controls of eustasy and diagenesis on the 238u/235u of … · 2018-10-09 · controls of eustasy...

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 242 (2018) 233–265

Controls of eustasy and diagenesis on the 238U/235Uof carbonates and evolution of the seawater (234U/238U)

during the last 1.4 Myr

Francois L.H. Tissot a,b,⇑, Cindy Chen a, Benjamin M. Go a, Magdalena Naziemiec a

Garrett Healy a, Andrey Bekker c, Peter K. Swart d, Nicolas Dauphas a

aOrigins Laboratory, Department of the Geophysical Sciences and Enrico Fermi Institute, The University of Chicago, 5734 South Ellis

Avenue, Chicago, IL, USAbDepartment of the Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, 77 Massachusetts Avenue,

Cambridge, MA, USAcDepartment of Earth Sciences, University of California, Riverside, 900 University Avenue, Riverside, CA, USA

dDepartment of Marine Geosciences, RSMAS/MGG, University of Miami, 4600 Rickenbacker Causeway, Miami, FL, USA

Received 19 January 2018; accepted in revised form 10 August 2018; Available online 23 August 2018

Abstract

Using a leaching protocol designed for the study of U isotopes in recent carbonates, we measured the U isotope compo-sition, both 238U/235U and 234U/238U, of modern and ancient corals (n = 6), a limestone and a dolostone, as well as 43shallow-water carbonate sediments from the ODP Leg 166 Site 1009 drill core, on the slope of the Bahamas platform.Although bulk corals record the seawater d238U value within ±0.02‰, differences of up to 0.30‰ in the d238U of individualleachates suggest a control of the coral structure and a more positive 238U/235U ratio in the centers of calcification.

The drill core d238U data shows that the 238U/235U ratio of shallow-water carbonates is controlled mainly by (1) variationsin sea-level through the mixing of different amounts of platform-derived sediments (with d238U �0.50–0.60‰ heavier thanseawater) and pelagic sediments (with seawater-like d238U values), (2) authigenic U enrichment via pore-water circulationand U reduction both on the platform and down to �5 m below the surface (mbsf) after deposition of the sediment, and,to a lesser extent, by (3) early diagenetic processes (i.e., carbonate dissolution and/or recrystallization) during sediment burial.The global effect of these processes leaves the d238U values of shallow-water carbonates offset relative to that of seawater byDCarbonates-SW = +0.24 ± 0.06‰ (95% CI, including all samples). This shift can be used in seawater paleoredox reconstructionsbased on carbonates deposited on shallow-water platform, shelf and slope environments (i.e., most of the carbonate sedimen-tary record prior to the Mesozoic) to account for the average effect of carbonate diagenesis. Assuming that the 238U/235U ratioof carbonate platform sediments directly records the seawater 238U/235U ratio would underestimate the extent of ocean-seafloor anoxia by at least a factor 10. The rapid fluctuations in d238U values due to sea-level changes (i) is a factor that shouldbe considered before interpreting d238U variations as reflecting changes in oceanic paleoredox conditions and (ii) reinforcesthe need for statistically meaningful data sets.

The d(234U) data suggest that the (234U/238U) ratio of the seawater has remained within �20‰ of the modern seawatervalue during the last 1–1.4 Myr. Furthermore, we find that small-scale (1–15‰) variations in seawater d(234U) mirrorsea-level changes during the penultimate glacial-interglacial period (�140 to �200 ka), thus confirming the record of lowerd(234U)SW during periods of low sea-level stand and expanding it to at least the last two glacial-interglacial events (i.e.,

https://doi.org/10.1016/j.gca.2018.08.022

0016-7037/� 2018 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: The Isotoparium, Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena,CA, USA.

E-mail address: [email protected] (F.L.H. Tissot).


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2018.08.022&domain=pdf

234 F.L.H. Tissot et al. /Geochimica et Cosmochimica Acta 242 (2018) 233–265

�0.23 Ma). Such fluctuations in d(234U)initial values should be taken into account when screening carbonate sediments U-Thages on the basis of the initial (234U/238U) ratios of the samples.� 2018 Elsevier Ltd. All rights reserved.

Keywords: Shallow-water carbonate; Bahamas; ODP drill core; U isotopes; Sea-level variations; Diagenesis

1. INTRODUCTION AND BACKGROUND

Reconstructions of Earth’s ancient atmosphere-oceanredox conditions rely on proxies such as the survival ofredox-sensitive detrital minerals (Ramdohr, 1958;Rasmussen and Buick, 1999), variations in the elementalabundance of redox-sensitive elements (e.g., Mo, Re, U;Anbar et al., 2007; Partin et al., 2013a, 2013b; Scottet al., 2008), or variations in the isotopic composition oftraditional stable isotopes (e.g., S, C, N; Holland, 2006).Since the mid-2000s, the so-called ‘non-traditional’ stableisotope systems (e.g., Cr, Fe, Mo, U) have emerged as pow-erful tracers of both high- and low-temperature geochemi-cal processes (reviewed in Teng et al., 2017). Of particularinterest for paleoredox studies is the 238U/235U ratio, whichhas the potential to track the global extent of oceanicanoxia (e.g., Weyer et al., 2008; Tissot and Dauphas, 2015).

Uranium has three naturally occurring isotopes: primor-dial 238U and 235U (t1/2 = 4468 Myr and 704 Myr, respec-tively, Jaffey et al., 1971), and the shorter-lived 234U(t1/2 = 245,620 yr, Cheng et al., 2013), which is part of thedecay chain of 238U. In terrestrial surface environments, Uexists in twomain oxidation states: soluble U6+ that behavesconservatively in the modern ocean (i.e., U concentrationvaries linearly with salinity, Ku et al., 1977; Owens et al.,2011), and insoluble U4+. Because the mean oceanic resi-dence time of U (�400 kyr; Ku et al., 1977) is much longerthan the global ocean mixing time (1–2 kyr), the salinity-normalized seawater composition is homogeneous withregards to both U concentrations ([U]SW = 3.22 ± 0.06 ng/g, for a salinity of 35 g/L, Chen et al., 1986) and U isotopes(d238USW = �0.39 ± 0.02‰, Tissot and Dauphas, 2015;d(234U)SW = 144.9 ± 0.4‰, Chen et al., 1986; Chutcharavanet al., 2018) (see Eqs. (1) and (2), for d-notations). The sea-water U concentration and isotopic composition at anygiven time is thus the balance between U input to the ocean,mainly from rivers, and U removal, mostly into biogeniccarbonates, anoxic/euxinic sediments and suboxic/hypoxicsediments (i.e., oxygen-minimum zones in continental mar-gin settings with high primary productivity; e.g., Dunk et al.,2002; Tissot and Dauphas, 2015).

In the ocean, d(234U) and d238U values are controlled bydifferent processes, making uranium a two-facetted system.On the one hand, the evolution of the seawater 234U/238Uthrough time both holds clues into continental weatheringand affects U-Th ages. Indeed, alpha-recoil during 238Udecay and preferential leaching of 234U over lattice-bound238U lead to 234U excesses in rivers and marine sedimentpore-waters, which are eventually transferred to the oceans(e.g., Chabaux et al., 2003). This results in a modernd(234U)SW value of �145‰ (e.g., Ku et al., 1977; Chenet al., 1986; Andersen et al., 2010). As removal of U from

the homogenized ocean into sediments and duringhydrothermal alteration does not significantly fractionate234U and 238U, changes in d(234U)SW predominantly reflectschanges in the source 234U fluxes to the ocean (e.g.,Henderson, 2002). To reconstruct d(234U)SW through time,carbonates, and particularly corals, are predominantlyused. These samples can faithfully record the ambient sea-water 234U/238U ratio at formation time (e.g.,Chutcharavan et al., 2018), and thus provide a way of con-comitantly dating the carbonate using U-Th and accessingthe d(234U)SW at the time of formation (e.g., Edwardset al., 2003). As even minimal sample alteration can, how-ever, lead to large and mostly positive shifts in the234U/238U of carbonates (e.g., Bard et al., 1991; Hamelinet al., 1991; Gallup et al., 1994; Stirling et al., 1995), disen-tangling changes in the seawater 234U/238U from minoropen-system behavior is far from straightforward, and pastworks concluded to a constant d(234U)SW throughout thelate Quaternary (Bard et al., 1991; Hamelin et al., 1991;Gallup et al., 1994). A growing body of evidence subse-quently suggested a 1–15‰ lowering of d(234U)SW in timesof lower than modern sea-level stand during the last glacial-interglacial period. Establishing the magnitude and timingof these variations and understanding their origins is thefocus of intense research (e.g., Robinson et al., 2004; Esatand Yokoyama, 2006; Andersen et al., 2007; Esat andYokoyama, 2010; Chen et al., 2016a; Chutcharavan et al.,2018; Arendt et al., 2018).

On the other hand, the evolution of the seawater238U/235U through time provides a direct record of the glo-bal oceanic redox history. Indeed, and unlike the parent-daughter pair of isotopes (234U and 238U), the two long-lived isotopes of U (235U and 238U) are not significantlyfractionated during weathering and transport to the ocean(e.g., Wang et al., 2015; Tissot and Dauphas, 2015) andd238USW is mainly controlled by isotopic fractionation dur-ing U removal into sedimentary sinks. While the isotopicfractionation factors (DSink-SW) associated with U removalinto most sinks are relatively small (|DSink-SW| � 0.25‰),nuclear field shift effects (Bigeleisen, 1996; Schauble, 2007;Abe et al., 2008) impart larger isotopic fractionation(DAnoxic/euxinic-SW � +0.60‰) during U removal into anox-ic/euxinic sediments (e.g., Andersen et al., 2014), leaving theseawater isotopically lighter than the riverine discharge.This fractionation forms the basis of the 238U/235U paleore-dox proxy: during periods of extensive anoxia, U sequestra-tion into anoxic/euxinic sediments will drive the d238USW

towards lower (238U-depleted) values.Initially, 238U/235U paleoredox reconstructions focused

on black shales (Montoya-Pino et al., 2010; Asael et al.,2013; Kendall et al., 2013; Kendall et al., 2015) in partbecause of their high U content and common occurrence

F.L.H. Tissot et al. /Geochimica et Cosmochimica Acta 242 (2018) 233–265 235

in the geological record. The DAnoxic/euxinic-SW value is, how-ever, not only large, but also highly variable (from 0.0 to+0.8‰) and dependent on depositional settings (e.g., openvs. restricted basin, extent of U loss from pore water, depthof oxygen penetration into the sediment, Andersen et al.,2014), which are difficult to assess for ancient sediments.Furthermore, carbonate-hosted U (Andersen et al., 2014)and/or detrital U (Asael et al., 2013; Noordmann et al.,2015) in shales can blur the authigenic signal and have tobe accounted for. These complexities result in large uncer-tainties and make d238USW paleoredox reconstructionsfrom black shales (e.g., Lu et al., 2017; Yang et al., 2017;Phan et al., 2018) more qualitative than quantitative.Though Fe-Mn crusts might record the d238USW value reli-ably and with a small isotope fractionation (DMetalliferous-SW

= �0.24‰, Goto et al., 2014; Wang et al., 2016), the scar-city of the record and slow growth of Fe-Mn crusts limittheir usefulness in the study of Earth’s deep past.

Carbonates appear as a more promising target as (i)their rock record spans most of Earth’s history, and (ii)the d238U values of modern primary carbonate precipitatesand well-preserved aragonitic corals up to 600 ka in age aregenerally indistinguishable from that of modern seawater(Stirling et al., 2007; Weyer et al., 2008; Romanielloet al., 2013; Andersen et al., 2014; Tissot and Dauphas,2015). Several recent publications have already utilizedthe carbonate record to track redox transitions during theOceanic Anoxic Event 2 (�94 Ma, Clarkson et al., 2018),at the end-Triassic extinction (�201 Ma, Jost et al., 2017),the end-Permian extinction (�252 Ma, Brennecka et al.,2011; Lau et al., 2016; Elrick et al., 2017; Zhang et al.,2018b; Zhang et al., 2018c), the Frasnian-Famennianboundary (�372 Ma, Song et al., 2017), the Ordovician-Silurian boundary (�444 Ma, Bartlett et al., 2018), theCambrian-Ordovician boundary (�485 Ma, Azmy et al.,2015), the late Cambrian ‘‘SPICE” event (�499 Ma, Dahlet al., 2014), the Ediacaran-Cambrian transition andearly-Cambrian (�550–520 Ma, Dahl et al., 2017; Weiet al., 2018; Zhang et al., 2018a), and the end of the SturtianSnowball Earth (�640 Ma, Lau et al., 2017). At the sametime, a number of findings are calling into question the reli-ability of the U proxy in carbonates. In seawater (i.e.,10.3 mM Ca2+, 53 mM Mg2+, 14 nM U6+, Chester andJickells, 2012) and for pH > 6, the predominant U6+ speciesare not UO2-CO3 complexes as previously thought(Langmuir, 1978) but ternary calcium uranyl Ca-UO2-CO3 complexes (Dong and Brooks, 2006; Endrizzi andRao, 2014). Bond length and/or U coordination numberdifferences between these aqueous complexes (e.g., Docratet al., 1999; Szabo et al., 2000; Bernhard et al., 2001;Elzinga et al., 2004; Kerisit and Liu, 2010) or changes dur-ing incorporation into different carbonate mineral struc-tures (Reeder et al., 2000; Reeder et al., 2001; Kelly et al.,2003; Kelly et al., 2006) could drive U isotope fractiona-tion, respectively, prior to or during, incorporation of thesecomplexes into the solid phase. In fact, calcite and arago-nite precipitation experiments under slightly alkaline condi-tions (i.e., pH � 8.5) found resolvable U isotope shifts inaragonite (but not in calcite; Chen et al., 2016b), suggesting(i) a control of U aqueous speciation on isotope fractiona-

tion, and (ii) that �0.11–0.23‰ d238U variations in carbon-ates could be due to changes in seawater pH, pCO2, Ca

2+

and Mg2+ concentrations during sample precipitation,rather than global oceanic redox conditions (Chen et al.,2017).

More importantly, the effects of diagenesis on carbonated238U values are poorly understood. Using Bahamian plat-form carbonate sediments down to �40 cm depth,Romaniello et al. (2013) found significant authigenic Uenrichment and d238U values +0.20 to +0.40‰ heavier thanseawater. These samples, however, are too shallow forporewater-seawater exchange to have ceased and it remainsunclear whether d238U values will be further modified asburial proceeds, especially as old, variably altered carbon-ates display non-systematic d238U variations over a rangeof up to �1.0‰, both in individual components within asingle hand specimen (Hood et al., 2016) and in bulk sam-ples (Hood et al., 2018). The depositional settings studiedby Romaniello et al. (2013) are restricted to the uppermostpart of the shallow-marine carbonate platform (smalllagoon open to the ocean, shallow tidal flat, and tidal pond)and it is unknown whether deeper-water carbonates woulddisplay similar isotopic shifts, as the effect of lithification oncarbonate d238U values is unconstrained.

In this work we aim to investigate the effect of early mar-ine phreatic diagenesis and lithification on the 238U/235Uratio of shallow-water carbonate sediments, on the meterto 100 m scale (i.e., 30 kyr to 1 Myr timescale) to constrainhow faithfully carbonates can record the U isotopic compo-sition of seawater. Shallow-water carbonate sediments werechosen because they make up the majority of the sedimen-tary carbonate record prior to the Mesozoic (Opdyke andWilkinson, 1988; Boss and Wilkinson, 1991; Holmdenet al., 1998; Walker et al., 2002; Ridgwell, 2005), and arewidely used to reconstruct Earth’s chemical and climatichistory. We measured Th and U concentrations as well asTOC content, organic and inorganic d13C, inorganic d18O,organic d15N, and U isotopes, both d(234U) and d238U, incarbonate samples from a core drilled in the Bahamas car-bonate platform (ODP Leg 166 Site 1009). As recent workin nearby drill cores (Higgins et al., 2018) revealed that sig-nificant open-system behavior affects samples in the top 10sof meter below the surface (mbsf), sampling depths for thepresent study ranged from the water–sediment interface tothe bottom of the drill core (�225 m), well beyond the pointwhere open-system behavior are thought to cease. In orderto separate carbonate-associated U from U hosted in non-carbonate phases and at ion-exchange sites, we developed aleaching protocol potentially applicable to carbonates of allages. Our results shed light on the mechanisms affecting thed238U values in carbonates as well as on the variability ofthe d(234U)SW over the last 1.4 Myr.

2. SAMPLES

2.1. Background on ODP Leg 166 Site 1009 drill core

To study early diagenesis, the Holocene and Pleistoceneshallow-water calcium carbonates from the western slope ofthe Great Bahamas Bank, Ocean Drilling Project (ODP)


Leg 166 Site 1009 (Eberli et al., 1997; Malone, 2000;Malone et al., 2001) were selected. Site 1009 (23�36.840N,79�03.000W) is located �100 km south of the main ODPBahamas Transect (Sites 1003 to 1007; Fig. 1) and 4.5 kmoff from the platform edge, in 308 m of water. The geother-mal profile linearly increases from 18 to 22 �C down thecore and the age at the base of the recovered section (226mbsf) is between 1.2 to 1.44 Myr as determined using cal-careous nannofossils and planktonic foraminiferalbiostratigraphy.

The moderate preservation of calcareous nannofossils inthe drill core (Eberli et al., 1997) might, however, affect theaccuracy of the age vs. depth profile constructed from thenannofossils record (black squares on Fig. 2; data fromTable 2 in Eberli et al., 1997). In particular, the accurateidentification of Emiliania huxleyi (whose first appearancecorresponds to Marine Isotope Stage 8, 0.25 Myr ago) isknown to be hindered by dissolution and relying on thisspecies in poorly preserved sediments can yield considerableuncertainties (Thierstein et al., 1977). We therefore built acomposite age vs. depth profile (blue line in Fig. 2; datain Table S1) using radiometric ages when available andbio-event ages otherwise. The radiometric ages wereobtained using both 14C (Slowey et al., 2002) and U-Th(Henderson, 2002; Robinson et al., 2002) data, and aremuch more reliable on their respective domains of validity(up to �40 kyr and 500 kyr, respectively) than the bio-event ages. The U-Th ages show that the base of the E. hux-leyi nannofossil datum was indeed incorrectly positioned at�45 mbsf and is actually much lower in the drill core, at�70 mbsf. Based on the dispersion of U-Th ages at a givendepth, a conservative ±7.5% relative error was assigned tothe entire age vs. depth profile. These errors are propagatedonto the 234U/238Uinitial ratios reported (see supplementary

Fig. 1. Locations of ODP Leg 166 drill sites (1003–1009). Platformboreholes Unda and Clino from the Bahamas Drilling Program arealso shown.

materials for details). Using this composite age profile, thesedimentation rates at the �1–20 meter-scale are found tovary between 4 and 275 cm/kyr in this drill core (Fig. 2).

The samples consist of unlithified to partially lithifiedpeloidal and bioclastic mudstone, wackestone, and pack-stone, with grainstone and floatstone intercalations and anannofossil ooze layer (Eberli et al., 1997). The carbonatecontent of the samples is typically between 90 and95 wt%, of which 11 to 93 wt% is in the form of aragonite(Fig. 3). The total organic carbon content (TOC) is gener-ally low (below 1 wt%), but the data is sparse with only12 samples measured for the whole drill core (Eberliet al., 1997). Several hardgrounds, some coinciding withseismic boundaries and Marine Isotope Stages (MIS,Table S2), occur in the drill core intervals correspondingto glacial periods or glacial-interglacial transitions(Malone et al., 2001). These hardground are cemented byhigh-Mg calcite with �14 mol% MgCO3, and are character-ized by anomalously high d18Oinorg and low d13Cinorg values(Fig. 3), all of which is consistent with precipitation nearthe seafloor from waters having present-day to slightlycolder temperatures (Malone et al., 2001). The fact thateither hardgrounds/firmgrounds or turbidites/current flowsare found at the top of most upward-coarsening sequencesis also consistent with the coarsening of the sedimentrecording decrease in sea-level and hardgrounds formingduring times of minimal sedimentation rates and sedimentbypass.

Shipboard interstitial water profiles show little change inthe top 30–40 m, likely reflecting fluid-buffered conditions(as defined by Higgins et al., 2018) as a result of the lateralseawater fluid flow of �5–10 cm/yr at this site (Hendersonet al., 1999). Below that depth, however, shallow-burial dia-genetic reactions are clearly seen (Fig. 4; Eberli et al., 1997;Malone et al., 2001). In the 0–70 mbsf interval, decreasingpH and sulfate concentration coupled with increasing totalalkalinity indicate degradation of organic matter. A slightincrease in Sr2+ concentration at constant Mg2+ concentra-tion and Sr2+/Ca2+ ratio in the 0–40 mbsf interval suggestsminor aragonite dissolution, while the sharp increase inSr2+ concentration and Sr2+/Ca2+ ratio and decrease inMg2+ concentration below 40 mbsf is indicative of activearagonite recrystallization into calcite and dolomite below�150 mbsf. The large increase in pore-water Cl� concentra-tion with depth suggests that little fluid circulation is occur-ring below 30–40 mbsf and that the solid + fluid form aclosed-system in which alteration/recrystallization pro-cesses are happening (i.e., sediment-buffered conditions,Higgins et al., 2018).

We selected 43 samples from Site 1009 for U, O, C, andN isotope analysis, with a sampling frequency of �5 malong the drill core. The selected samples cover the wholerange of mineralogical and geochemical (i.e., CaCO3 wt%,MgCO3 wt%, [Sr], d13Cinorg, and d18Oinorg) variability ofthe recovered section. The samples range from pristinearagonite to recrystallized facies, and include one hard-ground at 99.12 mbsf. Six samples were collected fromthe glacial packet of sediments representing several MIS,including the hardground sample at MIS 12, which coin-cides with a glacial lowstand �430 kyr ago and subaerial

Fig. 2. Age vs. depth profile of Site 1009, for the entire drill core (left panel) and for the top 80 m (right panel). Black squares representcalcareous nannofossils and planktonic foraminifera biostratigraphic datums (Eberli et al., 1997), while open circles, triangles, and diamondsare radiometric ages from, respectively, U-Th (Robinson et al., 2002), U-Th (Henderson, 2002), and 14C (Slowey et al., 2002). The blue curveis a composite age vs. depth profile built using radiometric ages when available and bioevents otherwise (see supplementary materials for moredetails). Sedimentation rates calculated from the composite profile as well as stratigraphic sequences (Eberli et al., 1997) are also shown. (Forinterpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 3. Bulk cumulative mineralogy (normalized to 100% carbonate), carbon and oxygen isotope compositions, and elemental geochemistrydepth profiles for Site 1009 samples (LMC: low-Mg calcite, HMC: high-Mg calcite, square = unlithified, crosses = partially lithified, andcircles = lithified; data are from Malone, 2000).


Fig. 4. Pore-water depth profiles at Site 1009 (U data from Henderson et al., 1999; other data from Eberli et al., 1997). Blue lines representseawater values. � = unlithified, � = partially lithified, � = lithified. (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)


exposure at that time (Malone et al., 2001). These sampleswere specifically targeted to assess the effect of sea-levelvariations on the 238U/235U ratios.

2.2. Other carbonate, carbonatite and shale samples

Uranium concentrations in carbonates vary from severalppm in aragonite to sub-ppm levels in calcite and dolomite(e.g., Rogers and Adams, 1969; Swart and Hubbard, 1982;Reeder et al., 2000; Lau et al., 2016, 2017). To obtainenough U to achieve high-precision d238U measurements(>150 ng of U), large sample masses (up to several gramsin some cases) are needed, which cannot realistically beobtained by micro-drilling. Bulk-rock measurements canpotentially be influenced by incorporation of carbonateaffected by secondary processes, such as exchange with cir-culating fluids and recrystallization. Furthermore, in car-

Table 1Details of samples used for development of leaching protocol.

Sample Location Age

Stony Coral Florida, USA <10,000 yrFlorida Coral M. Florida, USA ModernBermuda Coral Hamilton, Bermuda, USA ModernBoca Raton Coral Boca Raton, Florida, USA ModernLake Worth Coral Lake Worth, Florida, USA PleistoceneFlorida Coral P. Hollywood Beach, Florida, USA PlioceneLimestone Key Largo, Florida, USA �120 kyrDolomite San Salvador, Bahamas, USA Pliocene

CaCO3 and MgCO3 are approximate values estimated using a scanning

bonate rocks, U can also be associated with extraneouscomponents such as phosphates, iron oxides, silicates, andorganics. A step-digestion protocol using dilute acetic acidwas therefore developed in order to selectively recovercarbonate-associated uranium, and minimize the mobiliza-tion of U associated with non-carbonate components andsecondary carbonate overgrowth. Several samples wereused to test the various steps and the adequacy of this pro-tocol. They are listed in Table 1 and comprise a modernstony coral (from Florida, �99% CaCO3, <10,000 yr old,[U] =1.85 ppm, available from Ward Science), three othermodern corals, two older (Pleistocene and Pliocene) corals,one modern limestone, and one Pliocene dolomite.

To assess the effect of the digestion protocol on other U-rich lithologies, two additional samples were analyzed: theCOQ-1 carbonatite USGS geostandard (Canada), and theSBC-1 shale USGS geostandard (USA), which have U

CaCO3 (wt%) MgCO3 (wt%) [U] (ppm) Source

�99% �1.85 Wards Science85.9% 1.1% �2.5 Peter Swart86.3% 0.8% �3.1 Field Museum78.5% 1.1% �2.8 Field Museum83.2% 0.4% �2.9 Field Museum84.0% 0.7% �2.79 Field Museum87.6% 1.2% �0.6 Peter Swart55.6% 37.9% �2.0 Peter Swart

electron microscope.


concentration of 10.5 ± 1.5 ppm and 5.6 ± 0.6 ppm, respec-tively (Tissot and Dauphas, 2015).

3. METHODS

3.1. Notations

The 238U/235U ratios are reported in permil as d238U val-ues relative to the U standard CRM-112a (also namedSRM960 or NBL112-a; CRM-145 for the solution form):

d238U¼ 238U=235U� �

sample= 238U= 235U� �

CRM�112a�1

h i�103:

ð1ÞThe uncertainties are calculated as 2� rStandard=

ffiffiffin

p,

where 2� rStandard is the daily external reproducibility(2SD) of repeat measurements of the standard CRM-112abracketed by itself, and n is the number of solution mea-surements of the sample (see Tissot and Dauphas, 2015for more details on error propagation).

The 234U/238U ratios are expressed relative to secularequilibrium as:

d234U ¼ 234U= 238U� �

sample= 234U= 238U� �

Eq� 1

h i� 103;

ð2Þwhere (234U/238U)Eq is the atomic ratio at secular equilib-rium and is equal to the ratio of the decay-constants of238U and 234U, k238/k234 = (1.5513 � 10�10)/(2.8220 � 10�6) = 5.4970 � 10�5 (Cheng et al., 2013).Errors are reported in the same way as for d238U values.

The inorganic 13C/12C and 18O/16O isotope ratios weremeasured relative to NBS-19 (National Bureau of Stan-dards) and are reported as d13Cinorg and d18Oinorg relativeto V-PDB (Vienna Pee Dee Belemnite) using the conven-tional delta notation. Replicate analyses yielded a precision0.1‰ (2SD) for both d13Cinorg and d18Oinorg.

The organic 13C/12C and 15N/14N isotope ratios arereported as d13Corg and d15Norg values, relative to, respec-tively, V-PDB and AIR. Replicate analyses of an internalstandard of glycine (n = 54, d13Corg = �31.8‰ V-PDB,d15Norg = �1.0‰ AIR) yielded a precision of ±0.1‰(2SD) for both values.

3.2. Establishing a step-digestion protocol

Several studies (e.g., Bailey et al., 2000; Kuznetsov et al.,2005; Li et al., 2011; Liu et al., 2013; Liu et al., 2014; Zhanget al., 2015; Tostevin et al., 2016) have explored the poten-tial of dilute acid step-leaching dissolution as a means ofextracting primary geochemical signals (REE + Y patterns,87Sr/86Sr, 207Pb/204Pb-206Pb/204Pb, d26Mg values) frombulk carbonates. Though the observed release patterns varyas a function of the exact leaching protocol used and/or themineralogy of the samples (e.g., Liu et al., 2014; Zhanget al., 2015; Tostevin et al., 2016), all studies paint a gener-ally consistent picture. Between 0 and 30% of carbonatedigested, the geochemical signals of the dissolved fractionshow clear deviation from marine values due to contribu-tion from surface-adsorbed phases, ion-exchange sites in

organic matter, carbonates or clay minerals, salts, and/orsecondary carbonate overgrowths. Similarly, once morethan �70% of carbonate is digested, the signals again startto deviate from marine values due to partial dissolution ofthe non-carbonate residuum, (e.g., phosphates, Fe–Mnoxyhydroxides, clay minerals). In the ‘‘central cut”(between 30 and 70% of bulk carbonate digested), however,the geochemical proxies investigated display values in excel-lent agreement with, or closest to, those of seawater.

The U system could suffer similar contamination and/orU gain/loss issues that might affect d238U values. Recently,Zhang et al. (2018b) using a step leaching protocol withincreasingly stronger acetic acid solution (up to 10 % vol)reported d238U values varying by up to 0.50‰ between lea-chates of five carbonate samples (including one coral, twoBahamian shallow-water core sediments and twoPermian-Triassic carbonates). Because the d238U values ofthe bulk carbonate were indistinguishable from theweighted mean d238U of their respective leachates, theauthors concluded that the bulk carbonates were not iso-topically reset. This result, however, does not inform aboutthe extent of isotopic resetting of the sample. It simply con-firms that bulk digestion and step-leaching approach dis-solve the carbonate fraction to the same degree. Thed238U variations observed by Zhang et al. (2018b) duringstep leaching of carbonates might therefore be primary(i.e., different carbonate components formed with differentd238U values), secondary (i.e., d238U values were modifiedby U remobilization) or due to contamination from non-carbonate phases.

To understand the origin of the d238U variationsobserved during progressive dissolution of carbonates andassess whether a particular leaching step more faithfullyrecords the seawater U isotope composition, we conductedthe following experiments (see Section 3.5 for measurementdetails):

(1) To assess the amount of U released during MQ-waterand 1 M ammonium acetate washing, 2000 mg of astony coral sample powder were leached in a cen-trifuge tube for 24 h with 10 mL of each reagent, suc-cessively. The tubes were placed on a ThermoScientific MaxQ Shaker in a holder at a 45� angleto avoid settling of the solid. The samples were cen-trifuged three times on a Fischer Scientific Centrificcentrifuge (10 min at �2500 rpm). The supernatantswere pipetted out after each centrifugation and theresidues rinsed with MQ-water before the next stepto ensure full recovery of the dissolved species. Asimilar test was also conducted on the COQ-1 car-bonatite and the SBC-1 shale. Results are shown inTables S3 and S4 and indicate very limited U leach-ing (<2%) during both steps.

(2) To assess the amount and isotopic composition of Ureleased during step leaching with dilute (<20 wt%)acetic acid, two replicate digestions of the stony coralsample powder were done in 10% step increments(after Bailey et al., 2000; Zhang et al., 2015), using�2000 mg of powder each time. One of these powderaliquots was pre-cleaned with MQ-water and


ammonium acetate (see previous test). The generalcarbonate dissolution reaction can be written as:

Ca;Mg;Mn;Feð ÞCO3ðsÞ þ 2HþðaqÞ

! ðCa;Mg;Mn;FeÞ2þðaqÞ þ CO2ðgÞ þH2OðlÞ ð3Þwhere Ca;Mg;Mn;Feð ÞCO3ðsÞ account for all forms

of carbonates. To digest a given percentage, x, ofthe carbonate, the amount of acetic acid to add tothe sample depends on (i) the concentration of theacetic acid solution (CAcetic, here equal to 0.2 for a20 wt% solution), (ii) the total carbonate content ofthe sample (carb%), (iii) the molar weight of the car-bonate in the sample (MCarb) and the acetic acid(MAcetic = 60.05 g/mol), (iv) the density of the aceticacid solution (d = 1.01 g/mL for a 20 wt% solution),and (v) the mass of sample being used (msmp, in g).The volume, V (in mL), of acetic acid of concentra-tion CAcetic to add is calculated as:

V ¼ 2 msmp

MAcetic

MCarb

carb%

CAcetic dx ð4Þ

where the factor 2 accounts for the fact that 2 moles

of Hþ are required to digest 1 mole of carbonate.Though acetic acid is a weak acid, we calculate theamount of acid to add as if it were a strong acid(i.e., totally dissociated) because the consumptionof H+ ions during carbonate digestion drives the dis-sociation of CH3COOH to near completion. For apure carbonate sample like the stony coral, theamount of sample digested was typically accuratewithin ±5%. For each digestion step, the sampleswere placed on the MaxQ Shaker until reactionceased (typically 12–24 h). The supernatants werepipetted out after centrifugation as described above.The residues were left to dry and the next digestionstep was only started after complete drying: i.e., afterthe mass of the sample stopped decreasing (typicallyafter 12–24 h).The results of the leaching experimentsare shown in Table S5 and Fig. 5. The first few lea-chates (up to 20–30% of carbonate digested) aredepleted in U while the intermediate and late lea-chates show concentrations oscillating around thebulk value. Similarly, the isotopic composition of Ureleased in each step is variable and can differ fromthat of modern seawater (d238USW = �0.39± 0.02‰, Tissot and Dauphas, 2015; andd(234U)SW = 144.9 ± 0.4‰, Chutcharavan et al.,2018) by up to 0.20–0.30‰ for d238U and up to 4‰for d(234U) values. The low concentrations, highd238U values, and low d(234U) values observed inthe first few leachates might indicate either dissolu-tion of an internal coral component significantly dif-ferent from the bulk carbonate or disturbance to Uhosted at ion-exchange sites and/or in easily digestedfractions of the carbonate. For �30 to 85% digestion,the U isotopic composition matches that of modernseawater, exactly for d238U and within 1‰ ford(234U) values, showing that internal components ofcorals do record the isotopic composition of the

ambient seawater from which they precipitated.Deviation from seawater d238U value observed inthe last few percent of digested carbonate might bedue to either (i) an originally lower d238U value inthe most resistant fraction of the carbonate, (ii) con-tamination from non-carbonate residuum, or (iii) thecumulative effect of a weak preferential leaching of238U in the preceding steps, leaving a 235U-enrichedcarbonate fraction to be digested last (see Section 5.1for further discussion).

(3) The same step-digestion protocol was applied to car-bonatite COQ-1 (assuming that the sample is 100%carbonate) to assess the effect of acetic acid leachingon an igneous carbonate rock. The results are shownin Table S6. Less than 2% of the total U in the car-bonatite was released during the entire leaching pro-tocol. This result indicates that the weak chemicalattack does not affect the major U carriers in the car-bonatite, which are, apart from apatite, chemicallyresistant phases (e.g., perovskite, garnet), and thatour protocol only mobilizes U associated withcarbonates.

(4) A last test was performed in which a 3:1 mixture ofcarbonatite COQ-1 and shale SBC-1 (100 mg total)was placed in a centrifuge tube with excess acid toevaluate the effect of leaching of a detrital componentwith different acids. The amount of acid added to thepowder was enough to digest twice the sample mass ifit were 100% carbonate. An �8 wt% acetic acid solu-tion and a 1 M HCl solution were used for these tests.The results are shown in Table S7. The U release waslimited in both cases, yet the 1 M HCl solution lea-ched 4 times more U than the acetic acid solution.Only �1% of the U present in the mixture wasreleased by the acetic acid attack, indicating that astep-leaching protocol using this acid will minimizecontribution of silicate- and organic-bound uraniumand maximize release of carbonate-associated ura-nium, as intended.

3.3. Optimized digestion protocol

Several paleoredox reconstruction studies using U iso-topes in carbonates have been published, and almost asmany methods have been used to release carbonate-associated uranium: 1 M HCl (Brennecka et al., 2011;Wei et al., 2018; Zhang et al., 2018a; Zhang et al., 2018b;Zhang et al., 2018c), 1 M HNO3 (Song et al., 2017), 3 MHNO3 (Romaniello et al., 2013), <0.5 M HCl (Dahlet al., 2014), 1 M HCl followed by 1 M HNO3 (Azmyet al., 2015); 0.25 N HCl (Lau et al., 2016; Lau et al.,2017; Jost et al., 2017), 1 N acetic acid (Elrick et al.,2017; Bartlett et al., 2018), pH-buffered 1 M sodium acetate(pH = 5 at room T; Clarkson et al., 2018), and pH-buffered10% (1.7 M) acetic acid (pH > 3.6, T = 50 �C; Dahl et al.,2017). Only three of these sixteen studies report the resultsof leaching tests. Lau et al. (2016) showed that lower molar-ity (<0.5 N) HCl leaches have lower Mn/Sr ratios than lea-ches obtained with higher molarity HCl, thus suggesting

Fig. 5. (Left) Concentration and isotopic release pattern of U during step leaching of a stony coral with dilute acetic acid (<20 wt%) as afunction of the percent of carbonate dissolved. (Right) Same as left, but after pre-cleaning with MQ-water and 1 M ammonium acetate. Reddashed lines show bulk sample U concentration. Blue horizontal bars represent modern seawater values. Open circles show the weightedaverage of previous steps. The bulk carbonate and leachates collected after �25–30% of the carbonate was digested and leaving �10–15% ofcarbonate undissolved record the U isotopic composition of seawater from which carbonate precipitated, suggesting that different coralcomponents have different d238U values. (For interpretation of the references to color in this figure legend, the reader is referred to the webversion of this article.)


minimal digestion of diagenetically altered components.Dahl et al. (2017) studied the concentration and isotopiccomposition of U released from various proportions of apure carbonate powder (chalk) and a phosphate-rich shale.They found that the use of dilute HCl resulted in significantrelease (�70 %) of phosphate-associated U, while pH-buffered 10% acetic acid resulted in only minor (�30%)release of phosphate-associated U. Using a step-leachingprotocol with acetic acid solutions of increasing strength,Zhang et al. (2018b) found that the earliest (<�30% carbon-ate digestion) and latest (>95% carbonate digestion) lea-chates of sedimentary carbonates could have d238U valuesdifferent by up to 0.50‰ relative to the central cut (between�30 and 90% carbonate digestion). These differences couldreflect (i) primary variability within the carbonate compo-nents, or (ii) isotopic resetting of the most easily digestedphases (early leachates) and isotopic contamination fromnon-carbonate phases (late leachates).

Based on these previous studies of U isotopes, the exist-ing body of work on bulk carbonates digestion (e.g., Baileyet al., 2000; Kuznetsov et al., 2005; Li et al., 2011; Liu et al.,2013; Liu et al., 2014; Zhang et al., 2015; Tostevin et al.,2016) and the results of our own tests above (Fig. 5 andTables S3–S7) a simple, three-step leaching protocol wasderived for carbonate digestion that would maximizerelease of carbonate-associated U, while at the same timeminimizing U release from other components (e.g., silicates,organics).

Step 1: Approximately 20–30% of the sample is digestedusing dilute (<20 wt%) acetic acid in order to removeeasily mobilized U (i.e., contaminant U at ion-exchange sites or in secondary carbonates) whose U iso-tope composition might differ from that of seawater. Nowater or ammonium acetate pre-cleaning step was foundnecessary as they mobilize very little U.


Step 2: The bulk of the carbonate (�40–70% of the sam-ple) is digested using dilute acetic acid and the solutionrecovered is used for high-precision U isotope analysis.Although only HF and HNO3 acids will significantly

release silicate- and organic-bound U, at least 10% ofthe sample is left undigested to avoid leaching of suchdetrital phases, whose U isotopic composition will likelybe different from that of the carbonate.Step 3 (optional): The remainder of the carbonate isdigested, with dilute acetic acid, in order to study the fullU inventory of the samples.

Although not implemented here because of the lack ofphosphates in the samples, this protocol can be modifiedto minimize the contribution of phosphate-associated Uby buffering the pH of the acid to a value above 3.6, usingcalcium acetate (Jeppsson et al., 1985; Jeppsson et al., 1999;Dahl et al., 2017).

Given the absence of dolomite from our samples, andthe observation by Tostevin et al. (2016) that REE + Yrelease patterns are different in dolomite relative to calciteduring step-leaching, the relevance of our protocol to thestudy of dolomite samples remains to be assessed. Never-theless, we emphasize that this protocol is very similar tothose proposed to extract the most pristine REE + Y pat-terns, 87Sr/86Sr, 207Pb/204Pb-206Pb/204Pb and d26Mg valuesin an array of carbonate ranging from almost pure lime-stones to dolostones (Bailey et al., 2000; Kuznetsov et al.,2005; Li et al., 2011; Liu et al., 2013; Liu et al., 2014;Zhang et al., 2015; Tostevin et al., 2016), suggesting thatcombined analysis of REE + Y patterns, 207Pb/204-Pb-206Pb/204Pb, Mg, Sr, and U isotopes in the same lea-chates is a viable option for future studies of seawatercomposition.

3.4. Sample processing of modern and Site 1009 carbonates

The leaching protocol was applied to an array of car-bonates (Table 1) as well as the 43 Bahamas Bank samplesselected from the ODP Leg 166 Site 1009. The samples werecrushed into fine powder using agate mortar and pestle. Eq.(4) was used to calculate the amount of acid to add for eachdigestion step. For some samples (the corals, limestone, anddolostone listed in Table 1), no chemical composition datawas available and the carbonate content was estimatedfrom major element abundances determined on compactedpowder pellets using a JEOL JSM-5800LV scanning elec-tron microscope.

To ensure that the entire sample powder was well wettedprior to acid addition, the samples were first covered with10 mL of MQ-water, to which the required amount of 20wt% acetic acid was then added. The actual acetic acid con-centration used for carbonate digestion therefore variedbetween 2 wt% and 12 wt%. This approach has the benefitof avoiding direct contact of the 20 wt% acid solution withthe sample powder and therefore alleviates concern of pref-erential dissolution of the first grains being exposed to theacid.

Sample leaching was done in clean centrifuge tubesplaced on a Thermo Scientific MaxQ Shaker in a holder

at a 45� angle to avoid settling of the solid and protectionof carbonate grains by acetic-acid resistant phases, whichcould lead to low digestion yields (Dahl et al., 2017). Oncereaction ceased (typically after �24 h), the tubes were cen-trifuged three times for 10 min at �2500 rpm. The super-natants were pipetted out after each centrifugation andthe residues rinsed with MQ-water before the next step toensure full recovery of the dissolved uranium. The residueswere then dried in a laminar flow hood at T�30–40 �C(using heat lamps). Only limited heating was applied tothe samples during drying to avoid mineral transformationso some small amount of moisture might have remained.Nevertheless, once the sample mass stopped decreasing,the stabilized weight was used to calculate the amount ofsample digested before the next digestion step.

3.5. Sample spiking, uranium purification, and mass

spectrometry

After digestion, a small aliquot (2–5%) of each cut wastaken, diluted 10-fold with 0.3 M HNO3 and used for238U and 232Th concentration measurement (without anycolumn chemistry). Concentration measurements were per-formed on the Neptune MC-ICP-MS in wet plasma mode(i.e., spray chamber). Every third sample analysis wasbracketed by measurement of a standard solution of Uand Th at �2.5 ppb. The uncertainty on [U] and [Th] thusachieved was typically ±20% (see Table S8). After concen-tration check, a liquid aliquot of the ‘‘central cut” (�35–80% carbonate digestion) containing �240 ng of U wastransferred into a clean Teflon beaker and spiked withIRMM-3636 U double spike (50.46% of 233U and 49.51%of 236U; Verbruggen et al., 2008). Enough spike was addedto obtain a Uspike/Usample ratio of �3%. After spiking, thesamples were dried completely and taken back into concen-trated HNO3 before dilution to 3 M HNO3.

U purification and isotope measurements were done fol-lowing the procedure thoroughly described in Telus et al.(2012), Tissot and Dauphas (2015) and Tissot et al.(2017) (see supplementary materials for more details). Ura-nium procedural blank varied between 12 and 30 pg(<0.02% of sample uranium) and are therefore negligible.

3.6. d13Cinorg, d18Oinorg, d

13Corg, d15Norg, and TOC analyses

These analyses were performed following the routinemethods from Swart et al. (1991) and Oehlert and Swart(2014) (see supplementary materials for more details).

4. RESULTS

4.1. Geostandards

Some geostandards were analyzed to assess the accuracyof the U isotope measurements. Three total digestion repli-cates of the Columbia River basalt (BCR-2) were preparedfrom fresh powder aliquots, processed along with the sam-ples and gave reproducible d238U values of �0.24 ± 0.03‰,�0.25 ± 0.03‰, and �0.27 ± 0.03‰ (2SE). This is identicalto the average value of �0.27 ± 0.05‰ (95% CI), which


includes data from 11 studies (Tissot and Dauphas, 2015,supplementary material). Purified U cuts of three seawatersamples from a previous study (Tissot and Dauphas, 2015)were also measured (Table S5). The Faedra sample(Mediterranean Sea) was measured at �0.39 ± 0.05‰(replicate 1), �0.39 ± 0.04‰ (replicate 2), �0.38 ± 0.03‰(replicate 3), and �0.40 ± 0.07‰ (replicate 4), in agreementwith the value of �0.39 ± 0.05‰ previously reported forthe same sample. Similarly, the Abu Dhabi sample wasmeasured at �0.37 ± 0.04‰ (replicate 1) and �0.38± 0.04‰ (replicate 2), and the Dubai sample was measuredat �0.35 ± 0.04‰ (replicate 1), �0.36 ± 0.03‰ (replicate2), and �0.35 ± 0.06‰ (replicate 3), in agreement withthe values of, respectively, �0.34 ± 0.05‰ and �0.32± 0.05‰, previously reported.

4.2. Corals, limestone, and dolostone

The leachates U concentration and isotopic compositionof modern and ancient corals, the limestone, and the dolo-mite are presented in Fig. 6 and Table S5. Both modern andancient (Pleistocene and Pliocene) corals show a U release

Fig. 6. Concentration and U isotope release pattern for various modern cusing the optimized dilute acetic acid three-step leaching protocol (see S

pattern similar to that obtained for the stony coral(Fig. 5). Step 1 is U-poor relative to the bulk sample, and238U-rich compared to the rest of the carbonate (Steps 2and 3), while Steps 2 and 3 show similar U contents andd238U values, which are within error of the modern seawa-ter value. Modern corals have d(234U) values close to thatof modern seawater (for all steps) and consistent with theiryoung ages (less than 30 kyr). The Pleistocene and Pliocenecorals show lower than modern seawater d(234U) values,corresponding to model ages (i.e., decay interval startingfrom a modern seawater composition) of �100 kyr and70 kyr, respectively, which is consistent with the Pleistocene(10–1600 kyr) age of the first specimen, but indicates reset-ting of uranium systematics in the Pliocene (2.5–5.3 Myr)sample. For both samples, d(234U) values show resolvable,yet limited, variations among the three steps (within 10‰of each other). The 120 kyr old limestone shows constantU concentration and d238U values in all steps, but variabled(234U) values. The d238U values are 0.20‰ higher thanmodern seawater, while the d(234U) values start 30‰ lighterthan seawater in Step 1, increase to within 5‰ of seawaterin Step 2 and decrease to 16‰ below seawater in Step 3.

orals (a-c), ancient corals (d and e), limestone (f) and dolostone (g),ection 3.3). Symbols as in Fig. 5. See text for details.


For the dolomite sample, only Step 1 and half of Step 2were performed. U concentration and isotope abundancesare essentially constant in both steps, with d238U values0.45‰ higher than seawater and d(234U) values close to0‰ (i.e., at secular equilibrium).

4.3. Site 1009 samples

4.3.1. Matrix effects on d238U and d(234U) values

As the U double-spike consists almost entirely of 233Uand 236U, a simple quality control test for d238U measure-ments is to check the agreement between the d238UDS+SSB

Fig. 7. d238U values obtained with the double-spike data reductionmethod plotted against the raw d238U measured, all values aresample-standard bracketed (SSB). Disagreement of the two sets ofvalues after 1 column chemistry (grey) indicates that matrix effectsaffected the measurements (fore more details see Tissot andDauphas, 2015). Samples processed twice through column chem-istry (white circles) plot on the 1:1 line, indicating the absence ofmatrix effects.

Fig. 8. Comparison of U concentration and isotopic composition measurthe sample-standard bracketed data (Fig. 7) has, in most cases, little to ndouble-spike method. The only exception is sample 19.12, for which mchemistry pass (grey arrows).

values, obtained from the double-spike data reduction,and the d238USSB values, obtained from the raw 238U/235Uratios (corrected for the spike 234U, 235U and 238U contribu-tion, on peak zero, hydride formation and tailing of 238Uonto lighter isotopes) bracketed by standard measurementsbut without double-spike correction (Tissot and Dauphas,2015). Some samples from Site 1009 show disagreementbetween the d238UDS+SSB and d238USSB values (on average,the d238UDS+SSB-d

238USSB is �+0.3‰, but it covers a rangeof 1.41‰; grey symbols on Fig. 7). For half of the drill coresamples, a duplicate analysis was thus performed after asecond column chemistry to ensure complete matrixremoval. Measurements performed on this double-cleanedsolution show much better agreement between SSB andDS data (d238UDS+SSB-d

238USSB = 0.01‰ on average, cov-ering a range of only 0.43‰; open symbols on Fig. 7).For almost all samples [U] and d238U values obtained usingthe DS are identical (within uncertainties) after one or twopurifications (left and center panels; Fig. 8). Similarly,d(234U) values after one and two chemistry passes are ingood agreement, typically within 0.5 to 4‰ of each other(right panel; Fig. 8). These observations demonstrate therobustness of the DS method and imply that a single col-umn chemistry pass is sufficient to obtain reliable data.The only exception is sample 19.12, for which a differenceof 99‰ in the d(234U) value was observed after one andtwo chemistries. For this sample, a third replicate samplealiquot was processed through column chemistry twiceand measured again, yielding [U], d238U and d(234U) valuesidentical to the previous measurement made after two col-umn passes, showing that some matrix effect (most likelydue to the presence of Ca) was affecting the measurementafter one column chemistry. These reproducible values areused in the figures and discussion below.

4.3.2. Uranium concentration and isotopic composition

The U release pattern of the carbonate samples fromSite 1009 (Fig. 9 and Table S8) is very similar to that ofthe corals (Figs. 5 and 6). There is 2 to 40 times less U lea-ched during Step 1 than during Steps 2 and 3. Thorium

ed after one and two column chemistry. The matrix effect visible ino effect on the [U], d238U, and d(234U) values calculated using theatrix effect was still affecting the measurement after one column

Fig. 9. Results of the step-digestion of Site 1009 carbonates: fraction of carbonate digested (left), U concentration in the sample mass digested(center), and fraction of the total U released (right) in each of the three steps.


concentrations (and Th/U ratios) are low in Steps 1 and 2,but increase in Step 3 (Table S8), indicating leaching ofnon-carbonate phases in this last step. These results furthervalidate the adequacy of our leaching protocol. Step 1clearly removes a component that is not representative ofthe bulk sample, while leaving some of the carbonate undi-gested helps to avoid incorporation of U associated withnon-carbonate (i.e., detrital) phases.

The concentration and isotope composition of U in the‘‘central cut” (Step 2) are shown in Fig. 10 and Table 2.Uranium concentrations show significant dispersion from1.9 to 12.1 ppm, with no clear trend with depth. Similarly,no trend is observed in the 238U/235U ratios with depth,and d238U values vary from �0.94 to +0.13‰. Unlithifiedsamples (which mostly occur in the top portion of the core)have slightly higher U content and more dispersed d238Uvalues than partially lithified samples (which are most com-mon at the bottom portion of the core) (Fig. 10d and e).Almost all the samples are enriched in U (average [U]= 6.5 ± 0.7 ppm, 2se) compared to allochemical fragments(e.g., corals, calcareous algae, molluscs, and benthic for-ams; [U] = 2.3 ± 1.3 ppm, Dunk et al., 2002), and haved238U values ranging from seawater-like values to�0.50‰ higher than seawater. One exception is the highlylithified sample 99.12, which has one of the lowest d238Uvalues measured in a carbonate, �0.94 ± 0.03‰.

In contrast to the [U] and d238U profiles, there is a cleardecreasing trend in d(234U) values with depth (Fig. 10c),whereby the samples close to the water-sediment interfacehave seawater-like d(234U) values (�145‰), and the sam-ples at the bottom of the core have d(234U) values closeto secular equilibrium (�10‰). Our measurements are inexcellent agreement with published data obtained on sam-ples at a similar depth in the drill core (grey symbols onFig. 10c; Henderson, 2002; Robinson et al., 2002; seeTable 3). Some samples below �75 mbsf are systematicallyenriched in 234U (up to 100‰) compared to what would be

expected given the age of the samples (red curve onFig. 10c), indicating open-system behavior and U exchangewith a phase characterized by a high 234U/238U ratio, mostlikely, the circulating pore-water (Fig. 4).

4.3.3. TOC content, d13Cinorg, d18Oinorg, d

13Corg, and d15Norg

values

The bulk TOC content, d13Cinorg and d18Oinorg values,and organic d13Corg and d15Norg values are presented inFig. 11 and Table 4. The carbon and oxygen isotope dataare in excellent agreement with previous analyses (grey sym-bols in Fig. 11; from Malone, 2000) with d18Oinorg valuesvarying between �1.65 and +1.76‰ V-PDB, and d13Cinorg

values varying between +2.10 and +4.98‰ V-PDB. Par-tially and fully lithified samples have higher d18Oinorg andlower d13Cinorg values than their less lithified neighbors,and a shift to more positive d18Oinorg values is observedat �125 mbsf, corresponding to a general increase in lithi-fication, possibly due to recrystallization in the presenceof pore-water with more positive d18O values (Maloneet al., 2001).

The d13Corg and d15Norg values vary between �20.46 and�10.13‰ V-PDB and 0.49 and 2.92‰ AIR, respectively.Given that Bahamas platform carbonates have d13Cinorg

between +3‰ and +6‰ and d13Corg between �7‰ to�16‰, and pelagic materials have d13Cinorg between 0‰and +1‰ and d13Corg between �20‰ to �22‰ (Oehlertet al., 2012), the d13Cinorg and d13Corg values measured inthe drill core indicate that the Site 1009 sediments are mainlyplatform-derived rather than pelagic in origin (Fig. 12).

The TOC contents are low (below 1 wt%), as suggestedby the limited initial shipboard data (grey symbols inFig. 11, from Eberli et al., 1997), reflecting significant oxi-dation of organic matter during transport and burial(Oehlert et al., 2012). There is no statistically significantcorrelation between TOC content and d238U values(Fig. 13).

Fig. 10. Concentration (a) and U isotope composition (b, c and f) depth profiles in the ‘‘central cut” (Step 2) of Site 1009 carbonates.Concentration and d238U histograms are also shown (d and e). Orange vertical bars represent the average of the dataset and blue vertical barsshow the seawater value. Blue filled circles denote samples deposited during sea-level lowstands. Symbols: � = unlithified, � = partiallylithified, � = lithified. On panel (c) and (f), the green, blue, and red curves show the expected d(234U) depth profiles calculated assuming thatthe samples formed with the seawater composition and closed, respectively, 0, 5, and 10 m below the water-sediment interface. (Forinterpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)


In order to assess what controls the variability in U con-centration and isotope composition of the samples a scatterplot matrix of d238U, d13Cinorg, d18Oinorg, TOC, d15Norg

(this study) and geochemical data (U/Ca, Sr/Ca, U/Sr,Mg/Ca, and Na/Ca; Malone, 2000) is shown in Fig. 13. Sta-tistically significant correlations were evaluated by calculat-ing p-values and the Spearman’s rank correlationcoefficients (q) using the cor.test() function in R. The Spear-man’s rank test assesses how well the relationship betweentwo variables can be described using a monotonic (not nec-essarily linear) function. Because of its extreme composi-tion, the hardground sample was plotted, but not used forcalculation of p-values or correlation coefficients. Severalstatistically significant correlations (i.e., p-value < 0.05)are revealed (panels in full colors in Fig. 13).

5. DISCUSSION

5.1. Modern corals: possibly distinct d238U values in centers

of calcification

The low [U] and d(234U), and high d238U valuesobserved in the first digestion cut of modern and ancientcorals (Fig. 6, Table S5) could reflects disturbance of U at

ion-exchange sites and in heavily altered regions, whichare more easily digested. However, because the bulk mod-ern coral d238U and d(234U) values are indistinguishablefrom seawater within 0.02‰ and �1‰, respectively, thevariability in the U concentration and isotope release pat-terns (Figs. 5 and 6) more likely holds information aboutcoral U geochemistry. In a study of modern and fossil scle-ractinian corals, Robinson et al. (2006) showed that thecenters of calcification (COC, the dark bands where calcifi-cation is initiated, which run along the middle of coral septaand theca) are characterized by smaller crystal sizes (i.e.,fine-grained aragonite) than other parts of the coral, as wellas U/Ca ratios 2–3 times lower and d(234U) values higherthan other parts of the coral. These COCs are also charac-terized by fast precipitation rate, potentially due to highsaturation state or high pH conditions during calcification.Carbonate precipitation experiments have shown that highpH conditions result in 0.10–0.20‰ higher d238U values inaragonite relative to the liquid phase (Chen et al., 2016b),implying that higher d238U values should be observed inthe COC compared to the rest of the coral. This predictionis consistent with the seawater-like d238U value (�0.40± 0.03‰) observed in a COC-poor sub-sample of Carib-bean coral Acropora palmata and the �0.10‰ higher

Table 2Summary of U isotopic compositions and concentrations of Site 1009 carbonate samples measured in the second leachate (digestion Step 2).

Core 166, Used forchemistry

Double-spike data reduction Sample-standard bracketing

Sample Section 1009A Degree Cycles Col. d238U d(234U) Conc. Usp/ d238U d(234U)(mbsf) Interval Lithologya of lith.b #c #d (mg) n (‰)e (‰)f (ppm) Usmp (‰)e (‰)f

0.32 1H-1, 32–34 cm ul pel bio wack 50 1 104.08 7 0.01 ± 0.03 144.7 ± 0.3 4.50 ± 0.03 1.84% �0.29 ± 0.17 146.4 ± 0.460 2 104.08 4 0.01 ± 0.03 145.6 ± 0.3 4.50 ± 0.03 1.84% 0.12 ± 0.12 146.9 ± 0.3

5.12 2H-1, 32–34 cm ul pel wack 50 1 55.40 5 0.10 ± 0.02 142.9 ± 0.3 4.25 ± 0.12 3.64% �0.18 ± 0.31 144.5 ± 0.460 2 55.40 5 0.11 ± 0.03 145.4 ± 0.2 4.25 ± 0.12 3.64% 0.16 ± 0.34 145.5 ± 0.4

9.62 2H-4, 32–34 cm ul pel bio wack 50 1 52.44 7 0.09 ± 0.03 144.1 ± 0.2 8.86 ± 0.07 1.86% 0.09 ± 0.19 145.3 ± 0.360 2 52.44 6 0.09 ± 0.01 143.9 ± 0.2 8.86 ± 0.07 1.86% 0.04 ± 0.21 145.2 ± 0.3

14.62 3H-1, 32–34 cm ul pel wack-mud 50 1 73.26 8 �0.10 ± 0.03 143.2 ± 0.3 6.83 ± 0.05 1.72% �0.34 ± 0.67 144.7 ± 0.919.12 3H-4, 32–34 cm ul pel mud 50 1 65.50 9 �0.31 ± 0.02 27.8 ± 0.2 3.65 ± 0.09 3.47% �1.35 ± 0.41 32.4 ± 0.6

60 2 67.00 7 �0.43 ± 0.03 127.7 ± 0.2 5.02 ± 0.03 2.91% �0.53 ± 0.35 129.2 ± 0.460 2 68.70 8 �0.45 ± 0.02 123.3 ± 0.2 4.99 ± 0.03 3.22% �0.60 ± 0.42 122.0 ± 0.5

35.12 5H-2, 32–34 cm ul mud-wack 50 1 51.94 8 �0.14 ± 0.03 106.7 ± 0.3 9.71 ± 0.08 1.71% �0.48 ± 0.67 109.0 ± 0.938.61 6H-1, 31–33 cm floatstone PL 50 1 123.43 9 �0.37 ± 0.04 109.6 ± 0.2 3.66 ± 0.03 1.90% �0.75 ± 0.22 111.8 ± 0.3

60 2 123.43 5 �0.35 ± 0.03 110.5 ± 0.2 3.66 ± 0.03 1.90% �0.28 ± 0.34 112.0 ± 0.444.12* 7H-2, 32–34 cm ul pel wack 50 1 43.00 12 0.08 ± 0.03 92.3 ± 1.3 8.12 ± 0.29 2.51% 0.04 ± 0.18 93.8 ± 1.348.62 7H-5, 32–34 cm ul pel wack 50 1 35.45 9 �0.20 ± 0.04 89.4 ± 0.2 12.15 ± 0.11 2.00% �0.57 ± 0.22 91.9 ± 0.353.62 8H-2, 32–34 cm ul pel mud-wack 50 1 68.10 8 �0.11 ± 0.03 89.4 ± 0.3 6.19 ± 0.05 2.03% �0.54 ± 0.67 92.0 ± 0.958.12 8H-5, 32–34 cm ul pel wack 50 1 62.80 9 �0.03 ± 0.02 89.2 ± 0.2 4.94 ± 0.12 2.68% �0.44 ± 0.41 92.8 ± 0.6

60 2 72.30 8 �0.08 ± 0.02 90.4 ± 0.2 4.81 ± 0.03 3.44% 0.04 ± 0.42 92.1 ± 0.561.62 9H-1, 32–34 cm ul pel mud-wack 50 1 47.71 9 �0.12 ± 0.04 87.0 ± 0.2 8.81 ± 0.07 2.05% �0.49 ± 0.22 89.4 ± 0.364.62 9H-3, 32–34 cm ul bio wack 50 1 69.14 7 �0.23 ± 0.03 88.4 ± 0.2 6.19 ± 0.05 2.01% �0.21 ± 0.19 90.6 ± 0.369.12 9H-6, 32–34 cm ul pel mud 50 1 50.40 8 �0.11 ± 0.02 82.6 ± 0.3 7.11 ± 0.22 2.38% �0.39 ± 0.13 84.8 ± 0.374.12 11H-1, 32–34 cm ul pel wack 50 1 39.67 7 0.15 ± 0.03 60.8 ± 0.3 9.48 ± 0.08 2.27% �0.13 ± 0.17 63.8 ± 0.4

60 2 39.67 4 0.13 ± 0.03 61.3 ± 0.3 9.48 ± 0.08 2.27% 0.25 ± 0.12 63.4 ± 0.378.62 11H-4, 32–34 cm ul pel wack 50 1 42.84 8 �0.23 ± 0.03 66.3 ± 0.3 8.58 ± 0.07 2.33% �0.79 ± 0.67 69.2 ± 0.989.62* 13H-3, 32–34 cm ul foram pel pack 50 1 37.60 9 �0.18 ± 0.02 43.0 ± 1.0 8.55 ± 0.34 2.63% �0.50 ± 0.16 45.4 ± 0.994.12 13H-6, 32–34 cm ul pel wack-mud 50 1 34.58 8 �0.16 ± 0.03 52.7 ± 0.3 10.57 ± 0.10 2.34% �0.44 ± 0.67 55.5 ± 0.997.62 14H-2, 32–34 cm nanno ooze 50 1 66.86 9 �0.23 ± 0.04 51.1 ± 0.2 3.08 ± 0.07 4.16% �0.66 ± 0.22 52.9 ± 0.399.12* 14H-3, 32–34 cm hardground LITH 50 1 160.30 9 �0.98 ± 0.03 100.6 ± 1.5 1.91 ± 0.02 2.78% �1.06 ± 0.12 101.8 ± 1.5

60 2 175.00 6 �0.94 ± 0.03 101.4 ± 0.3 1.90 ± 0.01 3.53% �0.96 ± 0.49 103.1 ± 0.6103.62 14H-6, 32–34 cm ul pel wack 50 1 64.61 9 �0.47 ± 0.04 43.7 ± 0.2 5.02 ± 0.04 2.63% �0.79 ± 0.22 46.5 ± 0.3

60 2 64.61 4 �0.47 ± 0.03 44.5 ± 0.3 5.02 ± 0.04 2.63% �0.41 ± 0.12 46.7 ± 0.3108.62 15H-3, 32–34 cm ul pel mud-wack 50 1 44.90 9 �0.02 ± 0.02 37.6 ± 0.2 7.66 ± 0.26 2.35% �1.27 ± 0.41 42.7 ± 0.6

60 2 67.70 7 0.01 ± 0.03 38.0 ± 0.2 7.31 ± 0.03 3.27% �0.25 ± 0.45 40.9 ± 0.6113.59* 16H-CC, 29–31 cm ul pel wack 50 1 63.60 9 �0.32 ± 0.02 29.3 ± 1.0 4.98 ± 0.12 2.70% �0.55 ± 0.16 31.8 ± 0.9

60 2 63.60 7 �0.32 ± 0.03 29.3 ± 0.2 4.98 ± 0.04 2.70% �0.24 ± 0.35 31.7 ± 0.4115.62 17X-2, 32–34 cm ul float/ul wack w/lith 50 1 43.44 9 �0.17 ± 0.04 58.8 ± 0.2 7.30 ± 0.06 2.70% �0.54 ± 0.22 61.4 ± 0.3

60 2 43.44 4 �0.18 ± 0.02 60.1 ± 0.3 7.30 ± 0.06 2.70% �0.08 ± 0.26 61.9 ± 0.4120.12* 17X-5, 32–34 cm ul bio wack-float 50 1 50.40 12 �0.10 ± 0.03 39.8 ± 1.3 6.44 ± 0.19 2.66% �0.23 ± 0.18 42.3 ± 1.3

59 2 50.72 7 �0.11 ± 0.03 40.5 ± 0.2 6.12 ± 0.04 3.14% �0.42 ± 0.35 43.2 ± 0.4125.32* 18X-2, 32–34 cm ul bio wack 50 1 42.80 9 �0.37 ± 0.03 25.3 ± 1.5 6.08 ± 0.05 3.29% �0.58 ± 0.12 27.6 ± 1.5

60 2 42.80 5 �0.33 ± 0.03 27.1 ± 0.2 6.08 ± 0.22 3.29% �0.25 ± 0.34 29.0 ± 0.4(continued on next page)

F.L.H

.Tisso

tet

al./Geochim

icaet

Cosm

ochim

icaActa

242(2018)

233–265247

Table 2 (continued)

Core 166, Used forchemistry

Double-spike data reduction Sample-standard bracketing

Sample Section 1009A Degree Cycles Col. d238U d(234U) Conc. Usp/ d238U d(234U)(mbsf) Interval Lithologya of lith.b #c #d (mg) n (‰)e (‰)f (ppm) Usmp (‰)e (‰)f

133.22 19X-1, 32–34 cm pl bio wack PL 50 1 26.73 7 �0.19 ± 0.03 13.2 ± 0.2 11.60 ± 0.12 2.76% �0.32 ± 0.19 16.5 ± 0.3137.72* 19X-4, 32–34 cm pl bio wack PL 50 1 53.40 9 �0.28 ± 0.03 10.4 ± 1.5 5.66 ± 0.16 2.83% �0.49 ± 0.12 13.2 ± 1.5

60 2 70.70 9 �0.27 ± 0.03 10.4 ± 0.3 5.65 ± 0.03 3.46% �0.30 ± 0.27 13.3 ± 0.6142.82* 20X-1, 32–33 cm ul pel wack-mud 50 1 53.10 12 �0.12 ± 0.03 9.9 ± 1.3 5.37 ± 0.16 3.01% �0.38 ± 0.18 12.8 ± 1.3145.81 20X-3, 31–33 cm pl wack PL 50 1 38.66 7 �0.16 ± 0.03 56.2 ± 0.3 8.58 ± 0.07 2.58% �0.52 ± 0.17 59.1 ± 0.4

60 2 38.66 4 �0.18 ± 0.03 56.7 ± 0.3 8.58 ± 0.07 2.58% �0.25 ± 0.12 58.9 ± 0.3161.82 22X-1, 32–34 cm pl bio wack PL 50 1 62.60 8 0.02 ± 0.02 33.8 ± 0.3 5.00 ± 0.12 2.74% �0.30 ± 0.13 36.7 ± 0.3

60 2 62.60 4 0.04 ± 0.04 34.4 ± 0.3 5.00 ± 0.12 2.73% 0.07 ± 0.38 36.5 ± 0.5166.32* 22X-4, 32–34 cm pl pel wack-pack PL 50 1 47.90 12 �0.15 ± 0.03 91.9 ± 1.3 6.77 ± 0.22 2.64% �0.27 ± 0.18 93.4 ± 1.3

60 2 47.90 4 �0.19 ± 0.02 96.2 ± 0.3 6.77 ± 0.22 2.64% �0.19 ± 0.26 97.6 ± 0.4172.62 23X-2, 32–34 cm ul pel mud-wack 50 1 34.70 7 0.02 ± 0.03 22.3 ± 0.2 9.37 ± 0.09 2.63% 0.05 ± 0.19 25.2 ± 0.3

60 2 34.70 6 0.01 ± 0.01 22.5 ± 0.2 9.37 ± 0.08 2.63% �0.01 ± 0.21 25.0 ± 0.3177.12 23X-5, 32–34 cm pl pel wack PL 50 1 43.20 9 �0.28 ± 0.02 21.1 ± 0.3 7.09 ± 0.25 2.82% �0.53 ± 0.12 24.1 ± 0.2

60 2 43.20 4 �0.26 ± 0.04 21.7 ± 0.3 7.09 ± 0.25 2.81% �0.28 ± 0.38 24.1 ± 0.5181.92 24X-2, 32–34 cm ul-pl pel bio mud wack 50 1 50.50 5 �0.15 ± 0.02 8.2 ± 0.3 6.13 ± 0.19 2.78% 0.00 ± 0.31 11.0 ± 0.4

60 2 50.50 4 �0.15 ± 0.04 12.3 ± 0.3 6.13 ± 0.19 2.78% �0.26 ± 0.38 14.9 ± 0.5191.22 25X-2, 32–34 cm pl bio mud-wack PL 50 1 56.66 9 �0.08 ± 0.04 8.2 ± 0.2 5.23 ± 0.04 2.90% �0.50 ± 0.22 11.6 ± 0.3197.22 25X-6, 32–34 cm pl bio wack PL 50 1 73.20 9 �0.13 ± 0.02 9.0 ± 0.2 4.74 ± 0.10 2.48% �0.43 ± 0.41 12.9 ± 0.6201.82* 26X-3, 32–34 cm pl bio wack PL 50 1 74.20 9 �0.07 ± 0.02 10.1 ± 1.0 4.27 ± 0.09 2.72% �0.37 ± 0.16 13.1 ± 0.9207.92 27X-1, 32–34 cm pl bio wack PL 50 1 53.79 7 �0.10 ± 0.03 9.3 ± 0.2 5.78 ± 0.04 2.75% �0.29 ± 0.19 12.7 ± 0.3212.42* 27X-4, 32–34 cm pl bio mud-wack PL 50 1 65.90 9 �0.06 ± 0.02 12.7 ± 1.0 5.01 ± 0.12 2.61% �0.19 ± 0.16 15.4 ± 0.9216.92* 27X-7, 32–34 cm pl bio mud PL 50 1 97.60 8 �0.12 ± 0.03 15.0 ± 1.6 3.25 ± 0.05 2.71% �0.24 ± 0.13 17.7 ± 1.6219.26 28X-2, 106–108 cm pl bio wack PL 50 1 52.41 6 �0.26 ± 0.03 8.1 ± 0.3 5.83 ± 0.05 2.79% �0.63 ± 0.18 11.7 ± 0.4

60 2 52.41 4 �0.24 ± 0.03 8.2 ± 0.3 5.83 ± 0.05 2.79% �0.26 ± 0.12 10.9 ± 0.3222.26 28X-4, 106–108 cm pl bio mud-wack PL 50 1 70.80 9 0.00 ± 0.02 7.3 ± 0.3 4.59 ± 0.10 2.66% �0.25 ± 0.12 10.7 ± 0.2

All measurements were made using Aridus II + Jet Cones. Tailing from 238U on 236U, 235U and 234U was estimated as, respectively, 0.6 ppm, 0.25 ppm and 0.1 ppm. Hydride formation wascorrected using the value 238UH/238U = 7.3e�7 (see Tissot and Dauphas, 2015 for more details).* Cup configuration for 233/234/235/236/238U is L1/C/H1/H2/H3 (faraday cups with 1011 X resistor). For other samples, the configuration is L1/SEM/H1/H2/H3.a Shipboard lithology (from Malone, 2000) of samples at equivalent depth (within 5 cm, except for 219.26 and 222.26, for which the reference sample is 70 cm lower). ul = unlithified,

pl = partially lithified, pel = peloidal, bio = bioclastic, wack = wackestone, pack = packstone, mud = mudstone, forams = foraminifers, nannos = nannofossils.b PL = partially lithified, LITH = lithified.c Number of cycles of 4.194 s integration time each.d Successive number of times the sample was purified by column chemistry on U/Teva resin.e Values normalized to CRM-112a.f Activity ratio (234U/238U) of the sample relative to secular equilibrium. d(234U) = {(234U/238U)smp/(234U/238U)eq � 1} * 1000 where (234U/238U)eq is the atomic ratio at secular equilibrium

and is equal to k238U/k234U = 5.4970e�5 (Cheng et al., 2013).

248F.L.H

.Tisso

tet

al./Geochim

icaet

Cosm

ochim

icaActa

242(2018)

233–265

Table 3Comparison of U concentration, d(234U) values and ages of samples from Site 1009.

This study Literature values

Depth [U] d(234U) ‘‘Age” Depth 14C ages(mbsf) (ppm) (‰)a (kyr)b (mbsf) (kyr)

0.07 1.010 ± 0.5609.62 8.86 ± 0.07 143.95 ± 0.23 �2 12 5.350 ± 0.10014.62 6.83 ± 0.05 143.23 ± 0.27 �419.12 5.00 ± 0.03 125.47 ± 2.22 �51 21.86 10.000 ± 0.360

24.34 42.700 ± 0.580

Depth [U] d(234U) Corrected age

(mbsf) (ppm) (‰)c (kyr)

35.12 9.71 ± 0.08 106.74 ± 0.27 �108 36.68* 6.15 ± 0.02 102.48 ± 1.47 117.31 ± 5.5138.61 3.66 ± 0.03 110.53 ± 0.24 �96 39.26* 6.07 ± 0.02 109.48 ± 4.34 64.13 ± 8.35

43.19* 8.14 ± 0.02 96.51 ± 0.93 133.21 ± 2.9044.12 8.12 ± 0.29 92.27 ± 1.30 �160 44.19* 8.73 ± 0.02 105.76 ± 2.70 162.45 ± 3.15

46.19* 7.99 ± 0.02 85.72 ± 4.34 182.23 ± 6.1648.62 12.15 ± 0.11 89.45 ± 0.25 �171 48.18* 12.10 ± 0.03 98.16 ± 2.54 150.28 ± 2.05

52.69* 5.09 ± 0.01 83.24 ± 2.83 200.98 ± 6.3253.62 6.19 ± 0.05 89.43 ± 0.27 �171 �55** 5.039 ± 0.010 81.6 ± 7.0 192 ± 7

55.68* 5.17 ± 0.01 81.81 ± 0.88 191.55 ± 2.0656.35* 9.42 ± 0.02 88.89 ± 2.68 169.61 ± 3.1256.35* 5.63 ± 0.01 80.71 ± 0.90 197.13 ± 2.0756.68* 5.45 ± 0.01 80.36 ± 1.21 196.12 ± 2.28

58.12 4.81 ± 0.03 90.40 ± 0.22 �167 57.68* 5.01 ± 0.01 86.58 ± 4.85 190.69 ± 6.0761.62 8.81 ± 0.07 86.99 ± 0.25 �181 60.20* 4.45 ± 0.01 82.18 ± 2.68 199.36 ± 14.62

63.18* 5.66 ± 0.02 82.36 ± 8.81 208.35 ± 10.2064.62 6.19 ± 0.05 88.43 ± 0.19 �175 65.20* 3.38 ± 0.01 83.82 ± 2.97 196.23 ± 11.69

66.20* 6.71 ± 0.02 67.11 ± 0.88 234.95 ± 2.6466.69* 6.29 ± 0.02 72.92 ± 4.38 232.11 ± 12.7366.69* 6.51 ± 0.02 71.39 ± 0.97 231.69 ± 2.85

69.12 7.11 ± 0.22 82.62 ± 0.28 �199 �68** 6.750 ± 0.009 79.8 ± 4.8 247 ± 1370.69* 6.49 ± 0.02 70.21 ± 1.06 235.90 ± 3.29

74.12 9.48 ± 0.08 61.31 ± 0.26 �305 �75** 8.029 ± 0.007 58.8 ± 3.5 336 ± 18

* Robinson et al. (2002).** Henderson (2002).a Activity ratio (234U/238U) of the sample relative to secular equilibrium. d(234U) = {(234U/238U)smp/(234U/238U)eq � 1} * 1000 where

(234U/238U)eq is the atomic ratio at secular equilibrium and is equal to k 238U/k234U = 5.4970e�5 (Cheng et al., 2013).b Assumes the sample formed with the modern seawater d(234U) value of 144.9 ± 0.4‰, and that the difference in 234U/238U is exclusively

due to 234U decay.c As reported in original study, using half-lives from Cheng et al. (2000).

Fig. 11. Bulk carbon and oxygen isotope compositions, total organic carbon (TOC) content, and organic carbon and nitrogen isotopecomposition depth profiles for Site 1009. Symbols as in Fig. 10. Shipboard and previous data (Eberli et al., 1997; Malone, 2000) are shown ingrey.


Table 4Bulk TOC content, d13C and d18O values, and organic d13C andd15N values of Site 1009 samples.

Bulk Organic fraction

Sample d13C d18O d13C d15NV-PDB V-PDB TOC V-PDB AIR

(mbsf) (‰) (‰) (wt%) (‰) (‰)

0.32 4.13 �0.48 0.27 �13.26 1.265.12 3.92 �1.65 0.04 �14.27 1.199.62 4.76 �0.50 0.02 �14.12 1.0714.62 4.98 0.14 0.27 �15.11 0.8719.12 3.60 �0.01 0.09 �17.00 1.0135.12 4.61 0.35 0.45 �16.55 2.4138.61 3.41 1.68 0.11 �18.59 2.7044.12 4.10 0.59 0.65 �10.13 1.5548.62 4.51 0.33 0.75 �13.29 0.4953.62 4.79 0.34 0.16 �15.65 0.8658.12 4.77 0.59 0.21 �15.50 1.1261.62 4.66 0.40 0.24 �14.58 0.8764.62 4.08 �0.04 0.22 �17.17 1.9769.12 4.83 0.50 0.12 �15.73 1.0671.12 3.69 �0.77 0.20 �17.33 2.0574.12 3.97 �0.64 0.44 �16.17 1.4178.62 4.52 0.05 0.25 �15.42 1.0789.62 4.37 0.08 0.53 �11.81 0.8994.12 4.67 �0.30 0.25 �14.28 0.8997.62 4.55 0.13 0.30 �15.56 2.2599.12 2.10 1.76 0.05 �20.46 2.92103.62 3.79 �0.30 0.08 �17.09 2.03108.62 4.65 0.07 0.32 �13.35 1.18113.59 4.82 0.73 0.24 �14.29 1.94115.62 4.28 0.57 0.05 �15.68 2.03120.12 4.89 0.34 0.20 �16.83 1.89125.32 3.76 0.46 0.18 �16.62 2.83133.22 3.56 �0.07 0.37 �12.29 2.68137.72 3.51 0.51 0.41 �13.68 2.42142.82 4.68 0.79 0.22 �15.06 2.57145.81 4.06 �0.63 0.26 �14.89 2.00161.82 3.09 0.49 0.81 �10.69 1.33166.32 4.55 0.72 0.60 �11.21 1.14172.62 4.69 0.19 0.10 �13.67 1.50177.12 4.63 1.01 0.11 �17.24 2.71181.92 4.75 0.77 0.25 �14.85 2.48191.22 4.92 0.58 0.25 �14.15 2.14197.22 4.92 1.07 0.22 �14.52 2.28201.82 4.75 0.66 0.25 �15.12 2.02207.92 4.80 1.03 0.21 �15.51 2.46212.42 4.91 0.58 0.20 �15.12 2.27216.92 4.93 0.02 0.23 �14.99 2.54219.26 4.55 1.26 0.13 �16.20 2.41222.26 4.97 0.33 0.28 �16.41 1.90

Fig. 12. Carbon isotope composition of the organic fraction vs.inorganic fraction for Site 1009 samples. The orange box representsthe carbonate platform component (d13Cinorg between +3‰ and+6‰ and d13Corg between �7‰ to �16‰) and the blue boxrepresents the pelagic component (d13Cinorg between 0‰ and +1‰and d13Corg between �20‰ to �22‰) (Oehlert et al., 2012). (Forinterpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)


d238U value observed in a COC-rich sub-sample of the samecoral (Tomiak et al., 2016). Therefore, the low [U] and highd238U values observed in the first dissolution step of mod-ern corals in the present work could be a signature of thecenters of calcification with higher surface area to volumeratios, lower U/Ca ratios, and higher d238U values, dissolv-ing faster than other skeletal components. On anotherhand, the systematically lower d(234U) values observed inthe first digestion cut are inconsistent with the typicallyhigher d(234U) values observed in COCs of scleractinian

corals (Robinson et al., 2006; Tomiak et al., 2016). Thesehigher d(234U) values have been interpreted as the resultof 234U addition via diffusion along U concentration gradi-ents (Robinson et al., 2006) or during fluid alteration(Tomiak et al., 2016). The low d(234U) values observed inthe initial leachates in the present work might therefore rep-resent the original d234U value of the COCs, prior to 234Uenrichment. Alternatively, it could reflect the fact thatCOCs are not the only parts of the corals digested in the ini-tial leachates, and domains characterized by lower d(234U)values are also digested in those steps. Altered zones whereloss of U with preferential mobilization of 234U (which isnot lattice bound) occurred during fluid alteration couldrepresent such domains.

5.2. Older corals, and limestone, dolostone

The Pleistocene and Pliocene corals show lower thanmodern seawater d(234U) values and limited isotopic vari-ability among the three cuts, within 10‰ of each other.Taken at face value, these 234U/238U ratios would indicatethat the two samples are only �100 and 70 kyr old, respec-tively. The former number is consistent with the Pleistocene(10–1600 kyr) age of the first coral, but the latter estimaterequires recent resetting of U systematics of the Pliocene(2.5 –5.3 Myr) specimen. It did not affect, however, the238U/235U ratio, which has the same dissolution pattern asmodern corals and d238U values indistinguishable frommodern seawater. Together, the coral data shows that cor-als do record the U isotope composition of seawater, andare consistent with fully oxygenated oceans over the lastcouple million years.

The data for the limestone and dolostone shows d238Uvalues that are identical in all three steps, but �0.20‰and �0.45‰ higher than modern seawater, respectively.The limestone further displays variable d(234U) values, from

Fig. 13. Scatter plot matrix of U/Ca, d238U, d13Cinorg, d18Oinorg, TOC, d15Norg (this study), and bulk geochemical data (Sr/Ca, U/Sr, Mg/Ca,

and Na/Ca; Malone, 2000) for Site 1009. Spearman’s rank values (q, p-value) are listed in each panel. Panels in full colors show statisticallysignificant correlations (i.e., p-value < 0.05), whereas a transparency denotes non-significant correlation. The hardground sample (red) wasplotted, but not used for calculation of p-values or correlation coefficients. White = unlithified, grey = partially lithified, blue = sealevellowstands, red = hardground. (For interpretation of the references to color in this figure legend, the reader is referred to the web version ofthis article.)


110‰ up to 141‰ relative to secular equilibrium, whereasthe dolostone is essentially at secular equilibrium (valuesbetween �13 and +11‰). The bulk 234U/238U ratio of thelimestone is broadly consistent with the poorly definedage of the sample (�120 kyr), yet the high values observedin the second digestion step imply some recent 234U addi-tion, presumably from seawater or pore-waters (seeFig. 4). On the other hand, the Pliocene dolostone234U/238U ratio is consistent with a closed-system behaviorof the sample, suggesting that the offset of its 238U/235Uratio relative to seawater could be a feature acquired atthe time of dolomitization, or soon after. This is consistentwith Sr and O isotope and iodine concentration data forPrecambrian dolostones indicating that once dolostones

are formed, they act as a closed system (e.g., Hardistyet al., 2017).

Overall, the coral data is consistent with corals andother primary carbonate allochems (i.e., algae, mollusks,and ooids) recording a U isotope composition close to thatof the seawater from which they formed, as previouslyreported in several studies (Stirling et al., 2007; Weyeret al., 2008; Romaniello et al., 2013; Andersen et al.,2014; Zhang et al., 2018b), including data for stony corals(Tissot and Dauphas, 2015), and during calcite and arago-nite precipitation experiments at pH � 7.5 (Chen et al.,2016b). By comparing d238U values in coeval carbonatesequences from different depositional environments,Clarkson et al. (2018) recently showed that some calcite


successions deposited in deeper-shelf environments are infact capable of preserving this primary d238USW value ongeological timescales. The d238U values 0.20–0.45‰ higherthan seawater observed in the limestone and dolostone,however, add to a growing body of evidence that sedimen-tary carbonates do not always preserve the unfractionatedseawater 238U/235U ratio and that shifts to higher d238U val-ues are imparted onto these rocks either during their forma-tion (Chen et al., 2016b) or by secondary processes such asdiagenesis, burial, and dolomitization (e.g., Romanielloet al., 2013; Hood et al., 2016; Clarkson et al., 2018;Hood et al., 2018; this work, see Section 5.4).

5.3. 234U/238U profile from Site 1009

5.3.1. U closure time

The abundance of 234U in a sample is a function of thedecay constants of 234U and 238U (k234 and k238) and thetime since closure of the sample. At steady-state (i.e., in sec-ular equilibrium), the decay of 234U is balanced by the cre-ation of 234U from 238U decay, and the 234U/238U ratio inthe sample is equal to the ratio of the decay constants

k238=k234¼ 1:5513�10�10� �

= 2:8220�10�6� �¼5:4970�10�5

(Cheng et al., 2013). The d(234U) values in Site 1009 samplesfollow a decreasing trend with depth (Fig. 10c), fromseawater-like (�145‰) close to the water-sediment inter-face to values close to secular equilibrium (�10‰) at thebottom of the core. Assuming that the samples formed withthe modern seawater 234U/238U ratio, and that this valuewas constant over the past 1.5 Myr, knowledge of the ageof the samples is sufficient to calculate the expectedd(234U)p (where p stands for present) value for differenttime elapsed since closure of the 234U reservoir, as:

dð234UÞp ¼ dð234UÞSWe�k234 t ð5Þ

The composite age vs. depth profile built from radiomet-ric ages and bioevent datums (blue curve in Fig. 2) is usedto calculate the age of samples at any depth. The green,blue, and red curves in Fig. 10c and f show three d(234U)pvs. depth profiles calculated assuming that the samplesclosed, respectively, 0, 5, and 10 m below the water-sediment interface. To first order, the measured d(234U) val-ues in the drill core are in very good agreement with closureof the samples close to the water-sediment interface, andreturn to secular equilibrium in a closed system. A closerview of the top of the drill core (Fig. 10f) shows a betteragreement between the d(234U) values measured and closureof the samples 5 mbsf (blue fit) rather than 0 mbsf or 10mbsf (green and red fits). Open-system behavior down to5 mbsf implies (i) that the data of Romaniello et al.(2013), which extends to 40 cm depth only, does not recordthe entire carbonate platform diagenesis, and (ii) that anoffset will exist between 14C and U-Th ages in the Site1009 samples. Indeed, if the radiocarbon clock starts atthe death of the carbonate-secreting organism, it will effec-tively start before the U-Th clock. Assuming that the 14Cages in the Site 1009 samples are not significantly affectedby 14C marine reservoir effects (e.g., Bard et al., 1994) orother secondary processes (e.g., modern atmospheric CO2

contamination, bioturbations; Nadeau et al., 2001;Sepulcre et al., 2017), the actual age offset between 14Cand U-Th ages will depend on the U-Th closure depth ofthe sample and the sedimentation rate at the time of sampledeposition. For the top portion of the core, where 14C agesare available (Slowey et al., 2002) and can be used to calcu-late sedimentation rates (�250 cm/kyr; Fig. 2, Table S1), a5 m difference in closure time would result in 14C ages being�2–3 kyr older than U-Th ages.

Samples for which the measured and expected d(234U)disagree (i.e., samples that do not fall on the red curve inFig. 10c) are found mostly below �75 mbsf and are system-atically enriched in 234U (up to 100‰), indicating open-system behavior and U exchange with a reservoir character-ized by a high 234U/238U ratio, most likely the circulatingpore-water. Based on the 234U/238U pore-water profile(Fig. 4; data from Henderson et al., 1999) an upper limiton the amount of 234U addition to each sample can be cal-culated. Samples below 75 m are consistent with at most 7%of 234U addition, and even the 234U/238U ratio of the hard-ground sample (99.12 mbsf), which is 70‰ higher than theexpected value, can be explained with only 20% of 234Uaddition from pore-waters. Diagenetic processes involvingU exchange between the bulk samples and a fluid musttherefore have occurred mainly before closure of the systemat �5 mbsf, which based on the composite age vs. depthprofile (Fig. 2) corresponds to a duration of open-systembehavior of only �2–3 kyr.

The observation of closed-system behavior is based onU isotope analysis of the second leachate (30–80% carbon-ate digestion). Exchange with the circulating fluid mighttherefore continue below the �5 mbsf point, but only affectthe surface layer of the carbonate grains, which areremoved by the first leaching step (which digested about35% of the carbonate). This is broadly consistent with therecent findings of Higgins et al. (2018), who, based of Mgand Ca isotope data for Sites 1003, 1005, and 1007 (seeFig. 1), concluded that fluid-buffered, open-system condi-tions were most likely restricted to shallow depths (thetop 10s of meters of the sediment column) due to porosityand permeability limitations for driving fluid advection dee-ply into sediments. The shallower closure depth derivedfrom our data (�5 mbsf) relative to that of Higgins et al.(2018) (�10s of mbsf) likely reflect a solubility differencebetween U and Mg (or Ca) in the pore-water down thecore. Indeed, pore-water U concentrations decrease rapidlyas a result of U reduction to insoluble U4+, limiting the pos-sibility of significant sample U resetting below �50 mbsf,whereas Mg2+ and Ca2+ remain elevated even at the bot-tom of the drill core (Fig. 4; Henderson et al., 1999). Itcould also be due, in part, to a difference in sample process-ing, and specifically the use of a step-leaching protocolbefore U isotope analysis. It is possible that the leachingprotocol applied here has removed most of the fluid-buffered portion of the sediments, with overprinted Ca,Mg, and U isotope compositions, in the first digestion step.In this scenario, the excess 234U seen in the Step 2 leachateof some of the samples below �75 mbsf could be due toinsufficient leaching in the first step, as a result of a highersurface/volume ratio in these lithified samples.


5.3.2. Seawater (234U/238U) activity ratio during the last 1.4

My

Since the d(234U)p values are calculated based on theinferred ages of the samples and the assumption of a con-stant 234U excess in seawater, the remarkable agreementbetween the expected and most measured d(234U) values(Fig. 10c) implies that the inferred ages of the samplesand assumption of constant seawater 234U/238U ratio arecorrect to the first order. To further quantify the constancyof the seawater 234U/238U ratio, we use the composite agevs. depth profile and the measured d(234U) values to calcu-late the 234U/238U ratio at the time of carbonate precipita-tion, d(234U)initial (or ‘‘age-corrected 234U/238U ratios”)(Fig. 14 and Table S9). We find that the d(234U)initial is rel-atively constant and within 15–20‰ (dashed lines inFig. 14) of the modern seawater value (blue line inFig. 14), in agreement with earlier studies, which agreedon a seawater 234U/238U ratio within 15‰ of the modernvalue for the last 360 kyr (Henderson, 2002; Robinsonet al., 2004; Chutcharavan et al., 2018). Our results confirmthis broad constancy, within somewhat larger bounds (±20‰), and expand this conclusion to the last 1 Myr.

Beyond 1 Myr, errors in the age and/or the amount of234U contamination of the samples will lead to increasinglylarger shifts of the d(234U)initial values, making reconstruc-tion of the seawater d(234U) more difficult. For instance,for a 1 Myr old sample, a 1‰ 234U addition is equivalentto an error of 39 kyr on the sample age or a 17‰ shift inthe d(234U)initial. For a 1.5 Myr old sample, the same 1‰excess is equivalent to an error of 138 kyr in the sample

Fig. 14. Initial d(234U) values vs. age for Site 1009 samples. Ages are calcu(blue curve on Fig. 2). Symbols as in Fig. 10. The blue horizontal line reappears to have remained within �20‰ of the modern seawater value dseawater d(234U) mirror sea-level changes during the last two glacial-interlast 1.05 Myr as well as upward-coarsening sequences. Error bars incluuncertainties, with the latter accounting for more than 70% of the total errthis figure legend, the reader is referred to the web version of this article

age or a 69‰ shift in the d(234U)initial. For samples between1 and 1.2 Myr old (i.e., between �142 and 177 mbsf), thed(234U)initial calculated from the drill core samples are sig-nificantly offset from the modern seawater value with valuesas high as 2700‰. This is clearly due to 234U addition andopen-system behavior of the samples (see Fig. 10c). Belowthe 1.2 Myr old level (i.e., below 177 mbsf), smaller-amplitude departures from modern seawater are observedwith most d(234U)initial values being between 273 and365‰. Such effects can readily be explained by recent andvery limited open-system behavior of the samples resultingin an increase of only 4–6‰ of the present d(234U) value.Alternatively, this could indicate that the age of the samples(based on nannofossil biostratigraphic datums; Fig. 2) areoverestimated by �200–300 kyr or a higher than modernd(234U)SW value. Although it is possible that the constancy(within 15–20‰) of the seawater d(234U) observed in the 0to 1 Myr range extends all the way to 1.4 Myr, more workwill be needed to check if this is the case.

The most striking feature of the d(234U)initial record is,however, not its overall constancy (within 20‰) throughoutthe last 1 to 1.4 Myr, but the fact that there is some struc-ture to the small-scale variability seen in the record. Belowwe discuss possible causes of these 1–20‰ variations. Obvi-ously, part of the dispersion (i.e., noise) in the data must bedue to alteration by circulating subsurface fluids. However,these processes would result in random variations in thed(234U)initial over thick stratigraphic sections. Beside open-system behavior, it has been proposed that both long-term (glacial to interglacial) and short-term (on the order

lated based on depth in the drill core and the composite age profilepresents the modern seawater d(234U) value. The seawater d(234U)uring the last �1 Myr, however, small-scale (1–15‰) variations inglacial periods (see Fig. 15). The inset shows a close-up view for thede both the U isotope measurement and the age vs. depth profileor beyond 0.25 Myr. (For interpretation of the references to color in.)


of �10 kyr) sea-level variations could influence thed(234U)initial of corals and platform carbonates (Esat andYokoyama, 2006; Esat and Yokoyama, 2010). To assessthe effect of sea-level variations on the d(234U) of seawater,the d(234U)initial values and sea-level changes over the past0.6 Myr are plotted on Fig. 15 (top panel). Despite somescatter, likely due to subsurface alteration, there is a goodagreement between the variations in d(234U)initial andsea-level. In particular, there is a clear rising trend ind(234U)initial values, from 135 to 160‰, during the penulti-mate glacial and interglacial event (�140 to �200 ka),which strongly resembles the sea-level fluctuations. Thistrend is remarkably similar to the one observed duringthe last glacial and interglacial events (�30 to 140 kyr)based on corals data (e.g., Cutler et al., 2003; Robinsonet al., 2004; Esat and Yokoyama, 2006; Thompson et al.,2011) and our dataset (despite its scarcity) and which hasbeen interpreted as being the result of sea-level variations(Esat and Yokoyama, 2006).

The actual mechanism by which sea-level and the seawa-ter d(234U) are linked is highly debated, and proposedhypotheses fall in two broad categories: (i) storage of excess234U in near shore suboxic and anoxic sediments during gla-cial periods and its subsequent release with rising sea-level

Fig. 15. d(234U)initial (top) and d238U (bottom) values (circles) and sea-levand Lisiecki (2016). Following Esat and Yokoyama (2006), we placd(234U)initial = 153‰. Error bars on ages and d(234U)initial values includuncertainties, with the latter accounting for more than 70% of the total

during the next interglacial (Esat and Yokoyama, 2006;Esat and Yokoyama, 2010), and (ii) changes in weatheringregime resulting in an increase in the U flux and d(234U) ofhigh-latitude continental inputs (e.g., Robinson et al., 2004;Andersen et al., 2007; Andersen et al., 2013 ; Chen et al.,2016a; Chutcharavan et al., 2018; Arendt et al., 2018).

These models make predictions regarding the relativetiming of sea-level changes and seawater d(234U) variations,which can be tested with datasets of high temporal resolu-tion (e.g., Chen et al., 2016a). Although our dataset is toosparse to inform on which group of models is correct, itexpands the record of lower d(234U)SW in periods of lowsea-levels to at least the last two glacial-interglacial events(i.e., �0.23 Ma). In fact, despite the large uncertainties onthe age vs. depth profile (Fig. 2), the correspondencebetween fluctuations in sea-level and seawater d(234U) val-ues can be argued as far as �0.6 Ma. Deviation of the234U/238U from the modern seawater value has traditionallybeen used as a test of validity for U-Th ages for recent car-bonate sediments (e.g., Bard et al., 1991; Hamelin et al.,1991; Gallup et al., 1994; Stirling et al., 1995; Robinsonet al., 2002). With the discovery of 1–20‰ variations inthe d(234U)SW, the validity of this approach is being debated(e.g., Cutler et al., 2003; Robinson et al., 2004; Esat and

el (grey envelope) vs. age at Site 1009. Sea-level data is from Spratted the last interglacial high (3–5 m above the present level) ate both the U isotope measurement and the age vs. depth profileerror beyond 0.25 Myr.

Fig. 16. Fraction of U added to carbonates during pore-watercirculation and U reduction (freduced), calculated from [U] (left) andd238U (right), vs. depth at Site 1009. About 50% of the U incarbonates is non-primary. Symbols as in Fig. 10.


Yokoyama, 2006; Thompson et al., 2011). While samplerejection assuming a constant, modern d(234U)SW valuemight be viable with large enough screening bounds (e.g.,±25‰), a more appropriate approach could consist inscreening samples using a robust d(234U)SW vs. time record(e.g., Chutcharavan et al., 2018), with smaller screeningbounds (perhaps ± �5‰). A high-resolution study of thedrill core investigated here would provide a unique oppor-tunity to refine this record and expand it at least back to 1Myr ago (or MIS 29).

Finally, some of the small-scale variability in thed(234U)initial record could be due to the fact that the closuredepth of the samples did not remain exactly the samethrough time. Several factors could influence the closuredepth, including sedimentation rate and sea-level stand.In particular, development of hardground would result inthe creation of a barrier to fluid circulation that would tem-porarily move the closure depth closer to the water–sedi-ment interface until sedimentation rate increases.However, assessing the effect of such process on the Urecord is difficult because of the complex interconnectionbetween variations in sea-level, sedimentation rate, and clo-sure depth.

5.4. Site 1009 [U] and d238U profiles

5.4.1. Authigenic U enrichment

In the drill core, neither [U] nor d238U show obvioustrend with depth (Fig. 10). U concentrations show signifi-cant dispersion from 1.9 to 12.1 ppm (Fig. 10a) with highervalues in the top 80 to 100 m of the drill core (where sam-ples are mostly unlithified) than in the bottom portion(where samples are mostly lithified) (Fig. 10d). All samplesshow [U] values higher than in modern allochemical frag-ments (grey band, Fig. 10a) indicative of U addition.Almost all samples have d238U values close to or higherthan that of modern seawater, by up to 0.50‰, and definean average d238USite 1009 value of -0.16 ± 0.06‰ (95% CI),in agreement with data from shallow (0–40 cm) Bahamascarbonates (Romaniello et al., 2013) in which similar shiftshave also been observed.

Concentrations higher than those of primary carbonatesand d238U values higher than that of seawater are consistentwith U authigenic enrichment during pore-water circulationandU reduction below the depth of Fe reduction (e.g., Zhenget al., 2002; McManus et al., 2005; Morford et al., 2005;Romaniello et al., 2013; Andersen et al., 2014). Authigenicenrichment of U in the carbonates is, in fact, reflected inthe clear positive correlation between U/Ca and U/Sr ratiosof the solid samples despite variations in the Sr/Ca ratio ofthe samples (Fig. 13). A lower limit on the fraction of Uadded to the samples (freduced) during pore-water circulationand U reduction can be obtained by comparing the concen-tration of the samples ([U]measured) to that of a modern arag-onitic carbonate ([U]initial �3 ppm, Dunk et al., 2002) as:

freduced ¼ ½U�added½U�measured

¼ ½U�measured � ½U�initial½U�measured

ð6Þ

A second estimate for freduced can be obtained by com-paring the 238U/235U ratio of the samples (d238Umeasured)

to the initial isotopic composition of the samples (assumedto be identical to that of seawater, d238USW = �0.39± 0.02‰; Tissot and Dauphas, 2015; Noordmann et al.,2015; Holmden et al., 2015; Rolison et al., 2017), andassuming a fractionation factor during authigenic U addi-tion typical of U reduction in marine environments(DAnoxic-SW = d238Uadded � d238USW = +0.6‰; Andersenet al., 2014; Murphy et al., 2014; Noordmann et al., 2015;Holmden et al., 2015; Rolison et al., 2017; Bura-Nakicet al., 2018). The mass balance equation, d238Umeasured =freduced d

238Uadded + (1- freduced) d238USW, can be rearranged

as:

freduced ¼ d238Umeasured � d238USW

DAnoxic�SW

ð7Þ

Both equations provide only a lower limit on theamount of authigenic enrichment as they assume that thechange in U content and isotopic composition is solelydue to U reduction and incorporation into the sedimentsand neglect any U loss and/or decrease of the d238U valueduring sample dissolution and recrystallization. The valuesof freduced calculated using Eqs. (6) and (7) vary mostlybetween, respectively, 30 and 70% and 20 and 60%(Fig. 16), indicating that about half of the sample U isnon-primary and was added to the sample during early dia-genesis. Given that sample dissolution and recrystallizationdecrease U content and, possibly, the d238U value (seebelow) of the samples, the actual amount of authigenic Uin the samples is likely to be closer to the highest valuesof freduced calculated, �66 –78%. Some outliers with low(and even negative) freduced values are observed amongstthe lowstand samples, indicating the influence of sea-levelstand on the d238U record of carbonates (see the followingsection).

Despite clear authigenic enrichment in the samples,there is no correlation between TOC content and d238U val-ues (Fig. 13). Recently, Lau et al. (2016) used the low TOCcontent and the lack of correlation between d238U values


and TOC contents in the end-Permian carbonate samples toargue that the d238U values of the carbonates reflected theprimary seawater composition. In contrast, we observe astrong offset between 238U/235U ratios in carbonate samplesand seawater (�0.25‰). The argument put forward by Lauet al. (2016) does not take into account the fact that lowTOC content does not necessarily mean that there was noorganic matter in the samples to begin with, but could sim-ply indicate that significant oxidation of organic matter hasoccurred. We therefore argue that low TOC content and alack of correlation between TOC and d238U values in car-bonate samples cannot be used to infer how close the U iso-tope composition of carbonates and the coeval seawaterwere.

Fig. 17. d238U vs. sedimentation rate at Site 1009. Symbols as inFig. 10. Note positive correlation between d238U and sedimentationrate.

5.4.2. Sea-level control on the d238UUranium authigenic enrichment is not, however, the

only process influencing the d238U value of the samples,as shown by the statistically significant correlations betweend238U and several geochemical proxies insensitive tochanges in the redox potential of the pore-water: d13Cinorg,Sr/Ca, and d15Norg (Fig. 13). The Sr/Ca ratio or Sr contentprincipally reflects the proportion of aragonite present inthe sample. Change in the aragonite content can be dueto sea-level changes (with highstand sediments beingaragonite-rich and lowstand sediments being calcite-rich,Droxler and Schlager, 1985; Betzler et al., 1999), and disso-lution and recrystallization of aragonite into calcite (i.e.,lithification, Brand and Veizer, 1980; Gothmann et al.,2015). Similarly, on the Bahamas periplatform, long-termvariations in d13Cinorg values (on the scale of 0.5 My orlonger) reflect mixing of platform-derived and pelagic car-bonates that are linked to sea-level changes (Swart andEberli, 2005; Oehlert et al., 2012). On the other hand,shorter-term variations have been attributed to alterationand lithification of the samples (Malone et al., 2001).

To assess the role and relative importance of the pro-cesses affecting the 238U/235U ratio of carbonates, we usedthe available sea-level and geochemical data. There is a gen-erally good agreement between the variations in d238U val-ues and sea-level (Fig. 15, bottom panel), with sedimentsdeposited during periods of high sea-level (which are mainlyplatform-derived) having d238U values 0.50–0.60‰ higherthan samples deposited during periods of low sea-level(which are mainly pelagic). Distinct U isotopic composi-tions for platform-derived and pelagic carbonates are alsosupported by the correlations between d238U, d13Cinorg,d15Norg (Fig. 13), and d13Corg values (not shown) and Sr/Ca ratios. Indeed, Bahamas platform carbonates haved13Cinorg values between +3 and +6‰ (Oehlert et al.,2012), high Sr/Ca ratios (i.e., aragonite-rich), and near-zero d15Norg values (Swart et al., 2014), while pelagic car-bonates have d13Cinorg values between 0 and +1‰(Oehlert et al., 2012), low Sr/Ca ratios (i.e., calcite-rich),and d15Norg values around +3‰ (Swart et al., 2014). Over-all, samples with a large platform-derived component tendto have high U content and d238U values (�+0.10‰) com-pared to samples with a large pelagic component, whichhave d238U values of ��0.40‰ (i.e., close to the seawatervalue). This indicates that authigenic U enrichment in the

carbonate is more prevalent on the carbonate platform thanon the slope and explains the lower freduced values derivedfrom samples deposited in periods of low sea-level stand,when export of platform material to the slope is limited.

Because Na loss occurs during early diagenesis (Maloneet al., 2001), the lack of correlation between d238U and Na/Ca ratios (Fig. 13) suggests that sample dissolution andrecrystallization only minimally affected the carbonated238U values at Site 1009. On the other hand, the correla-tion between d238U and sedimentation rate (Fig. 17) furthersupports the idea that sea-level variations are the main con-trol of the d238U values of shallow-water carbonates as dur-ing periods of low sea-level the average sedimentation rateswill be lower, whereas during high sea-level sedimentationrates will be higher. This reflects a unique property of car-bonate depositional systems with respect to siliciclasticdepositional systems, where carbonate production andexport on a carbonate platform (carbonate factory) areenhanced during sea-level rise and shutdown during sea-level fall.

In fact, eustatic control on the 238U/235U ratios in plat-form carbonates might already have been documented,albeit unrecognized, in the rock record. In a recent studyof carbonate samples from the Guilin carbonate platform(Nanpanjiang Basin, China) at the Frasnian-Famennianboundary (�372 Ma), Song et al. (2017) reported negatived238U shifts of up to 0.40‰, which occurred concurrentlywith discrete and significant falls in sea-level (Chen andTucker, 2003; Song et al., 2017). In light of our findings,the interpretations of the U isotope signal solely in termsof extent of oceanic anoxia put forward by Song et al.(2017) might be invalid and sharp changes in d238U valuesmight instead be mainly responding to eustatic variations.

Comparison between sea-level and d238U is only possibleover the last 0.8 Myr, where the detailed sea-level recordextends (Spratt and Lisiecki, 2016). Assuming, however,that changes in sea-level were the main control for the Uisotope carbonate record over the last 1.5 Myr, the reducedd238U variability and lower [U] found in the partially lithi-fied samples in the lower portion of the drill core couldindicate more stable and generally low sea-level at the time

Fig. 18. U/Ca and U/Sr depth profiles at Site 1009. U concentra-tion from the ‘‘central cut” (Step 2, this work). Ca and Srconcentrations from bulk carbonate (Malone, 2000). Symbols as inFig. 10.


of sample deposition (i.e., between 1 and 1.4 Myr). Alterna-tively, the apparent d238U variability and lower [U] could bedue to more pronounced open-system behavior in this partof the drill core, consistent with the large 234U excess seen inthe samples between 140–180 mbsf. Though the effect ofalteration and/or recrystallization on the d238U appears tobe minor (see next section), in the absence of a detailedsea-level record, we cannot rule out the second hypothesis.

5.4.3. Effect of sample dissolution and recrystallization on the

d238UBased on the pore-water and sediment depth profiles

(Figs. 3, 4, and 10; data from Eberli et al., 1997;Henderson et al., 1999; Malone, 2000; and this study) twointervals can be identified in the drill core where authigenicenrichment, sample dissolution and recrystallization are atplay and could have affected the U systematics:

(1) In the 0–50 mbsf interval, degradation of U-richorganic matter is taking place as evidenced by thepore-water decreasing pH and sulfate concentration,increasing total alkalinity, and higher (3�) than sea-water U concentrations at constant Sr2+ concentra-tion. Authigenic enrichment of U in carbonates(Section 5.4.1) is consistent with the decreasingpore-water U concentrations (from 3 � to below1/3 of the seawater value) in this interval. At the sametime, grain dissolution is occurring (between 20 and55 mbsf), as shown by the lower than seawaterpore-water d(234U) values (Fig. 4; Henderson et al.,1999) and the slight increase in Sr2+ concentrationat constant Mg2+ concentration and Sr2+/Ca2+ ratio.Though dissolution should affect the bulk d(234U) ofthe sample by removing the outer layers which are234U-poor (due to alpha recoil), it had no impacton the d(234U) of the second leachate in our protocolas the outer layers of the grains were already digestedin the first leaching step.

(2) In the 50–225 mbsf interval, aragonite recrystalliza-tion into calcite (and dolomite below 150 m) andlithification become the dominant processes. This isshown by the sharp increase in pore-water Sr2+ con-centration and Sr2+/Ca2+ ratio and the decrease inMg2+ concentration. Authigenic U enrichment isnot happening at these depths; instead the samplesappear to be losing U. Compared to the 100–225 minterval (Fig. 10a), carbonates in the 50–100 m inter-val have higher U concentrations, U/Sr, and U/Caratios (Fig. 18), which do not correlate with Sr/Caratios (Fig. 13) and indicate U loss rather than Srgain at the deeper depth interval. Given that recrys-tallization is accompanied by Sr loss from the car-bonate, the lower U/Sr ratios in the 100–225 mdepth interval compared to the 50–100 m depth inter-val (Fig. 18) require that U loss was more pro-nounced than Sr loss. This is consistent with the�30 times lower partition coefficient of U in calcite(0.04–0.4; Kitano and Oomori, 1971; Meece andBenninger, 1993) compared to aragonite (1.7–9.8;Meece and Benninger, 1993; Gabitov et al., 2008),

and the only �10 times lower partition coefficientfor Sr in calcite compared to aragonite (0.1–0.35 incalcite; Gabitov and Watson, 2006, vs. 1.4 ± 0.53 inaragonite; Wassenburg et al., 2016). The largeexcesses of 234U recorded in some samples in the100–225 m depth interval (Fig. 14) might be the resultof U exchange with sediment pore-waters duringextensive recrystallization. Yet, other samples fromthis interval gave d(234U)initial identical to the modernseawater value, indicating for these samples thateither (1) no U exchange with sediment pore-watersoccurred during recrystallization, or (2) only the frac-tion of carbonate digested during the first step of ourleaching protocol had their 234U/238U ratio modified.

In light of these observations, the spread in U concentra-tion (by approximately a factor of 2) and d238U (of about0.2 to 0.5‰) for any given d13Cinorg, Sr/Ca, d

15Norg valuesor sedimentation rate, could reflect either (1) variableamount of secondary U addition (and associated isotopefractionation) via U reduction in sediments on the carbon-ate platform and in pelagic settings, or (2) U mobilizationand isotope fractionation during sample dissolution andrecrystallization. Evidence that the latter processes mightbe at play come from the shallow slope (� �0.09) of theSr/Ca vs. Mg/Ca data for the Site 1009, which is identicalto the slope of the diagenetic trend observed in altered por-tions of fossil corals (� �0.10) and different from the slopeof the primary trend resulting from substitution of Sr forMg in aragonite (slope � 1; Gothmann et al., 2015). Fur-thermore, U isotope fractionation during sample dissolu-tion and recrystallization provides a mean of explainingthe extremely low d238U value (�0.94 ± 0.03‰) of thehardground sample (99.12 mbsf). Indeed, if the composi-tion of this sample was the result of changes in the sea-level alone, it would require that the seawater d238U valuewas at least as low as that recorded in the sample. Thiswould imply a shift from modern oxygen levels to nearcomplete oceanic anoxia and back to modern levels on atimescale of �40 kyr. Such a scenario is highly improbable


and it is more likely that the low d238U value of the hard-ground sample is due to extensive and prolonged (due tothe slow sedimentation rate) diagenetic recrystallizationand neomorphism under fluid-buffered conditions(Higgins et al., 2018), prior to burial.

A summary of the controls on the U systematics in car-bonates from the Site 1009 is shown in Figs. 19 and 20.While primary carbonates and pelagic carbonates formingat the seafloor have 238U/235U ratios close to that of the sea-water (Romaniello et al., 2013; Clarkson et al., 2018 andthis study, Section 5.4.2), this value is rapidly overwritten

Fig. 20. Cartoon showing the main controls on the d238U of shallow-watexport of platform-derived sediments leads to high sedimentation rates another hand, periods of low sea-level (b) are characterized by lower sedimd238U values. In both cases, sample dissolution and recrystallization leacarbonates.

Fig. 19. Plot of 1/U vs. d238U at Site 1009 along with (a) compositionsediments, and (b) schematically shown impact of alteration and recrystacarbonate sediments. Vertical blue line indicates the seawater d238U valuethe reader is referred to the web version of this article.)

by authigenic U enrichment via pore-water circulationand U reduction. The process results in a significant(�0.50‰) increase of the d238U values and a factor 2–3enrichment in the carbonate U content (red arrows onFigs. 19 and 20). This diagenetic process is more prevalenton the carbonate platform than on the slope, where it cantake place as early as a few tens of centimeters below thewater-sediment interface (Romaniello et al., 2013), and con-tinues after sediment deposition down to �5 mbsf (thisstudy). During burial, grain dissolution (at 20–55 mbsf)and subsequent aragonite recrystallization (below 50 mbsf)

er carbonates at Site 1009. During periods of high sea-level (a), thed high [U] and d238U values in the peri-platform sediments. On theentation rates and [U] content in the sediments, and near seawaterd to U loss and, possibly, to a slight decrease in the d238U of the

of the expected platform (orange) and pelagic (blue) carbonatellization as well as addition of reduced U on the U systematics of. (For interpretation of the references to color in this figure legend,


lead to U loss and, possibly, a decrease in the d238U of thesamples. These last two processes also occur in the fluid-buffered region of the sediment column (i.e., top 10s ofmeters, Higgins et al., 2018), and will be more prevalentin times of low sedimentation rates. With the exception ofthe hardground sample (99.12 mbsf), all carbonates haved238U at or above the modern seawater value, indicatingthat the magnitude of the U isotopic fractionation associ-ated with sample dissolution and recrystallization is muchsmaller than the �0.60‰ value associated with U reductioninto anoxic sediments (e.g., Andersen et al., 2014; Tissotand Dauphas, 2015). Little data is available to put thisobservation into context, but calcite and aragonite precipi-tation experiments at pH �8.5 found a small, yet resolv-

able, isotopic fractionation of DAragonite-fluid = +0:07þ0:02�0:03‰

during aragonite precipitation, but not during calcite pre-cipitation (Chen et al., 2016b). If the recrystallization ofaragonite into calcite is an equilibrium process, it wouldresult in a lower d238U value and [U], consistent with thevery negative value observed in the hardground sample(99.12 mbsf). Due to the very minor dolomite content inthe studied samples (below 5 wt%; only present in 5 sam-ples), the effect of dolomitization could not be constrainedwith the data from the Site 1009.

5.4.4. Average effect of diagenesis on the d238U of carbonates

The combined and average effect of sea-level variationsand authigenic U enrichment, and, to a lesser extent, earlydiagenetic processes (i.e., sample dissolution and recrystal-lization) during sample burial, leaves the d238U value ofshallow-water carbonate sediments offset relative to thatof seawater by DCarbonates-SW = +0.24 ± 0.06‰ (95 CI,including all samples). Not including lowstand samplesgives an identical value of DCarbonates-SW = +0.28± 0.04‰. The former value, which captures the full vari-ability due to sea-level changes should be used in oceanicpaleoredox reconstructions based on carbonates depositedin shallow-water platform, shelf and slope environments(i.e., most of the carbonate sedimentary record prior tothe Mesozoic; Opdyke and Wilkinson, 1988; Boss andWilkinson, 1991; Holmden et al., 1998; Walker et al.,2002; Ridgwell, 2005) to account for the average effect ofcarbonate diagenesis. Assuming that the 238U/235U ratioof carbonate platform sediments directly records the238U/235U ratio of seawater would lead to underestimationof the extent of ocean-seafloor anoxia by at least a factor of10 (see models in Tissot and Dauphas, 2015; Lau et al.,2017).

Using the DCarbonates-SW value derived from the Site1009 samples implicitly assumes that the Bahamas plat-form carbonate is representative of all other platform car-bonates. As far as the Bahamas is the largest carbonateplatform dominated by aragonitic, non-skeletal material,this assumption is likely correct for other aragonite-dominated platforms. This interpretation is supported bythe 0.20–0.30‰ higher d238U values observed byClarkson et al. (2018) in shallow, aragonite-rich carbonatesuccession relative to the deeper-shelf pelagic calcitesuccession deposited during the Oceanic Anoxic Event 2(� 94 Ma). The Bahamas carbonate platform might, how-

ever, not be representative of all types of carbonate plat-forms, and future works should aim to measure Uconcentrations and isotope composition on othercarbonate platforms to assess the global relevance of theDCarbonates-SW proposed here.

5.5. Statistically meaningful datasets and seal-level

considerations for paleoredox reconstructions

As discussed above, the combined effect of sea-level vari-ations, authigenic U enrichment during pore-water circula-tion and U reduction, and diagenesis is not negligible andcan result in a 0.20–0.50‰ variability in the d238U valuesrecorded in recent carbonates. This variability is similarto, and sometimes larger than, the magnitude of U isotopevariations that have thus far been interpreted in terms ofoceanic redox (e.g., Dahl et al., 2014; Azmy et al., 2015;Dahl et al., 2017; Song et al., 2017; Bartlett et al., 2018;Clarkson et al., 2018). As such, our findings imply thatocean paleoredox reconstructions should rely on sufficientlylarge datasets and high-resolution sampling to allow foridentification of (i) statistically meaningful trends and (ii)variability due to sea-level changes. This is true even for lar-ger shifts in the rock record (on the order of 0.50‰), whichcould a priori be ascribed to variations in the redox state ofthe ocean.

To show the danger of directly interpreting d238U shiftsin term of extent of anoxia without considering the geo-logic context, four d238U depth profiles are presented inFig. 21. Each profile represents a subsampling of the Site1009 dataset, with variable sampling frequencies: 1 sampleevery 20 m (panels a, b, and c) or 1 sample every 40 m(panel d). Panels a, b, and c look different simply becausesubsampling started at different depths below the water-sediment interface: 5.12 mbsf (panel a), 0.32 mbsf (panelb), and 14.62 mbsf (panel c). Depending on which profileis considered, different conclusions can be reached regard-ing the redox state of the ocean: panel a would be inter-preted most readily as a shift to extreme extent ofanoxia (at least 20-fold) followed by a return to modernoxygenated oceans, panel b and c would be consistentwith a relatively (panel b) or very (panel c) redox-stableocean, while the panel d would seem to imply an oceanoxygenation event.

Such pitfalls can, to a large extent, be avoided by placingthe d238U record of carbonates in a broader geological con-text. In particular, attention must be brought to the identi-fication of sea-level controlled U isotope trends resultingfrom variable contributions of shallow platform and pelagicmaterials. The study of coeval carbonate successions depos-ited at different depths appears as a promising tool for suchassessments (e.g., Clarkson et al., 2018). When such com-parisons of depositional environments are not possible(e.g., available sediment successions were deposited in thesame environment), the potential influence of sea-levelchanges on carbonate d238U values has to rely on consider-ations of (i) the mineralogy of the carbonate sedimentsprior to lithification as inferred from mineralogical and pet-rographic observations, and (ii) diagenetic tracers (e.g.,d13Cinorg, d

18Oinorg, and Sr/Mn ratios).

Fig. 21. Effect of sub-sampling on the d238U vs. depth profiles at Site 1009. Each profile represents a resampling of the Site 1009 data-set(Fig. 10b): 1 sample every 20 m (panel a, b, and c) or 1 sample every 40 m (panel d). Depending on the profile considered, varying andincompatible conclusions could be reached regarding the redox state of the ocean. The blue vertical bar represents the modern seawater value.


6. CONCLUSIONS

A simple step-leaching protocol for bulk carbonate Uisotope analysis was developed (Section 3.3) and appliedto a series of modern carbonates (corals, limestone, dolo-stone) and 43 shallow-water carbonates from ODP Leg166 Site 1009. Though bulk corals do record the seawaterd238U value (within ±0.02‰), the step-leaching dissolutionshows variability in the concentration and isotopic Urelease pattern (Figs. 5 and 6), suggesting a control of thecoral structure with higher 238U/235U ratios in the centersof calcification.

The d(234U) data for the Site 1009 samples indicate thatU exchange with pore-water generally ceased within thefirst 5 mbsf (Fig. 10), equivalent to a closure time of �2–3 kyr after deposition. Based on the age vs. depth profileof the drill core (Fig. 2), the d(234U) values obtained hereinallow us to conclude that although the seawater (234U/238U)ratio remained broadly constant (within 15–20‰) over thelast 1–1.4 Myr (Fig. 14), small-scale variations (1–15‰)exist and mirror sea-level changes (Fig. 15). This expandsthe record of lower d(234U)SW in the periods with low sea-level stand previously observed for the last glacial-interglacial period to at least the last two glacial-interglacial events (i.e., �0.23 Ma). These permil-size varia-tions linked to sea-level changes should be considered whenscreening U-Th ages for carbonate sediments on the basisof the initial (234U/238U) ratios to avoid erroneous ageselection.

Correlation between d238U values and the geochemicaldata for the Site 1009 samples (e.g., d13Cinorg, Sr/Ca, andd15Norg, Fig. 13) reveal the dominant control of sea-levelstand on the d238U record (Figs. 13 and 15), through mixingof platform-derived carbonates (having d238U � 0.50–0.60‰ heavier than seawater) and pelagic carbonates (withseawater-like d238U values). The heavier U isotopecomposition and higher U content of the platform-

derived carbonates is attributed to U authigenic enrichmentvia U reduction on the platform. Sample dissolution andrecrystallization, which lead to U loss down the drillcore,appears to result in only minor modification of the carbon-ate d238U values.

The overall effect of sea-level variations and diagenesison the d238U of carbonates deposited on the shallow-water carbonate platform as well as in the shelf andslope environments results in a fractionation factor,DCarbonates-SW, of +0.24 ± 0.06‰ (95 CI, including allsamples) (Figs. 10, 19 and 20). This value should be usedwhen reconstructing the U isotope composition of seawa-ter from carbonates to avoid significant underestimationof the extent of oceanic anoxia (at least by a factor of10). Finally, the existence of relatively large (0.20–0.50‰) variations in the d238U of carbonates due tochanges in sea-level emphasizes the need for statisticallymeaningful data sets (Fig. 21) and implies that shifts inthe d238U carbonate record should not be ascribed tochanges in the extent of oceanic anoxia without properconsideration of the depositional environment with a par-ticular attention to sea-level control.

ACKNOWLEDGMENTS

FT thanks Anne M. Gothmann for valuable discussionsand comments on an earlier version of the manuscript. TheField Museum (Chicago) is thanked for providing severalcarbonate samples. Constructive criticisms from MortenB. Andersen, Chris Holmden, one anonymous reviewer,and editor Claudine Stirling greatly helped improve themanuscript. This work was supported by grants fromACS (52964-ND2), NSF (EAR1502591 andEAR1444951) and NASA (NNX17AE86G,NNX17AE87G, and NNX15AJ25G) to ND, a CrosbyPostdoctoral Fellowship to FT, and funding from NSERCDiscovery and Accelerator grant (RGPIN-316500) to AB.


AUTHORS CONTRIBUTIONS

FT and ND initiated the project. FT designed theresearch. FT, BMG, MN, GH, CC and PKS performedthe research. FT interpreted the data with inputs fromND, CC, AB and PKS. FT wrote the manuscript with con-tributions from all co-authors.

APPENDIX A. SUPPLEMENTARY MATERIAL

Supplementary data associated with this article can befound, in the online version, at https://doi.org/10.1016/j.gca.2018.08.022.

REFERENCES

Abe M., Suzuki T., Fujii Y., Hada M. and Hirao K. (2008) Anab initio molecular orbital study of the nuclear volume effects inuranium isotope fractionations. J. Chem. Phys. 129.

Anbar A. D., Duan Y., Lyons T. W., Arnold G. L., Kendall B.,Creaser R. A., Kaufman A. J., Gordon G. W., Scott C., GarvinJ. and Buick R. (2007) A whiff of oxygen before the GreatOxidation Event? Science 317, 1903–1906.

Andersen M. B., Romaniello S., Vance D., Little S. H., HerdmanR. and Lyons T. W. (2014) A modern framework for theinterpretation of U-238/U-235 in studies of ancient oceanredox. Earth Planet. Sci. Lett. 400, 184–194.

Andersen M. B., Stirling C. H., Porcelli D., Halliday A. N.,Andersson P. S. and Baskaran M. (2007) The tracing of riverineU in Arctic seawater with very precise U-234/U-238 measure-ments. Earth Planet. Sci. Lett. 259, 171–185.

Andersen M. B., Stirling C. H., Zimmermann B. and Halliday A.N. (2010) Precise determination of the open ocean U-234/U-238composition. Geochem. Geophys. Geosy., 11.

Andersen M. B., Vance D., Keech A. R., Rickli J. and Hudson G.(2013) Estimating U fluxes in a high-latitude, boreal post-glacial setting using U-series isotopes in soils and rivers. Chem.

Geol. 354, 22–32.Arendt C. A., Aciego S. M., Sims K. W. W., Das S. B., Sheik C.

and Stevenson E. I. (2018) Influence of glacial meltwater onglobal seawater delta U-234. Geochim. Cosmochim. Acta 225,102–115.

Asael D., Tissot F. L. H., Reinhard C. T., Rouxel O., Dauphas N.,Lyons T. W., Ponzevera E., Liorzou C. and Cheron S. (2013)Coupled molybdenum, iron and uranium stable isotopes asoceanic paleoredox proxies during the Paleoproterozoic ShungaEvent. Chem. Geol. 362, 193–210.

Azmy K., Kendall B., Brand U., Stouge S. and Gordon G. W.(2015) Redox conditions across the Cambrian-Ordovicianboundary: Elemental and isotopic signatures retained in theGSSP carbonates. Palaeogeogr. Palaeocl. 440, 440–454.

Bailey T. R., McArthur J. M., Prince H. and Thirlwall M. F. (2000)Dissolution methods for strontium isotope stratigraphy: wholerock analysis. Chem. Geol. 167, 313–319.

Bard E., Arnold M., Mangerud J., Paterne M., Labeyrie L.,Duprat J., Melieres M. A., Sonstegaard E. and Duplessy J. C.(1994) The North-Atlantic atmosphere-sea surface C-14 gradi-ent during the younger dryas climatic event. Earth Planet. Sci.

Lett. 126, 275–287.Bard E., Fairbanks R. G., Hamelin B., Zindler A. and Hoang C. T.

(1991) U-234 Anomalies in Corals Older Than 150,000 Years.Geochim. Cosmochim. Acta 55, 2385–2390.

Bartlett R., Elrick M., Wheeley J. R., Polyak V., Desrochers A.and Asmerom Y. (2018) Abrupt global-ocean anoxia during theLate Ordovician-early Silurian detected using uranium isotopesof marine carbonates. Proc. Nat. Acad. Sci. USA 115, 5896–5901.

Bernhard G., Geipel G., Reich T., Brendler V., Amayri S. andNitsche H. (2001) Uranyl(VI) carbonate complex formation:validation of the Ca2UO2(CO3)(3)(aq.) species. Radiochim.

Acta 89, 511–518.Betzler C., Reijmer J. J. G., Bernet K., Eberli G. P. and Anselmetti

F. S. (1999) Sedimentary patterns and geometries of theBahamian outer carbonate ramp (Miocene-Lower Pliocene,Great Bahama Bank). Sedimentology 46, 1127–1143.

Bigeleisen J. (1996) Nuclear size and shape effects in chemicalreactions. Isotope chemistry of the heavy elements. J. Am.

Chem. Soc. 118, 3676–3680.Boss S. K. and Wilkinson B. H. (1991) Planktogenic eustatic

control on cratonic oceanic carbonate accumulation. J. Geol.99, 497–513.

Brand U. and Veizer J. (1980) Chemical diagenesis of a multicom-ponent carbonate system. 1. Trace-elements. J. Sed. Petrol 50,1219–1236.

Brennecka G. A., Herrmann A. D., Algeo T. J. and Anbar A. D.(2011) Rapid expansion of oceanic anoxia immediately beforethe end-Permian mass extinction. Proc. Nat. Acad. Sci. USA

108, 17631–17634.Bura-Nakic E., Andersen M. B., Archer C., de Souza G. F.,

Margus M. and Vance D. (2018) Coupled Mo-U abundancesand isotopes in a small marine euxinic basin: constraints onprocesses in euxinic basins. Geochim. Cosmochim. Acta 222,212–229.

Chabaux F., Riotte J. and Dequincey O. (2003) U-Th-Rafractionation during weathering and river transport. Rev.

Miner. Geochem. 52, 533–576.Chen D. Z. and Tucker M. E. (2003) The Frasnian-Famennian

mass extinction: insights from high-resolution sequence stratig-raphy and cyclostratigraphy in South China. Palaeogeogr.

Palaeocl. 193, 87–111.Chen J. H., Edwards R. L. and Wasserburg G. J. (1986) U-238, U-

234 and Th-232 in Seawater. Earth Planet. Sc. Lett. 80, 241–251.

Chen T. Y., Robinson L. F., Beasley M. P., Claxton L. M.,Andersen M. B., Gregoire L. J., Wadham J., Fornari D. J. andHarpp K. S. (2016a) Ocean mixing and ice-sheet control ofseawater U-234/U-238 during the last deglaciation. Science 354,626–629.

Chen X. M., Romaniello S. J. and Anbar A. D. (2017) Uraniumisotope fractionation induced by aqueous speciation: implica-tions for U isotopes in marine CaCO3 as a paleoredox proxy.Geochim. Cosmochim. Ac 215, 162–172.

Chen X. M., Romaniello S. J., Herrmann A. D., Wasylenki L. E.and Anbar A. D. (2016b) Uranium isotope fractionation duringcoprecipitation with aragonite and calcite. Geochim. Cos-

mochim. Acta 188, 189–207.Cheng H., Edwards R. L., Shen C. C., Polyak V. J., Asmerom Y.,

Woodhead J., Hellstrom J., Wang Y. J., Kong X. G., Spotl C.,Wang X. F. and Alexander E. C. (2013) Improvements in Th-230 dating, Th-230 and U-234 half-life values, and U-Thisotopic measurements by multi-collector inductively coupledplasma mass spectrometry. Earth Planet. Sci. Lett. 371, 82–91.

Chester R. and Jickells T. D. (2012) Marine Geochemistry. Wiley.Chutcharavan P. M., Dutton A. and Ellwood M. J. (2018)

Seawater 234U/238U recorded by modern and fossil corals.Geochim. Cosmochim. Acta 224, 1–17.



http://refhub.elsevier.com/S0016-7037(18)30462-9/h0005


















































































































Clarkson M. O., Stirling C. H., Jenkyns H. C., Dickson A. J.,Porcelli D., Moy C. M., von Strandmann P. A. E. P., Cooke I.R. and Lenton T. M. (2018) Uranium isotope evidence for twoepisodes of deoxygenation during Oceanic Anoxic Event 2.Proc. Nat. Acad. Sci. USA 115, 2918–2923.

Cutler K. B., Edwards R. L., Taylor F. W., Cheng H., Adkins J.,Gallup C. D., Cutler P. M., Burr G. S. and Bloom A. L. (2003)Rapid sea-level fall and deep-ocean temperature change sincethe last interglacial period. Earth Planet. Sci. Lett. 206, 253–271.

Dahl T. W., Boyle R. A., Canfield D. E., Connelly J. N., Gill B. C.,Lenton T. M. and Bizzarro M. (2014) Uranium isotopesdistinguish two geochemically distinct stages during the laterCambrian SPICE event. Earth Planet. Sci. Lett. 401, 313–326.

Dahl T. W., Connelly J. N., Kouchinsky A., Gill B. C., Mansson S.F. and Bizzarro M. (2017) Reorganisation of Earth’s biogeo-chemical cycles briefly oxygenated the oceans 520 Myr ago.Geochem. Perspect. Lett. 3, 210–220.

Docrat T. I., Mosselmans J. F. W., Charnock J. M., Whiteley M.W., Collison D., Livens F. R., Jones C. and Edmiston M. J.(1999) X-ray absorption spectroscopy of tricarbonatodiox-ouranate(V), [UO2(CO3)(3)](5-), in aqueous solution. Inorg.

Chem. 38, 1879–1882.Dong W. M. and Brooks S. C. (2006) Determination of the

formation constants of ternary complexes of uranyl andcarbonate with alkaline earth metals (Mg2+, Ca2+, Sr2+,and Ba2+) using anion exchange method. Environ. Sci. Technol.40, 4689–4695.

Droxler A. W. and Schlager W. (1985) Glacial versus interglacialsedimentation-rates and turbidite frequency in the Bahamas.Geology 13, 799–802.

Dunk R. M., Mills R. A. and Jenkins W. J. (2002) A reevaluationof the oceanic uranium budget for the Holocene. Chem. Geol.

190, 45–67.Eberli, G.P., Swart, P.K., Malone, M., 1997. Sites 1008/1009. In:

Proc. Ocean Drilling Program, Initial Reports. pp. 347–373.Edwards R. L., Gallup C. D. and Cheng H. (2003) Uranium-series

dating of marine and lacustrine carbonates. Rev. Mineral.

Geochem. 52, 363–405.Elrick M., Polyak V., Algeo T. J., Romaniello S., Asmerom Y.,

Herrmann A. D., Anbar A. D., Zhao L. and Chen Z.-Q. (2017)Global-ocean redox variation during the middle-late Permianthrough Early Triassic based on uranium isotope and Th/Utrends of marine carbonates. Geology 45, 163–166.

Elzinga E. J., Tait C. D., Reeder R. J., Rector K. D., Donohoe R.J. and Morris D. E. (2004) Spectroscopic investigation of U(VI)sorption at the calcite-water interface. Geochim. Cosmochim.

Acta 68, 2437–2448.Endrizzi F. and Rao L. F. (2014) Chemical speciation of uranium

(VI) in marine environments: complexation of calcium andmagnesium ions with [(UO2)(CO3)(3)](4-) and the effect on theextraction of uranium from seawater. Chem.-Eur. J. 20, 14499–14506.

Esat T. M. and Yokoyama Y. (2006) Variability in the uraniumisotopic composition of the oceans over glacial-interglacialtimescales. Geochim. Cosmochim. Acta 70, 4140–4150.

Esat T. M. and Yokoyama Y. (2010) Coupled uranium isotope andsea-level variations in the oceans. Geochim. Cosmochim. Acta

74, 7008–7020.Gabitov R. I., Gaetani G. A., Watson E. B., Cohen A. L. and

Ehrlich H. L. (2008) Experimental determination of growth rateeffect on U6+ and Mg2+ partitioning between aragonite andfluid at elevated U6+ concentration. Geochim. Cosmochim.

Acta 72, 4058–4068.Gabitov R. I. and Watson E. B. (2006) Partitioning of strontium

between calcite and fluid. Geochem. Geophy. Geosy., 7.

Gallup C. D., Edwards R. L. and Johnson R. G. (1994) TheTiming of High Sea Levels over the Past 200,000 Years. Science263, 796–800.

Gothmann A. M., Stolarski J., Adkins J. F., Schoene B., Dennis K.J., Schrag D. P., Mazur M. and Bender M. L. (2015) Fossilcorals as an archive of secular variations in seawater chemistrysince the Mesozoic. Geochim. Cosmochim. Acta 160, 188–208.

Goto K. T., Anbar A. D., Gordon G. W., Romaniello S. J.,Shimoda G., Takaya Y., Tokumaru A., Nozaki T., Suzuki K.,Machida S., Hanyu T. and Usui A. (2014) Uranium isotopesystematics of ferromanganese crusts in the Pacific Ocean:implications for the marine U-238/U-235 isotope system.Geochim. Cosmochim. Acta 146, 43–58.

Hamelin B., Bard E., Zindler A. and Fairbanks R. G. (1991) U-234/U-238 mass-spectrometry of corals – how accurate is the U-Th age of the last interglacial period. Earth Planet. Sci. Lett.

106, 169–180.Hardisty D. S., Lu Z., Bekker A., Diamond C. W., Gill B. C., Jiang

G., Kah L. C., Knoll A. H., Loyd S. J., Osburn M. R.,Planavsky N. J., Wang C., Zhou X. and Lyons T. W. (2017)Perspectives on Proterozoic surface ocean redox from iodinecontents in ancient and recent carbonate. Earth Planet. Sci.

Lett. 463, 159–170.Henderson G. M. (2002) Seawater (U-234/U-238) during the last

800 thousand years. Earth Planet. Sci. Lett. 199, 97–110.Henderson G. M., Slowey N. C. and Haddad G. A. (1999) Fluid

flow through carbonate platforms: constraints from U-234/U-238 and Cl- in Bahamas pore-waters. Earth Planet. Sci. Lett.

169, 99–111.Higgins J. A., Blattler C. L., Lundstrom E. A., Santiago-Ramos D.

P., Akhtar A. A., Ahm A. S. C., Bialik O., Holmden C.,Bradbury H., Murray S. T. and Swart P. K. (2018) Mineralogy,early marine diagenesis, and the chemistry of shallow-watercarbonate sediments. Geochim. Cosmochim. Acta 220, 512–534.

Holland H. D. (2006) The oxygenation of the atmosphere andoceans. Philos. T R Soc. B 361, 903–915.

Holmden C., Amini M. and Francois R. (2015) Uranium isotopefractionation in Saanich Inlet: a modern analog study of apaleoredox tracer. Geochim. Cosmochim. Acta 153, 202–215.

Holmden C., Creaser R. A., Muehlenbachs K., Leslie S. A. andBergstrom S. M. (1998) Isotopic evidence for geochemicaldecoupling between ancient epeiric seas and bordering oceans:Implications for secular curves. Geology 26, 567–570.

Hood A. V., Planavsky N. J., Wallace M. W. and Wang X. L.(2018) The effects of diagenesis on geochemical paleoredoxproxies in sedimentary carbonates. Geochim. Cosmochim. Acta

232, 265–287.Hood A. V. S., Planavsky N. J., Wallace M. W., Wang X. L.,

Bellefroid E. J., Gueguen B. and Cole D. B. (2016) Integratedgeochemical-petrographic insights from component-selectivedelta U-238 of Cryogenian marine carbonates. Geology 44,935–938.

Jaffey A. H., Flynn K. F., Glendenin L. E., Bentley W. C. andEssling A. M. (1971) Precision measurement of half-lives andspecific activities of U-235 and U-238. Phys. Rev. C 4, 1889-+.

Jeppsson L., Anehus R. and Fredholm D. (1999) The optimalacetate buffered acetic acid technique for extracting phosphaticfossils. J. Paleontol. 73, 964–972.

Jeppsson L., Fredholm D. and Mattiasson B. (1985) Acetic-acidand phosphatic fossils – a warning. J. Paleontol. 59, 952–956.

Jost A. B., Bachan A., van de Schootbrugge B., Lau K. V., WeaverK. L., Maher K. and Payne J. L. (2017) Uranium isotopeevidence for an expansion of marine anoxia during the end-Triassic extinction. Geochem. Geophy. Geosy. 18, 3093–3108.

Kelly S. D., Newville M. G., Cheng L., Kemner K. M., Sutton S.R., Fenter P., Sturchio N. C. and Spotl C. (2003) Uranyl




































































































































incorporation in natural calcite. Environ. Sci. Technol. 37,1284–1287.

Kelly S. D., Rasbury E. T., Chattopadhyay S., Kropf A. J. andKemner K. M. (2006) Evidence of a stable uranyl site in ancientorganic-rich calcite. Environ. Sci. Technol. 40, 2262–2268.

Kendall B., Brennecka G. A., Weyer S. and Anbar A. D. (2013)Uranium isotope fractionation suggests oxidative uraniummobilization at 2.50 Ga. Chem. Geol. 362, 105–114.

Kendall B., Komiya T., Lyons T. W., Bates S. M., Gordon G. W.,Romaniello S. J., Jiang G., Creaser R. A., Xiao S., McFaddenK., Sawaki Y., Tahata M., Shu D., Han J., Li Y., Chu X. andAnbar A. D. (2015) Uranium and molybdenum isotopeevidence for an episode of widespread ocean oxygenationduring the late Ediacaran Period. Geochim. Cosmochim. Acta

156, 173–193.Kerisit S. and Liu C. X. (2010) Molecular simulation of the

diffusion of uranyl carbonate species in aqueous solution.Geochim. Cosmochim. Acta 74, 4937–4952.

Kitano Y. and Oomori T. (1971) The coprecipitation of uraniumwith calcium carbonate. J. Oceanograph. Soc. Jpn. 27, 34–42.

Ku T. L., Knauss K. G. and Mathieu G. G. (1977) Uranium inopen ocean – concentration and isotopic composition. Deep-

Sea Res. 24, 1005–1017.Kuznetsov A. B., Krupenin M. T., Ovchinnikova G. V., Gorokhov

I. M., Maslov A. V., Kaurova O. K. and Ellmies R. (2005)Diagenesis of carbonate and siderite deposits of the LowerRiphean Bakal Formation, the southern Urals: Sr isotopiccharacteristics and Pb-Pb age. Lithol Miner. Resour+ 40, 195–215.

Langmuir D. (1978) Uranium solution-mineral equilibria at low-temperatures with applications to sedimentary ore-deposits.Geochim. Cosmochim. Acta 42, 547–569.

Lau K. V., Macdonald F. A., Maher K. and Payne J. L. (2017)Uranium isotope evidence for temporary ocean oxygenation inthe aftermath of the Sturtian Snowball Earth. Earth Planet. Sci.Lett. 458, 282–292.

Lau K. V., Maher K., Altiner D., Kelley B. M., Kump L. R.,Lehrmann D. J., Silva-Tamayo J. C., Weaver K. L., Yu M. Y.and Payne J. L. (2016) Marine anoxia and delayed Earth systemrecovery after the end-Permian extinction. Proc. Nat. Acad. Sci.

USA 113, 2360–2365.Li D., Shields-Zhou G. A., Ling H. F. and Thirlwall M. (2011)

Dissolution methods for strontium isotope stratigraphy: guide-lines for the use of bulk carbonate and phosphorite rocks.Chem. Geol. 290, 133–144.

Liu C., Wang Z. R. and Raub T. D. (2013) Geochemicalconstraints on the origin of Marinoan cap dolostones fromNuccaleena Formation, South Australia. Chem. Geol. 351, 95–104.

Liu C., Wang Z. R., Raub T. D., Macdonald F. A. and Evans D.A. D. (2014) Neoproterozoic cap-dolostone deposition instratified glacial meltwater plume. Earth Planet. Sci. Lett.

404, 22–32.Lu X. Z., Kendall B., Stein H. J., Li C., Hannah J. L., Gordon G.

W. and Ebbestad J. O. R. (2017) Marine redox conditionsduring deposition of Late Ordovician and Early Silurianorganic-rich mudrocks in the Siljan ring district, centralSweden. Chem. Geol. 457, 75–94.

Malone, M., 2000. 14. DATA REPORT: GEOCHEMISTRYAND MINERALOGY OF PERIPLATFORM CARBO-NATE SEDIMENTS: SITES 1006, 1008, AND 10091, Pro-ceedings of the Ocean Drilling Program. Scientific results.Ocean Drilling Program, pp. 145–152.

Malone M. J., Slowey N. C. and Henderson G. M. (2001) Earlydiagenesis of shallow-water periplatform carbonate sediments,

leeward margin, Great Bahama Bank (Ocean Drilling ProgramLeg 166). Geol. Soc. Am. Bull. 113, 881–894.

McManus J., Berelson W. M., Klinkhammer G. P., Hammond D.E. and Holm C. (2005) Authigenic uranium: relationship tooxygen penetration depth and organic carbon rain. Geochim.

Cosmochim. Acta 69, 95–108.Meece D. E. and Benninger L. K. (1993) The coprecipitation of Pu

and other radionuclides with CaCO3. Geochim. Cosmochim.

Acta 57, 1447–1458.Montoya-Pino C., Weyer S., Anbar A. D., Pross J., Oschmann W.,

van de Schootbrugge B. and Arz H. W. (2010) Globalenhancement of ocean anoxia during Oceanic Anoxic Event2: a quantitative approach using U isotopes. Geology 38, 315–318.

Morford J. L., Emerson S. R., Breckel E. J. and Kim S. H. (2005)Diagenesis of oxyanions (V, U, Re, and Mo) in pore waters andsediments from a continental margin. Geochim. Cosmochim.

Acta 69, 5021–5032.Murphy M. J., Stirling C. H., Kaltenbach A., Turner S. P. and

Schaefer B. F. (2014) Fractionation of U-238/U-235 byreduction during low temperature uranium mineralisationprocesses. Earth Planet. Sci. Lett. 388, 306–317.

Nadeau M. J., Grootes P. M., Voelker A., Bruhn F., Duhr A. andOriwall A. (2001) Carbonate C-14 background: does it havemultiple personalities? Radiocarbon 43, 169–176.

Noordmann J., Weyer S., Montoya-Pino C., Dellwig O., NeubertN., Eckert S., Paetzel M. and Bottcher M. E. (2015) Uraniumand molybdenum isotope systematics in modern euxinic basins:case studies from the central Baltic Sea and the Kyllaren fjord(Norway). Chem. Geol. 396, 182–195.

Oehlert A. M., Lamb-Wozniak K. A., Devlin Q. B., Mackenzie G.J., Reijmer J. J. G. and Swart P. K. (2012) The stable carbonisotopic composition of organic material in platform derivedsediments: implications for reconstructing the global carboncycle. Sedimentology 59, 319–335.

Oehlert A. M. and Swart P. K. (2014) Interpreting carbonate andorganic carbon isotope covariance in the sedimentary record.Nature Commun. 5.

Opdyke B. N. and Wilkinson B. H. (1988) Surface area control ofshallow cratonic to deep marine carbonate accumulation.Paleoceanography 3, 685–703.

Owens S. A., Buesseler K. O. and Sims K. W. W. (2011) Re-evaluating the U-238-salinity relationship in seawater: implica-tions for the U-238-Th-234 disequilibrium method. Mar. Chem.

127, 31–39.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. and Lyons T. W. (2013a) Large-scale fluctuations in Precambrian atmospheric and oceanicoxygen levels from the record of U in shales. Earth Planet. Sci.

Lett. 369, 284–293.Partin C. A., Lalonde S. V., Planavsky N. J., Bekker A., Rouxel O.

J., Lyons T. W. and Konhauser K. O. (2013b) Uranium in ironformations and the rise of atmospheric oxygen. Chem. Geol.

362, 82–90.Phan T. T., Gardiner J. B., Capo R. C. and Stewart B. W. (2018)

Geochemical and multi-isotopic (Sr-87/Sr-86, Nd-143/Nd-144,U-238/U-235) perspectives of sediment sources, depositionalconditions, and diagenesis of the Marcellus Shale, AppalachianBasin, USA. Geochim. Cosmochim. Acta 222, 187–211.

Ramdohr P. (1958) New observations on the ores of the Witwa-tersrand in South Africa and their genetic significance. Geol.Soc. South Afr..

Rasmussen B. and Buick R. (1999) Redox state of the Archeanatmosphere: evidence from detrital heavy minerals in ca. 3250–
































































































































2750 Ma sandstones from the Pilbara Craton, Australia.Geology 27, 115–118.

Reeder R. J., Nugent M., Lamble G. M., Tait C. D. and Morris D.E. (2000) Uranyl incorporation into calcite and aragonite:XAFS and luminescence studies. Environ. Sci. Technol. 34, 638–644.

Reeder R. J., Nugent M., Tait C. D., Morris D. E., Heald S. M.,Beck K. M., Hess W. P. and Lanzirotti A. (2001) Coprecip-itation of uranium(VI) with calcite: XAFS, micro-XAS, andluminescence characterization. Geochim. Cosmochim. Acta 65,3491–3503.

Ridgwell A. (2005) A Mid Mesozoic Revolution in the regulationof ocean chemistry. Mar. Geol. 217, 339–357.

Robinson L. F., Adkins J. F., Fernandez D. P., Burnett D. S.,Wang S. L., Gagnon A. C. and Krakauer N. (2006) Primary Udistribution in scleractinian corals and its implications for Useries dating. Geochem. Geophy. Geosy., 7.

Robinson L. F., Henderson G. M., Hall L. and Matthews I. (2004)Climatic control of riverine and Seawater uranium-isotoperatios. Science 305, 851–854.

Robinson L. F., Henderson G. M. and Slowey N. C. (2002) U-Thdating of marine isotope stage 7 in Bahamas slope sediments.Earth Planet. Sci. Lett. 196, 175–187.

Rogers, J.J.W., Adams, J.A.S., 1969. Uranium. in Handbook ofGeochemistry, II/l (edit. by Wedepohl, K. H.), 92. Springer-Verlag, Berlin, 1969.

Rolison J. M., Stirling C. H., Middag R. and Rijkenberg M. J. A.(2017) Uranium stable isotope fractionation in the Black Sea:modern calibration of the U-238/(235) U paleo-redox proxy.Geochim. Cosmochim. Acta 203, 69–88.

Romaniello S. J., Herrmann A. D. and Anbar A. D. (2013)Uranium concentrations and U-238/U-235 isotope ratios inmodern carbonates from the Bahamas: assessing a novelpaleoredox proxy. Chem. Geol. 362, 305–316.

Schauble E. A. (2007) Role of nuclear volume in driving equilib-rium stable isotope fractionation of mercury, thallium, andother very heavy elements. Geochim. Cosmochim. Acta 71,2170–2189.

Scott C., Lyons T. W., Bekker A., Shen Y., Poulton S. W., Chu X.and Anbar A. D. (2008) Tracing the stepwise oxygenation ofthe Proterozoic ocean. Nature 452, 456-U455.

Sepulcre S., Durand N. and Bard E. (2017) Large C-14 age offsetsbetween the fine fraction and coexisting planktonic foramini-fera in shallow Caribbean sediments. Quat. Geochronol. 38, 61–74.

Slowey N. C., Wilber R. J., Haddad G. A. and Henderson G. M.(2002) Glacial-to-Holocene sedimentation on the western slopeof Great Bahama Bank. Mar. Geol. 185, 165–176.

Song H. Y., Song H. J., Algeo T. J., Tong J. N., Romaniello S. J.,Zhu Y. Y., Chu D. L., Gong Y. M. and Anbar A. D. (2017)Uranium and carbon isotopes document global-ocean redox-productivity relationships linked to cooling during the Fras-nian-Famennian mass extinction. Geology 45, 887–890.

Spratt R. M. and Lisiecki L. E. (2016) A Late Pleistocene sea levelstack. Clim. Past 12, 1079–1092.

Stirling C. H., Andersen M. B., Potter E. K. and Halliday A. N.(2007) Low-temperature isotopic fractionation of uranium.Earth Planet Sci. Lett. 264, 208–225.

Stirling C. H., Esat T. M., Mcculloch M. T. and Lambeck K.(1995) High-precision U-series dating of corals from Western-Australia and implications for the timing and duration of thelast interglacial. Earth Planet. Sci. Lett. 135, 115–130.

Swart P. K. and Hubbard J. (1982) Uranium in scleractinian coralskeletons. Coral Reefs 1, 13–19.

Swart P. K., Burns S. J. and Leder J. J. (1991) Fractionation of thestable isotopes of oxygen and carbon in carbon dioxide during

the reaction of calcite with phosphoric acid as a function oftemperature and technique. Chem. Geol. 86, 89–96.

Swart P. K. and Eberli G. P. (2005) The nature of the delta C-13 ofperiplatform sediments: implications for stratigraphy and theglobal carbon cycle. Sediment. Geol. 175, 115–129.

Swart P. K., Oehlert A. M., Mackenzie G. J., Eberli G. P. andReijmer J. J. G. (2014) The fertilization of the Bahamas bySaharan dust: a trigger for carbonate precipitation? Geology 42,671–674.

Szabo Z., Moll H. and Grenthe I. (2000) Structure and dynamics inthe complex ion (UO2)(2)(CO3)(OH)(3)(-). J. Chem. Soc.

Dalton, 3158–3161.Telus M., Dauphas N., Moynier F., Tissot F. L. H., Teng F. Z.,

Nabelek P. I., Craddock P. R. and Groat L. A. (2012) Iron, zinc,magnesium and uranium isotopic fractionation during conti-nental crust differentiation: the tale from migmatites, granitoids,and pegmatites. Geochim. Cosmochim. Acta 97, 247–265.

Teng F.-Z., Dauphas N. and Watkins J. M. (2017) Non-traditionalstable isotopes: retrospective and prospective. Rev. Mineral.

Geochem. 82, 1–26.Thierstein H. R., Geitzenauer K. R. and Molfino B. (1977) Global

synchroneity of late quaternary coccolith datum levels -validation by oxygen isotopes. Geology 5, 400–404.

Thompson W. G., Curran H. A., Wilson M. A. and White B.(2011) Sea-level oscillations during the last interglacial high-stand recorded by Bahamas corals. Nat Geosci 4, 684–687.

Tissot F. L. H. and Dauphas N. (2015) Uranium isotopiccompositions of the crust and ocean: age corrections, U budgetand global extent of modern anoxia. Geochim. Cosmochim. Acta

167, 113–143.Tissot F. L. H., Dauphas N. and Grove T. L. (2017) Distinct U-

238/U-235 ratios and REE patterns in plutonic and volcanicangrites: geochronologic implications and evidence for Uisotope fractionation during magmatic processes. Geochim.

Cosmochim. Acta 213, 593–617.Tomiak P. J., Andersen M. B., Hendy E. J., Potter E. K., Johnson

K. G. and Penkman K. E. H. (2016) The role of skeletal micro-architecture in diagenesis and dating of Acropora palmata.Geochim. Cosmochim. Acta 183, 153–175.

Tostevin R., Shields G. A., Tarbuck G. M., He T. C., Clarkson M.O. and Wood R. A. (2016) Effective use of cerium anomalies asa redox proxy in carbonate-dominated marine settings. Chem.

Geol. 438, 146–162.Verbruggen A., Alonso A., Eykens R., Kehoe F., Kuhn H., Richter

S. and Aregbe Y. (2008) Preparation and certification ofIRMM-3636, IRMM-3636a and IRMM-3636b. JRC Sci.

Technic. Rep..Walker L. J., Wilkinson B. H. and Ivany L. C. (2002) Continental

drift and Phanerozoic carbonate accumulation in shallow-shelfand deep-marine settings. J. Geol. 110, 75–87.

Wang X. L., Johnson T. M. and Lundstrom C. C. (2015) Isotopefractionation during oxidation of tetravalent uranium bydissolved oxygen. Geochim Cosmochim. Acta 150, 160–170.

Wang X. L., Planavsky N. J., Reinhard C. T., Hein J. R. andJohnson T. M. (2016) A cenozoic seawater redox record derivedfrom U-238/U-235 in Ferromanganese crusts. Am. J. Sci. 316,64–83.

Wassenburg J. A., Scholz D., Jochum K. P., Cheng H., Oster J.,Immenhauser A., Richter D. K., Hager T., Jamieson R. A.,Baldini J. U. L., Hoffmann D. and Breitenbach S. F. M. (2016)Determination of aragonite trace element distribution coeffi-cients from speleothem calcite-aragonite transitions. Geochim.

Cosmochim. Acta 190, 347–367.Wei G.-Y., Planavsky N. J., Tarhan L. G., Chen X., Wei W., Li D.

and Ling H.-F. (2018) Marine redox fluctuation as a potentialtrigger for the Cambrian explosion. Geology.



































































































































Weyer S., Anbar A. D., Gerdes A., Gordon G. W., Algeo T. J. andBoyle E. A. (2008) Natural fractionation of U-238/U-235.Geochim. Cosmochim. Acta 72, 345–359.

Yang S., Kendall B., Lu X. Z., Zhang F. F. and Zheng W. (2017)Uranium isotope compositions of mid-Proterozoic black shales:evidence for an episode of increased ocean oxygenation at 1.36Ga and evaluation of the effect of post-depositional hydrother-mal fluid flow. Precambrian Res. 298, 187–201.

Zhang F., Xiao S., Kendall B., Romaniello S. J., Cui H., MeyerM., Gilleaudeau G. J., Kaufman A. J. and Anbar A. D. (2018a)Extensive marine anoxia during the terminal Ediacaran Period.Sci. Adv. 4, eaan8983.

Zhang F. F., Algeo T. J., Romaniello S. J., Cui Y., Zhao L. S.,Chen Z. Q. and Anbar A. D. (2018b) Congruent Permian-Triassic delta U-238 records at Panthalassic and Tethyan sites:

confirmation of global-oceanic anoxia and validation of the U-isotope paleoredox proxy. Geology 46, 327–330.

Zhang F. F., Romaniello S. J., Algeo T. J., Lau K. V., ClaphamM.E., Richoz S., Herrmann A. D., Smith H., Horacek M. andAnbar A. D. (2018c) Multiple episodes of extensive marineanoxia linked to global warming and continental weatheringfollowing the latest Permian mass extinction. Sci. Adv. 4.

Zhang K., Zhu X. K. and Yan B. (2015) A refined dissolutionmethod for rare earth element studies of bulk carbonate rocks.Chem. Geol. 412, 82–91.

Zheng Y., Anderson R. F., Van Geen A. and Fleisher M. Q. (2002)Remobilization of authigenic uranium in marine sediments bybioturbation. Geochim. Cosmochim. Acta 66, 1759–1772.

Associate editor: Claudine Stirling





























controls of eustasy and diagenesis on the 238u/235u of … · 2018-10-09 · controls of eustasy...

Documents