Introduction
Mosasaurs are a widespread and diverse group of marine reptiles that inhabited the world’s oceans during the Late Cretaceous. Fossils of these animals have been found on all continents, including Antarctica (Martin et al., Reference Martin, Bell, Case, Chaney, Fernández, Gasparini, Reguero, Woodburne, Gamble, Skinner and Henrys2002), suggesting that they were well adapted to a number of different environments. Differences in tooth morphology show that some groups of mosasaurs specialised in specific prey items (Martin & Bjork, Reference Martin, Bjork, Martin and Ostrander1987; Massare, Reference Massare1987; Schulp, Reference Schulp2005; Martin & Fox, Reference Martin, Fox, Martin and Parris2007). Because of this wide geographic distribution and specialised diet of some taxonomic groups, certain mosasaurs may have had restricted ecological or environmental ranges based on the habitat requirements of their prey.
Previous attempts at identifying the habitat segregation of mosasaurs have been based on lithology, microfossil assemblages (Russell, Reference Russell1967; Bryan, Reference Bryan1992; Kiernan, Reference Kiernan2002) or macrofaunal associations (Lingham-Soliar, Reference Lingham-Soliar1995; Manning, Reference Manning2007). In some cases these methods do not agree, suggesting that the factors involved in determining habitat preference are probably more complex than any single method can reveal.
Applications of biogeochemistry in habitat determination have included research into stable carbon isotopes in tooth enamel and other skeletal material. These studies have shown that δ13C values in structural carbon typically become more depleted with distance of foraging habitat from shore (Clementz & Koch, Reference Clementz and Koch2001; Robbins et al., Reference Robbins, Ferguson, Polcyn and Jacobs2008; Robbins, Reference Robbins2010; Schulp et al., Reference Schulp, Vonhof, van der Lubbe, Janssen and van Baal2013; Strganac et al., Reference Strganac, Jacobs, Polcyn, Ferguson, Mateus and Schulp2014). These studies further concluded that this method for determining foraging habitat is complicated by biological factors such as respiratory mode, diving habits, body size and diet, in addition to other factors such as latitude, geologic age and the level of δ13C at the base of the food chain, which also impact δ13C signatures.
Patrick et al. (Reference Patrick, Martin, Parris and Grandstaff2004, Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007) demonstrated that certain rare earth element (REE) signatures may be used as a proxy for determining relative bathymetry during early diagenesis of fossils. Their study examined the REE content of labile sediments, pore waters and natural waters in several present-day marine environments. Representative light, medium and heavy rare earth elements (LREE, MREE, HREE) were selected from their data and plotted on ternary diagrams by percentage relative to each other. The results of their analysis showed that shallow water marine environments were relatively enriched in HREE whereas deeper water marine environments were depleted in HREE and enriched in LREE/MREE. Freshwater rivers and some estuaries tend to be relatively enriched in MREE/LREE (Patrick et al., Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007), but none of the geologic units in the present study are believed to represent these types of depositional environments (Mancini et al., Reference Mancini, Puckett, Tew and Smith1995; Staron et al., Reference Staron, Grandstaff, Gallagher and Grandstaff2001; Parris et al., Reference Parris, Shelton, Matrin, Martin and Parris2007). The explanation for these observed enrichments was that ‘HREE enrichment is due to enhanced solubility of HREE in relatively high pH and alkaline seawater, produced by preferential HREE carbonate complexing, and sorption of LREE by HFO [hydrous ferric oxide] and tests of planktonic organisms’ (Patrick et al., Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007, p. 75). The LREE/MREE enrichment of deeper ocean water is attributed to the release of these elements by dissolving planktonic tests and other particles in deep water environments. The region between the shallow and deep water depth showed a gradient from HREE to LREE/MREE due to mixing of shallower oxic and deeper anoxic seawaters. In the ternary diagrams, proximity to the HREE vertex was the best indicator of relative bathymetry. A simplified diagram of the different environments and the regions of relative REE enrichment can be seen in Fig. 1.
Fig. 1. Simplified diagram of natural water environments and the regions of relative REE enrichment based on studies by Patrick et al. (Reference Patrick, Martin, Parris and Grandstaff2004, Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007). A gradient from HREE-enriched shallow water to LREE/MREE-enriched deep water in marine environments is the result of mixing by oxic and anoxic waters. Freshwater rivers and some estuarine systems are often MREE/LREE enriched.
Following this interpretation of REE data these authors then analysed mosasaur fossils from the Pierre Shale Group of South Dakota to determine if LREE/MREE/HREE content could be used as a proxy for determining paleobathymetry of marine geologic formations at the time of deposition. The analysed mosasaur fossils were divided by the geologic unit from which they were recovered and their normalised LREE/MREE/HREE ratios were plotted on a ternary diagram. The plotted results were then used to develop a relative sea-level curve for the Pierre Shale Group formations. This interpreted sea-level curve compared well with previous interpretations of water depth developed using traditional stratigraphic methods (Fig. 2).
Fig. 2. Ternary diagram of normalised REE ratios of mosasaur fossils from South Dakota by geologic formation (top) and corresponding sea level curve developed by Patrick et al. (Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007) compared to previous studies (bottom). Representative LREE, MREE and HREE in ternary diagram are neodymium (NdN), gadolinium (GdN) and ytterbium (YbN), respectively. Shallow environments plot toward the YbN vertex whereas deeper environments plot away from the YbN vertex. (Figures adapted from Patrick et al., Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007.)
The present study employs the same principles of LREE/MREE/HREE distribution in marine environments, but categorises the results by mosasaur genera rather than geological units. Mosasaur preference for water of a certain depth range should be reflected in the REE signature that was acquired during the early diagenesis of their skeletal remains. It should be emphasised that depth range does not necessarily equate to proximity to shore as sea floor gradient in different locations may vary and shallow water structures such as carbonate platforms or submerged sea mounts may be located well offshore.
The REE ratios of mosasaurs from the Pierre Shale reported by Patrick et al. (Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007) were combined with the paleobathymetry estimates for the Pierre Shale Group determined by Haq et al. (Reference Haq, Hardenbol and Vail1987) and Hanczaryk (Reference Hanczaryk2002) in Fig. 2 to produce a ternary diagram with divisions based on approximate continental shelf depths (Fig. 3). This combined ternary diagram was used to determine the preferred habitats of mosasaur genera in this analysis.
Fig. 3. Correlation of shelfal marine and freshwater river/estuarine environments with LREE/MREE/HREE ratios based on research by Patrick et al. (Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007), Hanczaryk (Reference Hanczaryk2002) and Haq et al. (Reference Haq, Hardenbol and Vail1987) (see Fig. 2).
Methods and materials
Fossil mosasaur specimens that had been identified to the genus level were selected with permission from the collections of the Alabama Museum of Natural History (ALMNH) (Table 1). These specimens were recovered from a number of different localities in Alabama, with samples representing all outcropping Late Cretaceous marine formations in the central and western regions of the state (Fig. 4).
Table 1. List of mosasaur specimens sampled from the Alabama Museum of Natural History (ALMNH) collections
Fig. 4. Map of Alabama, USA, showing the region of outcropping Late Cretaceous marine formations (green area) and the collection localities of mosasaur fossils analysed in this study (red circles). Previously published REE data for mosasaur fossils from South Dakota (SD) and New Jersey (NJ) were also used.
The mosasaur specimens from Alabama were processed using methods adopted from Patrick et al. (Reference Patrick, Martin, Parris and Grandstaff2004) and Staron et al. (Reference Staron, Grandstaff, Gallagher and Grandstaff2001). Approximately 0.5 g of fossil material was removed from larger specimens using a Dremel® rotary tool equipped with diamond drill bits. Drill bits were cleansed with dilute trace metal grade nitric acid (HNO3) and rinsed with deionised water between samples to prevent cross-contamination. Sampling was restricted to the cortical region of the bone to reduce possible contamination or dilution of REE signatures by infilling matrix located in the trabecular bone.
Approximately 100 mg of powder per sample were dissolved with 3 ml of trace metal grade nitric acid (10%) and diluted with 7 ml of deionised water, and subsequently filtered through a 0.2 µm fibreglass medium prior to chemical analysis to remove any undissolved particles.
Prepared solutions were diluted by a factor of 100 before analysis with a Perkin Elmer Elan 6000® inductively coupled plasma mass spectrometer (ICP-MS). The ICP-MS was calibrated prior to analysis using ICP-MS Multi-element Standard manufactured by High Purity Standards whereas Claritas PPT Multi-element Standard manufactured by Spex Chemical was used as the quality control standard during testing. Each sample was analysed 20 times and automated analytical software processed the results. Analytical error for most specimens was within ±5% of certified values, and results are reported in parts per million (ppm) (Table 2). Resulting values were normalised with the North American Shale Composite (NASC) (Gromet et al., Reference Gromet, Dymek, Haskin and Korotev1984) (Table 3).
Table 2. Rare earth element (REE) content (in parts per million, ppm) of mosasaur specimens from Alabama
Table 3. Normalised REE values (REEsample/NASC) of mosasaur specimens from Alabama (North American Shale Composite values from Gromet et al., Reference Gromet, Dymek, Haskin and Korotev1984)
Additional REE data was acquired from published reports of mosasaur specimens from the Pierre Shale of South Dakota (Patrick et al., Reference Patrick, Martin, Parris and Grandstaff2004) and the Navesink and lower Hornerstown formations of New Jersey (Staron et al., Reference Staron, Grandstaff, Gallagher and Grandstaff2001). Of the previously reported data, only mosasaur specimens that could be identified to genus level were used in the present study.
Rigorous statistical analysis of the data in this study was not performed because most of the taxonomic groups are represented by a very small sample size. It is believed that statistical analysis would not produce any meaningful results beyond what can be observed in the ternary diagrams for each taxon. This is a preliminary study of a potential method for habitat determination that will require analysis of a much larger sample size in order to prove its value.
Analyses and results
Normalised neodymium (NdN), gadolinium (GdN) and ytterbium (YbN) values were used as respective LREE/MREE/HREE representatives. These particular REEs were selected to allow better comparison with previous studies and because their quantities are least altered by their chemical properties (Patrick et al., Reference Patrick, Martin, Parris and Grandstaff2004). The NdN, GdN and YbN quantities were summed and plotted on ternary diagrams by percentage of total. Other LREE/MREE/HREE combinations were tested but the ternary diagram patterns did not differ greatly from the REE representatives selected.
The resulting LREE/MREE/HREE ternary diagram patterns for the analysis are presented in Fig. 5. The REE pattern for Tylosaurus (Fig. 5) is the most widely dispersed of the five genera analysed in this study, with results ranging from deep water environments to moderately shallow water depths (Fig. 3). Tylosaurus also has the largest sample size of the taxonomic groups analysed in the study (N = 15).
Fig. 5. Ternary diagrams of normalised REE ratios for Tylosaurus, Clidastes, Mosasaurus, Platecarpus and Plioplatecarpus from all areas of study. Red squares = Alabama, black circles = South Dakota, blue triangles = New Jersey. The dispersed pattern for Tylosaurus suggests that it did not have a preferred habitat whereas the compact pattern for Mosasaurus suggests that it may have preferred deeper, middle shelf marine environments. The intermediate pattern for Clidastes is inconclusive. The ancestral Platecarpus plots farther away from the YbN vertex than its descendant Plioplatecarpus, suggesting an evolutionary shift from deeper to more shallow water environments.
The data points for Clidastes (Fig. 5) (N = 6) are as unconstrained as those of Tylosaurus, but they are arranged in a linear pattern directed toward the YbN vertex in the range of intermediate or moderate water depth. No other taxonomic group in this study has a linear pattern to this degree.
The REE pattern for Mosasaurus (Fig. 5) displays the tightest cluster of data points (N = 6) of the taxonomic groups analysed. This cluster is located in the range of deeper, outer-shelf marine environments.
Platecarpus and Plioplatecarpus (Fig. 5) are relatively well constrained, but the sample size for Platecarpus is exceptionally small (Platecarpus N = 3; Plioplatecarpus N = 6). The Platecarpus grouping is positioned in a deeper water portion of the ternary diagram than Plioplatecarpus.
Discussion and conclusions
The basic premise of the analysis presented here is that the majority of individuals within a vertebrate taxonomic group die and are buried in the environment in which they lived. In a marine environment, some individuals, especially those with high body fat content in very shallow water (Reisdorf et al., Reference Reisdorf, Bux, Wyler, Benecke, Klug, Maisch, Fornaro and Wetzel2012), may become buoyant with post mortem decompositional gasses and be transported by water or air currents to exotic locations well away from their typical habitat before burial. However, these ‘bloat and float’ individuals are here assumed to be exceptional rather than the standard and most specimens are preserved in or near their preferred habitat (Reisdorf et al., Reference Reisdorf, Bux, Wyler, Benecke, Klug, Maisch, Fornaro and Wetzel2012; Mateus et al., Reference Mateus, de Pablo and Vaz2013).
Once buried in marine sediment, the preserved skeletal remains begin to assimilate REEs located in the surrounding labile sediment and pore waters during the early stages of diagenesis (Trueman & Benton, Reference Trueman and Benton1997; Trueman, Reference Trueman1999). Patrick et al. (Reference Patrick, Martin, Parris and Grandstaff2004, Reference Patrick, Martin, Parris, Grandstaff, Martin and Parris2007) showed that REEs are fractionated to a certain degree in marine and freshwater environments due to the chemical properties of the individual REEs and the environmental conditions that are present in water of various depths. Because the ratios of LREE/MREE/HREE tend to be relatively uniform in marine sediments deposited in water of a given depth, and because most individuals of a taxonomic group are buried in the environment in which they lived, the LREE/MREE/HREE ratios of mosasaur fossils should also be uniform if their preferred habitat was restricted to water of the same depth.
In Fig. 6, two ternary diagrams illustrate the different hypothetical REE ratio patterns that would result if a given mosasaur taxon had a preferred habitat based on water depth (A) or was widely distributed with no preference of water depth (B). In diagram A, the majority of specimens are clustered tightly together, which represents the preferred habitat for this taxon, whereas only a few outliers represent ‘bloat and float’ carcasses or unusual individuals who ventured outside of the preferred habitat. In diagram B, there is no central core where the majority of specimens are clustered, suggesting that this taxon did not have a preferred habitat and was able to acquire food in multiple environments.
Fig. 6. Hypothetical ternary diagrams of normalised REE ratios showing expected distribution of data points if a given mosasaur genus had a preferred habitat based on water depth (A) or had no preferred habitat (B).
The REE ratios of taxa analysed in this study suggest that some mosasaur genera may have preferred a specific range of water depth whereas others appeared to be widely distributed with no preference of water depth. In Fig. 5, the specimens in the Tylosaurus diagram are not confined to a specific cluster and are widely distributed, which suggests they had the most variation in habitat of any genus in this analysis. This is not surprising considering that Tylosaurus did not possess specialised teeth or other derived morphological features that would limit it to a specific type of prey or environment. Martin and Bjork (Reference Martin, Bjork, Martin and Ostrander1987) documented one Tylosaurus specimen from South Dakota with the remains of a shark, fish, turtle, marine bird and a smaller mosasaur in the gastric region of its abdomen. The ability to thrive on a variety of prey items such as this would probably not have limited Tylosaurus to any specific marine environment. Although many of the Tylosaurus specimens in this study are identified as Tylosaurus sp. (including all of the specimens from South Dakota), it is likely that the majority of these specimens are T. proriger based on the stratigraphic range of this species (Everhart, Reference Everhart2005). The δ13C signatures in Tylosaurus tooth enamel possess a wide range of values, suggesting that they fed in a variety of habitats (Robbins, Reference Robbins2010), which agrees with the dispersed REE pattern for the genus in the ternary diagram.
The ternary diagram for Clidastes (Fig. 5) shows a peculiar linear pattern for the REE ratios in the specimens analysed. Although there is no clearly defined cluster signifying the preference of a specific water depth, the linear shape may imply that Clidastes had a wider range of habitats, from middle to inner shelf environments. Additionally, the elongated shape of the pattern may suggest an intraspecfic separation of individuals based on ontogeny, where juveniles live in separate environments from those of adults to avoid cannibalistic predation or so that the juveniles may obtain prey items more suitable for their smaller size. Of the Clidastes analysed in this study, the two samples plotting closest to the shallow water YbN vertex were taken from very large adult specimens whereas the four samples plotting further away from the YbN vertex, suggesting preference for deeper water, were taken from one juvenile and three subadult specimens. These findings are not in agreement with previous studies of δ13C values in tooth enamel of Clidastes, in which smaller individuals had higher values reflective of near-shore feeding and larger individuals had more negative values suggestive of feeding farther from the shore (Robbins et al., Reference Robbins, Ferguson, Polcyn and Jacobs2008; Robbins, Reference Robbins2010), although the findings of the present study may be influenced by the very low sample size (N = 6). The linear REE pattern for Clidastes in the present study may also be explained as an evolutionary shift in the genus from waters of one depth to another, given that the data were obtained from specimens over a relatively wide temporal span and not a single stratigraphic horizon. The results for Clidastes are currently inconclusive and will require additional analyses to define its habitat preference using additional analytical parameters.
The REE pattern for Mosasaurus is the most tightly clustered of the taxa in this study (Fig. 5) and most closely resembles the hypothetical pattern for specific habitat preference shown in Fig. 6. The data points for this genus are clumped in the area of outer middle shelf depth on the ternary diagram (Fig. 3), with a couple of outliers positioned in shallower water depths. Most of the individuals in this taxonomic group belong to M. hoffmanni, a large species of mosasaur with cranial adaptations for feeding on large prey (Lingham-Soliar, Reference Lingham-Soliar1995; Mulder, Reference Mulder1999; Harrell & Martin, Reference Harrell and Martin2014). Bryan (Reference Bryan1992) suggested that M. hoffmanni (= M. ‘maximus’) lived in deep marine water based on his interpretation of the lithology of the Prairie Bluff Chalk in Alabama and taphonomy of ALMNH PV 1988.0018 (see Table 1). Gallagher (Reference Gallagher2005) reported only Mosasaurus sp. from the Hornerstown Formation (Main Fossiliferrous Layer) of New Jersey, a geologic unit that is interpreted as being a mid to outer shelf deposit (Staron et al., Reference Staron, Grandstaff, Gallagher and Grandstaff2001). Lingham-Soliar (Reference Lingham-Soliar1995) and Manning (2007, as M. ‘maximus’) believed M. hoffmanni to be from shallower middle to inner shelf depth using invertebrate and vertebrate faunal associations to determine paleobathymetry. Recent analysis of stable carbon isotopes in tooth enamel by Robbins (Reference Robbins2010), Schulp et al. (Reference Schulp, Vonhof, van der Lubbe, Janssen and van Baal2013) and Strganac et al. (Reference Strganac, Jacobs, Polcyn, Ferguson, Mateus and Schulp2014) suggests that Mosasaurus may have fed in offshore/open ocean environments. The results of the present REE analysis for M. hoffmanni are in agreement with the interpretations of the carbon isotope studies.
The results for Platecarpus and Plioplatecarpus (Fig. 5) are combined on a single ternary diagram because of the probable ancestor–descendant relationship between the two genera (Holmes, Reference Holmes1996; Bell, Reference Bell, Callaway and Nicholls1997) and because their biostratigraphic ranges do not overlap (Russell, Reference Russell1967). Although only three Platecarpus specimens were analysed, their REE ratios plot close together in the deeper water, left side of the diagram. These three Platecarpus specimens were obtained from three different localities in Alabama, each representing a different geologic formation and age (Table 1). The more numerous Plioplatecarpus specimens also cluster relatively well, but on the shallower, right side of the diagram (Figs 5 and 3). This suggests an evolutionary shift from the older, deeper-water preferring Platecarpus ancestor to the younger, shallower-water preferring Plioplatecarpus descendant, although a high degree of uncertainty remains with only three Platecarpus specimens analysed. This shift in habitat preference may be partly due to a change in the prey preference of the two genera. The Platecarpus species analysed in this study have comparatively stout, conical marginal dentition and robust lower jaw whereas the Plioplatecarpus species analysed have slender, piercing marginal dentition and a more gracile lower jaw. Preserved gastric residues for Platecarpus include larger bony fish up to 1.2 m in length, whereas Plioplatecarpus has been found with the preserved remains of small, soft-bodied, belemnite cephalopods (Massare, Reference Massare1987). Analysis of stable carbon isotopes in mosasaur tooth enamel shows a similar evolutionary shift in foraging habitat preference in Platecarpus/Plioplatecarpus (Robbins, Reference Robbins2010; Schulp et al., Reference Schulp, Vonhof, van der Lubbe, Janssen and van Baal2013), with Platecarpus feeding farther from shore than Plioplatecarpus, but distance from shore does not necessarily relate to water depth.
Although the number of specimens analysed in this study is not high, the results for most genera are consistent with the findings of previous researchers. Research on stable carbon isotopes performed by Robbins et al. (Reference Robbins, Ferguson, Polcyn and Jacobs2008), Robbins (Reference Robbins2010), Schulp et al. (Reference Schulp, Vonhof, van der Lubbe, Janssen and van Baal2013) and Strganac et al. (Reference Strganac, Jacobs, Polcyn, Ferguson, Mateus and Schulp2014) produced some results that are supported by the findings of the REE study presented here. Future research to improve the usefulness of REE analysis will include increasing the sample size for each taxon and grouping specimens at the species level, as well as additional data pertaining to the ontogeny of specimens and intraspecific chronostratigraphic variations.
Acknowledgements
This study is a subset of a much larger REE analysis that is a part of dissertation research conducted by the first author under the supervision of the graduate advisory committee at the University of Alabama, Tuscaloosa, Alabama, USA. Thanks are extended to Jim Parham and Mary Bade, formerly of the Alabama Museum of Natural History, for the use and permanent loan of the fossil materials under their care that were destructively analysed in this study. A great deal of appreciation is extended to Michael Polcyn at Southern Methodist University for encouraging our participation in this volume and for critically reviewing an earlier draft, which resulted in considerable improvement. Two anonymous colleagues provided additional suggestions that further improved the manuscript. Sid Bhattacharyya, Brittany Hollon and Harold Stowell in the Department of Geological Sciences at the University of Alabama provided essential laboratory training, assistance and use of equipment. Sandy Ebersole at the Geological Survey of Alabama provided assistance with reference materials and locality information of fossil specimens. Rodrigo Pellegrini at the New Jersey State Museum (NJSM) in Trenton kindly provided mosasaur identifications of specimens in the NJSM collections analysed by Staron et al. (Reference Staron, Grandstaff, Gallagher and Grandstaff2001). Prescott Atkinson of Birmingham, Alabama provided considerable financial support to the first author for the presentation of this material during the Fourth International Mosasaur Meeting in Dallas, Texas, USA. Additional thanks are extended to the Graduate School, the College of Arts and Sciences, and the Department of Geological Sciences at the University of Alabama, as well as the Geological Sciences Advisory Board (GSAB) for providing generous financial support during the course of this study.
