Early life signatures in sulfur and carbon isotopes from Isua, Barberton, Wabigoon (Steep Rock), and Belingwe Greenstone Belts (3.8 to 2.7 Ga)
Published:January 01, 2006
- PDF LinkChapter PDF
N.V. Grassineau, P. Abell, P.W.U. Appel, D. Lowry, E.G. Nisbet, 2006. "Early life signatures in sulfur and carbon isotopes from Isua, Barberton, Wabigoon (Steep Rock), and Belingwe Greenstone Belts (3.8 to 2.7 Ga)", Evolution of Early Earth's Atmosphere, Hydrosphere, and Biosphere - Constraints from Ore Deposits, Stephen E. Kesler, Hiroshi Ohmoto
Download citation file:
Carbon and sulfur isotopes have been measured on samples from four Archean greenstone belts dating from 3.8 Ga to 2.7 Ga, in order to trace metabolic changes as life evolved over this one-billion-year period. In the Isua Greenstone Belt (3.8 Ga), Greenland, δ34S in sulfide minerals from sedimentary sequences range from −3.8‰ to +3.4‰. δ13Cred measured in BIFs, turbidites and conglomerates vary from −29.6‰ to −14.7‰; this range permits us to hypothesize the presence of hyperthermophilic and chemotrophic species in transient settings, or possibly pelagic photoautotrophic microbes, or both. In the Barberton Greenstone Belt, South Africa, sulfide minerals show δ34S values from +1.5‰ to +5.6‰. Black shales have δ13Cred values from −32.4‰ to −5.7‰, suggesting that oxygenic photosynthetic and sulfate-reducing bacteria were present by ca. 3.24 Ga. The δ13Cred measured in the stromatolites of Steep Rock Lake (3.0 Ga), Ontario, Canada, are from −30.6‰ to −21.6‰, giving clear evidence for occupation of a shallow water environment by cyanobacteria. The wide isotopic ranges for δ34S in sulfides from −21.1‰ to +16.7‰ and for δ13Cred in carbon-rich cherts and black shales from −43.4‰ to −7.2‰ in the Belingwe Greenstone Belt, Zimbabwe, indicate that photosynthetic microbial mat communities were well established at 2.7 Ga. In these well-preserved Late Archean formations, modern-style biological sulfur and carbon cycles may have been in operation. The δ34S and δ13Cred ranges, respectively 37‰ and 36‰, indicate a great variety of biological processes interacting with each other.
INTRODUCTION: USE OF STABLE ISOTOPES IN STUDYING EARLY LIFE
The best record of early life is in sedimentary rocks from Archean greenstone belts. This study uses isotopic analyses of carbon- and sulfide-rich deposits from four Archean belts to seek evidence for the onset of major metabolic processes. The Isua Greenstone Belt (ca. 3.8 Ga) in West Greenland and the Belingwe Greenstone Belt (2.7 Ga) in Zimbabwe are the two main areas studied. Additional δ13C and δ34S data were obtained from the Barberton Greenstone Belt in South Africa and Steep Rock Lake succession in northwestern Ontario, Canada.
Metabolic processes produce distinctive isotopic fractionations when selecting carbon and sulfur from chemical or organic substrates accessible to the organisms. The recognizable δ34S and δ13C signatures in the sediments after burial, both in residual organic matter and in associated minerals, are particularly useful in efforts to understand the nature and extent of microbial activity (e.g., Nisbet and Fowler, 1999; Grassineau et al., 2001a, 2002). These isotopic fingerprints are among the very few biological signals remaining in Archean rocks. Other evidence for early microbial phylogeny comes from molecular studies (Woese, 1987; Pace, 1997; review in Nisbet and Sleep, 2001). In this study, stable isotopes have been used to infer the presence of prokaryotic communities in relics of microbial mats from Archean rocks.
Carbon-bearing compounds in the modern earth system have very different δ13C values (Table 1). Biological fractionations produce wide δ13C ranges, with two distinctive signatures dominating modern organic carbon. The first, from −28‰ to −22‰, is due to the fractionation by the Rubisco I enzyme during photosynthesis (Sirevåg et al., 1977; Pierson, 1994), in aerobic and microaerobic environments. This characteristic carbon isotopic fractionation occurs when carbon supply is abundant (Erez et al., 1998) and taken from seawater in exchange with a CO2-rich atmosphere. The second is for methanogenic archaea that produce metabolic methane cycling, generating more fractionated δ13C, with a range of values mainly between −40‰ and −30‰. Methanotrophs or sulfate reducers that gain organic matter from this methane produce δ13C down to −80‰ (Coleman et al., 1981). An additional process is anoxygenic photosynthesis with fractionations that are usually smaller (as Chloroflexus: −20‰ to −10‰ [e.g., van der Meer et al., 2001]; or Rubisco II [e.g., Robinson et al., 2003; Schidlowski, 2001]). Conversely, the inorganic carbon byproduct of the photosynthesis has a narrow range represented by most Phanerozoic marine carbonates (Table 1).
In the Archean, carbon degassing from the mantle entered the atmosphere and thence the ocean. Carbon extracted by sedimentation from the Late Archean ocean/atmosphere system was fractionated between a relatively small 13C-depleted organic carbon portion (δ13C ∼ −28‰), and a larger residual reservoir of 13C-enriched inorganic carbon (δ13C ∼ 0‰) (e.g., Schidlowski et al., 1975; Abell et al., 1985a, 1985b). Studies of Archean biology (e.g., Hayes et al., 1983; Schidlowski et al., 1983) have recognized this distinct carbon isotopic fractionation as marking the onset of the global-scale processing of carbon by cyanobacteria, now known to be via Rubisco I. The rapid and successful production of cyanobacteria then induced the start of carbon management by oxygenic photosynthesis that may have been sudden in geological terms. As O2 is the process by-product, aerobic or at least microaerobic environments of deposition would be expected in rocks that showed the characteristic isotopic signature. Although an aerobic facies, organic δ13C signature, or δ13Ccarb signature of 0‰ may not be evidence of cyanobacteria individually, the presence of all three is diagnostic. The start date of photosynthetic processes in the Archean is very controversial, and more isotopic investigation is needed.
Anoxygenic photosynthesis predates the oxygenic process in standard phylogenetic models (see discussion in Nisbet and Sleep, 2001). Therefore, there may be an interval in the geological record where the δ13C fractionation between organic matter and carbonate was less than −28‰. However, it is still in debate whether large-scale carbonate deposition occurred before oxygenic photosynthesis. Most geologists concur that methanogenesis dates back to the Archean (Hayes, 1994; Nisbet and Sleep, 2001), and may well predate photosynthesis, though Cavalier-Smith (2002) has argued that methanogens did not appear until the late Proterozoic and hence a very “light methane” signature would not be expected. Thus the “geological” consensus that methanogenesis is ancient needs testing. Primary δ13C of marine carbonate and organic carbon in the Archean are still not well constrained, but the wide δ13Cred range so far obtained (−52‰ to −13‰; Strauss and Moore, 1992) suggests that biological processes were operational. A key target is to seek evidence for oxygenic photosynthesis (δ13Ccarb ∼ 0‰; δ13Cred ∼ −28‰) or anaerobic conditions within the Archean.
Two isotopically distinct primary sulfur reservoirs dominate today: homogeneous seawater sulfate, and sulfide derived from the mantle (Table 1). δ34S of sulfides in modern sediments ranges from −60‰ to +20‰ (Hoefs, 1997). Significant isotopic fractionations can be caused by inorganic processes, especially in hydrothermal fluids, but biological activities produce even greater fractionation (e.g., Schidlowski et al., 1983), particularly in repetitive recycled reactions (e.g., Canfield and Teske, 1996). The resulting fractionation between seawater sulfate and precipitated biogenic sulfides in sediments has been up to 80‰ since the beginning of the Phanerozoic, generating 34S-enrichment in seawater sulfate, which has averaged between +10‰ to +30‰ at different periods of Earth's history.
δ34S of Archean seawater sulfate remains unknown, as sulfate is very rare in most early sediments, but is estimated to be close to 0‰–2‰ (Ohmoto, 1992). Then oceans contained modest amounts of sulfate (e.g., Veizer et al., 1989; Canfield and Teske, 1996), and Habicht and Canfield (1996) suggested that high SO4 concentrations in seawater did not occur before the rise of atmospheric oxygen in post-Archean time. On the other hand, Ohmoto et al. (1993) interpreted the 9‰ δ34S range for 3.4–3.2 Ga pyrite in one specimen (South Africa) as evidence for microbial reduction of seawater sulfate, suggesting that seawater already had a high SO4 content. Shen et al. (2001) and Shen and Buick (2004) reached the same conclusion with values of +3‰ to +9‰ in 3.47 Ga sulfate samples from northwestern Australia. Thus there is currently no consensus.
The evolution of the sulfur cycle is also disputed, as there are few constraints on the timing of the first appearance of key parts of the cycle. Biochemical evidence from the “standard” model of microbial phylogenetic evolution (Woese, 1987; Pace, 1997) indicates strongly, though circumstantially, the great antiquity of both S-oxidation and S-reduction processes. Many biochemical processes rely on a S-containing enzyme, and thus at least some S-utilizing bacteria are Archean (e.g., Londry and Des Marais, 2003). Ohmoto and Felder (1987) suggested that microbial sulfate reduction was established by 3.5 Ga, and further examples have confirmed this (Barberton, South Africa, at 3.4 Ga [Ohmoto et al., 1993; Kakegawa and Ohmoto, 1999]; North Pole, northwestern Australia, at 3.47 Ga [Shen and Buick, 2004]). However, some authors consider that though sulfate reduction may be old, some of the sulfur cycle evolved only in the Proterozoic (Blank, 2004), and that the complete cycle with important S-isotope fractionation only appeared ca. 0.86–1.0 Ga (Canfield and Teske, 1996). Hitherto, the δ34S range of Archean sulfide minerals was considered to be less than 10‰ (Cameron, 1982; Habicht and Canfield, 1996), and up to 13‰ in mineral deposits (Ohmoto, 1992). However, Grassineau et al. (2001a, 2002) reported a wider range of 37‰ at 2.7 Ga, suggesting that the sulfur cycle was already well on its way to full operation in the Late Archean at 2.7 Ga, with evidence of sulfate-reduction, sulfur-oxidation, and possibly disproportionation. Furthermore Shen et al. (2001) found a range of 16‰ at 3.47 Ga. This demonstrates that the sulfur cycle is of great antiquity (Grassineau et al., 2001a).
δ13C for reduced carbon and δ34S were analyzed with a VG/Fisons/Micromass “Isochrom-EA” system, consisting of an elemental analyzer (EA1500 Series 2) online to an Optima mass spectrometer operating in He continuous flow mode (Matthews and Hayes, 1978, for carbon; Grassineau et al., 2001b, for sulfur). This high-resolution technique measures local isotopic fractionations in samples at sub-millimeter scale, which is necessary to detect specific activities of the biological communities, otherwise undetectable in larger samples.
A precision of ± 0.1‰ for δ34S was obtained on hand-picked sulfide minerals as small as 0.8 mg for pyrite. The six standards analyzed, including NBS123, NBS127, and IAEA-S3, cover a range from −31.6‰ to +20.3‰. The samples for reduced carbon analysis were first treated in 20% HCl at 120 °C for 12 h. A precision better than ± 0.1‰ in δ13C was obtained on hand-picked samples of 0.07 mg for pure carbon, to 30 mg for whole rock with 0.1 wt% C. The standards measured, including NBS21 and IAEA-CO9, cover a range from −47.1‰ to +3.3‰. Blank contamination from tin capsules is <34 ppm C, as measured in the laboratory, but all samples with less than 200 ppm C, where the blank isotopic effect might be significant for interpreting the results, have been rejected from the data set.
Pure carbonate samples (0.5 mg) were measured using an Isocarb automated carousel connected to a PRISM mass spectrometer. Impure carbonates were analyzed using a modified Micromass Multiflow connected to an Isoprime mass spectrometer. The system requires as little as 200 µg of pure carbonate or up to 50 mg of whole rock powder with 0.5% carbonate. Internal precision is better than ±0.07‰ for δ13C and ±0.10‰ for δ18O for both systems. The standards used are NBS19 limestone and an internal laboratory calcite.
EARLY ARCHEAN: THE ISUA GREENSTONE BELT, WEST GREENLAND
The Isua Greenstone belt (IGB), West Greenland, is a volcano-sedimentary relic exposed in an arcuate belt surrounded and locally intruded by a variety of tonalitic gneisses. The belt is dominated by thick sequences of mafic pillow lavas intercalated with numerous beds of iron formations (Appel et al., 1998; Fedo et al., 2001; Myers, 2001; Polat et al., 2003). Its minimum age constraint of 3.7 Ga comes from U-Pb of zircons in granitic sheets crosscutting the belt (Nutman et al., 1997). Sulfides in the IGB are mainly pyrrhotite and pyrite. They occur mainly as thin stratabound layers and disseminations throughout the rocks. Pyrite is most abundant in BIF and metacherts, often in association with fuchsite. Pyrrhotite dominates in pillow lavas, often associated with chalcopyrite in discordant quartz veins and veinlets (Appel, 1979). The IGB furthermore hosts sedimentary sequences, comprising mica schists or garnet mica schists, with or without staurolite as well as chemical sediments such as chert and iron formation. Rollinson (2002) observed five different metamorphic and structural domains, based on garnet study (Fig. 1). The southwestern zone has undergone two main metamorphic events (Rollinson, 2002) at 3.74 Ga (Frei and Rosing, 2001: Pb/Pb on metabasalts) and 2.84 Ga (Frei et al., 1999: Pb/Pb on magnetite). The latest episode was at high metamorphic grade with widespread metasomatic overprinting (Rose et al., 1996; Frei et al., 1999, Rollinson, 2002). Conversely, the northeast zone has experienced only one event of lower intensity, which took place at 3.69 Ga (Frei et al., 1999 [Pb-Pb on magnetite and tremolite]) and no deformation thereafter (Moorbath and Kamber, 1998). The latter authors suggest that this age might represent the depositional period. This low-strain area has well-preserved primary sedimentary and igneous features (Appel et al., 1998; Rollinson, 2002). Several carbonate alteration events took place in the belt. The earliest occurred during the deposition of the sediments and extrusion of lava flows (J.S. Myers, personal commun.). Later carbonate alteration especially affected the western part of the IGB from 3.7 to 2.8 Ga (Moorbath and Whitehouse, 1996; Rose et al., 1996; Frei and Rosing, 2001). The first author carried out fieldwork in the IGB in 1998 and 1999, as part of the Isua Multidisciplinary Research Project.
Sedimentary Formations Sampled
Sampling of the sedimentary rocks of the Isua Greenstone Belt focused on the regions that had the lowest metamorphic grade and were affected only by the early 3.74–3.69 Ga event, as these are most likely to preserve evidence of biological activity. Such areas are much rarer in the southwest than in the northeast (Rollinson, 2002). Eight formations were sampled in this study (Fig. 1). Six of the sampled units are banded-iron formations (BIF), three from the western and three from the northeastern part of the belt. The northeastern BIF samples are actinolite-bearing metachert and iron formation consisting of magnetite-rich bands alternating with grunerite-rich bands, or magnetite-rich bands alternating with quartz-bands. The largest example is the Iron Mountain Formation, a chert-BIF of alternating layers of quartz and magnetite with local amphibole-rich bands (Frei et al., 1999; Myers, 2001). The BIF samples from the west, including one of carbonate facies, are far more strongly deformed by higher-grade metamorphism than those from the northeast.
Another group of samples are metasedimentary rocks from the northwest that were formed by turbidity currents (Bouma sequence; Nutman, 1986). They consist of a series of medium-grained quartzites grading into fine-grained metapelites containing numerous small carbon particles (Rosing, 1999). In addition, a metaconglomerate was sampled in the northeast. The poorly sorted polymict assemblage consists mainly of quartz pebbles (metachert) with few pebbles of amygdaloidal volcanic rocks and BIF clasts (Fedo, 2000).
Previous Stable Isotope Studies at Isua
Monster et al. (1979) found a narrow range of −1.0‰ to +2.6‰ for sulfides in the IGB. More recently, Strauss (2003) gives a variation from −3.0‰ to +1.0‰, and Mojzsis et al. (2003) reported values of −0.9‰ to +2.2‰ on two pyrite grains (Fig. 2). These values around 0‰ are in the range of magmatic hydrothermal sulfides and provide no positive evidence that sulfur-utilizing bacteria were active during deposition.
Previous studies have obtained a wide range for δ13Cred (reduced carbon) from −28‰ to −6‰ (Fig. 3) (Perry and Ahmad, 1977; Schidlowski et al., 1979; Hayes et al., 1983; Naraoka et al. 1996; Rosing, 1999; Ueno et al., 2002; van Zuilen et al., 2002), suggesting biological fractionation. For example, Rosing (1999) proposed a biogenic origin for the graphite grains in the Bouma sequence, with δ13Cred of −19‰ close to the original depositional isotopic signature.
New Results Obtained for the Isua Belt
Sulfur Isotope Results and Discussion
The sulfide minerals analyzed from the sedimentary sequences (Table 2, Figs. 2A, 2B) are mainly disseminated pyrite, with only a few secondary pyrrhotites from the carbonate facies BIF. The δ34S obtained for the pyrites are from −3.8‰ to +3.4‰ (n = 19) (Grassineau et al., 2000). This 7.2‰ range is larger than previously measured in Isua. The extent of the range found in this study suggests that more diverse processes operated than hitherto thought. This result is particularly significant because the largest range of values is seen in the northeastern sulfide-rich BIF and Iron Mountain, which have the lowest metamorphic grade in the belt (Rollinson, 2002).
The δ34S has different ranges for each BIF. In the Iron Mountain, the values are only positive (+3.1‰ to +3.4‰; n = 3). In contrast, the two other northeastern BIFs show mainly negative values (−3.8‰ to +1.1‰; n = 9). The metaconglomerate with an average of +0.4 ± 0.5‰ (n = 4) is very homogeneous and close to 0‰, similar to the values of Monster et al. (1979). The only western BIF analyzed for δ34S gives an average of −0.7 ± 0.1‰ (n = 3).
In order to provide a basis for comparison, pyrites from igneous formations and a slightly later Tarssartôq dyke (3.47 Ga) have been analyzed. The δ34S ranges obtained are between −0.5 and +1.0‰ (n = 9) and would suggest that high-temperature fluids circulating in the belt had a signature of +0.4 ± 0.5‰ (Fig. 2C). This interpretation is supported by pyrite from metabasalts that show a narrow range of +0.5 ± 0.1‰ (n = 2).
Further comparison has been made with sulfides from two 3.7–3.8 Ga mineral deposits hosted by the western area of the IGB (Pb-Pb on galena; Frei and Rosing, 2001). These formed from low to moderate temperature hydrothermal fluids (Figs. 2D, 2E). δ34S from disseminated galena and sphalerite associated with gold mineralization in a tonalite sheet varies from −5.4‰ to +0.6‰ (n = 25) (Grassineau and Appel, 2000). More remarkably, the range of −10.0‰ to +3.0‰ (n = 21) obtained for disseminated sulfides associated with quartz-fuchsite deposits is the widest yet found in Isua. Such wide isotopic fractionation is most likely created by sulfate-sulfide reactions (Ohmoto, 1992).
The carbonate-facies BIF is dominated by siderite, with secondary pyrrhotite in veins and along fractures. The low to moderate oxygen fugacity and high pH, constrained by the presence of magnetite, kept the isotopic variation relatively small for the hydrothermal fluids generated during metamorphism (Ohmoto, 1986). If there was any δ34S signature left by organisms then it has been overprinted by the 2.84 Ga metamorphic event and re-homogenized as pyrrhotite. This is shown by the narrow range of values. Two main primary sources for sulfur can be suggested: the meta–pillow lavas and sulfur derived from seawater during the deposition of the BIF on the seafloor. The homogenized δ34S value of −0.7‰ obtained for the BIF is lighter than the magmatic sources (+0.5‰). Although the shift in values of 1.2‰ could be explained entirely by hydrothermal processes remobilizing magmatic sulfur, it is possible that another source caused this shift during re-homogenization, possibly primary sulfur in the BIF.
In contrast, the better preserved and less deformed metasediments in the northeast are more likely to preserve the original δ34S signatures. They show a larger δ34S range of 7.2‰ (n = 26), but values are different for each formation. In the polymict metaconglomerate, δ34S is homogenous (+0.4 ± 0.5‰; n = 4). The pyrites are from the matrix and the clasts and show similar δ34S. This is expected, as the clasts consist of mafic volcanic rocks and metacherts (Fedo, 2000). The detrital pyrites probably originated as magmatic sulfur, at ∼0‰. There is no sign of a biological signature, which should have given more isotopic variation.
The three samples analyzed from the Iron Mountain metachert are very homogeneous (δ34S of +3.2 ± 0.2‰). These δ34S values are heavier than those of the five other BIFs. The pyrites are idiomorphic and located close to or in veins, indicating that the secondary sulfides were introduced by hydrothermal fluids that may have been produced by the early metamorphism. The sulfur in these fluids could have originated from different sources: (1) sulfur already present in situ in the form of barite (barite traces exist in the metachert) reduced by interaction with hydrothermal fluids (e.g., Ohmoto and Godhaber, 1997); 2) sulfur from volcanic formations in the belt. This view is supported by Lepland et al. (2002) who interpreted REE patterns for apatites in the Iron Mountain as a result of a pervasive fluid from mixed sources.
Sulfide- and magnetite-rich BIFs from the northeast have a relatively wide and mainly negative range of δ34S (4.9‰; from −3.8‰ to +1.1‰; n = 9) (Fig. 2A). This heterogeneity stands out compared to the results of the two previously mentioned northeastern formations. Komiya et al. (1999) analyzed metabasites and suggested that Domain I had undergone retrograde metamorphism to greenschist facies, in which case sulfide would not have survived with a primary isotopic signature. Rollinson (2002), on the other hand, determined a temperature for the metamorphism not higher than 520 °C by analyzing metapelites. So it is likely that this fairly moderate event did not re-homogenize δ34S; hence the isotopic variation remained completely or partially preserved in the sulfides.
The sulfide-rich BIF contains alternating quartz and magnetite bands, but with up to 5% of pyrites in some zones, mostly in quartz but also within magnetite. Some pyrite crystals are agglomerated in clusters, suggesting that they are not recrystallized by a post-metamorphic event. The wider range of δ34S and the lower metamorphic grade make these BIFs the most likely to preserve a primary bacterial signature, and it is possible that a much wider range produced by bacterial sulfate reduction has been narrowed by metamorphic homogenization. It cannot be ruled out though that the range was produced by hydrothermal magmatic sulfur, or that there was a mixing between biogenic sulfur and hydrothermal sulfur (+0.5‰) during metamorphism. However, it would be difficult to fractionate inorganically the hydrothermal source to significantly lower δ34S without involving a second source of lighter sulfur, except at extremely low pH, or at higher f in the presence of sulfates (Rye and Ohmoto, 1974). Sulfate though hasO2 so far not been found in these particular samples.
Carbon Isotopic Results and Discussion
δ13C of reduced carbon from the sedimentary sequences has been measured after acid treatment (Table 2, Figs. 3A, 3B). The δ13Cred range is from −29.6‰ to −6.5‰ (n = 27) (Grassineau et al., 2000). The carbon content of the Isua rocks is between 0.02 and 0.52 wt% (n = 27). The carbon was not visible in hand specimen and the samples analyzed were whole rock chips. Where known from the literature the carbon is referred to as graphite; otherwise it is called reduced carbon. In addition, carbonates were measured with δ13Ccarb of −2.2 ± 1.1‰. The two northeastern formations analyzed are the sulfide-rich BIF and the polymict conglomerate with δ13Cred as light as −29.6‰ (0.03 wt% C) (n = 8). With one exception, the values are lighter than −22.6‰.
The δ13Cred range in the western formations is from −20.1‰ to −6.5‰ (n = 13). The results obtained from the dark layers of the Bouma sequence are from −18.4‰ to −14.7‰ (n = 7) with carbon contents from 0.09 to 0.52 wt%. The values are similar to those of Rosing (1999) (−19.1‰ to −11.4‰). The two small BIFs associated with ultramafic rocks have homogeneous δ13Cred of −11.1‰ ± 0.8‰ (n = 3). The carbonate-facies BIF shows the entire isotopic range obtained in the west, indicating that more than one process occurred. Carbonates from this formation, mainly siderite, give δ13Ccarb values from −3.9‰ to −0.4‰ (n = 8).
To differentiate between carbon in the sediments and the other formations, we measured 18 reduced carbon samples of igneous rocks and graphites from hydrothermally affected units in the northeast part of the belt (Fig. 3C). Carbon contents range from 0.02–3.5 wt%. Nine graphite samples from amphibolite facies give a wide range of δ13Cred from −20.3‰ to −10.2‰ (−14.8 ± 5.4‰). These values fall roughly within the range already reported by previous workers (Fig. 3). δ13Cred was also measured in five metacarbonate samples, with a narrow spread of −6.3 ± 1.3‰. Finally two ultramafic rocks were analyzed with values of −8.5 and −5.8‰, and two volcano-felsic rocks, with −24.2 and −8.6‰. The δ13Cred range (close to 20‰) is large in these formations, but two main sources can be pointed out. The heavier values between −8.6 and −5.8‰ are most likely of magmatic origin, whereas the lighter range of −14.6‰ to −10.2‰ may record the circulation of high-grade metamorphic fluids during the early event (Perry and Ahmad, 1977; Naraoka et al., 1996; van Zuilen et al., 2002, 2003). These authors suggested that the siderite decomposed to graphite at the high temperatures reached by the regional metamorphism. This graphite would be 13C-enriched during isotopic re-equilibration with the siderite (Ueno et al., 2002).
The δ13Ccarb range of metacarbonates for the northeastern part is wider (from −5.7‰ to +1.2‰) than for the western samples (from −3.9‰ to −1.0‰) (Fig. 3D). The homogenization of δ13C in the western carbonate facies BIF is consistent with the higher grade; in fact, most of the carbonates here were most likely remo-bilized during the 2.84 Ga metasomatic event (Rose et al., 1996). The wider variation shown in the northeastern carbonates reflects the lower grade of the Domain I metamorphism, where the isotopic overprint was less effective.
Comparison between the results shows that δ13Cred is generally lighter in the sedimentary rocks than in the ultramafic and mafic rocks, with a wider range of 23.1‰. This is not the case for the two small western BIFs; the value of −11.1 ± 0.8‰ indicates that the high-grade metamorphism has overprinted the δ13C, with a re-equilibration toward the generalized graphite value of −12‰ obtained in other parts of the belt.
Despite the proximity of the three western BIFs, the metamorphic imprint is less obvious in the carbonate facies BIF. In contrast to δ34S, the δ13Cred range for this BIF is wide, 13.6‰, indicating that the re-homogenization was not as efficient as in the other western BIFs, and that some areas of the formation may have partially retained the primary value. The lightest δ13C, at −20.1‰, is similar to the original value suggested by Rosing (1999) for kerogen in the Bouma sequence nearby. Though metamorphic overprinting is likely, −20.1‰ might be primary too: if so, it could represent the burial of organic matter within the BIF. On the other hand, δ13Cred at −6.5‰ from the same BIF appears to be related to the late (2.84 Ga) high-grade metamorphism. The carbonate in this rock is mainly siderite, which locally may have decomposed to graphite during the high temperatures reached during the 2.84 Ga event.
The last western formation studied is the sedimentary outcrop interpreted as turbidites displaying a Bouma sequence. The δ13Cred range measured in this study (−18.4‰ to −14.7‰; n = 7) might represent a primary biological signature, but the possibility of mixing with the −12‰ high-temperature metamorphic carbon cannot be ruled out. Rosing (1999), who obtained the least modified δ13C at −19.1‰, inferred that this value represents planktonic-like organisms. Though by no means proving the case for biogenicity, the evidence suggests that the most depleted δ13C (here −18.4‰) might record biological activity at 3.8 Ga.
The two northeastern sedimentary formations show lighter values. With the exception of two samples, all δ13Cred are more depleted than −22‰ (Fig. 3A). This is consistent with the lower metamorphic grade in this area, implying that more original signatures might have been preserved. The sulfide-rich BIF values average −25.7 ± 2.7‰, with one δ13Cred at −29.6‰, the lightest found in this study of Isua. These samples consist mainly of alternating quartz and magnetite bands. There is no visible carbon in the quartz-layers in thin sections, so it is likely associated with the magnetite. The range of 6‰ in samples with low reduced carbon content does not suggest a high-temperature fluid source (as with δ34S, a fluid would have left a more homogenized δ13C range for reduced carbon). Thus the preferred interpretation is that some of this reduced carbon had an organic origin. The δ13Cred distribution possibly records two carbon sources, syn-sedimentary organic carbon and a small component of inorganic CO2 from post-depositional hydrothermal fluids circulating through the belt. If so, because metamorphism can enrich the graphite in 13C by up to 20‰ at 450 °C (Schidlowski, 2001), the primary signature of the organic component in these BIFs may have been even lighter than −29‰. In this case, the biological environment could have been anaerobic.
Fedo (2000) described the polymict conglomerate as detrital with a mafic origin for some of the clasts. δ34S in this study (+0.4 ± 0.5‰) supports this interpretation, and the two δ13Cred at −8.6‰ are also similar to the results obtained for metabasalts (Figs. 3A, 3C). However, four values of −28.6‰ to −23.4‰ indicate a second carbon source. They are close to the sulfide-rich BIF values and may record an organic origin. Could traces of life be found in a conglomerate? Preservation is possible and the metamorphic grade that affected the metaconglomerate was low enough not to cause significant shift in primary signatures. The view that the original environment hosted life and preserved an organic signature is debatable. The shallow subaqueous setting for the conglomerate deposition suggested by Fedo (2000) could host life. Therefore the value of −28‰ might be close to the primary signature of the organic activity. It is interesting to compare this with the isotopic signature of oxygenic photosynthesis, though this is no more than speculation in the absence of other supporting evidence.
In summary, metamorphism has caused widespread overprinting of the original δ13C signatures and especially the δ34S values of the IGB. However, some areas have better preserved the primary isotopic compositions. This is the case for the western turbiditic rocks, and more particularly in the northeastern part, which was subjected to slightly less strain and lower-grade metamorphism (Rollinson, 2002, 2003). The results presented here are not inconsistent with the hypothesis that there is a record of planktonic life in the western part (Rosing, 1999). From the results obtained in the eastern BIFs and metaconglomerate, biogenic processes, possibly including methanogenesis and photosynthesis (Rosing and Frei, 2004), might have been in operation at 3.7 Ga.
MIDDLE ARCHEAN: THE BARBERTON GREENSTONE BELT
The Barberton Greenstone Belt, located in the Kaapvaal Craton in South Africa, is one of the best-preserved mid-Archean successions. The regional metamorphic grade is low to moderate, much lower than in the IGB, and though there are many shear zones and décollement horizons, the sequences of rocks generally have experienced only low strain (Viljoen and Viljoen, 1969). Among the components of the belt are the Onverwacht Group (mainly mafic and ultramafic formations) and two clastic and chemical sedimentary sequences, the Moodies and Fig Tree Groups (Paris, 1987).
This study is based on the Fig Tree Group, which consists of ferruginous cherts, greywackes, shales, BIFs, and pelites, deposited in a submarine setting (Paris, 1987). The material, collected by J. Kramers and C. Siebert (Bern, Switzerland), comes from a horizontal borehole composed of greywackes and shales, drilled from the Fairview mine in an ESE direction out of the orebody toward the Sheba mine, in the central part of the belt (Siebert, 2003; Kramers et al., 2004). The Fig Tree Group has an age between 3.26 and 3.22 Ga from the measurements of zircons in tuff layers (Lowe and Byerly, 1999).
Previous Stable Isotope Investigations
Many carbon and sulfur isotope measurements have been made on the sedimentary groups. Sulfides in the Fig Tree Group show δ34S ranges of −0.4‰ to +4.0‰ (Strauss and Moore, 1992), +1.2‰ to +3.9‰ (de Ronde et al., 1992), and −0.9‰ to +4.4‰ (Kakegawa and Ohmoto, 1999). An average of +3.4 ± 0.2‰ for barite (Strauss and Moore, 1992) suggests that mid-Archean sea-water sulfate was not highly fractionated. Studies in the Moodies and Onverwacht Groups show similar ranges (e.g., Ohmoto et al. 1993: −3.0‰ to +8.6‰ in pyrites).
Carbon-rich shales in the Fig Tree Group contain up to 14.0 wt% of organic carbon with δ13Cred from −29.5‰ to −9.5‰ (de Ronde and Ebbesen, 1996). Other studies give ranges of −32.8‰ to −24.3‰ (Hayes et al., 1983), or −35.4‰ to −27.0‰ (Strauss and Moore, 1992). The δ13C in carbonates are from −4.5‰ to −2.0‰ (de Ronde et al., 1992). All these authors came to the conclusion that photosynthesis involving marine organisms was occurring at the time.
New Isotopic Results and Discussion of the Fig Tree Group
The newly analyzed samples consist of three 10–16-cm-long cores of sulfide-rich carbonaceous shales. Similar ranges to those previously obtained have been found on samples analyzed at millimeter scale, with δ34S in pyrites and pyrrhotites ranging from +1.0‰ to +5.6‰ (n = 15) (Table 3). The sulfide minerals are an assemblage of agglomerated fine grains, as seen in Figure 4. δ13Cred varies from −32.4‰ to −5.7‰ (n = 18) in samples with carbon contents up to 0.25 wt%, with no direct correlation between the C content and δ13C. Of the samples analyzed, 73% are lighter than −20‰. Pervasive calcite in bands has a very narrow δ13Ccarb range of −4.9‰ to −4.4‰ (n = 6).
A detailed section of one of the samples, presented in Figure 4, illustrates the isotopic variations obtained. Core 279B has thin dark laminations, and a central sulfide- and carbonate-rich layer ∼5 cm thick. There is a wide range of δ13Cred, from −31.9‰ to −7.8‰, but at both ends of the core δ13Cred is homogeneous, with −31.4 ± 0.4‰; the δ13C tends to become heavier toward the center, reaching −7.8‰ inside the “central” layer. This layer has peripheral 5-mm-thick carbonate-rich bands with very homogeneous δ13Ccarb of −4.5‰, similar to the average value found by de Ronde et al. (1992). The sulfides, mainly pyrrhotite on the outside edges of this zone, and pyrites in the center in bleb-like shapes, give a relatively narrow range of δ34S averaging +1.8 ± 0.4‰. However, a variation of 3.64‰ over 1.5 cm in another sample suggests that re-homogenization of δ34S, if it took place, was not complete, especially considering the low-grade greenschist facies metamorphism that the rocks have experienced (Siebert, 2003).
Kakegawa and Ohmoto (1999), who obtained similar δ34S variations in pyrite grains at micro-scale within the Fig Tree Group, suggested both organic and inorganic origins for the pyrite crystals. They pointed out that coarse-grained pyrites associated with quartz veins, with δ34S from +1.1‰ to +3.6‰, have an inorganic origin and are more likely precipitated from high-temperature hydrothermal fluids. Although the values of pyrites in the present study are in a similar range, none are associated with quartz veins and they are agglomerates of fine rounded grains, which parallel the sedimentary bedding, suggesting that they are a sedimentary or diagenetic feature. It is impossible to rule out a partial re-homogenization of the sulfides, but the δ34S values of +1.5‰ to +5.6‰ found locally seem to indicate at least some microbial reduction of the seawater sulfate. This was suggested by Kakegawa and Ohmoto (1999), who concluded that the Barberton sea at 3.4–3.2 Ga was bearing an appreciable amount of sulfate.
The isotopically homogeneous carbonate (δ13Ccarb ∼ −4.5‰) does not crosscut the structure of the shale. It occurs with pyrite adjacent to a redox interface, outside which is pyrrhotite and reduced carbon with δ13C around −31‰. It is likely that this is a result of syngenetic or diagenetic hydrothermal processes. Shales enclosed by the carbonate-rich zones show some bleaching (Fig. 4), and significant exchange between the primary organic carbon and carbon from the hydrothermal fluid. The δ13Cred varies from −27.6‰ to −5.7‰ close to the carbonate. This suggests that only the δ13C values around −31‰ represent a primary organic carbon signature. An alternative explanation for the relatively heavy δ13Cred values is that they are a record of organisms using waste carbon dioxide or bicarbonate from methanotrophs. Either way they are evidence for biological activity.
The view that light δ13Cred values and δ34S values of sulfides are biogenic is in agreement with Kakegawa and Ohmoto (1999). The light δ13C of −31‰ suggest that the dark shales may be of deep-water facies, deposited in anoxic zone below the photic zone.
LOWER MIDDLE ARCHEAN: THE STEEP ROCK GROUP
The Steep Rock Group from the Wabigoon Greenstone Belt in northwestern Ontario, Canada, contains an Archean carbonate platform, the Mosher Carbonate Formation, which is 500 m thick and made of laminated carbonate, mainly limestone, in part stromatolitic (Wilks and Nisbet, 1985, 1988; Kusky and Hudleston, 1999). Regional metamorphic grade is broadly lower greenschist. The age of the Steep Rock Group is not fixed. It rests unconformably on a ca. 3.0 Ga gneissic terrane, the Marmion Complex (Davis and Jackson, 1985; Wilks and Nisbet, 1988). The upper unit of the belt includes metavolcanic rocks with an age of 2.93 Ga (Davis, 1993 cited in Kusky and Hudleston, 1999) but it is partly allochthonous. It is thus likely but not certain that the Steep Rock Lake stromatolites are ∼2.9–3.0 Ga old. The well-preserved stromatolites clearly show biogenic structures on both small and large-scales (Wilks and Nisbet, 1985; 1988). Cryptozoon structures, branching walled and unwalled columnar forms up to 20 cm, are succeeded at the top of the unit by spectacular domal structures up to 3 m in diameter. The presence of these very well preserved primary structures, ooliths and oncolites, indicates that strain was very low.
The samples studied are three 1-m-long hand drill cores taken at different sites from near the larger domal structures. The carbonate is bluish-white, a mixture of calcite and dolomite, with undulating bands of black material. These laminae are described as organic kerogen (Hayes et al., 1983). Few analyses have been carried out on the formation; Hayes et al. (1983) obtained three δ13Cred values with a range from −26.4‰ to −22.1‰, and Schidlowski et al. (1983), three carbonates with δ13Ccarb from +1.1‰ to +2.0‰. The authors inferred that these values were the result of biological activity.
Results from Steep Rock Lake Stromatolites
The δ13Cred results (Table 3) obtained for the stromatolites are from −30.6‰ to −21.6‰ (n = 32), and δ13Ccarb from +0.1‰ to +2.9‰ (n = 56), in agreement with the previously obtained values. The ranges for both reduced and carbonate carbon are narrow, especially the carbonate, with an average of +2.0 ± 0.6‰ (Fig. 5). The reduced carbon isotopic values have a wider spread of 9‰. This could be original variation but might also indicate minor effects of metamorphism. Only one pyrite has been found. It has a δ34S value of +5.0‰.
The δ13Cred average of −25.4‰ and the Δ13 Cred-carb fractionation of 26‰–31‰ are interpreted as clear evidence of biological activity in these stromatolites, more particularly of the carbon fractionation by Rubisco I. Consequently, considering that the biogenic structures were developed in shallow water, on the margin of the Marmion Complex (Kusky and Hudleston, 1999), it is possible to propose that oxygenic photosynthesis was already established in the Steep Rock Group at 3.0 Ga, and that cyanobacteria were fully active. This and the Mushandike Formation in Zimbabwe are perhaps the oldest rock units where an unambiguous “modern” carbonate isotopic ratio can be iden-tified (Abell et al., 1985b).
LATE ARCHEAN: THE BELINGWE GREENSTONE BELT
The Belingwe Greenstone Belt in Zimbabwe is one of the best preserved of all Archean successions. The two sequences, the Manjeri and Cheshire Formations, respectively the lowest and highest parts of the Ngezi Group (upper greenstones) contain carbon- and sulfide-rich sediments and stromatolitic limestones that have been affected only by low-grade metamorphism (Martin et al., 1980; Abell et al., 1985a; Bickle and Nisbet, 1993). Bolhar et al. (2002) obtained Pb-Pb ages of 2706 ± 49 Ma for the Manjeri Formation, and 2601 ± 49 Ma for the Cheshire Formation, which is in agreement with Pb-Pb and Sm-Nd ages for the Reliance Formation of 2692 ± 9 Ma from immediately above the Manjeri Formation (Chauvel et al., 1993).
The Manjeri sediments were deposited in a continental basin unconformably upon a 3.5 Ga tonalitic gneiss (Bickle and Nisbet, 1993; Hunter et al., 1998). Four units of this formation in the Nercmar drill core were investigated: at the base, 40 m of organic carbon-rich sediments with intercalated thin sulfide layers (Spring Valley and Shavi Members), 56 m of volcanoclastic rocks (Rubweruchena Member), and at the top, 10 m of organic carbon- and pyrite-rich black shales (Jimmy Member) of which the base is massive pyrite, capped by a thick mafic lava sequence, the Reliance Formation. The study is mainly based on the carbon- and sulfide-rich sedimentary units. The Spring Valley and Shavi Members represent a shallow-water environment (Hunter et al., 1998). At the same stratigraphic level, the discontinuous Rupemba Formation contains well-preserved stromatolites that were deposited on slopes (Martin et al., 1980; Abell et al., 1985a), and that have undergone low-grade metamorphism with a maximum temperature of 200 °C (Abell et al., 1985a). Deposition of the Jimmy Member occurred below wave base but still in the photic zone. It is complexly folded at the contact with the Reliance Formation (Hunter et al., 1998; Grassineau et al., 2002), but with well-preserved sub-rounded and elongate sulfide structures below this.
The Cheshire Formation contains mostly shallow-water sediments, mainly dark and ferruginous shales, ironstones, conglomerates, important limestone reefs, with alternations of lava and tuff. Two units were studied, the first one being the 1500-m-thick dark shales, above the basal volcanic sequence. Their deposition was in shallow water, indicated by ripple mark structures (Martin et al., 1980). The shales are well preserved and show evidence of only low-grade metamorphism. The second unit is the 24-m-thick Macgregor stromatolites, one of the largest reefs of the carbonate formation. They were deposited on a volcanic slope and comprise 22 cycles of environmental change (Martin et al., 1980) that occurred in the basin. Abell et al. (1985a) observed little textural evidence for metamorphism and from δ18Ocarb estimated the re-equilibration temperature for these rocks to be 80 °C.
Previous Stable Isotope Investigations
Until recently there were few δ13C data for the Manjeri black shale units, with ranges of −32.2‰ to −8.8‰ for organic carbon, and −7.2‰ to −5.7‰ for carbonate (from Strauss and Moore, 1992). In contrast, Abell et al. (1985a) made an extensive study on the Cheshire stromatolites (Table 4) with ranges of −0.6‰ to +1.0‰ for the carbonate, and −35.4‰ to −16.9‰ for the reduced carbon. They also obtained a few δ13Ccarb values for the Rupemba stromatolites. Some Cheshire black shales analyzed by Yong (1991) show a δ13Cred range of −40.8‰ to −32.0‰, and δ13Ccarb values from −13.3‰ to −8.0‰ for the carbonates in veins (Table 4). Some δ13Cred measurements from the Manjeri carbon- and sulfide-rich sediments, with a range of −38‰ to −17‰, have been presented by Grassineau et al. (2001a, 2002). Grassineau et al. (2002) also obtained values of −35.1‰ to −7.3‰ for the Rupemba stromatolites and δ13Cred values of −35.1‰ and 36.6 wt% C for a soft black “ball” ∼0.5 mm in diameter found in the carbonate. This sample is bitumen and likely to be close to a pure kerogen composition. Finally a detailed study of δ34S presented by Grassineau et al. (2001a, 2002) gives a range of −17.6‰ to +16.7‰ in the three sulfide-rich Manjeri units.
New Results Obtained and Discussion of the Belingwe Greenstone Belt
Further analyses have been made on the carbon- and sul-fide-rich shallow-water and subtidal sediments from the Shavi and Jimmy Members and the Cheshire dark shales (Table 4). The results, obtained in pyrites, increase the extensive δ34S range in the Manjeri units (Figs. 6A, 6B), from −21.1‰ to +16.7‰. This range of nearly 38‰ is the widest range yet recorded in Archean sediments.
In the Jimmy Member alone the δ34S range equals the total spread found in the belt. There is a bimodal distribution of the δ34S data (Fig. 6A). The sharp main peak is around 0‰. This value occurs mainly at the top of the Jimmy Member, toward the contact with Reliance volcanics. The proximity of the volcanism, which is taken to be stratigraphically directly above the Manjeri Formation (see discussion by Grassineau et al., 2002), suggests that this value records hydrothermal processes that happened during the deposition of the overlying volcanic unit or soon after, producing local re-homogenization of the primary δ34S. The second and smaller peak in the Jimmy Member histogram is around −5‰, with a skewed distribution on both sides. This value is close to the −6‰ recorded in the only pyrite found in the Cheshire shales. The δ34S range in the Shavi Member, even if smaller, still has a significant spread of nearly 24‰. The population is distributed around a less pronounced peak of −4‰ (Fig. 6B).
Such large δ34S ranges suggest microbial sulfate-reduction processes, but locally values have been recorded where δ34S in the pyrites are above +6‰. This occurs only in one black shale sample of 10 cm length (TR51 at 67.05 m), where the pyrites are in bleb-like shapes with internal concentric structure, which are interpreted to be biogenic features. These 34S-enriched pyrites (up to +17‰) are associated with adjacent blebs of dolomite, which have light δ13Ccarb (−12‰ to −10‰).
The reduced carbon for the Manjeri shales has a δ13C spread of 31‰. The distribution shows a main value around −29‰ (Fig. 7), and carbon concentrations from 0.04 to 20.0 wt‰. A few samples containing “soft” bitumen-like material have a high content of reduced carbon (from 14 to 20 wt%). In detail, the Shavi and Spring Valley Members show an average δ13Cred of −25.0‰ ± 7.6‰ (Fig. 7A), whereas the Jimmy Member is much lighter, with a δ13Cred of −32.8‰ ± 4.2‰ (Fig. 7B).
The wide range found in δ13Cred indicates primary bacterial activity. The main peak value around −30‰ suggests fractionation by Rubisco I (Pierson, 1994). Shavi and Spring Valley units appear to reflect mainly the enzyme signature for oxygenic photosynthesis, but the lighter values in the Jimmy Member more likely record carbon cycling via methanogenesis, in addition to photosynthesis (Grassineau et al., 2001a, 2002). In comparison, data for Cheshire shales are very light, averaging −39.5‰ ± 3.0‰ (Fig. 7C). These occur locally in association with light carbonates (down to −13.3‰; Table 4).
These results suggest methanogenic and methanotrophic activities at the time (e.g. Yong, 1991; Grassineau et al., 2002), with methane oxidation by methanotrophs and sulfate reducers to generate the light δ13Ccarb values at the oxic/anoxic interface (e.g., Peckmann and Thiel, 2004).
Detailed Study of BES49c
Numerous samples have been analyzed in detail at millimeter scale to search for evidence of microbial consortia, and large isotopic variations have been found for both δ34S and δ13Cred. BES49c, a 5-cm-long core section, roughly at the middle of the Jimmy Member is an example. It is a black shale with a gray chert layer at the bottom, and on the top, a 1.5-cm-thick sulfide layer (Fig. 8). The sedimentary layering in BES49c suggests significant diagenetic compaction.
The δ34S range is from −19.6‰ to −5.4‰, with large variations of more than 9‰ over distances of 2 or 3 mm. All sulfides are pyrite. In the upper part, they have slightly cataclazed bleb-like shapes, rather rounded, consisting of microcrystalline framboid-like agglomerations, with very small internal cavities, resembling a “sponge” structure. The δ34S value of −18.6‰ inside the top left large bleb seems to have been sealed from further exchange since the time of its deposition (Fig. 8). The δ34S variations are evidence of very prolific biological activity. The heaviest value, at −5.4‰, is from a layer of disseminated small pyrite crystals, interbedded in dark chert layers. It is difficult to say if these sulfides are primary, but the −5‰ value is very common in the Jimmy Member and partial homogenization by hydrothermal fluids could have occurred. Most likely the value is a genuine biological signature from isolated activity with a plentiful SO4 supply, as it is mainly found in scattered interbedded pyrites. The lighter values are typically in ∼1–2-cm-size blebs of aggregated microcrystallized sulfides and the isotopic compositions are zoned in these structures. As the variations imply a biogenic origin, these rounded and granulated shapes might be an illustration of what remains of the bacterial communities, with intense biological activity illustrated by large fractionations (Canfield and Teske, 1996). In these multiple micro-ecosystems once hosted by BES49c, the interactions between the bacteria created different fractionation values, depending on the neighboring bacteria and environment. Each value then could represent the identity of a unique bacterial community.
The δ13Cred values in BES49c, −27.5‰ and −31.4‰, are also evidence of biological activity (Fig. 8). They give an idea of the organic processes that had most likely taken place. At the bottom, the signature seems to reflect Rubisco I activity, and consequently the existence of photosynthetic processes, whereas at the top, near the sulfides, the lighter value more likely indicates a combination of photosynthesis and probably methanogenesis. Studying the isotopic variations at the scale of the sample gives an instant snapshot of the interactions between different bacterial activities present at 2.7 Ga.
Carbon isotopic results obtained for the stromatolites of the Rupemba Member at the base of Manjeri and of the Macgregor Member in the Cheshire are shown in Figure 9. The samples are 1-m-long hand-drilled cores, and they show superbly preserved biological textures. The values of the Macgregor Member are from Abell et al. (1985a) and McClory (1988). Considering the very narrow ranges for δ13Ccarb and δ13Cred (Table 4), we concluded that the stromatolites represent nearly unaltered biogenic sediments.
For the Macgregor stromatolites, the average δ13Ccarb is +0.2‰ ± 0.3‰ and the δ13Cred values are mostly between −34‰ and −24‰, with a −28.6‰ ± 3.3‰ average (Fig. 9). With these results, a mean value of 28.8‰ for the Δ13Cred-carb fractionation can be inferred. Given the stromatolitic textures, this fractionation is interpreted as the unambiguous signature of Rubisco I activity, indicating that the Macgregor stromatolites were deposited by a cyanobacterial microbial ecology based on oxygenic photosynthesis (Abell et al., 1985a; Grassineau et al., 2002).
Further analyses were made in the Rupemba stromatolites, which underwent more strain and are in a setting where metamorphic recrystallization has been more extensive. The δ13Ccarb values range from −0.7‰ to +1.4‰ and average +0.1 ± 0.5‰, which are similar to Macgregor stromatolite values. The samples also analyzed for δ13Cred give values mostly between −23.3‰ and −21.0‰. The <200 °C metamorphism mentioned earlier may have shifted δ13Cred slightly toward heavier values. Compared with the Macgregor stromatolites, the evidence for Rubisco I is less obvious, but it cannot be ruled out because of the probable effect of metamorphism, suggesting that the original values might have been lighter. Alternatively, the 22‰ difference between δ13Ccarb and δ13Cred may be a primary fractionation record of Rubisco II controlling early anoxygenic activity or may be a result of lower pCO2 in the water column. On the other hand, the single bitumen value of −35.1‰ with 36.6 wt% C (Table 4) from the Rupemba stromatolites indicates that methanogenic processes were occurring either at the base of stromatolitic layers, or post-depositionally if the limestone acted as a hydrocarbon reservoir.
The wide δ34S range of nearly 40‰ obtained for pyrite from the Manjeri Formation indicates that the biological sulfur cycle was mostly, if not completely, operational at 2.7 Ga (Grassineau et al., 2001a), well before the post-Archean increase of atmospheric oxygen took place at 2.2 Ga (e.g., Habicht and Canfield, 1996; Blank, 2004). Moreover, the isotopic variations at millimeter scale show that an important diversity of sulfur-dependent organisms, mainly sulfate-reducing bacteria, were present. It is also possible that sulfide-oxidizing bacteria were operational in restricted localities during Jimmy Member deposition, as some anoxygenic photosynthesizers can conduct sulfide oxidation (Cohen et al., 1989). At the interface between anoxygenic and oxygenic conditions, elemental sulfur is liberated and then re-reduced in part to sulfide, by sulfur disproportionation. At the interface, 13C-depleted carbonate can also be formed, as seen in TR51 sample (e.g., Wadham et al., 2004). More recent examples of depleted carbonates associated with sulfide mats and globular sulfides are found near methane seeps (Peckmann et al., 2004). The light carbon is a result of the anaerobic oxidation of the methane in association with sulfate reduction (e.g., Peckmann and Thiel, 2004). An alternative explanation of the positive sulfur values could be Rayleigh fractionation of a finite sulfate reservoir. The data for sulfide below TR51 in the core are on average 15‰ lighter, but they form a separate peak, which does not tail off to 34S-enrichment as expected by Rayleigh fractionation (Fig. 6A). Therefore, a change to an environment typical of more modern methane- and sulfur-rich hydrothermal events seems to be the likely explanation. The relatively smaller range in the Shavi Member (Fig. 6B) suggests that the rate of SO4 reduction was higher than in the Jimmy Member (e.g., Schidlowski et al., 1983). It is possible that in the Jimmy Member, the SO4 supply might have temporarily and locally changed, or been limited at the deposition sites, as illustrated by the TR51 sample (δ34S above +6‰).
The wide range of δ13Cred measured in the three Manjeri carbon-rich units reveals the existence of a well-developed carbon cycle. The oxygenic photosynthesis operated by Rubisco I existed in the Shavi and Spring Valley Members. This is consistent with the interpretation of these units as shallow-water sediments (Hunter et al., 1998). At the same stratigraphic level, the Rupemba stromatolites, despite some isotopic adjustment, were most likely generated by the same activity, directed by cyanobacteria. The δ13Cred in the Jimmy Member may represent a complex mixing of signatures from different bacterial processes, recording composite bacterial communities. The Jimmy Member includes finely laminated material probably deposited in quiet water below wave base, deeper than the Shavi Member and with conditions becoming dominantly anaerobic toward the top of the unit, with δ13Cred lighter than −30‰. Further evidence for the introduction of methane into a previously oxic environment comes from the δ13C of organic carbon. Ten centimeters below TR51, the δ13Cred drops to −37‰ with an associated increase in carbon content from <1 wt% to up to 18 wt% in the overlying 2.4 m. At this stratigraphic level the sulfide isotopic compositions shift from being strongly heterogeneous to having a restricted range typical of hydrothermal sulfur (Fig. 6A). The apparent presence of an “oxygenic photosynthesis” signature, at intervals mainly observed in the lower part of this 2.4 m, may record the infall of dead organisms from the water surface.
Highly evolved prokaryote mat communities operating complex metabolic processes existed at 2.7 Ga (Grassineau et al., 2001a, 2002). They are the result of various microbial ecosystems due to a change of environments, from the subtidal or tidal microbial mats of the Shavi Member to a deeper environment below wave base with the appearance of anoxic processes in the Jimmy Member.
Similar observations are noted in the Cheshire Formation. The dark shales give very light δ13Cred values, down to −44‰, which argue in favor of anaerobic environments. The processes involved might have been occasionally anoxygenic photosynthesis at the oxic/anoxic interface, but more likely methanogenesis and methanotrophy. The δ13Cred values in the Cheshire stromatolites show the fractionation produced by the Rubisco I enzyme, evidence of oxygenic photosynthesis by cyanobacteria, in a tidal to subtidal environment.
COMPARISON AMONG THE FOUR ARCHEAN GREENSTONE BELTS
Figure 10 (A, B) shows the narrow δ34S ranges obtained for the Barberton and Isua Greenstone Belts. The metamorphic events that affected the IGB, including the northeastern part, might have reduced the original range of values, but the average is probably representative of the primary signature. Even if there is no additional evidence, a biological cause for this 4.9‰ range cannot be ruled out. The fractionation is small, perhaps because of a low supply of sulfate, but is still consistent with the presence of sulfate reducers. The presence of high-temperature (>85 °C) hyperthermophiles (Woese, 1987; Nisbet and Fowler, 1996; Nisbet and Sleep, 2001) could result in such a small isotopic fractionation. Evidence for hydrothermal activity at Isua comes from methane- and carbonate-rich inclusions in vesicles from the least deformed pillow lavas (Appel et al., 2001), which are similar in composition to some current off-ridge axis vent fields (Touret, 2003). In such settings inorganic processes may have been sig-nificant, with elemental sulfur forming at vents (δ34S of 0‰ to +1‰). Presumably little sulfate was present in the Isua oceans and around these vents, and sulfur-oxidizing bacteria might have existed (Jannasch, 1989). Under this condition, the fractionation of sulfur due to biological activity will surely have been small, whatever metabolic processes were occurring.
The range of δ34S values obtained for Barberton samples in this study but also by previous authors (see list of references in the BGB section) is slightly bigger that the one found for Isua. The average is also enriched (+1.4‰) relative to the IGB (−0.2‰). In the detailed study, the growth of replacement carbonate within the formation and preferentially around the sulfide minerals may have been responsible for some re-homogenization. However, the recorded δ34S range is best explained as evidence of bacterial sulfate reduction, in sediments or in the anoxic water column, at 3.24 Ga (Kakegawa and Ohmoto, 1999).
The impact of metamorphism and secondary processes on the carbon isotopic compositions is not negligible, and perhaps only locally has the δ13C partially preserved its original signature. This has to be taken into account in assessing claims for biogenicity, especially in Isua, where Schidlowski et al. (1979), Schidlowski (1988, 2001) and Rosing (1999) stressed that results should be corrected for the metamorphism. Only the range of −30‰ to −22‰ in the Isua sedimentary sequences can be discussed (though these values could have been lighter originally), as it indicates the possibility of life. Rosing (1999) suggests the presence of organisms similar to modern plankton in the Isua sea (δ13Cred = −19‰). On the other hand, a process, probably anaerobic, is proposed from the results of −29‰ to −28‰ in the sulfide-bearing BIF and the metaconglomerate. Given that the isotopic range may have been reduced by metamorphism, values as light as −29.6‰ in this BIF suggest that methanotrophic activity took place during deposition, although it is also possible that photo-synthetic processes had a wider range of fractionations at the time (Watanabe et al., 1997). Thus a complex community might have been present as early as 3.8 Ga.
In the middle Archean, the δ13Cred range in the Barberton belt is larger and lighter than in the Isua belt, with values down to −32.4‰ (Figs. 11A, 11B). Again, allowance must be made for a small metamorphic shift, and for the possibility that the heavier values may be linked to the formation of replacement carbonates. However, the lower part of the range, −32‰ to −26‰, suggests that several bacterial activities are recorded, including photosynthesis and methane-utilizing processes.
By the late Middle Archean, a significant change had occurred. The δ13Cred averaging −25.4‰ and the Δ13Cred-carb fractionation around 28‰ in Steep Rock Lake (Fig. 11C) are interpreted as clear evidence of biological activity and more particularly of fractionation by Rubisco I. Therefore, as the biogenic structures were developed in shallow water associated with a carbonate platform, it is likely that oxygenic photosynthesis was already established around 3.0 Ga, and that cyanobacteria were fully active at that time.
By the Late Archean, the wide δ34S range obtained in the Manjeri units strongly indicates an operational biological sulfur cycle at 2.7 Ga (Fig. 10C), with mainly sulfate-reducing bacteria, but also possible sulfur oxidizers. The activities seem to depend on local supplies of sulfate, conditioned by the environment. A fast SO4 supply will benefit a high rate of sulfate reduction in the Shavi Member, but with a slower rate, additional but restricted sulfur-oxidizing processes could have occurred within the Jimmy Member at the oxic-anoxic interface. The wide δ13Cred range measured in the Manjeri Formation (Fig. 11D) reveals the existence of a well-developed carbon cycle with Rubisco I operational. The Shavi and Spring Valley Members and probably the Rupemba stromatolites indicate oxygenic photosynthesis (Fig. 11F). The lighter δ13Cred values (below −30‰) recorded in the Jimmy Member and Cheshire Formation suggest that processes in addition to photosynthesis were occurring, most likely methanogenesis and methanotrophy (Fig. 11E). Values for δ13Cred in the Macgregor stromatolites (Fig. 11G) indicate oxygenic photosynthesis with cyanobacterial management.
Over this one billion years of Earth's history, the evolution of life appears to have been marked by changes of environmental conditions, as suggested by Nisbet and Sleep (2001). The δ34S and δ13C obtained in the three Archean periods suggest that biological processes created wider isotopic fractionations from 3.8 Ga to 2.7 Ga. With better sulfate and oxygen supplies, organisms colonized a wide spectrum of environments on Earth, which increased the biological diversity by the Late Archean.
In the Early Archean at 3.8 Ga, the δ34S spread in sedimentary sequences of 7.2‰ around a 0‰ peak suggests that biological activity was focused around high-temperature hydrothermal vents. Sulfur processing could have existed, and possibly methanogenesis (δ13Cred of −29‰ after metamorphism). Photosynthesis might also have occurred.
In the Middle Archean, at 3.24 Ga, microbial activity seems to be more widespread, but mainly under anoxic conditions. Sulfate-reducing processes have been recognized, using seawater sulfate. Methanogenic archaea were probably present, and possibly sulfur oxidizers. For Steep Rock Lake at 3.0 Ga, there is unequivocal evidence of Rubisco I and therefore oxygenic photosynthesis, with a fractionation between organic carbon and residual carbonate of −28‰. Therefore cyanobacteria were well established by the end of the Middle Archean/early Late Archean. The colonization of the continental margins by the cyanobacterial mats was the beginning of a flourishing of activity that is well illustrated at 2.7 Ga.
In the basal Manjeri Formation and the Cheshire Formation in the 2.7 Ga Belingwe Greenstone Belt, the ranges for δ13Cred and δ34S are wide, respectively ∼37‰ and ∼38‰. These values indicate a large diversity of bacterial activities. The important variations found at millimeter scale in samples indicate that these bacterial communities were interacting by various pathways. Oxygenic photosynthesizers and sulfate reducers were operating in microbial mats in shallow water reefs under subtidal and tidal conditions. Organisms below wave base used mainly anoxygenic photosynthesis and sulfate-reducing processes. Methanotrophs and sulfide oxidizers existed at the oxic/anoxic interface, with methanogens below.
Various funding bodies financed this study: Natural Environment Research Council (NERC) and the Leverhulme Trust for the Belingwe work; the Geological Survey of Denmark and Greenland, the Isua Multidisciplinary Research Project, the Danish Research Council, the Bureau of Minerals and Petroleum in Greenland, the Commission of Scientific Research in Greenland, the GEODE project, and the Royal Society for the Isua part. A particular thank you is addressed to C. Siebert and J. Kramers for the Fig Tree samples, courtesy of R. LeRoux and C. Rippon (AVGold, ETC division). The authors thank H. Ohmoto, T. Kakegawa and, Y. Shen for their constructive reviews, and S. Kesler for thorough editing of the manuscript.
Figures & Tables
Evolution of Early Earth's Atmosphere, Hydrosphere, and Biosphere - Constraints from Ore Deposits
- Amitsoq Gneiss
- Arctic region
- Barberton greenstone belt
- Belingwe greenstone belt
- biogenic structures
- black shale
- Canadian Shield
- clastic rocks
- Eastern Canada
- Fig Tree Group
- isotope ratios
- Isua Belt
- North America
- sedimentary rocks
- sedimentary structures
- shallow-water environment
- South Africa
- Southern Africa
- stable isotopes
- Superior Province
- Wabigoon Belt
- West Greenland
- Steep Rock Lake