Ammonite habitat revealed via isotopic compositionand comparisons with co-occurring benthic andplanktonic organismsJocelyn Anne Sessaa,1, Ekaterina Larinaa,b,c, Katja Knolla,b, Matthew Garbb, J. Kirk Cochrand, Brian T. Hubere,Kenneth G. MacLeodf, and Neil H. Landmana

aDivision of Paleontology, American Museum of Natural History, New York, NY 10024; bDepartment of Earth and Environmental Sciences, Brooklyn College,Brooklyn, NY 11210; cDepartment of Earth Sciences, University of Southern California, Los Angeles, CA 90018; dSchool of Marine and AtmosphericSciences, Stony Brook University, Stony Brook, NY 11794; eDepartment of Paleobiology, National Museum of Natural History, Washington, DC 20013; andfDepartment of Geological Sciences, University of Missouri, Columbia, MO 65211

Edited by R. Mark Leckie, University of Massachusetts, Amherst, MA, and accepted by the Editorial Board October 9, 2015 (received for review April 23, 2015)

Ammonites are among the best-known fossils of the Phanerozoic, yettheir habitat is poorly understood. Three common ammonite families(Baculitidae, Scaphitidae, and Sphenodiscidae) co-occur with well-preserved planktonic and benthic organisms at the type locality ofthe upper Maastrichtian Owl Creek Formation, offering an excellentopportunity to constrain their depth habitats through isotopiccomparisons among taxa. Based on sedimentary evidence and themicro- and macrofauna at this site, we infer that the 9-m-thicksequence was deposited at a paleodepth of 70–150 m. Taxa presentthroughout the sequence include a diverse assemblage of ammon-ites, bivalves, and gastropods, abundant benthic foraminifera, andrare planktonic foraminifera. No stratigraphic trends are observedin the isotopic data of any taxon, and thus all of the data from eachtaxon are considered as replicates. Oxygen isotope-based tempera-ture estimates from the baculites and scaphites overlap with those ofthe benthos and are distinct from those of the plankton. In contrast,sphenodiscid temperature estimates span a range that includes esti-mates of the planktonic foraminifera and of the warmer half of thebenthic values. These results suggest baculites and scaphites livedclose to the seafloor, whereas sphenodiscids sometimes inhabitedthe upper water column and/or lived closer to shore. In fact, the rarityand poorer preservation of the sphenodiscids relative to the baculitesand scaphites suggests that the sphenodiscid shells may have onlyreached the Owl Creek locality by drifting seaward after death.

paleoecology | mollusk | Late Maastrichtian | ammonite |habitat reconstruction

Ammonites have constituted a primary data source for thefields of evolution, paleoceanography, biostratigraphy, and

paleoecology for more than a century; their ubiquity, diversity,occurrence in a wide variety of marine environments, and readilypreservable shell account for their utility in both paleontologicaland geological studies. Ammonites have been used extensively instudies of heterochrony because their shells preserve distinct on-togenetic changes that can be tracked in evolving lineages (1, 2);they are valued in paleoceanographic research because, like mostmollusks, they are inferred to have precipitated their aragoniticshells in isotopic equilibrium with the surrounding seawater (3, 4).Thus, shell chemistry may record temperature, via oxygen isotopes(δ18O) (5), and water mass properties, such as strontium isotopes(87Sr/86Sr), which are used to estimate numerical age (6). Am-monites are also a textbook example of an index fossil; besidesbeing abundant and widespread, they evolved rapidly, making themthe dominant Mesozoic tool for relative dating and correlation ofshallow water strata. For example, the 35-My-long stratigraphicrecord of Upper Cretaceous deposits in the US Western InteriorSeaway (WIS) has been partitioned into 66 ammonite zones (7).Finally, ammonites underwent a spectacular extinction at the closeof the Mesozoic. Explanations for why the ammonites, which wereflourishing immediately before the Cretaceous–Paleogene (K–Pg)

mass extinction (8), died out, whereas their relatives the nautiloidssurvived (9), have been used to understand the selectivity of marinemicrofossil groups across the K–Pg event (10), highlighting theimportance of ammonites in understanding extinction mechanisms.Despite the utility of ammonites to many disciplines, their ecol-

ogy remains poorly known. A challenge in reconstructing theirhabitat(s) is establishing if ammonites lived at the site from whichthey are recovered. Ammonite tissues could drop out after death,and the shell might float to the surface buoyed by relict air con-tained within the phragmocone (11). Empty shells of Nautilus arefound on beaches at remote distances from their actual habitat,documenting the potential for postmortem drift of positivelybuoyant shells (12). Similarly, Tanabe (13) mapped the distributionof Turonian ammonites along an onshore–offshore transect, andnoted that their postmortem distribution was broader than thesettings they inhabited during life. Uncertainty in ammonites’ pre-ferred habitat is especially concerning for temperature reconstruc-tions based on their occurrence or isotopes because temperaturevaries both with depth and with distance to the shoreline.A variety of studies have attempted to determine the ecology

of ammonites based on analogies with living relatives, shellmorphology, facies distribution, faunal associations, and isotopiccomposition. However, these studies have had limited success for

Significance

Because ammonites are one of the most diverse, abundant, andwell-preserved clades in the history of life, they are a mainstay inmacroevolutionary and biodiversity studies; however, their ecol-ogies are poorly understood, and it is unknown whether taxalived near the sea surface or seafloor. This uncertainty under-mines their use in paleoecological and paleoenvironmental re-constructions, which depend on knowledge of organisms’ depthpreferences. Here, we use a rare co-occurrence of exquisitely well-preserved ammonites and planktonic and benthic organisms toconstrain depth preferences of three common ammonite familiesby comparing the oxygen and carbon isotopic signatures of thesetaxa. The ammonites fall into two distinct depth habitats, en-hancing the utility of these families for highly refined paleoeco-logical and paleoclimatic studies.

Author contributions: J.A.S., E.L., and N.H.L. designed research; J.A.S., E.L., K.K., M.G., B.T.H.,K.G.M., and N.H.L. performed research; J.A.S., K.K., J.K.C., B.T.H., K.G.M., and N.H.L. analyzeddata; and J.A.S., J.K.C., B.T.H., K.G.M., and N.H.L. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. R.M.L. is a guest editor invited by the EditorialBoard.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. Email: jsessa@amnh.org.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1507554112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1507554112 PNAS Early Edition | 1 of 6

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

SEC

OLO

GY

both biological and geological reasons. Ammonites are extinct, andtheir closest living relatives, the octopods, squids, cuttlefish, andNautilus, are all in different orders/subclasses (14). Even amongliving cephalopods, a variety of behaviors are observed, includingvertical and lateral migrations (15). Other studies have sought toreconstruct ammonite habitat by comparing the isotopes recordedin their shells to those of co-occurring, or nearly co-occurring, taxaof known depth habitats. A powerful approach in theory, thesestudies have been limited in practice because ammonites are onlyrarely recovered with the planktonic and benthic organisms neededto establish a temperature-depth profile. Further, many studieswere undertaken in the WIS, where the water mass properties arepoorly understood and controversial (16, 17). We expand uponprevious isotopic studies by using exceptionally well-preservedammonites from the Owl Creek Formation (fm.) type locality innorthern Mississippi (Fig. S1). Ammonites are abundant at thissite and co-occur with bivalves, gastropods, and planktonic andbenthic foraminifera (Fig. 1 and Fig. S2), thus providing anexcellent opportunity to reconstruct a water column profile andestablish where the ammonites fall within it.

Results and DiscussionA total of 553 mollusk specimens were scored for taphonomicfeatures (SI Methods); of these, 405 specimens were evaluatedfor isotopic analysis: 196 were found to be well preserved, and

subsequent analyses resulted in 234 isotopic measurements(Table 1; SI Methods). Well-preserved foraminifera (116 plank-tonic and 72 benthic foraminifera) were picked and resulted inisotopic measurements for 11 planktonic and 14 benthic separates(Table 1); an additional 398 planktonic and 953 benthic forami-nifera were counted to constrain depth estimates (Table S1).

Determining Whether Ammonites Experienced Postmortem Drift.Wholespecimens of baculites, scaphites, bivalves, and gastropods arecommon and display low degrees of fragmentation, implying thatthese groups experienced similar taphonomic histories (SI Methods).The sphenodiscids, however, are highly fragmented; only 1 of the11 collected specimens was more than 50% complete. Epizoans arerare, occurring on less than 3% of specimens in each mollusk groupexcept for the sphenodiscids, where 27% (3 of 11 specimens) bearserpulid worm tubes or encrusting bryozoans. In both baculites andscaphites, delicate features, such as tubercles and the apertural mar-gin, are frequently observed (Fig. 1). Tubercles on some sphenodiscidshells are worn, and the apertural margin is always missing.Adult scaphites can be differentiated from juveniles by shell shape

and size, and adult males and females can be identified (18) (Fig. 1).For specimens where sex can be determined, 44% are males, 52%are females, and 4% are juveniles. These proportions are repre-sentative of a living community (19) and not of an egg-laying habitattransiently occupied by females (20). Predation indicators, such as

Fig. 1. Representative mollusks from the Owl Creek fm. (A) Left lateral and ventral views of a Discoscaphites iris macroconch AMNH 91329; (B) right lateraland ventral views of a D. irismicroconch AMNH 91335; (C) right lateral view of D. iris, showing a healed injury AMNH 77461; (D) ventral and right lateral viewsof Eubaculites latecarinatus AMNH 91330; (E) ventral and right lateral views of Eubaculites carinatus AMNH 91334; (F) left lateral view of Sphenodiscuspleurisepta; sutures are visible because most of the shell is missing AMNH 91520; (G) Gyrodes crenata AMNH 91333; (H) Nucula percrassa AMNH 91331.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1507554112 Sessa et al.

healed injuries, are present on 5% of scaphite specimens (Fig. 1); asimilar incidence of predation as observed in other scaphite accu-mulations interpreted as in situ (21). Although it is more challengingto determine sex in baculites than in scaphites, for those baculitespecimens where this assignment could be made, the proportion ofmales and females is equal. In contrast, little can be inferred aboutthe sphenodiscid’s population structure; only adults were recovered,and none could be assigned a sex because they were too incomplete.Sphendosicids are also an outlier in terms of abundance. Al-

though we did not count specimens from bulk samples, the numberof specimens in Table 1 reflects the general ammonite abundance.Baculites and scaphites occur at roughly equal frequency, andsphenodiscids are rare. Note that the relatively low numbers of bi-valves and gastropods in Table 1 reflects our bias toward collectingammonites. In a synthesis of ammonites from the Owl Creek fm.,Kennedy and Cobban (22) report relative abundances that areconsistent with our findings. The authors (22) studied four speci-mens of Sphenodiscus pleurisepta, noting that all were crushed tovarying degrees, one of which exhibits abraded tubercles (SI Meth-ods), 80 Eubaculites carinatus specimens, and 140 Discoscaphites irisspecimens. In summary, preservational features and populationcharacteristics suggest that, like the bivalves and gastropods, thebaculites and scaphites experienced little postmortem transport;they were living at or very near the studied site at the time of death.In stark contrast, the rarity, poor preservation, and recovery of onlyadult sphenodiscids indicate that these specimens could have ex-perienced postmortem drift.

Environmental Reconstruction. In a study of modern foraminiferaalong the North American Atlantic coastal margin, Gibson (23)found that the ratio of planktonic to benthic foraminifera is stronglycorrelated with depth, and this study has formed the basis for anextensive body of work estimating depth in modern and ancientsettings via these proportions. The proportion of benthic forami-nifera in the Owl Creek fm. ranges between 61% and 77% (TableS1). In the Gibson (23) dataset, the minimum recorded depth forassemblages with <80% benthics is 70 m, and for those with <50%benthics is 100 m. Assemblages with these proportions were largelyfound between 100 and 200 m, but some were recovered as deep as1,000 m (23). An Owl Creek paleodepth of greater than 200 m is

highly unlikely. The Owl Creek fm. is composed of dark grayglauconitic micaceous clayey silts to very fine quartz sands, inter-preted to represent a fully marine prodelta shelf (24, 25) thatinterfingers with shallow water chalks and sands (26). The absenceof sedimentary structures, with the exception of two beds withcentimeter-scale parallel laminations (at 5.2–5.4 m and 6–7 m),suggest extensive bioturbation (SI Methods).Comparison with the upper Maastrichtian deposits from Brazos

River (Rv.), TX, suggests that the Owl Creek section was shal-lower. Benthics comprise between 10% and 25% of the BrazosRv. foraminiferal assemblage (27), which translates to estimates∼150–250 m in the Gibson (23) dataset. Ashckenazi-Polivodaet al. (28), using the same taxa (Gavelinella sp., Globoheterohelixglobulosa, and Rugoglobigerina rugosa) as in our study, reportbenthic foraminifera temperatures that are ∼2.5 °C cooler, and adifference between planktonic and benthic values that is ∼2 °Cgreater, at Brazos Rv. than at the Owl Creek site. If these faunaland geochemical differences reflect changes in depth, then theOwl Creek site was shallower than Brazos Rv. The macrofauna atthese sites corroborates this depth relationship. Sessa et al. (29)categorized late Mesozoic Gulf Coastal Plain benthic molluskfaunas as shallow subtidal or offshore. The Owl Creek taxa andtheir abundance are transitional between these two settings,with components of both shallow subtidal assemblages (naticids,veneroids, crassatellids, and turritellids) and those of the off-shore (ostreoids and pectinoids), whereas the Brazos Rv. faunacomprises most of the offshore samples in Sessa et al. (29). Con-sidered as a whole, a conservative estimate of the paleodepthof the Owl Creek fm. is 70–150 m, with 100 m being likely.The δ13C data are consistent with this conclusion, as dis-cussed below.

Interpreting Oxygen Isotopes and Paleotemperatures.No stratigraphictrends are observed in the isotopic data for any taxonomic group(Fig. 2), and, therefore, all of the data from each taxon are con-sidered as replicate samples. The analyzed planktonic foraminiferaspecies, Planoheterohelix globulosa and Rugoglobigerina rugosa,display similar oxygen isotopic values (Table 1), yielding similartemperature estimates (∼26 °C; Fig. 2). These taxa have beeninterpreted as surface-to-subsurface mixed layer and/or near shore

Table 1. Total number of well-preserved specimens of each taxon, number of measurements made from these specimens, averagecarbon and oxygen isotopic composition and temperature, and temperature range

Aragonitic taxa Calcitic taxa

AmmonitesInfaunalbivalves Gastros

Epifaunaloysters

Benthic foraminifera Planktonic foraminifera

Baculites Scaphites Sphenos Lingulogav. sp. Gavelinella sp. P. globulosa R. rugosa

Totalno. specimens

65 67 7 22 32 3 33 39 84 32

Totalno. measurements

89 78 8 22 33 4 5 9 6 4

Mean δ13C −1.4 0.4 −3.9 1.6 1.7 1.8 0.8 0.9 1.6 1.8Mean δ18O −0.6 −0.7 −1.6 −1.0 −0.8 −1.7 −1.9 −1.8 −3.2 −3.3Mean

temperature, °C18.1 18.4 22.1 19.5 18.8 18.9 19.8 19.8 26 26.4

Benthos PlanktonMean

temperature, °C18.1 18.4 22.1 19.1 26.2

Range intemperature, °C

8.4 6 9 4.4 —

For the mollusks, multiple measurements were made on some specimens. For the foraminifera, individual specimens were combined to achieve the weightneeded for isotopic analysis. Temperatures are calculated using the equations given in SI Methods. The range in temperature is calculated as the differencebetween the means of the 10%warmest and 10% coolest values of each taxon, except for Sphenodiscus, where the maximum and minimum values were usedbecause of the small number of measurements. Benthos includes infaunal bivalves, gastropods, epifaunal oysters, and benthic foraminifera. Sphenos,Sphenodiscus; Gastros, Gastropods; Lingulogav., Lingulogavelinella; P. globulosa, Planoheterohelix globulosa; R. rugosa, Rugoglobigerina rugosa.

Sessa et al. PNAS Early Edition | 3 of 6

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

SEC

OLO

GY

dwellers, and, though they can exhibit distinct isotopic signaturesin open ocean settings, the variability in the inferred ecology ofP. globulosa is considerable (28, 30). For the Owl Creek specimens,isotopic overlap likely results from the relatively shallow setting,with both taxa living in a well-mixed portion of the upperwater column.

Both benthic foraminifera, Lingulogavelinella sp. and Gavelinellasp., yield temperature estimates of ∼20 °C (Fig. 2). Based onanalogy with modern gavelinellid taxa, these taxa probably lived atthe sediment-water interface or within the first few centimeters ofthe sediment in well-oxygenated environments (31, 32). The calciticoysters were also epifaunal, and display δ18O values that arecomparable to those of the benthic foraminifera (Table 1 and Fig.2). The gastropods and infaunal bivalves have aragonitic shells,and their δ18O values are similar to one another (Table 1).Temperature estimates from all benthic taxa are grouped around

19 °C (Fig. 2); they display a normal distribution (skew of −0.0004),suggesting that these data accurately capture seasonal temperaturefluctuations at the seafloor. The 7 °C temperature differencebetween the planktonic foraminifera (26 °C) and the benthictaxa (19 °C) may represent a seasonal maximum of the surface-to-bottom temperature gradient. Because modern planktonic fora-minifera usually have spring or summer peaks in abundance (33),the paleotemperature values reconstructed here may representsurficial water temperatures of warmer seasons, and the gradientcould have been reduced during winter months. Because thebenthic data are never as warm as planktonic values, however,finding ammonites with warm temperatures would provide strongevidence that they inhabited surface waters.Mean temperature estimates from the baculites and scaphites

are statistically indistinguishable (Kolmogorov–Smirnov test, P =0.3936), align with benthic estimates, and never approach plank-tonic estimates (Fig. 2). Baculites and scaphites also display largervariances than that of the benthos (Fligner–Killeen test of homo-geneity of variances P = 0.00004 for baculites vs. benthos; P valueof 0.003 for scaphites vs. benthos) and have asymmetric distribu-tions skewed toward cooler values (baculite skew, −0.4; scaphiteskew, −0.3; Fig. 2 and Table 1). The wider distribution may haveresulted from several interrelated causes; living above the seafloor,the ammonites may have experienced wider temperature fluctua-tions than the benthic taxa, and/or they may have had highermetabolic rates than the benthic taxa, resulting in faster growthrates and thus recording a less time-averaged temperature signalthan that of the sedentary benthic taxa. The skew toward coolvalues may result from preferential shell growth during the wintermonths, or from migration to/from deeper (and cooler) waterseasonally or through ontogeny. However, investigations of themorphology (muscle scars, shell shape) and distribution of LateCretaceous scaphites indicate that these animals were limited intheir mobility and may have remained at a single site for an ex-tended period, subject to current activity (34, 35, and referencestherein). These possibilities could be explored in future work byserially sampling the baculites and scaphites throughout ontogeny.Regardless of the relative importance of these alternatives, theconsistently high δ18O values among baculites and scaphites suggestthat they lived much closer to the seafloor than the sea surface.Temperature estimates from the sphenodiscids encompass a

broad range (9 °C). The warmest estimates for the sphenodiscidsoverlap with planktonic foraminiferal estimates, and their cool-est values span the warmer half of the benthic (and baculite andscaphite) estimates (Fig. 2 and Table 1). Though the smallnumber of measurements made from their shells prohibits sta-tistical comparisons, it seems apparent that the sphenodiscidsspent at least some portion of their lives in waters warmer thanthose where baculites and scaphites lived.

Interpreting Carbon Isotopes.Calcitic foraminifera generally secretetheir tests close to isotopic equilibrium with the dissolved inorganiccarbon (DIC) reservoir (32). The δ13C values of the planktonic(1.7‰) and benthic (0.9‰) foraminifera (Table 1) are consistentwith expected δ13CDIC depth gradients due to photosyntheticfractionation in the upper water column and remineralization ofsinking organic matter at depth. A 0.8‰ δ13CDIC depth variationis found in modern settings of ∼100 m depth (36).

10

15

20

25

30

Tem

p (°C

)Planktonic Foraminifera (c)Benthic Foraminifera (c)Epifaunal Bivalves (c)Infaunal Bivalves

GastropodsSphenodiscidaeBaculitidaeScaphitidae

0

-5

-4

-3

-2

-1

0

1

2

3

4

-60 8 00 80 88

δ13C

meters8

A

B

Fig. 2. (A) Temperature estimates and (B) carbon isotopic composition for alltaxa arranged by meter level. “(c)” indicates those taxa that secrete a calcitic shellor test; all other taxa secrete aragonitic shells. The stratigraphic scale is plottedalong the horizontal axis (0–8 m) and is repeated for each taxonomic group. Nostratigraphic trends are apparent for any taxon; the horizontal axis of thesphenodiscids is therefore compressed for visual ease. The baculites and scaphiteshave similar temperatures as the benthic taxa, whereas the sphenodiscids en-compass both planktonic temperatures and the warmer portion of the benthictemperature distribution. The δ13C values of the planktonic foraminifera and ofthe benthic foraminifera are consistent with the variation in δ13CDIC expectedwith depth. The δ13C values of the ammonites likely reflect physiological pro-cesses, and the consistent offset among the baculites, scaphites, and spheno-discids suggests differences in diet and/or lateral and depth differences in habitat.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1507554112 Sessa et al.

The δ13C of modern mollusk shells can reflect both the DICreservoir and metabolic pathways. Thus, environmental signalsare often overprinted by physiological processes that can changeseasonally and through ontogeny (37). Consistent with this pos-sibility, the scaphites, baculites, and benthic mollusks displayoffsets in their δ13C values relative to the δ13CDIC inferred fromanalyses of benthic foraminifera. On average, the gastropods andbivalves have δ13C values that are ∼1‰ higher than the inferredδ13CDIC (Table 1), as has been found in some modern taxa (38).The ammonites have δ13C values lower than the other taxa, withthe baculites intermediate between the scaphites and spheno-discids, suggesting differences in diet and/or position within thewater column. Tobin and Ward (39) also documented loweraverage δ13C values in Late Cretaceous ammonites relative tobenthic mollusks; however, these differences may be due tomethane-derived carbon incorporated into the shells (40). Thewarm temperature recorded by the sphenodiscids, combined withlow δ13C, supports the notion that these organisms generally livedclose to shore; if they lived in the surface waters of the study site,they would have been exposed to a higher δ13CDIC, as indicated bythe planktonic foraminifera δ13C. Thus, the marked δ13C differ-ences between sphenodiscids and planktonic foraminifera reinforcethe conclusion of offshore postmortem transport of sphenodiscidsbased on taphonomic observations.

Depth Habitats of Adult Baculites, Scaphites, and Sphenodiscids. TheOwl Creek fm. results match previous studies of baculites, sca-phites, and sphenodiscids well. Tsujita and Westermann (41)inferred a lower to middle water habitat for WIS baculites.Similarly, Henderson and Price (42) interpreted the baculiteSciponoceras as demersal because its δ18O values aligned withthose from benthic mollusks of the same formation. Tanabe (13)suggested that scaphites did not occupy nearshore or offshoresettings, but rather lived in an intermediate depth zone, as we inferfor the Owl Creek fm. Landman et al. (8) compiled the globalgeographic and facies distributions of Maastrichtian ammonites;they noted that sphenodiscids were restricted to nearshore facies(see also ref. 43), which strengthens our hypothesis that the OwlCreek fm. sphenodiscids lived in a more landward setting andfloated out to the site after death. Moriya et al. (44) comparedthe isotopic composition of ammonites with that of bivalves,gastropods, and planktonic and benthic foraminifera from bothmudstones (the microfossils) and calcareous concretions (themacrofossils) from throughout the 30-m-thick Yezo Group. Ademersal habitat was suggested for the Ancyloceratina, thesuborder that contains the baculites and scaphites, becausetheir isotopic values were similar to those of the benthos.Our comparison of the occurrence, preservation, and stable

isotopic composition of ammonites with that of coexisting ben-thic and planktonic organisms shows vertical and lateral habitatseparation between the Sphenodiscidae, and the Baculitidae andScaphitidae. The baculites and scaphites had similar taphonomichistories as the bivalves and gastropods, which did not experi-ence postmortem drift, indicating that these ammonites likelyinhabited the study site. Further, δ18O-based temperature esti-mates for the baculites and scaphites completely overlap withthose of the benthos, but are distinct from those of the plankton,indicating that they were demersal. The sphenodiscids, however,are rare and have poorer preservation relative to all of the othermollusks, implying that the sphenodiscids were transported tothe study site after death. These features, combined with oftenlow δ18O values (that is, warm, surface water-like tempera-tures), broad range in δ18O values, and low δ13C values supportthe interpretation that these sphenodiscids sometimes lived innearshore, perhaps even brackish, waters for some portion oftheir life.This refined model of ammonite habitat sets the stage for highly

detailed coastal-water reconstruction, whereby the sphenodiscids’

isotopic compositions likely reflect nearshore waters, and that ofthe baculites and scaphites represents near bottom, marine condi-tions of the areas in which they are preserved. Combining this im-proved ecological information with data from planktonic and benthicforaminifera could be a powerful tool for paleoceanographic andclimate modeling studies. Related is the possibility that the greaterrange of temperatures recorded in the baculites and scaphitescompared with the benthic mollusks is due to the ammonites pre-serving a less time-averaged record, again highlighting the strength ofusing ammonite shells as an archive of temperature variation. Forexample, in contrast to the well-resolved deep marine climate re-cords from middle to high paleolatitudes near the K–Pg boundary,temperature reconstructions from shallow marine settings at lowerpaleolatitudes are rare. Using ammonites from shelf sections like theOwl Creek fm. site would enhance our understanding of meridionalclimatic conditions just before this event.

MethodsGeologic Setting and Age. The study section consists of the upper 9 m of theOwl Creek fm. and 2 m of the overlying Danian Clayton fm. (SI Methods). Theregion is tectonically undeformed and was never deeply buried, resulting inunlithified sediments containing fossils that were not thermally altered (25).The Owl Creek fm. was assigned to calcareous nannofossil zone CC26b (45),which provides a conservative estimate that deposition occurred within thelast 1.3 My of the Cretaceous (46).

Specimen Collection. Fossils were collected throughout the lower 8 m of theOwl Creek fm. (Table 1; SI Methods). Five bulk samples for microfossil studywere collected between 0.5 and 7.5 m above the base of the outcrop (Table1). From these samples, foraminifera were isolated and concentrated usingstandard techniques (SI Methods), counted to determine planktonic:benthicratios, and picked for isotopic analyses.

Taphonomy. We tabulated the degree of completeness of each ammonitespecimen, noting delicate, easily broken features such as the phragmoconeand tubercles, the presence of epizoans, the proportion of macroconchs tomicroconchs (presumed to be females and males, respectively) (18), the sizeand ontogenetic stage of specimens, and any features suggestive of pre-dation (SI Methods). Parallel observations were made for bivalve and gas-tropod specimens, which do not experience postmortem drift.

Preservation, Isotopic Analysis, and Paleotemperature Determination. Preserva-tion of foraminifera and mollusk shell microstructure was evaluated beforeisotopic analysis (Figs. S2 and S3 and SI Methods). Isotopic results arereported in standard δ-notation and on the Vienna-PDB (VPDB) scale. Fifty-one mollusk specimens (59 measurements; some specimens were sampledmultiple times) were analyzed at the University of South Florida (USF) on aDelta V Isotope Ratio Mass Spectrometer coupled to a Gasbench II auto-matic preparation system using standard techniques (SI Methods). Addi-tional analyses of mollusks (155 specimens with 175 measurements,including 10 specimens also analyzed by USF), and all 25 foraminiferalanalyses were made at the University of Missouri using a Kiel III carbonatedevice attached to a Finnigan DeltaPlus Isotope Ratio Mass Spectrometer(IRMS) using standard techniques (SI Methods). Because some taxa ana-lyzed are calcitic, whereas others are aragonitic, and calcite and aragonitehave different fractionation factors, δ18O values were converted to tempera-ture using well-established formulas for each respective organism (SI Meth-ods). Converting to temperature allows all taxa to be compared on thesame scale.

ACKNOWLEDGMENTS. We thank J. Slattery for a sphenodiscid specimen;G. Phillips and R. Rovelli for field assistance; property owners A. Carroll andB. Carroll; American Museum of Natural History (AMNH) Master of Arts inTeaching 2013 graduates J. DeCosta, L. Hlinka, K. Lapenta, S. McFadden, andA. Nesheim for data collection; S. Haynes, M. Hill, S. Mahmood, and H. Tobinfor laboratory assistance; Z. Atlas for isotopic analyses; M. Foote, M. Hopkins,L. Petruny, and M. Tessler for helpful suggestions; E. Thomas for informative dis-cussion; and editor R. Mark Leckie and two anonymous reviewers for construc-tive comments. Support for this work was provided by a Katherine DavisPostdoctoral Fellowship at the AMNH and National Science Foundation (NSF)Grant NSF-DR K-12:1119444; a Lerner–Gray Scholarship (AMNH), a Richard K.Bambach Scholarship from the Paleontological Society, a James Welsch Schol-arship from the Association of Applied Paleontological Sciences, and NSF-GRFPGrant 2013171808 (to E.L.); and the AMNH Norman Newell Fund.

Sessa et al. PNAS Early Edition | 5 of 6

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

SEC

OLO

GY

1. Korn D, Klug C (2007) Conch form analysis, variability, morphological disparity, andmode of life of the Frasnian (Late Devonian) ammonoid Manticoceras from Coumiac(Montagne Noire, France). Cephalopods Present and Past: New Insights and FreshPerspectives, eds Landman NH, Davis RA, Mapes RH (Springer, Dordrecht, The Neth-erlands), pp 57–85.

2. Yacobucci MM (2015) Macroevolution and paleobiogeography of Jurassic-Cretaceousammonoids. Ammonoid Paleobiology: From Macroevolution to Paleogeography,Topics in Geobiology, eds Klug C, Korn D, De Baets K, Kruta I, Mapes RH (Springer,Dordrecht, The Netherlands), Vol 44, pp 189–228.

3. Lukeneder A, Harzhauser M, Müllegger S, Piller WE (2010) Ontogeny and habitatchange in Mesozoic cephalopods revealed by stable isotopes (δ18O, δ13C). Earth PlanetSci Lett 296(1–2):103–114.

4. Stevens K, Mutterlose J, Wiedenroth K (2015) Stable isotope data (δ18O, δ13C) of theammonite genus Simbirskites: Implications for habitat reconstructions of extinctcephalopods. Palaeogeogr Palaeoclimatol Palaeoecol 417(0):164–175.

5. Grossman EL, Ku TL (1986) Oxygen and carbon isotope fractionation in biogenicaragonite: Temperature effects. Chem Geol 59(1):59–74.

6. McArthur JM, Kennedy WJ, Chen M, Thirlwall MF, Gale AS (1994) Strontium isotopestratigraphy for Late Cretaceous time: Direct numerical calibration of the Sr isotopecurve based on the US Western Interior. Palaeogeogr Palaeoclimatol Palaeoecol108(1-2):95–119.

7. Cobban WA, Walaszczyk I, Obradovich JD, McKinney KC (2006) A USGS Zonal Tablefor the Upper Cretaceous Middle Cenomanian-Maastrichtian of the Western Interiorof the United States Based on Ammonites, Inoceramids, and Radiometric Ages. USGeological Survey Open File Report 2006-1250. Available at pubs.usgs.gov/of/2006/1250/pdf/OF06-1250_508.pdf.

8. Landman NH, et al. (2014) Ammonite extinction and nautilid survival at the end of theCretaceous. Geology 42(8):707–710.

9. Wani R, Kurihara Ki, Ayyasami K (2011) Large hatchling size in Cretaceous nautiloidspersists across the end-Cretaceous mass extinction: New data of Hercoglossidaehatchlings. Cretac Res 32(5):618–622.

10. Alegret L, Thomas E, Lohmann KC (2012) End-Cretaceous marine mass extinction notcaused by productivity collapse. Proc Natl Acad Sci USA 109(3):728–732.

11. Chamberlain JA, Jr, Ward PD, Weaver JS (1981) Post-mortem ascent of Nautilus shells:Implications for cephalopod paleobiogeography. Paleobiology 7(4):494–509.

12. House M (1987) Geographic distribution of Nautilus shells. Nautilus: The Biology andPaleobiology of a Living Fossil, eds Saunders B, Landman NH (Plenum, New York), pp53–83.

13. Tanabe K (1979) Palaeoecological analysis of ammonoid assemblages in the TuronianScaphites facies of Hokkaido, Japan. J Paleontol 22:609–630.

14. Jacobs DK, Landman NH (1993) Nautilus—a poor model for the function and behaviorof ammonoids? Lethaia 26(2):101–111.

15. Boyle PR (1983) Cephalopod Life Cycles: Species Accounts (Academic, New York), Vol 1.16. Wright EK (1987) Stratification and paleocirculation of the Late Cretaceous Western

Interior Seaway of North America. Geol Soc Am Bull 99(4):480–490.17. Dennis KJ, Cochran JK, Landman NH, Schrag DP (2013) The climate of the Late Cre-

taceous: New insights from the application of the carbonate clumped isotope ther-mometer to Western Interior Seaway macrofossil. Earth Planet Sci Lett 362(0):51–65.

18. Davis RA, Landman NH, Dommergues J-L, Marchand D, Bucher H (1996) Maturemodifications and dimorphism in Ammonoid Cephalopods. Ammonoid Paleobiology,eds Landman NH, Tanabe K, Davis RA (Plenum, New York), Vol 13, pp 464–543.

19. Westermann GEG (1996) Ammonoid life habitat. Ammonoid Paleobiology, edsLandman NH, Tanabe K, Davis RA (Plenum, New York), pp 608–710.

20. Callomon JH (1969) Dimorphism in Jurassic ammonites. Some reflections. Sexual Di-morphism in Fossil Metazoa and Taxonomic Implications, ed Westermann GEG(E. Schweizerbart’sche Verlagsbuchhandlung, Stuttgart), Vol l, series A1, pp 111–125.

21. Takeda Y, Tanabe K, Sasaki T, Landman NH (2015) Durophagous predation on sca-phitid ammonoids in the Late Cretaceous Western Interior Seaway of North America.Lethaia, 10.1111/let.12130.

22. Kennedy WJ, CobbanWA (2000) Maastrichtian (Late Cretaceous) ammonites from theOwl Creek Formation in northeastern Mississippi, U.S.A. Acta Geol Pol 50(1):175–190.

23. Gibson TG (1989) Planktonic benthonic foraminiferal ratios: Modern patterns andTertiary applicability. Mar Micropaleontol 15(1–2):29–52.

24. Pryor WA (1960) Cretaceous sedimentation in Upper Mississippi Embayment. AAPGBull 44:1473–1504.

25. Mancini EA, Puckett TM, Tew BH (1996) Integrated biostratigraphic and sequencestratigraphic framework for Upper Cretaceous strata of the eastern Gulf CoastalPlain, USA. Cretac Res 17(6):645–669.

26. Swann CT (1997) Geology of the Chalybeate Topographic Quadrangle, Tippah andAlcorn Counties, Mississippi. J Miss Acad Sci 42:142–150.

27. Schulte P, Speijer RP, Hartmut M, Kontny A (2006) The Cretaceous-Paleogene (K-P)boundary at Brazos, Texas: Sequence stratigraphy, depositional events and theChicxulub impact. Sediment Geol 184:77–109.

28. Ashckenazi-Polivoda S, et al. (2013) Paleoecology of the K-Pg mass extinction survivorGuembelitria (Cushman): Isotopic evidence from pristine foraminifera from BrazosRiver, Texas (Maastrichtian). Paleobiology 40(1):24–33.

29. Sessa JA, Bralower TJ, Patzkowsky ME, Handley JC, Ivany LC (2012) Environmentaland biological controls on the diversification and ecological reorganization of Late

Cretaceous and early Paleogene marine ecosystems in the Gulf Coastal Plain.Paleobiology 38(2):218–239.

30. Haynes SJ, Huber BT, MacLeod KG (2015) Evolution and phylogeny of mid-Cretaceous(Albian-Coniacian) Biserial Planktic Foraminifera. J Foraminiferal Res 45:42–81.

31. Alegret L, Thomas E (2004) Benthic foraminifera and environmental turnover acrossthe Cretaceous/Paleogene boundary at Blake Nose (ODP Hole 1049C, NorthwesternAtlantic). Palaeogeogr Palaeoclimatol Palaeoecol 208(1–2):59–83.

32. Wendler I, Huber BT, MacLeod KG, Wendler JE (2013) Stable oxygen and carbonisotope systematics of exquisitely preserved Turonian foraminifera from Tanzania—understanding isotopic signatures in fossils. Mar Micropaleontol 102(0):1–33.

33. Deuser WG, Ross EH (1989) Seasonally abundant planktonic foraminifera of theSargasso Sea: Succession, deep-water fluxes, isotopic compositions, and paleoceano-graphic implications. J Foraminiferal Res 19:268–293.

34. Landman NH, Cobban WA, Larson NL (2012) Mode of life and habitat of scaphitidammonites. Geobios 45:87–98.

35. Jacobs DK, Landman NH, Chamberlain JA (1994) Ammonite shell shape covaries withfacies and hydrodynamics: Iterative evolution as a response to changes in basinalenvironment. Geology 22:905–908.

36. Grossman EL (1984) Carbon isotopic fractionation in live benthic foraminifera—compar-ison with inorganic precipitate studies. Geochim Cosmochim Acta 48:1505–1512.

37. McConnaughey TA, Gillikin DP (2008) Carbon isotopes in mollusk shell carbonates.Geo-Mar Lett 28:287–299.

38. Beirne EC, Wanamaker AD, Jr, Feindel SC (2012) Experimental validation of envi-ronmental controls on the δ13C of Arctica islandica (ocean quahog) shell carbonate.Geochim Cosmochim Acta 84:395–409.

39. Tobin TS, Ward PD (2015) Carbon isotope (δ13C) differences between Late Cretaceousammonites and benthic mollusks fromAntarctica. Palaeogeogr Palaeoclimatol Palaeoecol428:50–57.

40. Little CTS, et al. (2015) Late Cretaceous (Maastrichtian) shallow water hydrocarbonseeps from Snow Hill and Seymour Islands, James Ross Basin, Antarctica. PalaeogeogrPalaeoclimatol Palaeoecol 418:213–228.

41. Tsujita CJ, Westermann GEG (1998) Ammonoid habitats and habits in the WesternInterior Seaway: A case study from the Upper Cretaceous Bearpaw Formation ofsouthern Alberta, Canada. Palaeogeogr Palaeoclimatol Palaeoecol 144(1–2):135–160.

42. Henderson RA, Price GD (2012) Paleoenvironment and paleoecology inferred fromoxygen and carbon isotopes of subtropical mollusks from the Late Cretaceous (Cen-omanian) of Bathurst Island, Australia. Palaios 27(9):617–626.

43. Ifrim C, Stinnesbeck W (2010) Migration pathways of the late Campanian andMaastrichtian shallow facies ammonite Sphenodiscus in North America. PalaeogeogrPalaeoclimatol Palaeoecol 292:96–102.

44. Moriya K, Nishi H, Kawahata H, Tanabe K, Takayanagi Y (2003) Demersal habitat ofLate Cretaceous ammonoids: Evidence from oxygen isotopes for the Campanian (LateCretaceous) northwestern Pacific thermal structure. Geology 31(2):167–170.

45. Landman NH, Johnson RO, Edwards LE (2004) Cephalopods from the Cretaceous/Tertiary boundary interval on the Atlantic Coastal Plain, with a description of thehighest ammonite zones in North America; Part 1, Maryland and North Carolina. AmMus Novit 3454:1–64.

46. Gradstein FM, Ogg JG, Schmitz M, Ogg G (2012) The Geologic Time Scale (Elsevier,Amsterdam).

47. Seton M, et al. (2012) Global continental and ocean basin reconstructions since 200 Ma.Earth Sci Rev 113:212–270.

48. Cochran JK, et al. (2010) Effect of diagenesis on the Sr, O, and C isotope compositionof Late Cretaceous mollusks from the Western Interior Seaway of North America. AmJ Sci 310(2):69–88.

49. Anderson TF, Arthur MA (1983) Stable isotopes of oxygen and carbon and their ap-plication to sedimentologic and palaeoenvironmental problems. Stable Isotopes inSedimentary Geology, Society of Economic Paleontologists and Mineralogists ShortCourse Series, eds Arthur MA, Anderson TF, Kaplan IR, Veizer J, Land LS (Soc EconPaleontol Mineral, Tulsa, OK), pp 1–151.

50. Erez J, Luz B (1983) Experimental paleotemperature equation for planktonic fora-minifera. Geochim Cosmochim Acta 47(6):1025–1031.

51. Tarutani T, Clayton RN, Mayeda TK (1969) The effect of polymorphism and magne-sium substitution on oxygen isotope fractionation between calcium carbonate andwater. Geochim Cosmochim Acta 33(8):987–996.

52. Landman NH, Waage KM (1993) Scaphitid ammonites of the Upper Cretaceous(Maastrichtian) Fox Hills Formation in South Dakota and Wyoming. Bull Am Mus NatHist 215:215–257.

53. Stephenson LW (1955) Owl Creek (Upper Cretaceous) fossils from Crowleys Ridgesoutheastern Missouri. Geological Survey Professional Paper 274-E:97–140.

54. Hewitt RA (1996) Architecture and Strength of the Ammonoid Shell. AmmonoidPaleobiology, Topics in Geobiology, eds Landman NH, Tanabe K, Davis RA (Springer,New York), Vol 13, pp 297–339.

55. Kruta I, Landman N, Rouget I, Cecca F, Tafforeau P (2011) The role of ammonites in theMesozoic marine food web revealed by jaw preservation. Science 331(6013):70–72.

56. Gupta NS, Briggs DEG, Landman NH, Tanabe K, Summons RE (2008) Molecularstructure of organic components in cephalopods: Evidence for oxidative cross linkingin fossil marine invertebrates. Org Geochem 39:1405–1414.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1507554112 Sessa et al.