Earth and Planetary Science Letters 389 (2014) 209–220

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Trace element content of sedimentary pyrite as a new proxy fordeep-time ocean–atmosphere evolution

Ross R. Large a,!, Jacqueline A. Halpin a, Leonid V. Danyushevsky a,Valeriy V. Maslennikov b, Stuart W. Bull a, John A. Long c,d, Daniel D. Gregory a,Elena Lounejeva a, Timothy W. Lyons e, Patrick J. Sack f, Peter J. McGoldrick a,Clive R. Calver g

a Centre for Ore Deposit and Exploration Science (CODES), University of Tasmania, Private Bag 126, Hobart, Tasmania 7001, Australiab Institute of Mineralogy, Russian Academy of Science, Urals Branch, Miass, Russiac School of Biological Sciences, Flinders University, PO Box 2100, Adelaide, South Australia 5001, Australiad Museum Victoria, PO Box 666, Melbourne, Victoria 3001, Australiae Department of Earth Sciences, University of California, Riverside, CA 92521, USAf Yukon Geological Survey, PO Box 2703 (K-102), Whitehorse, Yukon Y1A 2C6, Canadag Mineral Resources Tasmania, PO Box 56, Rosny Park, Tasmania 7018, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Received 29 July 2013Received in revised form 8 December 2013Accepted 15 December 2013Available online xxxxEditor: G. Henderson

Keywords:palaeo-oceanographysedimentary pyritetrace elementsocean chemistryoxygenation proxyselenium

Sedimentary pyrite formed in the water column, or during diagenesis in organic muds, provides anaccessible proxy for seawater chemistry in the marine rock record. Except for Mo, U, Ni and Cr,surprisingly little is known about trace element trends in the deep time oceans, even though theyare critical to developing better models for the evolution of the Earth’s atmosphere and evolutionarypathways of life. Here we introduce a novel approach to simultaneously quantify a suite of trace elementsin sedimentary pyrite from marine black shales. These trace element concentrations, at least in a first-order sense, track the primary elemental abundances in coeval seawater. In general, the trace elementpatterns show significant variation of several orders of magnitude in the Archaean and Phanerozoic,but less variation on longer wavelengths in the Proterozoic. Certain trace elements (e.g., Ni, Co, As,Cr) have generally decreased in the oceans through the Precambrian, other elements (e.g., Mo, Zn, Mn)have generally increased, and a further group initially increased and then decreased (e.g., Se and U).These changes appear to be controlled by many factors, in particular: 1) oxygenation cycles of the Earth’socean–atmosphere system, 2) the composition of exposed crustal rocks, 3) long term rates of continentalerosion, and 4) cycles of ocean anoxia. We show that Ni and Co content of seawater is affected by globalLarge Igneous Province events, whereas redox sensitive trace elements such as Se and Mo are affectedby atmosphere oxygenation. Positive jumps in Mo and Se concentrations prior to the Great OxidationEvent (GOE1, c. 2500 Ma) suggest pulses of oxygenation may have occurred as early as 2950 Ma. A flat todeclining pattern of many biologically important nutrient elements through the mid to late Proterozoicmay relate to declining atmosphere O2, and supports previous models of nutrient deficiency inhibitingmarine evolution during this period. These trace elements (Mo, Se, U, Cu and Ni) reach a minimum inthe mid Cryogenian and rise abruptly toward the end of the Cryogenian marking the position of a secondGreat Oxidation Event (GOE2).

! 2013 Elsevier B.V. All rights reserved.

1. Introduction

Whole rock variation of trace elements (TE) — in particular Mo,U, V, Cr and Ni — in black shales deposited in marine environmentshave been successfully used to interpret palaeo-redox conditionson the ocean floor (Lyons et al., 2003; Tribovillard et al., 2006;Reinhard et al., 2013), including temporal changes in oxygenation

* Corresponding author.E-mail address: ross.large@utas.edu.au (R.R. Large).

of the global ocean (Holland, 1984; Anbar et al., 2007; Scott etal., 2008; Partin et al., 2013). Most of the redox sensitive TE, aswell as multitude of other TE in black shales, are concentrated inearly-formed pyrite (Huerta-Diaz and Morse, 1992; Large et al.,2007, 2009; Gregory et al., in press). Laser Ablation-InductivelyCoupled Plasma Mass Spectrometer (LA-ICPMS) techniques devel-oped to study zonation in the composition of pyrite (FeS2) in var-ious ore systems (Danyushevsky et al., 2011) have demonstratedhow this information can be used to study potential sources ofmetals (Meffre et al., 2008), track changes in the chemistry of hy-drothermal fluids (Large et al., 2009), determine the evolution of

0012-821X/$ – see front matter ! 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.epsl.2013.12.020

210 R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220

ore deposits (Thomas et al., 2011) and establish the effectivenessof pyrite as a trap for TE in contaminated sediments (Gregoryet al., 2013, in press). We suggest here, that in the same way thathydrothermal pyrite tracks changes in chemistry of ore fluids, sedi-mentary pyrite, which is a common trace mineral in carbonaceousmarine black shales (Fig. 1a–d), can be used to track first-orderchanges in the chemistry of seawater and ultimately oxygenationof the ocean–atmosphere system. Whereas bulk shale studies re-quire samples of particular palaeo-redox facies (Lyons et al., 2003;Scott et al., 2008; Algeo and Rowe, 2012), the focus on early-formed pyrites developed here allows sampling of a wider rangeof redox conditions, enabling a more complete record of seawatervariation. For this study we have analysed over 1880 sedimentarypyrites from 131 black shale samples covering 77 time intervalsfrom 3515 Ma to present day (Supplementary Tables 1–4).

2. TE analysis of sedimentary pyrite as a proxy for seawaterchemistry

2.1. Incorporation of TE in sedimentary pyrite

Sedimentary pyrite forms in seafloor muds when H2S, pro-duced by the microbial reduction of marine sulfate, reacts withiron to form pyrite (FeS2), arguably by way of an iron mono-sulfide precursor, such as mackinawite or greigite (Rickard andLuther, 1997; Rickard, 2012). Experimental studies (Huerta-Diazand Morse, 1992; Morse and Arakaki, 1993) have shown thatmany TE are incorporated into the precursor iron monosulfideat an early stage, including (in approximate order of incorpora-tion) As, Hg, Mo, Co, Cu, Mn, Ni, Cr, Pb, Zn and Cd, which areabsorbed from seawater and local pore waters in seafloor muds(Larget et al., 2007, 2009; Gregory et al., in press). This TE en-richment process is controlled by the amount of pyrite producedand the amount of trace metals available from seawater and porewaters (Huerta-Diaz and Morse, 1992). However it is recognisedthat microbially controlled reduction and sequestration of TE andsulfate on the seafloor are complex biochemical processes. Vari-ables like Eh/pH conditions in the sediment/water column, de-gree of bio-productivity and availability of preferred electron ac-ceptors will influence which TE the microbes reduce and co-precipitate with pyrite at any given time (Kulp and Pratt, 2004;Mitchell et al., 2012).

Various textural forms of pyrite are typical of syngenetic/dia-genetic sedimentary origins, including framboids, microcrystallineclusters, small nodules and layers of fine crystals (Fig. 1a–d).Late diagenetic pyrites typically overgrow earlier pyrite genera-tions (Thomas et al., 2011; Rickard, 2012) and commonly havea rounded anhedral surface which may be expressed as elon-gate nodules or various spheroidal and nodular forms (Large etal. 2007, 2009). Recent studies have shown that many of theTE originally thought to be present in the structure of pyrite,are actually present as micro-inclusions (Ciobanu et al., 2009;Large et al., 2009; Deditius et al., 2011) abundant in early formsof pyrite (Fig. 1a–c). Whereas Ni, Co, and Se are commonly presentin the pyrite structure, As, Cu, Zn, Pb, Bi, Sb, Tl, Mo, Ag, Cd, Mn,Hg, and Te may be either in the structure or as sulfide micro-inclusions, and Ti, V, U, Ba, Sn, W, and Cr, are invariably presentwithin micro-inclusions of matrix material. This earliest formedinclusion-bearing sedimentary pyrite is the most TE rich, and sub-sequent, more crystalline generations with fewer inclusions (latediagenetic to early metamorphic) may contain lower TE concentra-tions and trend toward a pure FeS2 composition (Huerta-Diaz andMorse, 1992; Large et al., 2007; Large et al., 2009; Rickard, 2012).Trace elements that are loosely held in the pyrite structure or oc-cur as nano-inclusions (e.g., Zn, Cu, Pb, Ag, Mo) are readily expelledfrom the pyrite during metamorphic recrystallisation, whereas TE

Fig. 1. a–d, Various textures of sedimentary pyrite; a, cluster of framboids, b, cloudof fine framboids and microcrystals, c, nodule, d, fine microcrystals concentratedin a sedimentary bed. e–j, LA-ICPMS images of TE distribution in pyrite nodule(c above) and surrounding matrix. Note the high concentration of Se in pyrite rela-tive to matrix in (e), compared with low concentration of V in the pyrite relative tomatrix in j. k, time resolved LA-ICPMS spectra of selected TE in a sedimentary pyriteanalysis. The laser was switched on at 31 s after the start of the analysis. Ablationcontinued for 60 s. The intensity of the signal for each isotope is taken as the dif-ference between the mean signal with laser off (background, the first 30 s of eachacquisition) and the mean signal with the laser on. As the vertical axis is countsrecorded by the detector on ICPMS, differences in intensity of the signals from dif-ferent isotopes do not directly reflect differences in concentrations. The plot showsthe differing response from TE in the pyrite structure (Pb, Co, Sb and Se) comparedto TE in micro-inclusions (Ti, Zn and Cr). Copper shows a dual response.

R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220 211

Table 1Summary of TE variations in the Archean and Proterozoic oceans based on sedimentary pyrite composition.

Eon/Era Trace elements enriched* in ocean Trace elements depleted* in ocean Little change in mean trace elementconcentration*

Across GOE2 (c. 660 Ma) Mo, Se, Zn, Cd, Ni, Co, As, Cu, Mn,Hg, Pb, Sb, Te, U

Bi, Ba, W, Sn Ag, Cr, V, Tl

Neoproterozoic to 660 Ma# Cu, Bi, Ba Mo, Zn, Ni, As, Se, Cd, Tl, Hg, Cr, Ag,Sb, Te

Co, Mn, Pb, U, V, W, Sn

Mesoproterozoic As, Tl Mo, Co, Mn, Hg, U, Cr, Ba, V, W, Te Zn, Ni, Se, Cd, Cu, Pb, Bi, Sn, Ag, SbPalaeoproterozoic Cu, Bi, U, Cr, V, W Mo, Zn, As, Cd, Co, Tl, Hg, Ba, Ag, Sb,

TeNi, Se, Mn, Pb, Sn

Across GOE1 (c. 2500 Ma) Mo, Se, Tl, Mn, U Zn, Cd, Ni, Co, As, Cu, Ag, Sb, Hg, Te,Pb, Bi, Cr, Ba, Sn

V, W

Archean Ni, As, Co, Cu, Hg, Bi, Cr, W, Sn, Sb, Te Mo, Zn, Se, Cd, Mn, U, Ba Tl, Pb, V, Ag

* Compared with the range of average values through time (0 to 3.5 Ga).# Neoproterozoic prior to GOE2 (Tonian-Cryogenian).

that are stoichiometric in the pyrite structure (e.g., Co, Ni, As, Se)tend to remain and even become enriched during metamorphism(Large et al., 2007, 2011b).

2.2. Sample selection, rationale and analysis

In this study, samples of sedimentary pyrite were selected onthe basis of: (1) metamorphic grade of the host shale, (2) textureof pyrite and (3) Co/Ni ratio of pyrite, as outlined below.

1. Black shales of mid greenschist facies or below were chosen.Post lithification metamorphism and/or hydrothermal activityleads to recrystallisation of earlier pyrite forms and commonlyresults in pyrite overgrowths with euhedral outer surfaces andwell developed crystal growth zones with alternation of TEchemistry (Large et al., 2009, 2013; Thomas et al., 2011). Theselatter euhedral and zoned pyrite types have been studiouslyavoided in this current study, as their chemistry reflects thecomposition and conditions of the metamorphic or hydrother-mal fluids rather than the precursor seawater chemistry duringsedimentation. Samples with pyrrhotite were also rejected, asprevious studies have shown a marked loss of TE during con-version of pyrite to pyrrhotite (Thomas et al., 2011).

2. Pyrite was selected that retained textures indicative of syn-genesis or formation during early diagenesis. Textures indica-tive of diagenetic pyrite include; framboids, microcrystallinepyrite in clouds in the shale, microcrystalline pyrite alignedparallel to bedding, small cubic euhedral and bladed crystals,small porous patches and rounded or elongate zoned nod-ules (Fig. 1a–d) (Huerta-Diaz and Morse, 1992; Large et al.,2007, 2009; Guy et al., 2010). Pyrite textures indicative ofmetamorphic and/or hydrothermal growth, including mediumand large sized euhedral crystals (>1 mm), clear crystals with-out inclusions, and pyrite which cuts across the bedding asblebs or veins, were not selected for analysis. Textural dis-crimination between diagenetic and metamorphic pyrite tex-tures was assisted by etching pyrite with nitric acid (Largeet al., 2007), coupled with the LA-ICPMS mapping techniquewhen required (Fig. 1e–j, Supplementary Fig. 7) (Large et al.,2009, 2013; Thomas et al., 2011).

3. Chemical screening of the pyrite was undertaken following LA-ICPMS analysis. On a Co versus Ni plot, samples which showeda linear array parallel to Co/Ni = 1, and with Co/Ni < 2 wereselected as suitable. This is based on previous work (Bajwahet al., 1987; Large et al., 2009) that have shown diageneticpyrite commonly has Co/Ni ! 2, whereas hydrothermal pyritehas higher Co values that trend away from the Co/Ni = 1 line.More than 80% of our data was deemed acceptable on this ba-sis.

All LA-ICPMS analyses of pyrite were performed at the Cen-tre for Ore Deposit and Exploration Science (CODES), University ofTasmania (see Supplementary Information for LA-ICPMS method-ology). LA-ICPMS element maps clearly show those TE that areevenly distributed through the pyrite (e.g., Se in Fig. 1e), com-pared with those that are zoned (Cu, Fig. 1h), those with apatchy distribution (As, Fig. 1i), and those dominated by inclusions(Mo and Zn, Fig. 1f–g). The time-resolved signal of an individ-ual LA-ICPMS analysis (Fig. 1k) provides information on the likelyform of elements incorporation in the pyrite (Large et al., 2009;Thomas et al., 2011). Removal of the effects of matrix micro-inclusions leads to the pyrite compositions measured on severalpyrite grains from the same black shale sample commonly exhibit-ing a variation of less than one order of magnitude (ppm scale),compared with the overall dataset variation of up to four orders ofmagnitude for any given TE (see example in Supplementary Fig. 1).

2.3. Proof of concept

Five separate lines of evidence support our contention that vari-ations in the TE composition of early-formed sedimentary pyrite inblack shales are a good first order proxy for TE variations in theocean through time (Table 1). The physical/chemical connectionbetween early-formed sedimentary pyrite and seawater underpinsour first line of evidence. Syngenetic pyrite which forms in eu-xinic water columns adsorbs TE from the seawater (Huerta-Diazand Morse, 1992) is an obvious proxy for seawater TE chem-istry (Berner et al., 2013). However, diagenetic pyrite in anoxic todysoxic bottom water conditions forms by biogenic reduction ofsulfate in the pore waters of seafloor muds within the first fewcentimetres of the seafloor (Wilkin et al., 1996). In this environ-ment the muds contain from 60 to 85% seawater (Baldwin and But-ler, 1985; Harrold et al., 1999) and thus ensure a strong connectionbetween pyrite chemistry and seawater chemistry, even thoughsedimentary clays and organic matter are likely to play a role inpartitioning TE into pyrite. The seawater content of seafloor mudsremains over 40% in the top 100 to 200 m (Baldwin and Butler,1985) and thus diagenetic pyrites forming at considerable depthwill maintain a seawater connection influenced by matrix interac-tions. The significance of matrix/pore water interaction in effectinga change in sedimentary pyrite chemistry can be evaluated fromdata on pyrite TE depth profiles in recent sedimentary environ-ments from Huerta-Diaz and Morse (1990, 1992). Their data, col-lected from a range of anoxic to euxinic sedimentary environmentsin the Gulf of Mexico, shows that most TE in sedimentary pyriteincrease with depth from 0 up to 100 cm depth. However, the in-crease is variable, depending on sedimentary environment and theTE measured, and is commonly in the range of 1.4 to 6 times (lessthan one order of magnitude), compared to the 2 to 4 orders of

212 R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220

Fig. 2. a, Changes in mean LA-ICPMS analyses of Mo, As, Ni, Co and Se in the paragenetic stages of sedimentary pyrite from the Neoproterozoic Khomolkho Formation, Siberiafrom Large et al. (2007). Py1 = syngenetic to early diagenetic pyrite; py2 = early diagenetic pyrite; py3 = late diagenetic pyrite; py4 = metamorphic pyrite; py5 = latemetamorphic pyrite; pyrrhotite = late metamorphic pyrrhotite replacing pyrite. b, Range in TE content (Se, As, Ni and Mo) plotted against pyrite framboid size for a LatePermian black shale from the Perth Basin, Western Australia. The range in individual TE for each framboid size is similar.

magnitude variation exhibited by sedimentary pyrite in our sam-ples from different epochs and eras through time.

The second line of evidence comes from a study of the LA-ICPMS composition of a range of different textural and parageneticpyrite types, from syngenetic, through diagenetic to metamorphic,in the Neoproterozoic Khomolkho Formation in the Lena Gold-field, Siberia (Large et al., 2007). This study demonstrated a generaltrend of steady to decreasing mean TE content in pyrite with time,from syngenetic to diagenetic to metamorphic through the para-genesis (Fig. 2a). The changes during diagenesis for all five TE aresubstantially less than one order of magnitude. This suggests thatthe first order (2 to 4 orders of magnitude) temporal changes areunlikely to relate to diagenesis in the seafloor muds, but morelikely reflect the changes in TE content of seawater within andoverlying the muds.

The third line of evidence, based on our framboid data fromthe Perth Basin in Late Permian marine black shales (Thomaset al., 2004) shows there is no substantial difference in the TE(range ppm) of small (!5 µm) syngenetic framboids, comparedwith larger (6 to 35 µm) pyrite framboids (Fig. 2b). All sizes offramboids in this Late Permian black shale sample show similar TEcontents (40 to 800 ppm Ni and As; 3 to 80 Mo and Se; Fig. 2b)suggesting that early-formed sedimentary pyrite is a first orderrecorder of TE variations irrespective of its framboid size and modeof origin (syngenetic or diagenetic).

The fourth line of evidence is exhibited in Fig. 3, which demon-strates a first order positive relationship between the LA-ICPMS TEcomposition of sedimentary pyrite from an actively forming sed-imentary basin the Cariaco Basin on the Venezuela shelf (Lyonset al., 2003; Piper and Perkins, 2004), and the TE composition of

R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220 213

Fig. 3. Positive correlation (large dashed field) between mean TE concentrations incurrent ocean and mean TE concentrations in modern sedimentary pyrite from Cari-aco Basin. Vanadium, U and Ba do not fall in the same field as the other traces, asthey are concentrated in non-sulfide inclusions within the pyrite.

Fig. 4. Down hole comparison between LA-ICPMS Mo and Se analyses from wholerock and sedimentary pyrite in black shales from drill hole CD 12c, Carlin DistrictNevada (Large et al., 2011a).

modern ocean water (http://www.mbari.org/chemsensor/summary.html). Relative to seawater, TE in Cariaco Basin pyrite are enrichedbetween 5 and 8 orders of magnitude (Supplementary Table 1).Trace elements Mo, As, Ni, Zn, Cu and Se which are enriched inthe current ocean (>100 ng/kg) are also enriched in the currentlyforming sedimentary pyrite (>100 ppm), whereas Co, Pb, Sb, Tl,Cd and Ag which are in lower abundance in the ocean (>1 and<100 ng/kg) are also in lower abundance in sedimentary pyrite(>10 and <100 ppm). The least abundant TE in the ocean (Te,Bi, Au) are also the least abundant in Cariaco Basin sedimentarypyrite. Barium, V and U, which are present within silicate inclu-sions in the sedimentary pyrite, plot outside the pyrite correlationfield.

The final proof of concept for the approach proposed here in-volves comparing our data on the TE composition of pyrite withthe TE content of whole rock black shale samples, particularlyfor Mo, which is commonly used to interpret palaeo-ocean con-ditions and changes in oxygenation of the atmosphere/ocean sys-tem through time (Lyons et al., 2003; Tribovillard et al., 2006;Scott et al., 2008). A small group of black shale samples from theDevonian Popovich Formation, Nevada, reveal similar down-holetrends (Fig. 4); those TE concentrated in pyrite such as Mo andSe are commonly between 10 to 100 times the whole rock value,depending on the percentage of sedimentary pyrite and organicmatter in the sample.

In order to evaluate the deep time variability of TE in sedimen-tary pyrite in black shales, we compare Mo data on whole rockeuxinic and/or organic-rich black shales from Scott et al. (2008)with our data on the Mo content of sedimentary pyrite in blackshales over the same time interval (Fig. 5a and b). The two datapatterns are remarkably similar, with several times more Mo inpyrite compared with the whole rock values, even though Mo ispresent both in the sedimentary matrix and in the sedimentarypyrite (Fig. 1) (Chappaz et al., 2014). A second comparison is madewith the Ni/Fe ratio measured on whole rocks and minerals in ma-rine Banded Iron Formations (BIFs) (Konhauser et al., 2009) (Fig. 5cand d). Again the two datasets show similar patterns, although theKonhauser et al. (2009) plot contains no Phanerozoic data due tothe lack of BIFs in this era. In summary, our pyrite TE plots havereproduced the temporal trends in the previous whole rock dataand confirm the value of sedimentary pyrite as a proxy for deeptime variations of TE in the oceans.

3. Temporal trends in TE from sedimentary pyrite

Individual TE time curves derived from LA-ICPMS analysis of1885 pyrite spots on 131 black shale samples are given in theSupplementary Information (Supplementary Tables 2 and 3; Sup-plementary Figs. 3–5). We have selected a subset of the TE tofocus our discussion; Mo, Ni, Co, As and Se measured in sedi-mentary pyrite, and U and Cr measured in the sedimentary matrix(Figs. 6–8). In the summary figures to follow, we use a rolling TEmean to highlight the long-term TE trends (Fig. 6b–f, see also Sup-plementary Fig. 2) at the expense of the variation in individualsample sets (Fig. 6a). In general terms (see Mo, Ni, Co, As and Sein Fig. 6b–f), most TE exhibit significant cyclic variation of sev-eral orders of magnitude in the Archaean and Phanerozoic, but lessvariation on longer wavelengths in the Proterozoic. In the Precam-brian, there are four general TE trends:

Trend 1: a flat-to-broadly increasing pattern through the Archaeanand Proterozoic: only Mo, Zn, Mn, and Cd show this pattern.Although this pattern is not unidirectional, it suggests thatsupply of these TE to the ocean has exceeded drawdown intoseafloor muds through most of the Precambrian.

Trend 2: A variable, but generally decreasing pattern through theArchaean and Proterozoic: Ni, Co, As, Te, Tl, Hg, Sb, Pb, Bi, Cr,Sn, and W. This pattern suggests drawdown has exceeded sup-ply for the Precambrian.

Trend 3: a general increase through the Mesoarchaean to earlyPalaeoproterozoic followed by a decrease through the Meso-and Neoproterozoic: Se, U, and V. In the first half of the pe-riod supply exceeded drawdown, with reversal in the secondhalf.

Trend 4: a relatively flat pattern with no obvious trend throughthe Archaean and Proterozoic: Cu, Ag, and Ba. In this case TEsupply and drawdown were roughly matched through the pe-riod but with perturbations.

214 R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220

Fig. 5. Comparison of published data on temporal variation in whole rock Mo and Ni analyses on black shales and BIFs with LA-ICPMS analyses of sedimentary pyrite in blackshales. a, Mo whole rock black shale data from Scott et al. (2008) and b, Mo in sedimentary pyrite derived from this study. c, Ni/Fe whole rock and mineral data from BIFs(Konhauser et al., 2009) and d, Ni in sedimentary pyrite derived from this study.

All TE show high frequency and high amplitude variations throughthe Phanerozoic suggesting geologically rapid changes in oceanchemistry over this latter period of Earth history. It should be em-phasised that the four trends discussed above are a first attempt togroup the data, but in fact each TE shows its own individual pat-tern (Supplementary Figs. 3 and 5), that is probably controlled bya number of factors, which are beyond the scope of analysis here.A brief discussion of the temporal trends of the five TE shown inFig. 6, and comparison between these pyrite TE trends and thosepreviously published for whole rock datasets follows below.

Molybdenum (Trend 1) is a conservative TE in the currentocean, present as the soluble MoO4

= species (Helz et al., 1996).The general rise in the Mo time curve (Fig. 6b) is attributed tooxygenation of the atmosphere, particularly in the Neoarchean andPhanerozoic, leading to increased continental oxidative weatheringsuch that supply of Mo to the oceans exceeded the drawdowninto seafloor sediments (Anbar et al., 2007; Scott et al., 2008;Dahl et al., 2011). Our time curve shows that Mo in pyrite com-menced at a relatively low 0.5 to 1 ppm at 3500 Ma and increasedthrough a series of steps to reach 10 to 70 ppm at 1000 Ma. Thestep up at around 2500 Ma likely corresponds to the Great Oxi-dation Event (GOE1) (Farquhar et al., 2001; Holland, 2006) whenappreciable O2 first accumulated in the atmosphere. In the Neo-proterozoic there is a trough in the Mo time curve (referred to asthe Cryogenian TE trough hereafter) that concludes with the Stur-tian glaciation and proposed Snowball Earth (Hoffman et al., 1998).A rapid rise corresponding to a second Great Oxidation Event (re-ferred to here as GOE2) occurred during the Neoproterozoic whenoxygen rose to significant levels that ultimately supported the riseof animal diversity. Our TE data suggests a threshold was reachedby 660 Ma, earlier than recent estimates that place this event ei-ther shortly after the Marinoan Glaciation (630 Ma) (Sahoo et al.,2012) or prior to the late Ediacaran (580–550 Ma) (Fike et al.,2006; Canfield et al., 2007; Scott et al., 2008).

Nickel (Trend 2) is a nutrient TE present as Ni2+ in the mod-ern ocean. The Ni time curve (Fig. 6c) shows an overall decreasethrough the Precambrian to 660 Ma, but most of the drop oc-curs in the Archaean, decreasing from around 2500–10,000 ppm tojust under 1000 ppm through the Palaeoproterozoic. This pattern,also seen in Ni/Fe ratio in BIFs (Fig. 5c), is attributed to coolingupper mantle and decreased eruption of Ni-rich komatiite ultra-mafic rocks from about 2700 Ma (Konhauser et al., 2009). Cobalt

(Trend 2) shows a similar pattern to Ni, but with more varia-tion (Fig. 6d). Most of the segments on the time curve when Niand Co in pyrite exceeds 1000 ppm and 500 ppm respectively,overlap with episodes of Large Igneous Province (LIP) eruptions ofmafic lavas and associated intrusives, both subareal and submarine(Prokoph et al., 2004) (Fig. 7). The komatiitic LIPs of the Archaeanare particularly enriched in Ni and Co, and thus contributed to thesustained high levels of Ni and Co to the global oceans and sed-imentary pyrite (Fig. 7). The highest Phanerozoic Ni and Co weremeasured in sedimentary pyrite of 250 Ma from the Perth BasinWestern Australia; values up to 48,000 ppm Ni and 16,000 Cowere recorded, indicating the likelihood of widespread enrichmentof the global ocean in Ni and Co due to the Siberian Trap erup-tions. It is noteworthy that these Siberian eruptions are associatedwith the largest nickel ore bodies on Earth, Norilsk and Talnack.

Arsenic (Trend 2) (Fig. 6e) also exhibits a declining trendthrough the Precambrian. However, the As pattern has two seg-ments; 1) a continuous decrease from a mean of 5000 to 300 ppmin pyrite over the period 3000 to 1800 Ma; 2) followed by a rise to10,000 ppm at 1400 Ma and fall to the Cryogenian TE trough, ris-ing again at GOE2 around 660 Ma. The cause for the sharp declinein As through Late Archean and Early Proterozoic is not source-rock controlled, as unlike Ni and Co, As is not enriched in ultra-mafic or mafic rocks (Reimann and Caritat, 1998), and alternativeAs-bearing Archean source rocks are not known. A possible expla-nation for the As trend relates to the fact that As forms a solublearsenate complex (HAsO2"

4 ) in oxic to anoxic ocean conditions andsoluble polysulfide complex (H2As3S"

6 ) in euxinic ocean conditions(see Supplementary Table 5). Thus higher levels of volatile ele-ments, S and As, associated with a degassing Earth in the Archean,may have contributed to stable arsenic polysulfide complexes inthe mainly H2S-bearing Archean ocean, which declined in concen-tration through the Early Proterozoic as H2S levels in the oceandropped and As was drawn-down into seafloor pyrite. An increasein As through the Mesoproterozoic (Fig. 6e) supports a switch toanoxic conditions as the SO2"

4 /H2S ratio of the ocean continued toincrease, followed by an As decline as euxinic conditions took holdagain in the Neoproterozoic (Poulton and Canfield, 2011).

Selenium (Trend 3) (Fig. 6f), like Mo, is sourced from the ox-idative weathering of continental pyrite, with release of seleniteand selenate species (SeO2"

3 and SeO2"4 ; see Supplementary Fig. 6),

which accumulate in the ocean. Selenium is a nutrient TE, with

R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220 215

Fig. 6. a–f, Temporal trends of Mo, Ni, Co, As and Se measured by LA-ICPMS analysis of sedimentary pyrite in black shales. Each symbol represents a single pyrite analysis in(a). The curves presented in (b–f) are derived from a 19 point rolling mean (or moving average).

significant biological uptake and eventual accumulation on theseafloor in both organic matter and by substitution for S in sedi-mentary pyrite (Diener and Neumann, 2011; Mitchell et al., 2012).Se shows strong narrow peaks around 3000 and 2500 Ma, rising toa flat trend between 2500 and 1700 Ma, with a major perturbationfrom 1800 to 1300 Ma, then declines to a trough at 900 to 660Ma, before rising again at GOE2. The stoichiometric substitutionof Se for S in pyrite, its homogeneous distribution in sedimentarypyrite (Fig. 1e, Supplementary Fig. 7), high enrichment in pyriteversus the black shale matrix (Supplementary Table 4), low vari-ability in pyrite from the same sample (Supplementary Fig. 1), andtendency to remain in pyrite even after significant recrystallisation(Large et al., 2007; Thomas et al., 2011), indicate the Se contentof sedimentary pyrite may be a superior oxygenation proxy com-pared to other commonly used proxies, as further discussed be-low.

3.1. Selenium as a proxy for deep time atmosphere–ocean oxygenation

In Fig. 8 we compare the temporal trend of Se in pyrite withMo, U and Cr, three of the commonly employed whole rockproxies for atmosphere–ocean oxygenation (Anbar et al., 2007;Scott et al., 2008; Partin et al., 2013; Reinhard et al., 2013). Allfour redox-sensitive TE show related trends through the Archeanand Proterozoic.

The temporal trend of U (Fig. 8c) is broadly similar to Se(Fig. 8b), stepping up through the late Archean and Early Protero-zoic to a broad peak around 1700–1400 Ma, and then decreasinginto the Cryogenian TE trough from 1000 to 660 Ma. The broad Upeak around the mid Proterozoic may relate to the abundance ofhot U-bearing granites in the Paleo- and Mesoproterozoic (Wilsonand Akerblom, 1982; Wyborn et al., 1992) that would have pre-sented an enriched source rock of U for the contemporary oceans.Chromium (Fig. 8d), like Ni and Co (Fig. 7), is elevated in the

216 R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220

Fig. 7. Intervals of major LIPs (Prokoph et al., 2004), juxtaposed with the time curvesfor Co and Ni. Most highs for Ni and Co in sedimentary pyrite coincide with theperiods of high numbers of LIPs (green bars). The extreme minimum in Ni and Coaround 400 Ma is a good match for a period of minimum LIP activity from 300 to500 Ma, whereas the maximum at 250 Ma matches the Siberian Trap LIP eruptions.

Archean due to the abundant ultramafic/mafic crustal componentsand shows a decline from 2200 to 1200 Ma before rising to aplateau around 800 Ma. The Cr data shows a very similar trendto that recently published for organic-rich shales using whole-rockanalytical techniques (Reinhard et al., 2013).

Through the Archean, Se (Fig. 8b) averages around 10 ppm inpyrite but with a spike at 2970 Ma and steps upwards between2650 and 2500 Ma. The latter steps, to over 100 ppm Se, are manytens of millions years earlier than the accepted position of GOE1between 2450 to 2300 Ma (Farquhar et al., 2001; Holland, 2006;Pufahl and Hiatt, 2012) and suggest earlier oxygenation pulses(Anbar et al., 2007; Wille et al., 2007; Scott et al., 2008). Thespike at 2970 Ma, which is from pyrite in black shales from theWitwatersrand Basin, may define the first pulse of global oxygena-tion, and supports other recent evidence for an oxygenation pulsedetermined by Cr isotopes and redox sensitive TE in the PongolaSupergroup (Crowe et al., 2013). From 1700 to 700 Ma, Se showsa gradual decline (Fig. 8b). This trend suggests the possibility of agradual drop in atmosphere oxygen through the Meso- and Neo-proterozoic but with minor perturbations (particularly at 1700 to1300 Ma). Our U and Cr curves measured on the black shale matrix(Fig. 8c and d) show a similar decreasing trend during this period,consistent with published whole-rock data (Partin et al., 2013).Reinhard et al. (2013) have calibrated the palaeoredox proxies Crand Mo in the Proterozoic to delineate anoxic settings, where Cr isreduced and buried, compared to euxinic settings, where both Crand Mo are reduced/sulfidised and buried. On this basis, our data(Fig. 8a and d) may be interpreted to suggest that anoxic settingswere dominant in the Proterozoic from 2150 to 700 Ma, whereaseuxinic settings (anoxic + H2S) were dominant from 1200 to 800Ma. This interpretation fits with the concept of decreasing atmo-sphere oxygenation through the Meso- and Neoproterozoic, withminimum O2 levels around 800 Ma, corresponding to the Cryoge-nian TE trough.

Fig. 8. Temporal trends of a, Mo in sedimentary pyrite, b, Se in sedimentary pyrite,c, U in shale matrix, and d, Cr in shale matrix; measured by LA-ICPMS displayed asa 19 point rolling mean.

3.2. TE deficiency in the Meso- to Neoproterozoic Oceans

Anbar and Knoll (2002) present the hypothesis that under theconditions of a deep sulfidic ocean for most of the Meso- and Neo-

R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220 217

Fig. 9. Schematic of the major findings of this paper related to redox sensitive TE variations and oxygenation of the ocean–atmosphere. The red line is the estimatedoxygenation curve based on the Se content of sedimentary pyrite. The baseline % PAL (Present Atmospheric Level) is from Sahoo et al. (2012).

proterozoic (Canfield, 1998) nutrient TE would be scarce, thus lim-iting the ecological distribution of eukaryotic algae and adverselyeffecting marine evolution. Our data on the biologically importantnutrient TE (Mo, Zn, Co, Cd, Se, Ni, Ag, Mn, U, V and Cr; Sup-plementary Figs. 3–5) supports this hypothesis and shows thatalthough these TE exhibit various trends in detail, they all showa relative depletion from 1200 to 700 Ma compared to their earlyPhanerozoic values, and in most cases their Neoarchean to Palaeo-proterozoic values. For the redox sensitive TE that are employedas atmosphere–ocean oxygenation proxies, Mo, U, Cr and Se thereare two possible causes for these trends; (1) low values of theseTE represent gradual drawdown of these elements in a predom-inantly sulfidic deep ocean (Anbar and Knoll, 2002; Reinhard etal., 2013), with limited renewal by continental erosion due to lowlevels of tectonic activity over the period 1800 to 700 Ma. Atmo-spheric oxygen remains constant in this scenario (Canfield, 2005;Holland, 2006) Alternatively, (2) the decline in these redox sensi-tive TE signal a gradual decrease in atmosphere oxygenation from1800 to 800 Ma, with the lowest O2 levels forming the Cryoge-nian TE trough between 800 and 660 Ma, prior to the rise atGOE2. Other workers (Shen et al., 2003; Planavsky et al., 2011;Poulton and Canfield, 2011; Partin et al., 2013) provide evidencefor sustained low levels of O2 during the mid Proterozoic, althoughnone have suggested a significant O2 decline during this period.

In considering these two scenarios, there is little evidence tosupport a static tectonic regime, as this period of the middle to lateProterozoic saw the amalgamation (1300–1000 Ma) and breakup(900–700 Ma) of the super continent Rodinia (Li et al., 2008;Condie et al., 2011). Active erosion during these events wouldinvolve a continuous supply of nutrient TE to the oceans. How-ever the normalised 87Sr/86Sr curve of riverine runoff comparedwith mantle influence (Shields, 2007), indicates a decline in ero-sive supply of Sr which parallels the Se and U time curves. Onthe other hand, low O2 may have caused the nutrient deficiency,which is corroborated by carbon isotope evidence for low pro-ductivity (Anbar and Knoll, 2002). Canfield (2005) suggests thatatmospheric oxygen rose to a level of 5 to 10% Present Atmo-spheric Level (PAL) around 1800 Ma and remained constant untilabout 700 Ma. Planavsky et al. (2011) suggest a lower level of 1 to10% PAL throughout the Meso- and Neoproterozoic. Based on our

data, and these previous estimates, we speculate that atmosphericO2 may have decreased from a level around 10–20% PAL at 1800Ma to the Cryogenian TE trough at 1–5% PAL between 800 and 660Ma (Fig. 9).

4. Strengths, weaknesses and caveats of sedimentary pyriteproxies for ocean chemistry

The TE variation in sedimentary pyrite, determined by LA-ICPMS, has the potential to enable the development of sev-eral ocean–atmosphere proxies through time including: ocean TEchemistry, ocean pH, ocean sulfate content, ocean–atmosphereoxygenation, ocean anoxic events, ocean productivity and oceanhydrothermal activity. Although there are strengths associated withthis approach, there are also several weaknesses and caveats.

4.1. Strengths

Sedimentary pyrite concentrates many of the redox sensitiveand nutrient TE, in particular, Se, Mo, Cu, Ni, Ag, As, Cd, Co, Mn, Sbwell above their content in the sedimentary matrix of black shales(Fig. 4, Supplementary Information). This means, that by using theLA-ICPMS technique, the minimum (down to ppb levels) and max-imum (over wt% levels) of TE in pyrite can be measured accurately,enabling proxies for both maximum and minimum levels of TE inthe oceans to be developed. This has not been previously possible.The LA-ICPMS technique used here enables 22 TE to be analysedsimultaneously. This enables the trends of a suite of redox sensi-tive elements (e.g., Mo, Se, Cu, U and Cr) to be compared on thesame sample set. Furthermore, TE in both the pyrite and sedimen-tary matrix are quantified, and thus TE concentrated in the matrix,(e.g., U, V, W, Sn and Ba) can also be tracked through time.

Pyrite analyses involve only one mineral, and thus there are lesscomplications than whole rock analyses of black shales, where TEare partitioned amongst a number of minerals and organic matter,of varying proportions, plus seawater and pore fluids. Sedimentarypyrite forms in black shales from sub-oxic to euxinic environments,and thus sampling is not restricted to a particular redox facies inorder to apply the technique. This enables a much larger popula-tion of black shales to be included in deep time research. A de-tailed petrographic study of sedimentary pyrite allows the effects

218 R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220

of late diagenesis, metamorphism and hydrothermal overprints tobe fully assessed and thus avoided in sample selection.

4.2. Weaknesses

Sedimentary pyrite develops both in the water column (syn-genetic) and in the sea floor muds (diagenetic), and yet the tech-nique proposed here does not take into account the modificationof seawater TE chemistry by interaction with minerals, other thanpyrite, in the muds. Our analyses are a mixture of syngenetic andearly diagenetic framboidal pyrites plus nodular pyrites that de-veloped during late diagenesis, probably several meters below theseafloor. Diagenetic nodular pyrite that forms during late diagen-esis, commonly contains lower TE concentrations (Fig. 2; Large etal., 2007), but no more than an order of magnitude less than syn-genetic and early diagenetic pyrite in the same sedimentary unit.We argue that seawater is the principal component of pore flu-ids, and as framboidal pyrite develops in the top few centimetresof the muds, when seawater is >60% of the muds, then a strongchemical connection between pyrite and seawater is maintained.These same assumptions apply in the use of strontium, carbon andoxygen isotopes of sedimentary carbonate as a proxy for their re-spective isotope variations in the palaeo-ocean (Veizer et al., 1999;Shields and Veizer, 2002).

The TE in early formed (syngenetic and early diagenetic) pyritecan be strongly modified by later diagenetic, metamorphic or hy-drothermal processes and thus have little relationship to the orig-inal seawater chemistry. This is a major issue, especially for thoseelements loosely held in the pyrite structure, such as Mo (Chappazet al., 2014). However, the approach proposed here is to under-take petrographic studies of pyrite textures in order to screen outsamples with evidence of metamorphic and hydrothermal influ-ences (e.g., Thomas et al., 2011). Selenium in pyrite is very robustto metamorphic influences and is the best of the redox sensitive TE(Fig. 9). Black shales and associated pyrite are of little use as sea-water proxies for samples above middle greenschist facies due tomajor TE exchange with metamorphic fluids (Pitcairn et al., 2006;Large et al., 2011b).

Our data indicate a wide variation in the TE content of sedi-mentary pyrite from the same sample. However, many TE show avariation of less than one to two orders of magnitude. This degreeof variation could be considered to relate to ocean TE variation,or to the inherent variability of syngenetic and early diageneticpyrite, and the effects of diagenesis and low-grade metamorphism.The variation of one order of magnitude in a single sample issignificantly less that the variation of up to 4 orders of magni-tude in the full dataset through time. Studies on single drill core(Supplementary Fig. 1) show the variation in individual samplescompared to the mean variation through a stratigraphic section,and clearly demonstrate that robust trends in the TE are related tostratigraphic position.

In addition to the concentrations of TE in seawater, a numberof other factors may have an effect on the TE content of diage-netic pyrite, including; the matrix composition of the sediment,the amount of pyrite produced, pH of pore fluids, degree of biopro-ductivity and preferred electron acceptors. Although these factorsare assumed here to have a second order effect, they each needfurther evaluation.

4.3. Caveats

The most coherent TE data on early-formed sedimentary pyriteis obtained by the analysis of pyrite framboids, and these must bethe first priority in any analytical strategy. However many blackshales of greenschist facies do not contain framboidal pyrite, but

may contain diagenetic nodules or partially recrystallised fram-boids. Great care is required to record the textural and parageneticchanges in pyrite in order to undertake analyses on the early-formed pyrite or their partially recrystallised equivalents.

In order to place pyrite TE analyses in a temporal context, highquality geochronology is required. For Phanerozoic samples, fossilassemblages can yield detailed chronological resolution. In Pre-cambrian settings, the U–Pb analysis of zircon can provide highquality constraints on the timing of eruption of tuffs and/or max-imum depositional ages for silt- and sand-dominated strata thatmay be interbedded with black shale sequences. U-bearing au-thigenic mineral phases also provide at least a minimum age ofdeposition of the host rocks, and in some cases may date diagen-esis. Where available, we have preferred chronological constraintsbased on these techniques in order to minimise uncertainties re-lated to whole rock dating methods, however the age resolutionand uncertainties vary significantly (Supplementary Table 2). A ma-jor caveat of any temporal analysis such as this is the accuracy andresolution of chronological constraints.

5. Conclusions

The use of sedimentary pyrite as a proxy for TE trends inthe palaeo-ocean has opened a new window into deep time, andpresents exciting possibilities. Several lines of evidence indicatethat for any particular black shale depositional environment, syn-genetic pyrite that forms in the water column and diagenetic pyritethat forms in the top few meters of the seawater-saturated muds,will have a similar first order TE composition that relates to theTE composition of the seawater at that time. Processes during di-agenesis do not substantially change the mean first order TE con-tent of sedimentary pyrite. However, metamorphic processes thatcause recrystallisation of pyrite and conversion to pyrrhotite atmoderate to high metamorphic grades, may be accompanied bymajor changes in many TE concentrations. Therefore, this studyhas concentrated on early-formed (syngenetic to diagenetic) pyriteand avoided samples with obvious metamorphic or hydrothermalpyrite and/or pyrrhotite.

Our data suggest that for most TE, temporal variations in con-centration have been over 2 to 4 orders of magnitude. The concen-trations of some TE, like Ni, Co and Cr, generally decrease throughtime, but with significant variation, commonly related to changesin the composition of the exposed crust. Other redox sensitive TElike Mo and Se, generally increase through time, and are princi-pally controlled by oxygenation trends in the ocean–atmospheresystem.

A schematic diagram of our major findings is given in Fig. 9.This figure uses the Se content of sedimentary pyrite as a proxy forvariations in atmosphere oxygenation. This proxy is based on thepremise that oxidative weathering of pyrite in continental rocksleads to the release of selenium as both the selenate and selen-ite species. Under neutral to alkaline oxygenated conditions theselenate species remains highly soluble, where it can be read-ily transported via river systems to the ocean. However underlow oxygen conditions, continental weathering releases little sol-uble Se, because the reduced forms of Se (selenide, elemental Se0

and organo-selenium complexes) are relatively insoluble. Thus sig-nificant increases in atmospheric oxygen, accompanied by activeerosion, can lead to a major increase in the supply of soluble Seto the ocean, to be subsequently trapped in sedimentary pyrite inseafloor muds.

The Se curve derived from sedimentary pyrite, along withcurves for other redox sensitive TE (e.g., Mo, U, Tl, Mn, Cd, Ag),show significant increases in TE abundance of one to two ordersof magnitude at GOE1 and GOE2 (Fig. 9). However our data forsome TE shows sharp rises prior to GOE1, suggesting pulses of

R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220 219

oxygenation commenced as early as 2970 Ma. Through the Protero-zoic, TE variations in sedimentary pyrite are subdued, comparedwith the Archaean, suggesting steady state or declining oxygen lev-els reaching a minimum in the Cryogenian prior to GOE2. From1800 to 700 Ma a number of biologically important TE (e.g., Se, Cr,Ag, Mo, Cu, Ni, Mn) show a flat to decreasing trend suggesting thatnutrient TE deficiency may be the reason for limited eukaryoticdevelopment and radiation, thus adversely effecting marine evolu-tion during the middle to late Proterozoic. This period ended withthe Sturtian glaciation, and from 700 to 540 Ma many of the re-dox sensitive TE rose by 1 to 2 orders of magnitude in responseto GOE2, accompanied by diversification of phytoplankton and theappearance of eukaryotic clades (Shields-Zhou and Och, 2011); theprelude to the Cambrian Explosion.

Acknowledgements

The following colleagues have provided black shale samples to-ward this project: Robert Scott, Ray Coveney Jr., Jeffrey Abbott,Bradley Guy, David Huston, Lex Lambeck, Jeffrey Steadman, CharlesMakoundi, Sean Johnson, Richard Batchelor, Stuart Smith, MikhailKrupenin, Jerry Sharrock, Peter Sorjonen-Ward, Luba Leonova, Va-leriy Murzin and Sergey Karpov. Funding was provided by an ARCCentre of Excellence grant CE0561595 and AMIRA Internationalproject grant P923 to RRL. Ron Fuge (University of Wales), ThomasKulp (Binghamton University), Bernd Lottermoser, Patrick Quilty,Jocelyn McPhie (University of Tasmania), Brian Skinner (Yale Uni-versity) and Noel White (Consultant) are thanked for their com-ments on earlier versions of the manuscript. We thank B. Kamberand three other reviewers, and Editor G. Henderson, for excellentcomments for revision and improvement of the final manuscript.This is YGS publication number 019.

Appendix A. Supplementary material

Supplementary material related to this article can be found on-line at http://dx.doi.org/10.1016/j.epsl.2013.12.020.

References

Algeo, T.J., Rowe, H., 2012. Paleoceanographic applications of trace-metal concentra-tion data. Chem. Geol. 324, 6–18.

Anbar, A.D., Knoll, A.H., 2002. Proterozoic ocean chemistry and evolution: A bioinor-ganic bridge?. Science 297, 1137–1142.

Anbar, A.D., Duan, Y., Lyons, T.W., Arnold, G.L., Kendall, B., Creaser, R.A., Kaufman,A.J., Gordon, G.W., Scott, C., Garvin, J., Buick, R., 2007. A whiff of oxygen beforethe great oxidation event?. Science 317, 1903–1906.

Bajwah, Z.U., Seccombe, P.K., O!er, R., 1987. Trace element distribution, Co:Ni ratiosand genesis of the Big Cadia iron-copper deposit, New South Wales, Australia.Miner. Depos. 22, 292–300.

Baldwin, B., Butler, C.O., 1985. Compaction curves. Am. Assoc. Pet. Geol. Bull. 69,622–626.

Berner, Z.A., Puchelt, H., Nöltner, T., Kramar, U., 2013. Pyrite geochemistry inthe Toarcian Posidonia Shale of south-west Germany: Evidence for contrastingtrace-element patterns of diagenetic and syngenetic pyrites. Sedimentology 60,548–573.

Canfield, D.E., 1998. A new model for Proterozoic ocean chemistry. Nature 396,450–453.

Canfield, D.E., 2005. The early history of atmospheric oxygen: Homage to Robert M.Garrels. Annu. Rev. Earth Planet. Sci. 33, 1–36.

Canfield, D.E., Poulton, S.W., Narbonne, G.M., 2007. Late-Neoproterozoic deep-oceanoxygenation and the rise of animal life. Science 315, 92–95.

Chappaz, A., Lyons, T.W., Gregory, D.D., Reinhard, C.T., Gill, B.C., Li, C., Large, R.R.,2014. Does pyrite act as an important host for molybdenum in modern andancient euxinic sediments?. Geochim. Cosmochim. Acta 126, 112–122.

Ciobanu, C.L., Cook, N.J., Pring, A., Brugger, J., Danyushevsky, L.V., Shimizu, M.,2009. ‘Invisible gold’ in bismuth chalcogenides. Geochim. Cosmochim. Acta 73,1970–1999.

Condie, K.C., Bickford, M.E., Aster, R.C., Belousova, E., Scholl, D.W., 2011. Episodiczircon ages, Hf isotopic composition, and the preservation rate of continentalcrust. Geol. Soc. Am. Bull. 123, 951–957.

Crowe, S.A., Døssing, L.N., Beukes, N.J., Bau, M., Kruger, S.J., Frei, R., Canfield, D.E.,2013. Atmospheric oxygenation three billion years ago. Nature 501, 535–538.

Dahl, T.W., Canfield, D.E., Rosing, M.T., Frei, R.E., Gordon, G.W., Knoll, A.H., An-bar, A.D., 2011. Molybdenum evidence for expansive sulfidic water masses in#750 Ma oceans. Earth Planet. Sci. Lett. 311, 264–274.

Danyushevsky, L., Robinson, P., Gilbert, S., Norman, M., Large, R., McGoldrick, P., Shel-ley, M., 2011. Routine quantitative multi-element analysis of sulphide mineralsby laser ablation ICP-MS: Standard development and consideration of matrix ef-fects. Geochem., Explor. Environ. Anal. 11, 51–60.

Deditius, A.P., Utsunomiya, S., Reich, M., Kesler, S.E., Ewing, R.C., Hough, R., Walshe,J., 2011. Trace metal nanoparticles in pyrite. Ore Geol. Rev. 42, 32–46.

Diener, A., Neumann, T., 2011. Synthesis and incorporation of selenide in pyrite andmackinawite. Radiochim. Acta 99, 791–798.

Farquhar, J., Savarino, J., Airieau, S., Thiemens, M.H., 2001. Observation ofwavelength-sensitive mass-independent sulfur isotope effects during SO2 pho-tolysis: Implications for the early atmosphere. J. Geophys. Res., Planets 106,32829–32839.

Fike, D.A., Grotzinger, J.P., Pratt, L.M., Summons, R.E., 2006. Oxidation of the Edi-acaran ocean. Nature 444, 744–747.

Gregory, D., Meffre, S., Large, R.R., 2013. Mineralogy of metal contaminated estuarinesediments, Derwent estuary, Hobart, Australia: implications for metal mobility.Aust. J. Earth Sci. 60, 589–603.

Gregory, D.D., Meffre, S., Large, R.R., in press. Comparison of metal enrichment inpyrite framboids from a metal-enriched and metal poor estuary. Am. Mineral.http://dx.doi.org/10.2138/am.2014.4545.

Guy, B.M., Beukes, N.J., Gutzmer, J., 2010. Paleoenvironmental controls on the textureand chemical composition of pyrite from non-conglomeratic sedimentary rocksof the Mesoarchean Witwatersrand Supergroup, South Africa. S. Afr. J. Geol. 113,195–228.

Harrold, T.W.D., Swarbrick, R.E., Goulty, N.R., 1999. Pore pressure estimation fromMudrock porosities in tertiary basins, southeast Asia. Am. Assoc. Pet. Geol.Bull. 83, 1057–1066.

Helz, G.R., Miller, C.V., Charnock, J.M., Mosselmans, J.F.W., Pattrick, R.A.D., Garner,C.D., Vaughan, D.J., 1996. Mechanism of molybdenum removal from the seaand its concentration in black shales: EXAFS evidence. Geochim. Cosmochim.Acta 60, 3631–3642.

Hoffman, P.F., Kaufman, A.J., Halverson, G.P., Schrag, D.P., 1998. A Neoproterozoicsnowball earth. Science 281, 1342–1346.

Holland, H.D., 1984. The Chemical Evolution of the Atmosphere and Oceans. Prince-ton University Press, Princeton.

Holland, H.D., 2006. The oxygenation of the atmosphere and oceans. Philos. Trans.R. Soc. Lond. B, Biol. Sci. 361, 903–915.

Huerta-Diaz, M.A., Morse, J.W., 1990. A quantitative method for determination oftrace metal concentrations in sedimentary pyrite. Mar. Chem. 29, 119–144.

Huerta-Diaz, M.A., Morse, J.W., 1992. Pyritization of trace metals in anoxic marinesediments. Geochim. Cosmochim. Acta 56, 2681–2702.

Konhauser, K.O., Pecoits, E., Lalonde, S.V., Papineau, D., Nisbet, E.G., Barley, M.E.,Arndt, N.T., Zahnle, K., Kamber, B.S., 2009. Oceanic nickel depletion and amethanogen famine before the Great Oxidation Event. Nature 458, 750–753.

Kulp, T.R., Pratt, L.M., 2004. Speciation and weathering of selenium in Upper Cre-taceous chalk and shale from South Dakota and Wyoming, USA. Geochim. Cos-mochim. Acta 68, 3687–3701.

Large, R.R., Maslennikov, V.V., Robert, F., Danyushevsky, L.V., Chang, Z., 2007. Mul-tistage sedimentary and metamorphic origin of pyrite and gold in the GiantSukhoi log deposit, Lena Gold Province, Russia. Econ. Geol. 102, 1233–1267.

Large, R.R., Danyushevsky, L., Hollit, C., Maslennikov, V., Meffre, S., Gilbert, S., Bull, S.,Scott, R., Emsbo, P., Thomas, H., Singh, B., Foster, J., 2009. Gold and trace elementzonation in pyrite using a laser imaging technique: Implications for the timingof gold in orogenic and carlin-style sediment-hosted deposits. Econ. Geol. 104,635–668.

Large, R.R., Bull, S.W., Gilbert, S., 2011a. Multiple syngenetic metalliferous horizonsenriched in V–As–Mo–Se–Ni–Ag–Zn–Au in the Northern Carlin Trend: impli-cations for the genesis of world class gold-arsenic deposits. In: Steininger, R.,Pennell, B. (Eds.), Geological Society of Nevada 2010 Symposium: Great BasinEvolution and Metallogeny. DEStech Publications Inc., Reno, pp. 1–16.

Large, R.R., Bull, S.W., Maslennikov, V.V., 2011b. A carbonaceous sedimentarysource-rock model for carlin-type and orogenic gold deposits. Econ. Geol. 106,331–358.

Large, R.R., Meffre, S., Burnett, R., Guy, B., Bull, S., Gilbert, S., Goemann, K., Danyu-shevsky, L., 2013. Evidence for an intrabasinal source and multiple concentrationprocesses in the formation of the carbon leader reef, Witwatersrand Supergroup,South Africa. Econ. Geol. 108, 1215–1241.

Li, Z.X., Bogdanova, S.V., Collins, A.S., Davidson, A., De Waele, B., Ernst, R.E., Fitzsi-mons, I.C.W., Fuck, R.A., Gladkochub, D.P., Jacobs, J., Karlstrom, K.E., Lu, S., Nat-apov, L.M., Pease, V., Pisarevsky, S.A., Thrane, K., Vernikovsky, V., 2008. Assem-bly, configuration, and break-up history of Rodinia: A synthesis. PrecambrianRes. 160, 179–210.

Lyons, T.W., Werne, J.P., Hollander, D.J., Murray, R.W., 2003. Contrasting sulfur geo-chemistry and Fe/Al and Mo/Al ratios across the last oxic-to-anoxic transition inthe Cariaco Basin, Venezuela. Chem. Geol. 195, 131–157.

220 R.R. Large et al. / Earth and Planetary Science Letters 389 (2014) 209–220

Meffre, S., Large, R.R., Scott, R., Woodhead, J., Chang, Z., Gilbert, S.E., Danyushevsky,L.V., Maslennikov, V., Hergt, J.M., 2008. Age and pyrite Pb-isotopic compositionof the giant Sukhoi Log sediment-hosted gold deposit, Russia. Geochim. Cos-mochim. Acta 72, 2377–2391.

Mitchell, K., Mason, P.R.D., Van Cappellen, P., Johnson, T.M., Gill, B.C., Owens,J.D., Diaz, J., Ingall, E.D., Reichart, G.J., Lyons, T.W., 2012. Selenium as paleo-oceanographic proxy: A first assessment. Geochim. Cosmochim. Acta 89,302–317.

Morse, J.W., Arakaki, T., 1993. Adsorption and coprecipitation of divalent metals withmackinawite (FeS). Geochim. Cosmochim. Acta 57, 3635–3640.

Partin, C.A., Bekker, A., Planavsky, N.J., Scott, C.T., Gill, B.C., Li, C., Podkovyrov, V.,Maslov, A., Konhauser, K.O., Lalonde, S.V., Love, G.D., Poulton, S.W., Lyons, T.W.,2013. Large-scale fluctuations in Precambrian atmospheric and oceanic oxygenlevels from the record of U in shales. Earth Planet. Sci. Lett. 369–370, 284–293.

Piper, D.Z., Perkins, R.B., 2004. A modern vs. Permian black shale – the hydrog-raphy, primary productivity, and water-column chemistry of deposition. Chem.Geol. 206, 177–197.

Pitcairn, I.K., Teagle, D.A.H., Craw, D., Olivo, G.R., Kerrich, R., Brewer, T.S., 2006.Sources of metals and fluids in orogenic gold deposits: Insights from the Otagoand Alpine schists, New Zealand. Econ. Geol. 101, 1525–1546.

Planavsky, N.J., McGoldrick, P., Scott, C.T., Li, C., Reinhard, C.T., Kelly, A.E., Chu, X.,Bekker, A., Love, G.D., Lyons, T.W., 2011. Widespread iron-rich conditions in themid-Proterozoic ocean. Nature 477, 448–451.

Poulton, S.W., Canfield, D.E., 2011. Ferruginous conditions: a dominant feature of theocean through Earth’s history. Elements 7, 107–112.

Prokoph, A., Ernst, R.E., Buchan, K.L., 2004. Time-series analysis of large igneousprovinces: 3500 Ma to present. J. Geol. 112, 1–22.

Pufahl, P.K., Hiatt, E.E., 2012. Oxygenation of the Earth’s atmosphere–ocean system:A review of physical and chemical sedimentologic responses. Mar. Pet. Geol. 32,1–20.

Reimann, C., Caritat, P.d., 1998. Chemical Elements in the Environment: Factsheetsfor the Geochemist and Environmental Scientist. Springer-Verlag.

Reinhard, C.T., Planavsky, N.J., Robbins, L.J., Partin, C.A., Gill, B.C., Lalonde, S.V.,Bekker, A., Konhauser, K.O., Lyons, T.W., 2013. Proterozoic ocean redox and bio-geochemical stasis. Proc. Natl. Acad. Sci. USA 110, 5357–5362.

Rickard, D., 2012. Sulfidic sediments and sedimentary rocks. In: Van Loon, A.J. (Ed.),Developments in Sedimentology. Elsevier, p. 801.

Rickard, D., Luther, G.W.I.I., 1997. Kinetics of pyrite formation by the H2S oxida-tion of iron (II) monosulfide in aqueous solutions between 25 and 125 $C: Themechanism. Geochim. Cosmochim. Acta 61, 135–147.

Sahoo, S.K., Planavsky, N.J., Kendall, B., Wang, X., Shi, X., Scott, C., Anbar, A.D., Lyons,

T.W., Jiang, G., 2012. Ocean oxygenation in the wake of the Marinoan glaciation.Nature 489, 546–549.

Scott, C., Lyons, T.W., Bekker, A., Shen, Y., Poulton, S.W., Chu, X., Anbar, A.D.,2008. Tracing the stepwise oxygenation of the Proterozoic ocean. Nature 452,456–459.

Shen, Y., Knoll, A.H., Walter, M.R., 2003. Evidence for low sulphate and anoxia in amid-Proterozoic marine basin. Nature 423, 632–635.

Shields, G.A., 2007. A normalised seawater strontium isotope curve: possible impli-cations for Neoproterozoic-Cambrian weathering rates and the further oxygena-tion of the Earth. eEarth 2, 35–42.

Shields, G., Veizer, J., 2002. Precambrian marine carbonate isotope database: Version1.1. Geochem. Geophys. Geosyst. 3, 1–12.

Shields-Zhou, G., Och, L., 2011. The case for a neoproterozoic oxygenation event:Geochemical evidence and biological consequences. GSA Today 21, 4–11.

Thomas, B.M., Willink, R.J., Grice, K., Twitchett, R.J., Purcell, R.R., Archbold, N.W.,George, A.D., Tye, S., Alexander, R., Foster, C.B., Barber, C.J., 2004. Unique marinePermian-Triassic boundary section from Western Australia. Aust. J. Earth Sci. 51,423–430.

Thomas, H.V., Large, R.R., Bull, S.W., Maslennikov, V., Berry, R.F., Fraser, R., Froud,S., Moye, R., 2011. Pyrite and pyrrhotite textures and composition in sediments,laminated quartz veins, and reefs at Bendigo gold mine, Australia: Insights forore genesis. Econ. Geol. 106, 1–31.

Tribovillard, N., Algeo, T.J., Lyons, T., Riboulleau, A., 2006. Trace metals as paleoredoxand paleoproductivity proxies: An update. Chem. Geol. 232, 12–32.

Veizer, J., Ala, D., Azmy, K., Bruckschen, P., Buhl, D., Bruhn, F., Garden, G.A.F., Di-ener, A., Ebneth, S., Godderis, Y., Jasper, T., Korte, C., Pawellek, F., Podlaha, O.G.,Strauss, H., 1999. 87Sr/86Sr, !13C and !18O evolution of Phanerozoic seawater.Chem. Geol. 161, 59–88.

Wilkin, R.T., Barnes, H.L., Brantley, S.L., 1996. The size distribution of framboidalpyrite in modern sediments: An indicator of redox conditions. Geochim. Cos-mochim. Acta 60, 3897–3912.

Wille, M., Kramers, J.D., Nägler, T.F., Beukes, N.J., Schröder, S., Meisel, T., Lacassie,J.P., Voegelin, A.R., 2007. Evidence for a gradual rise of oxygen between 2.6 and2.5 Ga from Mo isotopes and Re-PGE signatures in shales. Geochim. Cosmochim.Acta 71, 2417–2435.

Wilson, M.R., Akerblom, G.V., 1982. Geological setting in geochemistry of uranium-rich granites in the Proterozoic of Sweden. Mineral. Mag. 46, 235–247.

Wyborn, L.A.I., Wyborn, D., Warren, R.G., Drummond, B.J., 1992. Proterozoic gran-ite types in Australia: implications for lower crust composition, structure andevolution. Trans. R. Soc. Edinb. Earth Sci. 83, 201–209.