Carbonate clumped isotope constraints on Silurianbiogeochem.wustl.edu/wp-content/uploads/2016/08/Cummins_Gotland... · Carbonate clumped isotope constraints on Silurian ocean temperature

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 140 (2014) 241–258

Carbonate clumped isotope constraints on Silurianocean temperature and seawater d18O

Renata C. Cummins a,⇑, Seth Finnegan b, David A. Fike c, John M. Eiler a,Woodward W. Fischer a

a California Institute of Technology, Geological and Planetary Sciences, MC 100-23, Pasadena, CA 91125, United Statesb Department of Integrative Biology, University of California, 1005 Valley Life Sciences Bldg #3140, Berkeley, CA 94720, United States

c Department of Earth and Planetary Sciences, Washington University, St. Louis, MO 63130, United States

Received 12 January 2014; accepted in revised form 19 May 2014; Available online 2 June 2014

Abstract

Much of what we know about the history of Earth’s climate derives from the chemistry of carbonate minerals in thesedimentary record. The oxygen isotopic compositions (d18O) of calcitic marine fossils and cements have been widely usedas a proxy for past seawater temperatures, but application of this proxy to deep geologic time is complicated by diageneticalteration and uncertainties in the d18O of seawater in the past. Carbonate clumped isotope thermometry provides anindependent estimate of the temperature of the water from which a calcite phase precipitated, and allows direct calculationof the d18O of the water. The clumped isotope composition of calcites is also highly sensitive to recrystallization and can helpdiagnose different modes of diagenetic alteration, enabling evaluation of preservation states and identification of the mostpristine materials from within a sample set—critical information for assessing the quality of paleoproxy data generated fromcarbonates.

We measured the clumped isotope composition of a large suite of calcitic fossils (primarily brachiopods and corals),sedimentary grains, and cements from Silurian (ca. 433 Ma) stratigraphic sections on the island of Gotland, Sweden. Substan-tial variability in clumped isotope temperatures suggests differential preservation with alteration largely tied to rock-buffereddiagenesis, complicating the generation of a stratigraphically resolved climate history through these sections. Despite thegenerally high preservation quality of samples from these sections, micro-scale observations of calcite fabric and trace metalcomposition using electron backscatter diffraction and electron microprobe analysis suggest that only a subset of relativelypristine samples retain primary clumped isotope signatures. These samples indicate that Silurian tropical oceans were likelywarm (33 ± 7 �C) and similar in oxygen isotopic composition to that estimated for a “modern” ice-free world (d18OVSMOW of�1.1 ± 1.3&). This result joins the growing body of evidence that suggests the d18O of Earth’s ocean waters has remainedbroadly constant through time.� 2014 Elsevier Ltd. All rights reserved.

1. INTRODUCTION

Despite more than half a century of study, the long-termtemperature history of Earth’s oceans remains poorly

http://dx.doi.org/10.1016/j.gca.2014.05.024

0016-7037/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 6504164224.E-mail address: [email protected] (R.C. Cummins).

constrained. The most widely used proxy for past oceantemperature is the carbonate-water oxygen isotopeexchange thermometer, also known as the carbonate d18Othermometer (Urey, 1947). This thermometer is based onthe temperature dependence of the fractionation betweenthe d18O of carbonate, here reported relative to the VPDBstandard, and the d18O of seawater, here reported relative


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2014.05.024&domain=pdf

242 R.C. Cummins et al. / Geochimica et Cosmochimica Acta 140 (2014) 241–258

to the VSMOW standard. The carbonate d18O proxy hastwo significant complications. First, as with manypaleoenvironmental proxies, it is often difficult to distin-guish the effects of diagenetic alteration from a primaryenvironmental signature. This is an acute challenge associ-ated with carbonate minerals, which commonly undergorecrystallization during diagenesis and burial, altering pri-mary geochemical signals and impacting the quality ofproxy data. Second, the measured d18O of carbonatesdepends not only on the temperature of the water fromwhich the carbonates precipitated, but also on the d18O ofthe fluid (seawater and/or pore fluids). The d18O of seawa-ter can be influenced by several processes, including localevaporation and mixing with meteoric water, as well asglobal changes in the volume and isotopic composition ofcontinental glaciers, all of which are difficult to constrainindependently in the geologic record.

Additionally, the d18O of seawater may undergo globalchanges over much longer timescales than those of glacialcycles. A longstanding and controversial hypothesis in geo-chemistry holds that the d18O of the ocean has increasedgradually but substantially through Earth’s history—from�8& in the Cambrian to its present value of �0&

(Veizer et al., 1999; Jaffres et al., 2007). Gradual changesin d18O are, in principle, possible given current knowledgeof the fluxes impacting the seawater d18O budget. Thelong-term average d18O of all the water on Earth’s surfaceis largely controlled by the balance between two classes ofprocesses within the silicate Earth: exchange during high-temperature alteration of the seafloor near mid-oceanridges, and exchange during low-temperature weatheringof oceanic and continental crust (Muehlenbachs, 1998).High-temperature exchange processes have small oxygenisotope fractionation factors and tend to increase thed18O of ocean water by equilibrating the water with theaverage mantle d18OVSMOW of 5.8& (Muehlenbachs,1998). Low-temperature exchange processes have largeroxygen isotope fractionation factors and tend to decreasethe d18O of seawater by preferentially moving 18O fromwater into rocks. Furthermore, we might reasonablyanticipate that the relative magnitudes of high- and low-temperature weathering have changed over time withcontinental assembly and rifting (Collins, 2003), changesin seafloor spreading and production of oceanic crust(Fornari and Embley, 1995; Muller et al., 2008), climate(Frakes et al., 2005), and sedimentation (Molnar, 2004).

The alternative hypothesis, that the d18O of seawater hasremained constant through time, asserts that the averagetemperature of water–rock interactions has also remainedconstant through time. Under this hypothesis, changes incontinental weathering rate are insignificant because themass of continental material weathered each year is muchsmaller than the mass of seafloor altered by hydrothermalaction (Muehlenbachs and Clayton, 1976; Muehlenbachs,1998); this view is supported by recent measurements ofcalcium isotopes (Caro et al., 2010). Assuming that theaverage temperature of hydrothermal alteration is con-trolled by the critical temperature of water, a physical prop-erty that remains constant unless the depth of the oceanschanges by multiple kilometers, the d18O of seawater is

essentially buffered by interaction with the isotopicallyhomogeneous mantle (Muehlenbachs and Clayton, 1976;Gregory and Taylor, 1981; Gregory, 1991; Muehlenbachs,1998; Jaffres et al., 2007). This alternative hypothesis issupported by measurements of the d18O of conodont apa-tite (Wenzel et al., 2000; Joachimski et al., 2004), ancientophiolites (Muehlenbachs et al., 2003; Turchyn et al.,2013), and fluid inclusions in evaporite minerals (Knauthand Roberts, 1991), all of which imply constant seawaterd18O through the Phanerozoic.

The most widely cited evidence for gradual change inseawater d18O is an observed ca. 8& increase in the averaged18O of marine carbonates from the Cambrian to today(Veizer et al., 1999; Jaffres et al., 2007). However, the com-plications associated with the carbonate d18O proxy allowconflicting interpretations of this trend in carbonate d18O,giving rise to a longstanding controversy. Several studieshave proposed that early Paleozoic carbonates may havelow d18O values because they are more likely to have under-gone diagenetic alteration (Clayton and Degens, 1959;Keith and Weber, 1964; Land, 1995), which tends todecrease the d18O of carbonate (Marshall, 1992; Knauthand Kennedy, 2009). In response to this concern, Veizeret al. (1999) measured the d18O of calcite fossils that werescreened for diagenetic alteration using trace metal compo-sitions and micro-scale studies of calcite fabrics. Thesescreened fossils also record increasing d18O through thePhanerozoic, arguing against diagenetic alteration as acause of the low d18O values in the early Paleozoic. Analternative hypothesis holds that oceans were much warmerin early Paleozoic time (Knauth and Epstein, 1976; Karhuand Epstein, 1986). Assuming that the d18O of the oceanswas the same as today, Ordovician tropical oceans musthave been �45 �C to produce carbonates with the observedaverage d18OVPDB of �5.9&. Such warm temperature esti-mates present an apparent paradox with evidence for alarge glaciation in the Late Ordovician (Veizer et al., 1986).

Carbonate clumped isotope thermometry supplies aunique insight into this problem because it provides datathat can help resolve the two main complications of the car-bonate d18O proxy: the clumped isotope composition of acarbonate is independent of the d18O of water and highlysensitive to diagenetic alteration at temperatures differentfrom those of original deposition. The extent to which13C and 18O occur together, or “clump” within the samecarbonate ion is a function of the temperature at whichthe carbonate precipitates (Eiler, 2011), and isotopic clump-ing in a wide diversity of shallow-water calcifiers has beenshown to record ambient water temperatures (Ghoshet al., 2006; Tripati et al., 2010; Saenger et al., 2012;Henkes et al., 2013). There is some uncertainty regardingthe uniformity of these calibrations, but all previous studiessuggest this phenomenon is a reliable and reasonably pre-cise thermometer when calibrated against carbonates ofthe same type as analyzed in the same lab. Clumped isotopetemperature estimates can in turn be used, with the d18O ofshell calcite and the experimentally determined temperaturedependence of the oxygen isotope fractionation betweenwater and calcite (Kim and O’Neil, 1997), to calculate thed18O of the water from which the calcite was precipitated.

R.C. Cummins et al. / Geochimica et Cosmochimica Acta 140 (2014) 241–258 243

Fossils that have been deeply buried in lithified sedi-ments can experience alteration at temperatures greaterthan those of original deposition. This alteration is oftenassociated with recrystallization, which can change thed18O and trace-metal content of carbonate mineral phasesbut is frequently difficult to detect. Clumped isotopes,however, can help to detect alteration at high temperatures,because any carbonate that is recrystallized at high temper-ature will have an elevated clumped isotope temperature,compared to the temperature of the pristine fossil calcite.If recrystallized material is sampled along with pristine fos-sil material during analysis (e.g. cements filling void spacewithin a fossil), the clumped isotope temperature of thesample will be significantly higher than the temperature ofthe seawater where the organism lived. We demonstratehere that carbonate clumped isotopes can provide a moresensitive indicator of diagenetic recrystallization than thed18O or the trace metal content of a given carbonate phase,particularly in rocks that undergo nearly rock-buffereddiagenesis.

In carbonate strata that have been deeply buried andsubjected to high temperatures for extended periods oftime, diagenetic alteration can also take the form ofclosed-system diffusive reordering of atoms within the min-eral lattice (Huntington et al., 2011; Dennis et al., 2013;Quade et al., 2013; VanDeVelde et al., 2013). During thisprocess, isotopes of carbon and oxygen diffuse throughthe calcite mineral lattice, which decreases the degree ofclumping and increases the measured temperature of thecalcite. The amount of diffusion that occurs is controlledby both the burial temperature and the length of time forwhich the sample remains at this temperature. The effectsof closed-system diffusive reordering can be difficult torecognize, because the reordering does not affect the bulkisotopic composition of the calcite, and it occurs withoutdissolution and re-precipitation, which means it is alsonot expected to change the calcite fabric or trace metal con-tent (Eiler, 2011). One way to recognize diffusive reorderingis through its effects on the distribution of temperatures in asample suite; diffusive reordering raises the mean tempera-ture without changing the variance in temperature. In addi-tion, the degree of diffusive reordering can be constrainedby independent indicators of any given sedimentary succes-sion’s time-temperature history, such as the conodont coloralteration index (CAI) or rock evaluation parameters oforganic matter maturity. Recent studies of closed-systemdiffusive reordering suggest the time-temperature condi-tions required to increase the clumped isotope temperatureof a sample by more than the measurement precision of�1 �C are more than sufficient to change the color of cono-dont elements, raising the CAI above the pristine value ofone (Epstein et al., 1976; Dennis and Schrag, 2010; Passeyand Henkes, 2012; Stolper et al., 2011). Therefore, insample suites with CAI = 1 or pre-oil window catagenesis,elevated clumped isotope temperatures can be reasonablyinterpreted to be the result of diagenetic alteration due torecrystallization.

In addition to being highly sensitive to diagenetic alter-ation, carbonate clumped isotope data, when combinedwith bulk O and C isotope analyses of the same materials,

can be diagnostic of specific diagenetic processes andregimes (recrystallization vs. closed-system diffusivereordering; high vs. low water/rock ratio conditions; mete-oric vs. marine pore fluids), and can be used to identify themost pristine end-member of a sample set (Eiler, 2011). Ofparticular interest for this study, as carbonates are burieddeeper and the temperature increases, the equilibrium oxy-gen isotope fractionation between water and carbonategrows smaller, preferentially driving 18O into the water.During rock-buffered diagenesis, this causes the d18O ofthe water to increase, while the d18O of the altered carbon-ate is not affected. In contrast, during a water-buffered dia-genetic regime, increasing temperature instead causes thed18O of the altered carbonate to decrease. Diagenesis inthe presence of meteoric water usually also causes thed18O of the carbonate to decrease because meteoric watertends to be depleted in 18O (Marshall, 1992; Wenzel et al.,2000; Jaffres et al., 2007). We demonstrate here that this dis-tinction has important implications for the estimation ofthe sea surface temperature and d18O of seawater in thepast.

2. GEOLOGICAL SETTING AND MATERIALS

Samples were collected from several sections exposedon the island of Gotland, Sweden that span the latestLlandovery through Wenlock epochs (~434 – 430 Ma) ofthe Silurian Period (Fig. 1). Ages of the formations andcorrelations between outcrops are constrained by cono-dont and graptolite biostratigraphy (Jeppsson et al.,2006). The sections record shelf, carbonate platform andbackreef lagoon sediments deposited on the margin ofthe Baltic basin, which was constrained to be 19� 5:1�

south of the equator during the Silurian Period (Torsviket al., 1992).

The oldest exposed strata on Gotland comprise theLower Visby Formation, which consists of marls interbed-ded with fine-grained limestones deposited below stormwave base (Fig. 2). Fossils are abundant throughout thisunit (Samtleben et al., 1996; Calner et al., 2004a). The con-tact between the Lower and Upper Visby formations ismarked by a condensed interval with abundant fossils ofthe large solitary rugose coral genus Phaulactis

(Samtleben et al., 1996; Jeppsson, 1997; Munnecke et al.,2003; Jeppsson et al., 2006). This contact occurs just abovethe Llandovery-Wenlock boundary as defined by conodontbiostratigraphy (Jeppsson, 1983; Mabillard and Aldridge,1985; Aldridge et al., 1993). We defined the Phaulactis

bed as a stratigraphic datum to align the sections, exceptthose in the Slite Formation, which are measured relativeto an arbitrary bed in the Slite Formation. The contactbetween the Lower and Upper Visby formations is alsocharacterized by a facies change that represents a maximumflooding interval (Calner et al., 2004b). Carbonate contentincreases in the Upper Visby Beds, with a decreasingfrequency of marl beds and the appearance of bioclasticlimestones and reef mounds. Ripple bedforms andcross-stratification indicate deposition above storm wavebase. The Upper Visby Formation is particularly fossilifer-ous (Calner et al., 2004a; Samtleben et al., 1996).

1,2 5

6,7

8

11

3,4,1213L & U Visby Fm.

Högklint Fm.

Hangvar Fm.

Slite Group

Halla Fm.

Tofta Fm.

Klinteberg Fm.20 km

N

Swed

en

Norw

ay

Poland

Gotland

Gotland

Fig. 1. Geologic map of the northern half of the island of Gotland, Sweden, modified from Calner et al. (2004a). Locations of stratigraphicsections are marked and labeled with a location number, which corresponds to the sample locations in Table EA1.

-2 0 2 4 6 -6 -5 -4

L. Visby

U. Visby

Högklint

Tofta

Slite

0 m12 m

47 m55 m

145 m

430.5 Ma

433.4 Ma

KeyInterbedded limestones and marlsOncoidal limestoneFossiliferous limestone

δ13CVPDB (‰) δ18OVPDB (‰)

Fig. 2. Lithology and isotope stratigraphy of the sampled sequence. On the left, the stratigraphic column shows the generalizedlithostratigraphic units sampled in this study. On the right are the carbon and oxygen isotopic compositions of brachiopods, collected fromthe same sections by Bickert et al. (1997). Isotope data is from Bickert et al. (1997).


Above the Upper Visby Beds, the Hogklint Formationconsists of algal and crinoidal limestones with some reefmounds, deposited above storm wave base (Riding andWatts, 1991; Samtleben et al., 1996; Watts and Riding,2000). The contact between the Hogklint Formation andthe overlying Tofta Formation represents a hypothesizedsequence boundary (Calner et al., 2004b). The ToftaFormation is characterized by oncoid-rich bedded lime-stone, and likely represents a shallow restricted setting, suchas a backreef lagoon (Riding and Watts, 1991; Samtlebenet al., 1996). The overlying Hangvar Formation consistsof interbedded marls and limestones with some reefmounds, indicating deposition in a less restricted, deeper-water setting. The youngest strata sampled for this studywere from the Slite Formation, which transitions from lime-stone-rich to marl-rich interbedded limestones and marls,likely representing deposition in deeper water (Calneret al., 2004a).

This section has long been recognized for its generallyexcellent fossil preservation. The strata on Gotland are verynearly flat lying, with a dip of less than 1�, display onlyminimal evidence of tectonic disturbance and have not been

deeply buried (Jeppsson, 1983; Calner et al., 2004a). Cono-dont fossils from this section yield a CAI of one (Epsteinet al., 1976; Jeppsson, 1983). The Silurian strata on Gotlandalso have an abundance of fossils, especially brachiopods,throughout the section. We sampled fossil phases forclumped isotope measurements in order to target calcitethat was first precipitated directly from seawater (asopposed to carbonate matrix, which may contain authi-genic cements). Many studies of paleoclimate have targetedthe secondary interior fibrous layer of the brachiopod shell,which is thought to precipitate in equilibrium with seawaterand to be relatively resistant to diagenetic alteration (Azmyet al., 1998; Samtleben et al., 2001; Auclair et al., 2003). Thefibrous calcite fabric characteristic of the secondary layerhas been well characterized in modern and fossil brachio-pods (Samtleben et al., 2001; Schmahl et al., 2004; PerezHuerta et al., 2007), providing a standard of comparisonfor the samples in our dataset. The trace metal compositionof brachiopods has also been extensively studied as an indi-cator of diagenetic alteration (Mii and Grossman, 1994;Grossman et al., 1996; Azmy et al., 1998; Brand, 2003;Shields et al., 2003; Brand et al., 2012). The other fossils


present in this section, including rugose corals, crinoids,bryozoans, and gastropods, are used less often thanbrachiopods for paleoenvironmental studies, but have beenexamined by other clumped isotope studies of paleotemper-ature (Came et al., 2007; Finnegan et al., 2011; Denniset al., 2013). Measurements of several fossil taxa allow com-parison of the relative preservation of different calciticmaterials and can help to detect taxon-specific “vitaleffects” if they are present.

A final attractive feature of the Silurian section onGotland is the previously observed positive excursion ind13C that initiates at the boundary between the Lowerand Upper Visby Beds (Fig. 2). This excursion, termedthe Ireviken event, is now recognized globally (Munneckeet al., 2003). In Gotland, a 3& excursion in d13C is coupledto a weakly resolved 0.4& positive excursion in d18O(Munnecke et al., 2003). It is somewhat surprising thatthe positive excursions coincide with a maximum floodinginterval, but it is difficult to distinguish the relative contri-butions of sediment supply and eustatic sea level rise to thisfacies change (Calner et al., 2004a). The excursions are alsobroadly coincident with minor extinctions in several benthicand pelagic marine taxa (Munnecke et al., 2003).

The coincident isotopic excursions and stratigraphicchange suggest that the Ireviken Event captures a climatechange phenomenon, the nature of which is still debated.The similarity between the isotopic signatures of theIreviken event and known glaciations, such as the LateOrdovician glaciation, have led several studies to hypothe-size a glacial cause for the Ireviken event (Azmy et al., 1998;Kaljo et al., 2003; Brand et al., 2006; Calner, 2008). Thishypothesis was supported by the presence of glacial tillitesin Brazil that were once interpreted to be early Wenlockin age (Grahn and Caputo, 1992), but these diamictiteswere later re-interpreted as re-workings of earlier glacialdeposits. Early Silurian glaciation in South America is cur-rently thought to have ended prior to the late Llandovery,before the Ireviken Event (Dıaz-Martınez and Grahn, 2007;Dıaz-Martınez, 2007). Many of the characteristics of theIreviken Event, although not necessarily its global nature,could also be explained by a shift from a humid period toan arid period in the tropics, which could accompany a gla-ciation or occur independently (Jeppsson, 1990; Samtlebenet al., 1996; Bickert et al., 1997; Munnecke et al., 2003).According to this model, during the humid period recordedby the Lower Visby Formation, increased precipitationwould have lowered the d18O of surface waters and alsostimulated the upwelling of deep water with low d13C ontothe shelves. Increased precipitation would have causedintense weathering, which could explain the dominance ofmarls over carbonate reefs during this time. During the suc-ceeding arid period beginning in the Upper Visby Forma-tion, evaporation would have increased the d18O ofsurface waters, and downwelling of saline water would havebrought surface water with high d13C onto the shelves. Lowterrigenous influxes would have allowed the growth of car-bonate reefs as observed in the Hogklint Formation. Thesehypotheses make different predictions for temperature andthe oxygen isotopic composition of seawater. Here we testour ability to resolve possible changes in ocean temperature

and/or seawater d18O through the Ireviken event usingstratigraphic samples from Gotland. Resolving a changethis small with the uncertainties associated with the carbon-ate clumped isotope thermometer is challenging because ashift of 0.4& in d18O, if caused by change in temperatureof carbonate precipitation at constant d18O of water, corre-sponds to a change in temperature of approximately 2 �C.Though the uncertainties for a single carbonate clumpedisotope temperature are commonly similar to or higher thanthis, replicate analyses can reduce these errors to ±1 �C oreven ±0.4 �C (Thiagarajan et al., 2011).

3. METHODS

3.1. Collection and preparation of samples

Stratigraphic sections were measured and samples werecollected from the island of Gotland, Sweden during thesummer of 2011. Each sample was labeled with ageographic location, as shown in Fig. 1, and a stratigraphicheight. For locations where the Phaulactis layer is exposed,stratigraphic height was expressed relative to the boundarybetween the Lower and Upper Visby formations (Fig. 2).Other locations were correlated on the basis of priorsequence-, bio- and chemostratigraphic work (Calneret al., 2004a).

Of the 22 brachiopods chosen for sampling, 20 likelybelonged to the superfamily Atrypoidea. Brachiopods inthis superfamily are often mistakenly identified as belong-ing to the single species Atrypa reticularis (Copper, 1973).We targeted brachiopods in this superfamily because theyare abundant in the sampled strata, their shells have a thicksecondary fibrous layer, and they were used in severalprevious paleoenvironmental studies in Gotland (Bickertet al., 1997; Samtleben et al., 2001; Munnecke et al.,2003). The species of the other fossils were not identified.

Most of the fossils were easily exposed by breaking thesurrounding layers of poorly-lithified marl. For fossils inmore strongly lithified beds, fossil material was accessedby cutting away the rock with a saw or microrotary drill.After fossils were separated from the rock, they weresonicated for two hours in a Cole Parmer 8891 ultrasoniccleaner to remove residual sand, silt and mud from thesurface of the fossils. Brachiopods, bryozoans, crinoids,gastropods and ostracods were sampled by flaking the sur-face of the skeleton or shell using a dental chisel. Everyflake was visually inspected under a stereomicroscope atca. 25� magnification. Flakes with visible pyrite grains orrough-textured recrystallized regions were rejected andnot included in the material for analysis. For brachiopods,only the flakes preserving the characteristic fabric of thefibrous secondary layer were chosen for analysis (seeFig. 3). Where flakes included a layer of primary layer cal-cite stuck to the secondary layer, the primary layer wasmanually removed, using fine-tipped tweezers, under thestereomicroscope, or the flake was rejected. Flakes werethen sonicated briefly, rinsed, dried in a 50 �C oven, andpowdered using a mortar and pestle. Only the apical endsof rugose corals were sampled, after removal of any sedi-ment from the surface using an abrasive bit. The apical

Fig. 3. Scanning electron microscope images of flakes of brachiopod shell. Images were taken in secondary electron mode. Panels (A)–(C)show flakes that were judged to be pristine and are representative of the material used for analysis. Panel (D) shows a flake that was judged tobe partially recrystallized and would not have been chosen for analysis. Clumped isotope temperatures of the samples for which representativeflakes are shown are: (A) 37 ± 3 �C; (B) 34 ± 7 �C; (C) 37 ± 3 �C.


end was then cut off using a diamond microsaw, and pow-dered using a mortar and pestle. The apical ends of the cor-als were targeted because the coral septa are closely spacedtogether there, leaving less room for sediment and cementbetween the septa. However, this sampling technique is stillnot as selective as the technique used for brachiopods, andcan include more cement and sediment along with skeletalcalcite.

All fossil samples shown in figures and used to estimateSilurian seawater temperature and d18O were prepared asdescribed above, with the aim of minimizing the inclusionof cement, sediment and altered material. However, in afew separate test cases we purposefully sampled cements,micrite and other textures that are independently knownto be more susceptible to diagenetic alteration, as a wayto assess possible diagenetic end-members. Micrite andcement samples were drilled from cut surfaces and thenpowdered using a mortar and pestle. During the prepara-tion of one brachiopod sample, flakes with large rough-tex-tured recrystallized regions were picked and analyzedseparately from the pristine flakes of fibrous secondarylayer as a measure of alteration.

While all fossils were inspected under a microscope priorto sampling, we were not able to inspect all fossils in thinsection, because carbonate clumped isotope measurementsrequire significant amounts of material, some of which islost during preparation of thin sections. However, 30 lm-thick uncovered, polished thin sections were obtained fromcuts through four representative fossils, two brachiopodsand two rugose corals. These thin sections were inspectedfirst in transmitted and reflected light to characterize calcite

textures, and were then examined using several differentmicro-scale spectroscopic techniques, described below.

3.2. Electron microscopy

Electron backscatter diffraction (EBSD) provides a mea-sure of the degree and style of recrystallization by mappingthe orientation of calcite crystals in a thin section (e.g.Bergmann et al., 2013). This allows more detailed charac-terization of calcite fabrics than is possible using opticalmicroscopy, and in the case of brachiopods, comparisonto pristine modern examples (Perez Huerta et al., 2007).Two thin sections, one rugose coral and one brachiopod,were analyzed using a Zeiss 1550VP Field Emission Scan-ning Electron Microscope (SEM) under variable pressurevacuum, with an accelerating voltage of 20 kV. The stagewas tilted 70�. The AZtec program was used in 4x4 binningmode, with frame averaging over one frame and a step sizeof 2 lm for the rugose coral and 1.2 lm for the brachio-pod. The data were analyzed using software from Oxfordinstruments.

Flakes of the secondary fibrous layer from twelve differ-ent brachiopods were also imaged via electron microscopyto define their morphology (but not orientation; i.e., wedid not perform EBSD on these flakes). This further inspec-tion helped us assess the preservation of original fibrouscalcite texture at a finer scale than available from thereflected light microscope (Fig. 3). The rinsed and driedflakes were prepared for imaging by coating them with alayer of carbon 12 nm thick. The coated flakes were imagedusing the secondary electron detector in the same SEM as


above, under high vacuum. The working distance was�6 mm and the accelerating voltage was 10 kV.

3.3. Wavelength-dispersive spectroscopy

The use of wavelength-dispersive spectroscopy onX-rays emitted during electron microprobe analysis, incombination with optical microscopy and EBSD, allowscharacterization of the trace metal contents of specific cal-cite textures. By comparing the trace metal concentrationsof primary textures to those of diagenetic textures, we canidentify the trace metal signature of diagenetic alterationin our sample set. The same thin sections as above werecoated with a layer of carbon 15 nm thick, and then ana-lyzed with wavelength-dispersive spectroscopy using theJEOL JXA-8200 Electron Probe Micro-analyzer (E-probe).Using the JEOL map analysis program, the beam wascalibrated on crystal standards of dolomite, siderite, stron-tianite and rhodochrosite. Qualitative maps of Ca, Mg, Fe,Sr and Mn concentration were made using a current of200 nA, voltage of 15 kV, and a counting time of 80 msper pixel. Quantitative spot measurements were made usingthe JEOL quantitative analysis program with the samebeam conditions as used for the qualitative maps andcounting times of 20 s on the elemental peak and 10 s onbackground.

3.4. Bulk trace metal analysis

The trace metal concentrations of bulk samples can beused as an indicator of the amount of sediment and recrys-tallized material included in the sample (Azmy et al., 1998;Came et al., 2007; Finnegan et al., 2011; Brand et al., 2012).Most studies of the trace elements in fossil calcite measureconcentrations of iron (Fe), manganese (Mn), and stron-tium (Sr), because diagenetic recrystallization is thoughtto enrich calcite in Fe and Mn and deplete Sr from calcite(Brand and Veizer, 1980; Shields et al., 2003). We measuredthe bulk concentrations of these three elements in a largeand representative subset of our samples. Between 1 and5 mg of each powdered sample were dissolved in a knownvolume of 4% nitric acid (HNO3). Standards were preparedfrom standard solutions of Sr, Mn and Fe with the sameconcentration of Ca and HNO3 as the sample solutions tominimize matrix effects. Solutions were sampled using aCetac ASX 260 autosampler, aspirated into the Ar plasmausing a peristaltic pump, and then analyzed on a ThermoScientific iCAP 6000 series inductively-coupled plasmaoptical emission spectrometer (ICP-OES). Between eachset of ten unknown samples, three standard solutions(10 ppb, 100 ppb, and 1 ppm) were measured. Reproduc-ibility was on average 3% for Fe, 2% for Mn and 3% for Sr.

3.5. Clumped isotope analysis and standardization

For each carbonate clumped isotope measurementapproximately 10 mg of powdered sample were dissolvedin phosphoric acid at 90 �C for 20 minutes, which producesCO2 with a known oxygen isotopic offset from the samplecarbonate. The subsequent purification of evolved CO2

followed procedures outlined by Passey et al. (2010).Cleaned CO2 was analyzed on a Finnigan MAT253 gassource mass spectrometer configured to collect masses 44through 49. The sample gas was measured for eight acqui-sitions of seven cycles each, where each cycle was bracketedby a measurement of reference CO2 with known d13C andd18O. The d13C, d18O, d45; d46; d47, and d48 were calculatedfor each sample relative to the reference CO2. Raw mea-sured values for D47 and D48 of each sample, defined asD47-½SGvsWG� and D48�½SGvsWG�, were also calculated relativeto the reference CO2 following equation 3 and 4 inHuntington et al. (2009). The D47-½SGvsWG� values were thenprojected into the Ghosh et al. (2006) reference frame andcorrected for the acid extraction at 90 �C by adding0.081& (Dennis et al., 2011). Each sample was measuredbetween one and three times depending on the amount ofavailable material.

All D47 values are projected into the absolute referenceframe (Dennis et al., 2011). The absolute reference frameis based on a thermodynamic equilibrium established intwo types of CO2 gases: heated gases and water-equilibrated gases. Heated gases are aliquots of CO2 thatwere sealed in quartz tubes, heated at 1000 �C for threehours, and then cooled rapidly to room temperature.Water-equilibrated gases are aliquots of CO2 that wereequilibrated with water at 25 �C for at least 24 hours.Heated gases were made from two separate reservoirs ofCO2 with different d18O compositions, and water-equilibrated gases were equilibrated with either of twoseparate waters, also differing in their d18O compositions.Heated and water-equilibrated gases were cleaned and mea-sured following the same procedure as the samples.

Before sample D47 values could be projected into theabsolute reference frame, we first needed to establishaccepted values in the absolute reference frame for twoCalifornia Institute of Technology intralaboratory carbon-ate standards. One of these standards is crushed Carraramarble and the other is granular travertine, called TV01.D47 values of the two standards were converted to the abso-lute reference frame using only data from weeks duringwhich heated gases and water-equilibrated gases were mea-sured along with the standards. For each of these weeks, atransfer function was constructed by plotting the acceptedabsolute reference frame D47 values of 0.0266& for heatedgases and 0.9252& for 25 �C water-equilibrated gasesagainst their observed D47 as determined in the Ghoshet al. (2006) reference frame (Dennis et al., 2011). Thistransfer function was used to convert the D47 values ofthe two intralaboratory carbonate standards from theGhosh et al. (2006) reference frame to the absolute refer-ence frame following equation C1 in Dennis et al. (2011).

The D47 values of all of the samples in this study wereprojected into the absolute reference frame using a separatetransfer function constructed for each week of measure-ments (Table EA2). For weeks when water-equilibratedgases were measured, this transfer function was constructedby plotting the accepted absolute reference fame D47 valuesof the two intralaboratory carbonate standards, heatedgases, and 25 �C water-equilibrated gases against theirobserved D47 as determined in the Ghosh et al. (2006)


reference frame. For weeks when water-equilibrated gaseswere not measured, this transfer function included onlymeasurements of the two intralaboratory carbonate stan-dards and heated gases. D47 values of the samples in theGhosh et al. (2006) reference frame were corrected usingthis transfer function following equation C1 in Denniset al. (2011) to produce the final value (D47-RF).

Although all samples were cleaned prior to introductioninto the mass spectrometer, the possibility for contamina-tion from hydrocarbons and chlorocarbons still exists. Toscreen for contamination, the d48 and D48 of each samplewere compared to a line of best fit through a plot of d48 ver-sus D48 of the heated gases run during the same week, all inthe Ghosh et al. (2006) reference frame. Samples whose D48

fell more than 1& off the line were discarded due to likelycontamination.

D47-RF values were converted to temperature using theGhosh et al. (2006) calibration line, which is projected intothe absolute reference frame in equation 9 in Dennis et al.(2011). As mentioned above, the Ghosh et al. (2006) cali-bration line is only one of several published relationshipsbetween D47 and temperature (Henkes et al., 2013;Saenger et al., 2012; Tripati et al., 2010). If temperatureswere calculated using the Henkes et al. (2013) calibrationline instead, cool temperatures would change less than war-mer temperatures. For example, a sample that is interpretedto be 33.1 �C using the Ghosh et al. (2006) line changes to34.8 �C when the Henkes et al. (2013) line is used, while asample that is 61.5 �C using the Ghosh et al. (2006) line isincreased to 100.1 �C using the Henkes et al. (2013) line.Use of the Henkes et al. (2013) calibration line would alsochange the d18O of water calculated from each sample: inthe first example, the d18OVSMOW of water changes from�1.1& to �0.7& and in the second example, thed18OVSMOW of water changes from 2.5& to 8.1&. At pres-ent, the reasons for the difference between calibration linesare not clear. We use the Ghosh et al. (2006) calibration linebecause it was developed using similar carbonates in thesame lab as this study.

The d18O of water was calculated from the temperatureand the measured d18O of the sample using the tempera-ture-dependent fractionation, a, of oxygen isotopes betweenwater and calcite, 1000 ln a ¼ 18:03ð103 T�1Þ � 32:42,where T is temperature in degrees Kelvin (Kim andO’Neil, 1997). D47 values of modern taxa, including brachio-pods, are thought not to be subject to disequilibrium vitaleffects (Ghosh et al., 2006; Tripati et al., 2010; Finneganet al., 2011), but there remains a possibility that the d18Oof rugose corals is influenced by vital effects (Finneganet al., 2011). In this case, the d18O of seawater calculatedfrom measurements of rugose corals would be inaccurate.

Standard errors represent the internal reproducibility andare equal to the standard deviation of the eight acquisitionscomprising one sample. Two-sigma standard errors based onall acquisitions in each measurement range from 0.0038& to0.0367& (averaging 0.0095&) for d13C, from 0.0047& to0.0393& (averaging 0.0167&) for d18O, and one-sigma stan-dard errors range from 0.014& to 0.045& (averaging0.029&) for D47. These errors in D47 are consistent with orgreater than counting-statistics limits of 0.013–0.016&

(Huntington et al., 2009). The errors in D47 propagate intotemperature errors ranging from ±3.4 to ±12.7 �C, with anaverage of ±8.0 �C. Errors in d18O and temperature propa-gate into errors in the d18OVSMOW of water ranging from±0.6& to ±2.3&, with an average of ±1.4&. For sampleswith more than one replicate measurement, the standarderror of the mean, representing external reproducibility, isalso reported. For a sample with n replicates, this error isequal to the standard deviation of the n measurement meansdivided by the square root of n. The external reproducibilityreflects heterogeneities in the sample carbonate and the aciddigestion environment, which are not captured by the inter-nal reproducibility. Errors arising from the heterogeneities inthe sample carbonate can be magnified by nonlinear mixingin D47 (Affek and Eiler, 2006).

4. RESULTS AND DISCUSSION

4.1. Stratigraphic trends

The d13C values of fossil material, micrite and cement arehighly consistent within beds (Table EA1 and Fig. 4). Weobserve a �3.5& positive excursion in d13C, beginning justbelow the boundary between the Lower and Upper Visbyformations, and continuing for >30 m. Our samples do notcapture the relaxation of this excursion, but samples fromthe Slite Formation all have consistently lower d13CVPDB val-ues of approximately �0.5&. The d13C trend through thesection is similar to the brachiopod d13C measurements fromBickert et al. (1997) and Munnecke et al. (2003).

The d18O values of fossil material are consistent withinbeds, but micrite and cements tend to have lower d18Ovalues than fossil material, often several permil lower thanfossils in the same bed (Table EA1 and Fig. 4). It is interest-ing that we were not able to resolve an excursion in carbon-ate d18O. This may simply be because our dataset containsfewer samples than that of Munnecke et al. (2003),although even with the additional samples, the previouslyreported excursion is poorly resolved. Alternatively, wemay not be able to resolve the excursion because theaverage preservation quality of the samples in this studyis worse than those of Munnecke et al. (2003); clumped iso-tope analysis requires a large mass of carbonate, so we maynot have been as careful during sample preparation as wecould have been if less material were required. The averaged18OVPDB value of the fossils remains fairly constant at�5.5& through the Lower and Upper Visby formations.These values are extremely similar to those of Munneckeet al. (2003) for the Lower Visby Formation, but our dataonly overlap with the lower end of the d18OVPDB valuesfrom Munnecke et al. (2003) for the Upper Visby Forma-tion. The low d18O values of Upper Visby fossils may becaused by increased contamination from secondarycements, which could be expected to accompany the transi-tion from marl to limestone. Higher in the section, the d18Ovalues of the fossils in this study are likely consistent withvalues reported by Bickert et al. (1997), although directcomparison is not possible because the absolute strati-graphic heights of the samples from the Slite Formationare not known. Due to this uncertainty, it is possible that

7060504030

brachiopod rugose coral crinoid bryozoan gastropod ostracod micrite cement

Munnecke et al. 2003

-7 -6 -5δ18OVPDB (‰)

30m

20m

10m

0m

420

50m

40m

30m

20m

10m

0m

δ13CVPDB (‰)

L. Visby

U. Visby

Högklint

Slite

Temperature (°C)

Fig. 4. The d13C, d18O and clumped isotope temperature of all the samples analyzed in this study, plotted against stratigraphic height. For theLower and Upper Visby and Hogklint formations, a height of zero meters is defined by the contact between the Lower Visby Formation andthe Upper Visby Formation. For the Slite Formation, a height of zero meters is defined as an arbitrary bed near the base of the outcrop, whichmay be near the base of the Slite Formation, or closer to the middle of the Slite Formation. Neither the base nor the top of the Slite Formationis exposed at any of the sampling locations. Brachiopod d13C and d18O measurements from Munnecke et al. (2003) are also included forcomparison. Error bars in clumped isotope temperature represent the one-sigma standard error of the acquisitions (internal reproducibility)for samples measured only once, and the standard error of the mean (external reproducibility) for samples measured more than once.


many of the samples in this section are in fact depleted ind18O compared to those reported by Bickert et al. (1997).

Clumped isotope temperatures of samples are highlyvariable within beds. Of the eight beds for which we mea-sured more than one fossil sample, four beds have temper-ature ranges of more than 13 �C (Table EA1 and Fig. 4).Assuming an error of 5 �C to account for environmentalvariability and measurement error, we used order statisticsto calculate that there is a probability of less than 1% thatone sample in these beds will have a temperature more than13 �C higher than the true depositional temperature, if allsamples record a primary temperature (see Appendix).Therefore, the within-bed scatter in clumped isotope tem-peratures suggests variable preservation quality amongthese samples. Even the secondary fibrous shell layer of bra-chiopods, which is commonly thought to be relatively resis-tant to diagenetic alteration, displays a range intemperature of more than 15 �C across the whole dataset.The variable preservation quality of these samples meansthat beds with fewer samples are less likely to contain apristine fossil with a reliable temperature—introducing asample bias. Because many of the beds in this study are onlyrepresented by a few samples, we are not able to strati-graphically resolve what we would interpret robustly as achange in seawater temperature with time.

4.2. Preliminary considerations of the diagenetic regime

Although we are not able to measure a change in seasurface temperature through the Ireviken Event, we canuse our data to constrain more broadly temperature andd18O of seawater by uncovering the mechanism of diage-netic alteration to identify the most pristine samples. As

described above, diagenetic regimes can vary in their fluidcomposition (meteoric vs. marine pore fluids) and in theirmechanism (closed-system diffusive reordering vs. recrystal-lization and incorporation of late-stage cements). We arguethat closed-system diffusive reordering did not stronglyaffect any of our samples. Gotland strata have a CAI ofone (Epstein et al., 1976; Jeppsson, 1983), which constrainsthe possible effects of diffusive reordering to an increase inthe clumped isotope temperature by a maximum of �1 �C(Dennis and Schrag, 2010; Passey and Henkes, 2012;Stolper et al., 2011). Closed-system diffusive reordering isalso incompatible with the observed variability in clumpedisotope temperatures because this process is expected topreserve the small variance resulting from environmentalvariability and measurement uncertainty (Fig. 5). The var-iance of our sample set is much larger than the maximummeasurement uncertainty of 4.4 �C, which is consistent withrecrystallization as the dominant diagenetic process. Thisleads us to hypothesize that the lowest temperature samplesare the most pristine, while the higher temperature sampleslikely represent recrystallization and incorporation oflate-stage cements during burial at elevated temperatures.

Comparisons of the clumped isotope temperature ofdifferent calcite textures within samples help diagnose thediagenetic regime that acted to increase the apparent tem-perature of more altered samples. We measured theclumped isotope composition of micrites adjacent to fourbrachiopods, and in all four cases, the temperature of themicrite was higher than the temperature of the neighboringbrachiopod (Fig. 6). Micrite is expected to be less pristinethan the brachiopod samples because it has high surfaceareas, far more commonly recrystallizes during burial,and often includes several generations of cements. Because

0

diffusivereordering

environmentenvironment + uncertaintyrecrystallization

prob

abilit

y de

nsity

prob

abilit

y de

nsity

Temperature anomaly Temperature (°C)

Num

ber o

f obs

erva

tions

30 35 40 45 50 55 60 650

2

4

6

8

10

12

14

Fig. 5. The effects of different diagenetic processes on the distribution of sample temperatures. (A) Recrystallization tends to increase thevariance of the distribution of sample temperatures above that expected from environmental variability and measurement uncertainty. (B)Closed-system diffusive reordering is not expected to affect the variance of temperatures. (C) A histogram of the temperatures of samples inthis study. The variance is much greater than the largest measurement uncertainty of 4.4 �C.

50

40

30

-2-1012

δ13C

min

eral

VPD

B (‰

)

-7

-6

-5

-4

δ18O

min

eral

VPD

B (‰

)Te

mpe

ratu

re (°

C)

brac

hiopo

d

micr

ite

brac

hiopo

d

micr

ite

brac

hiopo

d

micr

ite

brac

hiopo

d

micr

ite

Fig. 6. Comparisons between the d13C, d18O and clumped isotopetemperature of brachiopod samples and adjacent micrite. Micrite ismore easily altered than brachiopod shell material, so the consis-tently higher temperature of micrite supports the hypothesis thatthe lowest temperature samples are the most pristine. The generallydepleted d18O values of micrites compared to brachiopods suggestsome influence from fluid-buffered diagenesis.


the less pristine material, micrite, was associated withhigher temperatures, these comparisons are consistent withthe hypothesis that the lowest temperature samples are themost pristine. Interestingly, while the d13C values ofmicrites are similar to those of adjacent brachiopods, threeof the four micrites tested are depleted in d18O relative tothe adjacent brachiopods, suggesting that the micritesreflect recrystallization with a higher water-to-rock ratio.

4.3. Independent assessments of preservation quality

We tested the hypothesis that the lowest-temperaturesamples are the most pristine using independent methods,described above, to assess the preservation quality of repre-sentative samples. Bulk trace metal measurements, whichhave been widely used as indicators of diagenetic alteration(Azmy et al., 1998; Came et al., 2007; Finnegan et al., 2011;Brand et al., 2012), suggest that all the brachiopods in thisstudy are comparatively pristine. However, micro-scalestudies of calcite fabrics, as well as clumped isotope mea-surements, demonstrate that there is a range in preservationquality among the brachiopods that cannot be identified onthe basis of trace metal measurements alone. We firstexplore possible reasons for the insensitivity of trace metalconcentrations to preservation quality. Then we considerEBSD and SEM evidence for the inclusion of diageneticmaterial in the typical fossil sample.

Clumped isotope thermometry implies variable preser-vation among the brachiopods in this study, so we expectedto observe equivalent variations in trace metal content. Dia-genetic recrystallization of brachiopod skeletal calcite isthought to enrich Fe and Mn because these metals areinsoluble in oxic seawater but often found at elevated con-centrations in anoxic pore fluids (Brand and Veizer, 1980;Shields et al., 2003). Diagenetic recrystallization is alsothought to deplete Sr because brachiopod skeletons concen-trate Sr relative to its abundance in seawater and meteoricwaters (Brand and Veizer, 1980; Schrag et al., 1995;DePaolo, 2011). Previous studies have defined limits forthe Mn and Sr concentrations of brachiopods that can beconsidered pristine (Azmy et al., 1998; Mii et al., 1999;Korte et al., 2005; van Geldern et al., 2006). The bulkMn and Sr concentrations of all brachiopods in this study


fall within these proposed limits (Fig. 7); the small varia-tions in trace metal concentrations among the brachiopodswould not ordinarily be interpreted as variablepreservation.

We also expected to find a correlation between clumpedisotope temperatures and trace metal evidence for diage-netic alteration, as observed for fossils by Finnegan et al.(2011). We tested brachiopods, rugose corals, andmicrites/cements separately and combined for the robust-ness of a linear correlation between temperature and thelog of trace metal concentrations (Fig. 8). We performeda hypothesis test on each pair of variables to test theprobability (p) that the observed correlation is due torandom variation when the true correlation is zero. Weused a Bonferroni correction to adjust the significance level(a) to account for the multiple hypothesis tests. There areweakly significant positive correlations between tempera-ture and Fe concentration (r = 0.23) and also between tem-perature and Mn concentration (r = 0.28) when all samplesare combined. These correlations are largely due to differ-ences between fossil taxa and later phases; brachiopodshave the lowest average temperature, Fe, and Mn concen-trations, while micrites and cements combined have thehighest average temperature, Fe and Mn concentrations.There is also a weakly significant negative correlationbetween temperature and Sr concentration (r = �0.52)among rugose corals, which strengthens (to 99% confi-dence) when corals are combined with brachiopods(r = �0.49), and with micrites and cements (r = �0.42).

100

1000

1 10 100 1000

Sr (p

pm)

Mn (ppm)

brachiopod rugose coral micrite cement

this study Brand et al. 2012

van Geldern, 2006Mii et al. 1999Korte et al. 2005Azmy et al. 1998

Fig. 7. Bulk manganese and strontium concentrations of thebrachiopods, rugose corals, micrites and cements in this study. Thebrachiopods in this study have more Sr on average than thebrachiopods analyzed by Brand et al. (2012), which are also shownin this figure. Lines represent limits suggested by different authorsfor the range of pristine concentrations defined for brachiopods.Rugose corals generally have more Fe (not shown) and more Mnthan brachiopods, giving them more altered trace metal signatures,while cements and micrites tend to have the most highly alteredtrace metal signatures. Figure modified from Brand et al. (2012).

Trace metal concentrations reflect broad differencesbetween taxa and textures, however none of the othervariable pairs are significantly correlated within individualtaxa (Table EA3). These tests may be complicated by con-tamination from early void-filling cements, which have lowtemperatures but “altered” trace metal signatures becausethey precipitate from pore fluids. Variability in Mn concen-trations of low-temperature samples may obscure a positivecorrelation between temperature and Mn concentration,but low-temperature contamination does not appear toexplain the lack of correlation between temperature andFe or Sr. From these observations we suggest that tracemetal concentrations, perhaps with the exception of Sr inrugose corals, are relatively insensitive indicators ofalteration within fossil taxa under mild burial diagenesis,and in this case are only sensitive enough to reflect broaddifferences between taxa.

There are several possible explanations for the relativeinsensitivity of trace metal concentrations to diageneticalteration of these materials. Our sample set likely doesnot include a sufficient number of highly altered samplesto capture the full range in severity of alteration, becausevisibly altered fossils were rejected during visual screeningof samples and not included in the sample set. Variabilityin primary trace metal concentrations, which is large amongmodern brachiopods (Brand, 2003), may also mask anytrends superimposed by diagenetic alteration. A more com-pelling reason for the insensitivity of trace metal concentra-tions is that diagenetic recrystallization does not necessarilystrongly affect Fe and Mn concentrations. Indeed, even if asample set has been screened for diagenetic alteration usingtrace metal concentrations, there can still be outliers, suchas carbonate fossils with very low (<�15&) d18OVPDB val-ues, that have pristine trace metal signatures but have likelybeen diagenetically altered (Veizer et al., 1999). E-probeanalysis of a rugose coral from our sample set supports thispossibility. A recrystallized region of the coral septum isclearly visible as a dark area in transmitted light (Fig. 9A)and an area of aligned calcite crystals in an electron back-scatter diffraction (EBSD) image (Fig. 9B). However, theFe and Mn concentrations of the recrystallized region areindistinguishable from those of the pristine coral skeleton(Fig. 9C and D). This is only one example, but it highlightsrelatively large calcite domains within a sample that wererecrystallized but not strongly modified in trace metalconcentrations. The trace metal signature associated withrecrystallization likely depends on conditions duringrecrystallization, suggesting that this coral underwentclosed-system diagenesis, or recrystallization in the presenceof trace-metal-poor waters. If closed-system recrystalliza-tion is an important contributor to the elevated tempera-tures of fossil material in our sample set, trace metalsmay not be expected to be reliable indicators of diageneticalteration.

We therefore turn to micro-scale studies of calcite tex-tures as indicators of diagenetic alteration. Reflected-lightand EBSD images of rugose coral fossil skeletons revealsubstantial domains of late-phase diagenetic calcite, includ-ing recrystallized regions and cements (Fig. 9A and B). Asdiscussed above, recrystallized regions are visible as dark

10 100

1600

1400

1200

100010 10

010

00

60

50

40

30

102

103

104

brachiopod rugose coral micrite cement

Tem

pera

ture

(°C

)

Fe (ppm) Mn (ppm) Sr (ppm)

Fig. 8. Scatter plots showing the relationship between temperature and bulk trace metal concentrations. The negative correlation betweentemperature and the log of Sr concentration in rugose corals is weakly significant and strengthens when combined with the rest of the samples.There are also weakly significant positive correlations between temperature and the log of Fe concentration and between temperature and thelog of Mn concentration among all the samples combined. These plots demonstrate clear differences in trace metal content between the twofossil taxa and the later carbonate phases. Brachiopods have the lowest average temperature of all the sample subsets, and also tend to havethe most pristine trace metal signatures, with less Fe and Mn than the other samples. Micrites and cements have the highest averagetemperature and tend to have the most altered trace metal signatures, with high Fe and Mn concentrations.

Fig. 9. Calcite fabric and trace metal concentrations in a rugose coral. The four images show the same area of a rugose coral thin section, allat the same magnification. The black bar at the bottom of panel (B) is 500 lm long for scale. (A) Transmitted light. The dark triangle labeled“RC” was identified as a recrystallized domain of the coral septum. Cement in the void space between coral septa appears as blocky calcitecrystals. (B) Electron backscatter diffraction (EBSD) analysis. The orientation of calcite crystals is indicated by different shades. Therecrystallized domain is clearly marked by many calcite crystals with the same orientation. (C) Iron and (D) manganese concentrationsmeasured using the E-probe. The maps are qualitative, with darker shades representing higher concentrations. For both metals, therecrystallized domain is indistinguishable from the rest of the septum. The black lines show the location of an array of 62 spots where tracemetal concentrations were measured quantitatively. The concentration of Fe and Mn at these spots is plotted to the right of the respectivemap.


colors in reflected light and aligned calcite crystals in EBSDimages. Cements in Fig. 9A and B are characterized bylarge, subequant, interlocking calcite crystals.

In contrast to the rugose corals, the EBSD image of alow-temperature (37 �C) brachiopod thin section demon-strates excellent preservation of original fibrous calcitetextures (Fig. 10A). Within the brachiopod shell, calcite

crystals are elongated approximately parallel to the surfaceof the shell, as expected for the fibrous secondary layer.This preferential orientation is represented by the verticalband in the stereographic projection of crystallographicplanes (Fig. 10B). No large regions of recrystallization arevisible within the shell. There are large interlocking cementcrystals in the upper left hand corner of the EBSD image,

Fig. 10. Calcite fabric in a brachiopod. (A) Transmitted polarized light image of a thin section of G3 Lower Visby Float brachiopod 1. Thissample has a clumped isotope temperature of 37 �C. The boxed overlay is an electron backscatter diffraction (EBSD) image of the samesample. The orientation of calcite crystals is indicated by different shades. (B) Euler pole figure generated from the EBSD analysis. The centralband represents the preferential orientation of calcite crystals parallel to the outer surface of the brachiopod shell.

2 0 -2

-4

-6

-8

-10

brachiopod rugose coral crinoid bryozoan gastropod ostracod micrite cement

40 07 08 095020 30 60

−4

−6

−2

0

2

6

4

Temperature (°C)

δ18O

wat

er VS

MO

W (‰

)

δ18 O m

ineral VPDB (‰

)

Fig. 11. The clumped isotope temperature and calculated waterd18OVSMOW for all samples analyzed in this study. The contoursrepresent lines of constant carbonate d18OVPDB. We approximatethe correlated error as symmetric using ellipses that represent onestandard deviation of all measurement acquisitions for eachsample. The positive correlation between temperature and waterd18OVSMOW is diagnostic of rock-buffered diagenesis. Under thisregime, the lowest-temperature samples are the most pristine. Notethat some cement and micrite samples fall off the trend, suggestingthe effects of a more open-system diagenetic process.


but these formed in the void space between the two valves,and would not have been included in the sample forclumped isotope analysis. Based on visual inspection ofthe EBSD image of this brachiopod’s secondary fibrouslayer, we suggest that this sample contains a maximum of5% recrystallized material.

Scanning electron microscope (SEM) images of low-temperature brachiopod flakes also demonstrate excellentpreservation of original fibrous calcite textures (Fig. 3A–C). The original fibrous texture clearly contrasts with therough recrystallized texture covering part of a flake thatwould not have been chosen for analysis (Fig. 3D). The dif-ference between these two textures is visible under themicroscope used for picking brachiopod flakes, so flakeswith large recrystallized regions were easily identified andexcluded from the sample. However, small recrystallizedregions, especially if they occurred between layers of fibroustexture, may elude identification during picking. Therefore,although the representative flakes chosen for inspectionwith the SEM appear pristine, this does not guarantee thatall recrystallized material was excluded from the samples.

The micro-scale studies of calcite textures in Figs. 3, 9and 10 reveal relatively well-preserved samples, but donot rule out the inclusion of small amounts of late-phasecalcite. Variable fractions of late-phase calcite likely con-tribute to the range in temperature among fossil samples.Due to constraints on the amount of available materialfor analysis, we only performed SEM and EBSD analysison a small subset of samples, but an important conclusionthat can be drawn from the electron microscopy and spec-troscopy results is that the low temperature brachiopodsimaged in Figs. 3 and 10 probably contain at most onlytrace amounts of late-phase calcite, variable but likely lessthan 5% of the sample. We therefore look to where thesesamples fall on diagenetic trajectories to estimate Silurianocean temperature and the value of seawater d18O.

4.4. Constraining the temperature and d18O of Silurian

oceans

To estimate the temperature and d18O of Silurianoceans, we first need to identify the diagenetic regimeresponsible for the observed differential alteration of thesamples. As outlined above, different diagenetic processes

have characteristic effects on the relationship between tem-perature and the d18O of interstitial water. When theclumped isotope temperatures of the samples in this studyare plotted versus the d18O of water calculated for eachsample, there is a clear positive relationship (Fig. 11). Thisrelationship suggests that interstitial water grew increas-ingly enriched in 18O as the temperature increased, whichis characteristic of rock-buffered diagenesis. There are addi-tional, subtler trends in Fig. 11, including a trend towardmore negative calcite d18O. This trend could be the productof water-buffered diagenesis at high temperature, or diagen-esis in the presence of meteoric water. Importantly, in all ofthese diagenetic regimes, more altered samples have eitherhigher temperatures or lower calcite d18O, which is consis-tent with the hypothesis that the lowest temperature sam-ple, which also has one of the highest values of calcited18O, is the most pristine.


The lowest-temperature sample within our sample set, abrachiopod (G13 10.5–10.8 brachiopod 1, Slite Fm.), sug-gests that tropical Silurian oceans had a temperature of33 ± 7 �C and a d18OVSMOW of �1.1 ± 1.3& (Fig. 11).One caveat of this value is that it is based on only one mea-surement due to sample-size limitations. The same brachio-pod was picked a second time, but to collect enoughmaterial for analysis, some lower quality flakes had to beincluded, which yielded a temperature of 42 ± 13 �C (G1310.5–10.8 brachiopod 1 leftovers).

Due to the increase of observed sample temperaturesduring diagenesis, our estimates of the temperature andd18O of Silurian tropical oceans are upper limits. We arenot able to exclude the possibility that small amounts ofdiagenesis affected all the samples in our study, even themost pristine samples. However, micro-scale studies of thecalcite fabric of the lowest-temperature brachiopodsindicate that late-phase calcite is only a trace constituentof these samples, comprising a maximum of 5% of thematerial. Assuming closed-system diagenesis so that thelate-phase calcite is similar in bulk d13C and d18O tothe pristine fossil calcite, D47 values will mix linearly (Affekand Eiler, 2006). In this case, if the measured temperatureof the sample is 33 �C and 5% of the sample material wasrecrystallized at a maximum of 80 �C (constrained byCAI = 1), the primary temperature of the pristine fossilmaterial would be 31 �C. Therefore, the constraints pro-vided by the EBSD mapping of the brachiopod shell calcitedemonstrate that our estimate of the temperature of thetropical Silurian ocean is likely reliable to within 2 �C—similar to our measurement uncertainty for replicated sam-ples. Differences between different D47 calibration lines alsoadd only small uncertainties (�2 �C) at these temperatures.If we had used the Henkes et al. (2013) calibration lineinstead of the Ghosh et al. (2006) line, the temperature ofthe most pristine sample would be 35 ± 14 �C, correspond-ing to a seawater d18OVSMOW of �0.7 ± 2.6&.

These clumped isotope results indicate that Siluriantropical oceans were broadly similar to modern tropicaloceans in their temperature and d18O. Our estimate of thetemperature at a Silurian paleolatitude of 19� 5:1 � south(Torsvik et al., 1992) is within 3 �C of the maximumobserved temperature for the modern ocean at a similarlatitude (Reynolds and Smith, 1994). The averaged18OVSMOW of the modern ocean is approximately �0.3&

(Shackleton and Kennett, 1974), but without ice, theaverage d18OVSMOW would decrease to approximately�1.4& (Lhomme, 2005). As stated above, the volume ofcontinental ice is not known with certainty for all of earlySilurian time, but assuming Silurian ice volume was lessthan or equal to modern ice volume, our findings suggestthat the d18O of Silurian oceans was within 2& of themodern value.

The results of this study are also consistent with those ofan earlier clumped isotope study. Came et al. (2007) mea-sured the clumped isotope compositions of brachiopodsfrom the early Silurian on Anticosti Island, Canada, andreport temperatures of 30–35 �C and seawater d18OVSMOW

values of around �1&, which are similar to the results ofthis study. Both of these sedimentary basins were bathed

in tropical seawater, but each reflects a different inlandsea (Anticosti Island on the craton margin of a forelandbasin, and Gotland on the southern margin of the BalticShield). The strong similarity in temperatures and d18Obetween these localities lends some confidence to paleocli-mate reconstruction during Silurian time.

Our results rule out large (>2&) long-term changes inthe d18O of seawater and extremely hot (>40 �C) ocean tem-peratures as causes of the depleted d18O values of Siluriancarbonates. Instead, clumped isotopes demonstrate thateven among samples with well-preserved calcite fabricsand pristine trace metal signatures, there is a large rangein preservation quality. This means that the average fossilcalcite in almost any sample suite is likely altered from itsprimary composition in an asymmetric way. We note thatit is common to take the average d18O value of carbonatesamples when producing time series records of paleoclimate(Bickert et al., 1997; Veizer et al., 2000; Munnecke et al.,2003; Shields et al., 2003; Jaffres et al., 2007). If a carbonatesample set has undergone high-temperature fluid-buffereddiagenesis, causing more altered samples to have lowerd18O values, the average d18O of all the samples wouldnot accurately reflect primary ocean conditions. The best-preserved sample in our dataset has a d18OVPDB of�5.0&, which is close to the average d18OVPDB value(�4.6&) of compiled low-magnesium calcites from the Silu-rian Period (Veizer et al., 1999); however the same needsnot be true of all sample sets or intervals, particularly asone moves into older and more deeply buried strata.Indeed, the most pristine sample from the upper MississippiValley (pre-glacial Ordovician) measured by Finnegan et al.(2011) has a d18OVPDB of �4.9&, while the average of com-piled low-magnesium calcites measured globally from theOrdovician Period is �5.9& (Veizer et al., 1999).

The results from this study join a growing set of clumpedisotope studies (Came et al., 2007; Finnegan et al., 2011;Dennis et al., 2013), as well as studies of the d18O of cono-dont apatite (Wenzel et al., 2000; Joachimski et al., 2004),ancient ophiolites (Muehlenbachs et al., 2003; Turchynet al., 2013), and fluid inclusions in evaporite minerals(Knauth and Roberts, 1991), all of which suggest that thed18OVSMOW of seawater has been broadly constantthroughout much of Phanerozoic time, at around�1 ± 1& (Fig. 12). Several studies have hypothesized thathigh- and low-temperature alteration of the seafloor actsas a buffer for the d18O of seawater (Muehlenbachs andClayton, 1976; Gregory and Taylor, 1981; Gregory, 1991;Muehlenbachs, 1998). Observations from modern oceandrill cores suggest that the average temperature of exchangebetween the seafloor and seawater is 250–350 �C(Muehlenbachs, 1998). At these temperatures, an oceanwith a d18OVSMOW of �0& is in isotopic equilibrium withthe 5.8& mantle. This buffering capacity rests on theassumption that the flux of 18O exchange due to continentalweathering has remained insignificant in the overall 18Obudget throughout Earth’s history, and also that the aver-age temperature of exchange between the seafloor and sea-water has not changed through time. The results from thisstudy provide further evidence that these assumptions arevalid.

-15 -15

-10 -10

-5 -5

Muehlenbachs & Clayton 1976Veizer et al. 1999Wallmann 2001

Came et al. 2007Dennis et al. 2013

This studyFinnegan et al. 2011

-15 -15

-10 -10

-5 -5

500 400 300 200 100 0Age (million years)

δ18O

wat

er VS

MO

W (‰

)

δ18O

min

eral

VPD

B (‰

)

00

Fig. 12. Measurements and estimates of seawater d18O throughtime. In gray, the d18OVPDB of apparently well-preserved skeletalcalcites, compiled by Veizer et al. (1999) and updated in 2004. Thedata are available online: http://mysite.science.uottawa.ca/jveizer/isotope_data/index.html. Values from benthic foraminifera are notincluded because they tend to record much colder water than theother fossils. The black solid line represents one hypothesized timeseries for the d18OVSMOW of the oceans through time based on alinear least-squares fit (Veizer et al., 2000) through the skeletalcalcite data (Veizer et al., 1999). The straight dashed black linerepresents the hypothesis of Gregory and Taylor (1981) thatseawater d18OVSMOW has remained constant through time. Thecurved dashed black line represents the d18OVSMOW of seawaterthrough time as simulated by the box model in Wallmann (2001).All three black lines ignore short-term perturbations to seawaterd18O caused by glaciations. Finally, the black shapes are estimatesof seawater d18O from clumped isotope studies, measured at fivedifferent times by four separate studies. These clumped isotopeestimates suggest that seawater d18O has not changed by more than2& in the last 430 million years. These estimates also highlight thatthe average d18OVPDB in a given interval offers a biased estimate ofthe seawater composition.


5. CONCLUSIONS

We used clumped isotopes to constrain both thetemperature and d18O of Silurian oceans because thismethod provides a temperature that is independent of thed18O of the water from which the carbonate precipitated.Samples were collected from well-preserved flat-lying sec-tions on the island of Gotland, Sweden. By samplingwell-preserved, fossil-rich rocks that potentially record aclimate phenomenon associated with changing seawatertemperatures, we tested the ability of clumped isotopes toresolve potentially small climate change phenomenaassociated with the Ireviken Event. Despite the overall highquality of fossiliferous materials in our Gotland sections,we were not able to resolve a robust secular change in tem-perature through the section, due to the variable preserva-tion quality of the samples. We observed that trace metalconcentrations have only a limited ability to detect varia-tions in preservation among Gotland brachiopods, despite

a range in brachiopod clumped isotope temperatures, sug-gesting that clumped isotopes are a more sensitive indicatorof diagenetic alteration than trace metal compositions. Thisconclusion is supported by EBSD observations of fossil cal-cite that was recrystallized without a detectable change intrace metal concentrations. Micro-scale analysis of calcitefabrics, comparisons between brachiopods and micrite,and the positive relationship between temperature andd18O of water calculated from the samples all support thehypothesis that the lowest-temperature samples are themost pristine with alteration of fossil material largely tiedto rock-buffered diagenesis. Based on the most pristinematerials in our dataset, we estimate that tropical Silurianoceans had a temperature of 33 ± 7 �C and a d18OVSMOW

of �1.1 ± 1.3&. Our results, along with those from otherclumped isotope studies of paleoenvironment, suggest thatthe d18O of seawater (considering the fraction stored asice during the Pleistocene) has remained broadly constantover Phanerozoic time.

ACKNOWLEDGEMENTS

We acknowledge Nami Kitchen, Kristin Bergmann, DanielStolper, and the rest of the Eiler lab for help with the clumpedisotope measurements and J. Garrecht Metzger for assistance infield collection. We thank Chi Ma for his assistance with theSEM and the E-probe, and Joel Hurowitz for help with traceelement geochemistry. We also thank Mark Garcia and DavidMann for their help with sample processing, Clint Cummins forhelp with statistics, and Uno Granquist and Anders Birgesson atCementa for quarry access. We are also grateful to KarlisMuehlenbachs and one anonymous reviewer whose commentsgreatly improved the quality of our data analysis and manuscript.This work was supported by NSF grant EAR-1053523 and theAgouron Institute.

APPENDIX A

To demonstrate that within-bed temperature variabilityis unlikely to be explained only by the combination of envi-ronmental variability and measurement error, but instead ismost likely due to variable preservation quality, we usedorder statistics. We assumed an uncertainty ðrÞ of 5& toaccount for the combined effects of environmental variabil-ity and measurement error. We then assumed that all tem-peratures observed in a single bed were primary, and testedthe likelihood that the range of temperatures observedcould be explained by the assumed uncertainty of 5&.Given a bed with N = 2 samples, the probability ðpÞ thata range of 13 �C is observed between the coolest sampleand the warmest sample is equal to:

p ¼ N � p0 � ðl� p0ÞðN�1Þ ð1Þ

where p0 can be calculated in MATLAB using the equation:

p0 ¼ 1� normcdfðtemperature range=rÞ¼ 1� normcdfð13=5Þ ¼ 0:0047 ð2Þ

Substituting this value for p0 into the original equation,we get:

http://mysite.science.uottawa.ca/jveizer/isotope_data/index.html

http://mysite.science.uottawa.ca/jveizer/isotope_data/index.html


p ¼ N � p0 � ð1� p0ÞðN�1Þ ¼ 20:0047 � ð1� 0:0047Þð2�1Þ

¼ 0:0093 ¼ 0:93% ð3Þ

Therefore, there is a less than 1% chance that variabilityof 13 �C in a single bed is due to environmental variabilityand measurement error alone, assuming that the uncer-tainty due to these two factors is 5&.

APPENDIX B. SUPPLEMENTARY DATA

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.gca.2014.05.024.

REFERENCES

Affek H. P. and Eiler J. M. (2006) Abundance of mass 47 CO2 inurban air, car exhaust, and human breath. Geochim. Cosmo-

chim. Acta 70, 1–12.Aldridge R. J., Jeppsson L. and Dorning K. J. (1993) Early Silurian

oceanic episodes and events. J. Geol. Soc. London 150, 501–513.Auclair A. C., Joachimski M. M. and Lecuyer C. (2003)

Deciphering kinetic, metabolic and environmental controls onstable isotope fractionations between seawater and the shell ofTerebratalia transversa (Brachiopoda). Chem. Geol. 202, 59–78.

Azmy K., Veizer J., Bassett M. and Copper P. (1998) Oxygen andcarbon isotopic composition of Silurian brachiopods: implica-tions for coeval seawater and glaciations. Geol. Soc. Am. Bull.

110, 1499–1512.Bergmann K. D., Grotzinger J. P. and Fischer W. W. (2013)

Biological influences on seafloor carbonate precipitation.PALAIOS 28, 99–115.

Bickert T., Patzold J., Samtleben C. and Munnecke A. (1997)Paleoenvironmental changes in the Silurian indicated by stableisotopes in brachiopod shells from Gotland, Sweden. Geochim.

Cosmochim. Acta 61, 2717–2730.Brand U. (2003) Geochemistry of modern brachiopods: applica-

tions and implications for oceanography and paleoceanogra-phy. Chem. Geol. 198, 305–334.

Brand U. and Veizer J. (1980) Chemical diagenesis of a multicom-ponent carbonate system; 1. Trace elements. J. Sediment. Res.

50, 1219–1236.Brand U., Azmy K. and Veizer J. (2006) Evaluation of the Salinic I

tectonic, Cancaniri glacial and Ireviken biotic events: biochem-ostratigraphy of the Lower Silurian succession in the NiagaraGorge area, Canada and U.S.A.. Palaeogeogr. Palaeoclimatol.

Palaeoecol. 241, 192–213.Brand U., Jiang G., Azmy K., Bishop J. and Montanez I. P. (2012)

Diagenetic evaluation of a Pennsylvanian carbonate succession(Bird Spring Formation, Arrow Canyon, Nevada, U.S.A.)—1:brachiopod and whole rock comparison. Chem. Geol., 26–39.

Calner M. (2008) Silurian global events—at the tipping point ofclimate change. In Mass Extinctions (ed. A. M. T. Elewa).Springer-Verlag, Berlin and Heidelberg, pp. 21–57.

Calner M., Jeppsson L. and Munnecke A. (2004a) Field guide tothe Silurian of Gotland—Part I. Erlanger Geol. Abh. 5, 113–131.

Calner M., Jeppsson L. and Munnecke A. (2004b) Field guide tothe Silurian of Gotland—Part II. Erlanger Geol. Abh. 5, 133–151.

Came R. E., Eiler J. M., Veizer J., Azmy K., Brand U. andWeidman C. R. (2007) Coupling of surface temperatures andatmospheric CO2 concentrations during the Palaeozoic era.Nature 449, 198–201.

Caro G., Papanastassiou D. A. and Wasserburg G. J. (2010) 40K–40Ca isotopic constraints on the oceanic calcium cycle. Earth

Planet. Sci. Lett. 296, 124–132.Clayton R. N. and Degens E. T. (1959) Use of carbon isotope

analyses for differentiating fresh-water from marine sediments.AAPG Bull. 43, 890–897.

Collins W. J. (2003) Slab pull, mantle convection, and Pangaeanassembly and dispersal. Earth Planet. Sci. Lett. 205, 225–237.

Copper P. (1973) New Siluro-Devonian atrypoid brachiopods. J.

Paleontology 47, 484–500.Dennis K. J. and Schrag D. P. (2010) Clumped isotope thermom-

etry 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.

Dennis K. J., Cochran J. K., Landman N. H. and Schrag D. P.(2013) The climate of the Late Cretaceous: new insights fromthe application of the carbonate clumped isotope thermometerto Western Interior Seaway macrofossil. Earth Planet. Sci. Lett.

362, 51–65.DePaolo D. J. (2011) Surface kinetic model for isotopic and trace

element fractionation during precipitation of calcite fromaqueous solutions. Geochim. Cosmochim. Acta 75, 1039–1056.

Dıaz-Martınez E. (2007) The Sacta Limestone Member (earlyWenlock): cool-water, temperate carbonate deposition at thedistal foreland of Gondwana’s active margin, Bolivia. Palaeo-

geogr. Palaeoclimatol. Palaeoecol. 245, 46–61.Dıaz-Martınez E. and Grahn Y. (2007) Early Silurian glaciation

along the western margin of Gondwana (Peru, Bolivia andnorthern Argentina): palaeogeographic and geodynamic set-ting. Palaeogeogr. Palaeoclimatol. Palaeoecol. 245, 62–81.

Eiler J. M. (2011) Paleoclimate reconstruction using carbonateclumped isotope thermometry. Quat. Sci. Rev. 30, 3575–3588.

Epstein A. G., Epstein J. B. and Harris L. D. (1976) Conodont

Color Alteration: An Index to Organic Metamorphism. USGovernment Printing Office, Washington, DC.

Finnegan S., Bergmann K., Eiler J., Jones D. S., Fike D. A.,Eisenman I., Hughes N., Tripati A. and Fischer W. (2011) Themagnitude and duration of Late Ordovician-early Silurianglaciation. Science 331, 903.

Fornari D. J. and Embley R. W. (1995) Tectonic and volcaniccontrols on hydrothermal processes at the mid-ocean ridge: anoverview based on near-bottom and submersible studies.Geophys. Monogr. Ser. 91, 1–46.

Frakes L. A., Francis J. E. and Syktus J. I. (2005) Climate Modes of

the Phanerozoic. Cambridge Univ. Press.Ghosh P., Adkins J., Affek H., Balta B., Guo W., Schauble E.,

Schrag D. and Eiler J. (2006) 13C–18O bonds in carbonateminerals: a new kind of paleothermometer. Geochim. Cosmo-

chim. Acta 70, 1439–1456.Grahn Y. and Caputo M. V. (1992) Early Silurian glaciations in

Brazil. Palaeogeogr. Palaeoclimatol. Palaeoecol. 99, 9–15.Gregory R. T. (1991) Oxygen isotope history of seawater revisited:

timescales for boundary event changes in the oxygen isotopecomposition of seawater. In Stable Isotope Geochemistry: A

Tribute to Samuel Epstein (eds. H. P. Taylor, , Jr.J. R. O’Neiland I. R. Kaplan). Geochemical Society, pp. 65–76.

Gregory R. T. and Taylor, Jr., H. P. (1981) An oxygen isotopeprofile in a section of Cretaceous oceanic crust, Samailophiolite, Oman: Evidence for d18O buffering of the oceans bydeep (>5 km) seawater-hydrothermal circulation at mid-oceanridges. J. Geophys. Res. 86, 2737–2755.

Grossman E. L., Mii H. S., Zhang C. and Yancey T. E. (1996)Chemical variation in Pennsylvanian brachiopod shells; diage-



http://refhub.elsevier.com/S0016-7037(14)00353-6/h0010



























































































































netic, taxonomic, microstructural, and seasonal effects. J.

Sediment. Res. 66, 1011–1022.Henkes G. A., Passey B. H., Wanamaker, Jr., A. D., Grossman E.

L., Ambrose, Jr., W. G. and Carroll M. L. (2013) Carbonateclumped isotope compositions of modern marine mollusk andbrachiopod shells. Geochim. Cosmochim. Acta 106, 307–325.

Huntington K. W., Eiler J., Affek H. P., Guo W., Bonifacie M.,Yeung L. Y., Thiagarajan N., Passey B., Tripati A., Daeron M.and Came R. (2009) Methods and limitations of ‘clumped’ CO2

isotope (D47) analysis by gas–source isotope ratio massspectrometry. J. Mass Spectrom. 44, 1318–1329.

Huntington K. W., Budd D. A., Wernicke B. P. and Eiler J. (2011)Use of clumped-isotope thermometry to constrain the crystal-lization temperature of diagenetic calcite. J. Sediment. Res. 81,656–669.

Jaffres J. B. D., Shields G. A. and Wallmann K. (2007) Theoxygen isotope evolution of seawater: a critical review of along-standing controversy and an improved geological watercycle model for the past 3.4 billion years. Earth Sci. Rev. 83,83–122.

Jeppsson L. (1983) Silurian conodont faunas from Gotland. Fossils

Strata 15, 121–144.Jeppsson L. (1990) An oceanic model for lithological and faunal

changes tested on the Silurian record. J. Geol. Soc. London 147,663–674.

Jeppsson L. (1997) A new latest Telychian, Sheinwoodian andEarly Homerian (Early Silurian) standard conodont zonation.Trans. R. Soc. Edinburgh: Earth Sci. 88, 91–114.

Jeppsson L., Eriksson M. E. and Calner M. (2006) A latestLlandovery to latest Ludlow high-resolution biostratigraphybased on the Silurian of Gotland—a summary. Geol. Foeren.

Stockholm Foerh. 128, 109–114.Joachimski M. M., Geldern R., Breisig S., Buggisch W. and Day J.

(2004) Oxygen isotope evolution of biogenic calcite and apatiteduring the Middle and Late Devonian. Int. J. Earth Sci. (Geol.

Rundsch.) 93, 542–553.Kaljo D., Martma T., Mannik P. and Viira V. (2003) Implications

of Gondwana glaciations in the Baltic late Ordovician andSilurian and a carbon isotopic test of environmental cyclicity.Bull. Soc. Geol. Fr. 174, 59.

Karhu J. and Epstein S. (1986) The implication of the oxygenisotope records in coexisting cherts and phosphates. Geochim.

Cosmochim. Acta 50, 1745–1756.Keith M. L. and Weber J. N. (1964) Carbon and oxygen isotopic

composition of selected limestones and fossils. Geochim. Cos-

mochim. Acta 28, 1787–1816.Kim S. and O’Neil J. (1997) Equilibrium and nonequilibrium

oxygen isotope effects in synthetic carbonates. Geochim. Cos-

mochim. Acta 61, 3461–3475.Knauth L. P. and Epstein S. (1976) Hydrogen and oxygen isotope

ratios in nodular and bedded cherts. Geochim. Cosmochim. Acta

40, 1095–1108.Knauth L. P. and Kennedy M. J. (2009) The late Precambrian

greening of the Earth. Nature 460, 728–732.Knauth L. P. and Roberts S. K. (1991) The hydrogen and oxygen

isotope history of the Silurian–Permian hydrosphere as deter-mined by direct measurement of fossil water. In Stable Isotope

Geochemistry: A Tribute to Samuel Epstein (eds. , Jr.H. P.Taylor, J. R. O’Neil and I. R. Kaplan). Geochemical Society,pp. 91–104.

Korte C., Jasper T., Kozur H. W. and Veizer J. (2005) d18O andd13C of Permian brachiopods: a record of seawater evolutionand continental glaciation. Palaeogeogr. Palaeoclimatol. Palae-

oecol. 224, 333–351.Land L. S. (1995) Comment on “Oxygen and carbon isotopic

composition of Ordovician brachiopods: Implications for

coeval seawater” by H. Qing and J. Veizer. Geochim. Cosmo-

chim. Acta 59, 2843.Lhomme N. (2005) Global budget of water isotopes inferred from

polar ice sheets. Geophys. Res. Lett. 32, L20502.Mabillard J. E. and Aldridge R. J. (1985) Microfossil distribution

across the base of the Wenlock Series in the type area.Palaeontology 28, 89–100.

Marshall J. D. (1992) Climatic and oceanographic isotopic signalsfrom the carbonate rock record and their preservation. Geol.

Mag. 129, 143–160.Mii H. S. and Grossman E. L. (1994) Late Pennsylvanian

seasonality reflected in the 18O and elemental composition ofa brachiopod shell. Geology 22, 661–664.

Mii H. S., Grossman E. L. and Yancey T. E. (1999) Carboniferousisotope stratigraphies of North America: implications forCarboniferous paleoceanography and Mississippian glaciation.Geol. Soc. Am. Bull. 111, 960–973.

Molnar P. (2004) Late Cenozoic increase in accumulation rates ofterrestrial sediment: how might climate change have affectederosion rates? Annu. Rev. Earth Planet. Sci. 32, 67–89.

Muehlenbachs K. (1998) The oxygen isotopic composition of theoceans, sediments and the seafloor. Chem. Geol. 145, 263–273.

Muehlenbachs K. and Clayton R. N. (1976) Oxygen isotopecomposition of the oceanic crust and its bearing on seawater. J.

Geophys. Res. D: Atmos. 81, 4365–4369.Muehlenbachs K., Furnes H., Fonneland H. C. and Hellevang B.

(2003) Ophiolites as faithful records of the oxygen isotope ratioof ancient seawater: the Solund-Stavfjord Ophiolite Complex asa Late Ordovician example. Geol. Soc. London Spec. Pub. 218,401–414.

Muller R. D., Sdrolias M., Gaina C. and Roest W. R. (2008) Age,spreading rates, and spreading asymmetry of the world’s oceancrust. Geochem. Geophys. Geosyst. 9, 1–19.

Munnecke A., Samtleben C. and Bickert T. (2003) The IrevikenEvent in the lower Silurian of Gotland, Sweden—relation tosimilar Palaeozoic and Proterozoic events. Palaeogeogr. Palae-

oclimatol. Palaeoecol. 195, 99–124.Passey B. H. and Henkes G. A. (2012) Carbonate clumped isotope

bond reordering and geospeedometry. Earth Planet. Sci. Lett.,223–236.

Passey B. H., Levin N. E., Cerling T. E., Brown F. H. and Eiler J.M. (2010) High-temperature environments of human evolutionin East Africa based on bond ordering in paleosol carbonates.Proc. Natl. Acad. Sci. USA 107, 11245–11249.

Perez Huerta A., Cusack M. and England J. (2007) Crystallogra-phy and diagenesis in fossil craniid brachiopods. Palaeontology

50, 757–763.Quade J., Eiler J., Daeron M. and Achyuthan H. (2013) The

clumped isotope geothermometer in soil and paleosol carbon-ate. Geochim. Cosmochim. Acta 105, 92–107.

Reynolds R. W. and Smith T. M. (1994) Improved global seasurface temperature analyses using optimum interpolation. J.

Clim. 7, 929–948.Riding R. and Watts N. R. (1991) The lower Wenlock reef

sequence of Gotland: facies and lithostratigraphy. Geol. Foeren.

Stockholm Foerh. 113, 343–372.Saenger C., Affek H. P., Felis T., Thiagarajan N., Lough J. M. and

Holcomb M. (2012) Carbonate clumped isotope variability inshallow water corals: temperature dependence and growth-related vital effects. Geochim. Cosmochim. Acta 99, 224–242.

Samtleben C., Munnecke A., Bickert T. and Patzold J. (1996) TheSilurian of Gotland (Sweden): facies interpretation based onstable isotopes in brachiopod shells. Geol. Rundsch. 85, 278–292.

Samtleben C., Munnecke A., Bickert T. and Patzold J. (2001) Shellsuccession, assemblage and species dependent effects on the C/















































































































































O-isotopic composition of brachiopods—examples from theSilurian of Gotland. Chem. Geol. 175, 61–107.

Schmahl W. W., Griesshaber E., Neuser R., Lenze A., Reinhart J.and Brand U. (2004) The microstructure of the fibrous layer ofterebratulide brachiopod shell calcite. Eur. J. Mineral. 16, 693–697.

Schrag D. P., DePaolo D. J. and Richter F. M. (1995) Recon-structing past sea surface temperatures: correcting for diagen-esis of bulk marine carbonate. Geochim. Cosmochim. Acta 59,2265–2278.

Shackleton N. J. and Kennett J. P. (1974) Paleotemperature historyof the Cenozoic and the initiation of Antarctic glaciation:oxygen and carbon isotope analyses in DSDP Sites 277, 279 and281. Init. Rep. Deep Sea Drill. Proj. 29, 743–756.

Shields G. A., Carden G. A. F., Veizer J., Meidla T., Rong J. Y.and Li R. Y. (2003) Sr, C, and O isotope geochemistry ofOrdovician brachiopods: a major isotopic event around theMiddle-Late Ordovician transition. Geochim. Cosmochim. Acta

67, 2005–2025.Stolper D. A., Halevy I. and Eiler J. M. (2011) The kinetics of the

ordering of 13C-18O bonds in calcite and apatite. AGU Fall

Meeting, #PP51E-05 (abstr.).Thiagarajan N., Adkins J. and Eiler J. (2011) Carbonate clumped

isotope thermometry of deep-sea corals and implications forvital effects. Geochim. Cosmochim. Acta 75, 4416–4425.

Torsvik T. H., Smethurst M. A., Van der Voo R., Trench A.,Abrahamsen N. and Halvorsen E. (1992) Baltica. A synopsis ofVendian–Permian palaeomagnetic data and their palaeotecton-ic implications. Earth Sci. Rev. 33, 133–152.

Tripati A. K., Eagle R. A., Thiagarajan N., Gagnon A. C., BauchH., Halloran P. R. and Eiler J. M. (2010) 13C–18O isotopesignatures and ‘clumped isotope’ thermometry in foraminiferaand coccoliths. Geochim. Cosmochim. Acta 74, 5697–5717.

Turchyn A. V., Alt J. C., Brown S. T., DePaolo D. J., Coggon R.M., Chi G., Bedard J. H. and Skulski T. (2013) Reconstructingthe oxygen isotope composition of late Cambrian and Creta-

ceous hydrothermal vent fluid. Geochim. Cosmochim. Acta 123,440–458.

Urey H. C. (1947) The thermodynamic properties of isotopicsubstances. J. Chem. Soc., 562–581.

van Geldern R., Joachimski M. M., Day J., Jansen U., Alvarez F.,Yolkin E. A. and Ma X. P. (2006) Carbon, oxygen andstrontium isotope records of Devonian brachiopod shell calcite.Palaeogeogr. Palaeoclimatol. Palaeoecol. 240, 47–67.

VanDeVelde J. H., Bowen G. J., Passey B. H. and Bowen B. B.(2013) Climatic and diagenetic signals in the stable isotopegeochemistry of dolomitic paleosols spanning the Paleocene–Eocene boundary. Geochim. Cosmochim. Acta 109, 254–267.

Veizer J., Fritz P. and Jones B. (1986) Geochemistry of brachio-pods: oxygen and carbon isotopic records of Paleozoic oceans.Geochim. Cosmochim. Acta 50, 1679–1696.

Veizer J., Ala D., Azmy K., Bruckschen P., Buhl D., Bruhn F.,Carden G. A. F., Diener A., Ebneth S., Godderis Y., Jasper T.,Korte C., Pawellek F., Podlaha O. G. and Strauss H. (1999)87Sr/86Sr, d13C and d18O evolution of Phanerozoic seawater.Chem. Geol. 161, 59–88.

Veizer J., Godderis Y. and Franc�ois L. M. (2000) Evidence fordecoupling of atmospheric CO2 and global climate during thePhanerozoic eon. Nature 408, 698–701.

Wallmann K. (2001) The geological water cycle and the evolutionof marine d18O values. Geochim. Cosmochim. Acta 65, 2469–2485.

Watts N. R. and Riding R. (2000) Growth of rigid high-relief patchreefs, Mid-Silurian, Gotland, Sweden. Sedimentology 47, 979–994.

Wenzel B., Lecuyer C. and Joachimski M. M. (2000) Comparingoxygen isotope records of silurian calcite and phosphate—d18Ocompositions of brachiopods and conodonts. Geochim. Cosmo-

chim. Acta 64, 1859–1872.

Associate editor: Claire Rollion-Bard


















































































Documents

Carbonate clumped isotope constraints on Silurianbiogeochem.wustl.edu/wp-content/uploads/2016/08/Cummins_Gotland... · Carbonate clumped isotope constraints on Silurian ocean temperature