New iron isotope data document open-system formation of terrestrial iron oxide concretions and the potentially important role of iron-reducing bacteria in mobilizing iron. These terrestrial insights can provide valuable models for understanding extraterrestrial hematite spherules and their diagenetic history at Meridiani Planum, Mars. Whole-rock samples of Jurassic Navajo Sandstone host rock have δ56Fe values near 0 per mil (‰), whereas concretions typically have negative δ56Fe values. Negative δ56Fe values can be explained by complete oxidation and precipitation from aqueous fluids that had δ56Fe values of −0.5‰ to −1.5‰. The low δ56Fe values for the majority of concretions overlap those of Fe(II)aq and reactive ferric oxides in modern marine sediments where iron-reducing bacteria are actively cycling Fe, suggesting that Fe mobilization in the Navajo Sandstone occurred through bacterial reduction of Fe oxides. Variations in δ56Fe values support an open-system model of concretion formation where Fe is recycled via different chemical reactions involving reduction, mobilization, and precipitation. If the Mars concretions formed in a similarly open system during Fe mobilization and precipitation, their δ56Fe values should also deviate from δ56Fe = 0, dependent upon the pathway, but positive δ56Fe values would be expected for oxides in the absence of a role for microbial redox cycling.

Concretions are concentrated masses of mineral cements that are typically spherical, and they are very common and widespread in sedimentary rocks throughout the geologic record (e.g., Seilacher, 2001). Remarkably, the National Atmospheric and Space Administration (NASA) Opportunity rover discovered small hematite spherules in Meridiani Planum on Mars (dubbed “blueberries”) that have been interpreted as concretions (Squyres et al., 2004; Herkenhoff et al., 2004; Christensen et al., 2004). Iron oxide concretions on both Mars and Earth are valuable records of diagenetic, postdepositional changes in sedimentary units because they preserve evidence of ancient groundwater flow (Chan et al., 2004, 2005; Ormö et al., 2004; McLennan et al., 2005), fluid-rock interactions, element cycling, and the presence of redox boundaries. The concretions typically have spherical morphologies.

The Jurassic Navajo Sandstone exposed in southern Utah contains abundant, diverse, and well-documented examples of Fe oxide (e.g., hematite and goethite) concretions (e.g., Utah “marbles”) that have morphological similarities to the Mars “blueberries” in the deposits of Meridiani Planum (Chan et al., 2004, 2005). Although no single terrestrial analog is likely to be a perfect match for all the chemical and physical traits of sulfate mineralogy and Fe oxide concretions on Mars, the sediment-hosted Navajo concretions offer a natural, terrestrial laboratory that provides insights into the Meridiani Planum deposits. In this contribution, we present new Fe isotope data that allow us to constrain the Fe sources and pathways involved in concretion formation. These data document the open-system conditions that were required for concretion formation, the important role of hydrocarbons or Fe-reducing bacteria in mobilizing Fe, and the isotopic signatures of oxidation and precipitation at redox boundaries. These results provide insight into Fe mobility and redox transformations in terrestrial environments, and potentially constrain the range in Fe isotope signatures that may be found in oxide concretions on Mars in future in situ measurements or those obtained by sample-return missions.

Iron isotope geochemistry is a rapidly growing field, where it is now known that measurable isotopic fractionations occur in both high- and low-temperature environments (e.g., Welch et al., 2003; Beard and Johnson, 2004; Johnson et al., 2004; Dauphas and Rouxel, 2006). Some of the largest Fe isotope fractionations occur in low-temperature systems (<100 °C) during redox transformations, but only under conditions where separation of isotopically distinct Fe pools may occur (Johnson and Beard, 2006). For example, partial oxidation of aqueous Fe(II) generally leads to ferric Fe oxide or hydroxide precipitates that have higher 56Fe/54Fe ratios than the initial aqueous Fe(II). Complete oxidation, however, such as might occur when Fe(II)aq encounters atmospheric O2, or during weathering of Fe(II)-bearing igneous or metamorphic rocks, will produce ferric oxides or hydroxides that have essentially the same Fe isotope composition as the initial Fe(II) because oxidation ran to completion. This effect is well illustrated for surface weathering products, including suspended river loads, aerosols, and ferric Fe-rich clastic rocks that have low organic carbon or carbonate contents, all of which have Fe isotope compositions that are closely similar to those of the average crust (Beard et al., 2003b).

We report Fe isotope compositions using standard δ notation as the deviation in 56Fe/54Fe ratio of a sample relative to a reference reservoir, in units of parts per thousand, or per mil (‰):
where 56Fe/54 Fe rreference is taken as the average of igneous rocks (Beard et al., 2003a). On the igneous rock scale, the IRMM-014 standard produces a δ56Fe value of −0.09‰. The total spread in δ56Fe values of terrestrial samples is ∼4–5‰, and typical analytical precisions are ±0.05 to ±0.10‰ (2σ). Following standard convention, we describe the Fe isotope fractionation between two phases A and B as:
This may be related to the Fe isotope fractionation factor αA-B through the approximation:
Details on chemical separation and mass analysis methods are in Beard et al. (2003a) and Albarède and Beard (2004).

The porous and permeable characteristics of the Navajo Sandstone provide favorable conditions for the formation of Fe oxide concretions (Chan et al., 2004, 2005). The well-sorted, eolian quartz arenite (Beitler et al., 2005) has effective porosity that averages ∼17% (Cordova, 1978) and permeability up to 1 Darcy (Lindquist, 1988). Iron oxide mineralization can be summarized as a three-step process of diagenesis, which critically includes groundwater flow (Chan et al., 2000, 2004, 2005; Chan and Parry, 2002; Beitler et al., 2003, 2005). Diagenetic temperatures <100 °C are inferred based on an estimated burial depth of 2 km (Beitler et al., 2005) and mineralogical evidence (summarized in Chan et al., 2000), as well as supporting studies on the thermal burial history and uplift of the Colorado Plateau (Dumitru et al., 1994; Nuccio and Condon, 1996).

Iron Sources

Red Navajo sandstones (Fig. 1A) contain an average of 0.53 ± 0.34 wt% Fe2O3 (1 SD) distributed as thin Fe oxide films that coat individual sand grains (Beitler et al., 2005). This iron is likely to have been derived from detrital Fe-bearing silicate minerals within the sandstone during early weathering and diagenesis. Disseminated precipitation of this early diagenetic Fe oxide occurred shortly after deposition or during early burial via interaction with meteoric waters. In terms of expected Fe isotope compositions for these early Fe oxides, they probably had δ56Fe values near zero, similar to the igneous rock baseline that characterizes bulk continental crust, including low organic carbon (Corg) or carbonate (Ccarb) contents, clastic sedimentary rocks, and sandstones that are rich in disseminated ferric Fe oxide cements (Beard et al., 2003b; Beard and Johnson, 2004; Yamaguchi et al., 2005; Johnson and Beard, 2005, 2006).

Iron Mobilization

After burial, reduced fluids containing hydrocarbons derived from underlying units are thought to have flowed up preferential pathways such as faults and through the porous sandstone (Chan et al., 2000). Mobilization of the early oxides is envisioned to have occurred through reductive dissolution and transport as Fe(II)aq, leaving bleached, white sandstone (Fig. 1B). These bleached zones can be quite extensive, on the order of hundreds of square kilometers (Beitler et al., 2003). Bleached sandstones contain an average of 0.36 ± 0.26 wt% Fe2O3 (1 SD) as Fe oxide (Beitler et al., 2005). Based on total Fe contents of red and bleached sandstones, up to 30% of the Fe was removed. Reactions between hydrocarbons and ferric oxide may be expressed through reduction of Fe coupled with oxidation of organic matter as:
(e.g., Chan et al., 2000). The same reaction is applicable to reduction of ferric oxides or hydroxides by dissimilatory iron-reducing bacteria (DIRB) (e.g., Nealson and Saffarini, 1994). At temperatures below 120 °C, Fe(III) reduction by organic matter does not occur in the absence of DIRB (Lovley et al., 1991).

Iron isotopes are fractionated during reduction of Fe(III) oxides or hydroxides by DIRB, where Fe(II)aq has δ56Fe values that are 1–3‰ lower than the initial ferric oxide or hydroxide (Beard et al., 1999, 2003a; Icopini et al., 2004; Johnson et al., 2002, 2005; Crosby et al., 2005). Low-δ56Fe Fe(II)aq is balanced by a reactive Fe(III) layer on the oxide or hydroxide mineral surface that has high δ56Fe values, indicating that these minerals are dissolved incongruently by DIRB (Crosby et al., 2005). The low δ56 Fe fingerprint of Fe(II)aq that is produced by DIRB has been found in Fe(II)-rich pore waters in modern marine sediments (Severmann et al., 2006; Berquist and Boyle, 2006), and low δ56Fe ferric Fe oxides in modern marine environments have been ascribed to precipitation of DIRB-produced low δ56 Fe Fe(II)aq (Severmann et al., 2006; Staubwasser et al., 2006). The Fe isotope effects of abiotic reduction of ferric oxides or hydroxides by hydrocarbons are not known, and they likely depend on the dissolution mechanism; congruent dissolution is unlikely to produce any Fe isotope fractionation (Johnson et al., 2004), but formation of a new phase or reactive surface layer during incongruent dissolution may potentially produce an Fe isotope fractionation.

Although the form and occurrence of the Navajo Sandstone concretions are strikingly similar to the small, spherical “blueberry” hematite concretions (Figs. 1F–1H) in Meridiani Planum, the mechanisms by which Fe was mobilized may have been distinct. For example, Fe in the Mars system was probably mobilized through dissolution of primary silicates in basaltic rocks by acidic solutions (e.g., Morris et al., 2005). If this Fe mobilization occurred under reducing conditions, where aqueous Fe was largely Fe(II), no Fe isotope fractionation would be expected because no redox change would occur. Partial dissolution of silicates by organic ligands under oxic conditions can produce aqueous Fe(III) that has relatively low δ56Fe values (Brantley et al., 2001, 2004), although significant fractionations have only been observed at very low extents of dissolution (≤1%).

Concretion Precipitation

Precipitation of terrestrial concretions is thought to occur when Fe(II)-bearing (reduced) fluids intersected oxidizing groundwaters, where oxidation of Fe at near-neutral pH would produce immediate precipitation of Fe oxide at the mixing interface (Von Gunten and Schneider, 1991). Precipitation of Fe oxide would be concentrated within a spatially limited reaction front corresponding to this mixing interface. Concretions that precipitate within such a reaction front are commonly spheroidal in shape (Figs. 1C–1E). The average Fe concentration in the Fe oxide cemented concretions is 15.12 ± 9.99 wt% Fe2O3 (1 SD) (Beitler et al., 2005), where other constituents are largely SiO2 (avg. of 77.3 wt%) and Al2O3 (avg. of 2.2 wt%). Age determinations of related mineralization (Chan et al., 2001) suggest that some precipitation occurred ca. 25 Ma, but mineralization appears to have been episodic and may have included older or younger events.

The Fe isotope compositions produced during oxidation and precipitation will depend upon the extent of coupled oxidation-precipitation and the degree to which precipitation occurs under equilibrium conditions (Beard and Johnson, 2004; Johnson and Beard, 2006).

Partial oxidation of Fe(II)aq to Fe(III)aq, followed by complete precipitation of Fe(III)aq to ferric oxides or hydroxides should produce δ56Fe values for the ferric Fe precipitates that are ∼1–3‰ higher than those of the initial Fe(II)aq; the smaller fractionations are associated with a significant kinetic fractionation upon precipitation, whereas the maximum 3‰ fractionation occurs when precipitation occurs slowly under equilibrium conditions. In contrast, complete in situ oxidation and precipitation should produce δ56Fe values that are identical to those of the initial Fe(II)aq, which would allow the oxide concretions to be used as proxies for the Fe isotope compositions of the precursor aqueous Fe(II).

Fe(II)aq oxidation and precipitation of the Martian hematite concretions are thought to have occurred upon oxidation in a highly reactive chemical environment that included formation of evaporite and sulfate minerals (McLennan et al., 2005; Tosca et al., 2005). In both the Mars and terrestrial systems, however, the Fe isotope fractionations produced during oxidation would be expected to be similar, given the fact that generally similar fractionations are observed in a wide variety of oxidative pathways, including biological and abiological systems (Bullen et al., 2001; Croal et al., 2004; Balci et al., 2006). We therefore expect that the key component in determination of the Fe isotope compositions of concretions on Mars or Earth is the mechanism by which Fe was initially mobilized.

We analyzed representative samples from 16 localities of Jurassic Navajo Sandstone and 2 localities of Jurassic Entrada Sandstone across an ∼400 km traverse in southern Utah (Fig. 2). These included whole-rock samples of original red host sandstone (Fig. 1A) and bleached sandstone (Fig. 1B). Whole-rock samples where color, textural, and field characteristics indicated that no late secondary Fe mobilization occurred have δ56Fe values that lie within 0.2‰ of zero (avg. δ56Fe = +0.04‰), the same range observed in low-C and low-S clastic sedimentary rocks that have not undergone significant anoxic diagenesis (Table DR11; Fig. 3) (Beard and Johnson, 2004; Johnson and Beard, 2006). Bleached rocks that have lost Fe have a similar range in δ56Fe values, although their average is slightly higher than zero. One sample that is enriched in secondary oxides has a distinctly lower δ56Fe value of −0.6‰.

Iron oxide concretions have a large range of δ56Fe values from +0.9‰ to −2.0‰. The most remarkable characteristic of the concretion data is that most samples have negative δ56Fe values, which is generally unexpected for oxide minerals, which would be enriched in 56Fe/54Fe based on Fe isotope fractionation factors (Beard and Johnson, 2004). Comparison of δ56Fe values for individual concretions with whole-rock samples from the same locality generally show large isotopic contrasts, up to 1.5‰, which provides strong support for a model in which Fe was mobilized, oxidized, and precipitated in an open, fluid-rich system. There is no systematic correlation between δ56Fe values and oxide mineralogy (hematite or goethite). There are, however, sometimes significant variations in δ56Fe values (up to 0.8‰) within individual concretions. In several cases, the interiors of concretions have higher δ56Fe values than the rims (Fig. 3; Table DR1, Fig. DR1 [see 1]), but reverse zoning is also present.

The negative δ56Fe values for most of the oxide concretions coupled with the wide range in these values demonstrate that they did not form by simple, closed-system remobilization of Fe from earlier diagenetic oxides, but instead formed in an open system. Congruent dissolution of early diagenetic oxide minerals, followed by complete oxidation and precipitation in a closed system, would produce little change in δ56Fe values from their initial near-zero values. In principle, Fe oxides that have moderately negative δ56Fe values of −0.5‰ to −1.5‰ may be produced through oxidation and precipitation from an Fe(II)-bearing fluid that has a δ56Fe value of zero, but only under very special conditions. Assuming a Rayleigh process, the negative δ56Fe values for the majority of oxide concretions can only be produced by first oxidizing ∼70% to ∼90% of the Fe(II)aq initially present (that had an initial δ56Fe value of zero), and then selectively incorporating the last few percent of the incrementally produced oxide into the concretions (Fig. 4). This scenario seems unlikely for several reasons. The geological and geochemical evidence for the terrestrial model supports precipitation of Fe oxide at a reaction front produced by a mixing interface between reduced and O2-rich groundwater masses. If the Fe oxide was precipitating by a Rayleigh process at this interface, the great majority of the mass of the iron oxide present would have δ56Fe values significantly above 0‰ (Fig. 4). The opposite is observed; most concretions have negative δ56Fe values (Table DR1, see 1). Further, the terrestrial model appears to have operated at near-neutral pH conditions based on several geochemical characteristics: adsorbed species Zn, Co, Ni, As, V, and U that imply pH 6–8 for precipitation of Fe oxides (Beitler et al., 2005); co-existence of kaolinite and illite (Chan et al., 2000); and reasonable values of K+ in solution (0.001 M) that imply pH ∼ 6. At pH conditions of 6–8, oxidation will tend to run to completion at the reaction front where reduced fluids encounter O2-rich fluids. In such a scenario, selective incorporation of a narrow range of late-stage oxide products (e.g., Rayleigh process) into the concretions is not realistic.

Busigny and Dauphas (2006) analyzed several goethite-cemented concretions from the Navajo Sandstone in a few limited locality areas of southern Utah and suggested that negative δ56Fe values measured in the concretions could be explained by evolution of the fluid composition through successive precipitation (e.g., Rayleigh process) and/or adsorption of Fe. We regard the Rayleigh process as an unlikely mechanism for the reasons presented in the preceding paragraph. Moreover, our δ56Fe data on a broad range of concretion mineralogy types (many with complex layering and different δ56Fe core-to-rim trends) from different reaction fronts spanning a much larger regional area lead us to favor other models that involve open diagenetic systems and the potential role of Fe(III)-reducing bacteria.

As it seems more likely that the oxide concretions were produced by complete or near-complete oxidation of Fe(II)aq as reduced fluids encountered O2-rich zones, it is more appropriate to consider the integrated Fe oxide isotopic compositions (Fig. 4). In this case, however, the integrated Fe oxide can have δ56Fe values no lower than that of the initial fluid (Fig. 4). Hence, the negative δ56Fe values of −0.5‰ to −1.5‰ for the majority of oxide concretions is best explained by oxidation, followed by precipitation, of Fe(II)aq that had initial δ56Fe values that were negative. An end-member case would be one where the δ56Fe values of the oxide concretions are a direct proxy for the δ56Fe values of Fe(II)aq, which would require complete oxidation and precipitation. Alternatively, partial oxidation, but complete precipitation, may explain the data. For example, assuming an initial δ56Fe value of −2.0‰ for Fe(II)aq, ∼50%–100% oxidation may produce the range in Fe isotope compositions for the oxide concretions that have negative δ56Fe values (Fig. 4). Alternatively, the oxide concretions that have moderately negative δ56Fe values of ∼−0.5‰ could be explained by ∼100% oxidation and precipitation of Fe(II)aq that had an initial δ56Fe value of ∼−0.5‰. The few concretions that have positive δ56Fe values may be explained by ∼50% to ∼80% oxidation of Fe(II)aq that had an initial δ56Fe value of zero, again using the integrated Fe oxide compositions (Fig. 4). We conclude that the majority of oxide concretions were formed through complete or that initially near-complete oxidation of Fe(II)aq had negative δ56Fe values, perhaps averaging ∼−1.0‰, but ranging from ∼0.0‰ to ∼−1.5‰.

Origin of Low δ56Fe Aqueous Fe(II)

A number of processes may produce Fe(II)aq that has negative δ56Fe values. Approximately 30%–70% oxidation of Fe(II)aq that had an initial δ56Fe value of zero can, in principle, produce moderately negative δ56Fe values in the remaining Fe(II)aq of ∼−0.5‰ to ∼−1.5‰ (Fig. 4). Such a model, however, requires two stages of oxidation and precipitation to explain the negative δ56Fe values for the majority of oxide concretions: an initial partial oxidation step to produce low δ56 Fe Fe(II)aq, separation of Fe(II)aq from the initial oxide precipitates, followed by further oxidation (and precipitation) at the site of concretion formation. The core-to-rim variations in Fe isotope compositions for a number of oxide concretions (Fig. 3) provide some evidence for changing δ56 Fe values of Fe(II)aq during concretion formation, but these changes are most likely to have occurred at the site of concretion formation, rather than on a regional scale, as would be required in a “two-step” oxidation model.

It has been proposed that sorption of Fe(II)aq to Fe oxides may produce low δ56Fe values in the remaining Fe(II)aq (Icopini et al., 2004; Teutsch et al., 2005), but such proposals have been inferred from laboratory or field experiments that did not measure the sorbed Fe(II) directly. In contrast, Crosby et al. (2005) directly measured the Fe(II)aq–Fe(II)sorbed fractionation and found that this fractionation was small, between −0.3‰ (hematite) and −0.8‰ (goethite). At pH > 7, where the proportion of sorbed Fe(II) onto oxide minerals is at a maximum (e.g., Jeon et al., 2001), such a small isotopic fractionation would only be expressed at very low Fe(II)aq contents, and even under these conditions, the effect would be only several tenths per mil. Under mild to strong acidic conditions (pH < 6), the very low proportion of sorbed Fe(II) eliminates sorption as a mechanism for producing significant Fe isotope changes in Fe(II)aq, contrary to the suggestion of Balci et al. (2006). These considerations therefore make it very unlikely that sorption of Fe(II) to Fe oxide minerals is an explanation for the generally low δ56Fe values that are inferred for the Fe(II)aq that was the source for the oxide concretions.

Evidence for Fe(III)-Reducing Bacteria

The range in δ56Fe values estimated for Fe(II)aq that was the source of Fe for the Navajo concretions overlaps that measured for active diagenetic systems, including modern marine sediments (Fig. 4). Severmann et al. (2006) noted that porewater Fe(II) in suboxic sections of modern marine sediments (California margin) where DIRB are likely to play a major role in Fe cycling have low δ56Fe values, generally between −1.0‰ and −3.0‰ (Fig. 4). In addition, ferric hydroxides in the same sediments also have relatively low δ56Fe values, averaging ∼−1.0‰ (Fig. 4). Severmann et al. (2006) interpreted the negative δ56Fe values for reactive Fe(III) oxides to reflect complete or near-complete oxidation of low δ56Fe Fe(II)aq upon interaction with O2-bearing seawater, where the low δ56Fe values for porewater Fe(II) were generated by DIRB deeper in the sediment column. A similar interpretation has been made by Staubwasser et al. (2006) to explain moderately negative δ56Fe values for reactive Fe(III) oxides in modern sediments from the Arabian Sea. Based on the similarly negative δ56Fe values for the Navajo oxide concretions, and the inferred low δ56 Fe values for Fe(II)aq, we infer that bacterial Fe(III) reduction, coupled with hydrocarbon oxidation, was the major mechanism for mobilizing Fe in the Navajo Sandstone.

Other Mechanisms for Producing Low δ56Fe Values

It is unknown if low δ56Fe values for Fe(II)aq may be produced through abiotic reduction of early diagenetic Fe(III) oxides by hydrocarbons. Experimental studies by Shebl and Surdam (1996) demonstrated increases in porosity in oxide-bearing sandstone during heating (200–360 °C) in the presence of hydrocarbons, but these increases largely occurred through dissolution of carbonate cement, and there were no clear increases in Fe(II)aq contents during reaction. At the lower temperatures (<100 °C) estimated to have characterized Fe mobilization in the Navajo Sandstone, we would expect less reaction between hydrocarbons and sandstone cements. In addition, abiological reduction of Fe(III) oxides coupled with oxidation of organic matter was not observed in experiments run as high as 120 °C at circumneutral pH (Lovley et al., 1991). Therefore, our preferred interpretation is that the role of hydrocarbons in Fe mobilization in the Navajo Sandstone was primarily to serve as an organic carbon source for bacterial Fe(III) reduction rather than as an abiological reductant.

Finally, we consider the possibility that the oxide concretions may reflect in situ oxidation of pyrite concretions. Because many sedimentary pyrites in modern (Severmann et al., 2006) and ancient (Rouxel et al., 2005; Yamaguchi et al., 2005; Archer and Vance, 2006) environments have negative δ56Fe values, in situ oxidation of pyrite might explain the low δ56Fe oxide concretions. Such an explanation, however, is unlikely for several reasons. First, pyrite is rarely found in unbleached Navajo Sandstone, and only in trace abundances within bleached sandstones. There is no evidence of any remnant pyrite minerals in the numerous samples of Navajo concretions that have been examined, and if pyrite was important in the diagenetic history, some remnants would be expected. Furthermore, the Navajo Sandstone does not contain any evidence for sulfate minerals, nor is it rich in organic carbon, as is common in pyrite-bearing sedimentary rocks. Second, the widespread distribution, variety, abundance, and geometries of the Navajo concretions are very different from the features observed in pyrite-bearing sedimentary rocks, where pyrite is commonly localized within bedding and/or mineralization along fault zones. More importantly, the distribution of the Fe oxide concretions in the Navajo Sandstone is along geographically extensive redox fronts that are characterized by bleached zone boundaries that cut across primary structures. These features are not consistent with in situ oxidation of pyrite.

A Conceptual Model

We illustrate a conceptual model of redox cycling of Fe and formation of Fe oxide concretions in Figure 5. The δ56Fe values for the initial, early diagenetic Fe oxides should have been close to zero (Fig. 5A), as shown by our measurements of Fe oxide in unbleached Navajo sandstone and an extensive database for modern and ancient Fe(III)-rich weathering products. Iron mobilization through incongruent dissolution of the early diagenetic Fe oxides produced Fe(II)aq that has low δ56Fe values of ∼−0.5‰ to ∼−1.5‰ (Fig. 5B). On average, ∼30% Fe reduction occurred based on the contrast in Fe contents between red and bleached whole-rock samples. However, the proportion of mobilized Fe that was eventually sequestered as oxide concretions is difficult to constrain. Field measurements of this proportion are complicated by the highly nonuniform distribution of the oxide concretions, which likely resulted from the superposition of multiple reaction zones of variable scales (from centimeter to kilometer scales and greater) that migrated spatially and/or operated repeatedly over time.

Reductive dissolution most likely occurred by DIRB, where oxide reduction was coupled to hydrocarbon oxidation under anoxic conditions. Reduced, Fe(II)-bearing fluids were transported away from the site of initial oxide reduction (Fig. 5C), where anoxic conditions were likely maintained by dissolved organic carbon. Fe(II)-rich fluids encountered O2-bearing fluids either though ascent of reduced fluids or downward penetration of oxidizing fluids or both (Fig. 5D), which resulted in rapid and complete oxidation and precipitation of Fe(II)aq to ferric oxides and hydroxides at the mixing interface between these fluids. The distribution and size of oxide concretions indicate a major control by nucleation, suggesting that precipitation was rapid. Such a process should produce δ56Fe values for oxide concretions that are generally reflective of the δ56Fe values for the (Fig. 5E), although isotopic precursor Fe(II)aq zoning would be expected due to local (centimeter-scale) changes in δ56Fe values imposed on the remaining Fe(II)aq that existed between nuclei during the oxidation process and final precipitation of the remaining Fe.

Removal of low δ56Fe Fe(II)aq should have produced high δ56Fe values for bleached rocks, but this is not uniformly observed. The δ56Fe values for whole-rock samples of bleached rocks range from −0.16‰ to +0.24‰ (Table DR1, see 1). It is important to note, however, that it is difficult to estimate the δ56Fe value of the Fe inventory that was removed, given the fact that the oxide concretions represent only a small fraction of the Fe that was removed from the bleached zones. Moreover, the lowest δ56Fe values for Fe(II)aq generated by reductive dissolution of Fe oxides by DIRB would be produced only at low extents of reduction (a few percent) (Johnson et al., 2005; Crosby et al., 2005), which would produce negligible changes in the δ56Fe values of the remaining Fe oxide. It is therefore difficult to calculate a rigorous isotopic mass balance with the current data set, although this may be possible through much more extensive studies of the bleached zones.

Iron isotope compositions of oxide concretions in the Navajo Sandstone document reductive dissolution of early diagenetic oxides, aqueous transport, and finally precipitation in an open groundwater terrestrial system. Although the mineralogy and inferred fluid chemistry of the Navajo concretions have some important differences from Mars concretions, Fe isotopes should be similarly valuable in constraining Fe pathways and open- or closed-system behavior. In the Navajo system, the generally low δ56Fe values of the oxide concretions point to initial Fe mobilization through incongruent reductive dissolution of early sandstone oxides, probably by dissimilatory Fe(III)-reducing bacteria producing Fe(II)aq with δ56Fe < 0. In the Meridiani Planum system, Fe mobilization is thought to have occurred by acid sulfate weathering of primary silicate minerals in basaltic rock, and in this case, if the fluid reservoir was limited and large extents of dissolution occurred, the δ56Fe values of the fluid would probably lie closer to those of the original basalt, which would be within 0.1‰ of the δ56Fe value of terrestrial basalts (Poitrasson et al., 2004).

Precipitation of aqueous Fe in the Navajo Sandstone probably occurred when reduced groundwaters intersected O2-bearing fluids, and the extent of isotopic fractionation was dependent upon the extent of precipitation that occurred in an open system. Two of the 15 concretions analyzed have δ56Fe > 0, and partial precipitation must have occurred in an open fluid-flow system. However, 13 of the 15 concretions have δ56Fe < 0, which indicates that complete or near-complete precipitation occurred, largely preserving the low δ56Fe values of the fluid. Isotopic zonation within individual concretions would be expected in both terrestrial and Martian systems if local changes (centimeter scale) occurred in fluid composition between nuclei during oxidation and precipitation. In the Meridiani Planum system, where the inferred low pH is expected to have produced significant pools of coexisting aqueous Fe(III) and Fe(II) (Tosca et al., 2005), we might expect only partial precipitation to occur, which would tend to produce oxide concretions that had higher δ56Fe values, similar to the few high δ56Fe concretions found in the Navajo Sandstone.

GSA Data Repository item 2006236, Figure DR1, original isotopic traverse measurements of concretions (rim to interior), and Table DR1, original isotopic measurements of all data samples with description of the locality (keyed to Figure 2) and mineralogy, is available online at, or on request from or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA.

This publication is based upon work supported by the National Aeronautics and Space Administration (to Chan, Parry, and Bowman) under grant number NNG06GI10G issued through the Mars Fundamental Research Program, the National Science Foundation under grant EAR-0525417 (to Johnson and Beard), and the Bureau of Land Management–Grand Staircase Escalante National Monument (to Chan and Parry). We gratefully acknowledge the Bureau of Land Management, the National Park Service, and Utah State Parks for sampling permissions. We thank Aaron Shultis and René Wiesli for assistance in the analytical work. Three anonymous reviewers are thanked for their comments on the manuscript.

Attribution: You must attribute the work in the manner specified by the author or licensor (but no in any way that suggests that they endorse you or your use of the work).
Noncommercial ‒ you may not use this work for commercial purpose.
No Derivative works ‒ You may not alter, transform, or build upon this work.
Sharing ‒ Individual scientists are hereby granted permission, without fees or further requests to GSA, to use a single figure, a single table, and/or a brief paragraph of text in other subsequent works and to make unlimited photocopies of items in this journal for noncommercial use in classrooms to further education and science.