FIELDIANA GEOLOGY LIBRAE Geology NEW SERIES, NO. 41 The Morphology of Xenarthrous Vertebrae (Mammalia: Xenarthra) Timothy J. Gaudin en 2z September 30, 1999 Publication 1505 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY Information for Contributors to Fieldiana General: Fieldiana is primarily a journal for Field Museum staff members and research associates, although manuscripts from nonaffiliated authors may be considered as space permits. The Journal carries a page charge of $65.00 per printed page or fraction thereof. Payment of at least 50% of page charges qualifies a paper for expedited processing, which reduces the publication time. Contributions from staff, research associates, and invited authors will be considered for publication regardless of ability to pay page charges, however, the full charge is mandatory for nonaffiliated authors of unsolicited manuscripts. Three complete copies of the text (including title page and abstract) and of the illustrations should be submitted (one original copy plus two review copies which may be machine copies). No manuscripts will be considered for publication or submitted to reviewers before all materials are complete and in the hands of the Scientific Editor. Manuscripts should be submitted to Scientific Editor, Fieldiana, Field Museum of Natural History, Chicago, Illinois 60605-2496, U.S.A. Text: Manuscripts must be typewritten double-spaced on standard-weight, 8Vi- by 11 -inch paper with wide margins on all four sides. If typed on an IBM-compatible computer using MS-DOS, also submit text on S^-inch diskette (WordPerfect 4.1, 4.2, or 5.0, MultiMate, Displaywrite 2, 3 & 4, Wang PC, Samna, Microsoft Word, Volks- writer, or WordStar programs or ASCII). For papers over 100 manuscript pages, authors are requested to submit a "Table of Contents," a "List of Illustrations," and a "List of Tables" immediately following title page. In most cases, the text should be preceded by an "Abstract" and should conclude with "Acknowledgments" (if any) and "Literature Cited." All measurements should be in the metric system (periods are not used after abbreviated measurements). The format and style of headings should follow that of recent issues of Fieldiana. For more detailed style information, see The Chicago Manual of Style (13th ed.), published by The University of Chicago Press, and also recent issues of Fieldiana. References: In "Literature Cited," book and journal titles should be given in full. Where abbreviations are desirable (e.g., in citation of synonymies), authors consistently should follow Botanico-Periodicum-Huntianum and TL-2 Taxonomic Literature by F A. Stafleu & R. S. Cowan (1976 et seq.) (botanical papers) or Serial Sources for the Biosis Data Base (1983) published by the BioSciences Information Service. Names of botanical authors should follow the "Draft Index of Author Abbreviations, Royal Botanic Gardens, Kew," 1984 edition, or TL-2. References should be typed in the following form: Croat, T. B. 1978. Flora of Barro Colorado Island. Stanford University Press, Stanford. Calif., 943 pp. Grubb, P. J., J. R. Lloyd, and T D. Pennington. 1963. A comparison of montane and lowland rain forest in Ecuador. I. The forest structure, physiognomy, and floristics. Journal of Ecology, 51: 567-601. Langdon, E. J. M. 1979. Yage among the Siona: Cultural patterns in visions, pp. 63-80. In Browman, D. L., and R. A. Schwarz, eds., Spirits, Shamans, and Stars. Mouton Publishers, The Hague, Netherlands. Murra, J. 1946. The historic tribes of Ecuador, pp. 785-821. In Steward, J. H., ed.. Handbook of South American Indians. Vol. 2, The Andean Civilizations. Bulletin 143, Bureau of American Ethnology, Smithsonian Institution, Washington, D.C. Stolze, R. G. 1981. Ferns and fern allies of Guatemala. Part II. Polypodiaceae. Fieldiana: Botany, n.s., 6: 1- 522. Illustrations: Illustrations are referred to as "figures" in the text (not as "plates"). Figures must be accompanied by some indication of scale, normally a reference bar. Statements in figure captions alone, such as "X 0.8," are not acceptable. Captions should be typed double-spaced and consecutively. See recent issues of Fieldiana for details of style. All illustrations should be marked on the reverse with author's name, figure number(s), and "top." Figures as submitted should, whenever practicable, be SV2 by 1 1 inches (22 X 28 cm) and may not exceed 1 W2 by I6V2 inches (30 X 42 cm). Illustrations should be mounted on boards in the arrangement to be obtained in the printed work. This original set should be suitable for transmission to the printer as follows: Pen and ink drawings may be originals (preferred) or photostats; shaded drawings must be originals, but within the size limitation; and photostats must be high-quality, glossy, black and white prints. Original illustrations will be returned to the corre- sponding author upon publication unless otherwise specified. Authors who wish to publish figures that require costly special paper or color reproduction must make prior arrangements with the Scientific Editor. Page Proofs: Fieldiana employs a two-step correction system. The corresponding author will normally receive a copy of the edited manuscript on which deletions, additions, and changes can be made and queries answered. Only one set of page proofs will be sent. All desired corrections of type must be made on the single set of page proofs. Changes in page proofs (as opposed to corrections) are very expensive. Author-generated changes in page proofs can only be made if the author agrees in advance to pay for them. © This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). GEOLOGY UBRAFW FIELDIANA Geology NEW SERIES, NO. 41 The Morphology of Xenarthrous Vertebrae (Mammalia: Xenarthra) Timothy J. Gaudin Department of Biological and Environmental Sciences University of Tennessee at Chattanooga 615 McCallie Avenue Chattanooga, Tennessee 37403-2598 U.S.A.1 Research Associate Department of Geology Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2496 U.S.A. 1 Address to which correspondence should be sent. Accepted September 3, 1998 Published September 30, 1999 Publication 1505 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY © 1999 Field Museum of Natural History ISSN 0096-2651 PRINTED IN THE UNITED STATES OF AMERICA Table of Contents Abstract 1 Introduction 1 List of Abbreviations 4 Descriptive Anatomy 4 Cingulata 4 Euphracta (esp. Zaedyus pichiy) 4 Tolypeutes matacus 13 Other Cingulates 15 Vermilingua 17 Tamandua mexicana 17 Other Vermilinguas 19 Tardigrada 19 Bradypus variegatus 19 Hapalops 21 Other Tardigrada 24 Conclusions 26 Morphological Summary 26 Phylogeny and Evolution of Xenarthrous Vertebrae 27 Relationship of Xenarthra to Early Ceno- zoic Fossil Taxa 30 Acknowledgments 33 Literature Cited 33 Appendix: Summary of morphological Data 36 List of Illustrations lumbar vertebrae in anterior and poste- rior view 8 4. Zaedyus pichiy, stereophotographs of thoracic and lumbar vertebrae in right lateral view 10 5. Tolypeutes matacus (juvenile), thoracic and lumbar vertebrae in left lateral view 14 6. Tolypeutes matacus (juvenile), sacral vertebrae in dorsal, ventral, and right lateral views 16 7. Priodontes maximus, first lumbar verte- bra in anterior view 17 8. Tamandua mexicana, stereophoto- graphs of posterior thoracic vertebrae in right lateral view 18 9. Bradypus variegatus, stereophotographs of thoracic and lumbar vertebrae in right lateral view 20 10. Hapalops sp., posterior thoracic verte- brae in dorsal view 22 11. Hapalops sp., posterior thoracic verte- brae in left lateral view 22 12. Pronothrotherium typicum, isolated mid-thoracic vertebra in dorsal view 24 13. Diagrammatic representation of typical xenarthran intervertebral facets in ante- rior and posterior view 26 14. Distribution of vertebral character states on a phylogeny of the Xenarthra 28 Mephitis mephitis, thoracic and lumbar vertebrae in anterior, posterior, and left lateral views Zaedyus pichiy, Tamandua mexicana, and Bradypus variegatus, thoracic and lumbar vertebrae in dorsal view Zaedyus pichiy, Tamandua mexicana, and Bradypus variegatus, thoracic and List of Tables 1 . Ratio of maximum width to anteropos- terior length of anterior zygapophyseal facets in anterior thoracic vertebrae of Xenarthra 23 in The Morphology of Xenarthrous Vertebrae (Mammalia: Xenathra) Timothy J. Gaudin Abstract The presence of supplementary intervertebral articulations termed "xenarthrales" in the pos- terior dorsal vertebrae has been considered perhaps the most important diagnostic feature of the mammalian order Xenarthra. Xenarthrales are poorly understood, however, and substantial confusion exists in the literature over which facets are supplementary and which are not. Furthermore, much of the variation that exists in these joints, both within taxa and among the various xenarthran lineages, has gone unnoticed. Finally, the structural evolution of these facets has been inadequately treated. The goal of the present study is to describe the morphology of xenarthrous vertebrae in juvenile and adult extant xenarthrans and in extinct xenarthrans, to develop a model for the structural evolution of the supernumerary joints, and to use this in- formation to assess the affinities of several enigmatic groups of early Cenozoic taxa (Palaean- odonta, Ernanodon, and Eurotamandua) with purported ties to the Xenarthra. Vertebral mor- phology is described in detail for two armadillo species, one species of anteater, and one extant and one extinct species of sloth, with brief comments on other xenarthran taxa. The results suggest that all xenarthrans are characterized by two sets of zygapophyseal facets in the post- diaphragmatic vertebrae, one medial and one lateral to the metapophysis. In addition, the Xe- narthra is characterized primitively by a pair of xenarthrous facets on each side of the vertebra between the dorsal surface of the anapophysis and the ventral surface of the metapophysis of the succeeding vertebra. Other xenarthrous joints evolve within various xenarthran lineages. It is suggested that the supplementary facets developed initially in the diaphragmatic region of the vertebral column by means of a progressive widening of the zygapophyseal facets in the thoracic vertebrae and an increase in size of the metapophysis, which subdivided the zygapo- physeal facets into medial and lateral facets. Hypertrophy of the anapophyses and their contact with the metapophyses led to the formation of true xenarthrous facets. A review of vertebral morphology in the Palaeanodonta, Ernanodon, and Eurotamandua revealed few resemblances to undoubted xenarthrans beyond hypertrophy of the metapophyses and anapophyses — characteristics known to occur in many different groups of mammals. No supplementary intervertebral articulations could be documented unequivocally in any of these taxa. Thus, on the basis of vertebral morphology there is little evidence that would suggest a close phylogenetic relationship between true xenarthrans and palaeanodonts, Ernanodon, or Eurotamandua. Introduction The distinctive nature of the vertebral column in the mammalian order Xenarthra was recog- The single most important osteological character- nized in the earliest osteological descriptions of istic of Xenarthra is the presence of accessory ar- ^ ( Cuyi ,836a) In most mammals ticular processes or anapophyses, which articulate • ~7 . • ■ \ i«_ ventral to the metapophyses, or between them and successive vertebrae are jotned not only by an in- the transverse processes, of the following vertebrae tervertebral disc, but also by a single pair of sy- (Rose & Emry, 1993, p. 87). novial joints carried on more or less distinct ar- FIELDIANA: GEOLOGY, N.S., NO. 41, SEPTEMBER 30, 1999, PP. 1-38 1 ns B ns ns vc Fig. 1. Mephitis mephitis, utcm 521: thoracic and lumbar vertebrae shown in anterior, posterior and left lateral views (proceeding left to right). A, T4, B, L3. Scale bar = 1 cm. Abbreviations: ap, anapophysis; az, anterior zygapophyseal facets; dp, diapophysis; la, lamina; mp, metapophysis; ns, neural spine; pe, pedicel; pz, posterior zygapophyseal facet; rf, rib facet; sn, notch for emergence of spinal nerve; tp, transverse process; vc, vertebral centrum. ticular processes of the neural arches termed zyg- apophyses (Fig. 1). In addition to the typical zyg- apophyseal articulations, all xenarthrans possess one or more pairs of supplementary intervertebral articulations that are usually present between all lumbar and a variable number of posterior thorac- ic vertebrae. The supernumerary articulations, termed "xenarthrales," are well developed in xe- narthrans that span a wide range of locomotory habits, including a subterranean armadillo (Chla- myphorus), fossorial armadillos and anteaters (e.g., Dasypus, Euphractus, Myrmecophaga), ar- boreal climbing anteaters (Tamandua, Cyclopes), and semiarboreal (e.g., Hapalops; White, 1993a,b) to fully terrestrial (e.g., Mylodon, Me- gatherium) extinct ground sloths. Xenarthrales are present in the oldest well-known fossil xenarthran skeleton, the Casamayoran armadillo Utaetus (Simpson, 1948). The supplementary joints are strongly reduced only in the suspensory tree sloths, and they are absent only in the glypto- donts; in the latter the dorsal portions of the back- bone are fused into a bony tube used to support the massive carapace of the animals (Hoffstetter, 1958; Gillette & Ray, 1981). Because of the peculiar and complex nature of these xenarthrous articulations, their almost uni- versal presence among living and fossil xenar- thrans, and the near universal absence of similar supplementary intervertebral joints in other mam- mals (but see Scutisorex: Lessertisseur & Saban, 1967; Kingdon, 1984; Cullinane & Aleper, 1998; Cullinane et al., 1998), the presence or absence of xenarthrales has been used as a "litmus test" for determining phylogenetic relatedness to the Xe- narthra. The pangolins and aardvarks were origi- nally included with xenarthrans under the taxo- nomic grouping Edentata, but they were subse- quently removed to separate orders largely be- cause they lacked xenarthrales (Weber, 1904; see Hoffstetter, 1982, and Glass, 1985, for history of edentate classification). Several enigmaic groups of early Cenozoic mammals have been linked to the Xenarthra on the basis of "incipient" devel- opment of xenarthrales. These include the Pa- laeanodonta (Simpson, 1931), a group known from Paleocene to Oligocene deposits of North America and Europe, and Ernanodon (Ding, 1987), a Late Paleocene genus from China. Both Ernanodon and the palaeanodont Metacheiromys have enlarged anapophyses in the posterior dorsal vertebrae. Similarly enlarged anapophyses can be FIELDIANA: GEOLOGY found, however, among a number of unrelated groups of mammals, e.g., felids and geomyid ro- dents (Rose & Emry, 1993). Hence such process- es are not necessarily structural antecedents of true xenarthrous articulations. The purported Mid- dle Eocene anteater Eurotamandua, from the Messel fauna of Germany, allegedly possesses true xenarthrous articulations (Storch, 1981). Un- fortunately, several subsequent authors have been unable to verify the presence of accessory inter- vertebral articulations in this taxon (Rose & Emry, 1993; Szalay & Schrenk, 1994). Novacek and his colleagues (Novacek, 1986; Novacek & Wyss, 1986; Novacek et al., 1988) resurrected the taxonomic grouping Edentata, including pango- lins, palaeanodonts, and xenarthrans, in a com- mon supraordinal cohort based on the results of morphological studies of eutherian interordinal phylogeny. Their work has led them to suggest that the phylogenetic significance of xenarthrales may be overemphasized. Part of the difficulty in determining the phylo- genetic significance of xenarthrales, and in iden- tifying xenarthrales or incipient xenarthrales in early Cenozoic taxa potentially allied to Xenar- thra, lies in the fact that the morphology of these articulations among unquestioned xenarthrans is not well understood. As stated above, the presence of accessory in- tervertebral articulations in xenarthrans has been noted since the early nineteenth century. Cuvier (1836a)1 wrote the first brief description of such joints. They were not formally named, however, until Gill (1886, p. 66)2 coined the term "xenar- thral" (Gk., xenos = "strange," arthron = "joint") in order to distinguish xenarthran verte- brae from normal, "nomarthral" vertebrae. Simp- son (1931, 1948) used the adjective "xenar- throus" to refer to these accessory joints. This adjective is the one most commonly used in the recent literature (e.g., Emry, 1970; McKenna, 1 Curiously, the joints are not described in a contem- poraneous edition of Cuvier's Recherches sur les osse- mens fossiles, where, e.g., the vertebrae of the anteater Myrmecophaga are described as "unremarkable" (Cu- vier, 1836b, p. 208). 2 Glass (1985) cites Gill (1872) as the source of the term. I can find no mention of the term, however, in my copy of Gill's mammalian classification. Indeed, Gill's classification is incomplete, because it contains neither a discussion of family and subfamily characteristics nor a list of genera for any of the Ineducabilia, a group in- cluding the Bruta (= Xenarthra + Pholidota + Tubuli- dentata). Vertebral morphology is not mentioned in the ordinal level description of Bruta. 1975; Engelmann, 1985; Vaughn, 1986; Rose & Emry, 1993). The joints have also been referred to by the noun "xenarthrales" (e.g., Frechkop, 1949; Grasse, 1955; Hoffstetter, 1958; DeBlase & Martin, 1981; Storch 1981),3 and the condition of possessing such joints has been termed "xenar- thry" (e.g., Lessertisseur & Saban, 1967; Hoff- stetter, 1982; Novacek & Wyss, 1986). The first detailed morphological description of xenarthrous intervertebral articulations (Owen, 1851a) preced- ed Gill's work by some 20 years. Owen described the morphology of the supplementary interverte- bral articulations in at least two extant species from each of the three major xenarthran subor- ders, the Cingulata (armadillos), Vermilingua (Neotropical anteaters), and Tardigrada (sloths). Moreover, Owen described regional variation in the morphology of the extra intervertebral joints along the backbone of individual species. Owen (1851a) began each description with the anterior- most xenarthrous vertebra and then described how the morphology of the facets changed as one moved posteriorly along the spine. Flower's (1885) description was similar but much briefer than Owen's. He described the mor- phology of the xenarthrous articulations along the whole length of the vertebral column, but only in the vermilinguan Myrmecophaga (Flower, 1885). He also briefly summarized the form of the xe- narthrous facets in the sloth genus Bradypus. In- terestingly, Rower differed from Owen in decid- ing which facets to designate as the normal zyg- apophyseal facets and which to designate as sup- plementary. Owen (1851a) consistently recognized the medialmost pair of intervertebral facets, those lying medial to the metapophyses, as the normal zygapophyseal facets. Flower (1885, figs. 22-24) designated a set of facets lying lateral to the metapophyses as the homologues of the typical mammalian zygapophyses. As noted by Rose and Emry (1993), the con- fusion over which set of facets constitutes supple- mentary articulations and which are the normal zygapophyseal facets has persisted to the present. The designation of a lateral facet as the zygapo- physeal facet by Flower is followed by Grasse (1955) and Vaughn (1986). Owen's (1851a) iden- tification of the zygapophyses as lying more me- 1 Gaudin and Biewener (1992) and Gaudin (1993) re- fer to the joints as "xenarthrae," a term which I had assumed was standard usage. Unfortunately, I can now find no historical source for the term. Although I do not believe I invented the term, I have chosen to abandon it for the apparently more widely used "xenarthrales." GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE dially is followed by Hoffstetter (1958, 1982), Lessertisseur and Saban (1967), and Gaudin and Biewener (1992). Jenkins (1970) notes that in Ta- mandua the distinct facets designated as "zyg- apophyses" by Owen (1851a) and Flower (1885) are distinct only in the lumbar vertebrae. In the posterior thoracic vertebrae, the medial and lateral facets become confluent. The problem of identifying which facets are xe- narthrous and which represent the typical mam- malian zygapophyseal facets is exacerbated by variability in the number and form of the inter- vertebral joints among xenarthrans and by the lack of any modern comprehensive study of the problem. It is known that in some living xenar- thrans, e.g., the giant armadillo Priodontes, there are as many as six pairs of intervertebral articu- lations (Grasse, 1955; see below). In others the number of facets is reduced, so that in many ground sloths, e.g., Hapalops, there are only two pairs of intervertebral facets, one set of zygapoph- yses, and one set of supplementary joints (Scott, 1903-1904; see below). This variability has gone largely unnoticed, as most modern studies that make any mention of the morphology of xenar- thrales illustrate only a single joint in a single spe- cies, and describe this joint in a few brief sen- tences (Grasse, 1955; Hoffstetter, 1958, 1982; Vaughn, 1986; Gaudin & Biewener, 1992). Jen- kins (1970) describes regional variation in the morphology of xenarthrales, but only in Taman- dua tetradactyla. Finally, the problem of the origin of xenarthra- les has been inadequately treated in the literature. Gaudin and Biewener (1992), Gaudin (1993), and Gaudin and Fortin (unpubl. data) considered the functional reasons for the evolution of these facets but did not address their actual structural anteced- ents. MacPhee (1994, p. 174) suggested that the facets arose through the "sacralization" of the lumbar and posterior thoracic vertebrae, but he noted the paucity of evidence to substantiate this claim. To date no study has examined the struc- tural origin of xenarthrales from either a devel- opmental or paleontological perspective. The present study reexamines the morphology of xenarthrous vertebrae throughout the Xenar- thra. Following Owen (1851a), both regional in- traspecific variation and interspecific variation across the various xenarthran suborders are ad- dressed. Nearly all of the extant xenarthran genera have been examined in detail, as well as the fossil genera housed in the Field Museum of Natural History (fmnh) collections, plus an assortment of ground sloth genera from other North American museums. The descriptions provided below, how- ever, will focus on five representative taxa: the extant armadillos Zaedyus and Tolypeutes, the ex- tant anteater Tamandua, the extant sloth Brady- pus, and the extinct sloth Hapalops. Unlike the study reported by Owen (1851a), the present anal- ysis also incorporates ontogenetic data from ju- venile specimens, as well as paleontological in- formation unavailable to Owen. Based on this de- scriptive information as well as on functional (Gaudin & Biewener, 1992; Gaudin, 1993; Gau- din & Fortin, unpubl. data) and phylogenetic (En- gelmann, 1978, 1985; Gaudin, 1993, 1995) infor- mation garnered from other sources, a scenario is postulated for the structural evolution of xenar- throus intervertebral facets. Finally, this scenario is used to evaluate the phylogenetic affinity of several extinct early Cenozoic taxa with purported ties to Xenarthra. List of Abbreviations The following abbreviations will be utilized throughout the text: amnh, American Museum of Natural History, New York; fmnh, Field Museum of Natural History, Chicago; LI, L2, L3 . . . first lumbar vertebra, second lumbar vertebra, third lumbar vertebra, respectively, etc.; SI, S2, . . . first sacral vertebra, second sacral vertebra, etc.; Tl, T2, . . . first thoracic vertebra, second thoracic vertebra, etc.; utcm, University of Tennessee at Chattanooga Natural History Museum, Chatta- nooga. Descriptive Anatomy Cingulata Euphracta (esp. Zaedyus pichiy) The vertebral columns of six euphractan ar- madillos (sensu Engelmann, 1985) were exam- ined, two from the species Zaedyus pichiy (fmnh 23809, 104817), three from Chaetophractus vil- losus (fmnh 60467, 122623, 134611), and one from Euphractus sexcinctus (fmnh 152051). The description below is based primarily upon Z p. caurinus, fmnh 104817 (Figs. 2A, 3 A, 4A-D). It should be noted at the outset that vertebrae are bilaterally symmetrical structures. Nearly all of FIELDIANA: GEOLOGY the vertebral facets, foramina, and processes de- scribed in this paper are paired, with the midline neural spine constituting the single major excep- tion. In order to simplify the descriptions and avoid confusion, however, each vertebra is de- scribed unilaterally, with the implicit assumption that the same structures are present and exhibit the same morphology on opposite sides of the ver- tebra unless otherwise stated. Following Walker and Homberger (1992) and Wake (1979), the tho- racic vertebrae are defined as those vertebrae pos- sessing articulations with movable ribs, the sacral vertebrae as those vertebrae that articulate directly with the ilium (or are fused to these vertebrae pos- teriorly), and the lumbar vertebrae as those ver- tebrae that lie in between the thoracic and sacral vertebrae. All of Rower's (1885) euphractan armadillo specimens had 14 dorsal vertebrae — 3 lumbar and 11 thoracic. The fmnh specimens are somewhat more variable. Although most have 14 dorsal ver- tebrae, one Chaetophractus specimen (fmnh 60467) and the sole Euphractus specimen have 15, the former with 10 thoracic and 5 lumbar and the latter with 1 1 thoracic and 4 lumbar. The an- terior thoracic vertebrae of extant euphractans are very similar to those of other mammals, with small, depressed centra and elongated, posteriorly inclined neural spines (Fig. 1; see also Slijper, 1946; Walker & Homberger, 1992). As in many other mammals, the neural spines of the anterior thoracic vertebrae are dramatically longer than those of more posterior thoracic vertebrae. The anterior thoracic vertebrae of euphractans are un- usual, however, in several respects. The centra are functionally opisthocoelus. The main anterior ar- ticular surface of each centrum is nearly flat, but it is flanked by two small lateral facets (Fig. 3A). These facets face anterolaterally, creating a con- vex profile for the whole anterior surface of the centrum. Similarly, the posterior surface of each centrum bears two small lateral facets that face posteromedially (Fig. 3A). This creates a concave posterior articular surface to receive the convex anterior surface. In Zaedyus such functionally opisthocoelous centra are found in all thoracic vertebrae except the last two. The pedicels of the anterior thoracic vertebrae are extremely low and broad anteroposteriorly. Beginning with the third thoracic vertebra, these long, low pedicels occlude the intervertebral fo- ramina from which the spinal nerves typically emerge. Hence the neural arch of each thoracic vertebra from T3 through T10 is perforated by two foramina: one for the ventral branch of each spinal nerve, perforating the pedicel itself, and one for the dorsal branch of each spinal nerve, perforating the lamina of the vertebra just behind the root of the diapophysis (Fig. 4B). The latter process is much higher in euphractans than is typ- ical for mammals; this is due at least in part to the vertically depressed nature of the neural arch. The articular facet for the head of the rib is like- wise more dorsally situated. It lies between the anterodorsolateral part of the vertebral centrum and the posterolateral surface of the pedicel of the preceding vertebra. Finally, the zygapophyseal facets of the anterior thoracic vertebrae are unusual in certain respects. The anterior zygapophysis bears a typical flat, ovate facet situated immediately to one side of the midline. The long axis of this facet runs antero- posteriorly, and the facet faces dorsally. The an- terior zygapophyseal facet articulates with a sim- ilar ventrally directed facet on the underside of the lamina of the preceding vertebra. Both the an- terior and posterior zygapophyseal facets, how- ever, are contiguous or confluent with a second set of facets laterally (Figs. 2A, 3A). These atyp- ical lateral zygapophyseal articulations are also flat and ovate, but with their long axes oriented transversely. Their presence dramatically widens the zygapophyseal articular surface. The anterior lateral zygapophyseal facet faces dorsally and slightly laterally. It articulates with a ventrally facing facet borne on the underside of a rudimen- tary anapophysis (or accessory process) that pro- jects posteriorly from the root of the diapophysis (Fig. 3A). The anterior and posterior lateral zyg- apophyseal facets are present in all euphractan Fig. 2. Thoracic and lumbar vertebrae shown in dorsal view. A, Zaedyus pichiy caurinus, fmnh 104817 — T6, T7, T8, LI, L2, proceeding from right to left. B, Tamandua mexicana, fmnh 69597 — T12, T13, T14, LI, proceeding from right to left. C, Bradypus variegatus, fmnh 69589 — T14, T15, LI, L2, proceeding from right to left. All scale bars = 1 cm. Abbreviations as in Figure 1, plus alz, anterior lateral zygapophyseal facet; amz, anterior medial zygapophyseal facet; ax, anterior xenarthrous facet; ax/alz, fused anterior lateral zygapophyseal facet and anterior xenarthrous facet; D, diaphragmatic vertebra; plz, posterior lateral zygapophyseal facet; pmz, posterior medial zyg- apophyseal facet; px, posterior xenarthrous facet. GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE B amz ax/a I z L1 T14 az L2 Fig. 2. L1 FIELDIANA: GEOLOGY pmz T8 T7(D) r* dp ns ^ .alz -A-amz ,4 /^ yv px T6 ap r>^P pmz . ^ ns ns T 13(D) T12 ns T 15(D) Fig. 2. Continued. ,dp ^mp az T14 GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE ns mp px T7(D) caud T7(D) cran ns B mp '"As* vc T 13(D) caud Fig. 3. Thoracic and lumbar vertebrae shown in cranial and caudal views. A, Zaedyus pichiy caurimis, fmnh 104817 — T6 (cran.), T6 (caud.), T7 (cran.), T7 (caud.), T8, (cran.), proceeding from right to left. B, Tamandua mexicana. fmnh 69597 — T12 (caud.), T13 (cran.). T13 (caud.), T14 (cran.). proceeding from right to left. C, Bradypus variegatus. fmnh 69589 — T15 (cran.), T15 (caud.), LI (cran.), LI (caud.), L2 (cran.), proceeding from right to left. All scale bars = 1 cm. Abbreviations as in Figures 1 and 2, plus cran, cranial view; caud, caudal view: lea, lateral centrum articulation: If. lateral foramen for ventral branch of spinal nerve: rf/px, fused rib facet and posterior xenar- throus facet. FIELDIANA: GEOLOGY T6 caud T6 cran T13(D) cran T12 caud C^-. T15(D) caud T15(D) cran Fig. 3. Continued. specimens examined. In Zaedyus they occur from T3 posteriorly. The anteriormost xenarthrous articulations are between the sixth and seventh thoracic vertebrae. The small anapophysis of the sixth thoracic ver- tebra not only bears a lateral zygapophyseal facet on its underside, but also a small, flat, ovate facet on its dorsal surface (Figs. 2A, 3A, 4A). This lon- gitudinally elongated dorsal facet articulates with a ventrally directed facet carried on the underside of the metapophysis of T7, forming the first true xenarthrous joint (Figs. 3A, 4A). The facet on T7 is actually borne on a small anterior projection that emerges from the base of the metapophysis but lies well above the lamina of the neural arch. The anterior xenarthrous facet of T7, coupled with the normal horizontal zygapophyseal facet on the lamina of the neural arch, forms a slot that re- ceives the anapophysis of T6 (Fig. 2A). Owen (1851a) analogized this interlocking of vertebrae to a carpenter's "mortise and tenon" joint. The seventh thoracic vertebra is the diaphrag- matic vertebra, defined by Slijper (1946) as that in which the anterior zygapophyseal facets are GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE px pmz mp aP ax/px mp ax/px Fig. 4. Zaedyus pichiy caurinus. fmnh 104817: thoracic and lumbar vertebrae shown in right lateral view. A, stereophotographs of T6 and T7. B, stereophotographs of T7 and T8. C, T8 and T9. D, Tl 1 and LI. Scale bar = 1 cm. Abbreviations as in Figures 1 and 2, plus ax/px, xenarthrous intervertebral joint; ax 1/alz, fused anterior lateral zygapophyseal facet and anterior xenarthrous facet; ax 2/px 2, xenarthrous intervertebral joint between secondary anterior and posterior xenarthrous facets; px 1/plz, fused posterior lateral zygapophyseal facet and posterior xenar- throus facet; px 2, secondary posterior xenarthrous facet; spn, foramen/foramina for spinal nerve roots. horizontally oriented and the posterior zygapoph- yseal facets are roughly vertical. T7 is the anter- iormost vertebra to bear a distinct, although small, metapophysis. The metapophyses become pro- gressively elongated posteriorly (Figs. 2A, 4B- D). By the ninth thoracic vertebra, the metapoph- ysis is as long as the neural spine, and by the first lumbar vertebra, the metapophysis exceeds the neural spine in height. On the seventh thoracic vertebra, the base of the metapophysis lies pos- terior to the anterior zygapophyseal facet. On T8, however, the base of the metapophysis contacts the anterior edge of the lamina (Figs. 2A, 3A). This more fully divides the anterior zygapophyse- al facet into medial and lateral components. The portion of the anterior zygapophyseal facet that is medial to the base of the metapophysis (i.e., the anterior median zygapophysis) strongly resembles the postdiaphragmatic anterior zyg- apophyseal facets of other mammals (Pick & Howden, 1977; Walker & Homberger, 1992) and is homologized with these facets by Owen (1851a) and others (Hoffstetter 1958, 1982; Les- sertisseur & Saban, 1967; Gaudin & Biewener, 1992; Rose & Emry, 1993). I concur with this homology. The facet is concave and transversely elongated. The medial half of the facet is hori- zontal and faces dorsally. The lateral half is ver- tical, rolling up onto the base of the metapophysis and facing medially. This facet articulates with a 10 FIELDIANA: GEOLOGY spn pmz px2 spn ax2/px2 convex, ventrolaterally oriented facet on the pos- terior edge of the lamina of T7 (Fig. 3A). The portion of the zygapophyseal facet lateral to the base of the metapophysis in T8 is similar in position to the lateral anterior zygapophyseal facet of T7. It is, however, oriented much more obliquely (indeed, it is nearly vertical) and lies further laterally on the neural arch. It abuts a third intervertebral facet on its dorsomedial edge (Fig. 3A). The third facet, also positioned lateral to the GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 11 base of the metapophysis, is nearly identical to the anterior xenarthrous facet on T7. It is situated on the ventral surface of a small anterior projec- tion at the base of the metapophysis (Figs. 3A, 4B). The facet faces primarily ventrally, although it bears a concave medial lappet that extends onto the lamina and faces laterally. The two facets lat- eral to the metapophysis articulate with the an- apophysis of T7, forming the medial wall of a deep pocket that receives the anapophysis in mor- tise and tenon fashion (Figs. 3A, 4B). The an- apophysis on T7 bears a convex dorsal facet, the xenarthrous facet, as well as a flat, ventromedially oriented facet that appears to be the serial ho- mologue of the lateral posterior zygapophyseal facet of T6. Unlike T6, however, this lateral pos- terior zygapophyseal facet is not in contact with the medial posterior zygapophyseal facet. Rather, the two are divided by a distinct notch (Figs. 2 A, 3A). Because of its apparent serial homology with portions of the more anterior zygapophyseal fac- ets, I have not applied the term "xenarthrous" to the distinct lateral zygapophyseal articulation of T7/T8 and more posterior vertebrae in Zaedyus and other euphractans (as well as other xenar- thrans described below). Rather, in order to en- hance precision and avoid confusion, I will re- strict the term "xenarthrous articulation" to those accessory intervertebral joints that are clearly dis- tinct from the zygapophyseal system of facets. It should be noted that this usage differs markedly from that of most previous authors, who identify a single set of "normal" zygapophyseal facets, either medial or lateral, and then several sets of "xenarthrous" facets (Owen, 1851a; Flower, 1885; Grasse, 1955; Hoffstetter, 1958, 1982; Les- sertisseur & Saban, 1967; Vaughn, 1986; Gaudin & Biewener, 1992; Rose & Emry, 1993). This us- age also differs from that of Jenkins (1970), who merely identifies facets as dorsal, ventral, or in- termediate pre- and postzygapophyses without distinguishing "normal" from "xenarthrous" fac- ets. I believe there are two sets of "normal" fac- ets in xenarthrans, i.e., facets that can be homol- ogized with zygapophyseal facets in more anterior vertebrae. One set lies medial to the metapophy- sis, the other lateral. Previous authors disagree on which set is normal because they assume that only one set can be normal. I believe that in addition to these two sets of "normal" zygapophyseal fac- ets, xenarthrans possess extra "xenarthrous" fac- ets that lack serial homologues in more anterior vertebrae. The anapophysis of T7 is enlarged relative to that of T6, a trend that continues posteriorly (Fig. 4A-D). As noted by Owen (1851a), this enlarge- ment primarily represents an increase in vertical thickness, accompanied by a more modest in- crease in length. As the anapophysis becomes deeper, it participates in bearing, laterally, the fac- et for the head of the rib (starting at T8; Fig. 4C). Not only is the anapophysis of T8 larger than that of T7, but the two facets it carries on its inner surface, the lateral zygapophyseal and the xenar- throus facet, are confluent (Fig. 3A). The lateral zygapophyseal and xenarthrous facets are similar- ly confluent on the anterior edge of T9. The in- tervertebral joints of T9/T10 and T10/T1 1 are vir- tually identical to those of T8/T9. In each case there are two sets of intervertebral joints, one me- dial to the base of the metapophysis, representing the medial zygapophyseal joint, and one lateral to the base of the metapophysis, representing the conjoined lateral zygapophyseal and xenarthrous joints.4 The intervertebral joint between Til and LI differs from the T10/T1 1 intervertebral joint in a number of important respects. The diapophysis, which is progressively reduced posteriorly begin- ning with T8 (Fig. 4A-C), is rudimentary on Tl 1 (Fig. 4D). Concomitant with the reduction of the diapophysis, the foramen for the dorsal branch of the spinal nerve, which lies between the diapoph- ysis and anapophysis anteriorly, occupies a pro- gressively more caudal position on the side of the anapophysis of T8-T10 (compare Fig. 4B with C). It also changes orientation, from a nearly ver- tical course to a horizontal, posteriorly directed course. At the 1 1th and last thoracic vertebra, the foramen reaches the caudal edge of the anapophy- sis (Fig. 4D). The groove leading to this opening divides the articular facets carried on the anapo- physis into separate dorsal and ventral facets. In Chaetophractus a similar dorsal articulation on the ultimate thoracic vertebra contains two facets, a lateral zygapophysis and a xenarthrous facet (see footnote 4). In Zaedyus and Euphr actus, these facets are fused. As in more anterior xenar- 4 The point at which the lateral facets fuse in euphrac- tans is somewhat variable. In Euphractus, the two facets are never separate. In Chaetophractus, however, they re- main separate all the way back to the second lumbar vertebra. Moreover, in the latter genus the lateral zyg- apophysis is split into two facets starting at T7. These two lateral zygapophyseal facets remain separate until T10. Similar facets have also been observed in adult specimens of Tolypeutes (fmnh 121540, 153773). 12 FIELDIANA: GEOLOGY throus joints, the anapophyseal facet(s) above the spinal nerve contacts the ventrolateral surface of the metapophysis and the lateral surface of the neural arch. The serial homology of the ventral facet below the spinal nerve is more difficult to ascertain. Its articulation posteriorly with the dor- sal surface of the transverse process (a pleur- apophysis, not parapophysis, contra Owen, 1851a; see Tolypeutes below) suggests, however, homology with the facet on the lateral surface of the anapophysis that receives the head of the rib in the thoracic vertebrae. The anapophysis of Til appears much more elongate than that of T10 (Fig. 4C, D). This can be attributed to the reappearance of the interver- tebral foramen at Til and the concomitant nar- rowing of the pedicel. The intervertebral foramen is present in all the lumbar vertebrae. The mor- phology of the intervertebral articulations is vir- tually unchanged from Tll/Ll to L3/S1. Tolypeutes matacus (fmnh 124569 [juv.]) This specimen is a neonate. The neural arches are still unfused to the centra. Similarly, the cer- vical and lumbar ribs remain unfused.5 However, the left and right halves of the neural arches are fused in all but the cervical vertebrae. The spec- imen has 15 dorsal vertebrae — 11 thoracic and 4 lumbar. The numbers of thoracic and lumbar ver- tebrae are variable in adult members of the genus Tolypeutes. The fmnh collections include individ- uals with 1 1 thoracic and 4 lumbar vertebrae (fmnh 121540, 124570, 153773) as well as indi- viduals with 12 thoracic and 3 lumbar vertebrae (fmnh 122233). Flower (1885) characterizes To- lypeutes as possessing only 14 dorsal vertebrae (11 thoracic, 3 lumbar), but I suspect that he failed to include the last lumbar vertebra in his count. The last lumbar vertebra is typically fused to the first sacral vertebra in adult specimens. This condition is not uncommon in xenarthrans. It has been observed in the euphractans Euphractus and ' This confirms that the transverse processes of the lumbar vertebrae in adult armadillos (like those of other mammals) are pleurapophyses, i.e., rib attachments of the vertebra plus a fused rib (contra Owen, 1851a). The neonatal lumbar vertebrae lack any lateral projections, precluding the possibility that the adult transverse pro- cesses are parapophyses, i.e., lateral projections of the vertebra that serve as the site of attachment for the ven- tral head of two-headed ribs (Wake, 1979; Kardong, 1995). Chaetophr actus, the giant armadillo Priodontes, and in certain extinct genera of mylodontid sloths (Owen, 1842; Stock, 1925). The anterior thoracic vertebrae of Tolypeutes differ somewhat from those of Zaedyus. They lack the functionally opisthocoelus centra. Although the pedicels are low and longitudinally elongated, only the ventral spinal foramen is present (Fig. 5 A). The dorsal branch of the spinal nerve emerg- es through a notch between the large diapophysis and the rudimentary anapophysis. Only a single pair of zygapophyseal facets is present, and the zygapophyseal surface is narrower mediolaterally in Tolypeutes than in Zaedyus, especially in adult specimens. In fmnh 124569, the diaphragmatic vertebra is T7, and the anteriormost xenarthrous articulation lies between this vertebra and the preceding one, T6. As in Zaedyus, this joint is formed between the dorsal surface of a small anapophysis and the ventral surface of the metapophysis, which is ru- dimentary in this specimen (Fig. 5A). Unlike Zaedyus, this xenarthrous joint between T6 and T7 occurs only on the right side of the specimen. The position of this first xenarthrous joint is ap- parently somewhat variable in Tolypeutes. In fmnh 1 24540, the first xenarthrous joint also oc- curs between T6 and T7, but on the left rather than the right side. Moreover, the diaphragmatic vertebra in this specimen is T8 rather than T7. In fmnh 124570, the diaphragmatic vertebra is T7, but the first xenarthrous joint occurs between T7 and T8. As was the case with Zaedyus, in Tolypeutes the vertebra following the diaphragmatic vertebra bears a large metapophysis whose base reaches the anterior margin of the lamina. This creates three sets of intervertebral joints between T7 and T8: (1) a joint medial to the base of the metapoph- ysis, the medial zygapophyseal joint; (2) a joint lateral to the base of the metapophysis, formed by facets on the dorsal surface of the anapophysis and the ventral surface of the metapophysis; and (3) a joint lateral to the base of the metapophysis, formed by facets on the ventromedial surface of the anapophysis and the lateral surface of the neu- ral arch (Fig. 5A). It is not possible to determine in this juvenile specimen whether or not these two lateral joint surfaces are confluent in the thoracic vertebrae. They are separate, however, in the tho- racic vertebrae of at least one adult specimen (fmnh 153773), becoming confluent in the lumbar vertebrae of both the adult and juvenile speci- mens. As in Zaedyus, these two joints presumably GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 13 B ax/px spn Fig. 5. Tolypeutes matacus, fmnh 124569 (juv-)- A, series of thoracic vertebrae (T5-T8) shown in left lateral view. B, series of lumbar vertebrae (L1-L3) shown in left lateral view. Scale bar = 1 cm. Abbreviations as in Figures 1, 2, and 4, plus lrf, lumbar rib facet. represent xenarthrous and lateral zygapophyseal articulations. As occurs in Zaedyus, the anapophysis and me- tapophysis become progressively larger in the more caudal dorsal vertebrae in fmnh 124569. Also, the intervertebral articulations of the tho- racic vertebrae differ significantly from that of Tll/Ll and those of the lumbar vertebrae. The ventral spinal foramen resumes a typical interver- tebral condition in Tl 1 and all subsequent lumbar vertebrae (Fig. 5B). The dorsal branch of the spi- nal nerve, which emerges ventral to the anapoph- 14 FIELDIANA: GEOLOGY ysis in the majority of thoracic vertebrae, emerges medial to the anapophysis of Til and the lumbar vertebrae, i.e., between the anapophysis and the pedicel of the following vertebra. Below the open- ing for the dorsal branch of the spinal nerve, the anapophysis develops a laterally directed facet for articulation with the last thoracic and the lumbar ribs (Fig. 5B). This facet is absent in a slightly younger specimen (fmnh 124568, in which the right and left neural arches of the posterior tho- racic vertebrae are still unfused). In adult speci- mens it becomes a facet between the anapophysis and transverse process. MacPhee (1994, p. 174) suggested that xenar- throus intervertebral articulations are produced through "a kind of 'sacralization' of the lower part of the free spine," a hypothesis that he be- lieved could be confirmed by the discovery of ac- cessory processes or transitory supplementary in- tervertebral articulations in the early developmen- tal stages of some xenarthran or even non-xenar- thran mammal. The sacral vertebrae of juvenile specimen fmnh 124569 and the younger fmnh 1 24568 were examined for evidence of such pro- cesses or articulations. In the anterior sacral ver- tebrae the pedicels of successive vertebrae are broadly fused lateral to the metapophyses and dorsal to the intervertebral foramina (Fig. 6). It is unclear, however, whether these lateral areas of fusion represent anapophyses, fused sacral ribs, or some other type of structure. Moreover, the xe- narthrous articulations between L4 and S 1 , typical of older Tolypeutes (including fmnh 124569), had not yet developed in fmnh 124568. This suggests that sacral fusion and the development of xenar- thrales in the thoracic and lumbar vertebrae are unrelated. Other Cingulates The number of dorsal vertebrae is somewhat more variable among dasypodid (sensu Engel- mann, 1985) armadillos than in euphractans. The total number of dorsal vertebrae varies between 13 and 16. Thoracic counts range from 13 in Prio- dontes (fmnh 25271) to 9 in Dasypus hybridus (Flower, 1885). The number of lumbar vertebrae varies from 2 in Priodontes (perhaps 3, given the fusion of the last lumbar to the sacrum in this species) to 5 in Dasypus novemcinctus and some Dasypus hybridus (Flower, 1885; Gaudin & Biew- ener, 1992). Despite variation in vertebral number, the mor- phology of the xenarthrous articulations among living adult armadillos is remarkably constant. The anteriormost xenarthrous facets typically oc- cur in the vicinity of the diaphragmatic vertebra, between the diaphragmatic vertebra and either the preceding or the succeeding vertebra (although occasionally xenarthrous joints occur between the first and second prediaphragmatic vertebrae; see Tolypeutes above). The first xenarthrous joints form between the dorsal surface of a small an- apophysis and the ventral surface of a small me- tapophysis. As the metapophysis enlarges poste- riorly and its base reaches the anterior edge of the lamina (as almost always occurs in the first post- diaphragmatic vertebra), the zygapophyseal facet is divided in two. One portion comes to lie medial to the base of the metapophysis. It articulates with a facet on the posteromedial portion of the lamina of the diaphragmatic vertebra. The second portion comes to lie lateral to the base of the metapophy- sis, on the most lateral portion of the lamina. It articulates with a facet borne on the medial sur- face of the anapophysis of the diaphragmatic ver- tebrae. It usually becomes confluent with the xe- narthrous facet in the first few postdiaphragmatic vertebrae. The xenarthrous facet is carried on the ventrolateral surface of the metapophysis and ar- ticulates with the dorsomedial surface of the an- apophysis. These three types of joints — medial zygapophyseal, lateral zygapophyseal, and xenar- throus— are present in all the postdiaphragmatic thoracic vertebrae. A second type of xenarthrous joint is also usually present in armadillos. It is found in the lumbar vertebrae, where it forms be- tween the ventral surface of the anapophysis and the dorsal surface of the transverse process of the following vertebra. A very similar joint is found in the ultimate or penultimate thoracic vertebra, formed by the ventral surface of the anapophysis and the dorsal surface of the rib or transverse pro- cess of the following vertebra. Although the condition of the xenarthrous ar- ticulations cannot be ascertained in glyptodonts because of extensive fusion among the dorsal ver- tebrae (Hoffstetter, 1958; Gillette & Ray, 1981), the dorsal vertebrae remain unfused in several close relatives (following Engelmann, 1985; Pat- terson et al., 1989) of the glyptodonts, the pam- patheres and eutatine armadillos. In pampatheres typical xenarthrous articulations "begin to appear at the anterior end of the thoracic section, and are well developed in the posterior thoracic and lum- bar vertebrae" (Edmund, 1985, p. 88). A posterior thoracic vertebra illustrated by Edmund (1985, GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 15 Fig. 6. Tolypeutes matacus, fmnh 124569 (juv-)- A-C, sacral vertebrae shown in right lateral, ventral, and dorsal views. Scale bar = 1 cm. fig. 9) shows confluent xenarthrous and lateral zygapophyseal facets on the anapophysis and what is perhaps an anapophyseal facet for artic- ulation with the head of the rib. A specimen of Proeutatus (fmnh PI 29 12), a eutatine armadillo from the late early to early middle Miocene Santa Cruz Formation of Patagonia (Scott, 1903-1904), also has a typical cingulate pattern, fmnh PI 29 12 includes the last five thoracic and first lumbar ver- tebrae in articulation. The diaphragmatic vertebra is the second in the series. The most anterior xe- narthrous articulation occurs between the dia- phragmatic vertebra and the first prediaphragmatic vertebra, formed by the ventral metapophysis of the former and the dorsal anapophysis of the lat- ter. Facets medial and lateral to the metapophyseal base are present in the postdiaphragmatic verte- brae, with a third set of facets present between the anapophysis and transverse process in the last tho- racic and first lumbar vertebrae. Priodontes constitutes the only major exception to the cingulate pattern of xenarthrous articula- tions. The giant armadillo possesses all of the ar- ticulations described above. It is characterized, 16 FIELDIANA: GEOLOGY ax1/alz ax 3, 4, 5 Fig. 7. Priodontes maximus, fmnh 25271; first lumbar vertebra shown in cranial view. Scale bar = 1 cm. Ab- breviations as in Figures 1 , 2, and 4, plus ax 3, 4, 5, third, fourth, and fifth anterior xenarthrous facets, respectively. however, by several additional sets of facets in the posterior thoracic and lumbar vertebrae (Fig. 7). Anteriorly, these vertebrae have a single midline facet on the dorsal surface of the lamina, flanked by two pairs of facets also situated on the antero- dorsal edge of the lamina, cranial to the zyg- apophyseal facets. The extra facets articulate with corresponding facets on the posteroventral edge of the lamina of the preceding vertebra. The pres- ence of the extra facets is clearly a derived feature of Priodontes. Vermilingua Tamandua mexicana The vertebral columns of five specimens of the species Tamandua mexicana were examined (fmnh 22398, 58545 |juv.], 69597, 93095 [juv.], and 93176 [juv.]). The description below is based primarily on T. mexicana, fmnh 69597 (Figs. 2B, 3B, 8). The number of dorsal vertebrae tends to be higher in pilosans than in cingulates, particularly in the thoracic portion of the column. The actual numbers of thoracic and lumbar vertebrae in Ta- mandua vary, fmnh 69597 has 17 thoracic and 2 lumbar vertebrae; fmnh 22398 has 18 thoracic and 2 lumbar vertebrae. Both Cuvier (1836a) and Flower (1885) cite specimens of Tamandua with 17 thoracic and 3 lumbar vertebrae. The anterior thoracic vertebrae of Tamandua are less strongly modified than those of cingu- lates, with much taller, narrower pedicels and nor- mal intervertebral foramina. The diapophyses, though large, are less elevated than in the arma- dillos, and the head of the rib articulates exclu- sively with the anterodorsal portion of a single vertebral centrum (Fig. 8). These vertebrae are unusual, however, in four respects. First, the large neural spines are of uncommonly uniform height throughout the thoracic region, decreasing only slightly in height posteriorly. Second, these neural spines are markedly robust and elongated antero- posteriorly (Fig. 8). Third, the zygapophyseal ar- ticular facets are widely separated from the mid- line of the vertebral lamina (Fig. 2B). On the an- terior edge of the lamina this separation is pro- duced by a broad, rounded, midline indentation. Fourth, the zygapophyseal facets themselves are quite wide mediolaterally. Indeed, as can be seen in anterior and posterior views (Fig. 3B), the fac- ets extend further laterally than the vertebral cen- tra, a morphology in sharp contrast to that found in the armadillos (Fig. 3A) and to the generalized mammalian condition (Fig. 1; Flower, 1885; Jen- kins & Parrington, 1976; Jenkins & Schaff, 1988; Walker & Homberger, 1992). All of the thoracic vertebrae have distinct me- tapophyses that increase in size posteriorly. There GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 17 pmz px ap plz Fig. 8. Tamandua mexicana, fmnh 69597: stereophotographs of thoracic vertebrae shown in right lateral view. A, T12 and T13, B, T13 and T14. Scale bar = 1 cm. Abbreviations as in Figures 1, 2, and 4, plus alz/plz, lateral zygapophyseal joint. is no trace, however, of an anapophysis in any prediaphragmatic vertebra (Figs. 2B, 8). The di- aphragmatic vertebra is T13. The first xenarthrous articulation occurs between the diaphragmatic vertebra and T14. As in Zaedyus, the base of the enlarged metapophysis on the first postdiaphrag- matic vertebra reaches the anterior edge of the lamina, dividing the wide anterior zygapophyseal facet in half (Fig. 2B). The portion remaining me- dial to the metapophysis is a concave medial zyg- apophyseal facet that faces dorsomedially. It ar- ticulates with a convex, ventrolaterally oriented facet on the posterior edge of the lamina of T13 (Fig. 3B). This medial joint appears identical to the relatively vertical zygapophyseal articulations found in the posterior thoracic and lumbar verte- brae of non-xenarthran mammals. Lateral and ventral to the metapophysis of T14, a mortise-and-tenon-style articulation is formed with the well-developed anapophysis of T13. In anterior view, the facet lateral to the base of the metapophysis of T14 is parabolic in shape (Fig. 3B), with its dorsal portion facing ventrally and its ventral portion facing dorsally. The ventral portion is nearly identical in position, size, shape, and orientation to the lateral half of the anterior zygapophysis of T13, which is likewise borne on a lateral extension of the anterior vertebral lamina. I believe it to be the serial homologue of the outer half of the anterior zygapophyseal facet. The dor- sal portion is a true xenarthrous facet borne on the ventral surface of a small anterior process pro- jecting from the base of the metapophysis. The anapophysis of T13 has dorsal and ventral facets that are continuous medially. The dorsal facet is a xenarthrous facet. The ventral facet, though sep- 18 FIELDIANA: GEOLOGY arated from the medial zygapophyseal facet by a deep notch, is apparently a lateral zygapophyseal facet (Figs. 3B, 8). Nearly identical intervertebral articulations oc- cur between the vertebrae caudal to T13/T14. The major morphological changes in these more pos- terior vertebrae include a gradual increase in the size of the metapophyses and a gradual decrease in size of the diapophyses. Unlike the situation in Zaedyus, the anapophyses of the posterior thorac- ic vertebrae do not articulate with the ribs. It is therefore not surprising that articulations between the anapophyses and transverse processes at T17/ LI and L1/L2 are absent. There is, however, a third pair of intervertebral articulations that forms between L2 and S 1 . The anapophysis of L2 bears a ventrolaterally directed facet on its lateral edge. It articulates with a dorsomedially oriented facet on the dorsal surface of the sacral rib of S 1 . Several early juvenile specimens of Tamandua were available, including one very young speci- men (fmnh 58545) in which the left and right neu- ral arches of the sacral vertebrae are unfused (in contrast to both juvenile Tolypeutes specimens de- scribed above). This specimen has fairly typical adult-style xenarthrous articulations between the posterior thoracic and the lumbar vertebrae, be- ginning at the diaphragmatic vertebra and extend- ing back to the joint between LI and L2. How- ever, the lumbosacral joint differs markedly from the adult morphology. The last lumbar vertebra bears a large, anteroposteriorly broad transverse process. This in turn carries a very small an- apophysis that lies immediately lateral to a notch for the dorsal branch of the spinal nerve. The an- apophysis does not contact the fused transverse process/sacral rib of S 1 . The last lumbar and first sacral vertebrae are joined by a single pair of wide facets. These facets extend laterally from the base of the neural spine to the medial margin of the notch for the dorsal spinal nerve. In SI the facet passes around the base of the metapophysis, but it is much more extensive medial to this process. The intervertebral connections between subse- quent sacral vertebrae (S1-S3) are virtually iden- tical to those described at the lumbosacral joint. Although the sacral ribs of all three sacral verte- brae appear to carry small anapophyses, there is no indication that these participate in supplemen- tary articulations. A slightly older juvenile Tamandua (fmnh 93176) with unfused sacral vertebrae but fused neural arches was examined to ascertain ontoge- netic changes in the sacral intervertebral articu- lations. The sacral vertebrae of this specimen are joined not by synovial joints but by three areas of rugose bone. One joins the laminae near the mid- line, representing the aforementioned sacral joints of the younger Tamandua specimen. The second unites the vertebral centra. The third zone of at- tachment is a massive area of rugose bone situated lateral to the metapophysis and dorsal to the in- tervertebral foramina, presumably representing a broad zone of contact between the sacral ribs and their attendant anapophyses. It does not involve the metapophyses or vertebral laminae, and thus it bears little resemblance to the xenarthrous con- nections of the dorsal vertebrae. Other Vermilinguas As in its close relative Tamandua, the number of thoracic and lumbar vertebrae in the giant ant- eater Myrmecophaga is variable — 15 thoracic and 3 lumbar or 16 thoracic and 2 lumbar (Cuvier, 1836b; Owen, 1851a; Flower, 1885). The pygmy anteater Cyclopes has 16 thoracic and 2 lumbar vertebrae (Flower, 1885; fmnh 61853, 69969, 69971). The vertebrae of Myrmecophaga are vir- tually identical to those described above for Ta- mandua, except that the joint between the an- apophysis of the last lumbar and the sacral rib is also represented between the ultimate and penul- timate lumbar vertebrae. In these vertebrae, the joint lies between the ventrolateral surface of the anapophysis and the dorsal surface of the trans- verse process of the succeeding vertebra (Flower, 1885). The vertebrae of Cyclopes are likewise similar to those of Tamandua. Cyclopes lacks the joint between the anapophysis of L2 and the sacral rib. More interestingly, rudimentary anapophyses are present in the first prediaphragmatic vertebra (and sometimes in the second), and xenarthrous artic- ulations occur between the diaphragmatic and pre- diaphragmatic vertebrae. These xenarthrales are formed between the dorsal surface of the rudi- mentary anapophysis and the ventral surface of an anterior projection of the metapophysis. Tardigrada Bradypus variegatus The vertebral columns of 2 specimens of Bra- dypus variegatus were examined (fmnh 68919, GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 19 pmz plz I ap tp ap tp Fig. 9. Bradypus variegatus, fmnh 69589. Stereophotographs of thoracic and lumbar vertebrae shown in right lateral view. A, T15 and LI. B, LI and L2. Scale bar = 1 cm. Abbreviations as in Figures 1 and 2. 69589). The description below is based primarily on fmnh 69589 (Figs. 2C, 3C, 9), which has 15 thoracic and 3 lumbar vertebrae. The numbers of thoracic and lumbar vertebrae within the genus vary; there are between 14 and 16 thoracic ver- tebrae and either 3 or 4 lumbar vertebrae (Flower, 1885). The anterior thoracic vertebrae of Bradypus re- semble those of anteaters in that the pedicels are tall and narrow and normal intervertebral foram- ina are present. Similarly, no dramatic difference in the height of the neural spines exists between the anterior and posterior thoracic vertebrae, al- though the height of the neural spines does grad- ually diminish posteriorly. The neural spines are much shorter in Bradypus than in armadillos and anteaters (Figs. 3C, 9). The rib articulations occur at roughly the same level as was observed in Ta- mandua. The head of the rib articulates between successive vertebrae, as in armadillos but not ant- eaters. The lower rib facet is shifted somewhat anteriorly relative to that in armadillos, lying on the posterodorsal edge of the vertebral centrum and the anterior edge of the pedicel of the suc- ceeding vertebra. The zygapophyseal facets of the anterior thoracic vertebrae of Bradypus strongly resemble those of Tamandua. They are trans- versely oval and widely separated by a broad, rounded, midline notch on the anterior edge of the lamina (Fig. 2C). As in anteaters, these zygapoph- yseal facets extend lateral to the vertebral centra in anterior view.6 The lateral extension of the zyg- apophyses becomes increasingly prominent in more posterior thoracic vertebrae. As in Tamandua, there is no trace of an an- apophysis in any anterior thoracic vertebra. There are rudimentary metapophyses present in most thoracic vertebrae (from at least T3 posteriorly). These are situated on the anterodorsal edge of the 6 This is not the case in posterior view, due to the presence of elongated tubercles that extend dorsolater- al^ from the caudal edge of the centrum and articulate with the head of each rib. 20 FIELDIANA: GEOLOGY diapophyses, causing the diapophyses to be lon- gitudinally elongated in dorsal view (Fig. 2C). This is also quite reminiscent of the anteater con- dition. The first distinct metapophysis is small and borne by T14. The metapophyses of succeeding vertebrae are progressively larger (Fig. 2C). T15 is the diaphragmatic vertebra. It differs from that of other xenarthrans in lacking supple- mentary intervertebral articulations. T15 bears a small anapophysis that extends posteriorly from the base of the fused diapophysis/15th rib (Figs. 2C, 3C, 9). The anapophysis closely approximates a lateral projection of the lamina of LI. This ex- tension lies lateral to the base of the metapophysis of LI. As in the postdiaphragmatic vertebrae of other xenarthrans, the base of the metapophysis of LI reaches the anterior margin of the lamina. In contrast to other xenarthrans, however, a true synovial joint does not occur between the small anapophysis and the area lateral to the base of the metapophysis (Fig. 9A). The anapophysis of LI is much larger than that of T15, and it bears a flat, circular, ventromedially directed facet on its medial edge (Figs. 2C, 3C). This facet articulates with a similar facet located lateral to the base of the metapophysis of L2. This lateral facet on L2 is carried on a lateral extension of the neural arch, and it is narrowly divided from the curved, upright medial zygapophyseal facet by the base of the metapophysis (Fig. 2C). The articulation between the anapophysis of LI and the lamina of L2 is nearly identical in position, shape, and orientation to the lateral zygapophy- seal joints of armadillos and anteaters (Fig. 9), and it is hence homologized with these joints. In- terestingly, a dorsal facet on the anapophysis of LI, or an anterior projection or ventrally directed facet on the metapophysis of L2, is absent in Bra- dypus. There is apparently no true xenarthrous joint between these vertebrae. The intervertebral joints between L2 and L3 and between L3 and SI are virtually identical to that between LI and L2 in fmnh 69589. fmnh 68919, however, possesses an additional set of ar- ticular facets between L3 and SI. The transverse process of L3 in this specimen bears a wide, shal- low, ovate articular facet at its distal end. The fac- et forms a synovial joint with a similar facet lying on the anterior edge of the transverse process/sa- cral rib of SI. This joint provides the only evi- dence of nonzygapophyseal supplementary inter- vertebral articulations in this genus. The joint is of further significance because it does not involve the anapophysis at all, in contrast to the supple- mentary articulations found in other xenarthrans. The anapophysis of L3 in fmnh 68919 is situated well medially and forms a typical lateral zyg- apophyseal articulation. Hapalops The vertebral columns of four specimens of Hapalops from the Miocene Santa Cruz Forma- tion of South America (Scott, 1903-1904) were examined. These specimens included Hapalops longipalatus, fmnh PI 3 146, a mounted skeleton with a nearly complete series of vertebrae; Ha- palops sp., fmnh PI 3 145, a partially prepared (ventral surface only) block of articulated lumbar and sacral vertebrae; Hapalops sp., fmnh PI 3 133, a specimen with several discontinuous strings of articulated thoracic vertebrae; and Hapalops sp., fmnh PI 53 18, an articulated series of vertebrae composed of the first lumbar and ultimate and penultimate thoracic elements. The latter speci- men is the best preserved and is emphasized in the description below (Figs. 10, 11). Scott (1903-1904) estimated the number of thoracic vertebrae to be between 21 and 22 in the type specimen of Hapalops longiceps, with 3 lum- bar vertebrae present, fmnh PI 3 146 has 22 tho- racic and 3 lumbar vertebrae, although the column is incomplete, with several thoracic vertebrae and their accompanying ribs reconstructed in plaster. The anterior thoracic vertebrae are typically pi- losan. They have tall pedicels and well-developed intervertebral foramina. The zygapophyseal facets are remarkably wide, even more than in living pilosans. In pilosans and euphractan armadillos, the maximum width of the zygapophyseal facets is about twice the maximum anteroposterior length. In Hapalops the width is two-and-a-half times the length (Table 1 ). The anterior zygapoph- yseal facets are separated from one another by a rounded midline notch somewhat narrower than that observed in Bradypus and Tamandua. As in the living anteater and sloth, these facets extend further laterally than the vertebral centra. The fac- ets become even wider posteriorly and are slightly curved in a horizontal plane, with the medial por- tion extending out from the midline anterolater- al^ and the lateral part oriented almost directly laterally (Fig. 10). The anterior thoracic vertebrae of Hapalops are more reminiscent of anteaters than Bradypus in several respects. The neural spines are quite tall and elongated anteroposteriorly. They are of rel- GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 21 az ns pmz ?dp/tp Fig. 10. Hapalops sp., fmnh P15318; 19th through 21st thoracic vertebrae shown in dorsal view (cranial end toward the top of the page). Scale bar = 1 cm. Abbreviations as in Figures 1 and 2, plus ?dp/tp, pos- sible diapophysis or transverse process. atively uniform height, decreasing only slightly from the middle of the thoracic series posteriorly. As in Tamandua, the articulation for the head of the rib is positioned on the anterodorsal corner of a single vertebral centrum (Fig. 11). The diapoph- yses of Hapalops are much taller than those of the Bradypus. They are longitudinally elongated, as are those of other pilosans, due to the presence of rudimentary metapophyses extending from the anterior edge of the diapophyses. The metapoph- yses become progressively larger posteriorly. The first freestanding metapophysis occurs on T15 in fmnh P13146 (T18 in Scott, 1903-1904, pi. 30). As in Bradypus and Tamandua, anapophyses are absent on all anterior thoracic vertebrae. The diaphragmatic vertebra of fmnh PI 3 146 is T19, as in Scott's (1903-1904) specimens of Ha- palops longiceps and H. elongatus.1 Like Taman- dua, the first supplementary intervertebral articu- lation occurs between the diaphragmatic and post- diaphragmatic vertebrae. In addition, the dia- phragmatic vertebra is the anteriormost thoracic 7 There may be some variation in the position of the diaphragmatic vertebra in Hapalops. In fmnh P15318, the third and last vertebra in the series, the presumed 21st thoracic vertebra, has broken transverse processes and a very weak depression on the anterolateral surface of the centrum for articulation with the head of the rib. These rib facets are very strongly developed on T20 and T19 (Fig. 1 1). This raises the possibility that the vertebra labeled T2 1 is in fact the first lumbar, and the diaphrag- matic vertebra is the penultimate thoracic vertebra. az pmz ?dp/tp Fig. 11. Hapalops sp., fmnh P15318; 19th through 21st thoracic vertebrae shown in left lateral view. Scale bar 1 cm. Abbreviations as in Figures 1, 2, and 10, plus rh, rhachitomous foramen. 22 FIELDIANA: GEOLOGY Table 1 . Ratio of maximum width (parallel to long axis) to maximum anteroposterior length (orthogonal to long axis) of the anterior zygapophyseal facets of the prediaphragmatic thoracic vertebrae in Xenarthra. Maximum Total no. Maximum length Taxon of TV Vertebra width (mm) (mm) Ratio Zaedyus pichiy 11 T3 3.8 1.6 2.4 FMNH 104817 T6 3.5 1.7 2.1 Euphractus sexcinctus 11 T3 5.7 1.9 3.0 FMNH 152051 T5 5.9 2.6 2.3 T8 6.8 3.2 2.1 Chaetophractus villosus 10 T3 4.1 3.6 1.1 FMNH 60467 T5 6.2 2.4 2.6 T7 5.5 2.7 2.0 Tolypeutes matacus 11 fmnh 153773 (T4 measured) T4 2.8 1.4 2.0 fmnh 121540 (T6 measured) T6 2.4 1.7 1.7 Tamandua mexicana 17 T5 5.2 2.9 1.8 fmnh 69597 T8 5.4 3.2 1.7 Til 5.7 3.3 1.7 Bradypus variegatus 16 T4 5.3 2.8 1.9 fmnh 69589 T8 5.2 2.4 2.2 Til 5.3 2.8 1.9 Hapalops sp. 22? fmnh PI 3 133 (mid-TV measured) ?mid-TV 14.9 5.9 2.5 fmnh P15318 (T19 measured) T19 14.8 6.0 2.5 TV = thoracic vertebrae; Tl, T2, T3 = first thoracic vertebra, second thoracic vertebra, third thoracic vertebra, etc. vertebra that bears any trace of an anapophysis. T19 has a small anapophysis extending back from the posteromedial portion of the large diapophy- sis. The anapophysis bears a small flat articular facet ventrally. This facet is oriented nearly hor- izontally and separated by a wide notch from the posterior zygapophysis, which forms a semicylin- drical vertical joint surface lying more dorsal and closer to the midline (Fig. 10). The two posterior facets on T19 form synovial joints with corresponding surfaces on the anterior edge of T20. As in other xenarthrans, the two fac- ets on the anterior edge of T20 are apparently formed by division of the wide thoracic anterior zygapophyseal facet. The base of the metapophy- sis of T20 reaches the anterior edge of the verte- bra, dividing the zygapophyseal facet into medial and lateral portions. The medial zygapophyseal facet is concave. Its medial half lies horizontal, and its lateral half is oriented vertically. It artic- ulates in typical fashion with the semicylindrical posterior zygapophyseal facet of T19 (Fig. 10). The lateral zygapophyseal facet is ovate trans- versely, nearly flat (it is very slightly convex dor- sally), and oriented almost horizontally (it dips slightly ventrolaterally). It is borne on the dorsal surface of an anterolateral extension of the lamina and articulates with the anapophyseal facet of T19 (Fig. 11). The intervertebral joints between the more pos- terior thoracic and lumbar vertebrae, including that between the last lumbar and first sacral ver- tebrae, are nearly identical to those described for T19/T20. The only notable difference in the more posterior joints is an increase in the size of the anapophysis, corresponding to an increase in size of the thin, anteroposteriorly broad pleurapophy- sis of the lumbar vertebrae. As in Bradypus but in contrast to other xenar- thrans, Hapalops typically lacks true xenarthrous articulations, i.e., joints between the ventral and/ or lateral surface of the metapophyseal base and the dorsal or medial surface of the anapophysis. The postdiaphragmatic vertebrae of fmnh P15318 bear distinct depressions on the lateral surface of the metapophyseal bases immediately dorsal and medial to the lateral zygapophyseal facets (Fig. 11). These, however, do not have polished artic- ular surfaces, nor is there any evidence of corre- sponding facets on the anapophyses of T19 and T20. True xenarthrous facets are also absent in fmnh P13133 and P13145, as well as in Scott's (1903-1904) Hapalops material. Small xenar- throus articulations between the ventral surface of GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 23 the metapophysis and the dorsal surface of the anapophysis are sporadically present in fmnh PI 3 146 (e.g., on the right side only of the inter- vertebral joint between T22 and LI, and on both sides of L2/L3). Xenarthrous joints are clearly ab- sent in the thoracic intervertebral joints, even those bearing supplemental articulations. The presence or absence of xenarthrous joints could not be confirmed at L1/L2 or L3/S1 because of poor preservation. Other Tardigrada The extant tree sloth Choloepus and the extinct ground sloth families Mylodontidae, Megatheri- idae, Nothrotheriidae, and Megalonychidae are, like other pilosans, characterized by a relatively large number of thoracic vertebrae and a small number of lumbar vertebrae. Thoracic vertebral counts among the extinct forms range from 15 in Glossotherium and Paramylodon (Flower, 1885; Stock, 1925) to 22 in Hapalops. The highest num- ber among non-Santacrucian ground sloths is 19, which occurs in the Pliocene nothrotheriid Prono- throtherium (fmnh P14503) and the Pleistocene megalonychid Megalocnus (Matthew & Paula Couto, 1959, pi. 26). The number of thoracic ver- tebrae in Choloepus ranges between 22 and 24, the highest among living mammals (Flower, 1885). The tardigrades characteristically possess three lumbar vertebrae, although some Choloepus and some Hapalops have four (Flower, 1885; Scott, 1903-1904), and the Santacrucian genus Schismotherium has five (amnh 9244; Scott, 1903-1904). In several genera, one or more of the lumbar vertebrae are fused to the sacral vertebrae {Glossotherium, Owen, 1842; Paramylodon, Stock, 1925; Megalocnus, Matthew & Paula Cou- to, 1959; Acratocnus, amnh 177616-177619). The anterior thoracic vertebrae among tardi- grades are typically much like those described in Hapalops, with tall, anteroposteriorly elongated neural spines and diapophyses, and wide anterior and posterior zygapophyseal facets that extend further laterally than the vertebral centra and are separated by a small midline notch. The zyg- apophyseal facets appear to be particularly wide among the Santacrucian megalonychimorphs (sensu Gaudin, 1993) and the nothrotheriids (Fig. 12). The zygapophyseal facets of Eremotherium (Paula Couto, 1978) and Scelidotherium (Gervais, 1855) are unusual in that they are not elongated mediolaterally but instead are almost circular in Fig. 12. Pronothrotherium typicum, fmnh PI 4467: isolated mid-thoracic vertebra shown in dorsal view (cranial end toward the bottom of the page). Scale bar = 1 cm. Abbreviations as in Figure 1. shape. They are, however, very widely separated from one another on either side of the midline and, at least in Scelidotherium, their lateral mar- gins lie lateral to the outer margins of the centra. The anterior zygapophyseal facets of the Pleisto- cene West Indian megalonychid Megalocnus, al- though retaining the elongated ovate shape, are oriented anterolaterally at an angle of approxi- mately 45° to the midsagittal plane. The facets are also more widely separated on either side of the midline than is usual for ground sloths. In the ex- tinct Puerto Rican genus Acratocnus (Anthony, 1918) and the extant genus Choloepus (fmnh 147993, 127422), the facets are oriented almost directly anteriorly and separated by a very broad, rounded, midline notch very reminiscent of that in Bradypus. The thoracic vertebrae of these two taxa, particularly in Choloepus, also resemble Bradypus in the degree of reduction of the ver- tebral processes. One other noteworthy modification of the tho- racic vertebrae appears in the anterior and middle thoracic vertebrae of certain large mylodontid and megatheriid ground sloth genera. In these forms some of the thoracic vertebrae bear an additional articular facet both anteriorly and posteriorly. The facets are unpaired and lie on the midline. The anterior facet lies on the dorsal surface of the lam- ina, with its anterior margin situated approximate- 24 FIELDIANA: GEOLOGY ly at the level of the posterior margin of the an- terior zygapophyseal facets. The posterior facet is borne on the undersurface of the neural spine. These midline facets have been noted in Megathe- rium (Owen, 1851b), Eremotherium (Paula Couto, 1978), the large-bodied Santacrucian genus Plan- ops (Hoffstetter, 1961), the scelidotheres (Mc- Donald, 1987), and the mylodontines Paramylo- don harlani (Stock, 1925) and "Mylodon" gar- mani (Allen, 1913). They are absent, however, in the closely related Glossotherium (Owen, 1842). In almost all tardigrades accessory interverte- bral articulations make their initial appearance in the posterior thoracic vertebrae. As in Hapalops, accessory articulations usually occur as a single articular surface between the ventral surface of a rather indistinct anapophysis and the dorsal sur- face of an extension of the vertebral lamina lying lateral to the base of the metapophysis. These ar- ticulations extend from the joint between the di- aphragmatic and postdiaphragmatic vertebrae to the lumbosacral joint. In the majority of taxa, this lateral zygapophyseal articulation is flat, although in Planops (Hoffstetter, 1961) and Paramylodon harlani (Stock, 1925) it curves dorsomedially onto the lateral surface of the metapophysis and the medial edge of the anapophysis. A second accessory articulation, a true xenar- throus facet between the dorsal surface of the anapophysis and the ventral portion of the me- tapophysis, occurs only rarely among tardi- grades. It has, however, been observed in three genera from the Miocene Santa Cruz formation of Patagonia; Hapalops, where it may be vari- ably present, as noted above; Schismotherium, where again its presence is variable (observed in fmnh PI 3 137 but not in amnh 9244 or in speci- mens described by Scott, 1903-1904); and Pre- potherium, a genus closely related to Planops (Scott, 1903-1904). In most tardigrades separate medial and lateral zygapophyseal facets do not occur anterior to the diaphragmatic vertebra, but several taxa exhibit at least incipient separation of the facets in the pre- diaphragmatic vertebra. In the Santacrucian genus Pelecyodon (amnh 9240), the anterior zygapophy- seal surface of the diaphragmatic vertebra is formed by two contiguous facets oriented approx- imately 120° to one another. The medial facet is oriented horizontally, and the lateral facet slopes ventrolaterally. The posterior zygapophyseal fac- ets on the prediaphragmatic vertebra are similarly constructed. A very similar morphology has been described by Stock (1925) in Paramylodon har- lani, where these angled medial and lateral facets are variably expressed between the prediaphrag- matic and diaphragmatic vertebrae. The angled facets may appear bilaterally, on one side only, or be absent. These angled medial and lateral facets have also been observed in Megalonyx, including amnh FLA- 103- 1986, in which the abutting, an- gled medial and lateral posterior zygapophyses are present on one side, while on the other side the two posterior facets are widely separated, with the vertical medial facet and horizontal lateral fac- et typical of diaphragmatic vertebrae in other sloths. The accessory articulations are smaller and fewer in number in sloths compared to those of other xenarthrans. However, the articulations are reduced even further in several tardigrade gen- era. In the North American Pleistocene genus Nothrotheriops, the morphology of the articula- tions is much like that described above for Ha- palops. The diaphragmatic vertebra occurs in the posterior lumbar vertebrae, however, rather than in the posterior thoracic vertebrae. As in other sloths, Nothrotheriops has three lumbar verte- brae, the second of which is the diaphragmatic vertebra, so that the accessory intervertebral ar- ticulations occur only between L2 and L3 and between L3 and SI (Stock, 1925). In Choloepus the reduction is even more significant. The dia- phragmatic vertebra of Choloepus hoffmani (fmnh 127422, 147993) is the 21st, or antepen- ultimate, thoracic, a fairly typical position for a tardigrade. Metapophyses are well developed on T21 and more posterior vertebrae. The anapo- physis first appears as a small nubbin on T23 and becomes progressively larger on LI, L2, and L3. Nevertheless, the only accessory articulation pre- sent occurs at the lumbosacral joint, and even then only on one side of that joint (fmnh 127422 = left; fmnh 147993 = right). In a juvenile Chol- oepus specimen (fmnh 127421) only slightly younger than the juvenile Tolypeutes (fmnh 124569) described above (based on the degree of fusion of the neural arches in the midline), the accessory articulations are absent, as are meta- pophyses and anapophyses. When present, the accessory intervertebral joint of Choloepus re- sembles that of other tardigrades, taking the form of a lateral zygapophyseal articulation between the ventral surface of the anapophysis and the dorsal surface of a facet situated immediately lat- eral to the base of the metapophysis. GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 25 B 2. Medial zygapophyseal facet 3. Lateral 4. Xenarthrous zygapophyseal facet facet Fig. 13. Diagrammatic representation of the morphology of typical xenarthran intervertebral facets in the posterior thoracic or lumbar vertebrae. A, anterior view; B, posterior view. Abbreviations as in Figure 1. Conclusions Morphological Summary The morphological data described above are summarized in Appendix Table 1 (p. 36). The in- tervertebral articulations between the thoracic and lumbar vertebrae of xenarthrans can be placed into four distinct categories. 1. Normal zygapophyseal facets. The interver- tebral articulations between the anterior thoracic vertebrae in most xenarthrans differ little from the thoracic zygapophyseal facets found in other mammals (Fig. 1; Walker & Homberger, 1992). Morphological departures from the normal mam- malian pattern include widening of the anterior and posterior zygapophyses in pilosans and eu- phractan armadillos (Table 1, Fig. 3B), so that the zygapophyses extend further laterally than the vertebral centra. In addition, pilosans possess a broad midline notch that separates the anterior zygapophyseal facets (Fig. 2B,C). Typical mammalian zygapophyses are present only in the prediaphragmatic vertebrae of the ma- jority of xenarthrans. The exceptions are the liv- ing tree sloths Bradypus and Choloepus, in which the normal mammalian morphology extends from one {Bradypus) to six vertebrae {Choloepus) pos- terior to the diaphragmatic vertebrae. In addition, in euphractan armadillos and some extinct sloths, the zygapophyses of some prediaphragmatic tho- racic vertebrae begin to show evidence of incipi- ent division into distinct medial and lateral facets (see below). 2. Medial zygapophyseal facets. In all xenar- thrans (except glyptodonts, in which the vertebrae are fused), the postdiaphragmatic vertebrae bear on each side a curved, vertically oriented inter- vertebral facet that lies medial to the metapoph- ysis and adjacent to the midline anteriorly (Fig. 13). Posteriorly, the corresponding facet lies at the base of the neural spine adjacent to the midline. As noted above, the facets are homologized with the zygapophyseal facets of other mammals by Owen (1851a) and others (Hoffstetter, 1958, 1982; Lessertisseur & Saban, 1967; Gaudin & Biewener, 1982; Rose & Emry, 1993; contra Flower, 1885; Grasse, 1955; Vaughn, 1986). These facets are structurally identical to those pre- sent in the postdiaphragmatic vertebrae of other mammals (see Fig. 1). Given this fact, and their near-universal distribution within Xenarthra, I see no reason to doubt this homology. In several xenarthran taxa, the medial portion of the zygapophyseal facet of some prediaphrag- matic vertebrae is separated from the lateral por- tion by a small groove or ridge. This separation occurs in the first prediaphragmatic vertebra of several ground sloth taxa. In the euphractan ar- madillos it may occur as far forward as the third thoracic vertebra. 3. Lateral zygapophyseal facets. A majority of xenarthran taxa possess an anterior facet on each side of the postdiaphragmatic vertebrae that lies on a lateral extension of the vertebral lamina, sit- uated immediately lateral to the base of the me- tapophysis (Fig. 13). A corresponding facet is found posteriorly, on the ventromedial surface of the anapophysis. As noted by Jenkins (1970) for 26 FIELDIANA: GEOLOGY Tamandua, these facets closely resemble in struc- ture and position the lateral portion of the zyg- apophysis of the prediaphragmatic vertebrae. I therefore suggest that these facets are serially ho- mologous with the lateral parts of the zygapophy- seal articulations of the prediaphragmatic verte- brae. These facets are considered accessory zyg- apophyseal articulations, termed "lateral zyg- apophyseal facets," and are distinct from true "xenarthrous" facets. The homology of these lateral zygapophyseal articulations is less secure than that of the medial zygapophyseal facets because their taxonomic distribution is not entirely congruent with the tax- onomic distribution of the medial facet. Two taxa, the tree sloths Bradypus and Choloepus, lack lat- eral zygapophyseal articulations in at least some of the postdiaphragmatic vertebrae. However, the normal zygapophyseal articulations are poorly de- veloped in both of these taxa, with successive ver- tebrae fitting together only loosely. Moreover, the vertebral processes are weak in both taxa. It there- fore seems likely that the absence of lateral zyg- apophyseal facets in these taxa represents a sec- ondary reduction. 4. True xenarthrous facets. There are several different types of facets found in various xenar- thran taxa that can be categorized as true "xenar- throus" accessory intervertebral articulations. The most common is a facet borne, on each side, on the ventrolateral surface of an extension projecting from the base of the metapophysis; the corresponding posterior facet lies on the dorsal surface of the anapophysis (Fig. 13). Such facets are found on the diaphragmatic and postdiaphrag- matic vertebrae of all cingulates and vermilin- guas, and may extend as far forward as the second prediaphragmatic vertebra in these groups. These facets are also variably present in a few genera of early megalonychimorph sloths (Hapalops, Schis- motherium, Prepotherium), and hence they likely represent the primitive condition for tardigrades as well. They are often fused to the lateral zyg- apophyseal facets. A second type of facet occurs on the ventrolat- eral surface of the anapophysis of some thoracic vertebrae in armadillos for articulation with a rib. Such facets are found on T8-T10 (the penultimate thoracic vertebra) in Zaedyus (Fig. 4C) and T6 or T7-T9 (again the penultimate thoracic vertebra) in Dasypus, but have not been observed in any pilosan. These articular surfaces do not serve an intervertebral function. However, the anapophysis of the last thoracic vertebra of cingulates carries a ventrolateral articular facet that apparently cor- responds to the facet just described for more an- terior vertebrae. In the ultimate thoracic vertebra it articulates with a facet carried on the dorsal surface of the transverse process of LI (Fig. 4D). This type of articulation continues posteriorly through the lumbar vertebrae to the lumbosacral joint. A similar facet is present between the ven- tral portion of the anapophysis and the dorsal por- tion of the transverse process/sacral rib in Myr- mecophaga (at L1/L2 and the lumbosacral joint) and Tamandua (at the lumbosacral joint only), but not in Cyclopes. This type of facet is unknown in sloths, although some specimens of Bradypus are characterized by an articulation formed between the distal tip of the transverse process of L3 and the anterior edge of the sacral rib. Finally, the vertebrae of several groups of xe- narthrans possess unpaired midline facets that may be generally categorized as xenarthrous. The anterior facet is typically found on the dorsal sur- face of the vertebral lamina, posterior to the zyg- apophyseal articulations and anterior to the base of the neural spine. The posterior facet lies on the undersurface of the neural spine. These facets are particularly characteristic of large-bodied xenar- thrans, such as the giant armadillo Priodontes (Fig. 7) and megatheriid, scelidotheriine, and some mylodontine ground sloths. Phylogeny and Evolution of Xenarthrous Vertebrae When the data on intervertebral articulations are plotted on a phylogeny of Xenarthra (Fig. 14), a number of features may be identified that appear to be primitive characteristics of xenarthrous ver- tebrae. The prediaphragmatic thoracic vertebrae probably have abnormally wide zygapophyses, al- though this feature is absent in many cingulates. The metapophyses are large, especially in the pos- terior thoracic and lumbar vertebrae. In the post- diaphragmatic vertebrae, these large metapophy- ses split the zygapophyseal articulations into sep- arate medial and lateral joints. In addition, each metapophysis bears a ventrolateral articular facet that articulates with the dorsal surface of the an- apophysis of the preceding vertebra. Large an- apophyses may also be a primitive characteristic of xenarthrous vertebrae. However, the anapo- physes are only weakly developed in sloths and in the anterior xenarthrous vertebrae of cingulates. The goal of the present study was not only to GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 27 Cingulata l)Wide zygapophyseal facets (only in euphractans) 2) Enlarged metapophyses 3) Enlarged anapophyses (only in post. thor. and lumb. vert.) 4) Medial and lateral zygapophyseal facets in post-diaphragmatic vertebrae 5) Xenarthrous articulations between: metapophysis and anapophysis anapophysis and rib (thor. vert.) anapophysis and transverse process (lumb. vert.) neural spine and lamina (Tolypeutes & Priodontes only) Vermilingua l)Wide zygapophyseal facets 2) Enlarged metapophyses 3) Enlarged anapophyses 4) Medial and lateral zygapophyseal facets in post-diaphragmatic vertebrae 5) Xenarthrous articulations between: metapophysis and anapophysis anapophysis and transverse process (lumbo-sacral joint, Tamandua & Myrmecophaga only) Tardigrada l)Wide zygapophyseal facets 2) Enlarged metapophyses (in most taxa) 3) Weak anapophyses 4) Medial and lateral zygapophyseal facets in post-diaphragmatic vertebrae (in most taxa) 5) Xenarthrous articulations between: metapophysis and anapophysis (several Santacrucian genera) neural spine and lamina (large ground sloths only) PilOSa l)Wide zygapophyseal facets 2) Enlarged metapophyses 3) ? anapophyses 4) Medial and lateral zygapophyseal facets in post-diaphragmatic vertebrae XENARTHRA 5) Xenarthrous articulations DVVide zygapophyseal facets ^^ metapophysis and 2) Enlarged metapophyses anapophysis 3) Enlarged anapophyses, (in post thor. and lumb. vert.) 4) Medial and lateral zygapophyseal facets in post-diaphragmatic vertebrae 5) Xenarthrous articulations between metapophysis and anapophysis Fig. 14. Phylogeny of the Xenarthra (following Engelmann, 1985; Gaudin, 1993), showing the distribution of vertebral character states among the three major suborders, and the inferred character states at the ancestral nodes "Pilosa" and "Xenarthra." For further description of each character state, see text. Abbreviations: lumb. vert., lumbar vertebrae; post. thor. vert., posterior thoracic vertebrae; thor. vert., thoracic vertebrae. analyze the morphology of xenarthrous vertebrae across the Xenarthra as a whole, but also to un- derstand the structural evolution of these acces- sory intervertebral facets. Although determining the primitive morphology of xenarthrous verte- brae is an important step in understanding their evolution, a more complete explanation requires information on intermediate conditions leading to the appearance of fully developed xenarthrales. The difficulty in determining the structural gen- esis of xenarthrous intervertebral facets stems from what MacPhee (1994, p. 174) described as an "apparent absence of any recognizable inter- mediate condition between nomarthry and xenar- thry." As mentioned above, however, MacPhee (1994, p. 174) suggested that the development of accessory intervertebral facets may be the result of a process of "sacralization," whereby the pos- terior dorsal vertebrae become more solidly at- tached to one another in a manner analogous to the intimate union characteristic of mammalian sacral vertebrae. Such an analogy seems particu- larly apt for an animal such as Scutisorex (Les- sertisseur & Saban, 1967; Kingdon, 1984), in which the posterior thoracic and lumbar vertebrae are united laterally by closely interlocking bony 28 FIELDIANA: GEOLOGY excrescences.8 It is less clear whether the analogy can be usefully applied to xenarthrans, in which the vertebrae are not synostosed laterally or con- nected by tight fibrous joints but rather are con- nected by extra synovial joints that limit interver- tebral mobility (Gaudin & Biewener, 1992). Whereas MacPhee (1994) noted that the sacral el- ements in some mammals remain unfused throughout life and hence might serve as an ap- propriate xenarthran analogue, the fused condition is certainly primitive for xenarthrans, and perhaps for mammals as a whole (Jenkins & Schaff, 1988; Kielan-Jaworowska & Gambaryan, 1994; but see also unfused sacral vertebrae in Jenkins & Par- tington, 1976; Krebs, 1991; Marshall et al., 1995). In support of MacPhee 's (1994) hypothesis is the observation that in several xenarthran taxa some or all of the lumbar vertebrae are fused to the sacral vertebrae. This lumbosacral fusion oc- curs in the dasypodid armadillos Priodontes and Tolypeutes, in the euphractan armadillos Euphrac- tus and Chaetophr actus, in glyptodonts (Gillette & Ray, 1981), in some mylodontine sloths (Owen, 1842; Stock, 1925), and in some West Indian me- galonychid sloths (Matthew & Paula Couto, 1959). Nevertheless, the phylogenetic distribution of lumbosacral fusion suggests that it is a derived feature of these lineages and not a primitive fea- ture of Xenarthra. Moreover, in the juvenile spec- imens of Tolypeutes and Tamandua described above, the lumbosacral joint is the last to develop accessory intervertebral joints. This observation would seem to argue against a claim of some structural continuity between sacral and more an- terior vertebrae. MacPhee (1994) asserted that the demonstra- tion of accessory intervertebral joints in the sa- crum of developing xenarthrans (or indeed non- xenarthran mammals) would strongly support the sacralization hypothesis. No evidence of acces- sory synovial joints is present, however, in the skeletal remains of very young juvenile individ- uals representing all three major subgroups of xe- narthrans (armadillos, anteaters, tree sloths) in fmnh collections. Indeed, in the youngest Taman- dua specimen examined the sacral anapophyses do not contact the succeeding vertebrae. Rather, the sacral vertebrae were joined through wide 8 Such bony excrescences are also present in several fmnh specimens of Tamandua, e.g., fmnh 137419, 140912, and 150733. All, however, are zoo specimens. It therefore seems likely that this condition is an age- related pathological state. zygapophyseal facets reminiscent of those in the anterior thoracic vertebrae. In a young juvenile Choloepus specimen, the sacral vertebrae lacked anapophyses altogether. It would seem at present that there is little evidence in support of sacral- ization as the process through which xenarthrous articulations originated. In contrast to MacPhee's (1994) claim, I sug- gest that documented structural intermediates be- tween xenarthrous and nomarthrous vertebrae do exist. Among tardigrades, the xenarthrous articu- lations are reduced relative to the primitive con- dition for the order as whole. The reduction in- cludes not only a diminution in the number of intervertebral joints that possess accessory artic- ulations (see Bradypus, Nothrotheriops, and Chol- oepus, above), but also a simplification of the ac- cessory articulations themselves. The vast major- ity of sloths lack a xenarthrous joint between the ventral surface of the metapophysis and the dorsal surface of the anapophysis. Weakly developed an- apophyses are universally characteristic of sloth vertebrae. Sloths retain only the wide predia- phragmatic zygapophyseal facets, large met- apophyses, and medial and lateral postdiaphrag- matic zygapophyseal facets characteristic of prim- itive xenarthran vertebrae. Even these medial and lateral postdiaphragmatic zygapophyseal facets may be restricted to the last few lumbar vertebrae in several taxa, most notably the extant tree sloths. The presence of xenarthrous joints between the metapophysis and anapophysis in several relative- ly early sloth genera, e.g., Hapalops and Prepo- therium, suggests that the xenarthrales in sloths are reduced secondarily. Although the manner in which these facets are reduced need not parallel the process by which they originate, I believe that the sloth condition, as the only known structural intermediate between xenarthry and nomarthry, represents the best available tool for understand- ing the evolutionary origin of xenarthrous verte- brae. Using the condition in sloths as a model, I sug- gest that the first step in the evolution of xenar- throus articulations was a widening of the zyg- apophyseal facets and enlargement of the met- apophyses, both becoming more extensive in pro- gressively more posterior thoracic and lumbar vertebrae. The widening of the zygapophyses should increase the lateral moment of resistance of the joints and hence enhance the stability of the vertebral column in lateral bending (see be- low). The enlarged metapophyses would provide a larger area for origin of the epaxial transverso- GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 29 spinalis muscles. At the joint between the dia- phragmatic and first postdiaphragmatic vertebrae, the metapophyses would become sufficiently large to split the widened zygapophyseal facets perforce into separate medial and lateral zyg- apophyseal joints. These medial and lateral zyg- apophyseal joints would then continue posteriorly to the lumbosacral joint. The second step in the evolution of xenarthrales would involve the development of the other two primitive features of xenarthrous vertebrae, an en- larged anapophysis and a xenarthrous joint be- tween the dorsal surface of the anapophysis and the ventrolateral surface of the metapophysis. These two structurally linked modifications would serve to further enhance the stability of the spinal column, particularly in lateral and dorsal bending (Gaudin & Biewener, 1992). The final steps in the evolution of the xenarthrous vertebrae would in- volve the appearance of the various types of spe- cialized xenarthrous facets in the major xenar- thran subgroups, e.g., the joints between anapoph- yses and transverse processes in cingulates and some anteaters. These presumably function to provide additional stiffness to the vertebral col- umn while preserving a certain degree of mobil- ity. Secondary loss of some of these facets in sloths might be the result of their switch from a digging to a more terrestrial or semiarboreal hab- itus (Gaudin, 1993; White, 1993a,b). The above model contrasts with that of Mac- Phee (1994) in postulating that xenarthrales de- velop from the posterior thoracic vertebrae back rather than from the sacral vertebrae forward. The new model accords well with the observation that xenarthrous articulations in living xenarthrans tend to become more complex caudally, and that these more complex caudal articulations appear later in development (see Tolypeutes and Taman- dua above). It conforms with what is known about the mechanics of the mammalian spine. The mam- malian backbone tends to exhibit maximum flex- ibility in the mid-dorsal region, in the vicinity of the diaphragmatic vertebra (Slijper, 1946; Jenkins, 1974). Thus one might predict that adaptations to reduce spinal mobility might first appear in this region. Finally, the model fits well with what is known about the functional morphology of xe- narthrous vertebrae. Gaudin and Biewener (1992; see also Gaudin, 1993) have recently reaffirmed the idea that Xenarthra represents an offshoot of early placental mammals that were primitively specialized for digging. They note further that in digging mammals, the axial musculoskeletal sys- tem has a particularly important role to play in resisting the large dorsal and lateral reaction forc- es generated by digging. The axial skeleton of xe- narthrans is stiffened relative to the primitive mammalian condition in dorsal and lateral bend- ing. The appearance of wide zygapophyses in ear- ly xenarthrans would effectively increase the lat- eral moment of resistance of the vertebrae, stiff- ening the intervertebral joints in lateral bending. Because xenarthrans dig primarily with their fore- limbs, the vertebral column is loaded as a canti- lever, with loads increasing posteriorly. This would account for the progressive increase in width of the zygapophyseal facets in more caudal vertebrae. Large metapophyses would result from hypertrophy of transversospinalis muscles, which, according to Gaudin and Fortin (unpubl. data), play an important role in stabilizing the vertebral column during dorsal bending. Again, these mus- cles would be expected to increase in size poste- riorly in concert with the increase in dorsal bend- ing forces. It is of particular interest that the above model of xenarthrous vertebral evolution does not ini- tially include enlargement of the anapophyses. Enlarged anapophyses have been considered by several authors (Simpson, 1931; Ding, 1987; Storch, 1981; see below) as indicating incipient xenarthry. However, Gaudin and Fortin (unpubl. data) note that the longissimus dorsi muscles, which take their origin from the anapophyses, are not enlarged in xenarthrans relative to the primi- tive mammalian condition and likely play little or no role in enhancing dorsal and lateral stiffness. Further, Gaudin and Biewener (1992) were able to cut the anapophyses in the posterior lumbar vertebrae of Dasypus without significantly de- creasing either dorsal or lateral stiffness. Gaudin and Biewener (1992) presented evidence from strain gauge analyses suggesting that enlarged an- apophyses serve to reduce shear stress in dorsal bending, to augment the lateral moment of resis- tance in lateral bending, and to transmit forces from the forelimbs to the robust pelvic girdle and hind limbs in both dorsal and lateral bending. Nevertheless, the functional and phylogenetic ev- idence appears to indicate that enlarged anapoph- yses are not a necessary structural antecedent of xenarthrous intervertebral articulations. Relationship of Xenarthra to Early Cenozoic Fossil Taxa The final goal of the present study was to utilize the above conclusions on the structural evolution 30 FIELDIANA: GEOLOGY of xenarthrales to evaluate the taxonomic affinity of several enigmatic early Cenozoic taxa with al- legedly close ties to Xenarthra. Of particular con- cern are the Palaeanodonta, a group of fossorial mammals with reduced dentitions that is known from Paleogene strata of North America (Rose et al., 1991, 1992; Gunnell & Gingerich, 1993) and Europe (Hessig, 1982); Ernanodon, a taxon based on a single skeleton found in Late Paleocene sed- iments from China (Ding, 1987); and Eurotaman- dua, a purported anteater from the Middle Eocene Messel fauna of Germany (Storch, 1981). Simpson (1931) proposed a close phylogenetic link between palaeanodonts and xenarthrans based on various lines of evidence, including the morphology of the vertebral column (see Emry, 1970, for contrasting interpretations). In his de- scription of the vertebrae of the Eocene palaean- odont Metacheiromys, Simpson (1931, p. 334) stated that "While xenarthrous articulations are not definitely incipient in Metacheiromys, . . . [its morphology] seems ... an ideal point of departure for the origin of the secondary articulations." Yet Matthew (1918, p. 629), in his description of the Late Paleocene (or Early Eocene; see Gunnell & Gingerich, 1993) genus Palaeanodon, could find "no recognizable foreshadowing of the peculiar 'xenarthral' articulations" characteristic of true xenarthrans. In part the discrepancy results from the poor state of preservation of the vertebrae of Palaean- odon. Other Paleocene palaeanodonts lack pre- served vertebrae (Rose, 1978, 1979), and hence shed no light on the problem. Better preserved material exists from the Early and Middle Eocene, representing both families of palaeanodonts (Epoicotheriidae and Metacheiromyidae). In the epoicotheriid Alocodontulum and the metacheiro- myid Metcheiromys, certain resemblances to xe- narthran vertebrae can be observed (Simpson, 1931; Rose et al., 1992). The metapophyses are enlarged, particularly in the posterior thoracic ver- tebrae in the vicinity of the diaphragmatic verte- bra. Unlike xenarthrans, however, the metapophy- ses become progressively smaller in more poste- rior lumbar vertebrae. It is unclear whether or not the thoracic zygapophyseal facets of palaeano- donts are particularly enlarged transversely. Rose et al. (1992, p. 226) describe the lumbar prezy- panophyses of Alocodontulum as "broad." Broad , mediolateral width much greater than antero- .terior length) zygapophyseal facets have also ;n observed in the lumbar vertebrae of the Ear- ly Eocene epoicotheriid Pentapassulus (usnm 20028; Gaudin, pers. observ.) and the diaphrag- matic vertebra of Metacheiromys (usnm 26132; Gaudin, pers. observ.). They are not so broad, however, as in xenarthrans, in which width often exceeds anteroposterior length by a factor of two or more (Table 1). Simpson (1931, fig. 12c) fig- ured narrow anterior zygapophyseal facets on the 10th or 11th thoracic vertebra of Metacheiromys, but these facets are situated well lateral to the midline. In no case, however, are distinct medial and lateral zygapophyseal facets observed in any palaeanodont. The anapophyses of palaeanodonts are enlarged (Simpson, 1931; Rose et al., 1992), again in the vicinity of the diaphragmatic verte- bra. Although relatively deep on T12 and T13 of Alocodontulum (Rose et al., 1992, fig. 3), the an- apophyses of palaeanodonts tend to be dorsoven- trally narrow and spine-like. In this respect they more closely resemble the anapophyses of carni- vores than those of xenarthrans (Rose et al., 1992; Rose & Emry, 1993). Furthermore, there is no indication of articular surfaces between the ana- pophyses and the succeeding metapophyses or transverse processes. On balance, it would appear that palaeanodont vertebrae share few if any of the derived charac- teristics of xenarthran vertebrae unequivocally. The strongest similarity between the vertebrae of the two groups is the enlargement of the meta- pophyses. This specialization, however, is known to occur in many groups of fossorial mammals (Gaudin & Fortin, unpubl. data), and hence may likely be convergently acquired. Among other supposed similarities to Xenar- thra, Ding (1987, p. 92) noted the posterior tho- racic vertebrae of Ernanodon, which bear "lon- gitudinally elongated projections under the mam- millary process [= metapophysis], which are in- cipiently developed xenarthrous articulations." As in xenarthrans and palaeanodonts. the posterior thoracic and lumbar vertebrae of Ernanodon are characterized by well-developed metapophyses. A pair of large vertebral processes extend lateral to the metapophyses, but do not form joints with the metapophyses. These lateral processes (Ding, 1987, fig. 7) apparently represent large anapophy- ses. In the posterior thoracic vertebrae the ana- pophyses arise anteriorly from the back of rather low, weakly developed diapophyses, a condition that only vaguely resembles that characteristic of most true xenarthrans. In the lumbar vertebrae, the anapophyses arise from short, anteroposteri- orly elongated transverse processes, the latter reminiscent of those found in extinct sloths. The GAUDIN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 31 anapophyses are relatively deep in the thoracic vertebrae, but are oriented more dorsally than among unequivocal xenarthrans. The anapophyses of the lumbar vertebrae are not illustrated in lat- eral view, and the width of the zygapophyseal fac- ets in Ernanodon is unclear. They are described as "oval" in the anterior thoracic vertebrae, in- creasing in size posteriorly (Ding, 1987, p. 92). However, the zygapophyseal joints of the lumbar vertebrae are enrolled, as are those of pholidotans (Emry, 1970); such a condition is uncommon among true xenarthrans, where enrollment occurs in only a few derived taxa (e.g., Euphractus, Prio- dontes). There is no evidence of any accessory intervertebral articulations in Ernanodon, be they lateral zygapophyses or xenarthrous joints be- tween anapophyses, metapophyses, and transverse processes. As with the palaeanodonts, the case for strong shared derived similarity between the vertebrae of Ernanodon and undoubted xenarthrans is weak. The most notable resemblances are the enlarged metapophyses and anapophyses. However, the former are widespread among digging forms, as noted above. The latter differ slightly in their mor- phology from those of xenarthrans. Moreover, it is not clear that enlarged anapophyses are char- acteristic of Xenarthra primitively. Eurotamandua presents the most striking claim for derived resemblance between the vertebrae of undoubted xenarthrans and a non-South American early Cenozoic taxon. As in palaeanodonts and Ernanodon, the vertebrae of Eurotamandua pos- sess large metapophyses, beginning three to four vertebrae anterior to the diaphragmatic vertebra and extending posteriorly through the lumbar ver- tebrae (Storch, 1981). Similarly, enlarged an- apophyses are present in the diaphragmatic region of the vertebral column. Unlike what is found in palaeanodonts and Ernanodon, the anterior tho- racic vertebrae of Eurotamandua are very remi- niscent of armadillo (but not anteater) vertebrae, with elongated, anteroposteriorly narrow neural spines, broad laminae, and elevated diapophyses. The most remarkable aspect of the vertebral mor- phology in this genus, however, is its purported possession of true accessory intervertebral artic- ulations. The diaphragmatic vertebra of Eurotamandua, the last thoracic according to Storch (1981), has a large anapophysis that apparently bears both dorsal and ventral articular facets. By analogy with unequivocal xenarthrans, the former would represent either a xenarthrous facet between the anapophysis and metapophysis or a combined an- apophyseal/metapophyseal xenarthrous facet and a lateral zygapophyseal facet. The ventral articu- lation would then represent a xenarthrous joint be- tween the anapophysis and the transverse process of the first lumbar vertebra. Note that a joint be- tween the anapophysis and transverse process is not a primitive characteristic of xenarthran or ant- eater vertebrae, according to the phylogenetic dis- tributions presented above (Fig. 14). Storch (1981) suggests that less elaborate accessory in- tervertebral articulations can also be observed be- tween the diaphragmatic and the preceding ver- tebra, and perhaps even at the next anterior inter- vertebral junction. Unfortunately, there remain a number of diffi- culties that render Storch's (1981) morphological assessments open to question. Several subsequent workers have been unable to verify the presence of the xenarthrous articulations in Eurotamandua (Rose & Emry, 1993; Szalay & Schrenk, 1994). Szalay and Schrenk (1994, p. 48A) state that "xe- narthry cannot be corroborated (and not only be- cause of the nature of preservation)." The lumbar vertebrae, which on the basis of comparison with undoubted xenarthrans would be expected to show the most well-developed xenarthrous artic- ulations, are preserved with only the dorsalmost portions of the vertebrae exposed. This prevents observation of any lateral accessory joints. In ad- dition, the morphology of the accessory joints themselves, particularly those of the prediaphrag- matic vertebrae, is unusual (Storch, 1981, fig. 8). The two vertebrae anterior to the diaphragmatic vertebra have narrow, spine-like anapophyses, very similar to those found in palaeanodonts and unlike anything known in Xenarthra. These an- apophyses allegedly articulate with a small con- cave facet or depression on the pedicel of the suc- ceeding vertebra. This latter facet lies midway be- tween the base of the metapophysis dorsally and the base of a second process ventrally. The more ventral process apparently carries a facet for the rib tubercle distally, and hence it must represent a ventrally situated diapophysis.9 This type of ar- ticulation, between the anapophysis anteriorly and the metapophysis and diapophysis posteriorly, is not present in any known xenarthran. 9 Storch (1981) labels it a parapophysis, but I am not aware of any mammal that possesses a true parapophysis in the posterior thoracic vertebrae. Hence I believe it more likely to represent a ventrally displaced diapoph- ysis, articulating with the tubercle rather than the head of the rib. 32 FIELDIANA: GEOLOGY In summary, the case for derived resemblance between the vertebrae of Eurotamandua and un- doubted xenarthrans remains less than compel- ling. Like palaeanodonts and Ernanodon, Euro- tamandua does possess enlarged metapophyses and anapophyses, but these are of equivocal phy- logenetic utility. The width of the zygapophyses in Eurotamandua cannot be assessed from the published descriptions and figures. Storch (1981) claims that accessory intervertebral articulations are present, including a possible lateral zygapoph- yseal joint and xenarthrous joints between the an- apophysis and metapophysis, the anapophysis and transverse process, and the anapophysis/ metapophysis and diapophysis. However, the lat- ter joint is unknown in undoubted xenarthrans, and the presence of all such joints has been ques- tioned by subsequent workers. On the basis of vertebral morphology, there is at present little evidence that would clearly sug- gest a close phylogenetic relationship between true xenarthrans and palaeanodonts, Ernanodon, or Eurotamandua. Assessment of the vertebral morphology in the latter two taxa is hindered by the paucity of available material and the incom- plete preservation of the existing material. More complete information on the vertebral column of either Ernanodon or Eurotamandua (in particular the latter) would allow a more reliable assessment to be made of their potential ties with Xenarthra. Good material is available for Eocene and Oli- gocene palaeanodonts. The Paleocene material in this group is still poorly known, and thus it is difficult to arrive at a more complete understand- ing of any phylogenetic link between this group and the Xenarthra. In light of the present study, it is possible that restudy of the younger palaeano- dont material, focusing in particular on the mor- phology of the zygapophyses, might yield addi- tional derived characteristics linking xenarthrans and palaeanodonts. For now, however, these early Cenozoic taxa provide little help in elucidating the evolutionary history of the peculiar intervertebral articulations that characterize Xenarthra. Our best guide to understanding the structural evolution of xenarthrous vertebrae would appear to be contin- ued study of the functional, ontogenetic, and phy- logenetic history of these articulations among the xenarthrans themselves. Acknowledgments My thanks go first and foremost to Julia Scott, who so skillfully executed all the artistic rendi- tions in this report. For providing access to spec- imens under their care, including the loan of the figured specimens, I am grateful to Larry Heaney, Bruce Patterson, and Bill Stanley of the Division of Mammals, Field Museum of Natural History, and John Flynn and Bill Simpson of the Depart- ment of Geology, Field Museum of Natural His- tory. I thank Gerry Deluliis, Laura Panko, and two anonymous reviewers for their comments on an earlier draft of this manuscript. The research on which this report is based was supported by a UC Foundation Faculty Research Grant from the University of Tennessee at Chattanooga. Literature Cited Allen, G. M. 1913. A new Mylodon. Memoirs of the Museum of Comparative Zoology, 40: 318-346. Anthony, H. E. 1918. The indigenous land mammals of Porto Rico, living and extinct. Memoirs of the Amer- ican Museum of Natural History, 2: 331-435. Cullinane, D. M., and D. Aleper. 1998. The functional and biomechanical alterations of the spine of Scuti- sorex somereni, the hero shrew: Spinal musculature. Journal of Zoology, London, 244: 453-458. Cullinane, D. M., D. Aleper, and J. E. A. Bertram. 1998. The functional and biomechanical modifications of the spine of Scutisorex somereni, the hero shrew: Skeletal scaling relationships. Journal of Zoology, London, 244: 447-452. Cuvier, G. 1836a. Lecons d'anatomie comparee, vol. 1, 3rd ed. H. Dumont, Bruxelles, 655 pp. . 1836b. Recherches sur les ossemens fossiles, vol. 8, Part VI. Des ossemens d'edentes. 4th ed. E. D'Ocagne, Paris, 461 pp. DeBlase, A. E, and R. E. Martin. 1981. A Manual of Mammalogy with Keys to Families of the World. Wil- liam C. Brown Publishers, Dubuque, la., 436 pp. Ding, S.-Y. 1987. A Paleocene edentate from Nanxiong Basin, Guangdong. Palaeontologia Sinica, 173: 1- 118. Edmund, G. 1985. The fossil giant armadillos of North America (Pampathcriinae, Xenarthra = Edentata), pp. 83-93. In Montgomery, G. G., ed., The Ecology and Evolution of Armadillos, Sloths, and Vermilinguas. Smithsonian Institution Press, Washington, D.C. 451 pp. Emry, R. J. 1970. A North American Oligocene pan- golin and other additions to the Pholidota. Bulletin of the American Museum of Natural History, 142: 459- 510. Engelmann, G. 1978. The logic of phylogenetic analysis and the phylogeny of the Xenarthra. Ph.D. thesis, Co- lumbia University, New York, 329 pp. . 1985. The phylogeny of the Xenarthra, pp. 51- 64. In Montgomery, G. G., ed., The Ecology and Evo- lution of Armadillos, Sloths, and Vermilinguas. Smithsonian Institution Press, Washington, D.C. 451 pp. GAUDLN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 33 Flower, W. H. 1885. An Introduction to the Osteology of the Mammalia. Macmillan and Co., London. 382 pp. Frechkop. S. 1949. Replication biologiquc fournie par les Tatous. d'un des caracteres distinctifs des X6nar- thres et d'un caractere adaptif analogue chez les Pan- golins. Institut Royal des Sciences Naturelles de Bel- gique, 25: 1-12. Gal din. T. J. 1993. Phylogeny of the Tardigrada (Mam- malia. Xenarthra) and the evolution of locomotor function in the Xenarthra. Ph.D. thesis. University of Chicago, Chicago, 111., 226 pp. . 1995. The ear region of edentates and the phy- logeny of the Tardigrada (Mammalia, Xenarthra). Journal of Vertebrate Paleontology. 15: 672-705. Galdin, T J., and A. A. Bihwener. 1992. The functional morphology of xenarthrous vertebrae in the armadillo Dasypux novemcinctus (Mammalia, Xenarthra). Jour- nal of Morphology, 214: 63-81. Gervais. M. P. 1855. Part VII. Ordre des Edentes., pp. 44-53. In Animaux Nouveaux ou Rares Recueillis Pendant 1* Expedition dans las Parties Centrales de LAmerique du Sud, de Rio de Janeiro a Lima, et de Lima au Para. Vol. 1, Anatomic Chez P. Bertrand. Paris. Gill. T 1872. Arrangement of the families of mammals with analytical tables. Smithsonian Miscellaneous Collections. 230: 1-98. . 1886. Order I. Bruta or Edentata, pp. 46-67. In Kingsley, J. S., ed.. The Standard Natural History. Vol. V Mammals. S. E. Cassino & Co., Boston, Mass. Giuj/ite. D. D.. and C. E. Ray. 1981. Glyptodonts of North America. Smithsonian Contributions to Paleo- biology. 40: 1-255. Glass. B. P. 1985. History of classification and nomen- clature in Xenarthra (Edentata), pp. 1-3. In Montgom- ery. G. G., ed.. The Ecology and Evolution of Ar- madillos, Sloths, and Vermilinguas. Smithsonian In- stitution Press, Washington, D.C.. 451 pp. Grasse, P.-P 1955. Ordre des Edent6s. pp. 1182-1266. In Grasse\ P.P., ed.. Traits de Zoologie. Vol. 17. Mam- miferes. Masson et Cie, Paris. Gunnell, G. E, and P. D. Ginoerich. 1993. Skeleton of Brachianodon weslorum, a new Middle Eocene me- tacheiromyid (Mammalia. Palaeanodonta) from the Early Bridgerian (Bridger A) of the southern Green Rjver Basin. Contributions from the Museum of Pa- leontology, the University of Michigan. 28: 365-392. Hi.ssio, K. 1982. Ein Edentate aus dem Oligozan Sud deutschlands. Mitteilungen Bayerische Staatssamm- lung fur Palaontologie und Historische Geologic 22: 91-%. Hoiiyn her. R. 1958. Xenarthra, pp. 535-636. In Piv- eteau. J., ed.. Traite" de Paleontologic Vol. 2. no. 6. Mammiferes Evolution. Masson et Cie. Paris. 1961. Description d'un squelette de Planops (Gravigrade du Miocene de Patagonie). Mammalia. 25: 57-96. . 1982. Les cdent£s x6narthres, un groupe sin- gulier de la faune Neotropical, pp. 385-443. In Gal- litelli, E. M.. ed.. Paleontology. Essential of Historical Geology. STEM Mocchi Modena Press, Modena. It- aly. 524 pp. Jenkjns, F. A., Jr. 1970. Anatomy and function of ex- panded ribs in certain edentates and primates. Journal of Mammalogy, 51: 288-301. . 1974. Tree shrew locomotion and the origins of primate arborealism. pp. 85-1 15. In Jenkins. F. A., Jr., ed.. Primate Locomotion. Academic Press, New York. 390 pp. Jenkins, F. A., Jr., and E R. Parrington. 1976. The postcranial skeleton of the Triassic mammals Eozos- trodon, Megazostrodon, and Erythrotherium. Philo- sophical Transactions of the Royal Society of London, B273: 387-431. Jenkins. F A., Jr., and C. R. Schaff. 1988. The Early Cretaceous mammal Gobiconodon from the Cloverly Formation in Montana. Journal of Vertebrate Paleon- tology, 8: 1-24. Kardong, K. V 1995. Vertebrates. Comparative Anat- omy, Function, Evolution. William C. Brown Publish- ers, Dubuque, la., 777 pp. Kjelan-Jaworowska, Z., and P. P. Gambaryan. 1994. Postcranial anatomy and habits of Asian multituber- culate mammals. Fossils & Strata, 36: 1-92. Kingdon, R. 1984. East African Mammals. An Atlas of Evolution in Africa. Vol. IIA. Insectivores and Bats. University of Chicago Press, Chicago, 111., 341 pp. Krebs, B. 1991. Das Skelett von Henkelotherium gui- morolae gen. et sp. nov. (Eupantotheria, Mammalia) aus dem Oberen Jura von Portugal. Berliner geowis- senschaftliche Abhandlungen. A133: 1-110. Lisshriissi (R. J., and R. Sahan. 1967. Squelette axial, pp. 584-708. In Grasse\ PP., ed., Trait6 de Zoologie. Vol. 16, Mammiferes. Masson et Cie.. Paris, 1 167 pp. MacPhee, R. D. E. 1994. Morphology, adaptations and relationships of Plesiorycteropus, and a diagnosis of a new order of eutherian mammals. Bulletin of the American Museum of Natural History, 220: 1-214. Marshall, L. G., C. de Muizon, and D. Sigogneau- Russell. 1995. Pucadelphys andinus (Marsupialia, Mammalia) from the early Paleocene of Bolivia. M6- moires du Museum National d'Histoire Naturelle, Par- is, 165: 1-164. Matthew, W. D. 1918. A revision of the lower Eocene Wasatch and Wind River faunas. Part V — Insectivora (continued), Glires. Edentata. Bulletin of the Ameri- can Museum of Natural History, 38: 565-657. MATTHEW, W. D., and C. de Paula Cot to. 1959. The Cuban edentates. Bulletin of the American Museum of Natural History. 117: 1-56. McDonald, H. G. 1987. A systematic review of the Plio-Pleistocene scelidotheriine ground sloths (Mam- malia: Xenarthra: Mylodontidae). Ph.D. thesis. Uni- versity of Toronto, Toronto, Ontario, Canada, 499 pp. M< Ki nna. M. C. 1975. Toward a phylogenetic classi- fication of the Mammalia, pp. 21-66. In Luckett, W. P., and E S. Szalay, eds., Phylogeny of Primates. Ple- num Press. New York, 483 pp. Novackk, M. J. 1986. The skull of leptictid insectivor- ans and the higher-level classification of eutherian mammals. Bulletin of the American Museum of Nat- ural History. 183: 1-112. Novacek, M. J., and A. Wyss. 1986. Higher-level re- lationships of the recent eutherian orders: morpholog- ical evidence. Cladistics. 2: 257-287. 34 FIELDIANA: GEOLOGY Novacek, M. J., A. R. Wyss, and M. C. McKenna. 1988. The major groups of eutherian mammals, pp. 31-71. In Benton, M. J., ed.. The Phylogeny and Clas- sification of Tetrapods. Vol. 2, Mammals. Clarendon Press, Oxford, 326 pp. Owen, R. 1 842. Description of the skeleton of an extinct ground sloth, Mylodon robustus Owen. London. 176 pp. . 1851a. On the Megatherium. Part I. Preliminary observations on the exogenous processes of vertebrae. Philosophical Transactions of the Royal Society of London, 141: 719-764. 1851b. On the Megatherium. Part II. Vertebrae of the trunk. Philosophical Transactions of the Royal Society of London, 145: 359-388. Patterson, B., W. Segall, and W. D. Turnbull. 1989. The ear region in xenarthrans (= Edentata: Mamma- lia). Part I. Cingulates. Fieldiana: Geology, n.s., 18: 1-46. Paula Couto, C. de. 1978. Mamfferos Fosseis do Pleis- toceno do Espirito Santo. Anais, Academia Brasileira de Ciencias, 50: 365-379. Pick, T P., and R. Howden. 1977. Gray's Anatomy, 15th ed. Bounty Books, New York, 1,257 pp. Rose, K. D. 1978. A new Paleocene epoicotheriid (Mammalia), with comments on the Palaeanodonta. Journal of Paleontology, 52: 658-675. . 1979. A new Paleocene palaeanodont and the origin of the Metacheiromyidae (Mammalia). Breviora of the Museum of Comparative Zoology, 455: 1-14. Rose, K. D., and R. J. Emry. 1993. Relationships of Xenarthra, Pholidota, and fossil "edentates": The morphological evidence, pp. 81-102. In Szalay, E S., M. J. Novacek, and M. C. McKenna, eds.. Mammal Phylogeny. Vol. 2, Placentitis. Springer- Verlag, New York, 321 pp. Rose, K. D., R. J. Emry, and P. D. Gingerich. 1992. Skeleton of Alocodontulum atopum, an Early Eocene epoicotheriid (Mammalia, Palaeanodonta) from the Bighorn Basin, Wyoming. Contributions from the Mu- seum of Paleontology, the University of Michigan, 28: 221-245. Rose, K. D., L. Krishtalka, and R. K. Stucky. 1991. Revision of the Wind River Faunas, Early Eocene of Central Wyoming. Part II. Palaeanodonta (Mamma- lia). Annals of the Carnegie Museum, 60: 63-82. Scott. W B. 1903-1904. Mammalia of the Santa Cruz Beds. Part 1 : Edentata. Reports of the Princeton Ex- peditions to Patagonia, 5: 1-364. Simpson. G. G. 1931. Metacheiromys and the Edentata. Bulletin of the American Museum of Natural History. 59: 295-381. . 1948. The beginning of the Age of Mammals in South America. Bulletin of the American Museum of Natural History. 91: 1-350. Sluper. E. J. 1946. Comparative biologic-anatomical in- vestigations on the vertebral column and spinal mus- culature of mammals. Verhandelingen der Koninklijkc Nederlandsche Akadamie van Wetenschappen, AFD. Natuurkundc (Twecde Sectie), 17: 1-128. Stock, C. 1925. Cenozoic gravigrade edentates of west- ern North America. Carnegie Institute of Washington Publications, 331: 1-206. Storch, G. 1981. Eurotamandua joresi, ein Myrmeco- phagidae aus dem Eozan der "Grube Messel" bci Darmstadt (Mammalia, Xenarthra). Senckenbergiana lethaea, 61: 247-289. Szalay, F. S.. and F. Schrenk. 1994. Middle Eocene Eurotamandua and the early differentiation of the Edentata. Journal of Vertebrate Paleontology, 14: 48A. Vaughn, T A. 1986. Mammalogy. Saunders College Publishing, New York, 576 pp. Wake, M. H. 1979. Hyman's Comparative Vertebrate Anatomy. University of Chicago Press, Chicago. 111.. 788 pp. Walker. W E. and D. G. Homberger. 1992. Vertebrate Dissection, 8th ed. Saunders College Publishing, New York. 459 pp. Weber. M. 1904. Die Saugctierc. Gustav Fischer, Jena, 898 pp. White, J. L. 1993a. Indicators of locomotor habits in xenarthrans: Evidence for locomotor heterogeneity among fossil sloths. Journal of Vertebrate Paleontol- ogy, 13: 230-242. . 1993b. Functional and phylogenetic implica- tions of the postcranial skeleton of fossil sloths for the evolution of arborcality in tree sloths. Ph.D. thesis. State University of New York at Stony Brook, Stony Brook, NY, 835 pp. GAUDLN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 35 Appendix Table 1. Summary of Morphological Data Taxon No. Diaphrag- No. of of matic TV LV vertebra MP AP Spinal nerves Zaedyus pichiy 11 3 T7 fmnh 23809, 1048171 (104817) orT8 (23809) Chaetophractus villosus T7 FMNH 60467 ' 10 5 fmnh 122623, 134611 11 3 Euphractus sexcinctus 11 4 T8 FMNH 152051 Tolypeutes matacus T7 FMNH 121540, 11 4 (124569) 124568 (juv.), 124569 orT8 (juv.),1 124570, 153773 (124570) fmnh 122233 12 3 Tamandua mexicana T13 fmnh 69597 ■ 17 2 (69597) fmnh 22398 18 2 orT14 (22398) Cyclopes didactylus 16 2 T13 fmnh 69969, ■ 69971 (69971) orT14 (69969) Brady pus variegatus 15 3 T15 fmnh 68919, 695891 Choloepus hojfmani 23 3 T21 fmnh 127421 (juv.), 127422, 1 47993 ' Hapalops sp. 21-22 3 T19 FMNH PI 3 133, P13145, P13146, P15318 Ventral branch: intervertebral foramina be- tween pedicels of successive vertebrae, Tl and T2, Tl 1-L3; within pedicels, T3- T10. Dorsal branch: in lamina between ap and dp, T3-T10; medial to ap, T11-L3. Ventral branch: within pedicel, T1-T10; in- tervertebral foramina, T11-L4. Dorsal branch: perforates lamina, T2-T4; perforates ap, T5-T10 (two holes in T7- T9); medial to ap, Tl 1-L4. Ventral branch: within pedicel, T1-T10; in- tervertebral foramina, T11-L3. Dorsal branch: perforates lamina, T3-T7; perforates ap, T8-T10; medial to ap, T11-L3. Ventral branch: within pedicel T1-T10; in- tervertebral foramina, T11-L4. Dorsal branch: ventral to ap; T1-T6; perfo- rates ap, T7; medial to ap, T8-L4. T1-L2 T13-L2 Spinal nerves emerge through typical inter- vertebral foramina T1-L2 T13-L2 Spinal nerves emerge through typical inter- vertebral foramina T3-L3 T15-L3 Spinal nerves emerge through typical inter- vertebral foramina T21-L3 T23-L3 Spinal nerves emerge through typical inter- vertebral foramina T7-L3 T6-L3 T7-L5 T3-L4 T8-L4 T4-L3 T7-L4 T8-L4 T1-L3 T19-L3 Spinal nerves emerge through typical inter- vertebral foramina ALZ or alz = anterior lateral zygapophyseal facet; AMZ or amz = anterior medial zygapophyseal facet; AP or ap = anapophysis; AX = anterior xenarthrous facet; dp = diapophysis; (1) = left side only; LI, L2, L3 = first lumbar vertebra, second lumbar vertebra, third lumbar vertebra, etc.; LV = lumbar vertebrae; MP or mp = metapophysis; PLZ or plz = posterior lateral zygapophyseal facet; PMZ or pmz = posterior medial zygapophyseal facet; PX = posterior xenarthrous facet; (r) = right side only; SI, S2 = first sacral vertebra, second sacral vertebra, etc.; Tl, T2 = first thoracic vertebra, second thoracic vertebra, etc.; tp = transverse process; TV = thoracic vertebrae. 1 Information in table derived from this specimen, except for thoracic and lumbar vertebral counts. 36 FIELDIANA: GEOLOGY Appendix Table 2. Summary of Morphological Data Taxon AMZ and ALZ PMZ and PLZ AX PX Notes Zaedyus pichiy FMNH 104817 T3-S1 T2-L3 amz abuts alz pmz abuts plz from T3-T8 from T2-T7 Chaetophractus villo- sus FMNH 60467 T4(r)/T5(l)-L5, amz abuts alz from T4/5-T8 T3(r)/T4(l)-L4 pmz abuts plz from T3-T8 Euphractus sexcinctus T5-L4 T4-L3 fmnh 152051 amz abuts alz pmz abuts plz from T5-T9 from T4-T8 Tolypeutes matacus fmnh 124569 (juv.) T9-S1 T8-L4 Tamandua mexicana fmnh 69597 T14-S1 T13-L2 Cyclopes didactylus fmnh 69969 Bradypus variegatus fmnh 69589 T15-S1 L2-S1 T14-L2 L1-L3 Ventral mp: T7- Sl, confluent with alz from T9-S1; dorsal tp: Ll-Sl Ventral dp: T5- T6; ventral mp: T7-L5, confluent with alz from L3- L5; dorsal tp: L1-L5 Ventral mp: T7- L4, confluent with alz from T9-L4; dorsal tp: L1-L4 Ventral mp: T7(r)/T8(l)- Sl, confluent with alz in LV Ventral mp: T14-S1, con- fluent with alz in all; dorsal sacral rib: SI Ventral mp: T13-S1, con- fluent with alz in all Absent Dorsal ap: T6- L3, confluent with plz from T8-L3; lateral ap (for rib): T8orT9- T10; ventral ap: Tl 1-L3 Dorsal ap: T4- L4, confluent with plz from L2-L4; lateral ap (for rib): T8-T10; ven- tral ap: Tl 1- L4 Dorsal ap: T6- L3, confluent with plz from T8-L3; lateral ap (for rib): T8-T10; ven- tral ap: Tll- L3 Dorsal ap: T6(r)/ T7(l)-L4, confluent with plz in LV; ventral ap (for thoracic, lum- bar, and sacral ribs): T11-L4 Dorsal ap: T13- L2, confluent with plz in all; ventral ap: L2 Opisthocoelus cen- tra, T1-T9; rib head articulates between anterior centrum and pos- terior pedicel L5 fused to S 1 ; op- isthocoelus cen- tra; rib head as Zaedyus L4 fused to SI; me- dial zygapophy- ses enrolled as in pangolins; opisth- ocoelus centra; rib head as Zae- dyus Rib head as Zaed- yus Dorsal lamina: T12-T13; dor- sal ap: T14- L2 Absent Rib head articulates exclusively with anterior centrum; neural spines ro- bust, uniform height; anterior zygapophyses separated by mid- line notch; tho- racic zygapophy- ses wide, extending lateral to centrum Rib head, neural spines, zygapoph- yses as Taman- dua Rib head articulates between posterior centrum and ante- rior pedicel; all vertebral process- es reduced; neu- ral spines uni- form height; anterior zyg- apophyses sepa- rated by midline notch GAUDEN: THE MORPHOLOGY OF XENARTHROUS VERTEBRAE 37 Appendix Table 2. Continued Taxon AMZ and ALZ PMZ and PLZ AX PX Notes Choloepus hoffmani fmnh 147993 SI (r) L3(r) Absent Absent Hapalops sp. T20-S1 fmnh P13133, P13145, P13146, P15318 T19-L3 PI 3 146 only- ventral mp: LI (right side only), L3 PI 3 146 only- dorsal ap: T22 (right side only), L2 Vertebral processes as in Bradypus; anterior zyg- apophyses sepa- rated by midline notch as in Bra- dypus Rib head, neural spines as Taman- dua\ zygapophy- ses of anterior thoracics very wide, extending lateral to cen- trum; anterior zygapophyses separated by nar- row midline notch 1 From T7-T9, the lateral zygapophyseal facet is divided in two in this specimen. The "lateral" lateral zygapophy- seal facet articulates with the medial surface of the anapophysis of the preceding vertebra. 38 FIELDIANA: GEOLOGY A Selected Listing of Other Fieldiana: Geology Titles Available A Preliminary Survey of Fossil Leaves and Weil-Preserved Reproductive Structures from the Sentinel Butte Formation (Paleocene) near Almont, North Dakota. By Peter R. Crane, Steven R. Manchester, and David L. Dilcher. Fieldiana: Geology, n.s., no. 20, 1990. 63 pages, 36 illus. Publication 1418, $13.00 Osteology of Simosaurus gaillardoti and the Relationships of Stem-Group Sauropterygia. By Olivier Rieppel. Fieldiana: Geology, n.s., no. 28, 1994. 85 pages, 71 illus. Publication 1462, $18.00 Giant Short-Faced Bear (Arctodus simus yukonensis) Remains from Fulton County, Northern Indiana. By Ronald L. Richards and William D. Turnbull. Fieldiana: Geology, n.s., no. 30, 1995. 34 pages, 20 illus. Publication 1465, $10.00 Pachypleurosaurs (Reptilia: Sauropterygia) from the Lower Muschelkalk, and a Review of the Pachypleurosauroidea. By Olivier Rieppel and Lin Kebang. Fieldiana: Geology, n.s., no. 32, 1995. 44 pages, 28 illus. Publication 1473, $12.00 The Mammalian Faunas of the Washakie Formation, Eocene Age, of Southern Wyoming. Part III. The Perissodactyls. By Steven M. McCarroll, John J. Flynn, and William D. Turnbull. Fieldiana: Geology, n.s., no. 33, 1996. 38 pages, 13 illus., 5 tables. Publication 1474, $11.00 A Revision of the Genus Nothosaurus (Reptilia: Sauropterygia) from the Germanic Triassic, with Comments on the Status of Conchiosaurus clavatus. By Olivier Rieppel and Rupert Wild. Fieldiana: Geology, n.s., no. 34, 1996. 82 pages, 66 illus. Publication 1479, $17.00 The Status of the Sauropterygian Reptile Genera Ceresiosaurus, Lariosaurus, and Silvestrosaurus from the Middle Triassic of Europe. By Olivier Rieppel. Fieldiana: Geology, n.s., no. 38, 1998. 46 pages, 21 illus. Publication 1490, $15.00 Order by publication number and/or ask for a free copy of our price list. All orders must be prepaid. Illinois residents add current destination tax. All foreign orders are payable in U.S. dollar-checks drawn on any U.S. bank or the U.S. subsidiary of any foreign bank. Prices and terms subject to change without notice. Address all requests to: HELD MUSEUM OF NATURAL HISTORY Library — Publications Division Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2498, U.S.A. UNIVERSITY OF ILLINOIS-URBANA 3 0112 053323348 Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2496 Telephone: (312) 922-9410