SDAS 2002 Vol 81

Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 161

RARE EARTH ELEMENT SIGNATURES OF FOSSIL VERTEBRATES COMPARED WITH LITHOSTRATIGRAPHIC SUBDIVISIONS OF THE UPPER CRETACEOUS PIERRE SHALE, CENTRAL SOUTH DAKOTA

Doreena Patrick Department of Geology Temple University Philadelphia, PA 19122

James E. Martin Museum of Geology South Dakota School of Mines and Technology, Rapid City, South Dakota 57701

D.C. Parris Natural History Bureau New Jersey State Museum Trenton, NJ 08625

D.E. Grandstaff Department of Geology Temple University Philadelphia, PA 19122

Keywords

Rare earth elements, Pierre Shale, South Dakota, mosasaur, Verendrye, Cre- taceous, Hornerstown, Sharon Springs, paleoenvironments, New Jersey

ABSTRACT

Rare earth elements were measured in mosasaur bones collected from ﬁve members (Sharon Springs, Gregory, Crow Creek, DeGrey, and Verendrye) of the upper Cretaceous Pierre Shale at localities near the Missouri River in Brule, Buffalo, Hughes and Hyde counties. Fossils from each member of the Pierre Shale have different REE signatures. Signatures of fossils from individual members may be distinctive over wide areas; fossils from the Verendrye Member have REE signatures that are consistent over 250 square kilometers. Fossils from the Sharon Springs Member have distinctive REE signatures that may be further subdivided into three superposed groups that correspond with the upper, middle, and lower Sharon Springs Member. Because REE signatures may differ 162 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) among stratigraphic units, fossil bones eroded from their stratigraphic context may be assigned to their proper depositional unit based on REE signature com- parisons. We interpret changes in REE signatures among members as resulting from differences in mixing between two end members: oxygenated and anoxic seawaters. If differences in mixing are interpreted as depth differences, the lower Sharon Springs Member was deposited in deep, anoxic water. Water depths de- crease in the middle and upper Sharon Springs, but the overlying Gregory Member was deposited in shallow water. The overlying Crow Creek, DeGrey, and Verendrye members were then deposited in progressively deeper marine waters, but not as deep as the lower Sharon Springs. These interpretations are generally consistent with those based on faunal diversity and eustatic sea level curves. REE signatures in fossils from the Campanian Verendrye Member were also compared with those of similar fossils from the Maastrichtian Navesink For- mation of New Jersey. Although the signatures differ, they are sufﬁciently similar so as to indicate similar degrees of mixing of oxygenated and anoxic deep waters and suggest similar water depths.

INTRODUCTION

REE concentrations and neodymium isotopes in fossil bones and teeth have been used to infer paleo-redox conditions in marine waters (Wright et al., 1987), detect reworking of fossils (e.g., Trueman and Benton, 1997; Trueman, 1999; Staron et al., 2001), and to accomplish or test paleoenvironmental (Gi- rard and Albarède, 1996; Reynard et al., 1999) and paleogeographic reconstructions (Wright et al., 2002). Many of these and other studies have examined stratigraphic variations of REE and have attempted to infer paleoenvironmental or paleoredox conditions and their variations in time and space as well as chemical nature and origin of water masses producing those signatures. The total Rare Earth Element concentrations (ΣREE) in modern bones and teeth are generally less than 20 ppm (Chenery et al., 1996; Staron et al., 2001; Patrick et al., 2001). However, ΣREE in fossil bones may be greater than 1,000 to 10,000 ppm (e.g., Arrhenius et al., 1957, and Table 1). Therefore, more than ca. 95% of REE in fossil bone is diagenetically incorporated into bone post mortem. Because most REE are introduced post mortem, REE signatures in fossils do not reﬂect the diet, trophic level, or phylogenetic position of the organism. Different osteological materials from a single organism may have different concentrations of REE; however, the signatures in all of the bones are essentially the same within analytical error (Patrick et al., 2001, in preparation). Research indicates that REE and possibly other trace elements are incorporated within 3,000 to 10,000 years after deposition (Grandstaff et al., 2001; Patrick et al., 2001, in preparation; Millard and Hedges, 1995, 1996) during early diagenetic recrystallization of the bone apatite. After incorporation into the osteological material, the REE signature in fossils is apparently stable and serves as a record of depositional or early diagenetic conditions (Wright et al., 1987; Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 163 lable, - Not determined, Sh Spr = Sharon Springs Member (degree N) (degree W) 1 Plioplatycarpus2 Plioplatycarpus3 Verendrye Plioplatycarpus5 Verendrye 43.9335 Plioplatycarpus Verendrye 43.9335 99.5119 Verendrye 43.9335 99.5119 43.9335 99.5119 - 99.5119 - 214.7 - 31.27 55.70 - 128.2 161.1 8.11 27.70 21.70 42.68 32.67 6.20 85.56 5.91 7.34 17.56 34.94 23.55 1.70 4.29 5.40 4.62 9.43 25.05 36.04 1.13 1.50 3.89 7.95 6.51 9.92 26.50 23.09 1.02 2.20 6.31 3.24 6.95 19.30 20.31 6.67 1.65 2.78 3.24 0.98 17.43 5.04 5.94 2.85 0.71 0.98 4.51 0.76 7256 Tylosaurine65 Tylosaur (?) Lower Sh Spr53 Lower Sh Spr Mosasaur 43.701262 43.6967 Tylosaur Middle Sh Spr 99.401264 Mosasaur 43.7669 99.4119 Upper Sh Spr59 Mosasaur 9.31 Upper Sh Spr60 43.7782 99.4326 Mosasaur 13.86 18.18 43.7811 Gregory 23.23 Mosasaur 99.4298 3.11 10.04 Crow Creek 99.3989 3.65 19.15 19.64 12.04 DeGrey 19.98 * 4.60 3.34 * 47.79 13.24 4.79 19.14 1.06 44.32 1.7821 1.11 * 4.74 1.64 Plioplatycarpus 4.15 15.4236 * * 4.10 Plioplatycarpus 1.12 Verendrye 6.5933 3.51 0.49 0.47 Plioplatycarpus Verendrye 4.4534 1.37 0.75 2.50 43.9035 * 4953.2 Plioplatycarpus 2.38 469.9 Verendrye35 0.54 0.58 4987.5 3.73 0.40 43.9035 99.5384 825.6 Plioplatycarpus Verendrye 469.9 0.37 2.73 3.58 44.1841 0.55 0.92 22.82 172.2 99.5384 1932.2 Verendrye 0.86 0.42 44.1841 0.31 4.25 0.11 328.0 73.49 269.2 99.7289 - 0.11 44.1841 257.2 113.0 0.94 13.57 17.57 2.17 1.00 0.74 99.7289 50.84 422.3 731.7 0.67 214.4 69.06 7.47 0.11 0.48 3.26 0.09 99.7289 53.60 7.47 123.3 354.4 14.97 184.7 40.25 0.08 1.29 1.86 0.49 1060.4 221.1 46.22 28.31 291.4 4.15 208.0 3.61 321.8 0.11 0.24 48.59 3.44 194.0 37.29 6.27 318.3 23.50 25.71 1176.7 11.25 41.45 1.46 168.3 0.54 40.04 1.50 3.36 184.3 7.62 61.65 35.63 9.38 164.8 1251.5 0.31 25.49 8.19 28.43 8.80 221.6 34.22 8.30 51.96 6.63 4.61 1.31 61.08 8.13 46.63 7.53 21.47 30.32 13.32 8.64 46.23 6.74 51.22 3.06 5.34 40.26 1.92 11.35 6.97 47.45 18.53 5.71 34.41 10.35 5.79 49.40 3.30 35.85 31.65 4.87 10.86 0.83 5.56 33.47 4.58 30.62 5.19 4.78 30.35 4.74 0.87 31.36 4.68 4.85 Number Specimen Unit Latitude Longitude La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Table 1.Table of the Pierre Shale. selected members from REE in fossils REE concentrations in ppm, * South Dakota School of Mines and Technology Specimen, primary Latitude and Longitude data not avai REE concentrations in ppm, * South Dakota School of Mines and Technology 164 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002)

Henderson et al., 1983; Trueman, 1999). REE signatures in fossils from different stratigraphic units may differ, providing unique "fingerprints" of variations in their depositional or early diagenetic environments. In this paper we present results of a preliminary study of REEs in mosasaur fossils from five Campanian members (Sharon Springs, Gregory, Crow Creek, DeGrey, and Verendrye) from the lower part of the upper Cretaceous (Cam- panian to Maastrichtian) Pierre Shale at localities near the Missouri River in Brule, Buffalo, Hughes and Hyde counties, central South Dakota. This study was conducted to determine whether REE signatures in vertebrate fossils vary stratigraphically from one member to another in the Pierre Shale. We have also investigated whether REE signatures are similar laterally within in a single member. Where REE in fossils differ stratigraphically they may be used to dis- tinguish fossils from different units. Because the REE signatures reflect the orig- inal composition and source of water masses from which the REE were incorporated, we have attempted to draw inferences about paleoenvironmental conditions and the sources of marine waters in this part of the Western Interior Seaway. We introduce the use of ternary diagrams for interpreting variations of REE signatures in fossils and their paleoenvironmental implications.

STRATIGRAPHY OF THE PIERRE SHALE

The Pierre Shale, one of the best known lithologic units of the Western In- terior, was deposited during the upper Cretaceous (Campanian and Maas- trichtian Stages) and is primarily a sequence of organic-rich black and grey shales, and interbedded bentonites, most of which formed under relatively anoxic marine conditions. Originally named the Fort Pierre Group (Meek and Hayden, 1862), it is extensively exposed in South Dakota. Within central South Dakota, members recognized by Crandell (1958) have gained general accep- tance and were used in the research reported here. Validity of the subdivision has been confirmed by Hanczaryk (2002), whose comprehensive field studies were performed in association with our own. Eight members were recognized by Crandell (1958) of which only the five lower members are pertinent to our work thus far. In ascending order, these are the Sharon Springs, Gregory, Crow Creek, DeGrey, and Verendrye members. All are widespread within the region, and the Sharon Springs has been recognized throughout the Northern Plains. The stratigraphic column of the Pierre Shale in central South Dakota is shown in Figure 1. The Sharon Springs Member has been described in numerous reports not only in central South Dakota, but also in the vicinity of the Black Hills and else- where. The member consists largely of dark fissile shale, which weathers blue- grey, with many bentonite beds, and has abundant pyrite and gypsum derived from recent pyrite weathering. Vertebrate and invertebrate fossils are abundant, including mosasaurs, plesiosaurs, diving birds, and large fish. Great variations of its thickness have been reported, ranging from four to six meters in outcrops to as much as thirty meters recognized by some drillers (Crandell, 1958). With- in the present study areas, complete sections are uncommon but thicknesses Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 165 should be closer to the higher figure. Three superposed lithostratigraphic units have been noted (Martin, 1996) and are in- formally termed lower, middle, and upper Sharon Springs until they are formally named. The Sharon Springs Member contains moderately abundant iron Figure 1. Stratigraphy of the upper Cretaceous phosphate (vivianite) nod- Pierre Shale in central South Dakota (after ules. Hanczaryk, 2002). The Gregory Member is approximately 23 m thick in the Ft. Thompson section (Hanczaryk, 2002) and consists primarily of grey claystone and marl with a few bentonites. Much of its bentonitic content appears to have been redeposited and dispersed, which is probably responsible for its characteristic "popcorn" weathering surface. The unit has yielded significant numbers of invertebrate (Baculites gregoryensis) and a few vertebrate fossils. The Crow Creek Member is a calcareous siltstone, locally sandy, which is distinguishable from the other members by a light grey to yellow-brown color. In addition to being readily recognized by surface appearance, it often sup- ports a distinctive botanical cover. The member is essentially unfossiliferous, and only a single mosasaur sample from this member was available for REE analysis. The Crow Creek is 1.2 meters thick in the Fort Thompson section and is never a very thick unit. The DeGrey Member measured 10.1 meters in the Fort Thompson section and its lower portion is gray shale and interspersed with abundant iron-manganese concretions. It has yielded numerous vertebrate and invertebrate fossils. The upper portion of the member is less conspicuously concretionary and is mostly silver-gray shale with some bentonites. The Verendrye Member is primarily silver-gray shale within the study area, but only the lower 17 m is exposed in the Fort Thompson section. Vertebrate and invertebrate fossils are common. Within the primary study area a number of marker units are recognizable, giving additional accuracy to lithostratigraphic measurements. The relatively thin Crow Creek Member itself is one such stratum. Although not of such limited thickness, the iron-manganese concretionary levels of the DeGrey Member are readily recognized. A thin, but persistent, red stratum in the lower Verendrye Member provides a reliable reference datum essentially throughout the study area. Also, a recognizable terrace-forming concretionary stratum oc- curs about 20 meters above the DeGrey-Verendrye contact. Because a complete section of the Verendrye Member has not been found in the area, these marker beds are of considerable stratigraphic value. 166 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002)

Geochronological determinations of the Campanian Sharon Springs- Verendrye interval are of the order of ten million years, ranging from 81-71 Ma B.P. (Hanczaryk, 2002). In places, the lower part of the Crow Creek Member contains shocked quartz and feldspar grains which may originate from the 74.1 ± 0.1 Ma Manson impact structure (Izett et al., 1998). The biostratigraphic zona- tion of the Pierre Shale for this region, based on ammonites, is from the Bac- ulites obtusus Zone (oldest) through the Baculites reesidei Zone (youngest), a span of thirteen zones. The entire formation is interpreted to be of marine origin and appears to represent an ideal paleoenvironment for mosasaurs, highly aquatic lizards of the Late Cretaceous. The total thickness (Fig. 1) of this part of the Pierre Shale in central South Dakota, from the base of the Sharon Springs to the top of the Verendrye Mem- ber, is approximately 55 meters (Hanczaryk, 2002; Hanczaryk et al., 1999). In central South Dakota, the 17 meter-thick Verendrye Member is incomplete and has been truncated by an unconformity. In other areas of central South Dako- ta where the Verendrye is conformable with overlying Virgin Creek Member, the thickness of the Verendrye is about 46 meters. Similarly, the complete Sharon Springs section is probably about 30 meters. Therefore, the total thickness, compensating for unconformities, is about 110 meters. The time required for deposition of these members of the Pierre Shale was approximately 10 million years (Hanczaryk, 2002). Therefore, the average depositional rate is approximately 1.1 cm/1000 years.

METHODS

Mosasaur fossil sample numbers and localities are given in Table 1. Sam- ple preparation techniques follow those in Staron et al. (2001). For some specimens, small fragments that could be used for destructive analysis were available. These bone samples were immersed in water or a dilute acid solution in an ultrasonic cleaner and then dried. Other specimens were sampled by drilling into an intact bone using a Dremel variable speed electric drill. Be- tween samples the drill was cleaned with trace metal grade dilute (5%) nitric acid or trace metal grade acetone and rinsed in distilled water. Matrix and secondary minerals were removed by handpicking. An ultrasonic technique removed more of the matrix and secondary carbonate from the bone. Samples were then rinsed with distilled water and dried. Cleaned bone fragments were mechanically crushed in an acid-washed mortar and pestle. Approximately 0.2 gram of powder was weighed for each sample. Each sample was dissolved in

3 mL Ultrex-grade HNO3 and diluted to 50 mL with 5% HNO3. When only smaller masses of bone were available, the initial solution was diluted to smaller volumes in order to preserve the mass/volume ratio. Samples were diluted to appropriate levels with 5% ultra-pure HNO3, and an internal standard containing 10 ppb 145Nd and 2 ppb 171Yb was added. Samples were analyzed using a Finnigan MAT Element/1 Inductively Coupled Plasma – Mass Spectrometer (ICP-MS). Analytical procedures follow those in Field and Sherrell (1998). Re- sults of analyses are given in Table 1. The coefﬁcient of variation for most REE Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 167 is less than + 5% of the analyzed value. Analytical results presented in Figures 2–7 have been normalized relative to the North American Shale Composite (NASC) (Gromet et al., 1984).

RESULTS

Results of REE analyses of fossil bones are given in Table 1. REE signatures (patterns of REE concentration vs. atomic number) in bones from a single fossil are essentially identical within analytical error (Patrick et al., 2001, in preparation). REE signatures of fossils within individual members also appear near- ly identical. For example, Figure 2 shows REE signatures in bones from mosasaur fossils collected from the Verendrye Mem- ber at four sites. These REE signatures are light rare earth (LREE) depleted with a small negative Cerium anomaly. REE concentrations in various fossils differ, in part due to differences in bone type. Cortical bone tends to have higher concentrations than trabec- ular bone from the same or associated fossils (Patrick et al., 2001; in preparation). However, the signatures (relative concentrations of Figure 2. NASC-normalized REE signatures in REE) are essentially identi- mosasaur fossils from the Verendrye Member of cal. Verendrye mosasaur the upper Cretaceous Pierre Shale. Although REE fossils were collected at concentrations vary in different fossils, REE sig- three different localities natures in all of the fossils are the same. Fossils containing REE in this diagram were collected over an approximately 250 over an area of more than 250 km2 in central km2 area of central South South Dakota. This indicates that, at least under Dakota. Therefore, at least some conditions, REE signatures from a strati- in some marine environ- graphic unit may be similar over extensive areas. ments, REE signatures may have great lateral extent. However, REE signatures in fossils from neighboring stratigraphic units may be signiﬁcantly different. Figure 3 illustrates examples of REE signatures that indicate that fossils from various parts of the Pierre Shale have different REE signatures. In many cases the differences can be visually distinguished. Signatures in Figure 3 vary from Middle REE (MREE) enriched in the lower Sharon Springs, to Heavy REE (HREE) enriched in the Gregory Member, and to slightly Light REE (LREE) depleted in the Verendrye Member. To conﬁrm the visual impression (Fig. 3), REE data were analyzed by Dis- criminant Analysis using NCSS statistical software (Hintze, 1997). Discriminant 168 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002)

Analysis is a statistical technique (Davis, 1986) which ﬁnds linear combinations of variables (REE ratios) that produces the maximum or optimal separation between the deﬁned groups (members). Discriminant Analysis was conducted using NASC-normalized REE ra-

tios (e.g. GdN/YbN, Fig. 4). Ratios were used, rather than concentrations, to re- move the effect of differ- Figure 3. NASC-normalized REE signatures in mosasaur fossils from selected members of the ences in REE concentra- Pierre Shale. REE signatures from different strati- tions between different graphic units are visibly and statistically different. types of osteological materials and different localities. Discriminant Analysis was conducted using all REE ratios and by stepwise regression to select the best ratios. Results of Discriminant Analysis showed that the REE signatures (ratios) in the Verendrye and Sharon Springs members, for which several analyses are available, were statistically different. For example, linear combinations of

NdN/TmN and DyN/LuN ratios allowed accurate classification of the units. If REE signatures in fossils from different members or stratigraphic units are statistically different, then these differences may be used to establish or limit the provenience of fossils from unknown or questioned localities. In some cases, significant fossils may be found in float or field notes for fossils in museum collections may be incomplete or have become lost. In such cases the scientif- ic value of the fossil may be limited by the lack of stratigraphic information. Comparison of the REE signature of the fossil with REE signatures of fossils from candidate stratigraphic units may allow the provenience of the fossil to be established or limited. The differences within and between members may be illustrated by plotting ratios of various REEs as a function of stratigraphy. GdN/YbN from the various members are shown in Figure 4. These data show that GdN/YbN ratios differ from one member to another and indicate that different members can be distinguished based on REE signatures and ratios. Gd and Yb also show rela- tions between the MREE and HREE elements. The GdN/YbN ratio is high in the lower Sharon Springs Member, indicating MREE enrichment, decreases to a minimum in the Gregory Member, and then increases slightly to the Verendrye Member. The Verendrye Member shows a consistent REE pattern throughout the member, whereas the Sharon Springs displays significant REE variations within the member. The Sharon Springs Member may be further sub-divided into three parts, based on REE signatures. The subdivisions generally coincide with previously identified units by one of us (Martin) of upper, middle and lower Sharon Springs. Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 169

Another way of visual- izing variations in REE patterns is by use of a triangular or ternary diagram. Fig- ure 5 shows a ternary diagram with NASC-normalized Yb, Gd, and Nd (YbN,

GdN, and NdN) at the ver- tices. Yb is a representative HREE, Gd a MREE, and Nd a LREE. Further, Nd, Gd, and Yb are even-numbered elements, and therefore Figure 4. Shale-normalized Gd/Yb ratios (GdN/YbN) have higher concentrations in mosasaur fossils from selected members of the according to the Oddo- Pierre Shale. Harkins effect, and thus have generally more reliable analytical data. The LREEs, Ce and La, were not used because Ce, unlike most other REE, has two valence states and is sub- ject to effects of oxidation and reduction and La which, although a member of the REE, often has slightly different chemical variations than the others. Other REE combinations were tested, particularly substi- tuting Pr for Nd, but do not Figure 5. Triangular Diagram of NASC-normalized greatly alter the patterns. values of Yb (a heavy rare earth [HREE]), Gd The ternary diagram al- (MREE) and Pr (LREE) in fossils from the Pierre lows the basic shape of the Shale. Flat, shale-like REE patterns will plot in the REE pattern to be repre- middle of the diagram, samples plotting toward the Yb vertex will be HREE-enriched, whereas sented. NASC-normalized N samples plotting toward the PrN vertex will be samples that plot in the LREE- and those plotting toward the GdN vertex middle of the diagram (33% MREE-enriched. Mosasaur samples plot along a YbN, 33% GdN, and 33% straight trend line (from lower Sharon Springs to NdN) have equal amounts Gregory), which may result from mixing of two end-member waters, one enriched in MREE and of these elements and will LREE and the other enriched in HREE. Coastal have a ﬂat, "shale-like" REE seawater may represent the HREE-enriched water patterns (near Verendrye and deep, anoxic bottom water the MREE and signature, Fig. 3). REE in LREE enriched end-members. shale samples from the Pierre Shale do indeed plot 170 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002)

near the center of the diagram. Samples plotting toward YbN will be enriched in the HREE and signatures will have a positive slope (e.g., Gregory, Fig. 3), those plotting near NdN are enriched in LREE with a negative slope, and those plotting near GdN are enriched in MREE and have a bell shape (e.g., lower Sharon Springs, Fig. 3). Thus fossils from the lower Sharon Springs are relatively enriched in MREE, whereas those from the Gregory are enriched in HREE and relatively depleted in the other end-members. REE data for fossils from the various members lie near a straight line between the Gregory and lower Sharon Springs Members (Fig. 5). This indicates that, as a ﬁrst approximation, REE data can be explained by mixing or evolution between a Yb-rich (HREE) end-member near the Gregory Member and a relatively Gd-rich (MREE, with subordinate LREE Nd) end-member near the lower Sharon Springs. The amount of variation in REE signatures in the Pierre is as great as that in some terrestrial units. The REE variations in the fossils should be related to variations in marine waters from which the REE were obtained.

FACTORS INFLUENCING REE SIGNATURES IN FOSSILS

A long debate continues about the source of REE, the mode of REE uptake by fossils, and the relationship between the REE signature of the fossil and the REE composition of the water. For example, some workers have proposed that fossils obtain their REE primarily from surface waters, including sea bottom waters (Wright et al., 1987). In this case REE signatures in marine fossils may be used to infer paleo-oceanographic conditions. However, others have argued that, depending on sedimentation rate, REE may be derived more from ground or pore waters (e.g. Elderﬁeld and Pagett, 1986). In this case REE signatures will more likely reﬂect early diagenetic conditions. Previous workers have suggested that REE signatures measured in vertebrate fossils may be affected by a number of factors, including: mechanism of REE incorporation into fossils, REE composition and speciation of water providing trace elements, redox state, sedimentation rate, as well as late diagenetic alteration.

Incorporation mechanisms

Reynard et al. (1999) suggested that REE could be incorporated into apatite by two mechanisms (1) equilibrium crystal-chemical fractionation between apatite and water and (2) adsorption and apatite surface reactions. They suggested that some "bell shaped" REE signatures, such as those commonly found in Paleozoic ichthyoliths or conodonts (e.g., Wright et al., 1987; Girard and Al- barède, 1996; Reynard et al., 1999), could arise from equilibrium fractionation between seawater and the biogenic apatite. Reynard et al. (1999) calculated relative distribution coefﬁcients or fractionation factors between water and apatite. Although their equilibrium fractionation model could qualitatively explain the "bell shaped" patterns, it could not quantitatively explain all of the REE ra- Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 171 tios, nor explain other patterns. Therefore, REE incorporation in fossil bone apatite results primarily from surface reactions between apatite and water. Incor- poration may be controlled by sorption of REE on the apatite surface, followed by incorporation during diagenetic apatite recrystallization (Trueman, 1999; Armstrong et al., 2001).

Water Composition and Speciation

Concentrations and signatures of REE have been measured in a variety of natural waters (Johannesson and Xiaoping, 1997; Piepgras and Jacobson, 1992); however, many factors controlling REE in natural waters are still a matter of debate. The aqueous properties of the REE change systematically across the series. For example, there is a systematic increase in the strength of REE-carbonate ion complexes from light to heavy REE (Haas et al., 1995). This leads to fractionation between the REE in natural waters. REE concentrations from some representative natural waters are plotted on a Nd-Gd-Yb ternary (Fig. 6). REE concentrations in seawater are very low and are strongly HREE enriched (Piepgras and Jacobsen, 1992). Thus, seawater signatures (grey circles) plot toward the Yb end of the Figure 6. NdN-GdN-YbN Ternary of REE signatures ternary (Fig. 6, data from from selected natural waters. Data from: Baue et Westerlund and Ohman, al. (1997), DeBaar et al. (1988), Elderﬁeld et al. 1992). The HREE enrich- (1990), German and Elderﬁeld (1989), Hoyle et al. ment is due to enhanced 91984), Johannesson and Xiaoping (1997), Piep- solubility of HREE in rela- gras and Jacobsen (1992), Sholkovitz et al. (1992) and Westerlund and Ohman (1992). tively high pH and alkaline seawater, produced by preferential HREE-carbonate complexing, and sorption of LREE by hydrous ferric oxides (HFO) and tests of planktonic organisms (Piepgras and Jacobsen, 1992). The slight variations between oceans appear to be consistent with variations between deep and shallow ocean water and oceanic circulation patterns (e.g., Piepgras and Jacobsen, 1992). These open marine waters also have large negative cerium anomalies as a result of strong sorption of oxidized Ce4+ on HFO. 172 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002)

In contrast, river water signatures (e.g. Elderfield et al., 1990) are relatively enriched in MREE and LREE, and often exhibit bell-shaped signatures. These data plot toward the upper left of the triangular diagram. REE signatures in estuaries (not plotted in Fig. 6) are variable and intermediate between river waters and sea waters (Hoyle et al., 1984; Elderfield et al., 1990; Sholkovitz and Szymczak, 2000). Coastal sea waters tend to have somewhat higher REE concentrations than does open ocean sea water. Their signatures (open circles) are somewhat variable but, except in the area of estuaries, dissolved REE tend to plot in the same area as do open ocean waters. Although most open marine waters and coastal sea waters have similar REE signatures, anoxic or suboxic marine bottom or pore waters are more LREE and MREE enriched (German and Elderfield, 1989; Sholkovitz et al., 1992) and tend to plot more toward the upper left of Figure 6 than do oxic marine waters. The LREE and MREE enrichment is probably due to desorption of LREE or dissolution of REE-bearing HFO under anoxic conditions (Sholkovitz et al., 1992; Bau et al., 1997). These anoxic waters also have small cerium anomalies, in contrast with overlying oxygenated marine waters.

Sedimentation rate and timing of REE uptake

Sedimentation rate may affect the source of waters (surface vs. pore or ground waters) providing REE. Researchers such as Bernat (1975) and Elder- ﬁeld and Pagett (1986) have suggested that REE and other trace elements are incorporated into fossils over relatively short periods of time, perhaps as little as 3,000 – 10,000 years (Patrick et al., 2001). If a fossil incorporates REE only at the sediment-water interface (zero sedimentation rate) then its REE signature must be most closely related to the REE composition of surface (sea bottom) water. However, if the fossil becomes deeply buried during REE uptake, then the REE signature in the fossil will result from a combination of surface and pore water REE compositions, depending on the rate and timing of incorporation. Sholkovitz et al. (1992) and others have shown that, with increasing depth, submarine pore waters become more enriched in REE and that the LREE/HREE ratio increases with depth as LREEs are preferentially released or desorbed from solid phases. Therefore, in a marine system, fossils incorporat- ing REE in areas of slow sedimentation could have lower LREE/HREE ratios, reﬂecting the low LREE/HREE ratio in seawater, whereas higher sedimentation rates should produce signatures with higher LREE/HREE values. Marine waters typically have strong negative Ce anomalies; however, anoxic bottom waters, pore waters, and some coastal seawaters have only small anomalies. The source of REEs incorporated into bone and the timing of incorporation may be important in determining the resulting REE signature, including the size of any Ce anomaly.

Diagenesis

Most studies (see discussion in Trueman, 1999) have suggested that once REE are incorporated into the bone, they are retained and provide a stable sig- Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 173 nal reflecting the depositional or diagenetic environment from which they were incorporated. Apatite crystals in living organisms are very small, their lattices have low degrees of crystallinity (Person et al., 1995), and they contain carbonate, sodium, and other species which increase the solubility of the bone material. During fossilization, and while the REE are being taken up and incorporated into the bone, the bone apatite recrystallizes. Fossil apatite contains larger crystals which have more highly crystalline lattices. Incorporation of REE, fluoride, and other trace elements may also lower the solubility of the resulting fossilized material. Growth of larger, more highly crystalline materials, which contain REE, fluoride and other elements, reduces the solubility and po- tential rate of dissolution of the fossil materials. Thus the material is much less likely to be affected by dissolution and remobilization or alteration of REE. REE signatures will be altered only by dissolution and re-precipitation of the apatite, by solid state diffusion, or recrystallization during metamorphism (Armstrong et al., 2001).

DISCUSSION AND PALEOENVIRONMENTAL INTERPRETATION

Changes in REE signatures in vertebrate fossils from the various members (Figs. 3–5) should be explainable in terms of variations in REE concentrations in marine or pore waters, such as those shown in Figure 6. REE patterns in fossils may not be identical with those of waters from which the REE were obtained. They may differ from water compositions due to crystal-chemical effects (preferential incorporation of MREE, Reynard et al., 1999), differences in aqueous REE speciation, and relative efficiencies of adsorption of various REEs. However, for marine waters these factors should be relatively constant because pH and compositional variations in sea water and sea bottom waters are relatively small. Therefore, relative changes in fossil REE signatures should reflect similar relative changes in the compositions of natural waters from which they were obtained (see also Armstrong et al., 2001). If the average sedimentation rate in the Pierre Shale is approximately 1.1 cm/1000 years and the period of REE uptake less than ca. 3,000 to 10,000 years (Patrick et al., 2001), then the fossil would not be deeply buried and REE signatures probably most closely reflect composition of ocean bottom waters. REE signatures in fossils are different from those in the surrounding sediments, which have essentially flat patterns. This tends to support incorporation of REE largely from the sea bottom water. Changes in REE signatures in fossils from these members of the Pierre Shale might be explained either by (1) mixing between LREE/MREE enriched estuarine/river water (upper left in Fig. 6) and HREE enriched oxic open marine water (lower right in Fig. 6) end-members, or (2) by mixing between LREE/MREE-enriched anoxic water (also toward the upper left in Figure 6) and HREE-enriched oxic open marine water. There is no lithological evidence that the Pierre Shale in central South Dakota was deposited in an estuarine environment. As indicated above, the Pierre Shale consists of laterally extensive organic-rich, largely unbioturbated, 174 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) black and grey shales, silts, and bentonites. This suggests deposition in relatively anoxic conditions, with low deposition rates and relatively remote from estuaries. Therefore, the main trend observed in Figure 5 appears to be most consistent with incorporation of REE from mixtures between oxygenated open sea water and anoxic or suboxic marine water masses. For example, REE signatures in fossils from the lower Sharon Springs Member, which are relatively enriched in LREE and MREE (Fig. 5), and have low Ce anomalies (Fig. 3), are more similar to signatures in anoxic or seasonally anoxic marine waters such as the Cariaco Trench (DeBaar et al., 1988) or the Eastern Mediterranean (Bau et al., 1997). This also seems consistent with the nature of the black, pyrite-rich shale of the lower Sharon Springs. In contrast, REE signatures in fossils from the Gregory Member are highly HREE (YbN) enriched. Such signatures are more similar to those of open seawater. REE signatures in the overlying Crow Creek, DeGrey, and Verendrye Members become progressively more MREE and LREE enriched (Figs. 2 and 3), indicating progressively greater mixing with anoxic waters. Such changes in mixing between oxic and anoxic or suboxic marine waters in one locality might occur in three ways: (1) changes in water depth, (2) changes in organic productivity, and (3) changes in upwelling or intensity of marine currents bringing anoxic waters toward the surface. Although changes in marine currents, storms, or organic productivity cannot be ruled out, many researchers have proposed eustatic or relative changes in sea level during this period (Haq et al., 1987; Stoffer et al., 1998; Hanczaryk et al., 1999). If the data are interpreted primarily in terms of changes in relative water depth, then the lower Sharon Springs fossils were deposited in relatively deep water. REE patterns in the middle and upper Sharon Springs suggest shallowing and fossils in the overlying Gregory Member may have been deposited in relatively shallow water. Overlying members are then deposited in progressively deeper marine environments, but not as deep as the Sharon Springs. These results are generally consistent with paleoenvironmental interpretations by Hanczaryk (2002) and sea level variations proposed by Haq et al. (1987). Hanczaryk (2002) interpreted biofacies in members of the Pierre Shale as indicating shallow depths, perhaps 100 meters, for the Gregory, depths less than 100 meters in the Crow Creek, 100+ meters for the lower DeGrey, 200 meters for the upper DeGrey and 200 meters for the DeGrey-Verendrye boundary. Hanczaryk (2002) did not investigate the Sharon Springs Member; however, the Sharon Springs appears to have been deposited in deeper waters. Haq et al. (1987) proposed a general long-term eustatic shallowing between 81 Ma (Sharon Springs) and 71 Ma (Verendrye), with a short-term, abrupt shallowing about 75 Ma, generally consistent with HREE enriched Gregory and Crow Creek fossils. A short-term period of shallower waters around 80 Ma is not recorded in these fossils or in the lower Sharon Springs lithology, but might occur below the lowest Sharon Springs section preserved in central South Dakota. Although the REE signatures of mosasaur bones appear to correlate generally with members of the Pierre Shale, not enough samples have been obtained to demonstrate how closely the members and REE signatures correspond. More extensive sampling and better stratigraphic control will probably Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 175 show that REE signatures do vary within members, much as they do within the Sharon Springs.

COMPARISON OF REE IN THE VERENDRYE MEMBER WITH THE NAVESINK AND HORNERSTOWN FORMATIONS, NEW JERSEY

REE signatures in mosasaur fossils from the Pierre Shale may be compared (Fig. 7) with those from the Navesink and Hornerstown formations (Staron et al., 2001). The Navesink and Hornerstown formations consist of slowly deposited, olive-green to gray, pelletized glauconitic sand. These formations are thought to represent mid-shelf marine deposits. The glauconitic sediments were probably deposited under relatively anoxic conditions. The Navesink For- mation is Late Cretaceous (Maastrichtian) in age. The overlying Hornerstown Formation may span the Cretaceous-Tertiary boundary (Staron et al., 2001); its Main Fossiliferous Layer (MFL) is correlated with the K-T boundary and yields both vertebrate and invertebrate fossils. Because of some reworking of fossils between the Navesink and overlying Hornerstown Formation and MFL, only bones having cortical material have been used for comparison. These bones are less likely to have been reworked. These bones also have REE signatures that are characteristic of their formations and consistent with in-situ fossil deposition. The Figure 7. Comparison of YbN,PrN, and GdN in REE Verendrye and signatures in mosasaur fossils from the Verendrye Navesink/Hornerstown are Member of the Pierre Shale with the Navesink and not the same age. The Hornerstown Formations (upper Cretaceous Verendrye Member is Cam- [Maastrichtian] and lowest Paleocene [Danian]), at panian whereas the New Inversand, New Jersey. The Hornerstown Forma- tion contains a bone bed, the Main Fossiliferous Jersey formations are Maas- Layer (MFL), which is also shown in the Figure. trichtian to Danian. These Relative REE enrichments are similar in the data are plotted only for a Verendrye, Navesink, and Hornerstown Forma- general comparison of pa- tions. This suggests similar degrees of mixing of leoenvironments. Data near surface and anoxic bottom waters and may indicate similar depths of formation for these for- from the two sites are plot- mations. ted in a triangular diagram (Fig. 7). The degree of MREE and LREE enrichment is similar for the two 176 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) units. This suggests that, if the mixing end-members are similar, the Navesink and Hornerstown formations may have been deposited similar conditions of surface water/anoxic bottom water mixing.

CONCLUSIONS

Rare Earth Element signatures are essentially identical in bones of fossil mosasaurs from individual members or units of the Pierre Shale. However, REE signatures in fossils from different members are different and can be distinguished by statistical methods, such as Discriminant Analysis. Such unique REE signatures may enable paleontologists to determine the provenience of fossils of unknown or questioned origin, or fossils for which there are insufﬁcient cu- ratorial data. Ternary diagrams show that REE data for members of the Pierre Shale lie along a two-component mixing line. This line is interpreted as most consistent with mixing of oxygenated marine waters having HREE enrichment with anoxic or suboxic marine water having MREE and moderate LREE enrichment. If the REE trends are interpreted as being related to ocean depth, then the lower part of the Sharon Springs Member, the lowest member of the Pierre Shale, was deposited in relatively deep anoxic sea bottom waters. REE signatures in the lower, middle, and upper Sharon Springs members are different. The upper Sharon Springs member is more HREE enriched, possibly consistent with decreased depths and regression. The overlying Gregory Member was apparently deposited in shallow, possibly coastal waters. Overlying members (Crow Creek, DeGrey, and Verendrye) were deposited in progressively deeper water. These results are generally consistent with previous paleoenvironmental conclusions based on analysis of biofacies (Hanczaryk, 2002) and sea level curves of Haq et al. (1987). REE signatures from the Verendrye Member of the Pierre Shale were compared with signatures from the Maastrichtian Navesink and Maastrichtian-Dani- an Hornerstown formations. Based on similarities in LREE enrichments and as- suming that the mixing end-members have similar REE signatures, the two sites were deposited under similar degrees of mixing of oxic and anoxic marine waters.

ACKNOWLEDGMENTS

The authors wish to thank the U.S. Army Corps of Engineers for their on- going support of ﬁeld, laboratory, and research investigation at the South Dakota School of Mines and Technology. We also thank the Crow Creek Sioux Tribe for permission to collect on their land. Paul Field and Louise Bolge aid- ed ICP-MS analyses. Temple University provided partial support for this research with a grant from the Research Incentive Fund. This study represents part of a M.S. thesis by the ﬁrst author. Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 177

LITERATURE CITED

Armstrong, H. A., D. G. Pearson, and M. Griselin, 2001. Thermal effects on rare earth element and strontium isotope chemistry in single conodont elements. Geochim. Cosmochim. Acta, 54, 435-441 Arrhenius, G., M. N. Bramlette, and E. Picciotto.1957. Localization of radioac- tive and stable heavy nuclides in ocean sediments. Nature, 180, 85-86. Bau, M., P. Möller, P, and P. Dulski. 1997. Yttrium and lanthanides in eastern Mediterranean seawater and their fractionation during redox-cycling. Ma- rine Chemistry, 56, 123-131. Bernat, M., 1975, Les isotopes de l'uranium et du thorium et les terres rares dans l'environment marin. Cahiers ORSTOM Series Geol., 7, 65-83. Crandell, D.R. 1958. Geology of the Pierre Area, South Dakota. United States Geological Survey Professional Paper 307, 83 p. Chenery, S., T. Williams, T. A. Elliot, P. L. Forey, and L. Wergelin. 1996. Deter- mination of rare earth elements in biological and mineral apatite by EPMA and LAMP-ICP-MS. Mikrochimica Acta (supplement), 13, 259-269. Davis, J. C., 1986. Statistics and Data Analysis in Geology. John Wiley, New York, 646 p. DeBaar H. J. W., C. R. German, H. Elderfield, and P. vanGaans, 1988. REE dis- tributions in anoxic waters of the Cariaco Trench. Geochim. Cosmochim. Acta, 52, p. 1203-1219. Elderfield, H., and R. Pagett, 1986. REE in ichthyoliths: Variations with redox conditions and depositional environment. The Science of the Total Envi- ronment, 48, 175-197. Elderfield, H., R. Upstill-Goddard, and E. R. Sholkovitz, 1990. The rare earth elements in rivers, estuaries, and coastal seas and their significance to the composition of ocean water. Geochim. Cosmochim. Acta, 54, 971-991. Field, M. Paul and R. M. Sherrell. 1998. Magnetic Sector ICPMS with desolvat- ing micronebulization: Interference-Free subpicogram determination of Rare Earth Elements in natural samples. Analytical Chemistry, 70, 4480- 4486. German C. R., and H. Elderfield, 1989. Rare earth elements in Saanich Inlet, British Columbia, a seasonally anoxic basin. Geochim. Cosmochim. Acta, 53, 2561-2571. Girard C., and F. Albarède, 1996. Trace elements in conodont phosphates from the Frasnian/Famennian boundary. Paleogeograpphy, Paleoclimatology, Paleoecology, 126, 195-209. Grandstaff, D.E., Terry, D.O., Jr., and Patrick, D. 2001, Rare earth element (REE) signatures in fossils from marine and terrestrial environments (abstract). Jour. Vertebrate Paleo. 21, Supplement to No. 3, 56A. Gromet L.P., R. F. Dymek, R.F., L. A. Haskin, and R. L. Korotebv, 1984. The North American Shale Composite: Its composition, major, and trace element characteristics. Geochim. Cosmochim. Acta, 48, 2469-2482. Haas, J. R., E. L. Shock, and D. C. Sassani, 1995. Rare earth elements in hy- drothermal systems: estimates of standard partial molal thermodynamic properties of aqueous complexes of the rare earth elements at high pres- sures and temperatures. Geochim. Cosmochim. Acta, 59, 4329-4350. 178 Proceedings of the South Dakota Academy of Science,Vol. 81 (2002)

Hanczaryk, P.A. 2002. Integrated Stratigraphy of the Pierre Shale of Central South Dakota. Unpublished Master of Science Thesis, Rutgers University, New Brunswick, New Jersey. 156 pages. Hanczaryk, P.A., W. B. Gallagher, K. G. Miller, M. D. Feigenson, R. K. Olsson, J. E. Martin, D. C. Parris, and M. Depangher, 1999. Integrated stratigraphy of the Pierre Shale of the Big Bend region of the Missouri River in central South Dakota, Geol. Soc. America. Abstracts with Programs, 31, 235. Haq, B.U., J. Hardenbol, and P. R. Vail, 1987. Chronology of fluctuating sea levels since the Triassic. Science, 235, 1156-1167. Henderson, P., Marlow, C.A., Molleson, T.I., and Williams, C.T., 1983, Patterns of chemical change during bone fossilization: Nature, 306, 358-360. Hintze, J. L., 1997. NCSS Users Guide. Number Cruncher Statistical Systems, Kaysville, Utah, 404 p. Hoyle, J.. H. Elderfield, A. Gledhill, and M. Greaves, 1984. The behaviour of the rare earth elements during the mixing of river and seawaters. Geochim. Cosmochim. Acta, 48, 143-149. Izett, G. A., W. A. Cobban, G. B. Dalrymple, J. D. Obradovich, 1998. 40Ar/39Ar age of the Manson impact structure, Iowa, and correlative impact ejecta in the Crow Creek Member of the Pierre Shale (upper Cretaceous), South Dakota and Nebraska. Geol. Soc. America. Bull., 110, 361-376. Johannesson, K. H., and Z. Xiaoping, 1997. Geochemistry of the Rare Earth El- ements in natural terrestrial waters: A review of what is currently known. Chinese Jour. of Geochem. 16, 20-42. Martin, James E., 1996. Disconformities of the lower Pierre Shale (Cretaceous) of South Dakota. Geol. Soc. Amer. Abstracts with Programs, Rocky Moun- tain Section, 28, 16. Meek, F.B. and F. V. Hayden, 1862. Philadelphia Academy of Natural Sciences Proceedings, 13, 419-424. Millard, A.R., and Hedges, R.E.M., 1995. The role of the environment in uranium uptake by buried bone. Jour. Archaeol. Sci., 22, 239-250. Millard, A.R., and Hedges, R.E.M., 1996. A diffusion-adsorption model of uranium uptake by archaeological bone. Geochim. Cosmochim. Acta, 60, 2139-2152. Patrick, D., D. O. Terry, Jr. and D. E. Grandstaff, 2001. The influence of osteological material and time on rare earth element (REE) concentrations in vertebrate fossils (abstract). Journal of Vertebrate Paleontology, v. 21, Sup- plement to No. 3, p. 88A. Person, A., Bocherens, H., Saliège, J-F, Paris, F., Zeltoun, V., and Gérard, M., 1995, Early diagenetic evolution of bone phosphate: an x-ray diffractome- try analysis. Jour. Archaeological Sci., 22, 211-221. Piepgras, D. J., and S. B. Jacobsen, 1992. The behavior of rare earth elements in seawater: Precise determination of variations in the North Pacific water column. Geochim. Cosmochim. Acta, 56, 1851-1862. Reynard, B., C. Lécuyer, and P. Grandjean, 1999. Crystal-chemical controls on rare-earth element concentrations in fossil biogenic apatites and implications for paleoenvironmental reconstructions. Chem. Geol., 155, 233 - 241. Proceedings of the South Dakota Academy of Science,Vol. 81 (2002) 179

Sholkovitz, E. R., T. J. Shaw, and D. L. Schneider, 1992. The geochemistry of rare earth elements in the seasonally anoxic water column and porewaters of Chesapeake Bay. Geochim. Cosmochim. Acta, 56, 3389-3402. Sholkovitz, E., and R. Szymezak, 2000. The estuarine chemistry of rare earth elements: comparison of the Amagon, Fly, Sepik and Gulf of Papua systems. Earth and Planet. Sci. Lett., 179, 299-309. Staron, R.M., B. S. Grandstaff, W. B. Gallagher, and D. E. Grandstaff, 2001. REE signatures in vertebrate fossils from Sewell, NJ: Implications for location of the K-T boundary. PALAIOS, 16, 255-265. Stoffer, P. W., P. Messina, and J. A. Chamberlain, Jr., 1998. Upper Cretaceous stratigraphy of Badlands National Park, South Dakota: Inﬂuence of tecton- ism and sea level change on sedimentation in the Western Interior Seaway. Dakoterra, 5, 55-62. Trueman C. N., 1999. Rare earth element geochemistry and taphonomy of terrestrial vertebrate assemblages. PALAIOS, 14, 555-568. Trueman C.N., and M. J. Benton, 1997, A geochemical method to trace the taphonomic history of reworked bones in sedimentary settings: Geology, 5, 263-266. Westerlund Stig, and Peder Ohman, 1992. Rare earth elements in the Arctic Ocean. Deep-Sea Research, 39, 1613-1626. Wright C. A., C. R. Barnes, and S. B. Jacobsen, 2002. Neodymium isotopic composition of Ordovician conodonts as a seawater proxy: Testing paleo- geography. Geochem., Geophys., Geosys., 3, 2, 10.10029/2001GC000195. Wright J., H. Schrader, and W. T. Holser, 1987. Paleoredox variations in ancient oceans recorded by rare earth elements in fossil apatite. Geochim. Cos- mochim. Acta, 51, 631-644.