Precambrian Mississippi Valley–type deposits: Relation to changes in composition of the hydrosphere and atmosphere
Published:January 01, 2006
- PDF LinkChapter PDF
Stephen E. Kesler, Martin H. Reich, 2006. "Precambrian Mississippi Valley–type deposits: Relation to changes in composition of the hydrosphere and atmosphere", Evolution of Early Earth's Atmosphere, Hydrosphere, and Biosphere - Constraints from Ore Deposits, Stephen E. Kesler, Hiroshi Ohmoto
Download citation file:
We have evaluated the temporal distribution of Mississippi Valley-type (MVT) Zn-Pb deposits with special attention to the nature and number of deposits of Precambrian age. Our evaluation is based on the widely used model for MVT mineralization involving metal-bearing brines that lack reduced S and that deposit sulfides only where they encounter a reservoir of sulfide or where sulfate in the metal-bearing brine is reduced to sulfide. For MVT systems of this type, basins with abundant sulfate would be most favorable for development of MVT mineralization because these would allow transport of metals in sulfate-rich brines and deposition of metals in areas where the sulfate was reduced. Because abundant sulfate requires abundant atmospheric oxygen, the distribution of MVT deposits through time might reflect compositional changes in Earth's atmosphere, especially the suggested Great Oxidation Event (GOE).
A compilation of new data for the Bushy Park-Pering district in the Transvaal Supergroup of South Africa, the world's oldest known MVT province, and published information on other Precambrian MVT deposits in the Ediacara, Berg Aukas/Abenab, Gayna River, Warrabarty, Nanisivik, Kamarga (Century), McArthur River (Coxco), Ramah, and Esker districts shows that they are generally similar in geologic setting and mineralogy to those in Phanerozoic rocks. Fluid inclusions in some Neo-proterozoic deposits, including Berg Aukas/Abenab, Gayna River, Warrabarty, and Nanisivik, record higher temperatures and salinities than found in most Phanerozoic deposits, possibly reflecting igneous activity or a more proximal basinal setting during Precambrian time. Fluid inclusion leachate data for several Precambrian MVT deposits suggest that their parent brines formed by evaporation of seawater, and S isotope compositions indicate that the S was derived largely from coeval seawater sulfate. Comparisons of data from all deposits show no evidence for a gradual increase in temperature or salinity backward through time, such as might be caused by higher heat flow during early stages of Earth history, although the magnitude of this effect might be lost in the uncertainty of most fluid inclusion measurements. These observations confirm that MVT deposits reflect the chemistry of their source basins, which are as old as 2.6 Ga. No MVT deposits or suitable host rocks of an older age are known.
Precambrian MVT deposits do differ from their Phanerozoic analogues in the magnitude of mineralization. Precambrian deposits and districts formed at an estimated rate of 5.5 per billion years versus a significantly larger rate of ∼60 per billion years for Phanerozoic deposits, and the Phanerozoic deposits are considerably larger. Furthermore, the transition from low-magnitude, Precambrian-type to high-magnitude, Phanerozoic-type MVT mineralization took place at the beginning of Cambrian time rather than at the 2.3 Ga GOE. This appearance of widespread MVT mineralization is closer to the time at which sulfate concentrations in the world ocean are estimated to have reached present-day levels. Although these conclusions are subject to considerable uncertainty because of the limited number of Precambrian deposits, the lack of an increase in the frequency of MVT mineralization at the GOE suggests that widespread MVT mineralization requires higher levels of sulfate than could have been provided by this event, or that the appearance of sulfate in the ocean was considerably delayed. Finally, the presence of MVT deposits in basins that formed considerably before the GOE suggests that local sulfate concentrations were available at even early points in Earth's history.
Mineral deposits have an uneven distribution through time that has been related to long-term changes in global heat flow, tectonism, and compositions of the atmosphere and oceans (Meyer, 1988; Barley and Groves, 1992). This uneven temporal distribution is particularly distinct for deposits that formed at Earth's surface, such as uranium-bearing conglomerates, laterites, evaporites, and iron formations, all of which contain redox-sensitive elements that might reflect increasing oxygen contents in the atmosphere and sulfate contents in the hydrosphere (Holland, 1984).
Ore deposits that formed in the deep subsurface also show temporal variations that might be due, at least in part, to changes in the composition of the atmosphere and hydrosphere. Among subsurface environments, sedimentary basins are most likely to reflect the surface environment because their rocks originated at the surface. The most widespread hydrothermal mineral deposits that formed by sedimentary hydrothermal systems are sandstone and unconformity U deposits, which contain redox-sensitive U, and sedex, Irish-type, and Mississippi Valley-type (MVT) Zn-Pb deposits, which contain redox-sensitive S. Elsewhere in this volume, reviews for two of these deposit types, sandstone-U (Gauthier-Lafaye) and sedex Zn-Pb (Lyons et al.) suggest that their temporal distribution reflects increasing atmospheric oxygen and marine sulfate, respectively. Whereas sedex Zn-Pb deposits form on the seafloor and sandstone-U deposits form in subsurface zones dominated by meteoric water, MVT Zn-Pb deposits form at somewhat greater depths in zones dominated by formation waters. These deposits also show a distinct age distribution, with fewer Precambrian representatives (Fig. 1). Here, we review the characteristics of these Precambrian MVT deposits with the goal of determining whether this pattern reflects a change in the oxygen content of the atmosphere and sulfate content of the oceans.
GEOLOGICAL SETTING AND GEOCHEMICAL CHARACTERITICS OF MVT DEPOSITS
MVT deposits are part of a continuum of Zn-Pb deposits, also including sedex and Irish-type, that form in craton margin and intracratonic sedimentary basins (Sangster, 1990). Although all three deposit types have similar mineralogy, textures, host rocks, and geology settings differ considerably (Fig. 2). At one end of the spectrum are MVT deposits, which consist of coarse-grained sphalerite, galena, and other minerals (Fig. 3A) that fill breccia-type porosity in dolomitized limestone along the margins of sedimentary basins (Anderson and Macqueen, 1988). MVT sulfides contain primary fluid inclusions with temperatures of 75–200 °C and salinities of 10–30 equivalent wt.% NaCl that are interpreted to be basinal brines that were expelled from deeper parts of adjacent basins (Fig. 2). At the other end of the continuum are sedex deposits, which consist largely of fine-grained, well layered Zn-Pb and other sulfides (Fig. 3B) in black shales along rifted margins of basins (Scott, 1997). Although sedex sulfides do not provide much useful fluid inclusion data, other lines of evidence suggest that their parent fluids were slightly hotter and less saline than MVT fluids, that they were a mixture of basinal brine and seawater, and that they were driven at least partly by heat from mafic dikes that intruded rift-margin faults (Fig. 2). Irish-type deposits contain both layered and coarse-grained ores and appear to have formed largely along faults that cut rocks between the seafloor on which sedex deposits formed and the deeper reservoirs in which MVT deposits formed (Fig. 2).
We have focused on MVT deposits in this study because they form at greatest depth in the subsurface and are most likely to reflect the chemical composition of their host sedimentary basins. Most MVT deposits consist entirely or dominantly of galena and sphalerite with sparry dolomite, calcite, and local quartz and pyrite or marcasite. Some MVT deposits, which consist dominantly of fluorite, have been considered to form a distinct group in recent surveys and are not included in this summary (Sangster, 1990; Leach and Sangster, 1993). Similarly, barite is present in only a few deposits and, where present, it is not coeval with galena and sphalerite and thus does not provide a useful constraint on the composition of the metal-bearing MVT fluid. Other metals that are found in small amounts locally, including copper, cobalt, arsenic, and silver, are not sufficiently systematic in their distribution to classify MVT deposits further, and most were derived largely from nearby wall rocks (Burstein et al., 1992). Thus, insights into the processes that formed most MVT deposits are provided largely by constraints on the behavior of Pb, Zn, and S, and it is these metals that we must look to for information on the redox state of hydrothermal fluids in their host sedimentary basins.
Experimental studies show that chloride complexes are effective ligands for Zn and Pb in low-temperature, basinal brines and that bisulfide and organic complexes are less effective, particularly in the presence of abundant dissolved Ca and Na (Giordano, 1985; Barrett and Anderson, 1988; Sicree and Barnes, 1996). The presence of reduced (sulfide) S in MVT brines greatly reduces the solubility of chloride-complexed Zn and Pb (Anderson, 1983). The most widely applied model for formation of MVT deposits is based on this relation and involves metal-bearing brines that lack reduced S and that deposit MVT sulfides only where they encounter a reservoir of sulfide or where sulfate in the metal-bearing brine is reduced to sulfide (Anderson and Macqueen, 1988; Anderson, 1991; Leach and Sangster, 1993).
According to this “Anderson model,” basinal environments containing only reduced S could not form MVT deposits because their low-temperature brines would not transport sufficient metals. Conversely, basinal environments that lack S would allow extensive migration of metals in low-temperature brines, but would not have the capacity to deposit MVT sulfides. The most favorable basinal environments for formation of MVT deposits would be those containing sulfate-bearing evaporites and brines and local accumulations of sulfide formed from this evaporite sulfate. All other things being equal, the MVT-forming capacity of such sedimentary basins should increase with increasing amounts of sulfate that could be reduced locally to cause ore deposition. Possible mechanisms for sulfate reduction in Precambrian sedimentary settings include near-surface bacterial action and deeper thermochemical sulfate reduction, as is the case in modern settings (Ohmoto and Goldhaber, 1997; Habicht et al., 2002). Because basinal brines form largely by evaporation of seawater or dissolution of marine evaporites (Hanor, 1987), their sulfate contents should reflect the composition of their source ocean and the oxygen content of its coeval atmosphere.
Although the Anderson model provides a useful framework for evaluating possible relations between MVT ore formation and compositional evolution of the atmosphere and hydrosphere, it is not without complexities. Ohmoto and Goldhaber (1997) have described eight pathways by which seawater sulfate can be fixed as sulfide in ore deposits. Whereas some of these pathways involve single-stage reduction of sulfate by biogenic, thermochemical, or more direct inorganic processes, others involve multiple steps such as bacterial reduction of seawater sulfate during diagenesis, formation of pyrite, and its subsequent dissolution and transport. In general, biogenic and diagenetic sulfides have lower δ34S values than sulfides formed by thermochemical reduction of seawater sulfate, and some MVT deposits, such as the large Tri-State district, contain sulfides with low δ34S values that might be of this origin (Ohmoto and Rye, 1979; Ohmoto and Goldhaber, 1997; Leach and Sangster, 1993). Although the existence of numerous possible pathways makes it more difficult to identify the actual source of S, and the presence of districts such as Tri-State show that alternative pathways operated locally, they do not negate the generalization that high basinal sulfate contents favor formation of MVT deposits.
Other possible complications to the model used here involve solubility constraints. For instance, Ba cannot be transported in the presence of significant dissolved sulfate (Blount, 1977), and the presence of barite in MVT deposits has been cited as evidence that MVT brines lacked sulfate. However, as noted above, barite is actually absent from most MVT deposits and, where present, it was not deposited at the same time as galena and sphalerite. A more important complication involves the possibility that acid MVT brines might carry metals and reduced S together, thus obviating the need for sulfate-rich environments noted above (Sverjensky, 1981). Ohmoto et al. (1990) have suggested that MVT districts such as the Upper Mississippi Valley, where sulfides have uniform δ34S values and equilibrium fractionation, formed from fluids that contained both metals and sulfide, which requires that the fluid was either unusually hot or acidic. Wall-rock alteration in most MVT deposits does not provide evidence of highly acid brines, however, and maintenance of acidity in brines hosted by carbonate rocks requires high levels of CO2, which have been documented in only a few MVT deposits (Haynes et al., 1989). If this exception does apply, it is most likely for MVT districts that formed at high temperatures, such as the Upper Mississippi Valley district (McLimans et al., 1980; Ohmoto et al., 1990). As is apparent from this list of complications, the Anderson model probably is less clearly applicable to Irish-type and sedex deposits, and they are not included in this discussion (Sangster, 1990).
In addition to general acceptance of the Anderson model, use of Precambrian MVT deposits as possible sensors of oxygen and sulfate contents of the atmosphere and hydrosphere requires that (1) their geology and geochemistry be similar to Phanerozoic MVT deposits to which the Anderson model was originally applied, and (2) their age be constrained by geologic, paleomagnetic, or isotopic observations. In the following sections, we review Precambrian MVT deposits from these two perspectives.
GEOLOGIC FEATURES OF PRECAMBRIAN MVT DEPOSITS
MVT deposits that are hosted by Precambrian rocks and to which comparisons might be made include, in order of increasing host rock age, Ediacara, Berg Aukas/Abenab, Gayna River, Warrabarty, Nanisivik, McArthur River (Coxco), Kamarga, Esker, Ramah, and Bushy Park-Pering-Zeerust (Fig. 4). Geological features of these deposits and districts are summarized in Table 1 and discussed briefly in the next section.
Some deposits of possible MVT origin are not included in this survey because of inadequate data, because they appear to be transitional in character, or because metamorphic and other overprints obscure their original features. For instance, Neoproterozoic platform carbonate sequences in Brazil, including the Una (Irecê Basin) and Bambui (São Francisco Basin) Groups that originally covered an area of more than 300,000 km2, host Pb-Zn mineralization of uncertain type. The best-known deposits, including Morro Agudo, Vazante, and related deposits in the São Francisco Basin, have been classified as MVT, sedex, and Irish-type in different studies, but most recent descriptions suggest that they are least like MVTs and most like sedex (Iyer et al., 1992; Hitzman et al., 1995; Kyle and Misi, 1997; Misi et al., 2005). Similarly, the Balmat-Edwards district in New York has a general geologic setting typical of MVT deposits, including association with evaporites, but it is so highly metamorphosed that original textures necessary to determine its origin have been largely obliterated (Whelan et al., 1984, 1990).
MVT Deposits in Neoproterozoic Basins
Two of the four possible Neoproterozoic MVT districts, Ediacara and Berg Aukas/Abenab, have unusual metal contents. The Ediacara district, which is in the central Flinders Ranges of South Australia, is actually hosted by Cambrian shelf carbonates of the Ajax Limestone (Table 1), but is part of a dominantly Proterozoic sedimentary sequence (Christie-Blick et al., 1995). Mineralization consists of galena and pyrite with minor chalcopyrite and sphalerite and rare tetrahedrite and pearceite and has higher silver grades than most Phanerozoic MVTs (Drew and Both, 1984; McFarlane and Bone, 1994). The Berg Aukas/Abenab district, which is in the foreland thrust belt of the Damara orogen of Namibia, consists of galena, sphalerite, and sparry dolomite with trace amounts of pyrite, tetrahedrite, enargite, and chalcopyrite, as well as vanadium, silver, germanium, gallium, and cadmium at levels above those typical of Phanerozoic MVT deposits (Frimmel et al., 1996).
The Gayna River district, which is hosted by passive margin, platform carbonates of the Little Dal Group in Canada, is more typical of Phanerozoic MVTs in both composition and geologic setting (Table 1). The Little Dal Group is distinguished by its extensive stromatolites with internal structures typical of Precambrian reefs and by the presence of evaporites, including local salt casts (Hardy, 1979; Narbonne and Aitken, 1995; Turner et al., 2000). MVT mineralization is hosted by sedimentary and solution-collapse breccias that contain sphalerite and galena with sparry dolomite and local barite with snow-on-the-roof and colloform textures typical of Phanerozoic MVTs (Hardy, 1979; Hewton, 1982).
A final deposit that might belong in this group is Warrabarty in the Patterson orogen of Western Australia (Anderson et al., 2001, 2002). Warrabarty consists of Zn-rich mineralization in lower greenschist facies, metamorphosed carbonaceous dolostones, and limestones of the Broadhurst Formation, a part of the Meso-Neoproterozoic Throssell Group. Mineralization consists largely of sphalerite and galena with lesser pyrite, pyrobitumen, and sparry dolomite in breccias, veins, and disseminations in dolomitized wall rock (Smith, 1996).
MVT Deposits in Mesoproterozoic Basins
MVT districts are scarce in rocks of Mesoproterozoic age, the only good example being the Nanisivik district (Fig. 4). Although commonly classified as MVT, Nanisivik is somewhat unusual in form and mineralogy (Table 1). The ore-hosting Society Cliffs Formation is a platform carbonate sequence consisting largely of stromatolitic to massive dolostone with gypsum horizons that are among the oldest known extensive evaporite deposits (Kah et al., 2001). Solution-collapse and karst breccias are widespread in the upper part of the Society Cliffs Formation, but debate persists about their relation to MVT mineralization (Jackson and Ianelli, 1981; Olson, 1984; Ford, 1986). The ore zone is unusually sulfide-rich and ranges in texture from alternating layers of pyrite, sphalerite, sparry dolomite, and galena that fill open spaces to massive sulfides showing evidence of multiple stages of wall-rock replacement (Olson, 1984; Ghazban et al., 1990). Large iron sulfide bodies surround the Pb-Zn ore bodies and make up most of the district (Arne et al., 1991; Sutherland and Dumka, 1995). Iron-rich MVT mineralization forms from hot brines at temperatures of 200 °C or more, which is consistent with fluid inclusion evidence discussed below and might account for the unusual tube-like form of the main ore zone (St. Marie et al., 2001).
MVT Deposits in Paleoproterozoic Basins
Probable MVT districts in Paleoproterozoic rocks, including Coxco (McArthur River), Kamarga (Century), Ramah, and Esker, differ somewhat from Phanerozoic MVTs, especially in their apparently closer association with possibly coeval sedex mineralization. These relations are best displayed in the “Carpentaria zinc belt” of Australia, which contains both the McArthur River-Coxco and Kamarga districts.
MVT mineralization at Coxco is hosted by the latest Paleo-proterozoic McArthur (Coxco) and McNamara (Kamarga) Groups in north-central Australia. Both districts contain much better known sedex deposits, HYC in the McArthur Group and Century in the McNamara Group (Plumb et al., 1990). The McArthur River district and immediately adjacent area contains MVT mineralization at Coxco and possibly at Ridge and Cooley (Williams, 1978; Walker et al., 1983; Selley et al., 2001; Garven and Bull, 2000). Overlying middle and upper McArthur Group platform dolostones, which host MVT mineralization, contain layers of acicular, radiating carbonate fans (Coxco needles) thought to have been precipitated as aragonite (Winefield, 2000). Mineralization at Coxco is hosted by a karst system that formed during uplift and weathering of domal stromatolites and fine-grained dolomite of the Mara Dolomite Member, and is very similar to Phanerozoic MVTs (Walker et al., 1983). It contains a first stage of colloform sphalerite, galena, and pyrite-marcasite associated with abundant organic matter, which was covered by detritus from the overlying Lynnot Formation and then followed by a second stage of crosscutting, coarse-grained pyrite-marcasite, sphalerite, galena, sparry dolomite, and minor bitumen in veins and dolomite breccias of probable tectonic origin. Mineralization at Cooley and Ridge is similar to the second stage at Coxco, but includes late chalcopyrite, tetrahedrite, and bornite, which are more abundant than in most Phanerozoic MVT deposits (Williams, 1978).
The McNamara district hosts probable MVT mineralization at Kamarga, which consists of pyrite and sphalerite in veins, breccias, disseminations, and massive replacements in the Gunpowder Creek Formation (Jones, 1986). The Gunpowder Creek Formation contains abundant textures typical of evaporites that have undergone replacement and is thought to have been deposited in a sabkha environment (Jones et al., 1999).
The Esker district near the Arctic coast in Canada is hosted by passive margin shelf sediments of the Rocknest Formation, which include abundant silica pseudomorphs after halite, gypsum, and possibly anhydrite (Grotzinger, 1986a, 1986b, 1986c). MVT mineralization, consisting largely of sparry dolomite, sphalerite, and galena with minor chalcopyrite, is in ore-matrix breccias and disseminations in a regionally extensive stromatolitic reef zone, and several possibly coeval Cu-Co-Pb-Zn sedex showings are in the underlying Odjick Formation (Rhondacorp 2002; Wachowiak, 2001). Paragenetic relations indicate that MVT mineralization followed regional dolomitization, silicification, and sparry dolomitization (Wachowiak, 2001; Wachowiak et al., 1997, 1998).
The Ramah district in northeastern Labrador, Canada, is the most strongly deformed and metamorphosed of the deposits reviewed here, and this limits the certainty with which it can be placed in the MVT class. The district is hosted by shelf carbonates of the Reddick Bight Formation, part of the Paleoproterozoic Ramah Group (Morgan, 1975; Knight and Morgan, 1977, 1981; Mengel et al., 1991). These rocks were deformed and metamorphosed during the 1.86 Ga Torngat orogeny, reaching green-schist facies in the area of the MVT prospects (Korstgard et al., 1987; Mengel et al., 1991; Mengel and Rivers, 1994; Scott and Gauthier, 1996; Hayashi et al., 1997). Mineralization is hosted by breccias that are cemented by sparry dolomite, pyrite, sphalerite, galena, quartz, and calcite, as well as carbonaceous material thought to have been derived from bitumen (Archibald, 1992; Archibald and Wilton, 1994; Wilton et al., 1993, 1994). Feldspar gangue is present locally, although it might have formed during later remobilization, as is indicated by crosscutting veins with MVT-like mineralogy.
MVT Deposits in Archean Basins
With the exception of Gayna River and Nanisivik, the deposits described so far are not particularly compelling examples of MVT mineralization. MVT deposits in the Transvaal Supergroup of South Africa, however, are remarkably similar to Phanerozoic deposits. The Neoarchean Ghaap and Chuniespoort Groups in the lower part of the Transvaal sequence consist largely of carbonate platform sediments with abundant stromatolitic reefs and overlying banded iron formation and contain the Pering, Bushy Park, Zeerust, and other smaller MVT districts (Beukes, 1987; Altermann and Nelson, 1998; Eriksson and Altermann, 1998). These deposits are found in two remnants of the original Transvaal Basin, which are known somewhat confusingly as the Griquatown West and Transvaal Basins.
The Griquatown West Basin contains the Pering and Bushy Park Zn-Pb deposits. The two deposits are sufficiently far apart to be considered separate districts, although each district contains only one important deposit. Pering and surrounding prospects are largely in breccias of probable karst origin that are aligned along fracture systems (Wheatley et al., 1986b; Kruger et al., 2001). Mineralization at Pering includes early fine-grained, colloform sphalerite and subordinate galena and chalcopyrite associated with sparry dolomite and hydrocarbons, and a final stage of coarse-grained sphalerite, galena, sparry dolomite, quartz, and calcite (Greyling et al., 2001). Bushy Park, the more northerly of the two districts, is also hosted by breccias of probable collapse origin and consists largely of coarse-grained sphalerite, sparry dolomite, and galena (Wheatley et al., 1986a; Martini et al., 1995; Baugaard et al., 2001; Schaefer et al., 2001).
The Transvaal Basin contains the Zeerust district and scattered smaller deposits that surround the Bushveld Complex, which intruded the Transvaal sequence ca. 2.06 Ga (Altermann and Nelson, 1998; Eriksson and Altermann, 1998). Mineralization at Zeerust consists dominantly of fluorite in stratabound zones and breccia bodies and the smaller deposits consist largely of sphalerite and galena with variable amounts of fluorite (Martini, 1976; Roberts et al., 1993; Martini et al., 1995; Poetter, 2001). As discussed in the next section these deposits are thought to have formed as part of a large hydrothermal system around the Bushveld Complex. In view of their high fluorine content and igneous association, they are not included in this survey.
Age of Precambrian MVT Deposits
Geological relations provide the only constraints on the age of mineralization in most of the Precambrian MVT districts discussed here. At Ediacara, mineralization took place during formation of the Neocambrian–Cambrian sedimentary sequence (Drew and Both, 1984). At Berg Aukas/Abenab, mineralization is related to 0.75 Ga rifting along the northern margin of the Damara Basin (Frimmel et al., 1996). Age relations for Warrabarty are particularly poorly defined. No measurements are available on the deposit and the Throssell Group, with which mineralization is roughly coeval, is limited only by 1.08 Ga granites that it overlies unconformably and a post-ore 0.71 Ga metamorphic event (Blockley and Myers, 1990; Smith, 1996). Common Pb isotope models suggest that mineralization was synchronous with formation of sediment-replacement copper deposits in the area (Nifty) ca. 0.84 Ga (Smith, 1996).
At Gayna River, main-stage mineralization is younger than the 0.78 Ga dikes that cut it (Hewton, 1982). At Coxco (McArthur River), early mineralization was deposited while karst zones in the Reward Dolomite were being filled by clastic sediment of the overlying Lynott Formation, which constrains it to an age of ca. 1.64 Ga, and second stage mineralization at Coxco took place after deposition of the Lynott Formation (Walker et al., 1983; Page et al., 2000). No age constraints are recognized for mineralization at Cooley and Ridge, but they and Coxco should be roughly coeval with the 1.69 Ga enclosing sedimentary rocks if MVT mineralization is part of the HYC sedex system (Plumb et al., 1990). The age of mineralization at Kamarga has not been measured; its suggested relation to the Century sedex deposit constrains it to an age only slightly less than that of the 1.67 Ga McNamara Group host rocks (Plumb et al., 1990; Jones et al., 1999). Ages at Esker and Ramah are also not well known. At Esker, MVT mineralization is cut by faults related to the earliest phase of compressional deformation in the Coronation Group, making it probably post-1.90 Ga and pre-1.84 Ga in age (Hoffman and Bowring, 1984; Bowring and Grotzinger, 1992; Wachowiak, 2001). Breccia-hosted ore at Ramah is cut by structures related to the Torngat orogen, which formed during collision of the Nain and Rae terranes ca. 1.86 Ga (Wilton et al., 1993; Archibald, 1992).
Paleomagnetic and isotopic measurements supplement information from geologic relations at Nanisivik, where a diabase dike interpreted to cut ore has been correlated with the 0.72 Ga Franklin dike event (Olson, 1984; Heaman et al., 1992; Pehrsson and Buchan, 1999; Symons et al., 2000). Rb-Sr isotope compositions of sphalerite, dolomite, and leachates from Nanisivik do not fall on an isochron but are consistent with an age between ca. 0.75 and 1.25 Ga (Christensen et al., 1993). A paleomagnetic pole on recrystallized dolomite around the ore zone corresponds to an age of ca. 1.095 Ga on the North American apparent polar wander path (Symons et al., 2000). Although we have accepted a Mesoproterozoic age for Nanisivik here, some organic matter in the deposit appears to reflect a younger age (Gize, 1986), the age of the Franklin dikes has been challenged, and Ar-Ar measurements on MVT-related orthoclase yield Ordovician ages (Sherlock et al., 2004). Until these contradictory observations are resolved, inclusion of Nanisivik in this compilation should be regarded as tentative.
In the Transvaal Supergroup deposits of South Africa, the age of MVT mineralization is constrained by geologic relations and isotopic measurements. Mineralization is found largely in porosity related to an extensive karst system that developed on the Ghaap and Chuniespoort rocks in one or more intervals during Paleoproterozoic time. At Bushy Park-Pering, the age of ore-related karsting limits mineralization to pre-2.1 Ga (Martini et al., 1995). The Kalkdam and Katlani prospects, which are hosted by Ventersdorp lavas that underlie the Transvaal sequence in the Pering area, have similar Pb isotope compositions and yield a Rb-Sr isochron age of 1.977 Ga, which has been interpreted to indicate that mineralizing fluids were expelled during deformation of the 2.0 Ga Kheis Belt on the western edge of the Kaapvaal craton (Duane et al., 1991; Kruger et al., 2001). Relatively high 87Sr/86Sr ratios for calcite, dolomite, and sphalerite at Pering and Bushy Park require that the brine contacted evolved, potassium-rich rock outside the ore-hosting carbonate sequence, possibly during the Kheis event (Kruger et al., 2001; Schaefer et al., 2001; Kesler et al., 2003). Ar-Ar measurements on illite thought to coexist with ore minerals at Bushy Park yield an age of 2.145 ± 0.007 Ga, which is the best age estimate available at this time for Bushy Park-Pering mineralization (Schaefer, 2002).
In the Transvaal Basin, some Zn-Pb MVT deposits might have formed ca. 2.35 Ga, as indicated by relations at Genadendal where mineralization in Chuniespoort carbonate rocks forms a feeder for Zn-rich shale at the base of the overlying Pretoria Group (Martini, 1990; Eriksson et al., 2001a). However, fluorite mineralization at Zeerust yields a Sm-Nd isochron age of 2.06 Ga that is essentially the same as the Bushveld Complex, which dominated fluid migration in the Transvaal Basin (Kesler et al., 2003).
GEOCHEMISTRY OF PRECAMBRIAN MVT DEPOSITS
Fluid Inclusion Temperatures and Geochemistry
Temperatures of 75–200 °C are commonly cited for Phanerozoic MVT deposits with most of the variation related to position in the flow path relative to the source basin or possible sources of additional heat (Leach and Sangster, 1993; Rowan et al., 2001). Significantly higher heat flow in Precambrian time probably produced a hotter basinal environment. Estimates based on the history of global heat flow and resulting geotherms (Pollack and Chapman, 1977; Pollack, 1997) indicate that continental temperatures during Paleoproterozoic time were probably ∼100 °C higher at depths of ∼10 km, a level typical of the base of a thick sedimentary basin. A more conservative estimate of 50 °C is used here, assuming shallower source depths for mineralizing brines, lower temperatures in active sedimentary basins, and published estimates of ocean Precambrian temperatures (Knauth, 2005). This effect is shown in Figure 5 as a gradually increasing increment to the Phanerozoic temperature range. Increased ocean temperatures could have facilitated generation of basinal brines with salinities above the 10%–30% level typical of Phanerozoic deposits, although it is more difficult to estimate this effect because of the range of solution compositions observed in MVT deposits. An approximation based on the increase in salinity with temperature in the NaCl-H2O system (Sourirajan and Kennedy, 1962) results in salinities of ∼33% for late Archean brines with temperatures of 250 °C. This effect is shown as a similar continuous decrease in inclusion salinities though Precambrian time in Figure 6.
Homogenization temperatures for fluid inclusions, largely in sphalerite, from Precambrian MVT deposits are generally more complex, with higher temperatures and more CO2 and locally CH4, than their Phanerozoic counterparts (Fig. 5). Ediacara, the youngest deposit, has homogenization temperatures typical of Phanerozoic MVT deposits (Drew and Both, 1984), but temperatures for Berg Aukas (100–210 °C), Gayna River (156–231 °C), Warrabarty (165–245 °C for gray-stage sphalerite), and Nanisivik (87 to above 300 °C) are progressively higher (Misiewicz, 1988; Carriere and Sangster, 1992; Olson, 1984; McNaughton and Smith, 1986). Highest temperatures at Nanisivik, which are slightly above 300 °C, have been attributed to reheating by the diabase dike that cuts ore, but temperatures of 250 °C have been interpreted as primary, perhaps reflecting higher heat flow for the host rift-margin basin (Olson, 1984; McNaughton and Smith, 1986).
This trend of increasing homogenization temperatures with increasing age does not continue for Mesoproterozoic and older deposits, however. Inclusions in sphalerite from second-stage mineralization at Coxco (McArthur River) homogenize at 100–170 °C and those at Esker homogenize at ∼100–150 °C, both of which are typical of Phanerozoic MVT deposits. Homogenization temperatures for sphalerite at Ramah fall in the same range (120–180 °C), but inclusions in quartz and dolomite extend to 320 °C and contain CO2, probably reflecting a metamorphic over-print that might have corrupted the homogenization temperature record (Archibald, 1992). At Kamarga, fluid inclusions in sphalerite have homogenization temperatures of 270–320 °C and are accompanied by CO2-rich vapor inclusions that are not as clearly related to a post-ore overprint (Jones et al. 1999).
Fluid inclusions in Archean-hosted Zn-rich MVT deposits are more typical of those in Phanerozoic-hosted MVT deposits with the exception of high-temperature, gas-rich inclusions of uncertain origin that are present in some of the older deposits. Reconnaissance observations that we have made on sphalerite and sparry dolomite from Bushy Park yielded homogenization temperatures of 77–120 °C for primary and pseudosecondary aqueous inclusions and 130–195 °C for secondary aqueous inclusions, all in sphalerite (Table 2). Secondary aqueous inclusions from sparry dolomite homogenized at temperatures of 150–190 °C, slightly higher than temperatures of 100–175 °C reported by Wheatley et al. (1986b) for gangue carbonate at Pering. Schaefer et al. (2001) reported a similar range of homogenization temperatures (90–168 °C) for Bushy Park sphalerite and dolomite, with no systematic difference between inclusions in different minerals. They also reported primary vapor-rich inclusions containing variable proportions of CO2 and CH4 from Bushy Park, and used these inclusions to determine a pressure correction of ∼50 °C for the homogenization temperatures. Greyling et al. (2001) reported temperatures of 157–210 °C for Pering based on intersecting isochors in aqueous and carbonic inclusions.
Freezing temperatures (and salinities estimated from them) for fluid inclusions in sphalerite from some of the Precambrian MVT deposits extend to values in and above the high end of the range typical of Phanerozoic MVT deposits (Fig. 6). Whereas the range of freezing temperatures for inclusions at Gayna (−12 to −24 °C) falls in the center of the Phanerozoic range, those from Coxco II (McArthur River) (−22 to −28 °C) and Esker (−18 to −24 °C) are nearer the high end of the range (Walker et al., 1983; Wachowiak et al., 1997, 1998; Carriere and Sangster, 1992; Smith, 1996). Freezing temperatures from quartz and dolomite at Ediacara (−23 to −28 °C) are also near the top of the Phanerozoic MVT range. Similar patterns are seen in the deposits for which only salinities are quoted. Inclusions in sphalerite and dolomite from Warrabarty have salinities of 15–26 total salt and contain significant amounts of Ca (Smith, 1996); inclusions in quartz and dolomite from Ramah have salinities of 15–29 equivalent wt% NaCl (D. Wilton, 2002, written commun.), and inclusions from sphalerite and dolomite at Berg Aukas have average salinities of 23 equivalent wt% NaCl (Misiewicz, 1988). All of these are in the upper part of the Phanerozoic MVT range (Fig. 6). Inclusion fluids at Nanisivik have even higher salinities, with first-melting (eutectic) temperatures of −50 °C or more, unusually low last-melting temperatures of −25 to −45 °C, and estimated salinities of 24–35 equivalent wt% NaCl (McNaughton and Smith, 1986).
Our reconnaissance freezing measurements on primary and pseudosecondary inclusions in Bushy Park sphalerite indicate the presence of significant Ca or Mg in the fluids, and yield final-melting temperatures of −1 to −20 °C, corresponding to salinities of up to 22 equivalent wt% NaCl (Fig. 6). Secondary inclusions in sphalerite and sparry dolomite, which are not included in Figure 6, have final melting temperatures of −3 to −14 °C, corresponding to salinities of ∼5–17 equivalent weight percent NaCl. Schaefer et al. (2001) reported a similar range of salinities from 27 to 1 equivalent wt% NaCl for inclusions at Bushy Park and suggested that it reflected the presence of a high salinity ore fluid that mixed with meteoric water to form fluids with intermediate salinities, which is in agreement with our observations. Greyling et al. (2001) reported a similar large range of salinities (but not freezing temperatures) for Pering, extending from only a few equivalent wt% NaCl to values as high as 50%, reflecting abundant Ca and Mg in the inclusions.
Na-Cl-Br compositions of fluid inclusion leachates from most Phanerozoic MVT minerals fall on or near the seawater evaporation line in plots of Na/Br versus Cl/Br, suggesting that their source brines formed by evaporation of seawater (Kesler et al., 1995, 1996; Viets et al., 1996; Chi and Savard, 1997; St. Marie and Kesler, 2000). Of the Precambrian MVT deposits and districts included in this study, Na-Cl-Br leachate data are available for Nanisivik (Viets et al., 1996) and Berg Aukas (Chetty and Frimmel, 2000) and are reported here for Bushy Park (Table 3). Two leachates from sphalerite at Nanisivik and one from sparry dolomite at Berg Aukas plot just above the seawater evaporation line, with Nanisivik farther along the evaporation trend (Fig. 7). Leachates from Bushy Park sphalerite and sparry dolomite plot in about the same location as Berg Aukas, also just above the seawater evaporation line (Fig. 7). The similarity in position of Precambrian and Phanerozoic MVT deposits in Figure 7 suggests that the Na-Cl-Br composition of seawater and MVT brine-forming process were similar throughout most of Earth history.
Sulfur Isotope Geochemistry
Sulfides in most of the Precambrian MVT deposits have high δ34S values that approach those of sulfate in coeval evaporites or seawater, a pattern that is consistent with the Anderson model applied here (Fig. 8). Best agreement between MVT sulfides and rock sulfate is seen at Nanisivik, where δ34S values of 21‰–31‰ for MVT sulfides are almost identical to δ34S values of 22‰–32‰ for evaporite gypsum from the Society Cliffs Formation (Olson, 1984; Ghazban et al., 1990). Ghazban et al. (1990) showed that δ34S values of the sulfides in Nanisivik ore could be accounted for by deposition from a fluid containing S with δ34S values of 26 ± 1‰ and suggested that the S was derived from seawater, and more recent data of Kah et al. (2001) show that it could have come directly from Society Cliffs gypsum. Unusually low δ13C values of 6‰ to −12‰ for sparry dolomite in the Nanisivik ore assemblages suggest that the sulfate was reduced by reaction with organic matter, providing strong support for the Anderson model (Ghazban et al., 1990).
Sulfides at Coxco (McArthur River) have δ34S values significantly lower than coeval seawater. δ34S values of 1.3‰–21.7‰ are lower than the 20‰–32‰ estimated for sulfate in the McAr-thur Group from analysis of trace sulfate in carbonate rocks and barite that probably replaced sedimentary sulfates (Walker et al., 1983; Bottomley et al., 1992). Second-stage sulfides at Coxco have lower values of 0.9‰–16.1‰, and sphalerite, galena, and pyrite from the Cooley and Ridge deposits have even lower δ34S values that are similar to those in the HYC sedex deposit (Walker et al., 1983; Rye and Williams, 1981). A possible source of S with the necessary low δ34 S values is pyrite or H2S with δ34S values near 0‰ in nearby Tawallah Group black shales (Shen et al., 2002).
Sulfides of possible MVT origin from Ramah have δ34S values of ∼8–34‰ (D. Wilton, 2002, written commun.). No S isotope values are available for sulfate in the sedimentary host rocks, although δ34S values estimated for seawater sulfate of Paleoproterozoic age range from ∼15‰ to 20‰, which falls within the range of MVT sulfide values (Fig. 8). Sulfur isotope data are not available for ore at Esker, but trace sulfate in the Rocknest Formation has δ34S values of ∼12‰–25‰ (Ueda et al., 1991). Sulfur isotope analyses have not been published for the Gayna deposits. δ34S values for gray-stage sphalerite at Warrabarty range from 1.5‰ to 20.4‰, a somewhat larger range than seen in the other deposits, but have a “distinct mode” at 11‰–14‰ (Smith, 1996). Uncertainty about the age of Warrabarty mineralization limits the degree to which it can be compared with coeval seawater sulfate, although the best comparison is probably with that for Berg Aukas.
δ34S values obtained in this study for Bushy Park sulfides (Table 4) range from 15.2‰ to 16.6‰ for galena and 15.8‰–19.7‰ with one unusually high value of 24.1‰ for sphalerite (Fig. 8). Schaefer et al. (2001) reported lower δ34S values of 4.8‰–8.5‰ for sphalerite and galena and −9.7‰–26.7‰ for diagenetic(?) pyrite from Bushy Park. Sulfate evaporites are not known in the Transvaal sequence, but δ34S values of 13‰–17‰ have been reported for trace sulfate in carbonate rocks from the Malmani Subgroup (Buchanan and Rouse, 1982, in Strauss, 1993), similar to the range estimated by Canfield and Raiswell (1999) for late Archean seawater. This range is very similar to most of our values for sulfides at Bushy Park (Table 4), suggesting that MVT sulfide was derived largely from coeval seawater sulfate. The lower δ34S values observed for Bushy Park sphalerite by Schaefer et al. (2001) probably reflect additions of S from diagenetic pyrite or related sources.
SIGNIFICANCE OF PRECAMBRIAN MVT DEPOSITS TO EARLY EARTH ATMOSPHERE AND HYDROSPHERE COMPOSITIONS
Comparison of Precambrian and Phanerozoic MVT deposits
Our results show that Precambrian and Phanerozoic MVT deposits share many similarities. Deposits of both ages are along the margins of passive-margin and rifted basins, and are hosted largely by platform carbonate rocks containing extensive reefs. With the exception of Ramah and Esker, most of the Precambrian MVT deposits are in undeformed sequences or foreland thrust belts, as are most of their Phanerozoic counterparts. Both Precambrian and Phanerozoic deposits have generally similar and simple mineralogy. Where ore mineralogy is more complex and includes elements such as copper, silver, and cobalt, these are interpreted to have resulted from local features that contaminated the ore fluid (Frimmel et al., 1996; Wu et al., 1997). Fluid inclusion observations, including homogenization temperatures and salinities, are also generally similar for deposits of both ages, and Precambrian exceptions seem to be just that rather than indications of systematic temporal changes. For instance, only Gayna River and Nanisivik, among the unmetamorphosed Precambrian MVT deposits discussed here, have significantly higher temperatures than those of Phanerozoic deposits (which probably reflect more proximal basinal sources or later dike-related heating), and even older Precambrian deposits such as Coxco and Bushy Park have temperatures typical of Phanerozoic deposits (Fig. 5).
More detailed grade-tonnage and frequency comparisons, however, indicate that the two groups differ. The grade-tonnage plot for Precambrian and Phanerozoic MVT deposits (Fig. 9) can be divided into four quadrants at a grade of 7% Pb+Zn and a tonnage of 10 million that results in a nearly equal number of Phanerozoic deposits in all four quadrants (29%, 24%, 26%, and 21% for NW, NE, SE, and SW quadrants, respectively). Precambrian deposits are distributed much differently among the quadrants, however, with none (0%) in the NW, 13% in the NE, 62% in the SE, and 25% in the SW (Fig. 9). Viewed only from the perspective of grade, only 13% of the Precambrian deposits fall above the 7% division compared with 53% of the Phanerozoic deposits. However, 75% of the Precambrian MVTs have more than 10 million tons compared with only 50% of the Phanerozoic deposits. Thus, Precambrian deposits are generally lower grade but higher tonnage, which should mean that they have generally similar metal contents. This is confirmed by the fact that ∼50% of the Precambrian deposits plot above the 1 million ton diagonal in Figure 9, compared with ∼41% of the Phanerozoic deposits. Considering all of the uncertainties involved, this suggests that the amount of metal moved by Precambrian and Phanerozoic MVT-forming systems was about the same, but that fewer Precambrian systems made ore.
Frequency comparisons indicate further than there were not many Precambrian systems. Figure 1 shows that ∼11 MVT deposits and districts formed during Precambrian time versus at least 30 during Phanerozoic time. In view of the large difference in duration of these time periods, this amounts to a very different indicated “MVT-formation rate” of ∼5.5 per billion years for Precambrian deposits and districts versus 60 per billion years for Phanerozoic deposits and districts. If Nanisivik is not Precambrian in age, as noted above, the Precambrian MVT-formation rate would be even lower. The age of ore-hosting sedimentary units varies greatly for both Precambrian and Phanerozoic rocks, although the significance of this variation is complicated by uncertainty about the exact age of mineralization. For instance, Cambrian rocks contain a significant majority of deposits that probably formed later in Paleozoic time (Leach et al., 2001). Within the Precambrian, MVT formation appears to have been greatest during late and early Proterozoic time, with only Nanisivik from the intervening period. These complications can be minimized by comparing the abundance of deposits in 0.5-b.y. groups that bracket the range of possible depositional ages for most of the deposits. As seen in Figure 10, a plot of this type shows a steep rise in MVT formation for the last 0.5 b.y. of Earth history and a much smaller and not greatly changeable rate during earlier (Precambrian) time.
Frequency comparisons of this type should be normalized to the amount of favorable carbonate-bearing sedimentary basins that remain from each of the time periods of interest, thus taking into account both formation of appropriate host rocks and their preservation during later events. For instance, carbonate rocks are scarce in most early Archean greenstone belts and related sedimentary basins; significant volumes of carbonate sediment began to form only by ca. 2.7–2.4 Ga when favorable shelf environments developed for the first time (Eriksson et al., 1998; 2001a, 2001b). Thus, the lower frequency for Precambrian MVT deposits might simply reflect a lack of suitable shelf carbonates in which ore could have formed. Although quantitative data are not available on the change in shelf carbonates through time, they should be related to continental growth rates because shelf environments formed on the margins of flooded continents. Continental growth rates are themselves matters of significant debate (Condie, 2000), but even those based on episodic growth produce relatively smooth cumulative curves for the change in total crust volume with time. As can be seen in Figure 10, growth rates of this type are not similar to our estimate of the formation of MVT deposits through Precambrian time. This is not surprising in view of the other important factors that control shelf carbonate development, especially sea level and climate (Walker et al., 2002).
In the absence of global data, more detailed comparisons of shelf carbonate volumes must be confined to specific areas. One of the largest areas of early Precambrian carbonate sedimentation was the Transvaal-West Griqualand-Kanye Basins of the Kaapvaal Craton in South Africa, where platform sediments covered an area of ∼600,000 km2 (Beukes, 1987). If carbonates of the Kaapvaal Craton correlate with those of the Jeerinah Formation in the Pilbara Craton of Australia to form the Vaalbara terrane, as suggested by Cheney (1996), this platform probably covered at least an additional 100,000 km2. MVT deposits in this sequence can be compared with two important shelf sequences in North America: (1) Cambrian–Ordovician platform sediments of the Appalachian Basin (USA) with an area of at least 100,000 km2, and (2) Devonian platform sediments of the Lennard shelf (Australia) with an area of ∼50,000 km2 (Kesler, 1996; Vearncombe et al., 1996). MVT mineralization is widespread in the Vaalbara of South Africa but absent from correlative rocks in Australia (Blockley and Myers, 1990). In the Appalachian province, MVT mineralization is found in all areas, although it is most abundant in the south and progressively less abundant northward (Kesler, 1996). In the Lennard shelf, MVT mineralization is found largely in two areas, at the north and south ends of the belt (Vearncombe et al., 1996).
The density of MVT mineralization in these Precambrian and Phanerozoic basins differs greatly, however. Looking only at reserves and production for deposits and districts that have been or probably will be mined, the Vaalbara province contains ∼1 million tons of Zn compared with ∼4 million tons for the Appalachian province and 4 million tons for the Lennard shelf. If MVT Zn-Pb and F-Ba districts, rather than deposits and prospects, are counted, the Vaalbara contains approximately six districts compared with at least 25 for the Appalachian province and about six for the much less extensive Lennard shelf. Thus, when viewed in terms of the areas, Phanerozoic shelf sequences clearly contain more MVT ore than their Precambrian analogues.
If the scarcity of Precambrian MVT deposits cannot be attributed to a lack of reef carbonate rocks, perhaps it is the porosity of the carbonates that limited Precambrian MVT mineralization. Low porosity would have limited the volume of MVT minerals that could be deposited by invading fluids, and therefore the grade of the deposit. This might be suspected because Phanerozoic-age carbonate reefs were built by a variety of organisms, whereas Precambrian reefs are almost exclusively stromatolitic (Webb, 2001; Petrov and Semikhatov, 2001; Grotzinger, 1986a, 1986b, 1986c; Hoffman, 1989; Ricketts and Donaldson, 1989). Systematic comparisons of porosity in stromatolitic reefs versus reefs of other types are not available, but anecdotal information that can be gleaned from literature on oil fields does not offer much encouragement for this line of thought. Recent studies of stromatolitic reef rocks of a range of ages include descriptions of oil and MVT mineralization, as well as porosity estimates of 8%–11% (Counter, 1993; Whitesell, 1995; Kuznetzov, 1997; Grotzinger, 2000; Lemon, 2000; Osmond, 2000). Porosities at and below the low end of this range are sufficient to produce a deposit containing 7% Pb+Zn regardless of the relative abundances of galena and sphalerite.
These comparisons argue against the tempting possibility that the scarcity of Precambrian MVTs can be attributed solely to a lack of suitable Precambrian carbonate rocks. Furthermore, by including in this compilation several Precambrian MVT districts that have been explored but not produced, we have also minimized the likelihood that less extensive exploration in Precambrian terranes can account for their lower MVT endowment. Thus, we conclude that Precambrian MVT deposits are indeed less abundant than Phanerozoic deposits and must seek a reason beyond local geologic factors for this difference.
Relation between Atmosphere-Hydrosphere Compositions and MVT Mineralization
With no geologic reasons to account for the global MVT history shown in Figure 10, the pattern likely reflects changes in the composition of the atmosphere and hydrosphere. As noted at the outset of this study, the efficiency of MVT mineralization should be directly related to the sulfate content of basinal fluids and, by extension, the ocean in which the sediments were deposited. Sulfate in the ocean is linked, in turn, to the composition of the atmosphere and, especially, its oxygen content.
Two competing histories have been suggested for the evolution of the ocean and coexisting atmosphere. One history holds that the oxygen content of the atmosphere increased by as much as an order of magnitude from levels of less than 10−2 present atmospheric level (PAL) during the Great Oxidation Event (GOE) ca. 2.3 Ga (Holland, 1999; 2002; Bekker et al., 2004). Prior to the GOE, sulfate concentrations in the ocean are estimated to have been less than 1% of present levels and the main source of sulfate was disproportionation or photochemical oxidation of volcanic SO2 and oxidation of H2S by cyanobacteria (Hattori and Cameron, 1986; Canfield and Raiswell, 1999; Farquhar et al., 2000; Habicht et al., 2002; Holland, 2002; Farquhar and Wing, 2003). After the GOE, sulfate concentrations in the ocean are thought to have increased gradually through Proterozoic time, reaching present levels of 10–30 mM only near the end of Proterozoic time (Shen et al., 2002; Kah et al., 2004). The alternative view holds that there was no GOE and the oxygen content of the atmosphere and the sulfate content of the oceans have fluctuated between 10 and 30 mM since at least Neoarchean time (Ohmoto, 1997, 2004).
The two competing marine sulfate histories are shown in Figure 10, where it can be seen that our curve for the change in MVT mineralization through time nearly parallels the GOE-type curve. Similarity of the MVT and GOE-type seawater sulfate curves suggests that sulfate concentrations similar to those in the modern ocean are required for efficient MVT mineralization, and that a scarcity of sulfide-rich traps derived from this sulfate limited the formation of MVT deposits in Precambrian time. Evaporative brines derived from low-sulfate seawater were probably capable of dissolving and transporting metals in much the same amounts observed in Phanerozoic basins. However, low sulfate concentrations in basinal waters would have limited the amount of sulfide that could have formed from such brines (or their rare evaporites), whether by bacterial or thermochemical reduction. Local accumulations of sulfate-rich seawater, such as those proposed for Archean and Neoproterozoic time, might have been the necessary precursors to sulfide-rich traps for early MVT ore formation (Cameron, 1982, 1983; Kasting, 1991). In that context, it is interesting to note that some Precambrian MVT deposits, especially Nanisivik and Kamarga, were close to evaporite sequences, a factor that might have led to higher sulfate and derived sulfide concentrations.
Similarity of these curves does not exclude other possible controls on MVT genesis. Christensen et al. (1997) and Leach et al. (2001) have pointed out that Phanerozoic MVT deposits formed largely during Devonian and Permian time, possibly because of long-distance brine migration during episodic compressive or extensional tectonic events (Bradley and Leach, 2003; Kesler et al., 2004). In addition, the sulfate content of the world ocean during Phanerozoic time varied considerably in response to large-scale continent aggregation and disaggregation, although it did not reach levels as low as those estimated for Paleoproterozoic and Neoarchean time (Horita et al., 2002; Lowenstein et al., 2003). All of these factors operate at timescales considerably smaller than the 500 Ma interval used in Figure 10, however, and it appears that they were masked by a longer-term increase in the sulfate content of the ocean during Precambrian time.
Although the similarity of the MVT and GOE-type sulfate curves agrees with the “GOE-type history” for Earth's early atmosphere and hydrosphere, perplexing complications remain. In particular, the Neoarchean–Paleoproterozoic Transvaal Basin, which is significantly older than the GOE, generated MVT deposits very similar to those in Phanerozoic rocks. Perhaps these deposits, which probably formed ca. 2.0 Ga, took advantage of GOE-related sulfate. Or, maybe they did not form by the Anderson-type model advocated here. Neither of these alternatives seems geologically reasonable, however, and it is much more likely that the basinal brines that formed Bushy Park and Pering were part of the Transvaal sedimentary system and therefore representative of Neoarchean–Paleoproterozoic conditions. Thus, enough sulfate to form MVT deposits was almost certainly available, even at this early stage of Earth history. Whether this was sulfate that formed because of local environmental factors or these deposits are the sole remaining MVT signal of a more sulfate-rich early Earth remains to be determined.
Finally, although basin-hosted deposits are most likely to reflect atmosphere and ocean compositions, our results provide encouragement for study of other subsurface deposits. Deeper deposits related to igneous activity might also be useful. For instance, the oxidation state of magmas controls the form of S in the magma, which controls, in turn, whether metals form immiscible sulfides in the magma or separate into a vapor phase that could form hydrothermal ore deposits. Although current estimates suggest that fO2 of the mantle has remained within a single log-unit of present values (Canil, 1997; Delano, 2001; Holland, 2002), even these small differences can produce large changes in the SO2/H2S ratio of magmas and associated hydrothermal solutions (Core, 2004).
We are grateful to Derek Wilton, Nawojka Wachowiak, Bruce Ehlers, Peter McGoldrick, and Bruce Gemmell for sharing information with us on Ramah, Esker, Bushy Park, Kamarga, and Warrabarty, respectively. The manuscript has been improved by conversations with B.H. Wilkinson and Linda Kah, and challenging reviews by H.L. Barnes, M.B. Goldhaber, and H. Ohmoto.
Figures & Tables
Evolution of Early Earth's Atmosphere, Hydrosphere, and Biosphere - Constraints from Ore Deposits
- fluid inclusions
- host rocks
- isotope ratios
- lead ores
- lead-zinc deposits
- metal ores
- mineral deposits, genesis
- mississippi valley-type deposits
- ore grade
- ore-forming fluids
- South Africa
- South America
- Southern Africa
- stable isotopes
- Transvaal Supergroup
- zinc ores