FRONTISPIECE.—-Reconstruction of Glyptolherium anzonae. SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY • NUMBER 40

Glyptodonts of North America

David D. Gillette and Clayton E. Ray

ISSUED DEC 211981

SMITHSONIAN PUBLICATIONS

SMITHSONIAN INSTITUTION PRESS City of Washington 1981 ABSTRACT Gillette, David D., and Clayton E. Ray. Glyptodonts of North America. Smithsonian Contributions to Paleobiology, number 40, 255 pages, 97 figures, 70 tables, 1981.—All known North American glyptodonts belong in the genus Glyptothenum Osborn, 1903 (Family Glyptodontidae, Subfamily Glyptodon- tinae). Junior synonyms are Brachyostracon Brown, 1912; Boreostracon Simpson, 1929; Xenoglyptodon Meade, 1953; and all assignments of North American specimens to Glyptodon Owen, 1838. The ancestral species is Glyptothenum texanum from the Early Pleistocene Tusker (Arizona) and Blanco (Texas) local faunas of the Blancan Land Mammal Age; G texanum is smaller and lacks many of the exaggerated features of the descendant species. The descendant species are G. anzonae (Blancan? and Irvingtonian); G. floridanum (Ranchola- brean); and two species known from isolated localities in Mexico, G. cyltndricum and G. mexicanum. The taxonomic validity of G. mexicanum is questionable. The geographic distribution and faunal associations of Glyptothenum clearly indicate tropical or subtropical habitats. North American glyptodonts exhibit extreme tendencies toward hypsodonty and homodonty in the dentition, and they lack both incisiform and caniniform teeth. They probably fed on soft vegetation near permanent bodies of water. Graviportal limb proportions and details of the gross osteology suggest slow and cumbersome locomotion, which probably precluded occupation of upland habitats. A substantial expansion in the number of specimens available for study has extensively improved our knowledge of the gross osteology of Glyptothenum. especially for G. texanum and G. anzonae.

OFFICIAL PUBLICATION DATE is handstamped in a limited number of initial copies and is recorded in the Institution's annual report, Smithsonian Year. SERIES COVER DESIGN: The trilobite Phacops rana Green.

Library of Congress Cataloging in Publication Data Gillette, David D. Glyptodonts of North America. (Smithsonian contributions to paleobiology ; no. 40) Bibliography: p. Supt. of Docs, no.: SI 1.30:40 1. Glyptodontidae. 2. Paleontology—North America. I. Ray, Clayton Edward, joint au­ thor. II. Title. III. Series: Smithsonian Institution. Smithsonian contributions to paleo­ biology; no. 40. QE701.S56 no. 40 IQE882.E2] 560s [569'.31) 80-607040 Contents

Page Preface v Introduction 1 Previous Work 3 Systematics 5 Subfamily GLYPTODONTINAE 10 Genus Glyptothenum Osborn . . 10 Glyptothenum texanum Osborn . 11 Glyptothenum anzonae Gidley 12 Glyptothenum flondanum (Simpson), new combination 14 Glyptothenum cylindncum (Brown), new combination 15 Glyptothenum mexicanum (Cuataparo and Ramirez), new combination 16 Distribution 18 Osteology 27 Skull 27 Presacral Axial Skeleton 68 Front Limb 81 Pelvis 112 Hind Limb 124 Caudal Vertebrae 162 Carapace 168 Caudal Armor 194 Selected Aspects of Glyptodont Paleobiology 199 Functional Morphology 200 Habitat 207 Evolutionary History of Glyptothenum 208 Conclusions 209 Appendix: Tables of Measurements 212 Literature Cited 252

m We respectfully dedicate the present monograph to the late Ted Galusha, whose expert field work made this revision possible. Preface

"Mother," he said, "there are two new animals in the woods today, and the one that you said couldn't swim, swims, and the one that you said couldn't curl up, curls; and they've gone shares in their prickles, I think, because both of them are scaly all over, instead of one being smooth and the other very prickly; and besides that, they are rolling round and round in circles, and I don't feel comfy." "Son, son," said Mother Jaguar ever so many times, graciously waving her tail, "a Hedgehog is a Hedgehog, and can't be anything but a Hedgehog; and a Tortoise is a Tortoise, and can never be anything else." "But it isn't a Hedgehog, and it isn't a Tortoise. It's a little bit of both, and I don't know its proper name." "Nonsense!" said Mother Jaguar. "Everything has its proper name. I should call it 'Arma­ dillo' till I found out the real one. And I should leave it alone."—Rudyard Kipling, "The Beginning of the Armadillos"

Had Kipling chronicled the Pleistocene instead of modern times, he no doubt would have passed over the armadillos in this parable for the even more bizarre glyptodonts. For these extinct creatures of the past outwardly resem­ bled modern armadillos and turtles, and ecologically perhaps represented an animal somewhere between a land tortoise and a hippopotamus. Just as armadillos today are among the strangest of mammals, the glyptodonts of the Pleistocene, representing a branch of the same order, present a perplexing combination of unlikely characters. If their fossil remains had gone undiscov­ ered, glyptodonts could only have been the product of a poet's imagination, for their very existence could never be predicted within the confines of modern science and logic. Their improbable existence notwithstanding, glyptodonts comprised an important southern group of characteristic North American mammals of the Pleistocene Epoch. They were large and cumbersome, dwarfing the largest of modern armadillos. In many respects their anatomy reflects extreme adaptation for massive build and heavy weight. They have no living counterparts, except within our poetic imagination. Picture, if you will, a giant tortoise as a mammal, behaving as a lowland herbivore, with kinship to armadillos and the extinct ground sloths. Such are the glyptodonts, the North American representatives of which are reviewed here. FICURE 1.—Schematic illustration of skeleton oi Glyptothenum anzonae (modified for Glyptothenum after Burmeister and Hoffstetter). Glyptodonts of North America

David D. Gillette and Clayton E. Ray

Introduction found in other mammalian skulls. The bones surrounding the dental battery (maxilla, pala­ The extinct group of edentates commonly tine) are vertically expanded to accommodate the called "glyptodonts" (Greek: "carved tooth" or peculiar dentition, while bones of the basicran- freely, "grooved tooth") has perplexed students of ium are brachycephalic. The skull roof was pro­ natural history since their discovery early in the tected by an articulated set of dermal bones or 19th century. These former occupants of the New "cephalic shield." We propose that there was a World tropics and subtropics combine an unlikely large fleshy snout or a proboscis similar to that array of characteristics usually considered unique found in elephants, tapirs, and pyrotheres. to other vertebrates. The predominant feature is The legs, especially the hind limb, are elephan­ their turtle-like shell, or carapace (Figure 1). Dif­ tine. Glyptodonts surpass the proboscidean pro­ fering from the typical armadillo armor by being portions classically illustrated as graviportal and almost entirely immobile, the glyptodont cara­ represent one of the "most graviportal" groups pace afforded excellent protection. Attendant ever to have lived. Their large tail probably specializations in the axial skeleton are conver­ functioned as a counterbalance in locomotion, as gent with chelonian anatomy: extensive fusion of an accommodation for the lack of pelvic and most of the vertebral column and fusion of the axial mobility. The caudal vertebrae of the North pelvic girdle to the carapace, rendering the pelvis American representatives were encased in com­ entirely immobile. plete articulating rings of bony scutes. Despite The glyptodont dentition resembles in its con­ some reconstructions for North American glyp­ struction that found in certain rodents, notably todonts showing an elaborate terminal mace, this microtines and capybaras. But glyptodont teeth construction is found only in some of their South lack enamel, as do those of other edentates, and American relatives. Instead, the tail of North glyptodonts also lack incisiform and caniniform American representatives ended in a blunt ter­ teeth. Glyptodont cheek teeth are the "most hyp- minal tube formed by the fusion of two or three sodont" and the "most homodont" among terres­ caudal rings. trial mammals. Classically included as members of the extinct The glyptodont skull has modifications not Pleistocene megafauna, glyptodonts reached

David D. Gillette, Shuler Museum of Paleontology, Department of North America relatively late in the history of the Geological Sciences, Southern Methodist University, Dallas, Texas family. Already diversified and widespread in 75275. Clayton E. Ray, Department of Paleobiology, National Mu­ South America in the Miocene, it was at the close seum of Natural History, Smithsonian Institution, Washington, D. C. 20560. of the Pliocene Epoch or perhaps in the earliest SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Pleistocene that glyptodonts finally invaded for at least two other North American species, North America. There they experienced only and probably for two others as well. Without modest success, surviving until the Late Pleisto­ these specimens and the courteous assistance pro­ cene. The North American representatives are vided by members of the American Museum of descendants of probably a single invasion of their Natural History, in particular that of Mr. Galu­ South American relatives into Central America. sha, this study could never have been initiated, Subsequent evolution from this early Pleistocene much less concluded. ancestry resulted in the North American species The present review is weighted heavily toward reviewed in the present study. osteology and taxonomy in the belief that estab­ Ecologically, glyptodonts indicate tropical or lishing adequate morphological and taxonomic subtropical habitats, with quiet or standing water foundations for future studies is of paramount and dense, lush vegetation. These requirements importance. are indicated by their anatomy, distribution, ABBREVIATIONS.—Specimens from the follow­ faunal associations, and geologic occurrences. ing institutions and departments have been ex­ Although present in only a few North Ameri­ amined for this review: American Museum of can Pleistocene faunas, wherever they exist glyp­ Natural History, Department of Vertebrate Pa­ todont remains are not likely to be overlooked, leontology (AMNH) and Frick Collection (F: for the scutes comprising their carapace number AM); Charleston Museum (ChM); Midwestern some 1800 or more per individual. Once a pa­ University, Collection of Vertebrate Fossils leontologist has seen an isolated glyptodont scute, (MWU); Texas Memorial Museum (TMM); he or she is unlikely to forget its construction; University of Florida, Collection of Vertebrate scutes are unmistakable and can scarcely be iden­ Fossils (UF), and Collection of the Florida Geo­ tified as belonging to any other organism. logical Survey [UF(FGS)]; University of Michi­ North American glyptodonts, previously re­ gan Museum of Paleontology (UMMP); collec­ ferred to as many as five genera, are herein tions of the former United States National Mu­ assigned to one genus, Glyptothenum, and five spe­ seum, (USNM) in the National Museum of Nat­ cies, G. texanum, G. arizonae, G flondanum, G cylin- ural History, Smithsonian Institution; West dncum, and nominally, G. mexicanum. Until now, Texas State University (WT), and the Johnston classification of North American glyptodonts has Collection (JWT). been largely inferential, as material has been ACKNOWLEDGMENTS.—We are deeply indebted generally inadequate to establish credible taxo­ for support and assistance to many friends among nomic assignments. whom the following deserve special recognition: Through the courtesy of the Department of Claude Albritton, Jr.; Lynett Anderson; Cather­ Vertebrate Paleontology, American Museum of ine Casey; Walter W. Dalquest; Robert J. Emry; Natural History, several excellent glyptodont Ralph Eshelman; the late Ted Galusha; Robert specimens from Arizona, previously unreported, Habetler; Billy Harrison; the late Claude W provided a solid foundation for the present re­ Hibbard; Nicholas Hotton III; Lawrence Isham; view. This collection includes several complete Karoli Kutasi; Joshua Laerm; Wann Langston, and nearly complete carapaces, one of which is Jr.; Arnold Lewis; Everett Lindsay; Ernest L. associated with a nearly complete skeleton, the Lundelius, Jr.; Malcolm McKenna; William first such recovered in North America. These Melton; Gary Morgan; Stanley J. Olsen; Frank specimens were collected under the able direction Pearce, Robert W. Purdy; Albert Sanders; Bobb of the late Ted Galusha of the Frick Laboratory Schacffer; Bob H. Slaughter; Gladwyn Sullivan; in the years 1938-1939, 1943, 1948, and 1954. Beryl Taylor; Richard Tedford; S. David Webb; These Arizona glyptodonts are identified as Glyp­ John A. Wilson; and Jiri Zidek. We wish partic­ tothenum texanum, representing the ancestral stock ularly to thank the members of the Department NUMBER 40 f of Vertebrate Paleontology at the American Mu­ (1889), who named a new species, Glyptodon seum of Natural History for making available the nathorsti from Ejutla, Oaxaca, and by Brown superb material from Arizona, which constitutes (1912), who named and described Brachyostracon the principal basis for this study. As always, they cylindricus from Ameca, Jalisco, and proposed that have been generous with specimens, data, and Glyptodon mexicanus pertained to his genus. Mal- their time. Most of hundreds of hours of delicate donado-Koerdell (1948) rejected Glyptodon nathor­ laboratory preparation of these and other speci­ sti Felix and Lenk, considering it a junior syn­ mens for this study was done by Gladwyn B. onym of Brachyostracon mexicanus. Hibbard (1955) Sullivan with unfailing skill and good cheer. Law­ followed Maldonado-Koerdell in this regard, but rence Isham prepared Figures 1-4 and 95, and he he acknowledged that "until the range and vari­ provided expert advice on the preparation of ation of Brachyostracon mexicanus is known, how­ illustrations. Bonnie Dalzell prepared the frontis­ ever, there is as much justification for recognizing piece and assisted in the completion of some of B. nathorsti as for recognizing B. cylindricus" (Hib­ the figures. We are especially grateful to David bard, 1955:51). We recognize B. cylindricus (= Webb, Robert Emry, and Elaine Anderson for Glyptothenum cylindncum) and, nominally, B. mexi­ critical reviews of the manuscript. Bob Slaughter canus (= Glyptothenum mexicanum); the status of B. served as a catalyst in initiating our collaboration nathorsti (unseen) remains uncertain, and we also and in sustaining it through completion. provisionally follow Maldonado-Koerdell. David Gillette's 1974 research was supported Brown's (1912) descriptions were the last sub­ by a Smithsonian Institution predoctoral fellow­ stantial contributions on Mexican glyptodonts, ship, a National Science Foundation Doctoral although several others have recorded additional Dissertation Improvement Grant in the Field Sci­ recoveries of glyptodonts in Mexico (Hibbard, ences, and a Smithsonian Institution Short-Term 1955; Dalquest, 1961; Silva-Barcenas, 1969; Visitors Award. We are grateful for the opportu­ Mooser and Dalquest, 1975). New material of nities that these organizations have provided. good quality and reevaluation of the named Mex­ Besides those persons and institutions already ican taxa will be of great importance in the mentioned, we acknowledge gratefully the assist­ eventual improved understanding of glyptodonts ance of administration, scientific staff, and sup­ found in the United States, which form the pri­ porting personnel at the Institute for the Study of mary focus of the present investigation. Earth and Man, Southern Methodist University; Cope's (1888) report of an isolated glyptodont the Texas Memorial Museum, University of scute from Nueces County, Texas, which he Texas; Midwestern University, Wichita Falls, named Glyptodon petaliferus, has generally been Texas; the Charleston Museum; the National considered the first indication of glyptodonts Museum of Natural History (Smithsonian Insti­ north of Mexico. Two years earlier, however, tution); and the American Museum of Natural Cope (1886) had reported the recovery of several History. scutes and a phalanx from the Loup Fork beds of Kansas, designating them as a new genus and species Caryoderma snovianum. Apparently only Previous Work Flower and Lydekker (1891) acknowledged this Cuataparo and Ramirez (1875), two astute species; they identified the genus as Canoderma, civil engineers, were the first to report on the evidently a lapsus. Hay (1908:446) credited S. W. existence of glyptodonts north of South America, Williston for recognizing that these bones were in describing a complete carapace and associated reality chelonian; Hay assigned them to Testudo skeletal remains from the Valley of Mexico as and stated that they are in the collection at the Glyptodon mexicanus. Subsequent recoveries of Mex­ University of Kansas. We follow Hay's determi­ ican glyptodonts were reported by Felix and Lenk nation. SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Soon after Cope's reports, Osborn (1903) de­ established the certainty of Pleistocene glypto­ scribed Glyptothenum texanum from the Blanco beds donts in North America and the probability of of Texas. This is the genotypic species for all several taxa, glyptodonts received only sporadic known North American glyptodonts. The speci­ attention. Sellards (1940) maintained the unre­ mens described by Osborn were the first well- solved controversy concerning the existence of represented glyptodont remains recovered north Glyptodon in North America, referring several spec­ of Mexico. Osborn's description predated Brown's imens from Bee County, Texas, to Glyptodon petal­ (1912) description of Brachyostracon from Mexico, iferus. which is here regarded as a junior synonym of Meade (1953) appears to have been partly Osborn's genus although Brown's species is valid. influenced by prior remarks concerning the Hol- Following Brown's (1912) and Osborn's (1903) loman Gravel Pit (Oklahoma) glyptodont that definitive studies, an implicit controversy devel­ Hay and Cook (1930) had reported some years oped concerning the propriety of the South Amer­ earlier. Simpson (1929b) stated in a footnote that ican generic name Glyptodon for some of the North he had examined the Holloman specimens and American recoveries. The controversy has contin­ that they probably represent an undescribed spe­ ued up to the present time and, as demonstrated cies. Meade (1953), in reporting the Holloman below and in the taxonomy section, is not yet fauna, believed that the glyptodont belonged in adequately resolved. Hay (1916, 1926) reported a separate genus, and he accordingly established glyptodonts from two Texan localities, referring Xenoglyptodon fredencensis. them to Glyptodon petaliferus Cope. Simpson The discovery of glyptodonts in the Gilliland (1929b) commented on the improbability of the faunal horizon of the Seymour Formation in genus Glyptodon itself occurring in North America, northern Texas was reported first by Melton declared Glyptodon petaliferus a nomen nudum, and (1964) and soon thereafter by Hibbard and established for some Florida recoveries the genus Dalquest (1966). Melton briefly described this and species Boreostracon flondanus. He did not des­ extensive series of specimens, including excellent ignate a new name for the Texas fossils, however, skull material; he assigned them to the genus and their status remained unresolved until the Glyptodon and correctly recognized their identity present study revealed their identity as equivalent with the Holloman glyptodont. The comprehen­ to the Florida fossils. Simpson (1929b) also ques­ sive study of the Seymour faunas by Hibbard and tioned the status of the Arizona species Glypto­ Dalquest (1966) firmly established the contem­ thenum anzonae which Gidley had described a few poraneity of the Seymour and Holloman faunas, years earlier. Simpson apparently believed that lending support to Melton's specific determina­ the Arizona species represented a distinct and tion. Comparing the Seymour glyptodonts to unnamed genus. published accounts of South American represent­ At the time of Gidley's (1926) description of atives, Melton found the closest resemblance to Glyptothenum anzonae a second complete carapace the genus Glyptodon, in the classic sense, and he from the type locality had not been prepared, therefore referred to Xenoglyptodon as a junior syn­ and lay in storage in the American Museum of onym; he retained Meade's species, however, re­ Natural History. This carapace was prepared at cording the Seymour and Holloman glyptodonts approximately the time of publication of Gidley's as Glyptodon fredericensis. (1926) report, but he evidently never had the A comprehensive examination of North Amer­ opportunity to examine it, for if he had, he ican glyptodonts in the present study indicates certainly would have altered his ideas concerning that the Seymour and Holloman glyptodonts are the Arizona species. inseparable from Glyptothenum anzonae. At the Interest in North American glyptodonts soon time of Melton's (1964) and Hibbard and abated. Once Gidley, Hay, and Simpson had Dalquest's (1966) studies, it was believed that the NUMBER 40

Seymour fauna represents the mid-Pleistocene, South Carolina, identified here as Glyptothenum Kansan glacial age. Thus there was believed to flondanum; and that of Lundelius (1972), who have been representative species from the Blan­ discussed the glyptodont in the Late Pleistocene can {Glyptothenum texanum), early Irvingtonian (G. Ingleside fauna from San Patricio County, Texas. anzonae), late Irvingtonian [Glyptodon fredericensis), Lundelius's study is the most recent taxonomic and Rancholabrean (Boreostracon floridanus). With review of North American glyptodonts. the revision of the age of the Seymour fauna by Hibbard and Dalquest (1973), it is clear that the Curtis Ranch fauna (containing the early Irving­ Systematics tonian glyptodont) is not as greatly separated in time from the Seymour fauna as previously be­ Taxonomy of North American glyptodonts has lieved, for Hibbard and Dalquest (1973) have been based primarily on carapacial and pelvic proposed that the Seymour Gilliland fauna is pre- characters, a consequence of the limited number Kansan in age. The glyptodonts support their and character of specimens available. It is fortu­ contention, for there are no significant differences nate that glyptodonts were not recognized in between the Seymour and Curtis Ranch repre­ North America until late in the 19th century, sentatives. Thus Glyptodon fredericensis is here con­ when the number of proposed species and genera sidered to be a junior synonym of Glyptothenum for South American glyptodonts reached its ze­ anzonae. nith. At that time several authors (e.g., Burmeis- ter, 1874; Lydekker, 1894) sought to evaluate the Perhaps the principal cause of confusion re­ existing array of names from a more "balanced" garding the taxonomy of North American glyp­ perspective. The origin of the confusion lay in the todonts has been the general lack of cranial ma­ insistence of previous workers on establishing new terial. Only the incomplete skull remains of the species (and even genera) whenever a new cara­ Wolfe City, Texas, glyptodont described by Hay pace, or a partial carapace, was found. Appar­ (1916) were known with any measure of confi­ ently this early tendency of excessive "splitting" dence prior to Melton's (1964) report of the Sey­ occurred because no two individual glyptodont mour collection, which includes some excellent carapaces are exactly alike, nor are restorations skull material. Lacking adequate reference ma­ ever exactly alike, which undoubtedly affects a terial for the Seymour skulls, Melton resorted to description despite an author's awareness that a published descriptions of South American glyp­ particular aspect is not exactly correct. The pleth­ todont skulls. Melton's study was the first attempt ora of names evidently exasperated those who to place taxonomy of North American represent­ sought an overview, as expressed by Lydekker atives on a foundation comparable to that estab­ (1894:3): lished for South American glyptodonts, for which cranial and carapacial elements have received A large number of the so-called genera and species have approximately equal attention. been established on the evidence of such fragmentary and The addition of several excellent skulls of the imperfect specimens that it is frequently almost or quite impossible to determine to what forms they really belong, Arizona population oi Glyptothenum texanum in this and I have accordingly made no attempt to give the complete study has substantially modified the conclusions synonomy of a group whose study has been made unneces­ reached by previous workers concerned with sarily complex by incompetent workers. In addition to the North American glyptodonts and has revealed fact that many nominal genera and species have been formed the close relationship among all North American upon the evidence of specimens that ought never to have representatives. been described at all, others have been made by taking portions of the carapace from a different region from that to Recent reports on North American glyptodonts which the type of the form to which they really belong include that of Ray (1965), who recorded a par­ pertained. Others, again, have been established on the re­ tial skull of a glyptodont recovered from coastal mains of immature or young animals; and we thus have, in 6 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY what is really a rather small group, a host of meaningless minants. Skulls and dentitions have been largely names which it is almost impossible to correlate. unknown, even among South American taxa, It was approximately at the time of Lydekker's thus necessitating subordination of these tradi­ study that the existence of glyptodonts in North tionally important skeletal parts. We rely heavily America became established, and North Ameri­ on characteristics of the skulls and dentitions as can authors wisely were reluctant to counter Ly­ principal taxonomic features, thus placing North dekker's (1894) and Burmeister's (1874) admoni­ American glyptodont taxonomy on a footing sim­ tions. Indeed, only three names have been pro­ ilar to that of most other mammalian taxa. The posed for carapacial remains alone, to the exclu­ detailed study of the total osteology of North sion of noncarapacial characters. These are Glyp­ American representatives reveals their close, todon rivipacis Hay, Glyptodon petaliferus Cope, and rather than distant, relationships. Skulls and den­ Boreostracon flondanus Simpson. titions receive important consideration in the tax­ Glyptodon rivipacis Hay obviously was proposed onomy proposed here. This new information is improperly and has not been considered as valid. augmented strongly by the reexamination of car­ Because it was proposed for some of the glypto­ apacial and pelvic features, revealing that many dont scutes from Peace Creek, Florida, without of the differences previously considered to be designation or description, Simpson (1929b) cor­ fundamental indications of distant relationships rectly considered Glyptodon rivipacis a nomen nu­ are either misinterpretations or unjustified infla­ dum. tion of the relative importance of noted distinc­ Cope's species was based on an isolated and tions. fragmentary scute. Recognizing the futility of To the category of misinterpretations belongs taxonomic characterization based on isolated the taxonomic framework based on carapacial scutes, Simpson (1929b) also declared Glyptodon features. Improper restoration of the type speci­ petaliferus a nomen nudum, implicitly replacing mens of Glyptothenum anzonae, as an example, has this name with Boreostracon flondanus. (Simpson been important. Subsequent misinterpretations also believed that the Texas glyptodonts, referred of Gidley s (1926) description, owing to its incom­ to G. petaliferus by Cope, 1888, 1889, and by Hay, pleteness and a misleading photograph, soon fol­ 1916, 1926, actually belonged to a species distinct lowed, and became incorporated as fact. from the Florida species, but he did not designate Another example of misinterpretation of cara­ a new name.) pacial features resides in the series of specimens Boreostracon flondanus Simpson was erected for a that Simpson (1929b) designated as paratypes for large series of partial carapaces and isolated scutes Boreostracon flondanus. As postulated in the osteo- from Florida. Simpson effectively expanded logical description of the carapace, most of the Cope's procedure of reliance on carapacial fea­ specimens Simpson chose apparently represent tures, compounding the impropriety of taxo­ female individuals, while carapacial remains of nomic designation of type specimens without males were regarded as "abnormally large." Had noncarapacial remains. Because the hypodigm of his description included the numerous large Simpson's species is distinctive and fully repre­ scutes, there likely would have been early recog­ sentative, this name is here retained. nition of the congeneric relationship of his species Brown (1912), Gidley (1926), and Lundelius with other North American species. Instead, be­ (1972), among others, have considered noncara­ cause Simpson's series heavily favored the smaller pacial elements in their taxonomic determina­ scutes (because they were by chance in most tions. The unusual situation of having pelvic extensive articulation) those of the males have elements for almost all of the various North Amer­ been ignored, and Simpson's claim of generic ican populations has prompted the consideration distinction has been perpetuated. of pelvic characters as principal taxonomic deter­ To the second category, that of unjustified NUMBER 40 inflation of the importance of noted distinctions, imens with only two iliosacrals, the posterior lum­ belongs the usage of pelvic characters in separa­ bar vertebra participates more strongly in the tion of higher categories. For example, Brown pelvic structure than does the corresponding ver­ (1912), in part because of the nature of the pelvis, tebra of the specimens with three iliosacrals. Es­ not only established the genus Brachyostracon, but pecially in 53321, this vertebra possesses greatly he also placed this genus in a family separate expanded transverse processes that contact the from the only other North American species ilium broadly, allowing participation of this ver­ (Glyptothenum texanum) known at the time. Famil­ tebra as an anterior "pseudo-sacral" vertebra. ial and subfamilial separation on the basis of There is similar variation in the construction pelvic characters is unwarranted and even generic of the ischiosacral union in the same sample. In separation on the basis of pelvic characters is here all six specimens there are four ischiosacral ver­ denied, in the belief that carapacial and dental tebrae that participate in this union by fusion of similarities far outweigh the importance of pelvic their expanded transverse processes with the is- distinctions. This assertion is founded on the reex- chium. For three of the males (53321, 244905, amination of the carapaces, dentitions, and pelves 244906) the transverse processes of the terminal of North American representatives; the support­ sacral vertebra are solidly fused both with the ing arguments are drawn from the gross osteology ischium and with the transverse process of the of these and other skeletal regions. The question adjacent penultimate sacral vertebra. In one spec­ of relative importance of pelvic characters de­ imen (244905) the centrum of the last sacral serves further comment, however, since the pelvis vertebra is fused to that of the penultimate ver­ has received considerable attention. tebra; the centrum is unfused in the other two It is impossible to determine whether there is males despite the ankylosis of their transverse sexual dimorphism in glyptodont pelves on the processes. basis of the small North American sample. Nor is In marked contrast, the transverse processes of it possible with certainty to evaluate the differ­ the other male and the two females (256761, ences in construction and variations in the num­ 256760, and 49398, respectively) are fused only ber of vertebrae participating in the several com­ with the ischia and not with the transverse ponent regions of the pelvis. Because the pelves processes of the penultimate sacral vertebra. In are so well represented, and because earlier au­ all three specimens the last centrum is not fused thors have used characteristics of the pelvis in to the one anterior. Consequently, the terminal taxonomic determinations, it is appropriate to sacral vertebra, participating in the ischiosacral examine the pelves of two extant edentates. The union, more closely resembles the anterior caudal following digression is included to point up the vertebra (posterior "pseudo-sacral") than the ad­ weakness of dependence on pelvic characters in joining sacral vertebrae. Indeed, the fusion of the taxonomy of living edentates, and by analogy, of transverse processes with the ischia is incomplete the extinct glyptodonts. in 256760 and 256761, in which this contact In a small sample of six skeletal specimens of remains cartilaginous. In the fossil state, without the nine-banded armadillo, Dasypus novemcinctus, complete vertebral columns for reference, these of known sex (USNM Division of Mammals terminal sacral vertebrae could easily be mistaken 5332Id, 244905(5, 249906c?, 256760$, 25676Id, for the anterior caudal vertebra, and there would 49398?), there is significant variation in the ilio- therefore exist an artificial disparity in the ischio­ sacral and ischiosacral unions. In two males sacral count for these specimens. (256761 and 53321) there are only two iliosacral In all six specimens there are two free vertebrae vertebrae. In the other two males and in both in the sacral arch, although the participation of females there are three iliosacral vertebrae. Ap­ the transverse processes of the antepenultimate parently as partial compensation in the two spec­ vertebra is variable, from full participation, to an 8 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY estimated one-quarter participation (i.e., the ob­ To this point in the discussion, the congeneric turator foramen is more posteriorly situated). Re­ status of Boreostracon, Brachyostracon, and Glypto­ maining characteristics of the pelvis of D. novem- thenum has been mentioned. Glyptothenum Osborn, cinctus are relatively constant. The pubes are stout 1903, has priority, and this is the generic name in both sexes and they unite at the midline in a that is here applied to all known North American strongly ankylosed contact. There is little evident glyptodonts. variation in the compound curvature of the sac­ Xenoglyptodon Meade, 1953, was reduced to syn­ rum. In all six specimens the first two caudal onymy by Melton (1964), who correctly consid­ vertebrae resemble the "pseudosacral" construc­ ered the Holloman (Oklahoma) glyptodonts as tion of glyptodonts. identical to the Seymour (Texas) representatives Thus in the nine-banded armadillo, there is but referred the Seymour and Holloman speci­ considerable variation in the pelvis. None of the mens to the South American genus Glyptodon. variation is attributable either to age or sex; The presence of the genus Glyptodon in North instead, it appears to be simple individual varia­ America was first proposed by Cope (1888). Hay tion, and it evidently bears no taxonomic signifi­ (1916, 1926) and Sellards (1940) followed Cope's cance for this species. assertion, while Brown (1912), Gidley (1926), Similarly, in two male specimens of the giant Simpson (1929b), and Holmes and Simpson armadillo Pnodonles gtganleus (USNM Division of (1931), among early North American authors, Mammals 261024, 299630), there is a marked denied the existence of this genus in the North difference in the contribution of the terminal American fauna. Students of South American sacral vertebra. In specimen 299630 the trans­ glyptodonts, apparently invariably, have denied verse processes are fused with the adjoining trans­ the existence of Glyptodon in North America, rec­ verse processes of the penultimate vertebra and ognizing only Glyptothenum and Boreostracon. Mel­ with the ischia. In the other specimen the trans­ ton's (1964) resurrection of the generic name verse processes are free, remaining unfused with Glyptodon for the Seymour glyptodonts is therefore either the ischia or the adjoining transverse questionable from a historic standpoint. More processes. This variation cannot be an age-related important, however, is the fact that the North phenomenon, for the individual with the free American glyptodonts are all closely related, and posterior sacral vertebra appears to be older than apparently represent evolution from a single early the other. The pelves of these two individuals are Pleistocene immigrant. At least for three species, otherwise identical. Glyptothenum texanum, G. anzonae, and G. floridanum. Thus by analogy with armadillos, it is likely an ancestral-descendant relationship can be pro­ that the amount of intraspecific variation in the posed with substantial basis. The other two spe­ glyptodont pelvis is greater than Brown (1912) cies, G. cylindricum and G. mexicanum, of which the had supposed. Accordingly, the differences in the latter is nominally retained, are not as easily pelves of the North American representatives are placed in the sequence, primarily because they not necessarily indicative of supraspecific distinc­ are poorly known. They appear to represent a tion. Variation in vertebral participation in the species (or, nominally, two species) contempora­ sacral arch should not be accorded principal tax­ neous with the more northern Late Pleistocene G. onomic significance, nor should the size of the floridanum, if not actually pertaining to the same pubic cross bar be considered as a primary taxo­ species. Thus the North American glyptodonts nomic character. Variations in the pelvis are here are not only closely related, but they also appear proposed as no more than species distinctions, for to have an isolated history, apparently quite apart there is no justification for basing generic diag­ from South American relatives. Moreover, the noses on the pelvis. The amended diagnoses re­ early origin of the North American lineage (Early flect these contentions. Pleistocene-Blancan Land Mammal Age) coin- NUMBER 40 cides temporally with only one South American Finally, application of the generic name Glyp­ species of Glyptodon, G. laevis Burmeister; the re­ todon to North American fossils would suggest a maining South American species are listed by closer relationship than the evidence at hand Castellanos (1954) as Middle or Late Pleistocene warrants. The North American glyptodonts, in and, therefore, cannot be considered as ancestral all respects members of the subfamily Glyptodon- or closely related on temporal grounds. The re­ tinae, are distinct from their South American lationship of the ancestral North American spe­ relatives, and the existing lower level taxonomy cies, Glyptothenum texanum, with the apparently should reflect this situation by retention of a contemporaneous Glyptodon laevis (contempora­ separate generic name. The existence of so-called neous in the broad sense, and accepting Castel­ Boreostracon in South America is quite another lanos' (1954) affirmation of the validity of this matter, and there is no reason to deny the possi­ species) of South America seems to be no closer bility of secondary invasion of the North Ameri­ than that between G. texanum and other South can representatives back into South America. American species of this genus. Nor does there Although the taxonomy proposed here is a appear to be any closer relationship between the substantial alteration of the previously accepted North American species, collectively or individ­ classification of North American glyptodonts, the ually, with the species oi Glyptodon than with other changes are supported by a great expansion in South American species in the subfamily Glyp- the number of specimens, including skulls and todontinae. dentitions. The upper level classification follows The preceding comments assume validity of Simpson (1945) and Castellanos (1954). the genus Glyptodon, a matter that in itself, and quite apart from taxonomic problems in North Order EDENTATA Cuvier, 1798 America, deserves special attention. The status of Suborder XENARTHRA Cope, 1889 the generic name Glyptodon has been reviewed by Infraorder CINCULATA Illiger, 1811 Superfamily GLYPTODONTOIDEA Simpson, 1931 Hoffstetter (1955), who discussed the inappro­ Family GLYPTODONTIDAE Burmeister, 1879 priate circumstances surrounding the origin of Subfamily GLYPTODONTINAE Trouessart, 1898 this name. The generic name Glyptodon has been Genus Glyptothenum Osborn, 1903 applied, classically, to a group for which the Glyptothenum texanum Osborn, 1903 original description (Owen, 1838) does not apply, G. arizonae Gidley, 1926 at least in part. Hoffstetter recommended desig­ G. floridanum (Simpson, 1929) nation of a lectotype corresponding to the rear G. cylmdricum (Brown, 1912) foot that Owen described and that pertains in G. mexicanum (Cuataparo and Ramirez, actuality to the genus in the classic sense (that is, 1875) as has been accepted by subsequent common As acknowledged by Simpson (1945) and usage). Whether the recourse Hoffstetter has sug­ Hoffstetter (1958), among others, the family gested, which seems altogether reasonable, will name Glyptodontidae Burmeister, 1879, has gained the advantage of usage over Hoplophori- gain future acceptance is uncertain. For the pres­ dae Huxley, 1864. Except for the problematical ent, however, it would be premature to assign Early Tertiary genus Palaeopeltis Ameghino, 1895, North American glyptodonts on the basis of a all glyptodonts belong in the family Glyptodon­ South American genus of uncertain status. Cer­ tidae. The taxonomic position of Palaeopeltis has tainly the group of glyptodonts to which the not been resolved; some (e.g., Castellanos, 1954) name Glyptodon has been applied is a unified and assign this genus to a separate family, while others characteristic group, and it has been the subject (e.g., Simpson, 1945) conservatively declare its of several taxonomic reevaluations among which status to be incertae sedis. that of Castellanos (1954) is pertinent to the Simpson (1945) recognized four subfamilies of taxonomy of North American representatives. Glyptodontidae, whereas Hoffstetter (1958) rec- 10 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY ognized five by choosing to split off two genera conical or tuberculated; terminal armor anky- from Simpson's recognized subfamily Propalaeo- losed into a short tube (Hoffstetter, 1958, stated hoplophorinae Castellanos, 1932, to be included that the caudal armor consists of only seven in a separate subfamily. The remaining three individual rings; evidence presented herein indi­ subfamilies seem to be well established: Hoplo- cates that the number of caudal rings is variable); phorinae Weber, 1928 (this is Simpson's, 1945, (5) skull large, tall, and abruptly truncated be­ preference; Hoffstetter, 1958, retains the older cause of shortening of the nasals, premaxillae, name Sclerocalyptinae Trouessart, 1898, for the and anterior end of mandible; (6) narial aperture same assemblage); Doedicurinae Trouessart, trapezoidal and tall; (7) upper profile of the skull 1898; and Glyptodontinae Trouessart, 1898. straight and inclined forward; (8) zygomatic arch Brown (1912) believed that his genus Brachyo­ rather weak posteriorly; descending processes stracon belonged in a family separate from Glypto­ massive; (9) postorbital bar incomplete; (10) teeth thenum. Simpson (1945) formalized this belief by all trilobed, but anterior ones smaller; (11) osteo- placing Brachyostracon in the subfamily Hoplo- dentine cores form numerous secondary ramifi­ phorinae, and the remaining North American cations, a condition not known in the other taxa in the subfamily Glyptodontinae. Hoffstetter subfamilies; (12) humerus lacks the entepicon- (1958) recognized the closer relationship between dylar bridge, which is present in the other Brachyostracon and the other North American gen­ subfamilies; (13) manus with no more than four era, placing it also in the subfamily Glyptodon­ digits; pes with five complete digits. tinae. All of the North American glyptodonts exam­ Castellanos (1954) divided the subfamily Glyp­ ined in the present study conform to Hoffstetter's todontinae into three tribes, according to which (1958) characteristics for this subfamily. Some of the North American genera Glyptothenum and Bor­ the features listed below in the diagnosis for eostracon were separated. He apparently also be­ Glyptothenum might also pertain to the entire lieved that Brachyostracon belonged in a separate subfamily. subfamily, for he did not include it in the Glyp­ todontinae (Castellanos, 1953, 1954). According to the determinations of the present investigation, Genus Glyptotherium Osborn all North American glyptodonts belong in the "Glyptodon' of Cuataparo and Ramirez, 1875:362.—Cope, genus Glyptothenum Osborn, 1903, and the matter 1888:346.—Hay, 1916:107.—Sellards, 1940:1637.—Mel­ of tribal designation deserves reinterpretation. ton, 1964:131. Glyplolhtnum Osborn, 1903:492—Gidley, 1926:96— Aker- sten, 1970:13. Subfamily GLYPTODONTINAE Brachyostracon Brown, 1912:169. Boreostracon Simpson, 1929b:581.—Lundelius, 1972:36. Hoffstetter (1958) has provided the most com­ Xenoglyptodon Meade, 1953:455. prehensive review of the characters of this subfamily. The following features are some of the AGE.—Late Pliocene to Late Pleistocene; Late more pertinent ones that distinguish the Glypto­ Blancan through Rancholabrean Land Mammal dontinae from the other subfamilies: (1) carapace Ages. divided into long thoracic region and short lum- DIAGNOSIS.—Size medium to large; skull short, bosacral region; (2) scutes thick, with simple or­ truncated, dorsal profile flat, anteriorly inclined; namentation; (3) marginal scutes mostly differ­ nasal aperture trapezoidal; postorbital bar in­ entiated into a highly tuberculatcd construction; complete; digit I, manus, absent, digits II, III, IV (4) caudal armor composed of movable rings, large, digit V reduced; digits I and V, pes, small, each formed by two or three rows of plates, of reduced, digits II, III, IV large; humerus lacking which those in the posterior row are sometimes entepicondylar foramen; caudal vertebrae un- NUMBER 40 11 fused, each protected by individual, movable JWT 2387, 10 articulated scutes; JWT 1907, 9 rings, except the "sacrocaudals" and the last two articulated scutes; JWT 1706, 10 articulated or three, which may be fused into a short tube; scutes, 10 isolated scutes; JWT 2356, 2408, 2477, scutes of distal row of each caudal ring weakly to 2501, 2578, 22 isolated scutes; JWT 649, isolated strongly conical; first caudal ring incomplete; rear ungual phalanx. chevron distal articulation with caudal ring of Red Light Local Fauna, Love Formation, Hud- next posterior vertebra; carapace short, arched, speth County, Texas: TMM 40961-1, partial car­ usually with posterior recurved outline; cephalic apace and numerous isolated scutes; TMM boss may be present on carapace; carapacial 40664-245, distal end right humerus; TMM scutes firmly articulated, immovable except in 40664-109, ungual phalanx. anterolateral region; individual scutes polygonal, Hudspeth Local Fauna, Camp Rice Forma­ with rosette sculpturing; central figure larger tion, Hudspeth County, Texas: TMM 40254-1, than, or equal to, size of peripherals; one row 40254-2, 40254-3, 40255-2, isolated scutes. peripherals, with occasional intercalations of in­ Tusker Local Fauna, Graham County, Ari­ complete second row; central figures weakly con­ zona: F:AM 59581, carapace; F:AM 59583, par­ vex, flat, or weakly concave; peripheral figures tial skull; F:AM 59584, radius; F:AM 59589, two fiat; scutes of caudal border weakly to strongly scutes; F:AM 59599, carapace; F:AM 95737, conical; nuchal scutes usually unbossed; scutes of nearly complete skeleton and carapace; F:AM anterolateral region quadrilateral, movable; lat­ 59586, astragalus; F:AM 59584, phalanx; F:AM eral border scutes variously conical, pendant. 59585, radius. AGE.—Blancan Land Mammal Age, Late Pli­ ocene-Early Pleistocene. Glyptothenum texanum Osborn DIAGNOSIS.—Skull small; glenoid fossa smooth; paroccipital processes small; paroccipital canal Glyptothenum texanum Osborn, 1903:492, pi. 43.—Akersten, 1972:13, figs. 8d,e, 9a,b. lacking; basioccipital-basisphenoid contact in vertical plane of posterior nares; basioccipital- TYPE SPECIMEN.—AMNH 10704 carapace, basisphenoid union contiguous, in same plane; caudal vertebrae, caudal armor, pelvis, seven inferior border of foramen magnum entire, basi- chevrons. occipital not deeply clefted; squamous occipital TYPE-LOCALITY.—Llano Estacado, Texas; ex­ smooth and inclination shallow; occipital con- act locality uncertain; probably "Blanco Local­ dyles conical; hypoglossal canal lacking; ventral ity," Crosby County, Texas, of Johnston and extremity petromastoid rounded; medial parietal Savage (1955). region not elevated; mental foramen in plane of REFERRED SPECIMENS.—The following speci­ NT midlobe; inferior margin of horizontal ramus mens are referred to G. texanum Osborn. of mandible flattened; N~ ovoid; N~ trilobate; 8 Type-Locality: AMNH 10705, several scutes N--N anterior borders flattened; Ni elongate, (unseen); AMNH 20092, from Crawfish Draw, irregularly ovoid, situated on symphyseal curva­ Blanco Canyon, probably scutes (unseen). ture; Ng trilobate, submolariform; Ng submolari- Cita Canyon Locality, Randall County, Texas, form, obliquity of posterior lobe intermediate; N5 Bed no. 4: JWT 1723, tibiofibula; JWT 1712, anterior lobe relatively undeveloped, no anteroin- calcaneum; JWT 2330, fragmentary right and ternal sulcus; Ng -N7 lobes slightly oblique; Ng left mandibles, isolated teeth and tooth frag­ anteromedial and anterolateral faces convex, ments; JWT 1837, isolated rear upper tooth; apex blunt; presacral vertebrae unknown; acro- JWT 1715, nearly complete mandible with bro­ mion process of scapula symmetrical; propodials ken Ni-Ns; JWT 649, 13 articulated scutes, 119 and epipodials small and lacking greatly exagger­ isolated scutes; JWT 2319, 120 isolated scutes; ated features; humerus lacking supratrochlear SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY foramen; articular facets of manus and pes small, (paratypes), caudal vertebrae and caudal armor, curved; five or six vertebrae in lumbar tube; three portions of carapace, foot bones, broken teeth. iliosacral vertebrae; four free vertebrae in sacral TYPE-LOCALITY.—"About three miles east of arch; one or two ischiosacral vertebrae; pubic the Curtis Ranch, in Sec. 25, T. 18 S., R. 21 E., crossbar small; ischiac crests posteriorly conver­ Cochise County, Arizona" (Gidley, 1926). gent; compound curve of sacral arch not pro­ REFERRED SPECIMENS.—The following speci­ nounced; transverse processes of terminal sacral mens are referred to G. anzonae Gidley. vertebra directed obliquely rearward; anterior Type-Locality: AMNH 21808, complete cara­ surface of ilium strongly concavoconvex; cara­ pace and caudal armor, caudal vertebrae with pace small, symmetrical anteroposteriorly, not associated chevrons, pelvis, femur; AMNH 21809, highly arched; no cephalic boss on carapace; isolated scutes (unseen). border scutes not greatly enlarged; scutes small, Holloman Gravel Pit near Frederick, Tillman central figures convex, larger than peripherals, County, Oklahoma: TMM 934-37, fragmentary frequent deep hair follicles; posterior border of mandible with two broken teeth, two isolated carapace recurved weakly or not at all; central scutes; an unnumbered carapace in the Stovall figures of scutes usually convex, frequently with Museum, University of Oklahoma. deep medial depression; caudal and cephalic ap­ Seymour Formation, Gilliland Local Fauna, ertures taper laterally downward, not vertical; Knox County, Texas: MWU 1209, 1632, 1633, scutes of carapace disposed in transverse and 1667, 1668, 1798, 1799, 1861, 1862, 1980, 2307, oblique rows; posterior border scutes variable 2309-2315, 2317-2322, 2324, 2456, 2558, 2671, from flat to weakly conical; lateral border scutes 2672, 2674-2676, 2678, 7997, 7998, numerous only slightly enlarged, occasionally "pendant"; scutes, mostly isolated; MWU 1668, 2324, caudal nuchal scutes smooth, unbossed; caudal armor vertebrae, MWU 1634, sacral vertebra fragment; consisting of 11 separate rings of scutes, one ring MWU 1668, rib fragment, proximal end; UMMP per vertebra, including an incomplete anterior 33520, phalanx; UMMP 33524, tibiofibula; "accessory ring" and a terminal tube formed by UMMP 46474, isolated tooth; UMMP 46892, either the single terminal ring or by fusion of the 44704-44706, 44708, 44709, 22475, 46230, last two rings; scutes of caudal armor unbossed, 46240-46257. 46259-46264, 46267-46311, presenting uniformly scalloped distal outline, ex­ 46313-46237, 46341, 46343, 46346, 46347,46354, cept dorsal pair of scutes in rings 4-11, which are 46363, 46372, 46379, 46643, partial carapaces, conical and raised. caudal rings, isolated scutes; UMMP 44707, cau­ dal vertebra; UMMP 34826, skull, caudal verte­ brae, chevron bones, carapace scutes, caudal Glyptotherium anzonae Gidley rings, atlas, cervical tube; UMMP 38761, skull, Glyptothenum anzonae Gidley, 1926:96, fig. 4, pis. 40-44. mandible, complete rear foot, both radii, com­ "Glyptothenum" anzonae Gidley.—Simpson, 1929b:582. plete left front foot, incomplete right front foot; Glyptodon petaliferus Cope.—Hay and Cook, 1930:10. UMMP 46231, complete left rear limb; UMMP Xenoglyptodon fredericensis Meade, 1953:455, pi. 1. 46232, broken tibiofibula, two isolated scutes, rib; Glyptodon fredericensis (Meade). Melton, 1964:131, figs. 2-3, pis. 1-3. UMMP 46233, distal end of femur; UMMP 46234, fragmentary mandible and metatarsal II; TYPE SPECIMENS.—USNM 10536 (holotype), UMMP 46235, tibiofibula distal extremity; mandible with teeth, complete right front limb, UMMP 46236, patella; UMMP 46237, broken vertebral fragments including broken dorsal tube, mandibles and seven isolated scutes; UMMP carapace portions, isolated scutes from carapace 46238, caudal vertebrae; UMMP 46239, terminal and tail rings, complete right hind limb, incom­ caudal vertebra; UMMP 46258, mandible frag­ plete left hind limb; USNM 10537 and 10336 ments, rib and several isolated scutes; UMMP NUMBER 40 13

46329, metatarsal III; UMMP 46330, metatarsal with fused rib; dorsal tube formed by 10 thoracic III; UMMP 46331, navicular; UMMP 46332, vertebrae; acromion process of scapula asymmet­ navicular; UMMP 46333, phalanx I, digit I, rical; propodials and epipodials large and massive manus; UMMP 46334 metatarsal II with sesa- with exaggerated features; humerus with supra- moid bone; UMMP 46335, teeth fragments; trochlear foramen; articular facets, manus and UMMP 46376, distal end of femur; UMMP pes, large and curved; three iliosacral vertebrae, 46378, caudal vertebra; UMMP 46644, fragmen­ one or two ischiosacral vertebrae; pubic crossbar tary caudal vertebra. small; ischiac crests slightly convergent poste­ Specimens unseen: Glyptothenum sp. as listed in riorly; moderate sacral arch compound curve; Lindsay and Tessman (1974) in the Curtis Ranch transverse processes terminal sacral vertebra di­ fauna and equivalent faunules in Cochise rected obliquely rearward; anterior surface ilium County, Arizona. strongly concavoconvex; carapace large, highly AGE.—Irvingtonian Land Mammal Age. arched, divided into distinct preiliac and postiliac DIAGNOSIS.—Size large, skull very large; gle- regions; carapace with cephalic boss; scutes large, noid fossa with distinct foramen; paroccipital central figures of scutes slightly greater than half processes large, with deep paroccipital canal; the scute diameter, always slightly larger than basioccipital-basisphenoid contact posterior to peripherals, flat to weakly convex; border of an­ posterior nares; basioccipital-basisphenoid union terior aperture of carapace tapers laterally down­ obtuse, not contiguous in same plane; foramen ward and rearward, posterior aperture vertical; magnum inferior border entire; basioccipital Y- lateral border scutes large, conical; lateral scutes shaped; squamous occipital smooth and inclina­ of first interior row with rounded central boss; tion shallow; occipital condyles conical; hypo- caudal armor consisting of 11 or 12 rings, includ­ glossal canal lacking; ventral extremity petro- ing an incomplete anterior "accessory ring" and mastoid flattened; medial parietal region not el­ a terminal tube formed by either the terminal evated; mental foramen anterior to Ny; mandib- ring or by fusion of the last two rings; dorsal pair ular symphysis horizontal and flat; inferior mar­ of scutes of each ring very conical, other scutes of gin of horizontal ramus of mandible rounded; distal rows also conical and bossed; scutes of posterior margin ascending ramus of mandible anterior rings unbossed. not steeply inclined, parallel to anterior margin; REMARKS.—Large, isolated scutes from several N1 ovoid; N2 probably trilobate; N&-N8 anterior late Blancan to early Irvingtonian sites in Florida borders flattened; Ni molariform, nearly trilo­ most closely resemble the scutes of G. anzonae bate, situated behind symphyseal curvature; Ng from Texas and Arizona. We identify these scutes trilobate, molariform; Ng molariform, posterior as Glyptothenum sp., cf. G. anzonae Gidley. These lobe convex, markedly oblique, anterior lobe specimens and their localities are: Santa Fe River, squared; Nj fully trilobate, transverse axes of Gilchrist County, (UF localities SF 1, 4A, 8A), lobes oblique, posterior face concave, no anteroin- UF/FSM 10439, 10743, 10428, 10438, 19198, ternal sulcus; N5-N7 lobes oblique; Ng antero- 15122, isolated scutes, a broken tooth from a baby medial and anterolateral faces concave, anterior individual, and two caudal ring scutes; Inglis 1A, apex pointed, anterior lobe widest, middle and Citrus County, UF/FSM 20568, one scute; Rey­ posterior lobes progressively narrower, posterior nolds Point, Charlotte Harbor, Charlotte County, face concave, oblique; atlas free; alar processes UF/FSM 2774, three scutes; UF/FSM 10322, perpendicular, struts at anterior opening of inter- one scute. The age of the Inglis 1A fauna is early vertebral foramina, posterior openings interver- Irvingtonian Land Mammal Age (Webb, 1974). tebral foramina widely separated; cervical tube The Charlotte Harbor and Santa Fe faunas per­ formed by vertebrae nos. 2-7; axis articular facet tain to the late Blancan Land Mammal age oriented downward; first thoracic vertebra free, (Webb, 1974; MacFadden and Waldrop, 1980). 14 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Glyptothenum floridanum (Simpson), new Mefford Cave II, Marion County, Florida: UF/ combination FSM 6525, eight carapacial scutes. Glyptodon petaliferus Cope, 1888:346 [nomen nudum]; 1889: Piney Point, Manatee County, Florida: UF/ 662, fig. 2—Leidy, 1889b:25, pi. 4: fig. 9; pi. 5: figs. 11- FSM 11469, one carapacial scute, near anterior 12; pi. 6: figs. 1-3.—Hay, 1916:107, pis. 3-5; 1923:381; margin. 1924:2, 4; 1926:2, pi. 1: fig. 1; pi. 2: fig. 3; 1927:286.— Tomoka Park, Volusia County, Florida: UF/ Sellards, 1940:1637, pi. 2: fig. 3. FSM 14762, one carapacial scute. Glyptodon flondanus Leidy, 1889a [nomen vanum].—Cope, 1889:662. Terry Owen collection, Sarasota County, Flor­ Glyptodon rivipacis Hay, 1923:40 [nomen nudum]. ida: UF/FSM 11498, one carapacial scute. Boreostracon flondanus Simpson, 1929b:581, fig. 10.—Holmes Pinellas County, Florida: Near bridge at Trea­ and Simpson, 1931:405, figs. 14-21.—Gazin, 1950:399.— sure Island fill, St. Petersburg: UF/FSM unnum­ Lundelius, 1972:36, figs. 28-30, pi. 1. bered, nine carapacial scutes; R15E, T30S, ap­ TYPE SPECIMENS.—AMNH 23547 (holotype), proximately 35 carapacial scutes; 9th Street rear portion of carapace, including posterior bor­ South, St. Petersburg, Catalina Gardens: UF/ der; AMNH 23548-23562 (paratypes), topotypi- FSM unnumbered, two scutes. cal series of carapace portions and isolated scutes. Indian River County, Winter Beach, Florida: TYPE-LOCALITY.—Seminole Field, Pinellas UF/FSM 1677, isolated tooth from young indi­ County, Florida (see Simpson, 1929b; Holmes vidual, N-. and Simpson, 1931). DeSoto County, Peace River, 1/2 mile north REFERRED SPECIMENS.—The following speci­ Highway 70, near Arcadia: UF/FSM 19247, car­ mens are referred to G. floridanum (Simpson). apacial fragment, approximately 25 articulated Type-Locality: AMNH 23514, skull fragment scutes. (unseen); AMNH 23515, two juvenile teeth; Hendry County, Banana Creek, Caloosatchee AMNH 23586, isolated scutes; AMNH 23590, River, Florida: USNM 256751, caudal vertebra. mandible fragment; AMNH 95726-95728, 14 iso­ Edisto Beach and Edingsville Beach, Colleton lated scutes. County (formerly Charleston County), South Catalina Gardens locality, Pinellas County, Carolina: ChM-PV 2415, postglenoid cranial Florida: UF/FGS 6643, complete carapace, pel­ fragment; ChM-PV 2417, 2418, 2090, isolated vis, six anterior caudal vertebrae, fragmentary carapacial scutes. mandible and five isolated teeth, and isolated Nueces County, Texas: AMNH 14158, isolated scutes. scute, type specimen of Glyptodon petaliferus Cope, Melbourne, Brevard County, Florida: USNM 256750, in "Melbourne 1928" collection, five iso­ Wolfe City, Hunt County, Texas: USNM 6071, lated scutes. approximately 80 isolated scutes; two broken cra­ Sarasota, mouth of Hog Creek, Sarasota nial fragments with N2 and N4; several isolated County, Florida: USNM 11675, 11975, seven iso­ teeth; fragments of mandible; cervical tube; frag­ lated scutes; from Sarasota County, localities un­ mentary dorsal tube; pelvic and scapular frag­ known: UF11495, 3093, isolated scutes; from Sar­ ments; six caudal vertebrae; two humeri, frag­ asota County, localities unknown: UF11495, mentary; two ulnae; radius; femur; patella; tibi­ 3093, isolated scutes; UF/FSM 16830 isolated ofibula; several foot bones. a tooth ? N Aransas River near Sinton, San Patricio Orange County, Florida (exact locality un­ County, Texas: USNM 11378, 11379, seven known): USNM 11318, nearly complete mandi­ scutes; TMM 31141-15, numerous scutes; TMM ble. 31141-19, tibiofibula; TMM 31141-3, -16, -17, Waccasassa River, Levy County, Florida: UF/ -48, caudal vertebrae. FSM 16379, 18455, two isolated scutes. Ingleside Pit, San Patricio County, Texas: NUMBER 40 15

TMM 977-3, incomplete carapace and six ante­ trilobate, transverse axes of lobes perpendicular, rior tail rings; TMM 30967-2088, approximately posterior face convex, sulcus on anterointernal 50 articulated scutes from posterior region of face may be present; N5-N7 lobes nearly perpen­ carapace; TMM 30967-1926, nearly complete dicular; Ng anteromedial and anterolateral faces pelvis and six caudal vertebrae; TMM 30967- convex, anterior apex blunt, anterior lobe widest, 1814, left mandible with Nj-Ng. middle and posterior lobes progressively nar­ Bee County, Texas: TMM 31034-30, broken rower, posterior face weakly concave, slightly mandible and isolated scutes; TMM 31186-19, oblique; cervical tube formed by vertebrae nos. isolated tooth. 2-6; axis articular facet oriented forward and Hawley, Jones County, Texas: USNM 8644, downward; dorsal tube formed by nine thoracic approximately 15 isolated scutes. vertebrae, nos. 5-13; propodials and epipodials Laubach Cave, Williamson County, Texas: intermediate in size and in exaggeration of fea­ TMM 41343, approximately 80 scutes. tures; humerus lacking supratrochlear foramen; Port Lavaca, Calhoun County, Texas: TMM articular facets of manus and pes flattened; eight 41075-14, approximately 25 isolated scutes. to nine vertebrae in lumbar tube, three iliosacral Runnels Pierce Ranch, near Whorton, Mata- vertebrae, three or four free vertebrae in sacral gordo County, Texas: TMM 41530-6, three iso­ arch; two ischiosacral vertebrae; pubic crossbar lated scutes. small; ischiac crests posteriorly convergent; trans­ Specimens identified as Glyptothenum sp. cf. G. verse processes of terminal sacral vertebra perpen­ floridanum from Mexico are MWU 2246, four dicular to midline; anterior face of ilium uni­ carapacial scutes in articulation and MWU 2247, formly concave; carapace large, probably highly 2248, isolated carapacial scutes, from Vera Cruz arched, divided into distinct preiliac and postiliac (Dalquest, 1961); and MWU 9732-9740, isolated regions; carapace with at least some posterior scutes from the Cedazo local fauna, Aguascal- curvature; scutes small in females with marginal ientes (Mooser and Dalquest, 1975). For other sculpturing at suture; scutes large for males, with­ records of Mexican glyptodonts, see discussion out marginal sculpturing at sutures; central fig­ following the distribution accounts. ures of scutes approximately equal in size to AGE.—Rancholabrean Land Mammal Age. peripherals, usually slightly raised and weakly DIAGNOSIS.—Size intermediate to large; glenoid concave; anterior border scutes simple, lateral fossa with distinct foramen; squamous occipital border scutes conical; posterior border scutes of steeply inclined, with distinct median ridge; oc­ females with pointed conical bosses; posterior cipital condyles cylindrical; hypoglossal canal border scutes of males bluntly conical; caudal present, basioccipital-basisphenoid union contig­ armor as in G. arizonae. uous, in same plane; foramen magnum inferior margin deeply clefted; ventral extremity petro- Glyptothenum cylindricum (Brown), new mastoid crested; medial parietal region either combination elevated or smooth; mental foramen anterior to Ni; mandibular symphysis rounded and spout­ Brachyostracon cylindricum Brown, 1912:169, figs. 1-4, pis. 16- 18. like; inferior margin of horizontal ramus of man­ dible flattened; posterior margin ascending ramus TYPE SPECIMEN.—AMNH 15548, complete of mandible steeply inclined, not parallel to an­ carapace, 20 isolated teeth, atlas, hyoid fragment, terior margin; N2 trilobate; Ni ovoid, peglike, rib fragment, chevron, caudal ring fragment. situated behind symphyseal curvature; Ng trilo­ TYPE-LOCALITY.—Near Ameca, Jalisco, Mex­ bate and submolariform, or bilobate and irregu­ ico. lar; Ng molariform, posterior lobe convex and REFERRED SPECIMENS.—Known from type perpendicular, anterior lobe not squared; Nj fully specimen only. 16 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

AGE.—Rancholabrean Land Mammal Age. AGE.—Rancholabrean, presumably Wisconsin DIAGNOSIS.—Size large; N1 weakly sigmoid; N or Sangamon. irregular, middle lobe internal border undevel­ DIAGNOSIS.—Carapace large, similar to that of oped, anterior lobe pointed at medial apex; N5- G. cylindricum; anterior teeth trilobate but lobes N~ anterior borders rounded, convex; N2 trilo­ not expanded. bate, submolariform; Ng molariform, posterior STATUS OF Glyptothenum mexicanum (Cuataparo lobe convex and perpendicular, anterior lobe not and Ramirez).—The discovery of glyptodonts in squared; N5-N7 lobes nearly perpendicular; Ng North America was reported by two Mexican anterolateral and anteromedial faces convex, an­ civil engineers, who believed their specimen per­ terior apex blunt, lobes equally developed, pos­ tained to Glyptodon, but represented a distinct terior face concave, perpendicular; atlas free, alar species that they named Glyptodon mexicanus (Cua­ processes laterally directed, no struts at anterior taparo and Ramirez, 1875). Their description was opening intervertebral foramina; posterior open­ deficient in several important respects, and ing intervertebral foramina near midline; seven Brown (1912) partially amended their description to eight vertebrae in lumbar tube; four iliosacral of the carapace. The remaining skeletal parts, vertebrae; three free vertebrae in sacral arch; two reportedly including the skull, mandible, teeth, ischiosacral vertebrae; pubic crossbar stout, united at symphysis; sacral arch compound curve and sacrum, had already been lost at the time of pronounced; ischiac crests parallel; transverse Brown's study, but Brown did examine the cara­ processes of terminal sacral vertebra perpendicu­ pace firsthand. Describing carapacial and skeletal lar to midline; ilium anterior surface strongly remains of another specimen from a different concavoconvex; carapace large, highly arched, locality as Brachyostracon cylindricus, Brown referred divided into distinct preiliac and postiliac regions; the earlier named species to his genus. He in­ carapace with cephalic boss; anterior border cluded several photographs of the carapaces of scutes weakly bossed, lateral border scutes large, both species. There is no doubt, as Brown as­ conical; posterior border scutes conical, central serted, that the two species belong in the same figures of scutes slightly greater than half the genus, for the carapaces are nearly identical. scute diameter, always slightly larger than pe­ Brown chose to retain B. mexicanus as a valid ripherals, fiat to weakly concave; anterior and species, claiming that it differs from B. cylindricus posterior apertures vertical in lateral profile. in dental and carapacial features. Brachyostracon Brown, 1912, is a junior synonym oi Glyptothenum Osborn, 1903. Thus the two Mexican species are Glyptothenum mexicanum (Cuataparo and Glyptothenum cylindricum (Brown) and G. mexicanum Ramirez), new combination (Cuataparo and Ramirez). That the former is Glyptodon mexicanus Cuataparo and Ramirez, 1875:362, figs. specifically distinct from the other North Ameri­ 1-4. can species is well established. Whether G. mexi­ Brachyostracon mexicanus (Cuataparo and Ramirez).—Brown, canum is valid, however, is uncertain. 1912:168, pis. 13-15. Apparently the type specimen of G. mexicanum, TYPE SPECIMEN.—Carapace, skull, sacrum, and including the carapace, has been lost. If the spe­ isolated teeth (unseen), last known to be housed cies is valid, as assumed here, it will be necessary in the Mexico National Museum of Natural His­ for future workers to designate a neotype, provid­ tory; specimen number unknown. ing the type specimens are not located. By our TYPE-LOCALITY.—Drainage canal in the Valley retention of the species it is hoped that future of Mexico near Tequixquiac, Mexico. workers will recognize the type specimens if, and REFERRED SPECIMENS.—Known from type when, they should be found. Sinking the species specimens only. in synonymy, or as a nomen dubium, would likely NUMBER 40 17 result in its being forgotten. The following dis­ reveals a close similarity—especially James' figure cussion outlines the reasoning for its retention. 2 is remarkably similar. Even the casque scutes First, it is inconceivable that the original de­ that James illustrated show a close resemblance. scription of the species is entirely inaccurate. The The possibility of the presence of a chlamythere singular appearance of a glyptodont carapace skull among the original material of G. mexicanum and skeleton is so unique as to preclude total is also indicated by certain measurements, among conjecture, especially by nonpaleontologists at a which the 174 mm length of the tooth row, listed time (a century ago) when glyptodonts were un­ by the original authors, corresponds closely to the known in North America, and when comparative 170 mm length that James (1957) reported. These material was unavailable. Misrepresentation of figures are approximately equal to the tooth row the skeleton and carapace is understandable, and length of Glyptothenum texanum; they are greater their inaccuracies are insufficient justification for than G. floridanum and less than G. arizonae (Tables reducing the status of G. mexicanum to nomen 3-6). dubium. Indeed, their description of the cara­ Moreover, the construction of the descending pace, except for illustrating it in reverse end-to- process of the zygomatic arch figured by Cua­ end orientation, seems accurate. Even their mea­ taparo and Ramirez for G. mexicanum does not surements for the carapace, borne out by Brown, correspond with that of any other known glyp­ are reasonable and they correspond closely to the todont. As figured, the zygomatic arch of G. carapace of G. cylindricum. mexicanum is posteriorly open, rather than com­ Because at least the description of the carapace plete as in other glyptodonts. The doubly curved was accurate, the remaining portion of their de­ descending process in the figured skull bears a scription should also be considered as basically scalloped posterior outline, and its basal connec­ correct even though it may be erroneous in detail tion with the horizontal ramus is weak. This or interpretation. shape and construction closely resemble the upper Part of the dubious status of G. mexicanum cen­ surface of the zygomatic arch in chlamytheres ters around the description of its skull. As Brown (see James, 1957, fig. 2 for comparison). Hence, (1912) pointed out, the measurements given by there is a distinct possibility that the zygomatic Cuataparo and Ramirez are remarkable, indicat­ arch of a chlamythere skull (unknown to Cuata­ ing an extremely prolonged skull. Brown com­ paro and Ramirez) was taken as the descending pared the shape of the skull as figured in the process of a glyptodont skull. original description with that of the armadillo genus Eutaetus. This is certainly an appropriate The evidence outlined above supports the pos­ comparison, although comparison with the skull sibility that a chlamythere skull was associated of a chlamythere would have been even more with the glyptodont carapace and was erro­ appropriate. It now seems probable that Cua­ neously figured as pertaining to G. mexicanum by taparo and Ramirez described a composite of a the original authors. On the other hand, the facts glyptodont skull, properly belonging with the that they list only eight teeth per tooth row carapace, and a chlamythere skull, from the same (rather than nine, as in chlamytheres), and that deposit, which may account for much of the they figured and described what are obviously confusion regarding this species. glyptodont, rather than chlamythere, teeth indi­ Pointing toward the inclusion of portions of a cate that there was indeed a glyptodont skull or chlamythere skull are the drawing, several mea­ skull fragments in the same collection. Indeed, it surements, and part of the description. Compar­ is possible that portions of the chlamythere and ison of the drawing in restoration (Cuataparo and glyptodont skulls were inadvertently mixed in the Ramirez, 1875, fig. 1) with the illustrations pro­ restoration, and the original authors, unaware of vided by James (1957) for a chlamythere skull chlamytheres but knowing of glyptodonts, be- 18 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY lieved that all of these remains belonged to one ent pattern, a circumstance that adds consider­ species. able support to the proposed classification. Glyp­ From another standpoint, the possibility of a tothenum mexicanum and G. cylindricum are known chlamythere-glyptodont mixture is indicated by with certainty only from a single locality each, Silva-Barcenas' (1969) listing of the chlamythere both in central Mexico. On the basis of meager Holmesina in the Tequixquiac faunas, at least evidence, G. cylindricum seems to be of Early Ran­ provisionally demonstrating that such"an admix­ cholabrean Age, and G. mexicanum of Late Ran­ ture is possible. cholabrean Age. Glyptothenum cylindricum was re­ If the skull that the original authors described covered from near the western coast of Mexico. for G. mexicanum actually pertains to a chlamy­ To the best of our knowledge, however, glypto­ there, then most of the mystery surrounding the donts in the United States are unknown west of species is removed. The remaining descriptions, central Arizona. concerning the dentition, sacrum, and carapace Glyptothenum texanum appears to be restricted to are reasonable, and the anterior tooth, which the Late Blancan Land Mammal Age, and is Cuataparo and Ramirez (1875) illustrated in known only from southwestern United States in­ their figure 3, serves provisionally to distinguish cluding Texas and Arizona (Figure 2). Glyptoth­ the species. The carapace certainly indicates Glyp­ enum anzonae is mostly Irvingtonian in age, and tothenum. The validity of retention of a species on for the most part its geographic distribution is the basis of an anterior tooth is questionable, but contiguous with that of G. texanum, although the the one figured in the original description is so tentative records oi Glyptothenum sp., cf. G. anzonae distinctive as to warrant at least provisional ac­ from Florida (late Blancan and early Irvington­ ceptance of the species. The two teeth figured in ian) suggest an eastern distribution as well (Figure the original description are more fully discussed 3). in the osteology section following the dentition Glyptothenum floridanum is known from a rela­ descriptions. tively large number of localities mostly in Texas Finally, there is some indication that the ages and Florida, on the Atlantic and Gulf Coastal of the two Mexican species are different. As in­ Plains (Figure 4); all are late Pleistocene (Ran­ dicated in the taxonomies of G. mexicanum and G. cholabrean) in age. cylindricum, the latter species apparently is of Late Glyptothenum texanum (Figure 2).—The faunas Pleistocene age, and the former is somewhat containing G. texanum from the Blanco and Cita older, perhaps Early Rancholabrean if Silva-Bar­ Canyon localities in Texas have been extensively cenas' determinations are correct, and if the type- studied. Evans and Meade (1945), Meade (1945), localities are correctly identified. Johnston and Savage (1955), and Dalquest (1975) Therefore, accepting the authenticity of the have provided comprehensive treatment of these carapace and dentition of G. mexicanum as de­ faunas, which represent the classic Blancan Land scribed by Cuataparo and Ramirez (1875) and Mammal Age. improved upon by Brown (1912), it seems inad­ The Hudspeth local fauna (Strain, 1966) and visable to reject the species. Thus, it is here re­ the Red Light local fauna, both in western Texas, tained on a provisional basis, with the hope that are correlatives (Akersten, 1972), and include rep­ either rediscovery of the original material or fu­ resentative Blancan vertebrates, among which ture discoveries will resolve the problem. several apparently are restricted to the Blancan. According to the usage adopted by Akersten Distribution (1972), these two faunas represent the Early Pleis­ tocene portion of the Blancan Land Mammal The geologic and geographic distribution of Age. the five species of Glyptothenum fall into a consist­ The Tusker local fauna has not been as exten- NUMBER 40 19

FIGURE 2.—Occurrences by locality of Glyptothenum texanum. (1 = Crosby County, Texas, Blanco Canyon and Crawfish Draw; 2 = Randall County, Texas, Cita Canyon locality; 3 = Hudspeth County, Texas, Red Light local fauna and Hudspeth local fauna; 4 = Graham County, Arizona, Tusker local fauna.)

sively studied as the previously mentioned faunas. usage, late Pliocene. The nearby Benson fauna is (This name is applied in the broad sense to also listed by these authors as Early Blancan, and include the vertebrate recoveries from a wide may be considered roughly equivalent. The Ben­ geographic area in southeastern Arizona; there son fauna falls well within the Gauss magnetic have been no fully comprehensive studies of the polarity epoch (greater than 2.43 m.y. BP), and Tusker faunal horizon localities; glyptodonts sup­ has been dated at approximately 3.00 m.y. BP port a basic contemporaneity of these localities.) (Lindsay, et al, 1975). According to the occur­ In an unpublished dissertation, however, Wood rence of G. texanum in the Arizona Blancan, glyp­ (1962, MS.) established an Irvingtonian age on the todonts first appeared in the Tusker fauna, some­ basis of an extensive fauna that included a glyp­ time before the end of the Gauss magnetic polar' todont of undetermined identity. Lance (1960) ity epoch (2.43 m.y. BP). The absence of glypto­ and Seff (1962, MS.) have also considered the donts in the Benson fauna might indicate that fauna as Irvingtonian. More recently Downey the Tusker fauna is slightly younger (e.g., between (1970) agreed with a Middle Pleistocene age de­ 3.0 and 2.43 m.y. BP); of course sampling error termination (presumably Irvingtonian) for the and ecological differences might also account for rabbits of the Tusker fauna. Skinner (1942) and the lack of Glyptothenum in the Benson fauna. Cantwell (1969) have described an antilocaprine Lindsay (pers. comm. to Gillette) has expressed and sigmodontid rodents, respectively, from the doubt that the Tusker fauna is Irvingtonian, for Tusker fauna, but their data are inconclusive for several other reasons. First, there appears not to stratigraphic consideration. be as great a faunal separation between the un­ More recently, Lindsay and Tessman (1974) derlying Flat Tire faunal level, which is well have listed the Tusker fauna as belonging in the established as Blancan, and the Tusker faunal Early Blancan Land Mammal Age, by their level as has been proposed. Second, an ash that 20 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY occurs between the Tusker and Flat Tire faunal Therefore, the G. texanum representatives from levels has been K/Ar dated at 3.2 X IO6, well western Texas represent Middle and Late Blan­ within the Blancan, (Wood, et al., 1941; Evern­ can faunas. den, et al., 1964), and placing a lower absolute According to Lindsay, et al. (1975), the Mt. date on the Tusker fauna. Third, the younger Blanco type section has been dated as between Curtis Ranch fauna (containing Glyptothenum ar­ 1.4 and 2.4 m.y. BP by volcanic ash correlation. izonae) is approximately at the base of the Olduvai Hence, the early Pleistocene Blanco fauna of geomagnetic event of the younger Matayuma Texas is younger than the Tusker fauna of Ari­ magnetic polarity epoch, which Lindsay considers zona, although the difference in age might be no as latest Blancan. more than 100,000 years. We believe that the Mt. That the glyptodonts in the Tusker fauna are Blanco glyptodonts are somewhat more primitive. G. texanum is positively indicated by an excellent Despite the disagreement with the stratigraphic collection from the Frick Laboratory, American interpretation indicating that the Mt. Blanco Museum of Natural History, which was made fossils are younger, it is nevertheless tenable that available for the present study. These specimens, the Mt. Blanco glyptodonts exhibit more of the including three carapaces and a nearly complete primitive, or ancestral, traits of the species, while skeleton, were recovered from several localities the Tusker glyptodonts, although occurring some­ near Safford, Arizona, by Mr. Ted Galusha. what earlier, reveal more of the advanced traits These localities are centered around Dry Moun­ of the species. It is also possible that sampling tain, Southern Whitlock Mountains, and North­ error might account for differences between the ern Whitlock Mountains, Graham County, Ari­ two populations, or even that reinterpretation is zona, including the 111 Ranch area from which in order. most of the published Tusker faunal elements Lance (1960) regarded the Tusker and Curtis were recovered. Whether the horizons that have Ranch faunas as roughly equivalent. The latter yielded G. texanum are directly correlative with the fauna, containing the descendant species G. an­ Tusker faunal horizon (sensu stricto) has not been zonae, appears on the basis of the glyptodonts, and conclusively established. According to Galusha according to the stratigraphic relationships pre­ (pers. comm. to Gillette) the localities from which sented by Lindsay and Tessman (1974) and Lind­ he recovered the glyptodonts are at least roughly say, et al. (1975), to be younger than the Tusker equivalent stratigraphically, and, according to his fauna. determination, a Late Blancan age is the most Glyptothenum arizonae (Figure 3).—Extensively reasonable for the associated (unstudied) faunas. known from two localities, and poorlv known The type-locality for Glyptothenum texanum is from several others, the western geographic dis­ somewhere in the "Blanco Beds" of western tribution of G. arizonae overlaps that of G. texanum; Texas, presumably at or near the Mt. Blanco the species is tentatively identified in Florida as section, which is the type-locality for the Blancan well. The occurrence of G. arizonae appears to Land Mammal Age. Glyptothenum texanum also have followed closely that of G. texanum; these two occurs in abundance in the nearby Cita Canyon species occur much closer stratigraphically to. fauna. The fossils from Mt. Blanco are younger each other than either does to the remaining than those from Cita Canyon. According to Lind­ North American species. Glyptothenum anzonae say, et al. (1975) the Cita Canyon fauna is equiv­ does not appear to have existed contempora­ alent to the Benson fauna of Arizona (and hence neously with G. texanum in any of the known local to the Tusker fauna), and correlates with the faunas, although they are closely related and Gauss magnetic polarity epoch. The same authors represent ancestral-descendant species. correlate the Mt. Blanco faunas with the lower The earliest records of G. arizonae are the ten­ Matayuma epoch, prior to the Olduvai event. tative identifications of isolated carapacial scutes NUMBER 40 21

FIGURE 3.—Occurrences by locality oi Glyptothenum arizonae (localities 1-4) and Glyptothenum sp., cf. G. anzonae (localities 5-7). (1 = Cochise County, Arizona, Curtis Ranch; 2 = Knox County, Texas, Seymour Formation, Gilliland local fauna; 3 = Tillman County, Oklahoma, Holloman Gravel Pit near Frederick; 4 = Briscoe County, Texas, Rock Creek local fauna; 5 = Gilchrist County, Florida, Santa Fe River, UF localities SF1, SF4A, SF8A; 6 = Charlotte County, Florida, Reynolds Point; 7 = Citrus County, Florida, UF locality, Inglis 1A.)

and a tooth fragment from localities in the Santa series of G. arizonae remains are far to the west of Fe River, Gilchrist County, and Charlotte Har­ the Florida sites, in Arizona, Oklahoma, and bor, Charlotte County, Florida. The vertebrate Texas. The vertebrate fauna from the type-local­ faunas from these localities pertain to the Blancan ity of G. anzonae (Curtis Ranch locality, Cochise Land Mammal Age, as indicated by the presence County, Arizona) was preliminarily studied by of Namppus phlegon (MacFadden and Waldrop, Gidley (1922, 1926), who did not establish firmly 1980). Glyptothenum sp., cf. G. arizonae is also the age of the Curtis Ranch fauna. He believed known from isolated scutes in the fauna from the Curtis Ranch horizon to be late Pliocene in Inglis 1A, Citrus County, Florida. The Inglis 1A age, although he did not rule out an early Pleis­ fauna is considered earliest Irvingtonian (Webb, tocene determination. Gidley considered the 1974 and pers. comm.). fauna from the nearby Benson locality (lacking These isolated scutes are problematical for their glyptodonts) as somewhat older than that from age and large size; they are all provisionally the Blanco Formation in Texas (classic Blancan, referred to G. arizonae on the basis of sculpturing and containing G. texanum) and the fauna from (eliminating G. floridanum and G. cylindricum) and the Curtis Ranch locality as somewhat younger size (eliminating G. texanum). Presuming the iden­ (containing G. anzonae), although this belief was tification of these scutes to be correct, the Santa "not based on very tangible or direct evidence" Fe and Charlotte Harbor glyptodonts are the (Gidley, 1926:83). Gazin (1942) elaborated in an earliest occurrence for the species, representing a expanded review of the San Pedro Valley faunas population roughly contemporaneous with the but did not substantially alter or improve upon postulated ancestral species, G. texanum. Gidley's contentions. Lance (1960:157) consid­ The other localities that have yielded extensive ered the Curtis Ranch fauna to be Middle Pleis- 22 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY tocene, "either late Kansan or early Yarmouthian tion of G. anzonae remains, and their near contem­ in terms of the North American glacial chronol­ poraneity supports the contention that the glyp­ ogy." He also considered the Curtis Ranch fauna todonts are the same species. and the Tusker fauna (containing G. texanum) as Hibbard and Dalquest (1966) and Dalquest contemporaneous. As previously discussed, the (1977) regarded the Holloman local fauna (near Tusker fauna is Late Blancan, equivalent to the Frederick, Tillman County, Oklahoma) and the Benson fauna. The Curtis Ranch fauna according Rock Creek local fauna (Briscoe County, Texas) to Lindsay and Tessman (1974) and Lindsay, et as contemporaneous with the Gilliland local al. (1975) belongs in the early Irvingtonian Land fauna; both faunas include glyptodonts. It was Mammal Age, representing the early Matuyama the glyptodonts from the Holloman locality on magnetic polarity epoch, and occurring during which Meade (1953) based the genus Xenoglypto­ the Olduvai geomagnetic event (1.86-1.71 m.y. don. Melton (1964) later correctly assigned the BP). Lammers (1970, abs.) also regarded the Holloman and Gilliland (Seymour) glyptodonts fauna as Early Irvingtonian on the basis of bio­ to the same species. Morphological and strati­ stratigraphic evidence. The work of Lindsay and graphic evidence indicates that the material iden­ his collaborators has placed the Curtis Ranch tified by Melton as Glyptodon fredericensis (Meade) faunas in an absolute chronology, which supports actually belongs in Glyptothenum arizonae. the general concensus of an Early Irvingtonian The glyptodont remains that Hay (1927) re­ age; that determination is in turn supported by ported for the Rock Creek (Briscoe County, the glyptodonts, which are indeed younger than Texas) local fauna (unseen, probably scutes) are those of the faunas equivalent to the Benson. The probably G. arizonae. This local fauna is correla­ Curtis Ranch glyptodonts (G. anzonae) are de­ tive with the Gilliland and Holloman local faunas scendant, and advanced, in comparison with their (Hibbard and Dalquest, 1966). Because of this ancestors (G. texanum) in the older faunas. contemporaneity and the proximity of these three The glyptodonts recovered from the Seymour localities, it is reasonable to assume that the Rock formation and belonging in the Gilliland local Creek glyptodonts represent the same species as fauna are proposed here to represent the same the better known ones from the other two faunas. species as the Curtis Ranch glyptodonts, G. an­ Glyptothenum cylindricum (Figure 4, locality zonae. The Gilliland local fauna from Knox 25).—According to Brown (1912) the type speci­ County, north-central Texas, has classically been men of G. cylindricum was recovered from a richly considered as belonging in the Kansan glacial fossiliferous deposit near the town of Ameca, age, and because the vertebrates represent a Jalisco, Mexico. "warm" assemblage, the fauna has been assigned to an interstadial of the Kansan glaciation (Hib­ The Ameca River valley at this point is enclosed by bard and Dalquest, 1966). This assignment has moderately high mountains at the base of which, on either more recently been altered, however, by the same side of the river. Post Tertiary sediments are exposed in authors, who stated that the sediments in which terraces to a height of two hundred feet. the Gilliland local fauna occurs "were laid down The escarpments, of limited extent, are composed chiefly of volcanic ash, rhyolilic debris and gravel with an admix­ prior to the Kansan glaciation .... The warm ture of dialomaceous clay having the appearance of a river Gilliland local fauna is post-Blancan and pre- sediment. Apparently the outlet of the valley was obstructed Cudahy fauna in age" (Hibbard and Dalquest, during Pleistocene times when a shallow lake was formed 1973:270, 273). The Gilliland fauna belongs in over a considerable part of the valley. the Irvingtonian Land Mammal Age, probably Fresh water shells, fish teeth and bones, and turtle shells earliest Irvingtonian, and perhaps directly equiv­ were found in situ in the highest clay strata but vertebrate remains were chiefly found in the gravels. Many of these alent to the Curtis Ranch fauna. These two remains are identifiable only as to families (Brown, 1912: faunas have provided the most extensive collec­ 167). NUMBER 40 23

FIGURE 4.—Occurrences by locality oi Glyptothenum floridanum, G. cylindricum, and G. mexicanum. (G. floridanum, Texas: 1 = Cameron County; 2 = Nueces County, exact locality uncertain; 3 = San Patricio County, Ingleside local fauna; 4 = San Patricio County, Sinton near Arkansas River; 5 = Bee County, Berclair Terrace local fauna; 6 = Calhoun County, Port Lavaca "Eremotherium locality"; 7 = Matagorda County, Pierce Ranch locality; 8 = Jones County, near Hawley; 9 = Hunt County near Wolfe City; 10 = Williamson County, Laubach Cave; 11 = Harris County, Taylor Bayou local fauna. G. floridanum, Florida: 12 = Pinellas County, Seminole Field locality; 13 = Pinellas County, Catalina Gardens and other localities; 14 = Levy County, Waccasassa River; 15 = Brevard County, Melbourne locality; 16 = Orange County, exact locality uncertain; 17 = Hendry County, Caloosahatchee River, Banana Creek; 18 = Sarasota County, several localities; 19 = Marion County, Mefford Cave II; 20 = Indian River County, Winter Beach; 21 = DeSoto County, Peace River near Arcadia; 22 = Manatee County, Piney Point; 23 = Volusia County, Tomoka Park. G. floridanum, South Carolina: 24 = Colleton County (formerly Charleston County), Edisto Beach and Edingsville Beach. G. cylindricum: 25 = near Ameca, Jalisco, Mexico. G. mexicanum: 26 = near Tequixquiac, Valley of Mexico. Glyplolherium sp. cf. G. floridanum: 27 = Vera Cruz, Mexico; 28 = Aguascalientes, Mexico. For locality information concerning unsubstantiated records of glyptodonts in Mex­ ico, see Alvarez, 1965, and Silva-Barcenas, 1969.) 24 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Mammals included in Brown's preliminary Glyptodon clavipes from Tajo del Desague, and faunal list are: Sciuridae, Geomyidae, Cricetidae, Glyptodon sp. He also listed Glyptodon mexicanus, machaerodont, Equus sp., Elephas columbi, and the which Brown (1912) called Brachyostracon mexicanus glyptodont that he described as Brachyostracon cy­ (= Glyptothenum mexicanum), from the nearby lindricus (= Glyptothenum cylindricum). The only use­ Zumpango de Ocampo locality. It is uncertain ful record for stratigraphic purposes is the mam­ which of these listings, if any, refer to the occur­ moth, indicating an Irvingtonian or "Ranchola­ rence of the type specimen. According to Silva- brean age. Barcenas, all are Late Pleistocene in age. There­ Silva-Barcenas (1969) listed the following fore, a Rancholabrean age seems reasonable for mammals from the same stratigraphic level (Gran G. mexicanum. Canal) from three other nearby localities in Jal­ Glyptothenum floridanum (Figure 4, localities 1- isco: Bison sp.; Camelops sp.; Cervus intertuberculatus; 24).—As far as can be determined, all known Elephas sp.; Equus sp., cf. E. conversidens; Equus glyptodonts referable to G. floridanum are of Late mexicanus; Equus sp., cf. E. mexicanus; Equus sp.; Pleistocene age, although the time spanned may Felis sp.; Holmesina sp.; Mammuthus columbi; Neocho- be as much as 100,000 years or more. Many of erus sp.; Platygonus sp.; Tetrameryx sp.; and Ursus the localities that have yielded G. floridanum are sp. The presence of Bison sp. indicates a Ran­ poorly known both faunally and stratigraphi­ cholabrean age. The close similarity of Glyptoth­ cally, but enough is known from the Florida and enum cylindricum to the Late Pleistocene glypto­ Texas localities in general to establish a relatively donts to the north supports a Rancholabrean age late age for most of the G. floridanum records, determination for G. cylindricum. Silva-Barcenas dating from the early Wisconsin glacial age and considered the Gran Canal as pertaining to the therefore representing the late Rancholabrean Illinoian-Yarmouthian intervals in the glacial Land Mammal Age. chronology. If this determination is correct, then Two peculiarities of the geographic distribu­ G. cylindricum predates the occurrences of G. flori­ tion of G. floridanum are significant: the apparent danum in Texas and Florida. separation between Florida and Texas, and the Silva-Barcenas also listed several other Mexi­ near-coastal distribution. The latter characteristic can localities from which glyptodonts are known. is probably a manifestation of the habitat require­ Recognizing several taxa, the only recovery listed ments of glyptodonts, indicating the warm cli­ as Brachyostracon cylindricus (= Glyptothenum cylindri­ mate and lush vegetation necessary for their ex­ cum) was from the Ameca site, presumably in istence. The three inland recoveries in Texas reference to Brown's (1912) report, although it is (Wolfe City, Hunt County; Laubach Cave, Wil­ not included in the bibliography. liamson County; and Jones County) are proble­ Glyptothenum mexicanum (Figure 4, locality 26).— matical. They probably indicate the existence of Neither Cuataparo and Ramirez (1875) nor riparian corridors toward the inland areas from Brown (1912) provided definite locality informa­ the coast, apparently an uncommon situation as tion for the occurrence of G. mexicanum beyond indicated by the otherwise restriction to near- stating that the specimen was recovered in the coastal areas for this species. Valley of Mexico near Tequixquiac. Neither The apparent geographic separation between listed an associated fauna. the Florida (and South Carolina) population and Silva-Barcenas (1969) listed several Tequix­ the Texas population of G. floridanum is probably quiac localities, all of which fall within the Be- an artifact of sampling. If Pleistocene faunas were cerra stratigraphic level of Late Pleistocene age. as extensively known along the intervening Gulf Included in his faunal list in the vicinity of Te­ Coast (Louisiana, Mississippi, Alabama) as in quixquiac are Brachyostracon mexicanus (= Glypto­ Texas and Florida, G. floridanum would surely be thenum mexicanum) from Barranca de Acatlan, represented. There are no apparent geographic NUMBER 40 25 dispersal barriers along the Gulf Coast between The beach recoveries of G. floridanum from Ed- Florida and Texas, and there is no substantial isto Island, Colleton County (formerly Charleston reason to postulate one in the past. County), South Carolina, are the northernmost Faunal lists for several of the more extensive record for the species. Roth and Laerm (1980) faunas in Florida were compiled by Simpson have assembled a preliminary faunal study of the (1929a, 1929b) for the Seminole Field, Mel­ Quaternary reptiles and mammals from the is­ bourne, Sarasota, and Peace Creek localities. Ga­ land; chlamythere, Dasypus bellus, Neochoerus pinck- zin (1950) and Ray (1958) elaborated on the neyi, Canis dirus, and Mammuthus sp. are among Melbourne faunal list, and other authors have those mammals consistent with a late Pleistocene provided more limited contributions. Weigel age. The glyptodont remains support this age (1962) regarded the extinct fauna from the Vero assignment. Beach locality farther south as having lived dur­ In Texas the representatives of G. floridanum all ing the Wisconsin glacial age (from greater than appear to be from sediments of Wisconsin glacial 30,000 BP to 8200 ± 120 BP). The Vero Beach age. The fauna from the Ingleside locality, San fauna (lacking glyptodonts) was considered by Patricio County, has been comprehensively re­ Weigel as a correlative of the Melbourne and viewed by Lundelius (1972). This locality has Seminole Field localities, both of which contain yielded the most extensive representation of G. G. floridanmn. All of the taxa listed by Simpson floridanum in Texas. According to Lundelius, the (1929a) for the locality at the mouth of Hog Ingleside fauna is best considered as post-Sanga- Creek at Sarasota, Sarasota County, are present mon in age, probably early Wisconsin. He has at the Melbourne locality (Simpson, 1929a; Ray, also indicated (pers. comm.) that an approximate 1958), and a similar statement can be made for absolute age of 80,000 BP is reasonable, which the Peace Creek locality, DeSoto County. Thus may indicate that the Ingleside recoveries repre­ these four localities in Florida, which have yielded sent the earliest known occurrence of G. floridanum a large share of G. floridanum known in that state in the United States. Correlation of the Ingleside (Seminole Field, Sarasota, Melbourne, Peace fauna with the Florida faunas is rendered difficult Creek), contain essentially contemporaneous mid­ by the absence of adequate stratigraphic docu­ dle to late Wisconsin faunas (Webb, 1974). mentation in Florida. Whereas Lundelius (1972: No associated fauna has been reported from 6) stated that a late Wisconsin age is unlikely for the Catalina Gardens locality, Pinellas County, the Ingleside fauna, correlation of the several which provided the excellent and important G. glyptodont-bearing Florida localities with the floridanum fossils, described herein for the first Vero Beach locality as reported by Weigel (1962) time. Because of the close proximity of this local­ indicates that G. floridanum is likely to have been ity to the Seminole Field (type) locality, and the present in latest Wisconsin time. Further state­ certain identification of the glyptodont as G. flor­ ments regarding the Ingleside correlation with idanum, this record is here considered to be con­ the Florida records of G. floridanum must await temporary with the above-mentioned faunas. future investigations. For purposes of this study, Faunas from other Florida localities that have the Ingleside fauna is believed to predate, at least yielded G. floridanum are generally poorly known. partially, the Florida records of G. floridanum. That For example, the locality for USNM 11318, an this chronology is correct is suggested by a greater excellent mandible with most of the teeth (as­ expression of the postulated sexual dimorphism signed to G. floridanum), is known only to be in the Florida representatives, here taken to in­ somewhere in Orange County, Florida. Because dicate the culmination of a trend only suggested this specimen is G. floridanum, its age is assumed in the Texas population. Of course this statement to be equivalent to that for the other faunas, i.e., rests only on a postulated development, which middle to late Wisconsin. should not be considered as positively established, 26 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY and is here offered only in the absence of contra­ Sangamon age. By our determination the age dictory information. It is not absolute proof of an would be even younger, perhaps middle Wiscon­ earlier date for the Ingleside fossils. sin. Hay (1926) provided a faunal list for the Kaspar and McClure (1976) reported a single nearby Sinton locality, also in San Patricio scute of Boreostracon sp. in the Taylor Bayou local County. None of the faunal elements is contra­ fauna, Harris County of southeastern Texas. This dictory to a close correlation with the Ingleside is the easternmost record of glyptodonts in Texas. fauna. The Berclair Terrace local fauna, Bee The only species assigned to Boreostracon for North County, Texas, also not far from the Ingleside American glyptodonts is B. flondanus, which we fauna, was considered by Lundelius (1972) to be have found invariably to belong to Glyptothenum probable Wisconsin age. The glyptodonts re­ floridanum, according to our emended diagnosis. covered from these three localities (Ingleside, Sin- The authors regarded the fauna as equivalent to ton, and Bee County) represent the same species, that of Ingleside farther south, which contains further supporting approximate contemporane­ abundant G. floridanum. ity. No positive statement can be made for the The Port Lavaca, Calhoun County, specimen age of the Nueces County Record of G. floridanum of G. floridanum was associated with an excellent (Cope, 1888). skeleton of Eremothenum, and the site has been The specimen of G. floridanum from Cameron radiocarbon dated at roughly 23,000 BP and County in southern Texas is the southernmost 28,000 BP, both with rather large statistical errors record of the species in the United States. The of approximately 3000 years (Lundelius, pers. tentative identifications as Glyptothenum sp., cf. G. comm.). These dates are substantially younger floridanum for scutes from Vera Cruz and Aguas- than the assumed age for the Ingleside fauna, and calientes would be the southernmost records for represent a middle Wisconsin age. The Laubach the species in North America; these were reported Cave, Williamson County, record is also probably by Dalquest (1961) and Mooser and Dalquest middle to late Wisconsin in age. This recovery, (1975) as Brachyostracon mexicanus and Brachyostra­ the Jones County, and the Hunt County records con cf. mexicanus, respectively. Mooser and are the only known inland occurrences of G. Dalquest (1975) regarded the Aguascalientes fos­ floridanum. The age of the glyptodont from the sils from the Cedazo local fauna as Illinoian. We latter locality is unknown but is presumably also base identification of the glyptodonts as cf. G. Wisconsin. Other records of G. floridanum in Texas floridanum in this fauna on scute pattern, scute are probably also Wisconsin in age. size, and the presence of large and small adults Therefore, G. floridanum inhabited the Texas (male and female) corresponding to the sexual Gulf Coastal Plain from early to middle Wiscon­ dimorphism postulated for the species in Florida. sin time, while circumstantial evidence in Florida Because the dimorphism for the Aguascalientes indicates the presence of this species at least from glyptodonts is pronounced, an Illinoian age is middle Wisconsin through late Wisconsin time. inconsistent with our observed increase in the GLYPTODONTS IN MEXICO AND CENTRAL AMER­ expression of dimorphism from moderate in early ICA.— Besides the type specimens of Glyptothenum Wisconsin to very pronounced in late Wisconsin. cylindricum and G. mexicanum, and the specimens An Illinoian age for the fauna would be the tentatively referred to G. floridanum from Vera earliest occurrence for G. floridanum, if our identi­ Cruz and Aguascalientes above, there are several fication is correct, and would necessitate revision additional records of glyptodonts in Mexico. Hib­ of our hypothesis concerning the trend toward bard (1955) reported and illustrated several scutes increasing sexual dimorphism in the species. Al­ from the late Pleistocene Upper Becerra forma­ ternatively, the fauna might be younger; the tion from the Valley of Tequixquiac. These he original authors could not positively rule out a identified as Brachyostracon mexicanus, but he stated NUMBER 40 27 that scutes from at least one locality more closely tial information to our knowledge of North Amer­ resemble those of Boreostracon flondanus. For the ican glyptodonts south of the United States; it is purposes of this study we consider these specimens in this area where remaining secrets of glyptodont (unseen by us) tentatively as Glyptothenum sp. taxonomy and evolution are likely to be uncov­ indet. ered. Alvarez (1965) provided synonomies and local­ ity records for Brachyostracon cylindricus (= G. cylin­ Osteology dricum), Brachyostracon mexicanus (= G. mexicanum), and Glyptodon sp., which he correctly suggested Previous treatments of the osteology of North pertains to Brachyostracon (= Glyptotherium). These American glyptodonts have been limited to de­ records and localities have not been verified ex­ scriptions of single individuals, or populations, cept as mentioned in the discussion above. without extensive comparison to specimens from Silva-Barcenas (1969) also listed various North other localities. The excellent skeletal material American glyptodont taxa for several localities in from Safford, Arizona, of Glyptothenum texanum, Mexico not previously mentioned. Because he did made available for this study by the American not conform with established synonymies (he re­ Museum of Natural History, forms the basis for ported both Brachyostracon mexicanus and Glyptodon this comprehensive review of the comparative mexicanus, for example, as well as Glyptodon clavipes osteology of the five species of Glyptotherium. which, as far as we can determine is the only such record in North America and for which he made SKULL no literature citation), the identities of these spec­ imens cannot be suggested even tentatively. The Skulls of South American glyptodonts have lack of reported recoveries of Glyptotherium flori­ been described by several authors, including Hux­ danum (= Boreostracon flondanus to Silva-Barcenas) ley (1865) and Burmeister (1874), both of whom renders his identification suspect, for this species ascertained the general nature of most of the should be present in late Pleistocene deposits of bones of the cranium. Because of the massive, Mexico and Central America. solidly ankylosed nature of the cranial bones in Other records of Mexican glyptodonts include adult specimens, however, only Vinacci Thul references by Pichardo (1960). We have not ex­ (1945) has described completely the individual amined these materials. Glyptodonts from Val- bones of an entire skull. He described the skull of sequillo, Puebla (fauna currently under study by Glyptodon reticulatus from South America, outlining Russell Graham), possessed scutes far too small in rather general fashion the individual bones, and with sculpturing patterns too strikingly dif­ where distinguishable, and using a "geographic" ferent to suppose identity as Glyptotherium. Hence, approach. His description was not accompanied at least one other genus is present in the North by figures, and only three gross and imprecise American fauna, as yet unidentified. measurements were appended to the text: maxi­ This monograph is devoted to all glyptodonts mum vertical diameter 191 mm, maximum an­ of the United States and to those of Mexico that teroposterior diameter 302 mm, and maximum are important to the taxonomy of Glyptothenum; transverse diameter 244 mm. Corresponding di­ the only Mexican glyptodonts pertaining to G. mensions for the complete skull (AMNH 95737) floridanum of direct interest to this report are the oi Glyptothenum texanum are 147 mm, 248 mm, and ones reported by Hibbard (1955), Dalquest 187 mm, respectively. There is a rough propor­ (1961), and Mooser and Dalquest (1975). tionality in these gross measurements, the Glyp­ Undoubtedly, still other records of Central tothenum texanum skull apparently smaller by ap­ American and Mexican glyptodonts have been proximately 25 percent. Measurements of Glyp­ overlooked. Future discoveries will add substan­ totherium arizonae closely approximate those pro- 28 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY vided by Vinacci Thul. (These measurements for This skull was associated with the skeletal and the Glyptotherium texanum skulls are givtn only for carapacial remains, which are undoubtedly from comparison. Table 1 provides rriore precisely de­ a juvenile individual. Unfortunately the skull is fined measurements for these and other North heavily fractured throughout and the sutural con­ American skulls.) tacts are mostly obscured. This broken condition Others who have dealt extensively /With glyp­ seems to be the result of two disruptive circum­ todonts (e.g., Burmeister, Lydekker, Castellanos) stances: the apparent cause of death, and the have included skulls in their determinations of conditions of preservation. taxonomic rank. Especially diagnostic is the an­ The latter circumstance likely contributed to terior half of the skull roof, including the frontals, most of the broken condition of the skull bones, nasals, and premaxillary bones. The configura­ for there are innumerable small fractures almost tion of the frontals in their contribution to the "uniformly" covering the skull. Because the upper profile and the shape of the narial aperture braincase and nasal cavity were filled with sedi­ have been especially important in South Ameri­ ment, however, the fragments were essentially can generic descriptions. kept in place after fracturing. The fact that many The skulls of Glyptothenum compare favorably of the fractures are filled with the same matrix with illustrated examples (e.g., Burmeister, Ly­ that invests the cavities of the skull indicates that dekker, Huxley) of the South American genus the fracturing occurred after burial but before Glyptodon, in the classical sense (see Hoffstetter, consolidation of the sediment enveloping the skel­ 1955, for an excellent discussion of the taxonomic eton. status of this generic name). Lydekker's (1894) That this individual met a violent death is description wa$ amplified by Hoffstetter (1958) indicated by a pair of elliptical holes almost in a summary of the cranial characteristics of perfectly centered in the frontal bones (Figure Glyptoddn. The' skulls of Glyptotherium resemble 9a). The long axes of these holes are situated in those of Glyptodon. the parasagittal plane, evenly placed on either The Skull of F:AM 59583 (Glyptothenum texanum, side of the midline. The short axes, which are Figures 5-8, Table 1) is complete except for the approximately half the length of the long axes, region anterior to a frontal section through the do not intersect in their medial extensions, be­ anterior borders of the zygomatic arches; the cause the left hole is slightly more posteriorly anterior half of the palate is also missing. The situated than the right. Fragments of the skull upper dentition is fully represented. N Na of both roof from the region of the injury are preserved sides are preserved in place in the skull. Fragmen­ within the left hole, solidly held in place by the tary N1 and N2 from both sides, which are isolated enveloping matrix, thus indicating that the holes from the skull, complete the upper series. Each of are not a result of extrication or preparation. This the latter four teeth is represented by the crown contention is confirmed by the preparator, Mr. half of the tooth, including the occlusal surfaces. Gladwyn Sullivan, of the National Museum of This individual was probably a young to middle- Natural History (Smithsonian Institution), whose aged adult, for many of the sutures, though careful work brought these features to light. Each strongly united, are discernible. hole measures approximately 20 mm by 15 mm. Except for the right malar bone, the skull of Apparently this individual was attacked by a F:AM 95737 (G. texanum, Figure 9) is complete, large predator bearing enormous canines, which thus representing the most nearly complete single pierced the skull roof either simultaneously or in glyptodont skull in North America, the only one consecutive blows, probably resulting in imme­ that includes the prezygomatic region. The entire- diate death for the victim. Because this individual upper dentition is represented, although right N'J, was young, the cephalic shield, which was not and N1, N4, and N~ of the left side are missing. recovered, apparently was not fully developed, NUMBER 40 29

FIGURE 5.—Skull oi Glyptotherium texanum, F:AM 59583: a, dorsal; b, right side, with central zygomatic region broken away. (Bar = 5 cm.)

thus affording poor protection for the skull roof. the rear portion of the skull in glyptodonts was It appears that the predator, probably a medium- confluent with the anterior notch of the carapace, sized cat, attacked from the front, a curious anom­ in lateral aspect forming a continuous concave aly in view of the preference of most cats for a outline. Thus a rear attack would necessitate at rear attack. The reason for this assumption is that least an estimated 225° gape, an obvious impos- 30 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

P-ma>.? NUMBER 40 31 sibility. A front attack, however, seems more tiates nor refutes this possibility. If partial retrac­ likely, owing to the convex outline of the anterior tion was possible, then perhaps this individual end of the skull. was attacked frontally, and the blow penetrated A final consideration built upon this reasoning the portion of the skull that remained exposed. concerns the glyptodont's defense. As suggested Two skulls of G. arizonae provide adequate com­ elsewhere, glyptodonts were probably defenseless parison with those of G. texanum. The skull of except for their armor and had to rely on behav­ UMMP 34826 (Figure 10) was briefly described ioral mechanisms for protection from predators, by Melton (1964). It is complete except for the either by residing in water like a hippopotamus pre-zygomatic region. Another skull from the or by gregarious habits. In either case, it is incon­ same locality (UMMP 38761) is badly crushed, ceivable that a predatory cat would simply sur­ and, except for the dentition, offers little infor­ prise a juvenile individual by a frontal attack mation beyond that provided by UMMP 34826. with such a perfectly placed injury. When faced Skull fragments of G. floridanum from two local­ with danger, it seems that a normally healthy ities allow tentative comparisons with G. texanum individual would at least turn its head from an and G. arizonae. The skull of USNM 6071 (Figure impending blow, eliminating the possibility of 11), from Texas, was briefly described by Hay the symmetrically placed holes. It seems more (1916). It consists of two fragmentary portions, likely that this juvenile was stranded, perhaps in the posterior dorsal part of the braincase, and the mud, or was otherwise debilitated, unable to central region including most of the palate, the avoid an approaching predator. Some authors anterior zygomatic region, and the skull roof. (e.g., Huxley, 1865; Burmeister, 1874) have en­ Another skull fragment (ChM 2415, Figure 12), tertained the possibility that glyptodonts could consisting of the postglenoid region, was de­ retract their head, turtle fashion. Evidence pro­ scribed by Ray (1965), who was unable to make vided by North American fossils neither substan- a generic determination. This specimen compares favorably with USNM 6071 and is tentatively FIGURE 6.—Interpretation of skull oi Glyptotherium texanum (F: identified herein as G. floridanum. AM 59583) illustrated in Figure 5: a, dorsal, b, right lateral, Thus the skull of G. floridanum is poorly known. with most of zygoma removed. (Abbreviations as follow: N Comparison of associated upper teeth of USNM = nuchal crest; N^N8 = upper tooth row; POP = paroccip­ ital process; PT.TUB. = pterygoid tuberosity; SOP = su- 6071 with teeth of UF/FGS 6643 from Catalina praoccipital process; al. = alisphenod; b.oc. = basioccipital; Gardens, Florida, which is assignable to G. flori­ d.max. = descending process of zygomatic process of maxilla danum, supports the taxonomy suggested herein (= zyg. max.); fr. = frontal; fr.po. = postorbital process of for the Florida, South Carolina, and Texas Late frontal; inf.sq. = inferior region of squamous process of Pleistocene representatives. squamosal; jug. = jugal; jug.po. = postorbital process of Cuataparo and Ramirez (1875) included a jugal; lac. = lacrimal (inferred with reference to skull F:AM 95737—see Figure 8); m.f. = mandibular fossa; m.p. = largely erroneous description of the skull oi Brach­ mastoid process of squamosal; o.c. = occipital condyle; par. yostracon mexicanus (= Glyptotherium mexicanum). As = parietal; pt. = pterygoid; p.gl.sq. = postglenoid process of Brown (1912) and Ray (1965) have pointed out, squamosal; p.max.? = premaxilla (actual existence uncer­ their description and interpretation are highly tain); p.oc. = paroccipital process of squamosal; os. = orbito- suspect. Although no further mention will be sphenoid (suture with maxilla shown by dashed line); s.max. = superior process of maxilla; sag.fur. = sagittal furrow; made of the skull of G. mexicanum in the following sup.cr.sq. = superior crest of squamosal; sup.occ. = supraoc- descriptions, it should be stated that the skull of cipital; sup.sq. = superior region of squamous process of this species probably differs little from that of the squamosal; zyg.max. = zygomatic process of maxilla; zyg.sq. other species of this genus. The present location = zygomatic process of squamosal; p. = petromastoid; 2 = of the skull is unknown. posterior exit of infraorbital canal; 3 = channel formed by frontosquamosal flange; 4 = foramen rotundum region; 5 The skull of G. cylindricum (AMNH 15548) is = foramen ovale; 6 = pterygoid flange; 7 = petromastoid represented only by a fragment of the posterior fenestra.) extremity of the cranial vault, at the parietal/ 32 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 7.—Basicranial region oi Glyptotherium texanum (F:AM 59583): a, ventral; b, ventrolateral. (Bar = 5 cm.) NUMBER 40 33

FIGURE 8.—Glyptotherium texanum (F:AM 59583): a, line drawing interpretation of Figure la; b, line drawing interpretation of Figure lb. (Abbreviations as in Figure 6, with the following additions: PN = posterior narial aperture; b.o. and b.oc. = basioccipital; b.s. = basisphenoid; pal. = palatine; g.a. = glenoid articular facet; s.oc. = supraoccipital; 8 = hypoglossal foramen; 9 = foramen magnum.) FICURE 9.—Skull oi Glyptothenum texanum, F:AM 95737, from a juvenile individual: a, dorsal, showing a pair of holes presumed to have been the result of an attack by a large predator bearing piercing canines; b, lateral. (Bar = 40 mm.) NUMBER 40 35

FIGURE 10.—Skull of Glyptotherium arizonae UMMP 34826: a, right side, with prezygomatic region restored and incorrectly showing bulged superior profile, which should be straight instead; b, ventral, showing complete tooth row; c, detail of basicranial region. (Bars = 40 mm.) 36 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 1 1.—Skull oi Glyptothenum floridanum, from the Wolfe City, Hunt County, Texas, locality (USNM 6071): a, c, lateral and ventral aspects of cranial fragment from central region of skull, anterior to left; teeth shown in c are right N- and left N-; b, cranial fragment from parietal region of skull, dorsal aspect, anterior to left. (Bar = 40 mm.) NUMBER 40 37

laterally compressed element, with expansions at either end of the constricted shaft. Its length would have exceeded 91 mm. The following description of the skull of Glyp­ ; .^i«» itts tothenum texanum is based on the two Arizona skulls F:AM 95737 and F:AM 59583 (Figures 5-9). In the former, most sutural contacts are obscured by the fragmented condition of the cranial bones, despite the individual's young age (as determined & _*» from the postcranial remains). In the latter skull most sutural contacts are discernible, but they are not sufficiently indicated for complete delimita­ tion of all of the component bones of the skull, especially internally. As far as possible the follow­ ing descriptions are for individual bones; where contacts are not readily discerned the descriptions resort to a geographic context, adopting the pro­ cedure of Vinacci Thul (1945). Following the description of the skull of G. texanum are compar­ ative descriptions for G. anzonae and G. floridanum. NARIAL APERTURE.—The narial aperture is sit­ uated immediately anterior to the first tooth (N~). This aperture, which appears to be fully represented in F:AM 95737, bears the shape of an inverted trapezoid, with a concave upper mar­ gin. The legs of the trapezoid converge down­ ward, their apex forming an angle of approxi­ mately 60° The sides of the narial aperture are apparently formed by the maxillae, the lower edge by the inferred (reduced) premaxilla, which is missing in these specimens, and the upper edge apparently by the nasals. PREMAXILLA.—There is no overt indication of the premaxilla. Its presence in Glyptotherium tex­ anum is inferred by the nature of the inferior FIGURE 12.—Skull fragment oi Glyptothenum floridanum from (palatine) border of the narial aperture in F:AM coastal South Carolina, postzygomatic region (ChM 2415 95737. The edge of the bone forming the lower = CM 43.59 of Ray, 1965): a, ventral, basicranium; b, dorsal, border is somewhat thicker than the bone forming parietal region; c, posterior, occipital region. (Bar = 40 mm.) the lateral and superior borders of the nares. Its outline is also irregular, rather more suggestive of lambdoidal ridge contact. The inner surface of sutural contact than of breakage. If indeed there the bone is smoothly convoluted, offering little was a premaxilla, it was greatly reduced, for the diagnostic information. The outer surface lacks maxillae as preserved extend nearly to the ante­ vascular canals. Associated with this skull is an rior extremity of the narial aperture, which is incomplete hyoid bone, the only one known of a formed by the lower edges of the lateral borders. North American glyptodont. It is an elongate, It also appears that the premaxilla was confined 38 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY to the anterior extremity of the palate, either frontal flare laterally upward to contact the ex­ without or with extremely reduced, superior max­ panded and inflated lacrimal bone. illary processes. As Lydekker (1894) pointed out, The lacrimals occupy the superior anterolateral the bones forming the anterior extremity of the corners of the skull, forming the anterodorsal snout are poorly united and are frequently lost. roots of the zygomatic arches. The indistinct fron- Such appears to be the case in the present in­ tal-lacrimal contact is situated well within the stance. Vinacci Thul (1945) found reduced pre- flattened dorsal region of the skull roof. The maxillae in Glyptodon reticulatus and described in­ lacrimal contacts the zygomatic and nasal proc­ cisive foramina in his specimen. That incisive esses of the maxilla ventrally and anteriorly, and foramina were present in Glyptotherium texanum is it contacts the orbital process of the jugal poster- uncertain, but the symmetry of the anterior edge olaterally. The nasolacrimal duct is situated on of the maxillae suggests their presence. the lateral surface. The surface texture of the NASAL SEPTUM.—The narial aperture is filled lacrimal is finely pitted, in contrast with the with matrix in F:AM 95737. In F:AM 59583, smooth and faintly striated texture of the sur­ however, the passage is open. The broken edge of rounding bones. a broken narial septum, formed by the anterior FRONTAL.—The paired frontals occupy the extension of the perpendicular plate of the eth­ bulk of the upper surface of the cranium. The moid, extends as far forward as N~ in its preserved concavity of the transverse cross section in the condition. Its relatively stout construction at the anterior region becomes flattened rearward at the point of breakage indicates that it continued position of the postorbital process. From that considerably farther forward than its preserved position rearward the frontals are dorsally flat­ extent, probably to the opening of the narial tened and strongly convex laterally. The superior aperture. profile of the midline of the frontals, and therefore NASAL.—As for the premaxilla, the existence of of most of the upper region of the skull as well, is the nasals is questionable. The upper edge of the very weakly concave anteriorly between the zy­ narial opening is broken, but its reduced thickness gomatic roots, flat to weakly convex centrally in suggests that little is missing. This edge is here the plane of the temporal fossae, and again interpreted to be formed by the nasals. It is weakly concave on the posterior third extending possible, however, that the nasals occupied a more toward the parietals. The overall lateral profile is anterior position and that the preserved edge is almost straight; it tapers anteriorly downward at formed by the frontals. If so, the nasals are ex­ an angle of approximately 20° with respect to the tremely reduced, or they may be absent. The palate or tooth row. The construction of the interpretation that the nasals are present is not frontals therefore compares favorably with that supported by the existence of a nasal-frontal su­ described by Vinacci Thul (1945) for Glyptodon ture. There is, however, an indistinct textural reticulatus, although the anterior taper is less severe change on the surface of the bone in the frontal in Glyptotherium texanum. In the anterolateral po­ plane anterior to the zygomatic roots, suggesting sition on the frontal is a distinct postorbital an ankylosed contact, interpreted herein as the process, which fails to extend outward to meet its nasal-frontal suture. Vinacci Thul recognized na­ analog on the zygomatic arch. Thus the orbit sals in Glyptodon reticulatus. His description corre­ communicates freely with the temporal fossa, as sponds closely to the inferences provided above. in all representatives of the Glyptodontinae and LACRIMAL.—The transverse concavity of the Sclerocalyptinae, and unlike the Daedocurinae nasofrontal region reaches its greatest depth in (Hoffstetter, 1958). the frontal plane through the middle of the roots Behind the postorbital process the frontals ex­ of the zygomata. In this position, at the superior tend downward, lapping over the maxillae and anterolateral angle of the skull, the nasal and attaining a fully vertical aspect. This ventrolat- NUMBER 40 39 eral extension of the frontals, which is apparently open anteriorly and closed from the postorbital unique to glyptodonts, extends downward for plane rearward. Thus it appears that this char­ approximately the upper third of the vertical acteristic may vary with age and/or sex. There is diameter of the skull at the temporal opening. no sagittal crest. The margin of the maxillary process of the frontal PARIETAL.—The paired parietals complete the is rounded, extending downward from the rear portion of the skull roof. The median sagittal postorbital process and then upward toward the furrow continues rearward, terminating at the zygomatic process of the squamosal, in a more or nuchal crest. The parietals are relatively small less uniform outline. The flange continues onto and are confined to the dorsal region of the skull. the corresponding portion of the squamosal for a Their external surface is highly irregular, owing short distance before becoming confluent with to the pits, grooves, and ridges attendant with the the zygomatic process. Distinct grooves and ridges cephalic shield that covered these bones. mark the posterior half of the frontosquamosal The frontal-parietal suture is relatively straight flange. These are oriented in a posteroventral- and oriented perpendicular to the midline, as anterodorsal course; the grooves directed slips of already described. The squamosal-parietal suture the temporalis musculature from the lateral sur­ is a continuation of the squamosal-frontal suture. face of the skull to the coronoid process of the Rearward, this suture becomes irregular, gener­ mandible. ally tapering inward toward the occiput. Near The frontal-squamosal suture extends upward the nuchal crest the parietals expand outward in from the anterointernal angle of the zygomatic a pair of nuchal wings, which extend to the lateral process of the squamosal, curving rearward in a angles of the nuchal crest. The posterior border smooth arc on the dorsolateral surface of the skull of the parietals is situated on the nuchal crest, roof. The suture then becomes straightened in the where it contacts the squamous part of the occip­ anteroposterior direction for a short distance be­ ital bone, together forming the broadly V-shaped fore intersecting the frontal-parietal suture in a nuchal crest. This description of the parietal cor­ right angle. The frontal-parietal suture is con­ responds closely with that for Glyptodon reticulatus fined to the dorsal surface of the skull. It is a provided by Vinacci Thul (1945). relatively straight contact, perpendicular to the ZYGOMATIC ARCH.—The glyptodont skull is no­ midline. table for the greatly expanded zygoma, which The frontals are centrally constricted in the includes a massive descending process arising region of the temporal fossae, where the fronto­ from the anterolateral angle and extending down­ squamosal flange (= maxillary process of the ward well beyond the level of the tooth row. The frontal) laps onto the maxilla. The lateral surface development of this process is associated with the of the maxillary process is anteroposteriorly con­ horizontal orientation of the masseteric muscu­ cave. The transverse extent of the border of the lature, which originates on the expanded poste­ frontals is but slightly greater than at the central rior surface of the descending process. (See the constriction. The posterior third of each frontal is discussion of the masticatory apparatus for a marked by several large vascular foramina, which more extensive description.) The bones contrib­ penetrate from the external surface inward and uting to the zygomatic arch are the lacrimal, rearward. The paired frontals meet in a straight maxilla, jugal, and squamosal. The lacrimal was midline contact. At least from the central posi­ described in the previous section. tion, in the plane of the postorbital processes MAXILLA.—The construction of the maxillae is rearward, the suture is open, forming a distinctive partly responsible for the singular appearance of groove. This contact was called a sagittal furrow the glyptodont skull. For purposes of description, ("surco sagital") by Vinacci Thul (1945). In skull the maxilla can be divided into three portions: F:AM 95737, the sagittal furrow appears to be the greatly expanded superior maxillary, the pal- 40 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY atal process, and the massive zygomatic process, ramus (zygomatic process) of the jugal and the which contributes the bulk of the anterior root of zygomatic process of the squamosal in a contin­ the descending process of the zygoma. uous crest. The zygomatic process arises from the upper The posterior surface of the descending process two-thirds of the superior maxillary, from the of the zygoma is uniformly concave in vertical dorsolateral angle of the snout, where it is firmly section, reflecting the lower convexity of the an­ united with the lacrimal. The anteriormost extent terior surface. The upper half forms the anterior of the zygomatic process lies in the frontal plane border of the orbit. The transverse concavity from between N1 and N2, where it is united with the the orbital region diminishes downward, at the lacrimal in the dorsolateral position. The en­ level of the tooth row becoming flattened, obtain­ larged infraorbital canal, which extends rear­ ing a marked convexity on its lower half. ward, inward, and slightly upward through the The superior maxillary is a broad, flattened zygomatic process, divides the bone into two mas­ plate extending from the nares (as here inter­ sive struts. The infraorbital canal is elliptical; its preted) rearward to the plane of the glenoid fossa. diameter is smaller at its posterior emergence into It is greatly extended inferiorly, in maximum the orbit. vertical dimension approximately three-fifths the Nearly all of the anterior surface of the de­ diameter of the maximum anteroposterior length. scending process is formed by the maxilla. It As previously described, the anterior border of descends outward and slightly rearward. The an­ the maxilla appears to form the lateral borders of terior surface in lateral profile is relatively the nares. (The premaxilla may have contributed straight, and that of the lower half is curved short vertical processes at this border.) The pre- rearward in a weak convexity. At the level of the zygomatic portion (narial process) of the maxilla tooth row the anterior border of the process is is weakly convex above N- and N~. In anterior positioned in the plane of the anterior lobe of profile the maxillae in this region converge down­ N . The inferior extremity is situated in the plane ward in conformation with the construction of 4 of the anterior lobe of N The outer (lateral) the narial aperture. The zygomatic region of the border of the descending process is formed by the superior maxillary is modified by the root of the jugal, which is united with the descending process zygoma, which occupies all but the lower third of of the maxilla in a broad vertical contact nearly the bone. reaching the terminus of the process. The jugal Anteriorly, the upper border of the maxilla occupies approximately the lateral one-fourth of contacts the lacrimal in an obscure suture extend­ the descending process in transverse dimensions, ing through the upper portion of the zygoma and the upper half of the lateral side in vertical above the infraorbital canal. Behind the zygo­ dimensions. matic root the maxilla contacts the frontal at the The descending process of the zygomatic arch posteromedial angle of the orbit. Here the contact is anteroposteriorly compressed. The cross section forms nearly a right angle. At this position the is basically lens-shaped, with the posterior surface exposed portion of the maxilla reaches its maxi­ somewhat more convex than the anterior. At the mum superior extent. From the orbit rearward base of the zygoma the medial surface is rounded. the actual frontal-maxillary contact is hidden by From the level of the tooth row downward, the the descending maxillary process of the frontal inner surface forms a medial crest, which contin­ (the frontotemporal flange). The frontal-maxil­ ues to the blunt and rugose terminus. The lateral lary contact apparently diverges beneath the surface of the descending process, the upper half flange, for at the posterior emergence from be­ of which is formed by the jugal, is uniformly neath this process the maxilla is in contact with crested. This construction continues rearward, the orbitosphenoid, which is wedged between the forming the ventral margin of the horizontal maxilla and the frontal. At a position directly NUMBER 40 41 beneath the mandibular facet the border of the opening adjacent to N , the sulcus becomes maxilla contributes to the large foraminal com­ deepened rearward. Tubercle-like processes on plex, the bulk of which is formed by the foramen the medial borders nearly encompass the furrows. rotundum. Posteriorly the maxilla is overlapped The left palatine furrow enters the palatine fora­ by the sphenopterygoid complex. men adjacent to the anterior lobe of N-. The right The postzygomatic region of the superior max­ furrow becomes covered by a thin process of bone illary is a smooth, flat plate. It is weakly concave extending from the posterior lobe of N~ to the in both the vertical and anteroposterior direc­ anterior lobe of NA The right palatine foramen tions. Superiorly it forms the inner border of the opens posterior to this covering, opposite the mid­ longitudinal groove produced by the maxillary dle lobe of N5; thus the palatal foramina and the process of the frontal. This groove seems to course palatal furrows are neither equally constructed rearward and to enter the sphenomaxillary fissure nor symmetrically disposed. Ordinarily this con­ (foramen rotundum region). Presumably, the struction would receive little attention, for a small flange provided protection from the masseteric amount of asymmetry is expected. However, it is and temporalis musculature for the blood vessels notable that the skull UMMP 34826 (G. arizonae) and nerves, which were transmitted forward from exhibits an identical construction, except that the this foraminal complex (i.e., especially the optic left palatine furrow also possesses a bony enclo­ nerve, the foramen for which is apparently situ­ sure anterior to the foramen; the asymmetry in ated within this foraminal complex). This fora­ the position of the foramina is identical. minal complex is more extensively discussed in A pair of small, posterior palatine foramina is the description of the alisphenoid. situated adjacent to the anterior lobe of each From the middle of the tooth row rearward, N". The posterointernal border of the terminal the maxillae taper downward and outward, con­ alveolar process shares in the formation on each forming to the curvature of the alveoli, thus side of a large posterior palatine fissure paralleling producing the downward divergence (i.e., upward the contour of the alveolar bone. The posterior convergence) of the maxillae in frontal section. border of each fissure is formed by the palatine The palatal process of the maxilla occupies the bone, which is fused to the maxilla in a solid entire palatal region. The reduced palatines are contact. The posterior palatine fissure was noted confined to the region of the posterior narial by Vinacci Thul (1945), which he identified as aperture. Anteriorly the maxillae apparently are the posterior palatine foramen. In contrast, Glyp­ fused with the weak premaxillae, and they appear totherium texanum has well-defined palatal foram­ to form the posterior border of the incisive foram­ ina situated anterior to the fissures. The fissures ina, as discussed previously. In transverse section apparently transmitted blood vessels. This con­ the palate is weakly concave anteriorly, with more struction is probably a consequence of the ex­ pronounced concavity rearward. Posteriorly the tremely compressed condition of the posterior concavity is very deep owing to the downward part of the skull, providing vascular exit in an extensions of the aveolar processes. In sagittal otherwise extremely crowded region. section the palate is weakly concave for its ante­ The upper and lower teeth are discussed follow­ rior half, becoming markedly convex posteriorly. ing the description of the mandible. This construction conforms closely to the lateral JUGAL.—In vertical and transverse sections the profile of the occlusal surfaces of the teeth. descending process of the jugal is distinctly con­ There is a weak ridge at the midline at the cave. By virtue of this concavity, this region is position of N- to N-, and anteriorly, between N- separate from the maxillary part of the process. and N-, there is a distinct midline groove. Adja­ This roughened surface of the jugal forms the cent to the alveolar processes of each maxilla is a concave posterolateral face of the descending deep longitudinal sulcus. From its broad anterior process of the zygoma; it is situated half above 42 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY and half below the tooth row. The maximum Behind the temporal fossa and situated entirely dorsoventral diameter is approximately 65 mm upon the laterally directed (proximal) portion of on the preserved zygomatic arch of F:AM 95737; the zygomatic process is the glenoid articular it is approximately 15 mm wide. This area is a facet. This transversely elliptical facet is flat both principal region of attachment for the masseteric transversely and dorsoventrally, although on both musculature. (The other primary attachment sur­ G. texanum skulls the facet is very weakly concave face for the masseteric muscles is the tip of the laterally, and weakly convex medially, in trans­ descending process, which is heavily rugose.) verse section. The facet is confined to the lower The temporal process of the jugal parallels the half of the posterior surface of the zygomatic midline of the skull. It diminishes in thickness process. It is directed rearward and slightly down­ rearward, where it contacts the zygomatic process ward and outward. The postglenoid area is an of the squamosal. The sutural contact is obscured open region from the facet rearward to the par­ on F:AM 95737, the only one of the pair of occipital process. Thus the articular facet for the Arizona skulls with an intact zygomatic arch. condyle of the mandible is a flat surface, devoid From its contact with the temporal, the jugal is of confining flanges or processes, with a broad, inclined forward and downward with respect to open postglenoid region. the tooth row, forming an angle of approximately As indicated above, the squamous region of the 50°. The temporal process of the jugal is laterally squamosal is divided into superior and inferior compressed with a lens-shaped cross section. In surfaces by the continuation of the superior crest the region of contact with the squamosal (inferred of the zygomatic process rearward to the nuchal as the middle of the zygomatic arch), the zygoma crest. The superior surface is anteroposteriorly is weak and underdeveloped relative to the ante­ convex, with flattening of the profile rearward. rior and posterior regions. The transverse section is anteriorly concave, be­ SQUAMOSAL.—The squamosal is a large com­ coming flattened rearward, with the posterosu- pound bone. The squamous portion and the pe­ perior surface facing dorsolaterally and slightly tromastoid are fully represented in the G. texanum forward. From its vertical contact at the fronto- specimens. The tympanic bulla is not present. squamosal flange, the frontal-squamosal suture The squamous portion includes four distinct re­ extends vertically upward, and curves rearward gions: the zygomatic process; the squamous at the dorsolateral angle of the skull roof. From process, occupying the posterior dorsolateral re­ that position, the frontal-squamosal suture con­ gion of the skull; the paroccipital process; and the tinues posteriorly, becoming confluent with the postglenoid process. parietal-squamosal suture. The course of this su­ The zygomatic process arises from the lateral ture is difficult to ascertain posteriorly. It curves extremity of the squamous portion. The free an­ inward, where the parietals are transversely con­ terior surface of the zygomatic process forms the stricted, then turns outward and rearward, inter­ posterointernal border of the temporal fossa. The secting the superior crest of the squamosal and weakly crested, lower border of this surface is a becoming incorporated into the lateral angle of continuation of the inferior crest of the zygoma, the nuchal ridge, where the squamosal shares in which is confluent with the frontotemporal the formation of this crest with the squamous flange. The superior crest of the zygomatic process of the occipital. The posterior half of the process continues rearward and upward, curving superior surface of the squamosal is pierced by medially on the posterior third of the squamosal several large vascular foramina for the cephalic to become confluent with the nuchal crest. This shield similar to those on the frontals and parie­ continuation of the superior crest divides the tals. squamous portion into separate superior and in­ The inferior surface of the squamous process ferior surfaces. (the region below the superior crest) includes NUMBER 40 43 an expanded, laterally directed supraoccipital suture passes toward the foramen ovale for a short process forming the dorsolateral angle of the nu­ distance and then turns upward, passing above chal crest. The temporal-occipital contact extends the foramen. The posterior margin of the postgle­ laterally rearward and downward for a short noid process is markedly thickened, for open con­ distance. The bone in the region of the temporal- tact (nonsutural) with the anterior region of the occipital contact is greatly thickened. The tem­ petromastoid in two poorly defined, rearward poral-occipital contact diverges, exposing the pos­ facing concavities. This surface extends upward, terior portion of the petromastoid bone. More outward, and rearward toward the paroccipital laterally and ventrally the squamous portion and process. the mastoid process reconverge at the position of The petromastoid includes two distinct regions, the paroccipital process, half of which is contrib­ the petrous region and the mastoid process, which uted by the squamosal. The fenestra produced by form the anterolateral portion of the paroccipital this separation of the squamosal-occipital contact process. The mastoid process is heavily pitted and exposes the posterodorsal region of the petromas­ irregular. It is suturally united with the paroccip­ toid bone; there is no indication of any bony ital process of the squamosal anteriorly, and with covering over this opening. From the paroccipital the paroccipital process of the occipital poste­ process forward the squamous portion of the tem­ riorly. The participating bone in both contacts is poral diminishes in dorsoventral thickness, its greatly thickened. Thus the mastoid process oc­ upper and lower borders converging with the cupies an intermediate position between the squa­ zygomatic process above the mandibular fossa. mosal and occipital, the respective processes of The deeply concave mandibular fossa occupies each forming the compound paroccipital the open region ventral to the lower squamous process. Internally, both above and below, the portion. petrous portion is exposed. The upper exposure The mandibular fossa is bounded above by the of the petromastoid (posterolateral fenestra) was medial projection of the lower squamous bone. described above. The concavity thus formed maintains the shape The lower exposure is afforded by the lack of of a quarter sphere, opening outward, downward, the tympanic bullae in the preserved specimens. and forward. The anterior boundary is formed by The petrous squamosal arises from the antero- the zygomatic process, the interior surface by the medial region of the mastoid process. It projects postglenoid process. inward, forward, and slightly downward, par­ The postglenoid process of the squamosal is a tially filling the void between the basioccipital flattened, roughened region ventral to the squa­ and the postglenoid process of the squamosal. Its mous process and posterior and ventral to the shape is pyramidal, suggestive of a doubly convex zygomatic process. The surface of the bone bears prism. The external auditory meatus emerges an irregular contour; it generally faces laterally from its ventrolateral surface. A laterally directed and slightly ventrally. The maxillary-squamosal channel on the ventral surface of the mastoid suture continues inward and rearward from the process evidently communicated with the meatus zygomatic process. At approximately half the by a cartilaginous connection. The posterior lac­ distance between the foramen ovale and the zy­ erate foramen is bounded anteriorly by the pos­ gomatic process the alisphenoid overlaps both the terior surface of the petromastoid. This elliptical maxilla and the lower border of the squamosal in foramen is bounded posteriorly by the basi­ a forward directed process. The maxillary-squa­ occipital, and it communicates freely on its me­ mosal suture there passes beneath the alisphe­ dial border with the apparently open basitem- noid, and its further course is not evident. The poral vacuity. This vacuity is formed by the postglenoid process contacts the superior border failure of the petromastoid to enclose the poster­ of the alisphenoid in a well-defined suture. This olateral region. Its relatively large size is likely 44 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY due to the coalescence of several of the major illa, palatine, pterygoid, and alisphenoid partici­ foramina of the braincase resulting from the pate in the formation of the pterygoid processes. crowded condition of the component bones in These are situated posterior to the last tooth this region. alveolus on each side, at the lateral corners of the PTERYGOID-PALATINE-SPHENOID REGION.— vertical posterior narial aperture. Their construc­ Both Vinacci Thul (1945) and Burmeister (1874) tion is rather bosslike, each forming a prominent, described the bones of the basicranium and pos­ rounded, and rugose surface at this position, terior nares in some detail. Vinacci Thul identi­ which is essentially in the horizontal plane of the fied only the palatine and "sphenoids." He ac­ palate. The pterygoid processes taper upward curately described and named the spheno- along the margin of the nares, the rugosity and maxillary fissure, which contains the complex of massiveness giving way to the sharp smooth sur­ foramina including the foramen rotundum. He face of the aperture. also mentioned a round foramen posterior and The posterior narial aperture is a vertical, trap­ lateral to the sphenomaxillary fissure but did not ezoidal opening. The frontal plane of the aperture identify it; his description seems to indicate the lies in the same plane of contact as the basisphe- foramen ovale, as described herein. Burmeister noid-basioccipital suture. (The position of this correctly identified these two foramina, and he suture graphically indicates the extreme shorten­ discussed the palatine-pterygoid complex without ing of the bones in this region.) The upper border reaching a firm conclusion regarding their respec­ of the aperture is formed by the basisphenoid. tive contributions to the posterior narial aperture. The superolateral borders of the aperture are He suggested, however, that the pterygoid formed formed by the common angle of the alisphenoid- the posterointernal margins of the lateral walls of pterygoid contact. Inferiorly the pterygoid proc­ the narial aperture. Vinacci Thul indicated that esses occupy the lateral margins of the nares in a the palatines occupied this position instead. This blunt rugose surface. The inferior border of the region presents serious difficulties at interpreta­ nares is formed by the palatal (horizontal) proc­ tion in the Arizona skulls because of the strongly esses of the palatines, which are fused at the fused sutures, and because of the peculiar con­ midline. struction of the alisphenoid, as described below. PALATINE.—The reduced but stout palatines The following description concurs with Burmeis­ occupy the inferior border of the posterior narial ter's suggestions and is at variance with Vinacci aperture, which is situated immediately behind Thul's description. It should be pointed out, how­ the last alveolus. Each palatine possesses two ever, that until the taxonomic identities of the processes, the horizontal process and the vertical South American and North American genera are process. The posterior margin of the horizontal more satisfactorily established, the differences process is free, occupying the inferior border of noted herein may not indicate erroneous inter­ the narial aperture. Anteriorly the horizontal pro­ pretations; instead there may be significant dis­ cesses, which fuse at the midline, are firmly united tinctions between taxa, and the one or the other to the palatal processes of the maxillae. The of the skulls described by these two authors might anteroposterior diameter of the horizontal process not belong in Glyptodon proper. of the palatines at the midline is only about 1 cm; The cause for confusion in this region lies in thus the palatines contribute little to the total the extreme reduction in the anteroposterior ex­ length (approximately 18 cm) of the base of the tent of these bones, unlike any other mammal. palate. The palatine-maxillary suture is a straight This reduction is a consequence of the peculiar transverse contact, and it is limited to the region masticatory apparatus, for which the posterior behind the frontal plane of the posterior alveoli. region of the skull has been foreshortened in order The lateral extension of the maxilla-palatine con­ to optimize the effect of the massive masseteric tact is interrupted by the posterior palatine fis­ musculature. Accordingly, portions of each max­ sure, the posterior margin of which is contributed NUMBER 40 45 by the palatine. The horizontal process of the pterygoid. Ventrally, each vertically oriented palatine contacts the exterior surface of the su­ pterygoid contributes the major portion of the perior maxillary immediately behind the last al­ large rugose pterygoid tuberosity. The pterygoid veolus. This short contact terminates laterally in possesses the portion of the tuberosity that in­ a sharp angle at the intersection with the ptery­ cludes the greatest development of the tubercle. goid process of the alisphenoid, on the lateral The inferior and lateral contributions of the pal­ surface of the pterygoid tuberosity. atine and alisphenoid, respectively, participate as The posterior border of the lateral portion of adjacent bones, circumscribing in part the tu­ the palatine is firmly united with the alisphenoid berosity. The short pterygoid-maxilla suture on the inferior posterolateral surface of the pter­ passes through the lower extremity of the ptery­ ygoid tuberosity. This suture continues medially goid tuberosity. as the pterygoid-palatine contact on the inferior The pterygoid is an elongate, triangular pyra­ posteromedial surface of the pterygoid tuberosity. mid, apex upward. The basal contact with the The pterygoid-palatine suture extends vertically maxilla was described above. The three faces of on the inner surface of the posterior nares as the the triangular cross section face anteriorly, pos- posterior extremity of the vertical process of the teromedially, and laterally. The anterior side palatine. Superiorly the vertical processes fuse forms the pterygoid-palate contact in a strong with the basisphenoid in a right angle contact. suture. The posteromedially facing side forms the The vertical process is anteroposteriorly convex posterior extremity of the interior wall of the in an oblique direction, from anteroventral to nares. The posterior margin, which superiorly posterodorsal. Thus, viewed from the rear, the becomes a sharp crest, is parallel with the pala- walls of posterior nares diverge rearward; this tine-pterygoid suture. Dorsally the pterygoid is divergence is maintained by the pterygoids, united with the horizontal basisphenoid in a which form the posterior margin of the nares. right-angle suture. From the upper extremity of The anterior extent of the palatines cannot be the palatine-pterygoid suture the pterygoid-basi- determined. In the region of the anterior apex of the basisphenoid the vertical processes of the sphenoid suture extends in an oblique superior- palatines nearly come into contact. Their walls lateral-posterior direction, terminating in the diverge inferiorly in this area, thus creating in sharp apex of the pyramidal construction of the cross section a trapezoidal outline of the interior pterygoid. surface of the narial aperture in the frontal plane The posterolaterally directed face is overlapped of the apex of the basisphenoid. This cross section superiorly by the alisphenoid (i.e., the pterygoid becomes reversed posteriorly, owing to the con­ lies internal and inferior to the alisphenoid). The vexity of the vertical processes of the palatines inferior region is transversely thickened and and their consequent oblique posterior diver­ broadly rounded, forming on its exposed surface gence. Thus the cross section of the nares at the the major portion of the pterygoid tubercle, as contact with the pterygoids is an inverted trape­ described above. There is a distinct vertical flange zoid. The same cross section, with greater dimen­ on the lateral surface of the vertical crest of the sions, obtains at the posterior border of the nares. pterygoid. This flange expands ventrally with the Therefore, as interpreted here, the palatine is increasing thickness of the bone. This groove confined in its horizontal process to the region appears to have transmitted blood vessels and/or behind the last alveolus. The vertical process, nerves from the basioccipital-temporal fenestra, however, extends forward as the lateral wall of anterior to the petromastoid, downward to the the nares at least to the plane of N2, and perhaps muscles attaching to the pterygoid tuberosity. somewhat farther. Near its superior extremity, the anterior wall of PTERYGOID.—Occupying the internal posterior this groove is contributed by a flange of the extremity of the lateral wall of the nares is the alisphenoid. The posterolateral surface of the 46 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY pterygoid thus forms the bluntly rugose and cannot be ascertained. Presumably this union is greatly thickened posteroinferior terminus of the situated at the anterior constriction of the supe­ narial encasement. rior borders of the palatines. This construction of the pterygoid tuberosity, The narial surface of the basisphenoid is which forms a broad, blunt, rearward-facing sur­ smooth and flat. The basicranial surface is trans­ face for muscle attachment, is closely associated versely concave and rugose. with modifications attendant with the develop­ ALISPHENOID.—As interpreted here, the ali­ ment of the descending process of the zygoma sphenoid and orbitosphenoid are separate bones. and the consequent horizontal direction of the The alisphenoid occupies the lateral surface of masseteric musculature. the vertical wall of the posterior nares. The iden­ Vinacci Thul's (1945) description of the pala­ tification of the alisphenoid is founded on the tine corresponds closely to that provided here for presence of the foramen ovale, which is situated the pterygoid-palatine complex. Although we in the posterosuperior position of the bone. The have not examined the skull he described, it seems alisphenoid anteriorly contributes the bulk of the possible that the palatine-pterygoid suture was sphenomaxillary fissure, the borders of which are obscured because of the heavily rugose surface of in part shared with the superior maxillary and the pterygoid tuberosity. The palatine and pter­ the reduced orbitosphenoid. ygoid together could easily be taken for a single From its strong sutural union with the postgle­ bone, and because of their position (posterior to noid process of the temporal, the alisphenoid the maxillae), identification as palatine would be extends downward in a vertical process, which reasonable. The descriptions provided here for lies on the lateral surface of the pterygoid. This Glyptothenum texanum are in close agreement with thickness of the bone diminishes downward, ter­ Burmeister's (1874) suggested identifications of minating as a thin flange abutting the pterygoid the cranial bones of Glyptodon. near the superior apex of the posterior extremity BASISPHENOID.—The triangular basisphenoid of the vertical process of the palatine. The inferior occupies the roof of the posterior end of the nares extremity of the alisphenoid contributes the ru­ and the base of the cranium. Its lateral borders gose, superior lateral surface of the pterygoid converge anteriorly in an apex, owing to the process. anterior convergence of the palatines, with which The lateral-facing external surface of the ali­ the sides of the basisphenoid are fused in a right sphenoid is irregular. Beneath the foramen ovale angle. Posteriorly the basisphenoid contacts the the external surface is anteroposteriorly concave, basioccipital in a strong union. This suture is apparently providing a broad, downward, and situated in the vertical plane of the posterior forward-directed groove for the transmission of narial aperture. Both bones at this contact are the maxillary branch of the trigeminal nerve. In thickened medially, and the thickness decreases vertical section the external face of the alisphe­ laterally, forming a lens-shaped cross section at noid is superiorly concave, nearly attaining a the suture. The posterolateral apices of the basi­ horizontal aspect at the temporal suture. The sphenoid are free, contributing to the border of laterally directed inferior surface is fiat and ru­ each basitemporal vacuity. gose. The lateral borders of the basisphenoid are The posterior margin of the alisphenoid con­ united posteriorly with the upper extremities of tributes the anterior flange of the vertical ptery- each pterygoid in a short, right angle, sutural goid-alisphenoid groove, as described above. The contact. The basisphenoid appears also to contact alisphenoid-pterygoid suture lies within this horizontally the alisphenoid immediately behind groove. The suture extends downward and curves the foramen ovale, although the sutural contact anteriorly on the external surface of the pterygoid, is obscure. The anterior contact with the vomer where the thin descending process of the alisphe- NUMBER 40 47 noid laps over the pterygoid. This suture contin­ sphenoid, as the contact between these two pre­ ues forward and upward as the alisphenoid-su- sumed, separate bones is obscure. However, the perior maxillary suture. In this region also, the recessed position of the orbitosphenoid (as here alisphenoid overlaps the maxillary, so that the described) and its relatively smooth texture dis­ maxillary-pterygoid suture is covered by the ali­ tinguish this region from the main body of the sphenoid. The anterior border of the alisphenoid alisphenoid. If in fact the two bones are ontoge­ is irregular in shape, passing vertically upward netically fused, the orbitosphenoid process repre­ from the rounded lower margin toward the sents the orbitosphenoid proper. Hence these two sphenomaxillary fissure, where the alisphenoid is bones are here considered as separate entities. deflected inward as its posterior wall. The sharp ORBITOSPHENOID.—The anterior extension of anterior margin therefore shields the spheno­ the thin alisphenoid covers the surface of the maxillary fissure as a sharp flange. Situated on orbitosphenoid. In skull F:AM 59583 the alisphe­ the anterointernal wall of the alisphenoid, within noid processes are broken, exposing the reduced the fissure, is the foramen rotundum. Its elliptical orbitosphenoid. Only the exterior surface of the orifice opens downward and forward. The ante­ orbitosphenoid can be described. The orbitosphe­ rior lacerate foramen is situated immediately an­ noid contribution to the floor of the brain cavity terior to the foramen rotundum, within the fis­ cannot be determined. Lying beneath the ali­ sure. Its opening is directed laterally. The walls sphenoid flange, which contacts, but does not of the anterior lacerate foramen are formed by unite with, the posterior-inferior angle of the the superior maxillary and the alisphenoid. The frontal behind the maxillary process of the fron­ optic foramen probably also lies within the tal, the orbitosphenoid forms the upper surface of sphenomaxillary fissure, in the region contributed the anterior extension of the sphenomaxillary by the orbitosphenoid. fissure. It appears to be comprised of two parts, The sphenomaxillary fissure thus houses the a small horizontal process, continuous with and anterior lacerate foramen, the foramen rotun­ suturally united to the frontal, and a vertical dum, and probably the optic foramen. The pos- process, suturally contacting the superior maxil­ terosuperior margin is formed by the alisphenoid lary. The two processes are indistinctly separated flange, the anteroinferior margin by the recessed by the groove that passes anteriorly from the superior maxillary, and the anterosuperior corner sphenomaxillary fissure toward the frontomaxil- by the orbitosphenoid. The fissure opens laterally lary flange. The orbitosphenoid appears to be downward and forward, providing exit for nerves confined to the region behind and inferior to the passing forward beneath the exposed surface of maxillary process of the frontal. Because the latter the orbitosphenoid to the groove formed by the process overlaps the orbitosphenoid anteriorly, its maxillary process of the frontal. Thus the optic forward extension cannot be ascertained. It lies nerves and the anterior division of the trigeminal internal to the tubercles bordering the frontal are provided considerable protection by the com­ flange, and it occupies the intermediate position bination of the sphenomaxillary fissure and the between the sphenomaxillary fissure and the an­ maxillary process of the frontal on the side of the teriorly directed groove lying beneath the frontal face beneath the temporalis musculature. flange. The orbital foramen is not exposed; pre­ The alisphenoid is expanded into a short, an­ sumably it is situated within the sphenomaxillary teriorly directed flange, overlying the orbitosphe­ fissure. noid from the anterior region of the temporal OCCIPITAL.—For purposes of description the suture. The orbitosphenoid-alisphenoid contact is occipital can be divided into four regions: the hidden by this flange. It should be noted that the squamous region, the paroccipital region, the oc­ bone here identified as the orbitosphenoid may cipital condyles, and the basioccipital; these well- actually be ontogenetically fused with the ali­ developed regions are solidly fused into the single 48 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY occipital bone to form the large, posterior extrem­ tion by a weak external occipital crest, extending ity of the skull. obliquely forward from the dorsolateral margin The thin-boned, dorsal, squamous region is of the foramen magnum. The dorsal portion of broadly concave in the transverse direction and the paroccipital region is flat. Its free anterosu- anteroposteriorly flat. The dorsal surface is dis­ perior margin forms the rear border of the pos­ tinctly concave along the nuchal crest, where the terior petromastoid fenestra. Beneath the fenes­ bone is greatly thickened to meet the parietal and tra, in the region immediately lateral to the oc­ temporal bones. The squamous plate is inclined cipital condyle, is the paroccipital process. The forward at an estimated angle of 45° with the mastoid and paroccipital processes of the squa­ posterior palatal region. Thus the squamous oc­ mosal contribute the anterior third of the com­ cipital attains a significant horizontal component pound paroccipital process. Ventrally the mas- in its attitude, apparently unlike any other ter­ toid-paroccipital suture terminates at the lateral restrial mammal. In dorsal aspect the squamous extremity of the border of the basioccipital. The region forms a broad, concave plate, fusing an­ paroccipital process is separated from the occipi­ teriorly with the parietals and laterally with the tal condyle by a narrow, shallow groove. The squamous temporals. The apex of the nuchal external rugose surface of the process lies below crest forms approximately a right angle. The the plane of the condyle, and it maintains a posterolateral extensions of the nuchal crest be­ similar profile. come greatly expanded in the dorsal region of The basioccipital is a broad Y-shaped region contact with the squamous portions of the squa- forming the posterior basicranial extremity. An­ mosals to form the supraoccipital processes. These teriorly the basioccipital is medially united with massive tuberosities form the posterodorsal ex­ the basisphenoid in the vertical plane of the tremities of the skull. Apparently Melton (1964: posterior narial aperture. As described previously, 131) was referring to these when he stated that the bone on both sides of this suture is centrally "the external occipital protuberance sits well for­ thickened, diminishing in thickness laterally. The ward for the attachment of heavy neck muscles," basioccipital in this region is convex on both the although he may have intended to indicate the internal and external surfaces. Its anterolateral apex of the nuchal crest at the juncture with the extent is exceeded by the extent of the basisphe­ parietals. In actuality there is no external occipi­ noid, forming a notch at this constriction for tal protuberance, which is usually taken to des­ vascular or nervous passage from the basitem- ignate a median tubercle on the squamous occip­ poral vacuity into the nares. The lateral borders ital, sometimes contiguous with the nuchal crest. of the basioccipital are free, forming the internal Between the supraoccipital processes the squa­ margins of the basitemporal vacuities. These bor­ mous occipital is smooth and uniformly concave. ders are posteriorly divergent in an angle of ap­ The rugose supraoccipital processes are situated proximately 60°. The posterior extremity of the above the posterior petromastoid fenestra and basioccipital forms the inferior border of the for­ form the principal attachment surface for the amen magnum. A median line connecting the heavy neck musculature. superior and inferior borders of the foramen mag­ Posteriorly the squamous occipital contributes num passes forward and downward, intersecting the broad, superior border of the foramen mag­ the median posterior margin of the palate, and num. forms approximately a 45° angle with the palate. The paroccipital region includes that portion Thus the elliptical foramen magnum opens pos­ of the occipital that is situated posterior and teriorly downward. This construction is likely ventral to the supraoccipital process and the oc­ attendant with the modifications for the strong cipital contribution to the paroccipital process. neck musculature passing over the occiput to the This region is separated from the squamous por­ nuchal crest and the supraoccipital processes. NUMBER 40 49

Piercing each of the lateral wings of the basi­ anzonae is the presence of a deep canal behind the occipital near the occipital condyles are the round paroccipital process. This canal passes downward hypoglossal foramina. A notch on the posterior and inward from the region of the posterior "fe­ extremity of the free lateral border of each wing nestra" of the petromastoid toward the region of contributes the posterior margin of the posterior the (missing) tympanic bulla. In actuality, the lacerate foramen. posterior "fenestra" region of G. arizonae is covered The paired occipital condyles occupy the re­ by a thin sheet of bone from the mastoid process gion lateral to the foramen magnum, the lateral of the temporal bone, uniting with the mastoid margins of which are formed by the medial bor­ process of the occipital on the inner surface of this ders of the condyles. The transversely elongate channel. There is no indication of this covering condyles are relatively small. Their shape approx­ over the posterior fenestra in the skulls of G. imates the lower half of a conical section, the base texanum, nor is there any indication of a canal in of the cone medial. A pair of small accessory this position. facets projects medially from the ventral border Another important distinction occurs in the of each condyle; they do not meet at the midline. construction of the basioccipital region. Whereas These facets evidently provided a ventral stop for the basioccipital-basisphenoid contact lies in the the atlanto-occipital articulation. plane of the posterior nares in G. texanum, this The skull of G. anzonae (UMMP 34826, 38761, union appears to occur more posteriorly in G. Figure 10) differs from that of G. texanum in several arizonae. Moreover, the external surfaces of these important respects, especially in the postglenoid two bones meet in a poorly defined obtuse angle region. The prezygomatic region is unknown; it in G. arizonae, rather than being nearly contiguous is presumably similar to that of G. texanum as as in G. texanum. The median inferior margin of described above. Except for greater absolute size the foramen magnum extends farther forward in in all dimensions (Table 1), the preglenoid region G. arizonae, imparting a more deeply clefted Y- of G. arizonae closely resembles that of G. texanum. shaped basioccipital. The most significant distinctions occur in the Another distinction characterizing G. arizonae is basicranial region. the presence of a deep concave pit on the anterior Whereas the bone comprising the upper border border of the ventral extremity of the occipital of the mandibular fossa is smooth and entire in condyle. The hypoglossal foramen is situated G. texanum, there is a distinct glenoid foramen in within this recess. Concomitantly, the occipital G. arizonae. This foramen is apparently not a blind condyle is more extensively developed and the passage, but provides exit from the cranial cavity separation of the articular region from the non- to the glenoid fossa for nerves and blood vessels articular surface is more well defined. There is a to the muscles of the temporal-mandibular cap­ similar depression above each occipital condyle sule. We have been unable to establish the proper on the squamous occipital. These recesses presum­ homology of this foramen. Its apparent absence ably allowed greater vertical motion at the at­ in G. texanum may be due to the preservation in lanto-occipital joint. the skulls of this species. Because of the fragmentary nature of the skull The foramen ovale and the sphenomaxillary material assigned to G. floridanum, fewer distinc­ fissure appear to be identical in these two species. tions can be determined for this species. The The paroccipital processes exhibit a much greater cranial fragments of USNM 6071 (Figure 11) development in G. arizonae than in G. texanum, indicate no obvious differences in the palatal and although in the skulls of the latter species the central cranial regions. The basicranial region in processes are partially broken and their full extent this specimen indicates a closer resemblance to G. is therefore indeterminate. arizonae in the possession of the problematical A distinction in the paroccipital region of G. glenoid foramen. The well-preserved postglenoid 50 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY region of ChM 2415 (Figure 12), which overlaps associated with ChM 2415 (ChM 2090, 2417, and only slightly in its preserved parts with USNM 2418) indicate a close resemblance to both G. 6071, presents several important distinctions that anzonae scutes and G. floridanum scutes from other presumably serve to fix the characters of G. flori­ localities. This provisional assignment of the danum (assuming that the assignments of USNM South Carolina material to G. floridanum is sup­ 6071 and ChM 2415 to the same taxon as the ported also on geographic grounds, for G. anzonae remaining Florida and Texas G. floridanum mate­ and G. texanum are not certainly known from rial is correct). eastern United States, although isolated Florida The squamous occipital of ChM 2415 bears a material, mostly scutes, indicates the early pres­ median longitudinal ridge from the apex of the ence of glyptodonts, probably G. arizonae, in that nuchal crest to the foramen magnum. This ridge state. The associated vertebrate fauna for the distinctly separates the squamous occipital into Edisto Island cranium appears to be entirely late right and left halves. There is no corresponding Pleistocene (Roth and Laerm, 1980), consistent ridge in either G. texanum or G. arizonae. Also, the with our assignment to G. floridanum. squamous occipital is more steeply inclined with MANDIBLE (Table 2).—Mandibles of various respect to the cranial roof, forming an angle of North American glyptodonts have been described approximately 65° by Gidley (1926), Holmes and Simpson (1931), The occipital condyles are more nearly cylin­ Meade (1953), Melton (1964), and Lundelius drical in comparison with the cone-shaped con­ (1972). Except for the very fragmentary mandib­ dyles of G. texanum and G. arizonae. The inferior ular remains of UF/FGS 6643 from the Catalina and superior recesses bounding the condyles are Gardens site in Florida and a nearly complete intermediate in development between the deep hitherto unreported mandible from Orange recesses of G. arizonae and the flat surfaces of G. County, Florida (USNM 11318), there is no new texanum. The external-medial extremity of the material for study. However, there have been hypoglossal foramen extends as a short open canal several erroneous statements and consequent mis­ in G. floridanum; there is no canal-like exterior interpretations concerning the mandible of North construction in either G. texanum or G. arizonae. American representatives. Only Lundelius (1972) The basioccipital-basisphenoid contact more has attempted a comprehensive review, and his closely resembles the flattened contact of G. tex­ determinations were incomplete for lack of com­ anum. There is a pronounced cleft in the median parative materials. A primary misconception cen­ ventral border of the foramen magnum, contrast­ ters around Melton's (1964) erroneous assignment ing with the smooth contour in the other two of the root end of Ns, which had fallen from its species. A final noteworthy distinction of ChM alveolus and was cemented to the mandible, as 2415 is a pronounced downward extension of the the occlusal surface of Ns in the Seymour G. ventral extremity of the petromastoid, producing anzonae. This situation is more fully treated in the a distinctive crest at the terminal apex. This description of the dentition, which follows this region is flattened in G. arizonae and smoothly section. Another problem centers around Gidley's rounded in G. texanum. (1926) and Melton's (1964) failures to describe As Ray (1965) pointed out, ChM 2415 differs more thoroughly the mandibles of USNM 10536 from USNM 6071 in the construction of the and UMMP 38761, respectively, from the Curtis medial parietal region. This region is elevated in Ranch and Seymour faunas. A number of fea­ ChM 2415 and is flattened in USNM 6071 and tures considered by Lundelius (1972) as distinc­ in G. texanum and G. anzonae. Future recoveries tive for Boreostracon flondanus (= G. floridanum) are may substantiate this distinction between ChM actually present in the Arizona and Texas man­ 2415 and USNM 6071, necessitating reevaluation dibles, and though distinctive, the differences are of the taxonomy as here proposed. Three scutes not as great as previously supposed. NUMBER 40 51

Descriptions of mandibles of the South Amer­ prises a nearly complete composite mandible. ican genus Glyptodon (in the classic sense, see Comparisons of the mandibles of these two spec­ Hoffstetter, 1955) indicate a close resemblance imens indicate their close relationship and pro­ with the North American Glyptotherium. Because vide partial foundation for the determination of of the need for a modern comprehensive study of specific identity between the Curtis Ranch and South American glyptodonts, and for other more Seymour representatives. The mandible (TMM problematical reasons as discussed in the taxon­ 934-37) from the Holloman gravel pit, Okla­ omy section, the similarity of mandibular char­ homa, originally accorded generic distinction by acters between the North and South American Meade (1953) as Xenoglyptodon fredericensis, and representatives should not be so construed as to later recognized as identical with the Seymour indicate generic identity. Hence, comparisons of glyptodonts by Melton (1964) is a partial right mandibular characters, like comparisons of other mandible with portions of the horizontal and features, are here intended to indicate only the ascending rami, including Ng and N7 with missing close similarity. crowns, and partial alveoli for Nj, Ns, and Ng. Only the mandible of G. arizonae is completely Two scutes were associated with this mandible, known. USNM 10536 (Figure 13) is an entire and they exhibit no important distinction from mandible, with a full set of teeth. UMMP 38761 the Seymour representatives. The present study (Figure 14), in separate right and left sides, com­ agrees with Melton's (1964) contention that Xen-

'*-

\ ' \

i 7"T|}>

FICURE 13.—Mandible oi Glyptothenum arizonae from the Curtis Ranch fauna, Arizona (USNM 10536): a, dorsal; b, posterior; c, right side. (Bar = 10 cm; c slightly larger.) 52 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

rj

i

* i ^4 i/ !

^.:,>- if* a •**_•.

FIGURE 14. — Left mandible oi Glyptotherium arizonae from the Seymour formation, Gilliland local fauna, Texas (UMMP 38761): a, medial; b, lateral. (Bar = 10 cm.) oglyptodon fredericensis (Meade, 1953) is synony­ mous with the Seymour glyptodonts, which are here considered as Glyptothenum anzonae. The mandible of G. floridanum has been de­ scribed by Holmes and Simpson (1931), and more recently by Lundelius (1972), on the basis of a nearly complete left mandible, missing only Ny, the anterior extremity of the symphyseal ramus, the margin of the posterior angle, and the condyle (TMM 30967-1814, Figure 16). Mandibular frag­ ments and fragmentary isolated teeth referable to G. floridanum are known from several other Texas localities, including sites in Bee, Goliad, and Hunt counties (TMM 31034-30, Figure 15*; 31186-19; and USNM 6071, respectively); and one from Orange County. Florida (USNM 11318, Figure 17). The only known mandibular specimens of G. texanum are right and left jaw fragments with associated fragmentary Ny, Nj, N?-Ng (JWT 2330) and a partial left mandible with broken Ny-Ng (JWT 1715, Figure 15a) from the Cita Canyon Blanco beds. The mandible of G. cylindricum is unknown; however, the complete lower dentition is known FIGURE 15.—Left mandible: a, Glyptotherium texanum (JWT for this species (AMNH 15548). Comparing these 1715), medial side, reconstructed outline hypothetical; b, G. teeth with those of the other North American floridanum from Texas (TMM 31034-30). (Bar = 40 mm.) NUMBER 40 53

FICURE 16.—Left mandible oi Glyptotherium floridanum from Texas (TMM 30967-1814): a, lateral; b, medial. (Reconstructed portions crosshatched; bar = 10 cm.) 54 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

______*

».<*» ."*•

FIGURE 17.—Left mandible of Glyptotherium floridanum from Florida (USNM 1 1318): a, lateral; b, dorsal, slightly enlarged; c, medial. (Bar = 10 cm.) NUMBER 40 55 species, it appears that the mandible is probably The strut forming the postdental canal is bro­ not significantly different. ken on mandibles of G. arizonae (USNM 10536, The mandible of Glyptothenum is characterized Figure 13; and UMMP 38761, Figure 14). Its by its extremely massive build associated with presence is indicated, however, by the broken lengthening tooth row, open rooted teeth, and the nature of the bones in this region and by the unique masticatory apparatus involving the hor­ rearward continuation of the dental canal from izontal disposition of the masseteric musculature. the inner surface of the mandible. Thus, although The two sides are solidly ankylosed at the sym- not previously recognized, G. arizonae possesses a physis. The horizontal and ascending rami are dental canal and a postdental canal as well. The subequal in bulk. The condyle is situated well construction is similar in all respects to that of G. above the tooth row, fully twice the height of the floridanum. The dental canal in USNM 10536 is occlusal surfaces from the lower margin of the particularly well preserved and is relatively un­ horizontal ramus. The coronoid process and the damaged anterior to the broken strut. The open posterior angle of the jaw are well developed, and canal becomes covered below the seventh tooth. the symphysis bears a spoutlike construction pre­ It passes anteriorly into the horizontal ramus. sumably for the long tongue. Below the last tooth it issues several smaller fo­ Lundelius (1972) and Holmes and Simpson ramina, which pass inward to the alveolus. (1931) described the postdental canal for Boreos­ The dental and postdental canals are unknown tracon floridanus (— Glyptotherium floridanum) man­ for G. texanum owing to the fragmentary condition dibles. Neither Gidley (1926) nor Melton (1964) of available specimens. It can probably be safely mentioned the presence of the postdental canal assumed that this species is not very different in their descriptions of Glyptotherium arizonae and from G. arizonae or G. floridanum in this character. Glyptodon fredericensis (= G. arizonae), respectively. On a provisional basis, the position of the The postdental canal is indeed present in G. mental foramen may be used to differentiate anzonae, and it is not significantly different from between species of Glyptotherium. In G. texanum that of G. floridanum. The postdental canal is (WT 1715) the mental foramen is opposite the situated behind the last alveolus, extending midlobe of Ny. It is more anteriorly placed in G. "downward and inward and emerg(ing) on the arizonae (USNM 10536) and G. floridanum (USNM inner side of the mandible where it forms a 11318). In the Texas representative of the latter semicircular groove about 4 mm across and half species (TMM 30967-1814), the symphyseal re­ as deep" in the Texas G. floridanum (Lundelius, gion that includes the mental foramen, as indi­ 1972:36). Correct dimensions of 13 X 7 mm for cated by the Florida specimen, is missing. What the postdental canal were given by Holmes and appears to be a mental foramen in the Texas Simpson (1931) for the Florida representative. In specimen, located at the edge of the anterior both, the canal emerges on the inner side of the breakage, is probably one of the accessory foram­ jaw as the true dental canal, as observed by ina situated posterior to the mental foramen be­ Lundelius (1972). It branches below the last neath the second tooth. tooth. The postdental canal and the dental canal The upper margin of the rounded symphyseal are similar in the Florida and Texas representa­ region is level with the alveoli in G. arizonae. This tives of G. floridanum. As Lundelius (1972:37) ob­ profile is unknown for G. texanum; it was probably served, "The development of the post-dental similar to G. arizonae. In G. floridanum, the upper canal allows the formation of a buttress at the margin of the symphysis curves downward ante­ posterior end of the tooth row without crowding rior to Ny in a rounded anterior profile. This the nerves and blood vessels of the mandible condition is well indicated in USNM 11318, the between the posterior end of the tooth row and only G. floridanum mandible with a fully intact the posterior end of the jaw." symphyseal region. It is puzzling that the recon- 56 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY struction of TMM 30964-1814 includes a similar 934-37. Melton (1964), in establishing the syn­ profile, for the Florida mandible has never been onymy of Xenoglyptodon with the Seymour glyp­ reported. The reconstruction of the Seymour todonts, did not discuss this characteristic. In­ mandible, however, is essentially correct, assum­ deed, the mandible of TMM 934-37, from the ing the identity of the Texas and Florida speci­ Holloman fauna, does seem to have a more nearly mens. Hence, there is a fundamental difference vertically oriented anterior border than either the between G. anzonae, with a flat anterior profile, Seymour or Curtis Ranch mandibles. However, and G. floridanum, with a conspicuous, down- the broken condition of the Holloman specimen turned curvature in the symphyseal region of the and the absence of the upper half of the ascending mandible. ramus render comparison of this feature partially The upper surface of the symphysis is heavily inconclusive. The emergence of the anterior bor­ rugose, and it protrudes beyond the anterior pro­ der from the side of the tooth row is indeed file of the skull in a spoutlike construction. This vertical. Above the level of the tooth row, the surface probably provided attachment for strong anterior border is broken away and appears to lip musculature. In lateral profile the symphysis have been waterworn and abraded. As Melton in all three species tapers upward in a relatively (1964) determined, the Holloman and Seymour straight line. The symphyseal inclination is much glyptodont mandibular characters are nearly- steeper in G. arizonae than in G. floridanum. identical, on the basis of the dentition and shape For all three species (G. texanum, G. anzonae, G. of the horizontal ramus, but definite conclusions floridanum), the posterior extremity of the sym­ regarding the inclination of the ascending rami physis occurs in the plane of the anterior lobe of must await future discoveries from the Holloman Ny, and the ascending ramus emerges opposite locality. On the basis of the similarity in dentition Ng. The condyle is situated directly above the last and construction of the scutes, we agree with tooth in G. anzonae and G. floridanum, and probably Melton's (1964) contentions regarding the iden­ in G. texanum as well. tity of the Holloman glyptodont. Apparently the only useful taxonomic charac­ In all of the remaining specimens, the anterior ters of the mandible, other than the dentition and border of the ascending ramus is uniformly in­ the anterior profile of the symphysis, are the clined forward at an angle of approximately 60° shape of the lower margin of the horizontal ramus to the occlusal surface of the tooth row, with and the outline of the borders of the ascending increasing inclination above the tooth row, before ramus. becoming broadly rounded, leading upward to In G. arizonae the lower margin of the horizontal the angle of the coronoid process. The anterior- ramus is broadly rounded. Its maximum depth most extent of the anterior border occurs in the occurs in the vertical plane passing through N7 vertical plane of the rear lobe of Ny. and the coronoid process. From this position an­ The posterior margin of the ascending ramus teriorly and posteriorly the lower margin curves is more steeply inclined, forming an angle of upward and does not parallel the tooth row. Both approximately 40° with the occlusal surface of USNM 10536 and UMMP 38761 exhibit this the tooth row in G. floridanum, as discussed by shape, although the curvature is slightly more Lundelius (1972), and approximately 50° in G. pronounced in the former specimen. In marked anzonae. Thus the anterior and posterior margins contrast, both G. texanum (Figure 15) and G. flori­ of the ascending ramus are more nearly parallel danum (Figures \5b, 16, 17) possess a lower margin in G. arizonae than in G. floridanum. The slope of that is nearly parallel to the tooth row. the posterior margin is indeterminate for G. tex­ Meade (1953) based the diagnosis oi Xenoglyp­ anum. todon fredericensis in part on the inclination of the The lower extent of the posterior margin of the anterior border of the ascending ramus of TMM ascending ramus forms a greatly expanded pos- NUMBER 40 57

terior extremity of the mandible, which function­ Occlusal surfaces of individual teeth are hori­ ally replaces the underdeveloped angular process zontally flat (see schematic illustration in Figure as a primary region of muscle attachment. The 95); lower occlusal surfaces are directed upward posteriormost extremity occurs approximately at and somewhat medially; upper occlusal surfaces the level of the occlusal surface of the tooth row are directed downward and somewhat outward. in G. arizonae and G. floridanum. This construction Unfortunately, there are no complete mandible- seemingly distinguishes the North American glyp­ skull associations. It appears, however, that all todonts from South American representatives, for teeth of each series are in full contact when in which the posterior extremity is situated rather occlusion. The composite occlusal surface of the above the level of the tooth row, as observed by lower tooth row is flat along the anterior two- Holmes and Simpson (1931). thirds and distinctly concave along the posterior The inner surface of the lower half of the third. The composite occlusal surface of the upper posterior margin of the ascending ramus is tooth row is correspondingly convex posteriorly marked by distinct grooves for channeling the and flattened anteriorly. There are eight teeth in slips of the masseteric muscles. The outer surface both the upper and lower tooth rows. As in other is thickened into a massive boss for muscle at­ edentates, it is impossible to determine the dental tachment. formula. All teeth are distinctly molariform with The condyles are situated well above the tooth a basic trilobate, prismatic construction. The two row, reaching a height level with the coronoid anteriorieeth, in both the upper and lower series, process. The coronoid fossa is semicircular and are smaller and less distinctly molariform than relatively shallow. The condyles are transversely the more posterior teeth, but their construction is elongate and convex dorsoventrally. They are otherwise similar, and they doubtless functioned directed forward and slightly upward, occluding similarly, as the anterior members of the grinding loosely with the glenoid articular facets of the mill. The teeth are confined to the parallel por­ skull. There are no confining flanges on the con- tions of the palate and mandible, respectively; dylar process. The transverse elongation of the there are no teeth on the rounded symphyseal condyles indicates a relative freedom of transverse process, and there is no indication of upper teeth motion in the occlusal action, and the weak con­ in the standard incisor position. Thus, glypto­ struction and anterior direction of the articulation donts lack incisiform and caniniform teeth, and indicate a general lack of vertical stresses at the they therefore lacked any dental participation in temporomandibular joint. This construction is food procurement. more fully treated in the discussion of the func­ There is no evidence of the eruption sequences, tional anatomy of the masticatory apparatus. nor is there any indication that glyptodonts pos­ Complete descriptions of the upper and lower sessed separate milk and permanent series. There dentitions are provided in the following sections. are available for study three teeth from young DENTITION.—The teeth of all North American individuals (F:AM 23515 and UF/FSM 10623) species show a close correspondence in occlusal in which the apex is but slightly worn. From the patterns, differing in minor details and in size. little worn occlusal surfaces on these teeth, the Complete upper dentitions (Figure 18) are known alveolar portion rapidly increases in diameter and for G. texanum and G. anzonae; a nearly complete the trilobate construction appears only a short upper series, lacking only N~, is known for G. distance from the apex. The bases on all three are clylindricum; and isolated upper teeth are known broken away. Apparently these teeth simply in­ for both the Texas and Florida representatives of creased in diameter toward the pulp cavity, even­ G. floridanum. A complete lower series is known tually attaining full adult proportions through only for G. arizonae, although nearly complete growth and attrition. There is no indication that series are known for the other species (Figure 19). these are milk teeth, for they are but miniature 58 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY versions of adult teeth. Flower (1868) established tory motion, with secondary lateral motion (Fig­ the existence of tooth replacement in armadillos. ure 95). This is consistent with the nature of Hoffstetter (1958) suggested polyodonty as a pos­ construction of the mandible and its glenoid ar­ sible explanation for edentate departures from ticulation. normal mammalian dental formulae. Poly­ Development of the three-lobed construction of odonty, in the way of suppression of replacement each tooth lengthens the tooth row to the extent teeth and multiplication of existing tooth germs, that its length is fully three-quarters of the total is a convenient, though undemonstrable, expla­ length of the skull. This extreme proportion, the nation for the existence of near perfect homo- deep and massive construction of the mandible, donty in glyptodonts. and the brachycephaly of the basicranium distin­ Except for the reduced anterior two teeth in guish the glyptodont skull as unmistakably both the upper and lower series, individual teeth unique. Perhaps the closest comparison of glyp­ are three-lobed. The lobate construction is formed todont teeth with those of modern mammals is by two pairs of opposing longitudinal grooves on with certain rodents, notably microtines and cap- each side of the tooth; anterior teeth are progres­ ybaras. Thus the trilobate construction, the cor­ sively less deeply grooved. The teeth are all open responding elongation of the tooth row, and the rooted, and the pulp cavities remained unclosed elaboration of the osteodentine tracts can be throughout life. Except for alveolar thinning of viewed as an alternative to enamel. the outer osteodentine ridge, the prismatic teeth The following descriptions of individual teeth of adults are uniform in thickness and construc­ utilize Hoffstetter's (1958) means of designation: tion from the occlusal facet to the pulp cavity. N1 is the anteriormost upper tooth, N~ the next, There is no substantial increase in circumference to Na, the last upper tooth. Similarly, Ny is the toward the pulp cavity in adult teeth. The outer first lower tooth, Ng the second, to Ng. shapes of the teeth are generally maintained from Lundelius (1972) questioned Brown's (1912) the occlusal surface downward, although slight designations for the positions of the isolated teeth variation can be perceived with advancing age, associated with the type specimen oi Brachyostracon especially with regard to the posterior surface of cylindricus (= G. cylindricum). Lundelius suggested uppers. Because of the uniformity of individual that the teeth Brown designated as Ny and N3 are teeth and the lack of pulp cavity closure, it is more like N1 and N2 and that Brown's N2, N3, N4 impossible to determine relative ages of adult are more like Ns, Ns, Ny in other glyptodonts. individuals on the basis of dentition alone. Lundelius further suggested that if his contention In typical edentate fashion, glyptodont teeth regarding the positions of the teeth were correct, entirely lack enamel. Compensatory elaboration then "the lower dentition oi Brachyostracon cylindri­ of the osteodentine ridges provides an effective cus is very much like that oi Glyptotherium " In grinding mill, allowing for feeding habits associ­ actuality, Brown's designations were more erro­ ated with the ever-growing teeth. A thin outer neous than simple transposition of uppers for osteodentine rim completely encloses the inner lowers and vice versa, as indicated in the follow­ dentine-like core. Additional osteodentine tracts ing discussion and the accompanying charts. are situated in the center of each tooth, with Proper assignments for the upper dentition of branches penetrating into the lobes from a main G. cylindricum and Brown's original designations stem. Elaboration of the interior osteodentine are as follows (* figured in Brown, 1912:170, fig. tracts in North American glyptodonts is limited 1, upper series; ** same, lower series): to minor secondary branching. Brown's (1912) The outer and interior osteodentine ridges Actual position designation stand up in relief above the dentine core. Attrition N1 (left) Ny (right)** facets indicate a principal fore-and-aft mastica­ N2 (left) Nj (right)** NUMBER 40 59

3 N (left) N4 (left)* of G. cylindricum are based on direct comparison a N (right) N4- (right) with teeth of known position in the species G. (N- not present in collection) anzonae, G. texanum, and G. floridanum. Melton's N§ (left) same* § (1964) assignments of the lower teeth of UMMP N (right) N6 (right) Ns (left) same* 38761 (G. arizonae) were erroneous. The tooth a N (right) N6 (right) Melton considered as Ng was an upper, from the z N (left) same* rear of the associated skull, which was cemented Na (left) same* to the rear border of the posterior alveolus of the 8 N (right) same* mandible. This tooth and the cementing matrix Thus, as shown in this chart for the upper were not removed from the lower jaw until the dentition, Lundelius was correct in suggesting present study revealed this situation. Thus, the that Brown was mistaken in the assignment of Ny occlusal surface Melton identified and figured as - and N2 (right) and that these were instead the Ng was actually the unclosed alveolar end of N , first two teeth of the left upper series, respectively. and the teeth he identified as N5, Ng, and N7 were The tooth Brown identified and figured as N4 actually Ng, N7, and Ng, respectively, from the (left) is actually Na (left). Similarly, the tooth right mandible. Only the posterior lobe of N5 is identified (and unfigured) in the collection as N4 present on the right mandible; Ng is broken and (right) is actually N3 (right). There are no N4's its nature is indeterminate on the left mandible. represented in the collection; i.e., the teeth Brown Measurements provided in Table 8 are composite identified as N4 are actually N3. Brown's figured for the lower series, from the crowns of the left designations for N5-N8 (left) were correct, al­ N2-N4, and Ng-Ng for the right side. Ny is not though his designated labels for N& and N6 (right), present, nor is N5 adequately preserved for mea­ in the collection and unfigured, were apparently surement. a transposition from the actual order. Melton's (1964) assignments for the upper se­ Proper assignments for the lower dentition of ries of UMMP 38761 were correct. His illustra­ G. cylindrium and Brown's original designations tions and discussions of N2 and N3 were incorrect, are as follows (* figured in Brown, 1912:170, fig­ however. As discussed in more detail below, it ure 1, upper series; ** same, lower series): appears that N- is probably trilobed (similar to

Brown's (1912) those of F:AM 95737 and USNM 6071, and Actual f IOSI lion designation distinct from that of AMNH 15548), rather than bilobed, as Melton suggested. Similarly, N- is less (Ni not present in collection) z symmetrical than Melton indicated, having rela­ N5 (right) N (left)* 3 N5 (right) N (left)* tively underdeveloped buccal portions of the an­ (N< not present in collection) terior and posterior lobes, respectively. Moreover, NS (left) same** the lingual portion of the anterior lobe is more NS (left) same** obliquely oriented, rather than perpendicular to Ng (right) same N? (left) same** the long axis of the tooth row. 3 N? (right) same N~ and N therefore exhibit less departure from Ng (left) same** other North American glyptodonts than is indi­ Ng (right) same cated in Melton's descriptions. It must be stated As shown in this chart for the lower dentition, that our reinterpretation would not have been u_. ^. .„>„ J~~; „„ „r i v2 .,„, ! NT2 —„ „„*.,„U,, possible prior to the availability of the excellent Ng and Ng, respectively. Ny and N? are not rep­ comparative material used in the present study. 1 resented in the collection, and N5, Ng, N7, and Ng UPPER DENTITION (Figure 18, Tables 3-6).—N were correctly identified and figured by Brown. is submolariform, having a simple peglike, pris­ The present assignments of these isolated teeth matic ovoid construction. N- is intermediate be- 60 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

1 "x

G. floridanum G. cylindricum

FIGURE 18.—Upper teeth of North American glyptodonts, occlusal surfaces, left side (asterisk = reversed pattern, taken from corresponding tooth of right side): a, UMMP 34826: N1, N--a; UMMP 38761: Na, N4-; b, F:AM 95737; r, UF/FGS 6643; d, USNM 6071; e. AMNH 15548; relative sizes of individual teeth are accurate, but lengths of tooih rows are adjusted for comparison. (Bar = 5 cm.) tween the submolariform N~ and the nearly com­ the alveolar curvature, and with progressive loss pletely molariform N3. N4-N8 are completely mo­ of the outward curvature and outward orienta­ lariform, having rather symmetrical development tion of the wear facet. The outward curvature of of the inner and outer portions of each lobe. the alveolar portion is lost in NJ, which is nearly Occlusal surfaces of the anterior teeth are parallel straight. The occlusal surface is inclined forward to the plane of the palate. Posteriorly this ori­ with respect to the cross-sectional plane in N3-N". entation is relinquished for progressively more It is perpendicular in NlJ and inclined rearward outward-facing occlusal surfaces. in N1 and N8. The latter two teeth, especially N8, From the occlusal surface, the alveolar portion possess pronounced rearward curvature of the of N1 curves outward and slightly rearward to­ alveolar portion. The height of the teeth (pulp ward the pulp cavity. The occlusal surface is cavity to occlusal surface) increases gradually to inclined outward with respect to the cross-sec­ Nb and decreases dramatically from Nfi to N8; the tional plane of the tooth. N2 and N3 are similarly height of N8 is only about 70 percent of the height constructed, lacking the posterior component of of Nu (approximately 101 mm and 70 mm, re- NUMBER 40 61 spectively) for AMNH 15548. N8 is approxi­ the same configuration as in the other North mately the same length as N1 American species. The anterior lobe appears to The first two uppers, N1 and N2, are the most be slightly more pointed on the medial apex, and distinctive teeth of G. cylindricum. N1 is irregularly the lingual (inner) portion of the lobe is weakly ovoid, elongated in the anteroposterior direction. developed. This is the most distinctive tooth of G. The external face of the tooth is convex. The rear cylindricum and it serves to distinguish this species. half of the internal face is also convex, and the N2 of G. texanum and G. floridanum is trilobate, anterior half is recurved in a concave outline, with a nearly symmetrical middle lobe due to the producing an overall weakly sigmoid occlusal presence of two well-developed grooves on the pattern. This contrasts with the more pronounced internal face, producing a more typical, trilobate convexity of the internal face in G. anzonae and G. occlusal pattern. The external sides of the anterior Otexanum. In the latter two species, N1 is more and posterior lobes are underdeveloped relative nearly oval, with a less pronounced anterior con­ to the internal sides. Although nearly as long cavity on the internal face. The osteodentine anteroposteriorly as the succeeding teeth, N- is pattern in all is weakly sigmoid, with a pair of transversely smaller; it is intermediate in size short branches issuing from the center. In G. between N1 and N3. texanum and G. cylindricum, the osteodentine tract Because of the broken condition of both N~'s of issues short spinose branches along its main UMMP 34826, the exact configuration of this branch. The nature of the osteodentine tract in tooth in G. anzonae cannot be determined with G. anzonae is indeterminate; presumably it pos­ certainty. The posterolateral face is flattened, and 1 sessed similar branching. N is unknown for G. there is a distinct posterior apex. A portion of floridanum. right Na below the missing crown indicates that The N- occlusal pattern of G. cylindricum ex­ there was a constriction both externally and in­ hibits the most distinctive differences from known ternally, producing a posterior lobe. This portion N-'s of other species. The tooth here assigned as of the tooth more closely resembles that of G. N2 for G. cylindricum (Brown's Ns) appears to be texanum. Although Melton (1964) asserted that N2 properly designated as an upper tooth because of this specimen possessed anterior and posterior the occlusal pattern, the development of the lobes, lobes, and lacked a center lobe, the construction and the curvature of the tooth are all inconsistent anterior to the posterior lobe cannot be deter­ with any tooth of the lower series. Furthermore, mined with confidence. Because of the closer the assignments of the distinctive first two lowers similarity of the remaining teeth with G. texanum, are made with confidence, and these positions it is likely that N~ of G. arizonae is trilobate rather represent the only possible alternatives that the than either bilobate, as Melton suggested, or tooth here designated as N2 could possibly oc­ irregular as in G. cylindricum. The assumption of a cupy. This tooth is identified as the second in the trilobate construction should not be accepted as upper series because of the intermediate devel­ fact, however, since in G. floridanum, which ex­ opment of the lobes with respect to N1 and N~, hibits occlusal patterns otherwise almost identical and also because of the intermediate, and closely with G. cylindricum, N- is more similar to that of fitting, curvature and asymmetry of the alveolar G. texanum and quite distinct from N- of G. cylin­ portion of the tooth. dricum. Hence, resemblances of anterior teeth be­ More specifically, this N2 exhibits no lingual tween species are not necessarily consistent with (inner) development of the middle lobe, nor is resemblances of posterior teeth. there any indication of the development of this Osteodentine branches penetrate both the mid­ lobe with age. Instead, in this position, the occlu­ dle and posterior lobes with perpendicular rami sal pattern is only slightly convex. The outer in G. texanum and G. floridanum. A perpendicular portion of the middle lobe displays approximately ramus extends only externally into the medioex- 62 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY ternal lobe in G. cylindricum; this branch does not medial. The transverse axis of the anterior lobe is cross the main tract toward the internal face of somewhat obliquely oriented with respect to the the tooth. The anterior extremity of the osteoden­ middle and posterior lobes and to the long axis of tine tract curves medially into the anterointernal the tooth row. The middle lobe is the most widely lobe, giving off a short branch toward the under­ expanded. It extends farther laterally, but not as developed anteroexternal lobe as an unequal bi­ far medially in comparison with the anterior and furcation. There is no anterior bifurcation in G. posterior lobes. The posterointernal face of the cylindricum. There is weak spinose elaboration of posterior lobe is broadly convex. The posteroex- the osteodentine tract in G. cylindricum, G. texanum, ternal face bears a shallow sulcus, producing a and G. floridanum. Osteodentine patterns cannot weak concavity in the occlusal pattern at this be determined for G. anzonae. position. The characteristic asymmetry of this N is indistinguishable among the three species tooth is produced primarily by the oblique dis­ for which it is known. The assignment of N3 from position of the anterior lobe and the unequal the isolated teeth of G. cylindricum is based on transverse development of the lobes with respect direct comparison with this tooth in G. texanum to each other. and G. arizonae, to both of which it is identical. N4 is the first tooth to possess equal external There is an obvious disparity in the teeth in the and internal development of the lobes. The trans­ N- position between the two Seymour G. arizonae verse axes of the anterior and posterior lobes are skulls. The fragments of teeth replaced into the laterally convergent. The anterior face is slightly position of N3 in UMMP 34826 appear not to be convex to straight. The middle lobe possesses N~, nor does the shape of right N match that of equal development of both sides, thus distinguish­ left N3. Actual N3 is present and unbroken in ing this tooth from N- The rear lobe is posteroin- UMMP 38761, and it is from this specimen that ternally convex. There is a characteristic shallow the N3 characteristics of G. anzonae are derived. sulcus on the posteroexternal face, rendering a The tooth figured and described by Melton more pointed external apex for the posterior lobe. (1964) as N was the tooth for which the crown Besides the oblique orientation of the anterior was improperly placed in this position in UMMP and posterior lobes with respect to each other and 34826. This tooth is probably N4 or N5. the equal development of the middle lobe, N4 can N is present in the skulls of G. texanum. It is not be distinguished from the adjacent N by its lack known for G. floridanum; however, the alveolus for of outward curvature in the alveolar portion. N4 N3 in the skull of USNM 6071 indicates close is virtually identical in G. texanum, G. arizonae, and correspondence of the shape of this tooth with G. floridanum. It is notable that the N4 of UF/FGS those of the other species, especially with regard 6643 (G. floridanum) is identical to that of the to the shape and disposition of the anterior lobe Texas representative, USNM 6071. N4 is un­ and the relative lateral extents of the lobes. known for G. cylindricum; presumably it was simi­ The anterior border of the alveolus of N3 is lar. Osteodentine rami extend into each lobe, and situated opposite the anterior border of the root there is an indistinct branching of the ramus in of the zygomatic process of the maxillary. This is the posterior lobe. the first fully molariform, trilobate tooth of the The posterior teeth of the upper series, N&-N8, upper series. Each lobe is distinctly wider trans­ have nearly identical patterns, and they are iden­ versely than the poorly developed lobes of N2 and tical in size, except for the only known N8 of G. somewhat less expanded than those of N4. N3 is cylindricum (AMNH 15548), which is distinctly characterized by its distinct asymmetry in com­ smaller than the preceding teeth. This disparity parison with the more posterior teeth. The ante­ is probably due to the relatively late eruption of rior lobe is broadly rounded and convex. Its N8, however, as indicated by a full millimeter external side is relatively less expanded than the increase in length from the occlusal surface (19 NUMBER 40 63

mm) to the pulp cavity (20 mm) over a relatively posterior face and in having relatively blunt api­ short, tooth height diameter. Thus, it is likely ces of the lobes. This specimen is referable to G. that N- of G. cylindricum in the full adult condition texanum on the basis of the more nearly complete approximated the size of the preceding teeth. lower dentition. a Accordingly, this noted size difference in N of G. LOWER DENTITION (Figure 19, Tables 7-10).— cylindricum is of little diagnostic valve. The first lower tooth, though distinctly less mo­ Isolated posterior upper teeth are difficult to lariform than the succeeding teeth, is nevertheless position correctly. It is possible, however, to place rather more elaborate than the corresponding a complete series in proper order on the basis of upper tooth, at least in G. texanum and G. arizonae. the curvature of the alveolar portions that grade The succeeding teeth are increasingly more mo­ uniformly from the outward curvature, without lariform and exhibit increasing definition of the any rearward curvature in N5, to strong rearward trilobate construction. Ny-Ng are fully molari­ curvature, without any outward curvature in Na, form and completely trilobate; the last five teeth and on the orientation of the occlusal surfaces are so similar that occlusal patterns alone are with respect to the cross-sectional plane of the insufficient for determining proper position in the teeth, which increase in posterior inclination rear­ series. The dispositions of the occlusal surfaces ward from the nearly perpendicular orientation and the direction of alveolar curvature exhibit a of N~; N- occlusal surface is directed somewhat development complementary to the upper teeth, laterally. so that these features provide reasonably confi­ The occlusal patterns of N5-N8 exhibit a uni­ dent determination of proper positions of isolated form development of the lobes. The anterior bor­ teeth. ders are broadly convex to nearly straight. G. The alveolar portion of Ny curves strongly in­ arizonae and G. texanum have relatively flattened ward and rearward, and the occlusal surface is anterior borders, in contrast to the more rounded directed inward with respect to the horizontal anterior borders of G. cylindricum. A similar dis­ cross section of the tooth. Alveolar portions of N2- tinction is evident for the posterior lobes of the Ny are similarly constructed, but with decreasing lower dentitions of these species. The posteroin­ curvature, and the inward direction of the occlu­ ternal angle is rounded, and the posteroexternal sal surfaces diminishes rearward. There is no angle is pointed, owing to the presence of a inward curvature in the alveolar portion of Ng; it distinct sulcus on the posteroexternal face of each is nearly straight. Its occlusal surface is directed tooth. The laterally convergent transverse axes of rearward with respect to the cross section. Alveo­ the anterior and posterior lobes are slightly offset lar portions of Ng-Ng exhibit increasing rearward from a perpendicular orientation with the long curvature. The occlusal surface of N7 is directed axis of the tooth row. Osteodentine branches rearward, but not as distinctly as Ng. The occlusal penetrate into each lobe, and there is minor surface of Ng is perpendicular to the cross section. secondary branching. The osteodentine ridges are The occlusal pattern and the mandibular po­ marked by short spinose accessory ridges to vary­ sition of Ny are diagnostic. In G. texanum (Figure s 8 ing degrees. N to N are unknown for G. flori­ 19) Ny is elongate and irregularly ovoid, with a danum; presumably these teeth more closely re­ convex inner face and a pair of shallow grooves semble those of G. cylindricum than either G. tex­ on the outer face suggesting a tripartite division. anum or G. arizonae, as discussed above. On the posteroexternal face there is a shallow A single upper molar, probably N- or one concavity, producing an elongate posterior re­ adjacent, is known from the Cita Canyon Blanco gion. In the fragmented specimen JWT 1715, the Beds (JWT 1837). This tooth most closely resem­ only G. texanum mandible with anterior teeth, it bles that of G. texanum from Arizona in having a appears that Ny is situated near the beginning of relatively shallow posterointernal sulcus on the the curvature of the symphyseal ramus. Accord- 64 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

f g

G. floridanum G. cylindricum

FIGURE 19.—Lower teeth of North American glyptodonts, occlusal surfaces, left side (asterisk = reversed pattern, taken from corresponding toolh of right side): a, USNM 10536; b, UMMP 38761; c, JWT 2330; d, USNM 11318; e, UF/FGS 6643;/, TMM 30967-1814; £, USNM 6071; h, AMNH 11548; relative sizes of individual teeth are accurate, but lengths of tooth rows are adjusted for comparison. (Bar = 5 cm.)

ingly, the long axis of the tooth is obliquely developed, without medial expansion and only oriented with respect to the anteroposterior axis slight lateral expansion. The outer portion of the of the tooth row. middle lobe is well defined; the inner portion is Ny is unknown for G. floridanum, but mandible less well developed, but is nevertheless distinct. USNM 11318 (Figure 17) includes the alveolus The posterior lobe is obliquely oriented, extend­ for this tooth, indicating a considerable reduction ing posteroexternally into an elongate lobe. The in size. This tooth was apparently an ovoid, peg­ posterointernal apex of the posterior lobe is poorly like tooth, with a convex inner surface and two developed, and there is a shallow sulcus on the weak grooves on the outer surface, representing anteroexternal face. The alveolus for Ny is situ­ the vestige of the former three-parted division. ated posterior to the symphyseal curvature, as in The alveolus is situated well behind the symphy­ G. floridanum. seal curvature. In G. anzonae, the osteodentine core issues rami Quite in contrast, Ny of G. arizonae is strikingly in both directions at the middle lobe; it bifurcates molariform. The occlusal surface is more nearly posteriorly into a short internal and a long exter­ trilobate. The anterior lobe is relatively under­ nal branch; anteriorly the core bifurcates into NUMBER 40 65 two short rami in the anterior lobe. Secondary (these represent different species, but the closer spinose branching, though poorly developed, is resemblance of the other teeth of G. cylindricum distinguishable. The nature of the osteodentine with G. floridanum rather than G. anzonae supports core is indeterminate in the available specimens this contention); the tooth identified as Ng for of G. texanum. AMNH 15548, because of the alveolar curvature Ng is apparently a highly variable tooth. It is and the orientation of the occlusal facet, certainly distinguishable from Ny and Ng by its interme­ belongs in the position immediately anterior to diate molariform construction and alveolar cur­ the tooth identified as Ng. By indirect compari­ vature. The anterior and posterior lobes are well son, the tooth identified with confidence as Ng for defined relative to Ny, but their transverse devel­ G. cylindricum is identical to one of the teeth of the opment is less than in Ng, especially with respect Catalina Gardens specimen of G. floridanum, and to the inner apices of the anterior and posterior this tooth is therefore assigned as Ng. Thus, by lobes. reason of inference, Ng of G. cylindricum is identical In G. texanum the anterointernal surface of Ng to Ng of one specimen of G. floridanum. Both of is straight, nearly like that of Ny of G. arizonae. these seem to resemble closely Ng of the Ingleside The apices of the middle lobe in G. texanum are glyptodont (TMM 30967-1814), and all three of well defined, and the posterior lobe is well devel­ these differ considerably from Ng of the other oped, with prolongation of the posteroexternal Florida representative, USNM 11318, despite the lobe and corresponding obliquity in its orienta­ close similarity of Ng and the succeeding teeth in tion. The posterior surface is convex, becoming all four. (It is also notable that USNM 11318 somewhat more flattened with age. most closely resembles known G. floridanum man­ In G. arizonae, Ng is distinctly more molariform dibles and teeth on a quantitative basis, indicat­ in the elongation of the lobes and in the pro­ ing that if this specimen represents a separate nounced oblique disposition of the posterior lobe, species, it is not one of the others and is most which distinguishes it from the other species. Ng closely related to G. floridanum; see Tables 2 and of the Seymour glyptodont is identical to that of 10.) the Curtis Ranch glyptodont, USNM 10536 (type Therefore, Ng of G. cylindricum and of the Ca­ specimen), supporting the asserted identity of the talina Gardens (Florida) and Ingleside (Texas) glyptodonts from these two localities as G. an­ representatives bears a distinct trilobate construc­ zonae. tion, and, compared with the teeth of G. anzonae A broken Ng in the mandible of TMM 30967- and G. texanum, it more closely resembles Ny. Thus 1814 (Ingleside, Texas, respresentative) and a Ng is less distinctly molariform in these specimens complete Ng in USNM 11318 (Orange County, than in either G. arizonae or G. texanum. The an­ Florida, specimen), both representing G. flori­ terointernal border is straight, without medial danum, are strikingly different. An isolated Ng development of the anterior lobe. The lateral from the Catalina Gardens, Florida, locality (UF/ portion of the anterior lobe is slightly expanded, FGS 6643) more closely resembles the Texas spec­ while both portions of the middle lobe are well imen than the one from nearby Orange County. expanded. The rear lobe is expanded posterolat- The tooth here identified as Ng for G. cylindricum erally, with only slight medial development. The is identical to the one from Catalina Gardens. posterior lobe is obliquely oriented with respect Although it is possible that the isolated teeth here to the long axis of the tooth. This description is regarded as Ng for the Catalina Gardens and G. also accurate for the G. cylindricum and Catalina cylindricum specimens are actually Ny (thus resem­ Gardens specimens, which are identical. It is ac­ bling G. arizonae), the present assignment seems curate for the Ingleside specimen, at least for the most likely because Ng of USNM 11318 is iden­ anterior and middle lobes. The Ingleside Ng ap­ tical to the tooth assigned as Ng of AMNH 15548 pears to differ from Ng of USNM 11318 in much 66 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY the same fashion as AMNH 15548 (G. cylindricum) more closely approximates Ng of the other species. and UF/FGS 6643 (Catalina Gardens). The anteroexternal and posterointernal lobes are In USNM 11318, here tentatively identified as poorly developed. The disposition of the posterior G. floridanum, Ng is even less molariform. Its con­ lobe is somewhat oblique, rather intermediate struction resembles that of Ng in G. cylindricum in between the markedly oblique orientation in G. the reduction of the external portion of the middle anzonae and the nearly perpendicular orientation lobe. (The assignments for the upper teeth in G. in G. floridanum. cylindricum are made with confidence so that this Ng of G. arizonae and G. floridanum/G. cylindricum is a case of a parallel trend in reduction of anterior more closely resemble each other than either teeth, rather than suggestive of an incorrect as­ resembles that of G. texanum. In G. anzonae the signment for the tooth here identified as N2 for G. transverse development of the anterior lobe cylindricum.) Identification of USNM 11318 as G. nearly equals that of the middle and posterior floridanum is based on the posterior teeth. Thus Ng lobes. The anterior lobe obtains a rather squared in USNM 11318 represents a variation within the outline produced by weak concavities on the an- species and is not sufficient evidence alone to teromedial and anterolateral surfaces, as shown warrant the establishment of a new species. Only in USNM 10536. The middle lobe is set almost future discoveries will resolve this problematical perpendicular to the tooth row, and the convex situation. This tooth bears a sigmoid occlusal posterior lobe is markedly oblique, with prolon­ pattern, formed by prolongation, without expan­ gation of the posteroexternal lobe. The posteroin­ sion, of the anterointernal and posteroexternal ternal apex is sharp. The anteromedial and an­ lobes. The posterointernal face is a smoothly con­ terolateral surfaces are more rounded in UMMP vex, tapering surface from the weak expansion of 38761; this difference is attributable to geo­ the middle lobe; i.e., there is no internal devel­ graphic or temporal variation of the species. The opment of the posterior lobe. The outer portion relative development and oblique disposition of of the middle lobe is reduced to a faintly convex the lobes of Ng in UMMP 38761 are identical to surface connecting the weak lateral expansions of those of the Curtis Ranch representative. the anterior and posterior lobes. Ng of the Ingleside representative of G. flori­ The osteodentine core in all Ng's represented danum (TMM 30967-1814) is identical to the issues transverse rami into the middle and poste­ teeth assigned as Ng for G. cylindricum and the rior lobes and anteriorly the core issues a branch Orange County (Florida) representative of G. into the anterolateral lobe (except in USNM floridanum (USNM 11318). The anterior lobe is 11318, which has no anterolateral lobe). There is less distinctly squared in G. floridanum and G. minor secondary spinose branching at least in G. cylindricum. The anteroexternal face is convex, cylindricum, G. arizonae, and the problematical G. rather than concave. The transverse axes of all floridanum specimen—and probably in the others three lobes are nearly parallel, and their orien­ as well. tation is almost perpendicular to the tooth row. Thus, in conclusion for this tooth, Ng is appar­ The posterior face is broadly convex, the pos­ ently the most highly variable tooth in the lower teroexternal lobe is only weakly prolonged, and series. It is distinctly trilobate and molariform in the posterointernal apex is not sharply pointed. G. anzonae, somewhat less so in G. texanum, and Osteodentine rami extend transversely into variously developed in G. floridanum and G. cylin­ each lobe from the central core. Secondary spi­ dricum. nose branching appears to be limited to the an­ Ng occlusal pattern is distinctive for G. texanum terior lobe. and G. arizonae, and it is identical in G. cylindricum Ny is the first completely molariform tooth in and G. floridanum. In G. texanum, Ng is distinctly the lower series of G. anzonae and G. floridanum, less molariform than Ny; indeed, its development possessing a transverse development of the ante- NUMBER 40 67

rior lobe equal to that of the middle and posterior occlusal surfaces with the cross-sectional plane of lobes. Ny of G. texanum differs considerably in its the tooth: relatively steeply inclined, facing rear­ relatively shorter transverse development of the ward in Ng, with diminishing inclination rear­ anterior lobe. In fact, Ny of G. texanum (WT 1715) ward; N? is approximately perpendicular. These possesses an occlusal pattern that closely resem­ teeth are sufficiently different between species to bles Ng of G. floridanum, although there is a deep warrant separate descriptions, but the similarities sulcus on the anterointernal face. In G. texanum are by far greater than the distinctions. the lobes are nearly perpendicular to the long axis In all four species the lobes are equally devel­ of the tooth row, and the anterior lobe is relatively oped in the transverse dimension, and the trans­ underdeveloped. Ny of G. texanum differs from Ng verse axes are more or less perpendicular to the of G. floridanum in possessing a weakly concave long axis of the tooth row. There is a distinctive posterior face. flattening of the anterointernal face, uniformly Ny of G. arizonae and G. floridanum is fully tri­ developed on all three teeth for each species. This lobed. As in Ng, the anteromedial and anterolat­ flattening produces an anterior apex in the occlu­ eral surfaces are concave in USNM 10536, and sal pattern. The anteromedial surfaces of N5-N7 they are convex in UMMP 38761. In the latter are concave in G. arizonae, flattened (Ng, N?) to specimen also, the transverse extent of the ante­ weakly concave (Ng) in G. cylindricum, concave rior lobe is rather less than that of the middle and (Ng, Ng) to flat (N?) in G. floridanum, and all posterior lobes, thus contrasting with USNM flattened in G. texanum. A similar variability oc­ 10536. In both specimens of G. anzonae, however, curs for the anterolateral surfaces: all concave in the transverse axes of the lobes are obliquely G. arizonae and G. texanum, all straight in G. cylin­ oriented. The posterior face of USNM 10536 is dricum, and convex to straight in G. floridanum. concave; it is broadly convex in UMMP 38761. The development of the middle lobes of N5-N7 These differences between the Curtis Ranch and is identical in all four species. The posterior lobes Seymour representatives are likely attributable to exhibit a variability comparable to that of the variation. Ny's of the specimens from these two anterior lobes. The posterior lobes are concave localities more closely resemble each other than and slightly oblique (N5, Ng) to perpendicular either resembles Ny of G. floridanum. (N7) to the long axis in G. arizonae and G. texanum, In G. floridanum, Ny is also fully trilobate, al­ flat (Ns, Ng) to weakly convex (N7) in G. cylindri­ though the anterior lobe is slightly shorter than cum, convex in the Texas and Orange County, the middle and posterior lobes, as in the Seymour Florida, G. floridanum (TMM 30967-1814 and G. arizonae. The lobes are more nearly perpendic­ USNM 11318, respectively), and flattened (Ng, ular to the long axis of the tooth row, and the N?) in the Florida Catalina Gardens G. floridanum posterior face is distinctly convex. There is a wide (Ns not known). In G. cylindricum and G. floridanum range of variation in the development of the the orientation of the lobes is more nearly perpen­ sulcus on the anterointernal face—from deep in dicular to the long axis of the tooth row. USNM 6071, to shallow in TMM 30967-1814, Osteodentine rami extend into each lobe. to entirely absent in USNM 11318. There is no There is limited spinose secondary branching corresponding sulcus in either G. texanum or G. along the transverse rami, especially in the ante­ anzonae. In all three species osteodentine rami rior and posterior lobes. extend into each lobe. Secondary spinose branch­ Ng can be distinguished from the more anterior ing is apparently limited to the anterior lobe. teeth by its perpendicular occlusal surface with Ny is unknown for G. cylindricum. Within each respect to the cross section of the tooth, by the species the occlusal patterns of Ns, Ng, and N? are posteriorly diminishing medial extent of the lobes, almost identical. These teeth can be serially dis­ and by the distinctive concavity on the posterior tinguished in isolation by the orientation of the face of the tooth. 68 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

The anteromedial and anterolateral faces are sembles Ng of G. cylindricum. If this is indeed the concave in G. arizonae, producing an occlusal pat­ correct position, the anterior face is decidedly tern similar to the anterior teeth for this lobe, more concave in G. mexicanum; otherwise Ng is with a relatively pointed anterior apex. These virtually identical in these two species. The figure faces are convex in the other three species, and reveals that there is distinct but minimal second­ the anterior apex is rather more blunt in the ary spinose branching of the osteodentine core. occlusal pattern. The anterior tooth as figured by Cuataparo In G. arizonae and G. floridanum the middle and and Ramirez is unlike any teeth of the other posterior lobes are progressively shorter in the species. If indeed it is an Ny or N , G. mexicanum transverse dimension. In G. cylindricum the three possesses the most molariform anterior teeth of lobes attain nearly equal development, and there the genus. The tooth as figured is anteroposte­ is no rearward taper. The symmetrical middle riorly elongate and straight, rather than sigmoid lobes are otherwise similar in these three species. or peglike as in other species, and it is distinctly (The posterior two-thirds of Ng for G. texanum is trilobate. The three lobes are not greatly ex­ unknown.) panded transversely, and there is an unusual apex The posterior face exhibits a deep concavity in at either extremity. If the drawing is accurate, G. anzonae and G. cylindricum. In the latter species this tooth is sufficiently distinctive to indicate the the posterior lobe is perpendicular to the long validity of the species. Brown (1912) considered axis, whereas in G. arizonae it is somewhat oblique. the highly molariform teeth of G. mexicanum as The posterior surface of Ng of G. floridanum pos­ one of the principal features distinguishing this sesses a weak concavity; in occlusal outline this species, and we concur. concavity is barely evident. The posterior lobe is PRESACRAL AXIAL SKELETON slightly oblique, as in G. arizonae. In all three species the posteroexternal apex is The axial skeleton of glyptodonts exhibits a flattened, another feature distinguishing Ng from degree of fusion unparalleled among mammals. the more anterior teeth in the lower series, in Summarizing the general scheme of the axial which this apex is generally pointed or broadly construction for all glyptodonts, Hoffstetter rounded. (1958) stated that: the atlas is free, and usually DENTITION OF G. mexicanum.—As discussed un­ the axis and the next three or four cervical ver­ der "Systematics," the description provided by tebrae are united into a single element, the cer­ Cuataparo and Ramirez (1875) for the dentition vical tube; the sixth cervical is free; the posterior of G. mexicanum serves to fix provisionally the cervical and first two thoracic vertebrae are character of the species. That the specimens they united into a single element (occasionally referred described belong in Glyptotherium is well estab­ to as the "trivertebral" bone); and the remaining lished by the characters of the carapace, as de­ thoracics, usually numbering ten, are united into scribed elsewhere. According to their observations a compound dorsal tube. Because of the highly the tooth lengths vary from 21 mm anteriorly to fused condition of these vertebrae, determinations 25 mm posteriorly; hence the teeth are somewhat of vertebral count are difficult and are necessarily longer than those of G. cylindricum. The lower interpretive. Criteria for determination of verte­ tooth row measures 174 mm in total length, an bral count and the proper determination of in­ intermediate size in comparison with the other dividual elements in the following descriptions species. The authors figured occlusal views of two are (1) the presence of neural foramina emerging teeth, an anterior one and a posterior one, without laterally between fused transverse processes of designating their positions or whether they are adjoining vertebrae, analogous to the condition uppers or lowers. exhibited by armadillos; and (2) the presence of The drawing of the posterior tooth closely re­ rib facets or of an ankylosed first rib to distinguish NUMBER 40 69

thoracic vertebrae from cervical vertebrae in the ATLAS (Table 11).—The atlas of G. arizonae cervicothoracic region. To the extent that these (Figure 20) is the only free cervical vertebra. It is criteria are valid, rudimentary and incomplete short and complex, with a concave posterior facet comparisons between three North American taxa and a more deeply concave anterior facet. The are possible. dimensions are shortest in the anteroposterior The presacral vertebrae are unknown for G. direction, intermediate dorsoventrally, and most texanum. Only the atlas is known for G. cylindricum; greatly elongated in the transverse direction. The it exhibits minor differences from the atlas of G. alar processes extend upward from the antero- arizonae. In the latter species, the atlas is free; the dorsal angle of the atlas. They are obliquely cervical tube is formed by the fused axis and the compressed in the anterolateral-posteromedial remaining cervical vertebrae (i.e., nos. 2-7); there plane. The posterolateral face is concave, the is a free anterior thoracic vertebra bearing an anteromedial face roughened and irregular. The ankylosed, immovable rib; the second rib articu­ intervertebral foramina pass obliquely forward lates more or less freely with the second thoracic vertebra. The remaining vertebrae participating in the anterior dorsal elements, presumably cor­ responding to the so-called "trivertebral" element of South American glyptodonts, is unknown for G. arizonae. The dorsal tube is formed by at least 10 ankylosed thoracic vertebrae. The atlas of G. floridanum is unknown. The cervical tube consists of only five vertebrae, nos. 2-6. The sixth cervical and first thoracic verte­ brae are unknown. Presumably they were fused, and together supported the first rib in an immov­ able union, as in G. arizonae. The trivertebral element contains three dorsal vertebrae, presum­ ably thoracic vertebrae nos. 2-4. The dorsal tube includes nine thoracic vertebrae, nos. 5-13. Although there is little new material to report for the presacral axial skeleton of North American representatives, previous descriptions have been incomplete. The following descriptions are an attempt to consider more thoroughly this largely unknown and poorly understood region of the glyptodont skeleton. Brown (1912) described briefly the atlas of AMNH 15548 (G. cylindricum); this is the only known presacral vertebra for the species. Hay (1916) described the cervical tube, the triverte­ bral, and the broken dorsal tube of USNM 6071 (G. floridanum). Melton (1964) summarily de­ scribed the atlas, the cervical tube, and, according to his determination, the fused last cervical and first thoracic vertebrae (here considered as the FIGURE 20.—Atlas of Glyptotherium anzonae (UMMP 34826): free first thoracic) of UMMP 34826 (G. arizonae). a, anterior; b, dorsal; c, ventral. (Bar = 40 mm.) 70 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY and downward from a central position immedi­ upward motion beyond horizontal, while the an­ ately above the articular facets for the axis. These terior border of the odontoid expansion of the foramina emerge laterally and somewhat poste­ ventral arch functions as the ventral stop. riorly with respect to the alar foramina on the The atlas of G. cylindricum (Figure 21) differs dorsal arch. The alar foramina are more complex, little from that of G. arizonae. The alar processes coursing forward and mesially inward. These fo­ are more laterally directed than the near-perpen­ ramina emerge on the dorsal arch directly ante­ dicular orientation in G. anzonae. There are no rior to the alar processes, where a pair of thin struts bordering the anterior opening of the inter­ anteroposteriorly oriented struts flanks the upper vertebral foramina, and there is no mesially di­ borders of the foramina. A mesially directed rected groove passing toward the dorsal tu­ groove passes from the small opening formed by berosity. The intervertebral foramina emerge pos­ the medial strut of the alar foramen, and it teriorly much nearer the midline in G. cylindricum. terminates at the raised dorsal tuberosity. An­ The occipital facet is rather less concave, and the other groove passes from the same opening for­ odontoid facet faces entirely upward. In the atlas ward and medially to the anterior border of the of AMNH 15548 there is considerable evidence dorsal arch. These grooves presumably mark the of an arthritic condition at the occipital joint. position of vascular (nerves or blood) channels to The anterior extremity of the odontoid process the dorsal surface of the neck and rear portion of and the inner margins of the occipital facets are the head. marked by roughened, apparently pathologic, tu­ The dorsal and ventral arches are convex, the bercles. dorsal rather more massive than the ventral. The It is difficult to estimate the degree of transverse ventral tuberosity is an expansion of the ventral motion at the occipital joint. The atlas moves arch for the support and articulation of the odon­ freely in the transverse direction over the occipital toid process of the axis. The paired articular condyles through a half arc of approximately 15°. facets for the axis are confined to the posterior Because the atlas is so shortened (the anterior and face of the atlas. They are weakly concave in both posterior articular facets are so close that they the transverse and dorsoventral directions. They nearly contact internally), this figure seems ex­ are directed mesially and slightly rearward. These treme and a 10° half arc is probably more rea­ facets are confluent with the expanded facet for sonable. the odontoid process, which is confined to the The motion at the atlas-axis joint is more lim­ upper face of the ventral arch. It is weakly con­ ited. The convexity of the odontoid process al­ cave and faces upward and very slightly rear­ lowed a small degree of dorsoventral motion. The ward. relatively flattened articular facets permitted a The occipital facet is deeply concave dorsoven- greater transverse motion than in the anterior trally and concave transversely to a lesser degree. joint, the half arc apparently as great as 15°. The dorsoventral chord of each half of the facet Thus, the compound motion of the atlas, in func­ covers approximately 180°. The paired halves of tioning as the principal cervical articulation, is the occipital facet flare outward and forward, limited. forming a continuous marginal flange. Thus, the CERVICAL TUBE (Table 12).—In G. arizonae the primary motion at the occiput-atlas joint is dor­ cervical tube (Figure 22) is formed by the fused soventral; lateral motion was limited by the rel­ axis and cervical vertebrae 3-7 (i.e., the cervical atively flattened concavity in the transverse tube includes cervical vertebrae 2-7). This state­ plane. The deep dorsoventral concavity indicates ment is contrary to Melton's (1964) assertion that considerable dorsoventral motion of the head, the tube includes only cervical vertebrae nos. 2- from horizontal downward through an arc of 6, and that no. 7 is fused with the first thoracic. approximately 40°. The dorsal arch prohibits The neural arches and the extremities of the NUMBER 40 71

FIGURE 22.—Cervical tube of Glyptothenum arizonae (UMMP 34826): a, dorsal; b, anterior. (Solid line delimits articular surfaces as preserved; bar = 40 mm.)

transverse processes of the cervical vertebrae are absent. The neural tube is apparently closed above, as in G. floridanum, and it probably pos­ sessed short, fused neural spines. The convex odontoid process forms the anterior extremity. It is ovoid, and directed downward and somewhat forward, situated forward of the plane of articu­ lation of the lateral facets of the axis. The trans­ verse processes appear to increase in width rear­ ward, reaching maximum transverse dimensions at the last vertebra of the tube, this dimension exceeding that of the anteroposterior length. The vertebrae are solidly fused along the ventral arches, forming a flattened and transversely con­ cave lower surface. The only means of distinguish­ ing individual vertebrae in this specimen is by counting neural foramina. Externally there are five neural foramina, each presumably situated between transverse processes of the vertebrae, thereby indicating six cervicals. The anteriormost foramen, situated immediately posterior to the FIGURE 21.—Atlas of Glyptotherium cylindricum (AMNH articular facet of the axis, merges internally with 15548): a, anterior; b, dorsal; c, posterior; d, ventral. (Bar = the second neural foramen. Thus the first and 40 mm.) second neural foramina share a common internal opening for the neural canal, and there are only four internal neural foramina. 72 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

There are three posterior articular facets on along its upper and lateral borders. It is directed each side of the cervical tube for contact with the posteriorly and somewhat inward and upward. first thoracic vertebra and its fused rib. The The upper process is a posterior extension of the lateralmost facet is situated on the posterior face dorsolateral neural arch. It is situated directly of the last transverse process. It is concave and is above the lower facet, its lower surface forming directed posteriorly, receiving the corresponding the upper border of the neural canal at the cer- facet of the rib of the first thoracic vertebra. The vicothoracic joint. The articular facet faces down­ facet is transversely elongate, at least three times ward and posteriorly, receiving the opposing an­ longer than high; its lateral extent cannot be terior xenarthral process of the thoracic vertebra. determined from this specimen. In addition to The cervical tube of G. floridanum (Figure 23) this laterally situated facet for reception of the appears to be considerably different. Although modified head of the rib, there are two mesial the broken condition of the posterior extremity of facets, one above the other, for the articulation USNM 6071 precludes full comparison with between the centra of the last cervical and the UMMP 34826, it is sufficiently well preserved to first thoracic vertebrae. These facets are separated afford positive species distinction. The cervical from each other by the large neural foramen that tube of G. floridanum is formed by only five verte­ passes laterally and somewhat downward be­ brae, rather than six. This determination is tween the cervical and thoracic articulations. The founded on the presence of four external neural lower facet is also separated from the lateral rib foramina. As in G. arizonae, the first two foramina articular facet by the same obliquely directed merge mesially and share a common internal neural foramen. On the left side, there is a large opening for the neural canal. In USNM 6071 the and distinct anteroposteriorly directed foramen two anterior neural foramina bifurcate, sending situated in the angle between the three articular separate dorsal branches upward and rearward facets. This foramen communicates anteriorly to emerge above the lateral openings of the main with the adjoining neural foramen as a posterior foramina. The corresponding region in UMMP bifurcation of that canal. This foramen is entirely 34826 is broken. Hence the first two interior absent on the right side; there is a continuous foramina (coalesced nos. 1 and 2, and no. 3) bony cover in the corresponding position. communicate with distinct dorsal branches pass­ The lower of the internal facets is transversely ing upward and rearward from the neural canal. elongate and concave. It is situated on a necklike The determination of five vertebrae in cervical protuberance formed by the neural foramen tube USNM 6071 is contrary to Hay's (1916)

FIGURE 23.—Cervical tube oi Glyptotherium floridanum (USNM 6071): a, anterior; b, posterior; c, ventral; d, left side. (Bar =• 40 mm.) NUMBER 40 73 assertion that there are four (cervical nos. 2-5). it appears that the neural arch of the second Hay apparently counted only the interior open­ thoracic may have been immovably united with ings of the foramina into the neural canal. The that of the first; the broken condition of the bone determination presented here, that the vertebrae precludes certain determination of this contact, are nos. 2-6, is made in a fashion analogous to and what appears to be a broken contact might that for the G. arizonae specimens, in which the well be a relatively immobile articular surface cervical tube and anterior thoracic vertebrae are instead. A small, concave, posteriorly directed present, allowing more confident analysis. The facet on the first thoracic vertebra indicates an posterior cervical (no. 7) and the anterior thoracic articular contact with the head of the second rib. vertebrae are not preserved for USNM 6071, or The anterior articulations correspond to the for any other G. floridanum specimens, necessitat­ posterior facets of the cervical tube. The lower ing this determination of vertebral count by anal­ and inner facet is the smaller; it is situated on a ogy- prominent, knoblike, anteriorly projected tuber­ The dorsal arches of the cervical vertebrae of cle. The continuation of the neural foramen pro­ USNM 6071 are solidly fused, forming a medial duces a narrowed constriction posterior to the crest that increases in height rearward, reaching facet, effectively forming a distinct neck upon maximum elevation in the plane of the posterior which the centrum facet is situated. Above the extremity of the tube. The neural canal is nearly centrum facet, and separated from it by the circular above and laterally, and it is flattened neural foramen, is the anterior xenarthral process inferiorly. for articulation with the posterior process of the The anterior articular facets for the atlas are cervical tube. A flat, upwardly directed articular distinctive. The odontoid facet is inclined at ap­ facet is situated on the upper surface of the proximately 45°, thus facing forward as much as process. downward. This contrasts to the mostly down­ The rib facet is much larger than either of the ward direction in G. arizonae. Accordingly, the internal facets. It is convex and transversely elon­ odontoid facet of the atlas of G. floridanum (which gate, and it is directed anteriorly and somewhat is unknown) is probably also distinctive in pos­ laterally. The head of the rib is solidly fused with sessing a marked rearward-facing component in the ventrolateral angle of the vertebra, obscuring its orientation. the limits of these two bones. The shaft of the rib The posterior region of USNM 6071 is incom­ is expanded and flattened downward and for­ pletely preserved. The lateral rib facet is missing ward, forming a large, flat, vertically oriented on both sides. The partially represented centrum plate. A distinct tubercle ("unciform process"?) facets are separated by the large neural foramen marks the upper (inner) border in a position as in G. arizonae. The lower facet is directed rear­ midway between the head of the rib and the ward and outward, rather than somewhat inward midline of the body, giving a doubly concave, and upward. The upper facet faces posteriorly scalloped outline to the upper border of the rib. downward and outward, contrasting to the ori­ The lower border along the expanded plate is entation of this facet in G. anzonae. broken, and the distal end of the rib is absent, ANTERIOR THORACIC VERTEBRA.—Only the left obscuring the nature of the construction of the half of the first thoracic vertebra and the basal sternum. portion of the neural arch of the second are Because of the solid fusion of this rib with the known for G. arizonae (UMMP 34826, Figure 24). vertebra, it is certain that the vertebra is the first Most of the first rib is preserved. Its head portion thoracic and that this vertebra is apparently free. is solidly fused with the first thoracic vertebra. This contradicts Melton's (1964) conclusion that Anterior and posterior articular facets indicate the seventh cervical was fused with the first tho­ that the first thoracic vertebra was free. However, racic and that the rib was fused primarily with 74 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 24.—First thoracic vertebra with fused right rib oi Glyptotherium arizonae (UMMP 34826), oblique posteroventral aspect. (Bar = 40 mm.) the thoracic and in part with the cervical. To the consisting of the axis and five cervical vertebrae; contrary, there seems to be no indication of two one free thoracic with a single fused rib; and an vertebrae in the region of fusion of the rib. This indeterminate number of vertebrae in the re­ conclusion, that the vertebra is the first thoracic, maining thoracic series. rather than last cervical, is consistent with the The anterior thoracic vertebra and the last conclusion regarding the number of cervicals in­ cervical are unknown for G. floridanum. The pres­ corporated into the cervical tube of this species. ence of a free rib articular surface on the anterior- Measurements for thoracic vertebra 1 (free an­ most vertebra of the trivertebral element of terior thoracic with fused rib) of UMMP 34826, USNM 6071 indicates that the first thoracic ver­ for the left side, are as follows: anteroposterior tebra is indeed missing and is not incorporated diameter between rib articular facets, 47 mm; into the trivertebral element. The presence of an transverse diameter anterior rib facet, 40 mm; immovable first rib, fused to the first thoracic transverse diameter lateral extremity anterior rib vertebra, is apparently a characteristic common facet to internal extremity centrum facet, approx­ to most late Cenozoic glyptodonts. This fact, plus imately 62 mm. the similarity between the trivertebral elements The anterior vertebral count for G. anzonae is between G. anzonae and G. floridanum, insofar as as follows: one free atlas; fused cervical tube known, indicates that the first thoracic vertebra NUMBER 40 75 is indeed absent in USNM 6071. The lack of the described by Hay (1916) as the seventh cervical seventh cervical vertebra in USNM 6071 is also and first two thoracics, following the South Amer­ problematical. Either there was reduction in the ican supposition of homologies. Except for the number of cervical vertebrae, or there was a free fragment of the anterior vertebra of the triverte­ terminal cervical and a free anterior thoracic, or bral element associated with the free first thoracic the last cervical and first thoracic were united vertebra of UMMP 34826 (G. anzonae) discussed into a single element, together supporting the first above, the trivertebral element is unknown for rib. The latter suggestion is the most probable, the remaining North American taxa. However, at since in glyptodonts there is considerable varia­ least for G. arizonae, the trivertebral was probably tion in this region but apparently never any not significantly different, for the anterior artic­ reduction in the cervical count. ular facets of the trivertebral of USNM 6071 These elements are unknown for the other exhibit a construction complementary to the pos­ North American species. This presumed distinc­ terior facets of the first thoracic vertebra of tion between G. arizonae (a free first thoracic) and UMMP 34826; thus, at least the articulations are G. floridanum (probably fused last cervical and first similar, and probably the entire elements as well. thoracic) may prove to be important in further The trivertebral element of USNM 6071 (G. elucidation of evolutionary relationships of North floridanum, Figure 25) is shorter in the anteropos­ American glyptodonts. terior dimension, longest in the transverse dimen­ "TRIVERTEBRAL" ELEMENT.—The so-called sion as the following measurements indicate: "trivertebral" element of South American glyp­ maximum transverse diameter between articular todonts has usually been considered as consisting facets for anterior ribs, 128 mm; maximum trans­ of the last cervical vertebra and the first two verse diameter between articular facets for pos­ thoracics (Hoffstetter, 1958). Apparently these terior ribs, approximately 137 mm; maximum homologies were first established by Huxley slightly oblique dorsoventral diameter from trans­ (1865), who stated that the anteriormost rib ar­ verse tangent connecting lower edge of posterior ticular fossa on the trivertebral element is clearly articular facets to upper extremity dorsal spine, formed by the last cervical and first thoracic. This 75 mm; maximum anteroposterior diameter be­ identification assumed that the actual first rib tween vertebral articular facets, left side, 66 mm articulated at this position. Huxley unfortunately and right side, 65 mm; minimum anteroposterior had no complete anterior vertebral series to ex­ inside diameter between neural foramina, left amine, and apparently he was not aware of the side, 27 mm and right side, 19 mm; maximum vertebra anterior to the trivertebral, which bears dorsoventral diameter neural canal, 24 mm; max­ a fused rib, here identified as the first thoracic. imum transverse diameter neural canal, 42 mm. Apparently, Burmeister (1874) also missed this That there are indeed three vertebrae is indi­ important feature. Although it is possible that cated by the presence on each side of two neural there is here a fundamental distinction between foramina passing laterally downward and for­ the North and South American taxa, it is more ward. Each foramen bifurcates internally, send­ likely that the homologies for the participating ing a separate dorsal canal to the upper surface vertebrae of the trivertebral were erroneous, as of the transverse process and a ventral canal to first suggested by Huxley, and that these homol­ the lateral extremity. Above each ventral foramen ogies have remained unchallenged. At least is a complex rib articular facet, with a separation among the North American representatives, in­ between the head and tubercle portions. The sofar as known, the trivertebral is formed by the upper (head) portions of the facets occupy the second, third, and fourth thoracic vertebrae. lateral extremity of the fused transverse process. A complete trivertebral element is known for Each is elongate and concave in the anteropos­ G. floridanum (USNM 6071). This specimen was terior direction. The anterior head facet is more 76 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 25.—"Trivertebral" element of Glyptothenum floridanum (USNM 6071): a, left side; b, anterior; c, posterior; d, dorsal; e, ventral;/, trivertebral element (t = trivertebral element, d = dorsal tube). (Bar = 40 mm.) NUMBER 40 77 roughened and irregular, indicating less mobility teroposteriorly and flat transversely. Its contact than the posterior facet, which is relatively with the central facet of the dorsal tube indicates smoother. The head facets are circumscribed by a motion between these two elements that is small tubercles for muscle attachment, and they confined to the vertical plane. The construction are penetrated near their superior margin by of this articulation precludes any torsional or several vascular foramina. lateral motion. The four smaller tubercle facets are situated The dorsal spine is apparently formed by the beneath the head facets. They are concave and fused neural arches of the middle and posterior slightly oval in shape. One tubercle facet is lo­ vertebrae. It is inclined steeply rearward, its an­ cated anterior to the first neural foramen. It terior border forming an angle of approximately articulates with the corresponding facet on the 45° with respect to the horizontal axis of the first rib, which is (presumably) united to the first bone. Its superior extremity, terminating in a thoracic vertebra. Another tubercle facet, for the stout, blunt tuberosity, is situated immediately second rib, is situated posterior to the first neural above the central facet for the dorsal tube. This foramen and directly beneath the anterior head construction surely bolsters the articulation with facet on this bone. A third tubercle facet is located the dorsal tube and indicates the principal im­ external and slightly anterior to the posterior portance of this articular contact. foramen. It contacts the posterior head facet. A The neural canal is short. It is superiorly arched fourth tubercle facet, for the first rib of the ad­ in a rounded profile, and it is inferiorly flattened. joining dorsal tube, is situated on the posterior Its anterior opening is posteriorly inclined in surface of the transverse process. It is located roughly the same outline as that of the dorsal internally with respect to the other rib facets on spine, so that the superior extremity of the ante­ the trivertebral element, and, unlike the other rior opening lies well posterior to the inferior laterally directed tubercle facets, the fourth facet margin. faces downward and rearward. The ventral surface of the trivertebral element The anterior articular surface for the first tho­ is smooth and unmarked by tubercles or foram­ racic vertebra is poorly delimited. This rugose ina. The posterior surface of the dorsal spine is region is marked by a number of small tubercle­ deeply marked by three vertical ridges and four like processes, and there is no distinct articular vertical grooves. This roughened region appar­ region, indicating that the mobility in the artic­ ently provided an enlarged surface for attach­ ulation between the first thoracic vertebra and ment of the dorsal vertebral musculature con­ the trivertebral element, although free, is rela­ necting the trivertebral element with the dorsal tively minor. This contrasts with the posterior tube; this is another indication of the importance articulation with the dorsal tube, for which there of the trivertebral-dorsal tube articulation. This are three well-developed facets on the trivertebral articulation can pass through an arc of at least element. The two lateral facets are transversely 60°, although this figure is probably excessive by elongate and irregularly concave. They are ori­ 15°-20° over that possible in the living state. ented rearward and somewhat medially and up­ Nevertheless, a vertical articulation of at least 40° ward. The transverse and dorsoventral diameters is a considerable arc for a single contact. are approximately 42 mm and 19 mm, respec­ DORSAL TUBE.—Fragmentary dorsal tubes are tively. The upper facet, situated directly beneath known for G. floridanum and G. arizonae. Repre­ the dorsal spine and confined to the inferior sented by three broken portions, the dorsal tube surface of the neural arch, faces entirely down­ of G. floridanum (USNM 6071, Figures 25/, 26) ward. Viewed inferiorly, this facet is seen to oc­ appears to include only nine vertebrae, here iden­ cupy approximately half of the inferior surface of tified as thoracic vertebrae nos. 5-13. This deter­ the neural arch. It is crescentic in shape and mination is contrary to Hay's (1916) assertion transversely elongate. This facet is concave an­ that there are 10 vertebrae represented in this 78 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

for South American glyptodonts (Hoffstetter, 1958). The three broken portions of USNM 6071 apparently include nearly the entire dorsal tube. The anterior fragment includes the first three vertebrae and a fragment of the fourth (nos. 5-8) of the dorsal tube. The middle fragment ante­ riorly includes the posterior portion of a vertebra, here identified as the rear section of thoracic vertebra no. 8; three complete vertebrae, nos. 9- 11; and the anterior section of thoracic vertebra no. 12, here interpreted as the penultimate ver­ tebra. The posterior fragment includes the rear section of the penultimate vertebra and the com­ plete terminal vertebra, here interpreted as tho­ racic vertebrae nos. 12 and 13, respectively. Al­ though the three fragments cannot be reunited into a single element, there appears to be no entirely missing vertebra. Despite Hay's (1916) assertion that one vertebra is missing, at the breakage between the anterior and middle frag­ ments, there is no compelling justification for this assumption. Indeed the curvature and continuity of the entire dorsal tube is well maintained with­ out (imaginary) supposition of a missing vertebra. Moreover, a supposition of 10 vertebrae in the dorsal tube, bringing the count up to that for most South American representatives, would add another vertebra to the total thoracic count of 14 for G. floridanum, an even more drastic departure from the usual count of 12 in South American forms. Thus the three fragments of USNM 6071 are here assumed to include at least portions of all of the vertebrae incorporated into the dorsal tube of G. floridanum, representing thoracic verte­ brae 5-13. Except for the number of vertebrae, the dorsal tube of G. floridanum does not appear to FIGURE 26.—Dorsal tube oi Glyptotherium floridanum (USNM differ significantly from that of either G. arizonae, 6071), approximately anterior third: a, dorsal; b, ventral; c, described below, or Glyptodon asper, as figured by anterior. (Bar = 40 mm.) Burmeister (1874, pi. 30), and noted by Hay (1916). specimen. The count of nine vertebrae in the The anterior articulation with the trivertebral dorsal tube, rather than 10, and of a total thoracic element is accomplished by a distinct crescentic series of 13 rather than 12, indicate a contrast zygapophyseal facet and a pair of lateral facets. from the usual conditions of 10 vertebrae in the The dorsal zygapophyseal facet is convex both dorsal arch and a total of 12 thoracic vertebrae transversely and anteroposteriorly and is directed NUMBER 40 79 forward and upward, fitting neatly into the cor­ Posteriorly, the ventral surface is strongly convex responding facet of the trivertebral element. The in the transverse direction, conforming to the lateral facets, situated ventrally on the transverse interior shape of the neural canal. In the antero­ processes of the anterior vertebra, are semicylin- posterior direction the ventral surface maintains drical (to borrow from Hay's, 1916, apt descrip­ a relatively small compound curve, with an in­ tion) for articulation with the correspondingly creasing curvature in the concavity toward the concave lateral facets of the trivertebral element rear. in a tight contact. Concave facets for the heads of ribs 4-6 are The low and irregular dorsal spine is preserved situated ventrally on the transverse processes. posteriorly only to the thoracic vertebra 9. Al­ They are elliptical and are directed outward and though broken above 10-12, the process appears slightly forward. The head facet for rib 4 is to have been weak and its vertical extent insig­ located somewhat lateral and ventral with respect nificant. Above the terminal vertebra the spine is to the next two facets, which are situated pro­ reduced and forms only a prominent median gressively more superiorly and medially, in con­ ridge, which does not extend vertically as far as formation with the corresponding change in the the transverse processes. transverse process. The transverse processes are completely pre­ The tubercle facet for rib 4 is located on the served for the first three vertebrae of the dorsal corresponding region of the trivertebral element, tube. The transverse processes appear to taper as described above. Tubercular facets for the fifth gradually rearward from their maximum expan­ and sixth ribs are situated in the shallow recesses sion at thoracic vertebra 6 (second from front), as separating the respective transverse processes. indicated by the progressively more medial dis­ These two concave facets are somewhat smaller position of the neural foramina. Posteriorly the than the head facets; they are located anterior to, ventral surface of the coalesced transverse process and somewhat above, their respective head facets. becomes deflected laterally and compressed, at They are directed downward and posteriorly. The the terminal vertebra attaining a transverse ex­ remaining rib facets are missing. tent apparently only slightly greater than that of The posterior articular surface for contact with the neural canal. The neural foramina internally the sacrolumbar tube is formed by a pair of bifurcate, as in the preceding vertebrae, sending lateral wing-shaped facets and a pair of dorsal separate dorsal and lateral branches to the exte­ postzygapophyseal facets. These are separated by rior. The dorsal openings, on the upper surface of deep recesses for the corresponding prezygapo- the transverse process, are smaller and less con­ physeal facets of the lumbar tube. The small spicuous than the round lateral foramina, ante­ zygapophyseal facets of the dorsal tube are flat­ riorly emerging from the ventral surface of the tened and directed downward and outward and transverse process, posteriorly becoming more lat­ slightly rearward. The rearward-directed lateral erally situated and nearer the neural canal. facets, situated on the terminus of the only The neural canal increases in depth rearward, slightly expanded transverse processes, are rough changing gradually from a nearly circular shape and irregular, indicating a loose, primarily liga- anteriorly to an oblate and deepened shape at the mentous connection at this contact. posterior terminus. As Hay (1916) pointed out, Measurements for the dorsal tube of USNM the floor of the neural canal is approximately 1 6071, insofar as the broken condition permits, are mm thick at the anterior extremity and approxi­ as follows: estimated anteroposterior diameter, mately 5 mm thick at the posterior end. 370 ± 10 mm; maximum transverse diameter at Anteriorly the ventral surface is smooth and anterior extremity, 110 mm; maximum transverse broadly convex in the transverse direction. The diameter at posterior extremity, 46 mm; maxi­ ventral surface in the middle region is missing. mum transverse diameter neural canal anterior 80 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY extremity, 30 mm; maximum transverse diameter along a common breakage. The count of 10 ver­ neural canal posterior extremity, 22 mm; maxi­ tebrae in the dorsal tube of G. anzonae differs by mum dorsoventral diameter neural canal anterior one from that assumed for G. floridanum, as dis­ extremity, approximately 23 mm; maximum dor­ cussed above. However, due to the fact that the soventral diameter neural canal posterior extrem­ count for the latter species is not established with ity, 36 mm; maximum transverse diameter pre- certainty, distinction between these two species zygapophyseal facet (anterior extremity), 67 mm. on the basis of the number of vertebrae in the As Hay (1916) pointed out, there are several dorsal tube is unjustified. differences between USNM 6071 and the dorsal The anterior eight vertebrae of the dorsal tube tube figured by Burmeister (1874, pi. 30) for of G. arizonae (USNM 10536, Figure 27) can be Glyptodon asper, among which smaller size for the examined only from the ventral surface. The last North American species is perhaps the most sig­ two vertebrae, although broken, are free from nificant. Because Burmeister provided no com­ their plaster encasement. The anterior articular parable measurements, and since his illustrations facets closely resemble those of USNM 6071. The are somewhat idealized for his plate 30, figure 1, dorsal facet, although partially hidden, appears further comparison here, on the basis of his ac­ to possess a convex crescentic construction as in count, is not justified. G. floridanum. The lateral facets differ only in Gidley (1926) mentioned the recovery of the being slightly elevated above the surface of the dorsal tube of USNM 10536, stating that it in­ surrounding bone, with a better demarcation of cludes 10 fused vertebrae. He did not compare it the articular margin. with USNM 6071, however, in deference to a The centra of the first five vertebrae are pres­ comparison with the South American genera Pan- ent, and they retain discrete separation from ochthus, Glyptodon, and Hoplophorus (with 10, 11, adjacent centra. In lateral silhouette, the ventral and 12 vertebrae in the dorsal tube, respectively, outline is weakly scalloped, owing to slight thick­ according to Burmeister (1874)). The count of 10 ening of each centrum at the contact with the vertebrae appears to be correct, and we accept adjoining centra. The centra are not preserved this number because the specimen has deterio­ for USNM 6071, precluding comparison with G. rated somewhat from Gidley's time and is now in floridanum. two broken portions that cannot be matched Although somewhat damaged, the posterior

FIGURE 27.—Schematic interpretation of dorsal tube oi Glyptotherium arizonae (USNM 10536): a, right side, anterior to right; /;, cross section through section line in a. (Bar = 40 mm.) NUMBER 40

lumbar articular region is sufficiently well pre­ indicated by rib 1 and the free rib fragment served to indicate near identity with that of G. discussed above. floridanum. The only apparent difference is that the vertical height of the dorsal tube at its poste­ FRONT LIMB rior extremity is rather larger in G. arizonae. Its broken condition does not allow accurate mea­ The elements of the front limb are somewhat surement. less massive than those of the hind limb. Although The only notable difference between these two it exhibits a measure of specialization in the loss dorsal tubes lies in the nature of the fused trans­ of the first digit, the extreme reduction of the verse processes in the middle region of the bone. fifth digit, and the reduction of the trapezium In G. floridanum the transverse processes, as viewed and trapezoid, the front limb nevertheless retains from below, emerge from the centrum with a mostly a primitive construction correlated with distinct vertical orientation, whereas the ventral the animal's massive build: the scapula is an surface is more flattened in G. arizonae. Accord­ extremely broad plate, providing a large area for ingly, the positions of emergence of the ventral muscle attachment; there is no indication of a rami of the neural foramina reflect this difference. clavicle; the humerus is pillar-like; the radius In G. floridanum they emerge from the nearly length is only half that of the humerus, a propor­ vertical wall of the transverse processes near the tion indicating the extreme graviportal mechan­ centra. In G. anzonae these foramina are situated ics manifest in the front limb; the ulna shares more laterally, where the ventral surface is not so equally with the radius the transfer of weight to steeply inclined. Another difference, though cer­ the carpus; the carpal bones are unfused, with tainly not of taxonomic value, is that the anterior the nominal exception of disease conditions; there neural foramina are much larger in G. arizonae. is a full battery of digital sesamoids, a large The posterior foramina are approximately equal. palmar sesamoid, and a well-developed pisiform; Because of the broken condition of dorsal tube the metacarpals and phalanges are stout and USNM 10536 no accurate measurements are pos­ relatively simple. The construction of the manus sible. Its proportions and size closely resemble is suggestive of an elephant-like padded foot. those of USNM 6071. If future discoveries sub­ There appears to have been little capacity for stantiate the existence of 10 vertebrae in the digging, despite others' statements to the con­ dorsal tube of G. floridanum, rather than nine as trary. proposed here, there will be no definitive distinc­ Only Burmeister (1874) and Vinacci Thul tion in this element between these two species. (1943) have discussed in detail the osteology of RIBS.—Except for the first rib, which is fused the front limb of glyptodonts. Vinacci Thul to the first thoracic vertebra, ribs of North Amer­ (1943) provided an excellent description of the ican glyptodonts are mostly unknown. A single manus of Glyptodon, establishing that the missing fragment, representing the proximal extremity of digit is the first, rather than the fifth as asserted a rib of G. arizonae (MU 1668), possesses a dis­ by Burmeister (1874) and others. Unfortunately, tinctly flattened shaft in the typical glyptodont Vinacci Thul provided only one illustration and fashion. It bears two articular facets. Measure­ no measurements to accompany his description. ments for this fragment are as follows: transverse The front limb of the North American glypto­ diameter between facets, 74 mm; dorsoventral donts closely resembles that oi Glyptodon, although and transverse diameters of shaft distal to proxi­ all of the South American species seem to be mal expansion, 33 mm and 19 mm, respectively. sufficiently distinct from North American repre­ Because of the similarity of Glyptotherium with sentatives to preclude any assumption of identity. South American representatives, the ribs and ster­ SCAPULA (Table 13).—The broadly rounded num are here assumed to be similar, as tentatively scapula of Glyptotherium provides a widely ex- 82 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY panded surface for muscular attachment. The the greater portion of a nearly complete irregular straight scapular spine, increasing in height ven­ semicircle. The infraspinous fossa is smooth and trally, expands into a broad curved acromion. relatively flat, with the cranial border slightly The acromion process curves ventromedially be­ inflected laterally. yond the plane of the glenoid cavity. The spine The entire medial surface is smoothly concave. divides the scapula into a smaller cranial supra- The subscapular fossa is relatively smooth, with spinous fossa, and a larger caudal infraspinous a dorsoventral furrow extending from near the fossa. The caudal border is relatively straight to vertebral border toward the neck of the scapula its confluence with the neck. The thickness of the in the position opposite the scapular spine on the bone is expanded along this border. Although the outer surface. The bone is extremely thin over the scapular cartilage was apparently not extensive entire surface of the fossa except in the in F:AM 95737 (G. texanum, Figure 28), the strengthened region of the scapular spine. The broadly rounded vertebral border is incomplete bone in the neck region thickens toward the along the margin because of the incomplete ossi­ glenoid cavity and the coracoid process. The fication of the scapular margin in this subadult latter, broken in F:AM 95737, projects antero- individual. The caudal angle is sharp and well medially from the glenoid border. defined. The cranial border is poorly preserved. The glenoid cavity is broadly and irregularly It appears to have been confluent with the ver­ ovoid, the broadest expansion occurring at the tebral border, with no obvious cranial angle. If posterior border of the articular surface. The visually projected the vertebral border describes glenoid fossa is shallowly concave in both the transverse and anteroposterior directions. It is directed perpendicular to the surface of the ex­ panded portion of the scapula. The scapula of G. arizonae (USNM 10536, Fig­ ure 32a) is considerably larger and more massive. The caudal angle is decidedly more acute, and the surface of the infraspinous fossa bears several large prominences. As in G. texanum, the cranial border is broadly rounded, but the caudal border is more deeply concave owing to the acute caudal angle. The acromion process is less symmetrical in lateral aspect, with a considerably greater cra­ nial development and little caudal development of the curved extremity. The scapular spine is curved rather than straight. As Gidley (1926) pointed out, the scapula of USNM 10536 resembles that of the South Amer­ ican genus Glyptodon, although it differs consider­ ably in the anteroposterior extent, in the broadly rounded superior scapular border, and in the shapes of the external fossae. The scapula of F: AM 95737 differs much in the same way and is more similar to USNM 10536 than to the South American forms. Thus the scapula of each species resembles the other more closely than either re­ ''IGIIRE 28.—Right scapula of Glyptothenum texanum (F:AM sembles that of South American species, so far as 95737): a, lateral; b, ventral. (Bar = 40 mm.) known. NUMBER 40 83

HUMERUS (Table 14).—The humerus of Glyp­ its distal extremity. The teres tuberosity on the totherium is a stout, pillar-like bone with a massive posterior surface of the shaft is poorly developed. greater tuberosity. Only the right humerus of F: In cross section, the shaft at its midpoint forms AM 95737, a juvenile, is represented in North a nearly equilateral triangle, with rounded apices. American collections for G. texanum; the nature of The posterior apex is formed by the confluence the epiphyseal contacts indicates that growth was of the subdued ridge of the teres tuberosity with nearly complete at the time of death. The hu­ the proximal portion of the supinator ridge; the merus of G. texanum is considerably smaller than anterior apex is formed by the confluence of the that of G. anzonae, even allowing for additional medial supracondylar ridge as it crosses over the length in the full adult condition. The humerus anterior border in line with the distal portion of of G. floridanum is known only for the distal ex­ the deltoid tuberosity; and a posteromedial apex tremity, and it is unknown for G. cylindricum. is formed as the proximal extension of the poste­ In G. texanum (Figure 29) the hemispherical rior border of the medial supracondylar ridge. head presents a well-defined articular surface. In The lateral supracondylar ridge forms an ex­ proximal view the head is slightly elongate in the panded supinator plate, extending in a spiral anterolateral direction, with a right angle formed course from a posterior position near midshaft to by the convergent margins of the head along the a lateral position at its confluence with the lateral indistinct neck between the head and the tuber- epicondyle. The supinator ridge thickens distally, osities. The greater tuberosity extends nearly as forming a smooth concave anterior surface as the far proximally as the head, forming a massive distolateral border of the musculospiral groove. lateral tubercle. The lesser tuberosity, forming The posterior surface of the lateral supracondylar the anteromedial eminence of the proximal end, ridge is flattened. The medial epicondyle forms is separated from the greater tuberosity by a an expanded surface on the distomedial extrem­ shallow bicipital groove. With the head, the ru­ ity. Its rugose surface is penetrated by three small gose tuberosities comprise the proximal expan­ foramina. sion, the maximum diameter of which is approx­ The lateral epicondyle is continuous with the imately twice the minimum diameter of the shaft supinator ridge and it is not greatly expanded. Its at its midpoint. surface is directed anterolaterally. The coronoid The ridge of the greater tuberosity is rounded and radial fossae are confluent, forming a single and poorly defined, intersecting the deltoid ridge shallow excavation with a triangular outline. The approximately one-third the distance along the olecranon fossa is more deeply excavated, forming shaft from the proximal extremity. The anconeal a semicircular anterior boundary. The anterior crest and the deltoid ridge form the posterior boundary is the steepest, nearly at a right angle border of the well-developed deltoid tuberosity. with the shaft. There is no supratrochlear fora­ The lateral border of the deltoid tuberosity on men piercing the bony septum. F:AM 95737 is somewhat damaged; there was The trochlear facet is directed obliquely in a apparently a subdued flange along the course of distolateral direction. The capitular facet is the deltoid tuberosity in this position. The deltoid broader and provides a greater radius of articu­ tuberosity is directed anterolaterally and forms lation. The intercondylar groove is shallow. an isosceles triangle in lateral aspect, with the The most prominent features of the humerus distal apex positioned at the junction of the del­ of G. texanum are the expanded deltoid tuberosity toid tuberosity and the ridge of the greater tu­ and the broadened supinator plate. Although not berosity. greatly expanded at either extremity, there is The ridge of the lesser tuberosity is not as well sufficient development to deduce strong muscu­ developed. At a point on the medial surface lature in the upper limb. The long axis of the opposite the apex of the deltoid tuberosity, a humerus is directed from the most proximal ar­ projection of the ridge as a low eminence marks ticulating surface of the head through the center 84 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 29.—Right humerus oi Glyptothenum texanum (F:AM 95737): a, posterior; b. medial; anterior; d. lateral. (Bar = 40 mm.) of the shaft to the trochlear facet which is located tinction often mentioned in the literature is the in a central position on the distal extremity. presence of a supratrochlear foramen. This char­ The humerus of G. anzonae (Figure 32b) appar­ acteristic, however, actually seems to bear diag­ ently differs considerably from that of G. texanum. nostic value only as a specific distinction. Until a Besides being much larger and considerably more series of glyptodont humeri is evaluated for on­ massive, the humerus of G. arizonae presents a togenetic variation, this characteristic should be tremendous exaggeration of features related to used only as a provisional distinction, as indicated regions of muscle attachment. A prominent dis­ herein. Because the only complete humerus of G. NUMBER 40 85

texanum available for comparison (F:AM 95737, on the posterior surface of the shaft at its mid­ above) is from ajuvenile individual, the following point, in direct line with the proximal extension full description of the humerus of G. arizonae is of the lateral supracondylar ridge. included. Because of the age disparity the differ­ The cross-sectional aspect of the shaft at its ences may be more apparent than real, but until midpoint is a slightly eccentric ellipse, with the new specimens of both species are recovered, primary axis extending from the craniomedial separate descriptions are warranted. Indeed, the border to the caudolateral border. This cross distal end of the right humerus of G. texanum section obtains for only a short distance along the (TMM 40664-245) from the Red Light local middle portion of the shaft, giving way to irreg­ fauna is more nearly like that of G. arizonae ular outlines at either extremity. (USNM 10536), as pointed out by Akersten The trochlear and capitular facets are circular (1972). It lacks a supratrochlear foramen. in outline in lateral aspect, forming an arc of In G. anzonae (USNM 10536, Figure 326) as in approximately 300°. The trochlear facet forms G. texanum, the head presents a broad articulating the distal extremity. The medial epicondyle is surface, expanding beyond the thickness of the expanded and bears two distinct notches for mus­ shaft at the epiphysis. The greater and lesser cle insertions, one on the anterior surface, the tuberosities extend nearly to the most proximal other on the anteroproximal surface. extremity of the head and are separated from it The coronoid fossa is deeply excavated, with by an indistinct neck. The intertubercular groove the medial and lateral supracondylar ridges form­ (bicipital groove) is distinct only between the ing prominent borders on either side of the tri­ tuberosities, where it forms a narrow furrow. The angular fossa. The coronoid fossa undercuts the crests of both tuberosities are pronounced. The medial ridge. The bony septum in the coronoid crest of the lesser tuberosity extends along the fossa is extremely thin and is perforated in the medial surface to midshaft. The crest of the central position by the supratrochlear foramen. greater tuberosity, in cranial aspect, forms a From the lateral supracondylar ridge issues a broad surface for muscle attachment, curving prominent flange, the supinator ridge, concave downward across the anterior surface to intersect on its cranial surface. The supinator ridge extends the lateral margin of the deltoid tuberosity at along the lateral surface of the lateral supracon­ approximately one-third the distance from the dylar ridge, forming the distal boundary for the head along the shaft. The surface between the musculospiral groove. The lateral epicondyle is crest of the greater tuberosity and the deltoid directed distolaterally, forming a broadly obtuse ridge is concave, elongated parallel to the sagittal angle with the supinator ridge. The olecranon plane. The deltoid tuberosity presents a rugose fossa is roughly elliptical in shape, with the major triangular surface in lateral aspect, with the distal axis directed at an angle from proximolateral to apex located at the intersection of the crest of the distomedial. The fossa is deeply excavated; on the greater tuberosity with the deltoid ridge on an medial boundary the excavation undercuts the elevated surface. The caudal margin of the del­ medial supracondylar ridge. toid ridge is inflected, forming a prominent Only the distal ends of the humeri of G. flori­ flange. This deltoid flange extends from a point danum are known (USNM 6071, Figure 32«,/). near the head anteromedially across the proximal Their size and shape more closely approximate portion of the shaft, forming the anterior bound­ that of G. arizonae than G. texanum. There is no ary of the musculospiral groove, which wraps supratrochlear foramen, although this may not around the lateral border of the shaft to an be an important distinction, as indicated above. anterior position on the distal end. The inner This character is here utilized only as a provi­ surface of the musculospiral groove is formed by sional species distinction. The coronoid and ole­ the well-developed supinator ridge that originates cranon fossae are small and not deeply excavated, 86 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY rather intermediate in form between those of G. margin is nearly straight, with a gentle curvature arizonae and G. texanum. at the proximal extreme in contour with the ULNA (Table 15).—The ulna of G. texanum (F: rounded olecranon process. The coronoid process AM 95737, Figure 30) is essentially a straight, on the medial side of the semilunar notch pro­ laterally compressed rectangular prism, rounded vides an oblique surface for articulation with the on the posterior margin of the olecranon process. trochlear facet of the humerus. Lateral to the The rugose epiphysis of the olecranon constitutes coronoid process are two distinct facets: the flat­ the proximal extremity, with only slight expan­ tened radial notch, directed anteriorly, for recep­ sion from the shaft. The shaft is but slightly tion of the radius; and the capitular facet, form­ expanded transversely at either extremity; the ing the lateral surface of the semilunar notch for narrowest portion of the shaft occurs immediately reception of the capitulum of the humerus. below the semilunar notch. Distal to the semilu- The distal extremity is not greatly expanded. nar notch, the narrowed anterior margin of the Two facets provide the articular surface for the shaft is bowed medially. The rounded posterior carpals. The larger facet is lateral (anterior in border is nearly straight, tapering from a broad­ anatomical position) and provides the articular ened proximal surface to a narrowed ridge on the surface for the cuneiform. This semicircular facet distal end. In lateral aspect also, the posterior is flattened in the cross-sectional plane and forms

FIGURE 30.—Right ulna oi Glyptothenum texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 40 mm.) NUMBER 40 87

the distal extremity of the ulna. The facet for obliquely with respect to the articular facets. The articulation with the pisiform is medial and pos­ rugose styloid process is the most prominent fea­ terior to the cuneiform facet (laterally in anatom­ ture on the distal extremity. The dorsal tubercle, ical position), directed distomedially (distolater- although not as prominent as the styloid process, ally in anatomical position). The surface of the extends laterally to form a rounded rugose knob pisiform facet is oblate and convex, fitting loosely on the anterolateral surface of the distal extrem­ into the anterolateral articular surface of the ity. The distal articular surface consists of a single pisiform. concave facet for the reception of the lunar and The only remaining prominent feature of the trapezoid bones of the wrist. This surface is di­ ulna is a pronounced trough, with a roughened rected posteriorly at approximately 45° with the surface distal to the coronoid process of the sem­ cross-sectional axis of the shaft. ilunar notch and situated on the anteromedial As for the ulna, the radius of G. arizonae (USNM side of the shaft. 6701, 10536, UMMP 38761, Figure 32c) differs Except for greater size, the ulnae of G. floridanum only in being larger and possessing rather exag­ (Figure 32c) and G. arizonae (Figure 32d) differ gerated features, especially those related to mus­ only in the more greatly exaggerated features, a cle attachment. The apparent size disparity, as in condition that might be attributed to the age the ulna, may be due to the differences in age disparity represented by the available specimens between the juvenile specimen of G. texanum and of these two species (USNM 6071, 10536, respec­ the older individuals of G. anzonae; thus there are tively). no apparent diagnostic distinctions in the radii of RADIUS (Table 16).—The radius of G. texanum these two species. The radius of G. floridanum (F:AM 95737, 59585, Figure 31) is short and (USNM 6071, Figure 32a") appears to bear no stout, with both extremities expanded from the distinctive differences from that of G. arizonae, and narrow shaft; the distal extremity of the shaft is it differs from G. texanum much in the same way. the larger. The length of the radius is approxi­ CARPUS.—The carpal bones are arranged in mately half that of the humerus. The radius two rows, four bones in each row. The proximal crosses the anterior surface of the ulna at a low row includes the pisiform, and three tightly abut­ angle, permitting only a slight degree of rota­ ted elements, the cuneiform, lunar, and scaphoid. tional motion in the forearm. The glenoid cavity Together, the last three bones form a broad con­ of the radius presents an expanded articular sur­ vex proximal articulation with the radius and face for the reception of the broad capitulum of ulna. Distally, these elements articulate with the the humerus. Proximal articulation with the ulna bones of the distal row in a construction permit­ is limited to the flattened posterior face on the ting limited fore-and-aft motion. Transverse mo­ epiphysis. There is no radial tuberosity. The in- tion at the intercarpal joint was almost totally terosseous space between the radius and the ulna prohibited by the interlocking construction of the extends from the anterior epiphysis of the radius distal series, both with each other and with the distally to the point of articulation with the ulna proximal articulations from above. at the distal end of the shaft of the radius. This The distal series includes the trapezium, tra­ concave ulnar facet, on the posterior border of pezoid, magnum, and unciform. The trapezium the radius, includes only a small portion of the is considerably reduced relative to the other car- epiphysis. It is situated opposite the convex an­ pals; the trapezoid is also reduced, but to a lesser terior radial facet of the ulna, on which the degree; the magnum and unciform are large. The articulation is equally divided between the shaft distal carpals articulate with the metacarpals dis­ and the epiphysis. tally, also in a construction permitting limited The radius is approximately equidimensional fore-and-aft motion and little transverse motion. at midshaft. In lateral view the radius is straight. The carpals of the North American represent­ In anterior view the shaft of the radius is directed atives closely resemble the description of those of SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 31.—Right radius oi Glyptothenum texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 40 mm.)

Glyptodon by Vinacci Thul (1943). The resem­ descriptions of the carpal and metacarpal ele­ blance is so close, in fact, that there is little doubt ments of G. arizonae, in which there is considerable that the South American Glyptodon is a close rel­ evidence of pathologic conditions, in at least one ative of the North American genus. Unfortu­ instance including fusion of two bones owing to nately only a single illustration of the front foot, an arthritic condition. in complete articulation, was included with his Nearly complete carpal series are known for G. description, and he provided no measurements. texanum and G. anzonae. At least proportionally the carpals oi Glyptodon, as PISIFORM (Table 17).—The pisiform is situated described by Vinacci Thul, appear to be very lateral to the proximal series of carpals. Geome­ similar, if not identical, to those of Glyptotherium. trically the pisiform of G. texanum (F:AM 95737, A pertinent statement by Vinacci Thul (1943) Figure 41) is a distorted tetrahedron, with two concerning the trapezium and trapezoid is that articular and two nonarticular surfaces. Its con­ these two bones are sometimes fused. His state­ cave proximal articular facet receives the lateral ment is unclear as to whether the fusion occurs facet of the ulna. On the anteromedial surface, at occasionally in Glyptodon, or whether it is charac­ right angles with the proximal facet, is the artic­ teristic for certain taxa (i.e., certain species of ular surface for the reception of the corresponding Glyptodon, or other non-Glyptodon taxa). We suspect surface of the cuneiform. The distopalmar surface his intention was that this fusion is a manifesta­ is concave and rugose. The anterolateral surface tion of ontogenetic variability within the genus is similarly rugose, forming an oblique angle with and that fusion of these two reduced bones is not each of the articulating surfaces and an acute significant as a diagnostic character. The reasons angle with the distopalmar surface. for this deduction are more fully discussed in the The pisiform of G. arizonae (USNM 10536, NUMBER 40 89

FIGURE 32.—Glyptotherium arizonae (USNM 10536): a, left scapula, lateral; b, right humerus, anterior; c, right radius-ulna-pisiform in articulation, medial. Glyptotherium floridanum (USNM 6071): d, right radius-ulna in articulation, medial; e, anterior;/, posterior of distal extremity of right humerus. (Bar = 40 mm.) 90 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Figure 32c) is proportionately longer in the prox- latter bone. This arrangement in the carpal series imodistal plane. Otherwise it is identical to that thus exhibits specialization for increased effi­ of G. texanum. ciency in weight support by the elimination of a Vinacci Thul (1943) aptly compared the shape portion of an articular surface between the cu­ of the pisiform of Glyptodon to that of a human neiform and the unciform and by the sharing of external ear or pinna. A similar comparison can weight transfer to the metacarpal IV by these be made for the pisiform of both of these North two bones of the carpal series. American species. The cuneiform of G. anzonae (USNM 10536, CUNEIFORM (Table 18).—The cuneiform of G. Figure 42, UMMP 38761, Figure 43), besides texanum (F:AM 95737, Figure 41) is a transversely being half again larger, differs principally in the elongated concavoconvex disc. The proximal ar­ nature of the articular facets. The proximal artic­ ticular surface for reception of the ulna is convex ular facet is less rounded on the anterior margin in the plane of elongation and slightly concave in and extends nearly to the lateral metacarpal the plantar-palmar plane, with the anterior bor­ facet. The latter is flattened, rather than convex, der forming a raised lip. The proximal surface and is situated on the lateral margin of the bone bears a rugose nonarticular region on the lateral instead of occupying a knoblike projection. The margin, the lateral extremity of which forms an distal articular surface is more deeply excavated. eminence, providing a laterally directed convex The medial articular facet (lunar facet) is some­ articular surface for reception of metacarpal V. what more extensive in G. arizonae, and it is This is the only articulation for the reduced fifth essentially flattened. The most significant differ­ digit, with loss of unciform contact, a modifica­ ence is the relatively poor development of the tion indicative of the digit's vestigial character. pisiform facet. The anterior (plantar) surface is convex in LUNAR (Table 19).—Largest of the proximal proximal view, concave in lateral aspect, provid­ carpal series, the lunar of G. texanum (Figure 41) ing a groove for the reception of a strong plantar occupies a central position, receiving the bulk of tendon crossing from the medial surface of the weight transfer from the radius. It is elongated carpal series to the reduced fifth digit. The medial anteroposteriorly and bears four articular facets: surface of the cuneiform is convex, articulating a proximal facet for articulation with the radius; laterally with the corresponding articular surface a distal facet for articulation with the magnum; of the lunar. The posterior (palmar) surface is and a medial and a lateral facet for articulation slightly convex, articulating with the pisiform. with the adjacent proximal carpals. The medial The distal articular surface is concave, repeat­ and lateral articular facets are both concave in ing in parallel the convexity of the proximal the anteroposterior direction and in the proxi- surface. The distal articulation with the unciform modistal plane. The result of these concavities on occupies the entire distal surface except for the either side is a constriction, centrally located in lateral region distal to the lateral knob. Here, proximal aspect. From this vantage, the expanded lateral to the articular surface for the unciform anteroproximal articular surface is broadly and directed slightly anteriorly, is a facet for rounded and in contour with the corresponding articulation with metacarpal IV. The cuneiform, articular surfaces of the adjacent cuneiform and therefore, articulates not only with the unciform, scaphoid. This facet receives the lateral and pos­ which in turn articulates with metacarpal IV, but terior portion of the concave articular surface of also with the proximolateral surface of the fourth the radius. The nonarticular rear portion of the metacarpal. In this respect, the transfer of weight proximal surface is concave, extending to a prox­ directly to the fourth metacarpal from the cunei­ imal posterior flange, which serves as a heel by form is a reflection of the reduction of the fifth prohibiting the transfer of weight to the posterior digit, with its concomitant loss of articulation border of the lunar. with the unciform and reduction in size of the The distal facet, articulating with the magnum, NUMBER 40 91 is concave in its entirety, except for the anterior The scaphoid of G. arizonae (Figure 42) differs surface, which is recurved, forming a convex only in the nature of the posterior tubercle. facet. Like the proximal surface, the distal facet Whereas this tubercle is knoblike in G. texanum, it is constricted in the middle, expanding toward is tapered in G. arizonae and extends proportion­ the posterior and anterior borders. The posterior ally farther posteriorly from the border of the surface of the lunar is directed perpendicular to proximal facet. the long axis of the arm, in a palmar direction. TRAPEZIUM (Table 21).—Together with the tra­ Along with the posterior portion of the proximal pezoid, the trapezium carries most of the weight surface, this rear expansion forms the prominent from the scaphoid, with which it articulates. The heel. trapezium of G. texanum (Figure 41) is a small, The lunar of G. arizonae (Figure 42) differs only reduced bone, the medial member of the distal in its larger size and a proportionally greater carpal series. It is wedge-shaped, firmly articu­ transverse extent of the proximal articular facet. lating on its lateral surface with the trapezoid The lateral border of this facet protrudes in an and with the proximomedial articular surface of extension above the plane of contact formed by metacarpal II. The facets for these two articula­ the abutting surfaces of the lunar and cuneiform. tions are separated by a subdued ridge on the The distal articular facet (magnum facet) is some­ lateral surface. The convex proximal articular what more deeply excavated in G. arizonae. These surface meets the scaphoid on its distomedial differences are attributable to variation and are margin. The trapezium, with its dual articulation not diagnostic. between the trapezoid and metacarpal II, is SCAPHOID (Table 20).—The scaphoid com­ firmly locked into position on the medial side of pletes the proximal carpal series, occupying the the wrist, and although reduced in size, it is an medial position and receiving the articular sur­ important element. In G. texanum the size of the face of the radius. The anterior (plantar) two- trapezium is subequal to that of the trapezoid. thirds of the scaphoid of G. texanum (Figure 41) is In the Arizona representative of G. arizonae wedge-shaped, with three symmetrically disposed (USNM 10536, Figure 42) the trapezium is sim­ facets, and a fourth, located on the anterolateral ilar to that of G. texanum; although considerably surface, which interrupts the symmetry. The larger absolutely, it is nevertheless subequal with proximal surface bears a convex facet for recep­ the trapezoid. The broken and fragmentary tra­ tion of the medial surface of the radius. The pezium of the Texas G. arizonae (UMMP 38761, broad convexity of this articular facet is in con­ Figure 43), however, is considerably smaller than tour with the corresponding articular surfaces of the trapezoid, although its construction appears the lunar and cuneiform. On the rear portion of to be similar. the proximal surface is a raised nonarticular boss, The reduced trapezium and trapezoid are to­ occupying approximately the posterior third of gether approximately equal in size to the remain­ the bone. This boss forms a heel similar to that of ing distal carpals. Vinacci Thul (1943) indicated the lunar and is confluent with it. The lateral that there is some variability in the trapezium- surface of the scaphoid is convex, articulating trapezoid construction, including occasional fu­ with the concave medial surface of the lunar. The sion. Whether his statement was intended for only distal surface of the scaphoid is directed medially the genus Glyptodon is unclear. Comparing species and bears three concave articular surfaces. The of the North American Glyptotherium, it appears medial articular surface receives the reduced tra­ that there is indeed a high degree of variability pezium; the middle articular surface, occupying within a single genus. Because of the reduction in half the distal facet, receives the convex surface these two bones, especially in the trapezium, the of the trapezoid; and the distolateral facet artic­ difference in size between the Arizona (Curtis ulates with the posteromedial surface of a projec­ Ranch) and Texas (Seymour) representatives of tion of the magnum. G. arizonae is not here considered diagnostic. 92 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TRAPEZOID (Table 22).—The trapezoid of G. cupies the anteromedial extremity of the element, texanum (Figure 41), like the trapezium, is wedge- the position vacated by the reduction of the shaped and reduced. It is situated within a trough trapezium; it extends medially over the antero­ of the proximal surface of metacarpal II, with medial tubercle of metacarpal II, fully extending which it articulates on its laterally convex, an­ (in articular position) to the level of the medial teroposteriorly concave distal surface. The proxi­ borders of the scaphoid and metacarpal II, re­ mal articulation with the scaphoid firmly locks spectively. Thus, the trapezoid of the Seymour G. the distal carpal series on the inside. On the anzonae is considerably larger than in the Arizona lateral border the trapezoid is nonarticular but representatives of both G. arizonae and G. texanum, meets the interlocking wedge projection of the occupying the entire proximal surface of the magnum, where the latter articulates with the metacarpal. There is no indication of pathologic scaphoid. The anterior (plantar) surface is irregularity, nor is there any indication of artic­ broadly expanded from the distal articular con­ ulation with the reduced trapezium. cavity. As discussed for the trapezium, these differ­ The trapezoid of the Arizona G. arizonae (Figure ences in the trapezoid are seemingly unimportant 42) is characterized by its proportionately greater as taxonomic characters. Because of the loss of transverse diameter; otherwise it is similar to that digit I, it might be expected that the medial of G. texanum. carpals would exhibit more structural variation The trapezoid of the Texas Seymour G. arizonae than do the other carpals. Since the other carpals (Figure 43) is considerably different from either of the Seymour glyptodont are virtually identical of the preceding. Rather than being reduced in with the Arizona representative, the disparities in size, the trapezoid approximates the size of the the trapezoid-trapezium complex could be attrib­ unciform and magnum. Thus, in the Seymour uted to variation. representative, there are three primary distal car­ MAGNUM (Table 23).—In proximal view, the pals and a reduced (almost vestigial) trapezium. magnum of G. texanum (Figure 41) is rectangular, The trapezoid in proximal aspect is triangular with the articular facets for the lunar and scaph­ with rounded and irregular sides; the apices are oid occupying the entire proximal surface. The posterior, anterolateral, and anteromedial. The lunar facet is sigmoid in the anteroposterior di­ proximal (scaphoid) facet is transversely convex rection, concave and depressed anteriorly, convex and fan-shaped. With respect to the long axis of and raised posteriorly. This facet receives the metacarpal II the proximal facet is directed pos- corresponding distal surface of the lunar. The teroproximally and somewhat laterally. The distal articular facet of the magnum articulates proximal (scaphoid) and distal (metacarpal) with the proximal trough of the metacarpal III. facets are posteriorly convergent, the minimum The facet is concave, with an anteroposterior proximodistal diameter occurring at the posterior ridge separating the medial and lateral portions apex, the maximum at the anterior border. The of the facet, these portions articulating with the minimum transverse diameter occurs along the lateral shoulders of the trough of the metacarpal. anterior border. The medial border bears two articular facets, The distal (metacarpal) articular facet is deeply a proximomedial articulation with the disto­ concave in the transverse direction and concave lateral process of the scaphoid and a distomedial in an elongate sulcus in the anteroposterior direc­ articulation for the proximolateral surface of the tion. The facet becomes transversely convex near metacarpal II. The facet for this metacarpal is the lateral border, to conform with the corre­ concave and directed more distally than medially, sponding facet of the metacarpal. Beyond the serving to distribute weight from the center of the medial extremity of the distal facet there is a carpal series toward the medial digit. massive tubercle-like extension. This process oc­ The lateral surface is fiat, abutting with the NUMBER 40 93 corresponding surface of the unciform. The artic­ less deeply excavated, and the anterior margin of ulation is parallel to the long axis of the forearm the proximal articular facet is less distinct in G. and serves further to interlock the carpal series, arizonae. rather than for weight transfer. The anterior METACARPALS.—There are three functional, (plantar) surface is rugose and is nonarticular. weight-bearing metacarpals (numbers II, III, and The magnum of G. arizonae (Figure 42) bears IV) and one reduced nonweight-supporting lat­ little distinction from that of G. texanum other eral metacarpal (number V). Metacarpal I is than greater size. It does, however, possess a absent, as is digit I. Metacarpal V and digit V scaphoid facet, a feature that is partially missing are greatly reduced; these elements present the in the G. texanum specimen as preserved. This greatest variability in the forefoot of North Amer­ facet is concave and is directed posteromedially, ican glyptodonts. receiving the corresponding facet of the scaphoid. Metacarpal II is by far the largest, with pro­ UNCIFORM (Table 24).—The wedge-shaped un­ gressive decrease in size to metacarpal V All are ciform of G. texanum (Figure 41) articulates exclu­ basically rectangular prisms in form. Proximal sively with the cuneiform on its proximal surface, interlocking contacts with each other and with which is convex and directed laterally at approx­ the carpals preclude extensive carpal-metacarpal imately 45°. This articular surface is confluent mobility. with the proximolateral (cuneiform) facet of met­ Vinacci Thul's (1943) descriptions of the acarpal IV Thus the unciform shares with the metacarpals of Glyptodon indicate close similarity distal surface of the cuneiform the articulation with those of the North American representatives with metacarpal IV. The distal articulation with and could serve as proxy for the descriptions the medial two-thirds of the proximal facet of the provided here. For completeness, however, the metacarpal is perpendicular to the long axis of metacarpals of the North American genera are the digit. The proximal and distal articulations described in full detail. The accompanying mea­ intersect on the lateral surface, owing to the surements (lacking in Vinacci Thul's description) oblique orientation of the proximal facet. The should provide a better foundation for future distal articular surface is saddle-shaped, receiving comparisons with South American taxa. the central portion of the fourth metacarpal. On METACARPAL II (Table 25).—The proximodis- the distomedial surface of the unciform is an tally elongate metacarpal II of G. texanum (Figure additional facet for the reception of the antero­ 33) bears four proximal articular facets: two on lateral projection of the third metacarpal. This the proximal extremity, and one on each side in articulation serves to transfer considerable force the anteroproximal position. In anterior view the from the ulna through the cuneiform and unci­ outline of the proximal end is rendered asymmet­ form to the middle digit. This articulation is rical by the proximally raised facet articulating parallel to the oblique proximal facet of the un­ with the medial distal surface of the magnum. ciform and is so arranged as further to lock the This articular surface is laterally flattened and carpal series. anteroposteriorly sigmoid, strongly convex ante­ The unciform of G. arizonae (Figure 42) is dis­ riorly and weakly concave on the posterior cur­ tinctive in having a proportionately greater an­ vature. It terminates in a weak posterior flange. teroposterior extent, producing a subrectangular This facet is bordered medially by a shallowly proximal outline. The bulk of this extension in­ grooved trochlear facet for articulation with the volves the relatively greater development of the distal concavity of the trapezoid. This facet, posterior region, forming a tubercle-like construc­ which extends neither anteriorly nor posteriorly tion on the posterior border in contrast to the as far as the magnum facet, receives the entire attenuated posterior region in G. texanum. The distal surface of the trapezoid. The posterodistal distolateral articular facet (for metacarpal IV) is tubercle of the trapezoid fits snugly into a circular 94 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 33.—Right metacarpal II oi Glyptothenum texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) depression on the posterior border of the articular reception of a heel-like prominence on the ante­ facet of the metacarpal. Posteriorly beyond this rior end of the carina of the first phalanx. concavity is an expanded nonarticular region ex­ A pair of facets for the two proximal sesamoid tending to the posterior border of the shaft, which bones of digit II is situated on the epiphysis near it intersects in an acute angle; this border in the distal end of the shaft on its posterior surface. posterior (palmar) view is directed obliquely The larger facet, articulating with a large sesa­ downward toward the medial side. moid, is centrally located; the smaller facet is Viewed laterally, the shaft on its proximal nearer the lateral border and appears as an in­ extension is expanded along its posterior border conspicuous notch on the posterodistal extremity and flattened on its anterior margin. On the of the lateral trochlear facet. Vinacci Thul (1943) proximal end of the lateral surface of the shaft is did not recognize these sesamoid facets. Descrip­ a small distolaterally directed facet for articula­ tions of these sesamoid bones follow the descrip­ tion with a proximomedial facet of the third tion of digit II. metacarpal. Similarly, there is a vertical facet on Metacarpal II of G. arizonae (Figure 42) is con­ the proximal end of the medial surface of the siderably more robust, especially in the transverse shaft for articulation with the terminal portion of and anteroposterior dimensions. Another distinc­ the distolateral facet of the trapezoid. Posterior tion is the relatively deeper excavation of the and distal to this facet is a pronounced tubercle. shaft on its medial surface distal to the trapezoid A proximodistal furrow extends along the medial facet. surface of the shaft, distal to the trapezoid facet A feature not related to ontogenetic differences and anterior to the tubercle, toward the distal is the greater proximodistal extent of the articular epiphysis. facet for metacarpal III on the lateral side of the The distal articulation with the first phalanx is shaft. This facet extends fully half the length of saddle-shaped, elongate in the anteroposterior the shaft in G. arizonae, whereas in G. texanum, the plane. The lateral surface, with a longer radius of facet extends barely one-fourth the shaft length. curvature, extends farther distally than the me­ In addition, this facet in G. arizonae is vertically dial, and the central groove is offset slightly to­ situated, whereas in G. texanum it is oblique, di­ ward the medial side. The central furrow termi­ rected somewhat distally as well as laterally. nates anteriorly in a slight depression, for the METACARPAL III (Table 26).—Metacarpal III, NUMBER 40 95

G. texanum (Figure 34), like metacarpal II, is a molateral quarter of the lateral side of the meta­ modified rectangular prism. In block proportions carpal. the third metacarpal is nearly equidimensional, A small medial facet provides an articular sur­ only slightly longer in the proximodistal plane face for metacarpal II. This facet is situated on than in the transverse and anteroposterior planes. the anteroproximal corner of the lateral face of It can be further distinguished from the metacar­ the epiphysis and is directed medially and slightly pal II by the greater development of the lateral posteriorly. metacarpal articular facet, which is restricted to The distal articular facet is broader and more the anterior half of the lateral side and extends to extensive than the proximal facet. It is saddle- a point past midshaft, and by the lack of a shaped, with a shallow valley located between depression posterior to the articular facet on the slightly elevated ridges on the sides. This facet is proximal extremity. rather more flattened than rounded, and in prox­ The principal carpal articulation is with the imal view the shape is more nearly symmetrical magnum; the facet for the magnum is saddle- than for metacarpal II, and the medial articular shaped, transversely concave and anteroposte­ flange only slightly less expansive and elevated riorly convex. This articular surface extends the than the lateral. A poorly developed anterior heel full length of the anteroposterior dimension and provides a weak stop for hyperextension of the occupies nearly the entire proximal surface except digit. A posterior central prominence with a pair for the anterolateral corner. The posterior surface of flanking sesamoids buttresses the posterior bor­ of the proximal articular facet projects somewhat der of the distal facet. The sesamoids are situated beyond the shaft. on either side of the posterior articular promi­ Two facets provide lateral articulation with nence, fitting neatly into posteriorly directed metacarpal IV and the unciform, respectively. facets in this position. The anterolaterally directed unciform facet is The shaft of the metacarpal III is equidimen­ situated on the proximal anterolateral corner. Its sional, with slight expansion at the epiphyses. A facet forms an obtuse angle, with the proximal marked tendinal sulcus on the medial surface saddle articulation with the magnum, and an extends along the shaft, terminating at the distal acute angle, with the more laterally situated facet epiphysis. for metacarpal IV. The metacarpal IV articular Metacarpal III of G. arizonae (Figure 42) is facet presents a curved surface, directed distopos- distinctly more robust. In block proportions this terolaterally above and laterally below in its distal element is roughly cuboid, rather than resembling portion. This facet occupies the anterior proxi- a rectangular prism. There are two other differ-

FIGURE 34.—Right metacarpal III oi Glyptotherium texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) 96 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

ences in the available specimens (which are not A convex facet for articulation with metacarpal attributable to proportionality or age): the unci­ III on the medial surface in an anteroproximal form articular facet in G. arizonae is shallowly position receives the laterally directed antero­ concave in contrast to the slight convexity in G. proximally concave facet of metacarpal III, serv­ texanum, and similarly, the metacarpal IV articu­ ing as a locking mechanism as well as transferring lar facet is somewhat less concave, and it has a weight toward the more lateral digit. greater anteroposterior diameter, comprising The distal articular facet is flattened, with only roughly two-thirds the diameter of the shaft. a slight anterior-posterior curvature, fitting METACARPAL IV (Table 27).—Metacarpal IV loosely into the concave proximal facet of the first of G. texanum (Figure 35) is by far the smallest of phalanx. A tubercle located in a central position the three functional anterior metapodials, its on the posterior border of the facet is the only length approximately half that of the third met­ irregularity on the distal surface. Posterior to this acarpal and less than half that of the second. Its prominence is an irregular, triangular-shaped transverse and anteroposterior diameters are only nonarticular region. Relative to the proximal ar­ slightly compressed relative to the other metacar­ ticulation, the distal facet is oriented in a slightly pals. posterior direction, so that in lateral or medial The proximal articular facet of metacarpal IV aspect the outline is wedge-shaped, the apex of is convex and elongated in the anteroposterior which is formed by extension of the proximal and direction, for reception of the concave saddle- distal facets along the nonarticular surfaces to shaped facet of the unciform. This facet is di­ their posterior intersection. rected somewhat laterally, producing a central The shaft of the fourth metacarpal is longest peak, evident especially in anterior aspect; from on the flattened anterior surface. A prominent this peak the superior surface of the convexity of tubercle is located on the posterior surface, and the facet extends anterolaterally, intersecting the an elongated one is situated parallel to the distal common border of this facet with the lateral facet articular facet on the lateral surface. for the cuneiform. The anterior border of the Metacarpal IV of 6'. anzonae (Figure 42) closely proximal facet is directed anteroproximally, while resembles that of G. texanum. It is proportionately the posterior border is slightly upturned, oriented longer in the transverse and anteroposterior di­ in a proximal direction. rections, imparting a relatively more robust char­ The slightly convex cuneiform facet is located acter. The articular facets differ from those of G. on the proximal anterior portion of the lateral texanum only in being relatively more extensive in face of the shaft, and it is confluent with the the transverse and anteroposterior planes. lateral facet of the unciform, which together re­ In marked contrast, metacarpal IV of G. flori­ ceive the distal facet of the cuneiform. danum (Figure 44c) presents several distinctions.

FIGURE 35.—Right metacarpal IV ol Glyptotherium texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) NUMBER 40 97

This element is more nearly a modified rectan­ (G. arizonae, Figure 42), can be explained either as gular prism, the posterior region not tapering representing the natural ontogenetic coalescence rearward. The posterior portion of the proximal of these two elements or the fusion at this rela­ articular facet is flattened and directed postero- tively immobile joint due to an age-related ab­ laterally. The medial facet for articulation with normality. The relative proportions of this ele­ metacarpal III is flattened and vertically oriented. ment approximate the combined proportions of The distal articular facet is even more distinctive. these two elements of G. texanum. The nonarticular Whereas in G. texanum and G. arizonae the distal surfaces, primarily on the shaft, are heavily ru­ facet is weakly convex, in G. floridanum it is ante­ gose, and there is at least the hint of a separation riorly flattened and posteriorly concave, with a between the two presumed fused bones repre­ distinct separation between the lateral and me­ sented by a groove at the posteromedial position dial portions of the posterior region. Moreover, on the shaft. Because of the close correspondence the posterocentral prominence is absent; in its with the combined proportions of the fifth met­ place there is a simple elevated region at the acarpal and first phalanx of G. texanum, and be­ juncture of the two portions of the posterior part cause of the groove suggesting the actual separa­ of the facet. Also, the orientation of the distal tion of these two elements, it appears that the facet is nearly parallel to the proximal facet, bone was formed by fusion of the two elements rather than posteriorly convergent. rather than by the loss of one and corresponding METACARPAL V (Table 28).—The extreme re­ increase in size and proportions of the other. duction of the digit V of Glyptotherium is reflected Whether the fusion is representative of intra- by the small metacarpal. Despite its extreme or interspecific variation or is representative of an reduction, however, this bone nevertheless dwarfs osteological abnormality related to old age is the first phalanx. The distal articular facet of the problematical. It seems most parsimonious to at­ metacarpal V of G. texanum (Figure 39) is flat, its tribute this fusion to old age for two reasons: (1) shape in distal aspect is roughly circular, and it there is a very close similarity with G. texanum of occupies only the anterior two-thirds of the distal all the other bones of the forefoot, with little, if extremity. The concave proximal articular facet any, reduction, and no fusion of any of the ele­ articulates with the convex prominence project­ ments; and (2) several of the other phalanges of ing from the lateral surface of the cuneiform and the manus of USNM 10536 are suggestive of is elongate in the anteroposterior direction. The osteological disease or abnormality (e.g., phalanx lateral articulation with the cuneiform imparts I, digit III). In another specimen of G. arizonae an oblique orientation of digit V with respect to (UMMP 38761), there is also considerable degen­ the other digits. The lateral projection of the digit eration of the bones of the feet, as already men­ and the lack of interlocking contacts of metacar­ tioned. Although it is possible to attribute the pal V, as in the other metacarpals, indicate the coossification of these two elements to injury, it lack of importance of the lateral digit in weight seems more likely to be a condition of old age, transfer during locomotion. perhaps arthritis. Although Melton (1964) reported the recovery Unlike the concave proximal articular facet of of metacarpal V for the Seymour glyptodont (G. metacarpal V of G. texanum, the proximal facet of arizonae), this element was not located. The only this bone is flat, as is the lateral facet of the remaining metacarpal V of G. arizonae (USNM cuneiform with which it articulates. Also, unlike 10536) available for study is abnormally coossi- the concave distal articular facet of the first pha­ fied, as described in the following section. lanx of G. texanum, the distal articular facet of this COOSSIFIED METACARPAL V, PHALANX I, DIGIT bone is convex. Similarly, this facet articulates V, MANUS.—The fusion of the fifth metacarpal with the shallow, concave, proximal facet of the with the first phalanx, digit V, of USNM 10536 terminal phalanx in marked contrast to the dis- 98 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY tinctly convex facet of the ungual phalanx, digit had erroneously identified the elements of digit V, of G. texanum. Because of these seemingly dis­ V as belonging to digit I. Thus the North Amer­ tinct differences between the articular facets of ican glyptodonts are similar to G. reticulatus in the phalanges of G. arizonae and G. texanum, the lacking digit I; digit II is the medial member of conclusion drawn above that the fusion of the the manus. It consists of three phalanges and two elements of USNM 10536 was due primarily three associated sesamoid bones (Figure 36). Dig­ to an age-related abnormality must not be ac­ its II and III are subequal in size; digit II is cepted with finality. Recovery of additional spec­ slightly longer and relatively more slender; digit imens may alter this conclusion. The remaining III is slightly shorter and relatively more robust. features of the fused metacarpal V-first phalanx Digit IV is smaller, but nevertheless stout and of USNM 10536 correspond to those of these two important as the lateralmost fully functional elements in combination of G. texanum. Measure­ digit. Digit V is extremely reduced. Digit II is ments for this coossified element are as follows: directed somewhat medially with respect to the maximum proximodistal diameter, 26 mm; max­ midline of the forelimb. The phalanges display imum anteroposterior diameter, 23 mm; maxi­ asymmetry associated with the position of the mum transverse diameter, 19 mm. digit. Phalanges I and II are short and robust. DIGIT II, MANUS.—Digit I, manus, is lacking in The ungual phalanx is elongate and was covered North American representatives. Vinacci Thul by an ungual sheath over most of its surface. (1943) established the existence of digit V in PHALANX I, DIGIT II, MANUS (Table 29).—The Glyptodon reticulatus, correcting earlier statements first phalanx of the second digit, manus, of G. (e.g., Burmeister, 1874) to the effect that digit V texanum (Figure 36), is a distorted concavoconvex was the missing toe. Burmeister, among others, disc. Two anteroposterior proximal facets receive

FIGURE 36.—Right digit II, manus, oi Glyptothenum texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) NUMBER 40 99

the distal trochlear facet of the metacarpal. Both proximomedially. The medial eminence bears a facets are deeply concave, the lateral deeper than smaller facet for the medial sesamoid, which is the medial. These proximal facets are separated the smaller of the two at this joint. This facet is by a flattened interarticular carina. Both proxi­ directed proximolaterally. Between the two pos­ mal facets are buttressed by posterior projections terior projections is a deep concavity, the center extending beyond the terminus of the central of which is penetrated by several small foramina. interarticular carina. The medial facet is directed A prominent tuberosity on the anterior surface slightly outward, its deepest recess situated pos­ is situated in line with a proximodistal plane, terior to the transverse midline. On its anterior passing through the proximal carina and the extension the medial facet is confluent with an distal groove, and at a position level with the anteromedial prominence, which extends proxi­ anterior border of the lateral facet. mally to the level of the posterior lip of the facet. Another tuberosity is situated on the medial This anteromedial facet forms a heellike articular side, extending from near the anterior border to prominence for articulation with the anterior de­ the posterior extension of the medial prominence. pression of the central furrow of the metacarpal Except for its larger size, phalanx I, digit II, II. manus of G. arizonae (Figure 42) is similar in all The lateral facet is deeper than the medial and respects. extends farther anteriorly, where at its terminus PHALANX II, DIGIT II, MANUS (Table 29).—The is its deepest depression. This facet is directed second phalanx of the digit II, manus, of G. parallel to the long axis of the digit and receives texanum (Figure 36), repeats the form of the first the larger lateral facet of the trochlea of the phalanx. It is slightly smaller than the first pha­ metacarpal II. lanx, especially in lateral aspect, in which the two The distal articular surface is formed by a pair posterior prominences do not project as far pos­ of confluent convex articular facets, together pre­ teriorly. The articular facets on both extremities senting in distal aspect a dumbbell outline, with differ from those of the first phalanx only in the medial facet somewhat smaller than the lat­ presenting a slightly shorter radius of curvature eral, and with only a slight anteroposterior central and in covering a smaller surface area. The region constriction. The lateral facet is the larger; it is corresponding to that forming the anterior prom­ directed parallel to the long axis of the digit. The inence of the proximal articulation of the first medial facet is directed distolaterally, its medial phalanx is subdued in phalanx II. Combined border forming a distinct lip with the nonarticu­ with the shorter radius of curvature of the distal lar medial side. The rockers of the distal extremity rockers, the subdued anterior prominence on the occupy only the anterior half of the surface. A proximal facet allows for a somewhat greater pair of nonarticular projections extend posteriorly freedom of movement over the surface of the from the facets in a broadly confluent fashion, distal rockers of the first phalanx. The facets for forming the undersurface of the buttress for the the first phalanx/second phalanx articulation are proximal articulations. The distal surface of these proximodistally flattened, however, prohibiting a buttresses is directed posteroproximally, resulting large degree of rotation in this joint. The distal in a lateral view of the phalanx, which is deeper articular surface of the second phalanx is saddle- anteriorly and tapers rearward posterior to the shaped and presents a greater freedom of rotation concavoconvex depression. in articulation with the terminal phalanx; the The two posterior projections each bear a facet pair of symmetrically disposed facets, in lateral for the sesamoids, which are situated on the pos­ view, turns through nearly 180°, whereas the terior surface of the joint between the metacarpal distal facet of the first phalanx is distinctly more II and the first phalanx. The lateral projection flattened and less extensive. bears the larger sesamoid facet, which is directed In addition, the second phalanx is distinct from 100 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY the first in possessing three large foramina on the gate, ventrally flattened, and dorsally arched in anterior nonarticular surface; by comparison, the the transverse plane. Two proximal facets artic­ foramina on the first phalanx are much smaller. ulate with the corresponding pair of rockers of The posterior surface receives a large sesamoid, the second phalanx. These facets are separated which is situated at the joint with the ungual by a subdued carina. Both facets are roughly phalanx. The distal articulation allows consider­ triangular in proximal aspect, the outer margins able motion in this joint, indicating a relatively tapering anteriorly toward the prominent ante­ large freedom of rotation for the terminal pha­ rior tubercle. The taper is more pronounced on lanx. The large sesamoid is testament to the larger the margin of the medial facet, imparting an freedom of motion. It is this joint at which the asymmetrical proximal aspect. Both facets are greatest freedom of movement in the digit occurs. deeply concave in lateral aspect, comprising a Indeed, there was enough flexibility for the ter­ chord of approximately 120°, the anterior ex­ minal phalanx at maximum extension to abut tremity of each facet reaching a higher (more with the first phalanx, at which position further proximal) position than the posterior border. The hyperextension was prohibited. Small wear facets anterior tubercle bears a small foramen, situated on the tip of an anterior prominence on the on the inner surface of the tubercle at the termi­ proximal facet of the ungual phalanx and on the nus of the carina. anterior border of the lateral facet of the first A pair of sesamoid articular facets is located on phalanx indicate that this articulation acted as a the posteroproximal border. These articulate with lock to prohibit further movement of the ungual the large single sesamoid at the terminal joint on phalanx over the surface of the second phalanx the palmar (posterior) surface. These facets are at hyperextension. directed posteroproximally and are separated by Phalanx II, digit II, manus, of G. arizonae and a distinct tendinal groove, a furrow that is mir­ G. floridanum, differ in several respects. In G. flori­ rored on the articular surface of the sesamoid. A danum (Figure 44(/), the proximolateral articular deep foramen is situated within the groove, and facet is deeply excavated, the distal articular facet another is located on the border of the groove is a nearly perfect cylindrical section, and the and the articular facet. A third foramen pierces marginal borders of the distal trochlea are nearly the medial sesamoid articular facet near its distal parallel, rather than anteriorly convergent. border. Phalanx II, digit II, manus, of G. arizonae (Fig­ The free end of the phalanx was covered in life ure 42), has a distinctly greater elevation of the by a horny sheath, apparently more hooflike than medial portion of the proximal articular facet. clawlike. This sheath covered the entire phalanx This elevation is less pronounced in G. texanum, distal to the epiphysis except for the subungual and it is essentially absent in G. floridanum. region. That the sheath was hooflike is indicated In one specimen of G. arizonae (UMMP 38761, by the flattened ventral (subungual) region of the Figure 43), an extension of the posterior tubercle bone and the highly arched dorsal region. The appears to be the fused sesamoids, ankylosed to sheath conformed to the surface of the bone, and the phalanx in their appropriate positions and at the distal end it terminated bluntly; the sheath with each other. This construction is a manifes­ did not extend far beyond the distal tip of the tation of the degenerate condition in the front bone. foot of this individual, owing to an apparent In F:AM 95737, the juvenile individual, the arthritic malady. Other elements of this foot are subungual base is roughly semicircular in outline, even more severely affected. although there is the hint of an apex near the UNGUAL PHALANX, DIGIT II, MANUS (Table distal border. Its margins are not well defined, in 30).—The terminal phalanx of digit II, manus, of contrast to the extensions of the subungual base G. texanum (Figure 36), is proximodistally elon­ into a short hood process in older individuals. NUMBER 40 101

Two large subungual foramina are situated on three in Doedicurus. There are three sesamoids the medial and lateral borders of the subungual each on digits II and III in Glyptotherium, a prox­ base, respectively. The axis passing through the imal pair at the metacarpal-phalanx joint and a centers of these foramina is roughly parallel to single distal one at the terminal joint; one sesa­ the transverse axis of the proximal articular moid only on digit IV; and one sesamoid bone facets. Both penetrate deeply into the bone to­ on digit V. Thus there are eight digital sesamoids ward the center. in the manus of the North American representa­ The claw process bears numerous vascular fo­ tives. These bones are subject to considerable ramina piercing the otherwise smooth outer sur­ pathologic degeneration and/or fusion, as noted face. In cross section the claw process is half- earlier, and in the respective descriptions of the ellipsoid, with the long axis passing from dorso­ individual elements. lateral to anteromedial. In lateral aspect the pha­ The proximal pair of sesamoids of digit II, lanx is wedge-shaped, forming a nearly perfect manus, in G. texanum (Figure 41), are both irreg­ isosceles triangle distal to the epiphysis. Viewed ular in shape, approximating a distorted sphere dorsally, the lateral and medial borders both in configuration. The medial sesamoid is the curve toward the lateral side. The medial border larger, approximately twice the size of the lateral. has a shorter radius of curvature, producing a Both articulate with distinct sesamoid facets on terminal apex in direct line with the lateral bor­ the palmar surfaces of the metacarpal and pha­ der of the proximal articular facet. lanx at the joint. The medial sesamoid is centered The terminal phalanx of the digit II, manus, of approximately in line with the middle furrow of G. arizonae (Figure 42), differs in few details other the metacarpal and the carina of the phalanx. than its larger size. The subungual hood in the The lateral sesamoid is situated in line with the adult specimens is more well developed, extend­ lateral rocker of the metacarpal. ing as a projection around the circumference of The proximal sesamoids, digit II, manus, of G. the subungual base and barely extending on the arizonae (Figure 42), are distinctly more robust. lateral surface to the anterior tubercle. The single They are subequal in size, the medial only slightly characteristic distinguishing G. arizonae is the po­ larger than the lateral. The medial sesamoid bone sitioning of the subungual foramina. The axis more closely approximates equidimensional pro­ passing through the centers of these foramina is portions than the lateral sesamoid, which is rather obliquely situated with respect to the transverse elongate. axis of the articular facet. These two axes, if The distal sesamoid, digit II, manus, of G. extended, are medially convergent. texanum (Figure 41), is three times as large as the Ungual phalanx, digit II, manus, of G. flori­ medial sesamoid bone of the proximal pair. It is danum (Figure 44/), is similar in most respects to transversely elongate and tapers posteriorly from G. anzonae (Figure 43), although its proportions the dorsoventral and mediolateral surfaces to are less robust. The subungual foramina are form a broad arc in palmar aspect. Two distinct obliquely situated. The articular facet is not ex­ articular facets meet at a right angle on the tensive, the chord representing approximately anteroproximal (proximoplantar) border. The 90°. The shorter anteroposterior extent of the larger anterior facet articulates with the ungual articular facet is due primarily to a relatively phalanx, the smaller plantar (dorsal) facet with underdeveloped anterior tubercle. the second phalanx. A distinct tendinal groove is SESAMOID BONES, DIGIT II, MANUS.—Vinacci centrally situated at the juncture between the Thul (1943) did not recognize or describe any facets, separating the plantar (dorsal) facet into digital sesamoids in the manus of Glyptodon reticu­ a pair of smaller facets. latus. Burmeister (1874) found only one digital The distal sesamoid of G. anzonae (Figure 42) is sesamoid bone in the manus of Glyptodon asper and similar to that of G. texanum. The single distin- 102 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY guishing feature is the development on the non­ disc, with proximal concave and distal convex articular surface opposite the ungual facet of a articular facets. Two posterior projections provide distinct broad sulcus that extends anteroposte­ the rear buttress for the posterior extension of the riorly, with raised margins on the lateral and proximal facet. This concave articular surface is medial borders. divided in the sagittal plane by an unobtrusive Measurements for all of the sesamoid bones of raised area. On each side of this central rise the the manus are provided in Table 37 articular facets are shallowly excavated, receiving DIGIT III, MANUS.—Digit III is the central the two weakly developed distal rockers of meta­ member of the three functional digits of the carpal III. This facet is broadest in the transverse manus. It consists of three phalanges and three dimension near the anterior border; tapering pos­ sesamoid bones. The phalanges are more sym­ teriorly, its narrowest transverse diameter occurs metrical than those of the adjacent digits, and at the posterior extremity. The anterior border of they are larger and more robust. The middle digit the proximal facet is slightly recurved, forming a is slightly shorter than digit II, but it is distinctly heel to interlock with the metacarpal for limiting more robust, participating as the strongest hyperextension. weight-bearing member of the front foot. Phalan­ The distal articular facet is convex and covers ges I and II are short and robust, as in the a less extensive area than the proximal facet. It is adjacent digits, and the ungual phalanx was cov­ divided by a shallow central furrow, along which ered almost entirely by an ungual sheath. the articular surface is constricted anterolaterally, PHALANX I, DIGIT III, MANUS (Table 31).— producing an hourglass shape in distal view, the Phalanx I, digit III, manus, of G. texanum (Figure lateral facet slightly larger than the medial. Nei­ 37), is compressed in the proximodistal plane to ther facet extends posteriorly to include the distal the extent that it is an irregular concavoconvex surface of the posterior tubercles. The curvature

FIGURE 37.—Right digit III, manus, oi Glyptotherium texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) NUMBER 40 103 of the distal facet is more flattened than that of The second phalanx, digit III, manus, of G. the proximal facet. The distal facet is directed at arizonae (Figure 42), is distinct only in that the an angle somewhat posterior to the direction of proximal and distal facets are separated into me­ the proximal facet. A large medial tubercle and dial and lateral portions by poorly defined non­ a small lateral one are situated on the anterior articular anteroposterior furrows. Because the ar­ facet in the same sagittal planes as the medial ticular facets are otherwise identical in shape and and lateral pairs of articular facets. proportions to the juvenile G. texanum, this sepa­ Each of the posterior tubercles bears a poste- ration into lateral and medial regions in the adult roproximal facet for the reception of a sesamoid appears to be due simply to ontogenetic differ­ in the metacarpal-phalanx joint. ences. Except for size differences in the adult speci­ UNGUAL PHALANX, DIGIT III, MANUS (Table mens of G. arizonae (Figure 42) compared with the 32).—The ungual phalanx, digit III, of G. texanum juvenile specimen of G. texanum, phalanx I, digit (Figure 37), is the largest terminal element in the III, manus, is similar in all respects. USNM 10536 manus, its length nearly double that of the ungual bears extremely rugose nonarticular surfaces, and phalanx of digit IV and exceeding that of digit II the articular facets are rough and heavily worn. by more than 10 percent. The ungual phalanx of The distorted and misshapen posterior tubercles the third digit can be distinguished from that of in this specimen indicate a pathological condi­ digit II by its flattened and more symmetrical tion, probably an arthritic malady. cross section and by its nearly symmetrical ante­ PHALANX II, DIGIT III, MANUS (Table 31).— rior outline. Except for its larger size, this element Phalanx II, digit III, manus, of G. texanum (Figure is similar to the third phalanx, digit IV, in shape 37), is a smaller version of the first phalanx, with and cross-sectional and anterior aspects. A pri­ several distinguishing features. The proximal ar­ mary distinction, however, between the ungual ticular surface is more symmetrical in shape and phalanges of digits III and IV is the rotation of the medial and lateral portions are separated by the proximal articular trough for reception of the a deeper excavation between the posterior tuber­ second phalanx. In digit III this medial rotation cles. A small constriction on the anterior border of the articular facet of the ungual phalanx is not of the facet contrasts with the straight margin on as pronounced, so that in anterior aspect the the corresponding facet of the first phalanx. outline of the proximal end produces only a The distal articular surface has a shorter radius downward-directed slope toward the lateral mar­ of curvature, and the central constriction is more gin from the anteroproximal border, rather than pronounced. In sagittal section this facet resem­ a distinct trough with a raised margin on the bles a symmetrically disposed half hourglass fig­ lateral side as in the ungual phalanx of the digit ure, in contrast with the flattened aspect of this IV facet on the first phalanx. The medial portion of The proximal articulation consists of two con­ the distal facet on the second phalanx is slightly fluent concave facets separated by an indistinct more flattened than the lateral, and it extends to ridge. Both facets describe an arc of approxi­ a more proximal position at its anterior terminus, mately 120°, with most of the curvature anterior producing a slight asymmetry in the hourglass to the midline. The posterior terminus of the shape. articular facets does not extend beyond the level Like the first phalanx, the second is proximo­ of the subungual process, while the anterior ex­ distally compressed; its proximodistal dimension tension continues proximally, resulting in a pos­ approximates that of the first phalanx. There are terior orientation for the facets at their anterior no tubercles on the anterior face, nor are there border. The medial facet extends farther proxi­ any sesamoid articular facets on the posterior mally than the lateral; the latter possesses a tubercles. greater transverse diameter. A plane passing 104 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY through the low median ridge projects through nounced, and the anterior proximal tubercle is the anterior face of the claw process near the divided by a shallow groove, producing a larger medial border, indicating a rotational motion in medial prominence and a smaller lateral one. The the anteroposterior plane, which is oblique to the articular facets are weakly divided by an indis­ transverse axis of the roughly symmetrical claw tinct nonarticular area. process. The ungual phalanx, digit III, manus, of G. The claw process includes all of the distal floridanum (Figure 44g), exhibits a somewhat portion of the phalanx except for the semicircular greater departure from the features of G. texanum, prominence on the proximal end of the posterior especially with respect to the articular facet. As face formed by the subungual process. The prox­ in the ungual phalanx of the second digit, the imal borders of the claw process are indistinct in articular facet is less extensive and rather flat­ F:AM 95737, presumably because of the young tened, again primarily due to the relatively age of the individual. The cross section is an underdeveloped anterior proximal tubercle. anteroposteriorly compressed and roughly sym­ SESAMOID BONES, DIGIT III, MANUS.—Three metrical lens shape. On the posterior face the sesamoid bones are located on the palmar surface medial border is flattened near the proximal end. of the third digit: two small ones at the metacar- The two subungual foramina, located on either pal-phalanx joint and a large one at the terminal side of the subungual process, pass deeply into joint. the bone in a distal direction toward the midline. Of the proximal pair of sesamoids in G. texanum The diameter of the lateral foramen is approxi­ (Figure 41), the medial is slightly larger than the mately half that of the medial, a feature serving lateral. Both are irregularly spherical in shape, to further distinguish the ungual phalanx of digit with two poorly developed, flattened, articular III from those of the adjacent digits, in which the surfaces, a metacarpal facet, and a first phalanx diameters of the subungual foramina are approx­ facet. Only the lateral of the proximal pair of imately equal. sesamoids is known for G. arizonae (USNM 10536, Articular facets for the large posterior sesamoid Figure 42). As in G. texanum, this element is irreg­ on the joint between the second and third pha­ ularly spherical in shape, somewhat elongated in langes are located on the posterior border of each the transverse direction. The articular facet for proximal articular facet. These two sesamoid the first phalanx is likewise transversely elon­ facets are continuous with the proximal articular gated, and its surface is weakly sigmoid in the facet, forming a rounded, convex terminus for the transverse plane. posterior border of the larger facet. The large distal sesamoid of G. texanum (Figure Ungual phalanx, digit III, manus, of G. anzonae 41), located at the terminal joint, is approxi­ (Figure 42), exhibits a massive development in mately four times larger than the proximal ones. adult individuals, especially with respect to the Its shape is more well defined, transversely elon­ overgrowth of the subungual hood, which extends gate with a distinct second phalanx articular facet on the medial side to the anteromedial corner of on its proximal surface along the anterior border, the proximal extremity and on the lateral side and two somewhat less distinct terminal phalanx only to the margin. Thus the sheath covered the articular facets on the anterior face. Compared entire anterior surface and the greatest portion of with the terminal sesamoid of digit II, this ele­ the other surfaces as well. By inference, the same ment is slightly narrower and the proximal artic­ statement can be made for the adults of G. tex­ ular facet is narrower and more elongate in the anum. transverse plane. A minor difference between The shape and proportions of G. arizonae closely these elements occurs in the shape of the proximal approximate those of G. texanum. The flattened articular facet: broadly convex in the sesamoid of medial surface of the claw process is rather pro­ digit II, more distinctly divided into three sepa- NUMBER 40 105 rate portions in the terminal sesamoid of digit III. toward the medial margin. This surface is some­ The distal sesamoid bone of G. arizonae (Figure what asymmetrical, with a subdued central sulcus 42) closely resembles that of G. texanum. The only located medial to the midsagittal plane. The marked distinction is the exaggeration of the medial margin forms an acute angle with the tendinal sulcus on the nonarticular surface op­ shaft, while the lateral margin of the facet forms posite the articular facet for the ungual phalanx. a broadly obtuse angle with the lateral surface of DIGIT IV, MANUS.—The annular digit is the the shaft. Unlike the proximal articular surface, most lateral of the three functional weight-bear­ the distal facet does not extend posteriorly to ing digits of the manus (Figure 38). Digit V, include the entire region beneath the two poste­ located lateral to the fourth, is reduced in size rior tubercles formed by the posterior tendinal and length and is essentially vestigial. The fourth sulcus. digit is directed laterally from the midline and There is little shaft on the first phalanx, its participates as the lateral member of the manus. proximodistal length being so compressed that its The lateral orientation, approximately 45° from shape is that of a concavoconvex discoid with a the medial digit, is achieved through distortion broadly oblate outline. A tubercle is situated of the unciform in the distal carpal series. The laterally on the shaft, the only notable feature on distal articular facet of the unciform, articulating the nonarticular surface of this element. There with the fourth metacarpal, is rotated laterally, are no evident sesamoid articular surfaces. so that with respect to a sagittal plane passing Phalanx I, digit IV, manus, of G. arizonae (Fig­ through the proximal facet, the distal facet is ure 42), bears no significant distinction from that directed distolaterally and is restricted to the of G. texanum. In USNM 10536, the posterior lateral half of the element. tubercles are distorted and suggestive of disease, The fourth digit is the smallest of the three as in phalanx I, digit III. functional toes of the manus; each phalanx is PHALANX II, DIGIT IV, MANUS (Table 33).— somewhat reduced in size. There is only one Phalanx II, digit IV, manus, of G. texanum (Figure known sesamoid, a large palmar one at the ter­ 38), is not as extensively compressed in the prox­ minal joint. No proximal sesamoids at the meta- imodistal direction as phalanx I. Its articular carpal-phalanx joint were recovered with F:AM facets display a shorter radius of curvature, allow­ 95737, the only known North American specimen ing for greater rotational motion than in the with associated sesamoids, nor are there any facets metacarpal-phalanx joint. In proximal aspect the indicating their existence. It is possible, however, concave articular facet is butterfly-shaped, with that these sesamoids were unossified at the time the medial expansion rather narrower than the of death for this juvenile individual and would lateral. The lateral portion of the proximal facet have ossified at a later age. terminates anteriorly at the highest point on the PHALANX I, DIGIT IV, MANUS (Table 33).— anterior border. Phalanx I, digit IV, manus, of G. texanum (Figure The distal articular facet is trochlear, with the 38), is extremely compressed in the proximodistal medial facet separated from the lateral facet by direction. The proximal articular facet is bilat­ a shallow constriction. Both portions of the facet erally symmetrical about the sagittal plane. Its possess a tighter radius of curvature than the slightly concave surface is turned up on the corresponding surface of the first phalanx. The straight anterior border and to a lesser extent on medial portion is longer and narrower than the the rounded lateral borders. A deeply penetrat­ lateral in the anteroposterior direction. The lat­ ing, tendinal sulcus interrupts the posterior mar­ eral portion of the facet is transversely wider and gin. is rendered somewhat asymmetrical by having a The distal articular facet is the complement of broader surface on the posterior margin than on the proximal, its convex surface offset slightly the anterior. The medial portion of the facet is 106 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FICURE 38.—Right digit IV, manus, oi Glyptothenum texanum (F:AM 95737): a, anterior; b, lateral; c. posterior; d, medial. (Bar = 20 mm.) directed laterally, departing farther than the me­ the digit with respect to the midline of the front dially directed lateral facet from the normal sag­ limb. ittal orientation, thus imparting a marked asym­ The proximal articular facet is deeply concave metrical outline in anterior aspect. The central in the anteroposterior direction, forming a right sulcus is situated in a position medial to the triangle in proximal view, with legs of equal sagittal plane. dimension formed by the medial and posterior The shaft is extremely compressed, as in the borders, and the hypotenuse by the anterolateral first phalanx; no prominent tubercles mark its margin. This is the cross-sectional aspect as well, nonarticular surface. which obtains for the length of the element, The posterior margins of the two portions of becoming more rounded and anteroposteriorly the distal facet articulate with a large sesamoid compressed toward the free end. on the palmar surface of the terminal joint. This The free end of the terminal phalanx, like that articulation is confluent with that for the terminal of the other ungual phalanges, was encased by a phalanx. horny sheath, except for the raised subungual Except for greater size and more extensive region in the posterior proximal position. As with rugosity, in the adult specimens compared with the other terminal phalanges of the juvenile spec­ the juvenile of G. texanum, this element of G. imen F:AM 95737, the proximal borders of the arizonae (Figure 42) is similar in all respects. claw process are indistinct, presumably because UNGUAL PHALANX, DIGIT IV, MANUS (Table of the young age. Two large subungual foramina 34).—Ungual phalanx, digit IV, of G. texanum penetrate the bone on the medial and lateral (Figure 38), is the most asymmetrical phalanx of sides, respectively, of the subungual process. The the digit. It is the smallest of the three functional anterolateral surface is flattened, forming a sharp terminal phalanges and its cross-sectional aspect angle with the posterior border and a somewhat exhibits the farthest departure from bilateral sym­ more rounded angle with the medial border. metry as a reflection of the lateral orientation of The articular motion of the ungual phalanx is NUMBER 40 107

FIGURE 39.—Right digit V, manus, and metacarpal V oi Glyptothenum texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.)

in the sagittal plane passing through the proximal facet and the raised anterior tubercle. The leading edge of the phalanx is the anteromedial border, in keeping with the lateral position of the digit. A sesamoid articular facet is situated on the posteroproximal surface, forming an angle at the common border with the proximal facet. The terminal phalanx, digit IV, manus, of G. floridanum (Figure 44/z), differs in several impor­ FIGURE 40.—Right palmar sesamoid of Glyptotherium texanum tant features. As with the other ungual phalanges (F:AM 95737): plantar surface. (Bar = 20 mm.) of this species, the proximal articular surface is more extensive in the transverse direction; the surface of the claw process, in G. floridanum this maximum transverse diameter occurs through the rotation is less acute, forming an angle of roughly central low point of the facet rather than through 60°. the sesamoid facet on the posteroproximal border. The subungual hood of the adult specimen of Moreover, the radius of curvature of this facet is USNM 6071 (G. floridanum) is more well defined longer, producing a more flattened articular sur­ than that of the juvenile specimen of G. texanum, face. The reduced curvature of the proximal facet extending to the anteromedial and lateral corners is accentuated by the relatively underdeveloped of the proximal end. The subungual base is rela­ anterior tubercle. The proximal facet is more tively larger. The lateral subungual foramen is distinctly divided into lateral and medial por­ distinctly smaller than the medial in contrast to tions, and the outer shape of the proximal facet those of G. texanum, which are approximately is less triangular and somewhat suggestive of a equal in dimensions. The posterior (palmar) sur­ figure eight (8) shape. face of the claw process is transversely concave The most distinguishing feature in G. floridanum rather than convex. is the lesser degree of axial rotation of the articular The phalanx is unknown for either the Sey­ facet. Whereas in G. texanum the rotation of the mour or the Curtis Ranch representatives of G. ungual phalanx formed an angle of roughly 45° arizonae. with respect to the flattened anterolateral outer SESAMOID BONE, DIGIT IV, MANUS (Table 108 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FICURE 41.—Articulated right manus oi Glyptothenum texanum (F:AM 95737): a, anterior, with pisiform above (left), and palmar sesamoid above (right); b, posterior, including digital sesamoid bones, with palmar sesamoid above (left) and pisiform above (right). (Bar = 5 cm.)

FICURE 42.—Articulated right manus oi Glyptothenum anzonae (USNM 10536): a, anterior; b, posterior. (Bar = 40 mm.) NUMBER 40 109

37).—There is no indication of proximal sesa­ glyptodonts and their close South American rel­ moids at the metacarpal-phalanx joint, digit IV, ative is therefore only provisionally important as », ..... manus in known specimens of Glyptotherium. The a taxonomic distinction. single large sesamoid at the terminal joint of digit Digit V articulates with metacarpal V on the IV in G. texanum (Figure 41) is the smallest of the lateral edge of the foot and is directed laterally terminal sesamoid bones. Like the others, it is with respect to the long axis of the foot. transversely elongate, with articular facets for the PHALANX I, DIGIT V, MANUS (Table 35).—Pha­ second and ungual phalanges meeting at approx­ lanx I, digit V, manus, of G. texanum (Figure 39), imately a right angle on the anteroproximal bor­ is the smallest element of the forefoot except for der. This bone is the least distinctive of the ter­ the sesamoid bones. Its proximal aspect is semi­ minal digital sesamoids of the manus. circular, rounded along the anterior border. The This element was not recovered with any other flattened proximal articular facet bears the same specimens. shape, occupying the entire proximal extremity. DIGIT V, MANUS.—As discussed in the descrip­ The distal articular facet is transversely concave, tion of digit II, manus, digit V was not universally and in distal aspect it is likewise semicircular. recognized in the South American genus Glyptodon The distal facet articulates with the convex facet until Vinacci Thul (1943) established its exist­ of the second (terminal) phalanx; a small spheri­ ence. Some earlier authors (e.g., Burmeister, cal sesamoid bone is situated at the posterior 1874) had erroneously assigned the bones of digit border of the terminal joint. V to digit I and claimed that digit V was absent. Like the South American Glyptodon, the North American representatives have a well-defined digit V, but it is very reduced and may be consid­ ered as functionally vestigial. In Glyptothenum, digit V (Figure 39) consists of only two phalanges and a sesamoid bone at the terminal joint. This determination is at variance with Vinacci Thul's (1943) contention that Glyp­ todon reticulatus has three phalanges. He main­ tained that phalanges I and II were lost in his specimen and that the two elements he described were metacarpal V and phalanx III. His justifi­ cation for the belief that G. reticulatus has three phalanges was not stated, although it is possible that he based his statement on knowledge of other specimens. In the particular specimen he de­ scribed, there were indeed only two elements in this digit (including the metacarpal); his textual descriptions and the figure of both the metacarpal and the terminal phalanx indicate close similarity to phalanges I and II described here for North American representatives. It therefore appears possible that G. reticulatus has only two phalanges in digit V and that Vinacci Thul's statement to FIGURE 43.—Articulated left manus of Glyptotherium arizonae the contrary for G. reticulatus was erroneous. This (UMMP 38761), with digits I and II and phalanges I and II assumed difference between the North American of digit IV, anterior. (Bar = 40 mm.) no SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

There is no indication of fusion between the phalanx I with metacarpal V in one specimen of first and second phalanges of this digit. There is G. arizonae was discussed above. It does not appear no middle phalanx and no indication of its incor­ that phalanx I, digit V, of G. arizonae, differs poration by fusion into either of the other pha­ significantly from that of G. texanum. langes. PHALANX II (UNGUAL PHALANX), DIGIT V, Description of the abnormal coossification of MANUS (Table 36).—The ungual phalanx of the

FIGURE 44.—Isolated elements of manus of Glyptotherium floridanum (USNM 6071), anterior aspects: a, right metacarpal II; b, right metacarpal III; c, left metacarpal IV; d, second phalanx of digit II, right; e, second phalanx of digit III, right;/, ungual phalanx of digit II, right; g, ungual phalanx of digit III, right; h, ungual phalanx of digit IV, right. (Bar = 20 mm.) NUMBER 40 111 digit V, manus, of G. texanum (Figure 39), is between the subungual foramina as an eminence transversely compressed and proximodistally projecting from the subungual base. It bears a elongate. The small proximal articular facet is flattened surface of insertion at its extremity for convex in the transverse direction and semicir­ the tendon of the digit. cular in proximal view. The precise limits of the The second (terminal) phalanx of the fifth subungual base are indistinguishable from the digit, manus, of G. anzonae (Figure 42), differs in claw process. Two subungual foramina penetrate several important respects. First, as discussed pre­ deeply into the subungual base; the lateral fora­ viously (under the heading "Coossified Metacar­ men is slightly larger than the medial. In lateral pal V, Phalanx I, Digit V, Manus"), the proximal and medial views the portion of the bone distal articular facet is weakly concave in the antero­ to the epiphysis bears the shape of an isoceles posterior direction, rather than transversely con­ triangle, the apex at the terminus of the claw vex. Another important distinction is the dispo­ process. The smooth surface of the claw process sition of the subungual base. In USNM 10536, is distinct from the pitted surface of the subungual the subungual base is situated entirely at the base, indicating that the free end was encased in posteroproximal position of the medial surface of a horny sheath. the transversely compressed claw process. Accord­ A distinct posteroproximal tubercle is situated ingly, the "lateral" subungual foramen is situated

FIGURE 45.—Pelvis oi Glyptotherium texanum (AMNH 10704): a, dorsal; b, right side (reconstructed portions hatched); c, posterior; d, detail of acetabular region, ventral sightly enlarged. (Bar = 20 cm.) 112 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

in direct line with the posterior margin, and the into a single structure are the lumbar vertebrae, "medial" foramen is located at the proximocen- the sacral vertebrae, and the paired ilia, ischia, tral position on the medial side. Thus the terminal and pubes. The axial component includes two phalanx, digit V, of G. arizonae, bears a distinct primary contributions, the lumbar tube, consist­ asymmetry in marked contrast to the symmetrical ing of five to nine fused vertebrae (in North disposition of the subungual features of G. tex­ American glyptodonts); and the sacral tube, anum. which includes eight or nine vertebrae of which SESAMOID BONE, DIGIT V, MANUS (Table 37).— three or four are united with the ilia and one or The single sesamoid of digit V, manus, of G. two with the ischia. The lumbar tube is solidly texanum (Figure 41), is the smallest sesamoid bone ankylosed to the sacral tube. Fusion is so extensive of the forefoot. It is approximately spherical in in the axial portion of the glyptodont pelvis that shape and occupies a position on the posterior accurate determination of vertebral counts is border of the proximal extremity of the terminal nearly impossible for the lumbar tube and the phalanx, at the terminal joint. free sacral vertebrae ("free" in the sense that they The sesamoid of the digit V, manus, of G. are not united with the appendicular skeleton; anzonae (Figure 42), contrasts with that of G they are serially ankylosed in the sacral arch and texanum in its tetrahedral shape. There are two are therefore not "free" in the sense of maintain­ articular facets. One is a large, slightly concave ing discrete separation from the serially adjacent surface, which articulates and rests on a weakly vertebrae). convex surface posterior to the proximal articular The appendicular portion of the pelvis includes facet of the terminal phalanx. The other facet is the greatly expanded ilia and ischia and the smaller, meeting the larger facet at approximately relatively small pubes, which in at least one North a right angle. The smaller facet is transversely American species (G. cylindricum) are fused at the concave and elongate and articulates at the prox­ midline in a solid union. For the remaining spe­ imodistal surface of the fused metacarpal V, first cies, for which the extremities of the pubes are phalanx. There are two nonarticular surfaces; one not preserved, there was probably cartilaginous is a triangular-shaped surface on the lateral side, or ligamentous contact, if not actual fusion, at the other an irregular surface at the proximopos- the midline. Because of the small sample of pelves terior position. known for North American glyptodonts, it cannot PALMAR SESAMOID, MANUS.—In addition to the be determined whether there is sexual dimorph­ digital sesamoid bones, there is a large palmar ism in the construction of the pelvis. Brown's sesamoid in the front foot. In bulk this element (1912) classification was weighted heavily toward approximates that of metacarpals I and II and is the construction of the pelvis, in part separating thus a prominent bone of the manus. In both G. the two families Glyptodontidae and Sclerocalyp- texanum (Figures 40, 41) and G. arizonae, the pal­ tidae on the basis of differences in the pubes. As mar sesamoid is a nondescript, irregularly elon­ discussed below, the distinctions in the pelves gate bone. The surfaces are generally extremely cannot be relied upon to exhibit a constant set of rugose, although the sides are rather smooth, and characters, and in the present state of knowledge in several regions there are distinct striations. taxonomic determination therefore must not rely Length-width-height measurements are 62 mm, heavily on the nature of the pelvis. 38 mm, and 21 mm for G. arizonae (USNM 10536) Nearly complete pelves are known for all four and 44 mm, 24 mm, and 22 mm for G. texanum North American species. Osborn (1903) merely (F:AM 95737), respectively. mentioned the recovery of the pelvis of AMNH 10704 (G. texanum type specimen), without de­ PELVIS scription or illustration. Brown (1912) compared The glyptodont pelvis is remarkable for the this specimen with the one he described and degree of fusion of its component bones. United figured for AMNH 15548 (G. cylindricum type NUMBER 40 113

specimen), concluding that the differences in discussed under "Systematics," where a detailed these two pelves, among other distinctions, indi­ comparison with armadillo pelves is advanced to cate generic (and familial) separation. Another account at least in part for these variations in pelvis of G. texanum (F:AM 95737), from Arizona, vertebral count. Variation in the construction of is reported here for the first time. the components of the armadillo pelvis is consid­ Hay (1916) described fragmentary portions of ered in the same discussion, and an evaluation the pelvis of USNM 6071 (G. floridanum), includ­ and emended diagnoses for the pelves of the ing an acetabulum and the ischiosacral vertebrae. North American representatives are there pro­ Lundelius (1972) reported a nearly complete pel­ vided. vis of G. floridanum from the Ingleside locality. PELVIS OF Glyptotherium texanum (Table 38).— These two pelves from the Texas population of The pelvis of the type specimen of G. texanum the species are apparently identical. A hitherto (AMNH 10704, Figures 45, 46) has not been fully unreported pelvis, UF/FGS 6643, from the Ca­ described, despite its early recovery near the turn talina Gardens, Florida, locality, permits compar­ of the century. The addition of the pelvis of the ison between the Florida and Texas representa­ subadult individual F:AM 95737 (Figure 47) tives. serves further to fix the characters of Glyptotherium A nearly complete pelvis of G. arizonae, associ­ and allows positive identification of the Arizona ated with a carapace and a femur, is also de­ Blancan glyptodonts as G. texanum. scribed herein for the first time. Although this Osborn (1903) merely mentioned the pelvis specimen was collected many years ago, around recovered with the carapace AMNH 10704. 1925, it has been entirely overlooked except for Brown (1912), comparing this pelvis with AMNH Gidley's (1926) mention of its recovery. It was 15548, stated that in G. texanum there are 16 recovered from the Curtis Ranch locality, Ari­ vertebrae in the sacrolumbar tube. Our determi­ zona, from which the type specimen (USNM nation of 13 or 14 vertebrae in the enire arch is 10536) of the species was obtained. Thus this based on the apparent presence of five or six pelvis adds to the osteology of this species, for discrete neural arches in the lumbar tube and which the entire skeleton, except for some of the four free vertebrae in the sacral arch. As Brown thoracic vertebrae, is known. (1912) stated, there are three iliosacral and one The number of vertebrae participating in the ischiosacral vertebrae; thus, there are 13 or 14 various regions of the axial skeleton can be tab­ vertebrae in the pelvis, rather than 16 as Brown ulated as follows. assumed. Brown (1912) also thought that the penulti­ G. G. G. G. mate vertebra lacks transverse processes. In ac­ texanum anzonae cylindricum floridanum tuality the tip of each process is broken close to 1. Number of ver- 5-6 - 7-8 8-9 the centrum so that its lateral extent cannot be tebrae in determined. They appear to have been very short, lumbar tube and, as Brown assumed, they apparently did not 2. Number of il- 3 3 4 3 participate in the ischiac support. In contrast, the iosacral ver­ transverse processes of the penultimate vertebra tebrae of F:AM 95737 (Figure 48) extend to the ischium. 3. Number of ver- 4 - 3 3 or 4 Relative to the transverse processes of the termi­ tebrae in sac­ nal vertebra, however, they are small and if bro­ ral arch ken away, their lateral extent could not be deter­ 4. Number of is- 1 or 2 1 or 2 2 2 mined with certainty on the basis of their dimen­ chiosacral sions at the centrum. Therefore, in F:AM 95737 vertebrae there are two ischiosacral vertebrae, whereas in The significance of these counts is more fully AMNH 10704 there appears to be only one, 114 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 46.—Pelvis and carapace oi Glyptothenum texanum (AMNH 10704), right side. (Photo courtesy American Museum of Natural History.)

although the possibility of participation by the spines at the anterior iliosacral vertebra is ap­ penultimate vertebra cannot be discounted. proximately 130 mm. This same height occurs for Brown (1912) also assumed that in G. texanum the spine of the terminal sacral vertebra, while there is no pubic crossbar because the preserved the minimum height, located at the midpoint of proximal portion of the pubes of AMNH 10704 the sacral arch, is an estimated 110 mm. Because is extremely reduced. The distal region is missing the sacral arch and the lumbar vertebrae are not in this specimen, however, so that just as "there fully represented in F:AM 95737, no new insight is no indication of a crossbar," as Brown asserted, into this seemingly important character can be neither is there any indication, other than the established. However, it does appear that in the reduced size of the pubes, of the absence of a Arizona specimen the neural arches were rela­ crossbar. That there may have been a crossbar is tively low throughout, for the vertebral height at indicated in F:AM 95737 (Figure 48), in which the iliac union (130 mm) is approximately the the pubes are also reduced, but their distal ex­ same as in the terminal vertebra (140 mm). The tremities are preserved, and they were apparently curvature of the sacrolumbar tube therefore ap­ united at the midline in a cartilaginous connec­ pears to distinguish G. texanum from G. cylindricum. tion. With maturity from this subadult condition, As discussed below, however, the compound it is likely that the pubes would have become curve of the pelvis of G. cylindricum is not as ankylosed despite the small size of these bones. pronounced as Brown stated, and it is likely that Hence, at least in the Arizona representative, and the differences are not worthy of more than spe­ probably in the Texas representative, the pubes cific distinction. apparently were united in a weak contact. For purposes of comparison with other species, As Brown (1912) asserted, in G. texanum there and because the pelves AMNH 10704 and F:AM is a uniform height of the neural arches for the 95737 are representative of the type-species of entire length of the sacrolumbar tube, with con­ Glyptotherium, the following description of the pel­ sequent uniformity in the compound curve for vis of G. texanum goes into considerable detail. It the sacrolumbar arch. The height of the neural is a fortunate circumstance to have two nearly NUMBER 40 115 complete pelves of the same species from different are diminutive. They are but slightly expanded localities, one from a subadult individual (F:AM beyond the thickness of the lumbar tube. Because 95737), the other from a mature adult (AMNH of breakage, the development of the prezygapo- 10704), allowing for an assessment of intraspecific physeal facets cannot be accurately determined. and ontogenetic variation. The pelvis of the They appear not to have been very large, how­ subadult individual seems to be the least devel­ ever, indicating a close similarity in the thoraco- oped skeletal region, indicating, as expected, that lumbar articulation with that known for G. ari­ fusion and termination of growth in the glypto­ zonae. The height of the fused neural crests of the dont pelvis occurred relatively late. The lack of lumbar tube appears to be relatively uniform for fusion of the pelvic bones permits rather precise both specimens, and this height is approximately delimitation of the component bones in this re­ continuous rearward for the length of the sacro­ gion; hitherto, descriptions of glyptodont pelves lumbar tube. There is no indication of carapacial have been for completely fused (i.e., adult) spec­ contact. imens only. The iliosacral vertebrae, numbering three in The lumbar tube, incorporating five or six both specimens, are united with the body of the vertebrae as the preiliac axial component of the ilium by fusion of their neural arches and their pelvis, is complete in AMNH 10704. Only por­ transverse processes in a continuous ankylosed tions of the three lumbar vertebrae immediately contact. Whereas the two neural foramina are preceding the iliosacrals are preserved for F:AM present in the younger individual, they are ob­ 95737. The articular facets for the thoracic tube scured in the adult condition. The neural arch of

FIGURE 47.—Pelvis oi Glyptothenum texanum (F:AM 95737): a, left side of pelvis showing unfused contact (juvenile) of ilium with iliosacral vertebrae and incomplete pubic crossbar; b, posterior aspect, c, dorsal aspect of ishiosacral vertebrae; d, ventrolateral aspect, e, ventromedial aspect of the right ischium and pubis. (Bar = 20 cm.) 116 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY the middle iliosacral vertebra is the dominant ulation and sharing with it a common centrum element in the union of the neural arches with disc. Certainly the centrum contact, and probably the ilium. With respect to the vertical contacts the xenarthral articulation, would have fused between neural spines, which are unfused in the with growth, rendering a construction similar to young individual, the iliac-vertebral contact is that of the adult specimen. inclined anteriorly from below upward, crossing Articulation with the first caudal vertebra is the neural spines in a slightly oblique angle. achieved through a broad centrum contact and There are four free vertebrae in the sacral arch a well-developed xenarthral contact. The cen­ in the adult specimen, as indicated in part by the trum facet is transversely elongate and concave. neural canals. The neural spines are relatively The paired xenarthral facets are convex and are uniform in height; they were apparently united directed downward and outward in typical zyg­ with the carapace in a ligamentous connection. apophyseal fashion. A uniform transverse diameter of the neural crest In AMNH 10704, at the posterior angle formed is maintained from the iliac region to the posterior by the contact of the transverse process of the extremity of the sacral arch. terminal vertebra with the ischium, there is a The ischiosacral vertebrae, numbering two in weak, downward-directed concavity for reception the subadult specimen (Figure 47) and appar­ of the lateral extremity of the transverse process ently only one in the adult, are united with the of the first caudal vertebra. This weak lateral ischiac plates by fusion of their extended trans­ contact at the sacrocaudal union reflects a relative verse processes with the posteromedial surface of lack of importance of the first caudal vertebra in each ischium in a horizontal contact. The shaft the structure of the pelvis as a "pseudo-sacral" of the transverse process of the terminal vertebra vertebra. This region is underdeveloped in the of each specimen is stout and thickened. It is younger individual, F:AM 95737. The transverse dorsoventrally compressed and anteroposteriorly processes of the anterior caudal vertebra exhibit elongate. These processes pass slightly downward no indication of contact at the lateral ischiac and rearward from the centrum. The penultimate angle. With growth, however, there probably vertebra in both specimens bears transverse would have been contact, with the development processes. In the adult these are broken and their of a shallow articular concavity as in the adult lateral extent cannot be determined. The only condition, in contrast to the deep concavity in positive statement that can be made concerning other taxa. the penultimate transverse processes of AMNH The massive ilium of G. texanum is inclined 10704 is that they were small in this individual, anteriorly with respect to the cross-sectional axis but perhaps no smaller than in F:AM 95737, in of the body. The ilium is fused to the iliosacral which they extend to the ischium. In the latter vertebrae along its medial border. It is greatly specimen, the transverse processes of the penulti­ expanded transversely and dorsoventrally to con­ mate vertebra parallel the posterior ones, but they form to the undersurface of the carapace. In the are considerably smaller and their participation adult condition, the crests of the ilia became fused in the ischiosacral contact is minor. They do not with the carapace; in the subadult condition, the contact the posterior transverse processes, al­ ilia were apparently united by ligamentous or though it is possible that there would have been cartilaginous connection. The ilium presents two contact and subsequent fusion with growth. surfaces and three borders: a concavoconvex an­ The centrum and neural arch of the terminal terior (cranial) surface, and an opposing posterior vertebra of AMNH 10704 are fused. In the (caudal) surface; the medial border, including the subadult condition (F:AM 95737), however, the greater sciatic notch; the expanded superior bor­ terminal vertebra is unfused, contacting the pe­ der (iliac crest); and the ventrolateral border. At nultimate vertebra dorsally in a xenarthral artic­ its ventral extremity the ilium contributes to the NUMBER 40 117 acetabulum where it is united with the ischium tabular ramus of the ilium is free, occupying the and pubis. For purposes of description, the ilium anterior portion of the acetabular circumference. can be divided into the expanded superior region, The inferior epiphysis of the ilium forms the or "wing," and the narrower inferior region, the acetabular angle, and the anterior two-thirds of "body." the articular surface of the acetabulum is contrib­ The anterior surface of the expanded wing is uted by the ilium, as indicated in the subadult medially convex and laterally concave. The pos­ specimen. terior surface of the wing reflects the curvature of The ischium is a broadly expanded plate, ap­ the anterior surface: medially concave and lat­ proximating the ilium in size and surface area. It erally convex, although these inflections are less is transversely compressed and anteroposteriorly pronounced. The wing is an anteroposteriorly expanded. The ischiac plate is situated at approx­ compressed plate, gradually increasing in thick­ imately a right angle with respect to the wing of ness from the body toward the iliac crest, attain­ the ilium. The ischium consists of two regions: ing maximum thickness at its ventrolateral angle. the narrow acetabular ramus anteriorly and the The rugose iliac crest is arcuate. In the adult greatly expanded body forming the flat ischiac condition the transverse curvature of the crest plate posteriorly. The lateral, slightly posteriorly becomes more nearly flattened (Figure 47) in directed surface is the outer face of the ischiac comparison with the subadult curvature, which plate. The inner, slightly anteriorly directed sur­ is very rounded (Figure 45). Accordingly, the face is weakly concave. In the subadult specimen lateral angle of the iliac crest in the adult stage is F:AM 95737, a rough and broken region occu­ elevated nearly to the height of the iliosacral pying the middle third of the medial surface near region, whereas in the younger condition the the posterior border marks the surface with which lateral angle is decidedly lower. the ischiosacral vertebrae unite. The articulation On the continuous medial border of the body in this subadult condition was cartilaginous. In and wing, there is a marked thickening and ex­ the adult specimen AMNH 10704, this contact is pansion of the bone for contact and subsequent solidly fused. fusion with the united transverse processes of the The ischiac crest is broad and uniformly iliosacral vertebrae. This contact is unfused in the arched, conforming to the undersurface of the subadult condition; in the adult stage there is carapace in the anteroposterior direction. There complete fusion, obscuring these contacts. Be­ was a marked upward expansion of the crest with neath the region of contact with the iliosacral growth, as the bulk of the body and the carapace vertebrae, the medial border of the body is deeply expanded. Accordingly, in the adult condition, concave, forming the greater sciatic notch. the crest is less curved and is more nearly hori­ The body, or shaft, of the ilium is uniformly zontal than in the juvenile stage. In addition, convex on the anterior surface. It is flat on the there appears to have been posterior expansion of posterior surface, with a weakly developed con­ the ischiac plate with growth, more clearly sepa­ cave inflection on the medial border in the region rating the crest from the free posterior border. of the greater sciatic notch. The lateral border of The crest intersects the deeply concave anterior the ilium forms a uniformly concave arc in profile. border in an acute angle. The anterior border is At the inferior extremity of the medial border a continuation of the superior border of the ace­ there is a prominent iliopubic tuberosity. tabular ramus. The lower extremity of the ilium is united on The posteroinferior region of the ischium distal its flat anteromedial surface anteriorly with the to the obturator foramen is fused with the pubis. pubis and posteriorly with the ischium; it is Although the contact is not readily distinguish­ united at its posterolateral surface exclusively able even for the younger individual, it appears with the ischium. The lateral surface of the ace- to have been adjacent to the distinctly thickened 118 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY region. The entire body of the ischium is situated ity of a midline fusion cannot be firmly estab­ in a single plane. The greatest thickness of the lished. Because of the close similarity with F:AM ischiac plate is at the superior angle, with dimin­ 95737 in the proximal portions of the pubis, it ishing thickness toward the pubis. appears likely that there was indeed a complete The continuous dorsal and anterior border of pubic crossbar in adults. The pubis is divided the ischium (lesser sciatic notch) is straight along into the acetabular ramus, the middle body or the shaft, forming a deeply concave arch in the shaft, and the symphyseal ramus. expansion toward the ischiac crest. The shaft The acetabular ramus is an expansion that connecting the acetabular ramus with the body unites with the ilium and the ischium. From the is fiat on both sides and thicker in cross section acetabulum to the body of the pubis, the ace­ along the lesser sciatic notch than along the bor­ tabular ramus is expanded transversely, and, to der of the obturator foramen. a lesser extent, anteroposteriorly for participation Uniting with the ilium at its upper extremity in the acetabulum, although it is excluded from and with the pubis on its anteromedial surface, the actual articular surface. The thin, compressed the acetabular ramus contributes the posterior body forms the medioventral border of the obtu­ third of the acetabulum. In cross section the rator foramen. The symphyseal ramus in F:AM ramus is three-sided, presenting flattened antero­ 95737 extends in a broad arc toward the midline lateral and posteromedial nonarticular surfaces of the body to meet its opposite. The symphyseal and the concave articular surface of the acetab­ terminus of the pubis extends as a short, thin rod, ulum. the flattened rugose terminus indicating a carti­ The acetabulum is formed by the fused ace­ laginous continuation. This region is missing in tabular rami of the ilium, ischium, and pubis. the adult AMNH 10704, and a pubic crossbar is The latter bone is excluded from the articular not included in the restoration. Because of its surface, to which the ilium contributes the supe­ close similarity to the juvenile specimen, there is rior and lateral two-thirds and the ischium the no reason to believe that the crossbar was absent. posterior third. The deep cup-shaped acetabulum It is only missing from the specimen, and it is is slightly elongate in the anterolateral/postero- probable that it was simply a small, reduced bone medial direction. The inner circumference of the (as in the younger individual) in the adult con­ acetabulum is interrupted by the medial ace­ dition, fusing at the midline. A similar argument tabular fossa, which penetrates toward the center is advanced below for the pelves of G. floridanum from the acetabular margin on the ischial surface. and G. arizonae. The pubis of G. cylindricum is This nonarticular sulcus is marked by a shallow massive and unites at the midline in a strongly furrow bordering the iliac portion of the acetab­ ankylosed contact. ulum. Another fossa, usually termed the external In the region of ischiopubic fusion the pubis acetabular fossa, penetrates inward from the presents a flat posteromedially directed surface as outer margin. a continuation of the anteromedial surface of the The long slender pubis of G. texanum occupies body. Beyond the region of the ischiopubic fusion the lower position in its participation in the is- the crossbar becomes rounded. chiopubic plate. From the acetabular ramus, the Thus the pelves F:AM 95737 and AMNH pubis initially extends downward and outward, 10704 are qualitatively similar. As indicated in situated in direct line with the medial border of Table 38, quantitative measurements are remark­ the ilium. Distally the pubis curves toward the ably close between these two specimens, the only midline, in the subadult condition apparently important differences being in the smaller pubis uniting with its partner in a cartilaginous con­ in the younger individual—a distinction attrib­ nection. The distal extremity is broken in the utable to age. Moreover, these specimens more adult specimen (AMNH 10704), and the actual­ closely resemble each other, both qualitatively NUMBER 40 119

and quantitatively, than either resembles other determined whether these processes were free or specimens. It is partly on the basis of this com­ laterally fused, either to the transverse processes parison that these specimens, one from Texas, the of the last sacral, as in G. cylindricum, or directly to other from Arizona, are referred to the same the ischium, as in G. texanum and G. floridanum. species, despite their geographic separation. There is no obvious point of breakage on either These matters are discussed in greater detail else­ the ischium or the last transverse process, sug­ where. gesting that the penultimate vertebra did not In addition to those dimensions presented in participate in ischiac support. Table 38, several other potentially useful mea­ The neural crests are very high, exceeding those surements for the type specimen (AMNH 10704) of G. texanum by a factor of approximately 1.5 of G. texanum are: transverse diameter between (attributable to allometric size differences; see anterosuperior angles of ischiac crests, measured Table 38) and approximating those of G. cylindri­ in a straight line through neural arch, 500 mm; cum. For comparison, height measurements at the transverse diameter between posterosuperior an­ anterior iliosacral vertebra, at the posterior ilio­ gles of ischiac crests, measured in a straight line, sacral vertebra, at the midsacral arch, and at the 360 mm; transverse diameter sacrocaudal facet, ischiosacral vertebra for AMNH 21808 (G. an­ 58 mm; inferior-superior diameter, sacrocaudal zonae) and AMNH 15548 (G. cylindricum), respec­ facet, 39 mm; approximate height (inferior-su- tively, are as follows: G. anzonae 190 160 120 170 = 0.86; = 0.88; = 0.92; = 1.13 G. cylindricum 200 T80 730 T50 perior diameter) iliac crest from posterior border Thus, the compound curvature of the sacral arch of acetabulum, 300 mm; transverse inside diam­ of G. arizonae, as reflected in these measurements, eter between acetabula, 220 mm. A notable char­ is less than in G. cylindricum, for which the sacral acteristic is the posterior convergence of the is­ arch diminishes in height rearward more drasti­ chiac crests, an important feature distinguishing cally. G. texanum from G. arizonae and G. cylindricum, as Although the ischiac crests are broken away in discussed below and under "Systematics". AMNH 21808, they appear to be slightly conver­ PELVIS OF Glyptotherium arizonae (Table 38).— gent posteriorly but not as pronounced as in G. The pelvis of G. anzonae is known from a single texanum. The transverse process of the ischiosacral specimen only (AMNH 21808, Figure 48), col­ vertebra is directed slightly rearward, so that a lected from the type-locality for the species, Curtis line connecting the angles formed by the ischium/ Ranch, Arizona. The only mention of this impor­ transverse process contact passes somewhat pos­ tant specimen in print was Gidley's (1926) remark terior to the caudal articular facet. This orienta­ that the pubes appeared to be reduced. The pelvis tion is similar to that in G. texanum and distinct is partially fused to the carapace. The crests of from that in G. cylindricum and G. floridanum. both ischiac plates are missing, as are the distal The distal ends of the pubes are broken. De­ extremities of the pubes. The number of vertebrae spite their reconstruction to the contrary, they in the lumbar tube is indeterminate. There are were probably centrally united at the midline, as three iliosacral vertebrae, an indeterminate num­ assumed for G. texanum. Indeed, the shafts of the ber in the sacral arch, and either one or two pubes of this specimen are intermediate in size ischiosacral vertebrae. between those of AMNH 10704 (G. texanum type) The transverse processes of the three iliosacral and AMNH 15548 (G. cylindricum type), as indi­ vertebrae are solidly fused with the ilium, as in G. cated in Table 38. Thus the fact that the pubes texanum. The transverse processes of the penulti­ are even larger than in G. texanum, for which there mate sacral vertebra are broken. It cannot be is evidence of a crossbar, is indirect evidence that 120 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 48.—Pelvis of Glyptotherium arizonae (AMNH 21808): a. posterior region showing sacral arch, including ischiosacral vertebrae and superior crest of ischium; b, same, posterior; c, acetabulum, ischiac plate, and pubis, ventrolateral; (reconstructed portions hatched) d, enlarged detail of acetabulum. (Bar = 20 cm.)

G. arizonae possessed a crossbar also. In addition, crest of the ischium, which "over-rides" the trans­ the shape and contribution of the pubes more verse process. It undoubtedly received the extrem­ closely resemble those of G. cylindricum, adding ity of the transverse process of the first caudal further support for the assumption of a crossbar. vertebra. This participation by the first caudal The anterior surface of the ilium is concave in vertebra in the sacrocaudal complex as a "pseudo- the inferior-superior direction; transversely, it is sacral" is the most highly developed among North deeply concave for the medial two-thirds, and the American representatives. lateral third is strongly convex. There is a sub­ PELVIS OF Glyptotherium cylindricum (Table 38).— dued, but nevertheless distinct, vertical ridge on The pelvis of the type specimen of G. cylindricum the crest of the convex anterior surface, separating (AMNH 15548, Figure 49d) was originally de­ the inner concave region from the outer convex scribed by Brown (1912). He accurately reported region. This ridge is not present on the available a vertebral count of seven (but there may be ilia of G. texanum. The posterior surface of the eight) vertebrae in the lumbar tube, four iliosacral ilium is distinctive for its flat inner third; the vertebrae, three free vertebrae in the sacral arch, outer two-thirds is concave, corresponding to the and two ischiosacral vertebrae; total 16 (possibly convexity of the anterior surface. These two re­ 17) vertebrae in the pelvis. The number of ilio­ gions are separated by a vertical crest, similar to sacral vertebrae is indicated by the presence of the one on the anterior surface. four stout transverse processes on the inferior The articulation with the transverse process of arch, which are solidly fused with the ilia. the first caudal vertebra is formed by a deep The transverse processes of the two ischiosacral concavity at the juncture of the transverse process vertebrae are fused into a single shaft midway with the ischium. This downward-directed con­ between the ischia and the centra. The transverse cavity is formed by a posterior extension of the processes of the anterior ischiosacral vertebra (i.e. NUMBER 40 121

penultimate sacral vertebra) are slender in com­ terior surface. The crests of the ischia are parallel, parison with the posterior transverse processes. as in G. floridanum and in contrast to G. texanum The transverse processes of the terminal vertebra and G. arizonae. are massive; they are directed laterally and Only the fused centra, numbering seven or slightly downward from the centra, as in other eight, are present. The neural spines are broken representatives. Their rearward extension lies away, so that it cannot be determined whether only slightly posterior to the plane of the sacro­ the spines were fused with the carapace. The caudal articular facet. This orientation is inter­ undersurface of the scutes along the arch of the mediate between the greater posterior component carapace exhibits somewhat more rugosity than of G. texanum and G. arizonae and the perpendicular the adjoining scutes, but there is no indication of orientation in G. floridanum. ankylosis with the lumbar tube. The connection As Brown noted, the compound curve of the was probably ligamental. neural spines of the sacral arch decreases rear­ There are prominent articular facets at the ward with respect to the centra. The heights of angles formed by the transverse processes of the anterior neural spines measured from the lower ischiosacral vertebra and the ischiac plate for margin of the centrum are as follows: ischiosacral reception of the transverse processes of the first spine, 150 mm; midsacral arch, 130 mm; posterior caudal vertebra. These facets are oval and weakly iliosacral spine, 180 mm; anterior iliosacral spine, concave and are directed inward and rearward. 220 mm. Thus the height of the anterior neural In summary for G. cylindricum, the pelvis is spines is greater than the minimum height at distinctive in possessing four iliosacral vertebrae, midsacral arch by a factor of 1.5, rather than a in the massive pubes with a strong crossbar, and factor of 2 as stated by Brown (1912); and the the flattened posterior surface of the ilium. Its anterior height is greater than that of the ischio­ remaining characteristics are variously similar to sacral vertebra by a factor of approximately 1.3. those of the remaining North American species, As Brown (1912) indicated, the pubis is massive notably in the orientation of the ischiac crests, in comparison with the other North American the orientation of the transverse process of the representatives. It is solidly fused at the sym­ terminal sacral vertebra, the height and curvature physis, where its cross-sectional shape is elliptical. of the neural spines, the fusion of the transverse Anteroposterior and dorsoventral dimensions at processes of the ischiosacral vertebrae with each the pubic symphysis are 29 and 38 mm, respec­ other, and the general overall construction. tively. These dimensions are significantly greater PELVIS OF Glyptotherium floridanum (Table 38).— than those of the pubic shafts (Table 38), which, The pelvis of G. floridanum is known from several although considerably larger than those of the specimens, including a nearly complete one from other specimens, are not as different as Brown the Texas Ingleside locality (TMM 30967-1926, (1912) inferred. Brown considered the nature of Figure 49a,b,c), which was described by Lundelius the pubes as one of the principal distinguishing (1972); fragmentary remains from Hunt County, features of this specimen for his proposed taxon­ Texas (USNM 6071), which were described by omy. He was apparently correct in assuming that Hay (1916); and undescribed remains, compris­ this construction of the pubes is taxonomically ing a nearly complete composite pelvis, from the significant, but similarities of the remaining skel­ Catalina Gardens locality in Florida (UF/FGS etal elements to other North American species 6643). These remains allow positive comparison indicate a much closer relationship than the fa­ between the Texas and Florida representatives, milial separation that Brown proposed. in part forming the basis for establishing the The ilia differ slightly from those known for G. taxonomic identity of these two populations. Un­ arizonae in being more deeply concave on the fortunately, no pelvic remains are associated with anterior surface and rather flattened on the pos­ the type specimen. 122 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 49.—Pelvis of Glyptothenum floridanum (TMM 30967-1926): a, dorsolateral, including lumbar tube; b, dorsal; c, anterior of ilia; d, posterior aspect of pelvis of Glyptolhenum cylindricum (AMNH 15548). (Bar = 20 cm.)

The lumbar tube, as preserved in the Ingleside ida specimen; the number of iliosacrals is inde­ specimen, indicates no appreciable difference terminate for the Ingleside specimen. The con­ from that known for the other species. Because it struction of this region in the latter, however, is nearly complete, however, it adds to the knowl­ closely resembles that of the Florida pelvis, rather edge of this element in several regards. The paired than the region of the four iliosacral vertebrae in haemal arches are serially fused, forming a con­ G. cylindricum, so it is likely that there arc only tinuous haemal concavity from approximately three in TMM 30967-1926. In both the Florida the midposition of the tube rearward. The dorsal and Texas specimens, this region otherwise pre­ spines, although broken, are thin and solidly sents no outstanding features. fused. The centra of the lumbar tube are fused Only a fragment of the sacral arch of USNM into a straight, unarched unit. 6071 is preserved. Hay (1916) stated that this There are three iliosacral vertebrae in the Flor­ fragment represents the fourth and fifth sacral NUMBER 40 123 vertebrae, which is a reasonable identification. tebrae indicating the close relationship between Complete sacral arches are present for both these three specimens is the perpendicular ori­ TMM 30967-1926 and UF/FGS 6643. In the entation of the transverse processes. Thus the former specimen, the number of free vertebrae in ischiosacral vertebrae of G. floridanum are similar the sacral arch is obscured by the high degree of to those of G. cylindricum in the midpoint fusion of fusion. There are at least three free sacrals, and the transverse processes and their perpendicular probably four, in the Florida pelvis. orientation. Ischiosacral vertebrae are preserved for all The anterior surface of the ilium of G. flori­ three representatives, although they are complete danum, as indicated by both the Florida and In­ only in TMM 30967-1926. In this specimen the gleside specimens, is distinctive in being almost last two sacral vertebrae participate in ischiac uniformly concave in the transverse direction, support by their transverse processes. The proc­ rather than concavoconvex as in the other species. esses of the terminal vertebra are much larger The only convexity occurs near the lateral mar­ than those of the penultimate vertebra. The gin, where the body is rounded and thickened. processes are fused to each other from slightly There is otherwise no lateral convexity on the beyond their midpoint, as in G. cylindricum. The anterior surface. The posterior surface is weakly antepenultimate centrum in the Ingleside speci­ convex to nearly flat, with only a slight medial men possesses transverse processes that overlap concavity. the anterolateral angles of the centrum of the first The crest of the ischium is missing for the ischiosacral vertebra. These are fused with the Florida specimen; the ischium of the Ingleside centrum, although this fusion appears to have representative is complete. The crests of the ischia occurred at a relatively late age. The centra of are strongly convergent posteriorly. The ischium the two ischiosacral vertebrae are incompletely otherwise exhibits no peculiarities affording dis­ fused, similarly exhibiting a relatively late fusion. tinction from that of the other species. The transverse processes of the anterior ischio­ Lundelius (1972) suggested that the reduced sacral vertebra of UF/FGS 6643 are broken. They pubes of TMM 30967-1926, for which the distal are relatively small, as in the Ingleside specimen, extremities are not preserved, probably indicate and their proximal orientation is similar, indicat­ that there was no crossbar in this specimen. Al­ ing that they fused with the posterior transverse though the distal extremity of the pubis of the process before contacting the ischium. Florida specimen is also missing, enough is pres­ The last sacral vertebra (presumably sacral ent to indicate the likelihood of a crossbar. The vertebra 8, as identified by Hay, 1916) of USNM shaft of this pubis is preserved, including the 6071 is preserved as a separate element of the entire region of fusion with the ischium. It is sacrolumbar series. The region on this specimen broken beyond the ischiac contact; the breakage that Hay (1916) identified as "the hinder part of indicates that the pubes extended medially, main­ the seventh sacral" is actually the anterior part of taining approximately the same thickness toward the last centrum, for which the dorsal arches are the midline. Whether there was actual fusion or broken. The lateral extremities of the transverse a persistent cartilaginous or ligamentous connec­ processes of the penultimate (seventh) sacral ver­ tion cannot be determined, but at least some tebra are present as fused additions to the larger connection is certain, for it is unlikely that the transverse processes of the terminal vertebra. pubes would extend medially and terminate as Thus, in USNM 6071, as in TMM 30967-1926 blunt processes without distal support or connec­ and as inferred for UF/FGS 6643, the transverse tion. The similarity of the Ingleside pubes with processes of the two ischiosacral vertebrae are those of the Florida representative suggests that fused from their midpoints outward. there was indeed a crossbar, the relatively reduced Another characteristic of the ischiosacral ver­ size of the pubes notwithstanding. 124 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

HIND LIMB is gently rounded on the medial side and slightly concave on the expansion for the lateral epicon­ The bones of the hind limb oi Glyptothenum are dyle. The anterior surface of the shaft is irregu­ notable for their even more massive construction larly rounded in cross section. In sagittal section in comparison with their counterparts in the front the anteroproximal surface is weakly concave, limb. The femoral length exceeds that of the expanding in thickness toward the head and the humerus by a factor of approximately 1.25. The greater trochanter; the middle third is flattened; tibiofibula is a single fused element, and although and the distal third is gently concave, expanding only approximately half the length of the femur, toward the trochlea and the lateral epicondyle. it is a massive bone reflecting the graviportal In anterior aspect, the narrowest constriction of nature of the rear leg. The tarsals and metatarsals the shaft occurs near the middle, with marked are large and extremely compressed, unlike those expansion toward the extremities. The expansion of any other mammal. There are five symmetri­ of the shaft for the lateral supracondylar crest cally disposed, complete digits on the hind foot. presents a broad transverse inflection on the an­ As in the front limb, there is a full battery of terior surface. digital sesamoids, and there is no indication of The central longitudinal axis (sagittal plane) of fusion or loss of individual bones. The patella is the femur passes through the trochanteric fossa, so constructed as to form a "patellar lock" allow­ centrally through the narrowest constriction at ing full extension at the femoral-tibial joint in the midshaft, and emerges in the plane passing "resting" (probably related to feeding habits, i.e., through the lateral ridge of the distal trochlea. grazing or browsing) position. The massive lateral supracondylar crest is heavily Although the hind limb of South American rugose and presents a deeply pitted and flattened glyptodonts is fairly well known, there have been lateral surface. It occupies the entire distal third few comprehensive accounts other than Burmeis- of the lateral angle of the shaft as a continuation ter's (1874). As for the front limb, the hind limb of the obliquely oriented lateral epicondyle. The of Glyptotherium compares closely with that of the medial epicondyle is represented by an incon­ South American Glyptodon, in the classic sense. spicuous tuberosity on the medial angle. FEMUR (Table 39).—The femur of G. texanum The distal articular surfaces are represented by (Figure 50) is distinctive for its almost perfectly the trochlea and the medial and lateral condyles. straight, columnar construction. In block propor­ The trochlea is situated medially with respect to tions it is a rectangular prism, the transverse the long axis of the femur. Although the angle of breadth somewhat greater than half the length. the medial ridge is broken away, the trochlea The massive greater trochanter extends proxi­ appears to have been approximately symmetrical, mally as far as the head in this species, and in with the anteroposterior extent of the lateral ridge gross proportions it is approximately equal in slightly greater than that of the medial ridge. The size. The trochanteric fossa is shallow and smooth. trochlea is notable for its relatively short antero­ The short neck is unconstricted in the transverse posterior diameter. plane and only slightly constricted in the antero­ The obliquity of the distal articulation is ac­ posterior plane. centuated by the tremendous difference in size The underdeveloped lesser tuberosity extends between the condyles: the medial condyle is at as a rugose surface from the medial side of the least twice as large as the lateral one. The medial head to midshaft, occupying the proximal half of condyle is round in medial aspect, extending from the medial angle. A trochanteric ridge on the a relatively flat surface at the distal extremity to posteroproximal surface of the shaft is wanting. assume a posterior and somewhat proximal ori­ The proximal two-thirds of the posterior surface entation at its upper extremity. It is markedly of the shaft is distinctly flattened; the distal third elongate in the plane of articulation. The smaller NUMBER 40 125

FIGURE 50.—Femur oi Glyptotherium texanum (F:AM 95737): a, anterior; b, medial; c, posterior; d, proximal extremity; e, distal extremity. (Bar = 10 cm.) lateral condyle is separated from the medial con­ AMNH 21808). This larger size in the Seymour dyle by a deeply excavated intercondyloid fossa. representatives might indicate an actual size dif­ The lateral condyle is convex, and its boundaries ference between the two populations of G. arizonae, are approximately triangular in shape. or it might be variation attributable to the small The femora of G. arizonae and G. floridanum sample size. Because these differences seem to differ in minor details, mostly related to size indicate a continuum, rather than a well-marked differences. Although future recoveries will per­ separation, there is no foundation in these small haps fill in the existing gaps, the femora of these samples for taxonomic distinction between these two species are sufficiently distinct to allow sep­ two populations. arate comparisons with G. texanum, and with each On the other hand, the obvious size difference other, as follows. between these femora and the ones known for G. Melton (1964) noted the excessive length of the texanum is substantial. The femur of G. arizonae femur of UMMP 46231, implying taxonomic dis­ (Figure b\a,b), in addition to its larger size, differs tinction from the Curtis Ranch glyptodont. As from G. texanum in a number of relatively minor indicated under "Systematics," however, the di­ features. The plane of elongation of the head of mensions of this specimen are not significantly the femur in G. arizonae is oriented approximately different from those of USNM 10536 and AMNH perpendicular to the transverse axis of the proxi­ 21808. Despite the small sample size, however, mal extremity. If projected, this plane intersects the three femora known from the Seymour fauna the distal extremity through the medial facet of indicate a slightly greater size than the two known the trochlea approximately through its plane of from the Curtis Ranch fauna (USNM 10536 and elongation. In contrast, the plane of elongation of 126 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 51.—F'emur: a. right, Glyptothenum anzonae from Curtis Ranch, Arizona (USNM 10536), anterior; b, left, G. arizonae (AMNH 21808) from same locality, anterior (greater trochanter restored); c, right, G. floridanum (USNM 6071), anterior. (Bar = 80 mm.) the femoral head in G. texanum is oblique with Remaining distinctions between the femora of respect to the transverse axis of the proximal G. texanum and G. arizonae are attributable to size extremity, and in projection, this plane intersects and age differences in the available specimens. the medial condyle in an oblique angle with For example, there is a deeply excavated trian­ respect to its plane of elongation. Moreover, the gular supratrochlear fossa in the G. arizonae adult femoral head in G. anzonae is situated on a definite specimens; this region is unexcavated in the ju­ neck, which exhibits a greater degree of expansion venile (F:AM 95737) representative of G. texanum. from the shaft, especially on the anterior surface. The femur of G. floridanum (Figure 51c), repre­ Another distinction of G. arizonae is the massive sented by a complete specimen from Hunt development of the greater trochanter. It exhibits County, Texas, and a portion of the femoral head a relatively greater expansion from the shaft than from the Catalina Gardens, Florida locality (UF/ does that of G. texanum, and it is elevated some­ FGS 6643), is somewhat intermediate with re­ what beyond the level of the head. Pronounced spect to the femora described above. It is smaller anterolateral and proximal crests of the greater than those known for G. anzonae and larger than trochanter are separated by a distinct concavity, the ones known for G. texanum. The construction whereas this region in G. texanum presents a uni­ of the proximal extremity is similar to that of G. formly convex surface, without noticeable devel­ texanum, although the head is more elongate in opment of these crests. the Catalina Gardens specimen. Similarly, there NUMBER 40 127 is no development of the anterolateral and prox­ shortest. The latter dimension is greatest near the imal crests of the greater trochanter, as in G. proximal border, and least at the distal extremity. texanum, and in contrast with G. arizonae. The striated outer (cranial) nonarticular sur­ There is a shallow supratrochlear fossa in G. face is roughly cylindrical in shape, with the floridanum. A single feature distinguishing the fe­ lateral region somewhat flattened, especially near mur of this species is the construction of the the proximal extremity. In the proximocentral proximolateral surface of the shaft, which ex­ position there is a distinct tuberosity for insertion pands for the greater trochanter. Whereas in G. of the quadriceps musculature. arizonae and G. texanum there is a distinct crest The articular facet is confined to the posterior demarcating the anterior and lateral surfaces, this (caudal) surface. It is separated from the striated region is broadly rounded and convex in G. flori­ and rugose anterior surface by a distinct nonar­ danum. Furthermore, this region is distinctly ru­ ticular and nonrugose depressed region circum­ gose, including the anteriorly oriented region that scribing the articular surface and bearing vascu­ replaces the lateral crest. lar foramina of variable size. The articular facet In summary for the femur, there are only minor conforms loosely to the trochlea of the femur, differences among the three North American spe­ articulating with it tangentially. A rounded, cen­ cies for which it is known. The distinctions are tral crest, directed proximodistally, separates the likely attributable more to ontogenetic variation flattened medial region of the facet from the than is here recognized and, accordingly, should concave lateral portion. The proximolateral and not receive significant taxonomic consideration, distomedial borders extend as flared, concave despite others' (e.g. Hay, 1916; Melton, 1964) surfaces to form the proximal and distal extrem­ opinions to the contrary. ities of the facet, respectively, imparting a distinct PATELLA (Table 40).—The patella of Glypto­ asymmetrical shape to the articular surface. therium (Figure 52) is known for three species, The blunt distal apex fits loosely over the including the juvenile specimen of G. texanum (F: anteroproximal surface of the tibiofibula in exten­ AM 95737) and adult specimens of G. arizonae sion. Hyperextension is prevented partly by the (USNM 10536) and G. floridanum (USNM 6071). patella, which by the flared proximolateral sur­ Except for size differences the patella is identical face forms a "stopping" device or "patellar lock" for all three species. at the position of near-maximum extension (i.e., The patella is elongate in the proximodistal the femoral-tibial contact is straight at maximum plane. In transverse section the outline describes extension and allows a resting position for this a broad sector of a circle, with the apex situated articulation in the erect, standing posture). at the central crest of the articular facet. The The patellae of G. arizonae and G. texanum are proximodistal diameter is the longest; the trans­ identical in all features except for size. The patella verse diameter is intermediate; and the antero­ of G. floridanum (USNM 6071) differs in possessing posterior diameter (craniocaudal diameter) is the more deeply grooved striae on the cranial surface, especially on the distal half of the bone. The articular facet is relatively broader and more nearly rectangular, the proximolateral and dis­ tomedial flared extensions projecting less than in the other two species. The articular surface of the patella of G. floridanum is flattened; the transverse concavities are not as deeply excavated as in G. texanum and G. arizonae. TIBIOFIBULA FIGURE 52.—Patella oi Glyptothenum texanum (F:AM 95737): (Table 41).—The tibia and fibula a, proximal extremity; b, outer surface; c, inner surface. (Bar are ankylosed into a single, stout element, typical = 40 mm.) of glyptodonts. As a unit the tibiofibula is a 128 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY cylindrical body, its length approximately half uniformly concave. The facet is situated in a that of the femur. The length of the "cylinder" is posterior position, flaring weakly beyond the ex­ somewhat less than twice the diameter at either tent of the shaft on the posterior and medial extremity. The proximal and distal extremities margins. are nearly equal in outside dimensions, and they The entire anterior region of the proximal epi­ are only moderately expanded beyond the di­ physis is nonarticular. The caudal and cranial ameter of the "shaft" of the tibiofibular cylinder. intercondylar fossae are contiguous as a single The shaft obtains a narrow anterior aperture obliquely directed sulcus. The anterior terminus formed by the interosseous space that expands of the intercondylar sulcus intersects with the posteriorly. The tibial shaft is more massive than more vertically oriented tibial tuberosity, repre­ the fibular shaft, although the latter was impor­ senting the most distal extension of the epiphysis. tant in weight support. The tibial tuberosity overlies the anterolateral The separate identities of the proximal and extremity of the proximal end of the tibial shaft. distal epiphyseal plates are lost in the subadult The fibular facet, which receives the lateral stage, with the tibial and fibular epiphyses fused condyle of the femur, presents the shape of an with each other but not with the diaphyses. In equilateral triangle in proximal aspect, with api­ this young condition, the diaphyses are not in ces at the anterior, posteromedial, and posterolat­ proximal contact, the complete bridge being eral positions. The facet is saddle-shaped, antero­ formed by the epiphysis. Examination of the posteriorly convex, and transversely concave. The tibiofibulae of older individuals reveals that there medial extremity of the facet forms a prominent is little, if any, subsequent fusion of the proximal intercondyloid eminence, situated on the border extremities of the diaphyses with advancing age. of the facet with the intercondylar sulcus. A In G. texanum (Figure 53) the proximal extrem­ smaller intercondylar eminence is situated on the ity is divided into tibial and fibular articular anterolateral margin of the tibial facet, anterior facets by a distinct nonarticular intercondyloid and slightly medial to the larger fibular eminence. fossa. The tibial facet, for reception of the medial The shaft of the tibia is transversely com­ condyle of the femur, covers approximately twice pressed, and it is considerably deeper in the an­ the surface area of the fibular facet. Except for a teroposterior direction at the proximal end than weakly convex anterior margin, the tibial facet is at the distal terminus of the shaft. The tibial crest

FIGURE 53.—Right tibiofibula oi Glyptothenum texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, proximal extremity; e, distal extremity. (Bar = 10 cm.) NUMBER 40 129 extends obliquely from its anteroproximal conflu­ distal extremity of the fibular shaft is distinctly ence with the tibial tuberosity to a distomedial expanded for fusion with the distal epiphysis and position on the anterior surface of the shaft. The with the fibular process of the tibia. minimum transverse diameter occurs at approxi­ The distal epiphysis is subequally divided into mately midshaft. The medial surface (i.e., outside tibial and fibular portions with indistinct bound­ surface) is broadly convex at its proximal extrem­ aries. Unlike its proximal counterpart, the distal ity, becoming flattened at midshaft and remain­ epiphysis does not contact the diaphyses of the ing flat to the constricted distal extremity. The tibia and fibula as a discoid surface of fusion. lateral surface (i.e., interosseous surface) of the Instead, the shape of the epiphyseal plate reflects shaft is uniformly concave in the proximodistal approximately the sculpturing of the articular plane and irregularly concave in cross section. facets. The medial malleolus of the tibiofibula The cross section of the distal extremity is ap­ occupies the entire medial surface of the epiphysis proximately triangular, with apices formed an- and extends proximally onto the distolateral sur­ teromedially by the tibial crest, posteriorly by the face of the tibia. A prominent posterior malleolus vertical posterior crest of the shaft, and anterolat­ is situated in direct line with the shaft of the tibia. erally by the fibular process. The fibular process This posterior tuberosity is separated from the of the tibia extends from midshaft to fuse with medial malleolus by an obliquely oriented sulcus, the corresponding tibial process of the fibula at extending from the lateral (interosseous) surface the distal extremity. The anterior surfaces of these of the tibial shaft, downward over the postero­ processes are confluent, together forming the ex­ medial surface of the epiphysis, as a small mus­ pansive attachment surface for the massive tibi- cular groove. alis musculature for lateral rotation and flexion The lateral malleolus forms a somewhat more of the tarsus. rugose prominence, located on the anterolateral The shaft of the fibula, although somewhat less surface of the epiphysis and extending proximally massive than the tibial shaft, is nevertheless a well to form the distolateral angle of the fibular shaft. developed and integral part of the tibiofibula. The articular facet receives the trochlear facet Like the shaft of the tibia, this bone is markedly of the astragalus. The shape of the distal articular compressed in the transverse direction. There is facet in distal aspect is that of an hourglass, with little expansion at the proximal extremity, where a somewhat greater anterolateral expanse of the the shaft contacts the epiphysis. A lateral tu­ fibular portion. The facet is comprised of two berosity is situated on the proximal third of the parallel grooves, the smaller tibial (medial) shaft. The tibial process is the distal extension of groove and the larger and deeper fibular (lateral) the nearly straight anterior fibular crest and the groove. The fibular groove is deeper and its trans­ medial expansion of the distal extremity of the verse concavity is rather more pronounced. The shaft. As described above, this anterior surface of rotational axes of these facets are parallel, and the shaft is confluent with the corresponding the rotational motion imparted to the astragalus surface of the tibia, forming the distinctive mus­ is obliquely oriented, anteromedial to posterolat­ cular groove. The posterior crest of the fibular eral. Surrounding the entire articular facet, ex­ shaft is gently sigmoid in posterior aspect. The cept for the medial malleolus groove, is a contin­ medial surface (interosseous surface) of the fibular uous rugose nonarticular border, encasing the shaft is anteriorly concave and posteriorly convex. articular facet on the margin. The lateral surface (outside surface of the tibiofi­ The only known tibiofibula of a full adult bula) of the shaft is somewhat flattened at its individual of G. texanum (JWT 1723), from the proximal end, strongly convex at midshaft, and Cita Canyon fauna, is smaller than those known divided into flat lateral and anterior surfaces at for G. floridanum and G. anzonae. the distal extremity by the lateral malleolus. The The tibiofibula of G. arizonae (Figure 54) is 130 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

known from several specimens. Melton (1964) noted the somewhat greater lengths of the tibiofibulae of the Seymour glyptodonts in com­ parison with the Curtis Ranch specimen. (Speci­ mens 46231 and 33524 in Melton's (1964) table III should be reversed so as to be placed above their proper measurements.) In actuality, there is overlap in these measurements when taken be­ tween articular surfaces rather than including the proximal and distal tuberosities. The tibiofibula of G. arizonae differs from that of G. texanum primarily in features related to the size. Muscle attachment surfaces are distinctly more prominent and rugose. The tibial tuberosity and the malleoli are heavily rugose, the malleolar groove is more sharply defined, and the articular facets are deep and well defined. The interosseous surfaces of the tibial and fibular shafts bear a number of prominent ridges and grooves, and the bone is deeply pitted and rugose. A feature on USNM 10536 related to the exaggeration of the malleoli is a peculiar scoop-shaped sulcus issuing from the lateral tuberosity of the fibula on the proximoposterior surface of the shaft. This obliquely oriented sulcus is formed by the flared transverse expansion of the bone surrounding the origin surface for m. flexor digitalis longus in this position. From this origin, the muscle passed obliquely downward over the malleolar groove of the posterodistal surface of the tibia. A distinction of the tibiofibula of G. anzonae not related to age or size is the presence in USNM 10536 of a pair of deep excavations situated between the lateral margin of the distal articular facet and the flattened distal surface of the lateral malleolus. Both are subrectangular in shape, and the posterior one is the larger. They penetrate obliquely into the lateral malleolus, and they are separated by a bony strut connecting the lateral malleolus and the margin of the distal facet.

FIGURE 54.—Tibiofibula of 3 representalives oi Glyptothenum 33524), also from the Gilliland local fauna; c, right tibio­ arizonae, showing intraspecific variation: a, left tibiofibula fibula with proportions similar to a, but of slightly shorter with long and slender proportions (UMMP 46231) from the total length, from the Curtis Ranch local fauna, Arizona Gilliland local fauna, Seymour F'ormation, Texas; b, right (USNM 10536); left, anterior and ris^ht, posterior aspects in tibiofibula with short and robust proportions (UMMP a and b; c is anterior. (Bar = 40 mm.) NUMBER 40 131

The four available tibiofibulae of G. arizonae vicular and cuboid; and distally, the three cunei­ are variable in the orientation of the muscular forms. Articulation of the tarsus is a complex groove. In UMMP 46232 and USNM 10536 the system of buttress and support. muscular groove is less obliquely disposed than in The proximal trochlear facet of the astragalus UMMP 46231 and UMMP 33524. Apparently is the primary fore-and-aft articulation of the the more slender proportions and the lesser obliq­ tarsus. The torsion produced by the outward uity of the muscular groove are correlative, per­ rotation of the tibiofibula at thisjoint is succeeded haps indicating sexual dimorphism. The fifth by an additional displacement of the rotational tibiofibula, MU 2670, is not sufficiently complete axis at the astragalus-navicular joint. The latter to provide an adequate comparison. joint completes the serial torsion of the axes of The tibiofibula of G. floridanum (Figure 55) is rotation of the rear leg, resulting in a nearly known from only two specimens, both from Texas transverse rotation at this intertarsal joint, the (TMM 31141-19 and USNM 6071). These two most distal primary articulation of the rear leg. bones closely resemble the slender pair of G. The astragalar-calcaneum articulation is a ro­ arizonae tibiofibulae discussed above. They are tational buttress in the dorsoventral plane, sup­ slender and their muscular groove is less oblique porting the transversely directed astragalar-na- than in the more robust specimens. Distinctive vicular articulation. The latter joint is the major features of G. floridanum occur in the articular intertarsal articulation for weight transfer. An facets. The tibial (medial) portion of the proximal additional, but restricted, articulation in weight facet is somewhat elongate in the anteroposterior transfer is accomplished through the calcaneal- plane, rather than being a nearly circular depres­ cuboid joint. The three complete and separate sion as in G. arizonae and G. texanum. The distinc­ cuneiforms are tightly situated beneath the com­ tive distal articular facet is more asymmetrical at pressed and greatly expanded navicular. The lat­ the central constriction between the medial and erally situated cuboid lies adjacent to the ecto- lateral facets. cuneiform and navicular, equaling their proxi­ TARSAL SERIES.—The tarsal series (Figures 56- modistal thickness. Thus the distal extension of 62; and in articulation, Figures 74, 75) consists of the cuboid occupies a lateral position equivalent seven separate elements: the proximally situated to that of the cuneiforms; i.e., it functions as a calcaneum and astragalus; the intermediate na- separate fourth "distal" tarsal. The distal tarsals (cuneiforms and cuboid) form a tightly arranged series of functionally immobile elements of the tarsus. The flattened distal articular facets of the distal tarsals provide little rotational mobility at the tarsus-metatarsus joints. ASTRAGALUS (Table 42).—The astragalus pre­ sents a trochlear facet on the proximal and ante­ rior surface for articulation with the tibiofibula, two posterior articular facets for the calcaneum, and an anterodistal articular facet for the navic­ ular. The tibiofibular/astragalus articulation is directed obliquely with respect to the sagittal plane of the proximal articular facet of the tibi­ ofibula. This obliquity, measuring approximately FIGURE 55.—Right fragmentary tibiofibula of Glyptotherium 30° inward, is imparted mainly by the torsion of floridanum (USNM 6071): a, anterior, with most of fibular portion broken away; b, medial (inner surface) of tibia. (Bar the tibiofibula. = 40 mm.) In the astragalus of G. texanum (Figure 56), the 132 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 56.—Right astragalus oi Glyptothenum texanum (F:AM 95737): a, medial; b, proximal; c. anteroproximal; d, lateral; e, distal;/, posterior. (Bar = 20 mm.) lateral (fibular) rocker of the trochlear facet is border by a sulcus separating the articular surface somewhat larger and more elevated than the from the navicular tuberosity. medial (tibial) rocker. In anterior profile, the The posterior surface bears two facets for artic­ lateral rocker is more deeply convex than the ulation with the calcaneum. The lateral calcaneal medial, and the trough of the trochlea is situated facet is flat and irregular in outline, suggesting a closer to the lateral rocker. Thus the concavity of distorted ovoid shape, with the blunt end of the the anterior profile of the trochlea is asymmetri­ oval directed upward. This facet occupies the cal, somewhat steeper on the lateral side than on entire lateral half of the posterior surface of the the medial. This configuration conforms to the astragalus. It is separated from the medial calca­ deeper fibular facet, and the shallower tibial facet neal facet by a deep, obliquely directed, interar­ of the distal articulation of the tibiofibula. In ticular fossa. A small, laterally directed eminence, addition, the lateral (fibular) rocker of the astrag­ the lateral calcaneal tuberosity, projects from the alus covers a longer chord, although the radius of anterolateral border of this facet, beyond the curvature of each rocker is approximately equal. vertical plane of the lateral surface of the fibular The chord of the lateral rocker traverses nearly a (lateral) rocker of the trochlea. semicircle, the main addition relative to the op­ The medial calcaneal articular facet (susten- posing rocker being in the anterior extent. The tacular facet) is subrectangular in outline, nearly medial (tibial) rocker is bordered on its postero­ twice as long as wide. The long axis is situated medial corner by a pronounced eminence, the obliquely, in the distolateral/proximomedial medial calcaneal tuberosity, and on its anterior plane. The distal half of this facet is flat, its NUMBER 40 133 surface lying in a plane, which is oblique with cupying the same position as the concave upper respect to the adjacent lateral calcaneal facet, extremity of this facet in the younger individual. and directed slightly proximally and medially There is a suggestion of the incipient development away from it. The proximal half of the susten- of a similar separation between these two portions tacular facet is concave, flaring upward to ter­ of the facet in the left astragalus of F:AM 95737, minate immediately beneath the medial calca­ whereas the facet is continuous for the right neal tuberosity. This facet is uniformly flat in its astragalus. transverse plane (short axis of the rectangular Except for their greater size, the known astra­ outline). gali of G. arizonae are identical in all respects to The facet for articulation with the navicular those known for G. texanum. As shown in Table occupies most of the distal extremity, as an an- 42, there is a pronounced size difference between terodistal condyloid surface projecting from a UMMP 46231 and UMMP 38761, both from the short, indistinct neck. The navicular facet is same locality. Although both are adults, the na­ strongly convex in the transverse direction and vicular and calcaneal tuberosities of the latter are only slightly convex anteroposteriorly. In distal very underdeveloped with respect to those of the aspect its outline is nearly the shape of a quarter- former specimen. The sustentacular facet is con­ circle, the apex forming rather less than a right tinuous in G. arizonae. angle and situated at the proximal extremity of The astragalus of G. floridanum, known from a the facet. Immediately above the apex lies the single specimen only (USNM 6071), is interme­ navicular tuberosity. The long axis of the nearly diate in size between G. texanum and G. arizonae. It cylindrical surface of the navicular facet is di­ is distinct in possessing a more nearly symmetrical rected obliquely with respect to the rotational trochlear facet, in which the trochlear groove is axis of the trochlear facet, increasing the torsion very shallow. The rockers are nearly identical in of the lower limb at this joint. Thus, the additive anterior profile, although the medial (tibial) effect of torsion at the tibiofibular/astragalar and rocker is less extensive anteroposteriorly, as in the the astragalar/navicular joints produces a rota­ other species. In addition, the calcaneal facet is tional axis at the distal joint, which is nearly more nearly equidimensional, rather than elon­ perpendicular to the femoral/tibiofibular joint. gate. The anterior extremity of the navicular facet This perpendicular rotational motion of the pri­ and the posterior extremity of the sustentacular mary tarsal joint with respect to the rotational facet are broken in this specimen. These facets do motion of the knee is probably a consequence of not differ significantly from those of the other the immobility of the pelvis during locomotion. species. The navicular articular facet is the only distal CALCANEUM (Table 43).—The calcaneum of G. articulation of the astragalus. There is no cuboid texanum (Figure 57) bears three articular facets, articulation, nor are there any articulations with two for the astragalus and one for the cuboid. the distal tarsal elements. The transversely compressed calcaneal process As indicated in separate adult (F:AM 59586) (tuber calcis) is rectangular in lateral aspect, its and juvenile astragali (F:AM 95737) of G. tex­ proximodistal length approximately twice the anum, there is modification with increasing age, height, and with a slight constriction at the neck. despite only a slight increase in size. There is not In superior aspect the same constriction is evident only a relatively greater development of the tu­ on the medial surface of the neck, but the lateral berosities, but also there is a peculiar modification surface instead bears only an elongate dorsolat­ of the sustentacular facet in the adult stage. In eral concavity, with an indistinct lateral crest F:AM 59586, the sustentacular facet is evidently forming the lateral profile. The free end (distal separated into two portions: a flat posterior sur­ extremity of the calcaneal process) in the subadult face, and a rounded posteromedial surface, oc­ condition is formed by the unfused epiphysis, 134 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

F'IGURE 57.—Right calcaneum oi Glyptotherium lexanum (F:AM 95737): a, distal extremity; b, anterior; c, lateral; d, posterior; e, medial. (Bar = 5 cm.) which joins the neck at an oblique proximo- into the depression between the bifurcate poster­ lateral/distomedial angle. omedial tubercles of the navicular, inhibiting ex­ The flat and ovoid lateral astragalar facet is tension at this joint in the tarsal series. directed proximomedially with respect to the su­ The cuboid facet forms the lateral side of the perior crest of the calcaneal process. It is separated proximal end of the calcaneum. It is slightly from the proximomedially situated sustentacular concave, nearly ovoid in shape, and proximally facet by a deep sulcus and from the cuboid facet directed. A small lateral tubercle projects from only by a shallow interarticular region. the margin of this facet to form the lateral extrem­ The sustentaculum projects proximomedially, ity of the proximal end, and a large inferior forming a massive tuberosity, upon which is lo­ tubercle forms the lower extremity. The latter cated the sustentacular facet for articulation with tuberosity occupies a position above the superior the medial (sustentacular) facet of the astragalus. depression of the posterior tubercle of the cuboid, This facet reflects its partner on the astragalus. producing a locking mechanism, which aids in The subrectangular outline is rendered asymmet­ inhibiting extension of the calcaneal-astragalus/ rical by a proximal extension of the upper convex cuboid joint. portion, increasing the width of the outline of this The calcaneum of G. anzonae (Figure 75) differs facet at the superior extremity. The long axis of from that of G. texanum principally in features this facet is oriented in the proximolateral/disto- related to its greater size. In place of the dorso­ medial direction, forming an oblique angle with lateral concavity on the neck of the calcaneal respect to the lateral astragalar facet. The proxi­ process of G. texanum, there is a deeply excavated mal and inferior flat surface is directed proxi­ sulcus. This proximodistal excavation occupies mally and slightly laterally. The upper portion of the superior half of the lateral surface of the this facet is uniformly convex about the long axis calcaneal process. The inferior border of this of the facet, its chord traversing nearly a semicir­ groove is formed by a distinct lateral crest, curv­ cle. The extremity of this facet occupies the me­ ing proximally upward to participate as the bor­ dial surface of the sustentaculum. The entire facet der of the lateral tubercle. The furrow emerges is uniformly flat in the short-axis direction. The proximally between the astragalus facet and the inferior portion of the sustentaculum fits snugly lateral tubercle at the dorsolateral angle of the NUMBER 40 135

proximal extremity of the calcaneum. Although this furrow may simply reflect allometric changes resulting as a consequence of the achievement of larger size in the calcaneum of G. anzonae, this character seems to distinguish the species, for it is similarly developed in all three calcanea recog­ nized as G. arizonae (USNM 10536, UMMP 46231, UMMP 38761), and it is not present in G. floridanum, for which the size of the calcaneum approaches that of G. arizonae. The sustentacular facet is contiguous in two of the G. arizonae speci­ mens (USNM 10536 and UMMP 38761), and it appears to be separated into two portions by a weak angle in the other (UMMP 46231). The remaining characteristics of the calcaneum of G. arizonae are related to the larger size in this species, including elaboration of tuberosities and exagger­ ation of muscle attachment surfaces. The calcaneum of G. floridanum is known only for one individual, USNM 6071, for which both were recovered; one is nearly complete, the other is fragmentary. The size approximates that of G. arizonae. The distal extremity of the calcaneal process of G. floridanum bears a prominent down­ ward-directed tubercle on its inferior angle. Rel­ ative to the "horizontal axis," this tubercle ex­ tends below the level of the proximal extremity of the calcaneum. Moreover, a vertical groove marks the rounded extremity of the free end, passing over the posterior tubercle on its medial side, in contrast with the rounded and rugose free extremities in G. arizonae and G. texanum. The neck of the calcaneal process is smooth and unexca- vated, as in G. texanum, although there is a weak concavity. The astragalar facet is less elongate and the cuboid facet is more deeply concave than in either G. arizonae or G. texanum. NAVICULAR (Table 44).—The greatly expanded navicular of G. texanum (Figure 58) presents a single, broad cotyloid facet for articultion with the astragalus. In proximal aspect the shape of this facet is irregularly ovoid, and elongate in the FIGURE 58.—Right navicular of Glyptotherium texanum (F: AM transverse plane. The facet is concave in this 95737): a, proximal; b, distal. (Bar = 20 mm.) direction, and only slightly concave anteroposte­ riorly, imparting a close resemblance to a portion of the inner surface of a cylinder. 136 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Two massive posterior tuberosities characterize navicular. It is ovoid, slightly concave, and di­ the navicular. The outer (lateral) tuberosity is rected distomedially. It is located in direct line pointed, laterally compressed, and curved up­ with the medial extremity of the proximal facet, ward and medially, projecting farther from the occupying the central part of the "neck" portion articular facet than the medial tuberosity. The of the distal surface of the medial tuberosity. inner (medial) tuberosity is a blunt fusiform There is, in addition, an elongate, laterally process extending posterolaterally as a greatly directed facet for articulation with the cuboid. thickened surface for muscle attachment. On the This flattened facet occupies the lateral margin undersurface of the "neck" of this tuberosity is of the navicular. Continuing above the posterior the articular facet for the internal cuneiform. In extremity of the facet is the flattened lateral articulation, the sustentaculum of the calcaneum tuberosity, the lateral surface of which is situated fits snugly between these tuberosities. in the same plane as the cuboid facet. The net The distal facets of the navicular, one for each result of the three flat, distal articular facets of the three cuneiform bones, are situated directly produces in gross proportions a convex surface beneath the proximal facet. The lateral facet for parallel to the upper concavity of the proximal the external cuneiform is the largest. It is trian­ facet. Thus, the overall configuration of the na­ gular and flattened, conforming exactly to the vicular is that of a concavoconvex discoid, with proximal facet of the external cuneiform. There two posterior tuberosities. The maximum thick­ was little, if any, mobility at this joint. The shape ness of the disc is between the posterior border of of this facet closely resembles a right triangle with the proximal facet and the posterior margin of rounded apices, with the hypotenuse formed by the distal middle cuneiform articular facet. The the internal, anteroposteriorly directed border shortest proximodistal diameter is at the anterior facing the middle cuneiform. The rear portion of border of the same two facets. the facet grades imperceptibly into the distinct The navicular of G. arizonae is identical in most interarticular region separating this facet from respects to that of G. texanum, except for its larger that for the middle cuneiform. The lateral margin size and the corresponding accentuation of the of the facet forms a common angle with the articular facets and tuberosities. The proximal laterally directed facet for the cuboid. articular facet in G. arizonae is distinctly more The articular facet for the middle cuneiform is concave in the anteroposterior direction, espe­ trapezoidal and elongate in the anteroposterior cially near its posterolateral margin, where the direction. It is distally directed, oblique with facet is inflected upward. This articulation is respect to the somewhat laterally directed facet more nearly cotyloid than the cylindrical surface for the external and internal cuneiforms. The of the proximal facet of G. texanum. Similarly, the facet is situated in the plane of elongation of the distal facets, especially the facet for the internal lateral posterior tuberosity. The surface is irreg­ cuneiform, are markedly more concave in the ular, but mostly flattened, and slightly concave transverse direction. The inferior border of the transversely. This is by far the longest of the distal lateral posterior tuberosity is medially inflected, facets; its greatest transverse diameter is slightly providing a distinct channel beneath which the greater than that of the internal cuneiform facet. strong tarsal musculature passed. This inflection The lateral border of the middle cuneiform facet is only incipiently developed in G. texanum. With is slightly elevated above the interarticular region, this posterodistal inflection of the lateral tu­ separating it from the lateral cuneiform facet. berosity, the lateral profile of the navicular of G. The medial border contacts along its middle half arizonae is squared at this angle, rather than the lateral border of the internal cuneiform facet. rounded. The facet for articulation with the internal The navicular is unknown for G. floridanum and cuneiform is the smallest of the distal facets of the G. cylindricum. NUMBER 40 137

CUBOID (Table 45).—The cuboid of G. texanum other cuboid facets by shallow interarticular re­ (Figure 59) is an irregular bone, presenting three gions. articular facets and a posterior tuberosity. It is The flat distal articular facet is slightly convex anteroposteriorly elongate, in length nearly twice in the anteroposterior direction. Marginal projec­ the subequal transverse and proximodistal di­ tions of this facet produce a weakly constricted mensions. The cuboid functions as the posterolat­ "neck," upon which the facet is situated. A small eral element of the distal tarsal series and as a fiat facet, located on the anteromedial portion of lateral buttress adjacent to the navicular. The this "neck," is continuous with the distal facet as elongate facet for articulation with the navicular a transverse articulation with the metatarsal. is flattened and triangular in shape. It occupies The posterior tuberosity extends rearward to approximately half of the anteromedial surface of conform with the shape of the lateral tuberosity the cuboid, conforming exactly to the opposing of the navicular. It presents a flat triangular facet of the navicular. It is bounded on all sides extremity facing posteriorly and occupying the by distinct interarticular regions, a deeply exca­ posterolateral position of the distal tarsal series. vated inferior interarticular region separating this It is separated from the more anteriorly placed facet from the metatarsal facet. calcaneal and distal articular facets by a distinct The anterolateral facet of the cuboid articu­ neck. The neck of the posterior tuberosity is con­ lates with the distolateral facet of the calcaneum. stricted on its outer surface and bears on its inner It is flattened, with convex upper and lateral surface the elongate navicular articular facet. borders, fitting loosely into the opposing facet of The cuboid of G. arizonae differs from that of G. the calcaneum. Although this joint was undoubt­ texanum mainly in size and the accentuation of the edly important in weight transfer, it appears to articular facets and tuberosities. The only notable have been more important as a lever for trans­ distinction is the squared, rather than triangular, mission of calcaneal exertions through the tarsus shape of the posterior tuberosity in G. arizonae. to the metatarsus and digits. This facet faces The undersurface of the neck of the tuberosity is outward and upward and is separated from the clearly divided into two regions: a deep anterior

FICURE 59.—Right cuboid oi Glyptotherium texanum (F:AM 95737): a, distal; b, proximal; c, anterior. (Bar = 20 mm.) 138 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

sulcus passing transversely just posterior to the a convex anterior outline, a concave posterome­ distal articular facet, and a posterior roughened dial border, and an irregularly shaped posterolat­ portion behind the sulcus. eral border. The proximal facet is slightly concave In USNM 10536, the articular facet for the and is gradually elevated toward the anterolateral navicular is heavily rugose and tuberculated due angle, there producing the greatest proximodistal to degenerative arthritis (or some similar condi­ dimension of the bone. The posterior apex of the tion) in this individual. Moreover, this condition distal facet is somewhat more extensive than the indicates that little mobility was obtained proximal, and it projects farther rearward, result­ through the cuboid-navicular articulation. The ing in a downward taper of the interarticular arthritic condition is so extensive at this joint that surface between these apices at the posterior ex­ possibly there could have occurred a solid fusion tremity. of the cuboid with the navicular with advanced Two small laterally directed facets for the cu­ age. boid continue around the lateral angle from the EXTERNAL CUNEIFORM (Table 46).—The exter­ proximal and distal facets. There is a distinct nal cuneiform of G. texanum (Figure 60) represents anterolateral tubercle and a shallow transverse an asymmetrical section of a discoid, with the sulcus on the nonarticular anterior face. A pos- apex in the posterior position. The chord covers teromedially directed semicircular facet meets the approximately 60°, the internal leg of the proxi­ distal facet at a right angle on the internal border mal (and distal) profile is deeply concave, and of the external cuneiform. This facet occupies the the outer leg is nearly straight. The proximal and anterior extremity of the internal border for ar­ distal facets, for articulation with the navicular ticulation with the opposing facet of metatarsal and metatarsals, respectively, are parallel and II. nearly identical. Both are fiat and nearly trian­ The external cuneiform of G. arizonae bears all gular with equidimensional proportions, having the characters of that of G. texanum, with several

FICURE 60.—Righl ectocuneiform oi Glyptotherium texanum (F: AM 95737): a, anterior; b, proximal; c, distal. (Bar = 20 mm.) NUMBER 40 139 notable additions. The distal facet is slightly con­ cave in the transverse direction and slightly con­ vex anteroposteriorly. A nearly straight extension of the proximal surface over the concavity of the internal border produces a cup-shaped inter­ articular region for reception of the proximola­ teral angle of the second metatarsal. The lateral and medial facets for articulation with the cuboid and second metatarsal, respectively, are less well developed, although the surfaces of these poster­ olateral and posteromedial borders are more heavily rugose. The anterolateral tubercle on the anterior border is more massively developed, and the anterior border of the distal facet projects as an anterior flange, accentuating the transverse sulcus on the anterior border. The external cu­ neiforms of the Seymour glyptodonts are identical in size and construction to the Curtis Ranch specimen. This element is unknown for G. floridanum. MIDDLE CUNEIFORM (Table 47).—The middle cuneiform of G. texanum (Figure 61) is an antero­ FIGURE 61.—Right middle cuneiform oi Glyptotherium texanum posteriorly elongate bone, bearing a flat proximal (F:AM 95737): a, medial; b, distal; c, proximal. (Bar = 20 facet for the navicular and a concave distal facet mm.) for metatarsal II. The posterior half of the bone is fiat, with parallel upper and lower facets. The gate, posterior tubercle projects from the posterior distal concavity produces increasing proximodis­ face beyond the extent of the articular facets. tal thickness anteriorly. The lower facet is ^di­ Similarly on the anterior face, a distinct tubercle rected distally on its posterior half, becom.ng marks the anteromedial angle. The distal articu­ more medially directed toward the anterior face. lar facet is relatively less concave. The margins of This produces a quadrilateral anterior aspect both articular facets are irregular, rather than with three right angles and a lower oblique side straight, and the medial margin of the distal facet forming an acute distomedial angle and an obtuse is distinctly concave in outline. A notable distinc­ distolateral angle. Hence, the greatest proximo­ tion of USNM 10536 is a deep groove on the distal diameter occurs at the anteromedial angle. anterior face that extends downward and presum­ The proximal facet conforms with the opposing ably marks the position of a strong tarsal-meta- facet of the navicular, the distal facet with the tarsal tendon. This sulcus is absent in UMMP proximal facet of metatarsal II. There was little 46231 and UMMP 38761. mobility at either joint. Distinct vertical interar­ INTERNAL CUNEIFORM (Table 48).—The inter­ ticular regions separate the proximal and distal nal cuneiform of G. texanum (Figure 62) is the facets around the circumference of the bone. smallest of the distal tarsals. It occupies the dis­ The middle cuneiform of G. arizonae is similar tomedial position of the tarsus, providing the only to that of G. texanum, the only obvious differences articulation for the first metatarsal. The internal apparently related to size. The proximal facet cuneiform is directed medially and distally from does not cover the entire upper surface poste­ the proximomedial surface of the navicular. The riorly; instead, a prominent, transversely elon- articular surface for the navicular is oval and 140 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

tarsals II and IV are intermediate in size, and I and V are the smallest. They are arranged in a near-perfect semicircle. All are stout elements approaching cubic proportions; i.e., the proxi­ modistal diameters are only slightly greater than the transverse and anteroposterior diameters. Each of the three middle metatarsals possesses a pair of large plantar sesamoid bones. The two outer metatarsals each possess only a single sesa­ moid bone, considerably smaller than the internal sesamoids. All of the metatarsals are known for G. texanum and G. anzonae, and III and IV are known for G. FIGURE 62.—Right entocuneiform oi Glyptothenum texanum (F: floridanum. They exhibit a close resemblance to AM 95737): a, proximal; b, distal. (Bar = 20 mm.) those described by Burmeister (1874) for Glypto­ don. slightly convex. It covers nearly the entire convex Melton (1964) was the first to attempt an proximal surface of the bone. There was little enumeration of the elements of the pes, although mobility at this joint, and little weight was trans­ he did not compare the pes of the Seymour mitted to the first digit. A small, distomedially glyptodont with that known for the Curtis Ranch directed, convex distal facet for metatarsal I is representative, which was briefly described and oriented obliquely with respect to the proximal figured by Gidley (1926). Describing the Seymour facet. These facets are posteriorly convergent in specimens, Melton (1964:137-138) stated: an acute angle of approximately 45°. A distinct There are 39 bones in each of the complete left feet: 7 posterior tubercle projects beyond the extent of tarsals, 5 metatarsals, 15 phalanges, and 12 sesmoids [sic]. the proximal facet, and a small elongate tubercle Burmeister lists 14 phalanges and 10 sesmoids [sic] for Glyp­ on the medial side projects downward from the todon asper. He found no 2d phalanx and no distal sesmoids [sic] on digit I. In both of the feet recovered, digit I had 2 border of the proximal facet. phalanges and 2 distal sesmoids [sic]. The internal cuneiform of G. arizonae is identical in all respects except for larger size. In duality, Burmeister found two phalanges METATARSALS.—The five metatarsals articulate for digit I, accounting for the phalangeal count with the distal tarsal bones in flat contacts. Meta­ of 14. Moreover, there are 14, rather than 15, tarsals I, II, and part of III articulate with the phalanges in the Seymour pes. The count of 14 cuneiforms; part of metatarsal IV and all of V phalanges accounts for only two phalanges in articulate with the cuboid. There is a unique digit I. The two sesamoid bones Melton identified system of overlap between adjacent metatarsals as "distal" for digit I are in actuality single prox­ involving II-V. In each there is a proximolateral imal and distal elements, one articulating at the extension, which is supported beneath by a prox- metatarsal-phalanx joint, the other at the termi­ imomedial facet of the laterally adjacent meta­ nal joint. The correct sesamoid count for G. ari­ tarsal. This arrangement interlocks the meta­ zonae and G. texanum is 13, rather than 12 as tarsals at their proximal extremities, limiting Melton asserted. movement, and serving to distribute weight to­ Thus there is a closer similarity with the South ward the lateral digit. It is by this mechanism American genus than Melton recognized, for the that the symmetrical disposition of the digits and counts for the various elements of the pes are consequently their nearly perfect symmetry in identical. Burmeister's (1874) descriptions and function are achieved. figures do indeed indicate a close resemblance. The middle metatarsal is the largest. Meta­ Neither Melton (1964) nor Gidley (1926) de- NUMBER 40 141 scribed the metatarsals, although both provided articular facet. The articular surface for the pha­ illustrations. lanx is anteroposteriorly flat and transversely con­ METATARSAL I (Table 49).—The first meta­ cave. The rear portion of this facet, for articula­ tarsal of G. texanum (Figure 63) is a small, stout tion with the sesamoid, is convex anteroposte­ element bearing two articular facets. In proximal riorly and transversely flattened. A distinct view its shape is oval, rounded posteriorly and boundary delimits these portions of the distal rather more pointed anteriorly. The proximal facet. facet is shallowly concave, for reception of the The marginal tubercles are distinctly more well distal facet of the internal cuneiform. developed. A large medial groove and a smaller The distal facet of metatarsal I is anteroposte­ lateral groove mark the lower surface of the body riorly elongate, extending farther anteriorly than of the bone, immediately above the margins of the proximal facet, producing a tapered anterior the distal facet. The tubercles project over these border in lateral and medial aspects. The distal tendinal grooves, accentuating their depth. facet is convex in the anteroposterior direction SESAMOID BONE, METATARSAL I (Table 54).— and weakly concave in the transverse direction. One plantar sesamoid is situated at each of the The articular facets are basically parallel, but the joints of the first digit, rear foot of G. texanum. The distal facet is laterally offset so that its medial more proximal digital sesamoid, located at the border lies almost directly beneath the center of metatarsal-phalanx joint, is approximately twice the proximal facet. The distal facet articulates on as long as the distal one. It articulates primarily its anterior two-thirds with the first phalanx, digit with the posterior portion of the distal facet of I, and on its posterior third with a stout single the metatarsal in a slightly dorsally directed facet; sesamoid bone, which projects posteriorly from it articulates secondarily below with the rear the long axis of the digit. The posterior border of down-turned portion of the proximal facet of the the distal facet extends a short distance beyond first phalanx in a small lower facet. This articular the interarticular "neck" as a small flange along facet of the sesamoid is flat and situated at nearly the posterodistal margin. Small proximal and a right angle with respect to the long axis of the distal tubercles are present on the medial surface element; the lower portion is downward-directed of the interarticular area; a slightly raised tuber­ to meet the phalanx. This sesamoid bone extends cle is situated along the anterior margin; and a in the plantar direction as a somewhat trans­ prominent bosslike tubercle is located o, the versely flattened cylinder with a rounded, knob­ posterolateral angle of the shaft. like and medially inturned free end. It is heavily Metatarsal I of G. arizonae is distinct in having rugose on the nonarticular surfaces. a greater definition of the boundary between the The metatarsal sesamoid of G. arizonae is larger, phalangeal and sesamoid portions of the distal more transversely compressed, and more heavily

FIGURE 63.—Right metatarsal I oi Glyptothenum texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) 142 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY rugose than the same element in G. texanum. The outward rotation laterally beyond the long axis articular facets are more transversely expanded of the digit. Because of the tight-fitting contact and somewhat more dorsally directed. These dif­ with the adjacent tarsals and metatarsals, there ferences are within the expected range of onto­ appears to have been little proximal motion for genetic variation. The element is otherwise iden­ the second metatarsal. tical in these two species. Metatarsal II bears a saddle-shaped proximal This metatarsal sesamoid in UMMP 46231 articular facet, convex in the dorsoplantar direc­ exhibits considerable degeneration of the bone, as tion, and transversely concave, with the posterior do the remaining bones of the digit, as a manifes­ extension becoming transversely flattened. This tation of the apparent arthritic malady suffered facet extends laterally beyond the lateral surface by this individual. of the body covering the upper surface of the METATARSAL II (Table 50).—The second meta­ ectocuneiform process. The proximal articular tarsal in G. texanum (Figure 64) is a complexly facet conforms exactly with the opposing facet of modified rectangular prism, only slightly longer the middle cuneiform. in the proximodistal dimension than in the sub- The ectocuneiform tubercle is a squared exten­ equal dorsoplantar and mediolateral dimensions. sion of the proximolateral angle, with flat artic­ It bears proximal and distal facets for articulation ular surfaces on its upper side, as the extension of with the middle cuneiform and first phalanx of the proximal articular facet, and on its lateral the second digit, respectively. There is an addi­ and lower sides to interlock with the ectocunei­ tional pair of proximal articular facets, a lateral form and metatarsal III, respectively. The ecto­ one for vertical contact with the external cunei­ cuneiform facet occupies the entire lateral ex­ form, and a distally directed one beneath the tremity of the ectocuneiform process. It is rectan­ large ectocuneiform process on the proximolateral gular and flat, facing laterally and slightly out­ angle for articulation with metatarsal III. There ward. The long axis of the rectangle is directed are also two plantar-distal articular facets for the downward at approximately 45° with respect to pair of sesamoids located at the metatarsal-pha­ the long axis of digit II. The undersurface of the lanx joint of this digit. Proximal articulation of ectocuneiform process, for articulation with meta­ this bone includes contact with three other ele­ tarsal III, is semicircular and flat, forming an ments, the middle and external cuneiforms and external right angle with the ectocuneiform facet. the third metatarsal. The net effect of this artic­ Its surface is parallel with the upper anterior ulation is a strong lateral buttress preventing surface of the proximal facet, facing in the distal

FICURE 64.—Right metatarsal II oi Glyptotherium texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) NUMBER 40 143 and plantar direction, and slightly laterally. The excavated and considerably larger than the facet ectocuneiform process occupies the anterior half for the lateral sesamoid. of the proximolateral extremity of the bone. SESAMOID BONES, METATARSAL II (Table 54).— The shaft, or body, of metatarsal II bears four The two plantar sesamoids of the second meta­ modified surfaces. The lateral and medial surfaces tarsal articulate exclusively with this element. are irregularly concave. The anterior surface, be­ They are both transversely compressed and irreg­ tween the proximal and distal articular facets, is ular in shape. The medial sesamoid is nearly weakly concave, and the posterior surface is oc­ twice as large as the lateral one and has a consid­ cupied almost entirely by the sesamoid facets. erably wider articular surface. The articular Only a small blunt tubercle underlying the pos­ facets are concave in the sagittal direction and terior extension of the proximal facet represents transversely convex, conforming with the oppos­ the nonarticular portion of the rear surface of the ing facets of the metatarsal. The free ends of both body. sesamoids terminate in a blunt, knoblike exten­ The sesamoid facets are posterior continuations sion that is curved laterally, imparting a convex of the distal articular facet. They are proximodis­ medial and concave lateral surface for each bone. tally elongate, the larger medial one extending Except for the much larger absolute size in G. farther proximally, and the smaller lateral one arizonae these bones are identical to those of G. extending farther distally. They are transversely texanum. concave and are separated by an indistinct ridge, METATARSAL III (Table 51).—The third meta­ which is a continuation of the flattened lateral tarsal of G. texanum (Figure 65) is the largest and portion of the distal facet. stoutest of the metatarsal series. It presents pri­ The distal articular facet, for the first phalanx, mary proximal and distal articular facets, a pair digit II, is saddle-shaped, convex in the dorso­ of secondary facets for interlocking with the ad­ plantar direction, and transversely concave. It jacent metatarsals, and two posterior facets for narrows gradually anteriorly, reaching a rounded the plantar sesamoids. anterior extremity, which is directed as much The nearly flat, slightly convex, proximal artic­ dorsally as distally. The lateral fourth of the facet ular facet meets the corresponding facet of the is transversely flat, indistinctly separated from the middle cuneiform in a broad and expanded con­ medial concave portion by a weak angle, and tact. The anterior (dorsal) margin of this facet is extends somewhat farther in the distal direction. rounded. The medial border, which forms a com­ The second metatarsal of G. arizonae exhibits mon angle with the facet for metatarsal II, is minor modifications of the features described for straight and directed laterally and rearward from G. texanum. The proximolateral angle of the ecto­ the anteromedial corner of the anterior arch. The cuneiform process is marked by a deep dorsoplan­ lateral border, forming an acute common angle tar sulcus separating the proximal facet from the with the facet for metatarsal IV, is nearly straight, laterally directed ectocuneiform facet. Moreover, and it is parallel with the sagittal plane. The the ectocuneiform facet is convex, rather than posterior portion of this facet includes a squared flat, in the direction of its long axis. The most extension covering the posterior tubercle. The distinctive character of metatarsal II of G. arizonae proximal facet is directed anteriorly with respect is the relatively short ectocuneiform process, in to the distal articular facet, and these two facets absolute comparison approximately equidimen- are anteriorly convergent in an acute angle of sional with that of G. texanum, despite the other­ approximately 45°. The lateral portion extends wise pronounced difference in size between these over the lateral metatarsal process, the undersur­ two elements. The transverse concavity of the face of which is occupied by a convex articular proximal facet is less pronounced, while the artic­ facet. This facet narrows from before rearward ular facet for the medial sesamoid is more deeply and is oriented distolaterally. It interlocks in a 144 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 65.—Right metatarsal III oi Glyptothenum texanum (F":AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) tight articulation with the medial concave por­ proximodistally concave and transversely convex; tion of the proximal facet of metatarsal IV. and the posterior surface is occupied primarily by The medial metatarsal process is less extensive. the sesamoid facets and the posterior tubercle. A Upon its upper surface is situated the facet for less prominent tubercle is situated near the lateral metatarsal II, which exhibits a flattened, four- margin of the anterior surface, immediately be­ sided shape, the two external corners forming neath the proximal facet. right angles, and the two internal corners formed Metatarsal III of G. arizonae (Figure 75) differs by the obliquely directed common border with in possessing weakly concave, rather than flat, the middle cuneiform facet. This metatarsal facet proximal articular facets (cuneiform and meta­ is directed anteriorly with respect to the sagittal tarsal II facets), and the distal articular surface is plane and somewhat medially away from the deeply excavated in its posterocentral position, in cuneiform facet. place of the median furrow of G. texanum. The The distal articular facet is anteroposteriorly tubercles are more pronounced, and medial and convex and weakly concave in the transverse lateral excavations on the surface of the shaft direction. From the rounded anterior border the accentuate the greater size. facet narrows posteriorly. Its posterior border Metatarsal III of G. floridanum (Figure 73a) is joins the posterior sesamoid facets in a rounded, intermediate in size. It most closely resembles indistinct angle. A pair of small nutritive foram­ that of G. arizonae. It differs from G. texanum in ina is situated within the poorly defined central possessing the same additional features as those furrow of the facet, at the junction with the described for G. arizonae, but it also exhibits sev­ medial ridge of the sesamoid facets. eral important distinctions separating it from The sesamoid facets are proximodistally elon­ both species. The boundary between the proximal gate and convex, and each is concave in the articular facet and the medial facet for articula­ transverse direction. They are separated by a tion with metatarsal II is more sharply defined. sharp median longitudinal ridge, and they are The concavity of the proximal facet is as pro­ equally developed. These sesamoid facets occupy nounced as in G. arizonae, but its direction is more the distal two-thirds of the posterior surface of anteriorly oriented. The articular facet for meta­ the bone. Immediately above these facets is the tarsal II is more distinctly concave, and that for posterior tubercle. metatarsal IV lacks the proximodistal convexity The four-sided shaft of metatarsal III is irreg­ of the other two species. The sesamoid articular ular. The medial and lateral sides are posteriorly facets are relatively less elongate than in G. ari­ convergent; the expanded anterior surface is zonae. The distal articular facet entirely lacks the NUMBER 40 145 central nonarticular depression of G. arizonae, lation with the adjacent metatarsals. There is an which is incipiently developed in G. texanum. expanded distal facet for articulation with pha­ SESAMOID BONES, METATARSAL III (Table lanx I and a two-parted sesamoid facet for the 54).—The two plantar sesamoids of metatarsal two plantar sesamoids. The shape of metatarsal III in G. arizonae and G. texanum are symmetrically IV is irregular. Its maximum proximodistal di­ disposed, mirror-image equivalents. They are in mension approximately equals the anteroposte­ proximal contact along the median ridge of the rior (dorsoplantar) dimension, and both exceed sesamoid facet of the metatarsal. These are by far slightly the maximum transverse diameter. the largest of the proximal digital sesamoids of The proximal facet, for articulation with the the rear foot, a condition indicating the primary cuboid and the ectocuneiform, bears no evidence importance of digit III in weight support. These of the relative importance of these two proximal transversely compressed sesamoids each present articulations. They are situated in the same plane, lateral and medial surfaces and three borders: the and metatarsal IV evidently moved freely in the concave articular facet, the upper scalloped, dou­ transverse direction across their contact. The facet bly concave proximal (with respect to the proxi­ bears the shape of a parallelogram, with straight mal orientation of the metatarsal) border, and a marginal sides nearly twice the length of the lower uniformly curved border connecting the convex anterior and posterior borders. The ante­ distal end of the facet with the upper border. rior and posterior borders are directed generally Thus, in lateral aspect, the shape is that of a half- anteriorly and medially from the lateral angles. crescent. The external (parasagittal) surface is This outline is interrupted along the posterior weakly concave, while the internal surface (facing half of the medial border, which is irregular. The the sagittal plane of the metatarsal) is proximally anterior half of the same border forms a common flattened and rugose and is concave in a longi­ angle with the facet for metatarsal III. Similarly, tudinal furrow near the free border. This concav­ the anterior half of the lateral border of the ity is mirrored in the opposing sesamoid, produc­ proximal facet forms a common angle with the ing a proximodistal tendinal channel, which is facet for metatarsal V. The posterior half of the anteriorly and marginally bordered by the sesa­ proximal facet is a rearward extension beyond moids and open to the rear. The upper margin of the level of the shaft. It is buttressed beneath by the free end forms a right angle with the upper the blunt posterior tubercle. The proximal facet end of the articular facet and an acute angle with is weakly concave in both the transverse and the broadly curved lower margin. The bone is anteroposterior directions, with its central con­ thickened along the lower margin and along the cavity directed from the posterolateral angle for­ articular facet. The articular surface is proximo­ ward and downward, reaching a low point at the distally (with respect to this orientation of the anteromedial angle. metatarsal) concave and transversely convex. The facet for metatarsal III occupies the anter- Except for larger size and exaggeration of fea­ oproximal third of the medial surface of the shaft. tures in G. arizonae, these bones are identical in It is transversely concave and is directed medially both species. upward and forward, receiving the corresponding METATARSAL IV (Table 52).—The fourth meta­ facet of metatarsal III in a curved contact. Its tarsal of G. texanum (Figure 66) is approximately medial margin is rounded and extends as a flared the same size or slightly smaller than metatarsal process in a medial flange. The anterior apex II, as a reflection of the basic symmetry of the forms the distal extremity of the facet, and it is rear foot. It possesses three proximal articular separated from the distal articular facet by a facets: a primary one for articulation with the narrow interarticular region on the anteromedial cuboid and the lateral angle of the external cu­ angle of the shaft. neiform, and a pair of marginal ones for articu­ The lateral facet for metatarsal V is triangular 146 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 66.—Right metatarsal IV oi Glyptotherium texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) and weakly convex in the proximodistal direction. veloped anteroproximal, posteroproximal, and Its orientation is roughly parallel to the sagittal posteromedial tubercles. plane. Metatarsal IV of G. arizonae (Figure 75) is sim­ The distal articular surface, for phalanx I, is a ilar in most respects to that of G. texanum. It is four-sided facet, convex in both the anteroposte­ considerably larger, but all proportions are ap­ rior and transverse directions and occupying the proximately the same except for those in the entire distal surface of the bone. The straight transverse dimensions, which are somewhat medial border is nearly twice as long as the lateral longer. The distal articular facet is rather more border. Hence, the broadly rounded anterior mar­ flattened, and it bears in the central position a gin reaches its anterior extremity at the rounded nonarticular excavation. The facet for metatarsal anteromedial apex, from which it tapers laterally V is rectangular, rather than triangular, but its and rearward, forming a rounded anterolateral situation and disposition are otherwise identical. angle with the lateral margin. The marginal bor­ The tubercles and excavations are more pro­ ders are posteriorly convergent. They meet the nounced, in keeping with the larger size in this posterior sesamoid facets at the posterodistal an­ species. gle, the border with the shorter medial facet In UMMP 46231, there is a large foramen in extending farther posteriorly than that for the the shaft on the posterior surface immediately lateral facet. The intersesamoid crest of the sesa­ above the sesamoid facets, and directed anteriorly moid facet continues as a weak ridge on the and upward. There are, in addition, two deep posterocentral surface of the distal facet. foramina on the anterolateral surface of the shaft Both sesamoid facets are proximodistally elon­ above the distal facet, penetrating deeply inward. gate, parallel to the sagittal plane, and concave These are likely pathologic, as in other elements in the transverse direction. They unite at the crest in this specimen. There is a similar condition on of the intersesamoid ridge, which is situated ap­ the ungual phalanx. proximately in the sagittal plane. They are ap­ Metatarsal IV of G. floridanum (Figure 13b) is proximately equal in length, but the lateral one intermediate in size. Despite the fact that it is is distinctly wider transversely. Together they smaller than that of G. arizonae, it is more trans­ occupy the distal half of the posterior surface of versely expanded, and the proximal articular the shaft. facet is disproportionately wider. Its most signifi­ The nonarticular regions of the shaft are irreg­ cant character is in the nature of the sesamoid ular and pitted. There are shallow lateral and articular facets, which, rather than parallel to the medial excavations, and there are modestly de­ sagittal plane, are directed distally inward, form- NUMBER 40 147

ing an oblique angle with the distal articular other concave. It is anteroposteriorly elongate facet. The crest separating these two facets is and irregular in shape. similarly oriented. Moreover, the shapes of these The proximal articular facet is a uniform con­ facets are less symmetrical, and there are three cave half ellipse, with the straight border on the massive foramina situated on the shaft at their medial side, and the rounded outer margin ex­ proximal border, where only small foramina are tending from the anteromedial angle to the pos­ situated in the other two species. teromedial angle, conforming with the outer por­ SESAMOID BONES, METATARSAL IV (Table tion of the opposing facet of the cuboid. The 54).—The lateral sesamoid of metatarsal IV in G. posteromedial angle is upturned and forms a arizonae and G. texanum is considerably larger than small pointed prominence. The facet for contact the medial. Both sesamoids are transversely com­ with metatarsal IV is triangular, with a straight pressed, and they present articular facets that are proximodistal margin on the anteromedial corner proximodistally concave and transversely convex, of the bone, an upper anteroposterior margin these conforming exactly with the sesamoid facets forming a common angle with the proximal facet, of the metatarsal. The articular facet of the lateral and a diagonal extending downward and forward sesamoid is approximately twice the width of that from a position near the proximomedial tubercle of the medial, although it is only slightly longer of the proximal facet. The anteroproximal angle in the proximodistal direction. The free end of is approximately 90°. This facet is directed me­ each sesamoid forms a posteroproximal apex sit­ dially and slightly proximally with respect to the uated above the upper level of the articular facet. proximal facet. The plantar border of the lateral sesamoid is The distal articular facet is rather less concave thickened, forming an expanded convex free end, than the proximal one. It is oval in shape, with while that of the smaller medial sesamoid forms the medial margin somewhat straightened. The a more compressed, relatively sharp plantar an­ posterior border is upturned for the plantar sesa­ gle. The inner surface of each is partly concave, moid, forming a poorly defined knoblike facet. together in opposition forming the tendinal chan­ The distal facet is directed somewhat laterally nel, as in the other metatarsal sesamoids. The with respect to the proximal facet, their respective outer surfaces are irregular. Except for the concavities convergent posterolaterally. The max­ marked size difference, these elements are iden­ imum proximodistal diameter occurs at the an­ tical in the two species for which they are known. teromedial angle. METATARSAL V (Table 53).—The fifth meta­ The medial surface of the shaft is flattened, tarsal occupies the lateral position of the meta­ and the remaining outer border is rounded. A tarsus. It articulates proximally with the lateral blunt tubercle occupies the posterolateral posi­ half of the cuboid, proximomedially with the tion. fourth metatarsal in a flat contact, and distally Metatarsal V in G. arizonae (Figure 75) is pro­ with phalanx I, digit V. It possesses an articular portionally similar to that of G. texanum. The surface for the single metatarsal sesamoid, which proximal articular facet is identical. The meta­ also contacts phalanx I. In G. texanum, metatarsal tarsal facet differs in having rounded margins, V (Figure 67) is approximately the size of meta­ with a distally directed apex. It contacts the tarsal I. It is distinguished from the medial meta­ proximal facet over a relatively shorter distance tarsal by the articular facet on the shaft for because of its rounded anterior margin. The distal metatarsal IV; metatarsal I lacks articulation articular facet is more transversely expanded, and with the adjacent metatarsal. Metatarsal V is its concavity is irregular. There is a mediocentral further distinguished by the two concave primary depression on the distal facet, lacking in G. tex­ articular facets, rather than one convex and the anum. The sesamoid facet is more well defined. It 148 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 67.—Right metatarsal V oi Glyptotherium texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) is convex and extends posteriorly as a prominent AM 95737 (G. texanum). In USNM 10536 there is tubercle. The posterolateral tubercle is extremely a single sesamoid facet on metatarsal V in the well developed, extending far beyond the margins appropriate position, providing evidence of one of the articular facets. The surface of the shaft is sesamoid, rather than two or none. In all three G. rugose and pitted. arizonae specimens the lateral posterior tubercle of SESAMOID BONE, METATARSAL V (Table 54).— phalanx I is greatly overdeveloped, and the me­ The single plantar sesamoid associated with dial tubercle virtually lacking, as in G. texanum. metatarsal V in G. texanum is a small, irregular Thus a sesamoid bone was likely situated in the element. In F:AM 95737, the one associated with space between the metatarsal and phalanx II, left metatarsal V is rather more flattened and occupying the space left vacant by the reduction somewhat larger than the one for the right meta­ of the medial posterior tubercle of phalanx I. tarsal. The latter is a somewhat compressed sphe­ DIGIT I, PES.—The first digit of the rear foot roid, devoid of articular facets, thus accounting includes two phalanges and two plantar sesamoid for the indistinct sesamoid articular region of the bones, one at each of the two joints. The first metatarsal. In contrast, the left sesamoid, meta­ phalanx is smaller than either the first metatarsal tarsal V, in this specimen, possesses a flattened or the terminal (ungual) phalanx. The proximal articular facet and contacts the more well-defined sesamoid, situated at the metatarsal-phalanx articular region of the corresponding metatarsal. joint, as described above, is more than two times It is situated at the posteroproximal extremity of longer than the distal sesamoid, located at the the first phalanx and contacts the phalanx and terminal joint. This digit projects downward and the large sesamoid of phalanx II. medially from the center of the tarsus. This ori­ The sesamoid bone for metatarsal V is not entation is effected primarily by the medial ro­ known for either the Curtis Ranch or Seymour tation of the distal facet of the internal cuneiform. representatives of G. arizonae, in part explaining The succeeding articular facets are oriented the disparity in sesamoid counts between Mel­ obliquely with respect to the transverse axis of ton's (1964) determination and that proposed the digit, so that in anterior view the articular here. Although, of course, the existence of a ses­ surfaces are directed laterally downward. Hence, amoid bone for this metatarsal is questionable the transverse motion in the joints of this digit is until one is found, in all three otherwise complete functionally as important as the fore-and-aft mo­ specimens, USNM 10536, UMMP 38761, tion. UMMP 46231, including metatarsal V and the Digit I exhibits modifications commensurate complete digit for each, there is an appropriate with its position as the medial member of a series space for a sesamoid comparable to that for F: of the five symmetrically disposed digits arranged NUMBER 40 149 in a semicircle. Progressive modifications for this border projects beyond the lateral extent of the position, beginning with the internal cuneiform, proximal facet. The posterior border of the distal culminate in the terminal phalanx, which is so facet is upturned to form a large articular surface rotated that its flattened outer (dorsal) surface is for the distal sesamoid. medially directed with respect to the body of the The only notable distinctions of this element in animal. Similarly, the ontogenetically lateral bor­ G. arizonae are the transverse, rather than flat, der of the compressed ungual phalanx occupies concavity of the proximal facet and the more anatomically the anteriormost position. The fol­ proximal orientation of the sesamoid portions of lowing descriptions for this digit are based on the this same facet. The posterior tubercles are more ontogenetic reference, i.e., dorsal-anterior, plan­ massively developed, and the region separating tar-posterior, keeping in mind that the dorsal them is more deeply excavated. These differences surface is medially directed (outward with respect are attributable to ontogenetic variation with to the foot) and the plantar surface is laterally increasing size and are not taxonomically useful. directed (inward with respect to the foot). In UMMP 46231 there is considerable degen­ It is impossible to determine with confidence eration of the bone in both phalanges of this digit. whether an isolated rear ungual phalanx belongs In phalanx I this obvious pathologic abnormality to the first digit on one side (for example digit I, includes a tremendous enlargement of the poste­ right), or to the fifth digit of the other side (digit rior foramen. V, left). The reason for this difficulty is that the PHALANX II, DIGIT I, PES (Table 55).—The outer and inner ungual phalanges of each rear terminal phalanx of the first digit, pes, in G. foot are virtual mirror images, a consequence of texanum (Figure 68), is a triangular, dorsoventrally the marked bilateral symmetry of the rear foot. compressed element. An ungual sheath enveloped Digit I is known only for G. texanum (Figure 68) the entire bone except for the small subungual and G. arizonae (Figure 75). base. The proximal articular facet is transversely PHALANX I, DIGIT I, PES (Table 55).—The first elongate. It faces laterally and posteriorly, in phalanx of digit I, pes, of G. texanum, is a proxi­ anatomical position assuming a nearly vertical modistally compressed element, bearing proximal orientation. The weakly concave facet is dorsally and distal articular facets, each with posteriorly and laterally elevated, reaching an apex at the directed extensions for contact with plantar digi­ posterolateral angle, from which occurs the great­ tal sesamoids. Both articular facets present the est proximodistal diameter. The flattened dorsal shape of a half ellipsoid with rounded anterior surface of the ungual phalanx is rotated medially apices, the proximal more elongated than the with respect to the long axis of the digit, and the distal. The proximal facet is concave in the an­ lateral angle represents the leading edge of the teroposterior direction, transversely flat ante­ bone in fore-and-aft motion. Loose articulation riorly, and weakly convex near its posterior bor­ with the first phalanx occurred over a broad area, der. A smaller lateral and a larger medial exten­ indicating considerable mobility at this joint. sion of the proximal facet provide posteroproxi­ The subungual base in the juvenile specimen mally directed articular surfaces for the single F:AM 95737 occupies the proximal half of the digital sesamoid at the phalanx-metatarsal joint. posterior surface of the claw process. Its rounded A prominent posteromedial tubercle is situated outline continues medially and laterally, merging beneath the proximal facet, and a smaller lateral gradually with the proximomedial and proximo- tubercle is situated on the posterolateral angle. lateral angles. Two subungual foramina, a large The distal articular facet is anteroposteriorly lateral one and a smaller medial one, penetrate convex and transversely concave. It is laterally deeply into the plantar surface. The claw process convergent in an acute angle with the proximal is weakly convex on its dorsal surface in both the facet, and it is laterally offset so that its lateral transverse and proximodistal directions and more 150 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 68.—Digit I, right pes, with metatarsal I, of Glyptotherium texanum (F':AM 95737): a, medial, or plantar face of digit with respect to axial plane of foot; b, lateral, or palmar face of digit with respect to axial plane of foot; c, posterior, or trailing edge of digit wih respect to axial plane of foot; d, anterior, or leading edge of digit with respect to axial plane of foot. (Bar = 20 mm.) strongly convex on its posterior surface. The tri­ extremely rugose proximal extremity. The subun­ angular claw process terminates in an indistinct gual hood was probably similarly developed in apex. the adult condition for G. texanum. Ungual phalanx, digit I, pes, of G. arizonae As described in the general discussion of digit (Figure 75), is greatly expanded, especially in the I, pes, this bone bears an exact reverse identity transverse direction, in comparison with that of (mirror image identity), with the terminal pha­ G. texanum. The articular facet occupies relatively lanx of digit V on the same side of the body. less surface area. Its surface is laterally flattened Hence, it is identical to the ungual phalanx of and medially upturned in a small concavity. The digit V of the opposite foot. Therefore, caution subungual hood is proximally expanded and must be employed before determining that a heavily rugose, completely surrounding the prox­ given isolated rear ungual phalanx belongs to imal facet. Accordingly, the subungual foramina either the outer of one side or the inner digit of are situated well within the region of the subun­ the other side. gual hood on the proximal surface, rather than DISTAL SESAMOID BONE, DIGIT I, PES (Table on the border of the hood with the claw process. 64).—The distal sesamoid in G. texanum, located The subungual hood extends as a well-defined at the first phalanx-terminal phalanx joint, is a flange over the claw process in the adult condi­ modified sphere with a flattened articular surface tion, clearly demarcating the proximal limits of for contact with both phalanges, primarily the the ungual sheath. It is broadly arched on the first. It is also flattened on its lower surface, plantar surface of the bone, continuing upward forming with the articular facet a right angle. It to terminate on the medial and lateral angles. is rugose and pitted, presenting a transverse di­ The subungual hood did not extend onto the ameter nearly equal to that of the proximal ses­ dorsal surface. The ungual sheath apparently amoid and a long-axis diameter only approxi­ covered the entire dorsal surface except for the mately half that of the upper sesamoid. NUMBER 40 151

This element is embedded in the plaster mount border, a rounded anterolateral apex, and a for USNM 10536 (G. arizonae). In both UMMP nearly straight lateral border. The two posterior 38761 and UMMP 46231, the distal sesamoids tubercles are relatively small, and the intervening are identical to that of G. texanum in all respects groove is narrow. The medial tubercle is rather except for their larger size. more blunt and expanded than the lateral one. DIGIT II, PES.—As the inner of the three pri­ The proximal articular facet is concave in the mary weight-bearing digits of the rear foot, digit anteroposterior direction and fiat in the trans­ II includes the full complement of three phalan­ verse plane. The dorsal border is elevated as the ges, and along with the metatarsal, three plantar anterior extension of the concave facet, producing digital sesamoids, two proximal ones articulating in this position the greatest proximodistal diam­ with the metatarsal, described above, and a distal eter. The proximal facet is indistinctly separated one at the terminal joint. The articulations are into a narrower lateral portion and a broader generally flat and broad, with close contact and medial portion. The nearly flat distal articular little mobility, except at the terminal joint, which facet is weakly concave in all directions, thus has considerable fore-and-aft and transverse mo­ producing at its center the minimum proximodis­ tion. The proximal and middle phalanges are tal diameter. From this central position, a shallow compressed and discoid, the proximal one some­ and poorly defined nonarticular sulcus extends what larger than the middle one. The latter rearward to the posterior border. Thus the distal phalanx is wedge-shaped, its articular facets an­ facet is posteriorly divided into medial and lateral teriorly convergent. The medially and distally winged portions. A prominent transversely ex­ directed outer (dorsal) surface of the ungual pha­ panded tubercle marks the proximal half of the lanx is obliquely oriented at approximately 45° anterior (dorsal) surface of the compressed shaft. with respect to the sagittal and transverse planes. Phalanx I, digit II, pes, of G. arizonae (Figure Its proximal and distal length approximates the 75), differs in the relative proportions, being es­ combined lengths of the two preceding phalanges. pecially elongate in the transverse dimension. It All three phalanges are rounded medially and is so greatly expanded in the transverse dimen­ anteriorly and flattened laterally, fitting snugly sion, relative to the anteroposterior and proxi­ against the opposing surfaces of the middle digit. modistal dimensions, that it is transversely elon­ The sesamoids are stout. gate, rather than anteroposteriorly, as in G. tex­ Like the relationship between digit I and digit anum. This marked expansion may be regarded as V, the second digit bears a close reverse symmet­ the allometric product of increased size, for this rical identity with the fourth. The resemblance is bone is otherwise very similar to that of G. texanum especially marked for the terminal phalanges, except for the elaboration of features related to which are indistinguishable in isolation as being ontogenetic change: the anterior tubercle and the from digit II on one side or from digit IV from two posterior tubercles are more pronounced; the the opposite limb. This situation is analogous to articular facets are more well defined; and the that for the outer digits as discussed above. shaft is relatively more rugose and pitted in the Digit II, pes, is known only for G. texanum larger species. (Figure 69) and G. arizonae (Figure 75). PHALANX II, DIGIT II, PES (Table 56).—The PHALANX I, DIGIT II, PES (Table 56).—The first second phalanx of digit II, pes, in G. texanum phalanx of digit II, pes, of G. texanum (Figure 69), (Figure 69), like the first phalanx, is a modified is a modified and greatly compressed planocon­ plano-concave discoid, but the plantar articular cave discoid, with subequal transverse and an­ facet is the proximal one, and the concave facet teroposterior (dorsoplantar) dimensions. In prox­ is the distal one. It is smaller than the first pha­ imal and distal aspects the outline is anteropos­ lanx, and in articulation it represents a wedge- teriorly elongate, with a broadly curved medial shaped element separating the first phalanx from 152 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 69.—Digit II, right pes, oi Glyptotherium lexanum (F:AM 95737): a, anterior; b, lateral; c, slightly oblique posterior. (Bar = 20 mm.) the third. Its transverse diameter is slightly The second phalanx, digit II, pes, in G. arizonae greater than the anteroposterior diameter, and (Figure 75), although much larger, is proportion­ the proximodistal diameter is approximately half ally equivalent to that of G. texanum. It differs in of both. The flattened proximal facet presents details of the articular facets and in having medial two poorly defined portions: a smaller medial and lateral tubercles on the shaft. The proximal side, facing slightly away from center in the me­ articular facet is uniformly flat and not divided dial direction, and a larger lateral portion. The into two portions. There is a central constriction proximal facet extends rearward to the medial on the anterior face of the shaft that modifies the and lateral angles over the subdued posterior proximal facet into a wing-shaped surface, with tubercles. In proximal aspect the medial border concave anterior and posterior borders. The distal is broadly curved, and the anterior border is articular facet is divided into lateral and medial nearly straight, as is the lateral border. The distal flattened portions, each facing inward, by a lon­ articular facet is directed somewhat anteriorly gitudinal nonarticular furrow, corresponding to (dorsally) with respect to the proximal facet, im­ the central concavity of this facet in G. texanum. parting the wedge-like construction of the pha­ This furrow continues rearward, meeting a simi­ lanx. It is transversely concave and weakly convex lar groove on the posterior surface of the shaft, in the anteroposterior direction. The anterior bor­ which divides the sesamoid facet into medial and der nearly meets the anterior extremity of the lateral portions. The posterior tubercles are cor­ proximal facet. There is a gradual increase in respondingly more pronounced. proximodistal thickness rearward. The distal PHALANX III, DIGIT II, PES (Table 57).—The facet is confluent with the concave articular facet terminal phalanx of the second digit, pes, bears for the distal digital sesamoid. This latter facet a reverse symmetrical identity with the terminal occupies all of the posterior surface of the shaft phalanx of digit IV. Hence, it is impossible to except for a narrow interarticular region separat­ distinguish between terminal phalanx, digit II, ing the sesamoid facet from the posterior border and terminal phalanx, digit IV, of the opposite of the proximal facet. limb, as explained previously. NUMBER 40 153

This bone in G. texanum (Figure 70) is a trian­ tions of the outline of the facet, imparting an gular, wedge-shaped element, expanded in the overall bilobate shape to the articular surface. proximodistal and transverse directions and com­ This construction corresponds to the similar mod­ pressed in the anteroposterior (dorsoplantar) di­ ifications of the second phalanx. rection. It was entirely encased in an ungual The subungual foramina are situated well sheath except for the region of the subungual within the margins of the subungual base, and base on the plantar surface. The proximal artic­ both border the proximal facet. They are approx­ ular facet forms an acute angle (approximately imately equal in size. The free end is more elon­ 45°) with the dorsal surface, producing a trian­ gate in the proximodistal direction and suggests gular outline in lateral and medial aspects. The the shape of a trapezoid on edge (base of the proximal facet is ovoid, with a straight posterior trapezoid is the lateral border of the bone). The border. It is slightly convex in the transverse lateral, proximal, and medial borders are straight, direction, and there is an indistinct division into rather than rounded, and the roughened terminus medial and lateral portions. The facet extends to is weakly arched. The distal extremity of the the anteromedial angle, but its remaining borders phalanx occurs at the distolateral angle. The are flanked by nonarticular regions. The antero­ expanded posterior surface of the free end is lateral tubercle extends in the plane of the facet weakly concave in the transverse plane, and the as a prominent process. In anterior view, this lateral portion of the posterior surface is distinctly tubercle forms the expanded lateral portion of flattened. the proximal extremity above the epiphysis. Although these distinctions are in part attrib­ The subungual base presents the shape of a utable to ontogenetic changes, the differences half ellipsoid, with the subungual foramina oc­ between these two species seem to be consistent. cupying the medial and lateral borders. These DISTAL SESAMOID BONE, DIGIT II, PES (Table foramina are slightly elliptical and symmetrically 64).—The transversely elongate distal sesamoid disposed on the plantar surface of the bone. The bone in G. texanum articulates exclusively with the lateral foramen is slightly larger than the medial, second phalanx at the terminal joint. Its articular and both nearly contact the border of the proxi­ facet conforms exactly with the sesamoid facet of mal facet. The free end of the phalanx is trian­ phalanx II. Its surface is rectangular, with gular, with equidimensional, rounded sides. The rounded corners, and its transverse dimension is anterior surface is broadly and uniformly convex, approximately twice that of the proximodistal the posterior surface irregularly convex and some­ diameter. From a crest offset slightly medially what flattened near its lateral margin. The ter­ from center, the facet is laterally concave and minal apex is situated directly in line with the medially flattened. The former portion is oriented center of the proximal facet. The surface of the somewhat laterally, the latter somewhat medially. free end is distinctly smoother and less rugose The proximodistal diameter of this sesamoid in than the subungual base and the nonarticular articular position is approximately the same as regions of the proximal extremity. that of the phalanx. The plantar surface is weakly The terminal phalanx, digit II, pes, of G. ari­ divided into two proximodistally directed tendi­ zonae (Figure 75), differs from that of G. texanum nal grooves, and it extends distally to form a in its relatively greater transverse expansion and small median distal tubercle. in details of the articular facet and the free end. This bone is similar in G. arizonae. Its articular The proximal facet is distinctly divided into two facet is divided into two portions by a deep portions, a medial concave one, and a lateral furrow, corresponding to the division of the sesa­ flattened and somewhat elevated portion. The moid facet of phalanx II. This sesamoid is more division of these two portions of the facet is heavily rugose and more deeply excavated than accentuated by anterior and posterior constric­ in G. texanum. 154 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

DIGIT III, PES.—The middle digit is the largest try except for the rather greater development on of the hind foot, and each of the three phalanges the anterolateral surface of the anterolateral tu­ are respectively the largest of the pes. Digit III is bercle. The posterior tubercles are blunt and bilaterally symmetrical, with a broadly rounded rounded and set close together; they are separated anterior border and flattened lateral and medial by a relatively small intertubercular groove. The borders. As the largest weight-bearing digit it proximal articular facet is transversely flat, an­ articulates with the metatarsus as the central, teroposteriorly concave, reaching maximum ele­ anteriorly directed, element of the foot. Phalanx vation at the anterior margin. In proximal aspect I is discoid, and phalanx II is wedge-shaped. The this facet is ovoid, with gradual narrowing rear­ ungual phalanx is a stout, symmetrical terminus, ward from the expanded anterior border. The buttressed by a large plantar sesamoid on the facet extends onto the proximal surface of the posterior surface of the phalanx II. The total posterior tubercles. length of digit III is only slightly greater than The distal articular facet is more nearly circular that of metatarsal III, with which it articulates. and does not include the lower surface of the In anterior aspect the articular facets are elevated posterior tubercles. Its concavity is less pro­ slightly toward the lateral side, with respect to nounced. An indistinct nonarticular region in the the sagittal plane, thereby permitting determi­ central position of this uniformly concave surface nation of right or left for an isolated phalanx of affords identification as the distal facet. The ar­ this digit. ticular facets are posteriorly convergent, but the Complete middle digits, including the sesa­ angle so formed is very acute. Besides the antero­ moid, are known for G. texanum (Figure 70) and lateral and the two posterior tubercles, there is an G. arizonae (Figure 75). The first and ungual pha­ elongate tubercle on the lateral surface of the langes are known for G. floridanum. shaft. The nonarticular shaft is otherwise mildly PHALANX I, DIGIT III, PES (Table 58).—The rugose and pitted by nutritive foramina. first phalanx of the middle digit in G. texanum The only important distinction in phalanx I of (Figure 70) is a doubly concave discoid, somewhat G. arizonae (Figure 75) and G. texanum is the better elongate in the anteroposterior direction. This definition of the central nonarticular region of element displays almost perfect bilateral symme­ the distal facet. This region is somewhat de-

FIGURE 70.—Digit III, right pes, oi Glyptothenum texanum (F':AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) NUMBER 40 155 pressed and extends rearward to the posterior on the medial and lateral nonarticular surfaces of border of the bone in both species. The distal the shaft, which is otherwise mildly rugose and facet in G. floridanum is distinctly more concave pitted with nutritive foramina. than in either of the other species. The transverse The second phalanx of digit III, pes, of G. dimension is disproportionately larger in G. an­ arizonae, is similar in most respects. The differences zonae, in accordance with expected ontogenetic are primarily in degree of elaboration. Both the changes due to the larger size in this species. The proximal and distal articular facets are indis­ phalanx in G. floridanum, intermediate in size be­ tinctly divided into medial and lateral sides by a tween G. arizonae and G. texanum, is also interme­ shallow nonarticular depression. The medial and diate in the transverse expansion. lateral sides of the distal facet are more nearly PHALANX II, DIGIT III, PES (Table 58).—The flat, rather than concave, and the sesamoid facets second phalanx of the middle digit in G. texanum are somewhat more flattened. The latter are com­ is a wedge-shaped element, triangular in lateral pletely separated by the tendinal groove. The and medial aspects. The proximal articular facet medial portion of the proximal surface of USNM is flat, with only a subtle indication of convexity, 10536 exhibits a degree of degeneration and ex- thus conforming closely with the concavity of the ostosis, presumably due to the degenerative ar­ opposing facet of phalanx I. The proximal facet thritic condition as noted for some of the other is nearly rectangular in shape, with rounded an­ phalanges. terior corners and rounded posterior extensions at PHALANX III, DIGIT III, PES (Table 59).—The the posterior angles. The intertubercular groove terminal phalanx of digit III in G. texanum is the (tendinal groove) is widely expanded and rela­ only bilaterally symmetrical ungual phalanx of tively shallow. the rear foot. The only apparent departure from The distal articular facet is transversely con­ this symmetry occurs with the subungual foram­ cave and anteroposteriorly convex. The phalanx ina, of which the medial is rather larger than the exhibits equal medial and lateral proximodistal lateral, thus permitting identification of right or diameters at the posteromarginal angles of the left for an isolated specimen. The overall shape facet. The minimum diameter occurs in the cen­ of this bone is a broad oval in anterior aspect, tral position of the distal facet. The distal artic­ while the lateral and medial aspects present a ular surface is directed anteriorly at approxi­ modified triangular outline, with two concave mately 30°; thus its anterior convergence with short sides formed by the articular facet and the the proximal facet imparts the wedge-shaped distal end of the posterior surface, respectively, character of this element. The distal facet nearly and a longer convex side formed by the dorsal contacts the proximal one on its anterior border, surface. The articular facet is transversely convex with only a very narrow interarticular region and weakly concave in the anteroposterior direc­ separating the two facets on the anterior border. tion. It is oval and somewhat longer transversely. The two-parted sesamoid facet is transversely The facet is bounded on all sides by a nonar­ concave, the marginal expanded portions di­ ticular border, especially on the anterior margin, rected posteriorly and inward. The two portions which is extended in the same plane as the facet are continuous across a narrow constricted con­ as a broad, flattened anterior tubercle. The prox­ nection in line with the central concavity of the imal third of the dorsal surface is a flattened, distal facet. The sesamoid facet meets the distal crescent-shaped region, which, in the juvenile (as facet in a rounded angle. It is separated from the indicated in F:AM 95737), remained unsheathed. proximal facet by a narrow interarticular region. The dorsal surface of the free end is elevated The posterior tubercles of the proximal phalanx above this proximal crescent, and the two por­ are replaced by the expanded portions of the tions are clearly demarcated. sesamoid facet. Indistinct tubercles are situated The subungual base is transversely elongate 156 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY and poorly separated from the ungual process, rection. It is flattened on the margins, conforming from which it stands as an elevated process. The with the corresponding surfaces of the adjacent symmetrically disposed subungual foramina, of ungual phalanges. The distal terminus is straight which the medial is larger than the lateral, are rather than rounded. The anterior tubercle is situated on the border of the subungual base and relatively less pronounced. the ungual process. They do not contact the In UMMP 46231, this phalanx is rendered articular facet. almost fragmentary by the presumed arthritic The uniformly convex dorsal surface of the degeneration in this individual's foot. The claw ungual process is nearly circular in outline. The process is greatly reduced and certainly patho­ plantar surface is transversely convex to a some­ logic. what greater degree, and it is proximodistally Although the ungual phalanx, digit III, of G. concave for the elevation of the subungual base. floridanum (Figure 73c), resembles most closely the The articular face is situated at approximately ungual phalanx of G. arizonae, it differs in several 45° with respect to the dorsal border. In articu­ important details. Rather than presenting in lation this orientation produces an anterior sur­ overall outline a convex shape, the two portions face for the ungual phalanx, which is directed as of the articular facet are weakly concave, and much downward as forward, with the articular their posterior border is nearly straight. The su­ facet maintaining the standard orientation and bungual foramina are relatively smaller than in the plantar surface of the free end remaining in either of the other species, and there are three the plane of the posterior (plantar) surfaces of the large foramina situated in a row along the pos­ more proximal phalanges. terior margin of the articular facet. These are The terminal phalanx of G. anzonae differs pri­ lacking in G. texanum and G. arizonae, both of marily in its larger size, especially in the trans­ which possess smaller, irregularly spaced foram­ verse dimensions, and by the exaggeration of the ina in this position. The posterior concavity of same features possessed by the smaller species. the ungual process is more deeply excavated than The proximal nonarticular portion of the dorsal in G. arizonae, and the flattened margins are less surface is fused with the free end, the junction well developed. being indicated only by an obscure boundary DISTAL SESAMOID BONE, DIGIT III, PES (Table reflected by the nature of the pitted surfaces. The 64).—The distal sesamoid of the middle digit, dorsal surface of the ungual process thus includes pes, in G. texanum, articulates exclusively with all of the anterior surface and was entirely en­ phalanx II on its plantar surface at the terminal cased in the ungual sheath. In anterior (dorsal) joint. The sesamoid bone is transversely elongate, aspect the shape is more rectangular, with a approximately as long in this dimension as the somewhat greater expansion in the transverse phalanges. It possesses an elongate articular facet, plane. The articular facet is divided into medial occupying the full length of the bone, and a and lateral portions by a median ridge. These are generally rounded, fusiform nonarticular surface, anteroposteriorly flattened, rather than concave, tapering gradually toward the rounded margins. and transversely each faces outward to the same The articular facet conforms with the opposing degree as the marginal portions of the facet of G. facet of the phalanx. The facet is transversely texanum. The subungual base is relatively more elongate and faces anteriorly and proximally. The greatly expanded, incorporating within its lower margin of the facet is straight, the upper is bounds the subungual foramina. The subungual centrally constricted to produce the bilobate con­ hood is well developed and extends on the mar­ struction. The "wings" of the facet are directed gins upward to terminate on the proximomar- somewhat outward with respect to the sagittal ginal angles. The plantar surface of the ungual plane, and they are weakly concave in the trans­ process is centrally concave in the transverse di­ verse direction. Projecting proximally and poste- NUMBER 40 157 riorly is a blunt tubercle on the free margin of PHALANX I, DIGIT IV, PES (Table 60).—The this sesamoid. The plantar surface is rounded and proximal phalanx of the fourth digit, pes, in G. smooth, and the lower surface possesses several texanum, is easily distinguished from phalanx I, nutritive foramina set within a shallow, transverse digit II, by its nearly uniform proximodistal thick­ excavation. ness; its maximum proximodistal diameter is The distal sesamoid bone of digit III in G. nearly equal to the minimum diameter of pha­ arizonae is similar to that of G. texanum, except that lanx I, digit II. The proximal articular facet is the articular facet is separated into two distinct weakly concave and extends onto the upper sur­ lobes by an intervening groove. The two portions faces of the posterior tubercles. The margins of of the facet are relatively more extensive in the the facets parallel the surface of the shaft. The proximodistal direction. medial margin is straight, and the anterior and DIGIT IV, PES.—The fourth digit of the rear lateral margins form a broad continuous arc. The foot is the lateralmost of the three primary distal articular facet is nearly flat, and it possesses weight-bearing toes. Because phalanx I is rela­ at its center a shallow excavation, which opens tively short, digit IV is in total length somewhat posteriorly as a nonarticular surface, separating shorter than digit II, and both are considerably posteriorly the medial and lateral portions of the shorter than digit III. The fourth digit, along distal facet. The distal articular surface coalesces with its metatarsal, includes the full complement gradually with the nonarticular lower surface of of metatarsal and three phalanges plus one distal the posterior tubercles. The lateral tubercle is and two proximal sesamoid bones. The digit is blunt and broad; the medial tubercle is narrow medially flattened and broadly arched on the and more pointed. The intertubercular groove is continuous anterolateral outer surface. The prox­ constricted and narrow. imal and middle phalanges are extremely com­ Phalanx I, digit IV, pes, in G. arizonae, is similar pressed, and the ungual phalanx is stout and in most respects to that of G. texanum. The con­ pointed on its distomedial apex. cavity of the proximal facet is rather more pro­ Complete fourth digits are known for G. texanum nounced, especially near the medial border, and (Figure 71) and G. arizonae (Figure 75), and the the nonarticular central depression of the distal ungual phalanx is known for G. floridanum (Figure facet is more deeply excavated. The posterior 73d). tubercles are larger, and the intertubercular

FICURE 71.—Digit IV, right pes, oi Glyptotherium texanum (F:AM 95737): a, anterior; b, lateral; c, posterior; d, medial. (Bar = 20 mm.) 158 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY groove is more widely expanded. The body is description for the terminal phalanx, digit IV, heavily rugose and deeply pitted, and the anter­ pes, of G. texanum, is exactly the same as that for omedial tubercle is well developed. In overall the ungual phalanx of digit II, pes. By merely comparison, phalanx I of G. arizonae exhibits a interchanging "medial" and "lateral," that de­ proportionally greater development in the trans­ scription fully applies in every detail to ungual verse dimensions, probably in simple accordance phalanx IV. There are no differences in size, with the larger size in this species. proportions, or detail. PHALANX II, DIGIT IV, PES (Table 60).—The As for G. texanum, the terminal phalanx of the second phalanx of digit IV, pes, in G. texanum, is fourth digit, pes, of G. arizonae, bears a reverse somewhat smaller than phalanx I. It can be dis­ identity with phalanx III, digit II. It possesses tinguished from phalanx II, digit II, by its flat­ only two distinctions: its transverse dimensions tened proximal facet, the lesser development of are smaller, and the anterior margin of the artic­ the sesamoid facet, the lesser degree of concavity ular facet is smoothly rounded rather than cen­ of the distal facet, and its more nearly symmet­ trally constricted. Otherwise, these two elements rical shape. It is a wedge-shaped element, the are identical, and by merely interchanging me­ proximal and distal articular facets anteriorly dial and lateral, the description for this element convergent in an acute angle. The proximal ar­ is the same as for its symmetrical equivalent. The ticular facet is flat and occupies the entire proxi­ comparisons between the two species are also the mal surface, including the upper surface of the same as for digit II. (See "Phalanx III, Digit II, widely expanded posterior tubercles. The inter­ Pes.") tubercular groove is broad and widely expanded In UMMP 46231, the ungual phalanx exhibits to the rear. The distal articular facet is weakly considerable pathologic erosion of the terminal concave in the transverse direction and slightly border of the claw process. There is also erosion convex anteroposteriorly. It is directed somewhat of the proximolateral tubercle, resulting in a pro­ anteriorly with respect to the proximal facet. The nounced sulcus between the anterolateral angle proximal and distal facets nearly come into con­ and the anterolateral corner of the articular facet. tact on the anterior border of the bone. The A similar condition, but less pronounced, is evi­ marginal and anterior borders of the phalanx are dent in UMMP 38761. rounded, and the posterior tubercles are blunt The terminal phalanx of digit IV in G. flori­ and rounded. The sesamoid articular facet is danum bears a close similarity in size and propor­ centrally constricted, producing medial and lat­ tions to that of G. anzonae. Because the ungual eral, downward- and inward-directed flattened phalanx for digit II is unknown for G. floridanum, portions. a comparison between these two digits is not Phalanx II in G. arizonae is similar in most possible. There are only two minor distinctions respects to that of G. texanum. Like the other characterizing this element of G. floridanum. The phalanges, it is proportionally wider in the trans­ medial subungual foramen is smaller than the verse dimensions, and the angle formed by the lateral one, rather than being equal in size. The convergent proximal and distal articular facets is articular facet is distinctly more flattened, with less acute. The proximal facet is divided into only a slight transverse convexity, and its shape medial and lateral halves by an interarticular is almost perfectly ellipsoidal. depression. The distal facet is similarly divided DISTAL SESAMOID BONE, DIGIT IV, PES (Table and the sides are flattened and inward-directed. 64).—The distal sesamoid bone, digit IV, pes, of The sesamoid facet is two-parted also, with the both G. texanum and G. arizonae, articulates exclu­ central constriction separating completely the sively with the middle phalanx at the terminal two sides. joint. It closely resembles the distal sesamoid of PHALANX III, DIGIT IV, PES (Table 61).—The digit II, although it is somewhat smaller in all NUMBER 40 159 dimensions. The articular facet is two-parted and phalanx, digit V, pes, of G. texanum, is the most elongate transversely. The upper margin is con­ extremely compressed bone of the rear foot. The stricted, and the lower is straight. The bone is proximal facet is anteriorly flat and posteriorly fusiform in shape. There are several large foram­ convex, curving downward to contact the distal ina on the lower surface near the region of con­ facet in a common angle at the posterior border. striction of the articular facet. Posterior to these The distal articular facet bears an irregular, an­ is situated a blunt plantar tubercle occupying the teroposteriorly directed groove. The nonarticular posterior position. The only difference between surface of the shaft is reduced, forming a narrow the two species is that the sesamoid of G. anzonae irregular surface on the lateral and medial bor­ is considerably larger and more rugose. ders. The articulation of phalanx I is advanced DIGIT V, PES.—The lateral digit of the rear forward in such a way that its anterior margin foot contains three phalanges and two sesamoid contacts the anterior tubercle of the ungual pha­ bones. It occupies the lateral extremity of the lanx in the fully extended position of the digit, semicircle formed by the digits, in a position fully thus bypassing the middle phalanx, due to its opposite the first digit. The proximal and middle anterior wedge-shaped construction. The first phalanges are wedge-shaped in opposite direc­ phalanx therefore participates as the anterior por­ tions, together in articulation forming a more or tion of the doubly wedged discoid, formed by the less complete discoid. In this fashion, digit V proximal and middle phalanges in articulation. displays a remarkable convergence with digit I, Except for larger size and greater rugosity, this in which there are only two phalanges, the prox­ element is identical in G. arizonae. In UMMP imal and the ungual, rather than three. Digit V 46231 there is a deep foramen on the posterior is medially flattened, and laterally and anteriorly surface, owing to the pathologic condition in this rounded. Its outer surface is convex in the proxi­ specimen. modistal direction. Like the first digit, its mobility PHALANX II, DIGIT V, PES (Table 62).—The and capacity for weight support are limited. second phalanx completes the double-wedge con­ Complete fifth digits are known for G. texanum struction in the fifth digit. In G. texanum, the (Figure 72) and G. arizonae (Figure 75), and the primary facets taper anteriorly, and their anterior ungual phalanx is known for G. floridanum. borders are virtually in contact in a sharp angle. PHALANX I, DIGIT V, PES (Table 62).—The first The facets are faintly divided into two regions by

FIGURE 72.—Digit V, right pes, of Glyptotherium texanum (F:AM 95737): a, anterior or leading edge of digit with respect to axial plane of foot; b, medial or palmar face; c, posterior or trailing edge of digit with respect to axial plane of foot; d, lateral or plantar face. (Bar = 20 mm.) 160 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

C ' i -i

i, r I..'./' •

!Pd FIGURE 73.—Isolated bones of pes oi Glyptothenum floridanum (USNM 6071): a, right metatarsal III, anterior; b, left metatarsal IV, anterior; c, ungual phalanx, right digit I, anterior; d, ungual phalanx, right digit IV, anterior; e, ungual phalanx, right digit III, anterior. (Bar = 20 mm.) a weak central constriction. The two portions of the greater relative thickness at the posterior bor­ the proximal facet are each weakly concave, while der imparting a nearly equidimensional wedge the distal facet is uniformly concave in the trans­ shape in lateral outline. verse plane. The expanded posterior surface is PHALANX III, DIGIT V, PES (Table 63).—The occupied almost entirely by the sesamoid facet, ungual phalanx of digit V, pes, in G. texanum, is which is confluent with the distal facet in a blunt nearly a symmetrical equivalent of the terminal angle. In lateral or medial aspect this wedge- phalanx of the first digit. The only perceptible shaped phalanx suggests a 30°-60°-right triangle, difference is that in the fifth digit the terminal the hypotenuse formed by the proximal facet, the phalanx is somewhat broader transversely. As for 30° angle by the anterior border. The anterior the ungual phalanges of digits II and IV, by border reaches the midpoint only beneath the interchanging "medial" and "lateral" the descrip­ proximal phalanx, thus allowing partial contact tion for terminal phalanx, digit I, applies in every between phalanges I and III. detail to terminal phalanx, digit V Phalanx II, digit V, in G. arizonae, is similar in Similarly, the terminal phalanx of digit V, pes, most respects. The only distinctions other than in G. arizonae, is the symmetrical equivalent of the larger absolute size are the separation of the ungual phalanx, digit I. Unlike G. texanum, the sesamoid facet into lateral and medial halves and transverse dimensions are approximately equal in NUMBER 40 161 these two elements. The proximal facet is some­ II are further evidence of this apparently arthritic what smaller, and the medial subungual foramen pathologic condition. is relatively larger than the lateral in digit V. The ungual phalanx, digit V, pes, in G. flori­ Otherwise, by interchanging "medial" and "lat­ danum, closely resembles that of G. arizonae in size eral" the description for the ungual phalanx of and construction. Rather than convex, however, digit I, pes, applies in every detail to that of digit the articular facet is flat, and the medial extrem­ V. (See "Phalanx II, Digit I, Pes.") ity of the subungual hood is not as extensive as in In UMMP 46231 there is considerable degen­ G. arizonae. Also, the subungual foramina are eration of the free border of the claw process and approximately equal in size. obviously pathologic enlargement of the lateral DISTAL SESAMOID BONE, DIGIT V, PES (Table subungual foramen with concomitant exostosis of 64).—The distal sesamoid bone, situated in the the ungual hood in the region surrounding this plantar space between phalanges I and III and foramen. Similar conditions on phalanges I and articulating with phalanx II, is a proximodistally

FICURE 74.—Articulated right pes of Glyptotherium texanum (F:AM 95737): a, anterior; b, posterior, with digital sesamoid bones. (Bar = 5 cm.) 162 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

a ' •'• " ••' • -1

*** > FIGURE 75.—Articulated right pes of Glyptotherium arizonae (USNM 10536) a. anterior; b. posterior. (Bar = 20 mm.) compressed, transversely elongate element. Its size Figure 47, Table 65), which included one sacro­ and shape are similar to that in digit I. Its elon­ caudal vertebra, 13 free caudal vertebrae, and an gate articular surface is faintly divided into two imperfectly developed 14th terminal vertebra.

weakly concave portions. Of these vertebrae the posterior ten as appears from The distal sesamoid bone in G. arizonae is simi­ measurement and from the deflected transverse processes, lar, except that its articular facet is almost com­ were fitted within the tail sheath, there being thus a vertebra pletely separated into halves. Except for its larger for each ring, while the anterior three articulated with the size and greater development of rugosity on the peculiar sacrocaudal vertebrae, in which the greatly elon­ nonarticular surface, it is identical to that of G. gated transverse processes or ribs extend outward to coosify with the posterior plates of the ischia. The first free caudal texanum. has a transverse diameter of 302 mm., and distinct lateral articulations as facets of the posterior borders of the last CAUDAL VERTEBRAE sacrocaudal and of the ischium; the neural laminae are elevated, the pre- and post-zygapophyses are elevated and Osborn (1903) described the caudal vertebrae vertically placed; the neural spine is low; caudals 2 and 3 of the type specimen of G. texanum (AMNH 10704, were also well within the carapace, with transversely ex- NUMBER 40 163 tended spines; in caudals 4-11 the transverse processes are Osborn (1903) measured the caudal rings in deflected, downwardly and forwardly directed; the neural articulation at 620 mm for the type specimen. In arches, zygapophyses and spines diminish in distinctness. F:AM 95737, for which the caudal armor is iden­ Caudals 12-13 lack all processes. A single chevron, of the narrow type . . was found with the specimen; it measures tical to that of the type specimen, the length of 130 mm. vertically. Six stout chevrons with shallow, obtusely caudal vertebrae nos. 3-13 (the vertebrae that forked inferior processes, anteroposteriorly expanded dis­ received the protection of complete rings) is also tally, are placed beneath caudals 5-11 (Osborn, 1903:493- 620 mm. This measurement must correspond 494). closely to the length of the caudal armor (not Therefore, Osborn identified a total of 15 cau­ presently in articulation, therefore unmeasurable) dal vertebrae, including the sacrocaudal vertebra for F:AM 95737. Hence, by indirect comparison, and the imperfect terminal one. There are eight lengths of the caudal armor and of the last 11 rings of caudal armor and the terminal cone for vertebrae for both the type specimen and the the type specimen. As described below, the Ari­ Arizona representative are identical. zona specimen F:AM 95737 also possesses eight In F:AM 95737 the sacrocaudal vertebra pos­ caudal rings and a terminal cone, but the verte­ sesses expanded extremities of the transverse bral count is different, with one sacrocaudal ver­ processes for cartilaginous (subadult individual) tebra, 11 rather than 13 complete free caudals, contact with the ischiac plates. This contact and an incomplete terminal vertebra, for a total would have ankylosed in the adult condition, of 13 caudals instead of 15 as in the type speci­ incorporating the sacrocaudal vertebra as a strong men. participant in the sacrum. There is no associated In the Arizona specimen (Figure 76, Table 66) chevron with vertebra no. 1, and there is no the last three caudal vertebrae lay within the indication that it received protection of any der­ terminal cone, and all but the first two vertebrae mal armor. were protected by complete caudal rings. The The second caudal vertebra has expanded second vertebra was protected by the incomplete transverse processes for contact with the extrem­ accessory ring, and the first caudal vertebra (the ities of the sacrocaudal vertebra. Its transverse sacrocaudal vertebra) had no armor protection processes are not as stout as in either the sacro­ and lay well within the posterior aperture of the caudal vertebra or in the succeeding vertebrae. carapace. The difference in total counts between The neural spine is tall, and the lateral processes the Arizona and Texas representatives occurs in of the metapophyses are likewise elevated for the "pseudosacrals," of which there are three in contact with the accessory ring of armor, which the type specimen and only one in the Arizona did not extend laterally to contact the transverse specimen. This interpretation also explains why processes. the two specimens have the same number of Vertebrae 3-10 bear downturned processes at caudal rings, for only the last of the three pseu­ the extremities of the transverse processes for dosacrals in the type specimen would have armor articulation with the inner surface of the caudal protection in the incomplete accessory ring, the rings. In the Arizona subadult individual these succeeding eight vertebrae with complete rings, processes are underdeveloped and at the time of and the last three set within the terminal tube. death were united with the dermal armor only by The fact that the caudal rings are identical in cartilaginous contact. The xenarthral processes number and morphology between the Texas and for vertebrae 3-7 are strong and well developed. Arizona specimens is by far the more important Diminishing in all dimensions rearward, the pos­ consideration taxonomically; the disparity in ver­ terior xenarthral processes, although present, are tebral count is attributable to variation, a matter unimportant in vertebrae 6-7, and from vertebra which is discussed at length for modern armadil­ 8 rearward, the posterior region is absent entirely. los in our introductory remarks under "Sys- The anterior xenarthral processes also diminish tematics." in importance rearward, reduced to insignificant 164 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FICURE 76.—Caudal vertebrae of Glyptothenum texanum (F:AM 95737): a, articulated caudal series, dorsal; b, detail of representative vertebra, caudal vertebra 3 (dorsal aspect upper, and anterior aspect lower); c, chevron bones (upper row, 1-5, posterior aspect; middle row, 6-10, left side; lower row, 11, left, left side, and 12, right; all dorsal). (Bars = 50 mm.) tubercles on vertebra 9 and absent on the last two inferior processes, anteroposteriorly expanded dis­ vertebrae (12-13). tally . . . beneath caudals 5-U" (Osborn, 1903: The last three caudal vertebrae (11-13) are 493-494). In the Arizona specimen, F:AM 95737, fused by union of their intercentral discs. Verte­ there is a full set of 11 chevrons (Figure 76), of bra 13 terminates in a blunt point and is therefore which the anterior one is long and narrow and incomplete. These three vertebrae are 120 mm measures 113 mm. They diminish in length rear­ long, corresponding to the internal length of the ward, becoming stouter, and possessing blunt, terminal caudal tube. bifurcate inferior extremities for contact with the Osborn described seven chevron bones for the caudal rings. The first two are long and slender, type specimen. There are "a single chevron of the the next two are intermediate, and the last seven narrow type," measuring 130 mm vertically and are short and stout. The last four are wedge- "six stout chevrons with shallow, obtusely forked shaped and small. There is also a pair of (pre- NUMBER 40 165 sumed) ossified tendons associated with the ter­ 77a), from the type-locality, are identical in num­ minal vertebra and found within the terminal ber and construction to those of the type speci­ tube. The principal proximal articulation of the men. The caudal vertebrae of the Seymour rep­ first chevron was with the lower posterior surface resentative, UMMP 34826, are similar to those of the centrum of caudal vertebra 2, and its distal from the type-locality, but there are 13, rather contact was with the caudal ring that encircled than 12, in the caudal series. Although it was the third caudal vertebra. The succeeding chev­ proposed above that the Arizona representatives rons articulated similarly, so that each chevron perhaps had 13 caudal vertebrae rather than 12, articulated proximally with a vertebra and dis­ there nevertheless appears to be an actual dispar­ tally with the caudal ring of the next posterior ity in the vertebral counts as indicated by the vertebra. The articular facets are two-parted and caudal ring counts from these two localities. In narrow, relative to those in other species. addition, UMMP 34826 appears to have rather The total length of caudal vertebrae F:AM large caudal vertebrae. For example, vertebra 6 95737 in articulation (including the sacrocaudal in this specimen measures 92 mm centrum length vertebra) is approximately 750 mm. Because and 171 mm transverse diameter between angles there are two additional anterior vertebrae in the of transverse processes; corresponding measure­ type specimen AMNH 10704, the length (which ments for vertebra 6 in USNM 10537 are 82 mm as stated above is identical for the last 11 verte­ and 136 mm, respectively, while those of the next brae to that of the Arizona specimen) including anterior vertebra (5: 81 mm and 167 mm, respec­ the sacrocaudal is probably near 900 mm total. tively) more nearly approximate the dimensions Melton's (1964) erroneous statement that the tail of the sixth in the Seymour representative. It of the type specimen of G. texanum measures only must be remembered, however, that UMMP 620 mm was apparently based on the serial length 34826 is a large individual compared with others of the caudal rings of this specimen, as provided in the same population. by Osborn (1903). The probable 900-mm length The difference in vertebral counts appears to in G. texanum (maximum estimate) compares fa­ reside in the number of pseudosacrals, similar to vorably with the lower extreme in lengths ob­ the variation noted for G. texanum. In the Seymour served for G. arizonae. representative there are three pseudosacrals, Gidley (1926) gave an accurate count of 12 whereas in the Curtis Ranch specimens there are caudal vertebrae for paratype USNM 10537 (Fig­ only two. Whether this variation is attributable ure lib) of G. arizonae (Table 67). It appears, to sexual dimorphism is unclear, but that is a however, that there may have been 13 and that possibility. In all three specimens, caudal vertebra half of the 12th and all of the 13th are missing. no. 1 possesses expanded lateral extremities of the This condition is indicated by the construction of transverse processes, with upturned facets for con­ vertebrae 10-12. The 10th vertebra was free, and tact with the ischiac plates. However, the contact was encased by the last complete caudal ring. appears to have been an articular one rather than The last two, 11 and 12, are fused and lay within eventual ankylosis as in G. texanum. Thus the first the terminal tube. Their combined length is 115 caudal vertebra of G. arizonae participates more as mm. Apparently the terminal tube and the last a pseudosacral than as a sacrocaudal vertebra (by caudal ring were fused in this individual, so that comparison with G. texanum), for the transverse vertebrae 10-12 lay within the terminal tube, processes are not immovably united with the although vertebra 10 seems to have been unfused. ischia. The ischiac facets are dorsal expansions, Vertebrae 3-9 were encased by complete caudal facing upward and slightly outward and forward. rings. Vertebra 2 supported the first, or "acces­ The roughened surface indicates a general lack of sory" caudal ring, and 1 lacked dermal armor. mobility at this joint, but there is no indication The caudal vertebrae of AMNH 21808 (Figure of fusion. The transverse processes are also ex- 166 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 77.—Caudal vertebrae of Glyptothenum arizonae: a, AMNH 21808, without chevron bones, left side; b, USNM 10537, with chevron bones in articulation, but partially obscured by transverse processes, left side; both specimens from the Curtis Ranch locality, Arizona. (Bar = 15 cm.) panded rearward in the same horizontal plane, Succeeding chevrons are increasingly more nearly overlapping similar expansions of the next elongate and stout. Except for the first chevron vertebra. they are markedly bifurcate at their distal extrem­ The transverse processes of vertebra 2 in both ities, quite in contrast to the blunt or weakly Arizona specimens and of vertebrae 2 and 3 in bifurcate condition in G. texanum. Beginning with the Seymour specimen are also horizontally ex­ vertebra 3 in the Arizona specimens and vertebra panded for participation in the pelvic construc­ 4 in the Seymour specimens, the transverse tion as pseudosacrals. In all three specimens, a processes bear strongly downturned, forward- narrow chevron arises by principal articulation directed processes for internal contact with their from the posterior extremity of vertebra 2 and respective caudal rings. These processes are articulates distally with the first complete caudal stouter and more massive than in G. texanum. ring (which encases vertebra 3). Therefore, ver­ They diminish in proportions rearward, becom­ tebra 2 supported the incomplete "accessory" ing vestigial on the vertebrae set within the ter­ caudal ring, which partially encased caudal ver­ minal tube. The metapophyses and xenarthral tebra 2. articulations are similar to those of G. texanum. NUMBER 40 167 differing only in size. The remaining features of the caudal vertebrae of G. arizonae are similar to those of G. texanum. The greatest difference in the caudal vertebrae between G. texanum and G. ari­ zonae is the distinctly greater size in the latter species. Melton (1964) stated that the length of the caudal series in UMMP 34826 is 1102 mm. Be­ cause the specimen is now disarticulated, his de­ termination cannot be checked. Assuming that his measurement is correct, the tail of the Sey­ mour specimen is considerably longer than that of the Arizona specimens, which measure approx­ imately 900 mm. The difference in length is attributable to the different vertebral counts and is not taxonomically significant; however, it may be indicative of a trend toward larger size, al­ FICURE 78.—Isolated chevron bone oi Glyptotherium cylindricum though lack of adequate samples of G. arizonae (AMNH 15548): a, anterior; b, right side. (Bar = 20 mm.) precludes certain determination of this matter. A single chevron bone (Figure 78) is the only bly represent positions 5-10, although these as­ element of the caudal series known for G. cylindri­ signments might be wrong by one position in cum. It is large, corresponding in size and shape either direction. If they represent caudal verte­ to the anterior two chevrons in G. anzonae. It is brae 5-10, then the first 10 caudal vertebrae are "of the narrow type," to borrow Osborn's (1903) known for G. floridanum in composite sequence parlance, and its distal extremity is apparently between the Ingleside and Hunt County speci­ not bifurcate, indicating that it is probably chev­ mens. ron 1. Its articular facet is extremely rugose and The anterior six caudal vertebrae from Ingle­ marked with tubercles, indicating a pathological side are similar in all respects to those of G. (arthritic?) condition. Its proximodistal length of anzonae. Vertebrae 1-3 bear greatly expanded 145 mm compares favorably with the lengths of transverse processes, the first joining with the the anterior two chevrons in G. anzonae. It exhibits ischiac plates and 2 and 3 participating as pseu­ no substantial distinction from chevron no. 1 of dosacrals. Apparently the first participated in the USNM 10537. pelvis as a sacrocaudal, for the transverse Only fragmentary caudal vertebrae are known processes are expanded for contact. Vertebrae 3- for G. floridanum (Table 68), all of which are from 6 possess downturned lateral processes for artic­ the Texas population. Lundelius (1972) described ulation with the caudal rings. Thus the caudal caudal vertebrae nos. 1-6 for the Ingleside rep­ armor probably began with vertebra 2, with an resentative, associated with as many caudal rings. incomplete accessory ring, and ring 2 encased Hay (1916) briefly described six caudal vertebrae vertebra 3, which is the first to provide lateral from the Hunt County locality (USNM 6071); support for a complete ring. As Lundelius stated, these are from the basal region of the tail, as Hay the downturned processes posteriorly increase in stated, and represent the last six vertebrae that their anterior orientation. possessed complete caudal rings, exclusive of the Chevron bones for vertebrae 2-6 are identical vertebrae incorporated into the terminal tube. to those of G. arizonae. The chevron for vertebra Because the number of caudal vertebrae is un­ no. 2 distally supported the caudal ring encasing known for G. floridanum, the exact serial positions vertebra 3, further supporting the sequence pro­ of these six vertebrae are uncertain. They proba- posed above. 168 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

The six vertebrae from the Wolfe City speci­ istic of glyptodonts. The armor largely consists of men (USNM 6071), here assigned position 5-10, coalesced, tightly sutured individual scutes. The nearly complete the caudal series for G. floridanum. scutes covering the trunk region of the body are Although measurements for the first two (5 and polygonal, usually six- or four-sided, and they are 6 as here assigned) do not correspond exactly with so united as to inhibit trunk mobility. The struc­ 5 and 6 of the Ingleside specimen, the assignments ture so formed is the carapace, analogous to the nevertheless appear to be reasonable. Down- turtle carapace and more closely resembling it turned lateral processes are present on 6 and 7, than the mobile carapace of the related armadil­ but their extremities are broken. Lateral processes los. In Glyptothenum there are approximately 1600 are broken on the other vertebrae. Metapophyses scutes in the carapace, each firmly united with are preserved only for 6; they are large and the next except toward the margins of the cara­ expanded, as typical for the genus. Metapophyses pace, where the quadrilateral shapes of the scutes and transverse processes are vestigial on vertebra allowed limited mobility. The carapace covers 10. The diameter of the centrum of 10 suggests the entire trunk region except for the ventral that two additional vertebrae completed the cau­ surface of the body, providing nearly impenetra­ dal series and that these fit within the terminal ble protection from predators. In adults the scutes tube (unknown for this species). If the counts are thick, lateral ones measuring up to 60 mm or suggested here are correct, there are probably 12 more in thickness, interior scutes 30 mm or more, caudal vertebrae in G. floridanum. The important and, compared with armadillo scutes, they are matter is not the number of vertebrae, but the relatively porous and light. fact that the last several are free and unfused. By The disposition of the component scutes of the inference, therefore, the terminal caudal armor carapace into rows was established early in the (unknown) was composed of typical caudal rings ontogenetic development of the individual. Max­ for each vertebra except the last two (presum­ imum transverse dimensions of the component ably), which were protected by the terminal tube. scutes were achieved upon attainment of full The caudal vertebrae of G. floridanum therefore scute-to-scute contact, with subsequent growth appear not to differ in any substantial manner possibly limited to increasing the thickness of from those of G. arizonae. Lundelius (1972) consid­ each scute. ered the participation of three vertebrae in the Holmes and Simpson (1931) adequately pelvic complex as an indication of distinction treated the microstructure of the carapacial scutes from the older Seymour representatives. Actually of North American representatives. For the pur­ there are three pseudosacral vertebrae in the Sey­ poses of this study, no additional information has mour G. arizonae and only two in the Curtis Ranch been gathered, and their discussion appears to be representative, a disparity that we attribute to entirely accurate. Carapacial scutes exhibit a rel­ variation. A similar disparity exists for the rep­ atively uniform construction, variable only in resentatives of G. texanum, indicating that the quantitative and geometric details. The under­ number of pseudosacrals is not taxonomically surface of a scute is typically weakly concave, significant and may represent sexual dimorphism. with several large vascular foramina penetrating In summary for G. floridanum, caudal vertebrae, to the interior. Except for ankylosed attachment insofar as they are known, are not distinguishable in the pelvic region, the scutes generally occupied from those of G. arizonae; a similar conclusion is a dermal position, so that the undersurfaces are reached for the caudal armor. generally smooth, or, in some instances, weakly striated as an apparent indication of attachment by connective tissue. CARAPACE In Boreostracon |= Glyptothenum], like most glyptodonts, the The dermal armor ("coat of mail" of other scules are much more porous and lighter relative to their authors) is the single most outstanding character­ size than in the armadillos. The inner table is reduced to a NUMBER 40 169 thin periosteal layer of interwoven fiber groups, between slightly raised and weakly concave. which pass numerous canals. The latter become larger in Surface texture of the scutes is punctate, owing passing outward and communicate with the sinuses of the to numerous vascular channels that lay beneath cancellous central layers. Canals and sinuses are surrounded by lamellar bone. The outer table is fairly thick, but it, also, the scales for communication with the dermis and is not sharply distinguished from the trabecular part, and it epidermis. These pass into the interior of the scute contains numerous vascular canals. The vertical sutural and communicate with the cancellous central surfaces are not so dense as in armadillos, and the scutes layer. articulate by long spicules of bone, with prominent radial Also penetrating the external surface of the haversian systems. Spicules of adjacent scutes interlock some­ what as do the bristles of two brushes pressed together scutes are variable numbers of larger hair follicles (Holmes and Simpson, 1931:417). issuing from within the circular grooves at the boundary of the central figure. Typically, one In life the scutes were covered externally by follicle is set at the intersection of a radial groove, scales, similar to those of armadillos. Glyptodont with the circular groove on the anterior side of scutes are readily distinguished, however, by the the scute. Dorsal and anterior scutes possess as fact that each scute was covered by several scales many as five large follicles; the frequency dimin­ arranged in a generally symmetrical pattern, ishes laterally and rearward so that away from rather than being covered entirely by a single the dorsal region follicles are less numerous or scale as in armadillos. Grooves on the external absent. That these large pits lodged hairs, rather surfaces of the scutes mark the scale patterns; this than blood vessels, was argued convincingly by sculpturing has been extensively relied upon for Holmes and Simpson (1931). taxonomic determinations. For a typical hexago­ Scute thickness increases toward the margins, nal scute of Glyptotherium from the middle region and the peripheral figures predominate. The bor­ of the carapace, there is usually a central scale der scutes are variously modified into large coni­ and several peripheral scales symmetrically ar­ cal projections, effectively protecting the under­ ranged around the center one, producing a char­ surface of the body. acteristic rosette pattern for each scute, resem­ Besides the carapace, numerous ossicles in­ bling that of the South American genus Glyptodon. vested the skin of the legs and undersurface of the In Glyptotherium the rosette pattern is diagnostic. body. This condition is indicated by a large num­ The central figure is usually equal in size or ber of small isolated ossicles associated with the somewhat larger than the peripherals for interior limb elements of carapace F:AM 95737. scutes, with increasing importance of the central The North American species are known from figure in the scutes nearer the margins of the complete or nearly complete carapaces, and the carapace. There is only a single row of peripheral addition of several unreported specimens provides figures, varying from seven to 13 in number, important information regarding variation and surrounding the central figure. The peripheral species characteristics. Carapacial features of figures are usually confined to one scute, although North American species, though distinctive, are they occasionally overlap across sutural contacts. not as different as previous authors have believed. In G. texanum the central figures are larger than Besides the carapace, dermal armor covered the peripherals, and they are convex and slightly the head, in a cephalic shield, and the tail, in a raised above the level of the flattened peripheral series of movable rings of decreasing size. The figures. In G. arizonae and G. cylindricum the central head shield, composed of many so-called figures are relatively smaller, usually not much "casque" scutes, is not sufficiently well repre­ greater than half the scute diameter, but always sented for any North American species to warrant at least slightly larger than the peripherals and detailed description. These scutes protected the generally flat to weakly convex. In G. floridanum dorsal surface of the skull (but they did not afford the central figures are approximately equal in complete protection; see description of the skull); size to the peripherals, and they are usually their position is indicated on the frontal and 170 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY parietal bones by numerous vascular foramina. Apparently the cephalic shield abutted against the cephalic boss of the carapace, protecting the *:_£*:H neck. The tail of Glyptothenum was protected by rings of armor, one ring per vertebra, composed of double or triple rows of fused scutes. Except in G. texanum, for which the rings are smooth, the distal row of all the rings beyond the first two bore large conical bosses. The terminal ring is usually composed of the coalesced rings of the last two vertebrae. Measurements for the carapace are provided in Table 69. CARAPACE OF G. texanum.—Osborn (1903) es­ \^v Tv tablished the genus and species Glyptotherium tex­ anum primarily on the basis of carapace AMNH 10704. Although brief, his description was suffi­ ciently complete to provide a foundation for the genus. The addition of three nearly complete carapaces and the recoveries of carapace frag­ ments in southwestern Texas substantially im­ prove the state of knowledge for the carapace of this species. The carapace of the type specimen AMNH 10704 (Figures 47, 79) is considerably smaller than the known complete carapaces of other spe­ cies; it measures 145 cm in length and an esti­ mated 186 cm maximum half-circumference (marginal scute to marginal scute). The length/ width ratio (actually length/half-circumference ratio) is approximately 0.79. The scutes in this carapace are mostly unfused but all are fully developed, and thus the carapace belongs to a nearly mature adult of maximum size. The un­ fused condition might also indicate a limited mobility in the carapace, even in the adult. This FIGURE 79.—Carapace of type specimen oi Glyptotherium tex­ carapace is complete on the left side, and the anum (AMNH 10704): a, left side anterior to left; b, detail of cephalic and caudal apertures are fully repre­ nuchal notch, dorsal; c, detail of caudal notch, dorsal. (Bar sented, allowing a complete description. = 20 cm.) In lateral aspect the shape of the dorsal arch of the carapace is broadly elliptical, with near sym­ regions. The maximum height occurs at the cen­ metry between the anterior and posterior regions. ter of the carapace, and it is not as highly arched The posterior rows of scutes at the caudal notch as in the other species. are not recurved upward, nor is there a prominent Because this free-mount is somewhat distorted division of the carapace into preiliac and postiliac as restored, comparisons between the caudal and NUMBER 40 171 cephalic apertures must be inferred. As in other margins and internal apices for the sutural con­ glyptodonts, the posterior aperture is much the tacts with the interior scutes. Only the two scutes larger, measuring an estimated 60 ± 20 cm; the at the posterior notch are conical, and these are anterior aperture is only half as wide. The poste­ very low, poorly defined cones. These quadri­ rior aperture is distinctive in the lack of vertical lateral scutes are the largest of the border scutes, outline in lateral aspect: the outline of the border measuring approximately 45 mm in the trans­ scutes continues posteriorly from the posterolat­ verse (side-to-side) diameter and 55 mm in the eral angle upward to the posterior notch. Thus exterior-interior diameter. the posterior extremity of the carapace is the Most of the interior scutes are disposed into posterocentral border scute. The shape of the distinct rows allowed by the hexagonal shape of anterior aperture is similar, tapering upward, the scutes. Rows are distinguishable in three di­ forward, and medially from the anterolateral an­ rections. The primary direction is transverse, par­ gle. allel to the anterior and posterior apertures. The The border scutes of the anterior aperture are other two directions are oblique, offset at 30° in quadrilateral, uniformly simple, and unbossed. either direction from the sagittal plane. Laterally, They are fused along their lateral borders to the the scutes become quadrilateral and the row di­ adjacent border scutes. Together, they form a rections are transverse (parallel to the anterior smooth anterior outline. Sculpturing is indistinct; and posterior apertures) and sagittal (parallel to dermal scales individually covered each scute en­ the lateral margin). Thus, it appears that as much tirely, except, in a few instances, along the interior as the lower half of the side of the carapace suture. The anterior border scutes are small rela­ retained at least limited mobility, permitted by tive to interior scutes. There is no apparent indi­ the quadrilateral scutes. These scutes appear to cation of a cephalic boss at the anterior notch as be joined rather loosely, in the same fashion as in G. cylindricum. The position of the center of the the scutes of the anterolateral region. anterior notch is not distinguishable, for all the Sculpturing of the interior scutes is uniformly scutes along the anterior border are identical. punctate. Near the border row, the round central Border scutes at the anterolateral angle become figures occupy nearly all of the external surface, irregularly conical, but they remain small and and peripheral figures are absent or indistinct. quadrilateral. A few of the border scutes rearward Interiorly the central figures become smaller, but from the anterolateral angle become "pendant," they are always greater than half the diameter of i.e., sutural contacts with adjacent border scutes the scute. Also, the central figures are never are open, producing a scalloped outline for a equaled, or nearly equaled, in size by the periph­ short distance. From there (approximately 10 eral figures as in the other species. scutes from the anterolateral angle) the border The circular and radial grooves are generally scutes increase in size rearward; they are quadri­ shallow, and on some scutes these grooves are lateral and unbossed along the lateral margin. rather poorly defined. There are usually eight Anterior to the poorly defined posterolateral an­ (less frequently six, seven, nine, 10, 11) peripheral gle, the border scutes again become conical, figures in a single row around the central figure. pointing downward and rearward. The scutes are The central figure is everywhere convex along its generally three-sided, with a rounded or pointed outer margin, and about half are excavated in a free margin and a blunt apex formed by the shallow depression, usually to a level slightly sutures for the interior scutes of the first row. lower than the depth of the radial grooves. The Rearward from the posterolateral angle the bor­ other half are flat to slightly convex in the center. der scutes are again unbossed, but they become Similarly, the peripheral figures are generally increasingly larger toward the posterior notch. convex. Maximum side-to-side diameters of the They are generally five-sided, with rounded outer scutes vary between 40-45 mm, in this respect 172 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY somewhat smaller than in the other species. turned profile is not as pronounced as in the other Large, deep hair follicles occur generally in the species. The existence of this character in a cara­ anterior half of the circular groove; they vary in pace that is otherwise identical to the type speci­ number from one to five per scute. These are men emphasizes the intraspecific variability in most frequent in the anterodorsal scutes, where carapacial features. there was a central tract of stiff hairs along the The appearance of this recurved outline in the midline of the body. Although the follicles are Arizona G. texanum might indicate an intermedi­ present posteriorly along the dorsal arch, their ate condition in the trend toward G. arizonae. We density decreases rearward. Laterally, the scutes suggest an ancestral-descendant relationship be­ are mostly devoid of the hair follicles. tween the Blancan G. texanum, of Texas and Ari­ Identical in size to the carapace of the type zona, and the Arizona Curtis Ranch G. arizonae, specimen of G. texanum (see Table 69), carapace in which this outline is more pronounced, among F:AM 59599 (Figure 80) bears little obvious dis­ other changes in carapacial features. tinction and without doubt belongs in the same The remaining characteristics of carapace species. It is unfortunately not associated with F:AM 59599 are otherwise identical to those of any noncarapacial remains. As in carapace AMNH 10704, and it differs from the other spe­ AMNH 10704, the scutes are mostly unfused, but cies in much the same fashion. This carapace is they are fully developed and represent a mature not highly arched, and there is no prominent adult individual. preiliac and postiliac division. With only minor This nearly complete carapace from Arizona is exception, details of the scutes in F:AM 59599 important for the construction of its caudal ap­ are identical to those of AMNH 10704. The erture, exemplifying the posterior upturned cur­ posterior border scutes are all flat, however, with­ vature in lateral silhouette. The lateral outline out the pair of weakly conical scutes at the pos­ does not exhibit the broadly elliptical and nearly terior notch. As in the type specimen, the quad­ symmetrical silhouette of the type specimen. In­ rilateral shape of the scutes becomes prominent stead, beginning approximately four rows inter­ approximately midway between the dorsal arch nal to the caudal notch, the scutes near the and the lateral margin, and there appears to have posterior border are upturned to produce a re­ been considerable mobility in the anterolateral curved outline. The outline does not achieve a region. There is a lower frequency of the deep horizontal attitude, however, so that the up­ depressions in the central figures; only about one-

F'IGURE 80.—Carapace oi Glyptotherium texanum (F:AM 59599), right side, showing weakly recurved outline of posterior dorsal silhouette. (Bar = 20 cm.) NUMBER 40 173

F'ICURE 81.—Carapace of juvenile Glyptotherium texanum (F:AM 95737): a, region of posterolateral angle, left side; b, region of caudal notch, showing weakly conical border scutes of caudal aperture. (Bar = 10 cm.)

fourth of the scutes are so depressed. The largest terolateral angle. Unlike the previous carapace, scutes are barely 50 mm in side-to-side diameter; the scutes of the caudal aperture are all weakly most measure 40-45 mm. The external surfaces conical, with gentle concave outlines leading to of the scutes are not as markedly convex as in the apices of the cones. They are not highly AMNH 10704; most are flat, or weakly convex. elevated, but they are prominent features of the The complete carapace of the subadult individ­ posterior region. Sculpturing is uniformly punc­ ual F:AM 95737 (Figure 81) provides additional tate. Central figures are all much larger than the information on ontogenetic variation in G. tex­ peripherals, and they are only slightly depressed anum. This carapace, recovered not far from the in the middle. It appears that the development of preceding one, is from the Safford area, Arizona, deep depressions is an age-related phenomenon. and is associated with an essentially complete Central areas are slightly raised above the level skeleton. Because of the immaturity of this indi­ of the peripheral areas, and they are in places vidual, the component scutes of the carapace had weakly convex. Also like the previous specimen, not all achieved full articulation with the adja­ large hair follicles are common, especially ante­ cent scutes. Consequently this carapace cannot riorly and dorsally. be restored in a free mount; instead, it is separated Another Safford area specimen, F:AM 59581 into six sections. The scutes of the rear half of this (Figure 82), is represented by approximately one- carapace are solidly fused, while those of the third of the carapace, mostly from the anterior anterior half are mostly in loose contact, indicat­ and dorsal regions. This was a mature individual, ing that the rear scutes became fused ontogenet- as indicated by the uniform development of the ically sooner than the anterior scutes. Because it scutes, although the sutural contacts are largely is in sections, overall measurements for this cara­ unfused. All of the central areas are depressed, pace cannot be made. It does not appear to be some very deeply. There are generally five large much smaller than F:AM 59599, however, indi­ hair follicles, each situated in the circular groove cating that it is nearly of maximum size. at the juncture with the anterior radial grooves. The anterior border scutes are small and sim­ The scutes of the anterior notch are smooth and ple, forming a smooth anterior border. The lateral relatively unsculptured. border scutes are also small, increasing in size The three glyptodont scutes (not seen) reported rearward and becoming conical toward the pos­ by Wood (1962) from the Safford area, and which 174 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Reported recoveries of scutes of G. texanum in the "Texas Blanco Beds," in addition to that for the type specimen, include Johnston (1944), Meade (1945), and Johnston and Savage (1955). These exhibit no important properties distin­ guishing them from any of the carapacial mate­ rial described above for the species. Most numer­ ous are those from the Cita Canyon locality, as reported in detail by Johnston (1944) and men­ tioned by Johnston and Savage (1955). These scutes (JWT 649, 1706, 1907, 2319, 2356, 2387, 2408, 2477, 2501, 2578), some of which are in articulation, are the only large series of G. texanum scutes, other than those in the carapace of AMNH 10704, from the vicinity of the type specimen. CARAPACE OF G. arizonae.—Gidley (1926) briefly described the composite carapace consisting of paratypes USNM 10336 and USNM 10537, FICURE 82.—Carapace fragment of Glyptotherium texanum (F: which he chose for his diagnosis in deference to AM 59581), posterodorsal region near iliac attachment. (Bar the more fragmentary carapacial remains of the = 10 cm.) type specimen (USNM 10536) of G. arizonae. The he placed in the Tusker local fauna, are probably carapace of the type specimen is identical to the also G. texanum. portions of the two paratypes in the mounted Akersten (1972) reported isolated scutes and specimen (see Gidley, 1926, pi. 40; also Figure 83, approximately one-third of a carapace of G. tex­ herein) so that his description for the type mate­ anum in the Red Light local fauna from western rial remains accurate. However, Gidley s figure of Texas. This carapace remains unprepared, and the carapace is misleading in that the posterodor­ no information, other than that which Akersten sal region is restored and many of the anterodor- has provided, can be derived from the specimen sal scutes have been merely set in plaster, and in its present condition. The species assignment they are not in articulation. Although this latter appears to be correct, on the basis of a combina­ condition is not evident in the figure, the resto­ tion of features, including the size and construc­ ration of the posterior region of the carapace may tion of interior and marginal scutes, the disposi­ be incorrect. Because these individuals were iden­ tion and shape of the border scutes of the anterior tical in size, a detailed description of this com­ aperture, the pronounced disposition of the dorsal posite carapace is justified as accurately repre­ scutes into rows, and the nature of the sculpturing senting the fragmentary carapace of the type of the scutes. This carapace appears to be iden­ specimen. tical to those described above; it is important for Gidley (1926) mentioned the recovery of an­ extending the geographic range of the species. other carapace from the same locality (Curtis A few isolated scutes (TMM 40254-1, 40254-2, Ranch, Arizona), which was sent to the American 40254-3, 40255-2) in the Hudspeth local fauna, a Museum of Natural History. At the time of his direct correlative with the nearby Red Light local description of G. arizonae this carapace (AMNH fauna, were identified only to genus by Strain 21808, Figure 84) had not been prepared, and (1966). These are probably also G. texanum and Gidley knew little of its characteristics. It appears add no new information concerning the features from the American Museum specimen that the of the carapace. posterior restoration in USNM 10537/10336 in- NUMBER 40 175 correctly indicates a smoothly rounded, convex USNM 10537/10336 is nearly uniformly ellip­ lateral outline, similar to that in the type speci­ soidal, with a somewhat more pronounced ante­ men of G. texanum. Although it is possible that this rior curvature. The restoration of the posterodor­ restoration is accurate and that both the convex sal scutes begins from the region of iliac attach­ and recurved outlines occurred in this population, ment, precluding positive determination of the it is more likely that the lateral profile of the posterior curvature, and also precluding deter­ posterior region was recurved as in AMNH 21808. mination of whether there was a definite preiliac Most of the scutes in USNM 10537/10336 that and postiliac division in the curvature. are in true articulation are solidly fused. This The cephalic aperture is somewhat smaller condition may be partially attributable to the old than the caudal aperture in transverse dimen­ age of these individuals, which is indicated in the sions, and the caudal aperture reaches a much noncarapacial elements of the skeletons. The fu­ higher maximum elevation. Whereas the outline sion of scutes is in places so extensive as to obscure in lateral aspect of the curvature of the marginal the boundaries between scutes, and their sculp­ scutes leading to the anterior notch continues turing is correspondingly indistinct. anteriorly and upward, the outline of the poste­ Compared with AMNH 21808 (described be­ rior notch turns abruptly upward at the postero­ low), carapace USNM 10537/10336 is shorter lateral angle, producing a vertical posterior out­ anteroposteriorly by 15 cm, and the count of 16 line for the carapace. marginal scutes for the caudal aperture is half The scutes of the outer row of the cephalic that for the undescribed specimen (Table 69). In aperture are small and quadrilateral, laterally this composite carapace, the anteroposterior becoming larger and conical. The scutes of the length and the maximum half-circumference are second row are considerably larger and project 175 cm and 215 cm, respectively, giving a length- anteriorly ahead of the scutes of the first row. width ratio of approximately 0.81. None of these scutes are conical. In the lower As restored, the lateral outline of carapace fourth of the anterior half of the carapace (i.e., in

FIGURE 83.—Composite carapace of paratype of Glyplothenum arizonae (USNM 10336/10537), right side; scutes of entire lower half and all border scutes in proper articulation; scutes of anterior half of upper region improperly articulated jigsaw fashion and set in plaster; posterior half of upper region entirely restored (figures carved in plaster) except for border scutes; dorsal outline is probably inaccurate. (Bar = 10 cm.) 176 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY the region of the poorly defined anterolateral carapace is its dorsally depressed cross section. angle and rearward), the scutes are generally Anteriorly the dorsal region is broad and almost quadrilateral and unfused, attaining an imbri­ fiat, tapering rearward and becoming more cated construction. The marginal scutes in this rounded at the caudal aperture. Whether this region are not significantly larger than the inte­ shape is an accurate restoration is problematical, rior scutes, but they are variously conical to pen­ but it appears to be correct. dant (i.e., free, or open, contacts with the adjacent Other portions of these two specimens (USNM marginal scutes). The region of imbrication in­ 10537, 10336) and of the type carapace (USNM cludes only about four rows of scutes, tapering 10536) exhibit no outstanding differences from rearward to terminate at approximately the mid- this composite mount. All three are large, adult carapace position. Marginal scutes in the poste­ individuals. Marginal scutes are prominently con­ rior half are large and conical, with increasing ical, interior scutes sculptured as described above. size rearward. The conical projections are di­ Unfortunately, these remaining fragments are not rected downward and rearward. The marginals sufficiently complete to indicate the shape of the of the posterior notch are similarly conical, and posterior border. they project rearward. The largest marginal Because carapace AMNH 21808 is more nearly scutes occur at the posterolateral angle, and their complete and its composition less obscured by conical projections are somewhat recurved. These extensive fusion, a more detailed description is measure approximately 65 mm transversely, and provided below. the diameter from the interior border to the Carapace AMNH 21808 (Figure 84), from the extremity of the conical projection is approxi­ type locality of G. arizonae, was collected in 1924, mately 80 mm, of which all but 20 mm is occu­ and except for oblique reference to this specimen pied by the projection. The scutes of the second by Gidley (1926) and an apparent reference to it row, from the midpoint of the lateral margin to by Simpson (1929), this carapace has been ig­ the posterolateral angle, bear a smaller, but nev­ nored. As restored in free mount, this large cara­ ertheless prominent, outward-projecting boss. pace is unnaturally flattened transversely, so that Sculpturing of the lateral scutes is variously ab­ end views are somewhat distorted. The proper sent or limited to near the sutural contacts in shape would be rather more rounded than flat­ indistinct patterns. tened. Accordingly, the maximum transverse The interior scutes are polygonal, tending to­ half-circumference (from side to side) would be ward a six-sided ideal shape, although few of the increased by an estimated 30+ cm. Allowing for scutes exhibit symmetrical hexagon proportions. this alteration, the transverse half-circumference Interior scutes are large and thick, averaging measures approximately 260 cm; and the antero­ approximately 50 mm side-to-side diameter. As posterior length, which appears properly restored, Gidley (1926) stated, the central figures are larger measures 190 cm. Thus, the length-width ratio is than the peripherals, and they measure approxi­ approximately 0.73, intermediate between the mately half the scute diameter. Peripheral figures same ratio for carapace USNM 10537/10336, as are considerably smaller and they are generally described above, and for carapace AMNH 15548, indistinct, owing to these individuals' advanced the type specimen for G. cylindricum, described age. The number of peripheral figures varies from below. six to 11 or 12, with eight the most common. The In lateral aspect, the carapace is uniformly central figures generally are depressed, as Gidley arched in an ellipsoidal outline from the anterior (1926) observed, except near the marginal areas notch for approximately the anterior two-thirds of the carapace. The outer surfaces of the scutes of the curvature, reaching maximum height at are flat, rugose, and pitted. the position of the iliac attachment, approxi­ A predominant characteristic of this composite mately 120 cm from the anterior notch. From the NUMBER 40 177

F'IGURE 84.—Carapace oi Glyptotherium arizonae (AMNH 21808) from type-locality: a, posterior aspect showing caudal notch and caudal bridge (arrow); b, right side, showing recurved posterior outline of dorsal border. (Bar = 10 cm.) iliac position rearward, the lateral outline curves radius of curvature, forming in posterior aspect a more steeply downward, and then beginning ap­ semicircular caudal opening. proximately three scute rows inward from the Beneath the border scutes forming the caudal posterior notch, the outline becomes recurved arch is a massive bridge of bone, not represented upward, reaching a horizontal attitude at the in USNM 10537/10336, which is solidly anky­ posterior border. Thus the carapace is divided losed to the border scutes (Figure 84a). This into distinct anterior (preiliac) and posterior bridge appears to be formed by the fusion of two (postiliac) regions, occupying approximately two- transversely directed arches of bone. The bridge thirds and one-third of the curvature, respec­ tapers laterally, terminating approximately mid­ tively. way down the caudal aperture. The bone is ex­ The anterior aperture is distinctly smaller than ternally sculptured in an irregular pattern and, the posterior one, measuring an estimated 30 cm at least along the tapered lateral extremities, transverse diameter. In contrast with the posterior appears to have been formed by coalesced scutes. aperture, the border scutes of the anterior aper­ There are a few irregularly placed hair follicles. ture continue in a smooth curvature from the The inner surface of this bridge parallels the outer lateral scutes (rather than abruptly becoming surface, attaching to the undersurface of the bor­ vertical in lateral aspect) upward and forward. der scutes of the caudal aperture in a right angle. Hence, the anteriormost point of the carapace is By analogy with AMNH 15548 (G. cylindricum, at the anterior notch. with a similar bridge), for which the tail is un­ The posterior aperture is roughly twice as large known (but in which the exact spatial position of as the anterior opening, measuring an estimated the sacrum and thus the sacrocaudal vertebra 60 ± 10 cm. The serial arrangement of the border articular facet is indicated), at least two, and scutes of the caudal aperture becomes nearly perhaps three, caudal vertebrae were situated vertical in a relatively sharp angle at the poster­ beneath the carapace and were afforded protec­ olateral extremity of the carapace. The border tion by the caudal bridge. This observation fur­ scutes continue vertically upward in a smooth ther documents Melton's (1964) suggestion con- 178 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY cerning the position of the anteriormost caudal cessory scutes, completing the outline of the cau­ ring in G. arizonae. The bridge probably provided dal aperture and forming the connection for the considerable surface area for origin of the caudal ankylosis of the caudal bridge. Several of the musculature. large posterior border scutes are centrally de­ On all marginal scutes, the dermal scale cov­ pressed in a deep and smooth excavation occu­ ered all but the interiormost position. The scute pying most of the central region. occupying the anterior notch is unbossed and is The posterior border scutes are pentagonal, uniformly rounded inward. The adjacent border with a rounded or conical outer margin and an scutes are weakly bossed. Situated within the interior apex forming the suture for the apices of interior angle between the nuchal scute and the the scutes of the second row. The lateral border adjacent border scutes are a pair of enlarged, scutes anteriorly become increasingly more uni­ heavily bossed and projecting scutes, which form formly quadrilateral, facilitating the imbrication an anterior extremity composed of five symmet­ of the scutes in the anterolateral region of the rically arranged scutes. The upper outline of these carapace. five scutes is roughly squared and was likely As restored, apparently correctly, the border confluent with the scutes covering the skull roof scutes of this carapace along the anterolateral in the living animal. margin turn beneath the remaining outer border The scutes interior to this fiat compound boss of the carapace, downward and inward. (A simi­ and the adjacent border scutes are unbossed, lar construction is indicated in USNM 10537/ conforming with the surface of the interior scutes 10336.) Thus, the circumferential arc of the car­ of the carapace. The remaining border scutes of apace near the anterior extremity covers nearly the anterior aperture are small and unbossed. 270°, this extent decreasing gradually rearward The dermal scales covered each scute entirely. to become approximately semicircular at the pos­ Beginning at the poorly defined anterolateral terior aperture. Besides the border row of scutes, angle of the margin of the carapace, the border there are two rows of interior scutes forming the scutes become gradually more conical. The bosses "undercurved" anterolateral region of the cara­ thus formed increase in size rearward, becoming pace. The interior rows taper rearward to ap­ markedly elongate, externointernally compressed, proximately the midcarapace position, where the and forming a scalloped outline in lateral aspect. undercurved region gives way to the downward- The scute with the greatest elongation occurs 18 directed lateral scutes. These undercurved scutes scute positions from the posterior notch or ap­ appear to have maintained some mobility. The proximately six scute positions anterior to the fourth row of scutes forms a sharp angle, occu­ posterolateral angle. From this scute rearward, pying the actual anterolateral extremity of the the border scutes become conical, the boss reach­ carapace. This row continues rearward, poste­ ing maximum development at the posterolateral riorly becoming the first row interior to the border angle. Conical bosses of the caudal aperture occur scutes. Sculpturing of the undercurved scutes is only on the two lateral scutes of each side at the irregular. Radial grooves are variously indistinct extremity of the opening, adjacent to the poster­ or absent, and the central figure occupies nearly olateral angle. The remaining 10 scutes on each the entire scute. The central figure is generally side increase in size toward the caudal notch, convex and punctate; a few of the scutes include becoming rounded and extremely large (largest hair follicles, and one scute exhibits a central scutes of the carapace) at the posterior notch. excavation. The scutes forming the anterolateral Sculpturing, indicating the boundaries of the der­ angle of the carapace are generally quadrilateral, mal scales, remains confined to the area near the with rounded (convex) interior borders; they all interior border of each scute. Wedged between appear to have been solidly ankylosed with the the large posterior border scutes are smaller ac­ interior scutes. Sculpturing on these scutes is in- NUMBER 40 179

distinct, the dermal scales apparently occupying apex regions, where the scutes are irregular in the entire external surfaces. shape. Otherwise, there are seldom more than The four-sided scutes of the posterolateral por­ two peripheral figures separating adjacent central tion of the carapace are arranged in anteropos­ figures. The surface texture of the interior scutes terior and vertical rows. Dorsally and anteriorly is punctate, although some are rather smooth the rows become obscure and the scutes become with only a few indistinct small foramina. Side- generally six-sided (but frequently five and seven- to-side dimensions of the scutes are generally 45- sided) in the more dorsal regions of the carapace. 55 mm. A few are slightly larger, and some are Along the first row of interior scutes, posterolat- slightly smaller. erally and posteriorly, the central figures are large Compared with carapace USNM 10537/ (occupying at least 80 percent of the scute sur­ 10336, this carapace differs considerably in shape face), generally circular, and convex. The number and construction. Among the most notable dis­ of peripheral figures varies from seven to 12, and tinctions are the recurved posterior outline, the the radial groove in some scutes is indistinct. The cephalic boss, and the caudal bridge, none of external surface of the scutes is generally punc­ which are present in the composite carapace that tate, and the central regions are not elevated Gidley (1926) described and figured as the para­ above the level of the peripheral figures, a con­ types for G. arizonae, but which are present in dition characterizing all the interior scutes except carapace AMNH 15548 (G. cylindricum). There­ for those of the anterolateral angle and of the fore, while the differences between G. texanum and anterior cephalic boss. G. arizonae are valid distinctions, there is much Interior scutes are generally six-sided, and any less difference between these two species and G. disposition into rows is obscured by the complex cylindricum than has previously been supposed. pattern of the sculpturing. Interiorly the central Moreover, there are no features in the carapace figures become smaller, but their diameters are of G. anzonae that are not derivable, in a simple never less than half the diameter of the scutes, ancestor-descendant relationship, from G. tex­ and they are never smaller than the peripheral anum. figures, although in some scutes the peripheral From the Curtis Ranch locality, St. David figures and the central figure occupy approxi­ Formation, Lammers (1970:32) reported a few mately the same amount of surface area. The additional carapacial scutes (not seen), including central figures of perhaps one-fourth of the scutes some from "portions of the carapace apron," are excavated, forming generally shallow central presumably referring to the scutes of the antero­ depressions, and a few are rather deeply de­ lateral region. Besides the two carapaces de­ pressed. Large hair follicles are generally wanting; scribed above, this is the only additional glypto­ this individual (which is probably very mature to dont material collected from the Curtis Ranch old age) was mostly devoid of pelage in the site. carapacial region. Carapacial remains from the Seymour Forma­ The central figures of the interior scutes are tion, northern Texas, are referable to G. arizonae generally circular. The shapes of the peripheral on the basis of sculpturing and overall construc­ figures are more variable, although they are usu­ tion. Partial carapaces and isolated scutes from ally five- to six-sided and bilaterally symmetrical the Seymour fauna represent a large number of about the radius of the scute. The number of individuals, perhaps as many as 80 or more, peripherals varies from seven to 13, the lower according to Melton (1964). Particularly perti­ counts being more common. The radial grooves nent are Melton's remarks concerning the re­ are rarely continuous across the sutures. Occa­ curved outline of the posterior region of the car­ sionally an accessory peripheral figure occupies apace, and the measurement for the maximum the space between two peripherals, usually in the width of carapace UMMP 46320, which exceeds 180 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

214 cm. Both of these characteristics indicate a exterior surface; the central figures are flat to close affinity with AMNH 21808 and the com­ weakly concave, and occasionally deeply concave; posite type carapace USNM 10537/10336. The there are generally seven to 10 peripheral figures, details of the sculpturing of the scutes and their with shallow grooves delimiting them from the arrangement and disposition, which Melton ac­ central figure, and there are variable numbers of curately described, confirm the close similarity to relatively small hair follicles. These scutes do not Curtis Ranch specimens. Undescribed specimens differ from those known with more certainty as in the Midwestern University collection, re­ pertaining to G. anzonae. The similarity of these covered subsequent to Melton's study, do not add scutes with those of known G. arizonae specimens, any significant information. The noncarapacial the similarity of the mandible from the same osteology also indicates the identity of the Sey­ locality, and the general acceptance of near con­ mour representatives with those from the Curtis temporaneity of the Holloman fauna with the Ranch locality. Seymour fauna (Meade, 1953; Melton, 1964; The only remaining carapacial material per­ Dalquest, 1977; and Hibbard, pers. comm.) are taining to G. arizonae is from the Holloman gravel sufficient grounds for assuming specific identity pit, near Frederick, Tillman County, Oklahoma. of the glyptodonts from these two localities. A fragmentary mandible and two associated mar­ Hay (1927) reported unidentified glyptodont ginal scutes from this locality were the material remains (unseen, probably scutes) from Rock upon which Meade (1953) founded Xenoglyptodon Creek, Briscoe County. On the basis of the bio­ fredericensis, and upon which Melton (1964) sub­ stratigraphic equivalence of the Rock Creek sequently claimed synonymy with the Seymour fauna with the Gilliland and Holloman faunas representatives. These two marginal scutes (Hibbard and Dalquest, 1966), this occurrence is (TMM 934-37) are from the midlateral border, probably G. arizonae. Lundelius (1967, 1972) has and they are identical to corresponding scutes in reported glyptodont remains that he assumed to the better known carapaces of G. arizonae. be Late Pleistocene in age from Shafter Lake, An improperly restored carapace in the Stovall Andrews County, which lies on the southeastern Museum of Paleontology, University of Okla­ edge of the Great Plains to the south of the Blanco homa, has been mentioned by several authors, Beds. These scutes (unexamined), on geographic including Meade (1953) and Simpson (1929b). grounds perhaps represent G, arizonae, for such an Apparently only Simpson had the opportunity inland and western recovery of Late Pleistocene for direct observation, and referring to the Hol­ glyptodonts is improbable. loman specimen in his description of Boreostracon Finally, a small collection of isolated scutes flondanus, he stated in a footnote; "Through the from three localities along the Santa Fe River, kindness of C. N. Gould I have been able to Gilchrist County, Florida (Figure 86); four scutes examine parts of this important find. The plates from Charlotte Harbor, Charlotte County, Flor­ do not agree with those of any other North Amer­ ida; and one scute from the Inglis 1A locality. ican specimen known to me, and probably rep­ Citrus County, Florida, represent the possible resent an undescribed species" (Simpson, 1929b: occurrence of G. arizonae in eastern United States. 582). The restoration of this unnumbered cara­ These scutes are not associated with any noncar­ pace (Figure 85) is similar to that for the type apacial remains except for an isolated tooth from specimen of G. texanum. The scutes were im­ a baby individual from the Santa Fe River, which properly articulated in the restoration, randomly is without diagnostic value. These scutes are iden­ set in plaster, jigsaw fashion, with no proper tified as Glyptotherium sp., cf. G. arizonae on the alignment or positioning of the scutes. The scutes basis of their size and the nature of the external are all large (up to 55 mm in diameter), with a sculpturing. They are larger than scutes of G. large central figure occupying more than half the texanum. The central figures of representative in- NUMBER 40 181

•'

FIGURE 85.—Improperly reconstructed carapace of Glyptotherium arizonae from the Holloman locality, Oklahoma, unnumbered specimen, Stovall Museum, University of Oklahoma: a, postero(?)-lateral aspect of right(?) side; b, detail of carapace showingjigsaw fashion reconstruc­ tion. (Bar = 10 cm.)

terior scutes are considerably larger than the cylindricum. The closest resemblance is to G. ari­ peripheral figures, excluding G. floridanum, and zonae and G. cylindricum; the latter species is omit­ the central figures are weakly convex; external ted on geographic grounds. For discussion of the figures for two of the scutes bear deep central age and associated faunas for these Florida spec­ excavations as in G. arizonae, G. texanum, and G. imens, see "Remarks" under the taxonomic treat- 182 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

^%#g'

FIGURE 86.—Isolated scutes oi Glyptotherium sp. cf. G. arizonae from F'lorida: a, interior carapacial scute UF/F'SM 10438; b, two interior carapacial scutes UF/FSM 10743; c, isolated scutes UF/ F'SM 10428 (upper left, interior scute from near lateral or posterior border; upper right, interior carapacial scute; lower left, scute from distal row of first or second caudal ring; lower right, scute from proximal row of second or third caudal ring); d, UF/F'SM isolated scutes (left, interior carapacial scute near margin of carapace; right, border scute from caudal aperture; all from Sante Fe River, UF site 1). (Bar = 10 cm.) ment for G. arizonae and the distribution account decade earlier. Direct comparison of these two for the species. specimens reveals no less distinction now than CARAPACE of G. cylindricum and G. mexicanum.— when Brown described this carapace, but the Brown (1912) accurately described the carapace addition of the several other specimens of North of AMNH 15548 (Figure 87), designating it the American glyptodonts indicates a much closer type carapace for the genus Brachyostracon. At the relationship than the distant one Brown pro­ time there appeared to be a significant difference posed. As discussed elsewhere, AMNH 15548, between this carapace and that of Glyptotherium here designated Glyptotherium cylindricum, is distinct texanum, which had been established by Osborn a from other North American specimens primarily NUMBER 40 183 in the characters of the pelvis, although even in Brown briefly considered the description of this part of the skeleton the differences are not as Glyptodon mexicanus (= Glyptotherium mexicanum) by great as Brown perceived. Other regions of the Cuataparo and Ramirez (1875). As Brown body, so far as known, indicate a rather close pointed out, their description, although largely relationship with other North American species. deficient and at least in part incorrect, is never­ The dentition, although distinctive, is assuredly theless substantial enough to warrant at least similar; and the carapace, previously considered nominal retention of the species. Fortunately, to be highly characteristic and in itself partially Brown was privileged to examine the carapace; indicative of a distant relationship with other subsequent to his study the carapace was lost North American representatives, is actually quite along with the remainder of the two type speci­ similar to carapace AMNH 21808 (G. arizonae), mens, and the whereabouts of these fossils remains verifying the expected close relationship inferred unknown. Cuataparo and Ramirez mistakenly by comparison of dentitions and pelves. Carapace reversed the carapace end for end in their descrip­ AMNH 15548 is described below, following a tion and in their figure. Brown (1912), recogniz­ short digression on the carapace of G. mexicanum. ing this error and having examined the carapace

4r Z i 's-'

FIGURE 87.—Carapace oi Glyptothenum cylindricum (AMNH 15548): a, right side; b, caudal notch; c, detail of anterior aperture. (Bar = 10 cm.) 184 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY firsthand, provided a brief description and three there are frequent deep recessions in the central useful photographs of the type specimen of G. figure similar to those of the other two species mexicanum. These two descriptions together com­ under comparison. Peripheral figures vary in prise the only accurate information regarding this number from eight to 12, according to the original carapace, and the verification of its characters in authors; this determination is borne out by in­ Brown's photographs (pis. 13-15) justify an ex­ spection of Brown's photographs. tracted amplification. According to Brown, this carapace is distin­ Measurements for this carapace (unnumbered, guishable from that of G. cylindricum in several reported to be preserved in the Mexican National characters: larger central figures; transverse rows Museum of Natural History prior to its disap­ of scutes not continued far beyond border and pearance) are 183 cm length and 240 cm width, more firmly united; border scutes larger and more as given by the original authors, and providing a pendant. These differences are attributable to length-width ratio of 0.76. These measurements, variation and are insufficient as diagnostic fea­ if correct, are nearly identical to those for the tures. There appears to be no reason to assume type carapace of G. cylindricum (AMNH 15548). taxonomic distinction on the basis of these two Near the posterior border the scutes are recurved carapaces. As discussed in the taxonomy section, upward in a fashion identical to that in AMNH however, it is appropriate nominally to retain G. 21808 (G. arizonae) and AMNH 15548 (G. cylindri­ mexicanum, primarily on the basis of the dentition. cum.) The caudal aperture in lateral aspect The carapace of G. cylindricum, as originally achieves a near vertical outline, and the cephalic described by Brown (1912), remains the most aperture is much smaller. There is apparently a well-preserved and most perfectly restored one in preiliac and postiliac division of the carapace in North America. That it differs little from the lateral aspect, comprising roughly two-thirds and carapace of G. mexicanum was admitted by Brown, one-third of the dorsal arch, respectively. The but he nevertheless chose to consider the distinc­ marginal scutes are conical and some are pen­ tions as diagnostic. Carapace AMNH 15548 is dant. They increase in size rearward. Border large, measuring 170 cm anteroposteriorly and scutes of the anterior notch are simple and un­ 248 cm in the transverse (half-circumference) di­ bossed; those of the posterior notch are very large mension, with a length-width ratio of 0.68, which and conical. Along the lateral margin the scutes is slightly less than that estimated for AMNH of the first interior row bear knoblike bosses, 21808 (0.73), and somewhat greater than that for identical to those in G. cylindricum and G arizonae. either USNM 10537/10336 (0.81) or the G. mex­ There appear to be 24 marginal scutes along the icanum type specimen (0.76), all of which are posterior border, 14 along the anterior border, similar in carapacial features to AMNH 15548. and a total count of 48 along the right side, from This range in the length-width ratios, from 0.68 the anterior notch to the posterior notch. These to 0.81, is attributable to variation within the counts do not differ greatly from those of either genus. Therefore, the relative length-width pro­ G. arizonae or G. cylindricum. It cannot be deter­ portions should not be used as taxonomic criteria mined whether there is an anterolateral region of on the species level, although these proportions imbricated scutes; according to Brown's photo­ collectively may be indicative of generic charac­ graphs the anterolateral scutes are firmly united terization. Moreover, the 0.79 and 0.75 length- and continuous with those of the dorsal region. width ratios of the two mounted carapaces of G. Sizes of the scutes, as given by Cuataparo and texanum, discussed above, are not as radically dif­ Ramirez, and as shown in Brown's photographs, ferent from that of G. cylindricum as Brown sup­ correspond closely to the sizes in G. cylindricum and posed. G. arizonae. Central figures are generally large, In lateral aspect the shape compares favorably never smaller than half the scute diameter, and with AMNH 21808. The anterior dorsal arch is NUMBER 40 185

ellipsoidal, reaching a maximum elevation at the only a few ankylosed scutes occupying the ventral position of the iliac attachment approximately margins of some of the scutes of the caudal border. 105 cm from the nuchal border. From that point These are largest in the central position; laterally (approximately two-thirds the anteroposterior the bridge is represented by smaller accessory length from the anterior border) rearward, the scutes solidly fused to several of the dorsolateral carapace curves more steeply downward, becom­ border scutes. For this reason, and because of the ing recurved upward at the posterior aperture nature of the external sculpturing of the scutes in beginning at approximately the second interior this specimen, we believe that this individual was row of scutes from the posterior border. The an adult, and not an old-age representative as in outline in lateral aspect curves upward somewhat AMNH 21808; hence the presence and relative beyond horizontal at the posterior border. Thus, development of the caudal bridge are not diag­ as in AMNH 21808, the carapace is divided into nostic. distinct anterior (preiliac) and posterior (post­ Associated with this carapace, and in proper iliac) regions, occupying approximately two- restoration in th^e free mount, is the complete thirds (105/170) and one-third (65/170) of the pelvis. Iliac scars on the undersurface of the car­ curvature, respectively. apace mark the attachment of the iliac crests to The anterior aperture measures approximately the external armor. The sacrum is therefore in 45 cm. This figure is considerably larger than the proper position as restored. The sacrocaudal ar­ 30 cm estimate for AMNH 21808, although the ticulation lies some 22 cm inside the posterior figure for the latter could well be larger because notch. This figure corresponds closely with the of distortion in the free mount. Thus, the lateral anteroposterior diameter of the first three caudal extent of the anterior opening, the outline of vertebrae of AMNH 21808 (no caudal vertebrae which is almost vertical, is more easily identified are associated with carapace AMNH 15548), ten­ than in AMNH 21808, and the anteriormost tatively indicating by analogy that probably point of the carapace, although it is the nuchal three caudal vertebrae were situated beneath the scute, does not project far beyond the border posterior aperture, and corroborating Melton's scutes of the anterior opening, thus distinguishing (1964) suggestion that the first three caudal ver­ G. cylindricum from G. arizonae. The anterior aper­ tebrae probably lay interior to the caudal border. ture includes 26 border scutes between the an­ The border scutes of the anterior aperture are terolateral angles, approximately the same as that uniformly bossed in blunt projections dorsally; derived for G. mexicanum. laterally these projections expand to fuse with the The posterior aperture is much larger, measur­ adjacent scutes. None of the border scutes of the ing an estimated 70 cm in transverse diameter. anterior opening are large. The four scutes of the This measurement compares favorably with that first interior row at the anterior notch extend of G. arizonae, and, similarly, it is roughly twice as somewhat beyond the border scutes, and their large in transverse diameter as the anterior open­ anterior margins form a transversely straight out­ ing. The serial arrangement of the border scutes line and anteroposteriorly a sharp convex surface. of the caudal aperture becomes vertical, begin­ Thus, as in the similar bossed structure of AMNH ning in a relatively sharp angle at the posterolat­ 21808, it appears that the casque scutes of the eral position. As in AMNH 21808, the posterior rear portion of the skull fit snugly against this border scutes continue vertically upward to a projection, similarly protecting the short neck uniform curvature, forming a semicircular outline from exposure. The remaining border scutes of in posterior aspect. the anterior aperture laterally become more The massive posterodorsal bridge located in solidly ankylosed along their free lateral borders. the dorsal region within the caudal aperture of The open external boundaries in this specimen, AMNH 21808 is represented in this specimen by as compared with the closed borders in AMNH 186 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

21808, may be simply an age-related factor, and The interior lateral scutes become increasingly therefore should not be accorded any taxonomic more quadrilateral anteriorly. At least four lon­ significance. These small anterior scutes were cov­ gitudinal rows of anterolateral scutes are discern­ ered entirely by their dermal scales; hence these ible. The imbricated and apparently loose fit of scutes are devoid of sculpturing. None possess these scutes indicate limited mobility of the car­ hair follicles, and they are uniformly punctate. apace in this region. In contrast with AMNH Beginning at the relatively well-defined antero­ 21808, these scutes are not underturned; instead, lateral angle, the border scutes become increas­ they are vertically oriented, their external faces ingly more pointed and conical. The lateral scutes directed outward and anteriorly. on the anterior half of the carapace margin are The first row of scutes internal to the anterior noticeably "villiform," i.e., their boundaries with border is anteroposteriorly complete. At least four adjacent border scutes are deeply excavated, pro­ additional transversely oriented rows occur pos­ ducing the pendant scutes of the anterior portion terior to the first interior row, the dorsal extent of of the margin as described by Brown (1912) for the rows diminishing rearward. Here, also, the this specimen. As for the anterior border scutes, scutes are generally quadrilateral, as a continua­ these excavations (i.e., the pendant scutes) might tion of the apparently flexible construction of the well be present because of the relatively younger anterolateral region of the carapace. Additional age in comparison with AMNH 21808 and the transverse rows of scutes along the margin are type specimen of G. mexicanum. discernible, but none extend farther upward than The posterior lateral border scutes increase in the sixth or seventh scute from the border. There transverse (border scute suture to border scute are also at least faintly recognizable longitudinal suture) dimension. At approximately midposi- rows, parallel to the row of lateral border scutes. tion, the border scutes are elongate and bossed. Thus, the margins of this carapace were flexible Posteriorly the scutes become more conical, reach­ not only at the anterolateral angle, but also to a ing a well-defined, cone-shaped, external con­ lesser degree, along the lateral regions as well. struction at the posterolateral angle. These cone- Central figures on the scutes near the margins are shaped scutes project downward and posteriorly, uniformly large, and the peripherals are poorly increasingly more so rearward. The largest coni­ defined and small. The size of the central figures cal scutes are the four situated in the posterolat­ decreases interiorly, and the peripheral figures eral angle. Like the other border scutes, each of become more prominent. In the midregions of the the lateral border scutes was covered by a single carapace, the central figures are approximately dermal scale. There is no sculpturing or radial the same size as the peripherals, but never smaller. grooves, except along the interior borders of the A number of the central figures (at least half) are scutes near the interior sutures. weakly concave; a smaller number, perhaps one- The border scutes of the posterior aperture are fourth, are deeply depressed in the midregion of uniformly large and most bear prominent conical the central figure. The interior scutes are gener­ bosses occupying nearly their entire external sur­ ally six-sided; occasional five-, six-, eight-, nine-, face. Nine of the 27 scutes of the posterior aper­ and 10-sided scutes occur but with lesser fre­ ture bear deep depressions in place of the apices quency with increasing number. The scutes are of the cones, and several of these are punctuated uniformly punctate. There are no smooth-sur­ by a single large foramen. These large foramina faced scutes, quite in contrast to the large number appear to terminate blindly without entering into of smooth scutes in AMNH 21808. This possibly the internal portion of the scutes. The posterior age-related difference in surface texture perhaps border scutes are pentagonal, with conical outer corresponds to the greater number of hair follicles margins and interior apices forming the suture in AMNH 15548, although in this specimen also, for the corresponding apices of the scutes of the there are only a few. first interior row. The central figures of all the interior scutes are NUMBER 40 187 flat or depressed, as described above. They are Late Pleistocene species in Florida cannot be never convex and elevated above the level of the ruled out a priori, however, although this pre­ peripheral scutes as in G. texanum. The number of sumption is here maintained for lack of evidence peripheral figures of the interior scutes varies to the contrary. Moreover, chlamytheres, a group from six to 10, with eight the most frequent. Side- of giant dasypodid edentates comparable in part to-side diameter of the interior scutes falls consis­ to glyptodonts, often occur in the same fauna. tently between 45 and 50 cm, averaging slightly Although glyptodonts and chlamytheres are smaller than in AMNH 21808, and the diameters likely not ecologic equivalents, they are probably seem to be less variable. ecologically "analogous," and the likelihood of a In summary for the carapace of G. cylindricum, third ecological "analog" among glyptodonts there is only a single characteristic that separates seems remote. Therefore, despite the unfortunate it from that of G. arizonae—the lateral outline of lack of noncarapacial remains in the type mate­ the anterior aperture; and there is apparently no rial of G. floridanum, comparison on the basis of clear distinction from the carapace of G. mexi­ carapacial features, as described below, is an canum. The vertical outline of the cephalic aper­ adequate means of specific identification for Flor­ ture in G. cylindricum might also be attributable to ida glyptodonts. variation, indicating no valid species distinction Furthermore, identification of Texas represent­ in the carapaces of these three species. The cara­ atives of this species must rest ultimately on com­ pace of both G. cylindricum and G. mexicanum there­ parisons with the type specimens from Seminole fore differs from that of G. texanum in exactly the Field, although comparison with noncarapacial same fashion as does G. arizonae. remains of other Florida fossils that are circum­ CARAPACE OF G. floridanum.—The carapace of stantially identified as G. floridanum (on the un­ G. floridanum is known from several localities in desirable basis of carapacial features) must also Florida and Texas. It is ironic that this species, be considered. It is this latter consideration upon which is known from by far the greatest number which the Texas Late Pleistocene glyptodonts are of localities, is not represented by a single com­ referred to the same species as the Florida glyp­ plete carapace adequately preserved for mount­ todonts, in agreement with Lundelius' (1972) con­ ing. It was primarily on the basis of a large series clusions founded primarily on carapacial resem­ of scutes and partial carapaces that Simpson blance. (1929b) established Boreostracon flondanus. As dem­ It should be pointed out, however, that despite onstrated below, this material belongs in Glypto­ the apparent resemblance in carapacial features therium, but the carapacial characters are suffi­ between the Florida and Texas representatives, ciently well established in the type series to retain there is inconclusive evidence that there may be the specific assignment. Unfortunately, the Sem­ two species, for the mandibular and dental char­ inole Field type specimens were not associated acters are not identical. Therefore, the Florida with any noncarapacial elements, save for a frag­ population would retain the species designation mentary mandible, two teeth from an infant, and G. floridanum, while the Texas population would one adult tooth (the latter from nearby Sarasota, represent an unnamed species, unless the resur­ in a private collection). Hence, the relationship rection of the species G. petaliferus (Cope, 1888) of this material to other Late Pleistocene glypto­ were to be argued as valid, a matter that Simpson donts in Florida can be established only on car­ (1929b) partially resolved by declaring Glyptodon apacial comparisons and biostratigraphic corre­ petaliferus Cope a nomen nudem. However, Simp­ lation; therefore, comparison of noncarapacial son did not designate another species name for remains can be made only on an inferential basis. the Texas glyptodonts, which he considered spe­ Because Simpson's designation sufficiently de­ cifically distinct from the Florida representatives. fined the characters of the species, his assignment At present, the problem is only hypothetical, for is here retained. The possibility of more than one there is no compelling reason to consider the two 188 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY populations as different species, and, regardless of otype (AMNH 23547) and most of the designated inconclusive evidence indicating a specific dis­ paratypes (AMNH 23548-23562), although an tinction, the carapacial characters of the repre­ isolated scute (AMNH 23550) in this series, which sentatives from both states are identical. Simpson called "an unusually large scute" The carapacial fragments recovered from the (Holmes and Simpson, 1931:412), belongs to a Seminole Field locality (type-locality for G. flori­ male. Accordingly, Simpson's description of the danum) apparently can be separated into two dis­ species applies largely to the female characters, tinctive groups, those bearing the characters de­ among which relatively small size and mainte­ scribed by Simpson (1929b) and Holmes and nance of open sculpturing along the scute sutures Simpson (1931) for the types, and those closely are most prominent. Simpson apparently consid­ resembling the carapace of G. arizonae and G. ered the female carapacial fragments as repre­ cylindricum. These are here assumed to represent senting immature individuals. This contention is sexual dimorphism for the following reasons: (1) here rejected because there is full scute-to-scute they are associated in the same lots, and there is contact with firmly united, although largely un­ no reason to assume that they were recovered closed, sutures. This is an indication of maturity, from separate strata; (2) as discussed above, there and except for possible increase in thickness, there is no compelling justification to assume two glyp­ is no reason to assume these scutes would have todont species in the same fauna; (3) a similar become significantly larger with age. Further­ dimorphism occurs in the Texas population, thus more, the recovery of a large number of immature providing further evidence for rejecting a two- individuals at the same stage of development, species hypothesis, since sympatry should not be without size gradient, seems improbable. Simp­ expected to occur over such a great distance; and son's general description of the scutes for the (4) there is a general consistency, rather than species, here considered as pertaining to the fe­ separation into discrete groups, in the noncara­ males, cannot be improved upon. pacial remains, an unlikely circumstance in the case of sympatric species, i.e., if there are two The central scutes of this species show a typical pattern species recognizable according to carapacial fea­ which is variously modified in the more marginal regions. These scutes are generally hexagonal, although they may be tures, they should also be recognizable according very irregular and with four to eight sides. The inner surface to noncarapacial features, and this is not the case is concave, with numerous vascular foramina . The dorsal in either the Texas or the Florida populations. surface is divided by grooves into a nearly circular central Thus, there appears to be sexual dimorphism area and a number of smaller marginal areas, corresponding in carapacial features of G. floridanum. That this the overlying scales. The area of the primary scale is here truly central, generally somewhat depressed in the center, variation is attributable to sexual dimorphism and strongly punclale |vascular foramina]. The marginal rather than ontogenetic variation is indicated by areas, which are no! always well differentiated from each the more or less discrete separation into one or other, also have vascular openings and are especially marked the other of the two groups, without intermediate by a number of irregular radiating vascular grooves. On each plate these areas are generally six to nine in number, conditions, and by the representation of carapaces and in this central region they generally were separated by from mature adults that fall into only one or the marginal grooves |along the sutures) from their neighbors on other of the two groups. By analogy with other ihe next plate Follicles, usually two to four in number, mammals, the carapaces with small scutes, indic­ occur only between a primary scale and the surrounding ative of smaller individuals, are identified as fe­ intercalary scales and lend to occur only on the anterior and male and the larger ones as male. lateral margins of the primary scale. Most of the carapacial remains that Simpson Toward the borders the cenlral or primary scale area becomes relatively larger and, especially anterolaterally, may described and figured as the type and referred touch the posterior margin of (he scute Hair follicles specimens from the Seminole Field locality be­ are somewhat less numerous laterally, usually only one or long in the female group. These include the hol­ two on each scute, occasionally none at all. NUMBER 40 189

Along the borders of the carapace the scutes are regularly tures characterizing the species are (1) the rela­ arranged in transverse rows. In the more central part traces tively small size of the central figures of interior of regularity are discernible, but the segmented arrangement is masked by interpolation of accessory rows and individual scutes, diameters approximately half the scute scutes diameter, and not significantly larger than pe­ Normally the scutes are united by open sutures, ripheral figures; and (2) the maintenance of open crossed by numerous long interlocking spicules of bone which sutural contacts in the group here identified as hold the scutes firmly together female. The central figures are weakly concave All of the scutes of ihe marginal row were covered by a and raised slightly above the level of the periph­ single scale each. Those of the nuchal border, or anterior eral figures, and the surface of the scutes is gen­ notch were without bosses, and the first two rows, at least, of this part were somewhat movable against each other erally punctate. The number of peripherals varies The lateral borders are formed by projecting boss-like between six and nine, and for the interior scutes, scutes which are firmly united with each other and the apices they are nearly as large as the central figures. For of which point backward, downward, and slightly oulward. the females, the scutes are relatively small. Inte­ The scutes of the posterior border bear low pointed bosses rior scutes of the type specimen AMNH 23547 (Holmes and Simpson, 1931:408-410). measure approximately 30 mm transverse diam­ In addition, there is one important feature not eter between parallel sides and 20 mm thickness. indicated by Simpson. In the type specimen Measurements on the largest typical scute of the AMNH 23547 (Figure 88), which is from the posterior border (third from left) are 42 mm posterior notch, there is the indication of a re­ transverse diameter, 40 mm from free edge to curved posterior outline similar to that in the interior sutural apex, and 29 mm maximum other species and comparable to the weak cur­ thickness measured through the external boss. vature found in G. texanum. Because the curvature A statistical analysis of isolated, individual is weak and the series consists of only a small scutes would be falacious because it cannot be region of the carapace, the recurved outline is not determined how many are from separate individ­ immediately evident but is nevertheless present. uals. There is, however, a distinct bimodal sepa­ The similarity of these carapacial features with ration in scute size in the Seminole Field speci­ those of other species referred to Glyptotherium is mens. The smaller ones, as described above, are evident from Simpson's description and from the attributed to female individuals. Larger ones, recognition of the posterior recurved outline. Fea- with slightly different sculpturing, are attributa­ ble to males. None of the larger ones are in serial articulation. Another characteristic that seems to be limited to the female specimens is the pointed construction of the bosses on the posterior border scutes. Scutes attributable to males are characteristi­ cally larger, and apparently there is no open groove marking the boundaries between scutes, 'ft ••;•# • • r.&.-^M*i&m the sutures closing to the level of the peripheral figures. Therefore, the sculpturing delimiting the peripheral figures is continuous across sutural contacts rather than generally separated by the open marginal groove as in the females. A typical border scute from the posterior notch (AMNH 95728, Figure 89) measures 59 mm transverse FIGURE 88.—Carapace fragment of Glyptotherium floridanum holotype (AMNH 23547) from dorsal region of caudal bor­ diameter, 57 mm diameter from the free edge to der. (Bar = 10 cm.) the interior sutural apex, and 44 mm maximum 190 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

^jg, - thickness through the external boss. The latter figure does not adequately represent the size dis­ *YJ tinction, however, since in the males the external . )\_ _ u_r _*afcifV 'ijmf boss does not project as far from the level of the surrounding surface as in the females. In the male scute under consideration, the thickness at the interior border is 31 mm, whereas the correspond­ m ing thickness for the posterior border scute of the female described above is 24 mm. A typical male 3ffl_ i?O^^JBP^^^^SS8k interior scute from this same assemblage (AMNH #mhWm f^JfiW^^BSSk 95727, Figure 89) measures 51 mm in the trans­ >W%f ''*• M p»^fj i jJT'j'- ^ r5r_f '*W&^^^ verse diameter and 23 mm in thickness, substan­ A tially greater, especially transversely, than in the W%ur*'?i )** female. ^j^o^fjjr - \j& Although it cannot be determined from the available specimens whether the carapace of the FIGURE 90.—Interior section of carapace of the Catalina male possesses a recurved posterior outline, it is Gardens, Pinellas County, Florida, representative of probable that it does, on the basis of comparison Glyptotherium floridanum (UF/FGS 6643). (Bar = 10 cm.) with the females and with the Ingleside, Texas, specimen discussed below. females, the central figure for the interior scutes In comparison with the carapacial features of is but slightly larger than the peripherals, and it G. arizonae and G. cylindricum, the male represent­ is generally weakly depressed. atives in the Seminole Field population exhibit The virtually complete carapace from the no distinguishing characters for the species other nearby Catalina Gardens locality, Pinellas than in the nature of the sculpturing. As in the County (UF/FGS 6643, Figure 90) is referred to G. floridanum on the basis of comparison with the male scutes from the type-locality Scutes from comparable regions of the carapace are identical from these two localities when comparisons are made with scutes of the male variety from Semi­ nole Field. Posterior border scutes, anterior bor­ der scutes, and interior scutes are identical in every respect including size, and they differ from the female scutes from the type locality in exactly the same fashion as do the males. Unfortunately, this carapace is preserved in several flat sections of articulated scutes that are not sufficiently well preserved to allow for con­ struction of a free mount. Lateral border scutes are missing and the anterior and posterior aper­ tures are incomplete, precluding full comparison F'IGURE 89.—Scules oi Glyptothenum floridanum from type-lo­ with G. anzonae and G. cylindricum; nor can it be cality, Seminole Field, Pinellas County, Florida: upper row, determined whether there was a recurved poste­ three interior carapacial scules from an adull female (AMNH 95726); lower left, interior carapacial scute rior outline in this carapace. It appears, however, (AMNH 95727); lower right, posterior border scute (AMNH that the only distinction of this carapace from 95728) from adult males. (Bar = 5 cm.) those of the other species lies in the external NUMBER 40 191 sculpturing, which is identical to that of the Simpson (1929b) included a tooth from the Seminole Field representatives and is character­ "Moore Collection," which was apparently re­ istic for the species. covered from the same locality. Simpson's refer­ Gidley (in Hay, 1927:274) listed Glyptodon sp. in ence of that tooth, which he figured, was correct. the fauna from Melbourne, Brevard County, Whether he had access to carapacial material is Florida. Simpson (1929a) reported that these unclear, but at approximately the same time, glyptodon remains were referable to Boreostracon both Gidley and Hay received together a total of floridanus. Gazin (1950) has supplied the only seven scutes as gifts from the same Mr. Moore published information concerning the nature of from the Sarasota locality (USNM 11675, 11975). this glyptodont material, stating that it consists To the best of our knowledge these have not been of five or six scutes, and he concurred with Simp­ reported. There are two interior scutes, one with son's identification. These scutes are in the Na­ a relatively large central figure indicating a more tional Museum of Natural History (Smithsonian marginal proximity; two border scutes from the Institution), in the Melbourne, 1928 collection lateral margin; one border scute from the poste­ (USNM 256750). Two scutes are from the poste­ rior border; and one border scute from the pos­ rior border, one is a typical interior scute, and terolateral region of the anterior notch, and one two are small scutes probably from the cephalic interior scute from the anterolateral region. These shield. The two posterior border scutes are com­ are all large scutes, comparable to those of the parable in every detail to those of the female G. Catalina Gardens male G. floridanum. The poste­ floridanum. They are small, the largest measuring rior border scute is typically large, with only a 49 mm transverse diameter, 45 mm anteroposte­ low, blunt external boss; it measures 57 mm in riorly, and 32 mm maximum thickness, and they the transverse dimension, 59 mm in the external- possess distinctly pointed, cone-shaped bosses. All internal dimension, 37 mm maximum thickness, five scutes are likely from a single individual. and 28 mm thickness at the interior suture. The Leidy (1889b) reported glyptodont scutes from border scute from the posterolateral margin is Peace Creek, near Arcadia, DeSoto County, Flor­ typically pointed in a downward-directed trian­ ida, as Glyptodon petaliferus. Hay (1924) rejected gular projection, verifying the existence of conical Leidy's identification, apparently on grounds of posterolateral border scutes in this species. The geographic separation from the type locality for interior scutes are thick and large and bear the G. petaliferus and hastily erected G. rivipacis for the typical G. floridanum sculpturing. Peace Creek glyptodont. Simpson (1929b) re­ Two unreported glyptodont scutes from the jected Hay's species as a nomen nudem and went Waccasassa River, Levy County, Florida (UF/ on to describe the Seminole Field specimens as FSM 16379, 18455) are referred to G. floridanum the types for his new species, which, by implica­ on the basis of their sculpturing. tion, was to include the Peace Creek, Florida, The only eastern record of G. floridanum outside representative of G. petaliferus. The Peace Creek of Florida is from the Edisto Beach locality, South representative appears indeed to belong in Glyp­ Carolina. One heavily abraded scute (ChM 2417) totherium floridanum, the descriptions by Leidy associated with skull ChM 2415 (CM 43.59 as (1889b) and Hay (1924) indicating a close resem­ reported by Ray, 1965) was recovered from this blance to the carapacial materials described locality, and two other scutes (ChM 2418, 2090) above. (The scutes have not been examined for have been recovered from the nearby Edingsville the present study.) Beach. All three scutes exhibit typical G. flori­ Simpson (1929a) included Boreostracon floridanus danum sculpturing. The Edisto Beach scute is in the faunal list for the fossils recovered from the probably female, while the Edingsville scutes are mouth of Hog Creek at Sarasota, Sarasota larger and are probably from a male individual. County, Florida. In his description of the species, As in Florida, Late Pleistocene glyptodonts in 192 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Texas are known from a variety of localities, of peripheral grooves, located at the sutural contacts which several are recorded here for the first time. between scutes, are not as prominent in the Texas Lundelius (1972) has presented the most recent females, perhaps indicating either ontogenetic review of the Texas representatives. Primarily on variation or less dimorphism in the Texas popu­ the basis of carapacial features, he proposed syn­ lation. onymy of all the published Late Pleistocene rec­ Cope's (1888) type specimen of Glyptodon petal­ ords in Texas with Simpson's Florida species. The iferus from Nueces County, Texas (AMNH 14158, taxonomy proposed here is in essential agreement Figure 91c), is referred to G. floridanum. It consists with Lundelius that the Texas and Florida rep­ of little more than half of an interior scute, with resentatives belong in the same species, referred a relatively small central figure and four pre­ to Glyptotherium floridanum. served peripheral figures. There probably were The carapacial features in the Texas glypto­ eight peripheral figures in this scute before it was donts correspond closely with those from Florida. broken. Despite Cope's description, the undersur- There seems to be a separation into two groups, larger males and smaller females, although the differences are not as pronounced as in the Flor­ ida population. Unfortunately there are no sam­ ples large enough to assume representation of more than one or two individuals at the same locality, so that it is difficult to assert with confi­ dence that the postulated dimorphism is not in­ stead attributable to separate species. By analogy with the Florida population, particularly the rep­ resentation from Seminole Field, the Texas pop­ ulation of G. floridanum, which is identical in carapacial features (but not totally identical in noncarapacial osteology), is here assumed to ex­ hibit sexual dimorphism. The argument for this assumption is the same as that forwarded for the Florida population. Additional support for this hypothesis is the low probability of sympatry over such a great distance. Furthermore, if indeed the Texas and Florida populations prove to belong to different species (closely related, distinguishable only by their dentition), and if the dimorphism ¥$&&-' •;•' - •*>'• were taken to represent two species in each state, then there would be four species in the Gulf Coastal Plain, a geographic area for which there is little evidence of barriers promoting isolation. Hence, as in the Florida population, male-female distinction seems more reasonable, and accounts for the general similarity of noncarapacial re­ FIGURE 91.—Scutes from carapace oi Glyptotherium floridanum mains (unexpected if two species are represented). from Texas: a, carapace fragment from posterior region along caudal aperlure (TMM 30967-2088); b, border scute Compared with scutes from the Florida G, from caudal aperture (USNM 6071); c, interior carapacial floridanum, the Texas female scutes are not as scute fragment, holotype Glyptodon petaliferus (= Glyptotherium easily identified, except on the basis of size. The floridanum), AMNH 14158. (Bars = 5 cm.) NUMBER 40 193 face is not preserved, and its thickness therefore and that the Sinton and Wolfe City specimens cannot be measured. Its side-to-side diameter was likely belong in the same species. Hay mentioned at least 45 mm. Its size corresponds to the males two carapace fragments that Dr. Francis had represented in the Seminole Field population. donated to the United States National Museum, The central figure is raised slightly above the stating that they bear the catalog number USNM level of the peripheral figures, and there is a 11379. One of these, a single isolated scute, has suggestion of a weak central concavity. Three been located. Another fragment, consisting of hair follicles are preserved, and they are relatively approximately six fused scutes from the ischiac or small. Surface texture is pitted and punctate. iliac region, bears the number USNM 11378 and The carapace of the Wolfe City, Hunt County, is accompanied by a label that reads "gift of Dr. Texas, glyptodont (USNM 6071), which Hay Mark Francis. . .Aransas River." This latter spec­ (1916) referred to Cope's species, is represented imen is undoubtedly the second of the two that by approximately 80 isolated scutes, of which one Hay had stated bore the same catalog number. is from the posterior border, two are from the first Hay was correct in noting a difference in the interior row along the posterior aperture, and sculpturing from that of the Wolfe City speci­ several are from the anterior notch and caudal mens. The central figures are relatively large, and armor. Because there is no duplication of parts in they are depressed, similar to the depressions the noncarapacial remains it is safe to assume noted in other species. There are no peripheral that these represent only one individual. As in­ grooves at the sutural contacts in the two speci­ dicated by their size, these scutes belong to a male mens at hand, and the scutes are large. Sculptur­ individual. Measurements for the only posterior ing most closely resembles that of other G. flori­ border scute (Figure 9lb) are 57 mm transverse danum specimens, and because of the Late Pleis­ diameter, 52 mm exterior-interior diameter, 45 tocene age indicated by the associated fauna, mm maximum thickness, and 31 mm thickness at these are referred to G. floridanum. the interior suture. These dimensions are nearly Not far from the Sinton locality in the same identical with those given for the corresponding county is the Ingleside occurrence of G. floridanum, male scute from Seminole Field. Interior scutes which Lundelius (1972) identified as conspecific compare in exactly the same fashion. Sculpturing with the Wolfe City and Florida Late Pleistocene is identical to that in the Florida males of G. representatives. Lundelius accurately described floridanum, i.e., lacking peripheral sutural grooves the portion of the carapace that is mounted as a and otherwise typical for the species. display specimen (TMM 977-3). Apparently the Hay (1926) described carapace fragments from only part of this reconstruction that is composed the private collection of one Dr. Mark Francis of of actual bone is the posterior region of the cara­ College Station, Texas. These remains were as­ pace and the first six tail rings, as Lundelius sociated with fragmentary noncarapacial ele­ maintained. Notable features on this carapace ments including five mandible fragments and a include a healed injury over the left iliac region, complete femur. They reportedly came from the where the scutes are depressed and solidly fused; Aransas River, near Sinton, Texas, not far from the weak recurved outline of the posterior border the Nueces County record. The whereabouts of similar to that in the type specimen of G. flori­ these fossils is unknown. Hay detailed the differ­ danum; and the relatively large size, indicating a ences between the carapace fragments and teeth male individual according to the present propo­ with corresponding elements of the Wolfe City sitions. specimen, which he had described earlier, and A carapace fragment from the same locality, noted several distinctions in the anterior teeth consisting of three scutes from the posterior bor­ and in the sculpturing of the scutes. He concluded der and approximately 50 interior scutes (TMM that the differences probably represent variation, 30967-2088, Figure 91a), is from a second individ- 194 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY ual, for there is duplication in the regions pre­ Unreported recoveries of scutes referable to G. served between these two specimens. These ap­ floridanum in Texas include a collection of 80 parently represent a female, as indicated by their scutes from Laubach Cave, Williamson County size: posterior border scutes average 46 mm trans­ (TMM 41343), associated with a Late Pleistocene verse diameter, 43 mm external-internal diame­ fauna; approximately 25 isolated scutes (TMM ter, and 21 mm thickness at interior suture. These 41075-14) from the "Eremothenum locality," Port dimensions are nearly identical to those of the Lavaca, Calhoun County, Texas; and three small (female) type specimen of G. floridanum. These scutes from the Runnels Pierce Ranch, near scutes differ in possessing only blunt projections, Whorton, Matagordo County (TMM 41530-6). similar to those of males in the Florida popula­ With the exception of the Laubach Cave recov­ tion, and in mostly lacking development of pe­ ery, these are coastal records. The scutes from ripheral grooves at the sutural boundaries, al­ Laubach Cave are referable to G. floridanum on though there is the suggestion of this construction the basis of sculpturing. The locality is interme­ on the interior scutes. The age assignment of Late diate between the coastal records and the Wolfe Pleistocene as indicated by the associated fauna City, Hunt County site, and therefore represents further supports the G. floridanum identification. the third inland recovery of G. floridanum. The scutes Sellards (1940) mentioned from Bee County, Texas, not far from the San Patricio CAUDAL ARMOR recoveries, have not been examined for the pres­ ent study. The scutes were associated with a Osborn (1903) accurately described and fig­ fragmentary mandible (TMM 31034-30) that re­ ured the caudal armor for the type specimen of sembles in all respects, as far as can be deter­ G. texanum. The caudal armor of the Arizona mined, mandible TMM 30967-1814 from Ingle­ specimen is in most respects identical. However, side. This comparison, along with the Late Pleis­ there was also recovered the anterior "accessory tocene age of the fauna, indicates an assignment ring," here considered as the first caudal ring. as G. floridanum for the scutes and the mandible Although the accessory ring was not preserved from Bee County. with the type specimen, it is likely that there was The glyptodont remains that Hay (1924) re­ one, for these two specimens are otherwise iden­ ported from Jones County, Texas, include ap­ tical. The likelihood of an incomplete anterior proximately 15 isolated scutes (USNM 8644). ring, or accessory ring, can also be argued as a Hay identified these as Glyptodon petaliferus and necessary consequence of the peculiar construc­ Lundelius (1972) was unable to make a taxo­ tion and support provided by the vertebrae, in nomic assignment. Although heavily weathered, which dorsally the rings are supported by the they seem to resemble closely scutes of G. flori­ xenarthral processes of one vertebra, laterally by danum males in possessing relatively flat and punc­ the transverse processes, and ventrally by the tate surfaces for the central figures in addition to chevron articulating principally with the next their large size. These scutes are here referred to anterior vertebra. Thus the first vertebra to pos­ G. floridanum on the basis of the sculpturing and sess a chevron necessarily lacked ventral support the proximity of the locality to other recoveries of for a ring because its chevron supported the next this species. These scutes represent the second posterior ring; hence the necessity for an anterior inland recovery of G. floridanum in Texas. accessory ring as the first (but incomplete) caudal Lundelius (1972) reported a collection of scutes ring. An incomplete anterior ring is also known from Cameron County in southernmost Texas for G. arizonae. It is probably characteristic of the near the mouth of the Rio Grande. On the basis genus. of his determination these are referred to G. flori­ The accessory ring, or caudal ring 1 in G. danum. texanum (F:AM 95737, Figure 92), is a short dou- NUMBER 40 195

processes of the vertebra, and presumably at least in the subadult condition its connection was car­ tilaginous or ligamentous. This ring lay well within the caudal aperture. Ring 2 in F:AM 95737, the first complete caudal ring, is also the largest, measuring approx­ imately 210 mm transverse diameter in both the type specimen and the Arizona specimen. It is formed by two complete rows of 24 firmly sutured scutes. The distal scutes are larger, and their free margins are rounded, producing a scalloped out­ line. The scutes are flat externally, with indistinct sculpturing indicating a single large scale cover­ ing each scute. The textures are pitted and punc­ tate externally. Several large vascular pores mark the otherwise smooth internal surfaces. This ring, like the first, lay within the caudal aperture, as indicated by the pelvis-carapace geometry and by its lack of external ornamentation. Measure­ ments for this and the other rings are provided in Table 70. FIGURE 92.—Caudal armor of Glyptotherium texanum: a, iso­ lated caudal rings 2-11 of type specimen AMNH 10704 (2- Ring 3 (second complete ring) is similar to ring 7, dorsal aspect; 8-9, right side; 10/11, left side); b, articu­ 2 except that it is smaller (approximately 180 mm lated caudal armor including incomplete first ring of Arizona diameter) and some of the scutes of the distal row representative F:AM 95737, dorsal aspect. (Bar = 10 cm.) reach the proximal margin, crowding the proxi­ mal scutes and making the proximal row incom­ plete despite the same number of 20 scutes in ble row of small scutes. The proximal row in­ both rows. cludes seven small five-sided scutes and a small Rings 4-10 decrease in diameter and increase fragmentary one near the left extremity; the distal in maximum length rearward. The rings are all row includes nine large five-sided scutes, includ­ complete double rows, and the number of scutes ing the scutes forming the extremities. Scutes of in each row decreases from 19 in ring 4 to 10 in the proximal row are indistinctly sculptured with the penultimate ring (10). The scutes of the distal raised and convex upper surfaces. Each of the rows increase in absolute and relative size rear­ distal scutes bears a pair of large hair follicles, ward, while there is little change in the absolute and the proximal scutes bear one at their interior sizes of the scutes in the proximal rows. The apex near the contact with the distal scutes and dorsal pair of distal scutes in ring 4 are weakly one or two hair follicles at the sutural contacts conical, and the two adjacent are faintly conical. with adjacent scutes of the distal row. The un- The corresponding scutes in rings 6-8 are increas­ dersurfaces are weakly concave. The overall shape ingly more strongly conical, but the remaining and size compare with a flattened half section of outline of the free margin is scalloped as in the a banana peel. Its transverse length is 190 mm anterior rings. The dorsal pair of scutes of the and its central width (anteroposterior diameter) distal row in ring 9 are greatly enlarged and is 42 mm. In F:AM 95737 this accessory ring lay markedly conical. They are expanded rearward above caudal vertebra 2. There is no indication and fit snugly into a corresponding notch of the on its undersurface of contact with the support terminal tube. In the terminal tube three succes- 196 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY sive pairs of dorsal conical scutes complete the there may be considered nine complete rings plus ornamentation. the terminal tube. The accessory ring is only The penultimate ring and the terminal tube fragmentary and not well enough preserved for are basically similar in the Texas and Arizona restoration. Isolated scutes from the accessory ring representative specimens, but they differ in detail. are identical to those of AMNH 21808, from the Ring 9 in the type specimen is composed of only same locality. In the latter specimen the caudal one complete row of enlarged scutes, representing ring count is the same as for the type specimen: the distal row, and only two small scutes from the an anterior accessory ring, eight complete rings, proximal row. In F:AM 95737, there are instead and a terminal tube formed by the fusion of the two complete rows of scutes in the penultimate penultimate ring with the actual tube. Hence for ring, with full representation of the proximal row. both specimens there are 11 rings, of which the However, the scutes of the first row of the termi­ first is incomplete and the last two are conjoined nal cone in the Arizona specimen correspond to form the terminal base. exactly in their construction to those of the pe­ In UMMP 34826 there are 12 caudal rings, nultimate ring in the type specimen. Hence the including the anterior accessory ring and the long terminal cone in F:AM 95737 in actuality terminal tube. The penultimate ring is not fused appears to be the fused terminal cone and true to the terminal tube. The difference in total penultimate ring. That they are truly fused is counts between these three specimens is a conse­ indicated by fusion of the last three caudal ver­ quence of the greater number of caudal vertebrae tebrae. in the Seymour representative and is not neces­ The terminal cones are therefore different in sarily indicative of species characterization. The these two specimens, apparently because of the construction of the caudal rings is the outstanding lengthening of the tube by fusion in the Arizona feature emphasizing the similarity between the representative. In F:AM 95737, the terminus of Texas and Arizona representatives, the total cau­ the tube is formed by two incomplete and two dal ring counts notwithstanding. As indicated in complete alternating rows of scutes. In the Texas these three specimens, the scutes of the distal rows specimen the entire tube is formed by two com­ of all the rings except the first three are extremely plete rings and an incomplete ring at the tip, with tuberculate, in contrast to the smooth surfaces of two small accessory scutes on the proximolateral the scutes of the distal row in G. texanum (except­ angles. In both, the terminus is open, and it ing the pair of conical dorsal scutes). appears to have lacked a terminal scute. (Holmes The accessory ring, or ring 1, is formed by a and Simpson, 1931, reached a similar conclusion double row of smooth, quadrilateral scutes in the for G. floridanum.) Seymour specimen and by a single row, with a In G. texanum the caudal armor therefore ex­ few intercalated proximal scutes, in AMNH hibits minor variability, especially in the con­ 21808. It is semicircular, and laterally more ex­ struction of the terminal tube. It is tempting to tensive than in G. texanum. explain the lengthening of the tube in the Arizona Ring 2, the first complete ring, is composed of representatives as an evolutionary trend related a double row of smooth, flat scutes, with a scal­ to decreasing number of caudal vertebrae, and as loped outline for the free margin, similar except intermediate between the type specimen and G. for larger size to the corresponding ring of G. arizonae. texanum. Caudal ring no. 3 is similar to the one The caudal armor of G. arizonae (Figure 93) is preceding, but the scutes of the distal row are modified from that of G. texanum and is diagnostic variously raised and convex, some suggestive of for the species. In USNM 10537 there are eight the conical bosses characteristic of remaining complete rings and a terminal tube consisting of rings. As Melton (1964) suggested, the first three the actual tube and the penultimate ring. Thus rings evidently lay within the caudal aperture, as NUMBER 40 197

FIGURE 93.—Caudal armor of Glyptotherium arizonae from the Curtis Ranch, type-locality: a, USNM 10537, right side; b, AMNH 21808, right side. (Bar = 10 cm.)

indicated by the smooth surfaces of the scutes. 21808, a condition evidently attributable to in- This conclusion is also supported by the position­ traspecific variation. Scutes of the proximal rows ing of the pelvis. possess flat, oval figures covering approximately Caudal rings 4 to 10 are biseriate in the Sey­ the distal half of the external surface of each mour representative and in USNM 10537. Rings scute. These scutes are flat and unbossed. The 4 to 7 are incompletely triseriate in AMNH scutes of the distal rows are almost uniformly 198 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY tuberculated, each one possessing a pronounced similarity in the caudal armor of these two spe­ conical boss situated nearer the free margin than cies. The only differences, those of size and the the sutural border. The dorsal pair are distinctly development of tubercles, are indicative of species more conical in each ring, and the lateral and distinction, but further taxonomic separation of ventral ones are less well developed although the these two species on the basis of the caudal armor tubercles are distinctive. The free margins of the is unjustified. Indeed the basic similarities may distal rows are scalloped, as in G. texanum. be used as important counterargument consider­ The penultimate ring in all three specimens ations when evaluating other differences that (fused to the terminal tube in the Curtis Ranch alone are suggestive of greater than species sepa­ representatives) is biseriate, and each row is com­ ration. posed of six scutes. The dorsal pair in the distal Included in the collection of remains of the row are enlarged and extremely pointed. They type specimen of G. cylindricum (AMNH 15548), project rearward beyond the free margin of the but unmentioned and undescribed by Brown remainder of the distal row, fitting snugly into a (1912), is a series of 24 scutes, of which two groups corresponding notch of the terminal tube. The of three and seven are articulated. Of the remain­ lateral and ventral scutes of the distal row are ing 14 isolated scutes, at least eight can be artic­ indistinctly conical relative to corresponding ulated with confidence. The remaining scutes are scutes in the more proximal rings, and also rela­ of similar shape and size and apparently belong tive to the distal pair on the same ring, emphasiz­ with the articulated scutes. All belong to the same ing the marked asymmetry of the penultimate caudal ring, by virtue of the uniform curvature ring. of their undersurfaces and by the uniformity of The terminal tube is formed by three rows of the distal scutes, all of which lack bosses and are scutes with several accessory scutes. The dorsal smooth and flat, as in caudal ring 1 and 2 of pairs of scutes of the last two rings are conical, other North American glyptodonts. The construc­ thus maintaining the pattern of the enlarged tion of these scutes is no different from that of the conical pair of each distal ring to the terminus. A more completely known scutes of the first and small scute covers the tip. At its anterior margin second caudal rings of G. arizonae, with which there is a dorsal notch for the fit of the penulti­ they closely compare. mate ring. Holmes and Simpson (1931) mentioned the As Melton (1964:143) astutely observed, "in recovery of a few caudal scutes in the Seminole bony rings three to ten, the two ventral scutes are Field glyptodonts. These are the only caudal flat and smooth, and they join so that there is a scutes associated with the type specimens of G. noticeable V notch toward the rear where the floridanum; they afford no positive distinction from chevron bone in the next anterior vertebrae [sic] the corresponding scutes of G. anzonae, i.e., they articulated." This construction is not evident in are simple, with minimal sculpturing, and the G. texanum. scutes of the posterior row bear large conical The principal distinction in the caudal armor bosses. between G. texanum and G. arizonae, therefore, The same authors described a fragmentary cau­ appears to be the development of pronounced dal tube from the Sarasota locality and in the tubercles on the scutes of the distal row in each of possession of a private collector. Of these two the rings posterior to ring 3. The pair of enlarged caudal scutes, they stated that they came from dorsal tubercles on each distal row, the scalloped free outline, the nature of the accessory ring, the the terminal part of the tail, which apparently was tube-like general lack of fusion of adjacent rings, the simi­ and nol divided into definite movable rings. The beak-like larity in the terminal tube, and the nature of the posterior projection, prominent on the anterior rings, has flattened out. The more anterior of these two scutes forms a vertebral-chevron articulation all point to a basic 40 arc of a circle, indicating a ring or row of nine scutes, NUMBER 40 199 while the more posterior forms a 60° arc, indicating a row of six scutes. The two rows represented were apparently contig­ uous and the more anterior overlapped the other. The posterior end of the more distal row shows no sign of contact with other scutes, and nothing indicates the presence of bone beyond this point,,which would leave a very small opening at the end of the tail tube, possibly capped by a scale (Holmes and Simpson, 1931:410). These two scutes (not located for the present study) and the isolated fragmentary scutes are the only caudal scutes known from the Florida pop­ ulation of G. floridanum. The description of the two caudal scutes might apply to the ventral scutes of the terminal tube in G. arizonae and does not indicate primary distinction. (These are the only terminal tube remains in all of the G. flori­ danum population, Texas included.) Lundelius (1972) figured and described the six anterior caudal rings of G. floridanum (Figure 94) from the Ingleside locality, Texas. These probably correspond to caudal rings 2 to 7, as identified here, presuming that ring 1, the accessory ring, is missing and that the first ring present is ring 2. Measurements (Table 70) indicate that these caudal rings are slightly smaller than those of G. arizonae; this is the most notable distinction. The first ring (ring 2) is biseriate and the scutes are FIGURE 94.—Caudal armor of Glyptotherium floridanum from fiat. Ring 3 is also biseriate, with unbossed but the Ingleside locality, San Patricio County, Texas (TMM weakly pointed posterior scutes. Rings 4 to 7 are 977-3), dorsal aspect; first 6 rings are actual fossils, remainder biseriate, with an incomplete third ring formed are reconstructed. (Photo courtesy Texas Memorial Mu­ by the intercalation of proximal scutes. Scutes of seum; bar = 10 cm.) the distal rows in these rings are strongly bossed, conical, and similar to those in G. arizonae. The Selected Aspects of Glyptodont Paleobiology only distinction other than size (which is minor) is a qualitative one: the scutes of the distal rows Beyond occasional comments to the effect that are not as exaggerated in their conical develop­ glyptodonts were herbivorous, clumsy, and cum­ ment as in G. arizonae. This difference is seemingly bersome, few modern authors have considered trivial and is useless taxonomically. As presented them as living animals. In an obscure paper that in the discussion of the caudal vertebrae, G. flori­ has gone virtually unnoticed since its publication danum appears to have possessed individual rings more than a century ago, Senechal (1865) for all but the last vertebra, as in other species of breathed life into these altogether unforgettable Glyptotherium, rather than having a terminal tube creatures. of coalesced rings. Senechal described the biology of Glyptodon, as Therefore, on the basis of these inferences and he perceived it in life. He compared the size of incomplete data, the caudal armor of G. floridanum Glyptodon to that of a small hippopotamus, but its appears not to differ significantly from that of G. morphology to that of a giant tortoise. He also anzonae. proposed that Glyptodon possessed a mobile snout, 200 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY a conclusion that we have independently reached soft ground for food (primarily insects) is well for Glyptotherium. Senechal was also impressed by established (Talmage and Buchanan, 1954). Ar­ the cumbersome carapace and its consequent ef­ madillo dentition and mandibular construction fects on respiration, locomotion, feeding, overall are further reflective of their feeding habits. Al­ agility, and protection. He pictured Glyptodon as though glyptodonts also possess an unusually long restricted to flat terrain, an inoffensive quadru­ olecranon process, other regions of their anatomy pedal fortress, "approximating the most extreme are decidedly uncharacteristic of an entomopha- range of stupidity. One can wonder if a better gous diet. Moreover, glyptodonts are so much intellectual apparatus would not have been a larger than armadillos that digging for food seems superfluous luxury" (Senechal, 1865:23, loose improbable. Therefore, the apparent similarities translation). Among other suggestions in this between glyptodont and armadillo morphology charming discussion, Senechal proposed that relating to feeding habits, basically restricted to Glyptodon was adept at digging and was a seed- the construction of the olecranon process, are here eater. We have concluded that Glyptotherium was interpreted as representing a parallelism arising not restricted to granivory but instead ate vege­ from their common ancestry, not as indicating tation in marshy lowlands and near permanent similar feeding habits. bodies of water. In particular, the construction of glyptodont In the following discussions we hope to stimu­ teeth precludes any dietary habit other than her- late interest in glyptodonts and perhaps to invite bivory. The full dental battery, consisting of 32 debate on edentates in general. We give brief molariform teeth, presents a grinding mill similar consideration in the following pages to several to that found in the cheek teeth of microtine aspects of the functional morphology of rodents. The flat-crowned, hypsodont teeth ex­ Glyptotherium, the inferred habitat requirements, hibit an elaboration of the osteodentine core, and the evolutionary history of the genus. which compensates for the lack of enamel, as discussed in the osteology section. The teeth maintained unclosed pulp cavities throughout FUNCTIONAL MORPHOLOGY life, in this respect surpassing the classic hypso- Glyptodont anatomy is so peculiar as to invite donty so often discussed for horses. inquiry from many viewpoints, among which The maxilla and mandible are massive and masticatory function, locomotor function, and deep-bodied to accommodate the long hypsodont defense and protection promise to provide inter­ teeth. The mandibular symphysis is solidly an­ esting insights into glyptodont paleobiology. kylosed in a spoutlike construction, indicating MASTICATORY APPARATUS.—Clues to the feed­ that the masticatory motion of the mandible was ing habits of glyptodonts are found in their den­ unified and inflexible at the symphysis. The prin­ tition and in the construction of their mandible, cipal masticatory motion was a forward grinding portions of the skull, and axial skeleton. It is movement of the mandible, the upper and lower tempting on first examination to suppose that tooth rows maintaining surface-to-surface contact glyptodonts fed much as do the modern armadil­ during the grinding motion. That this motion los. Such a comparison is unwarranted, however, represents the primary movement is indicated by as indicated by the striking dissimilarities in the the wear facets of the occlusal surfaces of the skeletal regions pertaining to feeding. The closest teeth, by the construction of the mandible, and anatomical correspondence related to feeding by the horizontal disposition of the masseteric habits between armadillos and glyptodonts is the musculature. construction of the front limb. The long olecranon A detailed inspection of the attrition facets of process of the ulna of both groups is indicative of glyptodont teeth reveals convincing evidence of a strong digging ability. That armadillos dig in this proposed masticatory motion. The osteoden- NUMBER 40 201

tine rim and core stand up in subdued, but nevertheless distinct, relief above the interior den- tinal tracts (Figure 95). The dentine tapers grad­ ually downward and rearward in the lower teeth, reaching maximum depth immediately anterior to the transversely disposed osteodentine ridges. Attrition facets of upper teeth exhibit an opposite arrangement for the taper of the dentine tracts, which reach maximum depth immediately pos­ terior to the osteodentine ridges. This construc­ tion indicates that food particles were pushed toward the deep end of the dentine taper; as the forward motion of the mandibular teeth passed over the maxillary teeth, food particles were pushed rearward relative to the lower teeth and forward relative to the upper teeth, becoming lodged against the osteodentine ridges. Further motion crushed or sheared the food as each osteo- FICURE 96.—Schematic representation of glyptodont masti­ catory apparatus (the pterygoideus musculature is situated on interior side of ascending ramus and is indicated by diagonal, broken lines).

dentine ridge passed against an opposing one. This horizontal motion was effected by the horizontally disposed masseteric musculature (Figure 96). Sicher (1944) established that the descending process of the zygomatic arch is a modification permitting a principal horizontal component in the masticatory apparatus in tree sloths. By lowering the area of insertion for the masseteric musculature to the level of the tooth row, and below it, the descending zygomatic process provided a key innovation in the masti­ catory motion in several edentate lineages, in­ cluding tree sloths, ground sloths, to a lesser degree some armadillos, and to an exaggerated degree, the glyptodonts. It is interesting in this context to apply Turn- bull's (1970) classification of mammalian masti­ catory construction to glyptodonts. Glyptodonts belong in Turnbull's "Specialized Group III"; the FIGURE 95.—Schematic perspective view of occlusal surface "rodent-gnawing" or "anterior shift" type, rep­ of an idealized lower tooth of Glyptotherium (above) and a resented principally by rodents and lagomorphs. parasagittal section through same tooth (below) to show nature of attrition. (Occlusal motion relative to upper tooth This group is characterized by an overwhelming row indicated by arrows.) dominance of the masseter group, at the expense 202 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY of the temporalis and pterygoideus groups. proboscis-like, as the primary food procurement Capybaras exhibit the most extreme development device. Indeed, it appears that glyptodonts in of the masseter complex, which includes approx­ standing position (i.e., feeding position) were in­ imately 77 percent of the masticatory muscle capable of reaching the ground with the muzzle, mass, while the remaining 23 percent is unequally largely because of the extreme fusion and shorten­ divided between the pterygoideus group (approx­ ing of the cervical and thoracic vertebrae. imately 16 percent) and the temporalis group A final consideration concerning the mastica­ (approximately 7 percent) (data from Turnbull, tory apparatus is the function of the edentulous 1970:277). The resemblance between capybara symphyseal region of the mandible with respect dentition and the glyptodont dentition has al­ to the opposing premaxillary region. It is unlikely ready been mentioned; it is apparent that the that both regions were occupied by a toughened muscle mass proportions were also similar. The pad thick enough and large enough to occlude temporalis and pterygoideus muscles in glypto­ with the one opposing. In Glyptotherium floridanum donts were probably even more reduced, and the the symphyseal region is even anteriorly down- masseter group may have approached 90 percent turned, a construction unexpected if the opposing in relative mass. This extreme modification may symphyseal and premaxillary regions were occu­ be viewed as a consequence of the lack of anterior pied by pads. Moreover, the relatively weak ac­ teeth in glyptodonts, eliminating the necessity for cessory masticatory musculature and the reduced strong leverage in food procurement, a function (or absent) premaxillaries indicate that little food partially performed by the pterygoideus and tem­ could have been obtained by simple pad-on-pad poralis complexes. closure (if the existence of opposing pads is rea­ An additional consequence of the elaboration sonable, but which we believe is not). Such a of the masseteric complex, with its origin surface motion would necessarily involve unusually largely confined to the descending zygomatic strong closure and would necessitate a tearing process, is that the anterior facial surface was not function, rather than a shearing or cutting func­ occupied by the masseteric origin. That is, since tion characteristic of most herbivores. the masseter attached to the descending process, In summary, therefore, the glyptodont masti­ the pre- and suprazygomatic regions of the supe­ catory apparatus represents a somewhat aberrant rior maxillary and frontal bones, respectively, example belonging in Turnbull's (1970) Group were free of masseteric attachment. There is suf­ III; the aberration is the lack of anterior teeth. ficient surface area in these skull regions left The construction of the teeth, mandible, and vacant by the rearrangement of the masseteric relevant portions of the skull certainly indicate a complex to accommodate an expansion of the totally herbivorous diet. Because extensive cervi­ snout musculature. cal motion and the advantage of elaborate Thus, the lack of anterior teeth for participa­ enamel infolding in the teeth are seemingly req­ tion in food procurement and the arrangement of uisite for feeding on grasses, it is apparent that the masticatory musculature are here interpreted glyptodonts were not grazers, for they lacked as indicating the existence in glyptodonts of a these characteristics altogether. Instead they were well-developed snout musculature important in probably browsers, feeding selectively on vegeta­ feeding. Extensive fusion of the cervical and tho­ tion along water courses or perhaps in shallow racic vertebrae indicates a general lack of axial water, like capybaras. This postulated dietary mobility, precluding extensive axial motion in preference is supported by faunal associations and feeding. Without benefit of a highly mobile neck scdimentological characteristics of glyptodont oc­ region allowing reaching and bending during currences, as discussed elsewhere. feeding, it is again plausible to assume an elabo­ LOCOMOTION.—Several locomotor aspects of rate snout musculature, either "prehensile" or the skeleton deserve special consideration, for in NUMBER 40 203

many respects glyptodonts exhibit unique axial cated by maintenance of the thoracolumbar joint and appendicular osteology. A critical conse­ and by the free (unfused) scapular border. quence of an immobile carapace is the almost The large and flexible glyptodont tail may be total exclusion of the precaudal axial skeleton in considered as an integral part of the locomotor locomotor function. Glyptodonts represent the complex insofar as it probably functioned as a mammalian counterpart to turtles in this respect, counterbalance during movement. That is, the although the fusion is neither as extensive nor as heavy tail, which seems otherwise to be without immobile as in turtles. Nevertheless there is a significant function, was probably swung from close parallel between turtles and glyptodonts in side to side during movement to keep the center the functional participation of the axial skeleton. of balance as close to the midline as possible. It is A rigid carapace precludes extensive trunk mo­ likely, therefore, that the tail offset the locomotor bility. Hence, during locomotion there was no disadvantage of limited axial mobility, which axial participation beyond maintenance of a usually allows lateral flexure and torsion. Fur­ static and rigid support posture of the vertebrae. thermore, it may be postulated that the terminal Whereas fusion of the vertebrae to the carapace "mace" in the South American genus Doedicurus is an extreme condition exhibited by turtles, glyp­ originated because of this side-to-side motion of todont vertebrae are largely free of fusion to the the tail and provided additional protection from carapace. The trunk vertebrae, however, are ex­ rear-approaching predators. The North Ameri­ tensively fused to each other, with only a single can genus Glyptotherium lacked the terminal mace, weak articulation occurring at the thoracolumbar but this distinction has no significance to the joint. Cervical vertebrae exhibit various degrees locomotor function of the tail. of fusion in North American glyptodonts, and The limb proportions and their construction they have little bearing on locomotor function. lend considerable support to the intuitive notion The thoracic vertebrae are fused into two sections, that glyptodonts were clumsy, slow-moving ani­ the so-called "trivertebral element" and the long mals. As a guide to the significance of the various thoracic tube. The anteriorly situated trivertebral aspects of limb anatomy discussed below, How- element, like the cervical vertebrae, is unimpor­ ell's (1944) comprehensive treatment is indispens­ tant in locomotor function, its principal signifi­ able; the following comments are founded on the cance probably relating to cranial support. The tendencies Howell observed in his excellent cov­ long thoracic tube articulates with the anterior erage. The characteristics outlined below are extremity of the lumbar tube in a weak contact, mentioned only for the purpose of establishing and this is the only midtrunk axial articulation. the basic locomotor plan of the glyptodont ap­ The lumbar vertebrae are serially ankylosed, and pendicular skeleton, and the treatment is in no the entire lumbar tube is fused to the anterior way exhaustive. sacral vertebra. Thus the lumbar vertebrae are Both fore and hind limbs are heavy and mas­ incorporated into the rigid framework of the pel­ sive. The front limb is, as expected, somewhat vic skeleton, and the entire postthoracic and pre­ more mobile and slightly lighter in build in com­ caudal axial skeleton is a boxlike, immobile com­ parison with the hind limb. Both exhibit extreme plex of static support structures for the heavy modifications for weight support, with pro­ carapace and the massive soft-part anatomy. This nounced graviportal tendencies. construction is afforded even more rigidity pos­ The broad vertebral border of the scapula, teriorly by the solid fusion of the iliac and ischiac apparently representing an extreme development crests to the undersurface of the carapace, further among terrestrial mammals, indicates strong precluding axial mobility in the posterior region standing musculature, especially "the serratus an­ of the trunk. That there was limited anterior terior, the chief muscle supporting the anterior mobility of the axial skeleton, however, is indi­ half of the body when a quadruped is standing. 204 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

A heavy body needs a strong one; and the verte­ The patella is large and stout, and it forms a bral border of the scapula is very broad in distinctive "patellar lock" in standing posture. graviportal animals, while it is narrow in the This device allows the femur and tibiofibula to lighter, more agile ungulates" (Howell, 1944: maintain a vertical, straight-line orientation in a 142). The spherical head of the humerus and its standing, resting position. The tibia and fibula well-developed tuberosities are further indicative are equally developed. Both are large and mas­ of a graviportal construction for the upper portion sive, and they are proximally and distally fused of the front limb, while the nearly equal sharing to form the cylindrical tibiofibula. It appears that between the ulna and radius in the humeral glyptodonts are a singular exception to the gen­ articulation illustrates the strong support at the eralization that in placental mammals the fibula elbow joint. In addition, the distal articulation of is excluded from femoral articulation, for the the epipodials (ulna and radius) is also almost articular surface of the tibiofibula is expanded far equally divided, another indication of graviportal beyond the tibia to a position well above the construction. fibular shaft. This construction is more fully dis­ The elements of the manus are short, broad, cussed by Howell (1944), who also mentions mon- and stout; the principal articular facets are more otremes and some marsupials as exceptions. It flattened than rounded. The only indication of therefore seems likely that the fibula actually cursorial adaptation in the appendicular skeleton participates in this articulation, and, hence, glyp­ is the loss of digit I in the manus. Digits II and todonts as placental mammals represent an ex­ III are the largest of the manus, digit IV is ception to the usual condition. This modification somewhat smaller, and digit V is reduced and is certainly important in weight support and in functionally unimportant. The terminal phalan­ providing lateral flexibility, an important matter ges of the front foot were each encased by an for a clumsy, ponderous animal. Furthermore, ungual sheath more hooflike than either clawlike the fibular participation in the tarsal articulation or nail-like. The huge palmar sesamoid bone and is also unusual. Among mammals only elephants the full battery of digital bones at each of the and artiodactyls exhibit this arrangement; the joints in the manus indicate strong weight support condition in glyptodonts resembles that of the and an unusually large mechanical advantage former group for weight support. during articular motion. The elements of the tarsus are much more The hind limb exhibits an even more striking extensively modified for weight support than the exaggeration of graviportal features. Whereas the corresponding bones of the front limb. The distal pectoral appendage retains at least minimal cur­ tarsals, in fact, possess almost no curvature in sorial adaptation in the articular facets and re­ their articular surfaces. The flat articulations duction of lateral digits, the pelvic appendage has serve only to transfer weight and provide no lost all cursorial advantage. The ilium is vertical mobility to the distal region of the tarsus. Proxi­ and fused to the carapace. The ischium likewise mal tarsal elements, the calcaneum and astraga­ bears a distinct vertical component in its orienta­ lus, are stout and massive, with no indication of tion, and both of these pelvic bones serve to cursorial advantage. The tibiofibula/astragalus concentrate the weight of the rear half of the articulation in the fore-and-aft plane is situated body on the massive femur. The straight, large almost at a right angle to the astragalus-navicular femur possesses huge tuberosities and articulates articulation, an arrangement certainly adaptive equally with the tibia and fibula. The head of the for stability of the tarsus and decidedly ina- femur is spheroidal and is situated on a poorly daptive for running motion. defined, unconstricted neck. The lesser trochanter The metatarsals are nearly cubic in propor­ is underdeveloped, another feature peculiar to tions, again an extreme among mammals and graviportal animals. indicative of a graviportal condition. Lack of NUMBER 40 205 tarsal and metatarsal reduction or fusion is an percent computed for one specimen of Glyptothe­ indication of the "primitive," or generalized con­ rium texanum is an extreme among all terrestrial, struction, in which static and dynamic flexibility, quadrupedal mammals, living and extinct, if rather than stability, is the principal functional Howell's treatment is to be considered exhaustive. requirement. Thus the glyptodont tarsus and Discussion of other indices measuring other metatarsus permit the support of an enormous limb proportions would be redundant. The ines­ load, while they provide sufficient flexibility to capable conclusion is that glyptodonts possess an adjust to minor substrate irregularities, which extremely graviportal construction as indicated would be otherwise dangerous to so cumbersome not only by the morphology of individual ele­ an animal. ments but also by relative limb proportions. Thus The five symmetrically disposed digits are com­ the commonly held belief that glyptodonts were plete and highly modified for weight support. clumsy, cumbersome quadrupeds is here main­ The phalanges of the rear foot are even shorter tained, and no other conclusion seems possible. and stouter than those of the front foot, and the The locomotor apparatus of glyptodonts there­ terminal phalanges were encased in an ungual fore exhibits an unusual extreme for weight sup­ sheath rather intermediate between a nail and a port, with little modification for cursorial habits. hoof construction. The entire lower extremity of The feet are constructed in a fashion that permits the pes closely resembles that of elephants, and extension and retraction of the digits and meta- the component bones are even more extensively podials during advance and recovery of the foot, shortened and modified for weight support. It respectively. It is evident that this adaptation is appears that the spreading-retraction function of useless in upland, hard-ground terrain and that the digits is highly developed for the pes, and as the extreme graviportal proportions in the glyp­ expected, the pes is decidedly "more graviportal" todont limbs preclude occupation of irregular or than the manus. hilly regions. Finally, a brief consideration of glyptodont Glyptodonts probably inhabited lowland areas limb proportions, utilizing Howell's (1944) obser­ along primary water courses, frequenting lake vations, is instructive. In North American glyp­ and stream shorelines and perhaps even occupy­ todonts, the humeroradial index (radius/humerus ing shallow water, much as do modern hippos X 100) varies between 50 percent and 55 percent. and capybaras. In this habitat, the "spreading- This figure is more extreme than for any quad­ foot" adaptation would be most useful (for pass­ rupedal, terrestrial living mammal (approxi­ ing over mud and soft, marshy areas otherwise mately 100 percent is the generalized index, the uninhabited by most mammals, including higher the index the more cursorial and the lower predators), and there would be little necessity for the index the more graviportal). Hippos possess either agility, in order to negotiate an irregular the closest index among living mammals, a dis­ terrain, or running to escape from predators. tant 68 percent. Only the distinctly graviportal DEFENSE AND PROTECTION.—Glyptodonts notoungulate genus Pyrotherium among extinct found little advantage either in running or in mammals possessed a lower index, around 40 fighting to escape predation. Indeed it is unlikely percent. The femoral-tibial index (tibia/femur that they possessed either ability, for it appears X 100) in glyptodonts also falls between 50 per­ doubtful that glyptodonts could run at all, and cent and 55 percent. Again this figure is an they lack the necessary dental equipment and extreme compared to those of living animals, claw modifications for aggressive defense. The among which the decidedly graviportal hippos carapace, as an impenetrable covering, was cer­ and elephants (indices around 63 percent and 54 tainly the most important protective device. Its percent, respectively) fall closest to glyptodonts in inadequacies regarding total protection have this proportion. The femoral-tibial index of 50 been alluded to in the description of the skull, 206 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY and it is necessary to consider additional modes tection over a more upland, hard-substrate ter­ of defense and protection. rain that large predators could more easily nego­ The functional role of the glyptodont armor tiate. Moreover, occupation of shallow-water for predator protection is likely a secondary adap­ areas, even occasionally, would reduce suscepti­ tation. The glyptodont armor probably evolved bility to predation. Young individuals were likely from an ancestral investment of isolated dermal especially susceptible to predators, for it was not bones, perhaps for protection from coarse vege­ until attainment of full adult size that the cara­ tation. With increasing and more extensive cov­ pace became fully functional for protection. erage of the body, it is evident that the dermal Therein lies the reasoning behind the second armor would become bulky and heavy. This tend­ suggestion—that glyptodonts relied heavily on ency likely resulted in a concomitant reduction in behavioral means for predator avoidance. In par­ cursorial advantage in the limbs, further necessi­ ticular, we propose that glyptodonts were gregar­ tating increased predator protection afforded by ious, for the young individuals lacked even cara­ the rigid carapace. On the other hand, armadil­ pacial protection. Without the double advantage los, with a more flexible and lighter carapace, of gregarious habits and a favorable habitat, it is retain considerable running ability (compared difficult to imagine infant and juvenile survival with glyptodonts) as reflected not only in their in an animal so large and so poorly adapted for osteology but also in their observed behavior predator protection in the immature state. Oc­ (Talmage and Buchanan, 1954). Hence, it is no cupation of a terrestrial, shoreline habitat, fre­ accident that glyptodonts possessed, at the same quenting both shallow water and lowland areas time, extremely graviportal limb proportions and with dense vegetation would seem to be the op­ the heavy, rigid carapace. timal habitat for an animal with these combined The glyptodont carapace afforded excellent characteristics. protection from predators. The cephalic shield A related, but peripheral, matter concerns the and the elaborate caudal armor almost com­ glyptodont protection from external parasites. pletely covered the body, leaving only the lower Mostly lacking hair, at least dorsally, it is doubt­ limbs and anterior facial region unprotected. The ful that glyptodonts were favorable hosts for fleas limbs were protected by a heavy investment of and lice; a similar conclusion has been docu­ dermal ossicles, as mentioned in the description mented for armadillos by Talmage and Bu­ of the carapace, probably more for protection chanan (1954). The smooth, uncreviced carapace, from vegetation than from predators. The fact with its scaly covering, is also an obstacle to tick that the dermal armor was not absolutely impen­ infestation. The undersurface of the body and the etrable was mentioned in connection with a skull limbs, however, likely had considerably more hair of Glyptotherium texanum (F:AM 95737, Figure 9), than the carapace, and it seems that in these in which two well-positioned holes in the skull regions of the body, glyptodonts would be suscep­ roof indicate that the cephalic shield was not tible to ectoparasites. The general lack of axial totally protective. Obviously, therefore, glypto­ mobility and the absence of caniniform or incisi- donts possessed additional predator protection, form teeth surely presented serious difficulties at which we propose involved a combination of grooming (the lack of ectoparasites may have habitat preference and behavioral tendencies. been an antecedent condition requisite for loss of First, the habitat proposed here for Glyptothe­ anterior teeth; a similar statement could be ad­ num would be conducive to hiding or inaccessi­ vanced for the origin of alternative grooming bility rather than running for predator avoidance. habits prior to the total loss of anterior teeth). Occupation of soft-substrate terrain, in marshy or Perhaps if glyptodonts possessed a proboscis, as muddy lowland areas along primary water suggested elsewhere, it performed as a grooming courses, would be a decided advantage for pro­ device. Alternatively, it is conceivable that glyp- NUMBER 40 207 todonts had a close commensal association with Gilliland local fauna from the Seymour formation certain birds, much as that of oxpeckers with for G. anzonae (Hibbard and Dalquest, 1966). A many elements of the modern African mega- similar statement can be made for the absence of fauna. capybaras associated with G. floridanum in the Ingleside fauna of Texas (Lundelius, 1972). Other Late Pleistocene faunas in both Florida and HABITAT Texas have an unusually high frequency of this Much of the foregoing discussion has alluded association. The relationship is all the more strik­ to the habitat requirements of glyptodonts. De­ ing for the relative rarity of both capybaras and ductions from anatomical conditions and inferred glyptodonts in continental United States. In fact, behavioral adaptations point toward a semia- it may be reasonably assumed that where one is quatic, semiterrestrial habitat in a region of dense found, the other should be expected. vegetation. There is little reason to belabor the Partial explanation for the capybara-glypto- points and deductions already made concerning dont association lies in the neotropical origins the glyptodont habitat, and the following com­ and centers of dispersal. Both are warm temper­ ments are restricted to faunal associations and the ature animals, and the conclusion that they oc­ limited available evidence of sedimentology. cupied the same habitat is inescapable. Indeed, Specific faunas that include glyptodonts were the semiaquatic habits of capybaras are directly enumerated in the distribution section. The analogous to those suggested here for glyptodonts, faunas are invariably composed of animals with and it is no accident that the two are often southern affinities, either phylogenetic or eco- associated. logic, or both. The fact that glyptodonts are Another, almost expected, faunal association is neotropical representatives of the Pleistocene is with the giant tortoise genus Geochelone. These consistent with the overall composition of the turtles have received extensive attention as indi­ associated faunas. cators of a warm climate. Their common associ­ An overwhelmingly common association is the ation with glyptodonts substantiates the assump­ occurrence of glyptodonts with capybaras (the tion of a warm climate necessary for their ex­ rodent genera Neochoerus and Hydrochoerus, family istence. The fact that Geochelone is much more Hydrochoeridae), a circumstance which leads to common than glyptodonts in the Pleistocene of several important conclusions. The occurrence of North America is explained by the relative free­ capybaras with Glyptotherium texanum is known dom of land tortoises from close proximity to from the Tusker fauna in Arizona (Wood, 1962) water in comparison with glyptodonts, which are and from specimens (Ahearn and Lance, 1980) here asserted to have been limited to rather re­ collected at the same time and found in the strictive circumstances related to water accessi­ collections with the glyptodonts from the 111 bility and associated habitat. It is apparent that Ranch area in the Frick Collection, American the habitat requirements of Glyptotherium were Museum of Natural History. Capybaras are also more restrictive than for some of the species of associated with G. arizonae at least in the Curtis Geochelone, and the principal limiting factor was Ranch fauna of Arizona (Lammers, 1970). probably related to rainfall and humidity. Whether the Texas local faunas including G. Whereas glyptodonts, Geochelone, and capybaras texanum and G. arizonae also include capybaras is are all indicative of warm, mild winters and indeterminate, for small mammals are poorly moderate summer temperatures, it is evident that represented in these faunas; this statement is the distributional advantage oi Geochelone over the inclusive for the Texas Blanco fauna (Johnston other two groups is greater independence from and Savage, 1955) and Red Light local fauna water. (Akersten, 1972), both for G. texanum; and the Sedimentological evidence, insofar as known, 208 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY further substantiates these assertions. Both Wood four or five genera of North American glypto­ (1962) and Seff (1962) have stated that the Ari­ donts are now placed in a single genus Glypto­ zona occurrences of G. texanum in the 111 Ranch thenum, containing four well-defined species, G. area (Tusker local fauna) are associated with texanum, G. anzonae, G. cylindricum, and G. flori­ lacustrine and fluviatile sediments. Mr. Ted Tal- danum, and a fifth species, which is nominally usha (pers. comm.) has independently stated a retained, G. mexicanum. Originating from within similar belief for the glyptodont occurrences in the South American subfamily Glyptodontinae, the same area. Hibbard and Dalquest (1966) Glyptothenum first appears in the earliest Pleisto­ included stream and marshland communities in cene in Arizona and Texas. Glyptotherium texanum their discussion of the various habitats of the is the earliest species, and it represents the ances­ Gilliland local fauna, which contains G. anzonae, tral group for the remainder of the North Amer­ while Gidley (1926) and Lammers (1970) have ican taxa. stated that stream and lake communities are The ancestral species is small in comparison prevalent in the Curtis Ranch fauna of Arizona. with the descendant species; it nevertheless ex­ Similar conclusions have been reached by Lun­ hibits nearly all of the features of the other taxa, delius (1972) for the late Pleistocene sedimentary but with less exaggeration. Differing primarily in conditions at Ingleside, Texas, which includes G. larger size and attendant exaggeration of princi­ floridanum. In all, there is a commonality of lacus­ pal characteristics, G. arizonae soon replaced the trine and fluviatile conditions, indicating that ancestral species during Irvingtonian time, and, sedimentological evidence is not contradictory to in the broad sense, occupied almost exactly the the proposed habitat for Glyptotherium. same geographic area. Thus with respect to each Webb (1977, 1978) has presented a fertile sum­ other, the early Pleistocene taxa G. texanum and mary of the evolution of savanna vertebrates in G. anzonae are temporal, rather than geographic, the New World, in which Glyptotherium is listed as species. a member of the generalized Plio-Pleistocene "in- Because the paleontological record in southern teramerican mingled fauna," one of nine genera United States and Mexico is highly incomplete, of savanna grazers (Webb, 1978:404, 405). We it cannot be determined whether Glyptotherium suggest that Glyptotherium belongs instead in the continuously occupied the Arizona-New Mexico- category of "aquatic grazers" along with two Texas-Oklahoma region from the early Pleisto­ genera of capybaras, "which tended to graze cene portion of the Blancan Land Mammal Age 'hippo-style' in the adjacent savannas," and the through the Irvingtonian Land Mammal Age. manatee, Trichechus, which was of course "wholly Such an assumption can be neither proven nor aquatic." We would add Glyptotherium to the same denied. Since the faunas containing these two category as the capybaras, living and feeding in species were probably unaffected by glacial con­ lowlands adjacent to watercourses. ditions to the north, there is little reason to assume It is reasonable and consistent to assume that any pronounced habitat disruption in this region, Glyptothenum lived in the lake and stream com­ and the suggestion that Glyptotherium more or less munity, and that the climate was warm, mild, continuously occupied this area is at least a logical and relatively humid to support tropical or sub­ assumption. Stratigraphic details for the faunas tropical vegetation at least near permanent bodies containing G. texanum and G. arizonae have not of water. been satisfactorily determined, and much of the "old dogma" regarding these faunas has recently come under critical review (e.g., Hibbard and EVOLUTIONARY HISTORY OF Glyptotherium Dalquest, 1973; Lindsay and Tessman, 1974; According to the taxonomic revision proposed Lindsay, et al., 1975). Thus it cannot be deter­ in this study, what were previously considered as mined with certainty whether there is strati- NUMBER 40 209 graphic overlap between these two species. Our area is wanting. At least two species, G. cylindricum opinion is that these are strictly temporal species, and G. mexicanum, assuming that the latter is valid, and that specimens of intermediate age will also appear to be indigenous to Mexico and (likely) exhibit intermediate conditions. Thus it is our Central America during Middle and Late Pleis­ belief that G. texanum and G. arizonae are strati­ tocene times. Neither species is well known; both graphically useful for establishing relative ages of are probably derivative from either G. texanum or faunas containing either of these two species. It is from G. arizonae. impossible to claim categorically that one or the Finally, in the Late Pleistocene, during the other, or both, is restricted to either the pre- or latter part of the Rancholabrean Land Mammal the post-Blancan/Irvingtonian boundary, and ac­ Age, Glyptotherium reappeared in the United cordingly neither is unequivocally useful as an States. This time, however, the distribution was index species, although in conjunction with other limited to the Gulf Coastal Plain and the southern mammal taxa they are potentially valuable. Atlantic coast. The Late Pleistocene representa­ The existence of one of these two species during tive is G. floridanum, and although known from a the Early Pleistocene in Florida has also been large number of localities, this species is less well established; the age is known with a fair degree defined than either G. texanum or G. anzonae. There of confidence to be late Blancan and early Irving­ is little doubt, however, that G. floridanum evolved tonian. Nevertheless, identification of the early from the ancestral stock, probably from G. arizonae representation in Florida is here suggested to be rather than directly from G. texanum, since the G. anzonae. If correct, the Florida record may be former apparently survived longer than the latter. the earliest for this species, in which case a strong The population of G. floridanum became extinct argument may be advanced for initial geographic at the close of the Pleistocene, or sometime isolation from the ancestral population of G. tex­ thereabouts, but not before having evolved mor­ anum to the west. Hence, there could well have phological manifestation of sexual dimorphism, been contemporaneous existence of both species, a characteristic apparently lacking in the earlier G. texanum in the West, G. anzonae in the East, species. Glyptothenum floridanum probably also ex­ during Blancan time, with subsequent deploy­ tended well into Mexico during the Late Pleisto­ ment westward of the descendant species in Ir­ cene (correlating with the Wisconsin Glacial Age vingtonian time, thus replacing the ancestral spe­ to the north), although substantiated identifica­ cies. These thoughts are suggested as a parsimo­ tions from this area have not been established. nious explanation for the early evolution oi Glyp­ Concerning possible causes of extinction for tothenum in North America. In essence, this hy­ Glyptotherium, it is evident that climatic deterio­ pothesis couples initial geographic isolation, pro­ ration resulting from continental glaciation is ducing contemporaneous ancestral and de­ sufficient explanation. As proposed here, the hab­ scendant species, with further temporal speciation itat occupied by glyptodonts was unusually re­ later in the West. strictive, and even minor disruption of the low­ Glyptodonts apparently were absent in Florida land, warm temperature, humid and tropical cir­ during the Irvingtonian, and sometime during cumstances under which they lived would bring the course of this land mammal age, G. arizonae about their local demise. disappeared from the northern reaches of its dis­ tribution (United States), probably withdrawing Conclusions southward into Mexico as a consequence of de­ teriorating climatic conditions ensuing from con­ The intent of this study is to accomplish a tinental glaciation to the north. three-fold, comprehensive review of North Amer­ Glyptotherium is known only poorly in Mexico, ican glyptodonts: first, to review, in detail, the and accurate stratigraphic information for this gross osteology of these heretofore poorly known 210 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY animals; second, to establish a sound taxonomic 5. Gross osteology for all five species is known framework; and third, to delve into a few of the more or less comprehensively. Glyptotherium tex­ more important aspects of glyptodont paleobiol­ anum and G. anzonae are the most nearly com­ ogy, including their functional morphology, their pletely known; G. floridanum is reasonably well habitat requirements, and the history of their known; G. cylindricum is less well known, from a evolution on the northern continent. To the ex­ single specimen; and G. mexicanum is poorly tent that information is available, these objectives known. Osteological details indicate a fundamen­ have been addressed with a fair measure of con­ tally close relationship among all five species. fidence, and the following are some of the more 6. As derived from osteological and sedimen- important results of this review. tological evidence, and from faunal associations, 1. Insofar as can be determined, all glypto­ the habitat requirements of Glyptotherium were donts recovered in the Pleistocene of northern rather restrictive: proximity to standing water in Mexico and southern United States belong in a a region with extensive lowland terrain; lush, single genus, Glyptotherium Osborn, 1903. Included tropical or subtropical vegetation; a warm cli­ in this genus are four well-defined species, G. mate without excessive extremes of temperature; texanum, G. arizonae, G. cylindricum (= Brachyostracon and high, relatively constant moisture (for main­ cylindricus), G. floridanum (= Boreostracon flondanus), taining the permanent bodies of water and the and a fifth species of less certain standing, G. abundant vegetation). mexicanum (= Glyptodon mexicanus, = Brachyostracon 7 Glyptotherium was a strictly herbivorous, ex­ mexicanus). tremely graviportal, heavy-bodied representative 2. Xenoglyptodon fredericensis Meade, 1953, and of the Pleistocene megafauna. These glyptodonts Glyptodon fredericensis (Meade) are junior synonyms were largely devoid of predator protection except of Glyptotherium arizonae. The existence of the for their armor, and they lacked any capacity South American genus Glyptodon in North Amer­ whatsoever to flee from an approaching predator ica remains doubtful, and until a thorough revi­ beyond moving into shallow water. sion of South American glyptodont taxonomy is 8. Because of their restrictive habitat require­ accomplished, retention of the generic name Glyp­ ments, glyptodonts are relatively rare in the tothenum for the North American representatives northern extent of their range, which reaches to is the most advisable course. Arizona, Oklahoma, and South Carolina. Their 3. The five species of North American glypto­ center of abundance in North America was prob­ donts evolved from an initial population of the ably in Central America and Mexico, with occur­ ancestral species, Glyptotherium texanum, which mi­ rence in continental United States more restricted grated from South America during late Pliocene and possibly intermittent. or earliest Pleistocene time. Subsequent evolution It is impossible for any paleontological review resulted in what appear to be at least two succes­ to be totally comprehensive, and the present sive species here asserted to be primarily "tem­ study is no exception. Accordingly, it is appropri­ poral" species rather than "geographic" species: ate to point out what seem to be areas of special G. texanum, to G. arizonae, to G. floridanum. The interest in particular need of further investiga­ other two species are poorly known, and their tion. relationships are less certain, but they evolved Of course, the taxonomic framework proposed from within this complex. here, like any classification, is subject to further 4. A single invasion is sufficient to account for improvement. New material, especially skulls, all of the North American species through tem­ pertaining to the group here identified as G. poral and geographic speciation from the ances­ arizonae could well establish that this group ac­ tral stock. There is no reason to assume multiple tually includes two species, one from Arizona, the invasion from South America. other from Texas. That both groups are closely NUMBER 40 211 related is certain, and for lack of evidence to the Blancan representatives and the relationships of contrary, both are here assigned to the same the Mexican and Central American glyptodonts taxon. Conversely, new material could also estab­ to the more northern species. lish beyond doubt the identity of these two Further, the relationships between Glyptotherium groups, and there would thus be a strong argu­ and the South American glyptodonts remain to ment for near contemporaneity of the Curtis be established. This would of course entail a Ranch and Seymour faunas. Our opinion is that comprehensive review of the South American the latter circumstance will prove to be the case. Glyptodontinae and would be a rewarding, Another taxonomic problem concerns the loose though monumental, task, which must be carried assemblage here referred to Glyptotherium flori­ out before the North American immigrants can danum. Whether there are separate Texas and be fully understood. Florida species cannot be satisfactorily deter­ Concerning the paleobiology of glyptodonts, mined from the available specimens, nor is it many potentially fruitful aspects deserve detailed possible to establish with certainty the postulated inspection, including elaboration upon points sexual dimorphism in this species. What has been touched on here regarding the masticatory ap­ labeled sexual dimorphism could well represent paratus, locomotion, defense and protection, sympatric species, in which case a revision of the growth and development, and habitat require­ Late Pleistocene glyptodonts from the Gulf Coast ments. We hope that our descriptions, discussions, would be called for. hypotheses, and speculations will provide a stim­ Other taxonomic considerations that deserve ulus for renewed and more sophisticated research special attention when additional material be­ on these bizarre and unjustly neglected mammals. comes available are the identity of the Florida Appendix Tables of Measurements

TABLE 1.—Measurements (mm) of the skull of North American glyptodonts

G. texanum G. anzonae G. floridanum

Characters F:AM F:AM UMMP ChM 59583 95737 34826 2415

1. Maximum transverse diameter of frontals _ - 106 - 2. Transverse diameter between inner margins of - 90 130 - infraorbital foramina 3. Vertical length of descending zygomatic process _ 85/- 102/110* _ from lower margin infraorbital foramen (left/ right) 4. Anteroposterior diameter from anterior margin of 210/202 250/254 base of descending zygomatic process to posterior extremity, occipital condyle (left/right) 5. Maximum transverse diameter between lateral 176 254 extremities, zygomatic arches 6. Inferosuperior diameter from N? rear lobe occlusal 139V137 138V138* 179/179 surface to sagittal crest junction with lambdoidal crests (left/right) 7. Transverse outside diameter between paroccipital 115 117 li.li 132* processes 8. Transverse outside diameter occipital condyles 87 8H 108 101 9. Transverse diameter, left facet for occipital 30* 24 34 35 condyle 10. Transverse diameter, right facet for occipital 25 - 33 36 condyle 11. Inferosuperior diameter, left facet occipital 23 22 31 23 condyle 12. Inferosuperior diameter, right facet occipital 21 21 29 23 condyle 13. Maximum transverse outside diameter between 48 47 65 - crests of pterygoid tuberosities 14. Inferosuperior diameter from lower margin man­ /67 60*/60 86/87 _ dibular facet lo lower margin pterygoid tuberosity near N- alveolus (left/right) 15. Anteroposterior diameter, mandibular facet to in­ -/- 127/125 150/150 fraorbital canal, lower anterior margin (left/right) 16. Inside transverse diameter, foramen magnum 34* 38* 32* 17. Inside inferosuperior diameter, foramen magnum 22* 26* 20* 18. Maximum vertical diameter from centerof palate, 102 posterior extremity, to sagittal crest Approximate measurement.

212 NUMBER 40 213

TABLE 2.—Measurements (mm) of the mandible of North American glyptodonts

G. texanum G. anzonae G. floridanum

Characters JWT JWT USNM UMMP TMM USNM 2330 1715 10536 38761 30967- 11318 1814 1. Height of ascending ramus measured from con­ 230 245 195 175 dyle through posterior lobe of N7 to lower margin of horizontal ramus 2. Maximum transverse diameter, articular condyle 49" 39 - 3. Anteroposterior diameter, articular condyle 16a 17 - 4. Transverse diameter, horizontal ramus below N5 31 33 33 25 25 5. Transverse diameter, ascending ramus opposite 17b 18 19 - - N7 6. Vertical depth, horizontal ramus at N7 including 107 100b 77 51 tooth 7. Vertical depth, horizontal ramus at N5 including 70h 106 104 75 58 tooth 8. Vertical depth, horizontal ramus at N3 including 96 78 67 tooth a Left/right average. Approximate measurement. 214 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 3.—Measurements (mm; left/right) of the upper dentition of Glypothenum texanum

Axes Nib N? N3- Ni N5° N? N? N?

F:AM 59583 1. Maximum anteroposte­ 14.1/14.2 -/- 19.0/19.8 -/19.0 -/19.2 -/19.5 -/19.1 -/16.2 rior diameter measured through center (a-axis) 2. Maximum transverse 7.3/ 7.3 -/8.2 10.1/10.0 -/12.6 12.4/13.3 14.0/14.4 13.8/13.8 -/13.9 diameter, anterior lobe (b-axis) 3. Minimum transverse -/4.5 3.9/ 3.2 -/- 3.0/ 3.3 3.4/3.4 -/ 3.3 3.2/ 3.0 diameter, anterior constriction (c-axis) 4. Maximum transverse 7.8/7.9 9.8/10.2 -/l 1.2 11.8/11.2 11.9/12.4 -/12.4" 11.7/11.2 diameter, middle lobe (d-axis) 5. Minimum transverse 4.6/4.7 4.4/ 4.1 -/ 3.8 -/ 4.2 -/ 3.9 -/ 4.3 4.0/ 4.0 diameter, posterior constriction (e-axis) 6 Maximum transverse 6.9/6.9 9.1/ 9.1 -/l 1.2 -/12.9 -/13.3 12.9/12.9 10.8/10.6 diameter, posterior lobe (f-axis)

F:AM 95737d 1. a-axis -/15.2 16.6/16.7 17.2/17.8 -/18.4 18.9/- 18.1/18.1 -/18.0 15.4/16.0 2. b-axis -/ 7.0 7.5/ 7.0 10.5/ 9.8 /l 17 12.6/- 13.0/12.4 -/12.5 12.5/- 3. c-axis 4.7/ 4.8 4.5/ 4.1 -/ 3.4 3.3/- 3.4/ 3.6 -/ 3.3 -/- 4. d-axis 8.7/ 8.1 10.7/11.1 /l 1.6 12.2/- 12.0/12.2 -/12.2 11.2/10.9 5. e-axis 4.8/ 5.0 4.4/ 4.3 -/ 4.0 3.8/- 3.9/ 3.7 -/ 3.7 3.6/ 3.9 6. f-axis 7.3/ 7.0 9.6/ 9.7 -/10.6 12.0/- 11.8/- -/12.0 10.6/10.5

WT 1837" 1. a-axis -/19.8 2. b-axis -/14.8 3. c-axis -/ 3.4 4. d-axis -/13.0 5. e-axis -/ 3.9 6. f-axis -/13.3 a Approximate measurement. b a-axis: maximum length; b-axis: maximum width. c Anteroposterior diameters: N^-N- (g-axis); N--N- (h-axis); N?-N- (i-axis); N?-N- (j-axis) not applicable. Anteroposterior diameters N--N- (k-axis) = 40/43; N--N? (1-axis) = 87"/87. d g-axis = -/163; h-axis = -/80; i-axis = -/62; j-axis = 107/106; k-axis = 42/-; 1-axis = 81/- " Isolated posterior upper tooth, might be N- or N- instead. NUMBER 40 215

TABLE 4.—Measurements (mm; left/right) of the upper dentition of Glyptothenum arizonae (axes as defined in Table 3)

Axes Ni N? N5 N4 N5 N? N7- N?

UMMP 34826" 1. a-axis 19.4/18.6 -/- 22b/26b 23.9/24.8 25.5/24.6 24.2/25.0 23.6/24.1 22.5/23.0 2. b-axis 10.4/ 9.7 7.2/- -/15.6 17.8/17.3 18.0b/18.7 19.5/18.9 18.8/18.0 17.2/17.1 3. c-axis -/- 2.6/- -/- 4.3/- 3.8/ 4.2 3.9/ 4.6 4. d-axis -/- 17.9/- 18.4/18.9 17.9/17.7" 18.1/17.5 15.5/15.8 5. e-axis -/- -/- 5.2/- 5.0/ 5.0 5.0/ 4.5 3.9/ 3.5 6. f-axis /14.3 -/16.3b 17.2/17.6 17.5/16.5 17.7/17.5 15.3/14.8

UMMP 38761 1. a-axis -/25.7 -/25.8 -/26.2 2. b-axis -/13.2 -/19.4 -/- 3. c-axis -/ 5.3 -/- -/- 4. d-axis -/15.6 -/- -/- 5. e-axis -/ 5.6 -/- -/- 6. f-axis -/14.8 -/- -/- " g-axis = 200/200; h-axis = 92/93; -axis = 73/74; j-axis = 132/133; k-axis = 54/54; 1-axis 106/108. Approximate measurement.

TABLE 5.—Measurements (mm; left/right) of the upper dentition of Glyptotherium cylindricum, AMNH 15548 (axes as defined in Table 3)

2 Axes Ni* N * N3 N4 NS Nb- N? N? 1. a-axis 19.5/- 21.2/- 22.2/22.1 -/- 23.5/23.0 24.1/23.2 21.9/- 18.3/18.5 2. b-axis 9.1/- 11.6/- 13.9/13.5 -/- 15.9/16.5 15.5/15.5 13.9/- 12.5/12.6 3. c-axis 4.3/ 4.7 -/- 3.8/ 4.0 3.8/ 4.0 3.4/- 3.2/ 3.0 4. d-axis 12.5/12.4 -/- 13.7/14.2 13.9/13.6 12.8/- 11.3/11.1 5. e-axis 5.0/ 5.6 -/- 4.2/ 4.8 4.0/ 4.4 4.2/- 3.4/ 3.9 6. f-axis 11.5/12.5 -/- 14.2/15.4 14.2/14.5 13.0/- 11.4/10.7 a-axis and b-axis are maximum length-width, respectively.

TABLE 6.—Measurements (mm) of the upper dentition oi Glyptothenum floridanum (axes as defined in Table 3)

Axes USNM 6071 UF/FGS 6643 TMM 31186-19

N? N4- N4- N4(?) 1. a-axis 19.0 22.1 20.5 21.7 2. b-axis 8.6 14.9 12.5 12.3 3. c-axis 4.9 4.1 3.1 3.8 4. d-axis 9.6 15.8 12.7 13.0 5. e-axis 4.6 3.5 3.0 4.3 6. f-axis 8.0 13.4 11.3 11.8 216 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE—7.—Measurements (mm) of the lower dentition oi Glyptothenum texanum

Axes NT N2 Nt Nib Ne N'T

JWT 2330 (isolated teeth plus Ni, NT in mandible fragment) 1. Maximum anteroposterior diameter measured 16.9 20.9 20.4 17.5 18.7° 18.2 through center (a-axis) 2. Maximum transverse diameter, anterior lobe 8.8 7.3 9.4 11.4 12.7C 12.7 (b-axis) 3. Minimum transverse diameter, anterior con­ 4.1 3.7 4.1 3.9 3.8 striction (c-axis) 4. Maximum transverse diameter, middle lobe 8.7 12.1 12.6 13.4 13.2 (d-axis) 5. Minimum transverse diameter, posterior con­ 4.1 3.1 3.7 3.8 3.5 striction (e-axis) 6. Maximum transverse diameter, posterior lobe 8.9 13.5 14.0 13.9 13.5 (f-axis) JWT 1715d NT Ni NT NT NT Ne N7 Ns 1. a-axis 18.5 20.8 21.3 19.0 _ 18.9 20.4 2. b-axis 10.0'' 6.8 7.1 9.4 - 14. r 12.6 3. c-axis 4.2 5.0 4.1 - 4.8 3.9 4. d-axis 8.8 10.0 11.9 - 13.5 12.9 5. e-axis 4.6 4.5 4.3 - 4.1 3.0C 6. f-axis 9.5 11.4' 12.8 - 13.7 13.7 ° Position instead may be adjacent to Nj. Position instead may be NJ. ' Approximate measurement. Left mandible with broken teeth in alveoli; measurements taken on broken teeth w thin alveoli near alveolar border; anteroposterior diameters: NT Ni (g-axis) = 157 ± 5 (h-axis) not applicable this specimen; NT-NT (i-axis) = 80; N5-N? (j-axis) = 66. NUMBER 40 217

TABLE 8.—Measurements (mm; left/right) of the lower dentition of Glyptotherium anzonae (axes as defined in Table 7)

Axes NT N2 NT NT Ni Ni NT Ni

UMMP 38761" 1. a-axis -/- 21.7/- 24.07- 24.1/- -/- -/26.0 -/25.1 -/25.1 2. b-axis -/- 9.1/- 10.5/- 11.1/- -/- -/14.2 -/14.8 -/16.0 3. c-axis -/- 5.2/- 4.9/- 4.2/- -/- -/ 5.1 -/ 4.5 -/- 4. d-axis -/- 10.9/- 14.0b/- 14.8b/- -/- -/17.7 -/17.6 -/- 5. e-axis -/- 5.9/- -/- 4.6/- -/- -/ 4.8 -/ 4.6 -/- b 6. f-axis -/- 10.7/- 14.2/- 16.7b/- -/18.3 -/17.7 -/17.7 -/16.0 USNM 10536° 1. a-axis 24.0/- 21.7/23.2 24.0/23.8 24.8/- 23.0/23.2 24.0/23.5 23.5/- 24.5/- 2. b-axis 9.1/- 8.3/ 7.7 9.7/10.3 13.7/14.8 15.1/16.3 16.2/16.7 15.9/15.7 15.4/- 3. c-axis 4.4/ 4.8 4.5/ 4.7 4.6/- 4.7/ 4.6 -/ 5.3 4.6/ 4.7 4.0/- 4. d-axis 10.9/10.6 13.0/13.4 13.9/14.4 14.8/15.1 15.5/15.4 15.0/- 14.0/- 5. e-axis 4.9/ 5.0 4.0/ 4.6 4.6/- 4.3/- 4.6/ 4.2 4.5/- 3.8/- 6. f-axis 11.7/11.0 14.7/15.7 16.9/17.9 17.6/17.8 17.2/17.1 15.6/15.7 13.5/- a N2-N4 measurements from left mandible; Ni-Ni from right mandible; j-axis = 66. Approximate measurement. c g-axis = 197; h-axis = 178; i-axis = 95; j-axis = 76.

TABLE 9.—Measurements (mm; left/right) of the lower dentition of Glyptothenum cylindricum, AMNH 15548* (axes as defined in Table 7)

Axes NT N2 Ni NT Ni Ni N, Ni 1. a-axis -/- -/18.6 -/18.8 -/- 24.2/- 21.6/22.2 21.2/21.5 20.3/20.5 2. b-axis -/- -/ 6.0 -/ 7.5 -/- 13.6/- 12.7/12.5 12.6/13.4 12.5/12.4 3. c-axis -/ 5.0 -/ 4.0 -/- 4.0/- 3.8/ 3.8 3.3/ 3.6 3.3/ 3.2 4. d-axis -/ 9.0 -/10.0 -/- 14.5/- 14.3/14.4 13.0/13.8 12.7/12.1 5. e-axis -/ 5.1 -/ 4.5 -/- 4.1/- 3.8/ 4.4 3.5/ 4.5 3.1/ 3.1 6. f-axis -/10.0 -/l 1.0 -/- 14.5/- 14.3/14.5 13.7/14.7 12.2/12.0 * Isolated teeth. 218 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 10.—Measurements (mm) of the lower dentition of Glyplolhenum floridanum (axes as defined in Table 7)

Axes NT N2 NT N'T Ni Ni N7 Ni

TMM 30967-1814e 1. a-axis 18.9" 20.0 22.0C 21.7 - 21.8 19.6" 2. b-axis 6.3 7.2 10.0 - - 11.6 - 3. c-axis 3.6 3.4 - 4.9 - - 4. d-axis 9.6 11.8 11.6 12.7 11.7 - 5. e-axis 3.5 3.4 3.9 - - - 6. f-axis 11.2 12.7 13.3 13.3 12.3 - USNM 6071, UF/FGS 6643b 1. a-axis 19.9 20.6 20.0 19.9 21.6 2. b-axis 5.8 10.6 11.8 12.6 12.7 3. c-axis 4.5 3.3 - 3.1 - 4. d-axis 8.8 12.7 13.4 13.2 11.9 5. e-axis 4.5 3.5 - - - 6. f-axis 8.5 12.6 13.0 12.6 10.8

USNM 11318f 1. a-axis 11.2c'd 15.7 18.0 19.4 20.5 _ 20.6 _ 2. b-axis 7.0C- d 6.8 7.5 8.6 9.9 - 10.8 - 3. c-axis 5.3 1.1 3.5 3.3 - 3.3 - 4. d-axis 6.6 9.0 10.5 11.2 - 11.9 - 5. e-axis 4.5 4.3 3.3 3.2 - 4.2 - 6. f-axis 4.7 8.8 10.6 11.1 - 11.8 - a Measurement taken on alveolar border. hNi (right) for USNM 6071, isolated tooth; Ns, Ni, NT (left) and N« (right) for UF/FGS 6643, isolated teeth. c Approximate measurement. d Alveolar measurement. c h-axis = 157. f g-axis = 168c; h-axis = 155'; i-axis = 70; j-axis = 71. NUMBER 40 219

TABLE 11.—Measurements (mm) of the atlas of North American glyptodonts

G. arizonae G. cylindricum Characters UMMP 34826 AMNH 15548 1. Transverse outside diameter of occipital facets 116 99 2. Maximum dorsoventral diameter between tu­ 68 62 bercles of dorsal and ventral arches 3. Maximum transverse diameter, neural canal 46" 41a 4. Maximum dorsoventral diameter, neural canal 36 31a 5. Maximum transverse external-internal diame­ 38" 50 ter of left occipital facet 6. Maximum transverse external-internal diame­ 39 37" ter of right occipital facet 7. Dorsoventral diameter, posterolateral-ventral 74 61b tubercle to upper extremity, alar process left side 8. Dorsoventral diameter, posterolateral-ventral 58b tubercle to upper extremity, alar process right side 9. Dorsoventral diameter, odontoid process 13 15 10. Dorsoventral diameter, dorsal arch 12 14 11. Anteroposterior diameter, odontoid process 29 30 12. Anteroposterior diameter, dorsal arch 37 34 13. Transverse inside diameter between interver­ 83 36 tebral foramina on posterior surface 14. Maximum transverse diameter, lateral extrem­ 107 93 ities axis facets " Approximate measurement. b Upper extremity of alar process broken.

TABLE 12.—Measurements (mm) of the cervical tube of North American glyptodonts

G. arizonae G. floridanum Characters UMMP 34826 USNM 6071 1. Oblique anteroposterior diameter, centrum articular 110 80 facet to anterior extremity odontoid process 2. Maximum transverse diameter between lateral ex­ 102 79 tremities of atlas facets 3. Maximum transverse outside diameter, posterior 101* — centrum facets 4. Transverse diameter, left atlas articular facet 34* 23 5. Transverse diameter, right atlas articular facet 34* 22 6. Oblique anteroposterior/dorsoventral diameter, - 94 odontoid process to extremity of neural crest 7. Oblique dorsoventral diameter, lower margin left — 79 centrum facet to dorsal extremity neural crest * Approximate measurement. 220 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 13.—Measurements (mm) of the scapula of North American glyptodonts

G. texanum G. anzonae Characters F:AM 95737 USNM 10536 1. Greatest vertical depth, measured from glenoid facet 195 325 2. Greatest anteroposterior length 270 460 3. Maximum anteroposterior diameter, glenoid facet 68 83 4. Maximum transverse diameter, glenoid facet 45 59 5. Length of acromion, measured on curve 60* 120 6. Length of spine, including acromion, measured along 210 385 curve * Approximate measurement.

TABLE 14.—Measurements (mm) of the humerus of North American glyptodonts

G. texanum G. anzonae G. floridanum

Characters F:AM TMM USNM USNM 6071

95737 40664-245 10536 right left 1. Maximum length from proximal articulating surface of 258 - 365 - head to distal extremitv of trochlear facet 2. Maximum length from proximal articulating surface of 250 - 356 - head to distal extremity of capitular facet 3. Minimum diameter, proximal articulating surface of 249 - 354 - head to distal surface of intercondylar groove 4. Anteroposterior diameter of proximal extremity across 63 - 81 - - bicipital groove in sagittal plane 5. Maximum anteroposterior diameter of head measured in 51 - 76 - sagittal plane through lesser tuberosity 6. Maximum diameter from greater tuberosity across head 73 - KIT - 7. Minimum transverse diameter of shaft below deltoid 31 - 41 - ridge 8. Minimum anteroposterior diameter of shaft below deltoid 34 - 43 - ridge 9. Maximum transverse diameter of the proximal shaft 59 - 82 - across deltoid tuberosity 10. Maximum transverse diameter across condylar surface 56 71 72 62 63 11. Anteroposterior diameter of intercondylar groove 28 33 32 32 32 12. Maximum anteroposterior diameter of trochlear facet 37 45 49* 47 -48 13. Maximum anteroposterior diameter of capitular facet 31 39 45* 38 38 14 Maximum transverse diameter across epicondyles 85 119 114 103 104 15. Minimum diameter of medial supracondylar ramus mea­ 30 - 37 31 32 sured obliquely on anteromedial surface to coronoid fossa 16. Minimum diameter of medial supracondylar ramus mea­ 26 23 31 33 sured obliquely on posterior surface from medial margin to olecranon fossa 17. Maximum diameter of articular surface of head from 60 80* anterolateral apex bounding greater trochanter to poster­ omedial margin Approximate measurement. NUMBER 40 221

TABLE 15.—Measurements (mm) of the radius of North American glyptodonts

G. texanum G. anzonae G. floridanum

Characters F:AM F:AM USNM UMMP 38761 USNM 6071

95737 59585 10536 left right eft right 1. Maximum length measured from proximola­ 141 166° 184 174 teral extremity to distal extremity of styloid process 2. Maximum proximodistal diameter between 109 134 156 - 146 centers of articular facets 3. Maximum transverse diameter, proximal ar­ 43 56 59 56" 59 - 58' ticular facets 4. Maximum transverse diameter, distal extrem­ 53 - 65 55 55 ity styloid process to dorsal tubercle 5. Minimum transverse diameter of shaft 15 22 25 21 21 6. Minimum anteroposterior diameter of shaft 16c 24" 23 25 25 7. Maximum anteroposterior diameter, proxi­ 23 36 39' 31 32 mal extremity 8. Maximum midsagittal anteroposterior diam­ 26 - 42 33 33 eter of distal articular facet 9. Maximum transverse diameter, distal articu­ 41 55 52 19 lar facet measured obliquely from the facet border at styloid process to proximolateral border 10. Maximum anteroposterior diameter of prox­ 23 33 32* 27 imal articular facet 1 Approximate measurement. ' Minimum occurs near proximal end of shaft. Minimum occurs near distal end of shaft. Proximomedial tip broken away.

TABLE 16.—Measurements (mm) of the ulna of North American glyptodonts

G. texanum G. anzonae G floridanum USNM 6071 Characters F:AM USNM 95737 10536 right left 1. Maximum proximodistal length 194 270 239 237 2. Minimum anteroposterior diameter of shaft distal to 42 59 51 49 semilunar notch 3. Minimum anteroposterior diameter of shaft through 39 56 48 50 semilunar notch 4. Minimum transverse diameter of shaft distal to sem­ 18 22 19 18 ilunar notch 5. Minimum transverse diameter of shaft proximal to 20 23 23 23 semilunar notch 6. Transverse diameter of coronoid process to anconeal 47 59 49 48 process 7. Maximum diameter of cuneiform process 31 44 32 31 8. Maximum length of pisiform facet 24 23 22 22 9. Maximum width of pisiform facet 11 18 15 15 10. Length of olecranon above semilunar notch (anco­ 59 98 71 69 neal process to proximal extremity) 222 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 17.—Measurements (mm) of the pisiform of North American glyptodonts

G. texanum G. arizonae Characters F:AM 95737 USNM 10536 1. Maximum diameter medial angle to lateral angle 50* 58 2. Maximum diameter lateral angle to anterodistal angle 37* 51 3. Maximum diameter anterodistal angle to medial angle 29 33 4. Maximum diameter anteroproximal angle to anterodistal angle 15 22 5. Maximum diameter anteroproximal angle to medial angle 22 25 6. Maximum diameter anteroproximal angle to lateral angle 37* 58 Approximate measurement, tip of lateral angle broken.

TABLE 18.—Measurements (mm) of the cuneiform of North American glyptodonts

G. texanum G. anzonae

Characters F:AM USNM UMMP 95737 10536 38761 1. Maximum transverse diameter 37 50 45 2. Maximum anteroposterior diameter 26 35 33 3. Maximum proximodistal diameter 15 21 19

TABLE 19.—Measurements (mm) of the lunar of North American glyptodonts

G. texanum G arizonae

Characters F:AM USNM UMMP 38761

95737 10536 left right 1. Maximum transverse diameter 23 31 27 30 across proximal articular facet 2. Maximum transverse diameter of 25 35 33 33 posterior tubercle measured obliquely, medial to ventrolateral 3. Transverse diameter of constriction 17 25 - 28 1. Maximum anteroposterior diame­ 38 54 50 51 ter 5. Proximodistal diameter of posterior 21 32 27 25 tubercle 6. Minimum proximodistal diameter 13 16 - 19 of constriction 7. Maximum proximodistal diameter 17 22 25 through centers of proximal and distal articular facets 8. Maximum diameter scaphoid facel 32 41 30 32 NUMBER 40 223

TABLE 20.—Measurements (mm) of the scaphoid of North American glyptodonts

G. texanum G. arizonae

Characters F:AM USNM UMMP 95737 10536 38761 1. Maximum transverse diameter (measured slightly obliquely) 26 32 34 2. Maximum anteroposterior diameter, posterior tubercle to anterodistal 34 49 45 extremity 3. Maximum proximodistal diameter (radial facel to magnum facet) 18 24 28 4. Maximum transverse diameter, posterior tubercle 16 * 22 5. Maximum proximodistal diameter, posterior tubercle 14 * 15 * Tapers posteriorly, not knoblike, parameter unmeasurable.

TABLE 21.—Measurements (mm) of the trapezium of North American glyptodonts

G. texanum G. anzonae Characters K:AM USNM UMMP 95737 10536 38761

1. Maximum anteroposterior diameter 20 25 2. Maximum transverse diameter 16 21 * 3. Maximum proximodistal diameter 11 16 * * Not measurable, see text.

TABLE 22.—Measurements (mm) of the trapezoid of North American glyptodonts

G. texanum G. arizonae

Characters F:AM USNM UMMP 38761

95737 10536 right left 1. Maximum anteroposterior diameter, 24 33 31 29 anterolateral angle to posteromedial angle 2. Maximum proximomedial to disto­ 18 31 b b lateral diameter on anterior (dorsal) extremity 3. Maximum transverse diameter on an­ 16 22 38 terior (dorsal) extremity, distomedial to distolateral 4. Minimum proximodistal diameter 6 9 8 5. Minimum transverse diameter 10 16 b b b b 6. Proximal anteroposterior length, tra­ 13 22a. pezium facet 7. Maximum proximodistal diameter, b b 19 anterior face " Approximate measurement; anteroproximal tip broken. b Not applicable. 224 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 23.—Measurements (mm) of the magnum of North American glyptodonts

G. texanum G. arizonae Characters F:AM USNM UMMP 95737 10536 38761 1. Maximum anteroposterior diameter 31 40 39 2. Maximum proximodistal diameter through 16 20 center of articular facets 3. Maximum transverse diameter 22 28 29

TABLE 24.—Measurements (mm) of the unciform of North American glyptodonts

G. texanum G. anzonae

Characters F:AM USNM UMMP 38761

95737 10536 left right 1. Maximum proximodistal diameter, anterior face 13 18 23 26 2. Transverse diameter of anterior margin of proximal articular 24 23* 25 25 facet 3. Maximum anteroposterior diameter from posterior angle of lat- 24 40 29 31 erat articular facet to anterodistal angle of distal articular facet 4. Oblique diameter from posterior angle of lateral articular facet 26 35 36 36 to medial extremity on anterior border of proximal articular facet * Approximate measurement; anterior proximomedial tip broken.

TABLE 25.—Measurements (mm) of the metacarpal II of North American glyptodonts

G. texanum G anzonae Characters F:AM USNM UMMP 95737 10536 38761 1. Maximum proximodistal diameter between lat­ 46 47* 46 eral articular facets 2. Maximum proximodistal diameter between me­ 42 46 46 dial articular facets 3. Maximum transverse diameter, proximal artic­ 21 32 — ular facets 4. Maximum transverse diameter on posteroprox­ 25 ~ — imal extremity across proximomedial tubercle 5. Minimum transverse diameter of shaft 20 36* 31 6. Maximum transverse diameter across distal ar­ 25 32 32 ticular facets 7. Minimum anteroposterior diameter of shaft 22 28 25 8. Maximum anteroposterior diameter, proximal 31 39 36 extremity 9. Maximum anteroposterior diameter, distal ex­ 25 33 33 tremity Approximate measurement. NUMBER 40 225

TABLE 26.—Measurements (mm) of the metacarpal III of North American glyptodonts

G. texanum G. arizonae

Characters F:AM USNM UMMP 95737 10536 38761 1. Maximum proximodistal diameter between 36 40 34 lateral articular facets 2. Maximum proximodistal diameter between 35 38 35 medial articular facets 3. Maximum transverse diameter, proximal ar­ 28 39 38 ticular facets 4. Minimum transverse diameter of shaft 25 4(i 35 5. Maximum transverse diameter across distal 29 38 39 articular facets 6. Minimum anteroposterior diameter of shaft 23 32 29 7. Maximum anteroposterior diameter, proximal 28 37 36 extremity 8. Maximum anteroposterior diameter, distal ex­ 25 34 32 tremity

TABLE 27.—Measurements (mm) of the metacarpal IV of North American glyptodonts

G. texanum G. anzonae G. floridanum Characters F:AM USNM UMMP 38761 USNM

95737 10536 left right 6071 1. Maximum proximodistal length through 23 20 19 19 18 center 2. Maximum anteroposterior diameter mea­ 25 37 32 33 30 sured perpendicular to anterior face 3. Diameter from posterior tubercle to anter­ 29 37 34 - 31 odistal angle at center of angle 4. Maximum transverse diameter across dis- 29 37 33 35 27 toanterior lateral/medial margin 5. Maximum transverse diameter between lat­ 24 35 31 31* 27 eral and medial articular facets 6. Maximum proximodistal diameter, medial 15 16 - 15 10 articular facet 7. Maximum anteroposterior diameter, medial 11 18 - 17* 12 articular facet 8. Maximum proximodistal diameter, lateral 11 11, 10* - 11* articular facet 9. Maximum anteroposterior diameter, lateral 15* 15 17* - 16 articular facet 10. Maximum anteroposterior diameter, proxi­ 22 32 27 - 25 mal articular facet * Approximate measurement. 226 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 28. — Measurements (mm) of ihe metacarpal V of North American glyptodonts

G. texanum G. anzonae Characters F:AM USNM 95737 10536 1. Maximum proximodistal diameter on medial border 12 (see text) 2. Maximum proximodistal diameter on lateral border 9 3. Maximum proximodistal diameter through center of 8 articular facets 4. Maximum anteroposterior diameter 13 5. Maximum transverse diameter 20

TABLE 29.—Measurements (mm) of the phalanges I and II, digit II, manus, of North American glyptodonts

Phalanx I Phal inx II

G. texanum G. arizonae G. texanum G an zonae G. floridanum Characters F:AM USNM UMMP F:AM USNM UMMP USNM 95737 10536 38761 95737 10536 38761 6071 1. Maximum anteroposterior di­ 22 27 - 18 27 22 25 ameter of proximal articular facet, lateral side 2. Maximum anteroposterior di­ 19 28 30 18 26 25 25 ameter, distal articular facet, lateral side 3. Anteroposterior diameter, prox­ 23 16 22 26 23 imal extremity through carina to anterior tubercle 4. Maximum diameter posterolat­ 29 27 37 39 36 eral tubercle to anterior proxi­ momedial angle 5. Maximum transverse diameter 24 36 37 27 36 38 35 across posterior tubercles 6. Maximum transverse diameter 27 37 37 27 36 39 31 through center of articulation proximal facet TABLE 30.—Measurements (mm) of the ungual phalanx, digit II, manus, of North American glyptodonts

G. texanum G. arizonae G. floridanum

Characters F:AM USNM UMMP USNM 95737 10536 38761 6071 1. Maximum diameter, proximoanterior prominence to distal ex­ 56 76 74 68 tremity 2. Maximum diameter, posteroproximal tendinal groove to distal 49 65 66 62 extremity 3. Inside diameter between subungual foramina 14 18 20 21 4. Maximum transverse diameter across posteroproximal margin 25 36 38 35 through sesamoid facets 5. Maximum anteroposterior (palmar-plantar) diameter, anterior 25 36 35 35 border to subungual base 6. Maximum oblique transverse diameter, subungual base 18 43 43 39 7. Diameter from sesamoid tendinal groove to distal border, subun­ 27 33 35 29 gual base 8. Diameter from distal border of subungual base to distal extremity 23 34 34 33 of claw process

TABLE 31.—Measurements (mm) of the phalanges I and II, digit III, manus, of North American glyptodonts

Phalanx I Phalanx II

G. texanum G. anzonae G. texanum G. arizonae Characters F:AM USNM UMMP F:AM USNM UMMP 95737 10536° 38761 95737 10536 38761 1. Maximum anteroposterior diameter of 26 34 35b 23 26 25b proximal articular facet, lateral side 2. Maximum anteroposterior diameter of 20 26 24b 20 26 24b distal articular facet, lateral side 3. Anteroposterior diameter, proximal ex­ 22 28 - 16 20 - tremity through center 4. Maximum diameter, posterolateral tu­ 34 44 45 33 47 45 bercle to anterior proximomedial angle 5. Maximum anteroposterior diameter, 19 28 28 20 26 26 distal articular facet, medial rocker 6. Maximum transverse diameter through 29 41 36 28 41 38 center of articulation, proximal facet 7. Maximum transverse diameter, distal 27 39 36 25 37 34b articular facet 8. Maximum proximodistal diameter be­ 12 14 13 12 14 16 tween articular facets, anterolateral margin 9. Maximum proximodistal diameter be­ 10 11 10 12 15 17 tween articular facets, medial margin 10. Central proximodistal diameter be­ 10 13 — 11 14 - tween articular facets 11. Maximum anteroposterior diameter, 23 28 - 20 23 17b proximal articular facet, medial side a Extremely worn, pathologic, measurements estimated. b Approximate measurement. 228 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 32.—Measurements (mm) of the ungual phalanx, digit III, manus, of North American glyptodonts

G. texanum G. arizonae G. floridanum Characters F:AM TMM USNM UMMP USNM 95737 40664-109 10536 38761 6071 1. Maximum diameter proximoanterior promi­ 63 81 77 51* 72 nence to distal extremity 2. Maximum diameter posteroproximal tendinal 48 68 63 - 59 groove to distal extremity 3. Inside diameter between subungual foramina 15 22 21 - 22 4. Maximum transverse diameter across postero­ 27 41) 43 40 38 proximal margin through sesamoid facets 5. Maximum anteroposterior (palmar-plantar) 26 35 36 38* 34 diameter, anterior border to subungual base 6. Diameter from sesamoid tendinal groove to 25 31 34 - 27 distal border, subungual base 7. Diameter from distal border of subungual base 25 38 34 - 34 to distal extremity of claw process 8. Transverse diameter through deepest portion 29 43 43 40* 40 proximal articular facet Pathologic, measurements approximate.

TABLE 33.—Measurements (mm) of the phalanges I and II, digit IV, manus, of North American glyptodonts

Phalanx I Phalanx II

G. texanum G. anzonae G. texanum G. an zonae Characters F:AM USNM UMMP F:AM USNM UMMP 95737 10536 38761 95737 10536 38761 1. Maximum anteroposterior diameter 23 - 26 - - _ 2. Anteroposterior diameter through ten­ 18 24 23 16 22 24 dinal groove 3. Maximum transverse diameter 25 36 32 24 35 30 4. Maximum proximodistal thickness 7 - - 9 11 - through center of articular facets 5. Maximum proximodistal thickness, 8 10 9 10 12 9 medial margin articular facets 6. Maximum proximodistal thickness, lat­ 8 9 9 9 12 12 eral margin articular facets 7. Oblique diameter, posterolateral tuber­ 27 38 33 27 36 31 cle tip to anteromedial margin 8. Oblique diameter, posteromedial tu­ 24 35 31 24 34 30 bercle tip to anterolateral margin 9. Maximum anteroposterior diameter, - 29 25 20 26 25 medial side 10. Maximum anteroposterior diameter, - 32 20* 18 26 22 lateral side * Approximate measurement. NUMBER 40 229

TABLE 34.—Measurements (mm) of the ungual phalanx, digit IV, manus, of North American glyptodonts

G. texanum G. floridanum Characters F:AM USNM 95737 6071 1. Maximum proximodistal length from proximoanterior 53 57 prominence to distal extremity 2. Maximum proximodistal length from posteroproximal ten­ 44 52 dinal groove to distal extremity 3. Inside diameter subungual foramina 12 18 4. Maximum transverse diameter, proximal extremity 24 33 5. Maximum transverse diameter, sesamoid facets 19 20 6. Diameter from sesamoid tendinal groove to distal border, 25 26 subungual base 7. Diameter from subungual base, distal border, to distal 20 27 extremity, claw process 8. Maximum diameter anteroproximal prominence to distal 37 38 border, subungual base 9. Maximum anteroposterior diameter from anteromedial - 28 border to subungual base

TABLE 35.—Measurements (mm) of the phalanx I, digit V, manus, of North American glyptodonts

G. texanum G.. anzonae Characters F:AM USNM 95737 10536 1. Maximum anteroposterior diameter 11 (see text) 2. Proximodistal diameter through center of articular 4 facets 3. Maximum transverse diameter 13

TABLE 36.—Measurements (mm) of the ungual phalanx, digit V, manus, of North American glyptodonts

G. texanum G. anzonae Characters F:AM USNM 95737 10536 1. Maximum proximodistal length 32 43 2. Diameter from anterior border articular facet to su­ 21 34 bungual prominence 3. Diameter from subungual prominence to distal extrem­ 24 36 ity 4. Maximum transverse diameter 13 25 230 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 37.—Measurements (mm) of the digital sesamoid bones, manus, of North American glyptodonts (upper number = F:AM 95737, G. texanum; lower number = USNM 10536, G. arizonae)

Maximum M iximum Maximum Characters transverse proximodistal anteroposterior diameter diameter diameter

Digit II 1. Medial proximal sesamoid 11 7 (. 16 17 14 2. Lateral proximal sesamoid 11 111 10 18 14 14 3. Distal sesamoid 20 1(1 15 34 16 2(1

Digit III 1. Medial proximal sesamoid 9 6 8 15 12 11 2. Lateral proximal sesamoid 8 6 7 3. Distal sesamoid 21 9 13 34 19 20

Digit IV 1. Distal sesamoid 18 8 ')

Digit V 1. Distal sesamoid 7 6 4 16 14 10 1 Diameters arbitrarily defined as three mutually perpendicular axes. Measurements are for G. texanum F:AM 95737; element unknown for G. arizonae. NUMBER 40 231

TABLE 38.—Measurements (mm) of the pelvis of North American glyptodonts

G. texanum G. anzonae G. cylindricum ( J. floridanum Characters AMNH F:AM AMNH AMNH UF/FGS TMM USNM 10704° 95737 21808 15548 6643 30967 6071 1. Anteroposterior length of 490 500b 660b crest of sacral arch along curvature from anterior il­ iosacral vertebra to poste­ rior ischiosacral vertebra 2. Estimated anteroposterior 310b - - 270" - 360b length, lumbar tube 3. Minimum transverse diam­ 75/75 75/- -/127b 115/109 110/119 112/118 eter, ilium shaft 4. Anteroposterior diameter 35/34 30/- -/48b 38b/38b 42/37 39/36 (thickness) of ilium at posi­ tion of minimum transverse diameter 5. Maximum diameter (long 81b/80b 86/- 100/105 103b/102 89/89 -/89 -/91 axis) acetabular fossa 6. Minimum diameter, aceta­ 63/63 67/-b 85/84 85/85 74/- -/70b bular fossa, approximated by extension of circumfer­ ence through external notch formed by lateral acetabu­ lar fossa 7. Minimum inferior-superior 50/48 50/48 76/65 52/56 62/- 62/66 diameter, ischial neck 8. Minimum transverse diam­ 19/19 16/17 39b/32 30b/29 27/- 20/21 eter, ischial neck 9. Anteroposterior diameter of 230/220 -/171b crest of ischium measured along curvature 10. Transverse diameter, anter- 270/260 250b/- 270b/280 -/270b osuperior angle ischiac crest to neural spine, anterior is­ chiosacral vertebra (penul­ timate vertebra); i.e., "half- width" of pelvis at ischiosa­ cral vertebra 11. Minimum anteroposterior 9/9 9/8 -/12b 27/- diameter of transverse process of anterior ischiosa­ cral vertebra 232 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 38.—Continued G texanum G. arizonae G. cylindricum G. floridanum

Characters AMNH F:AM AMNH AMNH UF/FGS TMM USNM 10704" 95737 21808 15548 6643 30967 6071 12. Minimum anteroposterior 29/24 31/33 47/- 44/42 43/- 36/38 33/30 diameter of transverse pro­ cess of posterior ischiosacral i vertebra 13. Anteroposterior chord from 620 840b thoracolumbar facet to sac­ rocaudal facet b b b b 14. Transverse diameter, iliac 220/220 215/- 230 /230 230 /230 250b/- 280/290 crest measured along cur­ vature 15. Minimum transverse diam­ 11/- 5/6 -/6 14/15 11/- - - eter, pubic shaft 16. Minimum anteroposterior - 12/12 -/20 25/24 14/- - - diameter, pubic shaft 17. Transverse diameter be­ 370 460 tween posterior angles, is­ chiac crest 18. Transverse diameter be­ 460 440 tween anterior angles, is­ chiac crest 19. Transverse diameter be­ 210 190 tween pubes measured from inner surface of pubes at midshaft

1 Additional measurements in text. 'Approximate measurement. NUMBER 40 233

TABLE 39.—Measurements (mm) of the femur of North American glyptodonts

G. texanum G. floridanum G. arizonae

Characters F:AM 95737 USNM USNM AMNH UMMP UMMP UMMP

left right 6071 10536 21808 46231 46233 46376 1. Maximum proximodistal di­ 326 421 453 452 488 ameter head to medial con­ dyle 2. Maximum proximodistal di­ 331 443 484 527 ameter greater trochanter to lateral condyle 3. Oblique proximodistal di­ 344 454 523 542 ameter greater trochanter to medial condyle 4. Transverse diameter head 62 62 67 80 80 83 5. Anteroposterior diameter 74 74 84 95 97 103 head 6. Transverse diameter medial 156 219 244 248 border of head to lateral ex­ tremity greater trochanter 7. Proximal maximum trans­ 165 237 271 257 verse diameter between lesser and greater trochanters 8. Anteroposterior diameter 41 42 52 55 63 61 66 64" midshaft 9. Minimum transverse diame­ 65 62 81 90* 89 88* 101 103* ter midshaft 10. Maximum transverse diam­ - 46 50 65 53 60* 66 63 eter medial condyle 11. Maximum transverse diam­ - - 53 51 61* 63 67 66 eter lateral condyle 12. Length supinator ridge 155 202 227 _ _ 250 195 (third trochanter) to distal extremity lateral condyle 13. Maximum transverse diam­ 101 122 132 135 161 145 eter distal articular facet * Approximate measurement.

TABLE 40.—Measurements (mm) of the patella of North American glyptodonts

G. texanum G. arizonae G floridanum Characters F:AM USNM USNM 95737 10536 6071 1. Maximum proximodistal diameter 81 100 84 2. Maximum transverse diameter 55 83 72 3. Maximum anteroposterior (craniocaudal) 41 56 46 diameter 234 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 41.—Measurements (mm) of the tibiofibula of North American glyptodonts

G. texanum G arizonae G. floridanum

F:AM JWT USNM UMMP UMMP UMMP MU USNM TMM Characters 95737" 1723 10536 46232 46231 33524 2670 6071 31141- 19 left right 1. Maximum transverse di­ 105 107 111 135 146 _ 151 102 119 ameter, proximal ex­ tremity through center of articular facet b 2. Maximum anteroposte­ 87 85b 114b - - - 124 - 99 rior diameter, proximal extremity through tibial facet (excluding tibial tuberosity) 3. Minimum transverse di­ 78 78 99 112 112 140b 130 - 112 ameter of "midshaft," exclusive of tibial crest 4. Minimum anteroposte­ 47 46 49 51 66 - 61 - 56 rior diameter, fibular shaft 5. Minimum anteroposte­ 43 42 50 61 70 74b 78 - 48 60 rior diameter on medial surface of distal extrem­ ity of tibial shaft b 6. Maximum transverse di­ 105 105 116 133 131 151 140 - 118 ameter, distal extremity through malleoli 7. Maximum anteroposte­ 67 67 _ 78 _ 86 82 _ _ rior diameter, distal ex­ tremity through fibular facet b 8. Maximum anteroposte­ 50b 51 47b 66 73" 77 75 - 59 rior diameter of fibular facet, distal extremity 9. Maximum transverse di­ 65 64 69 84 94 93 88b 83 ameter, distal articular facet b b b b b b 10. Proximodistal tibial I16b 117 200 235 230 230 220 - 220 225 length measured from posterior margin of proximal articular facet to posterior malleolus 11. Fibular length measured 181 181 220b 266 280b - 250b - - from proximolateral in­ tercondylar prominence through lateral malleolus "Juvenile. b Approximate measurement. NUMBER 40 235

TABLE 42.—Measurements (mm) of the astragalus of North American glyptodonts

G. texanum G. arizonae G. floridanum

Characters F:AM 95737 F:AM USNM 10536 UMMP UMMP USNM

left right 59586 left right 46231 38761 6071 1. Maximum transverse di­ 59 59 64 75 77 85 80* 74 ameter, trochlear facet 2. Anteroposterior diameter, 42 42 47* 64 74 65 medial rocker, trochlear facet 3. Anteroposterior diameter, 50 50 51 65* 64 58* 59 lateral rocker, trochlear facet 4. Anteroposterior diameter, 43 - .42 62 60 70 - - navicular facet 5. Transverse diameter, na­ 42 44 - - 51 - 49 46 vicular facet 6. Maximum diameter, lat­ - 48 48* 54* 57 57 - 42 eral calcaneal facet 7. Oblique diameter, medial 69 69 69 87 105 81* calcaneal tuberosity to na­ vicular tuberosity * Approximate measurement. 236 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 43.—Measurements (mm) of the calcaneum of North American glyptodonts

G. texanum G. anzonae G. floridanum Characters F:AM 95737 JWT USNM UMMP UMMP USNM

left right 1712 10536 46231 38761 6071 1. Maximum diameter, astra­ 45 45 45 55 61* 55* 50 galar facet (long axis) 2. Width, astragalar facet (short - 28 27 33 34* 33* 31 axis) 3. Maximum length, sustenta­ - 31 27 40 - - 38 cular facet 4. Maximum width, sustenta­ - 20 17 25 - - 25 cular facet along convexity 5. Anteroposterior diameter, - 27 28 33 - - 35 cuboid facet 6. Transverse diameter, cuboid - 25 18 30 - - 30 facet (through constriction) 7. Diameter from astragalar - 47 49 61 64 64 55 facet, perpendicular through inferior tuberosity 8. Transverse diameter, susten­ 73 74 73 92 - - - tacular facet through cuboid facet to posterolateral tuber­ cle 9. Superior-inferior diameter, - 45 - 57 59 60* 50 astragalar facet to neck 10. Superior-inferior diameter, 47 45 - 54 55* 52* 45 tuber calcis above astragalar facet 11. Maximum proximodistal di­ - 99 - 137 - - 133 ameter * Approximate measurement. TABLE 44.—Measurements (mm) of the navicular of North American glyptodonts

G. texanum G. anzonae

Characters F:AM USNM UMMP UMMP UMMP UMMP 95737 10536 46231 38761 46332 46331 1. Maximum proximodistal diameter, 22 31 _ 30* 36 27 posterior border astragalar facel to posterior border middle cuneiform facet 2. Minimum proximodistal diameter, 8 13 14 10 15 13 anterior border astragalar facet to anterior border middle cuneiform facet 3. Minimum proximodistal diameter, 12 14 16 16 19 13 lateral angle astragalar facet to ex­ ternal corner ectocuneiform facet 4. Maximum anteroposterior diam­ 86 117* - - 132 - eter, anterior angle ectocuneiform facet to lateral posterior tuberosity 5. Anteroposterior diameter, astraga­ 44 59 67 58 68 55 lar facet 6. Maximum transverse diameter, as­ 51 57 - 55 60* 56 tragalar facet 7. Maximum transverse diameter, me­ 65 82 - 91 101 80 dial tuberosity to lateral tubercle at external boundary between cuboid facet and ectocuneiform facet 8. Maximum diameter (long axis), cu­ 44 49 - 50 65 - boid facet 9. Maximum transverse diameter, ec­ 24 28 60* 47 59 45 tocuneiform facet 10. Maximum anteroposterior diam­ 48 60 - 64 69 59 eter, middle cuneiform facet 11. Maximum transverse diameter, 22 30 29* 26 28* 26* middle cuneiform facet 12. Maximum transverse diameter, in­ 19 24 - 20* 27* - ternal cuneiform facet 13. Maximum anteroposterior diam­ 40 53 - 48 59 - eter, internal posterior tubercle * Approximate measurement.

TABLE 45.—Measurements (mm) of the cuboid of North American glyptodonts

G. texanum G. arizonae Characters F:AM USNM UMMP UMMP 95737 10536 46231 38761 1. Maximum anteroposterior diameter 71 63 81 78 2. Maximum transverse diameter, calcaneal facet 29 43 49 42 3. Maximum anteroposterior diameter, calcaneal facet 31 40 59 - 4. Maximum proximodistal diameter, calcaneal facet 29 37 49* 44* to distal articular facet 5. Maximum transverse diameter, posterior tubercle 28 36 35 38 Approximate measurement. TABLE 46.—Measurements (mm) of the ectocuneiform of North American glyptodonts

G. texanum G. anzonae Characters F:AM USNM UMMP UMMP 95737 10536 46231 38761 1. Proximodistal diameter at anterolateral angle 17 18 22 ' '• 22 2. Anteroposterior diameter, proximal facet from anterolatera angle to 30 35 34 33 posteromedial concavity 3. Anteroposterior diameter distal facet from anterolateral angle to 25 26 40 — posteromedial concavity 4. Anteroposterior diameter, anteromedial angle to posterior angle 44 57 75 61

TABLE 47.—Measurements (mm) of the middle cuneiform of North American glyptodonts

G texanum G. arizonae

Characters F:AM USNM UMMP UMMP 95737 10536 46231 38761 1. Maximum proximodistal diameter at 18 20 25 21 anteromedial angle of articular facets 2. Minimum proximodistal diameter be­ 7 10 9 7* tween posterior regions of articular facets 3. Maximum transverse diameter, ante­ 22 30 35 29 rior border 4. Maximum anteroposterior diameter 48 63 65 57 Approximate measurement.

TABLE 48.—Measurements (mm) of the internal cuneiform of North American glyptodonts

G. texanum G. anzonae

Characters F:AM USNM UMMP UMMP 95737 10536 46231 38761 1. Maximum anteroposterior di­ 35 41 52* 40* ameter 2. Maximum transverse diameter 19 24 25* 30* 3. Maximum proximodistal diam­ 24 30 34* 23* eter * Approximate measurement.

TABLE 49.—Measurements (mm) of the metatarsal I of North American glyptodonts

G. texanum Characters F:AM USNM UMMP UMMP 95737 10536 38761 46231 1. Maximum anteroposterior di­ 28 34 36 43 ameter 2. Maximum transverse diameter 20 29 25 28 3. Maximum proximodistal diam­ 18 32 24* 25 eter * Approximate measurement. NUMBER 40 239

TABLE 50.—Measurements (mm) of the metatarsal II of North American glyptodonts

G. texanum G. anzonae

Characters F:AM UMMP UMMP UMMP USNM 95737 38761 46231 46234 10536 1. Maximum anteroposterior diameter, proximal facet 42 50 55 51 52 2. Maximum transverse diameter, proximal facet 24 27 33 32 29 3. Maximum anteroposterior diameter, distal facet 36 26 38 35 41 4. Maximum transverse diameter, distal facet 25 31 27 34 37 5. Maximum proximodistal diameter 42 48 52* 47 52 Approximate measurement.

TABLE 51.—Measurements (mm) of the metatarsal III of North American glyptodonts

G. texanum G. arizonae C.Jloridanum Characters F:AM USNM UMMP UMMP UMMP UMMP usr >JM6071 95737 10536 38761 46231 46330 46329 left right 1. Maximum proximodistal 42 58 51* 63* 60 60* 48 47 diameter from posterolat­ eral angle proximal facet to posterior border distal facet 2. Minimum proximodistal di­ 20 28 25 27* 32 26 23 23 ameter, anteromedial angle proximal facet to anterior border distal facet 3. Anteroposterior diameter, 17 25 23* 26 30 30 22 21 metatarsal II articular facet 4. Anteroposterior diameter, 19 25 27 32 37 31 25 metatarsal IV articular facet 5. Maximum transverse di­ 33 53 46 49 56 46 45 ameter, cuneiform articular facet 6. Maximum transverse di­ 43 58 56 61 60* 62 51 51 ameter, proximal extremity 7. Maximum transverse di­ 33 45 41 - 42 49* 40 42 ameter, distal extremity 8. Maximum anteroposterior 36 45 47* - 52* 52 41 41 diameter, distal facet (to tip of sesamoid ridge) 9. Maximum transverse di­ 22 32 29* - 34 36 30 31 ameter, sesamoid articular facets 10. Maximum proximodistal 24 29 27* 29 27 26 diameter, sesamoid articu­ lar facets 11. Anteroposterior diameter, 41 58 56 60 61* 59 52 52 proximal extremity sesa­ moid facets to anterior bor­ der proximal facet Approximate measurement. 240 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 52.—Measurements (mm) of the metatarsal IV of North American glyptodonts

G texanum G. anzonae G. floridanum Characters F:AM USNM UMMP UMMP USNM 95737 10536 46231 38761 6071 1. Maximum proximodistal diameter, posterolateral cor­ 34 41 42 40 38 ner, proximal facet to posterolateral corner distal facet 2. Maximum proximodistal diameter between anterior 23 35 38 34 29 borders of proximal and distal facets at anterolateral angle 3. Maximum transverse diameter, cuneiform facet 19 36 33 31 34 4. Maximum anteroposterior diameter, cuneiform facet 40 41 44 47 42 5. Maximum posterolateral/anteromedial diameter, 24 31 28* 32 26 metatarsal III articular facet 6. Maximum anterolateral/posteromedial diameter, 21 25 25 24 20 metatarsal III articular facet 7. Maximum anteroposterior diameter, metatarsal V ar­ 13 14 20 20 25* ticular facet 8. Maximum anteroposterior diameter, distal articular 30 38 - , I — 35 facet 9. Maximum transverse diameter, distal articular facet 28 39 38* 36 36 10. Maximum proximodistal diameter, sesamoid facet 15 19 19* 17 22 11. Maximum oblique diameter, posterolateral tubercle 47 60 67 65 57 proximal extremity to anteromedial tubercle distal ex­ tremity 12. Maximum transverse diameter, outer border meta­ 32 46 50 46 45 tarsal III facet to metatarsal V facet Approximate measurement.

TABLE 53.—Measurements (mm) of the metatarsal V of North American glyptodonts

G. texanum G. arizonae

Characters F:AM USNM UMMP UMMP 95737 10536 46231 38761 1. Maximum anteroposterior di­ 30 40 40 43 ameter 2. Maximum transverse diameter 20 33 35 34 3. Maximum proximodistal diam­ 18 25 25 29 eter NUMBER 40 241

TABLE 54.—Measurements (mm) of the metatarsal sesamoids of North American glyptodonts

G. texanum G. arizonae

Characters F:AM USNM UMMP UMMP 95737 10536 38761 46231 J Sesamoid Bone, Metatarsal I

1. Maximum anteroposterior diameter 11 16 23 - 2. Maximum transverse diameter 13 23 18 - 3. Maximum proximodistal diameter 24 36 31

Medial Sesamoid Bone, Metatarsal II 1. Maximum anteroposterior diameter 24 - 17 30 2. Maximum transverse diameter 9 - 10 13 3. Maximum proximodistal diameter 15 18 27

Lateral Sesamoid Bone, Metatarsal II "• 1. Maximum anteroposterior diameter 24 29 29 30 2. Maximum transverse diameter 12 17 16 16 3. Maximum proximodistal diameter 26 31 43 ' 34

Medial Sesamoid Bone, Metatarsal III 1. Maximum length articular facet 26 30 28 - 2. Maximum transverse diameter articular 11 16 14 • , - facet 3. Maximum proximodistal diameter 25 30 31

Lateral Sesamoid Bone, Metatarsal III " 1. Maximum length articular facet 27 29 29* 35 2. Maximum transverse diameter articular 12 17 15 19 facet 3. Maximum proximodistal diameter 25 30 31 33

Medial Sesamoid Bone, Metatarsal IV 1. Maximum length articular facet 14 " 17 14 17 2. Maximum transverse diameter 8 11 11 15 3. Maximum anteroposterior diameter 21 24 29 33

Lateral Sesamoid Bone, Metatarsal IV 1. Maximum length articular facet 19 - 28 -

2. Maximum transverse diameter 12 -• 17 ' - 3. Maximum anteroposterior diameter 19 38

: • 1 Sesamoid Bone, Metatarsal V 1. Maximum anteroposterior diameter 7 - - - 2. Maximum transverse diameter 14 - - - 3. Maximum proximodistal diameter 12 - - - * Approximate measurement. TABLE 55.—Measurements (mm) of the phalanges I and II, digit I, pes, of North American glyptodonts

G. texanum G. arizonae

Characters F:AM USNM UMMP UMMP 95737 10536 46231 38761

Phalanx I 1. Maximum anteroposterior 21 37 40 27 diameter 2. Maximum transverse 17 32 25 23 diameter 3. Maximum proximodistal 17 22 20 22 diameter

Phalanx II 1. Maximum anteroposterior 17 22 56 52 diameter 2. Maximum transverse 29 50 23 24 diameter 3. Maximum proximodistal 34 60 58 65 diameter

TABLE 56.—Measurements (mm) of the phalanges I and II, digit II, pes, of North American glyptodonts

G. texanum G. anzonae

Characters F:AM UMMP UMMP USNM 95737 38761 46231 10536 Phalanx I 1. Maximum anteroposterior di­ 30 29 .28* 40 ameter, proximal facet 2. Maximum transverse diameter, 25 31 42 34 proximal facet 3. Maximum anteroposterior di­ 24 34* 29* 32 ameter, distal facet 4. Maximum transverse diameter, 26 35 47 36 distal facet 5. Maximum proximodistal diam­ 18 22 31 23 eter

Phalanx II 1. Maximum anteroposterior di­ 21 - - 29 ameter, proximal facet 2. Maximum transverse diameter, 27 34 47 36 proximal facet 3. Maximum anteroposterior di­ 21 ~ - 27 ameter, distal facet 4. Maximum transverse diameter, 25 39 45 36 distal facet 5. Maximum proximodistal diam­ 13 23 20* 20 eter * Approximate measurement. TABLE 57.—Measurements (mm) of the ungual phalanx, digit II, pes, of North American glyptodonts

G. texanum G. anzonae Characters F:AM USNM UMMP 95737 10536 38761 1. Maximum transverse diameter 47 66 50 2. Maximum transverse diameter, articular facet 30 39 36 3. Maximum anteroposterior diameter 23 28 49 4. Maximum anteroposterior diameter, articular 19 26 30* facet 5. Maximum proximodistal diameter 47 66 72 * Approximate measurement.

TABLE 58.—Measurements (mm) of the phalanges I and II, digit III, pes, of North American glyptodonts

G. texanum G. arizonae G. floridanum Characters F:AM USNM UMMP USNM 95737 10536 38761 6071

Phalanx I 1. Maximum transverse diameter, proximal facet 29 41 42 31 2. Maximum anteroposterior diameter, proximal extremity 36 41 44 34 3. Maximum transverse diameter, distal facet 32 42 43 31 4. Maximum anteroposterior diameter, distal extremity through 36 45 48 38 tubercles 5. Maximum proximodistal diameter, anterior border 17 24 23 20 6. Maximum proximodistal diameter, posterior border 11 16 11* 11 Phalanx II 1. Maximum transverse diameter, proximal facet 31 44 42 2. Maximum anteroposterior diameter, proximal extremity 24 32 36 3. Maximum transverse diameter, distal facet 31 43 42 4. Maximum anteroposterior diameter, distal extremity through 21 29 27 tubercles 5. Minimum proximodistal diameter, anterior border 4 7 3 6. Maximum proximodistal diameter, posterior border 14 23 22 * Approximate measurement.

TABLE 59.—Measurements (mm) of the ungual phalanx, digit III, pes, of North American glyptodonts

G. texanum G. arizonae G. floridanum Characters F:AM USNM USNM 95737 10536 6071 1. Maximum transverse diameter 40 70 62 2. Maximum transverse diameter, articular 32 48 43 facet 3. Maximum anteroposterior diameter 35 46 44 4. Maximum anteroposterior diameter, articu­ 22 27 27 lar facet 5. Maximum proximodistal diameter 48 58 62 6. Minimum inside diameter between subun­ 23 39 34 gual foramina 244 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 60.—Measurements (mm) of the phalanges I and II, digit IV, pes, of North American glyptodonts

G. texanum G. arizonae

Characters F:AM USNM UMMP UMMP 95737 10536 46231 38761

Phalanx I 1. Proximodistal diameter, anteromedial border 9 14 14 13 2. Maximum transverse diameter, proximal facet 23 36 41 37 3. Maximum anteroposterior diameter, proximal facet 30 36 - - 4. Maximum transverse diameter, distal facet 26 42 39 35 5. Maximum anteroposterior diameter, distal facet 24 - - - 6. Maximum anteroposterior diameter through tubercles 32 39 42 38

Phalanx II 1. Minimum proximodistal diameter, anterior border 4 5 6 5 2. Maximum transverse diameter, proximal facet 25 37 37 a 3. Maximum anteroposterior diameter, proximal facet 21 28 - - 4. Maximum transverse diameter, distal facet 27 39 35 35 5. Maximum anteroposterior diameter, distal facet 21 24 - - 6. Maximum anteroposterior diameter through tubercles 23 28 31 30 7. Maximum proximodistal diameter, lateral side 11 19 17 16 8. Maximum proximodistal diameter, medial side 9 19 19 18

TABLE 61.—Measurements (mm) of the ungual phalanx, digit IV, pes, of North American glyptodonts

G. texanum G. anzonae G. floridanum Characters F:AM USNM UMMP UMMP USNM 95737 10536 46231* 38761 6071 1. Maximum proximodistal di­ 48 62 57 65 63 ameter, medial border 2. Maximum transverse diam­ 45 59 56 58 65 eter, proximal extremity 3. Maximum transverse diam­ 27 36 32 33 39 eter, proximal facet 4. Maximum anteroposterior 18 22 23 25 26 diameter, proximal facet 5. Maximum anteroposterior 38 46 52 46 45 diameter, ungual hood * Pathologic: degenerative resorption. NUMBER 40 245

TABLE 62.—Measurements (mm) of the phalanges I and II, digit V, pes, of North American glyptodonts

G. texanum G. arizonae

Characters F:AM USNM UMMP UMMP 95737 10536 46231 38761

Phalanx I 1. Maximum anteroposterior diameter 19 23 34 32 2. Maximum transverse diameter 25 30 28 28 3. Maximum proximodistal diameter 8 10 11 10

Phalanx II 1. Maximum anteroposterior diameter 16 21 21 23 2. Maximum transverse diameter 20 28 28 28 3. Maximum proximodistal diameter 8 16 14 15

TABLE 63.—Measurements (mm) of the ungual phalanx, digit V, pes, of North American glyptodonts

G. texanum G. anzonae G. floridanum Characters USNM 6071 F:AM USNM UMMP UMMP 95737 10536 46231 38761 left right 1. Maximum anteroposte­ 37 57 61 53 48 39 rior diameter* 2. Maximum transverse di­ 22 40 52 45 40 36 ameter* 3. Maximum proximodistal 31 56 59 50 54 53 diameter * Anteroposterior and transverse diameters are ontogenetic references corresponding to transverse and anteroposterior diameters in anatomical reference, respectively. 246 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 64.—Measurements (mm) of the distal digital sesamoid bones, pes, of North American glyptodonts

G. texanum G. anzonae

Characters F:AM USNM UMMP UMMP 95737 10536 38761 46231

Digit I 11_- 1. Maximum anteroposterior diameter 9 r"'" 29* 14 2. Maximum transverse diameter 13 - 28 17 3. Maximum proximodistal diameter 11 23 20

Digit II 1. Maximum anteroposterior diameter 13 - 18* - 2. Maximum transverse diameter 24 - 28 - 3. Maximum proximodistal diameter 13 19

Digit III 1. Maximum anteroposterior diameter 14 - 24 - 2. Maximum transverse diameter 29 - 44 - 3. Maximum proximodistal diameter 11 13

Digit IV 1. Maximum anteroposterior diameter 12 18 18 16 2. Maximum transverse diameter 23 - 39 39 3. Maximum proximodistal diameter 12 12 13* 17

Digit V 1. Maximum anteroposterior diameter 7 15 18 - 2. Maximum transverse diameter 5 27 24 - 3. Maximum proximodistal diameter 12 - 14 - * Approximate measurement. NUMBER 40 247

TABLE 65.—Measurements (mm) of the caudal vertebrae of Glyptotherium texanum holotype AMNH 10704

Characters/Axes 1 2 3 4 5 6 7 8 9 10 11" 12" 13" 1. Anteroposterior diameter through 62 54 57 60 63 64 65 66 64 - 56 53 centrum at base (a-axis) 2. Maximum transverse diameter, cen­ 62 57 53 56 50 47 45 43 37 33 - - - trum anterior extremity (b-axis) 3. Maximum dorsoventral diameter, 43 45 50 50 49 46 43 41 39 35 - - - centrum anterior extremity (c-axis) 4. Maximum transverse diameter, cen­ 68 67 64 57 52 50 48 45 41 - - - trum posterior extremity (d-axis) 5. Maximum dorsoventral diameter, 53 54 54 54 48 47 46 43 40 centrum posterior extremity (e-axis) 6. Maximum transverse diameter be­ 318 287 261" 202b 164 135b 116b - 80 63 - - - tween angles of transverse processes (f-axis) 7. Maximum transverse diameter, an­ 52 77 80 82 71 72 56 50 50 31 26 19 terior xenarthral processes (g-axis) 8. Maximum transverse diameter, pos­ 35 38 39 33 27 20 11 9 terior xenarthral processes (h-axis) 9. Maximum dorsoventral diameter, - - - - - 53 39 27 26 - - - chevron (i-axis) 10. Anteroposterior diameter, chevron - - - - - 28 30 35 36 - - -:- through articular facet (j-axis) 11 Anteroposterior diameter, xenarthral 55 56 57 57 67 73 70 70 58 - - '- arch, midline (k-axis) " Centrum reconstructed. Tips restored, measurements include restoration. 248 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 66.—Measurements (mm) of the caudal vertebrae of Glyptotherium texanum, F:AM 95737 (axes as defined in Table 65)

Axes 1 2 3 4 5 6 7 8 9 10 11 12 13 . 57 58 64 64 65 64 62 60 59 58 1°3 I 2. b-axis 62 54 56 49 45 42 38 34 28 3. c-axis 44 - - 45 44 42 42 38 36 31 25 - 4. d-axis 64 - - 57 53 49 45 41 35 29 - - 5. e-axis 50 - - 46 45 45 43 41 35 29 - - 6. f-axis 267 236 212 169 151 124 97 80 66 48 - - 7. g-axis 64 69 71 73 68 63* 48 43 35 29 - - 8. h-axis 43 41 40 33 27 23 9. i-axis .rr 109 91 79 67 51 38 37 27 30 30 9 - 10. j-axis - 20 25 21 22 24 26 31 36 21 21 26 - 11. k-axis 48 46 53 59 - 72* 61 * Approximate measurement.

TABLE 67.—Measurements (mm) of the caudal vertebrae of Glyptotherium arizonae, UMMP 34826 (axes as defined in Table 65)

Axes 1 2 3 4 5 6 7 8 9 10 11 12 13 1. a-axis 77 80 84 85 89 92 90 88 86 81 82 75 85 2. b-axis 79 73 71 66 64 58 59 54 52 51 43 34 27 3. c-axis 55 63 66 65 64 63 61 54 53 47 43 36 32 4. d-axis 92 85 81 74 74 66 63 59 53 48 39 29 21 5. e-axis 66 65 65 64 63 63 58 55 52 42 38 27 20 6. f-axis 400 380 370 300 200 171 142 112 106 95 76 52 - 7. g-axis 91 112 133 137 111 107 89 79 66 55 42 28 21 8. h-axis 55 55 66 53 50 45 39 33 27 - - - - 9. i-axis - 168 138 115 89 77 56 43 31 25 19 21 - 10. j-axis - 34 33 34 32 31 26 23 20 20 22 22 - 11. k-axis 74 79 85 88 90 97 106 ------NUMBER 40 249

TABLE 68.—Measurements (mm) of the caudal vertebrae oi Glyptotherium floridanum (axes as defined in Table 65)

Axes 10 11 12

TMM 30967-1926 1. a-axis 63 71 76 80 82 81 2. b-axis 67 66 65 64 60 56 3. c-axis 62 63 63 39 57 52 4. d-axis 78 73 68 63 64 59 5. e-axis 64 62 ' 61 58 54 51

USNM 6071° 1. a-axis 67 75 78 82 82 82 2. b-axis 65 56 59 - 56 46 3. c-axis 64 58 57 - 48 38 4. d-axis - 61 59 - 45 41 5. e-axis 60 60 - - 55 38 6. f-axis - 240b 175b - - - USNM 10537 1. a-axis 67" 69 78 74 81 82 80 78 78 77 73 2. b-axis 76 66 68 64 60 58 52 48 47 42 34 3. c-axis 65 64 67 64 65 58 59 - - - 34 4. d-axis 77 77 80 77 62 58 55 49 44 34 29 5. e-axis 68b 65 63 67 63 56 55 51 - 35 34 6. f-axis 372 346 320b 219 167 136 101 78 - - - 7. g-axis - - 122 118 102 97 75 55 46 40 27 8. h-axis 49b 54b 56 40 30 21 12 8 - - - 9. i-axis - 170b 120b 95" 80h 70h 50b 25b 20b 18b 15 10. j-axis - 28 24 30 28 23 - - - - - 11. k-axis 68b 67 69 78 100b 105 98 62 - - -

AMNH 21808 1. a-axis 69 69 75 77 83 85 86 81 84 84 66 39b 2. b-axis 79 67 - - - - 50 43 27 3. c-axis 47 - - - - - 45 34 31 4. d-axis 70 - - - - - 53 44 31 5. e-axis 50 - - - - - 37 33 6. f-axis 440b 400" 320 260 200 150 120 110 7. g-axis 54 81 119 116 100 94 8. h-axis 46 59 52 45 36 - 9. i-axis - - - 98 85 - 40 33 19 17 b 10 j-axis 17 38 - 26 35 - 55 36 28 25 11 k-axis 66 66 67 78 89 - " Fused. b Approximate measurement. c Position of vertebrae 5-10 questionable; see text for discussion. 250 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

TABLE 69.—Measurements (mm) and numbers of scutes of the carapace of North American glyptodonts

G. texanum G. arizonae G. cylindricum G. mexicanum G. floridanum

Characters AMNH F:AM USNM AMNH AMNH type TMM 9773 10704 59599 10537/ 21808 15548 (restored) 10336 1. Anteroposterior length 1450 1440 1750 1900 1700 1830' 1800 along dorsal curvature 2. Maximum transverse 1860" 1920" 2150 2300" 2480 2400c 3600 half-circumference (mar­ ginal to marginal) 3. Number border scutes 28 23 16 31 35 24d _ along caudal aperture 4. Number border scutes, 56 54 34 47 54 48d - nuchal scute to posterior notch 5. Transverse diameter, ce­ 600" - 350" 300" 450 - - phalic aperture (±200) 6. Transverse diameter, 300" 420 700 caudal aperture (±200) 7. Number scutes, maxi­ 56 - - 50 63 - - mum transverse count 8. Number scutes along dor­ 33 - 35 36 36 - - sal arch maximum count " Approximate measurement. b Laterally compressed in free mount; if properly restored, this measurement would be an estimated 300 mm larger. c From Cuataparo and Ramirez (1875). dFrom Brown (1912). NUMBER 40 251

TABLE 70.—Measurements (mm) of the caudal armor of North American glyptodonts

Average Average Species inside Species inside Ring Vertebra diameter Ring Vertebra diameter « Glyptotherium texanum AMNH 10704 1 4 - AMNH 21808 I 2 - 2 5 210 2 3 365 3 (. 170 3 4 315 4 7 150 4 5 260 5 8 130 5 6 230 6 9 110 6 7 190 7 10 90 7 8 155 8 11 75 8 9 130 9 12 55 9 10 100 10 13/14 40 10/11 11/12 - F:AM 95737 1 2 190 UMMP 34826 1 2 - 2 3 210 2 3 380 3 4 190 3 4 300 4 5 160 4 5 240 5 6. 130 5 6 210 6 7 120 6 7 170 7 8 100 7 8 120 8 9 75 8 9 110 9 10 65 9 10 90 10 11/12/13 45 10 11 70 11 12 40 Glyptothenum arizonae 12 13 30 USNM 10537 1 2 - 2 3 310 Glyptothenum floridanum 3 4 270* TMM 977-3 1 3 - 4 5 240 2 4 280 5 6 200 3 5 270 6 7 180 4 6 260 7 a - 5 7 250 8 9 - 6 8 220 9 10 - 7 9 210 10/11 11/12 - * Approximate measurement. Literature Cited

Ahearn, M. E., and J. F. Lance 1888. Glyptodon from Texas. American Naturalist, 22:345- 1980. A New Species of Neochoerus (Rodentia: Hydro- 346. choeridae) from the Blancan (Late Pliocene) of 1889. The Edentata of North America. American Natural­ North America. Proceedings of the Biological Society of ist, 23:657-664, plates 31-32. Washington, 93:435-442, 2 figures. Cuataparo, J. N., and S. Ramirez Akersten, W. A. 1875. Descripcion de un mamifero fosil de especie des- 1972. Red Light Local Fauna (Blancan) of the Love conocida perteneciente al genero "Glyptodon." Formation, Southeastern Hudspeth County, Bolelin de la Sociedad Mexicana de Geografia y Estadis- Texas. Texas Memorial Museum Bulletin, 20: 53 tica (Mexico), series 3, 2:354-362, 4 figures. pages, 16 figures, 21 tables. Cuvier, G. Alvarez, T. 1798. Tableau elemenlaire de Thistoire naturelle des animaux. 1965. Catalogo Paleomastozoologico Mexicano. Instituto xvi -I- 710 pages, 14 plates. Paris: J. Bailliere. Nacional de Antropologia e Historia, Publicacwn (Mex­ Dalquest, W. W. ico), 17: 70 pages. 1961. Sylvilagus cuniculanus in the Pleistocene of Mexico. Ameghino, F 1895. Sur les Edentes Fossiles de l'Argentine (Examen Journal of Mammalogy, 42:408-409. critique, revision, et correction de l'ouvrage de M. 1975. Vertebrate Fossils from the Blanco Local Fauna R. Lydekker "The Extinct Edentates of Argen­ of Texas. Occasional Papers of the Museum of Texas tina." Revista del Jardin Zoologico de Buenos Aires, Tech University, 30: 52 pages, 7 figures, 1 table. III(4):97-192. [Also published in Ameghino Obras, 1977. Mammals of the Holloman Local Fauna, Pleisto­ XL447-909.] cene of Oklahoma. The Southwestern Naturalist, 22: Brown, B. 255-268. 1912. Brachyostracon, a New Genus of Glyptodonts from Downey, J. S. Mexico. American Museum of Natural History Bulletin, 1970. Middle Pleistocene Leporidae from the San Pedro 31(17): 167-177, 4 figures, plates 13-18. Valley, Arizona. United States Geological Survey Burmeister, H. Professional Paper, 700-B: 131 -136. 1874. Monographia de los glyptodontes en el Museo Evans, G. L., and G. E. Meade Publico de Buenos Aires. Anales del Museo Nacional 1945. Quaternary of the Texas High Plains. University of Buenos Aires, II: vi -I- 412 pages, 42 plates. Texas Publication, 4401:485-507, 9 figures, maps. 1879. Mammiferes vwanls el clients. In Description Physique Evernden, J. F., D. E. Savage, G H. Curtis, and G. T. James de la Republique Argentine , 3(1): vi + 556 1964. Potassium-Argon Dates and the Cenozoic Mam­ pages. Buenos Aires: Paul-Emile Coni. malian Chronology of North America. American Cantwell, R. J. Journal of Science, 262:145-198, 7 tables. 1969. Fossil Sigmodon from the Tusker Locality, 111 Felix, J., and H. Lenk Ranch, Arizona. Journal of Mammalogy, 50:375-378, 1899. Beitrage zur Geologie und Palaontologie der Re- 2 figures. publik Mexico. Leipzig, 2: 210 pages, 52 figures, Castellanos, A. 10 plates. 1932. Nuevos generos de Gliptodontes en relacion con su Flower, W. H. Filogenia. Physis, 11:92-100. 1953. Anotacoes e retificacoes ao genero Glyptodon Owen 1868. On the Development and Succession of the Teeth e a subfamilia "Glyptodontinae." Anales Academic in the Armadillos (Dasypodidae). Proceedings of the Brasileira Ciencias, 25(4):391-410. Zoological Society of London, 1868:378-380. 1954. Referenda a la subfamilia Glyptodontinae. Flower, W. H., and R. Lydekker Comples Rendus 19th International Geological Congress 1891. An Introduction to the Study of Mammals Living and (Algiers, 1952), 15:35-43. [English summary.] Extinct, xvi + 763 pages, 357 figures. London: Cope, E. D. Black. 1886. A Giant Armadillo from the Miocene of Kansas. Gazin, C. L. American Naturalist, 20(12): 1044-1046. 1942. The Late Cenozoic Vertebrate Faunas from the

252 NUMBER 40 253

San Pedro Valley, Arizona. Proceedings of the United quiac, Mexico, with Notes on Other Pleistocene States National Museum, 92:475-518. Forms. Contributions from the Museum of Paleontology, 1950. Annotated List of Fossil Mammalia Associated University of Michigan, 12:47-96, 5 figures, 9 plates, with Human Remains at Melbourne, Florida. 12 tables. Journal of the Washington Academy of Sciences, 40:397- Hibbard, C. W., and W. W. Dalquest 404. 1966. Fossils from the Seymour Formation of Knox and Gidley, J. W. Baylor Counties, Texas, and Their Bearing on the 1922. Preliminary Report on Fossil Vertebrates of the Late Kansan Climate of That Region. Contributions San Pedro Valley, Arizona, with Descriptions of from the Museum of Paleontology, University of Michigan, New Species of Rodentia and Lagomorpha. United 21(1): 1-66, 8 figures, 5 plates, 3 tables. States Geological Survey Professional Paper, 131E: 119- 1973. Proneoflber, a New Genus of Vole (Cricetidae: Ro­ 131, plates 34-35. dentia) from the Pleistocene Seymour Formation 1926. Fossil Proboscidea and Edentata of the San Pedro of Texas, and Its Evolutionary and Stratigraphic Valley, Arizona. United States Geological Survey Significance. Quaternary Research, 3(2):269-274, 2 Professional Paper, 140B:83-95, 13 plates, 4 figures. figures. Gillette, D. D. Hoffstetter, R. 1975. A Review of North American Glyptodonts (Ed­ 1955. Sur le genotype de Glyptodon Owen. Bulletin du entata, Mammalia): Osteology, Systematics, and Museum d'Histoire Naturelle (Paris), series 2, 27:408- Paleobiology. Dissertation Abstracts International, 35: 413, 1 figure. 1659. 1958. Xenarthra. In J. Piveteau, editor, Traite de Paleon­ Hay, O. P. tology, 6(2):535-636, 64 figures. 1908. The Fossil Turtles of North America. Carnegie In­ Holmes, W. W., and G. G. Simpson stitution of Washington Publication, 75: iv + 568 pages, 1931. Pleistocene Exploration and Fossil Edentates in 704 figures, 113 plates. Florida. Bulletin of the American Museum of Natural 1916. Descriptions of Two Extinct Mammals of the Or­ History, 59(7): 383-418, 21 figures. der Xenarthra from the Pleistocene of Texas. Pro­ Howell, A. B. ceedings of the United States National Museum, 1944. Speed in Animals, Their Specialization for Running and 51(2147):107-123, plates 3-7. Leaping, xii + 270 pages, 55 figures, 3 tables. 1923. The Pleistocene of North America and Its Verte- Chicago: University of Chicago Press. brated Animals from the States East of the Mis­ Huxley, T. H. sissippi River and from the Canadian Provinces 1863. Description of a New Specimen of Glyptodon Re­ East of Longitude 95°. Carnegie Institution of Wash­ cently Acquired by the Royal College of Surgeons ington Publication, 322: viii -I- 499 pages, 41 maps, of England (Abstract of Huxley, 1865). Proceedings 25 figures. of the Royal Society of London, 12:316-326. 1924. The Pleistocene of the Middle Region of North 1865. On the Osteology of the Genus Glyptodon. Philo­ America and Its Vertebrated Animals. Carnegie sophical Transactions of the Royal Society of London, Institution of Washington Publication, 322A: viii + 385 pages, 29 maps, 5 figures. 155:31-70, plates 4-9. 1926. A Collection of Pleistocene Vertebrates from Illiger, C. Southwestern Texas. Proceedings of the United States 1811. Prodromus Systematis Mammalium et Avium Additis Ter- National Museum, 68:1-18, 2 figures, 8 plates. mims Zoographicis Utriusque Classis. xviii + 301 1927. The Pleistocene of the Western Region of North pages. Berlin: C. Salfeld. America and Its Vertebrated Animals. Carnegie James, G. T. Institution of Washington Publication, 322B: 346 pages, 1957. An Edentate from the Pleistocene of Texas. Journal 21 maps, 12 plates, 19 figures. of Paleontology, 31:796-808, 4 figures, 9 tables, Hay, O. P., and H. J. Cook plates 101-102. 1930. Fossil Vertebrates Collected near, or in Association Johnston, C. S. with, Human Artifacts at Localities near Colo­ 1944. The Upper Pliocene Cita Canyon Fauna from Randall rado, Texas; Frederick, Oklahoma; and Folsom, County, Texas. D. E. Savage, editor. Unpublished New Mexico. Proceedings of the Colorado Museum of thesis, University of Oklahoma. Natural History, 9(2):4-40, 14 plates. Johnston, C. S., and D. E. Savage Hibbard, C. W. 1955. A Survey of Various Late Cenozoic Vertebrate 1955. Pleistocene Vertebrates from the Upper Becerra Faunas of the Panhandle of Texas, Part I: Intro­ (Becerra Superior) Formation, Valley of Tequix- duction, Description of Localities, Preliminary 254 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Faunal Lists. University of California Publications in Meade, G E. .Geological Sciences, 31:27-50, 6 figures, 5 maps. 1945. The Blanco Fauna. University of Texas Publication, Kaspar, T. C, and W. L. McClure 4401:509-556, 4 figures, plates 48-55. 1976. The Taylor Bayou Local Fauna (Pleistocene) near 1953. An Early Pleistocene Vertebrate Fauna from Fred­ Houston, Texas. 'The Southwestern Naturalist, 21:9- erick, Oklahoma. Journal of Geology, 61(5):452-460, 16, 2 tables. 1 plate. Lammers, G. E. Melton, W. G., Jr. 1970. The Late Cenozoic Benson and Curtis Ranch Faunas from 1964. Glyptodon fredericensis (Meade) from the Seymour the San Pedro Valley, Cochise County, Arizona. Unpub­ Formation of Knox County, Texas. Papers of the lished doctoral dissertation, University of Arizona, Michigan Academy of Science, Arts, and Letters, 49:129- Tucson. 146, 3 figures, 3 plates. Lance, J. F Mooser, O, and W. W. Dalquest I960. Stratigraphic and Structural Position of Cenozoic 1975. Pleistocene Mammals from Aguascalientes, Cen­ Fossil Localities in Arizona. Arizona Geological So­ tral Mexico. Journal of Mammalogy, 56:781-820, 12 ciety Digest, 3:155-159, 1 figure. figures, 4 tables. Leidy, J. Osborn, H. F. 1889a. Fossil Vertebrates from Florida. Proceedings of the 1903. Glyptotherium texanum, a New Glyptodont, from the Academy of Natural Sciences of Philadelphia, 41:96-97. Lower Pleistocene of Texas. Bulletin of the American 1889b. Description of Vertebrate Remains from Peace Museum of Natural History, 19(17):491-494, 1 plate. Creek, Florida. Transactions of the Wagner Free Insti­ Owen, R. tute of Science (Philadelphia), 2:19-31, plates 4-8. 1838. Note on the Glyptodon. In W. Parish, Buenos Ayres Lindsay, E. H., and N. T. Tessman and the Provinces of the Rio de la Plata, pages 178b-e. 1974. Cenozoic Vertebrate Localities and Faunas in Ar­ Pichardo del Barrio, M. izona. Journal of the Arizona Academy of Science, 9:3- 1960. Proboscideos Fosiles de Mexico, una Revision. In­ 24. stituto Nacional de Antropologia y Historia Mexico, In- Lindsay, E. H., N. M. Johnson, and N. D. Opdyke vestigaciones, 4: 63 pages, 9 figures, 22 plates. 1975. Preliminary Correlation of North American Land Ray, C. E. Mammal Ages and Geomagnetic Chronology. 1958. Additions to the Pleistocene Mammalian Fauna University of Michigan Papers on Paleontology, 12:111 — from Melbourne, Florida. Bulletin of the Museum of 119. Comparative Zoology at Harvard College, 119:419-450, Lundelius, E. L., Jr. 5 figures, 3 tables. 1967. Late Pleistocene and Holocene Faunal History of 1965. A Glyptodont from South Carolina. The Charleston Central Texas. In P. S. Martin and H. E. Wright, Museum Leaflet, 27: 12 pages, 4 plates. Jr., editors, Pleistocene Extinctions: The Search for a Roth, J., and J. Laerm Cause, 1967:287-319, 11 figures, 1 table. 1980. A Late Pleistocene Vertebrate Assemblage from 1972. Fossil Vertebrates from the Late Pleistocene Ingle­ Edisto Island, South Carolina. Bnmleyana, 3:1-29, side Fauna, San Patricio County, Texas. 77?? Uni­ 2 figures, 4 tables. versity of Texas Bureau of Economic Geology, Report of Seff, P Investigations, 11: vi + 74 pages, 59 figures, 52 1962. Stratigraphic Geology and Depositional Environments of tables, 1 plate. the 111 Ranch Area, Graham County, Arizona. Unpub­ Lydekker, R. lished doctoral dissertation, University of Arizona, 1894. Contributions to a Knowledge of the Fossil Ver­ Tucson. tebrates of Argentina, Part II: The Extinct Eden­ Sellards, E. H. tates of Argentina. Anales del Museo de la Plata, 3 1940. Pleistocene Artifacts and Associated Fossils from (Paleontologia Argentina): 118 pages, 61 plates. Bee County, Texas. Bulletin of the Geological Society MacFadden, B. J., and J. S. Waldrop of America, 51; 1627-1658, 7 figures, 2 plates. 1980. Nannippus phlegon (Mammalia, Equidae) from the Senechal, L. Pliocene (Blancan) of Florida. Bulletin of the Florida 1865. Notice sur I'armure ou le dermato-squelette et le systeme State Museum, Biological Sciences, 25:1-37, 16 figures. dentaire du Glyptodon clavipes, et particulariles biol- Maldonado-Koerdell, M. ogiques de eel animal, deduiles d'apres I'etude de ses restes 1948. Los Vertebrados Fosiles del Cuaternario en Mex­ fossiles. 24 pages. Paris. ico. Revista de la Sociedad Mexicana de Historia Natural, Sicher, H. 9(l-2):l-35. 1944. Masticatory Apparatus of the Sloths. Field Museum NUMBER 40 255

of Natural History, Zoological Series, 29(10): 161-168, Trouessart, E. L. figures 23-25. 1898. Catalogus Mammalium lam viventium quam fossilium. Silva-Barcenas, A. New edition (prima completa), part 5, pages 999- 1969. Localidades de Vertebrados Fosiles en la Repub­ 1264. Berlin: R. Friedlander und Sohn. lica Mexicana. Universidad Nacional Autonoma de Tumbull, W. D. Mexico Instituto de Geologia y Paleontologia Mexicana, 1970. Mammalian Masticatory Apparatus. Fieldiana: Ge­ 28: 34 pages, 1 map. ology, 18(2): 149-356, 48 figures. Simpson, G. G. Vinacci Thul, E. L. 1943. Nomenclature de los dedos de la mano de "Glyp­ 1929a. The Extinct Land Mammals of Florida. Annual todon." Physis, 19(53): 391-401, 1 figure. Report of the Florida State Geological Survey (20th 1945. Osteografia cefalica de Glyptodon reticulatus Ow. Annual Report), pages 229-279, plates 30-40, 1 Physis, 20(55):24-30. figure, 3 maps. Webb, S. D., editor 1929b. Pleistocene Mammalian Fauna of the Seminole 1974. Pleistocene Mammals of Florida, x + 270 pages. Field, Pinellas County, Florida. Bulletin of the Amer­ Gainesville: the University Presses of Florida. ican Museum of Natural History, 56(8):561-599, 22 Webb, S. D. figures. 1977. A History of Savanna Vertebrates in the New 1931. A New Classification of Mammals. Bulletin of the World, Part I: North America. Annual Review of American Museum of Natural History, 59:259-293. Ecology and Systematics, 8:355-380, 2 figures. 1945. The Principles of Classification and a Classifica­ 1978. A History of Savanna Vertebrates in the New tion of Mammals. Bulletin of the American Museum of World, Part II: South America and the Great Natural History, 85: xvi + 350 pages. Interchange. Annual Review of Ecology and Systematics Skinner, M. F. 9:393-426, 1 table. 1942. The Fauna of Papago Springs Cave, Arizona, and Weber, M. a Study of Slockoceros; with Three New Antiloca- 1928. Die Saugetiere. Volume 2 (systematic part), 2nd prines from Nebraska and Arizona. Bulletin of the edition, xxiv + 898 pages. Jena: Gustav Fischer. American Museum of Natural History, 80(6): 143-220, Weigel, R. D. 19 figures. 1962. Fossil Vertebrates of Vero, Florida. Florida Geolog­ ical Survey, Special Publication, 10: viii + 59 pages, 6 Strain, W. S. figures, 5 tables. 1966. Blancan Mammalian Fauna and Pleistocene For­ Wood, H. E., et al. mations, Hudspeth County, Texas. Bulletin of the 1941. Nomenclature and Correlation of the North Amer­ Texas Memorial Museum, 10: 55 pages, 8 figures, 13 ican Continental Tertiary. Bulletin of the Geological plates. Society of America, 52:1-48, lplate. Talmage, R. V., and G. D. Buchanan Wood, P A. 1954. The Armadillo (Dasypus novemcmctus). The Rice In­ 1962. Pleistocene Fauna from 111 Ranch Area, Graham County, stitute Pamphlet, 41(2): viii + 135 pages, 2 figures, Arizona. Unpublished doctoral dissertation, Uni­ 1 plate, 2 tables. versity of Arizona, Tucson.