RESEARCH ARTICLE10.1002/2013GC005074

Deep-sea redox across the Paleocene-Eocene thermalmaximumCecily P€alike1,2,3, Margaret L. Delaney4, and James C. Zachos5

1Institute of Geosciences, Goethe-University Frankfurt, Frankfurt, Germany, 2Biodiversity and Climate Research Centre(BIK-F), Frankfurt, Germany, 3Now at MARUM—Center for Marine Environmental Sciences, University of Bremen, Bremen,Germany, 4Ocean Sciences Department, Institute of Marine Sciences, University of California, Santa Cruz, California, USA,5Earth and Planetary Sciences Department, University of California, Santa Cruz, California, USA

Abstract Large amounts of 13C-depleted carbon were released to the oceans and atmosphere during aperiod of abrupt global warming at the Paleocene-Eocene thermal maximum (PETM) (�55 Ma). Investiga-tions of qualitative sedimentologic and paleontologic redox proxies such as bioturbation and benthicassemblages from pelagic and hemipelagic sections suggest transient reductions in bottom water oxygenduring this interval, possibly on a global scale. Here, we present bulk sediment manganese (Mn) and ura-nium (U) enrichment factors (EF) in Atlantic and Pacific deep-sea cores to constrain relative paleoredoxchanges across the PETM. Mn EF range from 1 to 9 in Atlantic sites, 1 to 35 in Southern Ocean sites, and areat crustal averages (EF 5 1) in Pacific sites. U EF range from 1 to 5 in Atlantic sites, 1 to 90 in Southern Oceansites, and are at crustal averages in Pacific sites. Our results indicate suboxic conditions prior to, during, andin the recovery from the PETM at intermediate depth sites in the Atlantic and Southern Ocean while thePacific sites remained relatively oxygenated. The difference in oxygenation between the Atlantic and Pacificsites leads us to suggest the source for isotopically light carbon release during the PETM was in the Atlantic.

1. Introduction

The Paleocene-Eocene thermal maximum (PETM) is characterized by extreme warming coupled with amajor perturbation to the global carbon cycle. The ocean’s surface temperature increased by 5–7�C, in lessthan 20 kyr [Sluijs et al., 2006; Zachos et al., 2003]. This event was the first and largest of a series of hyper-thermal events in the early Eocene, identified by sharp negative d13C excursions [Lourens et al., 2005; Stapet al., 2009; Zachos et al., 2010]. Several mechanisms have been posited to explain the globally synchronous�3& negative carbon isotope excursion (CIE) and implied carbon release: the rapid dissociation of methanerelease from gas hydrates [Dickens, 2000, 2011], volcanic release of CH4 and CO2 in the North Atlantic [Sven-sen et al., 2004], and various terrestrial sources of carbon [DeConto et al., 2012; Higgins and Schrag, 2006;Kurtz et al., 2003; Pancost et al., 2007].

The posited release of a large mass of reduced carbon, such as methane, directly into the oceancoupled with rapid or even slow warming of the upper ocean could alter processes critical to maintain-ing the dissolved oxygen content of the deep sea. This includes changes in the rate and style of meridi-onal overturning circulation (MOC), which would be highly sensitive to high-latitude warming [Luntet al., 2010]. Indeed, multiple indicators or proxies of bottom water redox conditions and model studiessuggest shifts toward lower redox state of bottom waters in multiple locations throughout the oceans[Dickson et al., 2012; Nicolo et al., 2010; Winguth et al., 2012]. For example, benthic foraminifer assemb-lages show reduction in the abundance of species that were intolerant of suboxic conditions [Thomas,2007]. Documenting the temporal and spatial patterns of redox change during the PETM is thus criticalto testing models for the origin of the event, as well as for assessing the sensitivity of ocean overturningto rapid warming in general.

Previous studies have either used dissolution patterns of seafloor calcium carbonate (CaCO3) records alongwith sediment models to evaluate the pattern and location of carbon release through the ocean basins[Panchuk et al., 2008; Zeebe and Zachos, 2007] or benthic foraminifera carbon isotope records [Nunes andNorris, 2006]. We previously documented changes in the distribution of redox-sensitive trace metals in adepth transect of Paleocene-Eocene Boundary (PEB) sections from Walvis Ridge, in the south Atlantic,

Key Points:� We measure concentrations of Mn

and U in marine PETM sediments� North Atlantic bottom waters were

suboxic relative to the Pacific duringthe PETM� The source of methane release

during the PETM was likely in theNorth Atlantic

Supporting Information:� Read_me� Figure S1� Table S1–S15

Correspondence to:C. P€alike,cpaelike@marum.de

Citation:P€alike, C., M. L. Delaney, and J. C.Zachos (2014), Deep-sea redox acrossthe Paleocene-Eocene thermalmaximum, Geochem. Geophys. Geosyst.,15, 1038–1053, doi:10.1002/2013GC005074.

Received 2 OCT 2013

Accepted 16 FEB 2014

Accepted article online 24 FEB 2014

Published online 11 APR 2014

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1038

Geochemistry, Geophysics, Geosystems

PUBLICATIONS

finding evidence of a transient shift toward suboxic conditions [Chun et al., 2010]. Our current study broad-ens the spatial patterns of change in the same redox-sensitive trace metals, specifically Mn and U. Oneobjective is to use these patterns to assess the methane dissociation hypothesis. If CH4 is oxidized in theoceans as posited, this should produce a pattern of oxygen deficiency in the deep oceans that is partlydetermined by the flow of carbon and ocean circulation [Dickens, 2000, 2003]. We use our proxies for bot-tom water redox as one test of the methane hydrate hypothesis. While alternate sources of carbon releaseto the ocean and atmosphere produce similar d13C and ocean temperature records, warming alone is insuf-ficient for major changes in dissolved oxygen [Dickens, 2000].

In the modern ocean, dissolved inorganic carbon (DIC) and bottom water oxygen are influenced in oppositeways along the path of deepwater circulation. The youngest deep waters are oxygen rich and DIC poor.Along the pathway of deep-ocean circulation, oxygen concentrations decrease as organic matter oxidationoccurs, increasing DIC concentrations. The increase in DIC results in deep waters that are more corrosive toCaCO3. Depletion of bottom water oxygen should be most pronounced in the reservoir where methane oxi-dation occurred [Dickens, 2000]. Methane oxidation in the water column will also cause DIC values toincrease, causing deep waters to appear temporarily older.

In this study, we test the theory that thermal dissociation of marine gas hydrates contributed to rapid cli-mate change at the PETM. We also aim to determine the geographic location of this gas hydrate reservoir.Our results are from a wide distribution of sites: two in the North Atlantic, two from the Southern Ocean,and two from the Pacific Ocean (Figure 1 and Table 1). Redox-sensitive trace metal enrichment factors aremeasured to infer the paleoredox conditions at time of sediment burial. We compare our results to otherproxies of bottom water carbon chemistry and oxygenation across the PETM.

2. Background

2.1. ProxiesTrace element enrichment factors of manganese (Mn) and uranium (U) are used to characterize the paleoredoxenvironment of the host sediments at or near the time of burial [Calvert and Pedersen, 1993]. Because of thecomplementary redox sensitivity of Mn and U in their solid and dissolved states, we use bulk trace elementratios to determine the position of the oxic/suboxic boundary [Mangini et al., 2001]. However, early diagenesis ofsubsurface sediments has the potential to remobilize primary redox signals [Froelich et al., 1979; Tribovillard et al.,2006]. We use modern day redox state to assess the likelihood of postdepositional diagenesis.

In general, Mn enrichments relative to crustal abundances represent Mn oxides (MnO2) indicating well-oxygenated conditions [Calvert and Pedersen, 1993]. However, Mn enrichments may also represent

180˚

180˚

210˚

210˚

240˚

240˚

270˚

270˚

300˚

300˚

330˚

330˚

0˚

0˚

30˚

30˚

60˚

60˚

90˚

90˚

120˚

120˚

150˚

150˚

180˚

180˚

-90˚ -90˚

-60˚ -60˚

-30˚ -30˚

0˚ 0˚

30˚ 30˚

60˚ 60˚

90˚ 90˚

12621266

1263

401

12581221

1209

690 738

9991001

180˚

180˚

210˚

210˚

240˚

240˚

270˚

270˚

300˚

300˚

330˚

330˚

0˚

0˚

30˚

30˚

60˚

60˚

90˚

90˚

120˚

120˚

150˚

150˚

180˚

180˚

-90˚ -90˚

-60˚ -60˚

-30˚ -30˚

0˚ 0˚

30˚ 30˚

60˚ 60˚

90˚ 90˚

55 Ma Reconstruction

Figure 1. Plate tectonic reconstruction for 55 Ma showing the location of the study sites. This map was generated by the Ocean DrillingStratigraphic Network (http://www.odsn.de/).

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1039

authigenic Mn carbonates (MnCO3) that precipitate within suboxic pore waters [Boyle, 1983; Gingele andKasten, 1994; Pedersen and Price, 1982]. We use a reductive cleaning procedure to operationally differentiatebetween MnO2 (oxic environment) and MnCO3 (suboxic environment) [Boyle, 1983]. In conjunction with MnEF, U enrichments above crustal averages represent uraninite (UO2), reflecting suboxic conditions [Calvertand Pedersen, 1993; Morford and Emerson, 1999]. Although these proxies have not been directly calibratedto oxygen concentrations, they can still provide useful information about the qualitative redox conditionswith time and among sites (see Chun et al. [2010] for more details).

2.2. Site SelectionThis study expands on our previous work in the South Atlantic to include sites from the North Atlantic,Southern Ocean, and Pacific Ocean. We selected six Deep Sea Drilling Program (DSDP) and Ocean DrillingProgram (ODP) sites where the PETM has been identified (Figure 1 and Table 1). The sites range in paleo-water depth from �1350 m on Kerguelen Plateau to �3300 m in the Equatorial Pacific, thus providing con-straints on intermediate to deepwater characteristics. Burial depths of late Paleocene age samples fromthese sites range from �150 meters below seafloor (mbsf) to �280 mbsf, sufficiently deep that convectiveflow through the sediments with oceanic crust has ceased [McDuff, 1981].

2.3. Age ModelOrbitally tuned age models exist for the PETM sections at ODP Sites 690 and 1263 [R€ohl et al., 2000, 2007].These age models are based on the identification of precession and eccentricity cycles in sediment lithol-ogy, Fe, and Ca; and they assign ages relative to the PETM to inflection points in the bulk calcite d13C. Weadopted the published tie points by Kaiho et al. [2006] for Site 1209 and by Paytan et al. [2007] for Site1221, reassigning the ages to R€ohl et al. [2007]. We visually correlated all other sites to the orbitally tunedage model using the major inflection points in bulk calcite d13C (Figure 2 and Table 2). The convention ofsubdividing the PETM into 50 kyr ‘‘preevent,’’ 80 kyr ‘‘core-CIE,’’ and 80–170 kyr ‘‘recovery’’ intervals is basedon the carbon isotope records and adapted from Chun et al. [2010].

To constrain time outside of the PETM interval, we used magnetostratigraphic and biostratigraphic datumswhere available with ages calculated by Westerhold et al. [2007] which are consistent with the time frame ofthe PETM as defined in R€ohl et al. [2007]. For the oldest age pick, we used the boundary between magneto-chron 24 reversal and magnetochron 25 normal (C24r/C25n) representing 1110 kyr prior to the onset of thePEB [Westerhold et al., 2007]. For an upper boundary past the recovery of the PETM, we used the base ofplanktonic foraminifera Zone P6 as defined by the last occurrence of Morozovella velascoensis as 1743.60kyr relative to the onset of the PETM [Zachos et al., 2004]. In all our sites, we assumed linear sedimentationrates between tie points for age assignments.

Table 1. Site Characteristics

Site (Leg)

Bay ofBiscaya

401 (48)

DemeraraRiseb

1258 (207)

Walvis Ridgec

Maud Rised

690 (113)

KerguelenPlateaue

738 (119)

EquatorialPacificf

1221 (199)

ShatskyRiseg

1209 (198)1262 (208) 1263 (208) 1266 (208)

Holes used in this study NA A B A/C/D B/C B C C BWater depth 2495 3192 4755 2717 3798 2914 2252 5174 2387Paleodepth 2000 2500 3600 1500 2600 2100 1350 3300 2400Latitude 47�26N 9�26N 27�11S 28�32S 28�33S 65�09S 62�43S 12�02N 32�39NLongitude 8�49W 54�44W 1�35E 2�47E 2�21E 1�12E 82�47E 143�42W 158�30EDepth range of samples used

in this study (mbsf)198.54–202.98 170.55–174.99 124.40–128.58 282.53–288.73 260.00–272.49 166.90–171.73 283.54–286.30 153.40–154.80 195.10–197.21

Sulfate minimum (mM) 24.3 0.7 21.19 23.32 19.13 20.1 22.8 29.2 23.5Depth to sulfate minimum (mbsf) 220 399.25 140.4 9.75 229 238.85 57.45 80.4 247.15Sulfate at depth of PETM (estimate) 24 15 22 24 22 22 22.8 NA 24

aEllis et al. [1979] and Montadert et al. [1979].bErbacher et al. [2004] and Sexton et al. [2006].cZachos et al. [2004, 2005].dBarker and Kennett [1988] and Kennett and Stott [1990].eBarron et al. [1989] and Barerra and Huber [1991].fLyle et al. [2002].gBralower et al. [2002].

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1040

Where bulk d13C values were not available (Site 401) we used benthic foraminifera stable carbon isotopes.Recent work has shown that the benthic foraminifera stable isotope records display a more abrupt CIE thanbulk carbonate records [McCarren et al., 2008]. Therefore, using d13C values from benthic foraminifera spe-cies (such as Nuttallides truempyi) that were not affected by the extinction event is the most suitablereplacement for bulk carbonate d13C records.

The benthic foraminifera carbon isotope record of Nuttallides truempyi from Site 401 [Nunes and Norris,2006] was correlated to the benthic records from Sites 1262, 1263, and 690 [Kelly et al., 2005; Kennett andStott, 1991; McCarren et al., 2008] using the assigned ages from R€ohl et al. [2007] (Figure 3 and Table 2). Toassign ages to depths below the onset of the PETM at Site 401, we used the sedimentation rate calculatedby Nunes and Norris [2006] for this portion of their age model, and extrapolated ages from the onset of theevent.

2.4. Samples and Analytical ProceduresSediment samples of �1 cm3 volume were taken from all sites for trace metal analysis. Sampling resolutionvaried among the sites and ranged from �1 sample every 1.5 cm (�1 sample/1.6 kyr) at ODP 738 to �1sample every 13 cm (�1 sample/13 kyr) at ODP 1258.

194

195

196

197

198

199

-1 0 1 2 3

1258 δ13C (‰, vPDB)

332

333

334

335

336

-1 0 1 2 3

1263δ13C (‰, vPDB)

164

166

168

170

172

-1 0 1 2 3

690δ13C (‰, vPDB)

210.5

210.7

211.0

211.2

211.4

-1 0 1 2 3

1209δ13C (‰, vPDB)

Dep

th (m

eter

s)

A-

A

C

FF

G

??

Figure 2. Bulk sediment carbon isotope records for Sites 1258, 1263, 690, and 1209 versus depth. Letters and associated lines indicatinglines of correlation are based on inflections in the carbon isotopes after Zachos et al. [2005]. Bulk isotope data for Site 1258 are out of rangeof the plot (this study), 1263 from Zachos et al. [2005], Site 690 from Bains et al. [1999], Site 1209 from Colossimo et al. [2006]. vPDB, ViennaPeeDee Belemnite; Sites 1258 and 1209 given in meters composite depth (MCD); Site 690 given in meters below seafloor (MBSF); and Site1263 given in revised meters composite depth (RMCD).

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1041

Bulk sediment samples were first freeze dried, crushed, and sieved with a 150 lm polypropylene mesh.Total sediment digestions and reductive cleaning of sample splits were processed in the same way asdetailed in Chun et al. [2010]. Total concentrations of manganese (Mn), titanium (Ti), and uranium (U) weremeasured on a Finnigan Element high-resolution inductively coupled plasma-mass spectrometer (HR-ICP-MS). Aluminum (Al) was measured on a Perkin-Elmer Optima 4300 DV inductively coupled plasma-opticalemission spectrometer (ICP-OES). All digestions and analyses took place at the University of California, SantaCruz, USA.

Two internal standards composed of ahomogenized mixture of ODP Leg 202and ODP Leg 208 samples were proc-essed and analyzed with each set ofdigestions to evaluate reproducibility.Mean concentrations 6 1 standard devi-ation of the ODP Leg 202 consistencystandard were 133 6 23 lmol g21 for Al,2.4 6 0.4 lmol g21 for Ti, 6.1 6 0.7 lmolg21 for Mn, and 1.1 6 0.3 nmol g21 for U(Table 3). Mean concentrations 6 1standard deviation of the ODP Leg 208consistency standard were 154 6 22lmol g21 for Al, 4.7 6 0.8 lmol g21 forTi, 4.8 6 0.6 lmol g21 for Mn, and0.30 6 0.2 nmol g21 for U (Table 3).

For sites that did not have publishedrecords of weight percent (wt %) CaCO3

or bulk stable carbon isotopes, we meas-ured them in this study. We measuredwt % CaCO3 on sediments from Site 401using a UIC, Inc. Coulometrics Model5012 CO2 coulometer at the Universityof California, Santa Cruz, USA. Relativestandard deviations of the mean of mul-tiple measures of a pure CaCO3 standardwere <1%. Weight % CaCO3

Table 2. Carbon Isotope Tie Points From ODP Site 690 and Assigned Ages Used for Correlation and Dating the Corresponding P-E Boundary Sectionsa

Time Relative toDepth

Tie PointsSite 690(mbsf)

PETM CIE (6kyr)[R€ohl et al., 2007] Site 401 (mbsf) Site 1258 (mcd)

Site 738(mbsf)

Site 1209 (mcd)[Kaiho et al., 2006]

Site 1221 (mbsf)[Paytan et al., 2007]

LO Morozovella velascoensis 318.08b 743.60 199.79 192.21 208.80H 166.13 196.87 283.78 153.40G 167.12 153.5 200.72 195.47 284.30F 169.05 94.23 201.48 196.38 284.62 211.04E 169.39 81.17D 169.56 71.25 153.90C 170.02 42.38 285.16 211.22 154.00B 170.33 21.9 154.10A 170.63 0.75 202.58 285.48 154.30PEB 170.64 0 198.15 211.26A- 171.24 230.8 286.02 211.31 154.40B- 172.81 2102.53C24rC25n boundary 153.11c 21110.00 216.28

ambsf 5 meters below seafloor and mcd 5 meters composite depth.bDepth for LO Morozovella velascoensis from Site 1263 in revised meters composite depth (rmcd).cfor C23r 25n boundary from 1262 in mcd.

199

200

201

202

203

-3 -2 -1 0 1 2 3

401

δ13C (‰, vPDB)

Dep

th (m

eter

s)

1262 benthic 1263 benthic 690 benthic1263 bulk

-50

0

50

100

150

200

250

-3 -2 -1 0 1 2 3δ13C (‰, vPDB)

Tim

e re

lativ

e to

PE

B (k

yr)

G

?

F

PEB

Figure 3. Benthic foraminiferal isotope correlation of Site 401 to Sites 1262,1263, and 690 versus depth (meters below seafloor). Benthic foraminiferal iso-tope records from Site 401 Nunes and Norris [2006], Sites 1262 and 1263 fromMcCarren et al. [2008], Site 690 record from Kennett and Stott [1991] and Kelleyet al. [2005]. Bulk isotope record from Site 1263, shown for comparison, fromZachos et al. [2005]. Depth of section for Site 401 is plotted on the left axis andrelative time before and after the PEB (Paleocene-Eocene Boundary) on theright axis. Ages for Sites 1262, 1263, and 690 from R€ohl et al. [2007]. vPDB,Vienna PeeDee Belemnite.

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1042

measurements for Site 1258 were made simultaneously during stable isotope measurements on the VGMicromass Optima ratio mass spectrometer with precision of 66%.

Bulk stable carbon isotopes were measured on 33 homogenized and crushed sediment samples from ODPSite 1258. Analyses were performed on the VG Micromass Optima ratio mass spectrometer at the Universityof California, Santa Cruz, USA. Results are reported in & relative to Vienna PeeDee Belemnite (vPDB) andcalibrated via NBS-19. Precision of d13C measurements is better than 0.1&.

3. Results

The results of bulk sample elemental analyses are presented in supporting information Tables S1–S15 andFigure S1. Average bulk Mn concentrations range from �3 lmol g21 at Site 738 to �171 lmol g21 at Site1221. Average U concentrations range from �0.5 nmol g21 at Site 738 to �5 nmol g21 at Site 1258. We nor-malize our results to Ti to isolate the enrichment over crustal values and to provide downcore comparisonsamong the sites. Trace metal enrichment factors (EF) for Mn and U were calculated as EF 5 (metal/Ti)sample/(metal/Ti)crust using the bulk crustal ratios Mn/Ti (mol/mol) 5 0.156 and U/Ti (mmol/mol) 5 0.061 [Rudnickand Gao, 2003].

Mn enrichment factors exhibit variation across the PETM. At the Atlantic and Southern Ocean Sites 401,1258, 690, and 738, Mn EF for bulk sample digestions have average values between 2 and 6 preevent andgenerally decrease during the core of the event to average values between 1 and 4 (Figure 4). Then in thecarbon isotope recovery, Mn EF for bulk sample digestions are at average values between 2 and 20. Site690 displays a gradual peak with increasing values from �95 kyr from onset of the PEB to a peak value of37 at �106 kyr, then a gradual decline. At the Pacific Sites 1221 and 1209, Mn EF for bulk sample digestionsaverage between 2 and 13 preevent, 21 and 47 during the core of the event, and 15 and 17 in the recovery.Site 1221 exhibits a sharp peak in Mn EF at �20 kyr to values of 350, lasting �4 kyr.

We compare Mn EF from bulk sediment digested before and after a reductive cleaning procedure to deter-mine the presence or absence of Mn oxides [Chun et al., 2010]. When Mn EF are >1 in bulk sediment beforethe reductive cleaning procedure and are then at crustal averages (Mn �1) after the cleaning and there isno U enrichment, the difference is interpreted to be Mn oxides representing an oxygenated environment.When Mn EF are >1 in bulk sediment before the reductive cleaning and remain relatively unchanged afterthe cleaning, this represents Mn carbonates (MnCO3), indicating an early diagenetic environment of suboxicpore waters.

At the Atlantic and Southern Ocean Sites 401, 1258, and 738 Mn EF on the subset of samples that werereductively cleaned prior to total digestion show little change compared to Mn EF before reduction. Site738 displays Mn EF values from reductively cleaned samples that are slightly higher than the untreated bulksamples. We suspect that this was due to sample loss during the reductive cleaning procedure as we didnot weigh samples again after the cleaning and prior to the sediment digestion. Average values at thesesites are between 1 and 7 before and after reduction. At Site 690, the gradual peak in Mn EF remainsunchanged in samples after reductive cleaning. Mn EF values postreductive cleaning at Site 690 are from�10 to 30 between �90 and 150 kyr relative to the PEB. At the two Pacific Sites 1221 and 1209, Mn/Ti onthe reductively cleaned samples are reduced to crustal averages throughout the sample section. At Site

Table 3. Analytical Figures of Merit

Element Concentration (lmol/g sediment)

Al Ti Mn Ua

Detection limitsb 18.52 0.32 0.78 0.44Reproducibility Leg 202 (consistency standard replicates)c 132.98 6 23 2.4 6 0.4 6.1 6 0.7 1.09 6 0.28Reproducibility Leg 208 (consistency standard replicates)c 154.26 6 22 4.71 6 0.81 4.82 6 0.55 0.30 6 0.15

anmol/g sediment.bDefined as three times the standard deviation of replicate measures of a blank of the same matrix as each digestion, and expressed

in equivalent concentration for a typical size sediment sample.cReproducibility defined as 61 standard deviation of multiple replicates of a solid sample included in each analytical run (n 5 14).

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1043

1221, the sharp Mn EF peak at �20 kyr is reduced to near crustal averages, Mn EF �3, postreductivecleaning.

U enrichment factors remain at or near crustal averages (U EF �2) at all sites during the PETM, with theexception of Site 690 (Figure 5). U EF are near crustal averages preevent and during the core at Site 690.Then, starting at �90 kyr from the onset of the PEB, U EF increase to a gradual peak of 90 at �106 kyr, coe-val with the Mn EF peak, then return to crustal averages.

As a part of this study, we also measured CaCO3 concentrations for Sites 401 and 1258. CaCO3 concentra-tions for Site 401 range between 56 and 68 wt % preevent (Figure 6). At the onset of the PETM, values dropto 45 wt % and remain in the range of 42–48 wt %. Just prior to the recovery interval, CaCO3 concentrationsdecrease to 28 wt % and remain between 30 and 40 wt % throughout the recovery. CaCO3 concentrationsfor Site 1258 average 33 wt % pre-CIE, then dropped to values of <1 wt % during the core of the event, andgradually return to values between 30 and 55 wt % in the recovery (Figure 6). Throughout the record,CaCO3 concentrations remain <55 wt %. We also generated bulk carbon and oxygen isotope data for Site1258. Bulk d13C values measured at Site 1258 average 1.58& pre-CIE, then decrease to values as low as26.3& at the base of the clay layer. During the recovery, d13C values average 1.02& (Figure 7).

4. Discussion

4.1. Paleoredox ChangesComparing Mn EF before and after a reductive cleaning procedure, combined with U EF measurements,shows paleoredox changes recorded in marine sediments across a global set of sites that record the PETM(Table 4). In the following, we evaluate how the geographic response of each site contributes to our under-standing of how the oceans responded to this abrupt global warming event.

Figure 4. Manganese enrichment factors (Mn EF) calculated from measured Mn concentrations and the crustal molar Mn/Ti of 0.156 fromRudnick and Gao [2003], plotted versus relative time before and after the PEB (Paleocene-Eocene Boundary) [R€ohl et al., 2007]. Scales forMn EF vary for each site. Site number is located at the top of each plot. Sites 1262, 1263, and 1266 are from Chun et al. [2010]. Dashed verti-cal line is EF 5 1 (no enrichment or depletion relative to presumed crustal source). Closed symbols indicate EF for bulk digests withoutreductive cleaning. Open symbols indicate EF postreductive cleaning. Reductively cleaned samples are at lower resolution than the bulkdigests. Gray shaded region indicates duration of core-carbon isotope excursion (CIE) �80 kyr.

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1044

-50

0

50

100

150

200

250

0 2 4 5 7

401

0 2 4 5 7

1258

0 2 4 5 7

126212631266

0 25 50 75100

690

0 2 4 5 7

738

0 2 4 5 7

1221

Tim

e re

lativ

e to

PE

B (k

yr)

Uranium Enrichment Factor

Tim

e re

lativ

e to

PE

B (k

yr)

0 2 4 5 7-50

0

50

100

150

200

2501209

Figure 5. Uranium enrichment factors (U EF) calculated from measured U concentrations and the crustal molar U/Ti of 0.061 from Rudnickand Gao [2003], plotted versus relative time before and after the PEB (Paleocene-Eocene Boundary) [R€ohl et al., 2007]. Site number islocated at the top of each plot. Sites 1262, 1263, and 1266 are from Chun et al. [2010]. Dashed vertical line is EF 5 1 (no enrichment ordepletion relative to presumed crustal source). Gray shaded region indicates duration of core-carbon isotope excursion (CIE) �80 kyr.

Figure 6. Bulk sediment weight % carbonate content plotted versus time (kyr) relative to the Paleocene-Eocene Boundary (PEB) using theage model of R€ohl et al. [2007]. Site number is located at the top of each plot. Site 401 (this study), Site 1258 (this study), Sites 1262, 1263,and 1266 from Zachos et al. [2005], Site 690 from Farley and Eltgroth [2003] (smoothed line); Kelly et al. [2005] (open triangles), and thisstudy (black X). Site 1221 from Murphy et al. [2006] and Site 1209 from Colossimo et al. [2005]. Note change in axis scale for Sites 690 and1209. Gray shaded region indicates duration of core-carbon isotope excursion (CIE) �80 kyr.

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1045

We have previously demonstratedthe utility of these proxies to pro-vide a paleoredox history on adepth transect on Walvis Ridgeduring the PETM [Chun et al., 2010].We found that prior to thePaleocene-Eocene Boundary (PEB),sites deeper than 2500 m paleo-water depth were oxygenatedwhile the shallow 1500 m site wassuboxic. During the core-CIE, theoxygen minimum zone expandedto include our deepest site (3600 mpaleowater depth), then returnedto preevent conditions during therecovery.

In the modern oceans, methanehydrates are predominately storedon continental margins at interme-diate water depths (�500–1000 m).DSDP Site 401 in the Bay of Biscayrepresents the northernmost Atlan-tic end-member in our global arrayof PETM sites. We selected interme-

diate water depth sites from the Atlantic basin to test the theory of methane hydrate dissociation in theNorth Atlantic [Dickens et al., 1995; Dickens, 2011]. Prior to the CIE, Mn EF at Site 401 are >1 and show nochange with reductive treatment, indicating the presence of Mn-carbonates (Figure 4). During the core ofthe CIE and recovery intervals, Mn EF values average <2 while U EF is �2 (Figure 5) throughout the intervalsuggesting that Site 401 pore waters became more reducing and Mn-carbonates were no longer produced.Thus, inferring relatively lower oxygen concentrations in the overlying deep water. Our results are in agree-ment with planktic foraminiferal and ostracode assemblage studies across the PETM at this site, whichrecord an increase in low oxygen tolerant taxa in the core of the CIE [Pardo et al., 1997; Yamaguchi and Nor-ris, 2012].

Calcium carbonate at Site 401 displayed a unique pattern not measured at the other PETM sites (Figure 6).The initial decrease in calcium carbonate at the onset of the CIE was expected as the global calcite compen-sation depth (CCD) shoaled in response to rapid acidification of the oceans due to the release of carbon to

Figure 7. Bulk carbon isotope records for Site 1258 plotted versus depth (meters com-posite depth) and time (kyr) relative to the Paleocene-Eocene Boundary (PEB) usingthe age model of R€ohl et al. [2007]. vPDB, Vienna PeeDee Belemnite.

Table 4. Summary of Proxy Results (Mean Values)

LocationSite (Leg)

Bay ofBiscaya

401 (48)

DemeraraRiseb

1258 (207)

Walvis Ridgec

Maud Rised

690 (113)

KerguelenPlateaue

738 (119)

EquatorialPacificf

1221 (199)

ShatskyRiseg

1209 (198)1262 (208) 1263 (208) 1266 (208)

Pre-CIE (250 to 0 kyr)Mn EF before reduction 5.81 1.51 5.37 2.49 5.33 3.33 5.69 2.52 12.71Mn EF after reduction 6.31 1.28 1.37 3.22 2.05 4.33 6.15 NA 6.28U EF 1.62 2.96 1.22 2.52 1.17 1.46 2.04 2.25 4.26Redox state of bottom waters Suboxic Suboxic Oxic Suboxic Oxic Suboxic Suboxic Oxic OxicCore-CIE (0–80 kyr)Mn EF before reduction 1.84 0.91 2.66 1.09 3.30 2.75 4.50 46.97 21.24Mn EF after reduction 2.52 0.79 0.88 1.20 1.24 3.35 4.38 2.08 6.16U EF 1.53 1.97 1.05 1.11 1.20 1.69 1.59 2.06 3.06Redox state of bottom waters Suboxic Suboxic Suboxic Suboxic Suboxic Suboxic Suboxic Oxic OxicRecovery (80–170 kyr)Mn EF before reduction 1.58 2.65 5.57 7.28 5.52 19.50 6.83 17.17 19.06Mn EF after reduction 1.17 2.34 1.81 8.41 2.22 15.24 7.97 1.70 8.02U EF 1.22 2.68 1.50 1.30 1.39 46.91 1.76 2.11 3.26Redox state of bottom waters Suboxic Suboxic Oxic Suboxic Oxic More suboxic Suboxic Oxic Oxic

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1046

the oceans and atmosphere. However, at most of the other measured PETM sites, calcium carbonatereturned to pre-CIE concentrations (or overshoots them) in the recovery as the source of light carbon ceasesand the CCD gets deeper again [Kelly et al., 2005; Zachos et al., 2005; Zeebe and Zachos, 2007]. Site 401exhibited the initial decrease in calcium carbonate at the onset of the CIE, but then calcium carbonateremained low for the remainder of the event and in the recovery (between 30 and 50 wt % calcium carbon-ate). One possibility could be that this site was situated at a water depth near the lysocline during and afterthe PETM, such that dissolution began to occur, but was not sufficient to form a clay layer.

Demerara Rise, ODP Site 1258, exhibits stable carbon isotopes of bulk carbonate that have more negativevalues than any of the PETM records in our study (Figures 7 and supporting information Figure S1). We sus-pect that these negative values in the clay layer are a result of early diagenesis under sulfate reduction anddo not reflect the primary signal. Sulfate reduction occurs at this site today [Erbacher et al., 2004] and giventhe location and sedimentation rates, might have altered the primary signal. Nunes and Norris [2006] alsosuggested diagenetic overprinting of the d13C values of benthic foraminifera from across the PETM atDemerara Rise because of the unusually low values and observations of calcite overgrowths on benthic for-aminifer tests.

The geographical proximity of Demerara Rise to ODP Sites 999 and 1001 in the Caribbean, which alsoexhibit thick clay layers and very low bulk carbonate d13C values at the PEB [Bralower, 1997], indicates thatthis diagenetic feature could be related to a regional influence. The paleodepth of the two Caribbean sites,2000–2500 m [R€ohl and Abrams, 2000], are similar to that of Site 1258 during the PETM, which means theymay have been influenced by the same water mass(es). With intensification of the O2 minimum zone atintermediate depths in the region, all three sites would have experienced enhanced sulfate reduction dur-ing the PETM, possibly to the point of driving carbonate authigenesis. Given the lack of biogenic carbonatepreservation, this authigenic carbonate dominates the carbon isotope signature of the bulk carbonate, lead-ing to the anomalously low values. Our EF data suggest that Site 1258 was close to suboxic throughout thelate Paleocene-early Eocene. Mn EF before and after reductive cleaning displays no change in the pre-CIEand recovery intervals, indicating the presence of authigenic MnCO3 formation (Figure 4). Mn EF values areat crustal averages during the core-CIE. U EF are slightly enriched (�3) during the sampled interval, also sug-gesting suboxic conditions throughout the PETM (Figure 5). Thus, just a slight drop in dissolved O2 levelswould have been sufficient to result in sulfate reduction.

ODP Site 690 located on Maud Rise in the Southern Ocean represents our southernmost end-member ofthe South Atlantic. Both Mn EF and U EF exhibit large broad peaks starting at �90 kyr after the onset of thePEB and lasting �50 kyr (Figures 4 and 5). The U EF peak in the range of 65–90 is the largest U EF measuredin all our global PETM Sites. Combined with Mn EF values in the range �20–35 that show no change afterreductive cleaning, these indicate highly suboxic conditions during the recovery phase at this site. Benthicforaminifera assemblage studies at 690 also indicate low oxygen conditions [Thomas, 2007].

Large Mn and U peaks in the recovery are coincident with a large shift in sedimentation rates at this site,identified through the accumulation of extraterrestrial 3He [Farley and Eltgroth, 2003] (Figure 8e). Sedimen-tation rates are �2 cm/kyr pre-CIE, within the core of the CIE they are �1.5 cm/kyr, and double in the recov-ery to �3 cm/kyr [R€ohl et al., 2007]. The increase in sedimentation rate for the recovery is considered part ofa negative feedback mechanism, brought on by enhanced continental weathering and runoff in responseto increased warming during the CIE [Dickens et al., 1997; Kelly et al., 2010], (Figures 8a–8d).

Our paleoredox data for Site 690 have important implications for hypotheses of local marine productivitychanges during the PETM. Specifically, an increase in mass accumulation rates (MAR) of biogenic barite atthis site (Figure 8f) has been proposed as evidence of an increase in local productivity and the rate of thebiological pump, thus acting as a strong negative feedback for the drawdown of atmospheric CO2 [Bainset al., 2000]. Increases in Sr/Ca in coccoliths also indicate a rise in growth rates and productivity [Stoll andBains, 2003; Stoll et al., 2007]. However, it has been suggested that the proposed peak in barite accumula-tion is not productivity related but instead caused by the thermal dissociation of gas hydrates. Dissociationof gas hydrates would create sulfate-reducing conditions prompting dissociation of barite crystals and rep-recipitation of authigenic barite higher in the sedimentary column [Dickens et al., 2003]. Other data such asa shift from nannoplankton assemblages indicating productive surface waters to more oligotrophic condi-tions are also in conflict with the notion of enhanced productivity during the CIE [Bralower, 2002; Kelly,

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1047

2002]. Finally, the peaks in biogenic barite and Sr/Ca occur during the core of the CIE and not in the recov-ery (Figure 8f) [Torfstein et al., 2010].

Our trace metal EFs indicate bottom water at Site 690 during the core of the CIE and recovery was suboxic(Figures 8g and 8h). We interpret the biogenic barite peak, which is coincident with the peak in Mn and UEF (Figure 8f), as an artifact of the sudden change in accumulation rates and bottom water redox conditions.This is somewhat analogous to the U EF peak documented at Walvis Ridge Site 1263 (Figure 4) [Chun et al.,2010], and is in agreement with modeling profiles of interstitial sulfate concentrations and barite fronts,where peaks in barite occur immediately below the interval of sulfate reduction [Dickens, 2001; Dickenset al., 2003]. Modern interstitial sulfate concentrations at this site for the depth of the PETM are �22 mM[Barker and Kennett, 1988] indicating sulfate reduction given modern seawater sulfate averages �28 mM.However, if average seawater Eocene sulfate concentrations were lower (14–23 mM) [Horita et al., 2002],then this site would not be sulfate reducing during the Eocene. Even if there is sulfate depletion in the porewaters presently, this represents sulfate reduction some time after initial deposition of the sediments.Finally, if paleoredox conditions were sulfate reducing, then barite would have been remobilized and possi-bly reprecipitated in nonsulfate-reducing depth horizons [Paytan et al., 2002].

The paleoredox features of ODP Site 738 on Kerguelen Plateau (paleodepth �1350 m [Barrera and Huber,1991]) resemble those of Walvis Ridge, ODP Site 1263; both are at intermediate paleowater depths and sub-oxic throughout the PETM. However, unlike Site 1263, Site 738 exhibits a gradual increase in Mn EF (averagevalue �4) during the core-CIE (Figure 4). Calcium carbonate concentrations at Site 738 do not go to zero,with minimum values of 70 wt % [Larrasoa~na et al., 2012], similar to Site 690, indicating this site remained

Figure 8. Geochemical data from ODP Site 690 plotted versus relative time before and after the PEB (Paleocene-Eocene Boundary) fromR€ohl et al. [2007]. (a) Carbon isotope record from Bains et al. [1999], vPDB, Vienna PeeDee Belemnite. (b) Bulk sediment weight % carbonatecontent from Farley and Eltgroth [2003] (smoothed line); Kelly et al. [2005] (open triangles), and this study (black X). (c) Weight percentcoarse fraction showing decrease of foraminiferal shells during recovery interval from Kelly et al. [2005]. (d) Changes in planktic foraminif-eral shell fragmentation from Kelly et al. [2005]. (e) 3HeET displays an increase in flux pre and core-CIE and a decrease in the recovery inter-val, from Farley and Eltgroth [2003]. (f) biogenic-barium mass accumulation rate from Bains et al. [2000]. (g) Uranium enrichment factors (UEF) calculated from measured U concentrations and the crustal molar U/Ti of 0.061 from Rudnick and Gao [2003]. Dashed vertical line isEF 5 1 (no enrichment or depletion relative to presumed crustal source). (h) Manganese enrichment factors (Mn EF) calculated from meas-ured Mn concentrations and the crustal molar Mn/Ti of 0.156 from Rudnick and Gao [2003]. Closed symbols indicate samples measuredbefore reductive cleaning. Open symbols indicate samples measured after reductive cleaning. Dashed vertical line is EF 5 1 (no enrich-ment or depletion relative to presumed crustal source). Gray shaded region indicates duration of core-carbon isotope excursion (CIE) �80kyr.

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1048

above the CCD throughout the sampled interval. Benthic foraminifera fauna also suggest decreased oxygenconcentrations at this site [Thomas, 1998]

ODP Site 1221 located in the equatorial Pacific is one of two Pacific sites in our global study of PETM loca-tions. Mn EF values of 185–350 during the core of the CIE are the highest measured in any of our PETM sec-tions (Figure 4). They were most likely caused by the presence of small manganese nodules as confirmed byshipboard measurements and visual inspection of the sediment [Lyle et al., 2002]. Biogenic barite crystalswere extracted from this section and close inspection of their morphology and sulfur isotopes do not sug-gest they were precipitated within the sediments during early diagenesis [Faul and Paytan, 2005]. Theseindicators suggest the sediments were buried in an oxygenated environment with no postdepositionalalteration.

Our geochemical results at ODP Site 1209 on Shatsky Rise indicate oxic bottom waters throughout thePETM and are contrary to benthic foraminiferal assemblage changes [Takeda and Kaiho, 2007]. The benthicforaminifera oxygen indicators display a shift in the percentage of ‘‘oxic’’ benthic foraminifera during thecore-CIE interval and suggest decreased oxygen levels caused the benthic foraminifera extinction. The testsize of planktonic foraminifera also increased during the core of the CIE, a phenomenon that is attributed tosurface water warming and enhanced stratification and oligotrophic conditions [Takeda and Kaiho, 2007;Zachos et al., 2003], a conclusion that is in agreement with independent studies of phytoplankton assemb-lages [Gibbs et al., 2006] and Sr/Ca measurements in individual coccoliths [Stoll et al., 2007].

While our paleoredox indicators disagree with the benthic foraminifera assemblage study at Site 1209, wealso point out that low sedimentation rates at this site (�0.25 cm/kyr) and oligotrophic surface water condi-tions during the PETM would favor oxygenated bottom waters and surface sediments. Reduced sedimenta-tion and decreased export of organic matter would lead to lower rates of organic matter oxidation andhigher bottom water oxygen concentrations. One alternative mechanism for the benthic extinction event atSite 1209 is decreased food supply [Thomas, 1998, 2003], with bottom waters remaining oxygenated. Theother is a short-lived episode of suboxic conditions that was not captured by our redox indicators.

4.2. Testing the Theory of Methane Hydrate Release at the PETMThe oxygen concentration of bottom waters is sensitive to changes in methane oxidation, deep-ocean ven-tilation and circulation, and primary productivity. Separating the contributions of each of these factors tothe oxygen concentration of bottom waters at any given location, however, requires additional constraintsthat do not yet exist for this interval. Methane hydrate dissociation and oxidation as a cause for the PETM istestable if, for example, single specimen carbon isotopes of planktonic and benthic foraminifera from ODPSite 690 have been used to infer a sequence of carbon release and oxidation that would be consistent withdirect emission to the atmosphere followed by gradual mixing into the deep sea [Thomas et al., 2002]. Ourstudy is the first to use redox-sensitive trace metals to assess bottom water oxygen concentrations acrossthe PETM at multiple locations in different ocean basins. With our spatial array of sites it should be possibleto distinguish local from global effects.

In theory, methane oxidation should occur first and most rapidly in the reservoir where methane is released[Dickens, 2000]. Previous empirical and theoretical observations point toward a possible Atlantic source ofmethane release during the PETM [McCarren et al., 2008; Zeebe and Zachos, 2007]. Our redox proxies indi-cate Atlantic intermediate water depths (Sites 1258 and 1263) experienced suboxic bottom waters duringthe PETM (Figures 6 and 7 and Table 4), and were thus most proximal to the methane source. The otherintermediate water depth Atlantic locations (Sites 401 and 690) were also suboxic throughout the PETM,but calcium carbonate at these sites indicate they remained above the CCD (Figure 4), therefore furtherfrom the source of isotopically light carbon. Below the CCD, the deeper Atlantic Walvis Ridge Sites (Sites1262 and 1266) remained well oxygenated both pre-CIE and in the recovery, suboxic only during the core-CIE [Chun et al., 2010]. Intermediate water depth sites (401 and 690) remain suboxic throughout the recov-ery, much longer than the direct influence of methane oxidation. We suggest these sites were situated inan oxygen minimum zone throughout the duration of the PETM. Ideally, comparing multiple sites on adepth transect (such as Walvis Ridge) would test this idea.

Moving away from the Atlantic, our intermediate depth (Site 738) in the Indian Ocean sector of the South-ern Ocean was also bathed by suboxic bottom waters, which is consistent with findings based on a

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1049

bioturbation index of bottom water oxygen in South Pacific intermediate water sites [Nicolo et al., 2010].Sites further north in the Pacific Ocean (Sites 1221 and 1209) experienced oxygenated bottom watersthroughout the PETM and thus were most distal from an inferred methane source. In short, these patternscould be attributed partly to methane oxidation, though it is likely that the oxygen minimum zoneexpanded as well in response to factors other than methane oxidation. Recent modeling work has shownthat with increasing atmospheric CO2 during the PETM, there was enhanced stratification and a decline inventilation of the Atlantic and Pacific, with the Atlantic deep waters more depleted in oxygen than Pacificdeep waters [Winguth et al., 2012].

We have used Mn and U EF paleoredox proxies to directly test the methane hydrate hypothesis during thePETM. However, these deep-ocean redox proxies are influenced by both methane oxidation as well aschanges in ocean circulation and ventilation. There are conflicting observational evidence as to whetherthere was a switch to Northern Hemisphere overturning at the start of the PETM [Nunes and Norris, 2006] ora predominant Southern Hemisphere overturning throughout the PETM [Thomas et al., 2003]. Recent mod-eling studies suggest predominantly Southern Hemisphere sources with sluggish overturning associatedwith deep-ocean warming [Lunt et al., 2011; Winguth et al., 2010]. We cannot form conclusions about deep-ocean circulation with our proxies alone.

In testing the size and pattern of carbon release during the PETM, Panchuk et al. [2008] have been able toreproduce global patterns of calcium carbonate concentrations by reducing bioturbation in the Atlanticduring the CIE. Our interpretation of decreased bottom water oxygen concentrations in the Atlantic relativeto the Pacific would also imply a lower intensity of bioturbation in the Atlantic, in agreement with theparameters set for their sediment model.

5. Conclusions

Using paleoredox conditions from a range of global sites, we infer the presence or absence of oxygenatedbottom waters and comment upon the possible release of methane from gas hydrates. Evidence from Mnand U EF in marine sediments indicate that North Atlantic intermediate water depth sites were the mostsuboxic during the PETM. Coupled with the pattern of CaCO3 during this interval, we suggest the source ofmethane release was located in the North Atlantic. Both of our Pacific Ocean sites are oxygenated before,during, and after the PETM. Therefore, we suggest the Pacific was not the source of methane release. Ourinterpretation of the present data reveals that intermediate water depth sites in the Atlantic are the mostsuboxic and likely closest to the site of methane hydrate release, while sites that are well oxygenated arefurthest from the source (Pacific locations). Furthermore, our study shows that Mn and U EF can be appliedas qualitative paleoredox proxies across transient climate states such as the PETM.

ReferencesBains, S., R. M. Corfield, and R. D. Norris (1999), Mechanisms of climate warming at the end of the Paleocene, Science, 285, 724–727, doi:

10.1126/science.285.5428.724.

Bains, S., R. D. Norris, R. M. Corfield, and K. L. Faul (2000), Termination of global warmth at the Palaeocene/Eocene boundary through pro-ductivity feedback, Nature, 407(6801), 171–174, doi:10.1038/35025035.

Barker, P. F., and J. P. Kennett (1988), Proceedings of the Ocean Drilling Program, Initial Rep., vol. 113, Ocean Drill. Program, College Station,Tex., doi:10.2973/odp.proc.ir.113.1988.

Barrera, E., B. T. Huber (1991), Paleogene and early Neogene oceanography in the southern Indian Ocean: Leg 119, foraminifer stable iso-tope results, in Proceedings of the Ocean Driling Program, Sci. Results, edited by J. A. Barron et al., Ocean Drill. Program, College Station,Tex., pp. 693–717.

Barron, J., B. Larsen, et al. (1989), Proc. ODP. Init. Repots, College Station, TX (Ocean Drilling Program), 119.

Boyle, E. A. (1983), Manganese carbonate overgrowths on foraminifera tests, Geochim. Cosmochim. Acta, 47(10), 1815–1819, doi:10.1016/0016-7037(83)90029-7.

Bralower, T. J. (1997), High-resolution records of the late Paleocene thermal maximum and circum-Caribbean volcanism: Is there a causallink?, Geology, 25, 963–966, doi:10.1130/0091-7613(1997)025<0963:HRROTL>2.3.CO;2.

Bralower, T. J. (2002), Evidence of surface water oligotrophy during the Paleocene-Eocene thermal maximum: Nannofossil assemblagedata from Ocean Drilling Program Site 690, Maud Rise, Weddell Sea, Paleoceanography, 17(2), doi:10.1029/2001PA000662.

Calvert, S. E., and T. F. Pedersen (1993), Geochemistry of recent oxic and anoxic marine sediments: Implications for the geological record,Mar. Geol., 113(1–2), 67–88, doi:10.1016/0025-3227(93)90150-T.

Chun, C. O. J., M. L. Delaney, and J. C. Zachos (2010), Paleoredox changes across the Paleocene-Eocene thermal maximum, Walvis Ridge(ODP Sites 1262, 1263, and 1266): Evidence from Mn and U enrichment factors, Paleoceanography, 25, PA4202, doi:10.1029/2009PA001861.

AcknowledgmentsWe thank R. Franks and the IMS foranalytical support. E. Stokes made thecarbonate measurements for Site 401and B. Murphy assisted with stableisotope measurements for Site 1258. B.Young and W. Kordesch provided helpin sample preparation. C.P. thanks R.Zeebe for pointing out the criticalneed for a global redox PETM study. B.Gill and an anonymous reviewer arethanked for their careful comments onthe manuscript. This research usedsamples and/or data provided by theOcean Drilling Program (ODP) andIntegrated Ocean Drilling Program(IODP). IODP is sponsored by the U.S.National Science Foundation (NSF) andparticipating countries undermanagement of the Consortium forOcean Leadership. Funding for thisresearch was provided by NSF grantEAR 0120727 to J.C.Z. and M.L.D. C.P.was also supported by GAANN andARCS fellowships through UCSC OceanSciences and a ‘‘Young scientistsin Focus’’ research grant at theGoethe-University Frankfurt.

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1050

Colosimo, A. B., T. J. Bralower, and J. C. Zachos (2006), Evidence for lysocline shoaling at the Paleocene/Eocene Thermal Maximum on Shat-sky Rise, northwest Pacific, in Proc. ODP, Sci. Results, 198 [Online], edited by Bralower, T. J., Premoli Silva, I., and Malone, M. J. Availablefrom World Wide Web: <http://www-odp.tamu.edu/publications/198_SR/112/112.htm>.

DeConto, R. M., S. Galeotti, M. Pagani, D. Tracy, K. Schaefer, T. Zhang, D. Pollard, and D. J. Beerling (2012), Past extreme warming eventslinked to massive carbon release from thawing permafrost, Nature, 484(7392), 87–91, doi:10.1038/nature10929.

Dickens, G. R. (2000), Methane oxidation during the late Palaeocene thermal maximum, Bull. Soc. Geol. Fr., 171(1), 37–49.Dickens, G. R. (2001), Sulfate profiles and barium fronts in sediment on the Blake Ridge: Present and past methane fluxes through a large

gas hydrate reservoir, Geochim. Cosmochim. Acta, 65(4), 529–543, doi:10.1016/S0016-7037(00)00556-1.Dickens, G. R. (2003), Rethinking the global carbon cycle with a large, dynamic and microbially mediated gas hydrate capacitor, Earth

Planet. Sci. Lett., 213(3–4), 169–183, doi:10.1016/S0012-821X(03)00325-X.Dickens, G. R. (2011), Down the Rabbit Hole: Toward appropriate discussion of methane release from gas hydrate systems during the

Paleocene-Eocene thermal maximum and other past hyperthermal events, Clim. Past, 7(3), 831–846, doi:10.5194/cp-7-831-2011.Dickens, G. R., J. R. O’Neil, D. K. Rea, and R. M. Owen (1995), Dissociation of oceanic methane hydrate as a cause of the carbon isotope

excursion at the end of the Paleocene, Paleoceanography, 10, 965–971, doi:10.1029/95PA02087.Dickens, G. R., M. M. Castillo, and J. C. G. Walker (1997), A blast of gas in the latest Paleocene: Simulating first-order effects of massive disso-

ciation of oceanic methane hydrate, Geology, 25(3), 259–262, doi:10.1130/0091-7613.Dickens, G. R., T. Fewless, E. Thomas, and T. J. Bralower (2003), Excess barite accumulation during the Paleocene-Eocene thermal maximum:

Massive input of dissolved barium from seafloor gas hydrate reservoirs, Spec. Pap. Geol. Soc. Am., 369, 11–23, doi:10.1130/0-8137-2369-8.11.

Dickson, A. J., A. S. Cohen, and A. L. Coe (2012), Seawater oxygenation during the Paleocene-Eocene thermal maximum, Geology, 40(7),639–642, doi:10.1130/g32977.1.

Ellis, R., J. Pine, and J. M. Gieskes (1979), Interstitial Water Studies, Leg 48, Initial Reports of the Deep Sea Drilling Project, 48, 297–303, doi:10.2973/dsdp.proc.48.110.1979.

Erbacher, J., D. C. Mosher, and M. J. Malone (2004), Proceedings of the Ocean Driling Program, Initial Rep., vol. 207, Ocean Drill. Program, Col-lege Station, Tex., doi:10.2973/odp.proc.ir.207.2004.

Farley, K. A., and S. F. Eltgroth (2003), An alternative age model for the Paleocene-Eocene thermal maximum using extraterrestrial 3He,Earth Planet. Sci. Lett., 208(3–4), 135–148, doi:10.1016/S0012-821X(03)00017-7.

Faul, K. L., and A. Paytan (2005), Phosphorus and barite concentrations and geochemistry in Site 1221 Paleocene/Eocene boundary sedi-ments, in Proceedings of the Ocean Drilling Program, Sci. Results, edited by P. A. Wilson, M. Lyle, and J. V. Firth, pp. 1–23, Ocean Drill. Pro-gram, College Station, Tex.

Froelich, P. N., G. P. Klinkhammer, M. L. Bender, N. A. Luedtke, G. R. Heath, D. Cullen, P. Dauphin, D. Hammond, B. Hartman, and V. Maynard(1979), Early oxidation of organic matter in pelagic sediments of the eastern equatorial Atlantic: Suboxic diagenesis, Geochim. Cosmo-chim. Acta, 43(7), 1075–1090, doi:10.1016/0016-7037(79)90095-4.

Gibbs, S. J., T. J. Bralower, P. R. Bown, J. C. Zachos, and L. M. Bybell (2006), Shelf and open-ocean calcareous phytoplankton assemblagesacross the Paleocene-Eocene thermal maximum: Implications for global productivity gradients, Geology, 34(4), 233–236, doi:10.1130/G22381.1.

Gingele, F. X., and S. Kasten (1994), Solid-phase manganese in Southeast Atlantic sediments: Implications for the paleoenvironment, Mar.Geol., 121(3–4), 317–332, doi:10.1016/0025-3227(94)90037-X.

Higgins, J. A., and D. P. Schrag (2006), Beyond methane: Towards a theory for the Paleocene-Eocene thermal maximum, Earth Planet. Sci.Lett., 245(3–4), 523–537, doi:10.1016/j.epsl.2006.03.009.

Horita, J., H. Zimmermann, and H. D. Holland (2002), Chemical evolution of seawater during the Phanerozoic: Implications from the recordof marine evaporites, Geochim. Cosmochim. Acta, 66(21), 3733–3756, doi:10.1016/S0016-7037(01)00884-5.

Kaiho, K., K. Takeda, M. R. Petrizzo, and J. C. Zachos (2006), Anomalous shifts in tropical Pacific planktonic and benthic foraminiferal testsize during the Paleocene-Eocene thermal maximum, Palaeogeography, Palaeoclimatology, Palaeoecology, 237(2–4), 456–464, doi:10.1016/j.palaeo.2005.12.017.

Kelly, D. C. (2002), Response of Antarctic (ODP Site 690) planktonic foraminifera to the Paleocene-Eocene thermal maximum: Faunal evi-dence for ocean/climate change, Palaeoceanography, 17 (4), 1071, doi:10.1029/2002PA000761.

Kelly, D. C., J. C. Zachos, T. J. Bralower, and S. A. Schellenberg (2005), Enhanced terrestrial weathering/runoff and surface ocean carbonateproduction during the recovery stages of the Paleocene-Eocene thermal maximum, Paleoceanography, 20, PA4023, doi:10.1029/2005PA001163.

Kelly, D. C., T. M. J. Nielsen, H. K. McCarren, J. C. Zachos, and U. R€ohl (2010), Spatiotemporal patterns of carbonate sedimentation in theSouth Atlantic: Implications for carbon cycling during the Paleocene-Eocene thermal maximum, Palaeogeogr. Palaeoclimatol. Palaeoe-col., 293(1–2), 30–40, doi:10.1016/j.palaeo.2010.04.027.

Kennett, J. P., and L. D. Stott (1991), Abrupt deep-sea warming, palaeoceanographic changes and benthic extinctions at the end of thePalaeocene, Nature, 353, 225–229, doi:10.1038/353225a0.

Kurtz, A. C., L. R. Kump, M. A. Arthur, J. C. Zachos, and A. Paytan (2003), Early Cenozoic decoupling of the global carbon and sulfur cycles,Paleoceanography, 18(4), 1090, doi:10.1029/2003PA000908.

Larrasoa~na, J. C., A. P. Roberts, L. Chang, S. A. Schellenberg, J. D. Fitz Gerald, R. D. Norris, and J. C. Zachos (2012), Magnetotactic bacterialresponse to Antarctic dust supply during the Palaeocene-Eocene thermal maximum, Earth Planet. Sci. Lett., 333–334, 122–133, doi:10.1016/j.epsl.2012.04.003.

Lourens, L. J., A. Sluijs, D. Kroon, J. C. Zachos, E. Thomas, U. Rohl, J. Bowles, and I. Raffi (2005), Astronomical pacing of late Palaeocene toearly Eocene global warming events, Nature, 435(7045), 1083–1087, doi:10.1038/nature03814.

Lunt, D. J., P. J. Valdes, T. D. Jones, A. Ridgwell, A. M. Haywood, D. N. Schmidt, R. Marsh, and M. Maslin (2010), CO2-driven ocean circulationchanges as an amplifier of Paleocene-Eocene thermal maximum hydrate destabilization, Geology, 38(10), 875–878, doi:10.1130/G31184.1.

Lunt, D. J., A. Ridgwell, A. Sluijs, J. Zachos, S. Hunter, and A. Haywood (2011), A model for orbital pacing of methane hydrate destabilizationduring the Palaeogene, Nat. Geosci., 4(11), 775–778, doi:10.1038/ngeo1266. [Available at http://www.nature.com/ngeo/journal/v4/n11/abs/ngeo1266.html.].

Lyle, M., et al. (2002), Proceedings of the Ocean Drilling Program, Initial Rep., vol. 199, Ocean Drill. Program, College Station, Tex., doi:10.2973/odp.proc.ir.199.2002.

Mangini, A., M. Jung, and S. Laukenmann (2001), What do we learn from peaks of uranium and of manganese in deep sea sediments?,Mar. Geol., 177(1–2), 63–78, doi:10.1016/S0025-3227(01)00124-4.

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1051

McCarren, H., E. Thomas, T. Hasegawa, U. R€ohl, and J. C. Zachos (2008), Depth dependency of the Paleocene-Eocene carbon isotope excur-sion: Paired benthic and terrestrial biomarker records (Ocean Drilling Program Leg 208, Walvis Ridge), Geochem. Geophys. Geosyst., 9,Q10008, doi:10.1029/2008GC002116.

McDuff, R. E. (1981), Major cation gradients in DSDP interstitial waters: The role of diffusive exchange between seawater and upper oceaniccrust, Geochim. Cosmochim. Acta, 45(10), 1705–1713, doi:10.1016/0016-7037(81)90005-3.

Montadert, L., and D. G. Roberts (1979), Site 401. Init. Rept. DSDP, 48, 73–123.Morford, J. L., and S. Emerson (1999), The geochemistry of redox sensitive trace metals in sediments, Geochim. Cosmochim. Acta, 63(11–12),

1735–1750, doi:10.1016/S0016-7037(99)00126-X.Murphy, B. M., M. Lyle, and A. Olivarez Lyle (2006), Biogenic Burial Across the Paleocene/Eocene Boundary: Ocean Drilling Program Leg 199

Site 1221, 1–12 pp., Ocean Drilling Program, College Station, TX.Nicolo, M. J., G. R. Dickens, and C. J. Hollis (2010), South Pacific intermediate water oxygen depletion at the onset of the Paleocene-Eocene

thermal maximum as depicted in New Zealand margin sections, Paleoceanography, 25, PA4210, doi:10.1029/2009PA001904.Nunes, F., and R. D. Norris (2006), Abrupt reversal in ocean overturning during the Palaeocene/Eocene warm period, Nature, 439(7072), 60–

63, doi:10.1038/nature04386.Panchuk, K., A. Ridgwell, and L. R. Kump (2008), Sedimentary response to Paleocene-Eocene thermal maximum carbon release: A model-

data comparison, Geology, 36(4), 315–318, doi:10.1130/G24474A.1.Pancost, R. D., D. S. Steart, L. Handley, M. E. Collinson, J. J. Hooker, A. C. Scott, N. V. Grassineau, and I. J. Glasspool (2007), Increased terrestrial

methane cycling at the Palaeocene-Eocene thermal maximum, Nature, 449(7160), 332–335, doi:10.1038/nature06012.Pardo, A., G. Keller, E. Molina, and J. Canudo (1997), Planktic foraminiferal turnover across the Paleocene-Eocene transition at DSDP Site

401, Bay of Biscay, North Atlantic, Mar. Micropaleontol., 29(2), 129–158, doi:10.1016/S0377-8398(96)00035-7.Paytan, A., S. Mearon, K. Cobb, and M. Kastner (2002), Origin of marine barite deposits: Sr and S isotope characterization, Geology, 30(8),

747–750, doi:10.1130/0091-7613(2002)030<0747:OOMBDS>2.0.CO;2.Paytan, A., K. Averyt, K. Faul, E. Gray, and E. Thomas (2007), Barite accumulation, ocean productivity, and Sr/Ba in barite across the Paleo-

cene-Eocene Thermal Maximum, Geology, 35(12), 1139–1142, doi: 10.1130/G24162A.1.Pedersen, T. F., and N. B. Price (1982), The geochemistry of manganese carbonate in Panama Basin sediments, Geochim. Cosmochim. Acta,

46(1), 59–68, doi:10.1016/0016-7037(82)90290-3.R€ohl, U., and L. J. Abrams (2000), High-resolution, downhole, and nondestructive core measurements from Sites 999 and 1001 in the Carib-

bean Sea: Application to the Late Paleocene thermal maximum, in Proceedings of the Ocean Drilling Program, Sci. Results, edited by R. M.Leckie et al., pp. 191–203, Ocean Drill. Program, College Station, Tex.

R€ohl, U., T. J. Bralower, R. D. Norris, and G. Wefer (2000), New chronology for the late Paleocene thermal maximum and its environmentalimplications, Geology, 28(10), 927–930, doi:10.1130/0091-7613(2000)28<927:NCFTLP>2.0.CO;2.

R€ohl, U., T. Westerhold, T. J. Bralower, and J. C. Zachos (2007), On the duration of the Paleocene-Eocene thermal maximum (PETM), Geo-chem. Geophys. Geosyst., 8, Q12002, doi:10.1029/2007GC001784.

Rudnick, R. L., and S. Gao (2003), Composition of the continental crust, in Treatise on Geochemistry, edited by H. D. Holland and K. K. Ture-kian, pp. 1–64, Pergamon, Oxford, U. K.

Sexton, P. F., P. A. Wilson, and R. D. Norris (2006), Testing the Cenozoic multi-site composite d18O and d13C curves: new Eocene monospe-cific records from a single locality, Demerara Rise (ODP Leg 207), Paleoceanography, 21, PA2019, doi: 10.1029/2005PA001253.

Sluijs, A., et al. (2006), Subtropical Arctic Ocean temperatures during the Palaeocene/Eocene thermal maximum, Nature, 441(7093), 610–613, doi:10.1038/nature04668.

Stap, L., A. Sluijs, E. Thomas, and L. Lourens (2009), Patterns and magnitude of deep sea carbonate dissolution during Eocene thermal max-imum 2 and H2, Walvis Ridge, southeastern Atlantic Ocean, Paleoceanography, 24, PA1211, doi:10.1029/2008PA001655.

Stoll, H. M., and S. Bains (2003), Coccolith Sr/Ca records of productivity during the Paleocene-Eocene thermal maximum from the WeddellSea, Paleoceanography, 18(2), 1049, doi:10.1029/2002PA000875.

Stoll, H. M., N. Shimizu, D. Archer, and P. Ziveri (2007), Coccolithophore productivity response to greenhouse event of the Paleocene-Eocene thermal maximum, Earth Planet. Sci. Lett., 258(1–2), 192–206, doi:10.1016/j.epsl.2007.03.037.

Svensen, H., S. Planke, A. Malthe-Sorenssen, B. Jamtveit, R. Myklebust, T. Rasmussen Eidem, and S. S. Rey (2004), Release of methane from avolcanic basin as a mechanism for initial Eocene global warming, Nature, 429(6991), 542–545, doi:10.1038/nature02566.

Takeda, K., and K. Kaiho (2007), Faunal turnovers in central Pacific benthic foraminifera during the Paleocene-Eocene thermal maximum,Palaeogeogr. Palaeoclimatol. Palaeoecol., 251(2), 175–197, doi:10.1016/j.palaeo.2007.02.026.

Thomas, D. J., J. C. Zachos, T. J. Bralower, E. Thomas, and S. Bohaty (2002), Warming the fuel for the fire: Evidence for the thermal dissocia-tion of methane hydrate during the Paleocene-Eocene thermal maximum, Geology, 30(12), 1067–1070, doi:10.1130/0091-7613(2002).

Thomas, D. J., T. J. Bralower, and C. E. Jones (2003), Neodymium isotopic reconstruction of late Paleocene-early Eocene thermohaline circu-lation, Earth Planet. Sci. Lett., 209(3–4), 309–322, doi:10.1016/S0012-821X(03)00096-7.

Thomas, E. (Ed.) (1998), Biogeography of the Late Paleocene Benthic Foraminiferal Extinction, pp. 214–243, Columbia Univ. Press, New York.Thomas, E. (2003), Extinction and food at the seafloor: A high-resolution benthic foraminiferal record across the initial Eocene thermal max-

imum, Southern Ocean site 690, Spec. Pap. Geol. Soc. Am., 369, 319–332, doi:10.1130/0-8137-2369-8.319.Thomas, E. (2007), Cenozoic mass extinctions in the deep sea: What perturbs the largest habitat on Earth?, Spec. Pap. Geol. Soc. Am., 424, 1–

23, doi:10.1130/2007.2424(01).Torfstein, A., G. Winckler, and A. Tripati (2010), Productivity feedback did not terminate the Paleocene-Eocene Thermal Maximum (PETM),

Clim. Past, 6(2), 265–272, doi:10.5194/cp-6-265-2010.Tribovillard, N., T. J. Algeo, T. Lyons, and A. Riboulleau (2006), Trace metals as paleoredox and paleoproductivity proxies: An update, Chem.

Geol., 232(1–2), 12–32, doi:10.1016/j.chemgeo.2006.02.012.Westerhold, T., U. R€ohl, J. Laskar, I. Raffi, J. Bowles, L. J. Lourens, and J. C. Zachos (2007), On the duration of magnetochrons C24r and C25n

and the timing of early Eocene global warming events: Implications from the Ocean Drilling Program Leg 208 Walvis Ridge depth tran-sect, Paleoceanography, 22, PA2201, doi:10.1029/2006pa001322.

Winguth, A., C. Shellito, C. Shields, and C. Winguth (2010), Climate response at the Paleocene-Eocene thermal maximum to greenhousegas forcing—A model study with CCSM3, J. Clim., 23(10), 2562–2584, doi:10.1175/2009JCLI3113.1.

Winguth, A. M. E., E. Thomas, and C. Winguth (2012), Global decline in ocean ventilation, oxygenation, and productivity during thePaleocene-Eocene thermal maximum: Implications for the benthic extinction, Geology, 40(3), 263–266, doi:10.1130/g32529.1.

Yamaguchi, T., and R. D. Norris (2012), Deep-sea ostracode turnovers through the Paleocene-Eocene thermal maximum in DSDP Site 401,Bay of Biscay, North Atlantic, Mar. Micropaleontol., 86–87, 32–44, doi:10.1016/j.marmicro.2012.02.003.

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1052

Zachos, J. C., M. W. Wara, S. Bohaty, M. L. Delaney, M. R. Petrizzo, A. Brill, T. J. Bralower, and I. Premoli-Silva (2003), A transient rise in tropicalsea surface temperature during the Paleocene-Eocene thermal maximum, Science, 302(5650), 1551–1554, doi:10.1126/science.1090110.

Zachos, J. C., et al. (2004), Proceedings of the Ocean Drilling Program, Initial Rep., vol. 208, pp. 1–112, Ocean Drill. Program, College Station,Tex.

Zachos, J. C., et al. (2005), Rapid acidification of the ocean during the Paleocene-Eocene thermal maximum, Science, 308(5728), 1611–1615,doi:10.1126/science.1109004.

Zachos, J. C., H. McCarren, B. Murphy, U. R€ohl, and T. Westerhold (2010), Tempo and scale of late Paleocene and early Eocene carbon iso-tope cycles: Implications for the origin of hyperthermals, Earth Planet. Sci. Lett., 299(1–2), 242–249, doi:10.1016/j.epsl.2010.09.004.

Zeebe, R. E., and J. C. Zachos (2007), Reversed deep-sea carbonate ion basin gradient during Paleocene-Eocene thermal maximum, Paleo-ceanography, 22, PA3201, doi:10.1029/2006PA001395.

Geochemistry, Geophysics, Geosystems 10.1002/2013GC005074

P€ALIKE ET AL. VC 2014. American Geophysical Union. All Rights Reserved. 1053