techniques for assessing spatial heterogeneity of carbonate

Techniques for assessing spatial heterogeneity of carbonated13C values: Implications for craton-wide isotope gradients

J . GARRECHT METZGER and DAVID A. FIKEDepartment of Earth and Planetary Sciences, Washington University, St. Louis, MO 63130, USA(E-mail: [email protected])

Associate Editor – Adrian Immenhauser

ABSTRACT

The sedimentary record of carbonate carbon isotopes (d13Ccarb) provides one

of the best methods for correlating marine strata and understanding the long-

term evolution of the global carbon cycle. This work focuses on the Late Ordo-

vician Guttenberg isotopic carbon excursion, a ca 2�5& positive d13Ccarb

excursion that is found in strata globally. Substantial variability in the appar-

ent magnitude and stratigraphic morphology of the Guttenberg excursion at

different localities has hampered high-resolution correlations and led to diver-

gent reconstructions of ocean chemistry and the biogeochemical carbon cycle.

This work investigates the magnitude, spatial scale and sources of isotopic

variability of the Guttenberg excursion in two sections from Missouri, USA.

Centimetre-scale isotope transects revealed variations in d13Ccarb and d18Ocarb

greater than 2& across individual beds. Linear d13Ccarb to d18Ocarb mixing

lines, together with petrographic and elemental abundance data, demonstrate

that much of the isotopic scatter in single beds is due to mixing of isotopically

distinct components. These patterns facilitated objective sample screening to

determine the ‘least-altered’ data. A d18Ocarb filter based on empirical d18Ocarb

values of well-preserved carbonate mudstones allowed further sample dis-

crimination. The resulting ‘least-altered’ d13Ccarb profile improves the under-

standing of regional as well as continental-scale stratigraphic relations in this

interval. Correlations with other Laurentian sections strongly suggest that: (i)

small-scale variability in Guttenberg excursion d13Ccarb values may result in

part from local diagenetic overprinting; (ii) peak-Guttenberg excursion d13Ccarb

values of the Midcontinent are not distinct from their Taconic equivalents;

and (iii) no primary continental-scale spatial gradient in d13Ccarb (for example,

arising from chemically distinct ‘aquafacies’) is required during Guttenberg

excursion-time. This study demonstrates the importance of detailed petro-

graphic and geochemical screening of samples to be used for d13Ccarb chemos-

tratigraphy and for enhancing understanding of epeiric ocean chemistry.

Keywords Aquafacies, carbonate diagenesis, chemostratigraphy, Guttenbergexcursion, isotope gradients, isotopic heterogeneity, sea water aging, sedi-mentation rate.

INTRODUCTION

General background

Secular variation in the carbon isotopic composi-tion of marine limestones can be used to map the

spatial and temporal patterns in sedimentationacross basins and to link these to environmentaland ecological changes. A time-varying d13C sig-nal in the marine dissolved inorganic carbon(DIC) reservoir can arise from changes in the fluxand isotopic composition of the major sources

© 2013 The Authors. Journal compilation © 2013 International Association of Sedimentologists 1

Sedimentology (2013) doi: 10.1111/sed.12033

and sinks of carbon to and from the ocean (Kump& Arthur, 1999). The variation in d13CDIC can becaptured in coeval sediment as carbonate carbon(d13Ccarb) or following biological uptake asorganic carbon (d13Corg). The resulting d13C dataprovide a means to correlate sedimentary stratawithin and between sedimentary basins by align-ing their d13C chemostratigraphic profiles (e.g.Knoll et al., 1986; Hayes et al., 1999) revealingstratigraphic information undetectable using con-ventional lithostratigraphic or biostratigraphicmethods. For example, discontinuous changes ind13Ccarb profiles provide a means to identify cryp-tic hiatal surfaces in sedimentary strata (e.g. Joneset al., 2011) because d13Ccarb should smoothlyvary when deposition is continuous. Even whendeposition is continuous, various post-deposi-tional processes, such as microbial oxidation oforganic carbon (Patterson & Walter, 1994) andmeteoric diagenesis (Allan & Matthews, 1982; Jo-achimski, 1994; Immenhauser et al., 2002), mayalter a primary d13Ccarb signal, typically towardmore 13C-depleted values. As such, screening fordiagenetic alteration is critical when usingd13Ccarb data for basinal correlation or for recon-structing the evolution of the global carbon cycle.The potential utility of d13Ccarb chemostratigra-

phy for correlations and environmental recon-structions depends on spatial homogeneity inthe isotopic composition of the marine DIC res-ervoir. Recently, it has been suggested that manymarine strata, particularly those formed in epei-ric settings with uncertain connections to theglobal ocean, have carbon isotope signatures inwhich local processes (for example, aging of thewater mass; Immenhauser et al., 2008) variablyoverprint a global d13C signal (Simo et al., 2003;Panchuk et al., 2005, 2006; Fanton & Holmden,2007; Brand et al., 2009), complicating regional-scale to global-scale correlations and attempts tounderstand the processes driving carbon cyclevariability.This work focuses on the Guttenberg carbon

isotope excursion, a positive ca 2�5& excursionin d13Ccarb during the Mohawkian Series of theUpper Ordovician, which provides an examplewhere chemostratigraphy has allowed for regio-nal and global correlation of strata from differentpalaeoenvironments across much of the UnitedStates (Patzkowsky et al., 1997; Ludvigson et al.,2000, 2004; Young et al., 2005; Bergstr€om et al.,2010a,b; Coates et al., 2010), portions of south-eastern Canada (Bergstr€om et al., 2010a,b),Sweden (Bergstr€om et al., 2004), Estonia(Martma, 2005; Kaljo et al., 2007), China (Young

et al., 2005; Bergstr€om et al., 2009a) and Malay-sia (Bergstrom et al., 2010c). However, differ-ences in the stratrigraphic expression of theGuttenberg excursion (including the shape ofthe d13Ccarb curve as a function of stratigraphicposition, as well as the pre-excursion, peak-excursion and post-excursion d13Ccarb valuesand the degree of scatter) have hampereddetailed correlation between sections andobscured a fundamental understanding of theenvironmental and/or ecological changes thatcaused the Guttenberg excursion (Patzkowskyet al., 1997; Ludvigson et al., 2004; Panchuket al., 2005, 2006; Young et al., 2005, 2008;Bergstr€om et al., 2010a).In North America, stratigraphic correlations

have been facilitated by the co-occurrence of theGuttenberg excursion and the two widely trace-able K-bentonites, the Deicke and Millbrig(Kolata et al., 1986, 1987, 1996, 1998, 2001).These bentonites act as time markers used to testchemostratigraphic and sequence stratigraphicmodels, providing a framework in which toinvestigate temporal and spatial variation in sed-imentation rates (Leslie & Bergstr€om, 1997), ero-sion (Railsback et al., 2003), facies migrationand sequence development (Holland & Patzkow-sky, 1998), as well as geochemical and isotopicheterogeneities in carbonates and their palae-oceanographic sources (Panchuk et al., 2006;Fanton & Holmden, 2007). Ordovician strata arereplete with bentonites (e.g. Kolata et al., 1996)and care must be taken in cross-continental cor-relation of K-bentonites, as apatite phenocrystgeochemistry (Sell & Samson, 2011) and radio-metric ages (e.g. Huff, 2008) of Laurentian andBaltoscandian K-bentonites suggest that someprevious bentonite-based correlations may notbe correct.The goals of this work are to: (i) document

cm-scale variability in d13Ccarb and d18Ocarb anduse petrographic and trace geochemical indica-tors of diagenesis to understand the processesby which isotopic signals can be altered as theyrelate to depositional setting; (ii) develop anobjective sample screening procedure to be usedin bulk carbonate analysis for identifying least-altered data; (iii) use d13Ccarb chemostratigraphyto correlate between two localities in Missouri,USA and re-evaluate existing stratigraphic rela-tions; and (iv) use this interpretative frameworkto compare the stratigraphic expression of theGuttenberg excursion in Missouri with existingrecords from across the continent to test hypo-theses on the existence of large-scale (103 to

© 2013 The Authors. Journal compilation © 2013 International Association of Sedimentologists, Sedimentology

2 J. G. Metzger and D. A. Fike

104 km) spatial gradients in d13Ccarb (i.e. ‘aqua-facies’ of Holmden et al., 1998) at this time.

Geological setting

Strata investigated in the present study occurwithin the Mohawkian Series of North America,which is roughly equivalent to the uppermostSandbian to lower Katian Global Series (Fig. 1;Bergstr€om et al., 2009b). Sections begin in theuppermost Turinian and end in the lower tomiddle Chatfieldian (North American Stages)based on correlations using globally extensive

K-bentonites and biostratigraphy (Thompson,1991; Leslie, 2000; Bergstr€om et al., 2009b,2010a,b). During this time, Missouri was locatedin sub-equatorial latitudes 15 to 30°S (Scotese &McKerrow, 1990) where sedimentation occurredin a warm, shallow sea (Fig. 2A). The Mohaw-kian contains a major change in sediment depo-sition across Missouri (Thompson, 1991), otherportions of the upper Mississippi Valley (Kolataet al., 1998) and the mid-South (Holland &Patzkowsky, 1998). In Missouri, the geographicthickness patterns of stratigraphic units becomeroughly inverted by early Chatfieldian time

Fig. 1. Proposed chronostratigraphic relations of select locations within the United States. Series and stages takenfrom Bergstr€om et al. (2009b). M4 and M5 refer to Mohawkian sequences (Holland & Patzkowsky, 1997, 1998).Biozones taken from Leslie (2000). Information for ‘Missouri (previous)’ taken from Thompson (1991). ‘Missouri(this report)’ constructed from new data and placement of the Castlewood Limestone into the Plattin groupaccording to Kolata et al. (1998). ‘Upper KL’ and ‘Lower KL’ correspond to new divisions of the Kings Lake Lime-stone used in this work for central-southern Missouri. Iowa stratigraphy adapted from Ludvigson et al. (2004).Kentucky and Ohio information taken from Leslie (2000) and McLaughlin & Brett (2007). New York informationadapted from Leslie (2000) and Mitchell et al. (2004). Horizontal bars with ‘X’ correspond to K-bentonites.CM = Carimona, CV = Curdsville, CW = Castlewood, D = Deicke K-bentonite, Fm = Formation, Gp = Group,HS = House Springs K-bentonite, KW = Kimmswick Limestone, Ls = Limestone, M = Millbrig K-bentonite. Wherecorrelations are ambiguous dashed lines and question marks are used.


Assessing spatial heterogeneity of carbonate d13C 3

(Kolata et al., 1998) as the carbonate depocentreshifted north towards Iowa. The uplift of theOzark Dome (modern day St. Francois Moun-tains) may have partially controlled sedimenta-tion patterns and is thought to haveintermittently interrupted regional carbonatesedimentation during the Early to Middle Ordo-vician, although sedimentation became morecontinuous during the Late Ordovician high-stand (McCracken, 1966; Thompson, 1991). Fur-ther, the rhyolites and granites of the Ozarkdome may have been an important siliciclasticsource for the present study sections (Fig. 2B).Although Mohawkian-aged strata in Missourihave been studied for over 100 years (e.g. Ulrch,1904; Kay, 1935; Templeton & Willman, 1963;Kolata et al., 1986), there is a dearth of literaturethat directly focuses on their deposition,sequence stratigraphy and chemostratigraphy,especially when compared to the time-equiva-lent Trenton-Black River succession of the east-ern United States (e.g. Brett et al., 2004;Mitchell et al., 2004; McLaughlin & Brett, 2007).Chronostratigraphic relations of selected

Upper Ordovician sections across the North

American Craton are shown in Fig. 1 and wereconstructed using K-bentonite stratigraphy,biostratigraphy and d13Ccarb stratigraphy. Thestudied interval ranges from the upper PlattinGroup through the lower Kimmswick Limestone.The Plattin Group (Platteville equivalent outsideof Missouri, Fig. 1) is thickest in nearby south-western Illinois where it reaches over 200 m(Kolata et al., 2001). The Plattin is ca 140 mthick in south-eastern Missouri near Cape Girar-deau and thins north-westward, where 300 kmnorth-west it is 14 m thick (Martin et al., 1961).The Plattin is uppermost Turinian and is equiv-alent to the upper Black River succession ofNew York, based on K-bentonite stratigraphy ofthe Deicke K-bentonite dated to 454�5 � 0�5 Ma(Tucker, 1992) and conodont biostratigraphy(Leslie, 2000; Brett et al., 2004). The PlattinGroup is interpreted to represent a major trans-gressive cycle as flooding spread north and westfrom the Sebree Trough, a bathymetric low thatruns north-east to south-west along the north-western Kentucky border (Witzke & Kolata,1988). Plattin Group strata are dominated bythin-bedded, somewhat nodular mudstones

A B

Fig. 2. (A) Palaeogeographic map showing general depositional environment and important geological structures(adapted from Scotese & McKerrow, 1990; Holland & Patzkowsky, 1998; Kolata et al., 2001; Simo et al., 2003).Stars represent sample locations HM = Highway MM and NL = New London. State abbreviations (italics) are asfollows: IA = Iowa, KY = Kentucky, MO = Missouri, NY = New York, OH = Ohio, PA = Pennsylvania, TN = Ten-nessee, VA = Virginia and WV = West Virginia. (B) Map of Missouri showing sample locations, generalized sur-face stratigraphy (under the Pleistocene alluvium) and major structural features of eastern Missouri. Surface strataadapted from Missouri Department of Natural Resources (2009). Structural features adapted from McCracken(1966). Strata are sedimentary except Precambrian igneous rocks.



characterized by fine-grained, heavily bioturba-ted, high-purity (>95% carbonate) limestoneswith partially dolomitized burrows (Thompson,1991).The overlying Galena Group is thickest in the

north-west and pinches out approaching theSebree Trough, where Plattin Group deposits arethickest; this represents an apparent restructur-ing of sediment deposition patterns from Plattintime (Witzke & Kolata, 1988). In Missouri,Galena Group strata comprise the Decorah For-mation (composed of the Glencoe Shale, KingsLake Limestone and Guttenberg Limestone) andthe overlying Kimmswick Limestone. The Glen-coe Shale contains dark, thinly bedded, fissile,carbonate-poor shales intercalated with planar,thin, sometimes discontinuous brachiopod-bryo-zoan calcarenites and calcirudites (Thompson,1991; Kolata et al., 1998), as well as the MillbrigK-bentonite, which is dated to 453�1 � 1�3 Ma(Tucker & McKerrow, 1995) and defines theTurinian-Chatfieldian boundary (Ludvigsonet al., 2004). The overlying Kings Lake Lime-stone is composed mostly of blue-grey carbonatemudstones and storm-bed packstones, both withthin blue shale partings. Kings Lake facies havebeen interpreted as transitional between theGlencoe Shale and the overlying GuttenbergLimestone in north-eastern Missouri (Kolataet al., 1986). In central and southern Missouri,the Kings Lake Limestone is unconformablyoverlain by the Kimmswick Limestone, while itis conformable with the overlying GuttenbergLimestone in northernmost Missouri. TheGuttenberg Limestone is only found north of St.Louis, Missouri and like the underlying KingsLake Limestone, is composed of mudstones withfrequent carbonate storm beds. In Missouri, thebrown mud and shale partings of the Guttenbergdistinguish the unit from the underlying blue-grey mud and shale partings of the Kings LakeLimestone. The contact between the DecorahFormation and the overlying Kimmswick Lime-stone is a minor unconformity with rip-up litho-clasts from the Guttenberg Limestone (northernMissouri) and Kings Lake Limestone (southernMissouri) found in the lowermost 20 cm of theKimmswick Limestone (Kolata et al., 1998).The Kimmswick Limestone is a well-cemen-

ted, medium to fine-grained crinoid-brachio-pod-bryozoan grainstone (Thompson, 1991;Kolata et al., 1998). Strata are thick to massivelybedded with evidence for crossbedding rare orabsent in upper portions (Thompson, 1991). TheKimmswick is a very pure limestone (>98%

carbonate) that contains abundant centimetrediameter burrows that weather to a ‘Swiss-cheese’ texture similar to the Plattin Limestone.It is exposed at the surface in northern Missouriand erosionally truncated and overlain by Mid–Late Paleozoic sandstones south of St. Louis.

Study locations

Two locations in eastern Missouri were chosenfor high-resolution chemostratigraphic study(Fig. 2). The first site is located 8 km south ofNew London, Missouri along the eastern front-age road (East Side Drive) off of Highway 61,which is located a few hundred m north ofSpencer Creek (Latitude: 39�522415°N; Longi-tude: 91�344629°W). The New London section isthe same as Location 53 in Kolata et al. (1986).The sampled interval begins with 7�5 m of theupper Plattin Limestone and continues throughthe Decorah Formation and extends 8 m into theKimmswick Limestone (Fig. 3). The second site,Highway MM, is a fresh road-cut exposurefound along Missouri State Highway MM nearBarnhart, Missouri (Latitude: 38�394101°N,Longitude: 90�543364°W) approximately 15 kmwest of Location 63 of Kolata et al. (1986). Thisstudy is the first published geological record forthis location that the authors are aware of. TheHighway MM section begins in the uppermost10 to 15 m of the Plattin Limestone and contin-ues through the Decorah Formation and into thefirst 5 m of the Kimmswick Limestone (Fig. 4).

METHODS

Sampling and petrology

Field sampling in both locations was completedover a lateral distance of ca 300 m along a singleoutcrop face. Individual beds (Highway MM:n = 80; New London: n = 69) were sampled atregular intervals where possible and some bedswere collected in replicate (Tables S1 and S2).No large faults, fractures, folds, joints, or igne-ous features were observed at either location.Stratigraphic columns were constructed usinga combination of the Dunham nomenclatureand Wentworth-Udden grain-size terminology(Fl€ugel, 2009). Stratigraphic units were identi-fied in the field using the criteria in Thompson(1991). Deicke, Millbrig and House SpringsK-bentonites were identified in the field basedon stratigraphic position (Kolata et al., 1986),



P = PackstoneG = Grainstone (fine)G = Grainestone (coarse)

B

A

G

FE

CD

G

F

E

DC

BA

c

New London

Gp.

/Fm

.

GGS MWPf

Millbrig

Deicke

NLM-2

NLKL-1

CW

GS

Mac

y L

imes

tone

Kin

gs L

ake

Kim

msw

ick

Lim

esto

neD

ecor

ah F

orm

atio

nP

latt

in G

roup

Gut

tenb

erg

8

12

16

4

0

20

24

Hei

ght (

m)

Mem

ber

Limestone

Dolomitized burrows

Bioclastic storm bed

Nodular or irregular limestone K-bentonite

Chert

Legend

‘Horseshoe’ stains

Black stained burrow

Curved cross- laminations

Rip-up clasts

Unconformity

S = ShaleM = MudstoneW = Wackestone

f

c

Dunham Key

Fig. 3. Lithology and selected facies for New London. See Fig. 1 for biozone, sequence and stratigraphic designa-tions. Scale bars (white) are 1 cm. Pen markings used for carbonate sampling. Samples are as follows: (A) NL59,lower Plattin wackestone; (B) NL21, burrowed calcarenite with dolomitic zones (green); (C) NL32, muddy tempes-tite wackestone; (D) NL37, muddy tempestite packstone; (E) NL49, cross-laminated mudstone; (F) NL56, lower-most Kimmswick with Guttenberg rip-up clasts; (G) NL65, clean Kimmswick grainstone with burrowing (brown).



A B

C D

E F

G H

Fig. 4. Lithology and selected facies for Highway MM. See Fig. 1 for biozone, sequence and stratigraphic designa-tions. See Fig. 3 for legend. Scale bars (white) are 1 cm. Pen markings used for carbonate sampling. Samples areas follows: (A) HM2, upper Plattin mudstone-wackestone with yellow dolomitized burrows; (B) HM18, coarse-grained calcarenite with dolomitic green clay; (C) HM25, fine-grained calcarenite; (D) HM27, tempestite packstone;(E) HM28, mudstone with black stained burrows; (F) HM55, wackestone showing dolomitic ‘wispy brushtrokes’;(G) HM61, zone i (0 to 3 cm) mudstone capped by hardground, zone ii (3 to 5 cm) is argillaceous and dolomitized,zone iii (5 to 8 cm) is a tempestite lens and zone iv (8 to 14 cm) is stylonodular mudstone; (H) HM42, burrowedlower Kimmswick grainstone.



thickness, colour, clay content, degree of lithifi-cation, bedding geometry and zircon abundance.In addition, several minor bentonites (HMMA,HMKL-1, NLKL-1 and NLM-2) were identifiedand named for the location/formation. Two ofthese bentonites (HMKL-1 and NLKL-1) weretentatively correlated in this study and may beequivalent to the Elkport K-bentonite of theupper Mississippi Valley (Kolata et al., 1986);however, no definitive correlation of these minorbentonites to equivalents outside of the studyarea is made.Forty-seven thin sections from 45 individual

beds were analysed under transmitted andreflected light microscopy on a Leica petro-graphic microscope (model DM-2500P; LeicaMicrosystems GmbH, Wetzlar, Germany) with amaximum magnification of 5009. Photographswere taken with a Leica (model DFC 295) micro-scope-mounted camera.

Geochemical methods

Stable carbon and oxygen isotope analysis ofcarbonateCarbonate samples were drilled from polishedrock slabs cut perpendicular to the beddingplane using a Woodtek Drill (model 109370;Woodworkers Supply Inc., Casper, Wyoming,USA) fitted with 1 to 2 mm carbide drill bit tips.Powdered samples were placed in vials, flushedwith He gas, and converted to CO2 through reac-tion with anhydrous phosphoric acid (Epstein &Mayeda, 1953). Acidified samples were heatedfor 1 to 23 h at 70˚C on a Gas Bench II (ThermoFisher Scientific, Waltham, MA, USA). EvolvedCO2 was analysed on a Thermo Fischer Delta VPlus Isotope Ratio Mass Spectrometer at Wash-ington University. Values are reported in permil (&) relative to the Vienna Pee Dee belemnite(VPDB) standard. All runs contained internalstandards as well as international standardsNBS-18, NBS-19 and LSVEC. For a single run,one standard deviation (1r) averaged 0�04& ford13Ccarb and d18Ocarb for all standards, whileaverage reproducibility (1r) for d13Ccarb of repli-cate samples for a single run was 0�07&. Repro-ducibility (1r) across all sampling days ford13Ccarb was 0�09& for NBS-18, 0�10& for NBS-19 and 0�25& for LSVEC, and for d18Ocarb was0�12& for NBS-18 and NBS-19 and 0�13& forLSVEC.Samples were dominantly micrite and skele-

tal grainstones with no large altered phases orfeatures (for example, spar, marl, stylolites, do-

lomitized burrows, etc.), although selected dia-genetic textures were analysed to test forisotopic alteration and trace element abundancepatterns as a function of distance from thealtered texture. The desired sampling resolutionin these sections precluded relying on a bra-chiopod-based approach (Veizer et al., 1997,1999; Brand, 2004). In total, more than 700

A

B

C

D

Fig. 5. Thin section photomicrographs and their cor-responding hand sample location. Scale bars in handsamples (white) are 1 cm. (A) Stained thin section ofHM2 in plane polarized light shows muddy and do-lomitized burrows. Calcite (CC) is red and dolomite(Dol) is white. Darker spots in hand sample are do-lomitized similar to yellow burrows. (B) Fine-grainedcalcarenite NL19 in plane polarized light. (C) NL14under plane-polarized light showing carbonate silt ormud aggregates that possibly formed by vadose zonealteration. (D) HM26 with plane-polarized light showswhole bioclast-bearing storm bed. Tr = trilobite (par-tially pyritized), Cr = crinoid arm and Br = bryozoan



samples (with over 100 duplicates) were drilledwith multiple parallel transects to identify lat-eral intra-sample variability and pinpoint themixing of isotopically distinct components.More than 450 samples were screened to con-struct bed-averaged isotope values for 66 bedsfrom Highway MM and 69 beds from NewLondon. Bed-averaged values were obtained bytaking an unweighted average of individualmicrodrilled samples identified as ‘least-altered’based on textural and petrographic analyses(see below).

Carbonate content (%carb), total organiccarbon content (TOC), d13Corg

Weight percent carbonate (wt% CaCO3) wasdetermined using a gravimetric method from thedissolution of 2 to 5 g of bulk material. Handsamples were cut in 2 cm vertical intervals todocument vertical variability across a singlebed. Diagenetic textures were avoided. Totalorganic carbon content (TOC) and d13Corg weremeasured on aliquots of rinsed and dried resi-dues of acid digestion by combustion to CO2 onan ECS 4010 Elemental Analyzer (Costech Ana-lytical Technologies Inc., Valencia, California,USA) and measured on a Thermo Fischer DeltaV Plus Isotope Ratio Mass Spectrometer(Thermo Fisher Scientific) at Washington Uni-versity. Values are reported in per mil (&) rela-tive to the Vienna Pee Dee belemnite (VPDB)standard. TOC was determined by comparingthe area of the evolved CO2 peak to standardsrun at varied masses and corrected for the car-bonate content of the sample. A total of 151samples were analysed for d13Corg. Average stan-dard deviations (1r) for d13Corg standards USGS24 (graphite), IAEA CH-6 (sucrose) and IAEACH-3 (cellulose) were 0�11&, 0�14& and 0�13&,respectively.

Elemental abundanceSamples were prepared for inductively coupledplasma optical emission spectrometry (ICP-OES) on an Optima 7300DV ICP-OES (Perkin-Elmer Inc., Waltham, Massachusetts, USA) atWashington University. Approximately 1�0 mgof carbonate powder was dissolved in 10%(Optima grade) acetic acid or with 5% HNO3

(Optima grade) in 15 ml Falcon centrifugetubes. Samples were placed on a shaker tableand left to dissolve overnight. Samples were fil-tered through a 0�2 lm nylon filter prior toanalysis. Standards were run at 1, 10, 20, 50,100 and 500 ppb concentrations for Mn and Sr,

while Ca and Mg were run using 1, 10, 50 and100 ppm standards. Costech multi-elementstandards were used in combination with singleelement standards with no first-order spectralinterferences. Standards for Ca, Mg, Mn, and Srhad an average reproducibility (1r) of <1%across concentrations. The spectral lines usedwere Ca = 317�933 nm, Mg = 279�077 nm,Mn = 257�61 nm and Sr = 407�771 nm.

PETROGRAPHIC, SEDIMENTOLOGICALRESULTS & INTERPRETATIONS

Plattin Limestone

ObservationsThe upper Plattin Limestone consists of the Macyand Castlewood Members. The Macy Member iscomposed of thin-bedded to medium-bedded,heavily burrowed mudstones and wackestoneswith occasional grainstone lenses (Figs 3A, 4Aand 4B). When weathered, differential cementa-tion of burrows and surrounding sedimentimparts a ‘Swiss cheese-like’ appearance to therock. The overlying Castlewood Limestone isheavily bioturbated, thick to massively beddedwith occasional stylolites. Burrows in the Plattinare large (ca 2 cm diameter) and pastel-yellow(Fig. 4A) or brown-grey (Figs 3A and 4A). Bur-rows are filled with euhedral dolomite crystals20 to 100 μm in diameter set in a fine-grainedcalcite matrix (Fig. 5A). Dolomite abundance ispositively correlated to bioturbation intensity,which decreases stratigraphically up within thePlattin Group. Thin section analysis shows dolo-mite abundance decreasing to <1% (by area) afew millimetres away from burrows. In general,bioclastic lenses and matrix materials werepredominantly calcite. Plattin strata occasionallycontain identifiable bryozoa, echinoderm frag-ments and trilobites set in a calcite micritematrix. Fossil texture preservation was poorerrelative to the Decorah Formation and averagewhole fossil size was smaller in the Plattin Lime-stone relative to other units in this study.

InterpretationThe Plattin Limestone is interpreted to representa sub-tidal environment with stable sea-levelconditions that terminated with the upwardshoaling Castlewood Member. The small grainsize and high burrow abundance is interpretedto represent deposition in oxygenated, intermit-tently calm conditions, while the absence of



evaporites and obvious subaerial exposure fea-tures are in agreement with a consistently sub-merged subtidal environment.

Glencoe Shale

ObservationsThe Glencoe Shale is composed of dark green togrey, thinly bedded, fissile shales and deep blueto purple calcarenites and calcirudites (Figs 4Band 5B) that sometimes pinch out over decime-tres to metres. Occasional bioclastic beds werepresent (Fig. 5C). Sedimentological structures(for example, cross-bedding and grading) and bur-rows were uncommon. Scouring created cm-scalerelief in some beds. Dolomite is absent or in traceamounts throughout most of the matrix, but isfound in yellow, red and green diagenetic tex-tures associated with bedding planes and burrowfill. Thin section analyses showed clasts to besupported in a granular cement fabric with highertotal cement content found in finer-grained grain-stones. The Glencoe Shale was nearly twice asthick at Highway MM compared to New Londonwith a larger amount of shale found at HighwayMM. The boundary between the Glencoe Shaleand overlying Kings Lake Limestone is moreabrupt at Highway MM and marked by a signifi-cant reentrant, while the lithological transition ismore gradational at New London.

InterpretationThe Glencoe Shale formed in a shallow sub-tidalenvironment (Thompson, 1991). The presence ofshales and calcareous tempestites suggests calm-water conditions punctuated by high-energystorm events that brought in allochthonousbioclasts.

Kings Lake Limestone

ObservationsThe Kings Lake Limestone differs lithologicallybetween sections. The entirety of the Kings Lakeat New London and the interval from the base ofthe Kings Lake to just above HMKL-1 at High-way MM is made of blue-grey wackestones andgrainstones (Figs 3B, 4C, 4D and 4E) with abun-dant thin shale partings. Above HMKL-1 theKings Lake Limestone is more mud rich (Fig. 4).Fossils are abundant in hand sample and thinsection throughout the entire Kings Lake andinclude trilobites, brachiopods, ostrocods andbryozoa (Figs 4D and 5D), while rare crinoidstalks were found in shale partings. At New

London, the Kings Lake contact with the overly-ing Guttenberg Limestone occurs just above abentonite horizon (NLKL-1). The contactbetween the Kings Lake and Kimmswick Lime-stones at Highway MM has relief up to 7 cmwith possible evidence for karstification in theuppermost 30 cm of the Kings Lake. A potentialhardground was identified 15 cm below the Ki-mmswick contact (Fig. 4G). In both locations,the Kings Lake contains little dolomite (<3%),except in the ‘wispy brushstroke’ textures,which are only found at Highway MM (Fig. 4F)and which vary in dolomite concentration up toca 40%.

InterpretationThe Kings Lake Limestone was deposited duringtransgression in a sub-tidal, shallow sea, likelydistal to a siliciclastic sediment source. At High-way MM, the Kings Lake represents a wider rangeof depositional environments than at New Lon-don, as evidenced by the lithological differencesabove and below HMKL-1. Hereafter, the zonesabove and below HMKL-1 are referred to as theLower and Upper Kings Lake, respectively. Thetransition to higher mud content and increasedfossil preservation in the Upper Kings Lake(Highway MM) is interpreted to represent deposi-tion in the deepest and most distal water facies atthis location, similar to conditions ascribed to theGuttenberg Limestone at New London. The LowerKings Lake at Highway MM is lithologically cor-relative with all of the Kings Lake at New Lon-don, while the Upper Kings Lake is lithologicallyequivalent with the Guttenberg Limestone at NewLondon (see below).

Guttenberg Limestone

ObservationsThe Guttenberg Limestone is only found at NewLondon and is lithologically similar to theUpper Kings Lake Limestone at Highway MM.This is seen in the mudstones and wackestone–packstone tempestites of the Upper Kings Lake(Highway MM) and Guttenberg (New London)(Fig. 3C, D and E), except that Guttenberg Lime-stone mud and shale fractions are brown (ratherthan blue), which is correlated with higher TOCcontent. Macrofossils and microfossils are abun-dant (Fig. 3C and D), while spar and dolomiteabundances are low. Burrowing textures arecommon, but textures are less obvious in handsample relative to other units in this study.Low-angle cross-laminations were observed in



δ δ δ

δ δ δ

A

B

Fig. 6. Isotopic and geochemical data for New London (top) and Highway MM (bottom). d13Ccarb and d18Ocarb aremicrodrilled samples. All other values are bed-averaged (Tables S1 and S2). Isotopic data are reported in & rela-tive to VPDB.



the upper 1 to 2 m with lighter laminations cor-responding to coarser grain sizes, decreasedorganic content and increased cement abun-dance (Fig. 3E). The Guttenberg Limestone isunconformably overlain by the KimmswickLimestone with a few centimetres of erosionalrelief present at the contact.

InterpretationSimilar to the Kings Lake Limestone, the Gutten-berg Limestone formed in a shallow epeiric seadistal to siliclastic sediment sources. Lithologi-cal similarities between the Guttenberg at NewLondon and the Upper Kings Lake at HighwayMM argue for similar depositional and environ-mental characteristics. This also suggests thatthe Upper Kings Lake and Guttenberg may be

partially time-equivalent. There is no evidenceto support erosion of the Guttenberg Limestonesouth of St. Louis as previously reported(Thompson, 1991; Kolata et al., 1998).

Kimmswick Limestone

ObservationsThe Kimmswick Limestone is a heavily bur-rowed, fine to coarse-grained limestone andappears somewhat saccharine in outcrop. Thedominant mineral is calcite with up to 1 to 3%disseminated dolomite in a granular mosaiccement fabric (sensu Fl€ugel, 2009). Grains aredominantly bioclastic. Burrows contain variabledolomite concentrations with darker, better-defined burrows bearing the highest dolomiteconcentrations (up to ca 40% dolomite) (Fig. 4H).The lowermost 20 cm of the Kimmswick containslarge (up to 10 cm) rip-up clasts from the under-lying Guttenberg (at New London) (Fig. 3F) andUpper Kings Lake (at Highway MM) strata.Porosity is highest in the upper portions of the

Kimmswick where pores are lined by a chalky yel-low dolomitic material. While Kimmswick bur-row structures are similar to those of the PlattinLimestone, the burrow walls are not as welldefined and overall mud content is much lower(Figs 3G and 4H). Cross-bedding was not observedin this study, but has been reported stratigraphi-cally farther up the Kimmswick from a sectionnear Barnhart, Missouri (Thompson, 1991).

InterpretationThe Kimmswick Limestone marked a return to ahigher energy environment impacted by wave-tide action, consistent with the presence of frag-mented bioclasts and mud-poor composition.The nature of the contact (including rip-upclasts) with the underlying Guttenberg Lime-stone (New London) and Kings Lake Limestone(Highway MM) is also consistent with a transi-tion to a higher energy environment with theonset of Kimmswick deposition. The stratigraph-ically thick, monotonous grainstones of theKimmswick Limestone suggest consistent envi-ronmental conditions during deposition.

GEOCHEMICAL RESULTS

Stratigraphic trends

Overall, d13Ccarb and d18Ocarb values range from�2 to 3& and �8 to �4&, respectively, with

δδ

δ δ

δ

Fig. 7. Geochemical cross-plot results for New Lon-don (left) and Highway MM (right). Values are iso-tope-texture screened and d18Ocarb filtered (seeMethods). r2 corresponds to linear least-squares fit;see Tables S1 and S2 for values. d13Ccarb, d

13Corg andd18Ocarb values are reported in & relative to VPDB.



similar profiles in both sections (Fig. 6; Table S1(New London); Table S2 (Highway MM). TheGuttenberg excursion dominates the d13Ccarb pro-file, rising from a baseline of d13Ccarb near 0& inthe base of the Kings Lake Limestone to a peakof 3& in the middle Kings Lake Limestone(Highway MM) and Guttenberg Limestone (NewLondon), before returning to 0& in the Kimms-wick Limestone (Fig 6). In addition, there is asmall (1 to 2&) negative excursion in d13Ccarb

found across the Deicke K-bentonite. d18Ocarb

averages –5�5& and �5& for New London andHighway MM, respectively, with lower values inthe Glencoe Shale (Highway MM only), KingsLake Limestone (New London only) andKimmswick Limestone. A rise in d18Ocarb accom-panies the rising limb of the d13Ccarb Guttenbergexcursion at both localities, while a less-pro-nounced fall in d18Ocarb is paired with the fall-ing limb of the Guttenberg excursion only atHighway MM.Superimposed on these first-order trends is per-

mil-level scatter associated with differenttextures and mineralogy. Spar, single large clastsand cement-rich and microclast-rich zones typi-

cally had lower d13Ccarb and d18Ocarb, while dolo-mite-rich regions had variable d13Ccarb andd18Ocarb, depending upon formation and texture.For example, yellow dolomitic burrow fill in thePlattin Limestone had d13Ccarb values similar tothe calcitic matrix and elevated in d18Ocarb by 0�2to 0�9&, while the smoky grey dolomite-rich zonein the same formation had d13Ccarb values andd18Ocarb up to 4& and 0�4& higher, respectively.Geochemical analysis ([Ca]/[Mg] ratios and [Sr]values) and thin section observations support thelink between observed isotopic variability anddolomitization and/or secondary cementation.Samples with significant dolomite (>5%) wereomitted from stratigraphic correlation.Isotopic variability can occur in data across

multiple spatial scales from the individual handsample to the formation level. Secondary carbon-ates of meteoric origin often contain lowerd13Ccarb and d18Ocarb values than primary material(e.g. Allan & Matthews, 1982; Veizer et al., 1997,1999; Brand, 2004) and high-temperature calcitehas a lower d18Ocarb signature owing to thestrongly temperature-dependent fractionation ofoxygen isotopes during calcite precipitation

A

B C

Fig. 8. Vertical and lateral trends in d13Ccarb (black symbols) and d18Ocarb (white symbols) in hand samples. Mul-tiple transects are split by symbol type (left = triangles, right = circles). All values are reported in & relative toVPDB. Scale bar (white) is 1 cm. (A) Calcarenite grading upwards to mudstone NL4. Linear regression of NL4showing correlation in d13Ccarb and d18Ocarb. Strong correlation results from linear mixing of a coarser, cement-richcomponent in base of rock with an isotopically heavier mud component near the top. Sample contains <1% dolo-mite. (B) Wackestone-Packstone NL14. Vertical enrichment primarily results from increasing fractions of isotopi-cally heavy mud, but contains more than two isotopically distinct components. (C) Mudstone HM57. Lowerd13Ccarb and higher d18Ocarb at bottom result from mixing with ‘wispy brushstroke’ texture, which contains abun-dant dolomite and affects the bulk isotope signals in the lower 2 cm.



A

B

C

D

Fig. 9. Plots of d13Ccarb vs. d18Ocarb for single hand samples. All values are reported in & relative to VPDB. Scale bar(white) is 1 cm. Filled green rectangles show 2r error (d13Ccarb = 0�18&, d18Ocarb = 0�26&) and represent inferred ‘pri-mary’ values. ‘Dolomitization’ refers to the trend for increasing amounts of dolomite in microdrilled samples; ‘second-ary’ refers exclusively to non-primary calcite. (A) Mudstone HM58 showing typical two-component mixing line. (B)Wackestone HM53 showing three components. (C) Mudstone-packstone HM52 showing similar trends to HM53. (D)Mudstone-Packstone HM26 showing different mixing line slopes. Convergence upon a single value is consistent withboth lithologies in the sample having common primary d13Ccarb and d18Ocarb values with differing diagenetic historiesassociated with the different Dunham classification.



(Epstein & Mayeda, 1953). To test for the influ-ence of secondary phases of meteoric origin,d13Ccarb versus d18Ocarb cross-plots were con-structed (Fig. 7). Data were divided into the threemain stratigraphic units: the Plattin, Decorah andKimmswick. No consistent formation-scalecovariance patterns were observed betweensections.The d13Corg values ranged from �33 to �24&

(Fig. 6); d13Corg is most negative beneath the De-icke K-bentonite and peaks in the Guttenberg(New London) (Fig. 6A) and Kings Lake Forma-tions (Highway MM) (Fig. 6B). From the initialbaseline values in the Plattin Limestone, d13Corg

increases twice, once each across the Deicke andMillbrig K-bentonites, above which it steadilyrises up to the contact with the Kimmswick Lime-stone. d13Corg rapidly returns to pre-excursionvalues (�30&) in the basal Kimmswick. The iso-topic offset between carbonate carbon and organiccarbon, D13C (=d13Ccarb � d13Corg), shows no sys-tematic relation between sections except in theinterval from the HMKL-1/NLKL-1 bentonite tothe Kimmswick contact, where D13C steadilydecreases from 30 to 29& (Fig. 6).Total organic carbon is less than 0�2% in most

of the Plattin Limestone, Glencoe Shale and Ki-mmswick Limestone. Highest TOC values arefound in the Decorah Formation (0�6% at High-way MM, 2�5% at New London). In contrast, car-bonate content was highest in the Plattin

Limestone and Kimmswick Limestone and low-est in the Decorah Formation. Two intervals ofdecreased carbonate content are observed: onenear the Millbrig K-bentonite and the other, afew metres above the HMKL-1/NLKL-1 bentonite(Fig. 6).Strontium concentrations ([Sr]) were consis-

tent and averaged ca 300 ppm in the PlattinLimestone. [Sr] also averaged ca 300 ppm in theGlencoe Shale, increased to around 1,000 ppmin the Lower Kings Lake Limestone (HighwayMM only) and Guttenberg Limestone (New Lon-don only), and consistently averaged ca200 ppm in the Kimmswick Limestone (TablesS1 and S2). Formational averages were similarbetween sections. Microdrilling results showdolomite-rich and recrystallized zones haddecreased [Sr].In summary, geochemical characteristics show

little variation throughout the Plattin (with theexception of the Castlewood Limestone) and Ki-mmswick Limestones (Fig. 6). These units arecharacterized by low TOC, [Sr], d13Ccarb, d

13Corg,d18Ocarb and high carbonate purity. This homoge-neous geochemical profile is paralleled by theobserved lithological homogeneity in these units.Geochemical patterns show clear signals ofincreasing d13Ccarb, TOC, [Sr] and decreasing %carb in the Decorah Formation. Within the Deco-rah the Upper Kings Lake Limestone at HighwayMM has geochemical trends similar to those seen

Fig. 10. Decision tree for isotope-texture screening and d18Ocarb filtering.



in the Guttenberg Limestone of New London(Fig. 6), but with lower TOC. The geochemicalsimilarities between the Upper Kings Lake Lime-stone (Highway MM) and Guttenberg Limestone(New London) are consistent with the lithologicalsimilarities for the same strata.

Intra-bed d13Ccarb and d18Ocarb variability

Centimetre-scale drilling transects were com-pleted to assess small-scale d13Ccarb and d18Ocarb

variability within individual samples to betterconstrain primary signatures used for correlationand carbon cycle reconstruction. These resultsrevealed significant isotopic heterogeneitywithin single beds (Fig. 8) and many samplesshowed vertical increases in d13Ccarb andd18Ocarb over just a few centimetres (Fig. 8A andB), while isotopic scatter in other samples wasrestricted to bedding planes (Fig. 8C). Excludingheavily dolomitized zones and other obviouslyaltered textures, the magnitude of change withinindividual beds in d13Ccarb was up to 2�0&,while the change in d18Ocarb was up to 3�0&.This isotopic variability was sometimes relatedto lithological transitions. For example, in Plat-tin sample NL4 (Fig. 8A), a fine-grained calcare-nite that grades upward to a mudstone, d13Ccarb

was 1& higher and d18Ocarb was 1�6& higher inthe mudstone portion. These data fell along amixing line (r2 = 0�99, n = 6) and correlate withcement content and grain size.Drilling transects across differing textures

(Fig. 9) allowed the identification of componentmixing where dolomite, cement and clasts influ-enced bulk isotope signatures. Coupled petro-graphic and isotopic analyses showed thehighest d13Ccarb and d18Ocarb samples to be mudsand dolomite-rich zones while cements and re-crystallized materials had the lightest d13Ccarb

and d18Ocarb (for example, Fig. 8A). A lineard13Ccarb to d18Ocarb trend is then an isotopemixing line where points on the line representgradational changes in the relative proportionsof two components. This logic was the basis forconstructing an ‘isotope-texture screen’ to selectthe ‘least-altered’ d13Ccarb and d18Ocarb valuesfrom populations within single samples.In some cases, plots showed changes in d13Ccarb

while d18Ocarb was invariant. For example, brownGuttenberg mudstones-packstones had higherd13Ccarb in darker brown micrite and lowerd13Ccarb in light brown micrite while d18Ocarb wasconstant. Petrographic analyses revealed a largeraverage crystal size, a likely diagenetic feature, inlighter shades of micrite that corresponded with

δ δ

Fig. 11. d13Ccarb correlations with screened and filtered data across the Guttenberg isotopic carbon excursion(GICE). Height is in metres. Colours correspond to chonostratigraphic intervals. ‘K-bent’ chronozone is boundedby K-bentonites and independent of d13Ccarb. The remaining intervals are defined by d13Ccarb chemostratigraphy.Bar graph shows relative sedimentation rate (RSR) of New London relative to Highway MM. Error bars representstandard deviation (1r) of bed-averaged values.



lower d13Ccarb, therefore the higher d13Ccarb valuesfrom darker zones were taken as more representa-tive of primary d13Ccarb. When no petrographic orgeochemical evidence could indicate which tex-tures/phases were most primary, an average of allvalues was taken.

d18Ocarb filter

While the ‘isotope-texture screen’ is useful inidentifying the ‘least-altered’ values within asingle sample, it does not necessarily help indeciding whether those values are themselveswell-suited for chemostratigraphic correlation.To address this problem, samples can be filteredusing their d18Ocarb value. d18Ocarb filtering isdone by excluding samples with d18Ocarb valuesbelow a certain threshold where the threshold isdefined as some value below the mean d18Ocarb

value for samples that passed the ‘isotope-tex-ture’ screen for a given formation. The d18Ocarb

cutoff value was arbitrarily picked as one stan-dard deviation (1r) lighter than the formationalaverage. The d18Ocarb cutoff values for each for-mation can be found in Table S3.

Hereafter, correlations and environmentalreconstructions use only d13Ccarb data thatpassed isotope-texture screening and d18Ocarb

filtering. A decision tree (Fig. 10) shows the pro-cess used to construct these ‘least-altered’d13Ccarb values. The full list of samples alongwith information as to whether they pass or failthe isotope-texture screening and d18Ocarb filtercan be found in Tables S1 and S2.

DISCUSSION

Intra-sample geochemical variability

Variations in bulk d13Ccarb values within andbetween individual samples can result fromanalysing mixtures of two or more isotopicallydistinct components (i.e. mud, cement andclasts). This was elegantly demonstrated bySwart (2008) who showed that Cenozoic peri-platform and ramp d13Ccarb values were decou-pled from open ocean signal due to mixing ofisotopically distinct aragonite from the shallowplatform with pelagic calcite materials. As

δδδ

A B C

Fig. 12. Plot of normalized Guttenberg d13Ccarb excursions in Laurentian sections where the duration of the excur-sion is normalized from 0 to 1, with 0 defined as the beginning of the excursion and 1 as the end. Generalizedglobal d13Ccarb values (heavy solid line) for the pre-Guttenberg, peak-Guttenberg and post-Guttenberg excursiontaken from Bergstr€om et al. (2009b). Instrumental error (1 standard deviation) is less than symbol size. (A) Nor-malized excursions for Missouri (MO) sections. Eureka (MOE, black diamonds) from Ludvigson et al. (2000), High-way MM (MOHM, red pentagrams) and New London (MONL, red hexagrams) from this report. (B) Normalizedexcursions ‘Midcontinent aquafacies’. Iowa (IA) subscripts refer to sample localities of Ludvigson et al. (2004).Note the general agreement in pre- Guttenberg, peak- Guttenberg, and post-Guttenberg excursion values and lowerdegrees of scatter in least-altered sections compared to altered sections. (C) Normalized excursions for all aquafa-cies with aquafacies designations taken from Young et al. (2005). Data for Virginia (VA), West Virginia (WV) andKentucky (KY) from Young et al. (2005). New York (NY) data is unpublished.



shown in Figs 8 and 9, varying mixtures ofmud, clasts, cements and dolomite explainmuch of the isotopic variability within indivi-dual hand samples. The largest range in d13Ccarb

values within an individual sample was greaterthan 2�0& (equal to ca 80% of the magnitude ofthe Guttenberg excursion), highlighting theimportance of both sample screening and high-resolution data for stratigraphic correlation andpalaeoenvironmental or palaeoceanographicreconstructions. From these results, it is appar-ent that d13Ccarb and d18Ocarb scatter within asingle sample is often unrelated to a primarymarine (i.e. water column) environmental signaland, instead, reflects the mixing of diageneticcomponents and primary materials. This frame-work provides an objective criterion for theselection of the ‘least-altered’ d13Ccarb values.Because secondary carbonates often contain

lower d13Ccarb and d18Ocarb, the heaviest d13Ccarb

values are usually considered more primary(Fig. 9A). For this work, accepting only theheaviest d13Ccarb would lead to incorporation ofsubstantial diagenetic artifacts because dolo-mite-rich samples occasionally had d13Ccarb

signatures and frequently had d18Ocarb signaturesheavier than the reconstructed ‘least-altered’ val-ues. When dolomite was present, plots with twolinear mixing lines provided an objective argu-ment for choosing most-primary d13Ccarb andd18Ocarb values (for example, Fig. 9B and C).Not all of the samples characterized fell along

a linear mixing line. This may be the result ofmulti-component mixing (for example, Fig. 9B, Cand D), natural environmental variation, or cryp-tic diagenetic alteration; samples that showedpoor covariation across d13Ccarb and d18Ocarb overa range in excess of analytical precision arebelieved to result from one of these causes. Forexample, Fig. 9D shows two intersecting mixinglines for the mudstone and packstone portions ofsample HM26. The intersection of these twolines is interpreted to represent primary d13Ccarb

and d18Ocarb signatures for all components, whilethe different downward trajectories represent theinclusion of different secondary cements possi-bly relating to different diagenetic histories.Such scatter prevented development of a well-constrained dolomite-rich d13Ccarb and d18Ocarb

signature and is thought to result either frommultiple generations of dolomite and/or calcitecement that precipitated under different condi-tions or from variable amounts of calcite/dolo-mite within the dolomitized zone. In such cases,‘least-altered’ samples were selected based on

qualitative trends in the d13Ccarb–d18Ocarb sample

population.Centimetre-scale isotope transects revealed

that the majority of samples had isotopic offsetsarising from variable amounts of secondary mate-rial. Only 2 of 11 samples (18%) were identifiedas ‘least altered’ in Fig. 9A, demonstrating thatmost samples carried a significant secondary sig-nal (i.e. offset from ‘primary’ by more than twiceinstrumental precision). The impact of secondaryalteration on d13C and d18O may be further testedon the micron-scale using secondary ion massspectrometry (SIMS) or similar instruments toallow for grain-specific isotope analysis. In sum,confident identification of ‘least-altered’ compo-nents requires a combination of petrographic,elemental abundance and isotopic analyses, par-ticularly when the magnitude of intrabed d13Ccarb

variability is similar to that of the stratigraphicsignal being investigated.

d18Ocarb filter

A d18Ocarb filter was applied to all samples thatpassed the above isotope-texture screening. Thisfilter was designed to exclude samples subjectedto pervasive resetting of d18Ocarb such as wouldoccur during meteoric diagenesis. The d18Ocarb

isotope filter provides an objective criterion forformation-scale sample discrimination in isoto-pically heterogeneous rocks, provided carefulpetrographic and chemical characterization hasbeen done. Because d18Ocarb is easier to resetthan d13Ccarb during diagenesis (Lohmann, 1988),a d18Ocarb filter provides a conservative methodfor isolating ‘least-altered’ d13Ccarb data. Withoutd18Ocarb filtering, some d13Ccarb data sets may beartificially noisy and the resulting chemostrati-graphic correlations and environmental recon-structions may be misleading or inaccurate.While careful screening and filtering can aid in

arriving at a ‘least-altered’ d13Ccarb profile a smallnumber of samples that displayed clear texturalevidence of post-depositional alteration wouldhave passed both the isotope-texture screeningand the d18Ocarb filter. This demonstrates that theisotope-texture and d18Ocarb filters remain imper-fect screens for identifying alteration and that pet-rographic and trace element abundance datashould be considered. Some beds yielded d13Ccarb

values different than those from beds above andbelow representing a break in a stratigraphicallyconsistent d13Ccarb pattern. If these aberrantd13Ccarb values were primary, they would repre-sent a complex and rapidly changing global car-



bon cycle (Kump & Arthur, 1999), but it is mecha-nistically simpler to invoke alteration by diage-netic fluids. While no samples in this work werediscarded based on stratigraphic continuity ofd13Ccarb alone this may be sufficient evidence toomit samples in other studies. If done, careshould be taken to avoid a model-driven interpre-tation of data that excludes real negative excur-sions in d13Ccarb.

d13C chemostratigraphic correlations andrelative sedimentation rates

Figure 11 shows the proposed chemostrati-graphic relations between Localities HighwayMM and New London using d13Ccarb data thatpassed both the isotope-texture screen andd18Ocarb filter. The Guttenberg excursion is aconspicuous feature of the d13Ccarb profile atboth sections, where it is preserved as a ca2�5& positive excursion, a magnitude similarto those previously reported from other Lauren-tian sections (Bergstr€om et al., 2010a). The con-tinuity of the d13Ccarb signal throughout theGuttenberg excursion at Highway MM suggeststhat the Guttenberg Limestone was neverdeposited at Highway MM and that any ero-sion was less substantial than previouslythought (Kolata et al., 1986, 1987; Thompson,1991) as the Guttenberg Limestone is the mem-ber that contains the Guttenberg excursion innorthern Missouri. A comparison of d13Ccarb

profiles shows that the difference in thicknessof the Kings Lake Limestone between NewLondon and Highway MM and the absence ofthe Guttenberg Limestone at Highway MM canbe best explained if the Upper Kings LakeLimestone (Highway MM) is a synchronoussouthern facies equivalent of the GuttenbergLimestone (New London). Figure 12A showsthe Guttenberg excursion curve for Missourisections normalized for excursion duration.The similarity in morphologies between High-way MM and New London during the fallingstage of the Guttenberg excursion shows thatsedimentation at Highway MM was as continu-ous as that at New London, but slower. Thisprovides further evidence against erosion of theGuttenberg time-equivalent strata at HighwayMM as erosion would produce a different nor-malized excursion morphology. These data arein strong agreement with lithological and geo-chemical observations.Comparison of relative sedimentation rate

(RSR) between localities can be used to under-

stand spatial differences in deposition rates. TheRSR is a unitless ratio of lithological thicknessof a given d13Ccarb interval from one section toanother (here the ratio of stratigraphic thicknessin New London strata relative to that inHighway MM strata). In the present case, theintervals are defined by a combination of K-bent-onites and features in the d13Ccarb profiles(Fig. 11). In this manner, the isotopes reveal thepartitioning of time between units and show thegeographic relation in relative sedimentationrates. The RSR is near unity during the ‘K-bent’chronozone and through the onset of the Gutten-berg excursion, arguing for similar sedimenta-tion rates at Highway MM and New London.The RSR increases above the HMKL-1/NLKL-1bentonite near peak Guttenberg excursion d13Cvalues, as net sedimentation rates at New Lon-don outpaced those at Highway MM. A possiblehardground 15 cm below the Kimmswick con-tact at Highway MM suggests that sedimentationwas very condensed at this locality but, asFig. 12A shows, not so condensed as to signifi-cantly change the normalized excursionmorphology at Highway MM. Lack of evidencefor subaerial exposure is consistent with acontinuously submerged environment at High-way MM and that the high RSR resulted, at leastin part, from declining sediment productionrates and/or a decrease in the creation of newaccommodation space at this location. The shiftin RSR is roughly coincident with the start ofthe Taconic Orogeny (Rodgers, 1971), and thedifferential subsidence may be the result of afar-field tectonic forcing on the inner craton (seediscussion in Holland & Patzkowsky, 1998) orchanging regional tectonics related to the OzarkDome. Future development of robust correla-tions in the overlying Kimmswick Limestonecould test whether these sedimentation patternscontinue in younger strata.

d13Corg

Sedimentary organic carbon is filtered through acomplex network of biological processing duringcarbon fixation, heterotrophic reworking andmicrobial respiration. As such, the d13Corg signalreflects not only variations in d13C from the par-ent dissolved inorganic carbon (DIC) source, butalso changing ecological and environmental (forexample, facies-dependent) factors. Therefore,fractionation can be influenced strongly by localsignals and resulting interpretations of d13Corg

data are less straightforward than d13Ccarb (Hayes



et al., 1999). The Guttenberg excursion is weaklyexpressed in d13Corg, where the signal is eithersuperimposed over varying biological fraction-ation or variable organic sourcing, as evidencedby the difference in stratigraphic expression ind13Corg relative to d13Ccarb (Fig. 6). The magni-tude of the Guttenberg excursion in d13Corg is ca2&, while the d13Ccarb excursion is ca 2�5& atboth locations. Other reported Guttenberg excur-sion d13Corg excursions were found to have mag-nitudes of ca 1& from West Virginia (Younget al., 2008), ca 3& from Pennsylvania (Patzkow-sky et al., 1997) and ca 8& from Iowa (Pancostet al., 1999). In this latter case, compound-specific analysis of TOC components was usedto correct for organic matter source mixing; theresulting reconstructed, source-independentmagnitude of the Guttenberg excursion was ca3�5& (Pancost et al., 1999). The d13Corg is partic-ularly variable in the zone above and below theMillbrig at Highway MM and New London mak-ing exact correlations difficult and the absence ofthe Millbrig in the Iowa core makes comparisonspeculative. Higher order trends in d13Corg areapparent at Highway MM, which may resultfrom source mixing (Pancost et al., 1999),unidentified diagenesis, a complex local signal,and/or decoupling of the d13Corg and d13Ccarb sys-tem.During catagenesis (i.e. thermal cracking), TOC

decreases as organic matter is converted to sim-ple hydrocarbons (for example, methane). Theseresulting hydrocarbons are often depleted in 13C,leaving the residual TOC enriched in 13C. Similartrends can be obtained by biological remineral-ization of organic matter by microbial respirationduring or after deposition. As such, cross-plots ofd13Corg versus TOC can provide insight into pro-cesses that alter d13Corg. A pattern of increasingd13Corg as TOC decreases can thus be indicativeof alteration in the d13Corg signal (e.g. Dehleret al., 2005) or mixture of isotopically distinctcomponents (e.g. Johnston et al., 2012). Instead,a trend toward low d13Corg with low TOC (this isparticularly noticeable at New London) isobserved; however, because the low TOC and13C-depleted samples come from strata belowand above the Guttenberg excursion (Plattin andKimmswick Limestones, respectively), while thehigh TOC, 13C-enriched samples come from stratacontaining the Guttenberg excursion, this patternmost likely reflects primary environmental vari-ability rather than alteration or contamination.Nearly all autochthonous marine organic

carbon is originally derived from the marine inor-

ganic carbon reservoir by way of photosynthesis.Assuming a constant fractionation factor duringphotosynthesis, d13Corg should be consistentlyoffset from d13Ccarb by a fixed amount. Thismeans that d13Ccarb and d13Corg should move inparallel. The isotopic offset between coeval car-bonate and organic carbon, D13C (=d13Ccarb �d13Corg), can be plotted to visually show if eithercarbon profile deviates from parallel behaviour.The D13C profiles (Fig. 6) are similar between sec-tions only from the HMKL-1/NLKL-1 bentoniteson the rising limb of the Guttenberg excursion tothe excursion termination. The D13C curves fromHighway MM and New London have a similarmorphology to the D13C curves from Iowa andPennsylvania (Patzkowsky et al., 1997) in that allcurves immediately decline following peak-Gut-tenberg excursion values. This is evidence of aspatially coherent D13C signal (on at least thecontinental-scale) with varying degrees of localoverprinting. The authors agree with Pancostet al. (1999), who argued that d13Corg correlationsduring this interval will be complicated bysource mixing and prefer to use (screened and fil-tered) d13Ccarb, rather than d13Corg for continentaland global correlations. Excluding the DecorahFormation, the lack of strong agreement in D13Cbetween sections suggests that existing pairedd13Ccarb and d13Corg data are insufficient toanswer questions of global biogeochemical Ccycling patterns across the whole study interval.More data, reflecting regionally reproducibletrends, will be needed to extract additionalmeaning and evaluate possible causes of D13Cvariation.

Correlation summary

The Guttenberg excursion is a clearly identifi-able chemostratigraphic feature in both sectionspresented here regardless of the degree of sam-ple screening and filtering. Correlations withunprocessed data have higher scatter (Fig. 6)and as such, the unprocessed signal has lesspotential to make the confident high-resolutioncorrelations needed to test geochemically basedhypotheses. For example, a compilation of alldata from this work (i.e. including those thatdid not pass the screening or filtering) wouldresult in a Guttenberg excursion morphologythat is relatively depleted in 13C and more simi-lar to that of the d13Ccarb profile from Eureka,Missouri, a location ca 13 km from HighwayMM (Ludvigson et al., 1996). It is possible thatdifferences between the ‘least-altered’ data and



the adjacent d13Ccarb profile from Eureka couldarise from local water column gradients; how-ever, based on both the scatter in theunscreened data and the similarity between thetwo ‘least-altered’ d13Ccarb profiles, it seemsmechanistically more plausible that the lowerd13C values and stratigraphic variability of theEureka section (Ludvigson et al., 1996) may inpart arise from sample selection, insufficientsampling density and/or diagenetic overprintingrather than the preservation of a primary d13Ccarb

signature. Detailed analysis of isotopic variabil-ity – in the context of petrographic and geo-chemical indicators for alteration – is necessaryto ascertain the origin of these disparities. High-resolution correlations using processed data(Fig. 11) show that carefully screened bulkd13Ccarb data can be used to correlate sectionsthat at first glance seem to have significant localoverprints, a problem that has hindered high-resolution correlations across Late OrdovicianLaurentia (e.g. Ludvigson et al., 2004; Younget al., 2005; Bergstr€om et al., 2010a,b; Coateset al., 2010). While it may not be required forlower-order correlations, detailed sample screen-ing can greatly increase knowledge of thesources of d13Ccarb variability and fine-tune theunderstanding of basin-scale sedimentary andgeochemical processes.

Depositional facies, sea-level, d13C and d18O

Facies likely exhibited both direct and indirectcontrols on d13Ccarb and d18Ocarb values at thebed scale and formation scale. Single beds (i.e.hand samples) showed differing isotope valuesrelated to burrow abundance and grain size. Forexample, samples from the peritidal PlattinLimestone contained high amounts of calciticmud with relatively invariant d13Ccarb andd18Ocarb values, while burrows were isotopicallymore variable. The burrows likely functioned asconduits for post-depositional fluids and repre-sent an indirect control of facies on isotope val-ues. Samples from more hydrologically energeticfacies, such as the Kimmswick Limestone, con-tained exhumed clasts from underlying forma-tions, a direct control on bulk isotope values.Grainstones also are more porous than mud-stones, facilitating inclusion of primary cements(direct control) and later recrystallization (indi-rect control). The more energetic facies limit theamount of carbonate mud, requiring sampling ofthe clast-cement mixtures that are associatedwith lower d13Ccarb and d18Ocarb values. In this

context grainstone facies are more susceptible toalteration (and an apparent isotopic offset) thanmuddier lithologies from less energetic facies.This seems to be the most likely cause for thelow d18Ocarb values in the grainstones of theGlencoe Shale and Kimmswick Limestone rela-tive to the Decorah Formation and Plattin Lime-stone.Facies controls on isotope values at the forma-

tion scale include the biological pump (Free-man, 2001) and ‘aquafacies’ model (sensuHolmden et al., 1998) and are the result of sea-level, microbial activity and connectivity withthe open ocean. As a result of the biologicalpump, DIC in shallow waters will become rela-tively enriched in 13C as 12C-rich organic carbonis formed and subsequently exported to the deepocean (where part of it may be oxidized back toDIC). In contrast, the aquafacies model invokeslocal oxidation of organic matter to generateshallow waters of the cratonic interior that arehydrologically restricted and depleted in 13C rel-ative to the open ocean. These models can becompared with d13Ccarb values observed forLocalities Highway MM and New London. Thehighest d13Ccarb values were observed in theKings Lake and Guttenberg Formations, both ofwhich are thought to represent the deepest andlowest energy facies. This result is opposite ofwhat the biological pump model predicts andsuggests that the biological pump is not control-ling the stratigraphic variation in d13Ccarb valuesin this interval in the Missouri area. The peakGuttenberg excursion values observed herematch those in other Laurentian localities anddo not match predicted d13Ccarb for the Missouriregion based on aquafacies-induced gradients ind13CDIC (Young et al., 2005). This suggests thatthe aquafacies model is not appropriate for thisinterval at these locations.Isotope gradients can also result from more

localized phenomena. Previous work on a slopeto platform top transect of Carboniferous carbon-ates from north-west Spain (Immenhauser et al.,2003) revealed a d13Ccarb gradient with heaviestvalues occurring in the deepest facies. Immenha-user et al., 2003 explain the heavier d13Ccarb val-ues as resulting from the increased proportion of13C-enriched ocean water mixed in during high-amplitude transgression and varying early diage-netic histories. A proportionally lower amountof 13C-enriched waters is mixed with the ‘aged’platform top water in offshore areas resulting ina relatively lower d13Ccarb profile in the moreshoreward facies. This mixing model is also not



appropriate for the Guttenberg excursion for sev-eral reasons. First, this excursion is confidentlycorrelated over multiple basins on the Lauren-tian craton and believed to be a global event,which is not the case for the middle Atokand13Ccarb shift (Immenhauser et al., 2003). Thissuggests that the mechanism responsible for theGuttenberg excursion would have to apply tothe entire craton, if not the globe, and thereforeincludes numerous different facies and at a spa-tial scale where it is unrealistic to expect river-ine waters to significantly impact d13Ccarb.Second, if the Guttenberg excursion representsthe simple mixing of ocean water with 13C-depleted, chemically-evolved epeiric platformwater then the open ocean baseline must havebeen +3&. (The absence of any abyssal Ordovi-cian-aged sea floor precludes establishing anindependent open ocean baseline.) If so, thenthe entire suite of existing records for this geo-logical period does not preserve an open oceansignal (e.g. Bergstr€om et al., 2009b). Finally,while sea-level is thought to have risen through-out the Late Ordovician (Munnecke et al., 2010),there is no corresponding, continuous increasein d13Ccarb of epeiric carbonates associated withthis increase in ocean connectivity. Therefore, itdoes not appear that a transgression thatincreased ocean connectivity everywhere wasthe responsible for the Guttenberg excursion.Another model explaining the relation

between sea-level and d13Ccarb has been pro-posed by Fanton & Holmden (2007), wherebysea-level rise was accompanied by an increasedflux of nutrient-rich open ocean water to theinner craton. This led to locally increased pri-mary productivity and organic carbon burial,which in turn increased d13CDIC. However, theagreement in d13Ccarb values during the Gutten-berg excursion across the Laurentian palaeocon-tinent argues that the origin of the higherd13Ccarb values is the result of global processes(for example, global organic carbon burial) ratherthan local or regional processes (for example,runoff contributions from nearby highlands,microbial oxidation of organic matter).

Regional and global implications

Correlations from screened and filtered data giveinsight into the geochemical nature of ancientepeiric seas by discriminating against secondarysignals. The correlations between New Londonand Highway MM are consistent with theGuttenberg excursion being an isochronous

excursion that occurred in an isotopically well-mixed ocean. The magnitude of the Guttenbergexcursion reported here (2�5&) is comparable tothose reported for the Guttenberg excursion fromIowa (2&, Pancost et al., 1999; 1�5 to 3&, Lud-vigson et al., 2004), Pennsylvania (3&, Patzkow-sky et al., 1997), Kentucky (2�0&, Coates et al.,2010), Virginia (2&, Young et al., 2005), WestVirginia (2&, Young et al., 2005) and Tennessee(2�0 to 2�5&, Bergstr€om et al., 2010a) (Fig. 12).The pre-excursion, peak-excursion and post-excursion values reported here also very closelymatch those reported for the ‘generalizedd13Ccarb curve’ of Bergstr€om et al. (2009b)(Fig. 12). Compilation of these data can be usedto understand geographic gradients and strati-graphic variability in d13Ccarb records and theirpossible syndepositional and diagenetic causes.Despite the general agreement in the afore-

mentioned Guttenberg excursion d13C records, adetailed examination of some reports revealslocal and regional variations in the stratigraphicexpression of the Guttenberg excursion. For asingle location in eastern Iowa, Ludvigson et al.(1996) argued that primary micritic d13Ccarb

might be decoupled from coeval brachiopodd13Ccarb signatures as a result of water columnstratification. Comparisons of bulk carbonated13Ccarb data with coeval brachiopods from St.Louis County, Missouri (Shields et al., 2003)show no systematic difference between the twomaterial types, although the stratigraphic resolu-tion of the brachiopods is low and the scatter ind13Ccarb is relatively high in the CastlewoodLimestone and Glencoe Shale equivalents,where the brachiopods were sampled. The pres-ent authors find no evidence for the stratifica-tion of the inner cratonic sea in the Missouriregion.Laterally restricted aquafacies with unique

water column isotopic signatures have been pro-posed as an explanation for regional patterns ind13Ccarb during the Guttenberg excursion (e.g.Young et al., 2005). The data set of Ludvigsonet al. (2004) shows apparent spatial variabilityin Guttenberg excursion d13Ccarb curves in sixcores from Iowa. Localites 4 and 5 of Ludvigsonet al., 2004 have a d13Ccarb profile very similar tothat of New London while Locality 1 (located<50 km away from Localities 4 and 5) shows adifferent morphology; cores farther away haveeven more divergent profiles (Fig. 12B). It ispossible that these cores reflect different watercolumn processes on the regional scale. Yet, thepresence of Guttenberg excursion curves in Iowa



that have the same pre-excursion, peak-excur-sion and post-excursion values as the Missourisections presented in this work (Fig. 12B) andfrom more distal locations (for example, WestVirginia) (Fig. 12C) suggests an alternativeexplanation, namely: local syndepositional andpost-depositional alteration is superimposedover a signal representing precipitation from anisotopically homogenous epeiric sea. A preva-lent mechanism by which such local variationcould result is the metabolic oxidation oforganic matter by microbes, which has been pro-posed as a contributing factor producing lowerd13Ccarb values in the pore waters and modernsediments of the Bahama Banks and Florida(Patterson & Walter, 1994). This results in scat-tered and lower average d13Ccarb values on thelocal scale. The present authors find this expla-nation more consistent with the data from thisreport than stratigraphically coherent gradientsin d13CDIC in the ancient ocean, which shouldnot produce the observed d13Ccarb trends charac-terized by high scatter and low d13Ccarb values.The Iowa cores with d13Ccarb profiles similar tothe two new Missouri sections (despite the>1000 km distance between the Iowa and Mis-souri sections and the open ocean) may be themost representative of open marine values,whereas the stratigraphically variable d13Ccarb

profiles in adjacent cores (despite their proxi-mity to each other) would result from diageneticalteration or a local water-column signal that isunrelated to any large-scale ‘ocean-to-innercraton’ gradient. Importantly, the sections withlow stratigraphic scatter in d13Ccarb all meet the‘generalized’ peak-Guttenberg excursion value(ca 2�5&) of Bergstr€om et al. (2009b), while sec-tions with higher scatter do not meet the peakvalue. The d13Ccarb scatter is preferentiallytoward lower d13Ccarb values. This differs fromtrue ‘noise’ (i.e. random scatter around a meanvalue), which should also have values that areheavier than the mean d13Ccarb, which is notobserved (Fig. 12B and C). In this view, the mid-continent Guttenberg excursion curve (corre-sponding to ‘midcontinent aquafacies’) of Younget al. (2005) was constructed using sections thatmay have been impacted by diagenetic over-printing and does not reflect a primary craton-scale d13Ccarb gradient. This interpretationsuggests that more locations may have under-gone more substantial diagenetic alteration thanpreviously assumed.This work does not argue that large-scale

isotopic gradients could not exist in another

time period or location, or for isotopic systemsother than carbon (for example, neodymium).This work does argue that for the specific caseof midcontinent records of the Guttenbergexcursion no long-range spatial gradient ind13Ccarb is required to explain the observedtrends. Instead, the existing Guttenberg excur-sion records are more parsimoniously inter-preted as reflecting differential alteration of aspatially homogenous primary d13Ccarb signal,rather than faithful preservation of what isessentially a spatially heterogenous d13Ccarb

primary signal. Figure 12B separates the mid-continent aquafacies into two categories, thosethought to track the global d13Ccarb record andthose that do not, the former being suitable forchemostratigraphic correlation and reconstruc-tions of ocean chemistry. Missouri sectionsHighway MM and New London (Midcontinentaquafacies) have the same basic Guttenbergexcursion values as the more hydrologicallyconnected Taconic and Southern aquafaciessections. Correlation with Iowa sections sug-gests that any pre-existing regional d13Ccarb gra-dient (Panchuk et al., 2006) was abolished inthe Missouri-Iowa region during Guttenbergexcursion-time coincident with the craton-wideMohawkian transgression (e.g. Kolata et al.,1998, 2001).It is clear that studies used to reconstruct spa-

tial or temporal variability in ocean chemistry orcarbon cycling need to be conducted at a highsampling resolution and data need to be placedin a rigorous depositional context and evaluatedbased on petrographic and geochemical indica-tors of alteration whenever possible. Because theunderstanding of oceanic connectivity and thebiogeochemical C-cycle hinge on d13Ccarb chemo-stratigraphic correlations, differing treatment ofdata sets can profoundly change the understand-ing of the Earth System.

CONCLUSIONS

This work supplies techniques for assessing dia-genetic alteration of d13Ccarb and d18Ocarb over alarge range of spatial scales and contributes tothe general understanding of carbon-isotopehomogeneity and hydrological connectivity inancient epeiric seas. Results from this workdemonstrate that the variable Guttenberg carbonisotope excursion profiles can be explained interms of local diagenetic alteration and are notconsistent with the previously proposed ‘aquafa-



cies’ model, which invoked an isotopically het-erogeneous epeiric sea that formed as a result ofa stable long-term and long-range ocean to cra-ton d13CDIC gradient. Specifically:1 Significant isotopic heterogeneity (up to 2&

in d13Ccarb and 3& in d18Ocarb) within singlehand samples can be superimposed over theGuttenberg carbon isotope excursion. These vari-ations typically result from the admixture ofmultiple isotopically distinct components of pri-mary and secondary origin, rather than fromdynamic carbon cycling or spatially heteroge-neous reservoirs. Screening of d13Ccarb datausing petrographic, geochemical and isotopicmethods explains apparent differences ind13Ccarb profiles between sections and the result-ing correlations have higher stratigraphic resolv-ing power than unprocessed data.2 Normalized d13Ccarb excursion profiles can

help identify alteration at the outcrop scale bycomparing d13Ccarb morphologies from within asingle basin. Normalized d13Ccarb curves for theGuttenberg excursion show that 13C-enrichedlocations have regionally reproducible d13Ccarb

profiles and low d13Ccarb scatter. In contrast, sec-tions that are relatively depleted in 13C are asso-ciated with poor regional reproducibility ofd13Ccarb trends and high d13Ccarb scatter, patternsconsistent with local diagenetic alteration.3 The strong agreement in absolute values of

d13Ccarb before, at the peak of, and after theGuttenberg excursion in sections identified as‘least-altered’ support an isotopically homoge-nous carbon reservoir.4 Paired d13Ccarb and d13Corg data yield D13C

profiles that are similar during the peak and fall-ing stage of the Guttenberg excursion; theregional reproducibility of this trend suggestspreservation of a primary D13C signature in thisinterval with possible global implications.

Chemostratigraphic correlations betweenLocalities Highway MM and New London giveinsight into the temporal and spatial relations ofUpper Ordovician strata in central and northernMissouri and relate them to other strata of simi-lar age.1 High-resolution d13Ccarb records show that

the Upper Kings Lake and Guttenberg Lime-stones in Missouri, previously thought to besuccessive, are largely coeval. The stratigraphi-cally smooth d13Ccarb profiles support a less sub-stantial erosional history in eastern Missouriduring post-Kings Lake, pre-Kimmswick timethan has been previously reported.

2 Lithological and chemostratigraphic evidencesuggests that a thickening of the Guttenbergexcursion interval in northern Missouri largelyresulted from an increased relative sedimenta-tion rate in northern Missouri during the lasthalf of the Guttenberg excursion.

ACKNOWLEDGEMENTS

The authors wish to thank R. Folkerts for herindispensable assistance in the field; D. Kolatafor guidance in the field and for discussions ofthe regional stratigraphy; D. McCay for help pre-paring and analysing samples; S. Young, R.Criss, F. Moynier and J. Catalano for discus-sions; P. Skemer for use of laboratory equip-ment; and B. Mahan for assistance withmicroscopy and photography. The manuscriptwas greatly improved by insightful commentsfrom L. Kump and an anonymous reviewer aswell as A. Immenhauser and T. Frank. Supportwas provided in part by funding from theAgouron Institute, the ACS Petroleum ResearchFund (Grant No. 51357-DNI2), and aHanse-wissenschaftskolleg Fellowship awardedto DAF. In addition, isotope analyses and travelwere partly funded by a DOSECC graduateresearch grant and a GSA graduate researchgrant awarded to JGM.

REFERENCES

Allan, J.R. and Matthews, R.K. (1982) Isotope signatures

associated with early meteoric diagenesis. Sedimentology,

29, 797–817.Bergstrom, S.M., Agematsu, S. and Schmitz, B. (2010c)

Global Upper Ordovician correlation by means of d13Cchemostratigraphy: implications of the discovery of the

Guttenberg d13C excursion (Guttenberg excursion) in

Malaysia. Geol. Mag., 147, 641–651.Bergstr€om, S.M., Huff, W.D., Saltzman, M.R., Kolata, D.R.

and Leslie, S.A. (2004) The greatest volcanic ash falls in

the Phanerozoic: Trans-Atlantic relations of the

Ordovician Millbrig and Kinnekulle K-bentonites. Sed.

Rec., 2, 4–8.Bergstr€om, S.M., Xu, C., Schmitz, B., Young, S., Jia-Yu, R.

and Saltzman, M.R. (2009a) First documentation of the

Ordovician Guttenberg d13C excursion (Guttenberg

excursion) in Asia: chemostratigraphy of the Pagoda and

Yanwashan formations in southeastern China. Geol. Mag.,146, 1–11.

Bergstr€om, S.M., Chen, X., Guti�errez-Marco, J.C. and

Dronov, A. (2009b) The new chronostratigraphic

classification of the Ordovician System and its relations to

major regional series and stages and to d13Cchemostratigraphy. Lethaia, 42, 97–107.



Bergstr€om, S.M., Schmitz, B., Saltzman, M.R. and Huff,W.D. (2010a) The Upper Ordovician Guttenburg d13Cexcursion (Guttenberg excursion) in North America and

Baltoscandia: occurrence, chronostratigraphic significance,

and paleoenvironmental relationships. In: Geol. Soc. Am.

Spec. Pap. (Eds S.C. Finney and W.B.N. Berry), 466 pp. 36

–67.Bergstr€om, S.M., Young, S. and Schmitz, B. (2010b) Katian

(Upper Ordovician) d13C chemostratigraphy and sequence

stratigraphy in the United States and Baltoscandia: a

regional comparison. Palaeogeogr. Palaeoclimatol.

Palaeoecol., 296, 217–234.Brand, U. (2004) Carbon, oxygen and strontium isotopes in

Paleozoic carbonate components: an evaluation of

original seawater-chemistry proxies. Chem. Geol., 204, 23–44.

Brand, U., Tazawa, J., Sano, H., Azmy, K. and Lee, X. (2009)Is mid-late Paleozoic ocean-water chemistry coupled with

epeiric seawater isotope records? Geology, 37, 823–826.Brett, C.E., McLaughlin, P.I., Cornell, S.R. and Baird, G.C.

(2004) Comparative sequence stratigraphy of two classic

Upper Ordovician successions, Trenton Shelf (New York-

Ontario) and Lexington Platform (Kentucky-Ohio):

implications for eustasy and local tectonism in eastern

Laurentia. Palaeogeogr. Palaeoclimatol. Palaeoecol., 210,295–329.

Coates, J.W., Ettensohn, F.R. and Rowe, H.D. (2010)

Correlations across a facies mosaic within the Lexington

Limestone of central Kentucky, USA, using whole-rock

stable isotope compositions. In: The Ordovician Earth

System (Eds S.C. Finney and W.B.N. Berry), Geol. Soc.

Am. Spec. Pap., 466 pp. 177–193.Dehler, C.M., Elrick, M., Bloch, J.D., Crossey, L.J.,

Karlstrom, K.E. and Des Marais, D.J. (2005) High-

resolution delta C-13 stratigraphy of the Chuar Group (ca.

770–742 Ma), Grand Canyon: implications for mid-

Neoproterozoic climate change. Geol. Soc. Am. Bull., 117,32–45.

Epstein, S. and Mayeda, T.K. (1953) Variations of O18

content of waters from natural sources. Geochim.Cosmochim. Acta, 4, 213–224.

Fanton, K.C. and Holmden, C. (2007) Sea-level forcing of

carbon isotope excursions in epeiric seas: implications for

chemostratigraphy. Can. J. Earth Sci., 44, 807–818.Fl€ugel, E. (2009) Microfacies of Carbonate Rocks: Analysis,

Interpretation and Application. Springer Verlag, Berlin,

984 pp.

Freeman, K.H. (2001) Isotope Biogeochemistry of Marine

Organic Carbon. In: Reviews in Mineralogy &

Geochemistry 43: Stable Isotope Geochemistry (Eds J.W.

Valley and D.R. Cole), 43, 579–605. Mineralogical Society

of America and Geochemical Society, Washington, DC.

Hayes, J.M., Strauss, H. and Kaufman, A.J. (1999) The

abundance of 13C in marine organic matter and isotopic

fractionation in the global biogeochemical cycle of carbon

during the past 800 Ma. Chem. Geol., 161, 103–125.Holland, S.M. and Patzkowsky, M.E. (1997) Distal orogenic

effects on peripheral bulge sedimentation: Middle and

Upper Ordovician of the Nashville Dome. J. Sed. Res., 67,250–263.

Holland, S.M. and Patzkowsky, M.E. (1998) Sequence

stratigraphy and relative sea-level history of the Middle

and Upper Ordovician of the Nashville Dome, Tennessee.

J. Sed. Res., 68, 684–699.

Holmden, C., Creaser, R.A., Muehlenbachs, K., Leslie, S.A.and Bergstrom, S.M. (1998) Isotopic evidence for

geochemical decoupling between ancient epeiric seas and

bordering oceans: implications for secular curves. Geology,

26, 567–570.Huff, W.D. (2008) Ordovician K-bentonites: issues in

interpreting and correlating ancient tephras. Quatern. Int.,

178, 276–287.Immenhauser, A., Kenter, J.A.M., Ganssen, G., Bahamonde,

J.R., Vliet, A.V. and Saher, M.H. (2002) Origin and

significance of isotope shifts in Pennsylvanian carbonates

(Asturias, NW Spain). J. Sed. Res., 72, 82–94.Immenhauser, A., Della Porta, G., Kenter, J.A.M. and

Bahamonde, J.R. (2003) An alternative model for positive

shifts in shallow-marine carbonate d13C and d18O.

Sedimentology, 50, 953–959.Immenhauser, A., Holmden, C. and Patterson, W. (2008)

Interpreting the carbon-isotope record of ancient shallow

epeiric seas: lessons from the recent. In: Dynamics of

Epeiric Seas (Eds B.R., Pratt, B.R., Holmden), Geol. Assoc.

Can. Spec. Pap., 48, 137–174.Joachimski, M.M. (1994) Subaerial exposure and deposition

of shallowing upward sequences: evidence from stable

isotopes of Purbeckian peritidal carbonates (basal

Cretaceous), Swiss and French Jura Mountains.

Sedimentology, 41, 805–824.Johnston, D.T., Macdonald, F.A., Gill, B.C., Hoffman, P.F.

and Schrag, D.P. (2012) Uncovering the Neoproterozoic

carbon cycle. Nature, 483, 320–324.Jones, D.S., Fike, D.A., Finnegan, S., Fischer, W.W., Schrag,

D.P. and McCay, D. (2011) Terminal Ordovician carbon

isotope stratiraphy and glacioeustatic sea-level change

across Anticost Island (Qu�ebec, Canada). Geol. Soc. Am.

Bull., 123, 1645–1664.Kaljo, D., Martma, T. and Saadre, T. (2007) Post-

Hunnebergian Ordovician carbon isotope trend in

Baltoscandia, its environmental implications and some

similarities with that of Nevada. Palaeogeogr.

Palaeoclimatol. Palaeoecol., 245, 138–155.Kay, G.M. (1935) Ordovician System of the upper

Mississippi Valley. In: Kansas Geological Society 9th

regional field conference, Upper Mississippi River Valley,

(Ed. A.C. Trowbridge), pp. 281–295. Kansas Geological

Society, Lawrence, KS.

Knoll, A.H., Hayes, J.M., Kaufman, A.J., Swett, K. and

Lambert, I.B. (1986) Secular variation in carbon isotope

ratios from Upper Proterozoic Successions of Svalbard and

East Greenland. Nature, 321, 832–838.Kolata, D.R., Frost, J.K. and Huff, W.D. (1986) K-bentonites

of the Ordovician Decorah Subgroup, upper Mississippi

Valley: correlation by chemical fingerprinting. IllinoisState Geol. Surv. Circular, 537, 30.

Kolata, D.R., Frost, J.K. and Huff, W.D. (1987) Chemical

correlation of K-bentonite bed in the Middle Ordovician

Decorah Subgroup, upper Mississippi Valley. Geology, 15,208–211.

Kolata, D.R., Huff, W.D. and Bergstr€om, S.M. (1996)

Ordovician K-bentonites of eastern North America. Geol.Soc. Am. Spec. Pap., 313, 84.

Kolata, D.R., Huff, W.D. and Bergstr€om, S.M. (1998) Nature

and regional significance of unconformities associated

with the Middle Ordovician Hagan K-bentonite complex

in the North American midcontinent. Geol. Soc. Am. Bull.,

110, 723–739.



Kolata, D.R., Huff, W.D. and Bergstr€om, S.M. (2001) The

Ordovician Sebree Trough: an oceanic passage to the

Midcontinent United States. Geol. Soc. Am. Bull., 113,1067–1078.

Kump, L. R. and Arthur, M.A. (1999) Interpreting carbon-

isotope excursions: carbonates and organic matter. Chem.Geol., 161, 181–198.

Leslie, S.A. (2000) Mohawkian (Upper Ordovician)

conodonts of eastern North America and Baltoscandia. J.

Paleontol., 74, 1122–1147.Leslie, S.A. and Bergstr€om, S.M. (1997) Use of K-bentonite

beds as time-planes for high-resolution lithofacies

analysis and assessment of net rock accumulation rate:

an example from the upper Middle Ordovician of

eastern North America. In: Paleozoic Sequence

Stratigraphy, Biostratigraphy, and Biogeography: Studies

in Honor of J. Granville (“Jess”) Johnson (Eds G.

Klapper, M.A. Murphy, and J.A. Talent), Geol. Soc. Am.Spec. Pap., 321, 11–21.

Lohmann, K.C. (1988) Geochemical patterns of meteoric

diagenetic systems and their application to studies of

paleokarst. In: Paleokarst (Eds N.P. James and P.W.

Choquette), pp. 416. Springer-Verlag, New York City.

Ludvigson, G.A., Jacobson, S.R., Witzke, B.J. and Gonz�alez,L.A. (1996) Carbonate component chemostratigraphy and

depositional history of the Ordovician Decorah Formation,

Upper Mississippi Valley. In: Paleozoic Sequence

Stratigraphy; Views From the North American Craton (Eds

B.J. Witzke, G.A. Ludvigson and J. Day), Geol. Soc. Am.Spec. Pap., 306, 67–86.

Ludvigson, G.A., Witzke, B.J., Schneider, C.L., Smith, E.A.,Emerson, N.R., Carpenter, S.J. and Gonz�alez, L.A. (2000)A profile of the mid-Caradoc (Ordovician) carbon isotope

excursion at the McGregor Quarry, Clayton County, Iowa.

Geol. Soc. Iowa Guidebook, 70, 25–31.Ludvigson, G.A., Witzke, B.J., Gonz�alez, L.A., Carpenter,

S.J., Schneider, C.L. and Hasiuk, F. (2004) Late

Ordovician (Turinian-Chatfieldian) carbon isotope

excursions and their stratigraphic and paleoceanographic

significance. Palaeogeogr. Palaeoclimatol. Palaeoecol.,210, 187–214.

Martin, J.A., Knight, R.D. and Hayes, W.C. (1961)

Ordovician System. In: The Stratigraphic Succession inMissouri (Ed. J.W. Koenig), Missouri Geol. Surv. Water

Resour., 40, 32–35.Martma, T. (2005) Ordovician carbon isotopes. In: Estonian

Geological Sections, Kerguta (564) Drill Core (Ed. A.

P~oldvere), Geol. Surv. Estonia Bull., 7, 25–30.McCracken, M.H. (1966) Major structural features of

Missouri [Map]. Missouri Department of Natural

Resources: Division of Geology and Land Survey, Rolla.

http://www.dnr.mo.gov/geology/adm/publications/map-

MajorStrucFeatures.pdf

McLaughlin, P.I. and Brett, C.E. (2007) Signatures of sea-

level rise on the carbonate margin of a Late Ordovician

foreland basin: a case study from the Cincinnati Arch,

USA. Palaios, 22, 245–267.Missouri Department of Natural Resources (2009)

Generalized Geology Map of Missouri [Map]. Missouri

Department of Natural Resources: Division of Geology and

Land Survey, Rolla. www.dnr.mo.gov/geology.

Mitchell, C.E., Adhya, S., Bergstr€om, S.M., Joy, M.P. and

Delano, J.W. (2004) Discovery of the Ordovician Millbrig

K-bentonite Bed in the Trenton Group of New York State:

implications for regional correlation and sequence

stratigraphy in eastern North America. Palaeogeogr.

Palaeoclimatol. Palaeoecol., 210, 331–346.Munnecke, A., Calner, M., Harper, D.A.T. and Servais, T.

(2010) Ordovician and Silurian sea-water chemistry, sea

level, and climate: a synopsis. Palaeogeogr.

Palaeoclimatol. Palaeoecol., 296, 389–413.Panchuk, K.M., Holmden, C. and Kump, L.R. (2005)

Sensitivity of the epeiric sea carbon isotope record to

local-scale carbon cycle processes: tales from the

Mohawkian Sea. Palaeogeogr. Palaeoclimatol. Palaeoecol.,228, 320–337.

Panchuk, K.M., Holmden, C.E. and Leslie, S.A. (2006) Localcontrols on carbon cycling in the Ordovician midcontinent

region of North America, with implications for carbon

isotope secular curves. J. Sed. Res., 76, 200–211.Pancost, R.D., Freeman, K.H. and Patzkowsky, M.E. (1999)

Organic-matter source variation and the expression of a

late Middle Ordovician carbon isotope excursion. Geology,27, 1015–1018.

Patterson, W.P. and Walter, L.M. (1994) Depletion of 13C in

seawater ΣCO2 on modern carbonate platforms:

significance for the carbon isotopic record of carbonates.

Geology, 22, 885–888.Patzkowsky, M.E., Slupik, L.M., Arthur, M.A., Pancost, R.D.

and Freeman, K.H. (1997) Late Middle Ordovician

environmental change and extinction: Harbinger of the

Late Ordovician or continuation of Cambrian patterns?

Geology, 25, 911–914.Railsback, L.B., Holland, S.M., Hunter, D.M., Jordan, E.M.,

D�ıaz, J.R. and Crowe, D.E. (2003) Controls on geochemical

expression of subaerial exposure in Ordovician limestones

from the Nashville Dome, Tennessee, U.S.A. J. Sed. Res.,73, 790–805.

Rodgers, J. (1971) The Taconic Orogeny. Geol. Soc. Am.

Bull., 82, 1141–1178.Scotese, C.R. and McKerrow, W.S. (Eds.) (1990) Revised

world maps and introduction. In: Paleozoic Paleogeography

and Biogeography, Geol. Soc. London Mem., 12, 1–21.Sell, B.K. and Samson, S.D. (2011) Apatite phenocryst

compositions demonstrate a miscorrelation between the

Millbrig and Kinnekulle K-bentonites of North America

and Scandinavia. Geology, 39, 303–306.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

of Ordovician brachiopods: a major isotopic event around

the Middle-Late Ordovician transition. Geochim.

Cosmochim. Acta, 67, 2005–2025.Simo, J.A., Emerson, N.R., Byers, C.W. and Ludvigson, G.A.

(2003) Anatomy of an embayment in an Ordovician

epeiric sea, Upper Mississippi Valley, USA. Geology, 31,545–548.

Swart, P.K. (2008) Global synchronous changes in the carbon

isotopic composition of carbonate sediment unrelated to

changes in the global carbon cycle. Proc. Natl Acad. Sci.

USA, 105, 13741–13745.Templeton, J.S. and Willman, H.B. (1963) Champlainian

Series (Middle Ordovician) in Illinois. Illinois State Geol.

Surv. Bull., 89, 260.Thompson, T.L. (1991) Paleozoic Successions in Missouri

Part II: Ordovician System. Missouri Geol. Surv. Rep.

Invest., 70, 110–213.Tucker, R.D. (1992) U-Pb dating of plinian-eruption ashfalls

by the isotope dilution method: a reliable and precise tool

for time-scale calibration and biostratigraphic correlation.

Geol. Soc. Am. Meeting, 24, A198.



Tucker, R.D. and McKerrow, W.S. (1995) Early Paleozoic

chronology: a review in light of new U-Pb zircon ages

from Newfoundland and Britain. Can. J. Earth Sci., 32,368–379.

Ulrch, E.O. (1904) The quarrying industry of Missouri. In:

Missouri Bureau of Geology and Mines Report Vol. II, 2ndseries (Eds E.R. Buckley and H.A. Buehler), 2, pp. 109–111. Missouri Bureau of Geology and Mines, Jefferson

City, MO.

Veizer, J., Bruckschen, P., Pawellek, F., Diener, A., Podlaha,O.G., Carden, G.A.F., Jasper, T., Korte, C., Strauss, H.,Azmy, K. and Ala, D. (1997) Oxygen isotope evolution of

Phanerozoic seawater. Palaeogeogr. Palaeoclimatol.

Palaeoecol., 132, 159–172.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 d18Oevolution of Phanerozoic seawater. Chem. Geol., 161, 59–88.

Witzke, B.J. and Kolata, D.R. (1988) Changing structural and

depositional patterns, Ordovician Champlainian and

Cincinnatian Series of Iowa-Illinois. Iowa Dept. Nat.

Resour. Guidebook, 8, 55–77.Young, S.A., Saltzman, M.R. and Bergstr€om, S.M. (2005)

Upper Ordovician (Mohawkian) carbon isotope (d13C)stratigraphy in eastern and central North America:

regional expression of a perturbation of the global carbon

cycle. Palaeogeogr. Palaeoclimatol. Palaeoecol., 222, 53–76.

Young, S.A., Saltzman, M.R., Bergstrom, S.M., Leslie, S.A.and Xu, C. (2008) Paired d13Ccarb and d13Corg records of

Upper Ordovician (Sandbian-Katian) carbonates in North

America and China: implications for paleoceanographic

change. Palaeogeogr. Palaeoclimatol. Palaeoecol., 270, 166–178.

Manuscript received 19 April 2012; revision accepted16 January 2013

Supporting Information

Additional Supporting Information may be found inthe online version of this article:

Table S1. Geochemical data for location New Lon-don.Table S2. Geochemical data for location Highway

MM. See Table S1 for column header definitions andstratigraphic abbreviations.Table S3. d18Ocarb filter table.



techniques for assessing spatial heterogeneity of carbonate

Documents