Assessing the utility of Fe/Al and Fe-speciation to record water column redox conditions in carbonate-rich sediments Clarkson, M.O.1, Poulton, S.W.2, Guilbaud, R.2 and Wood, R.1 1 School of Geosciences, University of Edinburgh, West Mains Road, Edinburgh, EH9 3JW, UK 2 School of Earth and Environment, University of Leeds, Leeds, LS2 9JT, UK Abstract Geochemical proxies based on Fe abundance (Fe/Al) and Fe-speciation have been widely applied to marine sediments in order to unravel paleo-depositional redox conditions though geological time. To date, however, these proxies have only been calibrated in relation to modern and ancient siliciclastic marine sediments. This clearly limits their use, particularly in relation to carbonate-rich sediments and rocks. To address this, we here explore the applicability of Fe-based redox proxies in carbonates through three approaches. First, we have compiled Fe/Al data for modern marine sediments to investigate variability in Fe-enrichments as a function of carbonate content and depositional setting. Second, we have expanded this approach with a compilation of new and existing Fe-speciation data for modern and ancient marine sediments deposited under oxic and euxinic (anoxic and sulfidic) water column conditions. Finally, we show new data from paired limestone and dolomite sample sets to demonstrate the potential significance of deep burial dolomitization on the Fe/Al and Fe-speciation redox proxies. Modern marine sediments deposited under oxic conditions show no relationship between Fe/Al and carbonate content. These sediments have an average Fe/Al ratio of 1 0.55 ± 0.11, with some higher values potentially being attributable to steady-state early diagenetic remobilization of Fe towards the sediment-water interface. In contrast, significant Fe/Al enrichments occur as a consequence of water column Fe mineral formation and deposition, either under anoxic conditions, or due to input of anoxic hydrothermal fluids into oxic seawater. Iron speciation data also show no direct correlation with carbonate content, and instead three groups can be distinguished based on total Fe (FeT) and organic C contents. Sediments deposited under oxic water column conditions, with FeT >0.5 wt%, generally plot below the 0.38 FeHR/FeT siliciclastic reference threshold for distinguishing oxic and anoxic environments, regardless of organic C content. Also consistent with siliciclastic calibrations, carbonate-rich sediments that contain significant organic matter (>0.5 wt%) and which were deposited under anoxic water column conditions tend to have FeHR/FeT ratios >0.38, independent of FeT content. In contrast, oxic carbonate-rich sediments with low FeT (<0.5 wt%) and low organic C (<0.5 wt%) routinely give a spuriously high FeHR/FeT ratio, suggesting that the use of Fe-speciation for such samples is not appropriate for evaluating water column redox conditions. Analysis of burial dolostones suggests that the Fe-speciation proxy may also be compromised by deep burial dolomitization, where there has been a clear source of mobile Fe to enrich rocks during recrystallization. This new assessment expands the utility of Fe-based redox proxies to also incorporate appropriate carbonaterich rocks, provided that care is taken to assess the possible impact of deep burial dolomitization. 2 1. Introduction Ancient redox reconstructions are a major focus of paleoenvironmental research and have greatly advanced our understanding of biogeochemical cycles, key evolutionary events, and past periods of environmental change (e.g., Canfield, 2005; Lyons et al., 2009; Lyons and Severmann, 2006; Meyer and Kump, 2008; Poulton and Canfield, 2011; Raiswell and Canfield, 2012). Two of the most widely utilized geochemical proxies in the redox toolbox are built upon the environmental behavior of Fe, through enrichments of total Fe relative to aluminum (Fe/Al), and highly reactive Fe to total Fe (FeHR/FeT) (Lyons and Severmann, 2006; Poulton and Canfield, 2011; Poulton and Raiswell, 2002; Raiswell and Canfield, 1998; Raiswell et al., 2001). FeHR refers to Fe minerals that are considered highly reactive towards biological and abiological reduction under anoxic conditions (Canfield et al., 1992; Poulton et al., 2004a), and includes carbonate-associated Fe (Fecarb; e.g., ankerite and siderite), ferric (oxyhydr)oxides (Feox; e.g., goethite and hematite), magnetite Fe (Femag) and Fe sulfide minerals (Fepy; e.g., makinawite and pyrite) (Poulton and Canfield, 2005). Sediments may be enriched in FeHR under anoxic marine conditions due to either export of remobilized Fe from the oxic shelf (Anderson and Raiswell, 2004; Duan et al., 2010; Raiswell and Anderson, 2005; Severmann et al., 2008), or under more widespread anoxia, due to upwelling of deep water Fe(II) (Poulton and Canfield, 2011). Precipitation of this mobilized water column Fe is then potentially induced through a variety of processes, including Fe sulfide precipitation if the Fe encounters water column sulfide, through biogenic or abiogenic oxidation of Fe(II) to form Fe(III)-containing minerals, or via direct precipitation of Fe(II) carbonates or phosphates (e.g., Canfield et al., 1996; Crowe et al., 2008; Jilbert and Slomp, 2013; Raiswell and Canfield, 1998; Zegeye et al., 2012). These processes have the consequence that FeHR/FeT ratios in deposited 3 sediments provide a particularly sensitive means to determine whether a depositional setting was oxic or anoxic. Calibration in modern and ancient marine environments suggests that FeHR/FeT <0.22 indicates oxic water column conditions, while FeHR/FeT >0.38 provides a robust indication of deposition from an anoxic water column (Poulton and Canfield, 2011; Poulton et al., 2002; Raiswell and Canfield, 1998; Raiswell et al., 2001). Values between 0.22-0.38, however, are somewhat equivocal, and care needs to be taken to determine whether such values are a consequence of masking of the additional anoxic water column flux of FeHR, either due to rapid sedimentation (Lyons and Severmann, 2006; Poulton et al., 2004b; Raiswell and Canfield, 1998), or due to post-depositional transformation of unsulfidized FeHR minerals to less reactive sheet silicate minerals (Cumming et al., 2013; Poulton et al., 2010; Poulton and Raiswell, 2002). By additionally examining the ratio of Fepy/FeHR, the Fe speciation technique has the unique advantage in that it allows the separation of anoxic settings into euxinic (sulfidic) environments (Fepy/FeHR >0.7-0.8) and non-sulfidic (Fe-rich; ferruginous) environments (Fepy/FeHR <0.7) (März et al., 2008; Poulton and Canfield, 2011; Poulton et al., 2004b). Fe/Al ratios provide a bulk measurement of this enrichment in Fe HR, which can allow anoxic and oxic depositional environments to be distinguished. The inclusion of ‘unreactive’ Fe (FeU), largely in the form of silicate-associated Fe, tends to make Fe/Al less sensitive than FeHR/FeT and more difficult to define a normal oxic baseline level for (this tends to vary quite considerably, dependent on the specific depositional setting and terrestrial sediment source; van der Weijden, 2002). Nevertheless, for Fe/Al it is common to consider that enrichments above the average oxic Phanerozoic shale value of 0.53 ± 0.11 denote anoxic conditions (Lyons and Severmann, 2006; Raiswell et al., 2008). Furthermore, a particular advantage of the Fe/Al proxy is that it does not suffer 4 from the possibility of post-depositional transformation of unsulfidized FeHR to less reactive minerals, and thus Fe/Al and Fe-speciation in combination provide a particularly powerful means to evaluate water column redox conditions (Cumming et al., 2013; Lyons and Severmann, 2006; Poulton et al., 2010). Normalization of Fe components to Al or FeT corrects for variable dilution by carbonate or biogenic material (Raiswell and Canfield, 1998), and also lessens the influence of variability in grain size and source mineralogy (Poulton and Raiswell, 2005), making the proxies more widely applicable to different sediment types. Nevertheless, the Fe-based redox proxies were developed and tested on siliciclastic-rich marine sediments, and whilst the utility of the technique has been demonstrated for a variety of chemical sediments, including banded iron formations (Poulton et al. 2004a; Poulton et al., 2010) and some carbonate-rich marine sediments (Kendall et al., 2010; März et al., 2008; Zerkle et al., 2012), the method has not yet been calibrated for carbonates. Indeed, Lyons et al. (2012) highlight that careful consideration of lithology is required when applying Fe-based redox proxies. Theoretical concerns with the application of these proxies to carbonates largely relate to the decreased detrital contents (and hence low FeHR and FeT), which ultimately means that carbonate sediments are much more sensitive to highly reactive Fe inputs that may originate from sources other than the detrital and anoxic water column inputs. These additional sources could include the incorporation of low concentrations of Fe HR into the carbonate lattice during carbonate precipitation under oxic conditions, or a post-depositional influx of dissolved Fe into the sediment profile, a process which is of particular concern during deep burial dolomitization (Warren, 2000). These possibilities are particularly important to evaluate, as the application of Fe-based redox proxies to carbonates could potentially provide a vast increase in spatial 5 and temporal understanding of ocean redox dynamics. As a lithology, carbonates account for ~25% of the rock record and often provide important complimentary information in the form of 13C and Sr isotope records, REE profiles, and carbonateassociated sulfur (CAS) estimates of the isotopic composition of contemporaneous seawater sulfate (Gill et al., 2007; Hurtgen et al., 2009; Newton et al., 2004; Planavsky et al., 2012). Additionally, carbonates often represent shallower water environments, which tend to be centers of biodiversity and therefore record important evolutionary events such as radiations and extinctions (Kiessling et al., 2010). Here, we present an assessment of the utility of Fe/Al and Fe-speciation to record water column redox conditions across a wide range in carbonate content. Firstly, we explore a compilation of modern marine Fe/Al data to evaluate variability in the Fe contents of carbonates from different depositional settings. Secondly, Fe-speciation data are presented for a selection of new and published modern and ancient carbonate-rich sediments deposited under oxic and euxinic water column conditions. Finally, the potential impact of deep burial dolomitization is evaluated with new data obtained across a dolomitization front in early Triassic carbonates from Oman. Together, this approach allows us to place preliminary constraints on the careful application of Febased redox proxies to appropriate carbonate-rich sediments. 2. Materials and Methods 2.1. A compilation of modern core-top Fe/Al data Data were compiled for modern (Holocene) oxic water column open-ocean and continental margin core-top sediments from the Pangaea database (Table 1). These data tend to represent shelf to basin environments and do not sample shallow marine 6 carbonate platforms. Nevertheless, the data provide important information on Fe systematics in a wide range of settings of differing carbonate content. Note that nearshore environments that receive unusually high inputs of highly weathered terrestrial sediment (e.g., those close to major river systems), such as the Amazon Shelf and Congo Fan mobile mud belts, were not included in this compilation. Although this includes relatively few samples in the Pangaea data-base, surface sediments from such sites tend to have unusually high Fe contents (and FeHR in particular), due to both the highly weathered nature of the sediment (Poulton and Raiswell, 2002) and due to intense diagenetic remobilization of Fe (Aller et al., 2004; Aller et al., 1986). We also avoided upwelling areas and other areas that tend to exhibit sporadic water column anoxia, and instead utilize published data from persistently anoxic basins (Lyons et al., 2003; Raiswell and Canfield, 1998) and from sites with significant hydrothermal Fe input (Lyle, 1986; Dubinin, 2006; Govin et al., 2012), focusing on samples for which carbonate concentration data were also available (Table 1). 2.2. Modern and ancient Fe-speciation data-set New data are presented for modern carbonate samples from a diverse range of environments, including shallow marine carbonate platforms (see Table 2). These include pure biogenic carbonates, abiotic ooids and carbonate sands from temperate and tropical environments. New data for ancient rocks come from Miocene carbonates from Spain, representing carbonates with a simple alteration history of uplift and meteoric weathering (Weijermars, 1991). Additionally, we incorporate modern and ancient Fe-speciation data (Table 3) from published calibration studies (Table 3; 7 Canfield et al., 1996; Lyons et al., 2003; Poulton and Raiswell, 2002; Raiswell and Canfield, 1998; Raiswell et al., 2008). The dolomitization study was performed on limestone and dolomite pairs sampled from the same beds across an oblique dolomitization front in the Early Triassic Maqam Formation, Oman (Richoz, 2006). The dolomitization front is clear from the orange colouration of altered samples, representing increased Fe. Carbon and oxygen stable isotope measurements across the front show a depletion of the original oxygen isotope signature in the dolostones compared to the limestones, whilst the carbon values were preserved (Atudorei, 1999) consistent with deep burial dolomitization (Richoz et al., 2010). 2.3. Geochemical methods Fe-speciation extractions were performed according to calibrated extraction procedures (Poulton and Canfield, 2005), whereby FeCarb was extracted with Na-acetate at pH 4.5 and 50°C for 48 h, FeOx was extracted via Na-dithionite at pH 4.8 for 2 h, and FeMag was extracted with ammonium oxalate for 6 h. FeT extractions were performed on ashed samples (8 h at 550°C) using HNO3-HF-HClO4. All Fe concentrations were measured via atomic absorption spectrometry and replicate extractions gave a RSD of <5% for all steps. Acid volatile sulphur (AVS) and pyrite were determined stoichiometrically from precipitated Ag2S after HCl and chromous chloride distillation, respectively (Canfield et al., 1986). Total inorganic carbon (TIC) was measured using a CM 5012 Coulometer. Total organic C (TOC) was measured on a LECO® carbon analyser after carbonate removal (two 25 % (vol/vol) HCl washes for 24 hours). Replicate analyses gave a precision of ± 0.09 wt% (2σ level). 8 3. Results 3.1. Modern core-top compilation Figure 1 shows the total Fe and Al contents of the modern data compilation as a function of CaCO3 content, with the data divided into oxic normal marine, anoxic marine (which are all euxinic environments), and hydrothermal settings. The data show the expected overall negative correlation between carbonate and both Fe and Al for all depositional settings, highlighting the simple dilution effect of the carbonate on major element concentrations. This results in very low FeT and Al contents at the highest concentrations of carbonate. In detail, Al tends to be more variable as a function of carbonate content for both oxic and anoxic settings, presumably due to a greater degree of variability in the lithogenic input, relative to Fe. For normal oxic marine sediments, the carbonate dilution effect causes the range in FeT to decrease at higher carbonate contents, due to a reduction in the relative impact of variability in the chemical composition of the lithogenic fraction, which is also seen to a lesser degree for Al. In contrast, for samples deposited beneath an anoxic water column, the lower lithogenic input at higher carbonate contents means that FeT concentrations are more significantly affected by relative variability in the rates of deposition of Fe minerals from the water column and rates of carbonate formation. This results in enhanced variability in FeT as carbonate increases (Fig. 1). Hydrothermal sediments might be expected to show enhanced variability in FeT throughout the entire range in carbonate content, relative to Al. This is because, in contrast to Al, FeT will be controlled by a balance between the rate of hydrothermal Fe mineral deposition and the rate of carbonate production, both of which 9 are highly variable on a global scale. There is some indication that this may be the case in Figure 1, but our data-set is not large enough to fully evaluate this suggestion. Figure 2 shows Fe/Al ratios for the different depositional settings as a function of CaCO3 content. The normal marine data exhibit a range in Fe/Al from 0.30 to 0.80 (with an average of 0.55 ± 0.11), and show no correlation with carbonate content. The combined euxinic data-set also shows no relationship with CaCO3 content, however, the Black Sea and Kau Bay data-sets show an overall increase in Fe/Al as carbonate content increases (see also Canfield et al., 1996; Raiswell and Canfield, 1998). Samples from the Black Sea and Kau Bay also tend to be significantly enriched in Fe relative to the normal marine data (Black Sea Fe/Al = 0.79 ± 0.10; Kau Bay Fe/Al = 0.87 ± 0.08). In contrast, euxinic samples from the Cariaco Basin have low Fe/Al ratios that are more similar to the normal marine data (0.49 ± 0.02). Fe/Al ratios are also high for hydrothermal sites close to mid-ocean ridges (Fe/Al = 3.03 ± 3.77), with the highest values (and largest range) occurring at higher carbonate contents (Fig. 2). 3.2. Modern Fe-speciation data New Fe-speciation data are presented in Table 4 and compiled with literature data in Figures 3 and 4. The anoxic modern and ancient data-sets generally plot above the anoxic siliciclastic FeHR/FeT threshold value of 0.38, and show an overall increase in FeHR/FeT with increasing carbonate content (Fig. 3). A few samples, largely comprising sediments from Kau Bay, Indonesia (Middelburg, 1991) plot below 0.38, but generally above the 0.22 FeHR/FeT threshold that is commonly taken as an upper value for robust identification of oxic water column conditions in ancient samples (this value is based on 10 an average FeHR/FeT ratio for oxic water column deposition during the Phanerozoic of 0.14 ± 0.08; Poulton and Canfield, 2011; Poulton and Raiswell, 2002). In contrast to the anoxic samples, sediments deposited from oxic bottom waters show no direct relationship with carbonate content (Fig. 3). In general, most of the oxic samples plot below 0.38, with 78% of the Phanerozoic oxic samples falling below the 0.22 threshold, as opposed to 37% for modern oxic samples. This decrease in the Phanerozoic average FeHR/FeT ratio, relative to the modern, likely arises due to loss of unsulfidized FeHR to sheet silicate minerals during diagenesis, as discussed by Poulton and Raiswell (2002). However, some oxic samples plot significantly above 0.38, and this is particularly the case for some modern samples with high carbonate contents. Consideration of these data in terms of total Fe, rather than carbonate content, shows that FeHR/FeT decreases at higher FeT for anoxic samples. In contrast, oxic samples display no direct correlation between these parameters. However, from Figure 4 it is apparent that the majority of oxic samples that plot above the 0.38 threshold have very low FeT (<0.5 wt%). 3.3. Paired Limestones and Dolomites The limestone samples from the Early Triassic Maqam Formation, Oman (Richoz, 2006) have low FeT (<0.52 wt%; Table 5), and consistent with the oxic modern and ancient compilation (Fig. 4), this results in elevated FeHR/FeT ratios, despite an inferred oxic depositional setting for these samples. Nevertheless, comparison with dolomitized samples from the same beds is instructive in terms of evaluating the potential role of burial dolomitization on Fe-speciation. Sample pairs 1 and 2 show little variation (within that expected for individual samples from the same bed) between the limestones 11 and dolomites, in terms of FeT, Fe partitioning between the different Fe pools, and FeHR/FeT (Table 5). However, sample pairs 3, 4 and 5 show an increase in FeT during burial dolomitization, which is particularly significant for samples 4 and 5. This increase tends to arise as a result of an increase in Fecarb (Table 5), which is consistent with Fe addition to the system during dolomitization, although for sample pair 4, approximately 50% of the additional Fecarb appears to be sourced from the Feox fraction. As a consequence of the additional Fe input to the system, FeHR/FeT ratios are elevated in dolostones relative to limestones for these samples. 4. Discussion 4.1. Behaviour of the Fe/Al paleo-redox proxy in carbonate-rich sediments Figure 2 provides the first compilation of Fe/Al as a function of carbonate content. The variability in Fe/Al observed for the oxic marine data-set (0.30-0.80) is larger than, but overlaps, values measured for modern siliciclastic-dominated sediments from oxic parts of the Black Sea, Effingham Inlet and Orca Basin (0.44-0.63; Lyons and Severmann, 2006). A potential explanation for some of the relatively high Fe/Al ratios in the modern compilation may relate to the sampling strategy used. To obtain a sufficient amount of data, and to be internally consistent, we utilized core-top data for our compilation. The potential for enrichment in total Fe in normal marine surface sediments is well-documented (e.g. Aller, 1980; Leslie et al., 1990; Trefry and Presley, 1982), and arises due to steady state remobilization of highly reactive Fe during anoxic diagenesis, followed by upwards diffusion and precipitation at the sediment-water interface. This process is likely particularly prevalent in organic-rich siliciclastic 12 sediments (i.e., sediments low in carbonate), which is consistent with the occurrence of the highest Fe/Al ratios at low CaCO3 in our compilation (Figure 2). Despite this complication, and more significantly in terms of the use of Fe/Al as a paleo-redox proxy, the mean of 0.55 ± 0.11 is entirely consistent with the Phanerozoic normal marine average Fe/Al ratio of 0.53 ± 0.11 (Raiswell et al., 2008). This Phanerozoic normal marine average is, however, based on siliciclastic sediments with average Al concentrations of 8.68 ± 2.94 wt% (Raiswell et al., 2008). Importantly, the lack of covariation between Fe/Al and carbonate content for modern normal marine sediments (Fig. 2) suggests that the proxy behaves in a consistent manner during deposition under oxic water column conditions, even when carbonate is high, and Fe T and Al concentrations are low (c.f., Fig. 1). The positive correlation observed between Fe/Al and CaCO3 for the independent euxinic Black Sea and Kau Bay data-sets is well-documented (Canfield et al., 1996; Raiswell and Canfield, 1998). This relationship arises because the organic matter that fuels sulfate reduction (and hence sulfide production) is derived from coccolithophorides in these settings, and thus sulfidized Fe that forms in the water column is intimately associated with their calcareous skeletons (Canfield et al., 1996; Raiswell and Canfield, 1998). However, as shown by Lyons and Severmann (2006) for an expanded suite of sediments from euxinic settings (for which CaCO3 data are not available), Fe enrichments may be decoupled from biogenic sediment inputs, and the overall controlling factor is simply precipitation of water column Fe under anoxic conditions. Therefore, no simple global relationship exists between Fe/Al and CaCO3 in anoxic settings. This is further exemplified in Figure 2, which also highlights one potential issue with the Fe/Al proxy. While the Kau Bay and Black Sea data-sets show clear enrichments in Fe/Al across a range of carbonate contents, relative to normal oxic 13 marine sediments, the Cariaco Basin euxinic sediments do not. However, euxinic Cariaco Basin sediments are in fact enriched in Fe/Al, relative to the lithogenic sediment supplied to the basin (Lyons et al., 2003). This highlights that local compositional variability in the terrestrial sediment influx into the marine environment can potentially mask sediment Fe/Al enrichments (relative to global average Fe/Al ratios) under anoxic conditions. The nature of Fe/Al ratios in carbonate-rich sediments from hydrothermal settings (Fig. 2) is also consistent with the Fe enrichment mechanism outlined above. At active spreading centres along the East Pacific Rise (E.P.R.) and Mid-Atlantic Ridge (M.A.R.), anoxic fluids enriched in dissolved Fe(II) are vented into oxic seawater. In sulfidic hydrothermal vent systems, Fe sulfides precipitate rapidly and are mostly deposited close to the vent (e.g., Feely et al., 1987; Mottl and McConachy, 1990). However, Fe(II) can continue to precipitate as Fe (oxyhydr)oxide minerals for some time in the neutrally buoyant plume that forms tens to hundreds of metres above the vents (Baker et al., 1985; Lupton and Craig, 1981; Reid, 1982). These plumes may be laterally advected from the ridge crest for hundreds of kilometres, with the result that Fe (oxyhydr)oxide minerals may continue to be deposited a considerable distance from the active vent (Baker et al., 1985; Klinkhammer and Hudson, 1986; Poulton and Canfield, 2006). The hydrothermal data presented in Figure 2 clearly demonstrate this process, and as might be expected due to the low lithogenic sediment input to such sediments, Fe enrichments may be particularly large. The hydrothermal data also highlight an important and often under-appreciated aspect of the Fe enrichment process. Iron enrichments under euxinic conditions require that Fe is transported into anoxic sulfidic waters and precipitated as sulfide minerals (e.g., Canfield et al., 1996). Similarly, under ferruginous conditions, Fe minerals may 14 form under strictly anoxic water column conditions (e.g., Zegeye et al., 2012). However, Fe (oxyhydr)oxide enrichments can also occur under oxic conditions, provided there is an anoxic mechanism to transport Fe(II) into the oxic setting. This may occur as a result of hydrothermal Fe(II) inputs into the ocean, as discussed above, but could also occur due to upwelling of anoxic Fe(II)-rich waters into shallow oxic surface waters, a process that may have been particularly prevalent during precipitation of some Precambrian banded iron formations (e.g., the 1.88 Ga Gunflint Formation; Pufahl et al., 2000). Sediments formed in this latter manner tend to be particularly enriched in ferric (oxyhydr)oxide minerals, and in these specific Fe(II) upwelling cases, enrichments actually imply that the adjacent deeper water column was anoxic, rather than the water column directly overlying the site of Fe enrichment. The above observations provide a framework for extending the potential application of the Fe/Al paleo-redox proxy to incorporate carbonate-rich sediments. Unfortunately, Fe/Al data is sparse for ancient sediments where paleo-redox conditions have been independently evaluated and where carbonate contents are also available. Nevertheless, initial calibrations of the Fe/Al proxy for siliciclastic samples were based on modern sediments deposited under different redox conditions (Lyons and Severmann, 2006; Lyons et al., 2003). Furthermore, the average modern and Phanerozoic Fe/Al ratios for normal oxic marine sediments are almost identical. This suggests that the oxic Phanerozoic siliciclastic Fe/Al ratio of 0.53 ± 0.11 (Raiswell et al., 2008) is also appropriate for carbonate-rich sediments, provided that deep burial dolomitization has not affected primary depositional Fe/Al ratios (see discussion below). However, the large relative standard deviation on this ratio (20%) highlights that the lithogenic sediment supplied to a particular locality can be highly variable in terms of chemical composition, which appears to be primarily related to enhanced 15 variability in Al contents, relative to Fe (Fig. 1). Thus, where possible, the best approach is to define an oxic baseline Fe/Al value for a particular setting (Lyons et al., 2003; Poulton et al., 2010). Nevertheless, our data suggest that Fe enrichments significantly above the normal oxic range (i.e., Fe/Al >0.64) can generally be used to identify anoxic depositional conditions in modern and ancient settings, for both siliciclastic and carbonate-rich sediments. 4.2. Fe-speciation in carbonate-rich sediments Figures 3 and 4 demonstrate that most sediments deposited from anoxic bottom waters have FeHR/FeT ratios above the 0.38 siliciclastic reference threshold for recognizing anoxia in modern and ancient sediments (Poulton and Raiswell, 2002; Raiswell and Canfield, 1998; Raiswell et al., 2001). This is a robust relationship that also holds for samples with very high carbonate (Fig. 3). In more detail, however, Fe HR enrichments tend to be more pronounced at higher CaCO3 in both modern and ancient sediments (Fig. 3). As discussed above, this feature may, in part, be due to sulfate reducing bacteria in the water column utilizing the organic matter associated with carbonate producers (Canfield et al., 1996; Raiswell and Canfield, 1998). However, the highest FeHR/FeT ratios also occur at low FeT, with a decrease as FeT increases (Fig. 4). Thus, consistent with the more recent suggestion that Fe enrichments are decoupled from biogenic sediment inputs (Lyons and Severmann, 2006), the degree of FeHR enrichment is perhaps better described more generally, as a balance between rates of water column FeHR deposition relative to the flux and composition of the lithogenic sediment fraction. Most of the anoxic sediments that plot below the 0.38 threshold are from Kau Bay. These sediments are unusual, in that despite low FeHR/FeT, Fe/Al ratios are high 16 and entirely consistent with deposition from an anoxic water column (Fig. 2). Masking of anoxic water column Fe enrichments due to rapid sedimentation is well-documented in both modern (Canfield et al., 1996) and ancient (Poulton et al., 2004b) settings, but such a process would reduce FeHR/FeT and Fe/Al ratios. In fact, the Kau Bay water column is not persistently euxinic, and instead alternates between euxinic and low-oxygen conditions (Middelburg, 1991). This gives two potential explanations for the low FeHR/FeT and high Fe/Al ratios. Firstly, under low oxygen, non-euxinic conditions, Fe minerals such as siderite and magnetite may ultimately be enriched in the deposited sediment (see Poulton and Canfield, 2011). This would give elevated Fe/Al, but these minerals are not extracted by the Fe extraction technique (Raiswell et al., 1994) used by Raiswell and Canfield (1998) to analyse these samples, and thus FeHR/FeT ratios would be low. A second possibility is that during low oxygen, non-euxinic intervals, FeHR minerals formed in the water column may escape sulfidation. In this case, early diagenetic transformation of unsulfidized FeHR to sheet silicate minerals would potentially reduce FeHR/FeT ratios, while maintaining high Fe/Al (e.g., Poulton et al., 2010; Cumming et al., 2013). The majority of oxic samples with FeT >0.5 wt% plot close to, or below, the 0.38 FeHR/FeT lower limit for anoxic siliciclastic sedimentation (Fig. 4). This suggests that when FeT is greater than 0.5 wt%, the thresholds derived for siliciclastic sediments deposited from an oxic water column are also appropriate for carbonate-rich sediments. In contrast, all oxic sediments with <0.5 wt% FeT have FeHR/FeT ratios >0.38 (Fig. 4). This is a generic feature of the data-set and does not solely relate to samples with very high carbonate, although particularly high FeHR/FeT ratios are evident for the purest oxic carbonates (Fig. 3). However, the low FeT for these samples does imply a relatively low 17 lithogenic fraction and hence a high biogenic fraction, whether it be carbonate or silica (or abiotic gypsum in the case of the anoxic Yazerez sample; Table 2). A review of literature data for Fe incorporation into a variety of calcifying marine organisms (including molluscs, scleractinian corals and planktonic gastropods), suggests that the FeT content may vary considerably in such carbonate biominerals (from 0.17 to 1540 ppm; Cravo et al., 2007; Foster and Chacko, 1995; Foster and Cravo, 2003; Keller et al., 2007; Kumar et al., 2010; Turekian et al., 1973), which is consistent with our own analyses (Table 4). Clearly, this concentration range is well below the lowest FeT contents of our oxic sample suite (Fig. 4). Hence, Fe uptake by marine calcifying organisms (which would be as FeHR) could potentially account for only a small fraction of the FeHR in our oxic samples that have spuriously high FeHR/FeT ratios. Instead, these samples have received FeHR from additional sources, which in some cases could include slowly depositing Fe (oxyhydr)oxide minerals from an oxic water column (a process that forms deep sea Fe-Mn nodules under oxic conditions), or due to Fe incorporation during early diagenetic mobilization of Fe and associated carbonate recrystallization (see discussion below). The oxic data therefore suggest that when FeT is <0.5 wt%, Fe speciation should not be used to recognize oxic sedimentation. However, Figure 4 highlights an additional important constraint on the use of Fe speciation for carbonate-rich sediments. Modern and ancient samples deposited from anoxic bottom waters, but with Fe T <0.5 wt% (i.e., mostly carbonate-rich samples; Fig. 3), give FeHR/FeT ratios that appropriately record anoxic sedimentation. In this case, however, the ratios are consistent with the overall negative trend of decreasing FeHR/FeT with increasing FeT, in contrast to the oxic samples with FeT <0.5 wt% (which scatter across a wide range in FeHR/FeT; Fig. 4). This suggests that Fe speciation may still be appropriate for recognizing anoxic depositional 18 conditions for samples with low FeT, but additional constraints are required to distinguish such samples from oxic, low FeT samples with spuriously high FeHR/FeT ratios. In this context, Raiswell et al. (2001) suggest that samples used to identify anoxic water column sedimentation via Fe speciation should generally be organic-C bearing, although no concentration limits were suggested in this study. To assess this for carbonate-rich sediments, Figure 5 recasts the FeHR/FeT data in terms of TOC rather than FeT, with the oxic low FeT samples (i.e., those with misleadingly high FeHR/FeT ratios; Fig. 4) distinguished as open triangles. Thus, based on organic C contents, the low FeT, high FeHR/FeT oxic samples can generally be distinguished from similar samples deposited under anoxic water column conditions. In fact, almost all oxic low FeT samples plot below 0.5 wt% TOC, whereas anoxic samples (including those with very high carbonate; Fig. 4) tend to have considerably higher TOC (with the exception of two samples), consistent with higher production and/or preservation of organic matter under anoxic conditions. Figure 5 also shows that oxic samples which do behave in a consistent manner with regard to their Fe speciation characteristics (i.e., those with Fe T >0.5 wt%), may also have very low TOC contents. Hence, it is not appropriate to simply define a TOC threshold for the use of Fe speciation, and instead, TOC and FeT concentrations should be considered in tandem. These combined constraints are demonstrated in Figure 6 and summarized in Table 6. Oxic samples behave appropriately with regard to Fe speciation, provided that FeT concentrations are >0.5 wt%, and no constraint on minimum TOC content is required in this case. The main exception to this concerns the unlithified stromatolite from Lagoa Vemelha, Brazil (Table 2), which has elevated TOC (2.12 wt%) but low FeT (0.23 wt%). However, the lithified stromatolite sample from the same environment 19 behaves more consistently (FeT = 0.02 wt%; TOC = 0.16 wt%), suggesting that unlithified stromatolites may be an anomaly, but such samples will not feature prominently in the geologic record. The Fe speciation technique also works appropriately for anoxic carbonate-rich sediments, regardless of FeT content, but these samples require a minimum TOC content of 0.5 wt% to be distinguished. In contrast, samples where both FeT and TOC are <0.5 wt% (open triangles; Fig. 5 and 6) do not record appropriate FeHR/FeT ratios, and hence Fe speciation should not be used in these cases. However, it is also interesting to note that these samples plot in their own distinct field on Figure 6, and thus we tentatively suggest that low FeT (<0.5 wt%) and low TOC (<0.5 wt%) contents in carbonate-rich samples may indicate oxic sedimentation, without the requirement for Fe speciation analyses. 4.3. Diagenetic alteration A potential caveat to the boundary conditions outlined above for the application of Fe speciation to carbonate-rich rocks and sediments concerns the possibility for Fe enrichment during diagenesis. Carbonate diagenesis is complex with numerous potential stages of recrystallization and cementation. Since the partition coefficient for Fe, with respect to calcite, is greater than unity, preferential scavenging of Fe occurs during precipitation (Barnaby and Rimstidt, 1989). This scavenging increases as Mg increases, because Mg and Fe have greater miscibility in a carbonate lattice than Ca and Fe. This leads to enrichment of Fe in the carbonate lattice compared to the fluid, although Fe can also be present between lattice planes, in lattice defects, along crystal boundaries, or can be added through adsorption (Tucker and Wright, 1990). For Fe to be incorporated into the carbonate lattice, however, it must be in the divalent state, 20 therefore the Eh of the pore fluid is the controlling factor on the incorporation of Fe (Barnaby and Rimstidt, 1989). Significant Fe incorporation is therefore not likely in primary precipitates or unconfined systems where oxygenated open marine and meteoric waters influence recrystallization, and this is reflected in the low Fe contents of primary abiotic and biotic calcites (Section 4.2). Anoxic pore waters, however, may promote the build-up of Fe2+ that can then be incorporated into carbonate during early cementation and recrystallization. Furthermore, the associated decrease in pH and increase in alkalinity that occurs during organic matter remineralization will enhance dissolution and reprecipitation of carbonate, therefore promoting the incorporation of Fe2+ into early cements. These early diagenetic processes would result in the transfer of FeHR between the different pools comprising the FeHR fraction (as is also the case with siliciclastic sediments), and particularly into carbonate phases during early recrystallization. When there is a significant lithogenic input of Fe, overall enrichments in FeHR/FeT are unlikely to occur, as the total FeHR pool should be conserved. However, when FeT is low, the spatial heterogeneity of these processes is more likely to result in FeHR enrichments. Potentially, if there is a significant external meteoric component to the anoxic diagenetic fluids, Fe will also be added to the system. Early recrystallization is exemplified by our modern carbonate mud and sand samples from Abu Dhabi (AD samples in Tables 2 and 4), which were deposited in a carbonate ramp environment and have undergone early dolomitization (see below) or have been affected by bacterial sulfate reduction within microbial mats (Lokier and Steuber, 2008). Samples with FeT >0.5 wt% do not show enrichments in FeHR/FeT above the 0.38 threshold (despite the fact that ratios may have been somewhat increased during early dolomitization), but samples with low Fe T do 21 show considerable enrichments, providing support for our FeT thresholds defined earlier. Since the incorporation of Fe is linked to the Mg content, the stabilization of high magnesium calcite (HMC) to low magnesium calcite (LMC) will reduce the tendency of a carbonate to take up Fe. LMC is the more stable form, and so ultimately calcite will tend towards this composition. This may result in a minor loss of Fe in confined systems during conversion, but more importantly, will aid the resistance of LMC carbonates to later Fe additions. The largest concern for Fe-speciation, however, is that if substantial quantities of Mg2+ are available during recrystallization then dolomitization occurs. Various kinetic problems tend to prevent the formation of dolomite as a primary precipitate in normal seawater (Land, 1998). These problems may be overcome by raising the Mg/Ca ratio of pore fluids (such as in sabkha settings, reflux dolomitization or through meteoric influence), or through microbial mediation or precipitation at higher temperatures (burial dolomites) (Warren 2000). Due to the greater stability of Fe and Mg, the effective distribution coefficient of Fe between dolomite and water is higher than for calcite. This leads to the preferential uptake of Fe, relative to Ca, during dolomitization (Tucker and Wright, 1990). Hence dolomites in the field can often be identified by their pink/orange colouration compared to limestones. However, the total Fe available for incorporation is again limited by pore water chemistry. For burial dolomites, Mg2+ may be supplied from a number of sources including bittern salt dissolution and clay mineral transformations (Boles and Franks, 1979; Kahle, 1965; Warren, 2000). This additionally supplies Fe from the unreactive fraction of autochthonous or allochthonous shales that is mobilized in anoxic pore waters, just as with siliciclastic diagenesis, and the total supply of Fe will be determined by the degree of open system behavior. The increased alkalinity of pore waters in carbonate-rich 22 sediments, however, promotes the precipitation of siderite, ferroan dolomite and ankerite, which augments the FeHR fraction (these minerals are extracted as part of the Fecarb fraction; Poulton and Canfield, 2005). Thus, deep burial dolomitization has a significantly greater potential to affect FeHR/FeT ratios across a wide range in FeT than early cementation, recrystallization or shallow dolomitization. Our data for samples collected across a deep burial dolomitization front in Early Triassic carbonates of the Maqam Formation, Oman (Table 5), clearly highlight the potential impact of this process on FeHR/FeT ratios. Thus, prior to any Fe speciation study of ancient carbonates, samples should be screened for additional Fe input via deep burial dolomitization. The extent of burial dolomitization is largely controlled by local permeability and, as such, is a localized process, creating sharp contacts between replaced dolomite and unaltered limestones (e.g., Carmichael and Ferry, 2008). Dolomitization may be readily discernable in outcrops from the distinct colouration of dolomites, but in addition, detailed petrographic distinction should be made between early and burial dolomitization phases, and saddle (ferroan) dolomites should not be included in Fe speciation studies. More detailed analyses are also possible, since incorporation of Mn and Fe into diagenetic cements can preserve a record of Fe mobilization, and this may be recognized via staining and cathodoluminescence (e.g., Dickson, 1965; Tobin et al., 1996). The measurement of oxygen isotopes can also give an estimate of recrystallization temperatures, and therefore depth, which may ultimately help to discriminate between burial dolomitization and early diagenetic alteration. 5. Summary We present the first compilations of Fe/Al and Fe-speciation data with carbonate content to investigate the utility of these proxies as redox indicators in carbonate-rich 23 sediments. Based on this we suggest new limits for the careful application of Fespeciation and Fe/Al, which are constrained by both FeT and TOC contents. When both TOC and FeT are <0.5 wt%, samples commonly record elevated FeHR/FeT ratios and thus Fe speciation cannot be used to identify paleo-depositional redox conditions in these cases, although low FeT and low TOC values in combination may indicate oxic sedimentation. For oxic samples, Fe-speciation behaves consistently, regardless of lithology, when FeT is >0.5 wt%. Fe/Al ratios in oxic carbonate-rich sediments are close to that of average Phanerozoic shale deposited under oxic water column conditions (0.53 ± 0.11; Raiswell et al., 2008), but the high relative standard deviation, caused primarily by enhanced variability in the Al content of the lithogenic fraction, suggests that a local oxic baseline should be defined where possible. Fe-speciation also allows carbonate-rich sediments deposited from anoxic waters to be identified, regardless of FeT content or lithology. However, an important caveat here is that the identification of water column anoxia additionally requires a TOC content of >0.5 wt%. This pattern is particularly clear in Jurassic Kimmeridge Clay samples (Raiswell et al., 2001), where high FeHR/FeT ratios and TOC contents are evident for low FeT (<0.5 wt%) samples deposited under anoxic water column conditions. The same principal is applicable to Fe/Al, where ratios >0.64 (i.e., 0.53 ± 0.11; Raiswell et al., 2008) combined with TOC contents of >0.5 wt% suggest an anoxic depositional setting. The impact of early diagenesis on Fe partitioning in carbonate-rich sediments is in many ways similar to that for siliciclastics. During early diagenesis, Fe may be transformed between the fractions that comprise the FeHR pool, but there will be a tendency for preservation of the mobilized Fe as Fecarb in carbonate-rich sediments. Early diagenetic recrystallization in carbonates may cause an increase in FeHR and FeT, but this only tends to cause spuriously high FeHR/FeT ratios when FeT is low (<0.5 wt%). 24 Late stage, deep burial diagenesis may significantly enhance FeHR/FeT ratios (and hence also Fe/Al). However, samples can be screened for these overprints, and hence careful sample selection means that these Fe-based redox proxies can be applied to carbonaterich sediments, within the geochemical framework outlined above. The above observations are entirely consistent with existing Fe speciation studies on ancient carbonate-rich sediments for which independent evidence of anoxic deposition exists, including the Toarcian (Jurassic) OAE (Raiswell et al., 2001) and the Cretaceous Coniacian-Santonian OAE3 (März et al., 2008). Therefore, with careful prescreening of samples, Fe-based redox proxies can now routinely be applied to carbonate-rich lithologies, opening up a rich potential archive for future reconstructions of water column redox conditions. Acknowledgements MOC acknowledges funding from the Edinburgh University Principal's Career Development Scholarship, International Centre for Carbonate Reservoirs and the Moray Endowment Award. RW and SWP acknowledge support from NERC through the ‘Coevolution of Life and the Planet’ scheme. Sylvain Richoz and Rob Newton are thanked for fieldwork assistance and sample collection, and we thank Rob Raiswell for compiling and sharing data. Thanks also to Stephen Lockier, Sandy Thudope, Simone Kasemann, Cees van der Land, Kate Darling, Simon Jung and André Bahr for supplying samples. 25 References Aller, R.C., 1980. Diagenetic processes near the sediment-water interface of Long Island Sound: Fe and Mn. Advances in Geophysics, 22: 351-415. Aller, R.C., Heilbrun, C., Panzeca, C., Zhu, Z.B. and Baltzer, F., 2004. Coupling between sedimentary dynamics, early diagenetic processes, and biogeochemical cycling in the Amazon-Guianas mobile mud belt: coastal French Guiana. Marine Geology, 208(2-4): 331-360. Aller, R.C., Mackin, J.E. and Cox, R.T., 1986. Diagenesis of Fe and S in Amazon inner shelf muds - apparent dominance of fe reduction and implications for the genesis of ironstones. Continental Shelf Research, 6(1-2): 263-289. Anderson, T.F. and Raiswell, R., 2004. Sources and mechanisms for the enrichment of highly reactive iron in euxinic Black Sea sediments. American Journal of Science, 304(3): 203-233. Atudorei, V., 1999. Constraints on the Upper Permian to Upper Triassic masine carbon isotope curve. Case studies from the Tethys, Lausanne University, 155 pp. Baker, E.T., Lavelle, J.W. and Massoth, G.J., 1985. Hydrothermal particle plumes over the southern Juan-de-Fuca Ridge. Nature, 316(6026): 342-344. Barnaby, R.J. and Rimstidt, J.D., 1989. Redox Conditions of Calcite Cementation Interpreted from Mn-Contents and Fe-Contents of Authigenic Calcites. Geological Society of America Bulletin, 101(6): 795-804. Boles, J.R. and Franks, S.G., 1979. Clay diagenesis in Wilcox sandstones of southwest Texas - implications of smectite diagenesis on sandstone cementation. Journal of Sedimentary Petrology, 49(1): 55-70. Canfield, D.E., 2005. The early history of atmospheric oxygen: Homage to Robert A. Garrels, Annual Review of Earth and Planetary Sciences. Annual Review of Earth and Planetary Sciences, pp. 1-36. Canfield, D.E., Lyons, T.W. and Raiswell, R., 1996. A model for iron deposition to euxinic Black Sea sediments. American Journal of Science, 296(7): 818-834. Canfield, D.E., Raiswell, R. and Bottrell, S., 1992. The reactivity of sedimentary iron minerals toward sulfide. American Journal of Science, 292(9): 659-683. Canfield, D.E., Raiswell, R., Westrich, J.T., Reaves, C.M. and Berner, R.A., 1986. The use of chromium reduction in the analysis of reduced inorganic sulfur in sediments and shales. Chemical Geology, 54(1-2): 149-155. Carmichael, S.K. and Ferry, A.M., 2008. Formation of Replacement Dolomite in the Latemar Carbonate Buildup, Dolomites, Northern Italy: Part 2. Origin of the Dolomitizing Fluid and the Amount and Duration of Fluid Flow. American Journal of Science, 308(8): 885-904. 26 Cravo, A. et al., 2007. Metals in the shell of Bathymodiolus azoricus from a hydrothermal vent site on the Mid-Atlantic Ridge. Environment International, 33(5): 609-615. Crowe, S.A. et al., 2008. Photoferrotrophs thrive in an Archean Ocean analogue. Proceedings of the National Academy of Sciences of the United States of America, 105(41): 15938-15943. Cumming, V.M., Poulton, S.W., Rooney, A.D. and Selby, D., 2013. Anoxia in the terrestrial environment during the late Mesoproterozoic. Geology, 41(5): 583-586. Dickson, J.A.D., 1965. A modified staining technique for carbonates in thin section. Nature, 205(4971): 587-&. Duan, Y. et al., 2010. Isotopic evidence for Fe cycling and repartitioning in ancient oxygen-deficient settings: Examples from black shales of the mid-to-late Devonian Appalachian basin. Earth and Planetary Science Letters, 290(3-4): 244253. Dubinin, A.V., 2006. Geochemistry of Rare-Earth elements in the Ocean. Nauka Publ., Moscow. Dubinin, A.V. and Rozanov, A.G., 2001. Geochemistry of rare earth elements and thorium in sediments and ferromanganese nodules of the Atlantic Ocean. Lithology and Mineral Resources, 36(3): 268-279. Feely, R.A. et al., 1987. Composition and dissolution of black smoker particulates from active vents on the Juan-de-Fuca Ridge. Journal of Geophysical Research-Solid Earth and Planets, 92(B11): 11347-11363. Foster, P. and Chacko, J., 1995. Minor and trace-elements in the shell of patella-vulgata (l). Marine Environmental Research, 40(1): 55-76. Foster, P. and Cravo, A., 2003. Minor elements and trace metals in the shell of marine gastropods from a shore in tropical East Africa. Water Air and Soil Pollution, 145(1): 53-65. Gill, B.C., Lyons, T.W. and Saltzman, M.R., 2007. Parallel, high-resolution carbon and sulfur isotope records of the evolving Paleozoic marine sulfur reservoir. Palaeogeography Palaeoclimatology Palaeoecology, 256(3-4): 156-173. Govin, A. et al., 2012. Distribution of major elements in Atlantic surface sediments (36 degrees N-49 degrees S): Imprint of terrigenous input and continental weathering. Geochemistry Geophysics Geosystems, 13. Gromov, V.V., 1975. Absorption of Iron Group Elements by Bottom Sediments of Pacific Ocean. Okeanologiya, 15(1): 59-65. Hurtgen, M.T., Pruss, S.B. and Knoll, A.H., 2009. Evaluating the relationship between the carbon and sulfur cycles in the later Cambrian ocean: An example from the Port au Port Group, western Newfoundland, Canada. Earth and Planetary Science Letters, 281(3-4): 288-297. 27 Jilbert, T. and Slomp, C.P., 2013. Iron and manganese shuttles control the formation of authigenic phosphorus minerals in the euxinic basins of the Baltic Sea. Geochimica et Cosmochimica Acta, 107: 155-169. Kahle, C.F., 1965. Possible roles of clay minerals in formation of dolomite. Journal of Sedimentary Petrology, 35(2): 448-&. Keller, N.B., Demina, L.L. and Os'kina, N.S., 2007. Variations in the chemical composition of the skeletons of non-zooxanthellate scleractinian (Anthozoa : Scleractinia) corals. Geochemistry International, 45(8): 832-839. Kiessling, W., Simpson, C. and Foote, M., 2010. Reefs as Cradles of Evolution and Sources of Biodiversity in the Phanerozoic. Science, 327(5962): 196-198. Klinkhammer, G. and Hudson, A., 1986. Dispersal patterns for hydrothermal plumes in the South-Pacific using manganese as a tracer. Earth and Planetary Science Letters, 79(3-4): 241-249. Kumar, S.K., Chandrasekar, N. and Seralathan, P., 2010. Trace Elements Contamination in Coral Reef Skeleton, Gulf of Mannar, India. Bulletin of Environmental Contamination and Toxicology, 84(1): 141-146. Land, L.S., 1998. Failure to precipitate dolomite at 25 degrees C from dilute solution despite 1000-fold oversaturation after 32 years. Aquatic Geochemistry, 4(Z3-4): 361-368. Leslie, B., Hammon, D., Berelson, W. and Lund, S., 1990. Diagenesis in anoxic sediments from the California continental borderland and its influence on iron, sulfur and megnetite behavious. Journal of Geophysical Research, 95(0148-0227). Lokier, S. and Steuber, T., 2008. Quantification of carbonate-ramp sedimentation and progradation rates for the late Holocene Abu Dhabi shoreline. Journal of Sedimentary Research, 78(7-8): 423-431. Lupton, J.E. and Craig, H., 1981. A major He-3 source at 15-degrees-S on the East Pacific Rise. Science, 214(4516): 13-18. Lyle, M.W., 1986. Major element composition of leg-92 sediments. Initial Reports of the Deep Sea Drilling Project, 92: 355-370. Lyons, T.W., Anbar, A.D., Severmann, S., Scott, C. and Gill, B.C., 2009. Tracking Euxinia in the Ancient Ocean: A Multiproxy Perspective and Proterozoic Case Study. Annual Review of Earth and Planetary Sciences, 37: 507-534. Lyons, T.W., Planavsky, N.J., Reinhard, C. and Raiswell, R., 2012. Revisiting iron-based paleoredox proxies in light of diverse primary and secondary controls and overprints: A story of cautious optimism., Fall Meeting, AGU, San Francisco, Calif. Lyons, T.W. and Severmann, S., 2006. A critical look at iron paleoredox proxies: New insights from modern euxinic marine basins. Geochimica et Cosmochimica Acta, 70(23): 5698-5722. 28 Lyons, T.W., Werne, J.P., Hollander, D.J. and Murray, R.W., 2003. Contrasting sulfur geochemistry and Fe/Al and Mo/Al ratios across the last oxic-to-anoxic transition in the Cariaco Basin, Venezuela. Chemical Geology, 195(1-4): 131-157. März, C. et al., 2008. Redox sensitivity of P cycling during marine black shale formation: Dynamics of sulfidic and anoxic, non-sulfidic bottom waters. Geochimica et Cosmochimica Acta, 72(15): 3703-3717. Meyer, K.M. and Kump, L.R., 2008. Oceanic euxinia in Earth history: Causes and consequences. Annual Review of Earth and Planetary Sciences, 36: 251-288. Middelburg, J.J., 1991. Organic-carbon, sulfur, and iron in recent semi-euxinic sediments of Kau Bay, Indonesia. Geochimica et Cosmochimica Acta, 55(3): 815-828. Mobius, J., Lahajnar, N. and Emeis, K.C., 2010. Diagenetic control of nitrogen isotope ratios in Holocene sapropels and recent sediments from the Eastern Mediterranean Sea. Biogeosciences, 7: 3901-3914. Mottl, M.J. and McConachy, T.F., 1990. Chemical processes in buoyant hydrothermal plumes on the East Pacific Rise near 21-degrees-N. Geochimica et Cosmochimica Acta, 54(7): 1911-1927. Newton, R.J., Pevitt, E.L., Wignall, P.B. and Bottrell, S.H., 2004. Large shifts in the isotopic composition of seawater sulphate across the Permo-Triassic boundary in northern Italy. Earth and Planetary Science Letters, 218(3-4): 331-345. Planavsky, N.J., Bekker, A., Hofmann, A., Owens, J.D. and Lyons, T.W., 2012. Sulfur record of rising and falling marine oxygen and sulfate levels during the Lomagundi event. Proceedings of the National Academy of Sciences of the United States of America, 109(45): 18300-18305. Poulton, S.W. and Canfield, D.E., 2005. Development of a sequential extraction procedure for iron: implications for iron partitioning in continentally derived particulates. Chemical Geology, 214(3-4): 209-221. Poulton, S.W. and Canfield, D.E., 2006. Co-diagenesis of iron and phosphorus in hydrothermal sediments from the southern East Pacific Rise: Implications for the evaluation of paleoseawater phosphate concentrations. Geochimica et Cosmochimica Acta, 70(23): 5883-5898. Poulton, S.W. and Canfield, D.E., 2011. Ferruginous Conditions: A Dominant Feature of the Ocean through Earth's History. Elements, 7(2): 107-112. Poulton, S.W., Fralick, P.W. and Canfield, D.E., 2004b. The transition to a sulphidic ocean similar to 1.84 billion years ago. Nature, 431(7005): 173-177. Poulton, S.W., Fralick, P.W. and Canfield, D.E., 2010. Spatial variability in oceanic redox structure 1.8 billion years ago. Nature Geoscience, 3(7): 486-490. 29 Poulton, S.W., Krom, M.D. and Raiswell, R., 2004a. A revised scheme for the reactivity of iron (oxyhydr)oxide minerals towards dissolved sulfide. Geochimica et Cosmochimica Acta, 68(18): 3703-3715. Poulton, S.W., Krom, M.D., Van Rijn, J. and Raiswell, R., 2002. The use of hydrous iron (III) oxides for the removal of hydrogen sulphide in aqueous systems. Water Research, 36(4): 825-834. Poulton, S.W. and Raiswell, R., 2002. The low-temperature geochemical cycle of iron: From continental fluxes to marine sediment deposition. American Journal of Science, 302(9): 774-805. Poulton, S.W. and Raiswell, R., 2005. Chemical and physical characteristics of iron oxides in riverine and glacial meltwater sediments. Chemical Geology, 218(3-4): 203221. Pufahl, P.K., Fralick, P.W. and Scott, J., 2000. Depositional environments of the Palaeoproterozoic Gunflint Formation. In: P. Fralick (Editor), Institute of Lake Superior Geology Field Guide, pp. 46. Raiswell, R. and Anderson, T.F., 2005. Reactive iron enrichment in sediments deposited beneath euxinic bottom waters: constraints on supply by shelf recycling. In: I. McDonald, A.J. Boyce, I.B. Butler, R.J. Herrington and D.A. Polya (Editors), Mineral Deposits and Earth Evolution. Geological Society Special Publication, pp. 179-194. Raiswell, R. and Canfield, D., 2012. The iron biogeochemical cycle past and present, 1. Geochemical Perspectives. Raiswell, R. and Canfield, D.E., 1998. Sources of iron for pyrite formation in marine sediments. American Journal of Science, 298(3): 219-245. Raiswell, R. et al., 2008. Turbidite depositional influences on the diagenesis of Beecher's Trilobite Bed and the Hunsruck Slate; Sites of soft tissue pyritization. American Journal of Science, 308(2): 105-129. Raiswell, R., Newton, R. and Wignall, P.B., 2001. An indicator of water-column anoxia: Resolution of biofacies variations in the Kimmeridge Clay (Upper Jurassic, UK). Journal of Sedimentary Research, 71(2): 286-294. Reid, J.L., 1982. Evidence of an effect of heat-flux from the East Pacific Rise upon the characteristics of the mid-depth waters. Geophysical Research Letters, 9(4): 381384. Richoz, S., 2006. Stratigraphie et variations isotopiques du carbone dans le Permien supérieur et le Trias inférieur de quelques localitiés de la Néothéthys (Turquie, Oman et Iran). Mémoires de Géologie (Lausanne) 46: 275. Richoz, S. et al., 2010. Permian-Triassic boundary interval in the Middle East (Iran and N. Oman): Progressive environmental change from detailed carbonate carbon isotope marine curve and sedimentary evolution. Journal of Asian Earth Sciences, 39(4): 236-253. 30 Severmann, S., Lyons, T.W., Anbar, A., McManus, J. and Gordon, G., 2008. Modern iron isotope perspective on the benthic iron shuttle and the redox evolution of ancient oceans. Geology, 36(6): 487-490. Tobin, K.J., Walker, K.R., Srinivasan, K. and Steinhauff, D.M., 1996. Suboxic to anoxic diagenesis of platform-marginal ooids and bladed-to-fibrous calcite from the Middle Ordovician Ottosee formation (east Tennessee). Geological Society of America Bulletin, 108(2): 155-167. Trefry, J.H. and Presley, B.J., 1982. Manganese fluxes from Mississippi delta sediments. Geochimica et Cosmochimica Acta, 46: 1715-1726. Tucker, M.E. and Wright, P.V., 1990. Carbonate Sedimentology. Blackwell Science, 482 pp. Turekian, K.K., Katz, A. and Chan, L., 1973. Trace-element trapping in pteropod tests. Limnology and Oceanography, 18(2): 240-249. Van der Weijden, C.H., 2002. Pitfalls of normalization of marine geochemical data using a common divisor. Marine Geology, 184(3-4): 167-187. Warren, J., 2000. Dolomite: occurrence, evolution and economically important associations. Earth-Science Reviews, 52(1-3): 1-81. Zegeye, A. et al., 2012. Green rust formation controls nutrient availability in a ferruginous water column. Geology, 40(7): 599-602. Figure Captions Figure 1: Plot showing Fe and Al contents as a function of CaCO3. The data are separated into oxic normal marine, anoxic marine and hydrothermal settings. Figure 2: Fe/Al ratios as a function of CaCO3, with the same classifications as in Fig. 1. Dashed lines represent the normal oxic marine average Fe/Al ratio (± 1 s.d.). Figure 3: Fe speciation data plotted against CaCO3 for modern (circles) and Phanerozoic (squares) oxic (open) and anoxic (closed) samples. Dashed lines at 0.22 and 0.38 FeHR/FeT, and represent the oxic and anoxic interpretative thresholds (Poulton and Canfield, 2011). Figure 4: Fe speciation data plotted against FeT for modern (circles) and Phanerozoic (squares) samples, with oxic (open) and anoxic (closed) settings distinguished. Dashed lines at 0.22 and 0.38 FeHR/FeT represent the oxic and anoxic interpretative thresholds (Poulton and Canfield, 2011). Figure 5: FeHR/FeT plotted as a function of TOC. Dashed lines at 0.22 and 0.38 FeHR/FeT represent the oxic and anoxic interpretative thresholds (Poulton and Canfield, 2011). Samples with FeT <0.5 wt% are plotted as triangles. 31 Figure 6: FeT plotted as a function of TOC. Open triangles represent oxic samples with FeT <0.5 wt% (these samples record spuriously high FeHR/FeT ratios). 32 Tables Table 1. Data sources for modern Fe/Al core-top calibration. Environment Normal Oxic Marine: Atlantic East Pacific North Atlantic North Pacific Mediterranean Reference Govin et al. (2012) Gromov (1975) Dubinin and Rozanov (2001) Dubinin (2006) Mobius et al. (2010) Anoxic Black Sea Cariaco Basin Kau Bay Raiswell and Canfield (1998) Lyons et al. (2003) Raiswell and Canfield (1998) Hydrothermal East Pacific Rise Mid-Atlantic Ridge Lyle (1986); Dubinin (2006) Dubinin (2006); Govin et al. (2012) 33 Table 2. Sample descriptions for new Fe-speciation data. Redox Oxic ID Ooids Pecten Razor Clam Coquina Coral Description Abiotic ooid Biogenic Locality Abu Dhabi Shallow water, UK Age Holocene Holocene Reference This study This study Biogenic Shallow water, UK Holocene This study Biogenic Biogenic Holocene Holocene Strom. lith Biogenic Strom. unlith Biogenic Sharks Bay, Aus. Caribbean Lagoa Vemelha, Brazil Lagoa Vemelha, Brazil AD xx Carbonate ramp sands & mud Abu Dhabi Holocene GBR xx Inter reef carbonate sands This study This study Vasconcelos et al., 2006 Vasconcelos et al., 2006 Lockier and Steauber, 2008 Scoffin and Tudhope, 1985 N.Uist 6 905/14 Azagador Anoxic Abad Marl BS carb BS ooze Yazerez Temperate Carbonate sands Carbonate slope sediment Temperate shallow water Limestone Pelagic Diatom/carbonate Authigenic carbonate Unit 1 coccolith ooze Deep water Carbonate/ gypsum Great Barrier Reef, Aus. North Uist, Scotland Oman Margin Holocene Holocene Holocene Holocene This study Holocene This study Mediterranean Miocene Weijermars, 1991 Mediterranean Black Sea Black Sea Miocene Holocene Holocene Weijermars, 1991 Bahr et al., 2009 Bahr et al., 2009 Mediterranean Miocene Stefano et al., 2012 34 Table 3. Published data-sets used for Fe speciation and Fe/Al compilations. Reference Raiswell and Berner, 1986 Canfield et al., 1996 Raiswell and Canfield, 1998 Raiswell et al., 2001 Dubinin and Rozanov, 2001 Lyons et al., 2003 Locality Robin Hoods Bay, Great Paxton, Lillingstone Lovell, Deanshanger, Tattenhoe, Snake Hill, Oslo, Cautley, Ohio Black Sea Framvaren, Black Sea, Orca Basin, Kau Bay Kimmeridge Clay, Jurassic, UK Trans Atlantic Cariaco Basin 35 Table 4. New Fe-speciation data for carbonate-rich samples. Ordered by ascending FeT. BD = below detection (0.05 wt% for TOC). Oxic Anoxic Sample Razor Clam GBR 38 Pecten Coquina Strom. lith Ooids Coral GBR 41 AD B15 AD S31 AD S25 AD B8 Azagador Strom. unlith AD S70 AD S36 AD S67 GBR 36 AD B14 AD S21 N.Uist 6 AD S64 905/14 Abad Marl BS ooze BS cement Yazerez BD BD 0.16 BD 0.06 BD BD BD BD BD BD 2.12 0.19 0.44 0.65 BD BD 0.09 0.12 0.31 1.98 0.31 CaCO3 % 98.64 89.85 94.77 92.62 93.31 89.15 92.73 96.84 85.42 54.86 64.51 60.54 90.79 59.33 33.45 30.37 52.02 93.31 87.68 32.24 77.84 68.66 66.31 41.53 FeT 0.004 0.008 0.014 0.021 0.023 0.035 0.068 0.073 0.085 0.089 0.117 0.132 0.201 0.231 0.383 0.408 0.415 0.439 0.641 0.644 0.648 0.737 0.923 2.086 Fecarb 0.028 0.033 0.018 0.020 0.026 0.023 0.068 0.052 0.049 0.087 0.064 0.061 0.072 0.071 0.063 0.101 0.047 0.412 wt% Feox 0.009 0.022 0.011 0.010 0.003 0.012 0.078 0.116 0.046 0.059 0.048 0.265 0.094 0.118 0.127 0.107 0.174 0.135 Femag 0.012 0.025 0.010 0.013 0.016 0.015 0.013 0.028 0.026 0.033 0.030 0.036 0.037 0.039 0.023 0.044 0.050 0.096 Fepy 0.036 0.006 0.012 0.004 0.043 0.018 0.004 0.021 0.046 0.085 0.043 0.013 0.015 0.022 0.019 0.012 0.086 0.123 FeHR/FeT 1.23 1.17 0.62 0.54 0.75 0.52 0.81 0.94 0.43 0.65 0.44 0.85 0.34 0.39 0.36 0.36 0.39 0.37 Fecarb 0.33 0.38 0.35 0.42 0.29 0.34 0.42 0.24 0.29 0.33 0.35 0.16 0.33 0.29 0.27 0.38 0.13 0.54 FeX/FeHR Feox Femag 0.10 0.14 0.26 0.29 0.22 0.20 0.21 0.28 0.03 0.19 0.17 0.22 0.48 0.08 0.54 0.13 0.28 0.15 0.22 0.12 0.26 0.16 0.71 0.10 0.43 0.17 0.47 0.15 0.55 0.10 0.40 0.17 0.49 0.14 0.18 0.13 Fepy 0.43 0.08 0.24 0.09 0.49 0.27 0.02 0.10 0.28 0.32 0.23 0.03 0.07 0.09 0.08 0.05 0.24 0.16 4.46 0.28 2.48 86.86 79.65 56.25 0.441 0.696 1.598 0.086 0.148 0.089 0.071 0.035 0.736 0.017 0.062 0.042 0.150 0.220 0.001 0.74 0.67 0.54 0.27 0.32 0.10 0.22 0.07 0.85 0.46 0.47 0.00 TOC BD 0 BD 0.05 0.13 0.05 36 Table 5. Fe-speciation data for limestone-dolomite sample pairs. Sample Dol.1 Lst. 1 Dol. 2 Lst. 2 Dol. 3 Lst. 3 Dol. 4 Lst. 4 Dol. 5 Lst. 5 FeT 0.364 0.438 0.350 0.367 0.422 0.316 0.826 0.521 1.050 0.030 Fecarb 0.065 0.054 0.060 0.057 0.276 0.032 0.663 0.011 0.930 0.011 wt% Feox 0.196 0.244 0.187 0.224 0.138 0.167 0.206 0.500 0.163 0.014 Femag 0.020 0.015 0.019 0.018 0.017 0.016 0.000 0.000 0.000 0.000 Fepy 0.002 0.001 0.001 0.001 0.005 0.001 0.018 0.003 0.013 0.000 FeHR/FeT 0.78 0.72 0.76 0.82 1.03 0.68 1.07 0.99 1.05 0.84 Fecarb 0.23 0.17 0.23 0.19 0.63 0.15 0.75 0.02 0.84 0.43 FeX/FeHR Feox Femag 0.69 0.07 0.78 0.05 0.70 0.07 0.75 0.06 0.32 0.04 0.77 0.07 0.23 0.00 0.97 0.00 0.15 0.00 0.57 0.00 Fepy 0.01 0.00 0.01 0.00 0.01 0.00 0.02 0.01 0.01 0.00 37 Table 6. Summary of threshold values for the use of Fe speciation and Fe/Al as paleoredox proxies in carbonate-rich sediments that have not experienced Fe addition during deep burial dolomitization. NA = not applicable (i.e., no threshold value required). Water column FeT TOC Redox (wt%) (wt%) FeHR/FeT Fe/Al Oxic >0.5 NA <0.22 0.53 ± 0.11 Anoxic NA >0.5 >0.38 >0.64 38