transvaal Basin in South Africa (Fig. 2). The formation is bracketed in age by the underlying ca. 2.22 Ga Hekpoort Lava and the overlying or intruding 2.06–2.05 Ga Rooiberg Felsite Group and Bushveld Complex, respectively (Walraven, 1997; Buick et al., 2001; Dorland, 2004). Further indirect age constraints might be inferred from the ca. 2.14–2.12 Ga U– Pb ages of authigenic metamorphic monazite in metashales, metasiltstones, and quartzites of the Chuniespoort Group and the older Central Rand Group of the Witwatersrand Supergroup reflecting recently recognized tectonic event in the Transvaal Basin (e.g. Rasmussen et al., 2007). This pre-Bushveld compressional event folded the upper part of the Pretoria Group including the Silverton and the overlying Magaliesberg formations in the far western part of the Transvaal Basin into open upright symmetrical folds (Hartzer, 1995, 2000; Bumby et al., 1998). Furthermore, units in the upper part of the Pretoria Group, above the Magaliesberg Formation overlying the Silverton Formation, are less mature with respect to the lower part of the Pretoria Group deposited on the passive continental margin indicating change to an active tectonic regime and a sediment provenance to the north of the Kaapvaal craton (Button, 1973, 1986; Schreiber et al., 1992). These data suggest that the Silverton Formation was deposited during the 2.22–2.1 Ga Lomagundi carbon isotope excursion and most likely shortly before the 2.14–2.12 Ga tectonic event. We therefore assume a ca. 2.15 Ga age for deposition of the Silverton Formation. In our study area, in the northwestern part of the Transvaal Basin, the Silverton Formation experienced only lower greenschist facies metamorphism. The deltaic to offshore marine Silverton Formation is

280

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

Fig. 1. Correlation chart between the Transvaal Supergroup in the Bushveld Basin of southeastern Botswana and in the northwestern part of the Transvaal Basin, South Africa (modified from Key, 1983 and Bekker et al., 2004 with the Rooihoogte–Timeball Hill age from Hannah et al., 2004 and carbon isotope data from this paper). The age of the Hekpoort Formation is inferred based on the correlation with the Ongeluk Andesite of the Griqualand West Basin and its whole rock Pb–Pb age (Cornell et al., 1996) together with the age of the youngest inherited zircons reported by Dorland (2004) for the Hekpoort Formation.

sandwiched between shallow-marine tidally-influenced quartz sandstones of the Daspoort and Magaliesberg formations (Button, 1973; Fig. 1). The lower contact with the Daspoort Formation is sharp

whereas the upper contact with the Magaliesberg Formation is gradational. The Silverton Formation is divided by basaltic lava into two thick units of shales and siltstones with minor chert and

Fig. 2. Schematic map of the Late Archean and early Paleoproterozoic sedimentary successions of South Africa, showing outcrop and sub-Kalahari sand distribution of the Chuniespoort (dark grey) and Pretoria (light grey) groups in the Transvaal structural basin and Ghaap (dark grey) and Postmasburg (light grey) groups in the Griqualand West structural basin of South Africa taken from the digital 1:1,000,000 geology maps of South Africa and Botswana with the additional information from the digital 1:2,500,000 map of sub-Kalahari formations (Council for Geoscience, 1997; Key and Ayres, 2000; Haddon, 2001). Similar color patterns are also used to label the correlative units in the Bushveld (B) and Kanye (K) structural basins, Botswana. All these structural basins used to belong to the same depositional basin. ‘S1’ indicates the location where carbonates in the lower part of the Silverton Formation with high carbon isotope values occur (Button, 1973; Buick et al., 1998). ‘S2’ and ‘Se’ point to the locations of sampled carbonate and black shale sections in South Africa and Botswana, respectively. ‘L’ refers to the location of sequence-stratigraphically correlative Lucknow Formation in the Griqualand West Basin that also contains carbonates with high carbon isotope values.

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

carbonate layers. It has a thick carbonate sequence at the top in the northwestern part of the Transvaal Basin at the border between South Africa and Botswana. The Silverton Formation is known in Botswana as the Sengoma Argillite Formation (Key, 1983; Figs. 1 and 2). Our study is focused on the carbonate sequence in both South Africa and Botswana and on the immediately underlying black shales of the upper unit in the drill core Strat 2 located near Lobatse in Botswana. The formation is between 500 and 700 m thick, whereas the carbonate sequence is up to 167 m at Sengoma Hill in Botswana (Key, 1983).

281

Two partial (~ 30 and 50 m thick) sections of the carbonate sequence were sampled in South Africa 1 to 2 km from the border with Botswana (Figs. 2 and 3). Carbonates, interlayered with shales, are partially dolomitized, laminated, and rarely display the features of a shallow-water depositional environment such as wave ripples, domal stromatolites, silicified nodules, soft-sediment deformation structures, water and gas escape structures, or small-scale crossbedding. Another partial (46 m thick) section of the carbonate sequence was sampled at Sengoma Hill, Botswana (Fig. 4). This

Fig. 3. Measured partial section of the Silverton Formation, South Africa near the border with Botswana with stratigraphic variations in carbonate and organic carbon isotope values shown as well as carbon isotope fractionation between carbonate and organic carbon (Δδ). Insets shown scatter plots for carbonate and organic carbon isotope values vs. Δδ values.

282

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

carbonate section displays wave ripple cross-laminations and flat laminations. In the lower part of this section sheet cracks (Fig. 5) are extensively developed and filled with recrystallized carbonate. These likely developed during early diagenesis as methane and CO2 produced during remineralization of organic matter escaped from sediments.

The drillcore Strat 2 (see Key, 1983) stored at the Geological Survey of Botswana was also logged and sampled (Fig. 6). It consists of two upward-shallowing cycles with organic-rich pyriferous shales in the lower parts of the cycles deposited in a prodelta environment above wave base and hematite-rich quartz sandstones at the top of the cycles deposited in a delta plain environment. Veins

Fig. 4. Measured partial carbonate section of the Sengoma Argillite Formation, Botswana in the type section of Sengoma Hills with stratigraphic variations in carbonate and organic carbon isotope values shown as well as carbon isotope fractionation between carbonate and organic carbon (Δδ). Insets shown scatter plots for carbonate and organic carbon isotope values vs. Δδ values.

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

283

Samples for δ13Corg and TOC analyses were prepared by decarbonating whole rock powders with 1 N HCl, followed by a thorough rinsing in distilled water and drying in an oven at 50 °C for 24 h. δ13Corg analyses were conducted on the Thermo Scientific Delta Plus XL instrument coupled to a Carlo Erba NC2500 Elemental Analyzer at the Environmental Isotope Geochemistry Laboratory, University of Illinois at Chicago. Samples marked with an asterisk in the Supplementary materials, Table 1 were prepared and analyzed at the Geophysical Laboratory, Carnegie Institution of Washington. Aliquots of powdered samples between 1 and 15 mg in weight, depending on TOC content, were decarbonated with 6 N HCl in muffled silver boats, dried overnight in a hood and, subsequently, in an oven at 50 °C, and analyzed on the Thermo Scientific Delta Plus XL instrument coupled to a Carlo Erba NC2500 Elemental Analyzer. All stable isotope values are reported in per mil vs. VPDB and USGS-40, USGS-41, caffeine UIC, and acetanilide (Costech Analytical Technologies) standards were used to monitor external and internal uncertainty that was better than 0.2‰. Total organic carbon abundances were calculated based on carbon content measured with Elemental Analyzer and loss during decarbonation. 2.1. Geochemical data and preservation of primary δ13C values

Fig. 5. Sheet cracks in carbonates of the Sengoma Argillite Formation at Sengoma Hills.

with carbonates, quartz, migrated pyrobitumen, and remobilized sulfides occur locally as, for example, in the 107–115 m depth interval. 2. Methods The least altered (i.e. lacking veins, discoloration, weathering rinds, and silicification) and finest-grained portions of polished thick sections of carbonates from the Silverton Formation were microdrilled with 2 mm in diameter diamond drill bits and reacted at 70°C using a Kiel III carbonate device directly coupled to a Thermo Scientific 253 gas isotope ratio mass-spectrometer in the Saskatchewan Isotope Laboratory, University of Saskatchewan. Each set of 6 samples was bracketed by an internal calcite standard calibrated against the (NIST) NBS-19 standard. Conversion to the VPDB scale was performed using the values −2.20‰ and 1.95‰ for δ18O and δ13C, respectively, for (NIST) NBS-19. The external uncertainty (±1σ) based on NBS-19 analyses is 0.10‰ for δ18O and 0.05‰ for δ13C. Carbonates from the Sengoma Formation were microdrilled and reacted at 70 °C using a GasBench carbonate device connected to a Thermo Scientific 252 instrument at the Geophysical Laboratory, Carnegie Institution of Washington. Each set of 6–7 samples was bracketed by the Geological Survey of Finland dolomite standard (Tytyri dolomite; Karhu, 1993) and conversion to the VPDB scale was performed using the values 0.78‰ and −7.07‰ for δ18O and δ13C, respectively, for this internal standard calibrated against the (NIST) NBS-19 and NBS-20 standards. The external uncertainty (±1σ) is based on the reproducibility of the Tyryri dolomite, which is better than 0.2‰ for δ18O and δ13C. Major and minor element analyses for both sets of samples were performed by ICP-AES using Thermo Iris instrument. The reproducibility of the analyses is ±5% (1σ).

Carbonate samples analyzed are dolostones, limestones, and marls with xenotopic mosaics of anhedral, coarsely-crystalline to very coarsely-crystalline grains with pseudospar, and rarely contain quartz or carbonate veins. Carbonates have a narrow range of δ13C values between +7.6 and +11.2‰ VPDB (Supplementary materials, Table 1). δ18O values show a much larger range from −16.7 to −5.2‰ VPDB. A correlation between δ13Ccarb and δ18Ocarb values, which has often been attributed to metamorphic decarbonation reactions (e.g. Valley, 1986; Baumgartner and Valley, 2001), is not present in our sample set. The carbonates have variable TOC contents ranging from 0.01 to 0.60% and a large range in δ13Corg values from −24.8 to −13.9‰. There is no correlation between TOC content and δ13Corg values as would be expected from the breakdown and volatilization of organic molecules by diagenetic and metamorphic processes (Hayes et al., 1983). Carbon isotope exchange between graphite and carbonate accounts for 13Cenrichment in graphite associated with marble, however graphite forms at temperatures above 500–600 °C (Valley, 2001), well above the lower greenschist facies metamorphism experienced by carbonates of the Silverton and Sengoma Argillite formations. Δδ values are also highly variable, ranging between 23.9 and 33.6‰, although the average (30.3‰) is close to 30‰, that is typical for Phanerozoic and Precambrian unmetamorphosed carbonates (Eichmann and Schidlowski, 1975; Knoll et al., 1986; Hayes et al., 1999). Mn and Sr contents and Mn/Sr ratios are also widely used to evaluate diagenetic effects in carbonate rocks (e.g. Kaufman et al., 1993). Carbonate samples analyzed in this study have Mn contents ranging from 27 to 1505 ppm (most samples contain less than 670 ppm), Sr contents vary from 26 to 119 ppm, and Mn/Sr ratios are between 0.5 and 26.5 (Supplementary materials, Table 1). These data are consistent with calcite or dolomite as the precursor mineralogy, rather than aragonite, and indicate only a minor to moderate amount of post-depositional alteration with respect to other Paleoproterozoic carbonates worldwide (e.g. Bekker et al., 2001, 2003a,b, 2005, 2006a; Melezhik et al., 1999; Veizer et al., 1992). The organic matter in the middle part of the Silverton Formation carbonate section, between 15 and 32 m from the base (Fig. 3), is 13Cenriched, with δ13C values of −18.8 to −13.9‰, compared to background values from −22.5 to −19.1‰. Although this stratigraphic interval has, on the whole, the lowest TOC contents in the section, there are some individual samples from outside of this interval with even lower TOC content that do not show a tendency towards 13C-enrichment compared to neighboring samples (see Supplementary materials, Table 1). In addition, it is difficult to interpret this interval as having been subjected to a higher degree of post-depositional alteration since the δ13Ccarb and

284

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

Fig. 6. Partial section of the Sengoma Argillite Formation, Botswana from the drill core Strat 2 with stratigraphic variations in organic carbon isotope values in two upward-coarsening cycles shown. Inset shows scatter plot for organic carbon isotope values vs. TOC content. Two samples that were strongly altered are shown with empty circles.

δ18O values, and trace element abundances do not appear themselves to be unusual. Alternatively, the observed 13C-enrichment through this stratigraphic interval might reflect high phytoplankton growth rates in the upwelling area where nutrients were delivered from the deep ocean. A similar interpretation was offered to explain the presence of 13C-enriched organic matter in stromatolitic phosphorites and carbonates of the ca. 2.0 Ga Jhamarkotra Formation, Aravalli Supergroup, Rajasthan, India (Banerjee et al., 1986). In our case, the relatively low TOC levels in this stratigraphic interval seem to be in conflict with the high productivity model unless organic matter was reworked from the high productivity area and transported to this depositional site. Regardless of the underlying cause of the 13C-enriched organic matter in this part of the section, it is important to note that there is no correlation between δ13Corg values and TOC contents in this section (see Supplementary materials, Table 1) and

even if these data were removed from the graph of δ13Corg vs. Δδ on the inset in Fig. 3, a significant correlation with a similar slope would still be evident, suggesting that data from this 13C-enriched interval most likely reflect primary values. δ13C values of carbonates in the sampled sections of the Silverton Formation are on average 1‰ more positive than carbonates sampled at Sengoma Hills (9.6 vs. 8.6‰; Figs. 3 and 4). Because these sections are relatively close to each other and are partial sections of much thicker carbonate sequence in both areas, we relate this discrepancy to a slight difference in the stratigraphic position of the sections within the carbonate sequence rather than a shelf gradient in δ13CDIC. The underlying shales of the Sengoma Argillite Formation have higher TOC content, ranging from 0.14 to 21.94 wt.%, and highly variable δ13Corg values from −33.7 to −20.8‰ with the majority of

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

values and the average value (−27.0‰) lower than −25‰ (Fig. 6; Supplementary materials, Table 1). The basal part of the section has the lowest δ13C values. The overlying strata become progressively 13Cenriched upsection reaching a maximum at 219 m. At this point the trend reverses with δ13C values declining to 205 m. From 205 m until to the top of the section there are no variations in δ13Corg values. Two shale samples at the top of the section between 107 and 115 m depth contain veins with carbonates, quartz, migrated pyrobitumen, and remobilized sulfides. Based on their elevated δ13Corg values, these samples appear to be more significantly altered than the samples without veins (Fig. 6). In summary, both the petrographic features and geochemical data suggest that carbonate and organic carbon fractions in the carbonates and shales were not significantly altered following deposition. Therefore, we conclude that δ13Ccarb values reflect the δ13C value of DIC in contemporaneous seawater in the Pretoria Basin. δ13Corg values correspond to the composition of organic matter deposited in this basin and may also reflect contemporaneous seawater DIC δ13C values, offset by some fractionation. Whether or not this was indeed the case depends on the relative contribution of organic matter sourced from primary (photosynthetic) and secondary (e.g. methanotrophic and sulfur-oxidizing) pathways that indirectly reflects redox stratification and CO2, CH4, and S cycling in the basin. 3. Discussion 3.1. Do high δ13Ccarb values of the Silverton–Sengoma carbonates reflect local or global conditions? Accepting that the Silverton–Sengoma δ13Ccarb values closely approximate seawater δ13CDIC values in the Pretoria epicontinental sea, the extent to which local scale carbon cycling processes may have overprinted the global-scale signature of the surface ocean remains unclear. Processes on both scales have been inferred to interpret the Phanerozoic carbonate carbon isotope records (e.g. Kump and Arthur, 1999; Melchin and Holmden, 2006). Modern (e.g. Florida Bay, Bahama Banks, and Shark Bay) and Phanerozoic carbonates show a large range in δ13C values that might reflect vital effects of carbonate-secreting organisms, local environmental conditions (e.g. hypersalinity, biological productivity, terrestrial runoff, and diagenetic mineralization of organic matter), and shelf gradients in epicontinental basins with the restricted connection to the open ocean. Local scale effects were cited by Melezhik et al. (1999) as an explanation for very positive δ13Ccarb values (N + 6 to +8‰) recorded by stromatolitic carbonates that may have been deposited during the Lomagundi Event in potentially hypersaline settings. Because we are interested in comparing fractionations between organic and inorganic carbon in sediments from shallow and deeper parts of the Pretoria basin, and we lack direct constraints on carbon isotope values of DIC in the deeper part of the basin, the following discussion is focused on whether the Silverton– Sengoma δ13Ccarb values reflect local or global signals. If global signals are implied, δ13Ccarb values in the shallow part of the basin can be used as a rough estimate for the carbon isotope values of seawater DIC in the deeper part of the Pretoria basin. Several models have been proposed to explain positive δ13Ccarb excursions in modern and Phanerozoic shallow-marine carbonates that are relevant to this discussion. These include i) weathering of carbonates during sea level lowstands, and the impact of this on the δ13C value of terrestrial runoff (Kump and Arthur, 1999; Melchin and Holmden, 2006); ii) enhanced productivity in restricted settings (e.g. Swart and Eberli, 2005); and iii) 13C-enrichment in evaporating brines (Stiller et al., 1985; but see Lazar and Erez, 1990 for a different view). With respect to the carbonate weathering model, the Silverton and Sengoma Argillite formations do not directly overlie an extensive carbonate platform, although the Late Archean Campbellrand–Malmani subgroups containing carbonates with δ13Ccarb values close to

285

0‰ occur significantly below these units. Enhanced cyanobacterial productivity has been inferred to explain 13C-enrichment beyond background levels during the Lomagundi Event in shallow-marine settings (e.g. Melezhik et al., 1999) with the Bahama Banks and Shark Bay used as modern analogues. High productivity on the Bahama Banks is maintained by red and green calcareous algae in shallow, restricted settings that are not analogous to an open and extensive carbonate platform in the Pretoria Basin where stromatolites reflecting cyanobacterial activity are rare due to the deeper-water depositional environment. Similarly, regardless of the validity of the evaporative model for the origin of extreme 13C-enrichments, there is no evidence for hypersaline conditions during the deposition of the Silverton or Sengoma Argillite formations in the Transvaal Basin. On this basis, we conclude that none of these models appears to satisfactorily explain the 13C-enrichment in the carbonates of these units, and we suggest that δ13Ccarb values in the nearshore settings mainly reflect the global, rather than local signals. Carbon isotope values in carbonates of the Sengoma Argillite and Silverton formations are indeed similar to those previously reported from these formations in other parts of the extensive Transvaal Basin (Master et al., 1993; Buick et al., 1998; Swart, 1999; Bekker et al., 2001) and in the sequence-stratigraphically correlative Lucknow Formation in the Griqualand West Basin (Coetzee et al., 2006; see Fig. 2 for location), and broadly correlative Paleoproterozoic carbonates deposited during the Lomagundi Event worldwide (Bekker et al., 2006a). We do not imply that small-scale shelf gradients were entirely lacking in the Paleoproterozoic epeiric basins; rather we suggest that high atmospheric CO2 levels predicted by climatic models likely caused them to be damped by the higher seawater DIC content and higher exchange rates between atmospheric CO2 and seawater DIC (cf. Panchuk et al., 2005). 3.2. Δδ in shallow-marine carbonate environment: implication for an enhanced methane cycle Δδ is an important parameter for characterizing carbon cycling during the 2.22–2.1 Ga Lomagundi δ13Ccarb excursion (Karhu and Holland, 1996). Whereas the δ13CDIC values are generally assumed to be homogeneous in open-marine settings and faithfully recorded by δ13Ccarb values, the carbon isotope record of marine TOC can be strongly influenced by local environmental conditions such as redox state, nutrient content, and aqueous CO2 levels, in addition to the seawater δ13CDIC value. It is therefore difficult to constrain an average organic carbon isotopic value on the global scale for any specific time interval. δ13Corg values or their stratigraphic trends in environments with high TOC contents could be strongly influenced by secondary productivity occurring in the anoxic water mass of the stratified water column (e.g. methanotrophs or S-oxidizing bacteria; Hayes et al., 1999), especially in the Precambrian ocean where dissolved oxygen levels were likely much lower (Slack et al., 2007). Intense biogeochemical carbon cycling in redox-stratified water columns thus complicates the interpretation of δ13Corg and Δδ values. Although organic-rich sediments that were strongly influenced by secondary productivity are unlikely to provide genuine records of global seawater δ13C values, sections with high TOC contents naturally attract biogeochemical studies and, therefore, bias the secular record of global average δ13Corg values. Shallow-water marine environments are better settings to look for faithful records of seawater δ13C trends, particularly for organic carbon isotopes, because secondary productivity there is confined to anoxic porewaters below the water– sediment interface, and direct comparison with carbonate carbon isotope values is possible in carbonate successions. There is relatively little data on Δδ values from carbonates deposited during the Lomagundi Event because most of these carbonate successions appear to have been deposited in shallow-marine and relatively welloxygenated waters and are, thus, extremely lean in organic carbon (≪1 mg C/g sample) with δ13C values of refractory organic matter

286

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

strongly affected by nearly complete loss of organic carbon (Bekker et al., 2006a). Carbonates of the Silverton Formation therefore provide a rare opportunity to measure Δδ values in shallow-marine carbonate environments during the Lomagundi Event. Assuming that δ13Ccarb values genuinely reflect seawater δ13CDIC values and that δ13Corg values represent average carbon isotope values of organic matter, fractionation between these two reservoirs on the shallow-marine carbonate platform was close to 30‰. Until now, the Lomagundi excursion has only been shown to be present in the carbonate fraction. Since older and younger Late Archean and early Paleoproterozoic carbonate successions show similar Δδ values (e.g. Eichmann and Schidlowski, 1975; Schidlowski, 1988; Des Marais et al., 1992), the secular trend of δ13Corg values from shallow-marine carbonate successions spanning the duration of the Lomagundi Event should record a positive δ13Corg excursion and, thus, coupling between the inorganic and organic carbon isotope records. If this interpretation is found to be correct on a global scale by further studies, then the hypothesis of an enhanced microbial degradation of organic matter associated with methanogenesis below the sediment–water interface in shallow diagenetic settings, and a decoupling of organic and carbonate carbon records during the progressive build-up of molecular oxygen in the oceans and atmosphere (Hayes and Waldbauer, 2006) is not a valid explanation for the origin of high δ13C values of carbonates deposited during the Lomagundi Event. By analogy with Phanerozoic, and younger Precambrian examples, a Δδ of ~ 30‰ suggests that the organic matter was mostly fixed by organisms utilizing a pathway of oxygenic photosynthesis in the water column, or perhaps at the sediment–water interface. The 13C-enrichment in both inorganic and organic carbon on the shallow-marine carbonate platform during the Lomagundi Event, therefore, provides strong evidence that the whole marine DIC reservoir was affected rather than just shallow porewaters below the sediment–water interface as predicted in the model of Hayes and Waldbauer (2006). 3.3. Chemical and redox stratification in the Silverton Basin Carbonates in the upper part of the Silverton–Sengoma Argillite formations and the underlying black shales of the same units were deposited during the 2.22–2.1 Ga Lomagundi Event based on their geochronologic constraints discussed in ‘Regional Geology and Stratigraphy’ section and positive carbon isotope values of carbonates in the lower and upper parts of the Silverton Formation bracketing the studied shales (Buick et al., 1998; this paper). Walther's Law dictates that in offshore settings, shales, that are interlayered with carbonates in the South African section (see Fig. 3), and conformable with an overlying carbonate sequence, are deeper-water time equivalents of shallow-water carbonates found higher in the succession. As discussed above, shales from the Strat 2 drill core in Botswana contain up to 22% TOC and are more organic-rich than the contemporaneous shallow-marine carbonates of the Silverton and Sengoma Argillite formations. The shales, therefore, suggest high organic carbon burial in offshore settings and, potentially, a strong biological pump to the deeper-water settings (cf. Hotinski et al., 2004) during the Lomagundi Event. In contrast to the associated organically lean carbonates, the TOC in the offshore shales displays significantly lower δ13C values. The apparent depth gradient in δ13Corg values between shallow-water carbonates and deeper-water shales is about 6‰, much larger than the water column gradient in the modern ocean (≤2‰; Kroopnick, 1985), even though the difference in depth of deposition for these carbonates and shales in the epicontinental basin was unlikely more than 100 m. Shore-to-basin gradients of about 4‰ have been documented across several epeiric platforms during the Hirnantian carbon isotope excursion (e.g. Melchin and Holmden, 2005). Although evidence for a seawater shelf gradient in δ13CDIC has not yet been convincingly demonstrated in Paleoproterozoic basins, we recognize that it is a challenging problem to amass the necessary evidence on account of

the lack of in situ carbonate production on the deeper shelf, and the potential overprinting of deeper shelf TOC records by secondary productivity, as discussed above. Furthermore, Hotinski et al. (2004) concluded based on box modeling that significant water column gradients would be unlikely to develop in the high CO2 atmosphere predicted by Precambrian climatic models (cf. Kasting, 1993) as vigorous CO2 exchange between the atmosphere and ocean would limit the impact of biological pump. Therefore, accepting that TOC of the Sengoma Argillite Formation with its low δ13C values does not reflect a global ocean water column gradient in δ13CDIC, two alternatives must be considered: i) either the DIC in offshore settings was locally affected by enhanced mineralization of organic matter, or ii) inputs from secondary productivity significantly contributed to the TOC of the offshore shales. The Black Sea is a good example for the first scenario in that it has a significant carbon isotope water column gradient in the DIC from +0.8‰ near the surface to −1.4‰ at 100 m depth and to about −6.3‰ at the bottom in the deep sulfidic part of the basin (Fry et al., 1991; Schouten et al., 2004). Therefore, the carbon isotope value of the DIC reservoir can be strongly affected in stratified basins where the exchange with the ocean is restricted at depth by sills. In contrast, areas with high organic productivity associated with upwelling zones on continental margins are vertically well-mixed and at most 1 to 2‰ lower in δ13CDIC values than the global surface ocean (e.g. Peru Upwelling Zone; Bidigare et al., 1997). Although mineralization of organic matter and methane cycling may have been important in the Pretoria basin, and bearing in mind the fact that sedimentary records of offshore δ13CDIC values are lacking or compromised, there are several lines of indirect evidence suggesting that seawater δ13CDIC values in offshore settings can be approximated by the sedimentary records of carbon in nearshore settings. First, as discussed above, it seems rather odd that a large, open-marine, tidally-influenced epicontinental basin under high Paleoproterozoic atmospheric CO2 levels would develop long-term conditions above the wave base under which δ13CDIC values could be so significantly 12C-enriched offshore. Low δ13C values similar to those observed in the thick black shale section in the drill core Strat 2 also occur in shales with lower TOC (mostly less than 1 wt.%) collected from the outcrop sections in the more proximal parts of the basin (Watanabe et al., 1997; Coetzee, 2001). Second, if a carbon isotope lateral gradient occurred in the Pretoria basin, as observed now in the Black Sea, one would expect to find carbon isotope values of carbonates to be highly variable and depth-controlled, with lower δ13C values at the base of the upward-shallowing cycles immediately above shales (see Fig. 3) and increasingly higher δ13C values at their tops. In contrast, our data do not show significant variations in carbon isotope values of carbonates with sea level changes and only highly positive carbon isotope values were found in this and previous studies (Buick et al., 1998; Swart, 1999). We therefore infer that δ13C values of DIC in the Pretoria basin were not significantly affected on the basin scale by mineralization of organic matter in sediments with high TOC content and use the δ13C values of Silverton–Sengoma carbonates as a proxy for carbon isotope values of DIC in settings where Sengoma shales accumulated. It has been suggested that Δδ values larger than 32‰ in the Late Neoproterozoic interglacial time intervals might be related to the activity of methane-oxidizing or sulfide-oxidizing bacteria incorporating light carbon either from methane or mineralized organic matter (Hayes et al., 1999). Returning to the Black Sea as a potential modern analogue for the Precambrian redox-stratified ocean, significant heterotrophic (secondary) and chemotrophic (e.g. sulfide-oxidizing and thiosulfate-disproportionating bacteria) production of organic matter in the water column there contributes ~45% of the total carbon flux to the sediments; the remainder is from primary productivity in the surface ocean above the redoxcline (Karl and Knauer, 1991). Bulk δ13Corg values for the organic carbon in the bottom sediments of the Black Sea are less than 4‰ lower in δ13Corg value than phytoplankton

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

in the surface waters (Fry et al., 1991; Hayes et al., 1999). This difference is consistent with the ~6‰ on the average lowering of δ13Corg values between nearshore and offshore settings in the Pretoria basin, suggesting that a significant portion of the organic matter in the deeper-water shales originated from heterotrophic and chemotrophic metabolism. This would also imply a stable redoxcline in this part of the basin (Fig. 7). The redoxcline with the pronounced underlying sulfidic conditions was indeed inferred independently for the Pretoria basin on the basis of Fe speciation and δ34S data (Shen and Bekker, 2006; Bekker et al., 2006b; Scott et al., 2008). Sedimentary structures in the black shale section in the drill core Strat 2 (see Fig. 6), such as wave ripples with internal laminations, climbing ripples, and softsediment deformation structures, however, suggest that the redoxcline developed in relatively shallow waters, above wave base, in contrast to the Black Sea analogue. It is plausible that deposition of these organic-rich shales reflects intense upwelling from the deep ocean, bringing nutrients to fuel primary productivity in the basin. The dramatic difference in carbon isotope fractionation between TOC and DIC in the Pretoria basin (~36‰ on the average) and in the Black Sea (~ 27‰ maximum) likely reflects a shallower redoxcline, reduced amounts of seawater sulfate and other oxidants and, consequently, extensive methanogenesis in the deeper anoxic part of the Pretoria basin with methanotrophy at the redoxcline. A related question, therefore, is whether other shallow-water environments spanning the Lomagundi Event show evidence for a pronounced redoxcline, sulfidic conditions, and significant 13Cdepletion in the preserved organic matter. Evidence from the Paleoproterozoic Francevillian Series of Gabon suggests that this may have been the case. Carbonates in the transgressive sequence with δ13Ccarb values ranging from +2.6 to +6.3‰ are interlayered with organic-rich shales with δ13Corg values as low as −46.2‰ (GauthierLafaye and Weber, 1989; 2003). The overlying volcanics were recently dated at 2083 ±6 Ma (Gauthier-Lafaye, 2006) establishing that the organic-rich shales with low δ13Corg values were deposited during the Lomagundi Event. Based on the speciation of Fe in the black shales (Scott et al., 2008), and basin analysis by Gauthier-Lafaye and Weber (2003), the intercalated shales were likely deposited under euxinic conditions in the epicontinental basin influenced, perhaps, by upwelling from the deep ocean. If these findings from two Paleoproterozoic basins may be taken as evidence for extensive euxinia in shallow-water epicontinental seas during the Lomagundi Event, the question arises as to the redox state of the deep ocean during that

287

time. Giant sedimentary Mn-deposits and associated minor iron formations deposited during the Lomagundi Event (e.g. GauthierLafaye and Weber, 2003; Bekker et al., 2003b) strongly suggest that the deep ocean was neither oxic nor euxinic but rather suboxic to anoxic at that time. Combining these observations, the ocean during the Lomagundi Event was likely redox-stratified, with biological activity coupled between oxic surface waters, euxinic intermediate waters, and suboxic to anoxic bottom waters (Fig. 7). Contrary to the generally held view (e.g. Melezhik et al., 2005; Hayes and Waldbauer, 2006) that the Lomagundi Event was not accompanied by the deposition of organic-rich shales consistent with a high relative burial rate of organic carbon, several basins in West and South Africa and Brazil contain organic-rich shales. Therefore, in spite of the fact that the signature of the Lomagundi excursion might be overprinted by secondary productivity in these deeper-water shales, their high TOC and low δ13Corg values suggest that significant amounts of organic carbon with light δ13Corg values were removed from the oceans during this time, helping to explain high δ13C values recorded in shallowwater carbonates. Assuming that our data for the Pretoria basin are typical for globally averaged δ13Ccarb and δ13Corg values of carbonates and organic matter, respectively, from the shallow-water environments, shales of the Sengoma Argillite Formation deposited in the deeper-water part of the basin provide some indication of the decrease in δ13C values of organic carbon below the redoxcline during the Lomagundi Event. It is therefore impossible to estimate a global average fractionation between organic and inorganic carbon during the Lomagundi Event from our dataset. However, because primary producers dominate the ocean organic carbon budget now (Hedges and Keil, 1995) and were likely similarly important during the Lomagundi Event and assuming that shallowwater environments are more likely to reflect primary productivity, we can bracket this parameter between 30 and 45‰; likely closer to 30‰. 3.4. Size of DIC and DOC reservoirs during the Lomagundi Event Our data provide a new perspective on carbon cycling during the Lomagundi Event and allow for a test of the possible effects of large DIC and DOC reservoirs on the operation of carbon cycle. It is generally accepted that except for relatively short but important time intervals in the Late Archean, Paleoproterozoic, and Neoproterozoic, the relative proportion of carbonate and organic matter burial was stable in the

Fig. 7. Model for biogeochemical cycling of carbon in the shallow-water and deeper-water environments of the Pretoria Basin, South Africa during deposition of the Silverton and Sengoma Argillite formations (modified from Bekker and Kaufman, 2007). Note the sulfidic oxygen-minimum zone impinging on the shelf in the shallow-marine tidally-influenced epicontinental basin overlain by the oxygenated upper part of the water column and underlain by the anoxic or suboxic deeper-waters. Organic matter in the shallow oxygenated part of the water column was dominated by primary productivity whereas secondary (heterotrophic) and chemotrophic productivity at and below the redoxcline contributed significantly to TOC in the deeper-water environments (cf. Coetzee et al., 2006; Bekker and Kaufman, 2007).

288

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

Fig. 8. a–c: Schematically shown fields for the Cenozoic carbon cycle and two end-members based on the carbon cycle dominated by either large DOC or large DIC reservoirs on the scatter diagrams of δ13Ccarb vs. δ13Corg (a); δ13Ccarb vs. Δδ (b); and δ13Corg vs. Δδ (c). Note different slopes and intercepts predicted by these states of the carbon cycle. d–f: Scatter diagrams for carbon isotope data of the Silverton and Sengoma Argillite formations. Note lack of correlation between δ13Ccarb and δ13Corg (d) and between δ13Ccarb vs. Δδ (e) but significant correlation between δ13Corg and Δδ (f).

Precambrian, and similar to that observed today based on the carbon isotope records of carbonates and organic matter (e.g. Schidlowski, 1988; Des Marais et al., 1992). A slightly different treatment of the carbon cycle involving ocean floor carbonatization, suggests that the relative burial rate of organic matter could have gradually increased over time if submarine hydrothermal weathering of the ocean crust

was more important on the early Earth. Nevertheless, even in this model the range of this gradual change is rather limited (Bjerrum and Canfield, 2004). While the endogenic carbon flux to surface environments, and the ratio of organic carbon to inorganic carbon burial fluxes did not change significantly since the Late Archean, DOC and DIC reservoirs could have been larger in the past due to weaker solar

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

luminosity and greenhouse requirements, and less efficient burial of organic carbon (Grotzinger and Kasting, 1993; Kasting, 1993; Rothman et al., 2003; Bartley and Kah, 2004). Under the present conditions, the residence time of DIC is on the order of 105 yr whereas that of DOC is at least several orders of magnitude shorter (Holland,1984). As a result, the smaller DOC reservoir responds faster to environmental perturbations such as changes in pCO2 or in the ocean redox state, whereas the larger DIC reservoir responds much more slowly. This relationship is illustrated schematically in Fig. 8a using a Cenozoic record that reflects a carbon cycle evolving quasistatically in a succession of steady states (Rothman et al., 2003). Note that for the last ca. 50 Ma average carbon isotope values of TOC consistently increased whereas those of carbonates were steady or decreasing. This results in the overall decrease in Δδ values through the Cenozoic (see Fig. 3 in Hayes et al., 1999). Laboratory experiments with phytoplankton show (see Hayes et al., 1999 for review) that as pCO2 increases δ13C values of photosynthetically-produced organic matter become more negative due to an enhanced biological fractionation. The increased isotopic discrimination between fixed organic carbon and DIC causes a corresponding positive shift in carbon isotope values of the seawater DIC pool that is reflected in higher δ13Ccarb values. Based on experiments with phytoplankton cultures, this relationship conflicts with the conventional treatment of the ocean carbon isotope mass balance (e.g. Des Marais et al., 1992; Schidlowski and Aharon, 1992), that assumes Δδ values were invariant over geological time. This assumption allows us to directly relate secular variations in δ13Ccarb values to changes in the relative burial rate of organic carbon over time. The linear relationship between Δδ and δ13Ccarb values was recently emphasized in the model of Rothman et al.'s (2003) for the Neoproterozoic carbon cycling, based on large variations in the carbon isotope fractionation between organic and inorganic carbon since the Neoproterozoic (Hayes et al., 1999). The latter model assumes that carbon isotope variations in carbonates reflect changes in Δδ values driven by environmental factors such as pCO2 or ocean redox changes rather than reflecting relative burial rate of organic carbon since the former changes faster than the latter. Considering that the size and residence times of DIC and DOC reservoirs could have been different in the Precambrian than today, it is useful to discuss two end-members with the surface ocean carbon cycle dominated by either large DIC or large DOC reservoirs (Fig. 8a–c). If the size and residence time of these reservoirs are dramatically different, one of them will be relatively inert to perturbations that would affect its carbon isotope value on a time scale shorter than its residence time, whereas the other with a significantly shorter residence time will respond more quickly. Over time scales that are much longer than the residence times of both reservoirs, their isotopic values will follow a curved path as has been shown for Phanerozoic seawater C and S reservoirs (Kump and Garrels, 1986). These different scenarios are depicted schematically in Fig. 8a–c showing C isotope responses to a perturbation in the carbon cycle whose duration is the intermediate between the DIC and DOC residence times. Using a similar approach, Rothman et al. (2003) interpreted δ13Corg and δ13Ccarb values reconstructed from the Neoproterozoic carbonate successions recording positive carbon isotope excursions as reflecting a dynamic system with a large DOC reservoir. In this case, carbon isotope values of carbonates are highly sensitive to external forcing whereas carbon isotope values of organic matter are less so (see Fig. 8a–c). The opposite would be true in case of a large DIC reservoir. Note that the Cenozoic carbon isotope data reflect an intermediate state of the carbon cycle between these two end-member models. The Silverton–Sengoma carbonate data (Fig. 8d–f) may be evaluated against these models. First, we agree with Rothman et al. (2003) and Bekker et al. (2003b) that the conventional steady-state model, and in particular its reliance on organic carbon burial as a way to change seawater δ13C values, cannot explain the extremely high δ13Ccarb values recorded by the Neoproterozoic and Lomagundi Events. Similarly to the Neoproterozoic carbon cycle, the carbon isotope

289

background for the Lomagundi excursion is +6 to +8‰, but higher carbon isotope values are not uncommon and even +28‰ values were observed in a relatively thick carbonate section deposited in an openmarine setting (Bekker et al., 2003b). The Silverton–Sengoma data plot on, or slightly above, the best-fit line on a plot of δ13Ccarb vs. Δδ for Neoproterozoic carbonate successions recording positive carbon isotope excursions and, arguably, a dynamic state of carbon cycle (Rothman et al., 2003). The Silverton–Sengoma data (Fig. 8d–f) show no correlation between δ13Ccarb and either Δδ or δ13Corg, but there is a significant correlation between δ13Corg and Δδ. The trend is consistent with the carbon cycle dominated by a large DIC reservoir during the Lomagundi Event. Carbonates are indeed abundant in the Paleoproterozoic sedimentary successions deposited during the Lomagundi carbon isotope excursion, and the long duration of the Lomagundi Event (N100 Ma) is consistent with the expected stability of carbon isotope values in carbonates at the time when the DIC reservoir was large. The DOC reservoir may have been larger than now as well, but due to a much larger size of the DIC reservoir, its effect on the carbon cycle is not apparent. How could the carbon cycle remain in a dynamic state regardless of how large the DIC reservoir was for the N100 Ma duration of the Lomagundi Event? We speculate that extensive intracratonic basins and epeiric seas, with restricted connections to a redoxstratified global ocean, developed during the protracted breakup of a Paleoproterozoic supercontinent (Bekker et al., 2006a). The paleogeography of the period, similar to that in the Neoproterozoic, favored high organic carbon burial rates, and this in turn caused a global-scale shift in the carbon cycle to higher δ13C values. Numerical simulations of carbon cycling study the effect of a flux perturbation by applying an instantaneous change from which the system recovers from, or a persistent forcing that drives the system into a new steady state (e.g. Kump and Garrels, 1986; Kump and Arthur, 1999). It is suggested here that under conditions of high organic carbon burial, carbon cycling operates in a dynamic rather than steady state due to perturbations occurring on a time scale intermediate between the residence times of DIC and DOC reservoirs. The carbon cycle could have been shifting from one non-steady dynamic state to another, never reaching a steady state until the breakup of the Paleoproterozoic supercontinent between 2.1 and 2.0 Ga that changed the locus of organic matter deposition to continental margins (Bekker et al., 2003b). In conclusion, our data support the existence of a large DIC reservoir in the oceans during the Lomagundi Event. The DIC reservoir likely decreased gradually as solar luminosity increased over time and, consequently, the carbon cycle shifted to one that was dominated by a large DOC reservoir sometime before the Late Neoproterozoic. At the end of the Precambrian, the DOC reservoir was drastically reduced as deep ocean oxygen levels increased (Logan et al., 1995; Rothman et al., 2003; Canfield et al., 2006; Fike et al., 2006) marking the transition to a modernstyle ocean carbon cycle. The DOC reservoir likely increased episodically in the Phanerozoic at the times when the ocean was stratified and anoxic. Shelf gradients likely waxed and waned under Phanerozic relatively low pCO2 levels as large epeiric seas developed in association with the tectonically-driven sea level changes (e.g. Melchin and Holmden, 2006). 3.5. Implications of Δδ as a diagenetic filter and of δ13Corg for reconstruction of seawater composition A number of workers have relied on carbon isotope values of TOC as a tool for evaluating the degree of post-depositional carbonate alteration and to infer seawater δ13CDIC values in Precambrian siliciclastic successions where carbonates are absent (e.g. Knoll et al., 1986; Kaufman and Knoll, 1995; Calver, 2000; Karlstrom et al., 2000; Walter et al., 2000; Kaufman and Xiao, 2003). While this approach has been generally successfully applied to Phanerozoic carbonate and siliciclastic successions, the results of this study suggest that it can lead to erroneous conclusions when applied to deeper-water deposits in the Precambrian, where the primary photosynthetic signature is

290

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291

susceptible to being overprinted by heterotrophic and chemolithotrophic productivity in redox-stratified oceans. Our study suggests that carbon isotope values of TOC during the Lomagundi Event were strongly influenced by environmental factors changing water column redox conditions and relative contribution of secondary productivity to TOC. As a result, changes in sea level and ocean circulation affecting position of the depositional site with respect to the redoxcline in the water column and upwelling currents may have influenced carbon isotope values of TOC in deeper-water settings to such an extent that they no longer behave as reliable indicators of DIC isotope trends in seawater. Since the ocean was likely redox-stratified during most of the Precambrian, these environmental controls might be equally applicable to other Precambrian successions. This situation is rather different from that in the modern oxidized ocean where δ13C values of TOC are relatively homogeneous and are generally coupled with δ13CDIC values. 4. Conclusions 1. During the Paleoproterozoic Lomagundi carbon isotope excursion, fractionation between carbonate and organic carbon on the shallow-marine carbonate platform was close to 30‰, but in deeper-water suboxic to anoxic and, locally, sulfidic settings it was larger, consistent with redox stratification in the ocean; 2. There is increasing evidence for deposition of organic-rich shales during the Lomagundi excursion in deeper-water settings; 3. The large magnitude of the Lomagundi carbon isotope excursion likely reflects the large isotope fractionation between bulk organic carbon and carbonate carbon, due to contributions from secondary productivity in addition to high relative burial rates of organic carbon. It is difficult at present to separate these two variables on the global scale; 4. Carbonate and organic carbon isotope data suggest that the carbon cycle was in a dynamic state dominated by a large DIC reservoir during the Lomagundi Event, however the DOC reservoir could have also been larger than at present, but its influence was dwarfed by a much larger DIC reservoir; 5. In redox-stratified Precambrian oceans, carbon isotope values of TOC in deeper-water facies were likely strongly influenced by contributions from heterotrophic and chemotrophic productivity and, therefore, estimates of post-depositional carbonate alteration and carbon isotope values of DIC made on basis of carbon isotope values of organic carbon should be cautioned. Acknowledgements This study was supported by the NASA-Ames Research Center, NASA, and PRF/ACS grants to H.D. Holland and, at a later stage, by NSF grant EAR-05-45484, NASA Astrobiology Institute award No. NNA04CC09A, and NSERC Discovery Grant to AB. Read Mapeo (University of Botswana) and El-El Coetzee (University of Johannesburg) guided in the field, Geological Survey of Botswana provided access to drill core material and permitted sampling, Marilyn Fogel provided access to mass-spectrometers and Seth Newsome patiently helped with analyses at the Geophysical Laboratory. Lee Kump and an anonymous reviewer provided valuable comments. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.epsl.2008.04.021. References Baker, A.J., Fallick, A.E., 1989a. Evidence from the Lewisian limestone for isotopically heavy carbon in two-thousand-million-year-old sea water. Nature 337, 352–354.

Baker, A.J., Fallick, A.E., 1989b. Heavy carbon in two-billion-year-old marbles from Lofoten-Vesterålen, Norway: implications for the Precambrian carbon cycle. Geochim. Cosmochim. Acta 53, 1111–1115. Banerjee, D.M., Schidlowski, M., Arneth, J.D., 1986. Genesis of Upper Proterozoic–Cambrian phosphorite deposits of India: isotopic inferences from carbonate fluorapatite, carbonate and organic carbon. Precambrian Res. 33, 239–253. Bartley, J.K., Kah, L.C., 2004. Marine carbon reservoir, Corg–Ccarb coupling, and the evolution of the Proterozoic carbon cycle. Geology 32, 129–132. Baumgartner, L.P., Valley, J.W., 2001. Stable isotope transport and contact metamorphic fluid flow. Rev. Mineral. Geochem. 43, 415–467. Bekker, A., Kaufman, A.J., 2007. Oxidative forcing of global climate change: a biogeochemical record across the oldest Paleoproterozoic ice age in North America. Earth Planet. Sci. Lett. 258, 486–499. Bekker, A., Karhu, J.A., Kaufman, A.J., 2006a. Carbon isotope record for the onset of the Lomagundi carbon isotope excursion in the Great Lakes area, North America. Precambrian Res. 148, 145–180. Bekker, A., Shen, Y., Scott, C., Kocanda, M., Lyons, T., Kenig, F., Anbar, A., Rouxel, O., 2006b. Redox state of the deep ocean during the 2.22–2.1 Ga carbon isotope excursion. EOS Trans. AGU, vol. 87. p. Jt. Assem. Suppl., Abstract U51A-03. Bekker, A., Kaufman, A.J., Karhu, J.A., Eriksson, K.A., 2005. Evidence for Paleoproterozoic cap carbonates in North America. Precambrian Res. 137, 167–206. Bekker, A., Holland, H.D., Wang, P.-L., Rumble III, D., Stein, H.J., Hannah, J.L., Coetzee, L.L., Beukes, N.J., 2004. Dating the rise of atmospheric oxygen. Nature 427, 117–120. Bekker, A., Sial, A.N., Karhu, J.A., Ferrerira, V.P., Noce, C.M., Kaufman, A.J., Romano, A.W., Pimentel, M.M., 2003a. Chemostratigraphy of carbonates from the Minas Supergroup, Quadrilátero Ferrífero (Iron Quadrangle), Brazil: a stratigraphic record of early Proterozoic atmospheric, biogeochemical and climatic change. Am. J. Sci. 303, 865–904. Bekker, A., Karhu, J.A., Eriksson, K.A., Kaufman, A.J., 2003b. Chemostratigraphy of Paleoproterozoic carbonate successions of the Wyoming Craton: tectonic forcing of biogeochemical change? Precambrian Res. 120, 279–325. Bekker, A., Kaufman, A.J., Karhu, J.A., Beukes, N.J., Swart, Q.D., Coetzee, L.L., Eriksson, K.A., 2001. Chemostratigraphy of the Paleoproterozoic Duitschland Formation, South Africa: implications for coupled climate change and carbon cycling. Am. J. Sci. 301, 261–285. Bidigare, R.R., et al., 1997. Consistent fractionation of 13C in nature and laboratory: growth rate effects in some haptophyte algae. Glob. Biogeochem. Cycles 11, 279–292. Bjerrum, C.J., Canfield, D., 2004. New insights into the burial history of organic carbon on the early Earth. Geochem. Geophys. Geosyst. 5, 1–9. Buick, I.S., Maas, R., Gibson, R., 2001. Precise U–Pb titanite age constraints on the emplacement of the Bushveld Complex, South Africa. J. Geol. Soc. (Lond.) 158, 3–6. Buick, I.S., Uken, R., Gibson, R.L., Wallmach, T., 1998. High-δ13C Paleoproterozoic carbonates from the Transvaal Supergroup, South Africa. Geology 26, 875–878. Bumby, A.J., Eriksson, P.G., Van der Merwe, R., 1998. Compressive deformation in the floor rocks to the Bushveld Complex (South Africa): evidence from the Rustenburg Fault Zone. J. Afr. Earth Sci. 27, 307–330. Button, A., 1986. The Transvaal sub-basin of the Transvaal sequence. In: Anhaeusser, C.R., Maske, S. (Eds.), Mineral Deposits of Southern Africa, pp. 811–817. Button, A., 1973. A regional study of the stratigraphy and development of the Transvaal Basin in the eastern and northeastern Transvaal. Ph.D. Thesis, Univ. Witwatersrand, Johannesburg, 352 pp. Calver, C.R., 2000. Isotope stratigraphy of the Ediacarian (Neoproterozoic III) of the Adelaide Rift Complex, Australia, and the overprint of water column stratification. Precambrian Res. 100, 121–150. Canfield, D.E., Poulton, S.W., Narbonne, G.M., 2006. Late-Neoproterozoic deep-ocean oxygenation and the rise of animal life. Science 315, 92–95. Coetzee, L.L., 2001, Genetic stratigraphy of the Paleoproterozoic Pretoria Group in the Western Transvaal [MSc. thesis], Rand Afrikaans University. Coetzee, L.L., Beukes, N.J., Gutzmer, J., Kakegawa, T., 2006. Links of organic carbon cycling and burial to depositional depth gradients and establishment of a snowball Earth at 2.3 Ga. Evidence from the Timeball Hill Formation, Transvaal Supergroup, South Africa. S. Afr. J. Geol. 109, 109–122. Cornell, D.H., Schütte, S.S., Eglinton, B.L., 1996. The Ongeluk basaltic andesite formation in Griqualand West, South Africa: submarine alteration in a 2222 Ma Proterozoic sea. Precambrian Res. 79, 101–123. Council for Geoscience, 1997, 1: 1,000,000 scale geological map of the Republic of South Africa and the Kingdoms of Lesotho and Swaziland. Council for Geoscience, Pretoria, South Africa. Des Marais, D.J., Strauss, H., Summons, R.E., Hayes, J.M., 1992. Carbon isotope evidence for the stepwise oxidation of the Proterozoic environment. Nature 359, 605–609. Dorland, H.C., 2004. Provenance ages and timing of sedimentation of selected Neoarchean and Paleoproterozoic successions on the Kaapvaal Craton. Ph.D. Thesis, Rand Afrikaans University, Johannesburg, South Africa, 326 pp. Eichmann, R., Schidlowski, M.,1975. Isotopic fractionation between coexisting organic carboncarbonate pairs in Precambrian sediments. Geochim. Cosmochim. Acta 39, 585–595. Fike, D.A., Grotzinger, J.P., Pratt, L.M., Summons, R.E., 2006. Oxidation of the Ediacaran Ocean. Nature 444, 744–747. Fry, B., Jannasch, H.W., Molyneaux, S.J., Wirsen, C.O., Muramoto, J.A., King, S., 1991. Stable isotope studies of the carbon, nitrogen and sulfur cycles in the Black Sea and Cariaco Trench. Deep-Sea Res. 38 (Suppl. 2), S1003–S1019. Gauthier-Lafaye, F., 2006. Time constraint for the occurrence of uranium deposits and natural nuclear fission reactors in the Paleoproterozoic Franceville Basin (Gabon). In: Kesler, S.E., Ohmoto, H. (Eds.), Evolution of Early Earth's Atmosphere, Hydrosphere, and Biosphere — Constraints from Ore Deposits, vol. 198. Geological Society of America, pp. 157–167. Memoir. Gauthier-Lafaye, F., Weber, F., 1989. The Francevillian (Lower Proterozoic) uranium ore deposits of Gabon. Econ. Geol. 84, 2267–2285.

A. Bekker et al. / Earth and Planetary Science Letters 271 (2008) 278–291 Gauthier-Lafaye, F., Weber, F., 2003. Natural nuclear fission reactors: time constraints for occurrence, and their relation to uranium and manganese deposits and the evolution of the atmosphere. Precambrian Res. 120, 81–100. Grotzinger, J.P., 1994. Trends in Precambrian carbonate sediments and their implication for understanding evolution. In: Bengston, S. (Ed.), Early life on Earth. Nobel Symposium, 84. Columbia University Press, New York, pp. 245–258. Grotzinger, J.P., Kasting, J.F., 1993. New constraints on Precambrian ocean composition. J. Geol. 101, 235–243. Haddon, I.G., comp., 2001, Sub-Kalahari geological map: Pretoria, South Africa Council for Geoscience, scale 1:2,500,000. Hannah, J.L., Bekker, A., Stein, H.J., Markey, R.J., Holland, H.D., 2004. Primitive Os and 2316 Ma age for marine shale: implications for Paleoproterozoic glacial events and the rise of atmospheric oxygen. Earth Planet. Sci. Lett. 225, 43–52. Hartzer, F.J., 2000. Geology of Transvaal Inliers in the Bushveld Complex, Council for Geoscience. Geological Survey of South Africa. Memoir 88, 222 pp. Hartzer, F.J., 1995. Transvaal Supergroup inliers: geology, tectonic development and relationship with the Bushveld complex, South Africa. J. Afr. Earth Sci. 21, 521–547. Hayes, J.M., Waldbauer, J.R., 2006. The carbon cycle and associated redox processes through time. Philos. Trans. R. Soc. Lond., B 361, 931–950. Hayes, J.M., Strauss, H., Kaufman, A.J., 1999. The abundance of 13C in marine organic matter and isotopic fractionation in the global biogeochemical cycle of carbon during the past 800 Ma. Chem. Geol. 161, 103–125. Hayes, J.M., Kaplan, I.R., Wedeking, K.W., 1983. Precambrian organic geochemistry, preservation of the record. In: Schopf, J.W. (Ed.), The Earth's Earliest Biosphere: Its Origin and Evolution. Princeton University Press, Princeton, N.J., pp. 93–134. Hedges, J.I., Keil, R.G., 1995. Sedimentary organic matter preservation: an assessment and speculative synthesis. Mar. Chem. 49, 81–115. Holland, H.D., 1984. The Chemical Evolution of the Atmosphere and Ocean. Princeton University Press, Princeton, N.J.. 582 pp. Hotinski, R.M., Kump, L.R., Arthur, M.A., 2004. The effectiveness of the Paleoproterozoic biological pump: a δ13C gradient from platform carbonates of the Pethei Group (Great Slave Lake Supergroup, NWT). Geol. Soc. Amer. Bull. 116, 539–554. Kah, L.C., Knoll, A.H.,1996. Microbenthic distribution of Proterozoic tidal flats: environmental and taphonomic considerations. Geology 24, 79–82. Karhu, J.A., Holland, H.D., 1996. Carbon isotopes and the rise of atmospheric oxygen. Geology 24, 867–870. Karhu, J.A., 1993. Paleoproterozoic evolution of the carbon isotope ratios of sedimentary carbonates in the Fennoscandian Shield: Espoo. Geol. Surv. Finland Bull. 371 87 p. Karl, D.M., Knauer, G.A., 1991. Microbial production and particle flux in the upper 350 m of the Black Sea. Deep-Sea Res. 38 (Suppl. 2), S921–S942. Karlstrom, K.E., Bowring, S.A., Dehler, C.M., Knoll, A.H., Porter, S.M., Des Marais, D.J., Weil, A.B., Sharp, Z.D., Geissman, J.W., Elrick, M.B., Timmons, J.M., Crossey, L.J., Davidek, K.L., 2000. Chuar Group of the Grand Canyon: record of breakup of Rodinia, associated change in the global carbon cycle, and ecosystem expansion by 740 Ma. Geology 28, 619–622. Kasting, J.F., 1993. Earth's early atmosphere. Science 259, 920–926. Kaufman, A.J., Knoll, A.H., 1995. Neoproterozoic variations in the C-isotopic composition of seawater: stratigraphic and biogeochemical implications. Precambrian Res. 73, 27–49. Kaufman, A.J., Jacobsen, S.B., Knoll, A.H., 1993. The Vendian record of Sr and C isotopic variations in seawater: implications for tectonics and paleoclimate. Earth Planet. Sci. Lett. 120, 409–430. Kaufman, A.J., Xiao, S., 2003. High CO2 levels in the Proterozoic atmosphere estimated from analyses of individual microfossils. Nature 425, 279–282. Key, R.M. and Ayres, N., 2000, Botswana Geology Map, Journal of African Earth Sciences, 30, CD-ROM Supplement. Key, R.M., 1983. The geology of the area around Gaborone and Lobatse, Kweneng, Kgatleng, Southern and South East Districts. District Memoir 5. Geological Survey of Botswana, Gaborone. 230 pp. Knoll, A.H., Hayes, J.M., Kaufman, A.J., Swett, K., Lambert, I.B., 1986. Secular variations in carbon isotope ratios from Upper Proterozoic successions of Svalbard and East Greenland. Nature 321, 832–838. Kroopnick, P.M., 1985. The distribution of 13C of ΣCO2 in the world oceans. Deep-Sea Res. 32, 57–84. Kump, L.R., Arthur, M.A., 1999. Interpreting carbon-isotope excursions: carbonates and organic matter. Chem. Geol. 161, 181–198. Kump, L.R., Garrels, R.M., 1986. Modeling atmospheric O2 in the global sedimentary redox cycle. Am. J. Sci. 286, 337–360. Lazar, B., Erez, J., 1990. Extreme 13C depletions in seawater-derived brines and their implications for the past geochemical carbon cycle. Geology 18, 1191–1194. Logan, G.A., Calver, C.R., Gorjan, P., Summons, R.E., Hayes, J.M., Walter, M.A., 1999. Terminal Proterozoic mid-shelf benthic microbial mats in the Centralian Superbasin and their environmental significance. Geochim. Cosmochim. Acta 63 (9),1345–1358. Logan, G.A., Hayes, J.M., Hieshima, G.B., Summons, R.E., 1995. Terminal Proterozoic reorganization of biogeochemical cycles. Nature 376, 53–56. Master, S., Verhagen, B.T., Bassot, J.P., Beukes, N.J., Lemoine, S., 1993. Stable isotopic signatures of Paleoproterozoic carbonate rocks from Guinea, Sinegal, South Africa and Zimbabwe: constrains on the timing of the ca. 2 Ga “Lomagundi” δ13C excursion,

291

Symposium: Early Proterozoic geochemical and structural constraints — metallogeny. Publication Occasionnelle 1993/23, 38-41: Dacar, Sénégal. Melchin, M.J., Holmden, C., 2006. Carbon isotope chemostratigraphy in Arctic Canada: sea-level forcing of carbonate platform weathering and implications for Hirnantian global correlation. Palaeogeogr. Palaeoclimatol. Palaeoecol. 234, 186–200. Melezhik, V.A., Huhma, H., Condon, D.J., Fallick, A.E., Whitehouse, M.J., 2007. Temporal constraints on the Paleoproterozoic Lomagundi–Jatuli carbon isotope event. Geology 35, 655–658. Melezhik, V.A., Fallick, A.E., Hanski, E.J., Kump, L.R., Lepland, A., Prave, A.R., Strauss, H., 2005. Emergence of the aerobic biosphere during the Archean–Proterozoic transition: challenges of future research. GSA Today 15, 4–11. Melezhik, V.A., Fallick, A.E., Medvedev, P.V., Makarikhin, V.V., 1999. Extreme δ13Ccarb enrichment in ca. 2.0 Ga magnesite–stromatolite–dolomite–‘red beds’ association in a global context: a case for the world-wide signal enhanced by a local environment. Earth Sci. Rev. 48, 71–120. Melezhik, V.A., Fallick, A.E., 1996. A widespread positive δ13Ccarb anomaly at around 2.33–2.06 Ga on the Fennoscandian Shield: a paradox? Terra Nova 8, 141–157. Panchuk, K.M., Holmden, C., Kump, L.R., 2005. Sensitivity of the epeiric sea carbon isotope record to local-scale carbon cycle processes: tales from the Mohawkian Sea. Palaeogeogr. Palaeoclimatol. Palaeoecol. 228, 320–337. Rasmussen, B., Fletcher, I.R., Muhling, J.R., Mueller, A.G., Hall, G.C., 2007. Bushveld-aged fluid flow, peak metamorphism, and gold mobilization in the Witwatersrand basin, South Africa: constraints from in situ SHRIMP U–Pb dating of monazite and xenotime. Geology 35, 931–934. Rothman, D.H., Hayes, J.M., Summons, R.E., 2003. Dynamics of the Neoproterozoic carbon cycle. Proc. Nat. Acad. Sci. 100, 8124–8129. Schidlowski, M., 1988. A 3,800-million-year isotopic record of life from carbon in sedimentary rocks. Nature 333, 313–318. Schidlowski, M., Aharon, P., 1992. Carbon cycle and carbon isotope record: geochemical impact of life over 3.8 Ga of Earth history. In: Schidlowski, M., Golubic, S., Kimberley, M.M., McKirdy, D.M., Trudinger, P.A. (Eds.), Early organic evolution: Implications for mineral and energy resources. Springer-Verlag, Berlin, pp. 147–175. Schidlowski, M., Eichmann, R., Junge, C.E., 1976. Carbon isotope geochemistry of the Precambrian Lomagundi carbonate province, Rhodesia. Geochim. Cosmochim. Acta 40, 449–455. Schidlowski, M., Eichmann, R., Junge, C.E., 1975. Precambrian Sedimentary Carbonates: carbon and oxygen isotope geochemistry and implications for the terrestrial oxygen budget. Precambrian Res. 2, 1–69. Schouten, S., Strous, M., Kuypers, M.M.M., Rijpstra, I.C., Baas, M., Schubert, C.J., Jetten, M.S.M., Sinninghe Damste, J.S., 2004. Stable carbon isotopic fractionations associated with inorganic carbon fixation by anaerobic ammonium-oxidizing bacteria. Appl. Environ. Microbiol. 70, 3785–3788. Schreiber, U.M., Eriksson, P.G., van der Neut, M., Snyman, C.P., 1992. Sedimentary petrology of the Early Proterozoic Pretoria Group, Transvaal Sequence, South Africa: implications for tectonic setting. Sediment. Geol. 81, 89–103. Scott, C., Lyons, T.W., Bekker, A., Shen, Y., Poulton, S., Chu, X., Anbar, A., 2008. Tracing the stepwise oxygenation of the Proterozoic biosphere. Nature 452, 456–459. Shen, Y., Bekker, A., 2006. Paleoproterozoic ocean chemistry: evidence from the Sengoma Argillite Formation (~ 2.2 Ga) in Botswana. Astrobiology 6, 166–167. Slack, J.F., Grenne, T., Bekker, A., Rouxel, O.J., Lindberg, P.A., 2007. Suboxic deep seawater in the late Paleoproterozoic: evidence from hematitic chert and iron formation related to seafloor-hydrothermal sulfide deposits, central Arizona, USA. Earth Planet. Sci. Lett. 255, 243–256. Stiller, M., Rounick, J.S., Shasha, S., 1985. Extreme carbon-isotope enrichments in evaporating brines. Nature 316, 434–435. Swart, P.K., Eberli, G., 2005. The nature of the δ13C of periplatform sediments: implications for stratigraphy and the global carbon cycle. Sediment. Geol. 175, 115–129. Swart, Q.D., 1999, Carbonate rocks of the Paleoproterozoic Pretoria and Postmasburg Groups, Transvaal Supergroup [M.S. thesis]: Johannesburg, South Africa, Rand Africaans University. Valley, J.W., 2001. Stable isotope thermometry at high temperatures. Rev. Miner. Geochem. 43, 365–413. Valley, J.W., 1986. Stable isotope geochemistry of metamorphic rocks. Rev. Miner. 16, 445–489. Veizer, J., Plumb, K.A., Clayton, R.N., Hinton, R.W., Grotzinger, J.P., 1992. Geochemistry of Precambrian Carbonates: V. Late Paleoproterozoic Seawater. Geochim. Cosmochim. Acta 56, 2487–2501. Walraven, F.,1997. Geochronology of the Rooiberg Group, Transvaal Supergroup, South Africa. Inf. Circ. 316. Economic Geology Research Unit, University of the Witwatersrand, Johannesburg, South Africa, 21 pp. Walter, M.R., Veevers, J.J., Calver, C.R., Gorjan, P., Hill, A.C., 2000. Dating the 840–544 Ma Neoproterozoic interval by isotopes of strontium, carbon, and sulfur in seawater, and some interpretative models. Precambrian Res. 100, 371–433. Watanabe, Y., Naraoka, H., Wronkiewicz, D.J., Condie, K.C., Ohmoto, H., 1997. Carbon, nitrogen and sulfur geochemistry of Archean and Proterozoic shales from the Kaapvaal Craton, South Africa. Geochim. Cosmochim. Acta 61, 3441–3459.