Mineralogical and geochemical data for ca. 1720 Ma Si-Fe-Mn seafloor-hydrothermal sedimentary rocks (exhalites) near the Jones Hill Zn-Cu-Pb-Ag-Au volcanogenic massive sulfide (VMS) deposit, northern New Mexico, provide valuable insights into the redox state of late Paleoproterozoic deep sea-water. Distal exhalites ~1200 m south of the deposit form beds 0.5–2 m thick composed of interlayered iron formation and metachert. The iron formation consists mostly of quartz and magnetite, and includes 0.3–3-cm-thick laminae of fine-grained garnet-quartz rock, which in places contains as much as 9.4 wt% MnO that resides chiefly in spessartine-rich garnet (coticule). Shale-normalized rare earth element data for an unaltered, low-Al quartz-magnetite iron formation show no Ce anomaly, which rules out fully oxic deep waters during exhalative mineralization. The garnet-quartz rocks and coticules mostly have small positive Ce anomalies, which are larger for calculated detrital-free compositions, thus precluding deposition in anoxic waters. Significant amounts of ferric iron are inferred for protoliths of the iron formation, based on the presence of abundant magnetite laminae, and of magnetite inclusions in cores of the spessartine garnets. Protoliths of the garnet-quartz rocks and coticules probably consisted largely of clays and Fe-Mn oxyhydroxides. Together these mineralogical and geochemical data suggest that the Jones Hill exhalites were deposited from deep sea-water having low concentrations of dissolved O2 corresponding to suboxic conditions, and not the sulfidic conditions proposed for late Paleoproterozoic deep seawater by other workers. Exhalites associated with Cu-rich VMS deposits, when effects of alteration and detrital components are considered, can be important proxies for evaluating the evolving redox state of ancient deep oceans.


Understanding the redox state of seawater during Precambrian time is important because it affected the evolution of early life on Earth and the secular distribution of banded iron formation (BIF) deposits (e.g., Anbar and Knoll, 2002; Huston and Logan, 2004; Holland, 2006; Canfield et al., 2007, 2008; Scott et al., 2008; Sleep and Bird, 2008). However, models for the redox state of Precambrian deep oceans remain controversial. Holland (1984) proposed a model in which anoxic conditions prevailed until ca. 1.8 Ga, becoming oxic thereafter. Canfield (1998) suggested a transition ca. 1.8 Ga from anoxic to sulfidic conditions, the latter similar to the redox state of the lower part of the modern Black Sea. Both models explain the end of BIF deposition ca. 1.8 Ga, either by deep-ocean oxidation of soluble ferrous to insoluble ferric iron, which prevented the transport of hydrothermally derived ferrous iron from ocean basins to continental margins and subsequent formation of oxide facies BIFs (Holland, 1984), or by precipitation of most of this ferrous iron in the water column as iron sulfide minerals as in the modern Black Sea, thus also terminating BIF deposition (Canfield, 1998). The Canfield model has been supported by diverse studies of Proterozoic black shales and related strata, using data for iron speciation, sulfur and molybdenum isotopes, and molecular biomarkers (Lyons et al., 2000; Shen et al., 2002; Poulton et al., 2004; Arnold et al., 2004; Brocks et al., 2005). A third model by Ohmoto and colleagues (e.g., Ohmoto et al., 2006) infers that the deep oceans have been oxic since 3.8 Ga. More recently, Slack et al. (2007) suggested that deep seawater from 1.74 to 1.71 Ga was at a transitional, suboxic state, with low concentrations (<5 μM) of dissolved O2 but no H2S, based on mineralogical and geochemical data for volcanogenic massive sulfide (VMS) related exhalites in central Arizona, mainly from the 1.74 Ga Jerome district. Such exhalites are generally interpreted as seafloor precipitates from hydrothermal plumes, having formed as integral parts of VMS systems (Isley, 1995; Spry et al., 2000; Peter et al., 2003; Grenne and Slack, 2005).

Our current study evaluates the redox state of deep seawater recorded in exhalites from a late Paleoproterozoic (ca. 1.72 Ga) volcanosedimentary sequence in the Pecos greenstone belt of northern New Mexico, testing whether the suboxic state was limited to selected marine basins in Arizona or was more widespread geographically and temporally. The Pecos exhalites are ~20 m.y. younger than those in the Jerome district, and their oceanic depositional position relative to the Laurentian craton, and to each other in the ocean, are unconstrained. Both volcanosedimentary sequences formed at unknown distances from the southern margin of Laurentia, prior to their accretion to this craton during the 1.71–1.68 Ga Yavapai orogeny (e.g., Whitmeyer and Karlstrom, 2007). The present distance of ~600 km between the Jerome district and the Pecos area therefore does not reflect their original position in the late Paleoproterozoic ocean. Hence, our data for the Pecos exhalites provide insights into the redox state of deep seawater for a different time period and a different location, compared to those for the Jerome district exhalites described in Slack et al. (2007).


Detailed mineralogical studies were done at the U.S. Geological Survey in Reston, Virginia, using transmitted and reflected light microscopy, and a JEOL JSM-840 scanning electron microscope (SEM) operated at 15 kV and a beam current of 40 nA. This instrument has a LaB6 electron emitter equipped with an EDAX energy-dispersive X-ray analytical system (EDS), which provides qualitative and semi-quantitative elemental data. Garnet and amphibole compositions were determined at the Geological Survey of Norway in Trondheim with a Leo 1450VP SEM-EDS instrument operated at 15 kV and a beam current of 10 nA, using both natural and synthetic mineral standards.

Samples for geochemistry were carefully trimmed of weathered surfaces and veins and pulverized in an alumina-ceramic mortar. All analyses were done at Activation Laboratories in Ancaster, Ontario. For most elements, rock powders were fused with lithium metaboratetetraborate in order to insure complete acid dissolution of minerals such as zircon, monazite, xenotime, and barite prior to analysis. Major, trace, and rare earth elements (REE) were determined on fused samples using inductively coupled plasma–mass spectrometry (ICP-MS). Instrumental neutron activation analysis was used for Au, Co, Sb, As, Se, Sc, and Cr. Determination of Be, V, Cd, Ni, Cu, Zn, Ag, and Pb was done by ICP-optical emission spectrometry, following a four-acid digestion of powders. All analyses were obtained using duplicate samples and 8–12 standards. Details of the various analytical methods are available at www.actlabs.com.


The Pecos greenstone belt in northern New Mexico contains a thick sequence of late Paleoproterozoic metavolcanic and metasedimentary rocks and contemporaneous subvolcanic intrusions (Fig. 1). Petrologic studies by Robertson and Condie (1989) indicate that protoliths of these rocks were mainly tholeiitic and calc-alkaline basalt, dacite, and rhyolite; volcaniclastic and siliciclastic sediments; and intrusions of tonalite and diabase-gabbro. Bowring and Condie (1982) obtained U-Pb zircon ages of 1720 ± 15 Ma for a “quartz-eye” rhyolite porphyry (a probable volcanic dome) from the Jones Hill mine area and 1718 ± 5 Ma for a sub-volcanic tonalite exposed along the Pecos River 5 km to the east. Trace element, REE, and Nd isotopic signatures of the volcanic rocks suggest formation in an evolved island arc or a backarc basin (Robertson and Condie, 1989). The volcanosedimentary and subvolcanic intrusive rocks of the greenstone belt show regional metamorphic assemblages of the greenschist to lower amphibolite facies, except in areas affected by high-temperature contact metamorphism from post–1700 Ma granitic intrusions.


The Pecos greenstone belt contains several VMS deposits and occurrences that form strat-abound lenses within late Paleoproterozoic metavolcanic and minor metasedimentary rocks (Fig. 1). Such deposits formed on or near the ancient seafloor by synvolcanic hydrothermal processes (e.g., Franklin et al., 2005). Recorded production has come only from the Pecos Zn-Pb-Cu-Ag-Au deposit, which from 1905 to 1939 was mined for ~2.0 Mt of ore at an average grade of 12.95% Zn, 4.00% Pb, 0.078% Cu, 3.4 ppm Ag, and 0.011 ppm Au (Krieger, 1932a, 1932b; Riesmeyer and Robertson, 1979; Thompson and McLemore, 1999). The larger Jones Hill Zn-Cu-Pb-Ag-Au deposit (Riesmeyer, 1978; Robertson and Fulp, 1987; Thompson and McLemore, 1999), 6.5 km to the west, was mined on a small scale in the 1930s but no production figures are available. Exploration drilling in the 1980s outlined two mineralized zones, an upper zone containing 5 Mt at 4.60% Zn, 2.47% Cu, and 0.072 ppm Au, and a lower zone containing 7 Mt at 3.81% Zn and 4.44% Cu (Thompson and McLemore, 1999). Figure 2 shows a generalized stratigraphic reconstruction of the Jones Hill area, which is structurally complex due to Proterozoic folding and Tertiary faulting. The deposit consists of massive to disseminated sulfides composed of sphalerite, chalcopyrite, galena, pyrite, and minor pyrrhotite and arsenopyrite, with lesser tetrahedrite, bornite, argentite, molybdenite, proustite, and gold; gangue minerals are mainly quartz, sericite, talc, calcite, and fluorite (Thompson and McLemore, 1999).


The Jones Hill VMS deposit has both proximal (formed near the hydrothermal vent) and distal exhalites. Stratigraphically above the deposit are carbonate-rich lenses as much as several meters thick composed mostly of calcite, chlorite, and quartz. Coeval exhalite units, exposed at approximately the same stratigraphic level 1200–1500 m to the south, consist of beds 0.5–2 m thick that can be traced along strike for ~100 m. Such stratigraphic continuity suggests deposition from a hydrothermal plume instead of from isolated diffuse vents (Grenne and Slack, 2005). The distal exhalites comprise metachert and interbedded iron formation, the latter composed of quartz + magnetite ± garnet ± Fe-amphibole ± chlorite ± apatite ± ilmenite. Reconnaissance electron microprobe analyses indicate that the Fe-amphibole is predominantly hornblende with minor intergrown grunerite. The magnetite-rich layers vary in thickness from 1 mm to 5 cm, and commonly alternate with layers of microcrystalline quartz. Carbonate is absent except in late veinlets. Some samples contain laminae 0.1–0.3 mm thick composed of fine-grained (10–50 μm) magnetite and quartz. Sulfide minerals are rare in these distal exhalites and are visible only microscopically as very small (<10 μm) grains of pyrite or pyrrhotite encased in quartz. The rarity of sulfides and their small grain sizes are consistent with precipitation >1 km from a hydrothermal vent, in contrast to near-vent (proximal) exhalites derived from modern seafloor-hydrothermal systems that in many cases contain minor to abundant, relatively coarse grained sulfides (e.g., Feely et al., 1994; German and Von Damm, 2003).

Some layers of quartz-magnetite iron formation have laminae 3 mm to 3 cm thick composed of fine-grained garnet + quartz ± Fe-amphibole ± apatite ± ilmenite. The garnet grains are mostly 10–30 μm in diameter and constitute ~60–70 vol% of the laminae. SEM imaging shows that these garnets typically are unzoned, with uncommon grains having darker (in backscattered electron images) growth zones occurring between lighter cores and rims. Electron microprobe analyses (Table 1) show that compositions of the garnets vary greatly among samples. The fine-grained garnet-quartz rocks are divided into two types, one containing predominantly almandine (66.1–73.7 mol%) with minor spessartine (11.8–12.5 mol%) and grossular (4.59–15.1 mol%), whereas the other type has a smaller content of almandine (42.7–59.1 mol%) with a major spessartine component (27.7–34.2 mol%) and a minor grossular component (2.83–25.2 mol%). The latter fine-grained, Mn-rich garnet-quartz rock is termed “coticule” (e.g., Spry, 1990). Disseminated garnet in the iron formation in contrast is unzoned, with subequal almandine and spessartine (39.0–41.1 mol% and 35.2–37.6 mol%, respectively) and lesser grossular (22.4–24.3 mol%). SEM imaging shows that these iron-formation garnets contain abundant inclusions of quartz + magnetite ± apatite (Fig. 3), whereas garnets in the garnet-quartz rocks and coticules have inclusions of quartz ± apatite ± Fe-amphibole; no carbonate inclusions were found in the garnets despite a detailed search using SEM. Para-genetically late hematite occurs in thin (<1 mm) veinlets and as variable replacements of magnetite grains in several samples of iron formation; the hematite is clearly a secondary feature produced after metamorphism, but whether it formed from postmetamorphic fluids or by surficial weathering (or both) is unknown. Sample JS-06-38A, a quartz-magnetite iron formation, lacks secondary hematite, as do all of the garnet-quartz and coticule samples.


The distal exhalites south of the Jones Hill VMS deposit show a wide range of compositions (Table 2202). Iron concentrations vary from 22.0 to 60.0 wt% Fe2O3T in the iron formations; a magnetite-rich layer cut from one of the iron formations has 90.4 wt% Fe2O3T. The three analyzed coticules contain 4.24–9.36 wt% MnO. Overall, the exhalites display a substantial variation in Mn/(Mn + Fe) ratios, from <0.01 to 0.45. Large ranges in concentration exist for Al2O3 (0.50–11.81 wt%), TiO2 (0.02–0.58 wt%), Sc (0.6–14.8 ppm), Zr (6–146 ppm), and Th (0.18–7.14 ppm), the higher abundances reflecting major detrital components in the samples (Table 2). A Ti-Al-Th plot (Fig. 4A) of the whole-rock data is consistent with the major detrital component being pelagic clay. Other ternary plots such as Ti-Sc-Th and Ti-Sc-Zr (not shown) rule out a significant component of locally derived volcaniclastic material. On a Th-Sm-Al plot (Fig. 4B), data for most of the exhalites are shifted toward the Al apex, relative to average pelagic clay, which we interpret as a component of hydrothermally derived Al (e.g., Alt and Jiang, 1991; Elderfield et al., 1993; German and Von Damm, 2003). The impure metachert (sample JS-06-42B) has a different detrital component, likely dominated by dacitic volcanic detritus. Figure 4B also documents an enrichment of Sm relative to local volcanic rocks of the Pecos greenstone belt and to average shale and average pelagic clay, reflecting the addition of a variable hydrothermal REE component to protoliths of the exhalites. Significant hydrothermal components in the magnetite iron formations and garnet-quartz rocks are similarly recorded by Fe-Al-Mn and Fe/Ti vs. Al/(Al + Fe + Mn) plots (Figs. 5 and 6).

All samples contain uniformly low Co (≤61 ppm) and Ni (≤35 ppm), and fall within the hydrothermal field on a ternary Fe-(Co + Ni)-Mn diagram (Fig. 7); none of the Mn-rich samples plots within the hydrogenous (manganese nodule) field, or in the diagenetic field. Concentrations of Cu are relatively low in the exhalites (7–62 ppm), except for one sample of quartz-magnetite-garnet iron formation that has 130 ppm; Zn contents are more variable, ranging overall from <1–350 ppm, whereas Pb is uniformly low (≤12 ppm). All samples have <50 ppm S.

REE data for the Jones Hill area exhalites display abundances mainly from ~0.2–0.8× post-Archean Australian average shale (PAAS) and PAAS-normalized La/Yb ratios of ~0.3–0.8 (Figs. 8A, 8B; Table 2). By comparison, average compositions of late Paleoproterozoic metarhyolite and calc-alkaline metabasalt in the Pecos greenstone belt, and of modern pelagic clay and subducting sediment, show variable patterns in terms of REE abundance levels and the presence or absence of Eu anomalies (Fig. 8C). VMS-related iron formations in the 1738 Ma Jerome district of Arizona have REE patterns similar to those of the Jones Hill area exhalites (Fig. 8D), but most are at lower abundance levels owing to minimal detrital material and dilution by greater amounts of quartz (Slack et al., 2007). For comparison, exhalites associated with other Proterozoic and Archean seafloor-hydrothermal systems show diverse REE abundances and La(SN)/Yb(SN) ratios (Fig. 8E). The REE patterns of such exhalites represent variable contributions from hydrothermal fluids, seawater, and detrital materials, including redox signatures recorded by Ce and Eu anomalies (Bau, 1993; Peter and Goodfellow, 1996).

Europium and Ce anomalies occur in most of the Jones Hill area samples (Figs. 8A, 8B), calculated using the formulas listed in Table 2 and normalized to PAAS (Taylor and McLennan, 1985). Small to moderate positive Eu anomalies reflect a component of reduced, high-temperature hydrothermal fluid (German et al., 1993; German and Von Damm, 2003; Chavagnac et al., 2005). The Ce anomalies record an integrated redox state, including that of ambient deep seawater influenced by adsorption onto Fe-Mn oxyhydroxide particles (e.g., Sherrell et al., 1999), detrital material, and postdepositional processes. All of the Ce anomalies of the exhalites are valid using the discrimination criterion of Bau and Dulski (1996), which involves systematics of Ce/Ce* and Pr/Pr* values (Table 2), and on this basis may be used to interpret the redox state of deep seawater during exhalite mineralization (Slack et al., 2007). However, meaningful use of the Ce anomaly in such exhalites requires careful evaluation of possible effects of alteration and weathering, and of detrital components (Bau, 1993). Two samples of altered magnetite iron formation, containing significant amounts of secondary hematite after magnetite, have small negative Ce anomalies (Table 2; Fig. 8B). Both samples (JS-06-39H and JS-06-39I) lack clearly positive Eu anomalies (Eu/Eu*SN > 1.1), which suggests postdepositional alteration given their major exhalative components (90.4 and 60.0 wt% Fe2O3T, respectively) and minimal detrital components (~2.3% and 1.2%). One sample of garnet-bearing magnetite iron formation (JS-06-41F) is distinctive in showing a moderately large positive Ce anomaly (2.39), but this sample also contains appreciable secondary hematite. Such Ce anomalies, especially positive anomalies, are well-documented products of oxidative surface weathering (e.g., Braun et al., 1990; Mongelli, 1993). REE data for these three samples therefore are not considered further in our study, because of likely diagenetic or later modification of original Ce anomalies (German and Elderfield, 1990; Bau, 1993; Pattan et al., 2005).


Analogy with hydrothermal sediments from the modern seafloor (e.g., Grenne and Slack, 2005) suggests that protoliths of the distal Jones Hill area exhalites were composed mainly of iron oxyhydroxide, silica, and one or more Mn-rich phases. The presence in these exhalites of abundant magnetite and thus high Fe3+/(Fe3+ + Fe2+) ratios (~0.67 in pure magnetite) implies that significant amounts of ferric iron existed in the primary sediments, as in the X-ray amorphous iron oxyhydroxide that characterizes Fe-rich precipitates from modern hydrothermal plumes (Hrischeva and Scott, 2007, and references therein). A similar protolith is widely accepted for ancient oxide facies iron formations (e.g., Bau and Dulski, 1996; Grenne and Slack, 2005). At 2 °C and pH of ~8, which are conditions typical of modern deep seawater, thermodynamic data indicate that iron oxyhydroxides form only above the Fe2+/Fe(OH)3 redox boundary (Glasby and Schulz, 1999); precipitation of ferrihydrite, goethite, and other crystalline ferric oxides such as hematite under these low-temperature conditions is kinetically restricted (e.g., Schwertmann et al., 1999). Primary hematite in modern seafloor- hydrothermal systems is rare, reflecting deposition from moderate-temperature (>115 °C) vent fluids with low ΣS concentrations (Hein et al., 2008). Although the water depth at which the Jones Hill VMS deposit formed is unknown, its chalcopyrite-rich (>1% Cu) assemblage suggests high-temperature (≥300 °C) deposition at a minimum water depth of 850 m (Slack et al., 2007), which is below the depths of photic zones (~200 m) and continental shelves (~600 m).

In anoxic Fe2+-rich fluids with very low ΣS and ΣCO3 contents, an increase in pH will precipitate the mineral amakinite [Fe(OH)2], which during later dehydration and deprotonation may convert to magnetite, in some cases including the intermediate formation of a mixed ferrous-ferric (“green rust”) compound (Murray, 1979). Such processes raise the possibility that magnetite in the Jones Hill area exhalites formed in anoxic seawater without a ferric oxyhydroxide precursor. However, amakinite does not precipitate from modern seawater or interstitial pore fluids, because saturation of FeCO3 is uniformly reached at a lower Fe concentration than is Fe(OH)2 saturation (Murray, 1979). Moreover, for a deep seawater composition at 2 °C, Fe(OH)2 is thermodynamically stable only below a pH of ~7.5 (Stumm and Morgan, 1996; Glasby and Schulz, 1999). The composition and pH of the late Paleoproterozoic deep ocean are poorly constrained, hence the possibility of an amakinite precursor to the Jones Hill magnetite cannot be rigorously evaluated. Evidence from Precambrian carbonates nevertheless offers some insights. Sedimentary textures in Archean and Paleoproterozoic carbonates, carbon isotope data for in situ analyses of Proterozoic microfossils, as well as climate models suggest high atmospheric pCO2 and bicarbonate saturation in seawater during early Precambrian time (Grotzinger and Kasting, 1993; Grotzinger and James, 2000; Sleep and Zahnle, 2001; Kaufman and Xiao, 2003). On this basis, we infer that amakinite is an unlikely precursor for Jones Hill magnetite in the late Paleoproterozoic ocean because seawater and diagenetic fluids then were probably saturated in FeCO3 before Fe(OH)2 saturation was reached.

Based on the mineralogy of modern and Holocene Mn-rich submarine deposits (Hein et al., 1997; Astakhov et al., 2006), the precursor of the coticule laminae may have consisted of an aluminous mineral such as kaolinite plus either Mn-rich oxyhydroxide or Mn-rich carbonate, which reacted to form spessartine during low-grade metamorphism (Grapes, 1978; Schreyer et al., 1992; Spry et al., 2000; Nyame, 2001). In modern plumes, hydrothermal Mn2+aq is adsorbed onto ferric oxyhydroxide, or coprecipitates with it as an amorphous manganese oxyhydroxide phase (Hrischeva and Scott, 2007). Mn- and Fe-rich oxyhydroxide precursors to the Jones Hill exhalites could have formed together in deep seawater under suboxic to oxic conditions, whereas precipitation of a Mn-rich carbonate precursor from seawater would be precluded due to insufficient [Mn2+aq] in the bottom waters, unless local Paleoproterozoic deep seawater had an Mn concentration several orders of magnitude higher than that of modern deep oceans (see Glasby and Schulz, 1999). Mn-rich carbonates in modern marine settings are restricted to anoxic sediments with pore fluids that typically have high alkalinities and high Mn2+aq concentrations, which favor precipitation of Mn-rich carbonate during early diagenesis (Calvert and Pedersen, 1993; Schulz et al., 1994; Glasby and Schulz, 1999). Because the Jones Hill iron formation is hosted by rhyolite and felsic volcaniclastic rocks, and not by organic-rich sedimentary rocks (Fig. 2), an Mn-rich carbonate protolith for the coticules is unlikely. This interpretation is consistent with the lack of carbonate inclusions in garnet crystals within the coticules.

Excluding likely altered and/or weathered samples, the only iron formation that is considered valid for evaluating the redox state of the ~1720 Ma deep ocean is JS-06-38A (Fig. 8B). This sample, containing ~5% clastic detritus (Table 2), has 1.77 wt% Al2O3, 1.5 ppm Sc, and a small positive Eu anomaly (Eu/Eu* = 1.18) but no Ce anomaly (Ce/Ce* = 1.02). The very low Sc content of this sample, in particular, makes it useful as a paleoredox proxy, relative to the Sc concentration of 2 ppm proposed by Bau (1993) as a maximum for such purposes. Other unaltered Jones Hill area samples of garnet-quartz rock and coticule show mostly small positive Ce anomalies, but all of these samples have high concentrations of Al2O3 (10.35–11.81 wt%) and Sc (12.1–14.8 ppm), and hence large amounts of detrital material (~28%–49%; Table 2). Because the Ce and other REE contents for these samples are probably greatly influenced by their detrital components, they cannot be used directly as paleoredox indicators (e.g., Bau, 1993). However, REE data calculated on a detrital-free basis are still useful for this purpose. Such calculations for the two garnet-quartz rocks, using PAAS as the detrital component and Th as the normalizing element, show moderately large positive Ce anomalies (1.37 and 2.06), which are greater than those of the uncorrected detrital-rich bulk samples (1.14 and 1.13). Larger positive Ce anomalies also occur in the corrected detrital-free portions of the coticules, relative to their bulk compositions. These moderately large positive Ce anomalies cannot reflect deposition in anoxic seawater, but instead record exhalative mineralization under suboxic conditions as inferred previously for the Jerome exhalites (Slack et al., 2007).

Figure 8E displays REE patterns for several exhalites associated with other Precambrian seafloor-hydrothermal sulfide deposits. In this plot, data are shown only for exhalites that are well documented in terms of depositional setting, especially the redox state of coeval bottom waters. VMS-related coticules typically have no or negative Ce anomalies (Spry et al., 2000), but redox conditions during deposition of the coticule protoliths are poorly constrained. To our knowledge, only coticules at the 1470 Ma Sullivan Pb-Zn-Ag deposit in British Columbia, Canada, have an independently established depositional redox state. Based on sulfur isotope data and organic C/S ratios of graphitic metasedimentary host rocks, Goodfellow (2000) determined that the bottom waters during mineralization at Sullivan were anoxic. Sullivan coticules lack Ce anomalies and are otherwise compositionally similar to those from the Jones Hill area (Slack et al., 2000). However, spessartine-rich garnets in the Sullivan coticules contain carbonate inclusions, which together with the graphitic nature of the host metasedimentary rocks, suggest a precursor composed dominantly of Mn-Fe carbonate (Slack et al., 2000), and not the Mn-Fe oxyhydroxide proposed here for the major precursor of the Jones Hill coticules. Among VMS-related Precambrian exhalites, only sulfide facies iron formation or sulfidic chert can be confidently assigned to seafloor anoxic depositional conditions. Modern, high-quality REE analyses for such samples are limited. Two examples of such sulfide facies exhalites, from the 1892 Ma Snow Lake VMS district of Manitoba and the 2725 Ma Bell Allard Zn-Cu-Ag VMS deposit in the Mattagami district of Quebec (Fig. 8E), both have low La(SN)/Yb(SN) ratios like modern seawater (Fig. 8F), but no Ce anomalies. The lack of Ce anomalies in these two Canadian exhalites is consistent with their formation under anoxic conditions, based on the absence of Ce anomalies in anoxic seawater of the modern Cariaco Basin (De Baar et al., 1988).

Oxic bottom waters of the modern ocean are characterized by large negative Ce anomalies (Fig. 8F) due to the strong fractionation of sea-water Ce into Mn nodules and Fe-Mn hydrogenous crusts (e.g., Byrne and Sholkovitz, 1996; Alibo and Nozaki, 1999). The Ce depletion in oxic seawater is partly reflected in Fe-Mn hydrothermal deposits, but the hydrothermal deposits have smaller negative Ce anomalies relative to seawater owing to preferential adsorption of Ce onto the oxyhydroxide particles in the hydrothermal plume (Barrett and Jarvis, 1988; Sherrell et al., 1999). The magnitude of these Ce anomalies is kinetically influenced (Bau, 1999), reflecting the rate and duration of adsorption of Ce from seawater. A small positive Ce anomaly with respect to that of surrounding oxic seawater records rapid precipitation and minimal Ce adsorption, whereas a large positive anomaly reflects slower precipitation and correspondingly greater Ce adsorption from seawater.

The lack of a negative Ce anomaly in unaltered exhalite JS-06-38A from the Jones Hill area (Fig. 8B) suggests that coeval seawater did not have the large negative Ce anomalies that characterize modern oxic ocean waters. Thus it is unlikely that the precursor of the Mn-rich spessartine in the associated coticule layers was an Mn oxide phase that precipitated directly from the hydrothermal plume or sea-water, because the Mn2+/MnO2 redox boundary occurs at a higher Eh than the Ce3+/CeO2 boundary (Byrne and Sholkovitz, 1996). These data, together with the presence of abundant magnetite in the iron formations, further suggest that the bottom waters during mineralization were not sulfidic, but suboxic, that is, above the Fe2+/redox boundary but below the Mn2+/Fe(OH)3 and Ce3+ MnO2/CeO2 boundaries (see De Baar et al., 1988; Glasby and Schulz, 1999). This proposed suboxic state of deep ocean waters is most consistent with a precursor formed from ferric oxyhydroxide plume particles that adsorbed appreciable amounts of hydrothermal Mn2+aq. Foustoukos and Bekker (2008) proposed a different model in which oxide facies iron formation of Archean age formed in anoxic deep seawater by the oxidation of Fe2+ during phase separation of hydrothermal fluids, associated with the development of shallow ridge crests and related high heat flow. In Archean oxide facies, Algoma-type iron formations (e.g., Gross, 1995), positive Ce anomalies are uncommon (Bau and Dulski, 1996; Klein, 2005; Kato et al., 2006), and those that are present may have formed after deposition during oxidative surface weathering (cf. Braun et al., 1990). We therefore infer that protoliths of magnetite iron formation in the Jones Hill area formed by partial oxidation in the water column of hydrothermally derived Fe2+. The lack of carbonate in the Jones Hill exhalites and of graphite in the host metavolcanic rocks suggests that the ferric oxhydroxides that likely dominated protoliths of these exhalites were transformed into magnetite prior to diagenesis and metamorphism, through non-redox reactions involving low-temperature (<200 °C), Fe2+-rich and H2S-poor fluids (see Ohmoto, 2003).

The abundance of ferric iron and manganese in the exhalite units near the Jones Hill deposit and their REE systematics indicate that ca. 1720 Ma the redox state of the deep ocean in this oceanic arc setting was suboxic, not sulfidic, as proposed in the model of Canfield (1998). This and subsequent studies (Lyons et al., 2000; Shen et al., 2002; Poulton et al., 2004; Arnold et al., 2004; Brocks et al., 2005; Canfield et al., 2007) have all invoked sulfidic conditions in the deep oceans during the late Paleoproterozoic and Mesoproterozoic based on data for black shale sequences. In Slack et al. (2007), it was emphasized that nearly all of these black shales were deposited in restricted basins (e.g., Belt Basin, Lyons et al., 2000; McArthur Basin, Shen et al., 2002), and, as noted by Meyer and Kump (2008), the black shales in each studied basin are volumetrically minor. Thus it is uncertain whether these sulfidic conditions were globally widespread and persistent, and whether they are representative of open-marine settings. An alternative view is that deep and shallow parts of the ocean were variably oxygenated while intracratonic basins and areas with high productivity on the continental shelves developed sulfidic conditions (Bekker et al., 2008; Scott et al., 2008).

Exhalative chemical sedimentary rocks associated with Cu-rich VMS deposits provide an independent proxy for the redox state of the deep oceans under open-marine conditions. Such conditions are characteristic of VMS mineralization that occurs in modern volcanic arc and backarc settings like those in the South Pacific (e.g., Hannington et al., 2005). Within a late Paleoproterozoic (1738 Ma) submarine arc sequence in central Arizona, VMS-related jasper and iron formations in the Jerome district have mineralogical features (abundant hematite or magnetite) and geochemical signatures (<0.1 wt% MnO; mainly small positive or no Ce anomalies) that suggest deep-marine Si-Fe mineralization under suboxic conditions (Slack et al., 2007). The oceanic arc-related exhalites near the Jones Hill VMS deposit in New Mexico also contain abundant magnetite, but include much higher concentrations of Mn (as much as 9.4 wt% MnO), and in one unaltered, nearly detrital-free sample, no Ce anomaly. Our data for the Jones Hill Si-Fe-Mn exhalites are thus also most consistent with suboxic depositional conditions, as inferred for exhalites in the Jerome district. The Mn-enriched nature of some of the Jones Hill exhalites, relative to those in the Jerome district, may reflect higher Mn/Fe ratios of the hydrothermal fluids, possibly as a function of differences in tectonic setting and spreading rate (see German and Von Damm, 2003). The high Fe/Mn ratios of the Jerome exhalites also could be due to very high temperatures, near the critical point, of the related vent fluids (see Von Damm et al., 2003). Large variations in whole-rock Mn/(Mn + Fe) ratios among exhalite samples from the Jones Hill area (Table 2) likely record spatial and/or secular changes in the composition of the local submarine hydrothermal fluids, especially Fe/H2S ratios (e.g., Hannington et al., 1995; German and Von Damm, 2003), and possibly rates of microbial scavenging of Mn from the hydrothermal plumes (Mandernack and Tebo, 1993; Hrischeva and Scott, 2007).

Oxide facies iron formation and hematitic chert are also associated with Cu-rich VMS deposits of late Paleoproterozoic and Mesoproterozoic age in volcanic arc-related sequences of southern Wyoming, southern Colorado, western Arizona, eastern Bolivia, southern Brazil, and the Cape Province of South Africa (1792–1241 Ma; Slack et al., 2007, and references therein). The preservation of abundant ferric iron in hematite and/or magnetite within these VMS-related exhalites suggests that many deep ocean basins during this time period were not sulfidic, but instead contained at least minor amounts of dissolved oxygen. Future studies of these and other VMS-related exhalites may yield important clues to evolving redox conditions of open-marine, deep sea-water during the Proterozoic, with significant implications for understanding biological evolution on early Earth.

We thank Mickey Fulp for supplying data from 1980s mineral exploration work in the Jones Hill area. Annabelle Lopez and the New Mexico Bureau of Geology and Mineral Resources provided access to drill cores and drill logs. Bob Osburn of Washington University shared his unpublished geologic mapping, and Johnny Lawrence of the U.S. Forest Service helped with logistics in the field. Bjørn Willemoes-Wissing of the Geological Survey of Norway assisted with electron microprobe analyses. We especially thank Paul Barton and Randy Koski (both at U.S. Geological Survey [USGS]) for thorough and constructive reviews of an early version of the manuscript, and Lee Kump, Paul Spry, and an anonymous referee for helpful journal reviews. Discussions with Jim Hein of the USGS are also appreciated. Art Schultz (USGS) and the USGS Mineral Resources Program supported field and analytical work by Slack and Grenne. Bekker's contribution was supported by the Natural Sciences and Engineering Research Council of Canada Discovery and Geological Survey of Canada TGI-3 programs.