Bone histology and growth of chasmosaurine ceratopsid dinosaurs from the late Campanian Kaiparowits Formation, Southern Utah

Ceratopsian dinosaurs are one of the most diverse dinosaur groups in the Cretaceous, and an outstanding question is how growth strategies of this group evolved in relation to their shift from small bipedal basal ceratopsians to larger quadrupedal ceratopsids. Previous bone histology studies have investigated several basal ceratopsians and centrosaurine ceratopsids (e.g., Centrosaurus, Pachyrhinosaurus, Einiosaurus), but no chasmosaurine ceratopsids have been investigated. I conducted histological analysis of humeri, ulnae, femora, tibiae, ribs, and ossified tendons from multiple specimens of two species of chasmosaurine ceratopsid dinosaurs from the late Campanian (Upper Cretaceous) Kaiparowits Formation of southern Utah, Kosmoceratops richardsoni and Utahceratops gettyi, to examine bone microstructure indicators of growth rate and maturity. I also reexamined the long-bone histology of the ceratopsian dinosaurs Psittacosaurus mongoliensis, Protoceratops andrewsi, and Centrosaurus apertus. All elements of Utahceratops and Kosmoceratops examined are dominated by densely vascularized tissue, indicative of sustained fast growth. Radially-oriented vascular canals as well as dense osteocytes from throughout ontogeny are further indicators of rapid growth. I identified juvenile (UMNH VP 20444 & UMNH VP 20454), subadult (UMNH VP 16681) and adult (UMNH VP 16860, UMNH VP 16861, UMNH VP 12198) specimens of Utahceratops, and two subadult to adult specimens (UMNH VP 17000 & UMNH VP 21339) of Kosmoceratops. I conclude that basal ceratopsians grew more slowly than the large quadrupedal ceratopsids, as evidenced by a generally higher number of definitive growth lines prevalent throughout development. In contrast, the presence of dense osteocytes, and reticular and radially-oriented vascular canals are rapid growth indicators shared by all sampled large ceratopsids, and imply an elevated metabolism for all ceratopsians. Sampled specimens of Utahceratops and Kosmoceratops do not preserve any evidence of annual lines of arrested growth (LAGs). Placed in context with the number of LAGs observed in Alaskan Pachyrhinosaurus, Centrosaurus from Alberta, and Einosaurus from Montana, these data suggest a latitudinal gradient in the number of LAGs, which suggests that bone growth is reacting to the climate.

BONE HISTOLOGY AND GROWTH OF CHASMOSAURINE CERATOPSID DINOSAURS FROM THE LATE CAMPANIAN KAIPAROWITS FORMATION, SOUTHERN UTAH by Carolyn Gale Levitt A thesis submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Master of Science in Geology Department of Geology and Geophysics The University of Utah May 2013 Copyright © Carolyn Gale Levitt 2013 All Rights Reserved The University of Utah Graduate School STATEMENT OF THESIS APPROVAL The following faculty members served as the supervisory committee chair and members for the thesis o f_______________Carolyn Gale Levitt_________________ Dates at right indicate the members' approval of the thesis. Randall B. Irmis_____________ , Chair March 8, 2013 Date Approved Allan A. Ekdale_______________ , Member March 7, 2013 Date Approved Scott D. Sampson______________ , Member March 11, 2013 Date Approved The thesis has also been approved by D. Kip Solomon_______________ , Chair of the Department of ________________Geology and Geophysics_________________ and by Donna M. White, Interim Dean of The Graduate School. ABSTRACT Ceratopsian dinosaurs are one of the most diverse dinosaur groups in the Cretaceous, and an outstanding question is how growth strategies of this group evolved in relation to their shift from small bipedal basal ceratopsians to larger quadrupedal ceratopsids. Previous bone histology studies have investigated several basal ceratopsians and centrosaurine ceratopsids (e.g., Centrosaurus, Pachyrhinosaurus, Einiosaurus), but no chasmosaurine ceratopsids have been investigated. I conducted histological analysis of humeri, ulnae, femora, tibiae, ribs, and ossified tendons from multiple specimens of two species of chasmosaurine ceratopsid dinosaurs from the late Campanian (Upper Cretaceous) Kaiparowits Formation of southern Utah, Kosmoceratops richardsoni and Utahceratops gettyi, to examine bone microstructure indicators of growth rate and maturity. I also reexamined the long-bone histology of the ceratopsian dinosaurs Psittacosaurus mongoliensis, Protoceratops andrewsi, and Centrosaurus apertus. All elements of Utahceratops and Kosmoceratops examined are dominated by densely vascularized tissue, indicative of sustained fast growth. Radially-oriented vascular canals as well as dense osteocytes from throughout ontogeny are further indicators of rapid growth. I identified juvenile (UMNH VP 20444 & UMNH VP 20454), subadult (UMNH VP 16681) and adult (UMNH VP 16860, UMNH VP 16861, UMNH VP 12198) specimens of Utahceratops, and two subadult to adult specimens (UMNH VP 17000 & UMNH VP 21339) of Kosmoceratops. I conclude that basal ceratopsians grew more slowly than the large quadrupedal ceratopsids, as evidenced by a generally higher number of definitive growth lines prevalent throughout development. In contrast, the presence of dense osteocytes, and reticular and radially-oriented vascular canals are rapid growth indicators shared by all sampled large ceratopsids, and imply an elevated metabolism for all ceratopsians. Sampled specimens of Utahceratops and Kosmoceratops do not preserve any evidence of annual lines of arrested growth (LAGs). Placed in context with the number of LAGs observed in Alaskan Pachyrhinosaurus, Centrosaurus from Alberta, and Einosaurus from Montana, these data suggest a latitudinal gradient in the number of LAGs, which suggests that bone growth is reacting to the climate. iv TABLE OF CONTENTS ABSTRACT....................................................................................................................... iii LIST OF TABLES ............................................................................................................vii LIST OF FIGURES........................................................................................................... viii ACKNOWLEDGEMENTS..................................................................................................x INTRODUCTION................................................................................................................ 1 Bone Histology.............................................................................................................. 3 Previous Studies............................................................................................................ 5 Goals of Present Study..................................................................................................6 GEOLOGIC SETTING........................................................................................................7 Laramidia......................................................................................................................7 Localities for Specimens in this Study..................................................................... 10 MATERIALS and METHODS......................................................................................... 18 Specimens................................................................................................................... 18 Preparation of Histologic Samples.............................................................................20 Analysis of Microstructure........................................................................................ 26 RESULTS.......................................................................................................................... 58 Utahceratops gettyi Bone Histology........................................................................ 58 Kosmoceratops richardsoni Bone Histology............................................................72 Summary of Histological Trends.............................................................................. 76 DISCUSSION.....................................................................................................................99 Comparison With Other Ceratopsian Taxa............................................................. 99 Comparison With Other Archosaurs....................................................................... 106 Evolution of Ceratopsian Body Size and Locomotion...........................................117 Metabolic Inferences.................................................................................................119 Implications for Ceratopsid Ontogeny......................................................................122 Latitudinal Variation in Growth.................................................................................123 CONCLUSIONS................................................................................................................. 134 REFERENCES.................................................................................................................... 137 vi LIST OF TABLES 1. Elements sampled in this study.................................................................................30 2. Limb bone measurements..........................................................................................33 3. Vascularity and osteocytes analysis.......................................................................... 34 LIST OF FIGURES 1. Map of Grand Staircase-Escalante National Monument........................................ 14 2. Stratigraphic Column of the Kaiparowits.................................................................16 3. Bones Sectioned from Utahceratops gettyi and Kosmoceratops richardsoni.......28 4. Outlines of Bones Where Sectioned............................................................................31 5. Box Count Example......................................................................................................42 6. Radial Canals, Longitudinal Canals, and Circumferential Canals............................44 7. Femora Graphs.............................................................................................................. 46 8. Tibiae Graphs............................................................................................................... 50 9. Humeri Graphs.............................................................................................................. 54 10. Cross Sections of Humeri Studied.............................................................................. 81 11. Cross Sections of Ulnae Studied.................................................................................83 12. Cross Sections of Femora Studied.............................................................................. 85 13. Woven Collagen Fiber Orientation.............................................................................87 14. Cross Sections of Tibiae Studied................................................................................89 15. Cross Sections of Ribs and Tendons Studied............................................................91 16. Remodeling in Kosmoceratops richardsoni............................................................. 93 17. Variation of Vascularity.............................................................................................. 95 18. Secondary versus Primary Osteons.............................................................................97 19. Comparison of Microstructure in Different Ceratopsian Species..........................128 20. Paleogeographic Map of Laramidia.......................................................................... 130 21. Femora and Tibiae: LAGS versus Circumference................................................... 132 ix ACKNOWLEDGEMENTS I would like to first and foremost thank my adviser Dr. Randy Irmis for helping me in developing this idea for my research project and for all of his helpful advice. I would like to thank my committee members Dr. Allan A. Ekdale and Dr. Scott Sampson for all of their advice and encouragement. I would like to thank Mike Getty, Dr. Mark Loewen, Eric Lund, Deanna Brandau, Alan Titus, Brian Switek, and Megan Crocker for all of their helpful discourse throughout this project. I would also like to thank Lab Technicians Quintin Sahratian from the Department of Geology and Geophysics at the University of Utah, Ellen-Therese Lamm and Carrie Ancell from the Museum of the Rockies in Bozeman, Montana for teaching me how to mold, cast, cut, and create thin sections of dinosaur bones. Also I would like to thank Lab Preparators Ed Lamb, Sharon Walkington, Fred Lacy, Ann Johnson, and Randy Johnson for cleaning and repairing the dinosaur bones for me to section. I could not have done this masters research project without the help of these wonderful volunteers who are invaluable to the Natural History Museum of Utah and to the scientists who do research on the dinosaurs bones housed within the museum. I would also like to thank Brian Baziak, Sarah Werning, Dawn Renee Farkes Prasad, Sally Potter and Marjorie Chan for all of their technical support with my large images. I would like to thank the following institutions for funding this project: Grand Staircase Escalante Partners, Association of Women Geoscientists, The Geological Society of America, and The Paleontological Society. I would like to thank my wonderful parents Gloria Evans Levitt and Bart Levitt for all of their love, support and encouragement throughout this project and throughout my whole life. I would like to thank Tim Bussian for his love, support, technical support with Excel and for letting me borrow his Dremel. I would also like to thank Karin Bussian for lending me her external hard drive for this project. She will always be remembered as a sister and a friend. xi INTRODUCTION Ceratopsids, or horned dinosaurs, are large-bodied, quadrupedal, herbivorous ornithischian dinosaurs (Dodson et al., 2004) that belong to the larger clade Ceratopsia, which is diagnosed by characteristics such as parrot-like beaks, dental batteries with shearing dentitions, hypertrophied narial regions, and ornamented parietosquamosal frills (Dodson et al., 2004; Sampson et al., 2010a). Ceratopsians include both small, basal forms like Psittacosaurus, and the larger quadrupedal ceratopsids exemplified by Centrosaurus and Triceratops (Ryan and Russel, 2001; You and Dodson, 2004; Dodson et al., 2004). Ceratopsidae is divided into two clades, the Centrosaurinae and the Chasmosaurinae. Chasmosaurines characteristically have parietosquamosal frills that are typically simply adorned and elongated (Lambe, 1915), whereas centrosaurines generally possess relatively shorter, highly adorned frills (Lambe, 1915; Dodson, 1993; Ryan and Russel, 2001). These clades radiated in a very short amount of time (<5 million years), and were extremely speciose in western North America during the latest Cretaceous (Sampson et al., 2010b). Furthermore, new evidence suggests they were largely endemic, with distinct forms living in each sedimentary basin of western North America (Sampson et al., 2010a). Ceratopsian dinosaurs were successful herbivores during the Late Cretaceous. In Campanian-Maastrichtian western North American terrestrial ecosystems, ceratopsids remains typically rank second to hadrosaurids in absolute abundance (Lehman, 1997; Brinkman et al., 1998; White et al., 1998). They played an important role in Late Cretaceous ecosystems as specialized primary consumers possessing dental batteries developed for shearing the plants that they ate (Farlow, 1987). The evolutionary history of ceratopsians includes several major morphological changes, and an outstanding question is how growth strategies of this group evolved in relation to their shift from small bipedal basal forms to the larger quadrupedal horned ceratopsids, a body size increase of several orders of magnitude (Carrano, 2006). Therefore, comparison of the growth rates and age of maturity among chasmosaurines, centrosaurines, and basal ceratopsids can elucidate how growth strategy changed in concert with locomotion and body size. This question relates more generally to how large animals get so large and the different ways in which they accomplish this, and ceratopsian dinosaurs are an excellent case study. Growth data can also provide a context for ontogenetic studies demonstrating that horn and frill characters were not fully expressed until individuals approached adult size (Sampson, 1997; Horner and Goodwin, 2008), and inform debates about whether certain species are distinct or different ontogenetic stages of the same taxon (e.g., Scannella and Horner, 2010, 2011; Farke, 2011; Longrich and Field, 2012). Finally, ceratopsids could be ideal for understanding how growth varied across latitude, because Campanian North American taxa range from Alaska to Mexico (e.g., Loewen et al., 2010; Erickson and Drunkenmiller, 2011; Fiorillo and Tykoski, 2012). Metabolic estimates for large chasmosaurine ceratopsid dinosaurs are not well constrained. A major question has been whether the thermophysiology of dinosaurs was more similar to extant birds and mammals or non-avian reptiles (Lee and Werning, 2 2008). It is also possible that they developed a thermal physiology that was uniquely their own, operating between ‘typical' reptilian and mammalian metabolic rates (de Ricqles, 1980). Emerging evidence suggests that many dinosaurs had metabolic strategies similar to modern birds and mammals (Horner et al., 2000; Erickson and Tumanova, 2000; Erickson et al., 2001; Padian et al., 2001, 2004; Horner and Padian, 2004; Klein and Sander, 2007; Lee and Werning, 2008; Erickson et al., 2009) therefore, did ceratopsian dinosaurs grow similarly or differently to these other dinosaurs? Specifically, the evolution of increased metabolic rates in dinosaurs is believed to have facilitated the evolution of gigantism by enabling them to build their skeletons swiftly (de Ricqles, 1980; Chinsamy 1993; Sander 2000); was this also the case for ceratopsian dinosaurs? Bone Histology The microstructure of bone can provide answers to the questions of how extinct animals grew, how old they were when they reached sexual maturity, and what the average age was when they died. Numerous studies have been conducted using histology of dinosaur bone and they reveal that the limb bones preserve an excellent record of the life history of the animal being studied (Erickson and Tumanova, 2000; Horner et al., 2000; Padian et al., 2001; Horner et al., 2004; Erickson, 2005; Chinsamy-Turan, 2005; Makovicky et al., 2007; Klein and Sander, 2007; Erickson et al., 2009; Erickson and Druckenmiller, 2011). Bone histological analysis is the best available method to answer questions regarding ceratopsian growth and physiology, because we cannot observe these taxa in life. The size of the skeleton can provide insight as to the age of the animal, but size does 3 not always vary with age, so size-independent criteria such as bone histology are better for assessing ontogenetic stages (Johnson, 1977). Bone deposits lines of arrested growth (or LAGs) that reflect periodicities in growth created by environmental modifications of endogenous rhythms (Francillon- Vieillot et al., 1990; de Ricqles et al., 1991; Castanet et al., 1993; Castanet et al., 2004). Research on extant animals demonstrates the existence of annual growth lines, which can be used with confidence to determine the ages of the animals, in a number of living species (Castanet et al., 1977; Castanet, 1978; Pascal and Castanet, 1978). LAGs in fossil animals, therefore, are also inferred to be annual (Peabody, 1961). Furthermore, the bone thickness between adjacent LAGs can inform us about these animals' bone growth rate, which allows for direct comparison to growth rates of extant vertebrates (Erickson et al., 2001; Padian et al., 2001; Lee and Werning, 2008). Unlike growth rings in trees, sampled bones do not always preserve all of the LAGs originally laid down, because they "remodel" themselves through resorption, so the early life history of the animal is often lost. These processes are presumed to be involved in the turnover of calcium and phosphorus in metabolism (Amprino 1967), where remodeling plays a direct and fundamental part in the growth of long bones (de Ricqles, 1980). The result is that some areas along the length of a bone contain a mosaic of microstructures that have different ontogenetic histories (Enlow, 1963). This remodeling means that a single limb bone lacks a complete life history record (Horner et al., 1999), but successive partial records of limb bones in an ontogenetic series are sufficient to reconstruct the history of bone growth (Chinsamy, 1990, 1993; Curry, 1999; Erickson and Tumanova, 2000; Horner et al., 2000; Sander, 2000; Erickson et al., 2009; 4 Horner and Padian, 2004; Bybee et al., 2006; Klein and Sander, 2007; Lee, 2007a,b,c; Makovicky et al., 2007; Reizner, 2010; Erickson and Druckenmiller, 2001). LAGs are not the only osteohistological data available for inferring growth trajectory. Vascular canal orientation and density within the bone cortex also relate to how rapidly animals grow. If the vascularity is dense, it indicates that numerous blood vessels were running through the bone and that the bone tissue was developing rapidly (de Margerie et al., 2002, 2004). Longitudinal canals are usually the most abundant vascular canals seen in a thin section and are circular in cross section (de Margerie et al., 2002). If vascular canals are oriented radially, it is indicative of fast relative growth (de Margerie et al., 2002, 2004). If vascular canals are oriented circumferentially, it is indicative of relative slow growth (de Margerie et al., 2004). The density of osteocyte lacunae (spaces for bone cells) is also useful data, because higher densities of osteocytes indicate faster growth (de Margerie et al., 2002) Previous Studies Limb bone histology and inferred growth curves have previously been investigated for smaller, bipedal ceratopsians Psittacosaurus mongoliensis (Erickson and Tumanova, 2000) and Psittacosaurus lujiatunensis (Erickson et al., 2009), small quadrupedal basal ceratopsians Protoceratops andrewsi (Lee, 2007; Makovicky et al., 2007), and the large quadrupedal centrosaurines Centrosaurus apertus (Lee, 2007a,b,c), Pachyrhinosaurusperotorum (Erickson and Druckenmiller, 2011, Fiorillo and Tykoski, 2012), and Einosaurusprocurvicornis (Reizner, 2010), but chasmosaurine ceratopsid dinosaurs have yet to be investigated. Small, basal forms (e.g., Psittacosaurus 5 6 mongoliensis, Psittacosaurus lujiatunensis, and Protoceratops andrewsi) show moderate rates of growth with a high number of LAGs (Erickson and Tumanova, 2000; Makovicky et al., 2007; Erickson et al., 2009). In contrast, large, quadrupedal centrosaurines (e.g., Centrosaurus apertus, Pachyrhinosaurus perotorum, and Einosaurus procurvicornis) show rapid growth with a small to moderate number of LAGs. Goals of Present Study I used osteohistological analysis of the chasmosaurines Utahceratops gettyi and Kosmoceratops richardsoni to provide insight into how these chasmosaurine ceratopsian dinosaurs grew and how this might relate to their thermal physiology. Placed in context with other previously sampled taxa, these data help evaluate how growth changed during the evolution of ceratopsians, and how these animals grew in comparison with other dinosaurs of similar size. This study analyzes for the first time the limb bone histology of chasmosaurine dinosaurs Utahceratops and Kosmoceratops, and reevaluate and compare the limb bone histology of Psittacosaurus mongoliensis, Protoceratops andrewsi and Centrosaurus apertus to the Utah forms to gain insight on the physiology, evolution and behavior of these animals. Utahceratops and Kosmoceratops are ideal for histological analysis because they are ceratopsids from the southern latitudes that provide good comparisons to their northern counterparts in Montana, Canada and Alaska, and are a test of the growth trajectories of co-occurring taxa in a single basin (cf., Fowler, 2011). Furthermore, known specimens of Utahceratops range from juvenile to adult, allowing me to reconstruct a nearly complete post-natal ontogenetic record of the species. GEOLOGIC SETTINGS Laramidia For approximately 27 million years of the Late Cretaceous (95-68 Ma), regional tectonics produced the Cretaceous Western Interior Seaway (KWIS), a shallow epeiric sea that flooded the central portion of North America, dividing the continent into eastern and western landmasses known as Appalachia and Laramidia, respectively (Kauffman, 1984; Roberts and Kirschbaum, 1995). The western landmass, Laramidia, was less than 20% the size of present-day North America at its maximum extent (Lehman 1997; Sampson et al., 2010a). Laramidia was occupied by a diverse assemblage of dinosaurs and other nonmarine vertebrate taxa (Gates et al., 2010; Sampson et al., 2010a,b). Among dinosaurs, the same major clades are present in the northern and southern part of Laramidia (e.g., hadrosaurids, ceratopsids, ankylosaurids, tyrannosaurids, ornithomimids), but the assemblages appeared largely distinct at the genus and species levels (Lehman, 1997, 2001; Gates et al., 2010; Sampson et al., 2010a,b). Some of the best information about southern Laramidian dinosaurs has emerged in the past 10 years from late Campanian sediments of southern Utah, including two new chasmosaurine ceratopsid dinosaurs, Utahceratops and Kosmoceratops (Eaton, 1999; Zanno and Sampson, 2005; Gates and Sampson, 2007; Sampson et al., 2010a,b,c; Gates et al., 2010; Getty et al., 2010). All specimens of Utahceratops and Kosmoceratops in this study were discovered in Grand Staircase-Escalante National Monument in southern Utah. Grand Staircase- Escalante National Monument (GSENM) (Fig. 1) encompasses 1.9 million acres of rugged terrain in southern Utah, and was the last major region within the contiguous United States to be mapped topographically (Foster et al., 2001). Formally designated in 1996, the Monument was established in large part to facilitate preservation and study of its diverse natural resources, both living and fossil (Designated by Presidential Proclamation on September 18, 1996, pursuant to the Antiquities Act of 1906). The most fossiliferous terrestrial geologic unit in GSENM is the Upper Cretaceous Kaiparowits Formation, deposited along the eastern margin of Laramidia within 100 km of the seaway (Roberts & Kirschbaum, 1995) in the Kaiparowits Basin, and now exposed on the Kaiparowits Plateau (Eaton, 1991). The Kaiparowits Formation (Fig. 2) is an unusually thick, ~860 m, package of Upper Cretaceous (late Campanian) strata exposed in Grand Staircase-Escalante National Monument of southern Utah, USA (Roberts, 2007). It is easily recognized by its distinctive, badland-forming blue-gray sandstones and mudstones, which are in stark contrast to the typical tan sandstones of the underlying early Campanian Wahweap and late Turonian-Santonian Straight Cliffs formations, and the overlying maroon conglomerates of the Maastrichtian Canaan Peak Formation (Roberts et al., 2005). This formation is part of a prograding clastic wedge that deposited vast quantities of sediment derived from sources in the Sevier orogenic belt, thrust sheets in southeastern Nevada and southern California, and the Mogollon slope in southwestern Arizona into the syn-evolving Sevier foreland basin (Goldstrand, 1992; Lawton et al., 2003; Roberts, 2007). 8 Kaiparowits strata represent a muddy-to-sand meandering fluvial system with paleoflow to the east and northeast in a relatively warm, humid paleoclimate (Eaton 1991; Goldstrand 1990, 1991, 1992; Little, 1995; Roberts et al., 2003, 2005, Roberts, 2007). The compositions of the sandstone indicate felsic volcanic, siliciclastic, metamorphic, and plutonic sources (Goldstrand, 1992). Radioisotopic dating of four bentonite horizons (Roberts et al., 2005) produced a late Campanian age of 76.6 - 74.5 Ma for the Kaiparowits Formation (Roberts et al., in press). The Kaiparowits Formation has among the highest sediment accumulation rates recorded in the Western Interior Basin (WIB) at 41 cm/ka (Roberts et al., 2005). Using the accumulation rate of 41 cm/ka calculated using these radioisotopic dates, the ca. 860-m-thick Kaiparowits Formation accumulated for ca. 2.1 Ma, from ca. 76.6 - 74.5 Ma (Roberts et al., in press). The Kaiparowits Formation is informally subdivided into three units (lower, middle, upper), based on distinct changes in alluvial architecture (Roberts et al., 2005). Nearly all of the fossils found in this area are found in the lower half of the formation with the age range of ~76.46 +/_014 Ma and ~75.51+/_015 Ma (Roberts et al., 2005; Roberts et al., in press). Many of the most fossiliferous vertebrate-bearing Campanian formations in the WIB are penecontemporaneous with the Kaiparowits Formation (Roberts et al., 2005). Specifically, the Kaiparowits Formation is partially coeval Dinosaur Park Formation, portions of the Judith River, the Two Medicine, Fruitland, and possibly Aguja formations (Goodwin and Deino, 1989; Eberth and Hamblin, 1993; Rogers et al., 1993; Fassett and Steiner, 1997; Rogers, 1994; Roberts et al., 2005; Jinnah et al. 2009). 9 10 Localities for specimens in this study UMNH VP locality 942 This site preserves partial remains of at least three individuals of Utahceratops, recognized by distinct size classes of recovered elements (Getty et al., 2010). The specimens were buried in a lag deposit at the base of a sandy channel, in a thin, coarse, pebbly conglomerate. The pebble conglomerate indicates that this material may be reworked from a previous, possibly larger depositional event (Getty et al., 2010). The skeletons were completely disarticulated and demonstrate characteristics consistent with considerable pre-depositional transport, including winnowing of most small elements and breakage and surface abrasion of preserved elements (Getty et al., 2010). UMNH VP locality 145 This locality preserves the skeletal remains of an individual disarticulated ceratopsian skeleton, the holotype of Utahceratops (UMNH VP 12198) (Sampson et al., 2010a). 280 individual elements and associated fragments were discovered spread over an area of approximately 29 square meters (Getty et al., 2010). The majority of missing bones were smaller-sized appendicular and axial elements such as distal phalanges, vertebrae, and chevrons, indicating that these parts of the body were scavenged or hydraulically removed (winnowing) prior to final deposition (Getty et al., 2010). The specimen is preserved in a fine-grained floodplain environment. Very limited evidence of sub-aerial weathering was observed and was only present on axial elements, whereas the prevalence of bone decomposition on the majority of the skeleton is indicative of significant sub-aqueous exposure of the skeleton in a pond environment prior to burial 11 (Getty et al., 2010). The subaqueous burial interpretation is supported by the influx of fine grained pond sediments, where the grain size decreases from muddy sandstone basally to a muddy siltstone (encasing the bones) and a capping silty claystone stratigraphically up-section (Getty et al., 2010). UMNH VP locality 945 UMNH VP locality 945 preserves the remains of a nearly completely articulated subadult ceratopsian skeleton (UMNH VP 20444), a partial disarticulated ceratopsian skeleton (UMNH VP 20454), and a partially articulated alligatoroid crocodylian skeleton, about 1 m in length (Irmis et al., in press), which appears to have been deposited on top of the articulated ceratopsian carcass prior to burial (Getty et al., 2010). The ceratopsian skeletons are most likely assignable to Utahceratops based on available prepared cranial remains from the articulated specimen. UMNH VP 20444 represents the most complete articulated skeleton of any animal found to date in a fine-grained facies (siltstone and mudstone) from the Kaiparowits Fm (Getty et al., 2010). Generally, articulated specimens in the Kaiparowits Fm. are associated with rapid burial in channel sandstone facies (Getty et al., 2010), so this well-preserved specimen in mudstone is unusual. UMNH VP locality 512 The ceratopsian skeleton at this site was disarticulated, but closely associated, and mostly predepositionally broken. UMNH VP 16865.1 is an unidentified ceratopsian perhaps being Utahceratops, but has to be identified with further investigations. "Water-rot" is present on the spongy parts of elements, suggesting a paludal environment. The 12 surrounding matrix contains degraded plant material, mostly carbonized bits of stems, and leaves. The matrix is gray-green, again supporting a water-logged environment, and is mostly homogenous siltstone and mudstone, with very little sand (only in thin lenses, not very laterally extensive) (E. Lund, personal communication). The sandy siltstone layers include iron concretions, small amounts of organic material and some gastropods (D. Brandau, personal communication). UMNH VP locality 890 This partial skeleton, UMNH VP 17000, is the holotype of Kosmoceratops. The specimen consists of a relatively complete skull, a significant portion of the axial skeleton from neck to tail, including part of the pelvic girdle, and at least part of one limb (Getty et al., 2010; Sampson et al., 2010a). The rest of the limbs and distal tail may have been lost either to scavenging or to rotting of the carcass prior to its deposition (Getty et al., 2010). This specimen was found in a silty sandstone channel facies and appears to represent an individual animal carcass that had been washed into a river channel and buried quickly (Getty et al., 2010). UMNH VP locality 1323 The ceratopsian material found at this site, UMNH VP 21339, is a single sub-adult to adult disarticulated individual of Kosmoceratops. Almost every element shows predepositional breakage. The matrix is stacked siltstones, and mudstones, with minor sandstones (E. Lund, personal communication). The matrix is suggestive of a pond 13 environment (E. Lund, personal communication). Several varieties of gastropods and bivalves are present, many of which still posses the original aragonite shells. 14 Figure 1. Map of Grand Staircase-Escalante National Monument. The Kaiparowits Formation is highlighted in green. The localities where the sampled specimens were found are indicated by the labeled black circles. Map modified from Roberts et al. (2005). 15 16 Figure 2. Stratigraphic column of the Kaiparowits Formation (Modified from Sampson et al., 2010) with sampled specimen localities indicated. The yellow circle indicates an Utahceratops specimen, which was not sampled but represents the extent of time in which Utahceratops lived. Indeterminant humerus UMNH VP 19490 is excluded from figure because of stratigraphic uncertainty. 17 MATERIALS AND METHODS Specimens I sampled four individuals from various stages of ontogeny from multiple bonebeds containing Utahceratops gettyi. Utahceratops is diagnosed as a chasmosaurine ceratopsid dinosaur by possessing the following autapomorphies: nasal horncore caudally positioned, almost entirely behind external naris; supraorbital horncores short, robust, dorsolaterally directed, and oblate in shape with blunt tip; episquamosals on mid portion of lateral frill margin low and extremely elongate and median portion of transverse bar of parietal rostrally curved. Utahceratops is further distinguished by mid-frill episquamosals with extremely elongate bases (some > 10 cm) (Sampson et al., 2010a). The Bluewash Bonebed, UMNH VP LOC 942, preserves partial remains of at least three individuals of Utahceratops. For this study, I sectioned three elements, a right tibia (UMNH VP 16681), a right femur (UMNH VP 16860), and a partial shaft from an unidentified limb bone (UMNH VP 16861) from this locality (Fig. 3) (Table 1). Because this is a bonebed filled with disarticulated skeletons, distinguishing which bone came from which individual is difficult. UMNH VP LOC 145 preserves the skeletal remains of an individual disarticulated ceratopsian skeleton of a referred specimen of Utahceratops (UMNH VP 12198) (Sampson et al., 2010a). 19 For this study, I sectioned a left femur, a rib fragment and a tendon fragment. UMNH VP locality 945 preserves the remains of a nearly completely articulated subadult ceratopsid skeleton (UMNH VP 20444) and a partial disarticulated ceratopsid skeleton (UMNH VP 20454), both referable to Utahceratops. From the articulated individual, I sectioned a left humerus (UMNH VP 20444.1), a left ulna (UMNH VP 20444.2), a left femur (UMNH VP 20444.4), and a right tibia (UMNH VP 20444.5). From the disarticulated individual, I sectioned a left humerus (UMNH VP 20454.8), a left ulna (UMNH VP 20454.1), a left femur (UMNH VP 20454.5), a right tibia (UMNH VP 20454.3), a rib fragment (UMNH VP 20454.7) and a tendon fragment (UMNH VP 20454.9). Two additional humeri were sectioned from two additional localities. Specimen known as UMNH VP 16865.1 is a left humerus from a single disarticulated individual of an indeterminate ceratopsid, perhaps Utahceratops, from UMNH VP locality 512. Specimen UMNH VP 19490 represents an isolated left humerus of an indeterminate ceratopsid. In addition to sampling Utahceratops, I sampled two femora from Kosmoceratops. Kosmoceratops is diagnosed as a chasmosaurine ceratopsid dinosaur by possessing the following autapomorphies: internal naris rostrocaudally abbreviated and caudodorsally inclined; nasal horncore transversely constricted, long-based, and blade­like, with flattened distal portion; supraorbital horncores dorsolaterally directed proximally, with a ventral curvature distally tapering to a point; parietosquamosal frill relatively short and broad (maximum width ~2 times maximum length), with small, caudally positioned parietal fenestrae; parietosquamosal frill with ten well developed processes on caudal margin composed on each side of three procurved epiparietals (ep1- 20 3), one procurved process on the parietosquamosal contact (esp), and one laterally to rostrolaterally directed episquamosal (es1) (Sampson et al., 2010a). Kosmoceratops is further distinguished by a total of 15 well developed horns or horn-like epiossifications (1 nasal horncore, 2 postorbital horncores, 2 epijugals, and 10 well-developed episquamosals and epiparietals), which makes it one of the most ornate skulls of any known dinosaur (Sampson et al., 2010a). UMNH VP 17000, a partial skeleton that is the holotype of Kosmoceratops from UMNH VP locality 890, consists of a relatively complete skull, a significant portion of the axial skeleton from neck to tail, including part of the pelvic girdle and at least part of one limb (Getty et al., 2010; Sampson et al., 2010a). I sampled the femur from two separate midshaft fragments. I also created a thin section from a femur UMNH VP 21339 (LOC 1323) from a new referred specimen of Kosmoceratops. UMNH VP locality 1323 consists of a femur associated with a disarticulated skull. Preparation of Histological Samples All specimens were cleaned, repaired, and prepared prior to study. Before creating thin sections, all of the bones were measured, photographed, molded and casted. Research-grade casts were made of each element prior to destruction to ensure all data of each element was preserved. The measurements of each bone sectioned were made using either digital calipers or a cloth metric measuring tape where necessary. The measurements taken include: the preserved length, the width and depth of the proximal end, the width and depth of the distal end, and the width, depth and circumference at the midshaft where the bone was to be sectioned (Fig. 4) (Table 2). Photographs of the bones prior to sectioning were taken using an Olympus Stylus 1200 12 megapixel digital camera. Prior to molding, the consolidant "Vinac" (a mixture of acetone and polyvinylacetate beads) was applied to the bones for protection and for easy removal from the putty. To create a mold of small bones such as ribs and tendons, Douglas & Sturgess Silpoxy Putty was used as the molding medium. The putty was placed onto wax paper and then the bone was pressed into the putty. Putty was pressed against the bone to ensure no gaps were present. While the putty was still soft, triangle shaped registration marks were indented in the putty. Vinac was then applied to the bottom part of the mold to act as a separator, and approximately 10 minutes was allowed for the vinac to dry. A ball of putty was placed on top of the bottom mold with the bone still inside, and then pressed all of the way down around the bottom mold all of the way down to the wax paper, creating the top part of the mold. After the putty cured, the two sections were de­molded. The mold is done at this point. The bones I molded using this method were rib UMNH VP 20454.7, tendon UMNH VP 20454.9, rib UMNH VP 12198, and tendon 12198. The larger molds for this project were done in the following manner. "Klean Klay" of a medium hardness was rolled out until it is about 1 cm thick. The bone was placed on top of the clay and the shape of the bone is outlined with a knife. The clay underneath where the bone is sitting was then removed. The bone was then placed in the hole, and the clay was then pushed up to the bone so that the clay is touching the bone itself. Excess clay was rolled into the shape of a cone and placed at one edge of the clay layout making sure the narrow end of the cone is touching the bone, creating a "pour 21 spout" for pouring in the eventual casting resin. Using a tool with a thin wire in the shape of a half circle at the end, a moat-like registration groove was cut into the clay all around the bone about a centimeter away. The desired amount of silicone (Mold Max 20 silicone rubber) was poured into a plastic tub (like a butter tub), mixed, and spread over the bone and on top of the clay. This is the first layer of the silicone. To strengthen the silicone mold, nylon panty hoes were cut up into ~ 5 cm by 5 cm squares and were placed into the wet silicone. At least two hours were allotted for the silicone to dry. Another batch of thin silicone created in the same manner was poured over the nylon pieces. This, too, dried from two hours. A thick layer of silicone was made with the addition of a thickening agent called "Thixotropic". The thick mixture was put on top of the other thin silicone. At least three hours were allotted for this layer to set. A "mother mold" for supporting the silicone mold was made using plaster and fiberglass. First, petroleum jelly was rubbed all over the silicone as a separating agent to ensure that the plaster does not permanently stick to the silicone mold. Pieces of various sizes were cut from closely woven sheets of fiberglass, dipped in wet plaster, and put on top of the silicone. This step was repeated until five layers of fiberglass and plaster were on the mold. When the plaster was dry, the whole mold was flipped over so that the clay was exposed. All of the clay was removed. A rasp was used to smooth out the mold. To create a separator between the silicone halves, "vinac" was used. Once that was dry, the steps for application of silicone and creation of a mother mold were followed again. Once this dried, holes were drilled through both layers of the plaster approximately three centimeters from the edge. The whole mold was knocked on a hard surface to separate the mother mold, and the silicone mold from one 22 side was carefully peeled away from the bone. Once the mold was made, a cast of the original bone was created. The mold was sprayed with mold release so that the casting resin did not stick to the silicone rubber. Bolts with washers and wing-nuts were screwed in the drilled holes to secure the mold together. The mold was then positioned vertically with the pour spouts pointing up. Polyester resin (TC-808") was mixed; working time with this resin is short as it cures very rapidly, in about 2 minutes. The mixture was poured into the mold via the pour spout. The whole mold was knocked against a hard surface to release any air bubbles. At least ten minutes was allotted before the newly made cast was removed. Once the cast was removed from the mold, a "Dremel" rotary tool was used to smooth out flashing along the mold seam. To section the bone, a water-cooled Felker 41-AR tile saw with Norton diamond blades to cut the element in half at the midshaft. A transverse section of bone no thicker than 2.2 cm was then cut from one of the bone halves. All exposed surfaces were then treated with Vinac consolidant and allowed to dry. This transverse section was then embedded in Silmar 95BA-41, a polyester resin. The bone section was placed into a hard plastic container that has been sprayed with mold release. Mixed resin was then poured into in the container with the bone. This container was then placed in an ABBESS Instruments, INC vacuum chamber and brought to the pressure of -72 Kpa (-23 inches of mercury) (with a Fischer High Vacuum Pump LAV-3 vacuum pump). The vacuum pump was then stopped and the chamber was kept at this pressure for three to five minutes; afterwards, the vacuum chamber was slowly released and brought back to room air pressure. The embedded specimen was then removed from the chamber left to fully 23 harden for twenty-four hours, and finally removed from the container. Initially, the resin block was cut in half with the tile saw, and then a thinner section was cut as thinly as possible from half of the embedded resin block. These sections were dried completely and then a thin cyanoacrylate glue, like PALEOBOND™ Penetrant/Stabilizer, was applied to provide support to the specimen, and allowed to dry again. Prior to mounting to a slide, this thinner section required ‘premount grinding' so that it could be glued onto a glass slide. This grinding was conducted using various coarsenesses of silicone carbide grit powder, water, and a glass plate. The section was successively ground using decreasing coarseness of 120, 320, and 600 grit powders. Each powder was applied to a different glass plate and water was sprayed on for lubrication. The section was placed flat on this water/powder mixture and was moved in an "infinity" symbol to make all sides received the same amount of grinding. Several sections made at the Museum of the Rockies used the same grit powders, but exchanged an Ecomet 4 Grinder-Polisher (Buehler Ltd.) for the glass plate. Once the specimens had been premount ground, they were mounted onto a glass slide using Devcon™ Two-Ton Epoxy. Two methods were utilized to reduce movement of the specimen during gluing. A weight (a cap filled with pennies embedded in resin) was put on top of the specimen. Alternatively, the section and glued slide were wrapped in wax paper, two extra slides were put on either side to "sandwich" it, and then each side was clamped with wooden clothes pins. The glue was left to dry for 12 hours. Once the large specimens were mounted, they were further thinned using the water-cooled tile saw. The slide was placed against the metal "bridge vice" about 3 mm away from the blade path, and slowly moved against the blade. The newly exposed bone 24 was allowed to dry completely, then glued again with cyanoacrylate PALEOBOND™ Penetrant/Stabilizer glue, and left to dry again. 1x2" and 2x3" slides were thinned using the Buehler Petro-Thin Sectioning System, with a 60 grit disk. The attached arm had a vacuum that allowed the slide to stay affixed and be moved against the disk. As the slide was brought closer and closer to the revolving disk, more and more of the specimen was shaved off until the slide reached the desired thickness. Two methods were used in this study to grind and polish the thin sections. The first, which was used in the Museum of the Rockies Histology Lab, used silicon carbide abrasive papers (Buehler Ltd.) and an Ecomet 4 Grinder-Polisher (Buehler Ltd.) lapidary wheel. The abrasive papers used were grits 60, 120, 180, 320, 600, and 800. I put the desired grit paper on the wheel being sure to secure it with the removable metal ring, turned the motor up to 120 rpm, and turned the water on. The slide was ground using successively finer grit papers as the section got thinner. I was sure to constantly check the thin section thickness with a micrometer. The second method of grinding at the University of Utah was used for most of the sections in this study, and replaced the Ecomet with grinding using glass plates, silicon carbide grit powders (120, 180, 320, 600, and 800), and water, as described above in the "premount grinding" step. As with the Ecomet, I successively ground down each section using decreasing coarsenesses of grit powder. At the MOR final polishing of each section used a paste on a wet platen consisting of about a tablespoon Buehler 5.0 micron aluminum oxide powder. The thin section was rubbed on the platen in a circular motion or the paste was rubbed onto the slide using a wet micro cloth. After this, a Buehler 1.0 micron aluminum oxide powder 25 solution was rubbed in a circular motion onto the slide until shiny. At the University of Utah, the slide was ground using very fine 600 and 800 silicon carbide grits until shiny. After polishing, a glass cover slip was permanently glued to each slide with Devcon™ Two-Ton Epoxy. These finished thin sections were then digitally imaged using a NIKON Optiphot Pol Microscope, PRIOR Optiscan II Automated Stage, a DELL Precision T3400 Computer, a NIKON DS-Fil camera, and NIKON NIS Elements BR 3.0 software. I scanned my slides at 4x magnification, manually focusing every 10 shots. Analysis of Microstructure The description of the bone histology of each element begins with large-scale features and works through to smallest-scale features. Large scale features include the nature of the medullary cavity and the extent of trabeculae. Then, because remodeling and secondary osteons were so prevalent, I focused on their extent and location within the thin section. Finally, the small scale features were divided into characters that do and do not change throughout the section. I specifically looked at extent of woven vs. parallel-fibered bone in crossed-polarized light, and how this changed from the medullary cavity to the periosteum. I evaluated the density and direction of vascular canals throughout the section, and described the proportion of simple canals vs. primary osteons vs. secondary osteons. All histological features are described starting from medullary cavity and working my way outwards towards the periosteum. I noted any evidence (or lack of) for Lines of Arrested Growth (LAGs), and External Fundamental System (EFS). If LAGs were present, I measured their circumference, counted how many there were, and 26 measured their spacing. I quantitatively calculated the density of both osteocytes and vascular canals (Table 3). In order to calculate the density of both the vascularity and the osteocytes, I focused on a standard column of bone from the anterior part of the midshaft. In Adobe Illustrator CS5.1, I then made uniform boxes of .6009 mm by .6009 mm in size and spaced them 0.5577 mm distance apart from each other. The size and spacing of the box was uniform for every slide analyzed (Fig. 5). I then counted the amount of longitudinal, circumferential, and radial canals (Fig. 6) present within each box for each slide. I did this throughout the sectioned column. I calculated the density of the osteocytes in the same way. I plotted these densities versus the radius of each box from the centroid of the element using Microsoft Excel Mac 2008. I created graphs for all of the femora I analyzed plotting radius vs. vascularity, radius vs. osteocytes, and vascularity vs. osteocytes (Fig. 7). I created graphs for all of the tibiae I analyzed plotting radius vs. vascularity, radius vs. osteocytes, and vascularity vs. osteocytes (Fig. 8). I created graphs for all of the humeri I analyzed plotting radius vs. vascularity, radius vs. osteocytes, and vascularity vs. osteocytes (Fig. 9). 27 28 Figure 3. Skeletons of Utahceratops gettyi and Kosmoceratops richardsoni. Red elements represent the bones sectioned for this study (modified from Sampson et al., 2010a). 29 Table 1. Elements sampled in this study. Specimen Species Humerus Ulna Femur Tibia Rib Tendon Limb Shaft UMNH VP 20444 Utahceratops X X X X UMNH VP 20454 Utahceratops X X X X X X UMNH VP 12198 Utahceratops X X X UMNH VP 16800 Utahceratops X X X UMNH VP 17000 Kosmoceratops X UMNH VP 21339 Kosmoceratops X UMNH VP 19490 Indeterminant X UMNH VP 16865 Indeterminant X 31 Figure 4. The outlines of the limb bones that were sampled and the areas where they were measured. A. Outline of a femur; B. Outline of a tibia; C. Outline of an humerus; D. Outline of an ulna. The black line labeled L indicates the full length measurement. The blue line labeled ‘pw' indicates the width of the distal end. The red line labeled ‘msw' indicates the midshaft width, where the circumference measurement was taken, and where the bone was sectioned. The purple line labeled ‘dw' indicates the distal end width measurement. The ‘P' arrow indicates the proximal orientation and the ‘L' arrow indicates the lateral orientation. 32 Table 2. Bone measurements before sectioning. SPECIMEN # LOC# NAME SPECIES ELEMENT PRESERVED LENGTH mm PROX WIDTH mm PROX DEPTH mm DIST WIDTH mm DIST DEPTH mm MIDSHAFT WIDTH mm MIDSHAFT DEPTH mm DISTANCE MIDSHAFT FROM CIRCUM DISTAL END mm mm UMNHVP 20444.1 945 KC&C UTAH HUMERUS LF 440.0 144.1 175.0* 44.5 131.9 42.5 55.2 52.3 160.0 150.0 UMNHVP 20444.4 945 KC&C UTAH FEMUR LF 605.0 . . . . . . 129.1 . . . 60.0 . . . . . . 245.0 UMNHVP 20444.5 945 KC&C UTAH TIBIA RT 465.0 136.4* . . . 148.6 . . . 54.4 . . . 189.0 240.0 UMNHVP 20444.2 945 KC&C UTAH ULNALF 322.0 111.4 27.2 111.1 46.2 61.1 22.9 55.1 27.9 136.0 120.0 UMNHVP 20454.8 945 KC&C UTAH HUMERUS LF 210.3 . . . 80.2 . . . 50.5 ___ 145.1 . . . UMNHVP 20454.5 945 KC&C UTAH FEMUR LF 291.0 97.8* 53.1* 118.8* 55.3* 80.9* 47.5* 230.0 . . . UMNHVP 20454.3 945 KC&C UTAH TIBIA RT 231.0 56.4* 73.9* 87.8 38.5 50.6 55.5 167.0 . . . UMNHVP 20454.1 945 KC&C UTAH ULNALF 321.0 93.1 34.7 64.6 25.4 50.5 22.7 121.0 110.0 UMNHVP 20454.7 945 KC&C UTAH RIB FRAG 104.4 31.9 . . . 27.4 . . . 31.4 . . . 75.0 . . . UMNHVP 20454.9 945 KC&C UTAH TENDON FRAG 92.8 17.0 . . . 14.9 . . . 14.1 . . . 40.0 . . . UMNHVP 12198 145 THE BLUES UTAH FEMUR LF 874.0 173.9* 90.0* 248.9* 96.4* 80.9* 73.2* 437.0 550.0 UMNHVP 12198 145 THE BLUES UTAH RIB FRAG 114.4 23.1 . . . 22.6 . . . 23.0 . . . 58.0 . . . UMNHVP 12198 145 THE BLUES UTAH TENDON FRAG 68.9 17.0 . . . 12.7 . . . 14.9 . . . 44.0 . . . UMNHVP 16681 942 BLUEWASH UTAH TIBIA RT 450.0 159.2 56.7 149.7 40.1 69.8 55.4 205.0 222.0 UMNHVP 16860 942 BLUEWASH UTAH FEMUR RT 488.0 390.3* 56.7 . . . . . . 95.4 58.5* 250.0 324.0 PROX UMNHVP 16861 942 BLUEWASH UTAH LIMB SHAFT 180.0 . . . . . . . . . . . . . . . . . . 180.0 . . . UMNHVP 19490 VARIOUS UTAH HUMERUS LF 167.6* . . . . . . . . . . . . 54.3 68.5 204.0 . . . UMNHVP 16865.1 512 KAC UTAH HUMERUS LF 545.0 131.0 83.6 152.4 47.1 61.9 56.7 189.0 182.0 UMNHVP 17000 890 KOSMO KOSMO FEMURFRAGA 118.8 68.5 . . . 66.0 . . . 72.2 . . . 211.0 . . . UMNHVP 17000 890 KOSMO KOSMO FEMUR FRAG B 80.9* . . . . . . . . . . . . 70.0 72.6 223.0 . . . UMNHVP 21339 1323 HEC 11-3 KOSMO FEMURFRAGA 670.0* 315.0 . . . . . . . . . 155.0 . . . . . . 440.0 PROX UMNHVP 21339 1323 HEC 11-3 KOSMO FEMUR FRAGB . . . . . . . . . . . . . . . . . . . . . . . . 440.0 PROX Table 3. Data from the analysis of the thin sections. This table includes the number and type of vascular canals seen in each element, and the change in vascular densities and osteocytes densities for each bone. Vascularity Osteocytes Specimen # Box # Total Longitudinal Radial Circumferential Osteocytes Ratio L:R Radius /mm2 /mm2 UMNH VP 16860A 1 14 14 0 0 633 - 5.15 38.77 1753.07 femur Utahceratops 2 3 3 0 0 423 - 6.31 8.31 1171.48 gettyi 3 7 7 0 0 653 - 7.46 19.39 1808.46 4 8 8 0 0 544 - 8.62 22.16 1506.59 5 12 11 1 0 800 11.00 9.78 33.23 2215.57 6 7 7 0 0 800 - 10.94 19.39 2215.57 7 13 11 0 2 868 - 12.10 36.00 2403.89 8 11 10 1 0 674 10.00 13.26 30.46 1866.62 9 10 10 0 0 600 - 14.42 27.69 1661.68 10 12 10 2 0 744 5.00 15.57 33.23 2060.48 UMNH VP 20444.4 1 3 3 0 0 127 - 18.27 8.31 351.72 femur Utahceratops 2 3 3 0 0 358 - 19.43 8.31 991.47 gettyi 3 3 3 0 0 450 - 20.59 8.31 1246.26 4 3 3 0 0 375 - 21.75 8.31 1038.55 5 5 5 0 0 344 - 22.91 13.85 952.70 6 3 3 0 0 400 - 24.07 8.31 1107.79 7 3 3 0 0 400 - 25.23 8.31 1107.79 8 3 3 0 0 451 - 26.38 8.31 1249.03 9 2 1 1 0 400 1.00 27.54 5.54 1107.79 10 4 4 0 0 400 - 28.70 11.08 1107.79 11 5 3 2 0 432 1.50 29.86 13.85 1196.41 12 7 7 0 0 357 - 31.02 19.39 988.70 13 5 5 0 0 400 - 32.18 13.85 1107.79 14 4 4 0 0 400 - 33.34 11.08 1107.79 4*. Table 3 Continued. Vascularity Osteocytes Specimen # Box # Total Longitudinal Radial Circumferential Osteocytes Ratio L:R Radius /mm2 /mm2 15 4 4 0 0 548 - 34.49 11.08 1517.67 16 2 2 0 0 500 - 35.65 5.54 1384.73 17 3 3 0 0 570 - 36.81 8.31 1578.59 UMNH VP 20454.5 1 5 5 0 0 500 - 13.06 13.85 1384.73 femur Utahceratops 2 16 16 0 0 490 - 14.22 44.31 1357.04 gettyi 3 6 6 0 0 400 - 15.38 16.62 1107.79 4 6 6 0 0 433 - 16.54 16.62 1199.18 5 13 13 0 0 460 - 17.70 36.00 1273.95 6 4 4 0 0 361 - 18.86 11.08 999.78 7 15 15 0 0 400 - 20.01 41.54 1107.79 8 15 15 0 0 400 - 21.17 41.54 1107.79 9 11 11 0 0 367 - 22.33 30.46 1016.39 10 18 18 0 0 458 - 23.49 49.85 1268.41 UMNH VP 12198 femur 1 2 2 0 0 346 - 24.28 6.78 1172.62 Utahceratops gettyi 2 3 3 0 0 500 - 25.36 10.17 1694.53 3 2 2 0 1 550 - 26.45 6.78 1863.99 4 3 3 0 0 470 - 27.54 10.17 1592.86 5 3 3 0 0 400 - 28.62 10.17 1355.63 6 4 4 0 0 420 - 29.71 13.56 1423.41 7 3 3 0 0 500 - 30.80 10.17 1694.53 8 3 3 0 0 500 - 31.88 10.17 1694.53 9 5 5 0 0 600 - 32.97 16.95 2033.44 10 9 8 0 1 600 - 34.06 30.50 2033.44 11 7 7 0 0 770 - 35.14 23.72 2609.58 12 7 7 0 0 485 - 36.23 23.72 1643.70 UMNH VP 20444.5 tibia 1 3 3 0 0 500 - 12.19 8.31 1384.73 Utahceratops gettyi 2 3 3 0 0 500 - 13.35 8.31 1384.73 Table 3 Continued. Specimen # Vascularity Osteocytes Box # Total Longitudinal Radial Circumferential Osteocytes Ratio L:R Radius /mm2 /mm2 UMNH VP 20454.3 tibia Utahceratops gettyi 3 7 7 0 0 500 - 14.51 19.39 1384.73 4 3 3 0 0 700 - 15.67 8.31 1938.62 5 7 7 0 0 500 - 16.82 19.39 1384.73 6 0 0 0 0 500 - 17.98 0.00 1384.73 7 1 1 0 0 900 - 19.14 2.77 2492.52 8 8 8 0 0 900 - 20.30 22.16 2492.52 9 9 9 0 0 700 - 21.46 24.93 1938.62 10 3 3 0 0 700 - 22.62 8.31 1938.62 11 7 7 0 0 800 - 23.78 19.39 2215.57 12 5 5 0 0 700 - 24.94 13.85 1938.62 13 3 3 0 0 700 - 26.09 8.31 1938.62 14 0 0 0 0 730 - 27.25 0.00 2021.71 15 4 4 0 0 700 - 28.41 11.08 1938.62 16 0 0 0 0 800 - 29.57 0.00 2215.57 1 6 6 0 0 600 - 11.82 16.62 1661.68 2 16 16 0 0 600 - 12.98 44.31 1661.68 3 7 7 0 0 567 - 14.14 19.39 1570.29 4 11 11 0 0 740 - 15.29 30.46 2049.40 5 11 11 0 0 638 - 16.45 30.46 1766.92 6 11 11 0 0 700 - 17.61 30.46 1938.62 7 9 9 0 0 600 - 18.77 24.93 1661.68 8 9 9 0 0 600 - 19.93 24.93 1661.68 9 9 6 0 3 770 - 21.09 24.93 2132.49 10 11 8 0 3 750 - 22.25 30.46 2077.10 11 7 7 0 0 600 . 23.40 19.39 1661.68 OS Table 3 Continued. Specimen # Box # Total Longitudinal Radial Circumferential Osteocytes Ratio L:R Radius Vascularity /mm2 Osteocytes /mm2 UMNH VP 16681 tibia Utahceratops gettyi UMNH VP 16865. IB humerus Indeterminant ceratopsian 12 9 9 0 0 673 - 24.56 24.93 1863.85 13 0 0 0 0 600 - 25.72 0.00 1661.68 14 4 4 0 0 700 - 26.88 11.08 1938.62 1 3 3 0 0 82 - 11.82 8.31 227.10 2 5 5 0 0 183 - 12.98 13.85 506.81 3 12 12 0 0 325 - 14.14 33.23 900.08 4 12 12 0 0 320 - 15.29 33.23 886.23 5 15 15 0 0 574 - 16.45 41.54 1589.67 6 13 13 0 0 660 - 17.61 36.00 1827.85 7 13 12 0 1 660 - 18.77 36.00 1827.85 8 16 16 0 0 570 - 19.93 44.31 1578.59 9 20 20 0 0 500 - 21.09 55.39 1384.73 10 12 12 0 0 500 - 22.25 33.23 1384.73 11 20 20 0 0 470 - 23.40 55.39 1301.65 12 16 16 0 0 600 - 24.56 44.31 1661.68 13 14 14 0 0 500 - 25.72 38.77 1384.73 14 21 21 0 0 700 - 26.88 58.16 1938.62 15 13 13 0 0 700 - 28.04 36.00 1938.62 16 12 12 0 0 700 - 29.20 33.23 1938.62 17 4 4 0 0 650 - 30.36 11.08 1800.15 18 6 6 0 0 581 - 31.51 16.62 1609.06 1 3 3 0 0 130 - 18.71 8.31 360.03 2 6 6 0 0 92 - 19.87 16.62 254.79 3 3 3 0 0 240 - 21.02 8.31 664.67 4 5 5 0 0 829 . 22.18 13.85 2295.88 Table 3 Continued. Vascularity Osteocytes Specimen # Box # Total Longitudinal Radial Circumferential Osteocytes Ratio L:R Radius /mm2 /mm2 5 7 7 0 0 655 - 23.34 19.39 1814.00 6 4 4 0 0 587 - 24.50 11.08 1625.67 7 3 3 0 0 473 - 25.66 8.31 1309.96 8 5 5 0 0 526 - 26.82 13.85 1456.74 9 4 4 0 0 724 - 27.98 11.08 2005.09 10 6 6 0 0 724 - 29.14 16.62 2005.09 11 10 10 0 0 767 - 30.29 27.69 2124.18 12 11 11 0 0 627 - 31.45 30.46 1736.45 13 12 12 0 0 600 - 32.61 33.23 1661.68 14 7 7 0 0 723 - 33.77 19.39 2002.32 UMNH VP 20454.8 1 8 8 0 0 500 - 15.70 22.16 1384.73 humerus Utahceratops 2 8 8 0 0 600 - 16.86 22.16 1661.68 gettyi 3 12 12 0 0 582 - 18.02 33.23 1611.83 4 12 12 0 0 670 - 19.18 33.23 1855.54 5 9 9 0 0 660 - 20.33 24.93 1827.85 6 13 13 0 0 700 - 21.49 36.00 1938.62 7 13 13 0 0 623 - 22.65 36.00 1725.38 8 18 18 0 0 600 - 23.81 49.85 1661.68 9 20 20 0 0 600 - 24.97 55.39 1661.68 10 14 6 3 5 500 2.00 26.13 38.77 1384.73 UMNH VP 19490 1 8 8 0 0 500 - 11.22 22.16 1384.73 Humerus indeterminant 2 7 7 0 0 544 - 12.38 19.39 1506.59 ceratopsian 3 10 10 0 0 376 - 13.54 27.69 1041.32 4 8 8 0 0 631 - 14.70 22.16 1747.53 5 14 14 0 0 521 - 15.86 38.77 1442.89 6 13 13 0 0 655 - 17.02 36.00 1814.00 00 Table 3 Continued. Vascularity Osteocytes Specimen # Box # Total Longitudinal Radial Circumferential Osteocytes Ratio L:R Radius /mm2 /mm2 7 17 17 0 0 600 - 18.17 47.08 1661.68 8 18 18 0 0 700 - 19.33 49.85 1938.62 9 12 10 2 0 700 5.00 20.49 33.23 1938.62 10 19 16 3 0 600 5.33 21.65 52.62 1661.68 1 9 7 2 0 517 3.50 17.70 24.93 1431.81 UMNH VP 17000 femur 2 6 5 1 0 475 5.00 18.86 16.62 1315.49 Kosmosaurus richardsoni 3 17 17 0 0 700 - 20.02 47.08 1938.62 4 10 10 0 0 565 - 21.17 27.69 1564.75 5 9 9 0 0 730 - 22.33 24.93 2021.71 6 9 8 1 0 670 8.00 23.49 24.93 1855.54 7 9 9 0 0 663 - 24.65 24.93 1836.15 UMNH VP21339B 1 6 6 0 0 252 - 4.03 16.62 697.90 femur Kosmoceratops 2 8 8 0 0 482 - 5.18 22.16 1334.88 richardsoni 3 4 4 0 0 500 - 6.34 11.08 1384.73 4 6 6 0 0 700 - 7.50 16.62 1938.62 5 8 8 0 0 566 - 8.66 22.16 1567.52 6 2 2 0 0 480 - 9.82 5.54 1329.34 7 5 5 0 0 730 - 10.98 13.85 2021.71 8 6 6 0 0 500 - 12.14 16.62 1384.73 9 7 7 0 0 741 - 13.30 19.39 2052.17 10 7 7 0 0 781 - 14.45 19.39 2162.95 11 4 4 0 0 746 - 15.61 11.08 2066.02 12 2 2 0 0 668 - 16.77 5.54 1850.00 13 1 1 0 0 456 - 17.93 2.77 1262.88 TMP 66.10.36 tibia 1 14 10 4 0 602 2.50 14.08 38.77 1667.22 Centrosaurus apertus 2 13 11 2 0 613 5.50 15.23 36.00 1697.68 Table 3 Continued. Specimen # Box # Total Longitudinal Radial Circumferential Osteocytes Ratio L:R Radius Vascularity /mm2 Osteocytes /mm2 TMP 79.11.56 tibia Centrosaurus apertus 3 13 10 3 0 673 3.33 16.39 36.00 1863.85 4 13 11 2 0 579 5.50 17.55 36.00 1603.52 5 14 10 4 0 845 2.50 18.71 38.77 2340.20 6 11 10 1 0 588 10.00 19.87 30.46 1628.44 7 14 13 1 0 671 13.00 21.03 38.77 1858.31 8 21 19 2 0 950 9.50 22.19 58.16 2630.99 9 22 19 3 0 901 6.33 23.35 60.93 2495.29 10 21 17 4 0 610 4.25 24.50 58.16 1689.37 11 23 22 1 0 626 22.00 25.66 63.70 1733.68 12 15 15 0 0 670 - 26.82 41.54 1855.54 13 8 3 5 0 400 0.60 27.98 22.16 1107.79 1 9 9 0 0 395 - 7.82 24.93 1093.94 2 12 11 1 0 514 11.00 8.98 33.23 1423.50 3 9 9 0 0 465 - 10.14 24.93 1287.80 4 14 12 2 0 500 6.00 11.30 38.77 1384.73 5 14 13 1 0 499 13.00 12.46 38.77 1381.96 6 12 12 0 0 458 - 13.61 33.23 1268.41 7 6 6 0 0 530 - 14.77 16.62 1467.82 8 14 14 0 0 570 - 15.93 38.77 1578.59 9 18 18 0 0 500 - 17.09 49.85 1384.73 10 16 16 0 0 448 - 18.25 44.31 1240.72 11 14 12 2 0 565 6.00 19.41 38.77 1564.75 12 19 19 0 0 456 - 20.57 52.62 1262.88 13 13 10 3 0 394 3.33 21.73 36.00 1091.17 14 12 11 1 0 300 11.00 22.88 33.23 830.84 4O*. Table 3 Continued. Vascularity Osteocytes Specimen # Box # Total Longitudinal Radial Circumferential Osteocytes Ratio L:R Radius /mm2 /mm2 15 11 9 0 2 200 - 24.04 30.46 553.89 16 16 14 2 0 740 7.00 25.20 44.31 2049.40 17 14 12 2 0 480 6.00 26.36 38.77 1329.34 femur F-2 Protocercitops 1 10 10 0 0 500 - 8.49 27.69 1384.73 andrewsi 2 11 11 0 0 534 - 9.29 30.46 1478.89 3 15 15 0 0 800 - 10.09 41.54 2215.57 4 14 14 0 0 700 - 10.89 38.77 1938.62 5 18 18 0 0 821 - 11.69 49.85 2273.73 6 15 15 0 0 1000 - 12.49 41.54 2769.46 MPC-D100/530 femur 1 8 8 0 0 566 - 19.04 22.16 1567.52 Protocercitops andrewsi 2 8 8 0 0 700 - 20.01 22.16 1938.62 3 9 9 0 0 800 - 20.97 24.93 2215.57 4 17 17 0 0 779 - 21.94 47.08 2157.41 5 13 12 1 0 700 12.00 22.90 36.00 1938.62 6 7 7 0 0 128 - 23.87 19.39 354.49 MPC-D 100/530 humerus 1 7 7 0 0 465 - 6.03 19.39 1287.80 Protoceratops andrewsi 2 18 18 0 0 474 - 7.03 49.85 1312.73 3 27 27 0 0 555 - 8.02 74.78 1537.05 4 12 11 0 1 558 - 9.01 33.23 1545.36 5 11 11 0 0 763 - 10.00 30.46 2113.10 6 9 5 4 0 638 1.25 10.99 24.93 1766.92 7 21 18 3 0 843 6.00 11.98 58.16 2334.66 PIN 698/1977 femur 1 11 4 7 0 900 0.57 - 30.46 2492.52 Psittacosaurus 2 19 13 6 0 800 2.17 - 52.62 2215.57 mongoliensis 3 20 4 16 0 851 0.25 - 55.39 2356.81 4 9 8 1 0 566 8.00 - 24.93 1567.52 42 Figure 5. UMNH VP 16860 femur illustrating how vascular canal density and osteocytes density were calculated. The blue circles indicate vascular canals. The white dots indicated osteocytes. The A arrow is pointing in the direction of the anterior side of the bone. The L arrow is pointing the direction of the lateral side of the bone. 43 44 Figure 6. This humerus, (UMNH VP 20454.8), illustrating the different types of vascular canals. The vascular canal outlined in yellow is a radial canal. The red horizontal lines represent the circumferential canals. The blue dots indicated longitudinal vascular canals. 46 Figure 7. Graphs that represent the data collected in the femora analysis of the different taxa of ceratopsians. A. The radius of the element plotted against the osteocytes densities. B. The densities of the vascularity seen plotted against the osteocyte densities. C. The radius of the element plotted against the vascular densities. 47 A 48 B Figure 7 Continued. 49 C Figure 7 Continued. 50 Figure 8. Graphs that represent the data collected in the tibiae analysis of the different taxa of ceratopsians. A. The radius of the element plotted against the osteocytes densities. B. The densities of the vascularity seen plotted against the osteocyte densities. C. The radius of the element plotted against the vascular densities. 51 RADIUS vs. OSTEOCYTES 3000 2500 £2000 50WH l» yc 1500 EdHVIO 1000 500 A A A A ▲■A ■ □0 A o o ▲ A ▲ AO <AO A ■ A 1 A □ □ C t A □ □ DA.□ O O o o 10 A 15 20 RADIUS (mm) 25 30 A UMNH VP 20444.5 Utahceratops gettyi a UMNH VP 20454.3 Utahceratops gettyi o UMNH VP 16681 Utahceratops gettyi mTMP 66.10.36 Centrosaurus apertus □ TMP 79.11.56 Centrosaurus apertus 35 52 B Figure 8 Continued. 53 C Figure 8 Continued. 54 Figure 9. Graphs that represent the data collected in the humeri analysis of the different taxa of ceratopsians. A. The radius of the element plotted against the osteocytes densities. B. The densities of the vascularity seen plotted against the osteocyte densities. C. The radius of the element plotted against the vascular densities. 55 A RADIUS vs. OSTEOCYTES 2500 2000 EE 1500 CdH> Uo bd (HZ!1000 o 500 X = X = o 1 » < " < ■ ■ < ■ < ■< ■ < ■ 1 X X X ' ' < » ■< 1 ■ X X ■ = 10 15 20 25 RADIUS (mm) 30 35 a UMNH VP 20454.8 Utahceratops gettyi - UMNH VP 16865.1 Indctcnninant ceratopsian ■ UMNH VP 19490 Indeterminant ceratopsian xMPC-D 100/530 Psittacosaurus mongoliensis 40 56 B Figure 9 Continued. 57 C Figure 9 Continued. RESULTS Utahceratops gettyi Bone Histology Humerus UMNH VP 20444.1. This bone (Fig. 10) has a mid-diaphyseal circumference of 160 mm and has a complete length of 440 mm. It was sectioned at the midshaft, 150 mm from the distal end. The medullary cavity is 21.16 mm by 17.91 mm (long axis medial-laterally oriented) and is not crushed. A post-mortem fungal alteration (which is described in Trueman and Martill, 2002) dominates this bone and all sampled elements from UMNH VP locality 945 (Brandau and Irmis, 2011, 2012). This gives the microstructure a gritty or dirty appearance and obscures some of its detail. The fungal alteration is pervasive enough that it is not possible to observe osteocytes continuously throughout the cross section. There were trabeculae present at one time but I cannot identify the extent of them; there is an abrupt transition between the matrix infilling the medullary cavity and the bone tissue of the cortex. The bone tissue is dominated by woven collagen fiber orientation, although this was difficult to assess due to the fungal alteration. Secondary osteons encircle the medullary cavity and are evenly distributed. Where this occurs, observed collagen fiber orientation reflects the orientation of the fibers of the secondary bone tissue. Longitudinal canals, circular or elliptical in shape, make up the majority of the vascular canals, but some radial canals are present near the periosteum. The canals do not exhibit any orderly change in density. The ratio of simple canals to primary and secondary osteons is 0:90:10, respectively. Histological details do not appear to change circumferentially, but they do change vertically from the medullary cavity to the periosteum. Evenly distributed secondary osteons are restricted to a narrow region encircling the medullary cavity. The bone does not preserve any LAGs. It does have a pronounced color change 2.5 mm from the periosteum (13 mm from the edge of the medullary cavity, and 22.2 mm from the middle of the medullary cavity), but this does not appear to be associated with any histological differences. UMNH VP 20454.8. This bone (Fig. 10) possesses a mid-diaphyseal circumference of 145.1 mm and a preserved length of 210.3 mm. I sectioned this humerus at the approximate midshaft (neither distal nor complete proximal end is preserved). The medullary cavity internal to the cortex is 24.37 mm by 15.21 mm (long axis medial-laterally oriented) and shows no evidence of crushing. Fungal alteration is present, but the microstructure can be observed clearly. Trabeculae are not abundant, but those present near the medullary cavity comprise primary bone and are oriented at right angles to the medullary cavity rim. There is an abrupt transition between the cortical bone tissue and the infilled medullary cavity. No remodeling is apparent in the cortex of the bone. Under cross-polarized light the entire cortex displays woven collagen fiber orientation. The proportion of simple canals versus primary osteons and secondary osteons is 10:90:0 respectively, where the simple canals are restricted to the periosteum and the primary osteons make up the majority of the cortex. Simple canals are restricted to the periosteum, and no secondary osteons are present. The vascular canals are predominantly longitudinal, except at the periosteum, where the ratio of longitudinal to circumferential to radial canals is 16:14:8 canals per mm2. From the edge of the 59 medullary cavity to the periosteum, there is a gradual increase in vascular canal density from 22 canals per mm2 to 55 canals per mm2. The vascular canals also change in size and shape, starting out fairly large encircling the medullary cavity, 0.111 mm by 0.049 mm, to more circumferential squat ellipses measuring 0.163 mm circumferentially and 0.018 radially. Osteocyte density is fairly uniform throughout the entire bone ranging from 1384 osteocytes per mm2 to 1938 osteocytes per mm2. A change of tissue characterized by elliptical, more reticular vascularity and more dense osteocytes is seen starting 2.5 mm internal from the periosteum (11 mm from the edge of the medullary cavity, 19 mm from the center of the medullary cavity). The matrix of the bone is mostly unchanged circumferentially except for some large vascular cavities, which are present on the anterior side of the bone and on both lateral sides, but not on the posterior side. They range in diameter from 3 ^m to 3 mm. No LAGs are present. Ulna UMNH VP 20444.2. This ulna (Fig. 11) has a mid-diaphyseal circumference of 136 mm and has a complete length of 322 mm. This element was sectioned at the midshaft, 12.0 cm from the distal end. The medullary cavity internal to the cortex is 25.0 mm by 5.09 mm (medio-lateral axis and antero-posterior axis). Fungal alteration is pervasive in this bone, which makes some of the microstructure difficult to observe. Trabeculae are present but are so fractured and altered by the fungal invasion that further histological details could not be determined. There is a gradual transition between the compact bone tissue of the cortex and the trabeculae infilled medullary cavity. The cortex is woven-fibered bone throughout, but this is difficult to ascertain because of the 60 fungal alteration. The approximate proportion of simple canals versus primary osteons versus secondary osteons is 20:80:0, respectively, with predominantly primary osteons through the inner cortex and simple canals restricted to near the periosteum. The vascular canals are predominantly longitudinal and uniformly circular in shape, becoming smaller in diameter as they approach the periosteum. This vascular canal size change is associated with a color change at 4.0 mm from the periosteum, easily visible as a line on the polished section. The fungal alteration is so severe in this bone that osteocytes cannot be observed. This makes the color and vascularity change harder to see. There is no evidence of LAGs. UMNH VP 20454.1. This element (Fig. 11) has a mid-diaphyseal circumference of 110 mm and has a complete length of 321 mm. This ulna was sectioned at the midshaft, 110 mm from the distal end. The bone is crushed antero-posteriorly into the medullary cavity, making accurate measurement of the cavity difficult. Like other specimens from this locality, fungal alteration is severe, making some of the microstructure difficult to observe. Neither cancellous bone nor trabeculae are present. There is an abrupt transition between the bone tissue of the inner cortex and the matrix infilling the medullary cavity. The collagen fiber orientation is dominated by woven fibered bone throughout the thin section. Vascular canals are predominantly longitudinal, comprised exclusively of primary osteons. The shape, size, and density of the vascular canals change throughout the section. Near the medullary cavity, canals are larger, more oblong and oriented radially. Approximately 3 mm away from the medullary cavity, the vascular canals become smaller, more circular, and more dense in a zone that is ~2 mm thick. For the final 2.5 mm adjacent to the periosteum, the vascular canals are more 61 circumferential and infilled with a black, opaque mineral. The fungal alteration is so severe that osteocytes are not clearly observable. No LAGs are present. Femur UMNH VP 16860. This femur (Fig. 12) has a mid-diaphyseal circumference of 250 mm, with a preserved femur length of 488 mm (the distal end is not preserved). The bone is antero-posteriorly crushed, which makes the medullary cavity a sigmodal shape instead of ovoid or circular. A small amount of cancellous bone is present, circumferentially outlining the medullary cavity. Preserved trabeculae comprise woven-fibered primary bone, and are arranged in an unordered web. The transition between the bone tissue of the cortex and the matrix-infilled medullary cavity is gradual because of the presence of trabeculae. Excluding the trabecular bone tissue, the majority of the tissue comprises secondary osteons, seen best in cross-polarized light. These secondary canals make the fiber orientation difficult to observe because the remodeling has obscured most primary bone tissue. No simple canals or primary osteons are visible. Longitudinal canals dominate, but radial canals are present sporadically between 9-16 mm away from the center of the medullary cavity. The density of the vascular canals does not seem to vary in an orderly fashion (Table 3). There are larger vascular canals, about 1 mm in diameter, which outline the medullary cavity and are deformed by the crushing. The density of the osteocytes generally increases towards the periosteum with osteocytes counts ranging from 1171 osteocytes per mm2 to 2200 osteocytes per mm2 moving nearer to the periosteum, but because the bone is primarily made up of secondary osteons, this density increase does not reflect the density of osteocytes of primary bone 62 tissue. No LAGs are present. UMNH VP 12198. The mid-diaphyseal circumference of this element (Fig. 12) is 437 mm, with a preserved femur length of 874 mm (the distal end is incomplete). Because of its substantial size, a complete cross section was not made. Instead an 18 mm wide core was removed from the anterior surface of the mid-diaphysis of the bone. As a result, the full extent of the medullary cavity is unknown. The trabeculae present are composed of secondary osteons. Woven collagen fibers dominate, but this represents mostly secondary bone tissue in the form of secondary osteons, which are present throughout the cortex all the way to the periosteum. They are concentrated towards the medullary cavity, but form approximately 75% of vascular canals throughout the cortex in comparison to primary osteons, and no simple canals are observed. The vascular canals are primarily longitudinal in orientation with a few radial canals at the periosteum. The vascular canals near the medullary cavity are larger, approximately 0.5 mm in diameter, becoming progressively smaller towards the periosteum. Osteocytes are consistently dense through the section. No LAGs are apparent. UMNH VP 20444.4. The complete femur length of this individual is 605 mm; I took the section (Fig. 12) from a location 245 mm from the distal end of the bone. The majority of the posterior side, however, is missing therefore a full circumference measurement was not possible, but it can be estimated at around ~200 mm. The medullary cavity is approximately 47 mm by 26 mm, although the second measurement was estimated because of incompleteness of the element. Minimal cancellous bone is present and no trabeculae are apparent. There is an abrupt transition between the cortical bone tissue and the matrix-infilled medullary cavity. The entire bone comprises woven 63 collagen fibers. Vascular canals are all primary osteons except outlining the medullary cavity where a few (10% of the total matrix) secondary osteons are present. Simple canals are rare if present. Vascular density remains consistent throughout the section, from the medullary cavity towards the periosteum, and the vascular canals are predominantly longitudinal in orientation, with a few radial canals in the inner cortex. Longitudinal canals are larger (about 0.19 mm in diameter) nearer to the medullary cavity and become progressively smaller (0.08 mm) and more uniform moving from the medullary cavity outward to the periosteum. The density of the canals ranges from 8 canals/mm2 to 19 canals/mm2, and average around 8 canals/mm2 throughout the section. Osteocyte density does not vary drastically, ranging from just under 1000 osteocytes per mm2 to 1578 osteocytes per mm2. Moving closer to the periosteum, 34.5 mm from the medullary cavity, average osteocyte density is higher so that it ranges from 1384 to 1578 per mm2. A change in tissue or a pause in growth may be present because both the thin section and polished section display at least two concentric color changes, one 10.72 mm from the periosteum, and the second 18.57 mm away. These color changes are linked with an alignment and size increase of vascular canals. Until this point, the vascular canals were unordered and not arranged in a certain way. At these color changes, the vascular canals organize and align themselves along the color band. These canals also become slightly larger near the color changes. The two bands are not diagnostic of a LAG because they do not birefringe under cross-polarized light or show any momentary cessation of growth (cf. Chinsamy, 1994). UMNH VP 20454.5. The femur has a mid-diaphyseal circumference of 230 mm 64 and a preserved length of 291 mm (as both the distal end and the proximal end are incomplete). The medullary cavity internal to the cortex is 61.55 mm (along the long axis of the cross section) by 28.16 mm and there is no evidence for crushing. Trabecular bone is extensive and circumferentially lines the exterior of the medullary cavity. The trabeculae comprise primary bone tissue and are arranged in a reticular pattern. There is an abrupt shift between the end of the trabeculae and the infilled medullary cavity. Fungal invasion is also present but does not obscure fine details such as osteocytes. The bone tissue possesses a woven collagen fiber orientation (Fig. 13). The extent of simple canals versus primary canals versus secondary canals is 5:65:30, respectively, with simple canals occurring predominantly near the periosteum, primary osteons making up the bulk of the cortex, and secondary osteons uniformly distributed around the circumference of the medullary cavity. Vascular canal density and orientation do not change throughout the section. Four zones unique in color, size and shape of vascular canals, and mineral infilling can be observed. Zone one is nearest the medullary cavity and is ~2.4 mm thick. Here, the vascular canals are primary longitudinal, vary in shape from circular to elliptical, and have a density ranging from 13-44 vascular canals/mm2. Zone two is lighter colored, ~3 mm thick, and is dominated by longitudinal canals that are ovoid in cross-section. The vascular density ranges from 16-36 canals/mm2. Zone three is darker, consisting of larger, denser longitudinal vascular canals, possessing an elliptical shape with the long axis of the ellipse oriented medio-laterally. This zone is 4.2 mm thick, with vascular densities varying from 11-41 canals/mm2. The exterior-most zone has a distinct appearance. In contrast to the aforementioned vascular canals, which are infilled with quartz, the canals in this zone are filled with some type of opaque black 65 mineral. The longitudinal vascular canals range from circular to elliptical in cross­section; a reticular organization is prominent, with the highest vascular canal density being ~50 canals per mm2. This zone is approximately 0.6 mm wide, and ends at the periosteum. Osteocyte density does not change through the entirety of the thin section (Table 3). There is no evidence of radial canals or LAGs. The cessation of zone one and the beginning of zone four can be observed on the surface of the polished section, but this does not exhibit the histological characteristics that define a true LAG. Tibia UMNH VP 16681. This element has a mid-diaphyseal circumference of 102 mm and a complete length of 450 mm. This tibia (Fig. 14) was sectioned at the midshaft, 22.2 cm from the distal end. The medullary cavity dimensions are 34.62 mm (long axis) by 14.34 mm. There is some posterio-medial crushing. This makes the medullary cavity subcircular and means the posterior edge of the bone is crushed, which makes the histology difficult to follow circumferentially. The trabeculae that are still attached to the lining of the medullary cavity are arranged in an unordered web. Preserved trabeculae are abundant, infill the entire medullary cavity, and are composed mainly of primary bone tissue. Osteocytes appear to have a consistent density throughout the cortex. Vascular canal density also does not seem to change through the section. Longitudinal canals dominate, although a few radial canals are present throughout the cortex, particularly in the vicinity of the periosteum. Reticular canals are also present but not dominant. The bone tissue is dominated by well defined secondary osteons, which rim the spaces between the trabeculae. Because the cortical bone is composed primarily of 66 secondary osteons, observing the collagen fiber orientation primary bone tissue is impossible. The ratio of simple to primary to secondary osteons is 10:0:85 respectively, with secondary osteons making up the majority of the bone extending circumferentially around the medullary cavity, and simple canals restricted to the periosteum. No LAGs or abrupt color changes were observed. UMNH VP 20444.5. This tibia (Fig. 14) has a mid-diaphyseal circumference of 189 mm, a complete length of 465 mm and shows no evidence of crushing. The element was sectioned at the midshaft, 24.0 cm from the distal end. The medullary cavity internal to the cortex is 26.30 mm (along the long axis) by 19.40 mm. Trabeculae are absent and the transition from the bone tissue of the cortex and the matrix-infilled medullary cavity is abrupt. Fungal invasion is present throughout the whole cross-section of this element, making some of the microstructure, specifically osteocytes, difficult to see. The collagen fiber orientation is extremely difficult to observe because of the fungal degradation, but it appears the bone is composed of woven collagen fibers. This histology is uniform circumferentially. The bone is almost completely composed of primary osteons and only a few secondary osteons are observed at the periosteum, making a thin rim outlining the edge of the bone. The ratio of primary to secondary osteons throughout the cortex is 95:5. The density of vascular canals changes across the radius of the bone, but not in any increasing or decreasing order. The vascular canal density ranges from less than 1 to 24 canals/mm2, with the cross-sectional shape of the longitudinal vascular canals being primarily circular in cross section. Moreover, the osteocytes do not exhibit an orderly transition in density, ranging from 1384 to 2492 osteocytes/mm2. Three circumferential color changes are present that were initially interpreted as 67 LAGs (Levitt, 2011). However, upon further investigation, these color changes do not exhibit the characteristics that define a LAG; that is, they do not show a distinct cessation in growth that is birefringent under cross-polarized light and can be observed around most or all of the bone's circumference. These color changes may reflect heterogeneities in preservation mode, but they do seem to be linked to growth, because the nature of vascularity changes slightly across the boundaries. The first zone of distinct color is approximate 3 mm thick and lies closest to the medullary cavity. The vascular canals are larger than the canals closer to the periosteum, ~0.009 mm in diameter, therefore there are fewer canals per mm2 (about 8 canals/mm2, located 12 mm from the center of the medullary cavity). Zone two, beginning at 15 mm from the center of the medullary cavity, is 9 mm thick, and has highly variable vascular canal density, from less than 1 to 25 canals/mm2. Zone three, starting 24 mm from the center of the bone, is slightly darker in color, approximately 2.5 mm thick, and has a vascular density of 8-19 canals/mm2. Zone four, starting at 26.5 mm from the center, is even darker, with larger vascular canals that are more elliptical in cross section, ranging from less than 1 to 11 canals/mm2. Finally, a rim of different colored bone tissue, composed mostly of secondary osteons, outlines the periosteum. Some radial canals are present in this area. UMNH VP 20454.3. This element (Fig. 14) has a mid-diaphyseal circumference of 167 mm and a preserved length of 231 mm. Both proximal and distal ends are missing. This tibia was sectioned at the midshaft, 167 mm from the distal end. The medullary cavity internal to the cortex is 19.34 mm in diameter and the cross section is circular in shape. Trabeculae are absent and there is an abrupt change between the bone tissue of the cortex and the matrix-infilled medullary cavity. This element shows 68 evidence of fungal effects but not so extensively as to obscure the microstructure. This tibia is composed entirely of woven collagen fibers. The density of osteocytes is fairly consistent throughout the whole bone, averaging 1760 per mm2, varying from 1570 per mm2 at a distance of 14 mm from the center, to 2132 osteocytes per mm2 at a distance of 21 mm from the center of the medullary cavity. Vascular canals are dominantly primary osteons. The vascular canal density is consistent through the inner cortex averaging ~ 24 canals per mm2, but then decreases to ~11 canals per mm2 in density near the periosteum. Vascular canal orientation is predominantly longitudinal, but circumferential vascular canals are also present between 1.92 mm and 4.81 mm from the edge of the periosteum. A change in color is present 1.74 mm from the edge of the periosteum. This is not associated with a change in the density or orientation of vascular canals, but there is a slight change in osteocyte density (from 1661 to1938 osteocytes per mm2) towards the periosteum. This dark brown color may be due to the high density of osteocytes or may represent a larger concentration of fungal invasion. The histology is not uniform circumferentially. On the anterior and right lateral side near the medullary cavity are large cavities ranging in size from 3 microns to 3 mm in diameter. Also, the density and size of the vascular canals decreases on the lateral posterior side of the element. No LAGs are present. Indeterminate limb shaft UMNH VP 16861. This element, recovered from UMNH VP locality 942, is crushed and broken to the extent that precise identification is impossible, but it is obviously some sort of limb bone shaft. This bone has a circumference of 180 mm where 69 it was sectioned and has a preserved length of 180 mm. The medullary cavity internal to the cortex is completely crushed, and anatomical orientation of the crushing cannot be determined. The trabeculae comprise largely primary bone tissue with occasional secondary osteons. No order can be seen in the orientation of the trabeculae because of the crushing. In crossed-polarized light, the bone tissue is entirely woven-fibered bone; however, the bone tissue is primarily secondary osteons, so the collagen fiber orientation of primary bone tissue is unknown. Vascular canals are predominantly longitudinal canals that are circular in cross section, and their density is fairly uniform throughout the cortex. Osteocytes also are dispersed evenly throughout the bone tissue and do not seem to increase or decrease in density through the section. No LAGs are present. Ribs UMNH VP 12198. The circumference of this rib shaft at its point of sectioning is 58.0 mm. A cavity preserved in the center of the bone is filled with cancellous bone. Trabeculae are composed of primary bone rimmed with secondary tissue. The majority of the element (Fig. 15) consists of remodeled bone (i.e., secondary osteons), only 10% of the vascular canals are primary osteons. The vascular canals are entirely longitudinal and uniform in density throughout the bone. The trabeculae form a uniform network of primary bone tissue in the center, and the secondary osteons make up the bulk of the cortex. Because this bone is dominantly composed of secondary osteons, the collagen fiber orientation of primary bone tissue is undistinguishable. Osteocytes are uniformly dense throughout and there is no evidence of LAGs. UMNH VP 20454.7. The circumference of this rib (Fig. 15) shaft is 75.0 mm. 70 Trabeculae comprise primary bone rimmed with secondary tissue, and make up an unordered web in the center of the cross section. Some fungal alteration is evident, but not enough to obscure the microstructure. The majority of vascular canals are longitudinally-oriented secondary osteons. Some primary osteons are present in the inner cortex, and their ratio with secondary osteons is 30:70. The diameter of the canals is larger in the interior of the bone, becoming smaller closer to the periosteum. The histology of this bone is uniform circumferentially. The preponderance of secondary osteons means that determining collagen fiber orientation of primary bone tissue is impossible. No LAGs are present. Ossified tendons UMNH VP 12198. The circumference of the sampled tendon fragment is 44.0 mm. It is composed entirely of remodeled bone (Fig. 15), and filled with compacted longitudinal canals, all circular in cross section, and no radial or circumferential canals. Because the tissue is composed of all secondary osteons, the orientation of primary tissue collagen fiber orientation is impossible to determine. No LAGs are present. UMNH VP 20454.9. The circumference of this tendon (Fig. 15) fragment is 40 mm. The section itself measures 16.12 mm by 8.14 mm. Fungal alteration is evident, but is not as severe as in most of the limb bones from this same site. No cavity is present in the element's interior, but vascular spaces are larger towards the center of the cross section. The few visible trabeculae comprise primary bone rimmed with secondary tissue. Vascular canals are 90% secondary osteons, and all canals are longitudinally oriented, which decrease in diameter and increase in density towards the periosteum. 71 72 Because most of the tissue comprises secondary osteons, determining primary tissue collagen fiber orientation is impossible. There is no evidence of LAGs. Kosmoceratops richardsoni Bone Histology Femur UMNH VP 17000. This femur (Fig. 12) was highly fragmented when recovered, and therefore the diaphyseal fragments used for this study may not have been from the exact mid-diaphysis. Based on the proportions of the femur and the sacrum of the specimen NHMUK R4948 of Chasmosaurus belli (Maidment and Barret, 2011), I estimate the femur length of UMNH VP 17000 to be approximately 100 cm, with a midshaft circumference of 45 cm. Two pieces were sectioned for study, both with incomplete cortex preservation, and thus their total circumference is unknown. The first fragment, UMNH VP 17000A with a maximum width of 72.2 mm, is dominated by cancellous bone that is fragmented by postmortem crystallization. This cancellous bone represents extensive trabeculae that underlie the cortex of the bone. The trabeculae are mainly composed of primary tissue, with some secondary osteons visible, and there is no apparent order to the trabecular organization. This bone's histology cannot be followed circumferentially because only one area preserves the original cortex. The cortex is almost entirely composed of secondary osteons with a concentrated rim of secondary osteons near the periosteum. The collagen fiber orientation of any primary tissue is hard to distinguish because of the abundance of secondary osteons, but the remodeled tissue is composed of woven fibered bone (Fig. 13). Both radial and longitudinal vascular canals are present, but the longitudinal canals dominate. There is no apparent order to the occurrence of radial canals, and canal density also varies widely with no apparent pattern, ranging from 16 canals/mm2 at 19 mm from center of the bone to 47 canals/mm2 20 mm from the center of the bone. Osteocyte density also lacks any apparent pattern, ranging from 1315 osteocytes/mm2 at 18 mm from the center of the bone to 2021 osteocytes/mm2 at 22 mm from the center of the bone. These osteocytes densities represent remodeled tissue. No LAGs are present. The second sectioned bone fragment from the femur, UMNH VP 17000B, has a preserved length of 80.9 mm (although this does not reflect the full dimensions of the femur) with a maximum width of 72.6 mm. The medullary cavity is infilled with matrix and cancellous bone that form extensive trabeculae; these have been fragmented by crystallization. The trabeculae are composed mostly of primary bone rimmed with secondary tissue. The trabeculae make up a reticular network of ellipses. The vascular canals, mostly longitudinal with some radial examples, are mostly secondary osteons, with some primary osteons present in the trabeculae. The secondary osteons are so dense that primary tissue collagen fiber orientation cannot bee seen. This vascular canal orientation is consistent across the thin section. Osteocytes are dense throughout the bone tissue. No LAGs are present. UMNH VP 21339. The preserved length of this femur is 670 mm (with the distal end missing). This bone (Fig. 12) was sectioned at the midshaft of the femur, 44.0 cm from the proximal end. This bone was crushed dorsoventrally, preventing an accurate measurement of the midshaft circumference. Based on the preserved portion of this element, I estimate the circumference at ~400 mm. Trabeculae are not very extensive but are composed primarily of primary bone and form an unordered network. The original 73 collagen fiber orientation cannot be surmised because the bone tissue is dominated by secondary bone tissue. Although dominated by secondary osteons (Fig. 16), the cortex consists of ~5% primary tissue. All vascular canals are longitudinal with their density varying widely without any apparent pattern, ranging 22 canals/mm2 at 5 mm from the center of the bone, 22 canals/mm2 at 9 mm from the center of the bone, and 3 vascular canals/mm2 at 18 mm from the center of the bone. Osteocyte density also varies in an unordered way, ranging from a density of 698 osteocytes/mm2 at 4.0 mm from the center of the bone, to 2162 osteocytes/mm2 at 14.5 mm from the center of the bone, and 1262 osteocytes/mm2 at 18 mm from the center of the bone. However, these values largely represent the osteocytes densities of remodeled tissue. The histology of the bone is uniform circumferentially around the bone, with the larger vascular canals encircling the crushed medullary cavity and then decreasing in size further out towards the periosteum. No LAGs are apparent. Indeterminate ceratopsid humeri UMNH VP 16865.1. This element has a mid-diaphyseal circumference of 189 mm and has a complete length of 545 mm. This humerus (Fig. 10) was sectioned at the midshaft, 182 mm from the distal end. The medullary cavity internal to the cortex is 28.98 mm at the long axis by 24.97 mm, and is not crushed. Trabeculae are an extensive network with spaces between the trabeculae made of circles and ellipses that extend throughout the medullary cavity. The trabeculae are composed mainly of primary bone tissue. Where primary tissue still dominates, it appears to be made up of woven collagen fibers, but the abundance of secondary osteons makes this uncertain. The proportion of 74 simple canals versus primary osteons versus secondary osteons is 0:60:40. Most of the secondary osteons are located at or near the medullary cavity. Through the section, vascular canals change in size, starting larger closer to the inner cortex, and smaller towards the periosteum. Similarly, the shape of the longitudinal vascular canals change from circles to flattened ellipses that are more circumferential near the periosteum. In contrast, canal density does not follow this pattern, ranging from 8 canals/mm2 at 18 mm from the center of the bone to 30 canals/mm2 at 31 mm from the center of the bone. The density of the osteocytes does not follow a clear pattern either, averaging ~1456 osteocytes/mm2, but varying from 254 osteocytes/mm2 at 20 mm from the center of the bone to 2295 osteocytes/mm2 at 22 mm from the center of the bone. Although no observable LAGs are present, a change in color is present, likely associated with a shift in microstructure. This change occurs 2 mm from the periosteum and is ~0.5 mm thick. In this zone, the osteocytes appear less dense, the vascular canals are more elliptical in shape, and vascular canal density decreases. The histology is uniform circumferentially with the secondary osteons encircling the medullary cavity and the rest of the bone comprising primary tissue. UMNH VP 19490. This humerus (Fig. 10) has a mid-diaphyseal circumference of 204 mm and has a preserved length of 167.6 mm (the proximal and distal ends are not preserved) and was sectioned at the midshaft. The medullary cavity is 40.29 mm by 23.65 mm. There has been a little crushing on the anterior side but it has not skewed the shape of the medullary cavity. Trabeculae composed of primary bone but rimmed with secondary tissue form an unordered web that fills the entire medullary cavity. The cortex is dominated by secondary osteons so the collagen fiber orientation of the primary tissue 75 cannot be observed. A few primary osteons occur near the medullary cavity, but they comprise only ~10% of the vascular canals. The vascular canals are primarily longitudinal, mostly circular in cross section, with some radial canals adjacent to the periosteum. The vascular canals gradually increase in density towards the periosteum; 11 mm from the center of the bone has 22 canals/mm2, whereas 21 mm from the center the bone has a density of 52 canals/mm2 (Fig. 17). The variation in osteocyte density shows no apparent order, ranging from 1041 osteocytes/mm2 at 13 mm from the center of the bone to 1938 osteocytes/mm2 at 19 mm from the center of the bone. However, these values represent the osteocyte density of largely remodeled tissue. The bone histology is uniform circumferentially. No LAGs or abrupt changes in tissue type are present. Summary of Histological Trends The study sample of Utahceratops elements derived from the articulated individual at UMNH VP locality 945 (UMNH VP 20444.4 - femur, UMNH VP 20444.5 - tibia, UMNH VP 20444.1 - humerus, and UMNH VP 20444.2 - ulna) are dominated by primary osteons that are oriented longitudinally. Both the tibia and humerus display a few radial canals near the periosteum. Although almost all of the elements show significant color changes throughout the section, none exhibit orderly changes in vascular canal or osteocyte density. There is no evidence of LAGs or EFS in any of the bones from this animal. In all of the elements present, the collagen fibers are primarily woven. All bones from this individual are composed of mostly primary bone tissue and primary vascular canals. The collagen fiber orientation in all of the bones is woven. Therefore, I conclude that this animal was growing rapidly at the time of death. Because this animal appeared 76 to grow rapidly, there are no lines of arrested growth present, there is a low amount of remodeling, and the maximum preserved lengths of these bones are small relative to the largest examples of this taxon, I conclude that this animal is a juvenile individual. The study elements from the disarticulated individual found at locality 945 (UMNH VP 20454.5 - femur, UMNH VP 20454.3 - tibia, UMNH VP 20454.8 - humerus, UMNH VP 20454.1 - ulna, UMNH VP 20454.7 - rib, and UMNH VP 20454.7 - tendon) are all dominated by longitudinal primary osteons. In all of these specimens, osteocyte density was consistent through the bone. The ribs consist of mostly secondary osteons, but this likely reflects a higher rate of remodeling for these elements relative to limb bones. Once again, although all of the limb bones show distinctive color changes, none of these bones contain LAGs. All of these limb bones are dominated by woven collagen fiber orientation. The femur, tibia and humerus all have trabeculae composed of primary bone. Because all of the bones from this animal are composed of primary bone tissue and primary vascular canals, and the collagen fiber orientation is woven, I conclude that this animal was growing rapidly at the time of death. Because of this rapid growth, the lack of remodeling, the absence of LAGs, and that the maximum preserved lengths of these bones are small relative to the largest examples of this taxon, I conclude that this animal was juvenile at the time of death. UMNH VP locality 942 is a disarticulated bonebed that contains the remains of a minimum of three individuals of Utahceratops. Although the bones from this site cannot be assigned to particular individuals, they do share several microstructure characteristics. All of the elements in the study sample (UMNH VP 16860 - femur, UMNH VP 16681 - tibia, and UMNH VP 16861 - indeterminate limb shaft) possess trabeculae that are made 77 of unordered primary bone. All three bones express woven collagen fiber orientation and are dominated by secondary osteons. Most of the vascular canals are longitudinal, but they all contain some radial canals as well. No LAGs were observed. The femur (UMNH VP 16860) as well as the limb shaft (UMNH VP 16861) are from adult individuals based on the high degree of remodeling and large size. The tibia sectioned from this site (UMNH VP 16681) represents a smaller, subadult animal based on the presence of some simple and radial canals, and its smaller size, indicating that its was growing faster than the femur at time of death. The elements sectioned from UMNH VP 12198 (femur, rib, and tendon) are from a single disarticulated individual of Utahceratops. All are dominated by secondary osteons with both the rib and the tendon being almost completely remodeled. Longitudinal vascular canals and dense osteocytes dominate all three elements. No LAGs are present. These characteristics, in addition to the large size of the femur, (complete length 874 mm and midshaft circumference 437 mm), suggests that this animal was an adult at the time of death. The femur fragments from the holotype specimen of Kosmoceratops (UMNH VP 17000) share many similar histological attributes with UMNH VP 12198. In both sections, trabeculae are extensive and fill the medullary cavity. Both are dominated by longitudinally-oriented secondary osteons. No LAGs are present. These data suggest the individual was an adult. UMNH VP 21339 is a femur from a referred disarticulated specimen of Kosmoceratops. This bone is dominated by secondary osteons that are primarily longitudinally oriented. The osteocytes are dense throughout the bone, although these values represent remodeled tissue. This animal has not deposited LAGs or EFS. 78 These data, along with the large size of the femur, suggest the individual was an adult. The indeterminate ceratopsid humerus (UMNH VP 16865.1) from UMNH VP locality 512 primarily consists of secondary osteons, but some primary osteons are present. The orientation of the canals changes from longitudinal to more circumferential moving nearer to the periosteum. No LAGs or EFS are present. Taken together, these data suggest the animal was still growing when it died. Perhaps this individual represents a subadult. This determination as a subadult is strengthened by the presence of more remodeling present in this bone than seen in the bones indicated as belonging to juvenile individuals. The second indeterminate ceratopsid humerus (UMNH VP 19490) possesses some similar histologic characteristics, but possesses a higher proportion secondary osteons, indicative of extensive remodeling. The vascular canals are predominantly longitudinal in orientation. This animal was most likely a subadult at the time of death based on the remodeling. In sum, the bones sampled in this study represent a distribution of ontogenetic stages for Utahceratops and Kosmoceratops. The two Utahceratops from UMNH VP locality 945 (UMNH VP 20444 and UMNH VP 20454) are juvenile individuals of Utahceratops based on the high degree of vascularity, the low degree of remodeling, and the relatively short limb bone lengths. Utahceratops specimens representing a range of age classes were preserved at UMNH VP locality 942. The femur sampled (UMNH VP 16860) as well as the limb shaft (UMNH VP 16861) are from adult individuals, based on the high degree of remodeling (Fig. 18) and large size. The tibia sectioned from this site (UMNH VP 16681) pertains to a smaller, subadult animal, with a more moderate amount of remodelling. UMNH VP 12198 is an adult Utahceratops, based on the density of 79 80 secondary osteocytes present in the cortex of this element as well as its overall size. Both individuals of Kosmoceratops (UMNH VP 17000 and UMNH VP 21339) represent adult individuals with extensive remodeling, and compare well histologically with the Utahceratops inferred adults. 81 Figure 10: Cross sections of the humeri in this study. A. UMNH VP 20454.8; B. UMNH VP 20444.1; C. UMNH VP 19490; D. UMNH VP 16865.1. Arrow A indicates the anterior direction of the element and the L points to the lateral direction of the element. 82 83 Figure 11. Cross sections of unlae sectioned for this study. A. Ulna UMNH VP 20454.1; B. Ulna UMNH VP 20444.2 84 30 mm A 85 Figure 12. Cross sections of femora sectioned for this study. A. UMNH VP 20444.4; B. UMNH VP 20454.5; C. UMNH VP 16860; D. UMNH VP 21339; E. UMNH VP 12198; F. UMNH VP 17000. A points towards the anterior side of the element and L points laterally. 86 87 Figure 13. Close up images of the collagen fiber orientation of three representative bones. All bone tissue exhibits woven collagen fiber orientation. A. Left femur, UMNH VP 20454.5, of a juvenile animal Utahceratops gettyi from LOC 945. B. Right Femur, UMNH VP 16860, of an adult Utahceratops gettyi from LOC 942. C. Femur fragment from UMNH VP 17000 of an adult Kosmoceratops richardsoni. 88 UMNH VP 16860 UMNH VP 17000 89 Figure 14. Cross sections of tibiae sectioned for this study. A. Tibia UMNH VP 20454.3; B. Tibia UMNH VP 20444.5; C. Tibia UMNH VP 16681. A points towards the anterior side of the element and L points laterally. 90 91 Figure 15. Cross section of the ribs and ossified tendons sectioned for this study. Note the secondary canals that dominate these bones. A. A rib sectioned from UMNH VP 12198; B. An ossified tendon from UMNH VP 12198; C. A rib sectioned from UMNH VP 20454.7; D. An ossified tendon section from UMNH VP 20454.9. The arrow pointing at A indicates the anterior side of the element. The arrow pointing at L indicates the lateral side of the element. 92 93 Figure 16. A. A cross section of the femur UMNH VP 21339 from Kosmoceratops richardsoni. B. A view of the bone microstructure at the periosteum. Note the secondary osteons that make up the bone tissue. 94 95 Figure 17. The change in vascularity in humerus UMNH VP 19490 from starting out large closer to the medullary cavity and becoming smaller and more dense near the periosteum. 96 97 Figure 18. Comparison between the microstructure of a juvenile Utahceratops gettyi, an adult Utahceratops, and an adult Kosmoceratops richardsoni. A. A femur, UMNH VP 20444.4, from a juvenile of Utahceratops gettyi from LOC 945. Note that the microstructure is composed of primary osteons. B. A femur, UMNH VP 12198, from an adult of Utahceratops. Note that the microstructure is composed entirely of secondary osteons. C. A femur, UMNH VP 17000, from an adult Kosmoceratops richardsoni. Note that the microstructure is composed entirely of secondary osteons. 98 DISCUSSION Comparison With Other Ceratopsian Taxa This study is the first to examine bone histology of chasmosaurine ceratopsid dinosaurs (Utahceratops gettyi and Kosmoceratops richardsoni), and therefore allows for the first time comparison with their sister clade Centrosaurinae (Centrosaurus apertus, Pachyrhinosaurus pertoreum, and Einosaurus procurvicornis), as well as with previously sampled smaller ceratopsians outside of Ceratopsidae (Psittacosaurus spp. and Protoceratops andrewsi). I reexamined first-hand the microstructure of Centrosaurus apertus based on slides first analyzed by Lee (2006, 2007a & c). Two tibiae were quantitatively examined for this study. The trabeculae in these tibiae are composed of primary bone. Simple canals and primary osteons are both present but secondary osteons dominate in a 5:15:80 ratio for the tibia TMP 79.11.56. These canals are primarily longitudinally oriented, but radial canals are also present, though less abundant. The vascular canals seem to alternate between cycles of longitudinal and radial orientation, as first observed by Lee (2007a). Circumferentially oriented canals are also present but not nearly as abundant as longitudinal or radial canals. For the elements I examined, the osteocytes present do not have an increasing or decreasing density from the medullary cavity to the periosteum. At least two lines of arrested growth (LAGs) are visible in the tibia TMP 79.11.56 (at 20.65 mm and 21.91 mm from center of the medullary cavity), and at least five LAGs visible in tibia TMP 66.10.36 (at 14.16, 15.67, 17.14, 18.88, and 19.94 mm from the medullary cavity). The bones sampled from Utahceratops and Kosmoceratops are similar in size to these bones sampled from Centrosaurus. Lee (2007a) observed a maximum of seven LAGs in Centrosaurus and concluded that Centrosaurus reached skeletal maturity in approximately 6 years. Vascular canal orientation and density, and osteocyte density of Centrosaurus match well with the condition in Utahceratops and Kosmoceratops. However, the repeated cycles of longitudinal and radial canals in Centrosaurus are not observed in either Kosmoceratops or Utahceratops. A major difference in the histology is the presence of multiple LAGs in Centrosaurus, and their absence in specimens of Utahceratops and Kosmoceratops, animals with the same or larger body size. Also, Utahceratops and Kosmoceratops possess more remodeling in the bone tissue of the adult animals then in Centrosaurus. In the adult specimens of Utahceratops (UMNH VP 12198 and UMNH VP 16861) and Kosmoceratops (UMNH VP 17000 and UMNH VP 21339), the secondary osteons permeate the entire bone tissue extending from the medullary cavity to the periosteum. In Centrosaurus (TMP 66.10.36, TMP 79.11.56, and TMP T-4) there is more primary bone throughout the bone tissue extending to near the periosteum. An even greater contrast to the Utah taxa is provided by Pachyrhinosaurus perotorum (Fiorillo and Tykoski, 2012), a centrosaurine ceratopsid from the Maastrichtian Prince Creek Formation in the North Slope of Alaska. Erickson and Druckenmiller (2011) sampled a femur, UAMES 3551, estimated to be ~74 cm in length, that contains 18 LAGs, a total of 19 growth stages, and there are indications that the individual was still actively growing when it died because there are partially formed 100 vascular canals at the periosteal surface. The bone tissue of this femur of Pachyrhinosaurus was composed largely of reticularly oriented primary osteons, with longitudinally oriented primary osteons comprising the final five bands. Secondary osteons are present near the medullary cavity and lessen in density towards the periosteum (Erickson and Druckenmiller, 2011)). Chinsamy (1994) provided a figure from an unidentified species of Canadian Pachyrhinosaurus. She notes the presence of LAGs and the figure shows dense osteocytes. The overall histology of Pachyrhinosaurus differs slightly from Utahceratops and Kosmoceratops as it contains less remodeling, more reticular vascular canals and specimen contains numerous LAGs. The only other histologically sampled centrosaurine is Einosaurus procurvicornis from the late Campanian of Montana; the study examined the tibiae of 16 individuals (Reizner, 2010). Einiosaurus contains highly vascularized bone tissue, predominantly longitudinally oriented canals, localized radial canals, and some circumferential canals present towards the periosteal surface (Reizner, 2010). No secondary osteons are present in the juveniles (average tibia circumference of 24.4 cm) but they are present in subadults (average tibia circumference 42.8 cm) (Reizner, 2010). This description of the animal's histology is very similar to that of Utahceratops and Kosmoceratops in that all three ceratopsians possess highly vascularized bone tissue, vascular canals that were predominantly longitudinally oriented, some radial canals, and dense secondary osteons throughout the bone tissue; however, no sampled Utahceratops and Kosmoceratops specimens contain LAGs. These data from three centrosaurines and two chasmosaurines suggest that all ceratopsids share the presence of rapid growth indicators such as dense osteocytes and 101 102 reticular and radially oriented vascular canals. Based on the number and spacing of LAGs, growth slowed for Einiosaurus at around 3-5 years of age, and the oldest individual in this study is about 6 years old (Reizner, 2010). Centrosaurus apertus also reached skeletal maturity in about 6 years (Lee, 2007). Because Utahceratops and Kosmoceratops do not have LAGs, an estimate of the age cannot be made based on the histology. The amount of remodeling differs in the large ceratopsians. Pachyrhinosaurus has secondary osteons near the medullary cavity that lessen towards the periosteum, Centrosaurus has secondary osteons that do not dominate the bone tissue of the adult animals, Einiosaurus has secondary osteons that do dominate the bone tissue of older subadults individuals but not in the juveniles, and the Utahceratops and Kosmoceratops adult individuals have secondary osteons that dominate the bone tissue. The presence of LAGs and other growth marks varies considerably in these animals, from 18 in Pachyrhinosaurus, 7 in Centrosaurus, 5 in Einiosaurus to none in Utahceratops and Kosmoceratops (Fig. 19). Only two nonceratopsid ceratopsians have been histologically sampled, Protoceratops and Psittacosaurus. Lee (2006, 2007a) and Makovicky et al. (2007) independently conducted histological studies of the basal neoceratopsian Protoceratops andrewsi from the Late Cretaceous of Mongolia, and I reexamined the specimens used in Lee's studies. Using LAGs observed in femora and fibulae, Makovicky et al. (2007) determined that Protoceratops' growth slowed around 9 to 10 years of age and ceased completely around 11-13 years; however, they did not provide detailed histological data on other microstructure (e.g., vascular canals and osteocytes). Lee (2006, 2007a) sectioned humeri, ulnae, femora, and tibiae and found that the circumferential bone growth rate for Protoceratops ranged from 1-15 mm/yr (Lee, 2007a) and that the age of skeletal maturity for Protoceratops was ~7 years (Lee, 2006). The Protoceratops specimens that I reexamined display few trabeculae that are composed of primary bone and are sometimes rimmed with secondary bone tissue. The cortical bone tissue is predominantly comprised of primary osteons with some simple canals dispersed throughout the whole cross section. These vascular canals are almost entirely longitudinally-oriented; in femur MPC-D100/530, there is a 12:1 ratio of longitudinal to radially oriented canals. Vascular canal density does not increase or decrease appreciably throughout each section; however, osteocyte density does increase throughout the section from the medullary cavity to the periosteum. For example the femur from PIN uncatalogued "F-2" has the lowest density of 1384 osteocytes/mm2 at 8.5 mm from the center of the medullary cavity, and 2769 osteocytes/mm2 near the periosteum, 12 mm away from the center of the section. LAGs are present in all the thin sections that were quantitatively re-examined. For example, there were a total of 4 LAGs in femur MPC-D100/530 (0.5, 11.5, 12.2, and 12.5 mm away from the medullary cavity center). An EFS was not observed in any of the specimens. Trabeculae that are comprised of primary bone rimmed with secondary bone tissue are histological features shared with Utahceratops and Kosmoceratops. However, there are many differences between the larger taxa and the smaller taxon. The larger taxa possess an abundance of secondary osteons, no consistent osteocyte density increase from the medullary cavity to the periosteum, and they do not possess LAGs. The smaller taxon possesses a majority of primary osteons (not secondary osteons), has an osteocyte density increase and does possess LAGs. 103 Erickson and Tumanova (2000) investigated the bone histology of Psittacosaurus mongoliensis, a basal ceratopsian from the Late Cretaceous of Mongolia. Sampled individuals ranged from juvenile through adult developmental stages (Erickson and Tumanova, 2000), and I reexamined these specimens. No secondary osteons are present in any of the elements, with the tissue composed entirely of primary osteons. In the femur PIN 698/1977, radial canals dominate. Erickson and Tumanova (2000) observed a pattern of predominantly longitudinal canal orientation in juvenile individuals shifting to predominantly reticular canal orientation in the subadult and the adult specimens possessing dominantly reticular canal orientation near the medullary cavity with radial canals nearer to the periosteum. In these thin sections, vascular canal density increases from the medullary cavity to the periosteum (Erickson and Tumanova, 2000; this study). Radial canals dominate the Psittacosaurus femur I quantitatively examined, PIN 698/1977 (length ~20 cm and estimated to have been about 8 years old), and no secondary osteons were present. Utahceratops and Kosmoceratops are composed of both longitudinal canals and secondary canals with rare radial canals, unlike Psittacosaurus. Erickson and Tumanova (2000) concluded that the specimens ranged in age from 3 to 9 years and Lee (2007) concluded that the same specimens grew to be 11 years, but these data are not directly comparable to Utahceratops and Kosmoceratops because the Kaiparowits Fm. taxa lack LAGs or other growth marks. Erickson et al. (2009) examined the bone microstructure of another species of Psittacosaurus, P. lujiatunensis from the Early Cretaceous Lujiatun bed of the Yixian Formation of Liaoning Province, China. In this study, fibulae from 26 specimens were sectioned (Erickson et al., 2009). The microstructure was not described in detail, but one 104 105 figured fibula (PKU VP 1056) (Erickson et al., 2009: fig. 3) appears to be highly vascularized. Lines of arrested growth are present but an EFS is not. According to Erickson et al. (2009), none of the specimens have an EFS, which indicates that these animals had not completely stopped growing. The specimens in this study were interpreted to range in age from neonates to 11 years old (Erickson et al., 2009). The lack of histological description and figures, as well as the sampling of different elements (e.g., fibulae vs. femur and tibia) make it difficult to compare with my data. It appears that basal ceratopsians grew more slowly than large quadrupedal ceratopsids. This is evidenced by a generally higher number of definitive growth lines prevalent throughout development of the basal ceratopsians as well as the smaller distances between those LAGs. In contrast, the large ceratopsids possess larger distances between LAGs, which indicate a lot more bone grew in one year for the ceratopsids than the basal ceratopsians. The vascular canal orientation is predominantly longitudinal for all species studied other than Psittacosaurus, which was dominated by radial canals. Utahceratops, Kosmoceratops, Protoceratops, and Psittacosaurus all possessed more radial canals than circumferential canals, Psittacosaurus having the most radial canals than all of the other ceratopsians combined. The vascular canal density for the femora studied of the smaller ceratopsians is an average 33.46 canals/mm2 for Protoceratops and 40.84 canals/mm2 for Psittacosaurus. This compared with the average vascular canal density for femora of the larger ceratopsids sampled with 18.67 canals/mm2 for Utahceratops and 18.68 canals/mm2 for Kosmoceratops. The average osteocyte density for the femora studied of the smaller ceratopsians is 1853 osteocytes/mm2 for Protoceratops and 2157.5 osteocytes/mm2 for Psittacosaurus. The average osteocyte density for the femora studied of the larger ceratopsids is 1437 osteocytes/mm2 for Utahceratops and 1650.3 osteocytes/mm2 for Kosmoceratops. Comparison With Other Archosaurs Extant crocodylians are poikilothermic ectotherms and grow significantly slower than birds (e.g., Case, 1978; Erickson et al., 2001; Padian et al., 2001). Crocodylian bones exhibit compact, dense tissue that contains sparse vascular canals and parallel-fibered collagen fiber orientation (Enlow and Brown, 1957, 1958), and numerous LAGs (Buffrenil, 1980; Ferguson et al., 1982). Utahceratops and Kosmoceratops did not grow in a similar way to extant or extinct crocodylians. Unlike most other reptiles, nonavian dinosaurs, including Utahceratops and Kosmoceratops, generally have complex, well-vascularized primary bone tissues and extensive secondary osteons, similar to birds and many large mammals today (Francillon-Vieillot et al., 1990; Padian et al., 2001). Hadrosaurid ornithopod dinosaur histology is well documented (Horner et al., 1999; Horner et al., 2000; Cooper et al., 2008). Maiasaurapeeblesorum and lambeosaurine Hypacrosaurus stebingeri from the late Campanian of Montana have limb bones that are predominantly composed of primary bone tissue and dense vascular canals (Horner et al., 1999; Horner et al., 2000; Cooper et al., 2008). These vascular canals are radial in orientation in younger individuals and change to longitudinal canals as the individuals get older. Trabeculae rimmed with secondary bone tissue, and the presence of LAGs and EFS in some animals, are all characteristics observed in both taxa (Horner et al., 1999; Horner et al., 2000; Cooper et al., 2008). Utahceratops and Kosmoceratops are similar in possessing dense vascularity, trabeculae rimmed with secondary tissue, and 106 rapid growth indicated by vascular canal density and orientation. However, sampled Utahceratops and Kosmoceratops differ in the absence of LAGs or an EFS. Chinsamy et al. (2012) conducted bone histological analysis of polar Edmontosaurus from the Prince Creek Formation of the Alaskan North Slope and compared it with Edmontosaurus from the high-latitude temperate region of the Horseshoe Canyon Formation in Alberta, Canada. The polar Edmontosaurus would have lived in a cold-temperate climate, where there would be no sunlight for up to 6 months, and reduced overwintering foliage (Chinsamy et al., 2012). Edmontosaurus specimens from Alaska (humeri, femora, and tibiae) possessed dense vascularity and a change in vascular canal organization from reticular to more circumferential in the compacta. These periodic textural shifts comprise up to 8 cycles of alternating reticular and circumferential bone tissue (e.g., femur of DMNH 22557). This shift from quickly deposited reticular canals to slow forming circumferential canals was interpreted as consistent with polar winter darkness, subsequent periodic reductions in nutritional quality of forage, and shifts in energy balance (Chinsamy et al., 2012). True LAGs were absent in these animals. No such shifts from reticular to circumferential vascular canals were observed in either Utahceratops or Kosmoceratops. The dense vascularity as well as the lack of LAGs is consistent with the histology observed in both taxa. These switches in vascular canal orientation were not consistently observed in the Horseshoe Canyon Formation sample of Edmontosaurus (Chinsamy et al., 2012). The cortical bone comprises reticular canals that switch to circumferential canals later in life; this is observed in all of the femora analyzed. In contrast, there are alternating cycles of circumferential and reticular vascular canals in the tibia, but they are not as numerous as 107 in the Alaskan sample. LAGs were also absent in these individuals. Chinsamy et al. (2012) propose that the variable histology observed in these animals reflect occasional difficulties in sourcing forage. Utahceratops and Kosmoceratops seem to grow more similarly to these temperate Edmontosaurus than to the polar dinosaurs, but lack this cycling of vascular canal orien