Fossil corals as an archive of secular variations in...

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 160 (2015) 188–208

Fossil corals as an archive of secular variationsin seawater chemistry since the Mesozoic

Anne M. Gothmann a,⇑, Jarosław Stolarski b, Jess F. Adkins c, Blair Schoene a,Kate J. Dennis a, Daniel P. Schrag d, Maciej Mazur e, Michael L. Bender a

a Princeton University, Geosciences, Princeton, NJ, United Statesb Institute of Paleobiology, Polish Academy of Sciences, Warsaw, Poland

c California Institute of Technology, Geological and Planetary Sciences, Pasadena, CA, United Statesd Harvard University, Earth and Planetary Sciences, Cambridge, MA, United States

e University of Warsaw, Department of Chemistry, Warsaw, Poland

Received 17 July 2014; accepted in revised form 16 March 2015; available online 25 March 2015

Abstract

Numerous archives suggest that the major ion and isotopic composition of seawater have changed in parallel with largevariations in geologic processes and Earth’s climate. However, our understanding of the mechanisms driving secular changesin seawater chemistry on geologic timescales is limited by the resolution of data in time, large uncertainties in seawater chem-istry reconstructions, and ambiguities introduced by sample diagenesis. We validated the preservation of a suite of �60unrecrystallized aragonitic fossil scleractinian corals, ranging in age from Triassic through Recent, for use as new archivesof past seawater chemistry. Optical and secondary electron microscopy (SEM) studies reveal that fossil coral crystal fabricsare similar to those of modern coralline aragonite. X-ray diffractometry (XRD), cathodoluminescence microscopy (CL), andRaman studies confirm that these specimens contain little to no secondary calcite. In order to screen for geochemical changesindicative of alteration, we measured 87Sr/86Sr ratios, clumped isotopes, and trace element ratios sensitive to diagenesis (e.g.,Mn/Ca). We retain samples when these tests either fail to identify any diagenetic modifications, or identify specific domainsfree of detectable alteration.

Using the validated fossil coral archive we reconstruct seawater Mg/Ca and Sr/Ca ratios, measured by Secondary IonMass Spectrometry (SIMS), back to �230 Ma. The effects of temperature on coral trace element incorporation cannot explainthe trends observed in our fossil coral Mg/Ca and Sr/Ca data. In agreement with independent records, seawater Mg/Ca molarratios inferred from corals are low (Mg/Ca �1) during the Cretaceous and Jurassic, and increase between the Early Cenozoicand present (Mg/Ca = 5.2). Seawater Sr/Ca ratios from corals vary systematically between �8 and 13 mmol/mol since230 Ma, with maximum values in the Cretaceous and Paleogene. The coral Sr/Ca record disagrees with records fromhydrothermal CaCO3 veins, but is similar to those reconstructed from other biogenic carbonates, especially benthic forami-nifera. The agreement between corals and other archives, for both Sr/Ca and Mg/Ca ratios, further validates our records. Inreturn, fossil coral records improve our understanding of past variations in seawater Mg/Ca and Sr/Ca.� 2015 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.gca.2015.03.018

0016-7037/� 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 (609) 964 9336.E-mail addresses: [email protected] (A.M. Gothmann), [email protected] (J. Stolarski), [email protected] (J.F. Adkins),

[email protected] (B. Schoene), [email protected] (K.J. Dennis), [email protected] (D.P. Schrag), [email protected] (M. Mazur), [email protected] (M.L. Bender).


mailto:[email protected]










http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2015.03.018&domain=pdf

A.M. Gothmann et al. / Geochimica et Cosmochimica Acta 160 (2015) 188–208 189

1. INTRODUCTION

There have been large changes in the major ion concen-tration and isotopic composition of seawater since theNeoproterozoic as indicated from studies of the mineralogyand geochemistry of marine evaporites, barites and bothinorganic and biogenic carbonates (Sandberg, 1983;Hardie, 1996, 2003; Paytan et al., 1998, 2004; Lowensteinet al., 2001, 2003, 2005; McArthur et al., 2001; Dickson,2002, 2004; Horita et al., 2002; Ries, 2004; Timofeeffet al., 2006; Farkas et al., 2007; Zhuravlev and Wood,2008; Coggon et al., 2010; Porter, 2010; Blattler et al.,2012; Misra and Froelich, 2012). An important finding ofthese studies is that the Mg/Ca ratio of seawater has variedby up to a factor of 5 throughout the Phanerozoic. Times ofhigher Mg/Ca ratios (‘aragonite seas’) are coeval with timescorresponding to greater deposition of inorganic aragonitecements, sea-level lowstands, and “icehouse” climates.Times of lower Mg/Ca ratios (‘calcite seas’) are coeval withtimes corresponding to greater deposition of calcitecements, sea-level highstands, and “greenhouse” climates(Sandberg, 1983; Hardie, 1996; Lowenstein et al., 2001).These results suggest a fundamental link between seawaterchemistry, carbonate mineralogy and Earth’s climate state.

A thorough understanding of the geologic and geo-chemical processes driving this link has yet to be achieved,but would provide valuable insight into the interactionbetween major domains in the Earth system. Existingrecords of paleo-seawater chemistry, while pioneering andprofoundly important, are limited by the resolution of datain time, uncertainties introduced by assumptions invoked inthe reconstructions, or ambiguities introduced by samplediagenesis. In order to supplement existing studies of pastseawater chemistry, we assembled a new archive of �60exceptionally well preserved fossil corals, documented theirpreservation for seawater paleochemistry studies, and usedthese samples to reconstruct past seawater Mg/Ca andSr/Ca ratios back to �230 Ma.

1.1. Coral background

Paleontological evidence and phylogenetic analysesbased on molecular markers of modern corals suggest thataragonitic scleractinian corals evolved during the EarlyPaleozoic (Stanley, 2003; Stolarski et al., 2011), but firstbecame a significant component of the rock record in theMiddle to Late Triassic. Consequently, corals have thepotential to yield a �230 Myr record of seawater chemistryspanning a full “cycle” of aragonite and calcite seas(Sandberg, 1983; Hardie, 1996). This feature, along withthe observation that corals generally incorporate manytrace elements in proportions that are related to their abun-dance in seawater (Lea et al., 1989; de Villiers et al., 1994;Cohen and McConnaughey, 2003) make corals attractivesubjects for paleochemistry studies. Aragonitic scleractiniancorals can either grow as large colonies that may form reefs(often harboring dinoflagellate symbionts calledzooxanthellae) or as solitary organisms (most frequentlywithout zooxanthellae). Their habitats range from thewarm, supersaturated surface ocean to the cold deep ocean.

A sketch of a hand sample and a thin section of a soli-tary coral, here modeled after the deep sea coralDesmophyllum dianthus, are shown in Fig. 1. The coralskeleton (or corallum) consists of a cup-like structure(calyx), which hosts the coral animal (polyp). The calyx ismade up of an outer wall called the ‘theca’ and radially ori-ented plates called ‘septa’, which provide further structure.

Two aragonite crystal morphologies exist within theskeleton, which are also distinguishable by their geochem-istry. Centers of Calcification (COCs), also called ‘RapidAccretion Deposits’ (Stolarski, 2003), appear to run downthe middle of the coral septa and theca when corals areviewed in cross section (Fig. 1). COCs may represent thezone where skeletal growth and CaCO3 precipitation isinitiated (Bryan and Hill, 1941; Cuif and Dauphin, 1998).Alternatively, COCs may be formed simultaneously withthe rest of the growing skeleton but with higher extensionrates (Stolarski, 2003; Brahmi et al., 2012; Domart-Coulonet al., 2014). It has been suggested that the extension rateof the COCs and the possible influence of an amorphous cal-cium carbonate precursor phase also may lead to greaterincorporation of organics or trace elements in this region(Cuif et al., 2003; Meibom et al., 2004, 2007; Sinclair et al.,2006). For the reasons described above, COCs are particu-larly susceptible to diagenetic alteration and recrystalliza-tion. Coral ‘fibers’, also called ‘Thickening Deposits’(Stolarski, 2003), exist as acicular needles extending outfrom the COCs and account for most of the mass of the coralskeleton (Cuif and Dauphin, 1998; Perrin, 2003; Tambutteet al., 2011). Typically, the fibers grow at slower rates thanthe COCs and are characterized by a different trace elementcomposition (Meibom et al., 2004; Gagnon et al., 2007).Identifying these crystal morphologies, accounting for theirdistinct chemistries, and studying their relative preservationis important for reconstructing environmental signals(Gagnon et al., 2007; Meibom et al., 2008).

If corals are to be utilized as archives of past seawaterchemistry, two requirements must be fulfilled. First, theremust be a quantifiable relationship between the propertyof interest (e.g., Mg/Ca) measured in the coral, and thatproperty in seawater. Second, fossil corals must be pre-served such that they retain their original geochemical com-position. For our purposes, it is not essential that aspecimen be entirely composed of pristine biogenic arago-nite. This condition is rarely, if ever, met (Allison et al.,2007; Griffiths et al., 2013). For example, secondary arago-nite overgrowths have been found in a modern coral asyoung as 40 years old, and secondary calcite cements havebeen found in fossil corals that lived less than one thousandyears ago (Sayani et al., 2011). Instead, we require thateither (1) the main coral skeleton retains its primary min-eralogy, crystal habit, and geochemistry regardless ofwhether secondary overgrowths are present, and/or (2) thatcorals possess domains of pristine aragonite that are largeenough to sample for our studies.

1.2. Corals as recorders of seawater properties

There is a strong foundation for reconstructing seawaterSr/Ca from corals, and some evidence that corals should

b c

1 cm

a

1 cm

calyx

septum

transversesection

theca

Fig. 1. Coral skeletal structure. (a) Sketch of Desmophyllum dianthus, a deep-sea coral, showing the arrangement of the theca and septa.Samples are sectioned in the transverse plane, perpendicular to the vertical axis of the coral calyx. The scale bar indicates the typical size ofsolitary corals that are part of our sample set. (b) Sketch showing the sectioning of a coral septum, and (c) a thin section image of the coralseptum, taken with a petrographic microscope. COCs, also called ‘Rapid Accretion Deposits’ (Stolarski, 2003), are oriented parallel to thecoral septum and appear opaque and dark in color. Coral ‘fibers’, or ‘Thickening Deposits’ (Stolarski, 2003), radiate outward from the COCs.

190 A.M. Gothmann et al. / Geochimica et Cosmochimica Acta 160 (2015) 188–208

accurately record past seawater Mg/Ca as well. Coral Sr/Caratios are higher than the seawater Sr/Ca ratio by only15–20%, consistent with the inorganic aragonite partitioncoefficient: Sr/Caaragonite/Sr/Casolution = 1.15 (Gaetani andCohen, 2006). In a pioneering study, Swart (1981) grew cor-als in solutions with varying Sr/Ca compositions and foundthat the Sr/Ca ratios measured in the newly grown skele-tons increased linearly with the Sr/Ca ratio of the growthsolutions, suggesting that coral Sr/Ca is dependent on sea-water Sr/Ca. They also found that Sr/Ca depended on Mg/Ca – a result that remains to be verified. The coral Sr/Catemperature dependence is small as well; for every 1 �Cincrease in temperature there is only a 1% decrease in Sr/Ca, indicating that temperature should not significantlybias seawater Sr/Ca reconstructions (Beck et al., 1992; deVilliers et al., 1994; Marshall and McCulloch, 2002;Correge et al., 2004; Gagan et al., 2012). In addition, whileSr/Ca ratios have been shown to vary over scales ofmicrons and tens of microns within the coral skeleton(Meibom et al., 2004; Gagnon et al., 2007), indicating thepresence of ‘vital effects’, bulk coral Sr/Ca ratios are lessvariable. For example, modern surface corals living inwaters between 20 and 30 �C give Sr/Ca ratios rangingbetween 8.8 and 9.5 mmol/mol (Beck et al., 1992; deVilliers et al., 1994; McCulloch et al., 1999), suggesting thatit should be possible to resolve changes in the seawaterSr/Ca ratio that are �1 mmol/mol or greater. These obser-vations show that seawater Sr/Ca likely is the dominantcontrol on the Sr/Ca ratio of the coral skeleton, and thatcorals are prime candidates for reconstructing past seawaterSr/Ca.

Some empirical evidence also suggests that coral Mg/Cascales with the seawater ratio. Ries et al. (2006), workingwith a small dataset, found that the Mg/Ca compositionof coral aragonite grown in culture solutions with differentMg/Ca ratios was positively correlated with the Mg/Caratio of the growth solution. Likewise, Lorens and Bender(1980) found a linear dependence between Mg/Ca in musselaragonite and the Mg/Ca of the water in which it was

grown. Finally, as we show below, the Mg/Ca ratios of fos-sil corals measured in this study track past seawater Mg/Caratios as inferred from other archives.

On the other hand, it is unclear whether Mg substitutesinto the aragonite crystal lattice, or whether it exists pri-marily in either organic or disordered inorganic compounds(Finch and Allison, 2008). Observations that Mg/Ca ratiosin modern coral fibers (excluding COCs) vary by about afactor of 3 on length scales 61 lm also highlight that finescale variations in Mg/Ca depend on vital effects thatremain to be understood (Meibom et al., 2004, 2008).Secondary Ion Mass Spectrometry (SIMS) analyses (spotdiameter �30 lm), which we use here to measure ele-ment/Ca ratios, serve to average out this considerable vari-ability at the 61 lm scale and are thus well-suited for ourpurpose. In this study, we assume a linear dependence ofthe Mg/Ca ratio of corals on seawater Mg/Ca. Thisassumption will be reassessed as coral chemistry is betterunderstood.

1.3. Fossil coral preservation

Scleractinian corals build skeletons of aragonite, a meta-stable polymorph of CaCO3. Most fossil corals undergorecrystallization to calcite, during which elements and iso-topes can be exchanged, altering the geochemistry of theskeleton and rendering the fossils unsuitable for paleo-chemical study (Pingitore, 1976; Brand and Veizer, 1980).However, it is possible to find specimens as old asTriassic in age that retain their original aragonitic min-eralogy (Bender, 1973; Stanley and Swart, 1995; Sorauf,1999; Getty et al., 2001; Ivany et al., 2004; Stolarski andMazur, 2005; Denniston et al., 2008; Mertz-Kraus et al.,2009; Griffiths et al., 2013). Indeed, preservation of coralaragonite may be more straightforward to assess than pre-servation of calcite because the very survival of aragonitesuggests that a specimen is unaltered. Stolarski et al.(2007) also recently identified a set of Cretaceous-age scle-ractinian corals composed of primary calcite and suggested


that these corals may have produced calcitic skeletons as aresult of the low Mg/Ca ratio of the Cretaceous ocean. Inthis study, we focus only on aragonitic corals.

While many previous workers have used petrographicand mineralogical techniques to assess coral preservation,geochemical tools have also been employed (Bender,1973; Allison et al., 2007; Griffiths et al., 2013). For exam-ple, Bender (1973) identified �45 well preserved Cenozoiccorals and dated them by U–Th/He. The aragonitic fossilcoral skeletons did retain most of their radiogenic He sug-gesting good preservation. By pairing petrographic andmineralogical tools with geochemical means of assessingpreservation, we show that it is possible to identify fossilcoral specimens that can be used to reconstruct originalenvironmental properties. Our validations also yield a setof specimens that can be used by others to study additionalproperties of seawater.

2. SAMPLES AND METHODS

Fig. 2 illustrates the range of geologic localities fromwhich samples were collected, and Table 1 lists the diage-netic tests performed on each specimen to validate their pre-servation. A complete summary of ages, locations, loaninginstitutions, and classifications of fossil coral specimensthat we deem fit for use in paleochemistry reconstructionsis given in the Electronic Annex EA-1. Electronic AnnexEA-2 gives a list of additional specimens that were exam-ined, but which did not pass our diagenetic tests.

2.1. Sample preparation

The majority of samples were prepared at PrincetonUniversity for chemical analysis. In preparation for opticalmicroscopy, Scanning Electron Microscopy (SEM) exami-nation, and trace element analysis by SIMS, thick sectionsapproximately 1 cm in width were cut from each hand sam-ple in the transverse plane perpendicular to the coral septa.Sections were ultrasonicated in deionized water for 20 min

PleistocenePlioceneMiocene

OligoceneModern

EoceneCretaceJurassiTriassicPaleocene

Fig. 2. (a) World physical map showing locations from which samples we(b) Europe and (c) North America. Maps taken from ArcGIS. (Source:

(3�). Thin sections were prepared either by AppliedPetrographic Services, Inc. or at the Institute ofPaleobiology in Warsaw, Poland as described in Stolarskiand Mazur (2005). Prior to SIMS analysis, thin sectionswere carbon-coated at the Geological and PlanetarySciences Division Analytical Facility at the CaliforniaInstitute of Technology.

For clumped isotope and Sr isotope analyses, �50 mgpieces of coral were chipped from hand samples, cleanedas described above, and powdered using a mortar and pes-tle. In the case of older specimens with coexisting sparrycalcite infill, powders were drilled using a Dremel tool fromfresh-cut surfaces of hand samples and/or thick sections. Inmany cases, we could not avoid secondary cements in thesesamples and we quantify the amount of secondary calcite inpowders using X-ray diffraction (XRD). Good clumped iso-tope and Sr isotope results for these mixed powders validatethe samples for SIMS measurements. However, we notethat bulk measurements of geochemical properties in thesesamples often reveal a diagenetic component.

2.2. X-ray diffraction (XRD)

Thin sections and aliquots of powder samples used forbulk geochemistry were examined for the presence ofaragonite and calcite using a Bruker D8 Discover X-raydiffractometer at Princeton University. Specimens that gaveXRD patterns with both aragonite and calcite peaksrequired further investigation by SEM and optical micro-scopy to determine if there were domains of well preservedmaterial appropriate for our study (Table 1, ElectronicAnnex EA-3, EA-4).

2.3. Imaging methods

Polished thin sections were studied with a Leica DMLPmicroscope (Princeton University) and a Nikon Eclipse 80imicroscope fitted with a DS-5Mc cooled camera head(Institute of Paleobiology, Warsaw) to look for textures

J1, J2J4 J3

E2

E2Pa2

Mi7

Tr3

E5Tr4, Tr5

Tr1, Tr2

E7K2, K3, K4

Pa3E3 E1

E4

E8

Ol2, Ol3, Ol4, Ol5, Ol6

Pa1

K1

Pl8

Pli3

Mi1, Mi2, Mi3

Mi6Pli1

Pl1, Pl2

Mi8M1

Pli2

a

c

b

ousc

Ol1

re collected (see also Electronic Annex EA-1) and enlarged maps ofUS National Park Service; Esri, DeLorme)

Table 1Summary of diagenetic tests performed on samples that we find to be well preserved enough to reconstruct seawater Mg/Ca and Sr/Ca. ‘–’ = test performed and passed, ‘X’ = test performed andnot passed, ‘ND’ = test not performed. ‘AC’ = acicular texture; ‘I’ = indurated with calcite or sediment; ‘SA’ = secondary aragonite overgrowth; ‘D’ = COC dissolution; ‘M’ = mud infilling;‘A’ = aragonite; ‘C’ = calcite; ‘S’ = silicates. Bold-faced samples correspond to those which failed one of our diagenetic tests, but which may still be suitable for study by microanalysis.

SampleID

Age(Ma)

Location E-SEM

Opticalmicroscopy

CL Raman Mineralogy:thin sectionXRD

Secondarymineralabundances(%): powderXRD

Sr-isotopeage(Ma)

ClumpedisotopesGhosh T(�C)

Average Mn/Ca(lmol/mol) for allSIMS spots belowMn/Ca threshold

Averagecoral Mg/Ca(mmol/mol)

Averagecoral Sr/Ca(mmol/mol)

M1 0.0 Guantanamo Bay, Cuba ND AC ND ND A <1% 0.5 ND 1.6 3.78 8.88M2 0.0 North Pacific ocean, off Agawa,

JapanND AC ND ND A <1% ND ND 1.2 4.15 9.23

Pl1* 2.9 Caloosahatchee Fm., Florida(26.0� N, 81.7� W)

– AC; I; M ND ND A + C + S 3% calcite 2.9 28 0.9 2.50 9.06

Pl2 1.4 Limon Group, Moin Fm.,Costa Rica. (10.0� N, 83.1� W)

ND AC ND ND A <1% 1.4 ND 2.9 3.43 9.20

Pl3 0.1 Barbados ND AC ND ND A ND ND ND 0.1 3.86 8.45Pl4 0.2 Barbados ND AC ND ND A <1% 0.2 ND 0.1 4.09 8.59Pl5* 0.8 Barbados ND AC ND ND A + C 8% calcite 0.8 ND 0.3 4.23 8.44Pl6 0.1 Barbados ND AC ND ND A ND ND ND 0.2 4.45 8.71Pl7 2.3 Caloosahatchee Fm., Florida

(26.0� N, 81.7� W)ND AC ND ND A <1% 2.3 ND 0.5 2.99 9.00

Pl8 2.2 Waccamaw Fm., NorthCarolina (33.9�N, 78.8� W)

ND AC ND ND A <1% 2.2 ND 4.9 3.11 9.32

Pl9 2.2 Limon Group, Moin Fm.,Costa Rica. (10.0�N, 83.1� W)

ND AC ND ND A ND ND ND 2.3 2.23 9.70

Pli1 3.5 Gurabo Fm., DominicanRepublic (19.5� N, 70.7� W)

ND AC ND ND A ND ND 32 1.0 2.97 9.04

Pli2* 3.8 Tamiami Fm., Pinecrest Beds,Florida (26.9� N, 82.0� W)

ND AC ND ND A 2% calcite 3.8 ND 0.6 3.74 8.82

Pli3 2.3 Yorktown Fm., Virginia (37.2�N, 76.9� W)

ND AC ND ND A <1% 2.3 24 2.9 2.29 9.79

Mi1 18.0 Chipola Fm., Florida (30.4� N,85.0� W)

– AC ND – A <1% 18.0 22 1.7 2.09 10.76


– AC; SA ND – A <1% 17.8 21 0.7 2.61 10.44


– AC ND – A <1% 18.2 27 1.4 1.67 10.21

Mi4 18.0 Enewetak Atoll, MarshallsIslands. (11.6� N, 162.3� E)

ND AC; M ND ND A ND ND ND 1.6 3.09 9.08

Mi5 18.0 Enewetak Atoll, MarshallsIslands. (11.6� N, 162.3� E)


Mi6 5.4 Dominican Republic (19.5� N,70.7� W)

ND AC ND ND A <1% 5.4 32 0.5 3.95 8.40

192A

.M.

Go

thm

ann

etal./

Geo

chim

icaet

Co

smo

chim

icaA

cta160

(2015)188–208

Mi7 14.0 Holy Cross Mountains,Korytnica Basin, Poland (50.6�N, 20.5� E)

ND AC ND ND A <1% 14.0 ND 0.4 2.16 10.11

Mi8 14.0 Florida ND AC; M ND ND A ND ND 29 1.8 1.68 10.29Mi9* 7.4 Enewetak Atoll, Marshalls

Islands. (11.6�N, 162.3�E)ND AC; M ND ND A 4% calcite 7.4 ND 0.3 3.77 8.50

Mi10 8.0 Enewetak Atoll, MarshallsIslands. (11.6�N, 162.3�E)

ND AC; SA ND ND A ND ND ND 0.8 3.41 8.77






– AC; SA; M ND ND A 1% calcite 9.4 ND 0.1 2.49 8.95


– AC; SA; M ND ND A 1% calcite 11.0 ND 0.1 2.61 9.57


– AC; SA; M ND ND A ND ND ND 0.1 3.81 8.59


ND AC ND ND A 1% calcite 10.5 ND 0.5 2.98 8.76


– AC; SA ND ND A ND ND ND 0.5 3.75 8.87



Ol1* 23.7 Aquitane, France (43.8�N,1.1�W)

– AC; D ND ND ND 6% calcite 23.7 ND 1.5 1.25 11.10

Ol2 37.7 Mississippi ND AC ND ND A <1% 37.7 30 3.5 1.63 10.33Ol3 31.8 Byram Fm., Mississippi

(32.0�N, 89.4�W)ND AC ND ND A <1% 31.8 ND 1.6 1.68 9.75

Ol4 32.4 Byram Fm., Mississippi(32.0�N, 89.4�W)

– AC ND ND A <1% 32.4 ND 2.3 1.73 9.95


ND AC ND ND A <1% 32.6 ND 3.2 3.03 9.70


ND AC ND ND A ND ND 34 2.3 1.92 9.88

E1 39.2 Gosport Sand Fm., Alabama(31.5�N, 87.9�W)

– AC ND – A <1% 39.2 12 1.7 1.49 10.99

E2 38.6 Siemien, Poland (51.2�N,22.6�E)

– AC; D ND ND ND <1% 38.6 ND 1.2 1.16 9.57

E3 41.0 Mississippi ND AC ND ND A <1% 41.0 23 2.3 1.38 10.47E4 46.7 Alabama ND AC ND ND A <1% 46.7 30 2.3 1.35 10.01E5 39.3 France ND AC ND – A + C 1% calcite 39.3 34 4.0 1.84 9.56E6 34.6 Ukraine ND AC ND ND A <1% 34.6 30 4.9 1.21 11.01E7* 48.9 Austria – AC; I ND – A + C 35% calcite 48.9 40 1.6 1.57 9.97

E8 36.9 Moodys Branch Fm.,Louisiana (31.5� N, 87.8� W)

ND AC ND ND A <1% 36.9 27 4.1 1.29 10.83

(continued on next page)

A.M

.G

oth

man

net

al./G

eoch

imica

etC

osm

och

imica

Acta

160(2015)

188–208193

Table 1 (continued)

SampleID

Age(Ma)

Location E-SEM

Opticalmicroscopy

CL Raman Mineralogy:thin sectionXRD

Secondarymineralabundances(%): powderXRD

Sr-isotopeage(Ma)

ClumpedisotopesGhosh T(�C)

Average Mn/Ca(lmol/mol) for allSIMS spots belowMn/Ca threshold

Averagecoral Mg/Ca(mmol/mol)

Averagecoral Sr/Ca(mmol/mol)

Pa1 61.0 Monrow Co., North Dakota ND AC ND ND A ND ND 33 6.9 1.43 10.95Pa2* 57.5 Babica Clays, Poland (49.9� N,

21.9 E)

– AC; D ND – A + S >1% silicates 32.8 ND 1.6 1.31 11.56

Pa3 60.1 Wills Point Fm., Texas (30.3�N, 97.7� W)

ND AC ND ND A <1% 60.1 31 3.3 1.47 10.92

Pa4 63.0 Sobral Fm., Seymour Island,Antarctica. (64.3� S, 56.8� W)

– AC; D ND ND A ND ND ND 6.9 1.57 11.54

K1* 71.4 Pierre Shales, Black Hills, SD.(45.1� N, 100.9� W)

– AC; I – – A + C + S 18% calcite 71.4 ND 5.1 0.91 13.99

K2* 86.7 Gosau, Austria (47.6� N, 13.5�E)

– AC; I – – A + C 15% calcite 86.7 33 1.2 0.92 10.86

K3* 85.8 Gosau, Austria. (47.6� N, 13.5�E)

– AC; I; D – – A + C 5% calcite 85.8 35 1.3 0.47 11.43

K4* 84.4 Gosau, Austria. (47.6� N, 13.5�E)

– AC; I; D ND – A + C 70% calcite 84.4 ND 0.3 0.88 10.92

J1 160.3 Ostromice, Poland (53.8� N,14.8� E)

– AC ND ND A 1% calcite 160.3 9 6.0 0.95 9.35

J2* 158.8 Ostromice, Poland (53.8� N,14.8� E)

– AC; I; D ND ND A + C 5% calcite 158.8 ND 2.7 0.72 9.53

J3 163.0 Lukow, Poland (51.9� N, 22.4�E)

– AC; I; D ND – A + C ND ND ND 3.9 0.94 8.49

J4 161.5 Hohenferchesar, Germany ND AC; SA ND ND A <1% 161.5 23 1.4 0.70 8.76Tr1* 206.4 Zlambach Fm., Austria (47.6�

N, 13.7� E)– AC; I; D – ND A + C 40% calcite 206.4 ND 6.2 1.04 11.52

Tr2* 220.0 Zlambach Fm., Austria (47.6�N, 13.7� E)

– AC; I; D ND ND A + C 75% calcite 200.9 ND 2.5 1.02 9.52

Tr3* 220.0 Alakir Cay, Turkey (36.6� N,30.3� E)

– AC; I; D ND – A + C 37% calcite 210.7 38 0.6 1.29 8.26

Tr4* 230.0 Alpe di Specie, Dolomites(46.6� N, 12.2� E)

– AC; I; D ND – A + C ND ND ND 0.8 1.00 8.97

Tr5* 230.0 Alpe di Specie, Dolomites(46.6� N, 12.2� E)

– AC; I; D ND – A + C ND ND ND 0.7 1.01 8.90

* Sample unsuitable for bulk analysis.

194A

.M.

Go

thm

ann

etal./

Geo

chim

icaet

Co

smo

chim

icaA

cta160

(2015)188–208


indicative of alteration. Observations were performed intransmitted, polarized, and reflected light. Regions thatwere determined to be best preserved were imaged inreflected light using a Zeiss Discovery V12 microscope(Princeton University Imaging and Analysis Center) to pro-vide a map for sampling during SIMS analysis.

An FEI Quanta FEG Environmental-SEM (E-SEM) atthe Princeton Imaging and Analysis Center was used toexamine a subset of thin sections, concentrating specificallyon those thin sections containing a mixture of aragoniteand calcite as indicated by XRD. E-SEM analyses wereconducted using both Secondary Electron (SE) andBackscattered Electron (BSE) modes, and were used toidentify regions that exhibit the best preserved crystalmorphologies.

A subset of specimens was also imaged by cathodolumi-nescence microscopy (CL) and Raman confocal microscopyas described in Frankowiak et al. (2013) to achieve a betterspatial understanding of calcite and aragonite distributionsin samples for which both polymorphs were present.Raman maps of calcite and aragonite polymorphs wereextracted from raw Raman data through a direct classicalleast squares (DCLS) modelling procedure performed onthe multidimensional spectral array. At each spectrumwithin the array the DCLS modelling procedure finds a lin-ear combination of the reference component spectra whichbest fits the data. The reference spectra were selected fromthe raw data at specific positions where only aragonite orcalcite were present, respectively.

2.4. Secondary Ion Mass Spectrometry (SIMS)

Trace element ratios were measured using a Cameca ims7f-GEO at the California Institute of TechnologyMicroanalysis Center. A negative oxygen ion (O�) beamwith a 25 lm raster size and a 15 nA beam current was usedto pre-sputter the sample surface for 240 s at each analysispoint. Subsequently, a 2 lm raster O� beam with 5 nAbeam current was used to sputter ions from the samplefor data acquisition. At each spot, 15 cycles were measuredfor 23Na (1 s/cycle counting time, �700,000 cps), 24Mg(1 s/cycle counting time, �7000 cps), 32S (6 s/cycle countingtime, �200 cps), 42Ca (1 s/cycle counting time,�100,000 cps), 88Sr (1 s/cycle counting time, 180,000 cps).55Mn (10 s/cycle counting time) count rates were low(�3 cps) in pristine samples, but up to �1000 cps whererecrystallization was significant. The total analysis timefor all 15 cycles (including pre-sputtering time and measure-ments of other isotopes not listed here) was 10 min. Foreach fossil coral thin section, at least one transect ofapproximately 10 sampling points was measured acrossthe coral septa. In most cases at least 3 transects were mea-sured. Achieved precisions for individual coral SIMS spotdata for Mg/Ca, Sr/Ca, Na/Ca, and S/Ca are <3% (2rS.E.). Precision for Mn/Ca ranges from <2–25% (2r S.E.)where recrystallization is significant.

Raw data were referenced to the carbonatite standard,OKA-II, which has been found to be an accurate standardfor Sr/Ca and Mg/Ca ratios for carbonate samples ofunknown composition (Gabitov et al., 2013). During the

course of an analytical session lasting 1 week, OKA-II wasmeasured between 10 and 15 times. Following Gabitovet al. (2013), samples were only referenced to standard dataacquired from within the sections of OKA-II that arehomogeneous with respect to Mg/Ca and Sr/Ca. Measuredelement/Ca intensity ratios for Mg/Ca and Sr/Ca measure-ments in OKA-II across two years of analytical sessionsare 0.163 ± 3% and 2.45 ± 2%, respectively (1r S.D.).Measured values for Mn/Ca, Na/Ca, and S/Ca in OKA-IIare 0.065 ± 3%, 0.378 ± 7%, and 1.3 � 10�5 ± 45% (1rS.D.) respectively. The large uncertainty in S/Ca is due toextremely low S concentrations in OKA-II and the resultingvery low count rates. Precision (reproducibility) of S countrates was good in our fossil corals. Relative S/Ca ratios arerobust but absolute ratios are highly uncertain. Sr/Ca andMg/Ca elemental ratios were measured in grain matchedsamples of OKA-II by Isotope Dilution InductivelyCoupled Plasma Mass Spectrometry (ID-ICP-MS)(Gabitov et al., 2013), and Mn/Ca and Na/Ca ratios inOKA-II were measured using a ThermoFinnigan Element2ICP-MS at Princeton University. S/Ca ratios in OKA-IIwere measured by Ion Chromatography at the CaliforniaInstitute of Technology as described in Paris et al. (2013).

Following SIMS analysis, each thin section was studieda second time with a petrographic microscope to determinethe placement of SIMS pits. Data from SIMS pits thatapparently sampled epoxy, calcite infilling, COCs, or sec-ondary aragonite overgrowths (instead of coral skeleton)were excluded from reconstructions of seawater Sr/Caand Mg/Ca. The full data set from SIMS measurements,including SIMS data from spots that sampled epoxy or sec-ondary materials, is presented in the supplementary materi-als (Electronic Annex EA-5, EA-6).

2.5. Clumped isotopes

A representative subset of 27 fossil coral specimens wasanalyzed for carbonate clumped isotope paleotemperaturesto test for burial diagenesis (Eiler, 2007; Dennis and Schrag,2010; Huntington et al., 2011; Passey and Henkes, 2012).With the exception of one coral that we determined to bepoorly preserved, R10 (see Electronic Annex EA-2),between 2 and 5 replicate measurements were made foreach sample (Electronic Annex EA-7). Analyses were per-formed at Harvard University, following the procedure ofDennis and Schrag (2010), with the data presented in theabsolute reference frame as described in Dennis et al.(2011). Improbably high clumped isotope temperatures(>40 �C) measured on a small number of samples are sug-gestive of alteration in a high temperature environment.

2.6. Sr isotopes

Approximately 70% of the fossil corals analyzed bySIMS were also measured for Sr isotopes (ElectronicAnnex EA-8). Powders of coral skeleton were dissolved in1 N HNO3, and un-spiked Sr2+ was isolated by ionexchange chromatography with an Eichrom Sr-spec resin.Dried separates of Sr samples were re-dissolved in 2 lL of0.1 M H3PO4. The dissolved samples were then combined


with a 2:1-part Ta-gel and 1 M H3PO4 mixture and loadedonto outgassed Re filaments. Sample load sizes wereapproximately 400 ng Sr.

87Sr/86Sr ratios were measured on an IsotopXPhoeniX62 Thermal Ionization Mass Spectrometer atPrinceton University. Measurements were made using a 3-sequence dynamic routine, where mass-87 was measuredin the Axial, H1, and H2 cups, and masses-84, 85, 86, and88 were measured in sequentially adjacent cups. Data werecollected in 15 blocks of 12 cycles each. Filaments werewarmed to 1 A in 30 min and ramped up to 2.8–3.2 A toachieve a beam intensity of 3–5 V for 88Sr and a filamenttemperature of �1350 �C. The Sr isotope standard NBSSRM 987 (87Sr/86Sr = 0.710248) was measured as a refer-ence throughout the period of our analytical sessions andgave an average value of 87Sr/86Sr = 0.710243 ± 0.000004(2r S.D.) (n = 12). A modern coral standard, with an iso-topic composition identical to that of modern seawater87Sr/86Sr = 0.70918 (Faure and Mensing, 2005), was passedthrough the entire chemical procedure with each group ofsamples, and exhibited a long term average composition of87Sr/86Sr = 0.709170 ± 0.000014 (2r S.D.) (n = 6).

3. RESULTS AND DISCUSSION

3.1. Screening samples for diagenesis

Samples must meet the following criteria in order tobe considered well preserved: (1) visual inspection of

a

b

Fig. 3. Characterization of diagenesis in fossil scleractinian corals by petrsection image (crossed-polars) of Upper Cretaceous coral, K4. Crystals wtypical of coralline aragonite. The space around the coral septum is infillepolars) of Middle Miocene coral, Mi14. An acicular habit is exhibited boutgrowths. Spaces are visible between needles of secondary aragonite, dspace. (c) Thin section image (transmitted light) of Upper Jurassic coraalteration of the COCs as evident by discoloration and texture. COCsPetrographic thin section image (crossed-polars) of Cretaceous coral, R2,was rejected from our well preserved sample set. (e) SEM image of recrimage of sample R6, showing dissolution and ‘fibrous micropores’ as in

hand samples must indicate good preservation of coralmorphology, (2) XRD must indicate specimens are pre-served as aragonite (except in cases where corals are visi-bly infilled with cement), (3) optical microscopy andSEM must show preservation of crystal ultrastructure,(4) Raman spectroscopy must suggest the absence of cal-cite within the coral skeleton, (5) results of SIMS andCL studies must indicate low Mn concentrations withinthe skeleton, (6) trace element patterns found in mostfossil corals must agree with those observed in mostmodern corals, (7) carbonate clumped isotope tempera-tures must be below 40 �C, and (8) stratigraphic ageinferred from the 87Sr/86Sr ratio must agree withindependently determined stratigraphic age. All of thesecriteria contribute valuable insight into fossil coralpreservation.

We investigated 99 samples in total using at least oneof the tests above, 63 of which were determined to be wellpreserved for our purposes (Table 1, Electronic AnnexEA-1, EA-2). Of those 63, two samples failed one of ourdiagenetic tests (the clumped isotope test and Sr isotopetest, respectively) but passed all other tests for diagenesis.Because both of these samples are infilled with calcite orsediment, and because the clumped isotope and Sr isotopetests were performed on bulk powders containing (forthese samples) a mixture of coral and infilling, we con-clude that they still are likely to be suitable for micro-analysis but are not well enough preserved to use forbulk geochemistry.

fibrous micropores

blocky calcite texture

c e

fd

ographic, optical microscopy, and SEM (a–f). (a) Petrographic thinithin the primary aragonite coral septum exhibit an acicular habit,d with calcite cement. (b) Petrographic thin section image (crossed-oth within the primary coral skeleton, and by secondary aragoniteistinguishing them from primary material. Black regions show porel, J2, showing both well preserved aragonite and dissolution andare avoidable by using microanalysis techniques to sample. (d)

exhibiting significant recrystallization of the skeleton. This specimenystallized sample, R16 showing blocky euhedral textures. (f) SEMBar-Matthews et al. (1993).


3.1.1. X-ray diffraction

We measured 15 thin section samples (out of the 63samples determined to be well preserved) with both arago-nite and calcite XRD peaks (Table 1). Three samples (Pl1,Pa2, K1) also showed peaks corresponding to silicate min-erals. All of these thin sections contained visible sedimentor calcite infilling when examined using a petrographicmicroscope, as in Fig. 3a. The presence of indurated cal-cite cement or sediment in Mesozoic and Early Cenozoicsamples is common; of the 18 fossil corals older than�50 Ma that we determined to be well-preserved, 13 arecharacterized by infilling of the coral skeleton with sec-ondary calcite spar as in Fig. 3a. In contrast, only 2 ofthe 46 specimens younger than �50 Ma that we deter-mined to be well preserved are infilled, although carbonatemuds are present in some specimens and secondary arago-nite is also sometimes visible as protruding needles(Fig. 3b). Small percentages of calcite were also quantifiedin drilled powder samples used for bulk geochemicalanalyses. The presence of calcite in these samples maybe related to partial recrystallization of the COCs or thepresence of carbonate muds. These samples (marked withasterisks in Table 1) may not be suitable for bulkgeochemistry.

Fig. 4. Characterization of preservation by cathodoluminescence and Ramcoral, K1. Yellow circles represent the locations of SIMS analysis spots. (K1. Regions of red luminescence correspond to regions composed of calskeleton, suggesting it is composed of aragonite and indicating that theimage (transmitted light) of Cretaceous coral, K2. Yellow circles represendistribution for the sample shown in (c). Areas shown in green corresponthe same sample shown in (c). Areas shown in red correspond to calcite-riclegend, the reader is referred to the web version of this article.)

3.1.2. Microscopy

To better quantify the spatial distribution of aragoniteand secondary minerals in infilled samples, we studied oursamples using petrographic microscopy, Raman micro-scopy, SEM, and CL. Petrographic microscopy and SEMare used to identify textures typical of either primary arago-nite or secondary calcite, that may suggest the absence orpresence of alteration, respectively (see Fig. 3, ElectronicAnnex EA-3). In well preserved samples (e.g., Fig. 3a–c)the acicular habit typical of primary coralline aragonite isvisible within the skeleton, suggesting good preservationof aragonite fibers. Poorly preserved specimens are oftencharacterized by the presence of a transparent, nongranulartexture within the area of the COC when viewed by opticalmicroscopy and blocky textures extending into the fibers(Fig. 3d). As the COCs become more pervasively altered,these zones take on the appearance and texture of calcitespar (Fig. 3d, e). In specimens for which dissolution andrecrystallization are limited to the COCs, it is possible toavoid these zones when choosing targets for SIMS spots(e.g., Fig. 3c). Other textural features that we used as exclu-sion criteria include dissolution pits, ‘fibrous micropores’(Bar-Matthews et al., 1993; Fig. 3f), and evidence ofmicro-boring (Webb et al., 2009; Nothdurft and Webb,

an. (a) Thin section image (transmitted light) of Upper Cretaceousb) CL image of the same region shown in (A) for Cretaceous coral,cite (mainly sediment). Red luminescence is absent within the coralregions analyzed were not affected by diagenesis. (c) Thin section

t the locations of SIMS analysis spots. (d) Raman map of aragonited to aragonite-rich zones. (e) Raman map of calcite distribution forh zones. (For interpretation of the references to colour in this figure


2009) (see samples R5–R8, Electronic Annex EA-2; EA-3).While petrographic microscopy and SEM studies help tobetter understand the preservation state of our specimens,we are unable to confirm the absence of calcite within thecoral skeleton using petrographic microscopy and SEMalone.

CL and Raman analyses allow us to better constrainthe spatial distribution of calcite in infilled samples.Samples passing our CL and Raman tests are generallycharacterized by the absence of luminescence (CL) andcalcite mineralogy (Raman) within the coral skeleton,despite the presence of highly luminescent and calcite-richinfilling (Fig. 4). Some samples that we designate as ‘wellpreserved’ do contain calcite-rich, luminescent zones nearthe COCs (Electronic Annex EA-9 to EA-12) consistentwith preferential dissolution and recrystallization of theCOCs as recognized in our petrographic microscopy stud-ies. While we retain these samples as part of our suite, wenote that they are unsuitable for bulk analysis and we dis-card any SIMS data that overlap with the luminescent orcalcite-rich zones. Samples failing our CL and Ramantests are typically characterized by domains of calciteintercalated with aragonite at the micron-scale, as shownin Frankowiak et al. (2013). Because both bulk andSIMS measurements of these corals would include a diage-netic component, we excluded such samples from our sea-water chemistry reconstructions (see Electronic AnnexEA-2).

8

10

12

14

1 12 23 4 5 6 7 3 4 5 6 7 8

100

200300

400500

4

2

6

0

Spot Number Across Transect

Sr/C

a (m

mol

/mol

)Mn/

Ca

(µm

ol/m

ol)M

g/C

a (m

mol

/mol

)

50 µm 50 µm

Sample K1, CretaceousWell preserved

Sample K3, CretaceousModerately altered

COChigh Mn/Ca

a b

0

High Mn Threshold

*all spots for this transecMn/Ca = 10.3 µmol/mol

Fig. 5. Mg/Ca, Mn/Ca, and Sr/Ca ratios from two Cretaceous age fossshown here are on average 3%, 1%, and 10%, respectively (2r S.E.)Trochocyathus egeri (White, 1879) from Pierre shale, Dry Creek, Blackscleractinian coral K3, Rennensismilia complanata (Goldfuss, 1826), fromscale) for sample K1. (d) Relationship between Mg/Ca and Mn/Ca (log scthan for sample K3, reflecting the more pristine nature of sample K1.

3.1.3. Mn/Ca ratios

Mn/Ca ratios measured concurrently with othermetal/calcium (Me/Ca) ratios using SIMS further ensurethat we identify small diagenetic domains that may nothave been detected by our other tests (SEM, petrographicmicroscopy, CL, and Raman). The partition coefficientfor Mn in calcite is greater than that for Mn in coral arago-nite (Pingitore, 1978; Shen et al., 1991) and so diageneticcalcite will usually be enriched in Mn. In addition, if recrys-tallization occurs in reducing waters, those diagenetic fluidsshould contain far more dissolved Mn than seawater, lead-ing to very high concentrations of Mn in the diageneticphase (Brand and Veizer, 1980).

For some samples older than Miocene in age, one ormore SIMS spot measurements had anomalously highMn/Ca ratios outside of the range measured in modern cor-als (Mn/CaModernCoral � 0.01–10 lmol/mol; Shen et al.,1991). We attribute these high Mn/Ca ratios to small-scalerecrystallization of the skeleton. The highest Mn/Ca ratiomeasured was 3700 lmol/mol for a Triassic age specimen.This is at least 3 orders of magnitude higher than the high-est Mn/Ca ratios reported for modern specimens. Similarlyhigh Mn/Ca ratios have recently been measured and identi-fied as the contaminant phase, kutnahorite, in foraminiferashells (Pena et al., 2005). As our criterion for small-scalerecrystallization, we used the maximum Mn/Ca ratio mea-sured in Shen et al.: Mn/Ca = 10.3 lmol/mol. Any spotswith Mn/Ca >10.3 lmol/mol were excluded from

9 10 1112

High Mn Threshold

t above

0.1

1

10

0.1 1 10 100 1000

Log(

Mg/

Ca)

mm

ol/m

ol

Log(

Mg/

Ca)

mm

ol/m

ol

Log(Mn/Ca) µmol/mol

0.1

1

10

0.1 1 10 100 1000 Log(Mn/Ca) µmol/mol

Sample K1, CretaceousWell preserved

Sample K3, CretaceousModerately altered

c

d High Mn Threshold

il corals. Standard errors for Mg/Ca, Sr/Ca, and Mn/Ca for data. (a) Transect across the septum of the scleractinian coral K1,

Hills (South Dakota, USA). (b) Transect across the septum ofGosau, Austria. (c) Relationship between Mg/Ca and Mn/Ca (log

ale) for sample K3. Fewer analyses are excluded from the sample K1


reconstructions of seawater Mg/Ca and Sr/Ca. In practice,decreasing our exclusion threshold to values lower than10.3 lmol/mol has very little impact on average Mg/Caand Sr/Ca ratios.

Fig. 5 shows Mn/Ca, Mg/Ca and Sr/Ca measurementsacross representative SIMS transects from two differentCretaceous age specimens (K1 and K3) and illustrateshow we used Mn/Ca to recognize fine scale recrystalliza-tion. Also shown are plots of Mg/Ca vs. Mn/Ca for all ofthe SIMS spot analyses measured for these samples acrossmultiple transects (Fig. 5c, d). Both K1 and K3 were deter-mined to be well preserved based on XRD, petrographicmicroscopy and SEM. Mn/Ca ratios in K1 fall below thethreshold value of 10.3 lmol/mol at all spots measuredacross the transect shown in Fig. 5a except for a singleSIMS point that overlaps with the COC. The COC isjudged to be recrystallized, and the average Mg/Ca andSr/Ca ratios for this sample are calculated from the remain-ing SIMS data. In contrast, Mn/Ca in K3 is elevated acrossthe entire transect shown in Fig. 5b (especially at spots 1–4,7, 10, and 11) although in other regions of sample K3 weidentify spots with Mn/Ca below 10.3 lmol/mol (Fig. 5d).For most of the spots where we observe high Mn/Ca ratios,we observe that Sr/Ca ratios are low and Mg/Ca ratios arehigh. This general pattern is consistent with diageneticalteration to calcite, which lowers Sr/Ca while raisingMg/Ca and Mn/Ca (Brand and Veizer, 1980; Webb et al.,2009).

We also find that rejected samples with micron-scale,intercalated domains of calcite and aragonite (as indicatedby CL and Raman; e.g., samples R1, R3, and R4) all haveMn/Ca >10.3 lmol/mol. This result shows that ourquantitative studies of Mn/Ca agree with CL imaging andRaman spectroscopy studies, and helps create a consistentpattern of preservation across our sample set.

10.5

11.5

12.5

13

12

14

0.5 1.5 2.5 0.8

8

42

10

10

12

11

1.0 1.6

Sr/C

a (m

mol

/mol

)

Sr/C

a (m

mol

/mol

)Eocene: E1 Cretace

Mg/Ca (mmol/mol) Mg/Ca (

Pleistocene: Pl1 Oligoce

Mg/Ca (mMg/Ca (mmol/mol)

Sr/C

a (m

mol

/mol

)

Sr/C

a (m

mol

/mol

)

Fig. 6. Representative plots of Sr/Ca vs. Mg/Ca spot measurements forsample set display an inverse relationship similar to that observed by Gagto low Mn/Ca ratios whereas data plotted in red represent spots correspohigh Mn/Ca spots represent SIMS analyses that were measured both inwithin the coral skeleton itself (representing recrystallization of the skelet3% and 1%, respectively (2r S.E.). (For interpretation of the references toof this article.)

We make three conclusions from our SIMS studies ofMn/Ca. First, despite careful study by XRD, SEM, andpetrographic microscopy, we document local alterationthrough SIMS measurements of Mn/Ca. CL and Ramanconfirm these results (Electronic Annex EA-9 to EA-12).Second, even when specimens are partially altered, it is pos-sible to access and characterize pristine domains, avoidingMn-rich regions according to the criterion described above.Third, bulk geochemical measurements for specimens withhigh Mn/Ca SIMS spots reflect a combination of primaryand secondary material, and thus cannot provide robustpaleoenvironmental information; microanalysis techniquesare necessary to recover the original geochemical signaturesfrom these specimens (these samples are marked by aster-isks in Table 1).

3.1.4. Trace element patterns

As another test of diagenesis, we determine whether ele-ment/Ca ratios in fossil corals exhibit the same patterns asin modern corals. Sr/Ca ratios vary inversely with Mg/Caratios in aragonite drilled from modern coral fibers,although this pattern is not observed in the COCs(Gagnon et al., 2007; Gaetani et al., 2011). We generallyexpect to find a relationship between Sr/Ca and Mg/Ca inour fossil corals if they are well preserved. However, wedo not reject samples that fail to show the relationshipdue to its absence in some modern corals measured bySIMS (e.g., Allison and Finch, 2010). NanoSIMS studiesthat measure Mg/Ca and Sr/Ca at a finer scale (spot size<1 lm) also fail to observe this pattern, suggesting thatthe relationship may be scale dependent (Meibom et al.,2007, 2008). In addition to the relationship between Mg/Ca and Sr/Ca, Recent corals and other aragonites alsoshow a positive relationship between S/Ca and Na/Caratios (Busenberg and Plummer, 1985; Bar-Matthews

9.5

10.5

1.61.2

2 4 6

Sr/C

a (m

mol

/mol

)

Mg/Ca (mmol/mol)

10 20 30

2.2* no clear relationship

Oligocene: Ol5

ous: K1

mmol/mol)

ne: Ol4

mol/mol)

5

10

Sr/C

a (m

mol

/mol

)

Mg/Ca (mmol/mol)

Triassic: Tr4

high Mn/Ca -shallower slope

low Mn/Ca

fossil corals of various ages. Approximately 80% of corals in ournon et al. (2007). Data shown in grey represent spots correspondingnding to high Mn/Ca ratios. In the case of the Triassic age sample,the calcite matrix in which the coral skeleton was embedded, andon). Standard errors for Mg/Ca and Sr/Ca SIMS spot analyses arecolor in this figure legend, the reader is referred to the web version


et al., 1993), which we expect to see in our samples if theyare well preserved.

Results of SIMS element/Ca analyses (Mg/Ca, Sr/Ca,Mn/Ca, Na/Ca, and S/Ca) for each sample in our studyset are listed in Electronic Annex EA-5. Approximately80% of all fossil corals exhibit an inverse relationshipbetween Sr/Ca and Mg/Ca, similar to that observed byGagnon et al. (2007), supporting the primary nature ofour samples (Fig. 6, Electronic Annex EA-13). Previousstudies have suggested that, while Sr substitutes directlyinto the aragonite crystal lattice, a significant portion ofcoral Mg may exist in either organic or disordered inor-ganic compounds (Finch and Allison, 2008). This dis-tribution of Mg raises concerns about the stability of Mgin corals over the timescales we are investigating. The pre-servation of Sr/Ca–Mg/Ca relationships across our fossilsuite supports the conclusion that both Sr and Mg are pre-served at a fine scale within the coral skeleton. Moreover,we fail to observe evidence for loss or gain of Mg fromthe boundaries of the coral skeleton that would raise con-cerns about preservation.

An inverse correlation between Sr/Ca and Mg/Ca couldalso be imparted by diagenetic alteration, as describedabove (Brand and Veizer, 1980). This diagenetic trend istypically distinct from primary patterns between Sr/Caand Mg/Ca in that the diagenetic Sr/Ca–Mg/Ca trendsexhibit much shallower slopes (Fig. 6). In addition, Mg/Ca ratios measured by SIMS in our modern corals varyby a factor of 2–5 while variability in altered fossils canbe much greater (e.g., the variation exhibited within thehigh-Mn/Ca zones of the Triassic age sample shown inFig. 6 is about a factor of 10).

We also observe a general positive correlation betweenNa/Ca and S/Ca in our fossil and modern corals. This pat-tern is similar to that observed by Busenberg and Plummer(1985) and Bar-Matthews et al. (1993) for various moderncorals, young fossil corals, and other inorganic aragonites(Fig. 7). Additional plots of Na/Ca vs. S/Ca relationshipsfrom within a few individual coral specimens can be found

S/C

a (m

mol

/mol

)

Na/Ca (mmol/mol)

Modern + Fossil coral (Bar-Matthews et al. 1993)

Modern coral (Busenberg and Plummer, 1985)

Oolitic aragonite (Bar-Matthews et al. 1993)

Fibrous aragonite (Bar-Matthews et al. 1993)

Acicular aragonite (Bar-Matthews et al. 1993)

0 10 20 30 40 50

a

25

0

5

10

15

20

Fig. 7. (A) S/Ca vs. Na/Ca ratios of modern and fossil corals (Busenbearagonites (Bar-Matthews et al., 1993). Bar-Matthews et al. data represenelectron microprobe. Uncertainties are 6% (2r) for Na/Ca data and 7%corals (this study). Each symbol corresponds to a single SIMS spot analy3%, respectively (2r). S/Ca and Na/Ca generally show a positive correla

in Electronic Annex EA-14. While corals and aragonitesmeasured in previous studies show a tight positive correla-tion (Fig. 7a), both modern and fossil corals from this studyexhibit a larger spread in both Na/Ca and S/Ca space(Fig. 7b). Our data represent individual SIMS spot mea-surements whereas data from Busenberg and Plummer(1985) and Bar-Matthews et al. (1993) represent bulk mea-surements made on powder samples and averages of elec-tron microprobe data, respectively. The different scales ofobservation explain some, but not all, of the greater vari-ability in our samples.

Data from Busenberg and Plummer (1985) and Bar-Matthews et al. (1993) also show that modern corals exhibitthe highest Na/Ca and S/Ca, while Pleistocene-age fossilcorals (younger than almost all of our fossil coral samples)are more depleted in Na and S relative to Ca (Fig. 7).Secondary aragonite and other inorganic aragonites(fibrous, acicular, and oolitic) are characterized by the low-est Na/Ca and S/Ca ratios. Bar-Matthews et al. (1993) sug-gest that this pattern may indicate a relationship betweenNa and S concentrations and preservation, where lowerNa/Ca and S/Ca ratios indicate poor preservation.Similar to Bar-Matthews et al., we find that Na/Ca andS/Ca ratios are highest in our Recent coral samples, butmost fossil corals ranging in age from Pleistocene throughJurassic retain S/Ca ratios within the range of our modernspecimens. This aspect of the data suggests good preserva-tion. Some SIMS measurements of S/Ca and Na/Ca ratiosin fossil corals also fall outside the modern field. Two fac-tors likely contribute to this departure. One is that ourmodern field is based on only 2 specimens and is unlikelyto cover all variability. The other is that departures areexpected as seawater [Ca2+] and possibly [SO4

2�] variesthrough time (Paytan et al., 1998, 2004; Lowenstein et al.,2003). Indeed, the low Na/Ca ratios in many Mesozoicsamples (as well as the lower S/Ca ratios for some samples)may reflect higher seawater [Ca] during the Mesozoic(Lowenstein et al., 2001, 2003, 2005; Horita et al., 2002;Dickson, 2002, 2004; Timofeeff et al., 2006; Brennan

S/C

a (m

mol

/mol

)

Na/Ca (mmol/mol)

25ModernPlio-PleistoceneMiocene

OligoceneEocenePaleocene

CretaceousJurassicTriassic

00

5

10

15

20

10 20 30 40 50

b

rg and Plummer, 1985; Bar-Matthews et al., 1993), and inorganict averages of �30 spot measurements (1 lm diameter) analyzed by(2r) for S/Ca data. (B) S/Ca vs. Na/Ca ratios measured in fossil

sis. Standard errors for Na/Ca and S/Ca are approximately 2% andtion.

??

no detectable calciteinfilled with calcite

only 1 measurement

rejected rejected

Geologic Age (Ma)

Gho

sh T

(˚C

)

0 50 100 150 2000

10

20

30

40

50

60

Fig. 8. Measured clumped isotope temperature in fossil corals vs.geologic age projected into the absolute reference frame of Denniset al. (2011) and calculated using the re-projected calibration ofGhosh et al. (2006). Temperatures above 40 �C indicate burialdiagenesis. Red arrows indicate samples for which clumped isotopetemperatures are unreasonably high. Samples E7 and Tr3 – bothmarked by a ‘?’ – are plotted as red squares in Fig. 10 to conveyuncertainties in their preservation. Data represent the average and1r S.E. (n > 2) or 1r S.D. (n = 2) of replicate measurements. (Forinterpretation of the references to colour in this figure legend, thereader is referred to the web version of this article.)


et al., 2013; Holt et al., 2014; this work). A few extremelylow S/Ca and Na/Ca measurements from Triassic coralsmay indicate inferior preservation and suggest that othercompositional data from these samples should be inter-preted with caution. In general, we do not see relationshipsbetween low S/Ca or Na/Ca ratios and Mg/Ca ratios thatmight indicate the latter ratio is diagenetically altered (seeElectronic Annex EA-15).

3.1.5. Carbonate clumped isotopes

Carbonate clumped isotope thermometry studies com-plement our independent tests of alteration by further sup-porting the good preservation of most of our samples, andby confirming the poor preservation of samples that weclassify as ‘altered’ by other independent tests (Fig. 8,Electronic Annex EA-2, EA-7). Aliquots of powders usedin clumped isotope measurements were also measured byXRD (see Table 1).

Coral clumped isotope temperatures fall into threegroups. First, most samples have clumped isotope tempera-tures around 25–30 �C, consistent with growth in warmsurface waters. The clumped isotope temperatures of thesesamples give additional evidence for their excellentpreservation.

Second, specimens E7, R1, R9, and R10 give averageclumped isotope temperatures P40 �C, which are likelyincompatible with early Cenozoic and Mesozoic surfaceocean temperatures (Pearson et al., 2001, 2007; Zachoset al., 2001; Keating-Bitonti et al., 2011). We assume R1,R9 and R10 underwent burial diagenesis, and exclude thesesamples from our seawater chemistry reconstructions.Sample E7 was shown to be well preserved based on ourother tests. This sample was infilled with calcite cement thatcould not be separated in drilling the powder sample, andthe high clumped isotope temperature probably results

from contamination by this secondary calcite (35%;Table 1). The powder from Sample Tr3, which also pro-duced a relatively high clumped isotope temperature(38 �C), is characterized by a high percentage of calcitederived from infilling cement (37%) as well. We retain bothsamples in our reconstructions of Sr/Ca and Mg/Ca basedon SIMS data, but flag the points in Fig. 10 (red squaresinstead of red circles) to indicate some ambiguity in thequality of their preservation.

Third, some specimens yield relatively cold clumped iso-tope temperatures. These temperatures could result fromgrowth in cool subsurface waters, meteoric diagenesis, orcoral vital effects; for some species of shallow water coral,clumped temperatures have been found to underestimatetrue growth temperatures by up to �12 �C (Ghosh et al.,2006; Saenger et al., 2012). We suggest that our clumpedisotope measurements give robust minimum temperaturesfor well-preserved samples but cannot be used to infer abso-lute temperatures until vital effects in shallow water coralsare better understood.

3.1.6. Sr Isotopes

Sr isotopes measured on a subset of fossil corals alsoindicate good preservation of our specimens (Fig. 9,Electronic Annex EA-8). Our Sr isotope criterion for goodpreservation is that the 87Sr/86Sr ratio measured in a coralmust agree (within analytical uncertainty) with the87Sr/86Sr ratio of seawater at the stratigraphic age of thesample. The ages of most of our samples are independentlyknown within ±5 My or better, but in some cases ages areonly constrained to a geologic epoch. While other studieshave found a small fraction of well preserved fossil arago-nites (though not corals) that apparently deviate from theSr isotope curve for reasons other than poor preservation(McArthur et al., 1994; Ivany et al., 2008; Marcano et al.,2009), we make the conservative assumption that anoma-lous Sr isotope ratios signify diagenetic alteration.

Using independently estimated ages, all but one of the�45 powders analyzed give 87Sr/86Sr ratios that are withinerror of those predicted by the McArthur et al. (2001) Srisotope curve (Fig. 9, Electronic Annex EA-8). It is surpris-ing that even samples containing some calcite give Sr iso-tope ages consistent with the expected age. There arethree possible explanations for this observation. First, andmost likely, the Sr content of the secondary calcite is toolow to shift the original Sr isotope ratio preserved in the pri-mary aragonite phase. Second, precipitation of secondaryminerals may have occurred relatively soon after the coralbuilt its skeleton in a solution with a similar compositionas contemporaneous seawater. Third, the calcite may haveprecipitated later, but from a diagenetic fluid with the sameSr isotope composition as the original skeleton.

A single coral, the Paleocene-age specimen Pa2, gave an87Sr/86Sr age that disagreed with the expected age; the87Sr/86Sr ratio measured was too high by about 0.00005.This deviation is small and the measured value falls nearlywithin the range of input data to the Sr isotope curve forthe Paleocene (McArthur et al., 2001). Leaching of radio-genic Sr from silicate minerals having a high 87Sr/86Sr ratiomay also account for the slightly elevated composition

0.7076

0.7080

0.7084

0.7088

0.7092

00 50 100 150 200 250 10 20 30 40 50 60Geologic Age (Ma) Geologic Age (Ma)

0.7070

0.7075

0.7080

0.7085

0.7090

Sr/

Sr Expected Age (Ma)

2-9% calciteInfilled sample, > 10% calciteInfilled sample, > 1% silicates

< 1% secondary mineralsMcArthur et al. (2001) a b

Fig. 9. (a)87Sr/86Sr of coral samples vs. geologic age, overlying the Sr isotope seawater curve of McArthur et al. (2001). (b) Enlarged versionof the last 60 Ma. Dashed black lines correspond to the mismatch between the Sr isotope age for sample Pa2 and the expected age.

Geologic Age (Ma)0 10 20 30 40 50

0

2

4

6

8

01

32

54

Coral M

g/Ca Se

awat

er M

g/C

a

250

Fossil Corals - passed all tests (this study)Fossil Corals - failing 1 test (this study)CaCO Veins (Coggon et al., 2010)CaCO Veins (Rausch et al., 2013)

Brine Inclusions (see caption) Echinoderms (Dickson, 2002; 2004)

Gastropods (Tripati et al., 2009)Gastropods (Sosdian et al., 2012)

Benthic Foraminifera (Lear et al., 2003)Fish Teeth (Balter et al., 2011)

Brachiopods, Belemnites, and Rudists(Steuber and Veizer, 2002)

Fossil Corals (Griffiths et al., 2013)Fossil Corals (Ivany et al., 2004)

Large Benthic Forams (Evans et al., 2013)

0

5

10

15

0

2

4

6

8

0 50 100 150 200Geologic Age (Ma)

0

5

10

15

20

0

1

2

3

4

5 Coral M

g/Ca (m

mol/m

ol)C

oral Sr/Ca (m

mol/m

ol)

Seaw

ater

Mg/

Ca

(mol

/mol

)Se

awat

er S

r/Ca

(mm

ol/m

ol)

modern seawater

a

b

c

modern seawater

modern seawater

Fig. 10. (a) Seawater Mg/Ca and (b) Sr/Ca inferred from various proxies. (c) Seawater Mg/Ca (as in A) for the last 50 Ma. Error barsrepresent 2r S.E. of the mean. Red circles correspond to results from well preserved fossil corals (this study). Red squares with black bordersrepresent fossil corals (this study) that passed all but one of our diagenetic tests. Brine inclusion data are from Lowenstein et al. (2001, 2003,2005), Horita et al. (2002), Brennan et al. (2004, 2013), Timofeeff et al. (2006).


observed, as the powder XRD pattern for this specimenindicates the presence of quartz and/or other aluminosili-cate minerals (Electronic Annex EA-4). We retain this sam-ple as part of our suite, but flag it in Fig. 10 (red squareinstead of red circle) to show that it has failed our Srisotope test.

Overall, our studies indicate that, by using multiplediagenetic indicators, it is possible to find well preservedscleractinian corals, or domains within skeletons exhibitinggood preservation, as old as Triassic in age. These speci-mens appear to retain their original chemical and isotopiccomposition and can be used for paleoenvironmentalreconstructions. Many samples with partially preserved

skeletons may not be suitable for bulk analysis of powders,but it is possible to obtain primary geochemical signaturesfrom these specimens using microanalysis techniques suchas SIMS (marked by asterisks in Table 1). We find thatsome other samples containing a mixture of primary andsecondary material (e.g., R1, R3 and R4) are too compro-mised to extract original paleoenvironmental signatureseven with the use of microanalysis techniques.

3.2. Secular variations in seawater Mg/Ca

The average Mg/Ca ratio of 2 modern surface dwellingcorals measured in this study using SIMS is

aMn/

Ca

(mm

ol/m

ol)

10

10

10

10

10

0

1

0.4

0.6

0.8

2

4

6

8

50 100 150 200 250Geologic Age (Ma)

Coral M

g/Ca (m

mol/m

ol)Log S

cale1

1

10

0.4

0.4

0.60.8

0.6

0.8

2

2

2030

4

68

4

6

8

b

c

Coral Mg/Ca including high Mn/Ca spots

Average coral Mg/Ca (log scale)

Minimum coral Mg/Ca (log scale)

Fig. 11. (a) Average coral Mg/Ca calculated including spots withhigh Mn/Ca ratios (log scale). Mg/Ca ratios for older specimensare skewed toward higher values. Error bars represent 2r S.E. (b)Average coral Mg/Ca (as in Fig. 10a) plotted on a log scale. Errorbars represent 2r S.E. (c) Minimum Mg/Ca ratios measured infossil corals, plotted on a log scale (error bars are not shownbecause each symbol denotes a single SIMS spot analysis andinternal errors are smaller than the symbol size).


4.0 ± 0.3 mmol/mol, which is similar to values reported inother SIMS studies of modern coral (Allison and Finch,2007) and previously unpublished bulk measurements ofmodern surface corals (n = 123) (Electronic Annex EA-16).Using these modern SIMS and bulk Mg/Ca data, we calcu-lated a partition coefficient KD

Mg/Ca = (Mg/Ca)coral/(Mg/Ca)seawater = 7.0 � 10�4, which we applied to reconstructseawater Mg/Ca from fossil coral data.

Our reconstruction of the seawater Mg/Ca ratio, calcu-lated from the average Mg/Ca composition of individualfossil corals and the distribution coefficient, is shown inFig. 10a. This history is qualitatively and quantitativelyconsistent with histories derived from other archives,including brine inclusions in halite, hydrothermal CaCO3

veins, fossil echinoderms, and a few studies of other fossilcorals. Seawater Mg/Ca ratios inferred from corals arelowest during the Jurassic and Cretaceous, and exhibit anearly 5-fold increase from the Late Mesozoic to thepresent.

Because coral Mg/Ca is sensitive to temperature(Mitsuguchi et al., 1996), we have assessed the impact oftemperature variability for our reconstruction. Coral Mg/Ca ratios are positively correlated with temperature andexhibit a dependence of �0.13 mmol/mol �C�1

(Mitsuguchi et al., 1996), or about 3% �C�1. Based onour knowledge of growth environments (see ElectronicAnnex EA-1), we estimate that fossil coral growth tempera-tures ranged between 20 and 35 �C for the bulk of our sam-ple set (Keating-Bitonti et al., 2011; Littler et al., 2011). Asa result, we should expect to find ±25% variability aboutthe average of our Mg/Ca ratios for corals of the same geo-logic age, which is roughly compatible with our data. Thisrange of ±25%, although large as illustrated by the spreadin the modern data, is far smaller than the Mg/Ca increasesince the Late Cretaceous (about 400%). There are also afew specimens in our dataset (Pa4 from Seymour Islandand Mi7 from Poland) that likely grew at temperatures<20 �C. Evidence for sample Pa4 comes from clumped iso-tope analyses of Eocene-age bivalve shells from SeymourIsland (Douglas et al., 2014). Evidence for sample Mi7comes from estimates of middle Miocene sea surface tem-peratures summarized in You et al. (2009). Cold growthtemperatures have the potential to bias Mg/Ca ratios inthese samples toward lower values. However, we find thatneither sample defines the minimum of our Mg/Ca recordand both fall well within the range of values measured forfossil corals of similar geologic age.

We also consider how Cenozoic cooling may affect fossilcoral Mg/Ca ratios. If we conservatively estimate thattropical temperatures were �4 �C higher during the EarlyCenozoic than today (Keating-Bitonti et al., 2011), and ifwe remove the temperature dependence of coral Mg/Caassociated with this change, we calculate that reconstructedMg/Ca ratios during the Cretaceous should be �13% lowerthan the measured values suggest. In this way, accountingfor the temperature change related to Cenozoic coolingslightly amplifies (by about 0.1 mol/mol) the differencebetween low seawater Mg/Ca ratios in the Mesozoic andhigh seawater Mg/Ca ratios in the late Cenozoic. This cal-culation confirms that trends observed in the coral Mg/Ca

data cannot be explained by a response to temperaturechange.

To further illustrate the robustness of the Mg/Ca changeinferred from fossil corals, we plot three additional repre-sentations of Mg/Ca ratios measured in corals against theirgeologic age (Fig. 11). Fig. 11a shows averages of coral Mg/Ca ratios including high-Mn analyses and demonstrates theimpact of using Mn/Ca as a diagenetic indicator. Data arecolored according to Mn content. While it may still be pos-sible to observe the general trend toward higher Mg/Caratios between the Cretaceous and today, many of the aver-age Mg/Ca ratios plotted (especially for samples of Eoceneage and older) are skewed toward higher values. The high-est Mg/Ca ratios plotted in Fig. 11a correspond to sampleswith high Mn/Ca ratios, consistent with the observationthat high Mn/Ca ratios occur with high Mg/Ca ratios (aswell as low Sr/Ca ratios: see Fig. 5). These observationsreinforce two conclusions made previously. First, Mn/Cais a strong indicator of small-scale recrystallization inaragonitic fossil corals. Second, it is clear that some gener-ally well preserved corals exhibit small-scale recrystalliza-tion. Microanalysis techniques such as those employedhere are essential for recovering primary geochemical infor-mation from these samples.

Fig. 11b plots average coral Mg/Ca ratios (as in Fig. 10)on a logarithmic scale. The variance in average Mg/Ca


measured in samples of similar geologic age does notappear to greatly change with time, further supporting thegood preservation of our samples. Fig. 11b also moreclearly shows that the rise in seawater Mg/Ca began around80 Ma. Parenthetically, the linear increase shown in sea-water Mg/Ca with time during the Cenozoic (when plottedon a logarithmic scale as in Fig. 11b and c) can be explainedby invoking a constant rate of [Mg] rise and [Ca] fall.

Fig. 11c shows the minimum Mg/Ca ratio (rather thanthe mean) of all SIMS spots measured in each coral speci-men. This representation would be appropriate if coralsprecipitate their skeletons from a closed batch of seawatervia a Rayleigh distillation process, and the most Sr-richand Mg-poor spot approximates initial precipitation fromunfractionated seawater. Fig. 11c still portrays a Mg/Cachange in fossil corals of similar magnitude and timing asobserved in the plot of average coral Mg/Ca (Fig. 11b),but the scatter is greater. Part of the scatter in minimumMg/Ca values may result from the fact that we have notalways captured the initial phase of the Rayleigh process.

Our more complete Mg/Ca record allows us to deter-mine when seawater Mg/Ca began to increase toward itspresent day value (�80 Ma), and suggests that theCenozoic transition from calcite to aragonite sea (Mg/Ca = 2 mol/mol according to Hardie, 1987) occurred at�40 Ma. At least half of the total change in seawaterMg/Ca between the middle Cretaceous and today appearsto have occurred within the last 15–20 My. Our data alsostrengthen the link between low Cretaceous seawater Mg/Ca and the presence of Late Cretaceous-age corals with pri-mary calcite skeletons (Stolarski et al., 2007).

Surprisingly, the coral archive suggests that Mg/Caratios during the Late Triassic were low – about 1/3 ofthe modern ratio (Figs. 10a and 11). This period is generallythought to be a time when inorganic CaCO3 precipitatedfrom seawater as aragonite (‘aragonite seas’), implying highseawater Mg/Ca ratios. Our results, along with data frombrine inclusions and fossil echinoderms (Fig. 10a;Lowenstein et al., 2001, 2003, 2005; Dickson, 2002, 2004;Horita et al., 2002; Timofeeff et al., 2006; Brennan et al.,2013), instead suggest that the Triassic transition from highMg/Ca seawater to low Mg/Ca seawater may haveoccurred prior to 230 Ma. The apparent mismatch betweenthe timing of Mg/Ca and mineralogical change during theTriassic, which is also observed for the Carboniferous(Holt et al., 2014), may be related to the observation thatcarbonate mineralogy also depends on seawater SO4

2�, tem-perature, alkalinity, saturation state, and pCO2 (Burtonand Walter, 1991; Morse et al., 1997; Lee and Morse,2010; Bots et al., 2011). This improved understanding ofthe timing of changes in seawater Mg/Ca may be usefulin helping to test key hypotheses that seek to determinethe geologic controls on seawater chemistry and long-termclimate (e.g., changes in the rates of dolomitization throughtime, changes in the rates of either low or high temperaturehydrothermal alteration, or changes in the river flux of Mgrelative to Ca: Wilkinson and Algeo, 1989; Hardie, 1996;Stanley and Hardie, 1998; Horita et al., 2002; Holland,2005; Broecker, 2013).

3.3. Secular variations in seawater Sr/Ca

We calculate seawater Sr/Ca ratios using a partitioncoefficient, KD

Sr/Ca = (Sr/Ca)coral/(Sr/Ca)seawater = 1.13,based on an average of modern coral SIMS and bulk mea-surements as in Section 3.2 for Mg/Ca (Fig. 10b; Allisonand Finch, 2004; Electronic Annex EA-16). Because coralSr/Ca ratios are also slightly temperature sensitive, (�1%decrease in Sr/Ca for every 1 �C increase in temperature),we have assessed the effects of temperature on fossil coralSr/Ca (Beck et al., 1992, 1997; de Villiers et al., 1994;McCulloch et al., 1994; Correge et al., 2004; Watanabeet al., 2011). Assuming (as noted earlier for the case ofMg/Ca ratios) that most fossil corals in our sample set livedin waters ranging in temperature from 20 to 35 �C, wewould expect a range in coral Sr/Ca of ±7.7% for coralsof similar geologic age. This range of variability is similarto the observed variability in reconstructed Sr/Ca of about±10% for corals of the same geologic age. We can also cal-culate the magnitude of temperature-induced Sr/Ca changeas a consequence of Cenozoic cooling. A cooling of 4 �Cbetween the Early Paleogene and today as estimated pre-viously would result in a �4% increase in coral Sr/Ca ratiosbetween the Paleogene and today. This temperature-induced change is much smaller than the observedCenozoic change and in the opposite direction, suggestingthat temperature effects do not significantly bias ourreconstructions.

Our fossil coral record is generally consistent with otherbiogenic carbonate-based records of seawater Sr/Ca (Learet al., 2003; Tripati et al., 2009; Balter et al., 2011;Sosdian et al., 2012; Evans et al., 2013; Fig. 10b) but dis-agrees with two independent records from hydrothermalcalcium carbonate veins (CCV) (Coggon et al., 2010;Rausch et al., 2013). We favor the fossil Sr/Ca recordsrather than the CCVs. The calculation of seawater Sr/Caratios from fossil biogenic carbonates is based on exhaus-tive observations of Sr/Ca in modern CaCO3 skeletons. Itis also based on empirical results from culture experimentsshowing that the Sr/Ca ratio of CaCO3 skeletons scaleswith that of seawater (Lewin and Chow, 1961; Lorensand Bender, 1980; Swart, 1981). Such a detailed empiricaland mechanistic basis is lacking for Sr/Ca in CCVs, andthe precipitation rate-dependence of the Sr/Ca ratio in inor-ganic calcites (e.g., Lorens, 1981; Tang et al., 2008) isproblematic.

Several fossil coral samples dating to the Early Cenozoichave Sr/Ca ratios �15% higher than modern and Sr/Ca infossil corals decreases steadily between the Early Cenozoicand present. A single sample dated to �71 Ma by Sr-iso-tope stratigraphy has the highest Sr/Ca ratio in our record.This result supports the inference of high seawater Sr/Caratios during the Late Cretaceous based on benthic forami-nifera and fossil fish teeth (Lear et al., 2003; Balter et al.,2011), but needs to be confirmed with additional samples.Our few samples older than 100 Ma have Sr/Ca ratios simi-lar to modern, consistent with a Mesozoic and Paleozoicrecord from brachiopods, belemnites and rudists (Steuberand Veizer, 2002).


Despite general agreement between biogenic carbonaterecords, there also are differences between the variousrecords during the Cenozoic. Although there are largeuncertainties in the gastropod seawater Sr/Ca records thatoverlap with the other biogenic records (Tripati et al.,2009; Sosdian et al., 2012), the gastropods give higher aver-age estimates for seawater Sr/Ca. This may be partlyrelated to uncertainties in interspecific offsets and tempera-ture sensitivities (Sosdian et al., 2012). In addition, there ismodest divergence between fossil corals, benthic foramini-fera, and fish teeth during the middle Cenozoic (Learet al., 2003; Balter et al., 2011). We suggest that the Sr/Ca history of seawater is likely bounded by our fossil coralsand the benthic foraminifera and fish teeth records.

4. CONCLUSION

Studies of the mineralogy, crystal habit, clumped isotopecomposition, Sr isotope composition, and trace element geo-chemistry (especially Mn/Ca) of fossil corals have allowed usto constrain the extent of diagenetic alteration in �60 wellpreserved fossil corals. SIMS studies of Sr/Ca and Mg/Cain these samples suggest that well preserved fossil coralscan produce useful records of seawater paleochemistry.These samples can also be used in future studies to exploreadditional seawater properties of interest. Many of our sam-ples are from museum or government collections and areessentially in the public domain. We find that seawater Sr/Ca was higher than today during the Late Cretaceous andEarly Paleogene, consistent with other biogenic records ofSr/Ca. Our Mg/Ca record agrees with existing records andsuggests low seawater Mg/Ca during the Mesozoic and highseawater Mg/Ca ratios today. The initial increase in Mg/Catowards the present value appears to occur at �80 Ma, andthe Cenozoic transition from calcite to aragonite seas occursin the Middle to Late Eocene. Sr/Ca and Mg/Ca recordsfrom well preserved fossil corals may help constrain majorelement cycling on Earth’s surface and the relationshipbetween seawater chemistry and Earth’s climate state.

ACKNOWLEDGEMENTS

We thank Yunbin Guan (California Institute of Technology)for his help with SIMS analyses, Gerald Poirier (PrincetonImaging and Analysis Center) for assistance with SEM andXRD, and Guillaume Paris (California Institute of Technology)for his help with sulfate concentration analyses. This manuscripthas benefitted tremendously from helpful discussions with JohnM. Eiler (California Institute of Technology), John A. Higgins(Princeton University), Alex C. Gagnon (University ofWashington) and helpful comments from Silke Severmann (AE),Tim Lowenstein, and two anonymous reviewers. We also thankStephen A. Cairns, Tim Coffer (Smithsonian Institution), RogerPortell (Florida Museum of Natural History), the USGS CoreResearch Center, Bill Thompson (WHOI), Gregory P. Dietl(Paleontological Research Institution), and Linda Ivany(Syracuse University) for contributing samples for this work. Thework of JS was supported in part by the Polish-NorwegianResearch Programme operated by the National Centre forResearch and Development under the Norwegian Financial

Mechanism 2009–2014 in the frame of Project Contract No Pol-Nor/196260/81/2013. Cathodoluminescence imaging was per-formed in the NanoFun Laboratory (Institute of Paleobiology)co-financed by the European Regional Development Fund withinthe Innovation Economy Operational Programme POIG.02.02.00-00-025/09. We gratefully acknowledge support from thePrinceton BP Amoco Carbon Mitigation Initiative, and from theFrank Harrison Tuttle Memorial Fund for Invertebrate studies.

APPENDIX A. SUPPLEMENTARY DATA

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.gca.2015.03.018.

REFERENCES

Allison N. and Finch A. A. (2004) High-resolution Sr/Ca recordsin modern Porites lobata corals: Effects of skeletal extensionrate and architecture. Geochem. Geophys. Geosyst. 5, 1–10.

Allison N. and Finch A. A. (2007) High temporal resolution Mg/Ca and Ba/Ca records in modern Porites lobata corals.Geochem. Geophys. Geosyst. 8, 1–11.

Allison N. and Finch A. A. (2010) d11B, Sr, Mg and B in a modernPorites coral: The relationship between calcification site pH andskeletal chemistry. Geochim. Cosmochim. Acta 74, 1790–1800.

Allison N., Finch A. A., Webster J. M. and Clague D. A. (2007)Palaeoenvironmental records from fossil corals: The effects ofsubmarine diagenesis on temperature and climate estimates.Geochim. Cosmochim. Acta 71, 4693–4703.

Balter V., Lecuyer C. and Barrat J. (2011) Reconstructing seawaterSr/Ca during the last 70 My using fossil fish tooth enamel.Palaeogeogr. Palaeoclimatol. Palaeoecol. 310, 133–138.

Bar-Matthews M., Wasserburg G. and Chen J. (1993) Diagenesisof fossil coral skeletons: Correlation between trace elements,textures, and 234U/238U. Geochim. Cosmochim. Acta 57, 257–276.

Beck J. W., Edwards R. L., Ito E., Taylor F. W., Recy J., RougerieF., Joannot P. and Henin C. (1992) Sea-surface temperaturefrom coral skeletal strontium/calcium ratios. Science 257, 644–647.

Beck J., Recy J., Taylor F., Edwards R. and Cabioch G. (1997)Abrupt changes in early Holocene tropical sea surface tem-perature derived from coral records. Nature 385, 705–707.

Bender M. L. (1973) Helium–uranium dating of corals. Geochim.

Cosmochim. Acta 37, 1229–1247.Blattler C., Henderson G. M. and Jenkins W. J. (2012) Explaining

the Phanerozoic Ca isotope history of seawater. Geology 40,843–846.

Bots P., Benning L. G., Rickaby R. E. M. and Shaw S. (2011) Therole of SO4 in the switch from calcite to aragonite seas. Geology

39, 331–334.Brahmi C., Kopp C., Domart-Coulon I., Stolarski J. and Meibom

A. (2012) Skeletal growth dynamics linked to trace-elementcomposition in the scleractinian coral Pocillopora damicornis.Geochim. Cosmochim. Acta 99, 146–158.

Brand U. and Veizer J. (1980) Chemical diagenesis of a multi-component carbonate system; 1, Trace elements. J. Sediment.

Res. 50, 1219–1236.Brennan S. T., Lowenstein T. K. and Cendo n D. I. (2013) The

major-ion composition of Cenozoic seawater: The past 36million years from fluid inclusions in marine halite. Am. J. Sci.

313, 713–775.



http://refhub.elsevier.com/S0016-7037(15)00160-X/h0005



















































Brennan S. T., Lowenstein T. K. and Horita J. (2004) Seawaterchemistry and the advent of biocalcification. Geology 32, 473–476.

Broecker W. (2013) How to think about the evolution of the ratioof Mg to Ca in seawater. Am. J. Sci. 313, 776–789.

Bryan W. H. and Hill D. (1941) Spherulitic crystallization as amechanism of skeletal growth in hexacorals. Proc. R. Soc.

Queensl. 52, 78–91.Burton E. A. and Walter L. M. (1991) The effects of pCO2 and

temperature on magnesium incorporation in calcite in seawaterand MgCl2–CaCl2 solutions. Geochim. Cosmochim. Acta 55,777–785.

Busenberg E. and Plummer L. N. (1985) Kinetic and thermody-namic factors controlling the distribution of SO32� and Na+in calcites and selected aragonites. Geochim. Cosmochim. Acta

49, 713–725.Coggon R. M., Teagle D. A. H., Smith-Duque C. E., Alt J. C. and

Cooper M. J. (2010) Reconstructing past seawater Mg/Ca andSr/Ca from mid-ocean ridge flank calcium carbonate veins.Science 327, 1114–1117.

Cohen A. L. and McConnaughey T. A. (2003) Geochemicalperspectives on coral mineralization. Rev. Mineral Geochem. 54,151–187.

Correge T., Gagan M. K., Beck J. W., Burr G. S., Cabioch G. andLe Cornec F. (2004) Interdecadal variation in the extent ofSouth Pacific tropical waters during the Younger Dryas event.Nature 428, 927–929.

Cuif J. P. and Dauphin Y. (1998) Microstructural and physico-chemical characterization of ‘centers of calcification’ in septa ofsome recent scleractinian corals. Palaontologische Zeitschrift 72,257–269.

Cuif J. P., Dauphin Y., Doucet J., Salome M. and Susini J. (2003)XANES mapping of organic sulfate in three scleractinian coralskeletons. Geochim. Cosmochim. Acta 67, 75–83.

de Villiers S., Shen G. T. and Nelson B. K. (1994) The Sr/Ca –temperature relationship in coralline aragonite: Influence ofvariability in (Sr/Ca)seawater and skeletal growth parameters.Geochim. Cosmochim. Acta 58, 197–208.

Dennis K. J. and Schrag D. P. (2010) Clumped isotope thermome-try of carbonatites as an indicator of diagenetic alteration.Geochim. Cosmochim. Acta 74, 4110–4122.

Dennis K. J., Affek H. P., Passey B. H., Schrag D. P. and Eiler J.M. (2011) Defining an absolute reference frame for ‘clumped’isotope studies of CO2. Geochim. Cosmochim. Acta 75, 7117–7131.

Denniston R. F., Asmerom Y., Polyak V. Y., McNeill D. F., KlausJ. S., Cole P. and Budd A. F. (2008) Caribbean chronos-tratigraphy refined with U–Pb dating of a Miocene coral.Geology 36, 151–154.

Dickson J. A. D. (2002) Fossil echinoderms as monitor of the Mg/Ca ratio of Phaneorzoic Oceans. Science 298, 1222–1224.

Dickson J. A. D. (2004) Echinoderm skeletal preservation: Calcite–aragonite seas and the Mg/Ca ratio of Phanerozoic oceans. J.

Sediment. Res. 74, 355–365.Douglas P. M. J., Affek H. P., Ivany L. C., Houben A. J. P., Sijp

W. P., Sluijs A., Schouten S. and Pagani M. (2014) Pronouncedzonal heterogeneity in Eocene southern high-latitudesea surface temperatures. Proc. Natl. Acad. Sci. 111,6582–6587.

Domart-Coulon I., Stolarski J., Brahmi C., Gutner-Hoch E.,Janiszewska K., Shemesh A. and Meibom A. (2014)Simultaneous extension of both basic microstructural compo-nents in scleractinian coral skeleton during night and daytime,visualized by in situ 86Sr pulse labeling. J. Struct. Biol. 185, 79–88.

Eiler J. M. (2007) “Clumped-isotope” geochemistry – The study ofnaturally-occurring, multiply-substituted isotopologues. Earth

Planet. Sci. Lett. 262, 309–327.Evans D., Muller R., Oron S. and Renema W. (2013) Eocene

seasonality and seawater alkaline earth reconstruction usingshallow-dwelling large benthic foraminifera. Earth Planet. Sci.

Lett. 381, 104–115.Farkas J., Bohm F., Wallmann J., Blenkinsop J., Eisenhauer A.,

van Geldern R., Munnecke A., Voigt S. and Veizer J. (2007)Calcium isotope record of Phanerozoic oceans: Implications forchemical evolution of seawater and its causative mechanisms.Geochim. Cosmochim. Acta 71, 5117–5134.

Faure G. and Mensing T. M. (2005) Isotopes: Principles and

Applications, third ed. John Wiley, New York.Finch A. A. and Allison N. (2008) Mg structural state in coral

aragonite and implications for the paleoenvironmental proxy.Geophys. Res. Lett. 35, L08704. http://dx.doi.org/10.1029/2008GL033543.

Frankowiak K., Mazur M., Gothmann A. M. and Stolarski J.(2013) Diagenetic alteration of Triassic coral from the aragoniteKonservat-Lagerstatte in Alakir Cay, Turkey: Implications forgeochemical measurements. Palaios 28, 333–342.

Gabitov R., Gagnon A., Adkins J. and Eiler J. (2013) AccurateMe/Ca ratio measurements in carbonates by SIMS andnanoSIMS and an assessment of heterogeneity in commoncarbonate standards. Chem. Geol. 356, 94–108.

Gaetani G. A. and Cohen A. L. (2006) Element partitioning duringprecipitation of aragonite from seawater: A framework forunderstanding paleoproxies. Geochim. Cosmochim. Acta 70,4617–4634.

Gaetani G. A., Cohen A. L., Wang Z. and Crusius J. (2011)Rayleigh-based, multi-element coral thermometry: A biomin-eralization approach to developing climate proxies. Geochim.

Cosmochim. Acta 75, 1920–1932.Gagan M. K., Dunbar G. B. and Suzuki A. (2012) The effect of

skeletal mass accumulation in Porites on coral Sr/Ca and d18Opaleothermometry. Paleoceanography 27, 1–16.

Gagnon A. C., Adkins J. F., Fernandez D. P. and Robinson L. F.(2007) Sr/Ca and Mg/Ca vital effects correlated with skeletalarchitecture in a scleractinian deep-sea coral and the role ofRayleigh fractionation. Earth Planet. Sci. Lett. 261, 280–295.

Getty S. R., Asmerom Y., Quinn T. M. and Budd A. F. (2001)Accelerated Pleistocene coral extinctions in the CaribbeanBasin shown by uranium–lead (U–Pb) dating. Geology 28, 639–642.

Ghosh P., Adkins J., Affek H., Balta B., Guo W., Schauble E. A.,Schrag D. and Eiler J. M. (2006) 13C–18O bonds in carbonateminerals: A new kind of paleothermometer. Geochim.

Cosmochim. Acta 70, 1439–1456.Goldfuss A. (1826) Petrefacta Germaniae. Arnz, Dusseldorf.Griffiths N., Muller W., Johnson K. G. and Aguilera O. (2013)

Evaluation of the effect of diagenetic cements on the element/Ca ratios in aragonitic Early Miocene (16 Ma) Caribbeancorals: Implications for ‘deep-time’ paleoenvironmental recon-structions. Palaeogeogr. Palaeoclimatol. Palaeoecol. 369, 185–200.

Hardie L. A. (1987) Dolomitization; a critical view of some currentviews. J. Sediment. Res. 57, 166–183.

Hardie L. A. (1996) Secular variation in seawater chemistry: Anexplanation for the coupled secular variation in the mineralo-gies of marine limestones and potash evaporites over the past600 Myr. Geology 24, 279–283.

Hardie L. A. (2003) Secular variations in Precambrian seawaterchemistry and the timing of Precambrian aragonite seas andcalcite seas. Geology 31, 785–788.























































































http://dx.doi.org/10.1029/2008GL033543

http://dx.doi.org/10.1029/2008GL033543






















































Holland H. D. (2005) Sea level, sediments and the composition ofseawater. Am. J. Sci. 305, 220–239.

Holt N. M., Garcia-Veigas J., Lowenstein T. K., Giles P. S. andWilliams-Stroud S. (2014) The major-ion composition ofCarboniferous seawater. Geochim. Cosmochim. Acta 134, 317–334.

Horita J., Zimmermann H. and Holland H. D. (2002) Chemicalevolution of seawater during the Phanerozoic: Implicationsfrom the record of marine evaporites. Geochim. Cosmochim.

Acta 66, 3733–3756.Huntington K. W., Budd D. A., Wernicke B. P. and Eiler J. M.

(2011) Use of clumped-isotope thermometry to constrain thecrystallization temperature of diagenetic calcite. J. Sediment.

Res. 81, 656–669.Ivany L. C., Peters S. C., Wilkinson B. H., Lohmann K. C. and

Reimer B. A. (2004) Composition of the early Oligocene oceanfrom coral stable isotope and elemental chemistry. Geobiology

2, 97–106.Ivany L. C., Lohmann K. C., Hasiuk F. J., Blake D. B., Glass A.,

Aronson R. B. and Moody R. M. (2008) Eocene climate recordof a high southern latitude continental shelf: Seymour Island,Antarctica. GSA Bull. 120, 659–678.

Keating-Bitonti C. R., Ivany L. C., Affek H. P., Douglas P. andSamson S. D. (2011) Warm, not super-hot, temperatures in theearly Eocene subtropics. Geology 39, 771–774.

Lea D., Shen G. T. and Boyle E. (1989) Coralline barium recordstemporal variability in equatorial Pacific upwelling. Nature 340,373–376.

Lear C. H., Elderfield H. and Wilson P. A. (2003) A Cenozoicseawater Sr/Ca record from benthic foraminiferal calcite and itsapplication in determining global weathering fluxes. Earth

Planet. Sci. Lett. 208, 69–84.Lee J. and Morse J. W. (2010) Influences of alkalinity and pCO2 on

CaCO3 nucleation from estimated Cretaceous compositionseawater representative of “calcite seas”. Geology 38, 115–118.

Lewin R. A. and Chow T. J. (1961) La Enpreno de Strontio enKokolitoforoj. Plant Cell Physiol. 2, 203–208.

Littler K., Robinson S. A., Bown P. R., Nederbragt A. J. andPancost R. D. (2011) High sea-surface temperatures during theEarly Cretaceous Epoch. Nat. Geosci. 4, 169–172.

Lorens R. (1981) Sr, Cd, Mn and Co distribution coefficients incalcite as a function of calcite precipitation rate. Geochim.

Cosmochim. Acta 45, 553–561.Lorens R. B. and Bender M. L. (1980) The impact of solution

chemistry on Mytilus edulis calcite and aragonite. Geochim.

Cosmochim. Acta 44, 1265–1278.Lowenstein T. K., Hardie L. A., Brennan S. T., Hardie L. A. and

Demicco R. V. (2001) Oscillations in Phanerozoic seawaterchemistry: Evidence from fluid inclusions. Science 294, 1086–1088.

Lowenstein T. K., Hardie L. A., Timofeeff M. N. and Demicco R.V. (2003) Secular variation in seawater chemistry and the originof calcium chloride basinal brines. Geology 31, 857–860.

Lowenstein T. K., Timofeeff M. N., Kovalevych V. M. and HoritaJ. (2005) The major-ion composition of Permian seawater.Geochim. Cosmochim. Acta 69, 1701–1719.

Marcano M. C., Mukasa S., Lohman K. C., Stefano C., TavianiM. and Andronikov A. (2009) Chronostratigraphic andpaleoenvironmental constraints derived from the 87Sr/87Sr and18O signal of Miocene bivalves, Southern McMurdo Sound,Antarctica. Global Planet. Change 69, 124–132.

Marshall J. F. and McCulloch M. T. (2002) An assessment of theSr/Ca ratio in shallow water hermatypic corals as a proxy forsea surface temperature. Geochim. Cosmochim. Acta 66, 3263–3280.

McArthur J. M., Kennedy W. J., Chen M., Thirlwall M. F. andGale A. S. (1994) Strontium isotope stratigraphy for LateCretaceous time: Direct numerical calibration of the Sr isotopecurve based on the US Western Interior. Palaeogeogr.

Palaeoclimatol. Palaeoecol. 108, 95–119.McArthur J. M., Howarth R. J. and Bailey T. R. (2001) Strontium

isotope stratigraphy: LOWESS Version 3: Best fit to the marineSr-isotope curve for 0–509 Ma and accompanying look-up tablefor deriving numerical age. Geology 109, 155–170.

McCulloch M. T., Gagan M. K., Mortimer G. E., Chivas A. R.and Isdale P. J. (1994) A high-resolution Sr/Ca and d18O coralrecord from the Great Barrier Reef, Australia, and the 1982–1983 El Nino. Geochim. Cosmochim. Acta 58, 2747–2757.

McCulloch M. T., Tudhope A. W., Esat T. M., Mortimer G. E.,Chappell J., Pillans B., Chivas A. R. and Omura A. (1999)Coral record of equatorial sea-surface temperatures during thepenultimate deglaciation at Huon Peninsula. Science 283, 202–204.

Meibom A., Cuif J. P., Hillion F., Constantz B. R., Juillet-LeclercA., Dauphin Y., Watanabe T. and Dunbar R. B. (2004)Distribution of magnesium in coral skeleton. Geophys. Res.

Lett. 31, L23306. http://dx.doi.org/10.1029/2004GL021313.Meibom A., Mostefaoui S., Cuif J. P., Dauphin Y., Houlbreque F.,

Dunbar R. B. and Constantz B. R. (2007) Biological forcingcontrols the chemistry of reef-building coral skeleton. Geophys.

Res. Lett. 34, L02601. http://dx.doi.org/10.1029/2006GL028657.

Meibom A., Cuif J., Houlbreque F., Mostefaoui S., Dauphin Y.,Meibom K. and Dunbar R. (2008) Compositional variations atultra-structure length scales in coral skeleton. Geochim.

Cosmochim. Acta 72, 1555–1569.Mertz-Kraus R., Brachert T. C., Jochum K. P., Reuter M. and

Stoll B. (2009) LA-ICP-MS analyses on coral growth incre-ments reveal heavy winter rain in the Eastern Mediterranean at9 Ma. Paleogeogr. Paleoclimatol. Paleoecol. 273, 25–40.

Misra S. and Froelich P. N. (2012) Lithium isotope history ofCenozoic seawater: Changes in silicate weathering and reverseweathering. Science 335, 818–823.

Mitsuguchi T., Matsumoto E., Abe O., Uchida T. and Isdale P. J.(1996) Mg/Ca thermometry in coral skeletons. Science 274,961–963.

Morse J. W., Wang Q. and Tsio M. T. (1997) Influences oftemperature and Mg:Ca ratio on CaCO3 precipitates fromseawater. Geology 25, 85–87.

Nothdurft L. D. and Webb G. E. (2009) Earliest diagenesis inscleractinian coral skeletons: Implications for palaeoclimate-sensitive geochemical archives. Facies 55, 161–201.

Paris G., Sessions A. L., Subhas A. V. and Adkins J. F. (2013) MC-ICP-MS measurement of d34S and D33S in small amounts ofdissolved sulfate. Chem. Geol. 345, 50–61.

Passey B. and Henkes G. (2012) Carbonate clumped isotope bondreordering and geospeedometry. Earth Planet. Sci. Lett. 351–

352, 223–236.Paytan A., Kastner M., Campbell D. and Thiemens M. (1998)

Sulfur isotopic composition of Cenozoic seawater sulfate.Science 282, 1459–1462.

Paytan A., Kastner M., Campbell D. and Thiemens M. (2004)Seawater sulfur isotope fluctuations in the Cretaceous. Science

304, 1663–1665.Pearson P. N., Ditchfield P. W., Singano J., Harcourt-Brown K.

G., Nicholas C. J., Olsson R. K., Shackleton N. J. and Hall M.A. (2001) Warm tropical sea surface temperatures in the LateCretaceous and Eocene epochs. Nature 413, 481–487.

Pearson P. N., van Dongen B. E., Nicholas C. J., Pancost R. D.,Schouten S., Singano J. M. and Wade B. S. (2007) Stable warm





























































































http://dx.doi.org/10.1029/2004GL021313

http://dx.doi.org/10.1029/2006GL028657

http://dx.doi.org/10.1029/2006GL028657












































tropical climate through the Eocene Epoch. Geology 35, 211–214.

Pena L. D., Calvo E., Cacho I., Eggins S. and Pelejero C. (2005)Identification and removal of Mn-Mg-rich contaminant phaseson foraminiferal tests: Implications for Mg/Ca past tempera-ture reconstructions. Geochem. Geophys. Geosyst. 6, Q09P02.http://dx.doi.org/10.1029/2005GC000930.

Perrin C. (2003) Compositional heterogeneity and microstructuraldiversity of coral skeletons: Implications for taxonomy andcontrol on early diagenesis. Coral Reefs 22, 109–120.

Pingitore N. E. (1976) Vadose and phreatic diagenesis: Processes,products and their recognition in corals. J. Sediment. Petrol. 46,985–1006.

Pingitore N. E. (1978) The behavior of Zn2+ and Mn2+ duringcarbonate diagenesis; theory and applications. J. Sediment. Res.

48, 799–814.Porter S. M. (2010) Calcite and aragonite seas and the de novo

acquisition of carbonate skeletons. Geobiology 8, 256–277.Rausch S., Bohm F., Bach W., Klugel A. and Eisenhauer A. (2013)

Calcium carbonate veins in ocean crust record a threefoldincrease of seawater Mg/Ca in the past 30 million years. Earth

Planet. Sci. Lett. 362, 215–224.Ries J. B. (2004) Effect of ambient Mg/Ca ratio on Mg fractiona-

tion in calcareous marine invertebrates: A record of the oceanicMg/Ca ratio over the Phanerozoic. Geology 32, 981–984.

Ries J. B., Stanley S. M. and Hardie L. A. (2006) Scleractiniancorals produce calcite, and grow more slowly, in artificialCretaceous seawater. Geology 34, 525–528.

Saenger C., Affek H. P., Felis T., Thiagarajan N., Lough J. M. andHolcomb M. (2012) Carbonate clumped isotope variability inshallow water corals: Temperature dependence and growth-related vital effects. Geochim. Cosmochim. Acta 99, 224–242.

Sandberg P. A. (1983) An oscillating trend in Phanerozoic non-skeletal carbonate mineralogy. Nature 305, 19–22.

Sayani H. R., Cobb K. M., Cohen A. L., Elliott W. C., Nurhati I.S., Dunbar R. B., Rose K. A. and Zaunbrecher L. K. (2011)Effects of diagenesis on paleoclimate reconstructions frommodern and young fossil corals. Geochim. Cosmochim. Acta 75,6361–6373.

Shen G. T., Campbell T. M., Dunbar R. B., Wellington G. M.,Colgan M. W. and Glynn P. W. (1991) Paleochemistry ofmanganese in corals from the Galapagos Islands. Coral Reefs

10, 91–100.Sinclair D. J., Williams B. and Risk M. (2006) A biological origin

for climate signals in corals - Trace element “vital effects” areubiquitous in Scleractinian coral skeletons. Geophys. Res. Lett.

33, L17707. http://dx.doi.org/10.1029/2006GL027183.Sorauf J. E. (1999) Skeletal microstructure, geochemistry, and

organic remnants in Cretaceous scleractinian corals: SantonianGosau beds of Gosau, Austria. J. Paleontol. 73, 1029–1041.

Sosdian S. M., Lear C. H., Tao K., Grossman E. L., O’Dea A. andRosenthal Y. (2012) Cenozoic seawater Sr/Ca evolution.Geochem. Geophys. Geosyst. 13, Q10014. http://dx.doi.org/10.1029/2012GC004240.

Stanley, Jr., G. D. (2003) The evolution of modern corals and theirearly history. Earth-Sci. Rev. 60, 195–225.

Stanley S. M. and Hardie L. A. (1998) Secular oscillations in thecarbonate mineralogy of reef-building and sediment-producingorganisms driven by tectonically forced shifts in seawaterchemistry. Palaeogeogr. Palaeoclimatol. Palaeoecol. 144, 3–19.

Stanley, Jr., G. D. and Swart P. K. (1995) Evolution of the coral-zooanthellae symbiosis during the Triassic: A geochemicalapproach. Paleobiology 21, 179–199.

Steuber T. and Veizer J. (2002) Phanerozoic record of plate tectoniccontrol of seawater chemistry and carbonate sedimentation.Geology 30, 1123–1126.

Stolarski J. (2003) Three-dimensional micro-nanostructural char-acteristics of the scleractinian coral skeleton: A biocalcificationproxy. Acta Palaeontol. Polon. 48, 497–530.

Stolarski J. and Mazur M. (2005) Nanostructure of biogenic versusabiogenic calcium carbonate crystals. Acta Palaeontol. Polon.

50, 847–865.Stolarski J., Meibom A., Przeniosto R. and Mazur M. (2007) A

Cretaceous scleractinian coral with a calcitic skeleton. Science

318, 92–94.Stolarski J., Kitahara M. V., Miller D. J., Cairns S. D., Mazur M.

and Meibom A. (2011) The ancient evolutionary origins ofScleractinia revealed by azooxanthellate corals. BMC Evol.

Biol. 11, 316.Swart P. K. (1981) The strontium, magnesium and sodium

composition of recent scleractinian coral skeletons as standardsfor palaeoenvironmental analysis. Palaeogeogr. Palaeoclimatol.

Palaeoecol. 34, 115–136.Tambutte S., Holcomb M., Ferrier-Page‘s C., Reynaud S.,

Tambutte E., Zoccola D. and Allemand D. (2011) Coralbiomineralization: From the gene to the environment. J. Exp.

Mar. Biol. Ecol. 408, 58–78.Tang J., Kohler S. J. and Dietzel M. (2008) Sr2+/Ca2+ and

44Ca/40Ca fractionation during inorganic calcite formation: I.Sr incorporation. Geochim. Cosmochim. Acta 72, 3718–3732.

Timofeeff M. N., Lowenstein T. K., Martins da Silva M. A. andHarris N. B. (2006) Secular variation in the major-ion chemistryof seawater: Evidence from fluid inclusions in Cretaceoushalites. Geochim. Cosmochim. Acta 70, 1977–1994.

Tripati A. K., Allmon W. D. and Sampson D. E. (2009) Possibleevidence for a large decrease in seawater strontium/calciumratios and strontium concentrations during the Cenozoic. Earth

Planet. Sci. Lett. 282, 122–130.Watanabe T., Suzuki A., Minobe S., Kawashima T., Kameo K.,

Minoshima K., Aguilar Y. M., Wani R., Kawahata H., SowaK., Nagai T. and Kase T. (2011) Permanent El Nino during thePliocene warm period not supported by coral evidence. Nature

471, 209–211.Webb G. E., Nothdurft L. D., Kamber B. S., Kloprogge J. T. and

Zhao J. (2009) Rare earth element geochemistry of scleractiniancoral skeleton during meteoric diagenesis: A sequence throughneomorphism of aragonite to calcite. Sedimentology 56, 1433–1463.

White C. A. (1879) Contribution to paleontology No. 1:Cretaceous fossils of the western states and territories. Annu.

Rep. U.S. Geol. Geogr. Surv. of the Territ. 11, 273–319.Wilkinson B. H. and Algeo T. J. (1989) Sedimentary carbonate

record of calcium and magnesium cycling. Am. J. Sci. 289,1158–1194.

You Y., Huber M., Muller R. D., Poulsen C. J. and Ribbe J. (2009)Simulation of the Middle Miocene Climate Optimum. Geophys.

Res. Lett. 36, L04702. http://dx.doi.org/10.1029/2008GL036571.

Zachos J., Pagani M., Sloan L., Thomas E. and Billups K. (2001)Trends, rhythms, and aberrations in global climate 65 Ma topresent. Science 292, 686–693.

Zhuravlev A. Y. and Wood R. A. (2008) Eve of biomineralization:Controls on skeletal mineralogy. Geology 36, 923–926.

Associate editor: Silke Severmann



http://dx.doi.org/10.1029/2005GC000930







































http://dx.doi.org/10.1029/2006GL027183




http://dx.doi.org/10.1029/2012GC004240

http://dx.doi.org/10.1029/2012GC004240



































































http://dx.doi.org/10.1029/2008GL036571

http://dx.doi.org/10.1029/2008GL036571







Fossil corals as an archive of secular variations in...

Documents

Transcript of Fossil corals as an archive of secular variations in...