550 5 GEOLOGY LIBRARY r iELDIANA Geology NEW SERIES, NO. 45 The Cranial Anatomy of Placochelys placodonta Jaekel, 1902, and a Review of the Cyamodontoidea (Reptilia, Placodonta) Olivier Rieppel £J October 31, 2001 {== Publication 1514 CO Z PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY «EWLUB at iirbANA-CHAMPAIGN HI un QpQLOQYlnfermation for Contributors to Fieldiana Generall|Jjk/zvL is primarily a journal for Field Museum staff members and research associates, although manuscripts fromnonaffiliated 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 /o 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 oi 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, 8%- 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 5%-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). n For papers over 100 manuscript pages, authors are requested to submit a "Table of Contents, a List ot 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 "X0.8 are not acceptable. Captions should be typed double-spaced and consecutively. See recent issues of Fieldiana for details of All illustrations should be marked on the reverse with author's name, figure number(s), and "top." Figures as submitted should, whenever practicable, be 8V2 by 1 1 inches (22 X 28 cm) and may not exceed 1 IV* 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 ot ANSI/NISO Z39.48-1992 (Permanence ot Paper). FIELDIANA Geology NEW SERIES, NO. 45 The Cranial Anatomy of Placochelys placodonta Jaekel, 1902, and a Review of the Cyamodontoidea < Kept i I in, Placodonta) Olivier Rieppel Department of Geology Field Museum of Natural History 1400 South Lake Shore Drive Chicago, Illinois 60605-2496 U.S.A. Accepted March 7, 2000 Published October 31, 2001 Publication 1514 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY © 2001 Field Museum of Natural History ISSN 0096-265 1 PRINTED IN THE UNITED STATES OF AMERICA no Table of Contents «-VU/UTUfcfft*ft, u-r Abstract 1 Introduction 1 Material Included in This Study 2 List of Abbreviations Used in the Figures 3 The Cranial Anatomy of Placochelys pla- codonta 3 Measurements of the Holotype 5 Morphological Description of the Skull 6 Morphological Description of the Lower Jaw 21 Dermal Ornamentation of the Skull 23 Comparison of the Cranial Anatomy of Placochelys placodonta with That of Other Cyamodontoid Placodonts 23 The Cranial Anatomy of Cyamodus ros- tratus (Minister, 1839) 24 The Cranial Anatomy of Cyamodus muensteri (Agassiz, 1839) 33 The Cranial Anatomy of Cyamodus "/a- ticeps" (Owen, 1858) 35 The Cranial Anatomy of Cyamodus kuhnschnyderi Nosotti and Pinna, 1993a 37 The Cranial Anatomy of Cyamodus hilde- gardis Peyer, 1931a 39 The Cranial Anatomy of Henodus chel- yops v. Huene, 1936 39 The Cranial Anatomy of Macroplacus raeticus Schubert-Klempnauer, 1975 .. 49 The Cranial Anatomy of Protenodonto- saurus italicus Pinna, 1990b 56 The Cranial Anatomy of Psephoderma alpinum H. v. Meyer, 1858 59 Autapomorphies in the Skull of the Cy- amodontoidea 63 Evolution of the Rostrum and of the Den- tition Within the Cyamodontoidea 65 Cladistic Analysis of Cyamodontoid In- terrelationships 67 Systematic Paleontology 71 Cyamodontoidea Nopcsa, 1923 71 Cyamodontida, new taxon 72 Henodus Huene, 1936 72 Henodus chelyops Huene, 1936 72 Cyamodontidae Nopcsa, 1923 73 Cyamodus Meyer, 1863 73 Cyamodus hildegardis Peyer, 1931a 73 Cyamodus kuhnschnyderi Nosotti and Pinna, 1993 74 Cyamodus muensteri (Agassiz, 1839) 75 Cyamodus rostratus (Munster, 1839) 76 Placochelyida, new taxon 78 Macroplacus Schubert-Klempnauer, 1975 .. 78 Macroplacus raeticus Schubert-Klemp- nauer, 1975 78 Unnamed Taxon 78 Protenodontosaurus Pinna, 1990b 78 Protenodontosaurus italicus Pinna, 1990b .. 79 Placochelyidae Romer, 1956 79 Placochelys Jaekel, 1902 79 Placochelys placodonta Jaekel, 1 902 79 Psephoderma Meyer, 1858 85 Psephoderma alpinum Meyer, 1858 86 Placochelys and Potential Turtle Rela- tionships of the Cyamodontoidea 87 Paleobiogeography and Paleoecology of Cyamodontoid Placodonts 92 Acknowledgments 95 Literature Cited 95 Appendix I 101 List of Illustrations 1 . Skull of Placochelys placodonta Jaekel (holotype) 4 2. Skull of Placochelys placodonta Jaekel (paratype) 5 3. Skull of Placochelys placodonta Jaekel (holotype) 8 4. Skull of Placochelys placodonta Jaekel (holotype) 10 5. Skull of Placochelys placodonta Jaekel (paratype) 12 6. Braincase of Placochelys placodonta Jaekel (holotype) 14 7. Occiput of Placochelys placodonta Jae- kel (paratype) 18 8. Occiput of Placochelys placodonta Jae- kel (holotype) 19 9. Lower jaw of Placochelys placodonta Jaekel (holotype) 22 10. Lower jaw of Cyamodus hildegardis Peyer 23 1 1 . Skull of Cyamodus rostratus Munster (holotype) 24 12. Skull of Cyamodus rostratus Munster (holotype) 26 13. Skull of Cyamodus rostratus Munster (holotype and referred specimen) 28 14. Pterygoid of Cyamodus rostratus Mun- ster (referred specimen) 31 in 15. Lower jaw of Cyamodus rostratus Miinster (referred specimen) 32 16. Skull of Cyamodus muensteri (holo- type) 33 17. Skull of Cyamodus muensteri (holotype of C. "laticeps" Owen) 34 18. Skull of Henodus chelyops v. Huene (syntype) 40 19. Skull of Henodus chelyops v. Huene (syntype and referred specimen) 41 20. Skull of Henodus chelyops v. Huene (syntypes) 42 2 1 . Skull of Henodus chelyops v. Huene (syntype and referred specimen) 44 22. Premaxillary denticles in Henodus chelyops v. Huene 44 23. Right side of dermal palate in Henodus chelyops v. Huene 46 24. Suspension of left quadrate in Henodus chelyops v. Huene 47 25. Lower jaw of Henodus chelyops v. Huene 48 26. Skull of Macroplacus raeticus Schu- bert-Klempnauer (holotype) 50 27. Skull of Protenodontosaurus italicus Pinna (holotype) 54 28. Skull of Psephoderma alpinum H. v. Meyer 59 29. Skull of Psephoderma alpinum H. v. Meyer 60 30. Left lateral braincase wall of Placodus gigas Agassiz 64 31. Most parsimonious unrooted network for Placodontoidea 67 32. Most parsimonious reconstruction of placodont interrelationships 68 33. Most parsimonious reconstruction of placodont interrelationships, with Cy- amodus hildegardis included 69 34. Strict consensus tree for placodont in- terrelationships, with the Negev speci- men included 71 35. Skull of Placochelys alpis sordidae Broili (holotype) 81 36. Skull of Placochelys alpis sordidae Broili (holotype) 81 37. Holotype of Placochelys stoppanii Oss- wald 83 38. Dermal ossifications referred to IPla- cochelys 84 39. Area cladogram for the cyamodontoid placodonts 94 List of Tables 1 . Measurements of maxillary and pala- tine tooth plates of Placochelys placo- donta (holotype) 5 2. Measurements of the dentary tooth plates of Placochelys placodonta (holo- type) 21 3. Measurements of the palatine tooth plates of Cyamodus rostratus (holo- type) 29 4. Measurements of the dentary tooth plates of Cyamodus rostratus (referred specimen) 32 5. Dentitional characters of species of the genus Cyamodus 33 6. Measurements of the tooth plates of Cyamodus muensteri 37 7. Proportions of the posterior palatine tooth plate throughout the Cyamodon- toidea 39 8. Skull proportions of cyamodontoid pla- codonts 51 9. Measurements of the maxillary and palatine tooth plates of Macroplacus raeticus (holotype) 53 10. Measurements of the maxillary and palatine tooth plates of Protenodonto- saurus italicus (holotype) 58 1 1 . Data matrix for the analysis of placo- dont relationships 66 12. Measurements of the maxillary and palatine tooth plates of Placochelys mal- anchinii Boni, 1947 84 13. Measurements of the maxillary and palatine tooth plates of Psephoderma alpinum 85 14. Data matrix for the implementation of the Brooks parsimony analysis in the reconstruction of cyamodontoid histori- cal biogeography 85 IV The Cranial Anatomy of Placochelys placodonta Jaekel, 1902, and a Review of the Cyamodontoidea (Reptilia, Placodonta) Olivier Rieppel Abstract The skull of Placochelys placodonta Jaekel is described in detail and compared with all other cyamodontoid skulls kept in public repositories. Cladistic analysis based on a character set derived from cyamodontoid skull anatomy results in a reconstruction of placodont interre- lationships as follows: (Paraplacodus (Placodus ((Henodus, Cyamodus) (Macroplacus (Pro- tenodontosaurus (Placochelys, Psephoderma)))))). The monophyly of the Cyamodontoidea is very robust, supported in particular by a suite of derived braincase characters. On the basis of present evidence, Henodus is the sister taxon of Cyamodus, and the monophyletic genus Cy- amodus includes C. hildegardis. The monophyly of placochelyids, including Placochelys and Psephoderma, is strongly supported also. A detailed comparison of skull anatomy provides no basis for a hypothesis of close phylo- genetic relationships of turtles and cyamodontoid placodonts. Any similarities between the two clades, particularly with respect to the development of extensive dermal armor, must be con- vergent. The historical paleobiogeography of cyamodontoid placodonts can largely be understood as a sequence of vicariance events that involved an early bifurcation establishing separate clades in the Germanic Basin and on the Eurasian carbonate platform. Subsequent vicariance estab- lished separate clades in the northern Alpine Triassic and in the southern Alps on the Hungarian platform, with further subdivision of the clades within the latter. Introduction Cyamodontoid placodonts are a clade of marine reptiles that occurs in shallow epicontinental and nearshore deposits of Middle and Upper Triassic age throughout the western Tethyan faunal prov- ince. The earliest cyamodontoid placodont to ap- pear in the fossil record is Cyamodus tarnowitz- ensis Giirich, 1884, a skull (now lost) from the Karchowice Beds of Tarnowskie Gory, Poland (formerly Tarnowitz in Upper Silesia), which be- long to the uppermost lower Muschelkalk (lower Illyrian, lower Anisian). Another early occurrence is a skull fragment (Brotzen, 1957; Rieppel, Ma- zin, & Tchernov, 1999) of a cyamodontoid from the lower Muschelkalk of Makhtesh Ramon, Ne- gev, Israel (Middle Member of the Gevanim For- mation, upper Bithynian, upper lower Anisian: Druckman, 1974). The latest occurrences of cy- amodontoids are from the Rhaetian of the north- ern (Psephoderma: Meyer, 1858a, b; Broili, 1921; Macroplacus: Schubert-Klempnauer, 1975) and southern (Psephoderma: Osswald, 1930; Boni, 1946 [1947], 1947 [1948]; Pinna, 1975, 1976a, b, 1978, 1979; Pinna & Nosotti, 1989) Alpine Tri- assic and from the Rhaetian of England (Meyer, 1858a, b; Storrs, 1994; see also Pinna, 1990a). Other localities that have yielded cyamodontoid placodonts are in the Anisian of Transylvania (Jurcsak, 1982; Huza et al., 1987); the upper Mu- schelkalk and Keuper of southern Germany (Cy- amodus rostratus, Cyamodus muensteri, and Cy- amodus "laticeps" from the upper Anisian [Ag- assiz, 1833-45; Minister, 1839; Owen, 1858; FIELDIANA: GEOLOGY, N.S., NO. 45, OCTOBER 31, 2001, PP. 1-104 Meyer, 1863 J; Cyamodus kuhnschnyderi from the lower Ladinian [Nosotti & Pinna, 1993a]; Pse- phosaurus suevicus from the upper Ladinian [Fraas, 1896]; Henodus chelyops from the Carni- an [Huene, 1936]) and of the Lorraine, France (Corroy, 1928; Rieppel & Hagdorn, 1999); the Ladinian of the southern Alps {Cyamodus hilde- gardis, Peyer, 1931a); the middle Carnian of the Tre Venezie area of northeastern Italy (Pinna & Zucchi Stolfa, 1979; Dalla Vecchia, 1993; Proten- odontosaurus italicus, Pinna, 1990b); the Ladini- an of northeastern Spain (Rieppel and Hagdorn, 1998); and Middle Triassic (?Anisian, Ladinian) localities on the northern Gondwanan shelf (Haas, 1959, 1975; Gorce, 1960; Beltan et al., 1979; Vickers-Rich et al., 1999). Cyamodontoid placodonts were a widespread and taxonomically diverse group characterized by the development of extensive dermal armor, which enhanced their chances of representation in the fossil record. In its most derived condition, this dermal armor consisted of a solid carapace, linked to a ventral armor by a lateral dermal body wall (Haas, 1969). This resulted in a remarkably turtle-like appearance of cyamodontoid placo- donts, so much so that, based on his study of the cranial anatomy and dermal armor of Placochelys placodonta, Jaekel (1902a, b, 1907) proposed a derivation of turtles from cyamodontoids. The hy- pothesis of a relationship of turtles to placodonts was later rejected by Gregory (1946), who noted that convergent evolution is remarkable in these two groups, especially with regard to the dermal armor. The more recent finding that turtles may be the sister-group of Sauropterygia among crown-group Diapsida (Rieppel & deBraga, 1996; deBraga & Rieppel, 1997; Rieppel & Reisz, 1999) has brought cyamodontoid placodonts back into fo- cus. Although a broad-based analysis of turtle re- lationships over a wide range of taxa confirmed that the similarities shared by turtles and cyamo- dontoids are convergent (Rieppel & Reisz, 1999), a more in-depth analysis of the cranial anatomy of cyamodontoid placodonts and its comparison with that of turtles appears desirable in the at- tempt to discover further similarities or differen- ces between the two groups. Placochelys placo- donta was selected as primary focus for this pro- ject not only because Jaekel (1902a, b, 1907) based his hypothesis of a turtle-placodont rela- tionship on this taxon, but also because it repre- sents one of the best-preserved cyamodontoid skulls. The cranial anatomy of Placochelys pla- codonta will also be compared in detail with the cranial anatomy of all other cyamodontoids for which skull material is available in an effort to analyze the phylogenetic interrelationships within the Cyamodontoidea. This will provide the nec- essary framework for the identification of the bas- al cyamodontoid skull morphology and the anal- ysis of evolutionary changes of skull morphology within this group of fascinating reptiles. Material Included in This Study The following is a list of material included in the present study. Institutional abbreviations are: bmnh: British Museum (Natural History), now The Natural History Museum, London; bsp, Bay- erische Staatssammlung fiir Palaontologie und historische Geologie, Munich; fafi, Magayar Al- lami Foldtani Intezet (Geological Institute of Hun- gary, Budapest); gpit, Geologisch-Palaontolo- gisches Institut, Universitat Tubingen; HUJ-Pal., Paleontological Collections, Department of Evo- lution, Systematics and Ecology, Hebrew Univer- sity, Jerusalem; mb.r., Museum fiir Naturkunde der Humboldt Universitat, Berlin, fossil reptile collection; mbsn, Museo Brembano di Scienze Naturali, San Pellegrino; mfsn, Museo Friulano di Storia Naturale, Udine; msnb, Museo Civico di Scienze Naturali "E. Caffi," Bergamo; msnm, Museo Civico di Storia Naturale di Milano; pimuz, Palaontologisches Institut und Museum der Uni- versitat Zurich; smf, Senckenberg Museum, Frankfurt a.M.; smns, Staatliches Museum fiir Na- turkunde, Stuttgart; st, Museo della Vicaria di S. Lorenzio, Zogno (Bergamo, Italy); umo, Urwelt- Museum Oberfranken, Bayreuth. Cyamodus hildegardis Peyer, 1931a: pimuz T4763 (holotype), T4768 (original of Peyer, 1935, PI. 46, Figs, la-c, and Pinna, 1992, Fig. 6), T4771 (original of Pinna, 1992, Fig. 7), T2796 (original of Kuhn-Schnyder, 1959, PI. I, and Pinna, 1992, Fig. 8). Cyamodus "laticeps" (Owen, 1858): bmnh R 1644 (holotype). Cyamodus kuhnschnyderi Nosotti and Pinna, 1993a: smns 15855 (holotype), smns 16270 (par- atype); mhi 1294 (incomplete skull). Cyamodus muensteri (Agassiz, 1839): bsp AS VII 1210 (holotype, original of Minister, 1830, skull no. II; Meyer, 1863, PI. 31, Figs. 1-2). Cyamodus rostratus (Munster, 1839): umo BT FIELDIANA: GEOLOGY 748 (holotype, original of Drevermann, 1928; Kuhn-Schnyder, 1965a). Cyamodus cf. rostratus: smns 17403 (incom- plete skull, original of Nosotti and Pinna, 1993b, Fig. 3); umo BT 2172 (isolated lower jaw, original of Drevermann, 1928, PI. 23, Fig. 2); smf R-4040 (isolated lower jaw, original of Drevermann, 1928, PI. 23, Figs. 3a-d, and of Rieppel, 1995a, Fig. 31). Henodus chelyops v. Huene, 1936: "Specimens I and II," syntypes of Huene (1936); "specimens IV and VI," collected in 1959 (Fischer, 1959). All specimens are kept at the gpit, uncatalogued. Macroplacus raeticus Schubert- Klempnauer, 1975: bsp 1967 I 324 (holotype). Placochelyanus malanchinii Boni, 1947 (1998): msnm uncatalogued (cast of holotype). Placochelyanus stoppanii Osswald, 1930: bsp AS I 1457 (holotype). Placochelys alpis sordidae Broili, 1921: bsp 1921.1.3 (holotype). Placochelys placodonta Jaekel, 1902b: fafi Ob/ 2323/Vt.3. (holotype); mb.r. 1765 (paratype). Placodus gigas Agassiz, 1839: umo Bt 13 (skull, original of Sues, 1987, and Rieppel, 1995a). Protenodontosaurus italicus Pinna, 1990b: mfsn 1819GP (holotype), mfsn 1923GP (referred second specimen). Psephoderma alpinum H. v. Meyer, 1858: bsp AS I 8 (holotype, carapace); msnm V471 (skull; referred specimen); msnb 4884a-b (juvenile skull, original of Pinna, 1979). Psephosaurus mosis Brotzen, 1957: HUJ-Pal. 220 (referred specimen). fcst facet (on quadrate) receiving the shaft of the stapes f-jug jugular foramen f.l lacrimal foramen f.p.dl posterior dental lamina foramen f.pin pineal foramen f.trig trigeminal foramen f.vest vestibular (oval) fenestra in internal naris ju jugal m maxilla nu- Meckel's canal ll nasal op opisthotic P parietal Pi palatine I > in premaxilla po postorbital pof postfrontal pq.rc palatoquadrate cartilage recess pra prearticular prf prefrontal pro prootic pt pterygoid pt.f posttemporal fenestra pto.f pteroccipital foramen q quadrate qj quadratojugal sang surangular so supraoccipital sp splenial sq squamosal sq.bt squamosal buttress (receiving the distal end of the paroccipital process) stc sella turcica V vomer List of Abbreviations Used in the Figures ang angular ar articular bo basioccipital c coronoid cci canal for internal carotid d dentary ds dorsum sellae eo exoccipital ep epipterygoid ep.o epiotic ossification f frontal f.cc foramen for cerebral carotid f.ch.t. chorda tympani foramen The Cranial Anatomy of Placochelys placodonta Remains of Placochelys placodonta were first collected in 1899 by Desiderius Laczk6 in the Also Keuper (upper Middle Triassic) of Jeruzsa- lemhegy (Jerusalem mountain) near Veszprem, a small town in west central Hungary, located on the south slopes of the Bakony Mountains over- looking Lake Balaton. Subsequent collecting ef- forts yielded two skulls, several carapace frag- ments, and scattered remains of the postcranial skeleton. The new genus and species was de- scribed by Jaekel in 1902, who subsequently pre- sented the material in a comprehensive mono- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA Fig. 1. Skull of Placochelys placodonta Jaekel (holotype, fafi Ob/2323/Vt.3): A, dorsal view; B, ventral view; C, left lateral view. Scale bar = 20 mm. graph (Jaekel, 1907). The holotype (skull, speci- men I of Jaekel, 1907) is three-dimensionally pre- served (Fig. 1), but was incompletely prepared at the time of its original description. A cast of the holotype in its original condition, that is, as de- scribed by Jaekel (1902a, b, 1907), is kept at the Geological Institute of Hungary. The second skull (Fig. 2) is dorsoventrally compressed and was fig- ured in ventral view only by Jaekel (1907, PI. III). The holotype was later sent to Frankfurt a.M. for further preparation by Christian Strunz, be- cause Fritz Drevermann planned to study the specimen in greater detail. Strunz separated the lower jaw from the cranium and fully exposed the braincase. Drevermann never got around to de- scribing the specimen, but it was briefly dealt with in a publication by Huene (1931). The figures published by Huene (1931) are erroneous in many details, as will be discussed below. In his mono- graph, Jaekel (1907) made only passing reference to the second skull, which later received no fur- ther attention other than a photograph included in Kuhn-Schnyder (1965b, Fig. 8) and Miiller (1968, Fig. 235). Parts of the postcranial remains of Placochelys placodonta were lost during World War II (West- phal, 1975). The material now missing comprises the following elements illustrated by Jaekel FIELDIANA: GEOLOGY Fig. 2. Skull of Placochelys placodonta Jaekel (paratype, mb.r. 1765): A, dorsal view; B, ventral view. Scale bar = 20 mm. (1907): PI. V, Figs. 2-7; PI. PI. VI, Fig. 4; PI. VII, Figs. 3, 5-6, 8, 10-11; PI. VIII, Figs. 1-18. Casts of the right femur (bmnh R 4070, 4074) and of the left humerus (bmnh R 4069) are kept at The Natural History Museum, London. Measurements of the Holotype The anterior tip of the rostrum is broken in the holotype. All measurements are given as pre- Table 1. Measurements of maxillary and palatine tooth plates of Placochelys placodonta (holotype, mafi Ob/2323/Vt.3). All measurements in mm; approximate values in parentheses. left right longi- tudinal 0 trans- verse 0 longi- tudinal 0 trans- verse 0 anterior maxillary tooth 8.2 - - 6.7 intermediate maxillary tooth 8.9 7.4 8.9 7.2 posterior maxillary tooth 15.7 11.4 - 11.8 anterior palatine tooth 13.8 11.0 14.0 11.0 posterior palatine tooth 27.2 20.8 (26) 21.0 served in the fossil. Measurements in parentheses are those of the right side of the skull. Measure- ments for the maxillary and palatine tooth plates are given in Table 1. Tip of rostrum to occipital condyle: 1 18.3 mm Tip of rostrum to mandibular condyle of quadrate: 119.2 (116.4) mm Tip of rostrum to posterior margin of skull table: 112.8 mm Maximal length of skull: 148 mm Tip of rostrum to anterior margin of external nar- is: 21.2 (20.5) mm Tip of rostrum to anterior margin of internal naris: 32.0 (32.0) mm Tip of rostrum to anterior margin of orbit: 39.1 (38.5) mm Tip of rostrum to anterior margin of upper tem- poral fossa: 83.5 (82.1) mm Longitudinal diameter of external naris: 12.5 (13.0) mm Transverse diameter of external naris: 9.8 (10.0) mm Longitudinal diameter of internal naris: 7.1 (7.0) mm RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA Transverse diameter of internal naris: 4.3 (4.1) mm Longitudinal diameter of orbit: 26.1 (27.4) mm Transverse diameter of orbit: 27.4 (24.8) mm Longitudinal diameter of upper temporal fossa: 52.1 (51.5) mm Transverse diameter of upper temporal fossa: 32.9 (32.4) mm Distance from external naris to orbit: 7.0 (7.0) mm Distance from orbit to upper temporal fossa: 18.5 mm Bony bridge between external nares: 6.3 mm Bony bridge between internal nares: 4.0 mm Bony bridge between orbits: 10.6 mm Bony bridge between upper temporal fossae: 8.2 Morphological Description of the Skull All cyamodontoid placodonts with the excep- tion of Henodus (Huene, 1936) have a skull with a broadly expanding temporal region and a nar- row, tapering rostrum. Proportions vary from a rather short and blunt rostrum in Cyamodus (No- sotti & Pinna, 1996) to a very elongate, narrow rostrum in Psephoderma (Pinna & Nosotti, 1989; see also further discussion below). Placochelys is intermediate between these two extremes, with an elongate and narrow rostrum formed by the pre- maxillae (Fig. 3A). The tip of the rostrum is bro- ken in the holotype, and the rostrum is altogether missing in mb.r. 1765. The rostrum is sufficiently well preserved in the holotype, however, to allow the conclusion that the premaxilla of Placochelys was edentulous. In ventral view, distinct longitu- dinal grooves can be seen running from the an- terior margins of the external nares anteriorly up to the broken tip of the rostrum. These grooves are delineated laterally by the maxilla (in their posterior part) and by the premaxilla (in their an- terior part), and medially by a raised ventral crest running along the medial edge of each premaxilla. The premaxilla broadly enters the anterior and lateral margin of the external naris. It forms a dis- tinct (autapomorphic) posterior process that ex- tends backward from the posteroventral corner of the external naris to a level shortly behind the anterior margin of the orbit, embraced both dor- sally and ventrally by the maxilla (Fig. 4A). A comparable process is not observed in other cy- amodontoid taxa. Dorsomedially, the premaxilla meets the nasal between the two external nares in a posterolaterally trending suture. In ventral view (Fig. 3B), the contact of the premaxilla with the maxilla posterolaterally and the vomer posteriorly is obscured by paint cov- ering the bone surface. In dorsal view, however, the anterior end of the maxilla can be seen to ex- pand medially to form most of the dermal floor of the external naris. The ventral outline of the anterior end of the maxilla has been reconstructed accordingly (broken lines in Fig. 3B), which in- dicates that the maxilla met the vomer along the anterior margin of the external naris, excluding the premaxilla from the latter. The maxilla forms a slender anterior process that runs along the ventrolateral margin of the ros- trum to a level well in front of the anterior margin of the external naris (Fig. 4A). It remains sepa- rated from the lateral margin of the external naris by the posterior process of the premaxilla. The maxilla expands medially below and deep to this posterior process of the premaxilla, as it forms most of the floor of the external naris. Between the external naris and the orbit, the maxilla forms a distinct ascending process with a pointed dorsal tip wedged in between the nasal anteriorly and the prefrontal posteriorly. The maxilla narrowly en- ters the anteroventral margin of the orbit, but fur- ther posteriorly it is excluded from the ventral margin of the orbit by the anterior process of the jugal. The posterior end of the maxilla forms an essentially vertical and deeply interdigitating su- ture with the jugal at a level somewhat in front of the posterior margin of the orbit but behind the level of the midpoint of the longitudinal diameter of the orbit. The lateral surface of the maxilla shows five to seven superior labial foramina (Fig. 4A). In ventral view (Fig. 3B), the maxilla is seen to enter the anterolateral margin of the internal naris. It contacts the vomer anteromedially in front of the external naris (the suture between the two bones is distinct at the anterior margin of the internal naris), and the palatine lateral to the ex- ternal naris. The maxilla remains excluded from the anterior margin of the subtemporal fossa by a lateral process of the palatine, which contacts the jugal. Each maxilla carries three tooth plates, of which the posteriormost one is distinctly larger than the two anterior ones (Table 1). In contrast to the holotype, mb.r. 1765 shows a distinct dental lamina foramen located on the palatine-maxillary suture posteromedial to the posterior maxillary tooth plate. FIELDIANA: GEOLOGY The nasals are paired, triangular elements that define the posteromedial margin of the external nares. They meet each other along the dorsal mid- line of the skull, separating the premaxilla from the frontal (Fig. 3A). The anterior tips of the na- sals lie at a level behind the anterior margin of the external nares. In the holotype, a narrow but deep cleft separates the nasals from one another posteriorly, exposing the underlying frontal. Giv- en the overall solid ossification of the skull and the tendency for the sutures to fuse in the dermal skull roof, it seems unlikely that the narrow cleft between the nasals reflects incomplete ossification of these latter elements. In no other cyamodontoid skull are the posterior parts of the nasals separated by a deep cleft exposing the underlying frontal. The posterolateral margin of the nasal runs from the posterior margin of the external naris in a pos- teromedial direction, contacting the ascending process of the maxilla, the anterior margin of the prefrontal, and the short anterolateral process of the frontal. The anterolateral process of the frontal therefore remains separated from the ascending process of the maxilla by the nasal and prefrontal (Fig. 3A). The prefrontal is a rather small element located at the anterodorsal margin of the orbit. A medial ventral process forms the anteromedial margin of the orbit. The location of the lacrimal foramen cannot be identified unequivocally in Placochelys. The anteroventral corners of the orbits are not pre- served in mb.r. 1765. The anteroventral margin of the right orbit is subject to breakage in the holo- type. In the left orbit of the holotype, the prefron- tal is seen to extend further down than in Cyamo- dus kuhnschnyderi and Protenodontosaurus, where the prefrontal remains excluded from the lacrimal foramen (Nosotti & Pinna, 1996, 1998; see further discussion of the latter two taxa be- low). In Placochelys the medial ventral process of the prefrontal reaches the maxilla and closely ap- proaches the anterior process of the jugal without quite reaching it (Fig. 3A). The position of the lacrimal foramen is again obscured by breakage and compression. However, this break might pass through an area of weakness indicating the posi- tion of the lacrimal foramen, in which case the prefrontal might have entered its dorsal margin. Breakage likewise obscures the location of the fo- ramen for the passage of the infraorbital division of the maxillary branch of the trigeminal nerve (infraorbital foramen sensu Oelrich, 1956) in the anteroventral corner of the orbit. However, Pla- cochelys does not show a distinct groove running along the anteroventral margin of the orbit, iden- tified as "basiorbital furrow" in Cyamodus kuhn- schnyderi by Nosotti and Pinna (1996). The frontals are paired in Placochelys, as in all other cyamodontoids. Short anterolateral process- es of the frontal are embraced between the pre- frontals and nasals. These anterolateral processes of the frontal are shorter (i.e., less well developed) in Placochelys than in some other cyamodontoids. The concave lateral margin of the frontal broadly enters the dorsal margin of the orbit between the prefrontal and the postfrontal. A massive break running obliquely through the skull table and the right postorbital arch obscures sutural details of the frontoparietal suture. However, a posterolat- eral lappet of the frontal is clearly identifiable on the left side of the skull, indicating that the fron- toparietal suture was located at a level between the posterior margin of the orbit and the anterior margin of the temporal arch (Fig. 3A). The postfrontal is a broad element that defines the posterodorsal margin of the orbit. Its ventral process tapers off along the posterior margin of the orbit but remains separated from the jugal by the postorbital. The posterior process is rather broad and extends backward to about the level of the anterior margin of the temporal fossa; it re- mains narrowly excluded from the anteromedial margin of the upper temporal fossa by a contact of the postorbital with the parietal. Posteriorly the postfrontal meets the parietal in an interdigitating suture, which slightly trends in an anteromedial direction. The medial margin of the postfrontal is more or less straight. The posterolateral margin of the postfrontal is deeply concave and angled in Placochelys, as it also is in Cyamodus kuhnschny- deri (Nosotti & Pinna, 1996), but unlike in Cy- amodus rostratus (Kuhn-Schnyder, 1965a), Pro- tenodontosaurus, or Psephoderma, where this bone has a less concave and more evenly curved posterolateral margin. The unpaired (fused) parietal forms a rather rectangular and flat skull table characterized by extensive dermal ornamentation (Fig. 3A). The parietal skull roof carries four distinct tubercular protuberances. Posterolateral processes of the pa- rietal define the posteromedial margins of the up- per temporal fenestrae as well as the deeply con- cave occiput and meet the squamosal in an inter- digitating suture that runs across an osteodermal encrustation located at the posteromedial margin of the upper temporal fossa (left side of skull in Fig. 3A). The lateral margin of the skull table nar- rowly projects laterally beyond the descending RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA Fig. 3. Skull of Placochelys placodonta Jaekel (holotype, fafi Ob/2323/Vt.3): A, dorsal view; B, ventral view. Scale bar = 20 mm. For abbreviations, see p. 3. flange of the parietal, which participates in the formation of a secondary lateral wall of the brain- case in a manner described in more detail below. The presence and position of the pineal fora- men in Placochelys remain uncertain. Jaekel (1907, PI. I) figured a relatively small pineal fo- ramen located between the parietals at the level of the anterior margin of the upper temporal fossa but clearly behind the frontoparietal suture. Jae- kel's (1907, PI. I) drawing includes an element of reconstruction, however, because a massive break passes through the skull at precisely this level. FIELDIANA: GEOLOGY Fig. 3. Continued. Huene (1931, PL I) nevertheless followed Jaekel's (1907) lead but increased the size of the pineal foramen located in the same position (i.e., behind the frontoparietal suture). Other cyamodontoids (Cyamodus, Protenodontosaurus, Psephoderma: see below) all have an equally large and anteriorly placed pineal foramen, but the frontal reaches far- ther back and narrowly approaches, or even en- ters, its anterior margin. Although Huene's (1931) reconstruction appears plausible by comparison with other cyamodontoids, the cast of the skull of the holotype before its preparation by Strunz shows a distinct splint of bone embedded in the break in exactly the location where Jaekel (1907) and Huene (1931) placed the pineal foramen. The two possible conclusions are that this splint of RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA V.0 sq B Fig. 4. Skull of Placochelys placodonta Jaekel (holotype, fafi Ob/2323/Vt.3): A, left lateral view; B, occipital view. Scale bar = 20 mm. For abbreviations, see p. 3. bone is in a natural position, in which case Pla- cochelys would lack a pineal foramen, or alter- natively, the splint of bone is in a displaced po- sition, in which case Jaekel's (1907) and Huene's (1931) reconstructions might be correct. The skull roof of mb.r. 1765 is severely damaged, with its anterior part missing. The break through the skull table lies at a level of the anterior margin of the upper temporal fossa. This again might indicate an area of relative weakness, perhaps caused by a large and anteriorly placed pineal foramen. The postorbital broadly enters the posteroven- tral margin of the orbit, from where it extends posteriorly to define the anterior and the greater part of the lateral margin of the upper temporal fossa. Huene (1931) believed the postorbital to extend far back along the medial margin of the upper temporal fossa, exposed in dorsal view lat- 10 FIELDIANA: GEOLOGY eral to the parietal. This configuration of the post- orbital could not be confirmed for the holotype, where the postorbital meets the parietal at the an- teromedial margin of the upper temporal fossa, or for mb.r.1765. In lateral view, however, the post- orbital can be observed to form a posterior ver- tical flange that extends backward for a consid- erable distance below the overhanging rim of the parietal skull table (Fig. 4A), overlapping the ven- tral flange of the parietal and completing the an- terodorsal part of the secondary lateral wall of the braincase dorsal to the epipterygoid. At the pos- terodorsal corner of the orbit, the postorbital forms a stout ventromedial process that abuts the lateral surface of the anterodorsal corner of the epipterygoid. The posterolateral process of the postorbital extends along the lateral margin of the upper temporal fossa to a level behind the mid- point of the longitudinal diameter of the latter. This postorbital process does not narrow signifi- cantly at its posterior extremity, and meets the squamosal in an interdigitating suture (but see dis- cussion below). The jugal carries a narrow anterior process that forms most of the ventral margin of the orbit dor- sal and medial to the maxilla (Figs. 3A, 4A). At the anterolateral margin of the subtemporal fossa, the jugal forms a distinct posteroventral lappet, ornamented with a pattern of radiating grooves and ridges. Behind the orbit, up to a level of about the midpoint of the postorbital arch, the jugal is sutured to the postorbital. More posteriorly, how- ever, the jugal separates from the postorbital and narrows to a pointed tip located somewhat in front of the posterior end of the postorbital, at about the level of the midpoint of the longitudinal diameter of the upper temporal fossa and about the lower third of the height of the temporal arch. Together, the dorsal margin of the jugal and the ventral mar- gin of the posterolateral process of the postorbital define a distinct V-shaped sutural pattern (the tip of the V pointing forward), thus embracing the anterior process of a bone whose identity remains to be discussed (see discussion below and Fig. 4A). The squamosal of cyamodontoid placodonts is a very complex bone. Its body defines the pos- terolateral margin of the upper temporal fossa as well as the dorsal, lateral, and ventral margins of the posttemporal fossa. A dorsomedial process of the squamosal meets the posterolateral process of the parietal at the posteromedial margin of the up- per temporal fossa anteriorly and at the dorsal margin of the posttemporal fossa ventrally. These sutural relations are very distinct, both in the ho- lotype of Placochelys and in mb.r. 1765 (Fig. 5A). In dorsal view, the contact of the squamosal with the parietal is bridged by an elongated der- mal encrustation located at the posterior margin of the upper temporal fossa (left side of the skull in Fig. 3A). This dermal encrustation tends to ob- scure the squamosal-parietal suture in the holo- type, particularly on the right side of the skull, whereas mb.r. 1765 clearly shows this interdigi- tating suture traversing the posterolateral part of the dermal encrustation (Fig. 5A). In lateral view and inside the temporal fossa, a narrow process of the squamosal can be followed along the dorsal margin of the posttemporal fossa, meeting the posterodorsal process of the epipterygoid in the anterodorsal corner of the posttemporal fossa. A similar contact of the epipterygoid with the squa- mosal at the anterodorsal corner of the posttem- poral fossa is also observed in Cyamodus rostra- tus (Kuhn-Schnyder, 1965a), but this character re- mains unknown for Cyamodus kuhnschnyderi (Nosotti & Pinna, 1996), Protenodontosaurus, and Psephoderma. A ventromedial process of the squamosal curves around the lateral margin of the posttem- poral fossa, thereby establishing a broad ventro- lateral contact with the quadrate. Specimen mb.r. 1765 shows particularly well how this process of the squamosal extends anteromedially along the anterior aspect of the paroccipital process (Fig. 5B). Sutured to the opisthotic, this process of the squamosal forms the posterolateral margin of the pteroccipital foramen (sensu Nosotti & Pinna, 1993b, to be discussed in detail below), which is bordered anteromedially by the opisthotic. A neomorphic process of the squamosal, the otic process sensu Nosotti and Pinna ( 1 993b), ex- tends anterolateral^ from the lower margin of the posttemporal fossa along the dorsal margin to the dorsomedial flange of the quadrate ramus of the pterygoid and meets the prootic at the anterolat- eral corner of the pteroccipital foramen. As a re- sult, the otic process of the squamosal forms the lateral margin of the pteroccipital foramen, which is bordered anterolaterally by the prootic (Figs. 3A, 6A, 6B). It is the anterior and lateral relations of the squamosal that remain the most controversial as- pect of the dermatocranium in placodonts (Pinna, 1989; Zanon, 1989). The debate largely results from the difficulty of separating the squamosal from the quadratojugal within the temporal arch of placodonts. Pinna (1989; see also Pinna & No- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 11 Fig. 5. Skull of Placochelys placodonta Jaekel (paratype, mb.r. 1765): A, dorsal view; B, ventral view. Scale bar 20 mm. For abbreviations, see p. 3. sotti, 1989; Nosotti & Pinna, 1993b) reconstructed a very narrow squamosal that is barely exposed in the lateral view of the temporal arch both in Placodus and cyamodontoids. This leaves room for an expansive quadratojugal that broadly enters the lateral margin of the upper temporal fossa be- tween the postorbital and squamosal. Examination of all Placodus skulls in public repositories (Riep- pel, 1995a) did not yield conclusive evidence to support Pinna's ( 1 989) reconstruction of the squa- mosal and quadratojugal in that taxon. In fact, none of the specimens shows a distinct and un- equivocal suture line separating the squamosal from the quadratojugal within the temporal arch (Rieppel, 1995a). The delineation of quadratoju- gal and squamosal in cyamodontoids is further 12 FIELDIANA: GEOLOGY Fig. 5. Continued. complicated because the lateral surface of the pos- terior part of the temporal arch is subject to the encrustation of dermal tubercles that obscure su- tural patterns. The only specimen apparently sup- porting Pinna's (1989) reconstruction of a small squamosal in cyamodontoids is Cyamodus cf. ros- tratus smns 17403 (Nosotti & Pinna, 1993b, Fig. 3), which shows what looks like a V-shaped con- tact of the squamosal with the quadratojugal at the posterolateral corner of the upper temporal fossa (but see the detailed description of the specimen below). A quadratojugal is unquestionably present in Placochelys, and it covers the lateral surface of the quadrate as in all other cyamodontoids. The suture separating the quadratojugal from the shaft of the quadrate is distinct in the occipital view of the skull (Fig. 4B). Equally distinct is the suture between quadratojugal and squamosal on the pos- terolateral aspect of the skull, at the level of the RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 13 pq. re Fig. 6. Placochelys placodonta Jaekel (holotype, fafi Ob/2323/Vt.3); right lateral braincase wall, partially recon- structed. Scale bar = 20 mm. For abbreviations, see p. 3. dorsal head of the quadrate. More anteriorly and laterally, this suture disappears under a dermal tu- bercle. Squamosal and quadratojugal therefore re- main indistinct from one another in the posterior part of the lateral surface of the temporal arch of the holotype (Fig. 4A). The medial surface of the temporal arch likewise offers no further clues. In mb.r. 1765, the same sutural relations be- tween postorbital and jugal can be observed as were described for the holotype. The two bones again embrace a V-shaped anterior process of a posterior element (Fig. 5A). The pointed posterior tip of the jugal again lies somewhat in front of the posterior end of the postorbital and above the 14 FIELDIANA: GEOLOGY ventral margin of the temporal arch. In this spec- imen, however, a suture line can be observed to extend backward from the posterior tip of the ju- gal in a more or less horizontal direction until it disappears under dermal encrustation on the pos- terolateral aspect of the temporal bar. In addition, the right side of the skull of MB. r. 1765 rather clearly shows the anterior tip of the quadratojugal tapering off along the ventral margin of the lower temporal arch below the jugal, reaching far for- ward to a level in front of the anterior margin of the upper temporal fossa (Fig. 5A). A similar an- terior extension of the quadratojugal can be re- constructed for the left temporal arch of the ho- lotype (Fig. 4A). A horizontal suture extending backward from the posterior tip of the left jugal can also be identified in "Macroplacus" raeticus (Schubert-Klempnauer, 1975; Rieppel, 1995a, Fig. 22), and, as in Placochelys, it appears to hor- izontally subdivide the posterior part of the tem- poral arch into a dorsal squamosal, broadly enter- ing the upper temporal fossa, and a ventral quad- ratojugal. This, in any case, is the most plausible reconstruction of the relationship of these two bones when observations on the holotype of Pla- cochelys placodonta and mb.r. 1765 are com- bined. All that is required is to link the suture that runs backward from the posterior tip of the jugal with the suture that separates the quadratojugal from the squamosal at the posterodorsal corner of the skull, a connection that is obscured by dermal encrustation of the temporal region of the skull. The anterior extent of the quadratojugal may be autapomorphic for Placochelys, but this character remains insufficiently known in other cyamodon- toids for meaningful comparison. If the quadratojugal is reconstructed, as argued above, to form most of the ventral margin of the temporal arch and to be separated horizontally from the squamosal, the jugal and postorbital are left to embrace between themselves the anterior tip of the squamosal. In Placochelys, the anterior tip of the squamosal also reaches to a level in front of the anterior margin of the temporal fossa, an autapomorphy of the genus in comparison to other cyamodontoids. In ventral view the skull shows the paired in- ternal nares to be separated by paired vomers (Fig. 3B). The sutural contact of the vomer with the maxilla is distinct at the anterior margin of the internal naris on both sides. The ventral surface of the rostrum is painted with resin, obscuring fur- ther details of the relationships of the vomer with the maxilla and premaxilla. However, there is no indication of large anterior processes of the vo- mers, entering deeply between the premaxillaries, as is indicated in the figure published by Huene (1931). Posteriorly, the vomer contacts the pala- tine at the posteromedial corner of the internal naris. As in all placodonts, the palatine is enlarged at the expense of the pterygoid. In Placochelys, it carries a smaller anterior and a much enlarged posterior tooth plate (Table 1 ). The palatine forms most of the posterior and lateral margin of the internal naris. Anteriorly and laterally, the palatine contacts the maxilla. At the anterior margin of the subtemporal fossa, the palatine carries a distinct lateral process that embraces the posterior end of the maxilla and contacts the jugal laterally (Fig. 3B). This lateral process cannot represent the ec- topterygoid, as indicated in the figure published by Huene (1931), because it carries on its dorsal surface the anterior tip of a groove that housed the palatoquadrate cartilage in the living animal. Posteriorly, the palatine meets the pterygoid in an interdigitating suture that curves around the pos- terior margin of the posterior palatine tooth plate. Within a distinct depression on the palatine-pter- ygoid suture lies the large, transversely oriented dental lamina foramen, located posteromedial ly to the posterior palatine tooth plate (Fig. 3B). A small dental lamina foramen is located postero- medially to the anterior palatine tooth plate in mb.r. 1765 (Fig. 5B), but a similar foramen is not distinct in the holotype. In both specimens, irreg- ular suture lines are observed at the palatine-pter- ygoid contact, delineating a triangular area in the midline of the palate between the two bones (Figs. 3B, 5B). It appears possible that the dermal palate includes a small heterotopic ossification between palatines and pterygoids. In lateral view, the pal- atine can be seen to extend backward medial to the pterygoid to meet the quadrate along the lat- eral margin of the palatoquadrate cartilage recess (described in more detail below); more anteriorly, the dorsal surface of the palatine carries a groove that housed the palatine process of a cartilaginous palatoquadrate, which persisted in the adult. The pterygoids complete the posterior part of the dermal palate. They meet one another in an interdigitating suture, which in the holotype shows an irregularly curved course (Fig. 3B). Lat- erally, the pterygoid forms a prominent, longitu- dinally oriented ventral flange and extends ante- riorly to the level of the posterior third of the lon- gitudinal diameter of the posterior palatine tooth plate. The ventral pterygoid flange forms a single RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 15 ventral projection. Compared to other cyamodon- toids (see below), the palatal exposure of the pter- ygoid is relatively long in Placochelys: dividing the distance from the posterior margin of the pter- ygoid to the posterior dental lamina foramen by the distance from the posterior dental lamina fo- ramen to the posterior margin of the internal naris yields a quotient of 0.45. Posterolateral^, the pterygoid forms a short quadrate ramus that in ventral view meets the anteromedial aspect of the quadrate in an interdigitating suture. The quadrate ramus of the pterygoid carries a broad dorsal flange that extensively overlaps a broad antero- medial flange of the quadrate, thus obliterating the cranioquadrate passage anteriorly. The overlap of these two bones is well exposed in lateral view, as will be discussed in more detail below. The presence or absence of an ectopterygoid in cyamodontoid placodonts again remains a matter of debate. Although generally assumed to be pre- sent (Huene, 1931; Pinna & Nosotti, 1989; No- sotti & Pinna, 1996), its delineation from the pal- atine and pterygoid has been notoriously difficult. The presence of an ectopterygoid in Psephoderma could not be ascertained (personal observation), and an ectopterygoid is positively absent in Pro- tenodontosaurus (Nosotti & Pinna, 1998; personal observation). The lateral view of the skull (holo- type) of Placochelys shows the pterygoid to be sutured to the quadrate posterodorsally and to the palatine anterodorsally (Figs. 6A, 6B). As is also seen in ventral view, the anterolateral tip of the pterygoid reaches to about the posterior third of the longitudinal diameter of the posterior palatine tooth plate. Exposed in lateral view in front of the pterygoid, a suture line appears to separate from the palatine a small, splintlike element that might represent an ectopterygoid. If so, the ectoptery- goid would only line the anteromedial margin of the subtemporal fossa with hardly any ventral ex- posure at all. However, the supposed palatine-ec- topterygoid suture seen in lateral view could also represent a crack, since no clear separation of an ectopterygoid from the palatine can be seen in ventral view. Two elements ossify in the palatoquadrate of reptiles, the quadrate and the epipterygoid. The quadrate of Placochelys can be described as being composed of a shaft and a broad anteromedial wing. The posterior aspect of the quadrate shaft is distinctly concave. The lateral surface of the shaft is covered by the quadratojugal (Fig. 3A). The mandibular condyle of the quadrate is bipar- tite, a central concavity matching the saddle- shaped articular surface of the lower jaw. A shal- low stapedial recess is located on the anteromedial aspect of the quadrate shaft, located narrowly above the mandibular condyle (Figs. 3B, 5B); it must have received the (cartilaginous?) distal end of the stapes. In the holotype, a relatively large foramen can be identified, located lateral to the dorsal head of the quadrate, between the latter and the squamosal (Fig. 4B). The quadratojugal is ex- cluded from this foramen on the left side but nar- rowly enters its ventral margin on the right side. The foramen may have served the passage of a lateral branch of the internal carotid or stapedial artery respectively to the temporal musculature. A comparable foramen is not distinct in mb.r. 1765 (obscured by deformation of the skull?) and is ab- sent in all other cyamodontoids (but see the dis- cussion of Macroplacus, below). The broad an- teromedial wing of the quadrate forms the sloping posterior wall of the temporal fossa from which the posterior part of the external jaw adductor must have originated. It is well exposed in lateral view, as it demarcates the posterolateral margin of a deep palatoquadrate cartilage recess (Fig. 6). The broad and complex epipterygoid is the dominant element in the secondary lateral wall of the braincase. The bone can be described as con- sisting of two parts, an anterior portion with a deeply concave lateral surface and a posterior por- tion with a distinctly convex lateral surface (Fig. 6). These two parts of the epipterygoid may cor- respond to the anterior "palatal ramus" and pos- terior "quadrate ramus" of the epipterygoid of Cyamodus kuhnschnyderi (Nosotti & Pinna, 1996), although the opening that separates these two parts of the epipterygoid in the latter taxon is absent in Placochelys (the lateral opening within the epipterygoid of Cyamodus kuhnschnyderi ap- pears to be the result of incomplete ossification; Nosotti & Pinna, 1996: 27). The posterior part of the epipterygoid of Placochelys shows a deeply concave posterior margin that defines the anterior margin of the trigeminal incisure, enclosed be- tween the epipterygoid and the prootic. Postero- dorsally, the epipterygoid is extended into a long, slender process that runs across the dorsal margin of the prootic and meets the squamosal in the an- terodorsal corner of the posttemporal fossa (Fig. 6). Posteroventrally, the epipterygoid narrowly contacts the prootic on the left side of the skull (of the holotype), thus closing the trigeminal in- cisure ventrally; a similar contact is absent on the right side of the skull. The ventral margin of the posterior part of the epipterygoid shows a surface 16 FIELDIANA: GEOLOGY of unfinished bone, which overhangs the medial margin of the palatoquadrate cartilage recess. The anterior part of the epipterygoid is sutured to the dorsal surface of the palatine and reaches far for- ward to a level dorsal to the posterior palatine tooth plate (Figs. 5A, 5B). The anteroventral tips of the epipterygoids of both sides converge to- ward the midline of the skull. The prootic is exposed in lateral view (Fig. 6) as it emerges from below the narrow posterodor- sal process of the epipterygoid. Its lateral surface is distinctly convex. The anterior margin of the prootic defines the posterior margin of the trigem- inal incisure. The ventral margin of the prootic broadly meets the neomorphic otic process of the squamosal in an oblique suture. These two ele- ments together form the ventral margin of the posttemporal fossa, which also represents the an- terior margin of the pteroccipital foramen (Figs. 3A, 6A). Laterally, the otic process of the squa- mosal together with the ventral margin of the pro- otic define the posteromedial margin of the pala- toquadrate cartilage recess. The palatoquadrate cartilage recess is a char- acter of all cyamodontoid placodonts included in this study whose skulls are adequately preserved and prepared to reveal the morphology of the lat- eral braincase wall (Figs. 3A, 4A, 6). It is repre- sented by a characteristic cleft with a triangular outline of its posterior part, bordered laterally by the medial wing of the quadrate and by the pala- tine, bordered medially by the squamosal, prootic, and epipterygoid, and floored by the pterygoid posteriorly and by the palatine anteriorly. This re- cess continues anteriorly as a distinct groove on the dorsal surface of the palatine, which in its an- terior part turns laterally toward the anterior cor- ner of the subtemporal fossa. In the living animal, this recess and groove must have housed cartilage of the palatoquadrate, which persisted in the adult in its classic position dorsal to the dermal palate (pterygoid and palatine) and which connected the two ossifications within the palatoquadrate (i.e., the quadrate and epipterygoid). The pteroccipital foramen (Nosotti & Pinna, 1993b) is another synapomorphy shared by all cy- amodontoids included in this study with adequate- ly preserved and prepared skulls. Whereas Pla- codus retains a complete cranioquadrate passage (Rieppel, 1995a), the latter is obliterated in cy- amodontoids in its anterior part by fusion of the dermal palate to the basicranium and by the broad overlap of the dorsal wing of the pterygoid with the anteromedial wing of the quadrate. This over- lap also closes a gap which in the skull of Pla- codus persists between the quadrate and pterygoid laterally and the lateral braincase wall medially, and through which the stapedial (temporal) artery reached the jaw adductor musculature. In cyamo- dontoid placodonts, the stapedial (temporal) ar- tery reaches the jaw adductor musculature through the pteroccipital foramen, located at the ventral margin of the posttemporal fossa (Figs. 3, 5, 6). It is bordered anteriorly by the otic process of the squamosal and the prootic, and posteriorly by the squamosal and the opisthotic, the latter ex- tended into the paroccipital process. As the sta- pedial (temporal) artery branches off the internal carotid within the posterior part of the crani- oquadrate passage, it turns dorsally to pass through the pteroccipital foramen in front of the paroccipital process; it reemerges from the same foramen at the ventral margin of the posttemporal fossa, through which it gains access to the tem- poral musculature. The occipital view of the skull features the rel- atively large posttemporal fossae, bordered dor- sally by the occipital flange of the parietal and squamosal and ventrally by the slender paroccip- ital process (Fig. 4B). The supraoccipital is a trap- ezoidal element carrying a low median crest. It is obliquely inclined, contacts the parietal in an in- terdigitating suture anteriorly, and defines the dor- sal margin of the foramen magnum posteriorly. The lateral margin of the foramen magnum is formed by the exoccipital. The exoccipitals of ei- ther side do not meet dorsal to the occipital con- dyle, which is formed by the basioccipital only. The occipital condyle of mb.r. 1 765 shows a dis- tinct notochordal pit on its posterior surface (Fig. 5B). The metotic (jugular or vagus) foramen (Rieppel, 1985) is enclosed between the exoccip- ital and the opisthotic. Neither the holotype nor mb.r. 1765 shows an internal subdivision of the metotic foramen that would have separated the exit of the roots of the glossopharyngeal, vagus, accessory, and hypoglossal nerves. Below the me- totic foramen and lateral to the occipital condyle, the braincase forms two ventral processes on ei- ther side of the skull. The medial one, close to the occipital condyle, is the ventrally directed ba- sioccipital tuber. A ventral process of the opis- thotic is located lateral to the basioccipital tuber. The two ventral processes together enclose an opening or foramen, which may have trapped the internal carotid at the entry into the cranioquad- rate passage. In the holotype, the basioccipital tu- ber and the opisthotic process do not extend ven- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 17 Hfr "■M . §5 ' '''•'•'■ ■ ^H '''': ^« ■ ^ ■■ ._ SS^m''-y^: ^K^> , iHbP^Pj^B t,lBRn'i!''!-y- ■■» v •'--.""'' • : Ik^Se^- ■ ■\^B 5? W^Lsr v V' Byjfc. ■"*>« Fig. 7. Placochelys placodonta Jaekel (paratype, mb.r. 1765); contact between the left paroccipital process (opisthotic) and the left squamosal. Scale bar = 20 mm. trally to meet the basisphenoid or the posterior margin of the pterygoid, and the two structures also fail to make a ventral contact with each other, such that the passage of the internal carotid be- tween the opisthotic and basioccipital is open ven- trally (Fig. 4B). In mb.r. 1765, the opisthotic pro- cess and the basioccipital tuber are more strongly developed and meet each other ventral to the pas- sage of the internal carotid, which is thereby cap- tured in a closed foramen (Fig. 5B). Whether the close approximation of these two processes to the basisphenoid and pterygoid reflects natural rela- tions or results from dorsoventral compression of the skull is difficult to determine. But given the relatively larger size of both structures in the specimen mb.r. 1765, it appears possible that the area was damaged during preparation of the ho- lotype. In both skulls of Placochelys, the opis- thotic is pierced by a small foramen located just above the ventral flange. The function of this fo- ramen remains unknown; separate exit(s) for the root(s) of the hypoglossal nerve would be ex- pected to be located within the exoccipital medial to the passage of the internal carotid. Laterally, the opisthotic extends into a slender paroccipital process. The distal end of the paroc- cipital process abuts a distinct buttress, located on the lower surface of the squamosal medial to the dorsal head of the quadrate (Figs. 3B, 5B, 7). In mb.r. 1765, but not in the holotype, the squamosal buttress can be seen to extend into a medially di- rected ridge on the lower surface of the squamo- sal, which follows the anterodorsal aspect of the paroccipital process and meets the opisthotic in an oblique suture at the posterior margin of the pteroccipital foramen (see above for a detailed de- scription of the squamosal). Medial to this con- tact, mb.r. 1765 also displays the vestibular (oval) fenestra of the otic capsule, enclosed by the pro- otic anteriorly and the opisthotic posteriorly (Fig. 5B). Opisthotic and prootic remain separate, as is indicated by a distinct suture at the dorsal margin of the vestibular fenestra. The vestibular fenestra lies deep inside the posterior part of the cranio- quadrate passage. In front of it, the cranioquadrate passage is obliterated by the fusion of the dermal palate to the basicranium and by the broad overlap of a dorsal flange of the pterygoid with the an- teromedial flange of the quadrate. No stapes is preserved in mb.r. 1765, but an imaginary straight line that extends from the vestibular fenestra in a posterolateral direction connects the latter with the stapedial recess on the quadrate. The location of the pteroccipital foramen behind this imaginary line would seem to indicate that the stapedial ar- tery, which branches off from the internal carotid after the latter has entered the cranioquadrate pas- sage through the gap between the basioccipital tu- ber and the ventral opisthotic process, would have passed behind the stapes on its way to the tem- poral region of the skull. However, the location of the pteroccipital foramen with respect to the par- occipital process in the three-dimensionally pre- served holotype of Placochelys suggests that the stapedial artery passed in front of the stapes in- stead. The left side of the occiput of the holotype shows what looks like a distally bifurcated par- occipital process, or a sturdy process emerging from behind and below the paroccipital process (Figs. 4B, 8). This element was tentatively iden- tified as the distal end of a massive stapes by No- sotti and Pinna (1996: 33). Following this inter- pretation, cyamodontoid placodonts would have lost an impedance matching middle ear and in- creased the volume of the stapes in a system that relied on sound transmission through bone only. Close inspection of the holotype shows, however, that the supposed stapes is not distinctly separated from the paroccipital process. Indeed, the two structures appear to be separated by breaks only, and where breaks are absent, the two structures appear to be confluent. In search of alternative explanations for the supposed stapes, reference may be made to the "lateral tubercle" on the par- occipital process of Cyamodus kuhnschnyderi: "By its postero ventral margin, close to the squa- mosal, the paroccipital process bears a small, downward and posteriorly projecting tubercle" (Nosotti & Pinna, 1996: 19). This structure, how- FIELDIANA: GEOLOGY Fig. 8. Placochelys placodonta Jaekel (holotype, fafi Ob/2323/Vt.3); the skull in occipital view, showing the left paroccipital process. For further discussion see text. Scale bar = 20 mm. ever, is much smaller than the supposed stapes in the holotype of Placochelys, and a comparable "lateral tubercle" is absent on the perfectly pre- served paroccipital processes of mb.r.1765. In- stead, the latter specimen shows a slight expan- sion of the ventral surface of the paroccipital pro- cess into a weakly protruding flange with a rugose surface indicating muscle attachment, probably of the longissimus capitis muscle. Even if broken, this flange would not be large enough to match the supposed stapes in the holotype. However, mb.r. 1765 also shows a distinct ridge on the ven- tral surface of the squamosal, originating at the squamosal buttress for the paroccipital process and extending medially along the anterodorsal as- pect of the latter (Fig. 5B). A comparable ridge is not easily identified in the holotype, and perhaps the supposed stapes could represent a broken seg- ment of this ridge. In the end, however, the iden- tity of the bone fragment tentatively identified as stapes in the holotype of Placochelys remains de- batable, as was also noted by Nosotti and Pinna (1996). Should it indeed represent the stapes, it would add further weight to the argument that the stapedial (temporal) artery passed in front of the latter on its way through the pteroccipital fora- men. Breakage of the skull roof in mb.r. 1765 ex- poses details of the basicranium in dorsal view (Fig. 5A). Easily identified is the sella turcica on the basisphenoid-parasphenoid complex, located in front of a distinct dorsum sellae and pierced by two foramina through which the cerebral carotids must have gained access to the cranial cavity. The basisphenoid-parasphenoid complex is broken shortly in front of the sella turcica. The passage of the cerebral carotids through foramina in the sella turcica indicates that the internal carotid must have pursued an intracranial course after having traversed the posterior part of the cranio- quadrate passage and after having given rise to the stapedial (temporal) artery. Unfortunately, none of the available skulls of cyamodontoid pla- codonts allows detailed reconstruction of the course of the internal carotid canal through the basicranium. Only in Cyamodus kuhnschnyderi is there some indication that the internal carotid en- tered the basicranium between the basisphenoid and the otic capsule at the depth of the cranio- quadrate passage. The internal carotid likewise passes between the basisphenoid and the otic cap- sule in the eosauropterygian genera Nothosaurus and Simosaurus (Rieppel, 1994a). Continuing an- teriorly within this basicranial canal, the internal carotid must have bifurcated, the dorsal branch emerging through the sella turcica as cerebral ca- rotid and the anterior branch continuing within the basicranial canal as palatine artery. This recon- struction of the course of the internal carotid rais- es the question as to how the palatine artery gained access to the soft palate, which it supplies with arterial blood. The only foramina in the der- mal palate located at a level in front of the sella turcica are those identified above as dental lamina foramina. The large foramina located posterome- dial to the posterior palatine tooth plates lie in close proximity to the foramina for the cerebral carotids through the sella turcica, which indicate the level of bifurcation of the internal carotid. It is conceivable that the palatine artery could have gained access to the soft palate by passing through these posterior dental lamina foramina. Similar questions relate to the passage of the lateral head vein and the profundus branch of the trigeminal nerve through the cavum epiptericum, which, in a generalized reptile skull, constitutes the anterior continuation of the cranioquadrate RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 19 passage, obliterated in cyamodontoids. mb.r. 1765 shows the basisphenoid-parasphenoid complex to lie dorsal to the dermal palatines (Fig. 5A). No ossified base of the pila antotica (clinoid process) can be seen ascending dorsally from the lateral margin of the basisphenoid at the level of the sella turcica or dorsum sellae. There appears to be no ossification in the primary lateral wall of the braincase in Placochelys, in contrast to Placodus, where such an ossification was described as an "alisphenoid bridge" by Broili (1912). The epipterygoid is located well lateral to the outer margin of the basisphenoid-parasphenoid complex, the latter representing the level of the primary lateral wall of the braincase. The body of the epipterygoid is obliquely inclined with respect to the dorsal surface of the palatine, meeting the latter at an acute angle. Through this relation of the bones, a narrow space becomes enclosed be- tween the medial surface of the epipterygoid and the dorsal surface of the palatine, as can also be seen in a horizontally split skull fragment of a cyamodontoid from the Muschelkalk of Makhtesh Ramon, Negev (HUJ-Pal. 220; Rieppel, Mazin & Tchernov, 1999, Fig. 7). mb.r. 1765 is unusual, however, in that a vertical lamina of bone appears to descend from the margins of the broken skull roof to the dorsal surface of the palatine medial to the epipterygoid but lateral to the lateral margin of the basisphenoid-parasphenoid complex (Fig. 5A). This vertical lamina of bone appears to be separated from the epipterygoid by a gap that opens in a minute "foramen" or cleft close by the anterior tip of the epipterygoid on its medial side. Because it has no relationship to the lateral mar- gin of the basisphenoid-parasphenoid complex, this vertical lamina of bone cannot represent an ossification of the primary lateral wall of the braincase. Also, the anterior "foramen" located between the vertical lamina and the epipterygoid appears to be too small to represent the anterior opening of a cavum epiptericum enclosed be- tween the vertical lamina and the epipterygoid. It therefore appears that the epipterygoid has been longitudinally split because of dorsoventral com- pression of the skull, resulting in a vertical lamina that appears to be separated from the laterally placed part of the epipterygoid. Alternatively, it might be assumed that the epipterygoid broke along a horizontal line and that the dorsal part of the bone, which is very thin at its anterior margin in the holotype, was pushed ventrally medial to the base of the epipterygoid because of dorsoven- tral compression. This interpretation is supported by the observation that neither the holotype of Placochelys nor the horizontally split skull of the cyamodontoid from Makhtesh Ramon (HUJ-Pal. 220) show a vertical lamina of bone separated from, and located medial to, the basal part of the epipterygoid that, because it cannot be an ossifi- cation of the primary lateral wall of the braincase, would have to represent a vertical downgrowth from the parietal, frontal, postorbital, or a com- bination thereof. Whereas the maxillary and mandibular branch- es of the trigeminal nerve would have emerged from the trigeminal incisure behind the epiptery- goid, the profundus branch would have emerged anteriorly from the cavum epiptericum (i.e., from the gap between the epipterygoid and the lateral margin of the basisphenoid-parasphenoid com- plex). The course of the lateral head vein, which in reptiles passes through the cranioquadrate pas- sage and the cavum epiptericum, is less easily re- constructed for cyamodontoids because of the obliteration of the anterior part of the cranioquad- rate passage (Nosotti & Pinna, 1996). Kuhn- Schnyder (1960) suggested that the lateral head vein would have escaped the posterior part of the cranioquadrate passage through the pteroccipital foramen, rather than passing through the cavum epiptericum. This certainly remains a possibility, but it must be remembered that there is a certain plasticity in the differentiation of the cranial ve- nous system. During embryonic development, the primary head vein (vena capitis medialis) forms loops and sinuses surrounding the roots of the cra- nial nerves and the auditory sac as it becomes replaced, at least in part, by the lateral head vein (vena capitis lateralis), and as differences in the differentiation of these veins persists among rep- tiles in general, and in squamates in particular (van Gelderen 1924; Goodrich, 1930), the cranial veins of cyamodontoids may have been differen- tiated in a pattern that may not be easily compared with that seen in extant reptiles. Another unsolved problem of the cranial anat- omy of cyamodontoid placodonts is the course of the palatine and hyomandibular branches of the facial nerve. In reptiles, the facial nerve usually exits through a foramen located between the com- missura basicapsularis and the commissura prae- facialis of the endocranium, which link the otic capsule with the basal plate and which both ossify as part of the prootic. However, the prootic of Placochelys lacks a foramen for the exit of the facial nerve (Fig. 6), as is also the case in Cy- amodus kuhnschnyderi (Nosotti & Pinna, 1996). 20 FIELDIANA: GEOLOGY Table 2. Measurements of the dentary tooth plates of Placochelys placodonta (holotype, mafi Ob/2323/ Vt.3). All measurements in mm; approximate values in parentheses. left right longi- :uJnul0 trans- verse 0 longi- tudinal 0 trans- vcrsc 0 anterior dentary tooth 19.1 12.0 18.0 12.0 posterior dentary tooth 29.0 20.3 28.5 (21) One specimen (mfsn 1923GP) of Protenodonto- saurus was observed to show an internal subdi- vision of the trigeminal incisure by a vertical strut of bone (Nosotti & Pinna, 1996). Should the pos- terior division of the incisure have served the exit of the hyomandibular branch of the facial nerve, the latter would have to have reached the middle ear region through the pteroccipital foramen (No- sotti & Pinna, 1996). Similarly, the paratype of Cyamodus kuhnschnyderi (smns 1 6270) shows, on the right side of the skull, a foramen at the ventral margin of the prootic closely behind the trigemi- nal incisure. Should this foramen have served as the exit of the hyomandibular branch of the facial nerve, it would again have to be assumed that the latter reached the middle ear cavity through the pteroccipital foramen. Alternatively, the hyoman- dibular branch might have exited the prootic in close proximity to the vestibular (oval) fenestra deep in the cranioquadrate passage, a region not well exposed in any of the cyamodontoid skulls available for study. The palatine branch of the fa- cial nerve, on the other side, appears to have pur- sued an intracranial course, joining the internal carotid and palatine artery respectively on their way through the basicranium (Nosotti & Pinna, 1996). Morphological Description of the Lower Jaw The lower jaw (Fig. 9) was separated from the skull by Strunz after the description of the holo- type by Jaekel (1907). As is true for the rostrum, the tip of the mandibular symphysis is incomplete. The retroarticular processes also appear to be in- complete, probably because of damage inflicted on them through Strunz's removal of dermal tu- bercles that were fused to the articular. The tip of the right coronoid process is incomplete, whereas the tip of the left coronoid process, although sub- ject to some damage, appears to retain its original height at least at its anterodorsal corner. The total length of the (better preserved) left mandible is 122 mm; its total height at the anterodorsal mar- gin of the coronoid process is 56.5 mm. Each den- tary carries on its posterior part two tooth plates, of which the posterior one is distinctly larger than the anterior one (Table 2). The dentaries form the deep, massive, and elon- gated mandibular symphysis, with some partici- pation of the splenials at the ventral margin of the latter. Posteriorly, the dentary extends to a level well behind the anterior margin of the large cor- onoid process, thus carrying the large posterior tooth plate to a position partially medial to the anterior part of the coronoid process. The dentary is broadly exposed in lateral view, whereas the splenial gains only a limited lateral exposure along the ventral margin of the anterior part of the mandible. These relations are reversed on the medial aspect of the lower jaw, with a relatively narrow exposure of the dentary and a broad ex- posure of the splenial. The two bones are sepa- rated from one another by the prearticular. The coronoid process is formed exclusively by the large coronoid, the dominant element on the lateral surface of the lower jaw. As is typical for most cyamodontoids, the coronoid reaches far down, closely approaching the ventral margin of the lower jaw (Fig. 9A). An anteroventral process of the coronoid, which overlaps the posterior part of the dentary, tapers to a pointed tip at a level below the anterior dentary tooth plate. The pos- teroventral margin of the coronoid is rounded in- stead (distinct on the right mandible only). The lateral surface of the coronoid is turned outward along its ventral margin, forming a distinct shelf that overhangs the contact of the splenial with the angular along the ventral margin of the lower jaw. This shelf delineates the ventral extent of the area of insertion for superficial jaw adductor muscle fibers on the lateral surface of the coronoid. The posterior part of the lower jaw includes the large angular, which extends anteriorly to a level in front of the apex of the coronoid process, where it tapers off in an overlapping sutural contact with the splenial. Behind the coronoid process, the lat- eral exposure of the angular is larger (deeper) than that of the surangular. The suture between the sur- angular and angular is not very distinct, however, but seems to correspond to a distinct and curved ridge on the lateral surface of the lower jaw. Con- cave ventrally, this ridge delineates the insertional facet for the superficial portion of the pterygoi- deus muscle. Along the dorsolateral edge of the RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 21 Fig. 9. Lower jaw of Placochelys placodonta Jaekel (holotype, fafi Ob/2323/Vt.3): A, lateral view; B, medial view. Scale bar = 20 mm. For abbreviations, see p. 3. mandible and behind the coronoid process, the surangular forms a distinct, dorsally protruding rim lateral to the articular surface of the lower jaw, but does not appear to participate in the for- mation of this articular surface itself. The articular surface of the mandibular joint is saddle-shaped, matching the surface of the mandibular condyle of the quadrate. The articular extends posteriorly into a short retroarticular process. The retroarti- cular process is very distinct in the placodontoid genera Placodus and Paraplacodus (Rieppel, 1995a). In the better known genus Placodus, the retroarticular process is long, slender, and turned slightly upward. In contrast, the retroarticular pro- cess of cyamodontoid placodonts is generally short, stout, and has a posteroventrally sloping surface, as is seen in a perfectly preserved left mandible (Fig. 10) of Cyamodus hildegardis (pi- muz T2796; Kuhn-Schnyder, 1959, PI. 1, Fig. b; Pinna, 1992, PI. 10, Fig. 8). In Placochelys the chorda tympani foramen is located on the dorsal surface of the retroarticular process closely behind the articular facet within the articular. The prearticular is a large element that, together with the angular, closes the large Meckelian canal medially (Fig. 9B). The latter opens posteriorly on the medial aspect of the lower jaw, where it forms a deep adductor fossa. The adductor fossa has a considerable longitudinal extension medial to the coronoid process and extends downward to 22 FIELDIANA: GEOLOGY ang Fig. 10. Lower jaw of Cyamodus hildegardis Peyer (pimuz T2796) in lateral view. Scale bar = 20 mm. For abbreviations, see p. 3. the ventral margin of the lower jaw. Two parallel, longitudinally oriented ridges are located on the dorsal surface of the angular at the bottom of the adductor fossa. These may have secured the at- tachment of tendinous plates of the bodenaponeu- rosis, into which inserted the fibers of the medial and deep portions of the external jaw adductor muscle. Similar ridges are seen at the bottom of the adductor fossa in Placodus (Rieppel, 1995a). Dermal Ornamentation of the Skull Cyamodontoid placodonts generally show a tu- bercular dermal ornamentation of the temporal bones of the skull. A closer look at Placochelys reveals two different components in the dermal ornamentation of the skull. The postfrontal, post- orbital, and the parietal in particular show a pat- tern of ornamentation that might be called dermal encrustation, resulting in the formation of projec- tions or bosses (Figs. 3A, 4A). The ossification center of the postfrontal and postorbital is elevat- ed into a low and blunt apex, from which grooves and ridges radiate toward the margins of the bone. Reflecting the growth pattern of the underlying bone, this dermal encrustation appears to have de- veloped simultaneously with the ossification of the underlying bone itself. The same appears to apply to the four relatively low and ill-defined tu- bercular encrustations observed on the parietal skull table and to the encrustations across the pa- rietal-squamosal suture at the posteromedial mar- gin of the upper temporal fossa. Whether these projections or bosses reflect areas of epidermal scales remains unknown, although it is likely. A much more pronounced pattern of dermal or- namentation is found on the surface of the pos- terior part of the squamosal and quadrat ojugal in the form of distinct dermal tubercles that are par- ticularly well-developed at the posterolateral mar- gins of the temporal region of the skull (Figs. 3, 4). These tubercles have a sharply defined base, which appears to have secondarily fused to the underlying bone. Their position correlates in no way with the growth pattern of the underlying bone. In specimens of Cyamodus hildegardis (pi- muz T4763, T4771; Peyer, 1931a), it can be seen that the large temporal tubercles located on the posterior margin of the squamosal (pimuz T4771; Pinna, 1992, Fig. 7, right side of skull) match those aligned along the lateral margins of the car- apace (pimuz, T4763; Peyer, 1931a, PI. 15, right posterolateral margin of dorsal shield) both in size and shape. To judge from the mode of their at- tachment to the surface of the underlying tempo- ral bones of the skull, it appears that these large tubercles initially developed independently from the underlying bone and only later fused to the surface of the latter. The inference that these large and distinct der- mal tubercles are secondarily fused to the under- lying bone is supported by Jaekel's (1907) figure of the holotype of Placochelys, which shows on both sides of the skull two such tubercles fused to the dorsal surface of the retroarticular process closely behind the mandibular joint, formed by the articular. As the latter represents an endoskel- etal element, dermal tubercles must have fused to it after its ossification. The dermal tubercles on the retroarticular process were removed through preparation by Strunz. Comparison of the Cranial Anatomy of Placochelys placodonta with That of Other Cyamodontoid Placodonts A number of well-preserved and well-prepared skulls of cyamodontoid placodonts are available that allow a detailed comparison with the skull of Placochelys placodonta. These include Cyamodus rostratus (Miinster, 1839) (umo BT 748; original of Drevermann, 1928, and Kuhn-Schnyder, 1965a; smns 17403, referred specimen); Cyamo- dus kuhnschnyderi Nosotti and Pinna, 1993a (smns 15855, smns 16270; see also Nosotti & Pin- na, 1996); Henodus chelyops v. Huene, 1936 (gpit "specimens I and II," syntypes of Huene, 1936; "specimens IV and VI," collected in 1959). Ma- croplacus raeticus Schubert-Klempnauer, 1975 (bsp 1967 I 324); Protenodontosaurus italicus RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 23 Fig. 11. Skull of Cyamodus rostratus Miinster (holotype, umo BT 748): A, dorsal view; B, ventral view; C, left lateral view. Scale bar = 20 mm. Pinna, 1990b (mfsn 1819GP, mfsn 1923GP; see also Nosotti & Pinna, 1998), and Psephoderma alpinum Meyer, 1858a, b (msnm V471, see also Pinna & Nosotti, 1989). All of the taxa mentioned above have been subject to personal observation and form the basis of the following comparative analysis. The Cranial Anatomy of Cyamodus rostratus (Munster, 1839) Cyamodus rostratus is the type species of the genus Cyamodus (H. v. Meyer, 1863). It is rep- resented by an incomplete skull (umo BT 748, ho- lotype) from the lower upper Muschelkalk (mol) of Bayreuth (Bindlach, Lainecker Hohenzug), Ba- varia, Germany (Fig. 11). umo BT 2172 and smf R-4040 are two isolated lower jaws from the same locality, smns 17403 is an incomplete skull from the lower upper Muschelkalk (Trochitenkalk, mol) from the Burrer Quarry, Gaismuhle near Crailsheim, southern Germany. Cyamodus rostratus may represent the most generalized cyamodontoid with respect to many features of its cranial anatomy (Figs. 12, 13). The skull of Cyamodus rostratus (umo BT 748) is in- completely preserved and was redescribed by Kuhn-Schnyder (1965a). Both temporal arches are missing, as is the right suspensorium. The basi- 24 FIELDIANA: GEOLOGY cranial length (tip of rostrum to basioccipital con- dyle) measures 106 mm. The longitudinal diam- eter of the left orbit measures 27.0 mm; the trans- verse (vertical) diameter of the same orbit mea- sures 25.0 mm. The longitudinal diameter of the left upper temporal fossa approximates 60 mm; its transverse diameter can be estimated to mea- sure 43 mm. By comparison with other cyamo- dontoids, the skull of Cyamodus rostratus shows a relatively high and narrow temporal region and deep orbits facing laterally. The premaxillaries form a broad, short rostrum. Each premaxilla bears two teeth, which are bul- bous and rounded but still form an anterior trans- verse crest reminiscent of chisel-shaped anterior premaxillary teeth seen in the sister-taxon of cy- amodontoids, Placodus (Rieppel, 1995a). Short posterior (nasal) processes of the premaxillae en- ter between the external nares and meet the nasals in a V-shaped suture at about the level of the mid- point of the longitudinal diameter of the external nares (Fig. 12 A). In ventral view, distinct yet slen- der posterior (vomerine) processes of the premax- illae meet the equally slender vomers at a level in front of the midpoint of the longitudinal diameter of the internal naris (Fig. 12B). The margin of the internal naris is fairly complete on the right side of the skull and shows that the premaxilla remains excluded from the anterior margin of the external naris by the vomer and maxilla (the suture be- tween the latter two elements is indistinct). In lat- eral view, the transverse process of the premaxilla reaches backward to a level slightly in front of the posterior margin of the external naris. It forms the anteroventral margin of the external naris and meets the maxilla in a V-shaped suture (the apex pointing backward). The nasals are broad, leaf-shaped structures (Fig. 12A). They define the posteromedial margin of the external naris and meet each other along the midline, thereby broadly separating the pre- maxilla from the frontal. Each nasal forms a short and tapering posterior process; together these em- brace an anteromedial process formed by the fron- tals. Posterolaterally, the nasal meets the prefron- tal, thus separating the frontal from the maxilla. The frontals are paired and elongated elements that posteriorly reach to a level behind the anterior margin of the upper temporal fossa (Fig. 12A). Anterolaterally, each frontal forms a distinct yet slender anterolateral process that enters between the prefrontal and the nasal. Between the prefron- tal and the postfrontal, the frontal broadly enters the dorsal margin of the orbit. The orbital margin of the frontal is distinctly concave. Although closely approaching the pineal foramen, the fron- tals remain excluded from it by the parietal. The posterior end of the frontal is rather broad and meets the parietal in a more or less transversely oriented, weakly interdigitating suture. Each fron- tal carries three distinct grooves on its posterior (postorbital) part that converge toward the pineal foramen. The parietal forms a flat skull table with con- cave lateral margins, owing to a posterior con- striction of the skull table. Distinct yet short an- terolateral processes of the parietal are embraced by the posterior ends of the frontal and of the postfrontal. The relatively large pineal foramen is located close to the anterior margin of the parietal (Fig. 12A). The medial suture, separating the orig- inally paired parietals, is still visible at the ante- rior margin of the pineal foramen. The laterally descending flange of the parietal is distinct. It contributes to the formation of a secondary lateral wall of the braincase as it meets the epipterygoid ventrally and the squamosal posteroventrally. The dermal encrustations on the parietal skull table re- semble those of Placochelys: an unpaired postero- medial one, and two protuberances along each side of the skull table. The prefrontal is small but relatively broadly exposed in dorsal view by comparison with Pla- cochelys. Located at the anterodorsal margin of the orbit, it contacts the frontal dorsally, the nasal anteriorly, and the maxilla ventrally. It closely ap- proaches but remains excluded from the lacrimal foramen, which is enclosed entirely by the max- illa. The postfrontal is a broad, plate-like and rough- ly triangular element broadly entering the poster- odorsal margin of the orbit (Fig. 12A). The pos- terolateral margin is deeply concave and angled. The posterior process tapers to a slender tip that lies alongside the parietal, extending backward to the same level as the frontal and separated from the anteromedial margin of the upper temporal fossa by a broad posterodorsal process of the post- orbital. The postorbital again is a relatively large ele- ment that broadly enters the posteroventral margin of the orbit (Fig. 1 2A). A relatively broad poster- odorsal process meets the parietal at the antero- medial margin of the upper temporal fossa at a level shortly in front of the posterior tip of the postfrontal, but well behind the level of the an- terior margin of the upper temporal fossa. In con- trast to Placochelys, the postorbital only margin- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 25 Fig. 12. Skull of Cyamodus rostratus Munster (holotype, umo BT 748): A, dorsal view; B, ventral view. Scale bar = 20 mm. For abbreviations, see p. 3. ally overlaps the laterally descending flange of the parietal and does not extend further backward on the descending process of the parietal than does its dorsally exposed part. Each maxilla carries two small, bulbous teeth, located at the level of the internal naris (Fig. 12B). Only the left internal naris is preserved, and its posterior margin appears smooth and natural. This indicates that the posterior maxillary tooth is lo- cated in front of the posterior margin of the in- ternal naris. The maxilla enters the anterolateral margin of the internal naris, but the precise lo- cation of its contact with the vomer along the an- terior margin of the internal naris cannot be as- sessed. The palatine meets the maxilla along the posterolateral margin of the internal naris at a lev- el of about the midpoint of the longitudinal di- ameter of the latter. Behind the internal naris, the maxilla can be followed in ventral view as a ta- pering shelf of bone that meets the jugal well in front of the anterior margin of the subtemporal fossa (Fig. 12B). In lateral view, the maxilla is seen to broadly enter the posteroventral margin of the external naris (Fig. 13 A). Between the external naris and the orbit, the maxilla forms a distinct yet small 26 FIELDIANA: GEOLOGY Fig. 12. Continued. and slender ascending process, wedged in be- tween the nasal and prefrontal. Further back, the maxilla broadly enters the ventral margin of the orbit, enclosing the lacrimal foramen at the antero- ventral corner of the orbit (Figs. 12A, 13 A). Pos- teriorly the maxilla meets the jugal in an essen- tially vertically oriented, sigmoidally curved, and interdigitating suture at a level somewhat behind the posterior margin of the orbit. Unlike in other cyamodontoids, the maxilla contacts the postor- bital at the posteroventral margin of the orbit (Fig. 13 A). The jugal is incompletely preserved because of breakage of the temporal arch. In lateral view, the jugal is seen to form a short, small, anterior pro- cess that enters between the postorbital and max- illa but remains excluded from the orbital margin by the contact of the latter two bones (Fig. 13A). In ventral view, the jugal is seen to define the anterior margin of the subtemporal fossa. The right side of the skull shows an anterior palatal process of the jugal that enters deeply between the maxilla and the palatine, reaching anteriorly to a level slightly in front of the anterior palatine tooth plate (Fig. 12B). This corresponds to the level of anterior extension of the postorbital along the pos- teroventral margin of the orbit. Preservation is not good enough to ascertain whether the jugal be- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 27 fc.st 28 FIELDIANA: GEOLOGY Table 3. Measurements of the palatine tooth plates of Cyamodus rostratus (holotype, umo BT 748). All measurements in mm; approximate values in parenthe- ses. right palatine longi- tudinal 0 trans- verse 0 first toothplate 8.7 8.6 second toothplate 9.5 (9.8) third toothplate 27.5 23.2 comes exposed in dorsal view in the floor of the orbit between the maxilla and the postorbital lat- erally and the palatine medially, or whether the palatine meets the maxilla and postorbital dorsal to this anterior process of the jugal. In lateral view, the jugal can be observed to form a distinct, tapering posteromedial process that extends along the anteromedial margin of the subtemporal fossa, where it enters between the palatine dorsally and the pterygoid ventrally (Fig. 13 A). A similar pos- teromedial process of the jugal has not been re- corded for other cyamodontoids. The ectoptery- goid would be expected to be located at the an- teromedial margin of the subtemporal fossa, but it is absent in Cyamodus rostratus. The palatal view of the skull is somewhat dif- ficult to interpret because of damage due to pres- ervation and preparation (Fig. 12B). The vomers are paired elements that separate the internal nares from one another, but other than that little can be said about the precise nature of their contacts to neighboring bones. The palatines are the domi- nant elements in the dermal palate, each carrying three tooth plates, of which the posteriormost plate is much larger than the two anterior ones (Table 3). The nature of the dentition in Cyamo- dus rostratus has generated considerable contro- versy because some of the teeth are obviously glued to the underlying bone surface at places where they may not have been originally located. This is particularly true of the anteriormost left palatine tooth. Kuhn-Schnyder (1965a) recapitu- lated the controversy and concluded that Cyamo- dus rostratus has two premaxillary, two maxil- lary, and three palatine teeth, the same tooth count proposed by H. v. Meyer (1863). I concur with this conclusion. The palatine bones themselves are badly broken, and the medial suture between them can be only partially identified. The left pos- terior palatine tooth plate is incompletely erupted. The posterior dental lamina foramina are distinct, located posterior to the palatine tooth plates on the palatine-pterygoid suture. The right pterygoid is incompletely preserved, but the left pterygoid indicates that the length of its palatal exposure is short relative to the length of the palatine. The distance from the posterior margin of the dermal palate to the posterior dental lamina foramen is slightly larger or equal to 10 mm; the distance from the (left) posterior dental lamina foramen to the posterior margin of the left internal naris approximates 58 mm. The ratio of pterygoid length to palatine length thus is approx- imately 0.17. The (left) pterygoid forms a distinct, longitu- dinally oriented ventral flange that unfortunately is broken along its ventral edge. It is therefore impossible to unequivocally ascertain whether there was a single or a double ventral projection on the pterygoid flange. However, the bipartition of the broken bone surface by a small interme- diate stretch of finished bone with a slightly con- cave surface strongly suggests a double ventral projection, as is also observed in specimen smns 17403 (see below). The left lateral view of the skull shows rather distinctly the anterior extent of the pterygoid along the medial margin of the sub- temporal fossa (Fig. 13 A). The pterygoid extends to a level slightly behind the anterior margin of the posterior palatine tooth plate and approaches the anterior margin of the subtemporal fossa more closely than in other cyamodontoids. The quadrate has a weakly concave posterior margin and is covered laterally by the quadrato- jugal. The suture separating the quadratojugal from the quadrate and the squamosal is distinct in occipital view; in lateral view, cracks and dermal encrustations obscure the dorsal delineation of the quadratojugal from the squamosal. The dominant element in the lateral wall of the braincase is the epipterygoid (Fig. 13 A). As in other cyamodontoids, the epipterygoid can be de- scribed as consisting of two parts, a posterior part Fig. 13. A, Skull of Cyamodus rostratus Miinster (holotype, umo BT 748) in left lateral view. B, Left side of the occiput of Cyamodus rostratus Miinster (holotype, umo BT 748). C, Right side of the occiput of Cyamodus rostratus Miinster (referred specimen, SMNS 17403). Scale bar = 20 mm. For abbreviations, see p. 3. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 29 with a convex lateral surface and an anterior part with a concave lateral surface, trending toward the midline of the skull. The ventral margin of the posterior part of the epipterygoid is formed by unfinished bone and overhangs the palatoquadrate cartilage recess. The ventral margin of the anterior part is sutured to the dorsal surface of the palatine. The dorsal margin of the epipterygoid meets the laterally descending flange of the parietal. Unlike Placochelys, Cyamodus rostratus shows no dis- tinct ventromedial process of the postorbital abut- ting the lateral surface of the anterior dorsal part of the epipterygoid. Instead, the ventral process of the parietal is exposed as a narrow strip of bone between the posteromedial process of the postor- bital (lining the anteromedial margin of the upper temporal fossa) and the dorsal margin of the epi- pterygoid. Between these two bones, the descend- ing process of the parietal can be followed ante- riorly up to the posterior margin of the foramen interorbitale (paired openings between the orbits filled, in life, by the cartilaginous interorbital sep- tum; Gaffney, 1972), such that the epipterygoid contacts the parietal rather than the postorbital at the posterodorsal corner of the foramen interor- bitale. The posterior margin of the epipterygoid is deeply concave and forms the anterior margin of the trigeminal incisure. The posterior margin of the trigeminal incisure is formed by the prootic. Behind and dorsal to the trigeminal incisure and above the prootic, the epipterygoid forms a dis- tinct, slender posterior process that meets the squamosal at about the midpoint of the dorsal margin of the posttemporal fossa (Fig. 13 A). The posterodorsal, posterior, and posteroventral margin of the posttemporal fossa is formed by the squamosal (Fig. 13B). The otic process of the squamosal is shorter in Cyamodus rostratus than in Placochelys; it meets the prootic at about the midpoint of the ventral margin of the posttem- poral fossa, corresponding to the midpoint of the anterior margin of the paroccipital foramen (Fig. 12A; not exposed in lateral view in Cyamodus rostratus). At the same time, the prootic and the otic process of the squamosal form the dorsal margin of the posterior part of the palatoquadrate cartilage recess. The ventral margin of the palatoquadrate car- tilage recess is formed by the palatine anteriorly and by the quadrate posteriorly. Other than in Pla- cochelys, the palatine does not contact the quad- rate lateral to the recess in Cyamodus rostratus (Fig. 13 A). This results is a greater width of the palatoquadrate cartilage recess, with the pterygoid forming its lateral margin between palatine and quadrate. The right side of the occiput is very poorly pre- served in the holotype of Cyamodus rostratus. On the left side (Fig. 13B), the posttemporal fossa is larger than in smns 17403 or in Cyamodus kuhn- schnyderi (see descriptions below). It is possible, however, that the lateroventral margin of the post- temporal fossa is incomplete (broken); parts of the squamosal may be missing, such that the pteroc- cipital foramen forms a deep cleft located in the lateroventral corner of the posttemporal fossa. On the other hand, breakage of the squamosal cannot be unequivocally ascertained, such that the con- figuration of the pteroccipital foramen in Cyamo- dus rostratus may represent the plesiomorphic condition relative to other cyamodontoids (see discussion of the derivation of the cyamodontoid braincase from that of Placodus, below). Other than at the lateroventral corner, the margins of the posttemporal fossa are complete, indicating a rel- atively large size. The holotype of Cyamodus rostratus shows the distal tip of the posterior dorsal process of the epipterygoid to become exposed at the dorsome- dial corner of the posttemporal fossa in occipital view (as in smns 17403, see below). In addition, the specimen shows the bilaterally symmetrical presence of an additional separate ossification in the occiput, a narrow strip of bone that extends from the medial corner of the posttemporal fossa toward the juncture of supraoccipital and exoccip- ital (Fig. 13B). The ventral suture, separating this element from the opisthotic, is distinct on both sides of the skull; the dorsal suture, separating this element from the supraoccipital, is distinct on the left side of the skull but appears partially fused on the right side. No comparable epiotic ossifi- cations are known in other cyamodontoid skulls. Kuhn-Schnyder (1965a) described the presence of postparietals and tabulars in the occiput of the holotype of Cyamodus rostratus; however, the supposed suture separating the (dorsal) parietal from the (ventral) postparietal corresponds to the margin of a dermal encrustation situated on the posterior margin of the parietal skull table. The sutures indicated by Kuhn-Schnyder (1965a) to be located between the postparietals and tabulars re- main enigmatic. The supraoccipital is a relatively broad plate carrying a low median crest and defining the dor- sal margin of the foramen magnum. The lateral margins of the foramen magnum are formed by 30 FIELDIANA: GEOLOGY the exoccipitals. The basioccipital, forming the occipital condyle, is badly eroded and difficult to separate from the exoccipitals. However, a dis- tinctly thickened rim runs along the ventral mar- gin of the foramen magnum that seems to be bi- partitioned at the midline of the skull. If indeed composed of the exoccipitals, this rim would in- dicate that the latter bones meet dorsal to the ba- sioccipital, as they do in Cyamodus kuhnschny- deri. The opisthotic forms the paroccipital process, which is severely damaged by a horizontal crack passing through it (Fig. 13B). This crack also ob- scures the vagus foramen. In ventral view, a me- dioventral extension of the squamosal appears su- tured to the anterior aspect of the paroccipital pro- cess (as in smns 17403, see below). The distal end of the paroccipital process expands into a distinct, posteroventrally directed tubercle, as is also ob- served in smns 17403, as well as in Cyamodus kuhnschnyderi (Nosotti & Pinna, 1996). There is, in Cyamodus rostratus, no distinct buttress on the squamosal to receive the distal end of the paroc- cipital process. Specimen smns 17403 is an incomplete skull that was used by Nosotti and Pinna (1993b) in support of their interpretation of the relations of the squamosal to the quadratojugal in the tempo- ral arch. The specimen is supposed to show a dis- tinctive suture separating the squamosal from the quadratojugal in the posterolateral corner of the upper temporal fossa. This suture starts laterally as a crack, below which the suture cannot be pur- sued on the lateral surface of the temporal arch. On the narrow dorsal surface of the temporal arch, the suture forms a "V" with the apex pointing backward. The lateral shank of that "V" crosses a dermal tubercle, but because these dermal tu- bercles secondarily fuse to the posterolateral as- pect of the temporal arch, one would expect the suture to be concealed by this tubercle rather than passing through it. The medial shank of the suture meets the margin of the temporal fossa, but from there the suture cannot be followed onto the me- dial surface of the temporal arch. It is for these reasons that I consider this supposed suture be- tween squamosal and quadratojugal not to be un- equivocally distinct from a break or an artifact of preparation. By contrast, there is on the lateral surface of the temporal arch, at about the level of the dorsal head of the quadrate and below a pos- terolaterally placed dermal tubercle, a horizontal groove that might likewise be interpreted as a su- ture between the (dorsal) squamosal and the (ven- Fig. 14. Cyamodus rostratus Miinster (referred spec- imen, smns 17403); double ventral projection of the pterygoid flange. Scale bar = 10 mm. tral) quadratojugal in a position comparable to that seen in Placochelys. The right lateral wall of the braincase again shows rather distinctly the posterior dorsal pro- cess of the epipterygoid, which meets the squa- mosal at the anterodorsal margin of the posttem- poral fossa, and from below which emerges the prootic. The ventral view of the skull shows the large posterior palatine tooth plates with the den- tal lamina foramina located posterior and postero- medial to the latter. The longitudinally oriented ventral flanges of the pterygoid are well preserved on both sides of the skull, and both show a dis- tinctly concave ventral margin. This results in a double ventral projection of the pterygoid flange, the anterior one located at the level of the poste- rior dental lamina foramina, the posterior one lo- cated at the posterolateral corners of the dermal palate (Fig. 14). The most important information that can be ob- tained from specimen smns 17403 relates to the well-preserved right side of the occiput (Fig. 13C). Because of an increased occipital exposure of the parietal, squamosal, and opisthotic, the posttemporal fossa appears reduced by compari- son with the holotype of Cyamodus rostratus and more closely resembles the occiput of Cyamodus kuhnschnyderi, which again shows reduced post- temporal fossae. Because the pteroccipital fora- men is also of conventional size and position (at the lower margin of the posttemporal fossa) in smns 17403, the question arises again whether the lateroventral corner of the posttemporal fossa of the holotype of Cyamodus rostratus was subject to damage. As in the holotype of Cyamodus ros- tratus, however, smns 17403 shows the posterior dorsal process of the epipterygoid gaining an oc- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 31 Fig. 15. 20 mm. Cyamodus rostratus Munster (referred specimen, umo BT 2172); lower jaw, occludal view. Scale bar cipital exposure at the dorsomedial corner of the posttemporal fossa (Fig. 13C). The right paroccipital process of specimen smns 17403 is intact. Its distal end is expanded into a distinct posteroventral tubercle, a character shared by the holotype of Cyamodus rostratus and by Cyamodus kuhnschnyderi (Nosotti & Pinna, 1996). Again, there is no buttress on the squa- mosal to receive the distal end of the paroccipital process. Instead, the distal end of the paroccipital process (opisthotic) meets the squamosal in a broad, interdigitating suture that trends ventrolat- erally from the lateroventral corner of the post- temporal fossa (Fig. 13C). A similar sutural re- lation of the distal end of the paroccipital process to the squamosal appears to be present in the ho- lotype of Cyamodus rostratus and in Cyamodus kuhnschnyderi (Nosotti & Pinna, 1996). Specimen umo BT 2172 (Drevermann, 1928, PI. 23, Fig. 2) is a lower jaw with both rami in articulation (Fig. 15) that may be referred to Cy- Table 4. Measurements of the dentary tooth plates of Cyamodus rostratus (referred specimen, umo BT 2172). All measurements in mm. left right longi- tudinal0 trans- verse 0 longi- tudinal 0 trans- verse 0 anterior dentary tooth 14.7 12.2 15.7 12.1 posterior dentary tooth 34.2 27.3 35.8 26.7 amodus rostratus. The bone surface is badly erod- ed, and the tips of the retroarticular processes as well as the apex of both coronoid processes are incomplete. Each mandible carries an anterior tooth of a generally bulbous shape but retaining a transversely oriented anterior crest. This anterior tooth is followed, after a short diastema, by two dentary tooth plates, of which the posterior one is distinctly larger (Table 4). The only unusual fea- ture of the specimen is that the elongated lower jaw symphysis is deeply excavated (hollow) if looked at in posterior view. Specimen smf R-4040 is a much better pre- served but smaller (Tjuvenile; Drevermann, 1928) left mandible referable to Cyamodus rostratus, which was described by Drevermann (1928, PI. 23, Figs. 3a-d) and Rieppel (1995a, Fig. 31). It shows all the cyamodontoid characteristics, such as the large coronoid process formed by the cor- onoid bone, which closely approaches the ventral margin of the lower jaw, and the posterior dentary tooth plate, which is partially located medial to the coronoid process. The suture between the sur- angular and angular is distinctive in this speci- men, perhaps because of its possible juvenile sta- tus (Rieppel, 1995a, Fig. 31). It shows that in the area behind the coronoid process, the lateral ex- posure of the surangular is larger (deeper) than that of the angular, in contrast to Placochelys (but note the difficulty in delineating the angular-sur- angular suture in the latter specimen). In smf R- 4040, part of the angular-surangular suture is ob- 32 FIELDIANA: GEOLOGY i • V M ■1 3 Fig. 16. Skull of Cyamodus muensteri (holotype, bsp AS VII 1210): A, dorsal view; B and C, ventral view. Scale bar = 20 mm. Table 5. Dentitional characters for the species of the genus Cyamodus. posterior palatine tooth-plate maxilla palatine long. 0 trans. 0 C. rostratus 2 3 1.19 C. muensteri 3 2 1.32 C. laticeps 3 2 1.38 C. tarnowinensis 3 2 1.41 C. kuhnschnyderi 2 2 1.16-1.29 scured by a laterally protruding boss, which may represent a dermal encrustation on the lower jaw (less distinct and perhaps abraded in umo BT 2172). smf R-4040 shows four dentary teeth in- creasing in size from front to back, of which the posteriormost one is again much larger than the three anterior ones. This may represent ontoge- netic tooth variation. Kuhn-Schnyder (1959) de- scribed the ontogenetic reduction of tooth posi- tions in Cyamodus hildegardis, which also affects dentary teeth. The diastema observed between the anterior and the two posterior dentary teeth in umo BT 2172 may thus have formed by reduction of one dentary tooth, the second from the front end of the mandible in smf R-4040. The Cranial Anatomy of Cyamodus muensteri (Agassiz, 1839) Agassiz figured and named (Placodus miin- steri) the holotype on plate 71, published as part of volume II of his Recherches sur les Poissons fossiles in 1839; the accompanying text was pub- lished in chapter VI, 2nd part, volume II in 1844 (Brown, 1890). The holotype of Cyamodus muen- steri is a small cyamodontoid skull (bsp AS VII 1210) that is badly preserved and poorly prepared (Fig. 16). The maximum length of the skull is 106.5 mm; the maximum width is 1 15 mm. Large parts of the skull are reconstructed and heavily painted. Agassiz (1833-1845) diagnosed the spec- imen mainly on the basis of its dentition, yet not- ed the unusually short rostrum and the wide tem- poral arches (both reconstructed in plaster, how- ever). Indeed, the entire rostrum was missing in the original skull. Attempting a reconstruction of the skull, Miinster rounded off its anterior end with RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 33 Fig. 17. Skull of Cyamodus muensteri (Agassiz; holotype of C. "laticeps" Owen, bmnh R 1644): A, dorsal view; B, ventral view. Scale bar = 20 mm. For abbreviations, see p. 3. plaster (as indicated by a dotted line in the figure published by Munster 1 830) and arranged a series of small teeth (three on either side; the first right "maxillary" tooth is now broken) along its mar- gin, as figured by Agassiz (1833-1845). The re- construction of the skull by Munster is certainly unnatural, as was already noted by Meyer (1863), and the only indication of what the specimen may originally have looked like is Munster's (1830) first illustration. This shows the two enlarged, posterior palatine tooth plates in situ. Dividing the longitudinal diameter of the better preserved right posterior palatine tooth (25.9 mm) by its trans- verse diameter (19.6 mm) yields a ratio of 1.32, which is larger than the equivalent ratio for Cy- amodus rostratus but very close to the values ob- tained for Cyamodus "laticeps" (see below). In front of the left posterior palatine tooth plate is the smaller, anterior palatine tooth plate, aligned with the left posteriormost maxillary tooth, which is located lateral to it. The right side of the skull shows three marginally positioned teeth increas- ing in size from front to back, as is also the case for the three maxillary teeth of Cyamodus "lati- ceps" (Owen, 1858). As noted by Meyer (1863, PI. 31, Fig. 2), the anteriormost one of these three marginal teeth (now broken) is located slightly more medially than the posterior ones, as is again the case for the anteriormost maxillary tooth of Cyamodus "laticeps." Cyamodus muensteri can therefore be reconstructed to share three maxillary teeth with Cyamodus "laticeps" (and Cyamodus tarnowitzensis Giirich, 1884). Dentitional charac- ters (Table 5) therefore suggest synonymy of Cy- amodus muensteri and Cyamodus "laticeps" (the first name takes priority; the holotype of Cyamo- 34 FIELDIANA: GEOLOGY B Fig. 17. Continued. dus tarnowitzensis Giirich, 1 884, can no longer be located today, and the schematic illustration of the poorly preserved specimen is not diagnostic at the species level). The Cranial Anatomy of Cyamodus "laticeps" (Owen, 1858) The holotype of Cyamodus "laticeps" (bmnh R 1644) is an incomplete skull from the Upper Mu- schelkalk (mol) of Bayreuth (Fig. 17). The right posterior lateral part, comprising the skull table and the temporal region, is missing. Owen's (1858, Pis. IX and X) illustrations represent a mir- ror image of the specimen. The skull is incom- pletely prepared, especially the left side of the braincase and the preserved left side of the occi- put. It also appears that parts of these regions have been reconstructed and smoothed over using a mixture of glue and ground bone substance, as was commonly done with fossils for sale from Bayreuth, but without X-ray analysis, the extent of reconstruction is impossible to discern. The maximal length of the skull, as preserved, is 186 mm; the maximal width is 154 mm. In size and general appearance, the skull close- ly resembles that of Cyamodus kuhnschnyderi (see below). The basicranial length can be recon- structed to approximate 150 mm in Cyamodus "laticeps," whereas the transverse diameter of the (left) upper temporal fossa may have approximat- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 35 ed 65 mm. Dividing the basicranial length by the transverse diameter of the upper temporal fossa yields a ratio of approximately 2.3 (i.e., very close to Cyamodus kuhnschnyderi) and distinctly small- er than the ratio in cyamodontoids outside the ge- nus Cyamodus, which all have a relatively nar- rower upper temporal fossa. Cyamodus "lati- ceps" differs from Cyamodus kuhnschnyderi in its dentition, however, as well as in the relation of the nasal and premaxillary. Because of damage done to the bone surface during early preparation, suture lines can be only partially identified on the skull of Cyamodus "laticeps," primarily in the preorbital region of the skull. In the snout, the maxilla appears to meet the premaxilla in a suture that enters the anterolateral corner of the external naris. If correctly identified, this suture forms an anteriorly wide-open V as it trends toward the lat- eral margin of the upper jaw. The maxillary-pre- maxillary suture lies in a distinctly more anterior position in Cyamodus "laticeps" by comparison with Cyamodus rostratus and Cyamodus kuhn- schnyderi (Nosotti & Pinna, 1996). The possibil- ity remains, however, that what appears to be the premaxilla-maxilla suture may, in fact, be a break. The external nares of Cyamodus "laticeps" are elongate and kidney-shaped. Their lateral (ven- tral) margin is formed almost exclusively by the maxilla. The nasal broadly enters the posterior and posteromedial (posterodorsal) margin of the external naris, whereas the premaxilla forms its anterior and anteromedial (anterodorsal) margin. Other than in Cyamodus kuhnschnyderi, the na- sals are paired in Cyamodus "laticeps," and they are separated from one another by elongate pos- terior (nasal) processes of the premaxillae, which meet the frontal at a level shortly in front of the anterior margin of the orbit (Fig. 17 A). The fron- tal forms distinct anterolateral lappets, which re- main separated from the maxilla by a broad con- tact of the nasal with the prefrontal. In Cyamodus kuhnschnyderi, the nasals are fused and meet the frontal in an almost straight transverse suture at a level shortly behind the level of the posterior mar- gin of the external naris. The prefrontal of Cyamodus "laticeps'" is dis- tinct on the right side of the skull, located at the anteromedial (anterodorsal) margin of the orbit and excluded from the lacrimal foramen, which is enclosed by the maxilla only. Closely comparable to Cyamodus kuhnschnyderi, a distinct "basior- bital furrow" (Nosotti & Pinna, 1996) lines the lateral (ventral) margin of the orbit, but within it only two foramina can be identified in Cyamodus "laticeps" (Fig. 17 A). The larger anterior fora- men represents the lacrimal foramen, whereas the smaller posterior one must represent the infraor- bital foramen (sensu Oelrich, 1956), transmitting the infraorbital nerve. Subsequent to preparation and preservation, the elements bordering the bas- iorbital furrow cannot be identified. Other than these, few sutural details can be gleaned from the dorsal view of the skull. The anterior tip of the left postfrontal can be identified at the posteromedial (posterodorsal) margin of the orbit, which by comparison with the right pre- frontal indicates that these two elements were sep- arated from one another by the frontal along the dorsal margin of the orbit. The frontal is unpaired (fused). The lateral part of the suture separating the maxilla from the jugal is distinct on the left side of the skull, but this suture cannot be traced to the lateral (ventral) margin of the orbit. The posterior tip of the maxilla can be seen to extend to a level somewhat behind the midpoint of the longitudinal diameter of the orbit but in front of the posterior margin of the latter. Extrapolating from the identifiable part of the maxillary-jugal suture, the latter must have lined at least the pos- terior part of the basiorbital furrow. There is a weak indication of the suture separating the post- orbital from the jugal at the posterolateral (pos- teroventral) corner of the orbit, but this could also represent a crack. The left temporal arch is dam- aged, as the dorsoventral compression of the skull resulted in the formation of a deep trough, but the unaltered dorsal (medial) part of the surface pre- serves the interdigitating suture separating the postorbital from the squamosal. Large tubercles have secondarily fused to the surface of the squa- mosal along the posterior and posterolateral mar- gin of the upper temporal fossa. The lateral braincase wall reveals very little structural detail. The most conspicuous feature is the trigeminal incisure, located between the epi- pterygoid (anteriorly) and the prootic (posterior- ly). The contours of these two elements, as well as the palatoquadrate cartilage recess and the pter- occipital foramen, remain indistinct, which raises the suspicion that at least part of the lateral brain- case wall has been reconstructed. The same is true for the preserved (left) part of the occiput. The ventral view of the skull reveals even few- er sutural details (Fig. 17B). Identifiable are the ventromedial suture line between palatines, vo- mers, and premaxillae, and the suture between the quadrate ramus of the (left) pterygoid and the 36 FIELDIANA: GEOLOGY Table 6. Measurements of the tooth plates of Cy- amodus muensteri (holotype of C. "laticeps," bmnh R 1644. All meaurements in mm. kmgi- tudmal0 trans- 1st pre maxillary 10.6 9 2nd pre maxillary 8 8.2 1st maxillary 11.3 10.4 2nd maxillary 13.2 11.5 3rd maxillary 17.2 17 1st palatine 21.5 17.8 2nd palatine 44.3 33 quadrate. The internal nares are relatively small, oval openings that appear to be located a little further posteriorly than in Cyamodus kuhnschny- deri. However, this could also be an impression created by differences of the palatal dentition. The posterior dental lamina foramen is located behind the posterior palatine tooth plate. The suture sep- arating the pterygoid from the palatine can be seen to extend from the lateral corner of the left posterior dental lamina foramen in an anterolat- eral direction to the medial margin of the subtem- poral fossa. There is no indication of the presence of an ectopterygoid. The sutural relations of the palatine and jugal at the anterior margin of the subtemporal fossa remain obscure. Cyamodus "laticeps" shows the well-preserved double ven- tral projections of the longitudinally oriented pter- ygoid flange that are characteristic for the genus (Fig. 17B). The palatal dentition is well preserved and cor- responds to the description given by Owen (1858). Cyamodus "laticeps" shows two premax- illary teeth (a replacement tooth is located medial to the second tooth on the right premaxilla), whereas the holotype and paratype of Cyamodus kuhnschnyderi show only a single premaxillary tooth. Two premaxillary teeth are preserved in a third, recently described specimen (Rieppel & Hagdorn, 1999), indicating variability of the pre- maxillary dentition in the latter species. However, all three specimens known of Cy- amodus kuhnschnyderi show two maxillary teeth only, whereas Cyamodus "laticeps" shares with Cyamodus muensteri three maxillary teeth (Fig. 17B). These increase in size from front to back (Table 6), and, as described by Owen (1858), the anteriormost maxillary tooth is located a little more medially than the succeeding two teeth. Fi- nally, Cyamodus "laticeps" has two palatine teeth, a feature it shares with Cyamodus muensteri and Cyamodus kuhnschnyderi. In all three species, the (smaller) anterior palatine tooth plates, togeth- er with the posterior maxillary teeth, are aligned in a transverse, anteriorly slightly concave row. Dividing the longitudinal diameter of the enlarged posterior palatine tooth plate by its transverse di- ameter results in a ratio of 1 .34, which is distinct- ly higher than the ratio in Cyamodus rostratus and somewhat higher than the ratio in Cyamodus kuhnschnyderi, but very close to the value ob- tained for Cyamodus muensteri. The Cranial Anatomy of Cyamodus kuhnschnyderi Nosotti and Pinna, 1993a Cyamodus kuhnschnyderi (smns 15855, 16270) has been the subject of a recent monographic de- scription (Nosotti & Pinna, 1996). The species shares with Cyamodus rostratus the relatively wide upper temporal fossae. Dividing the basicra- nial length (tip of snout to occipital condyle) by the transverse diameter of the upper temporal fos- sa yields a ratio of 2.3 in Cyamodus kuhnschny- deri (smns 15855, with parts of the left temporal arch preserved) and a ratio of approximately 2.45 for Cyamodus rostratus. All other cyamodontoids have relatively narrower upper temporal fenestrae, with a corresponding ratio ranging from 3.4 to 4.95. The posterior (nasal) processes of the premax- illae meet the nasal at the level of about the mid- point of the longitudinal diameter of the external naris. Unlike all other cyamodontoids included in this study, the nasals are fused in Cyamodus kuhn- schnyderi. In smns 16270, a partially fused suture is still apparent at the posterior margin of the na- sal, but the nasals are clearly fused in their ante- rior part. The nasal broadly contacts the prefron- tal, thus separating the frontal from the maxilla. The frontals are paired but may be partly fused at their posterior margin. They meet the (fused) pa- rietal in a deeply interdigitating suture but remain narrowly excluded from the pineal foramen, which lies in an anterior position within the skull table. As in Cyamodus rostratus, the parietal forms short anterolateral processes, which are em- braced by the frontal and the postfrontal. In ad- dition to these anterolateral processes, each pari- etal forms a more medially located anterior pro- cess, which is embraced by the frontal only. Sim- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 37 ilar processes are absent in Cyamodus rostratus. The posterolateral margin of the large postfrontal of Cyamodus kuhnschnyderi is deeply concave and angulated. The postfrontal closely approaches the anteromedial margin of the upper temporal fossa, separated from the latter by a narrow con- tact of the parietal with the postorbital. The lateral view of the skull is very incomplete in both specimens, and the identification of su- tures is rendered difficult by the extensive resto- ration of the specimens. The short jugal indicated in the reconstruction of the skull by Nosotti and Pinna (1996) is almost certainly incorrect and re- flects the incompleteness of the temporal arches in both specimens. The postorbital of Cyamodus kuhnschnyderi does not form a posteromedial pro- cess that overlaps the laterally descending flange of the parietal, as seen in Cyamodus rostratus. The epipterygoid contacts the parietal at the pos- terodorsal margin of the foramen interorbitale. Neither of the specimens is well enough preserved to allow the assessment of whether or not the epipterygoid forms a posterior dorsal process meeting the squamosal at the dorsal margin of the posttemporal fossa. As in Cyamodus rostratus, the palatine fails to meet the quadrate lateral to the palatoquadrate cartilage recess (smns 15855). Other aspects of the lateral wall of the braincase of Cyamodus kuhnschnyderi have been discussed in comparison with Placochelys in the descriptive section above, such as the incomplete ossification of the epipterygoid in both specimens and the possible course of the facial nerve through a fo- ramen at the ventral margin of the prootic, close behind the trigeminal incisure (smns 16270). In ventral view, the premaxillaries can be seen to remain excluded from the internal naris by a contact of the maxilla and vomer. There is no ev- idence for the presence of an ectopterygoid in Cy- amodus kuhnschnyderi. The palatal exposure of the pterygoid is short relative to the length of the palatine. Dividing the length of the pterygoid (from its posterior margin to the dental lamina foramen on the palatine-pterygoid suture) by the distance from the pterygoid-palatine suture to the posterior margin of the internal naris yields values ranging from 0.2 to 0.3 for Cyamodus kuhnschny- deri. The longitudinally oriented ventral flange of the pterygoid is prominent and bears two distinct ventral projections: the anterior at the level of the posterior dental lamina foramen, the posterior at the posterolateral corners of the dermal palate (smns 15855, right side of skull). Cyamodus kuhnschnyderi also differs from all other cyamodontoids with the exception of Mac- roplacus raeticus (Schubert-Klempnauer, 1975) and the specimen smns 17403 (referred to Cy- amodus rostratus above) by a significant reduc- tion in the size of the posttemporal fossa. This results from an expansion of the occipital expo- sure of the parietal, squamosal, and opisthotic. The exoccipitals meet dorsal to the occipital con- dyle in Cyamodus kuhnschnyderi. The right me- totic (jugular, vagus) foramen of smns 16270 pre- serves an internal subdivision by a vertical strut of bone that separates a smaller anterior passage (transmitting the glossopharyngeal nerve?) from a larger posterior division (transmitting the vagus complex?). The paroccipital processes are rela- tively well preserved in specimen smns 15855, which shows the anterior aspect of the distal end of the right paroccipital process to be broadly su- tured to the squamosal. The postero ventral aspect of the distal end of the paroccipital process is ex- panded into a distinct tubercle that projects into the space representing the posterior opening of the cranioquadrate passage. Cyamodus kuhnschnyderi shows a distinct "basiorbital furrow" (Nosotti & Pinna, 1996). This is a groove running along the ventrolateral margin of the orbit on the inside of the anterior process of the jugal and of the maxilla, bordered medially by the palatine. Along the ventrolateral margin of the orbit, three foramina can be iden- tified. The posteriormost one lies within the basi- orbital furrow, between the jugal and the palatine; the intermediate one lies at the anterior end of the basiorbital furrow, between the maxilla and the jugal; and the anterior one lies at the anteroventral corner of the orbit in front of the basiorbital fur- row, entirely within the maxilla. Nosotti and Pin- na (1996) identified the anterior foramen as the lacrimal foramen and the intermediate one as the infraorbital foramen (sensu Oelrich, 1956), trans- mitting the infraorbital nerve. The interpretation of these foramina proposed by Nosotti and Pinna (1996; see above) leaves the posteriormost fora- men unexplained. I concur with Nosotti and Pinna (1996) that the infraorbital nerve would have come to lie in the basiorbital furrow, but instead of passing across the posteriormost foramen lo- cated in that furrow, it most probably passed through it on its way to the superior alveolar ca- nal. Following this interpretation, the anterior two foramina in the anteroventral corner of the orbit would have transmitted the branches of an ante- riorly bifurcating lacrimal duct. Rather than the three foramina in Cyamodus kuhnschnyderi, there 38 FIELDIANA: GEOLOGY are only two foramina in the equally distinctly differentiated basiorbital furrow of Cyamodus "laticeps." This indicates some variation in the soft anatomy structures relating to the basiorbital furrow in Cyamodus, which renders the unequiv- ocal identification of the function of these foram- ina difficult if not impossible. The Cranial Anatomy of Cyamodus hildegardis Peyer, 1931a Although several skulls of Cyamodus hildegar- dis are available, its cranial anatomy remains very poorly known because of the severe dorsoventral compression of the material (Peyer, 1931a, 1935; Pinna, 1992). The dentition, and its ontogenetic variation, was analyzed by Kuhn-Schnyder (1959). Personal inspection of the available ma- terial of Cyamodus hildegardis did not result in the collection of new data. One of the better pre- served skulls (pimuz T4771) shows large dermal tubercles fused to the posterolateral side of the squamosal, similar to those observed in Cyamo- dus kuhnschnyderi. Doubts have been expressed that the three species of the genus Cyamodus could be treated as congeneric (Kuhn-Schnyder, 1960), and Nosotti and Pinna (1996) have treated the genus Cyamodus as paraphyletic. This prob- lem cannot be solved easily, because the cyamo- dontoid material from the Germanic Muschelkalk consists almost exclusively of skulls, whereas Cy- amodus hildegardis is best known from its post- cranial skeleton. If the three carapace fragments from the Muschelkalk have been correctly as- signed to Cyamodus kuhnschnyderi by Nosotti and Pinna (1996, Fig. 14), they would indicate that the dermal armor of the latter taxon is distinct from that of Cyamodus hildegardis. The enlarged posterior tooth plates of Cyamodus hildegardis are more distinctly elongated than those of the Cyamodus from the Germanic Muschelkalk and approach proportions otherwise typical for the Al- pine genus Psephoderma (Table 7; see also com- ments below). As incomplete as our current knowledge is of the cranial anatomy of Cyamodus hildegardis, the taxon can be coded for 35.2% of the 54 characters used in the cladistic analysis discussed below. The addition of Cyamodus hildegardis to the analysis did not result in a loss of resolution and corrob- orated, on the basis of cranial characters, the monophyly of the genus Cyamodus, including C. hildegardis. Table 7. Proportions of the posterior palatine tooth plate throughout the Cyamodontoidea. posterior palatine longitudinal 0 tooth-plate transverse 0 Cyamodus rostratus 1.19 Cyamodus kuhnschnyderi 1.16-1.29 Cyamodus laticeps 1.34 Cyamodus muensteri 1.32 Cyamodus hildegardis 1.33 - approx. 1.4 Protenodontosaurus italicus 1.24-1.26 Phcochelys placodonta approx. 1.23- 1.3 Macroplacus raeticus 1.41 Placocheylanus malanchinii 1.57 Psephoderma alpinum (juvenile) L27 Psephoderma alpinum (adult) 1.43 - 1.48 Placochelys alpis sordidae 1.44 Placocheylanus stoppanii 1.73 The Cranial Anatomy of Henodus chelyops v. Huene, 1936 Henodus chelyops is known from a total of eight specimens (comprising seven skulls), all of which come from the same deposit and locality, the upper Gipskeuper (Carnian) of Lustnau near Tubingen, southwestern Germany. Not all the skulls are currently accessible for investigation, because some are mounted on permanent exhibit, and preservation is not equal in all available skulls. The present description is based on the skulls of specimens I and II (syntypes, Huene, 1936), and the skulls of specimens IV and VI, which were collected in 1959 (Fischer, 1959) and have never been the subject of a detailed descrip- tion before (Stein, 1993). Specimen I (Fig. 18) is a skull from which the left mandible has been removed and the right mandible has been left in articulation. This is one of the best specimens for study of the preorbital region of the skull in dorsal view. Although the skull roof is poorly preserved, this specimen shows good detail in the dermal palate and is the best skull for study of the quad- rate suspension and the occiput. Specimen II (Fig. 19A) is the best skull for study of the skull roof and the temporal region behind the orbits. It also provides good detail on the structure of the dermal palate and some information on the occiput. Both RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 39 Fig. 18. Skull of Henodus chelyops v. Huene (syntype, gpit uncatalogued, specimen I of Huene, 1936): A, dorsal view; B, ventral view. Scale bar = 20 mm. lower jaws have been removed from the skull and preserve most information on the structure of the mandible. The skull of specimen IV (Fig. 19B) shows good detail of the structure of the dermal skull in the pre- and postorbital region. The oc- ciput and palate are rather poorly preserved, as are the mandibles. The skull of specimen VI is generally poorly preserved but provides excellent detail on the nature of the contact between frontal, postfrontal, and parietal. Henodus is a highly autapomorphic cyamodon- toid in every aspect of its anatomy (Fig. 20). Un- like in all other cyamodontoids, the rostrum of Henodus is extremely short yet broad, and the postorbital segment of the dermatocranium forms a broad, flat covering of the braincase. Both the structure of the dermal palate and its dentition dif- fer markedly from what is seen in other cyamo- dontoids. The skull of Henodus initially creates the false impression of being extremely de- pressed, but closer inspection shows that it is dis- tinctly curved relative to the long axis of the low- er jaw. The preorbital region is steeply inclined relative to the skull table, such that the external nostrils and the orbits face more or less anteriorly relative to the long axis of the lower jaw. The premaxillae and maxillae form a short yet broad, spatulate rostrum. Immediately behind the premaxillae the skull is strongly constricted, a trait that corresponds to the rostral constriction observed in other sauropterygians. The skull reaches its narrowest dimension just behind the orbits. As the elements of the skull roof and cheek region together form an essentially horizontal der- mal cover of the skull, the ancestral cheek emar- gination results in a weakly expressed concavity of the lateral margins of the skull table. The skull table reaches its greatest lateral extension just above the quadrate suspension. The posterior mar- gin of the skull table is deeply concave, the squa- mosals projecting backward far beyond the level of the occiput, as is also the case in other cyamo- dontoids. The paired premaxillae are the principal ele- ments forming the rostrum, and have a complex morphology (Fig. 21 A). Anteriorly the premaxil- lae form a transversely oriented, vertically de- scending flange that terminates in a ventral cutting 40 FIELDIANA: GEOLOGY Fig. 19. Skull of Henodus chelyops v. Huene. A, Syntype, gpit uncatalogued, specimen II of Huene (1936), dorsal view. B, gpit uncatalogued, specimen IV, skull with lower jaw, left lateral view. Scale bar = 20 mm. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 41 ppf /P° Fig. 20. Skull of Henodus chelyops v. Huene. A, Left lateral view, partial reconstruction based on specimen I of Huene (1936). B, Dorsal view, specimen I of Huene (1936). C, Ventral view, reconstruction based on specimens I and II of Huene (1936). Scale bar = 20 mm. For abbreviations, see p. 3. 42 FIELDIANA: GEOLOGY Fig. 20. Continued. edge. This cutting edge is reinforced on its ante- rior surface by a series of incompletely individu- alized denticles (Fig. 22), a strip of enamel lining the anterior surface of the cutting edge and mim- icking the presence of a row of minute teeth in the more or less regularly spaced indentations and the development of pulp cavities (Stein, 1993, 1995; Reif & Stein, 1999). In dorsal view the pre- maxilla is exposed as a rather narrow strip of bone that defines the entire anterior margin of the ex- ternal naris and meets the maxilla in the antero- lateral (anteroventral) margin of the external naris, as is also the case in other sauropterygians. Pos- teriorly the premaxillae carry ascending nasal pro- cesses that define the medial margins of the ex- ternal nares and project posteriorly up to the level of the anterior margin of the orbits, entering be- tween the anterior parts of the frontal s (Figs. 21 A, B). The nasal is a small, triangular element located at the posteromedial margin of the external naris, its apex pointing posteriorly. Together with the posterior (nasal) process of the premaxilla, the na- sal embraces a well-developed anterolateral pro- cess of the frontal. Posterolaterally, the nasal con- tacts the medial (ascending) process of the maxilla (Figs. 21 A, B). In dorsal view, the maxilla caps the premaxilla laterally as it forms the lateral margin of the broad, spatulate rostrum. Behind the premaxilla, RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 43 pm n Fig. 21. Skull of Henodus chelyops v. Huene. A, Dorsal view (specimen IV, gpit uncatalogued). B, Dorsal view (specimen I of Huene, 1936). Scale bar = 20 mm. For abbreviations, see p. 3. Fig. 22. Premaxillary denticles in Henodus chelyops v. Huene (specimen II of Huene, 1936). at a level between the external naris and the orbit, the maxilla shows a distinct lateral constriction (Figs. 20, 21). The ascending process of the max- illa is medially directed and forms the posterolat- eral (posteroventral) margin of the external naris before entering between the nasal and prefrontal. The nasal and prefrontal generally remain sepa- rated from one another by a contact of the as- cending process of the maxilla with the anterolat- eral process of the frontal, although this contact may be very narrow (specimen II, Fig. 20B). Pos- teriorly the maxilla is separated from the jugal by a suture that in dorsal view appears V-shaped (the apex pointing backward) and is located at about the level of the midpoint of the longitudinal di- ameter of the orbit (Fig. 20A). The maxilla there- fore defines the anterolateral (anteroventral) mar- gin of the orbit, where it forms a pronounced basi- orbital furrow (Figs. 20A, 21 A, B), otherwise re- corded for Cyamodus kuhnschnyderi and C. "laticeps" only among cyamodontoids (Nosotti & Pinna, 1996). In Henodus, two distinct foramina are located in the anteiror half of the basiorbital furrow, of which the anteiror one may correspond to the lacrimal foramen, whereas the posterior one may have served the passage of the maxillary 44 FIELDIANA: GEOLOGY branch of the trigeminal nerve. In this character, Henodus resembles Cyamodus "laticeps" more closely than Cyamodus kuhnschnyderi. A lacrimal is absent in Henodus. The prefrontal is a curved and rather slender element that forms most of the anterior margin of the orbit. It contacts the maxilla anteriorly and laterally (ventrally) and the frontal medially (dor- sally). It always remains separate from the post- frontal along the dorsal margin of the orbit, which is formed by the concave lateral margin of the frontal (Figs. 20B, 21 A). The frontals are paired elongate and rather slen- der elements. They reach their maximal width at the level of the anterior margin of the orbit. Pos- teriorly, each frontal forms a narrow and elongate posterolateral process, each of which is embraced by equally narrow and elongate anterior processes of the parietal. The details of the contact between frontals and parietal are well preserved in speci- mens II (Fig. 20B), IV (Fig. 21 A), and VI (Fig. 2 IB). The parietal is unpaired (fused). It forms an anteromedial process that enters between the pos- terolateral processes of the frontal. More laterally, the parietal forms a more elongate and equally slender anterior process that enters between the posterolateral process of the frontal and the post- frontal and may extend anteriorly up to a level close to the posterior margin of the orbit. The postfrontal broadly enters the posterome- dial (posterodorsal) margin of the orbit. Shortly behind the orbit the postfrontal is distinctly con- stricted, which results in an angulation of its lat- eral margin (Fig. 21 A). The posterior process of the postfrontal, which extends between the an- terolateral process of the parietal and the postor- bital, appears broad at its posterior end as it meets the parietal in a deeply interdigitating suture in specimen IV By contrast, the postfrontal tapers to a blunt tip posteriorly in specimens I and II. The postorbital is a large, broad, platelike ele- ment that defines the posterolateral (posteroven- tral) margin of the orbit and partially invades the space of the upper temporal fenestra (Fig. 20B). The element covers most of the postorbital area of the skull in front of the parietal and is bifur- cated posteriorly as it embraces the tapering an- terior end of the squamosal. The posteromedial process of the postorbital extends along the post- frontal and meets the parietal in a deeply inter- digitating suture. This posteromedial process may (specimen II, Fig. 20B) or may not (specimen IV, Fig. 21 A) exceed the postfrontal in length; it cor- responds to that part of the postorbital that in the plesiomorphic condition lines the anteromedial margin of the upper temporal fossa. The postero- lateral process of the postorbital may (specimen IV) or may not (specimen II) exceed the postero- medial process in length, and in the plesiomorphic condition contributes to the formation of the tem- poral arch as it lines the anterolateral margin of the upper temporal fossa. The jugal enters the lateral (ventral) margin of the orbit between the maxilla anteriorly and the postorbital posteriorly. It forms the anterior part of the lateral margin of the (horizontally exposed) cheek before it starts to taper off posteriorly. This posterior process of the jugal may (specimen II, Fig. 20B) or may not (specimen IV, Fig. 21 A, left side of skull) reach as far back as the posterior tip of the postorbital. Specimen IV, however, shows that this character is subject to variation between the left and the right side of the skull. None of the specimens shows a clear demar- cation of the squamosal from the quadratojugal. In front of the dorsal head of the quadrate, three anterior prongs can be identified within the (hor- izontally exposed) cheek region of the skull. The two medial prongs are here interpreted as repre- senting the squamosal, embracing the posterolat- eral process of the postorbital. The squamosal al- ways remains shorter than the third, lateral prong, which forms the posterior part of the lateral mar- gin of the (horizontally exposed) cheek and which is interpreted as the quadratojugal (by analogy to other cyamodontoids, Fig. 20B). The quadratoju- gal reaches anteriorly to a level close to the pos- terior margin of the orbit (specimen IV, Fig. 21 A). Large osteoderms have secondarily fused to the surface of the squamosal and quadratojugal along the posterolateral and lateral margins of the skull table. The upper temporal fossa is obliterated in Hen- odus (at least in the adult). Closure of the upper temporal fossa is mainly due to lateral spreading of the parietal ossification in a fan-shaped manner, best seen in specimens II (Fig. 20B) and IV (Fig. 21 A). The only specimen that retains a vestigial upper temporal opening with a smooth, natural margin is specimen II (left side of the skull). The hole in the left side of the skull roof of specimen I (Fig. 18) is certainly an artifact of preservation. At this juncture, a brief comment on the skull of specimen III appears to be in order, because it was claimed by Huene (1938) to show well-developed temporal fossae on both sides of the skull. Al- though difficult to access because it is mounted on permanent exhibit, the skull of specimen III RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 45 allows some observations. The skull is not smaller than that of any of the other specimens. The so- called temporal fenestrae have irregular contours, and these differ on the two sides of the skull, sug- gesting that these temporal openings are the result of breakage, as is unquestionably the case with the opening in the left side of the skull roof in specimen I. In fact, the corresponding area of the skull roof appears relatively weak and prone to breakage in all Henodus skulls. The area of the conjectural temporal fossa is strongly depressed on the right side of the skull in specimen II, and it has been severely damaged in specimens IV and VI. Weakness of this area of the skull roof may result from a lateral thinning of the spreading of the parietal ossification responsible for the oblit- eration of the plesiomorphic temporal fenestra. Most of the skulls have sustained a longitudinal break, obscuring the pineal foramen and making it difficult to assess whether the parietal is paired or fused. Specimen IV indicates, however, that the parietal is fused at least behind the pineal fora- men. Specimen II shows the elongate and narrow pineal foramen to be located at the anterior end of the parietal (Fig. 20B). The frontal remains ex- cluded from the pineal foramen. It remains un- clear, however, whether the parietal is fused or whether it shows a midline suture in front of the pineal foramen. The structure of the dermal palate of Henodus is again highly autapomorphic (Fig. 20C). The premaxilla is the dominant element in the broad rostrum in ventral view. The cutting edge of the premaxilla overbites the anterior cutting edge of the dentary. In the lateral part of its ventral sur- face, the premaxilla shows a distinct, transversely oriented trough which receives the lateral part of the cutting edge of the dentary. The broad pre- maxilla appears to enter the anterior margin of the internal naris. None of the specimens shows com- plete sutural contact between the premaxilla and the vomer. A short stretch of an obliquely oriented suture is seen on the left side of the palate of specimen II, located between the internal nares and indicating that the vomer enters the medial margin of the latter. The maxilla reaches its maximum width in its anterior part, where it forms a lateral projection of the rostrum and enters the lateral margin of the internal naris. The ventral surface of the maxilla carries the highly characteristic, sigmoidally curved groove (Figs. 20C, 23), which was be- lieved by Huene (1936) to have carried keratinous structures similar to baleen. Huene (1936) even Fig. 23. Right side of dermal palate in Henodus chelyops v. Huene (specimen I of Huene, 1936). claimed to have prepared such structures from the maxilla of one specimen, but this piece can no longer be located today. None of the specimens shows in ventral view the sutural separation between the jugal and the maxilla, which in dorsal view is located at the level of the midpoint of the longitudinal diameter of the orbit. The suture between the maxilla and the palatine is distinct, however, running from the posterolateral corner of the internal naris posteri- orly and slightly laterally and terminating at about the posterior two-thirds of the maxillary groove (Fig. 20C). In none of the specimens is it possible to assess whether the posterior third of this groove is still formed by the slender maxilla extending backward between the jugal and palatine or whether this groove continues behind the maxilla either on the jugal or on the palatine. There is no evidence for the presence of an ec- topterygoid in Henodus. The palatine is an elongate and rather slender element that is broadest in its anterior part, where it forms the posterior margin of the internal naris. A single small palatine tooth plate is located on 46 FIELDIANA: GEOLOGY Fig. 24. Suspension of left quadrate in Henodus chelyops v. Huene (specimen I of Huene, 1936). Scale bar = 20 mm. the posterior part of the palatine, at the back end of the maxillary groove (Fig. 20C). The dental lamina foramen is located posterior and postero- medial to the palatine tooth plate. The paired vomers are broad elements that form the medial margins of the internal nares, from where they extend backward to meet the elongated pterygoids. The vomers separate the an- terior parts of the palatines from one another. Un- like in any other cyamodontoid, the posterior parts of the palatines are separated by the paired ptery- goids (Figs. 20C). At the level behind the palatine tooth plate and its dental lamina foramen, the pterygoid extends laterally to cover the entire width of the dermal palate. The pterygoid also forms a weakly ex- pressed, longitudinally oriented flange posterior to the palatine tooth plate with a single ventral pro- jection. The short quadrate ramus of the pterygoid emerges from above this ventral flange and ex- tends posterolaterally to contact the quadrate. In contrast to other cyamodontoids, the pterygoids broadly extend anteriorly between the palatines and meet the vomers at approximately the mid- point of the length of the palatines (Fig. 20C). The quadrate and its suspension are again of a peculiar structure in Henodus, best observed in specimen I (Figs. 20A, 24). The mandibular con- dyle is transversely expanded and bipartite, to match the saddle-shaped surface of the mandibu- lar articulation. The medial articular facet on the mandibular condyle is distinctly larger than the lateral facet. The lateral surface of the shaft of the quadrate is covered by a descending process of dermal bone, presumably part of the quadratoju- gal. Below the skull table, the quadrate expands into a large, elongated, posterior (suprastapedial) process, accentuating the concavity of its poste- rior margin. The posterior tip of that dorsal ex- pansion of the quadrate abuts a distinct flange de- scending from the ventral surface of the squa- mosal. Between the dorsal expansion of the quad- rate, the skull roof, and the descending flange of the squamosal there persists, in all specimens of Henodus, a gap that leads into the temporal va- cuity of the skull. None of the specimens of Henodus shows a well-preserved occiput, but some information can be obtained from specimens 1 and II. The post- temporal fossae are large (i.e., not reduced as in Cyamodus kuhnschnyderi or Macroplacus). The exoccipitals meet dorsal to the basioccipital in the occipital condyle. The sutural contact of the ex- occipital to the supraoccipital, to the basioccipital lateral to the occipital condyle, and to the opis- thotic remains indistinct. As in other cyamodon- toids, the opisthotic forms a distinct ventral flange, which together with the basioccipital tuber enclosed the passage of the internal carotid. The ventral opisthotic flange remains separate from the basioccipital tuber as well as from the basi- cranium. There is evidence for the expansion of the distal end of the paroccipital process into a ventrally directed tuber (now broken), comparable to the same character observed in Cyamodus kuhnschnyderi. The distal end of the paroccipital process abuts the medial surface of the flange descending from the lower surface of the squamosal. On the left side of the skull of specimen II, this ventral pro- cess of the squamosal expands into a ventral flange that meets a dorsal flange originating from the quadrate ramus of the pterygoid. The squa- mosal and pterygoid thus define the closed lateral margin of the posterior opening of the cranio- quadrate passage, but at the same time conceal the point of articulation of the stapes with the quad- rate in the posterior or ventral view of the skull. The lateral braincase wall is rather poorly pre- served in all specimens available. As in other cy- amodontoids, the most prominent element is the broad epipterygoid. Details of its relation to neighboring elements cannot be ascertained. A pteroccipital foramen can be identified in speci- men I, but the sutures between the surrounding elements again remain indistinct. A palatoquad- rate cartilage recess is present but still filled with matrix on the left side of the skull of specimen I. It is bordered ventrally by the palatine and pter- ygoid and posteriorly by the quadrate. It remains RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 47 ang sang ang sang Fig. 25. Lower jaw of Henodus chelyops v. Huene (partially reconstructed, based mainly on specimens I and II of Huene, 1936). A, Lateral view; B, medial view; C, dorsal view. Scale bar = 20 mm. For abbreviations, see p. 3. unclear whether the palatine contacts the quadrate lateral to the palatoquadrate recess. The ventral margin of the palatoquadrate cartilage recess is complete, however, and does not show the gap that exists in these cyamodontoids where the pal- atine remains separate from the quadrate. Specimens II and III show well-developed and well-ossified hyoid elements overlying the poste- rior part of the dermal palate. The lower jaw of Henodus is unusually deep and massive (Fig. 25). The anterior end of the dentary is sharply turned inward to form a trans- versely oriented cutting edge. No toothlike struc- tures have ever been identified on the dentary. The mandibular symphysis is narrow and delicate, presumably a secondary development correlated with the development of a spatulate snout. Behind the anterior, transversely oriented cut- ting edge the dentary carries a sigmoidally curved groove, the counterpart of the maxillary groove (Fig. 25C). At the posterior end of the dentary, behind that groove, is located the single and rel- atively small dentary tooth plate, opposing the palatine tooth plate with an occlusal surface that tilts somewhat medially. The coronoid process is less well developed in Henodus compared with other cyamodontoids, which reflects a lesser degree of durophagy in this genus. As in other cyamodontoids, the coronoid process is formed by the coronoid only, which also defines the anterior margin of the adductor fossa. Ventrally, the coronoid expands across the lateral surface of the mandible in a fan-shaped manner, but not to the same degree as is observed in other cyamodontoids, so that it remains rather broadly separated from the ventral edge of the lower jaw (Fig. 25 A). The right mandible of spec- imen II shows a relatively broad contact between the dentary and the coronoid dorsally and be- tween the splenial and angular ventrally. Both these latter elements gain a broad exposure on the lateral surface of the lower jaw. The lateral sur- face of the mandible furthermore shows a char- acteristic relief in that the surangular and angular form a distinct vertical step at the level of the anterior end of the adductor fossa. Superficial jaw adductor muscle fibers must have glided across the dorsal margin of that step (formed by the sur- angular) as they expanded to insert into the lateral surface of the coronoid, surangular, and dentary. The ventral margin of the mandible, composed of 48 FIELDIANA: GEOLOGY splenial and angular, is distinctly sculpted and un- even in all specimens of Henodus. The adductor fossa is a deep yet anteroposte- riorly relatively short trough located medial to and behind the coronoid process. Anteriorly, the ad- ductor fossa extends to a level well in front of the posterior margin of the dentary tooth plate. The articular facet of the mandibular joint clos- es the adductor fossa posteriorly. The articular surface itself is strongly saddle-shaped and bicon- cave to accommodate the mandibular condyle of the quadrate. The chorda tympani foramen is dis- tinct and located on the dorsal surface of the re- troarticular process just behind the articular facet (Fig. 25C). The retroarticular process is very prominent. It is deep, its dorsal surface slants posteroventrally, and it is distinctly sculpted to facilitate the attach- ment of muscle fibers or tendons. The medial surface of the lower jaw (Fig. 25B) is covered by the prearticular, angular, splenial, and dentary. The prearticular and angular meet the dentary and splenial in an almost vertically ori- ented suture at the level of the coronoid process. The dentary and splenial close Meckel's canal in medial view. The canal opens anteriorly just be- hind the medial bent of the anterior end of the dentary. The Cranial Anatomy of Macroplacus raeticus Schubert-Klempnauer, 1975 Macroplacus raeticus is based on an incom- plete skull (bsp 1967 I 324; Fig. 26) from the Ba- varian Alps (Rhaetian Koessen-Formation, Hin- terstein bei Hindelang im Allgau: Schubert- Klempnauer, 1975). The skull is relatively large (maximal length as preserved: 181 mm; maximal width: 180.6), with a broad and relatively high temporal region. The rostrum is missing, with an oblique anterior break passing through the left ex- ternal naris and just in front of the right external naris. The skull was subject to considerable ero- sion. It was prepared with acid and subsequently heavily coated with varnish. This, together with a tendency toward fusion of the bones, renders the unequivocal identification of sutures difficult if not impossible in some areas of the skull. The most prominent feature of the skull is the enor- mous posterior palatine tooth plates. Their dis- tinctly elongate shape may have been the reason why Pinna (1978) considered Macroplacus a ju- nior synonym of Psephoderma {Psephoderma raeticus). Unfortunately, the most diagnostic feature of the skull, the rostrum, is incomplete in Macropla- cus. In Psephoderma, the rostrum carries paired grooves on its lower surface that lead up to the internal nares, and the maxilla carries a distinct anterior process entering the rostrum in ventral view along the lateral margin of this ventral groove. In Macroplacus, the ventral surface of the preserved proximal part of the rostrum is deeply concave, forming a single longitudinal groove or trough (Fig. 26C). Likewise, the maxilla does not extend as far anteriorly along the lateroventral margin of the rostrum as it does in Psephoderma. Finally, the proportions and shape of the upper temporal fossae of Macroplacus are distinctly dif- ferent from those of Psephoderma alpinum, which shows elongated but relatively narrow temporal fossae (see Table 8). The dorsal view of the skull of Macroplacus (Fig. 26B) shows greatly enlarged posterior (na- sal) processes of the premaxillae, that define the entire dorsal (medial) margin of the external nares and extend backward to the level of the anterior margin of the orbits, where they meet the frontals in an interdigitating suture. The nasals are rela- tively small, triangular elements that define the posterior margin of the external nares, and remain separated from one another by the broad posterior (nasal) processes of the premaxillae. The nasals narrow toward their posterior ends as they meet the prefrontals, thus separating the frontal from the maxilla. The posterior tip of the nasal lies at the same level as the posterior tip of the nasal process of the premaxilla. The two bones embrace a relatively small yet distinct anterolateral process of the frontal. The maxilla forms most of the ventral and pos- teroventral margin of the external naris (Fig. 26A). Unlike in Placochelys and Cyamodus, the maxilla of Macroplacus does not expand medially to floor the external naris. Between the external naris and the orbit, the maxilla forms a distinct and pointed ascending process that is embraced by the nasal (anteriorly) and the prefrontal (pos- teriorly). Further back, the maxilla enters the an- teroventral margin of the orbit. The lacrimal fo- ramen is located in the anteroventral corner of the orbit and is fully enclosed by the maxilla (Fig. 26B). Posteriorly, the maxilla meets the jugal at the level of the midpoint of the longitudinal di- ameter of the orbit. Immediately below the ventral margin of the orbit, the suture separating the max- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 49 Fig. 26. Skull of Macroplacus raeticus Schubert-Klempnauer (holotype, bsp 1967 I 324): A, left lateral view, B, dorsal view; C, ventral view. Scale bar = 20 mm. For abbreviations, see p. 3. 50 FIELDIANA: GEOLOGY Fig. 26. Continued. Table 8. Skull proportions of cyamodontoid placodonts. Measurements based on the holotype except ***), which is msnm V471. Abbreviations: basicranial, distance from tip of snout to occipital condyle; long., longitudinal; temp.f., upper temporal fossa; trans., transverse; *) approximate values as reconstruction of the skull or severe preservational distortion of the skull is involved; **), measurements based on the right side of the skull. basicranial long. 0 temp.f. basicranial long. 0 temp.f. long. 0 orbit trans. 0 temp.f. Cyamodus rostratus* 1.76 2.46 2.22 Cyamodus kuhnschnyderi* 1.55 2J 2.5 Cyamodus ktticeps* 1.97 2.3 2.46 Cyamodus hildegardis* 2.2 4 2 Macroplacus 1.72 Placochetys placodonta 2.28 3.62 1.94 Protenodontosaurus italicus** 2.07 3.38 1.65 Psephoderma alpinum*** 2.46 4.95 1J9 RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 51 ilia from the jugal is V-shaped, the apex pointing backward. More ventrally, the maxillary-jugal su- ture becomes vertically oriented and interdigitat- ing. It curves around the ventral margin of the skull shortly behind the posterior maxillary tooth (Fig. 26A). As in all other cyamodontoids, the prefrontal is located at the anterodorsal margin of the orbit and displays limited dorsal exposure. It descends far down along the anterior margin of the orbit, close- ly approaching but not quite reaching the lacrimal foramen. The delineation of the postfrontal and postor- bital is difficult. The figures of the specimen (Figs. 26 A, B) therefore entail an element of re- construction. On the right side of the skull, a dis- tinct, slightly interdigitating suture can be fol- lowed from the posterodorsal (posteromedial) margin of the orbit in a posterolateral direction at first, before the suture abruptly turns in a postero- medial direction, continuing its course up to the level of the anterior margin of the upper temporal fossa, where the suture turns laterally and be- comes more deeply interdigitating. By compari- son with other cyamodontoids, this suture must demarcate the medial and posterior margin of the right postfrontal, the latter establishing a medial contact with the frontal and a posterior contact with the parietal. A distinct and interdigitating suture running into the posteroventral corner of the left orbit must delineate the anteroventral tip of the left postorbital. From the orbit this suture trends in a posterodorsal direction as it extends into the tem- poral arch. As it approaches the upper margin of the temporal arch, it forms a distinct vertical step, but then continues horizontally below the dorsal margin of the temporal arch before entering the margin of the temporal fossa at a level well be- hind the midpoint of the longitudinal diameter of the upper temporal fossa. This suture delineates the posterior lateral process of the postorbital, which in Macroplacus reaches far back within the temporal arch, as it also does in Placochelys and Psephoderma. What remains unclear on both sides of the skull is the suture separating the postorbital from the postfrontal within the postorbital arch. Along the posterior margin of the left orbit there is a rather distinct step that might indicate the anteroventral tip of the postfrontal. Beyond that, however, it re- mains unclear whether the posterolateral margin of the postfrontal was smoothly curved or deeply convex and angulated, and whether the postfrontal was broadly or narrowly separated from the an- teromedial margin of the upper temporal fossa by a contact of the postorbital with the parietal (the suture in the drawing represents a reconstruction without observational basis). The jugal meets the maxilla at the level of the midpoint of the longitudinal diameter of the orbit. It therefore defines the posterior half of the ventral margin of the orbit. In Macroplacus, the jugal does not extend anteriorly beyond the level of the midpoint of the longitudinal diameter of the orbit as it does in Placochelys, where the jugal closely approaches the lacrimal foramen. Whether the ju- gal forms a distinct ventral expansion behind the maxilla as in other cyamodontoids cannot be es- tablished because of breakage. The jugal extends into the temporal arch to a level in front of the midpoint of the longitudinal diameter of the orbit, distinctly less far posteriorly than the postorbital. The posterior end of the jugal tapers to a blunt tip; this tip and the postorbital together embrace the anterior end of the squamosal (Fig. 26 A). Along the ventral margin of the temporal arch, the jugal meets the quadratojugal at a level well be- hind the anterior margin of the upper temporal fossa, that is, further back than in Placochelys. Macroplacus is an important specimen for re- constructing the relations of the quadratojugal and squamosal within the temporal arch. The V- shaped suture (apex pointing anteromedially) that separates the squamosal from the parietal along the posteromedial margin of the upper temporal fossa is distinct on the left side of the skull. No suture can be observed on either side of the skull that would separate the squamosal from the quad- ratojugal at the posterolateral margin of the tem- poral fossa, as would be required on the basis of Nosotti and Pinna's (1993b) reconstruction of the temporal region of cyamodontoids (see also Pin- na, 1989). Instead, the left side of the skull shows a horizontal line or groove that extends backward from the posterior end of the jugal and appears to represent the suture separating the dorsal squa- mosal from the ventral quadratojugal within the temporal arch (Fig. 26A). More posteriorly, this suture disappears below dermal encrustations on the posterolateral aspect of the temporal arch. The frontals are paired, although the suture be- tween the two elements may be partially ob- scured. Between prefrontal and postfrontal, the frontal forms the concave dorsal margin of the orbit. Unlike in any other cyamodontoids, the frontal widens conspicuously behind the orbit be- cause of the concave medial margin of the post- 52 FIELDIANA: GEOLOGY frontal (Fig. 26B). The frontoparietal suture re- mains obscure, which makes it impossible to es- tablish whether or not the frontal enters the pineal foramen. The large pineal foramen is located at the level of the anterior margin of the temporal fossa. A longitudinal crack runs through the parietal skull table close to its midline, but fusion of the pari- etals is indicated by the absence of any trace of a suture either at the anterior or at the posterior mar- gin of the pineal foramen. The parietal skull table is slightly constricted at its posterior end. Dermal encrustations are present, but ill-defined and weakly expressed on the skull table. The ventral view of the skull (Fig. 26C) dis- plays the dentition of Macroplacus, which com- prises two maxillary teeth and two palatine tooth plates. The right maxilla is well delineated in ven- tral view. Although shorter than in Placochelys, Protenodontosaurus, and, especially Psephoder- ma, the maxilla carries an anterolateral process that tapers off along the lateral margin of the ros- trum— unlike in Cyamodus, where the maxillary- premaxillary suture is transversely oriented (C. kuhnschnyderi) or even trends in a posterolateral direction (C rostratus). Unfortunately, the ante- rior delineation of the vomers is difficult, which makes it impossible to unequivocally assess whether the premaxilla enters the internal naris in Macroplacus. The suture separating the maxilla from the palatine enters the lateral margin of the internal naris at about the level of the midpoint of its longitudinal diameter. From there the suture curves around the anterior palatine tooth plate and then trends posterolaterally toward the pointed posterior tip of the maxilla, which remains ex- cluded from the anterior margin of the subtem- poral fossa by a contact of the palatine with the jugal. The maxilla carries two tooth plates, the posterior one somewhat larger than the anterior one (Table 9). A nutritive foramen is located me- dial to the anterior maxillary tooth entirely within the maxilla. The nutritive (dental lamina) foramen of the posterior maxillary tooth plate is located behind the tooth on the maxilla-palatine suture. At the level of the anterior margin of the sub- temporal fossa, the palatine forms a lateral pro- cess that meets a medial process of the jugal (Fig. 26C). The jugal reaches to the anteromedial cor- ner of the subtemporal fossa but does not extend backward along the medial margin of the latter as it does in Cyamodus rostratus. The palatines are paired, much enlarged ele- ments, each carrying the hypertrophied posterior Table 9. Measurements of the maxillary and pala- tine tooth plates of Macroplacus raeticus (holotype, bsp 1967 I 324). All measurements in mm. left right tudinal0 trans- verse 0 longi- tudinal 0 trans- verse 0 anterior maxillary tooth - ■ 15.0 15.6 posterior maxillary tooth - - 22.6 16.9 anterior palatine tooth 217 19.8 21.0 20.7 posterior palatine tooth 68.5 483 - palatine tooth plate along with a smaller, anterior palatine tooth (Table 9, Fig. 26C). The available specimens of Psephoderma indicate a relative size increase in the posterior palatine tooth plates dur- ing ontogeny. Given the relatively large size of the skull of Macroplacus, the hypertrophy of the posterior palatine tooth plates may be the result of their positive allometric growth. The right pos- terior palatine tooth plate is missing. A posterior dental lamina foramen cannot be identified behind or within its wide tooth socket. The left posterior dental lamina foramen is represented by a narrow groove, located on the palatine-pterygoid suture posteromedial to the posterior palatine tooth plate. Because of the large size of the palatine, the pterygoid gains a narrow palatal exposure only. Dividing the distance from the posterior margin of the (left) pterygoid to the left dental lamina foramen (13 mm) by the distance from the left dental lamina foramen to the posterior margin of the internal nares (76.5 mm) yields a quotient of 1.7. The pterygoid forms a distinct longitudinally oriented ventral flange, but erosion on both sides of the skull renders it impossible to assess wheth- er they formed a single or double ventral projec- tions. Along the medial margin of the subtemporal fossa, the pterygoid extends anteriorly to the level of the posterior third of the longitudinal diameter of the posterior palatine tooth plate. There is no evidence for the presence of an ectopterygoid in Macroplacus. The quadrate ramus of the ptery- goid is short and stout, and meets the quadrate in an interdigitating suture. The articular surface on the mandibular condyle of the quadrate is bicon- cave, matching the saddle-shaped surface of the mandibular articulation. The lateral wall of the braincase is poorly pre- served on both sides of the skull. A number of interesting observations can nevertheless be re- corded. The palatoquadrate recess is distinct, but RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 53 Fig. 27. Skull of Protenodontosaurus italicus Pinna (holotype, mfsn 1819GP): A, right lateral view; B, dorsal view; C, ventral view. Scale bar = 20 mm. For abbreviations, see p. 3. 54 FIELDIANA: GEOLOGY Fig. 27. Continued. as in Cyamodus, the palatine fails to contact the quadrate at the lateral margin of the latter (Fig. 26B). This allows the pterygoid to enter the lateral margin of the palatoquadrate cartilage recess be- tween palatine and quadrate. The exact contours and relations of the epiterygoid and prootic are difficult to establish, although the trigeminal in- cisure is distinct on the left side of the skull. The right side of the skull shows a distinct medio- ventral process of the postorbital that, as in Pla- cochelys but unlike Cyamodus, abuts the lateral aspect of the epipterygoid at the posterodorsal corner of the foramen interorbitale. The posttemporal fossa is even more reduced in Macroplacus than in Cyamodus kuhnschnyderi by the expansion of the occipital exposure of the parietal, squamosal, and opisthotic. The relations of the epipterygoid and squamosal along the dor- sal margin of the posttemporal fossa remain ob- scure. The ventral margin of the posttemporal fos- sa, which coincides with the anterior margin of the pteroccipital foramen, is formed by the squa- mosal in the posterior half and by the prootic in the anterior half. The otic process of the squa- mosal therefore remains relatively short in Macro- placus, similar to that of Cyamodus rostratus, but unlike that of Placochelys, where it forms the en- tire ventral margin of the posttemporal fossa and extends beyond the anteromedial corner of the lat- ter. The occiput of the skull of Macroplacus is bad- ly eroded. Only rudiments of the paroccipital pro- cesses can be identified, with their distal tips su- tured to the squamosal. Basioccipital, exoccipi- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 55 tals, and supraoccipital are all missing. A gap be- tween the rudiments of the left opisthotic and squamosal represents the pteroccipital foramen. As preserved, this foramen is located behind and below the squamosal-prootic bridge, which de- fines the ventral margin of the posttemporal fossa in lateral view. With respect to this character, Ma- croplacus resembles Cyamodus rostratus, but as in the specimen of the latter taxon it remains un- clear to what degree this character is the result of erosion or breakage. Macroplacus is unique among cyamodontoids in that a foramen pierces the shaft of the quadrate just above the mandibular condyle (exposed in posterior view: Schubert-Klempnauer, 1975, PI. 5, Fig. 3). This foramen has smooth edges and is bilaterally symmetrical. Its function remains un- known, as does the function of the foramen that pierces the suspensorium between quadrate, quad- ratojugal, and squamosal in Placochelys. The Cranial Anatomy of Protenodontosaurus italicus Pinna, 1990b Protenodontosaurus italicus is known from two skulls (mfsn 1819GP, 1923GP) from the Carnian of Chiout Zuguin east of Dogna, Udine, north- eastern Italy (Pinna, 1990b). These skulls were recently the subject of a detailed description by Nosotti and Pinna (1999). The skull of Proteno- dontosaurus (Fig. 27) is distinctly higher than that of other cyamodontoids. The taxon differs from all other cyamodontoids by the presence of a sin- gle maxillary tooth (Fig. 27C). The persisting maxillary tooth is located lateral to the anterior palatine tooth, which is the position of the pos- teriormost maxillary tooth in all other cyamodon- toids except Cyamodus rostratus. This indicates that Protenodontosaurus has reduced the anterior maxillary tooth (teeth) and retained the posteri- ormost one for biomechanical reasons. Reduction of the maxillary dentition leaves a distinct diaste- ma separating the maxillary from the premaxil- lary dentition (Nosotti & Pinna, 1996). A similar diastema, but less distinctly developed, is present in those specimens of Cyamodus kuhnschnyderi that retain a single pair of premaxillary teeth only (character 8 of Nosotti & Pinna, 1996: 35). In a third specimen of the same species (mhi 1294; Rieppel & Hagdorn, 1999) with two pairs of pre- maxillary teeth, no diastema is present between premaxillary and maxillary teeth. Given the lack of premaxillary teeth in Placochelys and Pse- phoderma, a diastema separating premaxillary and maxillary teeth is autapomorphic for Proten- odontosaurus among cyamodontoids and is cor- related with the reduction of the maxillary denti- tion. The premaxillaries form a short and rounded rostrum. The posterior nasal processes of the pre- maxillaries are rather short and meet the nasal be- tween the external nares in a V-shaped suture (with the apex pointing backward). The nasals are paired, slender, triangular elements that form the entire dorsal (medial) margin of the external na- res. The tapering posterior tips of the nasals ex- tend backward to a level behind the anterior mar- gin of the orbits. A narrow contact of prefrontal and nasal separates the frontal from the maxilla. A distinct anterolateral process of the frontal is embraced by the prefrontal and the nasal (Fig. 27B). The prefrontal is located rather high on the an- terodorsal margin of the orbit and has a relatively narrow dorsal exposure. It descends along the an- terior margin of the orbit but does not approach the anterior tip of the jugal as closely as in Pla- cochelys. The maxilla meets the premaxilla at the anter- oventral (anterolateral) margin of the external nar- is. It forms all of the ventral (lateral) and posterior margin of the external naris. Because of the skull proportions of Protenodontosaurus, the maxilla is almost as high as it is long (Fig. 27 A). The dis- tinct ascending process narrowly enters the anteroventral (anterolateral) corner of the orbit, and its pointed tip is embraced by the prefrontal and maxilla. In the anteroventral margin of the orbit, two foramina can be identified, both fully enclosed by the maxilla (Fig. 27B). The larger foramen is lo- cated medial to the ventral marginal rim of the orbit and represents the lacrimal foramen. The smaller foramen lies on the lateral aspect of the maxilla, that is, lateral to the ventral marginal rim of the orbit. The infraorbital foramen (sensu Oel- rich, 1956), transmitting the infraorbital nerve would have to lie inside the orbit, but it cannot be located in Protenodontosaurus (Nosotti & Pin- na, 1996). Nosotti and Pinna (1998) concluded that the infraorbital nerve must have passed me- dial to the preorbital bridge formed by the pre- frontal, maxilla, and palatine. This interpretation leaves the lateral foramen located within the max- illa unexplained, which raises the question of whether Protenodontosaurus had an anteriorly bi- furcating lacrimal duct, the branches of which 56 FIELDIANA: GEOLOGY passed through paired foramina in the maxilla at the an tero ventral corner of the orbit. Cyamodus kuhnschnyderi is unique in that it shows three fo- ramina along the lateroventral margin of the orbit, and a bifurcating lacrimal duct may be hypothe- sized to have been present in this taxon (see above). Although one of the two lacrimal foram- ina is not in exactly the same topological position with respect to surrounding bones and the orbit in the two taxa, a bifurcating lacrimal duct might have been a character shared by Protenodonto- saurus and Cyamodus kuhnschnyderi. Alterna- tively, Protenodontosaurus might have had a sim- ple lacrimal duct passing through the larger fo- ramen located on the inside of the orbital rim, whereas the smaller foramen located on the max- illa, close to but lateral to the orbital rim, might represent an unusually high placement of the pos- teriormost superior alveolar foramen. The vertical portion of the interdigitating suture between maxilla and jugal lies at about the level of the midpoint of the longitudinal diameter of the orbit in Protenodontosaurus (Fig. 27 A), compa- rable to what is seen in Macroplacus but in a somewhat more anterior position than is observed in Placochelys, in which the vertical suture be- tween maxilla and jugal lies behind the level of the midpoint of the longitudinal diameter of the orbit. The frontals of Protenodontosaurus are paired and meet the parietal in a deeply interdigitating suture at a level between the orbit and the upper temporal fossa (Fig. 27B). The orbital margin of the frontal, between pre- and postfrontal, is straight rather than concave. The frontal closely approaches but does not enter the pineal foramen (Nosotti & Pinna, 1998). The postfrontal is a broad, triangular element with a weakly concave and evenly curved posterolateral margin. It re- mains broadly separated from the upper temporal fossa by the postorbital, which meets the parietal at the anteromedial margin of the upper temporal fossa (Fig. 27B). A broad separation of the post- frontal from the upper temporal fossa is also ob- served in Psephoderma and, to a lesser extent, in Cyamodus rostratus, but not in Placochelys or Cyamodus kuhnschnyderi, where the contact of postorbital and parietal is narrow. The postorbital broadly enters the postero ventral margin of the orbit. It extends backward into the temporal arch, forming the anterior and anterolateral margin of the upper temporal fossa. The posterior tip of the postorbital remains restricted to a level in front of the longitudinal diameter of the upper temporal fossa. The jugal broadly enters the ventral margin of the orbit and closely approaches but does not en- ter the lacrimal foramen (Fig. 27A). Posteriorly it extends along the ventral margin of the temporal arch to about the same level as the postorbital does along the dorsal margin of the temporal arch. Together, these two elements embrace the anterior tip of the squamosal, which remains restricted to a level behind the anterior margin of the upper temporal fossa in Protenodontosaurus and hence does not reach as far anteriorly as in Placochelys. The parietal skull table, formed by the fused parietals, is broad and shows straight lateral mar- gins. The large pineal foramen is located anteri- orly in the skull table (Fig. 27B). Dermal encrus- tations are present but poorly defined. As far as can be determined, they resemble those of Pla- cochelys, with a pair of encrustations along either side of the parietal skull table and an unpaired posteromedial encrustation located at the posterior margin of the skull table. The diverging posterior processes of the parietal meet the squamosals along the posteromedial margins of the upper temporal opening in an interdigitating suture, which is distinct in the second specimen (mfsn 1923GP) only. Dermal encrustations create a dis- tinct step at the posterior corner of the upper tem- poral fossa on both sides of the skull that might be mistaken for a suture separating a small squa- mosal from the large quadratojugal. Whether the quadratojugal entered the posterolateral margin of the upper temporal fossa remains indistinct in Protenodontosaurus because of dermal encrusta- tions on the temporal arch, which obscure the su- tural relations, and because of the incompleteness of the specimens. The ventral view of the skull (Fig. 27C) shows the much enlarged vomers, which extend far for- ward into the rostrum, an autapomorphy of Pro- tenodontosaurus. The enlarged vomers exclude the premaxillaries from the internal nares. Poste- riorly the vomers do not extend beyond the level of the posterior margins of the internal nares as they do in Cyamodus kuhnschnyderi and Placo- chelys (unknown in Cyamodus rostratus and Pse- phoderma). The maxilla narrowly enters the an- terolateral margin of the internal naris, between the anterior vomer and the posterior palatine. A distinct nutritive (dental lamina) foramen is locat- ed within the maxilla behind the maxillary tooth plate. In ventral view, the posterior end of the maxilla tapers to a pointed tip that remains ex- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 57 Table 10. Measurements of the maxillary and pal- atine tooth plates of Protenodontosaurus italicus (holo- type, mfsn 1819GP). All measurements in mm. left right longi- tudinal 0 trans- verse 0 longi- tudinal 0 trans- verse 0 maxillary tooth 20.2 13.4 20.0 13.3 anterior palatine tooth 12.8 13.3 13.5 13.2 xjsterior palatine tooth 36.0 28.5 36.0 29.0 eluded from the anterior margin of the subtem- poral fossa by a contact of the jugal and palatine. On the left side of the skull of the holotype (mfsn 1819GP), a lateral process of the palatine meets the jugal, as is also the case in Placochelys; on the right side of the same specimen, the jugal is seen to extend medially to meet the palatine. In Protenodontosaurus, the jugal does not extend backward along the anteromedial margin of the subtemporal fossa, as it does in Cyamodus rostra- tus. As in other cyamodontoids, the palatine carries two tooth plates, of which the posterior one is much larger than the anterior one (Table 10). The posterior dental lamina foramina are located be- hind the posterior palatine tooth plates on the pal- atine-pterygoid suture. Although the posterior palatine tooth plates are not as disproportionally enlarged as in Macroplacus, the relative palatal exposure of the pterygoid is very short in Proten- odontosaurus. The ratio of the length of the pter- ygoid (from its posterior margin to the dental lam- ina foramen on the palatine-pterygoid suture) to the distance from the pterygoid-palatine suture to the posterior margin of the internal naris is 0.2 for Protenodontosaurus. This indicates that the rela- tive length of the palatal exposure of the ptery- goid is not simply a reflection of the relative size of the posterior palatine tooth plates. The ptery- goid forms a longitudinally oriented ventral flange with a single ventral projection. The pterygoid ex- tends anteriorly along the medial margin of the subtemporal fossa to the level of about the mid- point of the longitudinal diameter of the posterior palatine tooth plate. An ectopterygoid is absent in Protenodontosaurus. The lateral view of the braincase shows the dis- tinct palatoquadrate cartilage recess with a com- paratively broad contact of the palatine with the anteromedial wing of the quadrate along its lateral edge (Fig. 27B). The posterodorsal margin of the palatoquadrate cartilage recess is formed by the otic process of the squamosal, which in Proteno- dontosaurus extends far anteriorly, closely ap- proaching the trigeminal incisure located between the prootic and epipterygoid. As mentioned above, the second specimen (mfsn 1923GP) shows an internal subdivision of the trigeminal incisure by a vertical strut of bone. The epitery- goid is a broad element that in the second speci- men (mfsn 1923GP) is incompletely ossified. Un- fortunately, the dorsal relations of the epiptery- goid remain incompletely known for Protenodon- tosaurus. In particular, it is impossible to ascertain whether the epipterygoid forms a posterior dorsal process that meets the squamosal at the dorsal margin of the posttemporal fossa. The posttemporal fossa is large in Protenodon- tosaurus, with the pteroccipital foramen located at its ventral margin. As in Placochelys, the pter- occipital foramen appears slightly recessed below and behind the ventral margin of the posttemporal fossa (Fig. 27B). The exoccipital forms the lateral margin of the foramen magnum and between it- self and the opisthotic encloses the jugular fora- men. The jugular foramen is not subdivided in- ternally in Protenodontosaurus. The exoccipitals meet each other dorsal to the occipital condyle formed by the basioccipital alone, a character shared with Cyamodus kuhnschnyderi (Nosotti & Pinna, 1996) but absent in Placochelys (and Pse- phoderma: see below). As in other cyamodon- toids, the anterior aspect of the relatively slender paroccipital process is lined by the squamosal, but a distinct squamosal buttress receiving the distal end of the paroccipital process, as seen in Pla- cochelys (and Psephoderma: see below), is absent in Protenodontosaurus. The foramen for the in- ternal carotid is closed ventrally in Protenodon- tosaurus by a ventral contact of the basioccipital tuber and the ventral process of the opisthotic, yet the opisthotic pedicel remains widely separated from the posterior margin of the basicranium or pterygoid, respectively. Enlarged temporal tubercles secondarily fused to the underlying bone are absent on the lateral surface of the temporal arch of Protenodontosau- rus, whereas they do occur in Cyamodus kuhn- schnyderi and Placochelys. Unfortunately, the posterior margin of the upper temporal fenestra is poorly preserved on both sides of the skull of Protenodontosaurus, so that it is impossible to as- certain whether such temporal tubercles were re- stricted to a posterior position as in Psephoderma or were altogether absent in Protenodontosaurus. 58 FIELDIANA: GEOLOGY Fig. 28. 20 mm. Skull of Psephoderma alpinum H. v. Meyer (msnm V471): A, dorsal view; B, ventral view. Scale bar = The Cranial Anatomy of Psephoderma alpinum H. v. Meyer, 1858 Psephoderma alpinum was originally based on an isolated carapace (H. v. Meyer, 1858) from the Rhaetian Koessen-Formation of the Bavarian Alps (Winkelmoos Alpe), but today it is known from several articulated specimens, two of which are complete (Pinna & Nosotti, 1989; Renesto & Tin- tori, 1995). All of these specimens have sustained severe dorsoventral compression, which has ob- scured some details of skull anatomy. The speci- men yielding the most information on the cranial anatomy of Psephoderma is the isolated skull from Monte Cornizzolo (msnm V471; Pinna, 1976a; Pinna & Nosotti, 1989). This skull too was subjected to some dorsoventral compression, however, which resulted in extensive breakage in the preorbital region. Also, the specimen is in- completely prepared: both orbits and the left tem- poral fossa are still filled with matrix, and the lat- eral wall of the braincase is only incompletely ex- posed in the right temporal fossa (Fig. 28). Other than by the morphology of the rostrum (see above), the genus is unique among cyamo- dontoids in the relative proportions of the upper temporal fossa. The skull is depressed yet narrow, and the temporal fossae in particular are relatively short and distinctly narrower than in other cy- amodontoids (Table 8). The paired premaxillaries form the elongated, narrow, edentulous rostrum (Fig. 29). Posteriorly the premaxillaries extend into distinctly enlarged posterior (nasal) processes that project backward far beyond the level of the posterior margins of the external nares, entering deeply in between the frontals (Pinna & Nosotti, 1989). The premaxillae thus completely separate the nasals from one an- other, a character that Psephoderma (Fig. 29B) shares with Macroplacus. The nasals are reduced to small, elongate elements located entirely pos- terior to the external nares, between the premax- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 59 qj Pp 60 FIELDIANA: GEOLOGY Fig. 29. Continued. ilia and the prefrontal, although their precise out- line can no longer be determined because of breakage. A break runs transversely through the posterior (nasal) processes of the premaxillaries at the level of the anterior margin of the external nares, creating a distinct step that might be mis- taken for a separation of the premaxillaries from large nasals. The exact contours of the prefrontal are also difficult to identify because of breakage, but, as in all other cyamodontoids, a lacrimal is absent. The maxilla is an elongated yet very low element that extends anteriorly into the rostrum (see above) and posteriorly to a level slightly behind the midpoint of the longitudinal diameter of the orbit, where it meets the jugal in a V-shaped su- Fig. 29. Skull of Psephoderma alpinum H. v. Meyer (msnm V471 ): A, left lateral view, B, dorsal view; C, ventral view. Scale bar = 20 mm. For a list of abbreviations see p. 3. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 61 ture (with the apex pointing backward; Fig. 29A). An ascending process of the maxilla cannot be identified, but even if this should be attributable to breakage, the process must have been rudimen- tary, given the low profile of the skull. The pos- terior end of the maxilla extensively enters the anterior ventral (lateral) margin of the orbit. Be- cause of incomplete preparation, the lacrimal fo- ramen cannot be identified. The right side of the skull would seem to indicate that the jugal did not extend anteriorly along the ventral margin of the orbit beyond the level of the midpoint of the lon- gitudinal diameter of the latter, although if true, the maxilla would extend along the medioventral margin of the orbit much farther back than is in- dicated by the lateral exposure of the bone. The left side of the skull shows no indication of the anterior tip of the jugal. The frontals are paired, each forming a distinct anterolateral process that is embraced by the pre- maxilla and prefrontal (Fig. 29B). A contact of the prefrontal with the nasal separates the frontal from the maxilla. Between prefrontal and post- frontal, the frontal forms the concave dorsal mar- gin of the orbit. The large and elongate pineal foramen lies at the anterior end of the parietal, at a level between the orbit and the upper temporal fossa. On the right side of the skull, a narrow entry of the frontal into the anterior margin of the pineal foramen can be identified. The postfrontal is a comparatively small, triangular element lo- cated at the posterodorsal corner of the orbit. Its posterolateral margin is concave but evenly curved rather than angulated. A contact of the postorbital with the parietal broadly separates the postfrontal from the anteromedial margin of the upper temporal fossa. The postorbital is a large element that forms most of the broad postorbital arch (Figs. 29A, B). It defines the posteroventral (posterolateral) mar- gin of the orbit. Along the lateral margin of the upper temporal fossa, the postorbital extends backward to a level well behind the midpoint of the longitudinal diameter of the upper temporal fossa, a character that Psephoderma (Fig. 29B) shares with Placochelys and Macroplacus but not with Protenodontosaurus, in which the postorbital extends only to about the midpoint of the longi- tudinal diameter of the upper temporal fossa (this character remains unknown for Cyamodus). The parietals of Psephoderma are fused and form a broad parietal skull table with distinctly concave lateral margins due to a constriction of its posterior part. Instead of dermal encrustations, the skull table is ornamented with a pattern of ridges and grooves radiating from its center of ossification to the margins of the bone. Behind the pineal foramen, the concave anteromedial margin of the fused parietal forms a distinct step that con- tinues laterally along the anteromedial margin of the postorbital. In this way the pineal foramen comes to lies in a distinct depression, which is a unique character of Psephoderma. The right side of the skull shows a distinct yet slender anterior process of the parietal that extends beyond this step medial to the postorbital, postfrontal, and frontal to form the lateral margin of the pineal foramen. The left temporal arch shows particularly well the sutures delineating the posterior processes of the postorbital and jugal, together embracing the anterior process of the squamosal which, unlike in Placochelys, extends anteriorly to a level be- yond the anterior margin of the upper temporal fossa (Fig. 29A). The quadratojugal extends an- teriorly along the ventral margin of the temporal arch beyond the level of the midpoint of the lon- gitudinal diameter of the upper temporal fossa, but it does not reach beyond the level of the an- terior margin of the upper temporal fossa, as is the case in Placochelys. Because of breakage of the bone surface, the relations of the vomers remain obscure in the ven- tral view of the skull of Psephoderma (Fig. 29C), but, especially on the left side of the skull, it ap- pears that the premaxilla enters the anterior mar- gin of the internal naris, whereas it remains ex- cluded therefrom in all other cyamodontoids, with the possible exception of Macroplacus. The max- illa enters the anterior lateral margin of the inter- nal naris, between the premaxilla anteriorly and the palatine posteriorly. The left side of the skull shows a distinct medial process of the jugal, which meets the palatine at the anteromedial mar- gin of the subtemporal fossa and thus excludes the maxilla from the latter. There is no indication that the jugal extends backward along the medial mar- gin of the subtemporal fossa, as it does in Cy- amodus rostratus. The posterior palatine tooth plates of Psepho- derma are distinctly elongated (at least in the adult, see Tables 7 and 1 1 ), a character also shared by Macroplacus. The posterior dental lamina fo- ramina are located on the pterygoid-palatine su- ture posteromedial to the posterior palatine tooth plates. In spite of the elongation of the palatine 62 FIELDIANA: GEOLOGY tooth plates, the pterygoid retains a relatively long palatal exposure compared with Protenodontosau- rus (without elongation of the posterior palatine tooth plates) and Macroplacus (with much en- larged posterior palatine tooth plates). The relative length of the palatal exposure of the pterygoid can therefore be treated as a character that, to some degree at least, is independent from the relative size of the posterior palatine tooth plates (see also the discussion above). Dividing the length of the pterygoid (from its posterior margin to the dental lamina foramen on the palatine-pterygoid suture) by the distance from the pterygoid-palatine suture to the posterior margin of the internal naris yields a ratio of approximately 0.36 for Psephoderma (msnm V471). The pterygoid forms a longitudi- nally oriented ventral flange with a single ventral projection. The left pterygoid is seen to extend anteriorly along the medial margin of the subtem- poral fossa to the level of the posterior third of the longitudinal diameter of the posterior palatine tooth plate. An ectopterygoid is absent (Fig. 29C). The lateral wall of the braincase is only par- tially exposed, and little anatomical detail is re- vealed because of poor preservation and prepa- ration. The palatoquadrate cartilage recess is dis- tinct, however, and the palatine contacts the an- teromedial lamina of the quadrate along its lateral edge. The occiput of Psephoderma is deeply exca- vated. The occipital exposure of the braincase was subject to erosion, which obscured structural de- tail. The quadratojugal is clearly demarcated from the quadrate in occipital view. There is a distinct buttress on the lower surface of the left squamo- sal, abutted by the distal tip of the slender par- occipital process (Fig. 29B). A comparable squa- mosal buttress is present in Placochelys, but not in Cyamodus or in Protenodontosaurus. The ven- trally descending flange of the opisthotic is dis- tinct and in Psephoderma contacts the posterior margin of the pterygoid. Among the other cyamo- dontoids included in this study, a comparable con- tact is observed only in the Berlin specimen of Placochelys (mb.r. 1765). Compared with Cyamodus and Placochelys, en- larged temporal tubercles secondarily fused to the underlying bone are reduced in Psephoderma, where they are restricted to the posterior extremity of the squamosals but do not appear on the lateral surface of the posterior part of the temporal arch (as indicated in the reconstruction of Pinna & No- sotti, 1989). Autapomorphies in the Skull of the Cyamodontoidea In addition to highly diagnostic characters of the postcranium, such as the development of an extensive dermal armor or dermal encrustations on the skull, the cranial anatomy offers additional evidence of the monophyly of the Cyamodonto- idea (Rieppel & Zanon, 1997). The palatal den- tition is reduced to two tooth plates on the palatine (one in Henodus: Huene, 1936). The postfrontal remains excluded from the upper temporal fenes- tra by a contact of the postorbital with the parietal (the postfrontal enters the anteromedial margin of the upper temporal fossa in Placodus [Rieppel, 1995a], a condition that is also plesiomorphic for Eosauropterygia [Rieppel, 1997]). The pineal fo- ramen is large, displaced anteriorly, and may bor- der on the frontoparietal suture. The jugal extends far backward in the temporal arch, contacting the quadratojugal and the squamosal posteriorly (per- haps also in Placodus, but not in Paraplacodus [Rieppel, 1995a]; a short jugal was reconstructed for Cyamodus kuhnschnyderi by Nosotti & Pinna, 1996, but the temporal arch is incomplete in both specimens). The ectopterygoid is missing (but see the discussion of Placochelys above). Most dis- tinctive for cyamodontoids, however, is the struc- ture of the secondary lateral wall of the braincase, which incorporates the palatoquadrate. The epi- pterygoid is very broad. It extensively overlaps the dorsal surface of the palatine and is connected to the quadrate by palatoquadrate cartilage, which persisted in the adult. The palatine contacts the quadrate lateral to the palatoquadrate cartilage re- cess in some taxa, and a groove on the dorsal surface of the palatine accommodates the persis- tent anterior (palatine) process of the palatoquad- rate. Posteriorly, the palatoquadrate cartilage re- cess is floored by the pterygoid, the dorsal wing of which broadly overlaps with an anterior wing of the quadrate. Together with the fusion of the palatobasal articulation, this broad overlap of pterygoid and quadrate obliterates the anterior part of the cranioquadrate passage. As a conse- quence, the stapedial (temporal) artery reaches the jaw adductor musculature through the pteroccipi- tal foramen. Finally, the internal carotid enters the cranioquadrate passage through a foramen located between the basioccipital tuber and a ventral pro- cess of the opisthotic and continues anteriorly in a basicranial canal. This whole character complex is unique among amniotes but is shared by all cyamodontoid placodonts. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 63 Fig. 30. Left lateral braincase wall of Placodus gigas Agassiz (umo BT 13). Scale bar = 20 mm. For abbrevia- tions, see p. 3. Because the clade that includes Placodus and Paraplacodus is the sister group of the Cyamo- dontoidea, it is interesting to compare the derived structure of the cyamodontoid braincase with the more generalized braincase anatomy of Placodus (the braincase of Paraplacodus is not known). The best-preserved and best-prepared skull of Placodus is the specimen umo BT 13, originally described by Sues (1987) and redescribed by Rieppel (1995a). In contrast to cyamodontoids, Placodus has an epipterygoid with a broad base but a relatively narrow dorsal process. Broili (1912, PI. 14, Fig. 6) described a recess located between the posteroventral aspect of the base of the epipterygoid (covered by unfinished bone) and the anteromedial aspect of the quadrate, above the pterygoid. Huene (1931) correctly assumed that this recess must have housed persistent palato- quadrate cartilage. The specimen umo BT 13 shows nicely (Fig. 30) the extensive dorsomedial flange of the quadrate ramus of the pterygoid, which anteriorly supports the broad, fan-shaped base of the epipterygoid. The dorsal head of the epipterygoid abuts the descending flange of the parietal at the posterodorsal corner of the foramen interorbitale. A bony process rises up along the medial aspect of the anterior margin of the epi- pterygoid, which is clearly located lateral to the rostrum basisphenoidale and hence cannot repre- sent an ossification of the primary lateral wall of the braincase. Its base is broken in umo BT 13, but Broili's (1912) specimen suggests that this dorsal process originates from the pterygoid. Oth- er than in cyamodontoids, the cranioquadrate pas- sage persists in Placodus; it extends from the space between the pterygoid and the dorsolateral aspect of the basioccipital tubers into the cavum epiptericum medial to the epipterygoid (Rieppel, 1995a). The cavum epiptericum is represented by the space between the epipterygoid and the ros- trum basisphenoidale. The rostrum basisphenoi- dale carries the sella turcica, which is pierced by two foramina for the passage of the cerebral ca- rotids into the braincase. From the lateral margin of the sella turcica originates, on either side of the skull, a clinoid process that represents an ossifi- cation of the primary lateral wall of the braincase (base of the pila antotica). The clinoid process rises up in a dorsolateral direction and contacts the dorsal tip of the dermal process lining the an- teromedial edge of the epipterygoid. The anterior opening of the cavum epiptericum thus is well- defined by the clinoid process and the dorsal pro- cess of the ?pterygoid lining the anteromedial 64 FIELDIANA: GEOLOGY margin of the epipterygoid. The internal carotid must have entered the cranioquadrate passage be- tween the pterygoid and the basioccipital tube. It must have subdivided within the posterior part of the cranioquadrate passage, giving rise to the sta- pedial artery, which must have reached the tem- poral musculature by passing through a gap be- tween the dorsomedial flange of the quadrate ra- mus of the pterygoid and the descending flange of the parietal, lateral to the otic capsule and an- terior to the paroccipital process (Fig. 30). As in cyamodontoids, the palatine-pterygoid suture can be traced to the ventral tip of the lon- gitudinally oriented ventral flange of the ptery- goid. From there it trends in an anterodorsal di- rection until it disappears under the epipterygoid. In Placodus, the epipterygoid does encroach on the posterior margin of the palatine, but it does not invade the dorsal surface of the palatine to the same extent as the much broader epipterygoid of cyamodontoids. Posteriorly, the base of the epi- pterygoid expands over the dorsomedial flange of the pterygoid, terminating in a surface of unfin- ished bone. As in cyamodontoids, the quadrate of Placodus also carries a distinct anteromedial flange that overlaps the dorsomedial flange of the pterygoid and again terminates in an unfinished margin opposing the posteroventral margin of the base of the epipterygoid. In between these two bones, the pterygoid flange must have been cov- ered by persistent palatoquadrate cartilage in the live animal (Fig. 30). In contrast to cyamodon- toids, there is no evidence that a cartilaginous an- terior palatal process of the palatoquadrate per- sisted in Placodus. As mentioned above, the stapedial artery of Placodus passed through a gap between the dorsal flange of the pterygoid and the descending flange of the parietal. The prootic is exposed at the an- teromedial corner of this gap as it emerges from behind the epipterygoid and from below the de- scending flange of the parietal. Posterolateral to the prootic, the opisthotic extends into the par- occipital process, again exposed in lateral view between the pterygoid and the parietal. The gap for the passage of the stapedial artery is closed posteriorly by the squamosal. In cyamodontoid placodonts, this gap for the passage of the stapedial artery is reduced to a small pteroccipital foramen (Nosotti & Pinna, 1993b). Closure is the consequence of the for- mation of a neomorphic otic process of the squa- mosal, which extends anteriorly along the dorsal margin of the pterygoid flange and meets the prootic, which expands backward. These two bones (in Cyamodus rostratus, see above), or the otic process of the squamosal alone (in Placo- chelys), form the anterior margin of the pteroc- cipital foramen. In a parallel development, there is an expansion of the squamosal anteromedially along the anterior aspect of the paroccipital pro- cess; this, together with the opisthotic, forms the posterior margin of the pteroccipital foramen. Evolution of the Rostrum and of the Dentition Within the Cyamodontoidea The most distinctive characters that differ among the cyamodontoid taxa relate to the struc- ture of the rostrum and the dentition. The genus Cyamodus retains a short, broad rostrum formed by tooth-bearing premaxillae. There are two teeth in each premaxilla of Cyamodus rostratus (Fig. 12B), and between one (Nosotti & Pinna, 1996) and two (Rieppel & Hagdorn, 1999) teeth on the premaxilla of Cyamodus kuhnschnyderi. Ontoge- netic variations of the dental formula of Cyamo- dus hildegardis have been described by Kuhn- Schnyder (1959). The rostrum of Macroplacus is incomplete, but what is preserved shows a deep trough, or con- cavity, on its ventral surface between the two pre- maxillaries (Fig. 26C). The maxilla carries an an- terolateral, relatively short yet distinct process that tapers off along the lateral margin of the rostrum well in front of the level of the anterior margin of the internal nares. This sutural pattern contrasts with that of Cyamodus, where the suture between the premaxilla and maxilla is more transversely oriented (C. kuhnschnyderi) or even trends in a posterolateral direction (C. rostratus). The ante- rior tip of the premaxilla does not extend anteri- orly beyond the level of the anterior margin of the internal nares in Cyamodus. The rostrum of Protenodontosaurus is more elongated than that of Cyamodus, and each pre- maxilla carries an anterior tooth, plus a second posterior alveolus, which may have carried a sec- ond premaxillary tooth or which may represent a replacement pit (Nosotti & Pinna, 1999). As in Macroplacus, the maxilla of Protenodontosaurus forms an anterior process that participates in the formation of the rostrum. However, this antero- lateral process of the maxilla is somewhat longer in Protenodontosaurus, and its ventral surface shows a low ridge that, together with its counter- part from the other side of the skull, delineates a RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 65 Table 1 1 . Data matrix for the analysis of placodont interrelationships. For further details, see text. 1 1 2 3 4 s 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0 2 1 2 2 2 3 1 Ancestor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 Placodus 1 0 0 0 0 0 0 0 0 0&1 0 1 0 0 0 0 1 1 0 0 0 0 0 3 Paraplacodus 0 0 0 0 0 0 9 0 9 9 9 9 0 9 9 0 0 9 0 9 0 9 0 4 C. rostratus ? 0 0 0 0 0 1 0 1 1 0 2 9 0 1 0 1 1 5 C. hildegardis 2 ? 0 0 ? 9 9 9 9 ? 9 9 1 ? ? 9 ? 9 9 f 0 1 6 C. kuhnschnyderi 2 0 0 0 0 0 1 1 1 1 0 1 9 1 9 1 1 7 Henodus 2 0 2 9 0 1 0 9 0 1 0 9 9 9 9 0 0 0 8 Macroplacus 2 0 0 ? 1 1 0 0 0 0 0 0&1 ? ? 1 2 1 9 0 9 Placochelys 2 1 1 1 0 0 1 0 0 0&1 0 0 1 1 1 1 1 0 0 1 0 Protenodontosaurut 2 0 0 1 1 0 0 0 0 0 0 0 1 0 2 0 2 9 0 0 1 1 Psephoderma 2 1 1 0 ? 1 9 9 0 2 0 0 0 0 0 2 1 1 ? 0 0 1 2 Israel 2 ? 9 ? ? 9 9 ? ? ? ? 9 9 1 9 ? 9 ? ? 9 9 1 3 laticeps ? 0 0 0 9 1 ? 9 ? ? 0 9 9 9 9 ? 0 1 9 1 1 broad and shallow trough leading to the internal nares (Fig. 27C). In Placochelys, the rostrum becomes distinctly elongated and narrow (Fig. 3B). The premaxilla is edentulous. The maxilla again carries an ante- rior process participating in the formation of the rostrum and together with the premaxilla delin- eates a distinct sulcus on either side of the ventral surface of the rostrum, leading up to the anterior margin of the internal naris. In Psephoderma, the rostrum is even more elongate and narrow (Fig. 29C). The premaxilla is edentulous. The maxilla carries an anterior pro- cess that participates in the formation of the ros- trum and that carries a low ridge on its ventral surface located lateral to the deep premaxillary sulcus. Each premaxilla carries a distinct ridge on its ventral surface along its entire length, defining a deep sulcus on the ventral surface of the rostrum that is united anteriorly but becomes paired by medial ventral ridges on the premaxillae posteri- orly as these grooves approach the anterior mar- gin of the external nares. Pinna (1979) described two premaxillary teeth in a juvenile specimen of Psephoderma alpinum, an observation that could not be confirmed on personal inspection of the specimen (msnb 4884a-b). With the exception of Henodus (Huene, 1936), the rostrum of cyamodontoid placodonts can thus be described in terms of a morphocline, progress- ing from a short, rounded and tooth-bearing struc- ture to an elongated, narrow, edentulous rostrum with deep grooves on its lower surface leading up the internal nares. Some authors reconstruct the lateral margins of these grooves to have been cov- ered by a horny sheath comparable to the rham- photheca of turtles. Henodus, by contrast, shows a broad and flat rostrum, the anterior cutting edge of which is furnished with a series of denticles as described above (Stein, 1995; Reiff & Stein, 1999). Among the species of the genus Cyamodus, the Table 1 1 . Continued. 2 24 2 5 2 6 2 7 28 29 3 0 3 1 3 2 3 3 34 35 3 6 37 38 3 9 4 0 4 1 4 2 4 3 4 4 45 4 6 1 Ancestor 0 0 0 0 0 0 0 0 0 0 0 0 0 ? 0 0 0 0 0 0 ? 0 0 2 Placodus 0 0 0 0 0 0 0 0 0 1 1 0&1 1 1 0 0 0 0 1 0 0 0 0 3 Paraplacodus 0 9 0 ? 0 9 ? 0 0 0 1 0 0 1 0 9 9 0 1 9 ? 0 ? 4 C. rostratus 1 0 1 1 0 0 0&1 0 2 0 2 1 0 0 0 0 0 0 2 0 1 0&1 5 C. hildegardis ? 9 ? 9 ? 9 9 ? 0 2 0&1 0&1 1 &2 0 0&1 9 ? 9 0 ? 0 ? 9 6 C. kuhnschnyderi ? 1 1 1 0 0 0 2 0&1 2 2 0 0 0 0 0 0 2 0 1 1 7 Henodus ? 0 9 1 9 ? ? ? 9 4 3 ? 1 9 0 0 0 ? 1 0 0 8 Macroplacus ? 0 1 0 0 0 ? ? 1 2 2 0 1 1 1 0 0 1 0 9 1 9 Placochelys 1 0 2 0 1 1 1 ? 9 1 2 0 0 1 1 1 0 1 1 0 0 1 0 Protenodontosaurut ? 0&1 2 0 1 1 0 1 1 3 2 0 0 1 1 0 0 1 &2 0 0 0 1 1 Psephoderma ? 0 ? 0 1 9 1 ? ? 2 2 0 1 1 1 1 0 1&2 1 0 0 1 2 Israel ? 0 ? ? 0 0 9 ? 9 ? ? ? 0 9 9 9 9 ? 1 9 0 1 3 laticeps ? 9 ? 1 9 9 9 0 2 0 1 2 0 0 9 0 0 0 ? 9 1 9 66 FIELDIANA: GEOLOGY Table 1 1 . Continued. 3 4 7 48 4 9 50 5 1 5 2 53 54 1 Ancestor 0 0 0 0 0 0 0 0 2 Placodus 0 0 0 0 0 0 0 0 3 Paraplacodus ? ? ? f 0 0 0 0 4 C. rostratus 0 1 1 0 0 1 ? 5 C. hildegardis ? ? ? ? 0 1 2 6 C. kuhnschnyderi 0 1 1 0 0 1 2 7 Henodus 0 1 1 0 1 0 2 8 Macroplacus ? ? ? ? ? ? 2 9 Placochelys 1 0 0 1 1 1 2 1 0 Protenodontosaurui 0 0 1 1 0 ? ? 1 1 Psephoderma 1 0 0 0 1 1 1 1 2 Israel ? 0 0 0 ? ? 2 1 3 laticeps ? ? ? ? ? ? 2 specimens shows posterior palatine tooth plates that are 1 .4 to 1 .5 times as long as they are wide. Interestingly, there is an ontogenetic change of proportions in Psephoderma alpinum, with dis- tinctly more elongated palatine tooth plates in the adult than in the juvenile (Table 7). Macroplacus raeticus shares the elongated posterior palatine tooth plates, which may be the reason why Pinna (1978) treated this taxon as a junior synonym of Psephoderma alpinum. Cladistic Analysis of Cyamodontoid Interrelationships number of maxillary teeth varies between two and three in adult specimens (Kuhn-Schnyder, 1959, 1965a; Nosotti & Pinna, 1996). Placochelys shows three maxillary teeth, Psephoderma shows two maxillary teeth, and Protenodontosaurus shows a single maxillary tooth plate. All adult specimens of Cyamodus, Placochelys, Proteno- dontosaurus, and Psephoderma carry a small an- terior and a much larger posterior tooth plate on each palatine. Henodus is autapomorphic by the loss of maxillary teeth and by the reduction of the palatine teeth to one relatively small posterior tooth plate. The proportions of the posterior pal- atine tooth plates vary among cyamodontoid taxa. Cyamodus from the Germanic Triassic (i.e., C. rostratus, C. kuhnschnyderi), as well as Placo- chelys placodonta and Protenodontosaurus itali- cus have posterior palatine tooth plates that are between 1.2 and 1.3 times as long as they are wide (Table 7). The southern Alpine species Cy- amodus hildegardis shows somewhat more elon- gated palatine tooth plates (1.3 to 1.4 times as long as they are wide), and in this character ap- proaches Psephoderma alpinum, which in adult Henodus C. rostratus C. kuhnschnyderi Macroplacus Protenodontosaurus Placochelys Psephoderma Placodus Paraplacodus Fig. 31. Most parsimonious unrooted network for Placodontoidea. For further discussion, see text. The cladistic analysis of cyamodontoid rela- tionships initially included as terminal taxa a hy- pothetical all-0-ancestor (which roots the tree on the presence of characters), Placodus, Parapla- codus, Cyamodus rostratus, Cyamodus kuhn- schnyderi, Henodus chelyops, Macroplacus, Pla- cochelys, Protenodontosaurus, and Psephoderma, and was based on 53 cranial (see Appendix I) and one postcranial (absence or presence of carapace) characters (Table 1 1 ). Because of the incomplete knowledge of their cranial anatomy, Cyamodus muensteri (coding based on the holotype of C. laticeps Owen), Cyamodus hildegardis and the cyamodontoid skull from the Muschelkalk of Makhtesh Ramon, Negev, Israel (provisionally re- ferred to IPsephosaurus by Brotzen, 1957) were added to the analysis in separate steps. All search- es employed PAUP version 3.1.1. (Swofford, 1990; Swofford & Begle, 1993), with the branch- and-bound search option implemented. Diagnostic characters listed below are those obtained by DELTRAN character optimization if not other- wise indicated. Unambiguous synapomorphies, optimizing the same way in ACCTRAN or DEL- TRAN, are indicated by an asterisk (*). Bootstrap analysis is based on 2,000 replications using the branch-and-bound search option. The reconstruction of an unrooted network (characters 1, 10, 20, 27, and 30 were uninfor- mative and hence ignored) resulted in a single tree with a tree length of 91 steps (Fig. 31). Although it does not indicate phylogenetic relationships, the network does suggest that the Cyamodontoidea could potentially include two separate monophy- letic subclades, one comprising the taxa from the Germanic Triassic (Henodus, Cyamodus rostra- tus, Cyamodus kuhnschnyderi), the other compris- ing Placochelys and the taxa from the Alpine Tri- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 67 Fig. 32. Most parsimonious reconstruction of pla- codont interrelationships (single most parsimonious tree with a TL of 103 steps, a CI of 0.757, and an RI of 0.706). For further discussion, see text. assic (Macroplacus, Protenodontosaurus, Pse- phoderma). At the very least, the network indi- cates that grouping some but not all taxa from the Germanic Triassic with those of the Alpine Tri- assic and vice versa will result in a paraphyletic grouping (for the results obtained on including Cyamodus hildegardis, see below). Indeed, the cyamodontoid relationships suggested by the net- work are exactly those obtained either by rooting the analysis on the sister group of the Cyamodon- toidea, the Placodontoidea, or by rooting all ter- minal taxa on an all-0-ancestor. With the branch-and-bound search option im- plemented, a single most parsimonious tree was obtained when the analysis was rooted on the sis- ter group of the Cyamodontoidea, that is, the Pla- codontoidea (Placodus, Paraplacodus; characters 1, 10, 20, 27, and 30 were uninformative and hence ignored). Tree length (TL) was 91 steps, the consistency index (CI) was 0.769, and the reten- tion index (RI) was 0.677. Rooting the analysis of placodont interrelation- ships on the hypothetical all-0-ancestor (all char- acters informative) yielded one single most par- simonious tree with a TL of 103 steps, a CI of 0.757, and an RI of 0.706 (Fig. 32). Interestingly, the tree suggests paraphyly of the Placodontoidea, with Placodus being more closely related to the Cyamodontoidea than to Paraplacodus. This re- sult is potentially important, because Paraplaco- dus had previously been claimed to be the most generalized placodont known (Peyer & Kuhn- Schnyder, 1955), a claim that was refuted by an earlier analysis of placodont interrelationships (Rieppel & Zanon, 1997). Characters that Placo- dus shares with the Cyamodontoidea but that are absent in Paraplacodus are: 17* (1) [ci = 0.5], the postorbital extending along the lateral margin of the upper temporal fossa to a level behind the midpoint of the longitudinal diameter of the upper temporal fossa; 33* (1) [ci = 1], chisel-shaped anterior premaxillary teeth. This latter character is somewhat problematic, as the premaxillary teeth of Protenodontosaurus are not as slender and pointed as those of Paraplacodus, but neither are they as broad and chisel-shaped as those of Pla- codus. On the basis of the current data matrix, the paraphyly of the Placodontoidea obtains only if the premaxillary teeth of Protenodontosaurus are coded as chisel-shaped (1). If they are coded as pointed (0), two trees are obtained, with unre- solved relationships of Placodus and Paraplaco- dus relative to the Cyamodontoidea. The same re- sult is obtained if the teeth of Protenodontosaurus are coded as intermediate between the chisel- shaped premaxillary teeth of Placodus and the bulbous premaxillary teeth of Cyamodus. The monophyly of the Placodontoidea is recovered if two postcranial characters shared by Placodus and Paraplacodus are added to the matrix (i.e., the hyposphene-hypantrum articulation and the strongly bent lateral gastral rib elements). For those reasons, support for a relationship of Pla- codus closer to cyamodontoids than to Parapla- codus is weak; the node collapses in a tree one step longer (TL = 104 yields a total of five trees, 80% of which retain Placodus relatively closer to cyamodontoids than Paraplacodus); the percent- age of trees retaining the node in the bootstrap analysis is 69%. Obviously, the hypothesis of a paraphyletic Placodontoidea will have to be tested by including postcranial characters in the analysis. Conversely, the monophyly of the Placodontoidea currently is weakly supported. By contrast, the monophyly of the Cyamodon- toidea is highly supported; the node breaks only in a tree eight steps longer (TL = 111), the per- centage of trees retaining the node in the bootstrap analysis is 100%. The characters supporting monophyly of the Cyamodontoidea are: 1 * (2) [ci = 1], carapace present; 10(1) [ci = 1], pineal fo- ramen displaced anteriorly; 16* (2) [ci = 0.667], postfrontal broadly excluded from temporal fossa; 19* (1) [ci = 1], broad dorsal process of epiptery- goid; 20* (1) [ci = 1], epipterygoid sutured to palatine; 24* (1) [ci = 1] epipterygoid meets 68 FIELDIANA: GEOLOGY squamosal at dorsal margin of posttemporal fossa; 26* (1) [ci = 1], otic process of squamosal pre- sent; 27* (1) [ci = 1], palatoquadrate cartilage recess present; 30* (1) [ci = 1], pteroccipital fo- ramen present; 35* (2) [ci = 1], two maxillary teeth; 36* (2) [ci = 0.75], two palatine tooth plates; 37* (0) [ci = 1], anterior palatine tooth plates round; 53* (1) [ci = 1], retroarticular pro- cess short, with sloping surface; and 54* (2) [ci = 1], tubercular osteoderms secondarily fused to squamosal. ACCTRAN optimization adds 8 (1) [ci = 0.5]; 42 (0) [ci = 0.5]; 43 (1) [ci = 1]; 49 (1) [ci = 0.5]; and 52 (1) [ci = 0.5]. Within the Cyamodontoidea, Henodus groups with Cyamodus rostratus and Cyamodus kuhn- schnyderi (Fig. 32). The node is weakly support- ed: it breaks in a tree one step longer (TL = 104), and the percentage of trees retaining the node in the bootstrap analysis is less than 50%. Weak sup- port for this grouping results from the autapo- morphic structure of the Henodus skull, which re- sults in frequent noncomparable character states (coded as [?]), and from the fact, recognized ear- lier, that Henodus shares a number of similarities with the placochelyids (Rieppel & Zanon, 1997). Nevertheless, the grouping of Henodus with the other cyamodontoids from the Germanic Triassic is supported by the following characters: 8 (1) [ci = 0.5], the jugal does not extend far anteriorly along the ventral margin of the orbit; 1 2 ( 1 ) [ci = 0.5], parietal with distinct anterolateral process; 15* (1) [ci = 0.5], lateral margin of postfrontal deeply concave and angulated; 48* (1) [ci = 1], posteroventral tubercle present on distal end of paroccipial process; and 49 (1) [ci = 0.5], exoc- cipitals meet above the basioccipital condyle. ACCTRAN optimization adds 9 (1) [ci = 1]; 28 (1) [ci = 0.5]; 33 (2) [ci = 1]; 34 (0) [ci = 0.667]; and 43 (2) [ci = 1]. Cyamodus rostratus and Cyamodus kuhnschny- deri group as a well-supported monophyletic tax- on. The node breaks in a tree three steps longer (TL = 106), and the percentage of trees retaining the node in the bootstrap analysis is 97%. Char- acters diagnostic for the genus Cyamodus are: 7* (1) [ci = 0.5], maxilla floors external naris; 9(1) [ci = 1], jugal extends backward along the an- teromedial margin of subtemporal fossa (needs confirmation for Cyamodus kuhnschnyderi); 13* (1) [ci = 1], frontals extend posteriorly beyond the level of the anterior margin of the upper tem- poral fossa; 22* (1) [ci = 1], relatively broad up- per temporal fossa; 23* (1) [ci = 1], upper tem- poral fossa at least twice as long as orbit; 33 (2) Fig. 33. Most parsimonious reconstruction of pla- codont interrelationships, with Cyamodus hildegardis in- cluded (single most parsimonious tree with a TL of 109 steps, a CI of 0.752, and an RI of 0.693). For further discussion, see text. [ci = 1], premaxillary teeth bulbous, with trans- verse ridge; 43 (2) [ci = 1], jugal extends medi- ally to meet palatine at anterior margin of subtem- poral fossa; 45* 7(1) [ci = 1], pterygoid flange with double ventral projection; and 52 (1) [ci = 0.5], coronoid closely approaches the lower mar- gin of the mandible. ACCTRAN optimization adds 46 (1) [ci = 0.667]. These characters diagnose the genus Cyamodus without inclusion of Cyamodus hildegardis. Kuhn-Schnyder (1960: 96) expressed some doubt as to whether hildegardis should be included in the genus Cyamodus, and Nosotti and Pinna (1996, Fig. 24) show the genus to be paraphyletic, with Cyamodus hildegardis more closely related to the Protenodontosaurus-p\dcoche\yid clade than to the Germanic representatives of the genus. At least one character of this analysis would sup- port this conclusion — the relatively narrow upper temporal fossa in Cyamodus hildegardis (see Ta- ble 8). However, inclusion of Cyamodus hilde- gardis (coded for 35.2% of the characters) in a separate analysis yielded one single most parsi- monious tree with a TL of 109 steps, a CI of 0.752, and an RI of 0.693 (Fig. 33). Cyamodus hildegardis came out as the sister-taxon of Cy- amodus rostratus plus Cyamodus kuhnschnyderi on the basis of the following characters: 23* ( 1 ) [ci = 1], upper temporal fossa at least twice as long as orbit; 33 (2) [ci = 1], bulbous premaxil- lary teeth; and 52 (1) [ci = 0.5], coronoid closely approaching the ventral margin of the mandible. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 69 ACCTRAN optimization added 7 (1) [ci = 0.5]; 13 (1) [ci = 1]; 45 (1) [ci = 1]; and 46 (1) [ci = 0.667]. Cyamodus muensteri (coding based on the ho- lotype of C. laticeps Owen) could be coded for 46.3% of the characters used in this analysis. If this taxon was included but Cyamodus hildegardis deleted from the analysis, a total of six equally parsimonious trees would be obtained with a TL of 106 steps, a CI of 0.736, and an RI of 0.699. The strict consensus tree shows unresolved rela- tionships of Placodus and Paraplacodus relative to the Cyamodontoidea and unresolved relation- ships among the three species of Cyamodus in- cluded (C rostratus, C kuhnschnyderi, and C. muensteri). If both Cyamodus muensteri and Cy- amodus hildegardis are included in the analysis, a total of 30 equally parsimonious trees are ob- tained with a TL of 1 12 steps, a CI of 0.732, and an RI of 0.688. The strict consensus tree retains only two nodes, that is, the monophyly of the Cy- amodontoidea, and the sister-group relationship of Placochelys and Psephoderma within the latter. All other cyamodonoids fall into an unresolved polytomy. In a tree only one step longer than the most parsimonious reconstruction (TL = 104; Fig. 32, incomplete taxa not included), the node that re- lates Macroplacus to the Protenodontosaurus- placochelyid clade collapses (the percentage of trees retaining the node in the bootstrap analysis is 57%). The lability of the relationships of Ma- croplacus within the cyamodontoids is due to the fact that the skull, although rather incompletely preserved, shares important characters with Cy- amodus, as indicated in the morphological de- scription above. These are the short anterior ex- tent of the jugal along the ventral margin of the orbit (8 [1]), the relative shortness of the otic pro- cess of the squamosal (26 [1]), and the reduced posttemporal fossa (46 [1]). However, the most parsimonious reconstruction places Macroplacus as sister-taxon to Protenodontosaurus and the Placochelys-Psephoderma clade on the basis of the following characters: 18* (2) [ci = 0.5], the suture between the maxilla and the jugal below the midpoint of the orbit; 21* (1) [ci = 1], post- orbital forms medioventral process that abuts the epipterygoid; 39* (1) [ci = 1], maxilla with an- terolateral process tapering off along lateral mar- gin of the rostrum; 40* (1) [ci = 1], ventral sur- face of rostrum concave; and 43 (1) [ci = 1], pal- atine extends laterally to meet jugal. ACCTRAN optimization adds 4 (1) [ci = 0.5]; 5 (1) [ci = 0.5]; 12 (0) [ci = 0.5]; 14 (1) [ci = 0.667]; and 50 (1) [ci = 0.5]. Protenodontosaurus is the sister-taxon of the Placochelys-Psephoderma clade ("placoche- lyids") in the most parsimonious reconstruction, although the node is weakly supported. The node collapses in a tree one step longer (TL = 104 yields five trees, of which 80% retain Proteno- dontosaurus as sister-taxon of Placochelys and Psephoderma), and the percentage of trees retain- ing the node in the bootstrap analysis is a mere 52%. The sister-group relationship of Placochelys and Psephoderma is retained in 80% of the trees of the bootstrap analysis; but in a tree only three steps longer (TL = 106), Protenodontosaurus, Placochelys, and Psephoderma all fall into an un- resolved polytomy with Macroplacus and Heno- dus. The sister-group relationship of Protenodon- tosaurus with the placochelyids is supported by the following characters: 26* (2) [ci = 1], otic process of squamosal extends beyond the medial margin of the posttemporal fossa; 29* (1) [ci = 1], palatine contacts the quadrate lateral to pala- toquadrate cartilage recess; 31* (1) [ci = 1], pro- otic exposed in posterior view of skull. ACCT- RAN optimization adds 8 (0) [ci = 0.5]. Finally, Placochelys is found to be the sister- taxon of Psephoderma, a dichotomy that breaks in a tree three steps longer (TL = 106) but is supported by 80% of the trees of the bootstrap analysis. Several synapomorphies support a sister- group relationship of Placochelys and Psepho- derma: 2* (1) [ci = 1], skull depressed; 3* (1) [ci = 1], rostrum narrow and distinctly elongated; 18* (1) [ci = 0.5], suture between maxilla and jugal located below posterior half of the orbit; 32* (1) [ci = 1], premaxillary teeth absent; 41* (1) [ci =1], ventral surface of rostrum with distinct grooves; 44* (1) [ci = 0.5], palatal exposure of pterygoid relatively long; 47* (1) [ci = 1], squa- mosal buttress receives distal tip of paroccipital process; 51* (1) [ci = 0.5], anterior tip of dentary edentulous; 52 (1) [ci = 0.5], coronoid closely approaching ventral margin of mandible. ACCT- RAN optimization adds 5 (0) [ci = 0.5]; 7 (1) [ci = 0.5]; 49 (0) [ci = 0.5]. Jaekel (1907: 78; Pla- cochelidae), Nopcsa (1923: 12; Placochelynae), Nopcsa (1923: 172; Placochelyinae), and Peyer and Kuhn-Schnyder (1955: 475; Placochelydae) all introduced a higher yet monotypic category to include Placochelys only. Romer (1956: 670) was the first to propose a family Placochelyidae to in- clude Placochelys and Psephoderma, along with Psephosaurus (no cranial material known) and the 70 FIELDIANA: GEOLOGY Fig. 34. Strict consensus tree for placodont interre- lationships, with the Negev specimen included (four equally parsimonious trees with a TL of 104 steps, a CI of 0.75, and an RI of 0.701). For further discussion, see text. enigmatic Saurosphargis (see Rieppel, 1995b, for comments on the latter genus). An incomplete cyamodontoid skull known from the lower Muschelkalk of Makhtesh Ramon, Negev, Israel, was provisionally referred to IPse- phosaurus by Brotzen (1957). This specimen could be coded for 29.6% of the characters only. If it is entered into the analysis without inclusion of Cyamodus hildegardis, four equally parsimo- nious trees result with a TL of 104 steps, a CI of 0.75, and and RI of 0.701. The strict consensus tree shows the Negev cyamodontoid to fall into an unresolved polytomy with Macroplacus, Pro- tenodontosaurus, and the placochelyids (Fig. 34). Three of those four trees retain Protenodontosau- rus as sister-taxon of the placochelyids. If both the Negev specimen and Cyamodus hildegardis are entered into the analysis, a further drop in res- olution is observed, with six equally parsimonious trees of TL = 1 10 steps (CI = 0.745; RI = 0.689). The strict consensus tree shows the Israeli speci- men to fall into a polytomy with Macroplacus, Protenodontosaurus, the placochelyids, and the clade from the Germanic Triassic. Four of those six trees show the Israeli specimen in a trichoto- my with Macroplacus and the Protenodontosau- ras-placochelyid clade. Inclusion of the Negev taxon and Cyamodus muensteri, but excluding Cyamodus hildegardis, resulted in 24 equally par- simonious trees (TL = 107; CI = 0.729; RI = 0.695), with the three species of Cyamodus in an unresolved trichchotomy, while the Israeli speci- men falls into a polytomy with Macroplacus, Pro- tenodontosaurus, and the placochelyids. However, Henodus is retained as sister-group of Cyamodus, and the two are retained as sister-taxon of all other cyamodontoids. If the Negev specimen is includ- ed in the analysis along with Cyamodus hildegar- dis and Cyamodus muensteri, a total of 166 equal- ly parsimonious trees are obtained (TL = 1 13; CI = 0.726; RI = 0.648), the strict consensus tree of which retains only two nodes, the monophyly of the Cyamodontoidea and the sister-group relation- ship of Placochelys and Psephoderma. What little evidence there is suggests that the cyamodontoid from the Muschelkalk of Makhtesh Ramon is rel- atively more closely related to the taxa from the Alpine Triassic than to the clade from the Ger- manic Triassic. Systematic Paleontology Sauropterygia Owen, 1860 Placodontia Zittel, 1887-1890 Cyamodontoidea Nopcsa, 1923 definition — A monophyletic taxon including the Cyamodontida and the Placochelyida. diagnosis — Dermal armor forming a carapace; palatoquadrate cartilage persisting in adult in a re- cess between epipterygoid and quadrate; broad dorsal process of epipterygoid; epipterygoid su- tured primarily to palatine; epipterygoid meets squamosal at dorsal margin of posttemporal fossa; squamosal with otic process enclosing the pter- occipital foramen anteriorly; pineal foramen dis- placed anteriorly; postfrontal broadly excluded from temporal fossa; two maxillary tooth plates; two palatine tooth plates; anterior palatine tooth plates round; retroarticular process short, with sloping surface; tubercular osteoderms secondar- ily fused to squamosal. distribution — Middle to Upper Triassic, west- ern Tethyan faunal province (Europe and northern Gondwanan shelf)- comments — The lower (lower Anisian) and up- per (upper Anisian, lower Ladinian) Muschelkalk of the Germanic Triassic has yielded a number of Cyamodus skulls, skull fragments, or lower jaws (Giirich, 1884; Meyer, 1863), but only three car- apace fragments have been recorded from the up- per Muschelkalk (mo2, referred to Cyamodus kuhnschnyderi by Nosotti & Pinna, 1996), and RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 71 none has ever been collected in the lower (mu) or lower upper Muschelkalk (mol). As carapace fragments show a high probability of fossilization at other localities where cyamodontoid placodonts occur, it remains uncertain whether all represen- tatives of the genus Cyamodus (C. rostratus, C. muensteri, C. tarnowitzensis) in fact developed a carapace. This systematic section provides diagnoses of monophyletic taxa based primarily on cranial anatomy. Species diagnoses will be based on au- tapomorphies. Not included in this systematic sec- tion is Psephosaurus suevicus. Fraas, 1896, be- cause this taxon is known from its carapace only and therefore cannot be diagnosed on the basis of skull anatomy. A proper diagnosis for this taxon, along with the description of a new genus of cy- amodontoid placodont, will be presented in the context of a comprehensive review of the cyamo- dontoid dermal armor. The palatoquadrate cartilage recess is here cod- ed and treated as an autapomorphy of Cyamodon- toidea. This accounts for its complex structure, defined by the prootic, epipterygoid, quadrate, pterygoid, and palatine and including an anterior groove holding a persistent palatine process of the palatoquadrate. Palatoquadrate cartilage also per- sists in Placodus and cryptodire turtles, and its simple presence in the adult could also be treated as synapomorph at a more inclusive level. Cyamodontida, new taxon definition — A monophyletic taxon including the Cyamodontidae and the genus Henodus Hue- ne, 1936. diagnosis — Jugal not extending far anteriorly along the ventral margin of the orbit; parietal with distinct anterolateral process entering between frontal and postfrontal; lateral margin of postfron- tal deeply concave and angulated; paroccipial pro- cess with posteroventral tubercle at its distal end; exoccipitals in contact above the basioccipital condyle. distribution — Middle and Upper Triassic (low- er Anisian through Carnian), Germanic Basin and southern Alps. Henodus Huene, 1936 type species — Henodus chelyops Huene, 1936, from the upper Gipskeuper (Carnian) of Lustnau near Tubingen, southern Germany. definition — A monotypic taxon including the species chelyops. diagnosis — Skull broad and flat; rostrum short and broad; anterior cutting edge of rostrum (pre- maxillaries) lined by a series of incompletely in- dividualized denticles; maxillary without tooth plates but with deep grooves (possibly supporting baleens); palatine with single posterior tooth plate; upper temporal fenestra vestigial or absent; parietal broad and fan-shaped; cephalic condyle of quadrate posteriorly expanded and abutting a ventral flange of the squamosal; palatines sepa- rated from one another by broad vomers and pter- ygoids; dentary with deep groove and single pos- terior tooth plate; coronoid small, forming small coronoid process and remaining widely separated from lower margin of lower jaw. distribution — Upper Gipskeuper (lower Car- nian, Upper Triassic), southern Germany. Henodus chelyops Huene, 1936 1936 Henodus chelyops, Huene, pp. 99, \25jf., Figs. 1-23, Pis. 9-12. 1937 Henodus chelyops, Reiff, pp. 534, 536 ff. 1938 Henodus chelyops, Huene, p. 105 ff., Figs. 1-3, PI. 17. 1938 Henodus chelyops, Schmidt, p. 62/, Fig. 1152b. 1939 Henodus chelyops, Kuhn, p. 277. 1942 Henodus chelyops, Kuhn-Schnyder, p. 175. 1942 Henodus chelyops, Reiff, p. 31 ff., Figs. 1-5. 1946 Henodus chelyops, Gregory, p. 315, Fig. 31. 1947 Henodus chelyops, Vialli, p. 112. 1949 Henodus, E.v" Huene, p. 78, Fig. 3b. 1955 Henodus chelyops, Peyer and Kuhn- Schnyder, p. 477/, Figs. 20-21. 1956 Henodus chelyops, Huene, p. 341 ff, Figs. 420-422. 1958 Henodus chelyops, Huene, p. \65ff., Figs. 1-6, Pis. 3-4'. 1959 Henodus chelyops, Fischer, p. 24\ ff. 1961 Henodus chelyops, Kuhn, p. 20. 1963 Henodus, Kuhn-Schnyder, p. 74. 1965a Henodus, Kuhn-Schnyder, p. 278. 1965b Henodus, Kuhn-Schnyder, p. 153. 1967 Henodus chelyops, Westphal and West- phal, p. 252#, Figs. 5-6. 1968 Henodus chelyops, Muller, p. 205 ff., Figs. 239-242. 72 FIELDIANA: GEOLOGY 1969 Henodus chelvops, Kuhn, p. 17/, Fig. 6.6-6.12. 1975 Henodus chelyops, Westphal, p. 114j^, Figs. 12-13. 1976 Henodus chelyops, Westphal, p. 36 ff.. Figs. 5-6. 1978 Henodus chelyops, Jurcsak, p. 24. 1980a Henodus chelyops, Pinna, pp. 298, 302. 1980b Henodus chelyops, Pinna, p. 311. 1989 Henodus chelyops, Mazin, p. 729, Fig. 5B. 1990a Henodus chelyops, Pinna, p. 146, Fig. 1. 1990b Henodus chelyops, Pinna, p. 11. 1992 Henodus chelyops, Pinna, p. 20. 1993b Henodus chelyops, Nosotti and Pinna, p. 109. 1993 Henodus chelyops, Mazin and Pinna, p. 83# 1993 Henodus chelyops, Pinna and Mazin, p. 126, Fig. 1. 1995 Henodus chelyops, Stein, p. 35j/( Figs. 3-5. 1996 Henodus chelyops, Nosotti and Pinna, pp. 22, 36, 38, Fig. 24. 1997 Henodus chelyops, Rieppel and Zanon, pp. 202-213, Figs. 2E, 3E. 1999 Henodus chelyops, Pinna, p. 42 ff. lectotype — "Specimen II" of Huene, 1936. paratype — "Specimen I" of Huene, 1936. stratum typicum — Estherienschichten, upper- most Gipskeuper, lower Carnian, Upper Triassic. locus typicus — Lustnau near Tubingen, south- ern Germany. diagnosis — Same as for genus, of which this is the only known species. distribution — Same as for genus, of which this is the only known species. referred specimens — Specimens III through VIII (GPIT, uncatalogued; Huene, 1938; Fischer, 1959). Cyamodontidae Nopcsa, 1923 definition — A monophyletic taxon including the genus Cyamodus Meyer, 1863. diagnosis — Anterior end of maxilla extended medially to floor the external naris; jugal extend- ing backward along anteromedial margin of sub- temporal fossa; frontals extending posteriorly be- yond the level of the anterior margin of the upper temporal fossa; upper temporal fenestra relatively broad; upper temporal fenestra at least twice as long as orbit; premaxillary teeth bulbous, with transverse ridge; jugal extends medially to meet palatine at anterior margin of subtemporal fossa; pterygoid flange with double ventral projection; coronoid closely approaches the lower margin of mandible. distribution — Middle Triassic (Anisian, Ladi- nian), Germanic Basin and southern Alps. comments — The species Cyamodus tarnowitz- ensis Giirich, 1884, is based on an incompletely preserved and poorly illustrated skull from the Karchowice Beds (uppermost lower Muschelkalk (lower Illyrian, lower Anisian) of Tarnowitz in Upper Silesia (now Tarnowskie Gory, Poland), but this skull is now lost. Fragmentary cranial ma- terial of a cyamodontoid placodont from the lower Muschelkalk (lower Anisian) of Makhtesh Ra- mon, Negev, Israel, is not diagnostic at the genus level but has been thought to perhaps belong to the genus Cyamodus (Brotzen, 1957; Rieppel, Mazin, & Tchernov, 1999). If so, this occurrence would greatly increase the area of distribution of the genus Cyamodus. However, the cladistic anal- ysis reported above does not support inclusion of the Negev cyamodontoid in the genus Cyamodus. Cyamodus Meyer, 1863 1839 Placodus (partim), Miinster, p. 119. 1839 Placodus (partim), Agassiz, PI. 71, Figs. 1-5. 1844 Placodus (partim), Agassiz, p. 220. 1858 Placodus (partim), Owen, p. 169. 1862 Placodus (partim), Braun p. 8. type species — Cyamodus rostratus (Miinster, 1839), from the lower upper Muschelkalk (mol) of Bayreuth, southern Germany. definition — A monophyletic taxon including the species hildegardis, kuhnschnyderi, muensteri, and rostratus. diagnosis — As for family, of which this is the only known genus. distribution — As for family, of which this is the only known genus. Cyamodus hildegardis Peyer, 1931a 1 93 1 a Cyamodus hildegardis, Peyer, p. 6, Textfig. 5, Pis. 15, 16, 17, Fig. 1. 1933 Cyamodus hildegardis, Kuhn, p. 11/ 1935 Cyamodus hildegardis, Peyer, p. 20, PI. 46. 1936 Cyamodus miinsteri, Huene, pp. 129, 147. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONT A 73 1942 Cyamodus hildegardis, Kuhn-Schnyder, p. 175. 1944 Cyamodus hildegardis, Peyer, pp. 60-63, Figs. 61, 63. 1947 Cyamodus hildegardis, Vialli, p. Ill, Ta- ble 1. 1955 Cyamodus hildegardis, Peyer and Kuhn- Schnyder, p. 475, Figs. 15-16. 1956 Cyamodus hildegardis, Huene, p. 371. 1959 Cyamodus hildegardis, Kuhn-Schnyder, p. 174#, Fig. 1, PI. 1. 1960 Cyamodus hildegardis, Kuhn-Schnyder, p. 9\ff., Fig. 1. 1961 "Cyamodus" hildegardis, Kuhn, p. 19. 1965a Cyamodus hildegardis, Kuhn-Schnyder, pp. 260, 275. 1968 Cyamodus hildegardis, Muller, p. 199, Fig. 234. 1968 Cyamodus hildegardis, Peyer, p. 153, Fig. 98. 1969 "Cyamodus" hildegardis, Kuhn, p. 15. 1974 Cyamodus hildegardis, Kuhn-Schnyder, p. 69#, Fig. 47. 1975 Cyamodus hildegardis, Westphal, p. Ill/, 121, Fig. 10. 1976 Cyamodus hildegardis, Westphal, p. 38. 1979 Cyamodus hildegardis, Kuhn-Schnyder, p. 29, Fig. 23. 1979 "Cyamodus" hildegardis, Pinna, p. 200/ 1979 "Cyamodus" hildegardis, Pinna and Zuc- chi Stolfa, p. 311. 1980a Cyamodus hildegardis, Pinna, p. 275 ff., Figs. 1-6, Pis. 4-6. 1980b Cyamodus hildegardis, Pinna, p. 313. 1988 Cyamodus hildegardis, Westphal, p. 162, Fig. 11. 1989 Cyamodus hildegardis, Mazin, p. 727, Figs. 2C-E. 1989 Cyamodus hildegardis, Pinna, p. 150, 154/ 1989 Cyamodus hildegardis, Tschanz, p. 168, Fig. 8. 1990a Cyamodus hildegardis, Pinna, pp. 146, 149, 153, Fig. 1. 1 990b Cyamodus hildegardis, Pinna, p. 1 1 . 1990c Cyamodus hildegardis, Pinna, p. 138. 1992 Cyamodus hildegardis, Alafont, pp. 77, 81. 1992 Cyamodus hildegardis, Pinna, p. \ ff., Figs. 1-23. 1993 Cyamodus hildegardis, Dalla Vecchia, pp. 53, 55, Fig. 4A-B. 1993 Cyamodus hildegardis, Mazin and Pinna, p. 83 ff., Figs. 3, 4, 6a. 1993 Cyamodus hildegardis, Pinna and Mazin, p. 126, Fig. 1. 1993 Cyamodus hildegardis, Rieppel, p. 136. 1996 Cyamodus hildegardis, Nosotti and Pinna, p. 3ff, Fig. 24. 1997 Cyamodus hildegardis, Rieppel and Zan- on, p. 212. 1999 Cyamodus hildegardis, Pinna, p. 46 ff. holotype — pimuz T4763; articulated skeleton. stratum typicum — Grenzbitumen-horizon, An- isian-Ladinian boundary, Middle Triassic. locus typicus — Val Porina, Monte San Gior- gio, southern Switzerland. diagnosis — Two premaxillary teeth; three or four maxillary teeth; two or three palatine tooth plates; posterior palatine tooth plates elongate; body heavily armored with bipartite carapace (dorsal shield and tail shield). distribution — Middle Triassic (Anisian-Ladi- nian boundary), southern Alps. referred specimens — pimuz T58, articulated skeleton; T1285, postcranial fragment; T2796, subadult skull; T2797, juvenile skull; T2804, tooth-bearing elements as stomach content of Lar- iosaurus buzzii (Tschanz, 1989); T4764, postcra- nial remains; T4765, fragmentary mandible; T4766, mandibular teeth; T4767, palatine and dentary teeth; T4768, skull; T4770, partial skull; T4771, skull; T4772, isolated teeth, msnm V458, juvenile skeleton; V478, skull. comments — The diagnosis of this taxon is in- complete. Pinna (1999) considered Cyamodus hildegardis a probable synonym of Cyamodus la- ticeps, but because the latter name was recognized as a junior synonym of Cyamodus muensteri above, all three species would have to be synon- ymized. The dentition of Cyamodus muensteri falls into the range of variation of that of Cyamo- dus hildegardis. The latter taxon is known to have developed extensive dermal armor, which remains largely unknown for the other species of this ge- nus. Although this may be a diagnostic feature, the apparent lack of body armor in the species of Cyamodus from the Upper Muschelkalk of Bay- reuth (Cyamodus muensteri and Cyamodus ros- tratus) may also be due to the incompleteness of the fossil record. Cyamodus kuhnschnyderi Nosotti and Pinna, 1993 1928 Cyamodus sp., Berckhemer, p. xix, Fig. 1956 Cyamodus, Huene, p. 371, Fig. 415. 74 FIELDIANA: GEOLOGY 1959 Cyamodus sp., Kuhn-Schnyder, p. 184, Figs. 2c, 3b. 1960 Cyamodus sp., Kuhn-Schnyder, p. 96, Figs. 2, 4, 7b, PI. 7. 1969 Cyamodus, Kuhn, p. 16. 1974 Cyamodus sp. (Crailsheim), Kuhn-Schny- der, p. 70, Fig. 49. 1979 "Cyamodus" sp. (Crailsheim), Pinna and Zucchi Stolfa, p. 312. 1989 Cyamodus sp. (Crailsheim), Mazin, p. 728, Fig. 2B. 1990a "Cyamodus" sp., Pinna, p. 149, Fig. 1. 1990b Cyamodus sp. (Crailsheim), Pinna, p. 1 1. 1993 Cyamodus sp., Mazin and Pinna, p. 84. 1993a Cyamodus kuhn-schnyderi, Nosotti and Pinna, p. 847 ff., Figs. 1-2. 1993b Cyamodus sp., Nosotti and Pinna, pp. 109, 112, Fig. 3. 1993 Cyamodus Khunschnyderi, Pinna and Ma- zin, Fig. 1. 1994b Cyamodus kuhnschnyderi, Rieppel, p. 42. 1996 Cyamodus kuhnschnyderi, Nosotti and Pinna, p. \ff., Figs. 1-19, 21-23. 1997 Cyamodus kuhn-schnyderi, Rieppel and Zanon, p. 212, Figs. 1C, 2B, 3B. 1999 Cyamodus kuhnschnyderi, Pinna, p. 26/ holotype — smns 15855; skull. paratypes — smns 16270, skull; 18380, partial lower jaw. stratum typicum — Hohenlohe Subformation of the Meissner Formation, Discoceratitenschi- chten, upper Muschelkalk (mo2), lower Ladinian, Middle Triassic. locus typicus — Crailsheim, southern Germany. diagnosis — Nasals fused; anterolateral process of frontal absent; anteromedial process of the pa- rietal embraced by the frontal; a basiorbital furrow present (also in Cyamodus "laticeps" and Heno- dus), with three foramina (two in Cyamodus "la- ticeps" and Henodus); the epipterygoid is incom- pletely ossified in the adult (may occur conver- gently in Protenodontosaurus). distribution — Upper Muschelkalk (mo2, lower Ladinian), southern Germany. REFERRED SPECIMENS — SMNS 15891c, 16725, 81600; carapace fragments. Cyamodus muensteri (Agassiz, 1839) 1830 no name, Munster, p. 3, Fig. 2. 1839 Placodus Munsteri, Agassiz, Vol. 2, PI. 71, Figs. 1-5. 1839 Placodus Munsteri, Munster, p. 120. 1840 Placodus munsteri, Braun, p. 120 {fide Pinna, 1999). 1843 Placodus Munsteri, Munster, p. 126 (fide Pinna, 1999). 1844 Placodus Munsteri, Agassiz, Vol. 2, p. 220/ 1858 Placodus laticeps, Owen, p. 169, PI. 9, Figs. 1-2, PI. 10, Fig. 1. 1861 Placodus laticeps, Meyer, p. 57. 1861 Placodus Munsteri, Meyer, p. 57. 1862 Placodus laticeps, Braun, p. 9. 1862 Placodus Munsteri, Braun, p. 8. 1863 Cyamodus laticeps, Meyer, pp. 179, 2\9ff. 1863 Cyamodus Munsteri, Meyer, pp. 179, 2\5 jf., PI. 31, Figs. 1-2. 1884 Cyamodus laticeps, Giirich, p. 140. 1884 Cyamodus Munsteri, Giirich, p. 139. 1890 Cyamodus laticeps, Lydekker, p. If., Fig. 1. 1890 Cyamodus muensteri, Lydekker, p. 7. 1922 Cyamodus laticeps, Drevermann, p. 100. 1928 Cyamodus Munsteri, Corroy, p. 125. 1928 Cyamodus laticeps, Drevermann, p. 292/ 1928 Cyamodus Munsteri, Drevermann, p. 292/ 1928 Cyamodus Munsteri, M. Schmidt, p. 410/, Fig. 1151. 1931a Cyamodus laticeps, Peyer, p. 5. 1931a Cyamodus munsteri, Peyer, p. 5. 1931b Cyamodus laticeps, Peyer, p. 274. 1931b Cyamodus munsteri, Peyer, p. 274. 1933 Cyamodus munsteri, Kuhn, p. 1 1. 1936 Cyamodus munsteri, Huene, pp. 129, 147. 1947 Cyamodus munsteri, Vialli, Table 1. 1956 Cyamodus laticeps, Huene, p. 371. 1956 Cyamodus munsteri, Huene, p. 371. 1959 Cyamodus laticeps, Kuhn-Schnyder, p. 184#, Fig. 2b. 1959 Cyamodus munsteri, Kuhn-Schnyder, p. 184/ 1960 Cyamodus laticeps, Kuhn-Schnyder, p. 92. 1960 Cyamodus munsteri, Kuhn-Schnyder, p. 92. 1961 Cyamodus munsteri, Kuhn, p. 19. 1965a Cyamodus Munsteri, Kuhn-Schnyder, p. 259. 1969 "Cyamodus"' laticeps, Kuhn, p. 15. 1969 "Cyamodus" muensteri, Kuhn, p. 15. 1979 "Cyamodus" laticeps, Pinna and Zucchi Stolfa, p. 312. 1979 "Cyamodus" munsteri. Pinna and Zucchi Stolfa, p. 312. 1980a Cyamodus laticeps, Pinna, p. 278. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 75 1980a Cyamodus muensteri, Pinna, p. 278. 1988 Cyamodus laticeps, Westphal, p. 159. 1988 Cyamodus muensteri, Westphal, p. 159. 1989 Cyamodus laticeps, Mazin, p. 733. 1989 Cyamodus muensteri, Mazin, p. 733. 1989 Placodus laticeps, Nosotti and Pinna, p. 42, Fig. 8, PI. 4, Fig. 1. 1989 Placodus miinsteri, Nosotti and Pinna, p. 37, Fig. 4. 1989 Cyamodus laticeps, Nosotti and Pinna, pp. 67, 82. 1989 Cyamodus miinsteri, Nosotti and Pinna, pp. 67, 82. 1990a Cyamodus miinsteri, Pinna, p. 149, Fig 1. 1990b Cyamodus laticeps, Pinna, p. 11, Table 1. 1992 Cyamodus muensteri, Pinna, p. 12. 1993 Cyamodus laticeps, Mazin and Pinna, p. 84. 1993 Cyamodus muensteri, Mazin and Pinna, p. 84. 1993 Cyamodus muensteri, Pinna and Mazin, p. 126, Fig. 1. 1996 Cyamodus laticeps, Nosotti and Pinna, p. 36, Fig. 24. 1996 Cyamodus muensteri, Nosotti and Pinna, p. 3, Fig. 24. 1997 Cyamodus miinsteri, Rieppel and Zanon, p. 212. 1999 Cyamodus laticeps, Pinna, p. 45/ 1999 Cyamodus muensteri, Pinna, p. 22 ff. holotype — bsp AS VII 1210; incomplete skull. stratum typicum — Trochitenkalk Formation and the lower part of the Meissner Formation (atavus through postspinosus biozone), upper Muschelkalk (mol), uppermost Illyrian, upper Anisian, Middle Triassic. locus typicus — Bayreuth, southern Germany. comments — As discussed above, the skull of Cyamodus muensteri is incomplete and heavily reconstructed. The skull of Cyamodus laticeps Owen, 1858, from the same locality is more com- plete, somewhat larger than that of C. rostratus, and distinctly larger than that of C. muensteri. Ac- cording to Kuhn (1933), Meyer (1863) considered C laticeps a junior synonym of C muensteri, but no such indication can be found in Meyer's (1863) text. Drevermann ( 1 928) was the first to claim that C. laticeps and C. tarnowitzensis are junior syn- onyms of C. muensteri, an arrangement that was also accepted by Huene (1956) for C. laticeps and C. muensteri. Table 5 summarizes the dentitional characters of the species of the genus Cyamodus and shows that the presence of three palatine tooth plates, located entirely behind the posteriormost maxillary tooth, is autapomorphic for C. rostra- tus. All other species have two palatine tooth plates, of which the anterior one is located lateral to and slightly behind the posterior maxillary tooth. The posterior maxillary and anterior pala- tine teeth thus are aligned in an anteriorly slightly concave arch. C. muensteri, C. laticeps, and C tarnowitzensis differ from C. kuhnschnyderi by the presence of three maxillary teeth and by a slightly (C. muensteri, C. laticeps) or distinctly (C. tarnowitzensis) elongated posterior palatine tooth plate. However, C. hildegardis also has three maxillary teeth (four in one specimen) and elongated posterior palatine tooth plates. Whereas similarities in the dentition corroborate Drever- mann's (1928) conclusion that C. laticeps is a ju- nior synonym of C. muensteri in spite of size dif- ference, comparison with C. hildegardis indicates that the incomplete and poorly preserved skulls of C. muensteri do not permit a diagnosis of the spe- cies on the basis of dentitional characters. The skull of C. laticeps is unique among its genus in that the posterior (nasal) processes of the premax- illae meet the frontal, thereby separating the na- sals from one another; this character is unknown for the holotype of C. muensteri. It may therefore be prudent to treat this species as a metaspecies sensu Gauthier, Estes, and DeQueiroz (1988; see also Archibald, 1994). The synonymy of C. tarnowitzensis is less eas- ily resolved. The incomplete skull came from a different horizon and a different locality (lower Muschelkalk, Upper Silesia) than C. muensteri (upper Muschelkalk, Bayreuth, southern Germa- ny) and can no longer be located today. The il- lustrations provided by Giirich (1884) are sche- matic and not useful in establishing species valid- ity or synonymy. For these reasons, the species tarnowitzensis is here treated as nomen dubium. Cyamodus rostratus (Miinster, 1839) 1839 Placodus rostratus, Miinster, p. 119, PI. 15, Figs. 1-6. 1839 Placodus rostratus, Agassiz, Vol. 2, PI. 71, Figs. 6-12. 1840 Placodus rostratus, Braun, p. 74 (fide Pinna, 1999). 1843 Placodus rostratus, Miinster, p. 126 (fide Pinna, 1999). 1844 Placodus rostratus, Agassiz, Vol. 2, p. 221. 76 FIELDIANA: GEOLOGY 1856 Placodus rostratus, Owen, p. 170, PI. 11, Fig. 4. 1861 Placodus rostratus, Meyer, p. 57. 1862 Placodus rostratus, Braun, p. 8. 1863 Placodus rostratus, Braun, p. 10. 1863 Cyamodus rostratus, Meyer, pp. 179, 21 \ff., PI. 23, Figs. 1-2. 1890 Cyamodus rostratus, Lydekker, p. 7. 1922 Cyamodus rostratus, Drevermann, p. 100. 1928 Cyamodus rostratus, Berckhemer, p. xx. 1928 Cyamodus rostratus, Corroy, p. 125. 1928 Cyamodus rostratus, Drevermann, p. 291 ff., Figs. 1-2, PI. 23, Figs. la-e. 1928 Cyamodus rostratus, M. Schmidt, p. 410, Fig. 1150. 1931a Cyamodus rostratus, Peyer, p. 5. 1933 Cyamodus rostratus, Kuhn, p. 10/ 1936 Cyamodus rostratus, Huene, pp. 110, 129, 147. 1947 Cyamodus rostratus, Vialli, p. 112, Ta- ble 1. 1955 Cyamodus rostratus, Peyer and Kuhn- Schnyder, p. 474, Figs. 12-14. 1956 Cyamodus rostratus, Huene, p. 371, Fig. 415. 1959 Cyamodus rostratus, Kuhn-Schnyder, p. 184# 1960 Cyamodus rostratus, Gorce, p. 23. 1960 Cyamodus rostratus, Kuhn-Schnyder, p. 92#, Figs. 3, 4, 5a, 6a, 7a. 1961 Cyamodus rostratus, Kuhn, p. 19. 1961 Cyamodus rostratus, Kuhn-Schnyder, p. 107, Fig. 7b. 1965a Cyamodus rostratus, Kuhn-Schnyder, p. 257 ff, Figs. 1-6, Pis. 116-118. 1968 Cyamodus rostratus, Miiller, Figs. 232- 233. 1969 Cyamodus rostratus, Kuhn, p. 14/., Fig. 7. 1974 Cyamodus rostratus, Kuhn-Schnyder, p. 70/, Fig. 50. 1975 Cyamodus rostratus, Schubert-Klemp- nauer, p. 50, Figs. 7A, 8A. 1976 Cyamodus rostratus, Jurcsak, p. 75. 1979 Cyamodus rostratus, Pinna and Zucchi Stolfa, p. 312^ 1987 Cyamodus rostratus, Sues, p. 143. 1988 Cyamodus rostratus, Westphal, p. 159, Fig. 10. 1989 Cyamodus rostratus, Mazin, p. 727, Figs. 2A. 1989 Placodus rostratus, Nosotti and Pinna, p. 37, Fig. 4, PI. 6. 1989 Cyamodus rostratus, Nosotti and Pinna, pp. 67, 82, Fig. 14.1. 1990a Cyamodus rostratus, Pinna, p. 149, Fig. 1. 1990b Cyamodus rostratus. Pinna, p. 11, Table 1. 1992 Cyamodus rostratus, Alafont, p. 78. 1993 Cyamodus rostratus, Dalla Vecchia, p. 55, Fig. 4C. 1993 Cxamodus rostratus, Mazin and Pinna, p. 84. 1993b Cyamodus rostratus, Nosotti and Pinna, pp. 109, 112. 1993b Cyamodus cfr. rostratus, Nosotti and Pin- na, Figs. 2A, 4. 1993 Cyamodus rostratus, Mazin and Pinna, p. 84. 1993 Cyamodus rostratus, Pinna and Mazin, p. 126, Fig. 1. 1995a Cyamodus rostratus, Rieppel, p. 36. 1996 Cyamodus rostatus, Nosotti and Pinna, p. Iff, Fig. 24. 1996 Cyamodus cfr. rostatus, Nosotti and Pin- na, Fig. 20. 1997 Cyamodus rostatus, Rieppel and Zanon, p. 212. 1999 Cyamodus rostratus. Pinna, p. 24 ff. holotype — umo BT 748; skull. stratum typicum — Trochitenkalk Formation and the lower part of the Meissner Formation (atavus through postspinosus biozone), upper Muschelkalk (mol), uppermost Illyrian, upper Anisian, Middle Triassic. locus typicus — Bayreuth, southern Germany. diagnosis — Grooves on the postorbital part of each frontal converging toward the pineal fora- men; maxilla extends backward in lateral view to level well behind the posterior margin of orbit; jugal excluded from the orbital margin in lateral view; anterior palatal process of jugal entering deeply between maxilla and palatine; jugal ex- tending backward along the anteromedial margin of the subtemporal fossa (possibly also present in Cyamodus kuhnschnyderi); tip of the posterior dorsal process of the epipterygoid exposed at the dorsomedial corner of the posttemporal fossa in occipital view; occiput with a separate, heterotro- phic "epiotic" ossification located between supra- occipital and opisthotic; three palatine tooth plates, located entirely behind the two maxillary tooth plates. distribution — Upper Muschelkalk (upper An- isian), southern Germany. referred specimens — smns 17403; incomplete RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 11 skull, umo BT 2172; isolated lower jaw. smf R- 4040; isolated lower jaw. Placochelyida, new taxon definition — A monophyletic taxon including the genera Macroplacus Schubert-Klempnauer, 1975, Protenodontosaurus Pinna, 1990b, and the Placochelyidae. diagnosis — Vertical suture between maxilla and jugal below midpoint of orbit; postorbital with medioventral process abutting the epiptery- goid; maxilla with anterolateral process tapering off along lateral margin of rostrum; ventral sur- face of rostrum concave; palatine extends laterally to meet jugal. distribution — Upper Triassic, central Europe and southern Alps. Macroplacus Schubert-Klempnauer, 1975 type SPECIES — Macroplacus raeticus Schubert- Klempnauer, 1975, from the Rhaetian of the Ba- varian Alps. definition — A monotypic taxon including the species raeticus. diagnosis — Posterior palatine tooth plates hy- pertrophied; posterior (nasal) processes of the pre- maxillaries enlarged and extending backward to reach the frontal, thereby separating the nasals from one another (convergent in Psephoderma); posttemporal fossae greatly reduced; a foramen piercing the shaft of the quadrate just above the mandibular condyle. distribution — Rhaetian (Upper Triassic) of the Bavarian Alps. Macroplacus raeticus Schubert-Klempnauer, 1975 1975 Macroplacus raeticus nauer, p. 35, Figs. 1-6. 1976a Macroplacus raeticus, 1976b Macroplacus raeticus, 1977 Macroplacus raeticus, 1978 Psephoderma raeticus, 1989 Macroplacus raeticus, 1 989 Macroplacus raeticus 154, Fig. 5. 1990a Macroplacus raeticus, Schubert-Klemp- , 7C, 8C, Pis. 4-5. Pinna, pp. 1 1, 35/ Pinna, p. 107. Jurcsak, p. 14. Pinna, p. 350. Mazin, p. 728. Pinna, pp. 150, Pinna, p. 149. 1990a Psephoderma raeticus, Pinna, p. 149, Fig. 1. 1992 Psephoderma raeticus, Alafont, p. 75. 1993 Psephoderma raeticus, Mazin and Pinna, p. 84. 1993 Macroplacus raeticus, Pinna and Mazin, p. 126, Fig. 1. 1993b Macroplacus raeticus, Nosotti and Pinna, pp. 110, 112. 1995a Macroplacus raeticus, Rieppel, p. 17, Fig. 22. 1996 Macroplacus raeticus, Nosotti and Pinna, p. 32/ 1997 Macroplacus raeticus, Rieppel and Zanon, p. 215. 1999 Psephoderma raeticum, Pinna, p. 40. holotype — bsp 1967 I 324; skull. stratum typicum — Kossener Schichten, Rhae- tian, Upper Triassic. locus typicus — Hinterstein (Sonthofen), All- gau, Bavaria, Germany. diagnosis — Same as for genus, of which this is the only known species. distribution — Same as for genus, of which this is the only known species. REFERRED SPECIMENS None. Unnamed Taxon definition — A monophyletic taxon including Protenodontosaurus Pinna, 1990b, and the Pla- cochelyidae. diagnosis — Otic process of squamosal extends beyond the medial margin of the posttemporal fossa; palatine contacts quadrate lateral to pala- toquadrate cartilage recess; prootic exposed in posterior view of skull. distribution — Upper Triassic, central Europe and southern Alps. Protenodontosaurus Pinna, 1990b type species — Protenodontosaurus italicus Pin- na, 1990b, from the Carnian of Dogna, Udine, It- aly. definition — A monotypic taxon including the species italicus. diagnosis — Single (posterior) maxillary tooth, separated by a wide diastema from premaxillary tooth (teeth?); maxilla almost as high as it is long in lateral view because of distinct ascending pro- 78 FIELDIANA: GEOLOGY cess; prefrontal not extending far down along the anterior margin of the orbit; orbital margin of frontal rather straight, postorbital not extending beyond the midpoint of the upper temporal fossa along its lateral margin; vomers much enlarged and reaching far anteriorly into rostrum; exoccip- itals meet dorsal to the occipital condyle (conver- gent in Cyamodus); basioccipital tuber and ventral flange of opisthotic meeting each other ventral to the passage of the internal carotid, and meeting the posterior margin of the basicranium (conver- gent in Placochelys). distribution — Carnian (Upper Triassic) of the southeastern Alps (northeastern Italy). Placochelyidae Romer, 1956 definition — A monophyletic taxon including the genera Placochelys Jaekel, 1902, and Pse- phoderma Meyer, 1858. diagnosis — Skull depressed; rostrum narrow and distinctly elongated; suture between maxilla and jugal located below posterior half of orbit; premaxillary teeth absent; ventral surface of ros- trum with distinct grooves leading to internal na- res; palatal exposure of pterygoid relatively long; distal tip of paroccipital process abuts squamosal buttress; anterior tip of dentary edentulous. distribution — Upper Triassic, central Europe and southern Alps. Protenodontosaurus italicus Pinna, 1990b 1990b Protenodontosaurus italicus, Pinna, p. 6, Figs. 1-4. 1990a Protenodontosaurus italicus, Pinna, p. 151. 1992 Protenodontosaurus italicus, Pinna, p. 12. 1993 Protenodontosaurus italicus, Dalla Vec- chia, pp. 51, 53, Fig. 5C. 1993 Protenodontosaurus italicus, Mazin and Pinna, p. 84. 1993b Protenodontosaurus italicus, Nosotti and Pinna, pp. 110, 112, Fig. 2B. 1993 Protenodontosaurus italicus. Pinna and Mazin, p. 126, Fig. 1. 1994 Protenodontosaurus italicus, Sirna et al., p. 262. 1996 Protenodontosaurus italicus, Nosotti and Pinna, p. 22j^( Fig. 24. 1997 Protenodontosaurus italicus, Rieppel and Zanon, p. 213. 1998 Protenodontosaurus italicus, Nosotti and Pinna, p. 1 ff. 1999 Protenodontosaurus italicus. Pinna, p. 28/ holotype — mfsn 1819GP; skull. stratum typicum — Carnian, Upper Triassic. locus typicus — Chiout Zuguin near Dogna, Province of Udine, Tre Venezie area, Italy. diagnosis — Same as for genus, of which this is the only known species. distribution — Same as for genus, of which this is the only known species. referred specimens — mfsn 1923GP; skull. Placochelys Jaekel, 1902 type species — Placochelys placodonta Jaekel, 1 902, from the Carnian of Veszprem, Hungary. definition — A monotypic taxon including the species placodonta. diagnosis — Three maxillary teeth; posterior process of premaxilla extending backward beyond the level of the anterior margin of the orbit; fron- tal narrowly exposed between the posterior parts of nasals; anterolateral processes of frontal weak- ly developed; maxilla flooring the external naris (convergent in Cyamodus); squamosal and quad- ratojugal extending anteriorly within the temporal arch to a level in front of the anterior margin of the upper temporal fossa; foramen piercing the suspensorium between the quadrate, squamosal, and quadratojugal; basioccipital tuber and the ventral flange of the opisthotic meet ventral to the passage of the internal carotid, and meet the pos- terior margin of the basicranium (convergent in Protenodontosaurus, unknown in Psephoderma). distribution — Carnian (Upper Triassic) of cen- tral Europe (Hungary). Placochelys placodonta Jaekel, 1902 1901 Placochelys, Jaekel, p. 57. 1902a Placochelys placodonta, Jaekel, p. 4, Figs. 1-5 {fide Pinna, 1999). 1902b Placochelys placodonta, Jaekel, p. 127, Fig. on p. 130. 1903 Placochelys placodonta, Freeh, p. 17, PI. XV. 1906 Placochelys placodonta, Arthaber, p. 428. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 79 1907 Placochelys placodonta, Jaekel, p. \jf., Pis. I— III, V-X. 1910 Placochelys placodonta, Jaekel, p. 335, Fig. 4. 1911 Placochelys placodonta, Huene, p. 45, Fig. 51. 1914 Placochelys placodonta, Jaekel, p. 210. 1915 Placochelys, Drevermann, p. 403 ff. 1921 Placochelys placodonta, Broili, p. 316. 1922 Placochelys placodonta, Broom, Fig. 2C. 1922 Placochelys, Drevermann, p. 292 ff. 1 924 Placochelys placodonta, Broom, p. 54, Figs. 4a, 5a. 1928 Placochelys, Drevermann, p. 98 ff. 1930 Placochelys placodonta, Goodrich, p. 333, Fig. 345. 1931 Placochelys placodonta, Huene, p. 7, PI. I— III. 1933 Placochelys placodonta, Kuhn, p. 12. 1935 Placochelys placodonta, Peyer, p. 21. 1936 Placochelys placodonta Huene, pp. 123, 128, 147. 1947 Placochelys placodonta, Boni, p. 2. 1 947 Placochelys placodonta, Vialli, p. Ill, Table 1. 1 955 Placochelys placodonta, Peyer and Kuhn-Schnyder, p. 476/, Figs. 17-18. 1956 Placochelys placodonta, Huene, p. 373, Figs. 416-417. 1956 Placochelys placodonta, Sacchi Vialli, p. 96. 1957 Placochelys placodonta, Brotzen, p. 167/ 1958 Placochelys placodonta, Huene, p. 210. 1 959 Placochelys placodonta, Kuhn-Schnyder, p. 186. 1960 Placochelys placodonta, Gorce, p. 23. 1 960 Placochelys placodonta, Kuhn-Schnyder, p. 96, Figs. 5b, 6b. 1961 Placochelys placodonta, Kuhn, p. 19. 1 96 1 Placochelys placodonta, Kuhn-Schnyder, p. 107, Fig. 7b. 1 963 Placochelys placodonta, Kuhn-Schnyder, Figs, lb, 2b, 3b, 4. 1965a Placochelys placodonta, Kuhn-Schnyder, p. 257# 1 965b Placochelys placodonta, Kuhn-Schnyder, Fig. 8. 1967 Placochelys placodonta, Haas, p. 332. 1967 Placochelys placodonta, Kuhn-Schnyder, Figs. 7, 11. 1968 Placochelys placodonta, Miiller, p. 199, Figs. 235-237. 1969 Placochelys placodonta, Kuhn, p. 15/, Figs. 3, 6. 1975 Placochelys placodonta, Haas, p. 452. 1 975 Placochelys placodonta, Schubert-Klemp- nauer, pp. 50, 53, Figs. 7B, 8B. 1975 Placochelys placodonta, Westphal, p. 113, 121, Fig. 11. 1976a Placochelys placodonta, Pinna, pp. 10, 36# 1976 Placochelys placodonta, Westphal, p. 35, Fig. 3C. 1977 Placochelys placodonta, Jurcsak, p. 8. 1978 Placochelys placodonta, Jurcsak, p. 24. 1978 Placochelys placodonta, Pinna, p. 347 ff. 1979 Placochelys placodonta, Pinna and Zuc- chi Stolfa, p. 307 ff, PI. 13. 1980 Placochelys, Kuhn-Schnyder, p. 164, Fig. 9.6. 1980a Placochelys placodonta, Pinna, p. 298 ff. 1986 Placochelys placodonta, Nosottti, Fig. 2. 1987 Placochelys, Sues, p. 141. 1989 Placochelys placodonta, Mazin, p. 728, Fig. 3. 1989 Placochelys placodonta, Nosotti and Pin- na, pp. 50, 83, Pis. XIII-XIV. 1989 Placochelys placodonta, Pinna, pp. 141, 150, 153, Fig. 4. 1990a Placochelys placodonta, Pinna, p. 150/, Fig. 1. 1990b Placochelys placodonta, Pinna, p. 11, Ta- ble 1. 1990c Placochelys placodonta, Pinna, p. \AQff. 1992 Placochelys placodonta, Pinna, p. \2ff. 1993 Placochelys placodonta, Dalla Vecchia, pp. 51, 55, Fig. 5 A. 1993 Placochelys placodonta, Mazin and Pin- na, p. 83 ff., Figs. 7-9. 1993b Placochelys placodonta, Nosotti and Pin- na, p. 109. 1993 Placochelys placodonta, Pinna and Ma- zin, p. 126, Fig. 1. 1994 Placochelys placodonta, Sirna et al., p. 262. 1995a Placochelys placodonta, Rieppel, p. 37, Fig. 47. 1 996 Placochelys placodonta, Nosotti and Pin- na, p. Aff., Fig. 24. 1 997 Placochelys placodonta, Rieppel and Zanon, p. 215, Figs. ID, 2C, 3C. 1999 Placochelys placodonta, Pinna, p. 30 ff. holotype — fafi Ob/2323/Vt.3; skull and post- cranial remains. stratum typicum — Physiocardia-beds (Artha- ber, 1906), Also Keuper (equivalent to the Raibler Schichten), Carnian, Upper Triassic. 80 FIELDIANA: GEOLOGY Fig. 35. Scale bar = Skull of Placochelys alpis sordidae Broili (holotype, bsp 1921.1.3): A, dorsal view; B, ventral view. 20 mm. locus typicus — Jeruzsalemhegy (Jerusalem Mountain) near Veszprem, Bakony Mountains, Hungary. diagnosis — Same as for genus, of which this is the only known species. distribution — Same as for genus, of which this is the only known species. REFERRED MATERIAL — MB.R.1765; skull. comments — Placochelys placodonta Jaekel, 1902, is a monotypic genus represented exclu- sively by the material collected in the Keuper near Veszprem, Hungary. No other vertebrate fossils were ever collected in the same strata. Nonethe- less, a number of placodont remains from the Al- pine Triassic have been referred to the genus Pla- cochelys by various authors, mostly on the basis of stratigraphic considerations. Placochelys alpis sordidae Broili, 1921, is based on an incomplete skull (bsp 1921.1.3; Figs. 35, 36) from the Rhaetian Koessen-Formation of pot P1 f.p.dl Fig. 36. Skull of Placochelys alpis sordidae Broili (holotype, bsp 1921.1.3): A, dorsal view; B, ventral view. Scale bar = 20 mm. For a list of abbreviations see p. 3. RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 81 the Bavarian Alps (Kothalpe am Wendelstein), which Broili (1921) believed to be of equivalent age as the Keuper of Veszprem. The skull is dis- tinctly smaller than the holotype of Placochelys placodonta and is very incomplete, for which rea- son it is not diagnostic at the species level, nor is its referral to the genus Placochelys unequivocal, as was already noted by Broili (1921). The skull is transversely broken through the middle of the orbits, and the anterior part is missing. The max- imal length of the preserved (posterior) part of the skull measures 54 mm. The right side of the skull shows the postorbital arch in lateral and dorsal view (Figs. 35A, 36A). Along the posterior mar- gin of the orbit, the postfrontal can be demarcated from the postorbital, and posteroventrally the postorbital can be delineated from the jugal. Un- fortunately, it is impossible to ascertain whether the posterolateral margin of the postfrontal was deeply concave and angulated (as in Placochelys) or only weakly concave and smoothly curved (as in Psephoderma). Likewise, it is impossible to es- tablish whether the postfrontal remained broadly separated from the anteromedial margin of the up- per temporal fossa or narrowly approached it. The right temporal arch is complete but broken and distorted, as is the right upper temporal fossa. Little sutural detail can be made out, but the pos- terior tip of the postorbital is clearly indicated along the medial margin of the temporal arch at a level behind the midpoint of the longitudinal diameter of the upper temporal fossa. This is an important character shared by Placochelys and Psephoderma (see below for further discussion). The skull table is of rather square or weakly rectangular contours as in Placochelys, without the constriction in its posterior part that is ob- served in Psephoderma. It shows dermal encrus- tations similar to those of Placochelys, that is, two protuberances along each lateral side of the skull table and a posteromedial one in front of which there is another medial protuberance (Figs. 35A, 36A). The lateral walls of the braincase have under- gone breakage and distortion, but the contact of the palatine with the quadrate lateral to the pala- toquadrate cartilage recess is distinct on the right side (Figs. 35A, 36A). The left side of the skull shows the tilted epipterygoid in lateral view, with an indistinct depression, possibly representing the trigeminal incisure, separating the anterior epi- pterygoid from the posterior prootic. The occiput is incompletely preserved. The only identifiable elements are the exoccipitals, which form vertical struts indicating the lateral margins of the foramen magnum. As preserved, the exoccipitals are wide- ly separated from one another. There is certainly no evidence that they may have met above the occipital condyle (basioccipital). Splints of bone located below the posterior margin of the skull table may represent parts of the supraoccipital. The ventral view of the skull (Figs. 35B, 36B) displays the palatine and pterygoid bones along with the palatal dentition. Sutural details are dif- ficult to determine. The ventral margin of the right temporal arch shows the posterior tip of the jugal, which appears to meet the quadratojugal (?) at an unusually posterior level. The mandibular condyle of the right quadrate is somewhat eroded, but in front of it the suture separating the anteromedial wing of the quadrate from the quadrate ramus of the pterygoid can clearly be identified. The lon- gitudinally oriented flange of the right pterygoid is distinct but deflected toward the midline of the skull, and the anterior tip of the right pterygoid is distinct, located at the level of the posterior third of the longitudinal diameter of the posterior pal- atine tooth plate. There is no indication of the presence of an ectopterygoid bone. The medio- ventral suture between the palatines is distinct; that between the pterygoids is partially obscured. The posterior dental lamina foramina are distinct and located posteromedial to the posterior palatine tooth plates. By comparison with other cyamo- dontoids, the position of the posterior dental lam- ina foramina also indicates the position of the transverse suture between palatines and ptery- goids. This in turn allows the estimation of the relative length of the palatal exposure of palatine and pterygoid. Dividing the distance from the posterior margin of the dermal palate to the pos- terior margin of the posterior dental lamina foram- ina (10/10.5 mm) by the length of the palatine (28.2/27.5 mm) yields a quotient of approximately 0.37. As discussed below, a quotient larger than 0.3 is synapomorphic for Placochelys and Pse- phoderma. Each palatine preserves a small anterior and a much larger posterior palatine tooth plate (Figs. 35B, 36B). The left palatine tooth plate is not completely exposed; its posterolateral part is cov- ered by bone. The right palatine tooth plate is ful- ly exposed. Dividing its longitudinal diameter (18.4 mm) by its transverse diameter (12.8 mm) yields a ratio of 1 .44. This ratio is larger than the corresponding value for Placochelys yet falls into the lower range of variation for Psephoderma, a genus diagnosed by distinctly elongated posterior 82 FIELDIANA: GEOLOGY Fig. 37. Holotype of Placochelys stoppanii Osswald (bsp AS I 1457) in ventral view. Scale bar = 20 mm. palatine tooth plates. As discussed in more detail below, there is a positive allometric size increase in the posterior palatine tooth plate in Psepho- derma, and the fact that P. alpis sordidae is a relatively small skull may explain why the pro- portions of its posterior palatine tooth plate fall into the lower range of variation observed in Pse- phoderma. The relatively pronounced elongation of the posterior palatine tooth plates, together with the posterior extent of the postorbital along the lateral margin of the upper temporal fossa and the relatively broad palatal exposure of the pterygoid, indicates that specimen bsp 1921.1.3 is best re- ferred to the genus Psephoderma. The holotype of the genotypical species, Psephoderma alpinum (bsp AS I 8), likewise comes from the Rhaetian Koessen-Formation of the Bavarian Alps (Win- kelmoos Alpe). bsp 1921.1.3 is too incomplete however, to be diagnostic at the species level. The species and subspecies names alpis and sordidae therefore are nomina dubia. A fragmentary cyamodontoid palate (bsp AS I 1457) from the Rhaetian Koessen-Formation of the Bavarian Alps (Plankensteinsattel, S-Tegern- see) was preliminarily described as Placochelys stoppanii by Osswald (1930); a full description of the specimen was never published (Fig. 37). The specimen consists of a fragmentary dermal palate that shows a small right anterior, a large right pos- terior, a small left anterior, and a fragment of the left posterior palatine tooth plates. The right an- terior palatine tooth plate has a longitudinal di- ameter of 7.1 mm and a transverse diameter of 6.5 mm. Dividing the longitudinal diameter (25.1 mm) of the right posterior palatine tooth plate by its transverse diameter (14.5 mm) yields a ratio of 1.73, which indicates a distinct elongation of the tooth plate, a character diagnostic for the genus Psephoderma. Zapfe (1950) later referred two iso- lated tooth plates from the Rhaetian of the Aus- trian Alps to that same taxon. All of this material is too incomplete, however, to be diagnostic at the species level. The species name stoppanii there- fore is a nomen dubium. The same is true of an isolated tooth plate, again from the Rhaetian of the Bavarian Alps, which was originally described as Placodus zitteli by Ammon ( 1 878) and later referred to the genus Placochelys by Broili (1921). Another isolated tooth from the Upper Triassic (Norian or Rhaetian) of the Austrian Alps was referred to Placochelys by Rosenberg (1935). The Bayerische Staatssammlung fur Palaonto- logie und historische Geologie holds two dermal ossifications (bsp AS I 1463) from the Rhaetian (Upper Triassic) of Kotalm in the Bavarian Alps, which were referred to ?Placochelys (Fig. 38). One of these fragments is a conical ossification with a broken tip, a circular cross section, and a lightly striated surface. Its length, as preserved, is 27.8 mm. The second fragment is a triangular os- sification, flattened, again with a broken tip. One surface is slightly convex, the opposite surface is slightly concave. The surface of the bone is again slightly striated; the length (as preserved) is again 27.8 mm. Both these ossifications are in fact quite different from the conical tubercular osteoderms of the carapace of Placochelys, which show deep furrows radiating from the apex toward the base of the tubercle. Assignment of the ossifications (bsp AS I 1463) to Placochelys is therefore highly conjectural and certainly not based on close sim- ilarity, let alone diagnostic features. The occurrence of Placochelys in the northern Alpine Triassic cannot therefore be established on the basis of diagnostic material. The same is true for the southern Alpine Triassic. An isolated cy- amodontoid tooth plate, almost certainly re- worked, from the Liassic of Arzo was tentatively referred to Placochelys by Peyer (1931b), but again cannot be considered diagnostic. Isolated tooth plates from the Norian, Carnian, and Rhaetian of the southern Alps were tentatively re- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 83 Fig. 38. Dermal ossifications referred to 1 Placochelys (bsp AS I 1463) from the Rhaetian of the Bavarian Alps. Scale bar = 20 mm. ferred to Paraplacodus broilii, Psephoderma al- pinum, Psephoderma cfr. alpinum, and Placoche- lys placodonta, by Nosotti (1986), but again, the material is not diagnostic and most probably rep- resents Psephoderma alpinum. Other incomplete cyamodontoid remains (tooth plates, dorsal ver- tebra, rib fragment), again from the Carnian, No- rian, and Rhaetian of northern Italy (Province Bergamo), have been referred to Psephoderma al- pinum, Psephoderma cfr. alpinum, and Placoche- lys placodonta respectively by Nosotti (1987). Because of its incompleteness, all of this material is not diagnostic at the generic or species level. Given the absence of any diagnostic material of Placochelys placodonta from the Upper Triassic of the southern Alps, however, all of this frag- mentary material most likely represents Psepho- derma alpinum. Boni (1946 [1947]) reported Placochelys mal- anchinii from the Rhaetian of Val Sambuco near Valcava (Monte Albenza, Lombardy); the species Table 12. Measurements of the maxillary and pal- atine tooth plates of Placochelys malanchinii Boni, 1947. Cast kept at msnm. All measurements in mm. left right longi- tudinal trans- verse 0 longi- tudinal trans- verse 0 anterior maxillary tooth 9.3 5.6 - - posterior maxillary tooth 11.6 5.3 12.1 5.8 anterior palatine tooth 11.0 8.0 10.1 8.4 posterior palatine tooth 36.0 23.2 (33) (22) was based on an incomplete skull that he later described in detail (Boni, 1947 [1948]). The orig- inal skull (Pinna, 1976, Pis. IV-V) was in the pri- vate collection of Luciano Malanchini of Berga- mo, and can no longer be located today. A rather poor cast kept at the Museo Civico di Storia Na- turale di Milano indicates that the skull is broken obliquely through the right temporal arch and transversely immediately in front of the orbits. Preserved are the left temporal region of the skull, two teeth on the left maxilla, the posteriormost tooth on the right maxilla, and the anterior tooth plate on the left palatine. The posterior tooth plate on the left palatine is badly broken but still indi- cates a distinctly elongated (i.e., oval) shape, sim- ilar to the posterior tooth plates of Psephoderma alpinum but unlike the more rounded shape of the posterior palatine tooth plates of Placochelys pla- codonta (measurements for the specimen are giv- en in Table 12; comparative measurements for Placochelys placodonta are presented in Table 1). The dorsal view of the skull shows the presence of a large pineal foramen in an anterior position, close to the frontoparietal suture. A second, beautifully preserved skull (msnm V471), collected in 1974 from the Rhaetian of Monte Coraizzolo, Lombardy, was preliminarily identified as that of "Placochelyanus malanchi- nii" by Pinna (1975). In so doing, he referred mal- anchinii to the genus Placochelyanus, following a suggestion made by Kuhn (1969). Later, Pinna (1976) considered the incomplete skull from Monte Albenza {Placochelys malanchinii of Boni, 1946 [1947]), to represent the same taxon as the 84 FIELDIANA: GEOLOGY Table 13. Measurements of the maxillary and pal- atine tooth plates of Psephoderma alpinum (msnm V47 1 . All measurements in mm. left right longi- tudinal0 trans- verse 0 longi- tudinal trans- verse 0 anterior maxillary tooth 5.3 4.0 - - >osierior maxillary tooth 93 6.8 11.0 73 anterior palatine tooth 7.5 6.4 73 63 posterior palatine tooth 752 17.0 243 16.9 skull from Monte Cornizzolo, and furthermore treated malanchinii as a junior synonym of stop- panii, which he referred to the genus Placoche- lyanus Kuhn, 1969. The skull from the Rhaetian of Monte Cornizzolo was thus described as that of Placochelyanus stoppanii (new combination, Pinna, 1976). Still later, another fragmentary skull was collected in the Rhaetian of Monte Rena near Bergamo (Pinna, 1978, PI. LXXI-LXXII) from deposits that also yielded carapace fragments re- ferred to Psephoderma alpinum (Pinna, 1978; the specimens are kept at the Museo Civico di Scien- ze Naturali "E. Caffi" in Bergamo). Pinna (1978) considered the skull fragments to be associated with the carapace fragments, and at the same time established the taxonomic identity of the skull fragments from Monte Rena (again with a large and anteriorly placed pineal foramen and elon- gated posterior palatine tooth plates) with the skulls from Monte Albenza and from Monte Cor- nizzolo (msnm V471; Table 13). Placochelyanus stoppanii thus became a junior synonym of Pse- phoderma alpinum. The identification of the well- preserved skull from Monte Cornizzolo as that of Psephoderma alpinum has since been corroborat- ed by the discovery of articulated whole skeletons (Pinna & Nosotti, 1989; Renesto & Tintori, 1995). The synonymy of the incomplete skulls from Monte Albenza and Monte Rena with this taxon is likely. Finally, an incomplete and strongly depressed skull from the uppermost Ladinian or lowermost Carnian of Fusea near Tolmezzo (Province of Udine, northeastern Italy; Zucchi Stolfa, 1975) was identified as Placochelys placodonta by Pin- na and Zucchi Stolfa (1979). The skull originally belonged to the Museo Friulano di Storia Naturale in Udine, but can no longer be located today. It shows little morphological detail, and in particular it lacks the rostrum. General resemblance is more with Cyamodus than with Placochelys placodon- ta, however. The same layer that yielded this skull also yielded a complete cyamodontoid carapace, carapace fragments, and skull fragments. The car- apace is clearly different from that of Placochelys placodonta in osteoderm structure and arrange- ment, suggesting that the material from Fusea rep- resents a new cyamodontoid taxon (Rieppel & Dalla Vecchia, 2001). The only possible occurrence of Placochelys placodonta outside the type locality is in the Mus- chelkalk of Makhtesh Ramon, Negev, Israel. Iso- lated osteoderms from that locality, of upper An- isian or lower Ladinian age, have the same struc- ture as the enlarged tubercles arranged in longi- tudinal rows and surrounded by a ring of smaller tubercles in the carapace of Placochelys placo- donta. Similarly large tubercles are not known from the dermal armor of any other cyamodon- toids, although isolated osteoderms cannot be considered diagnostic of a specific taxon. Psephoderma Meyer, 1858 1975 Placochelyanus, Pinna, p. 92. 1976 Placochelyanus, Pinna, pp. 10, 13. 1976 Placochelys (partim). Pinna, p. 13. 1978 Placochelyanus, Pinna, p. 346. type species — Psephoderma alpinum Meyer, Table 14. Data matrix for implementation of the Brooks parsimony analysis in the reconstruction of cyamodon- toid historical biogeography. 1 1 2 3 4 s e 7 a 9 1 0 1 1 1 Germanic Basin 1 1 0 0 0 0 i 0 0 0 2 northern Alps 0 0 1 0 0 0 0 0 0 1 3 southeastern Alps 0 0 0 1 0 0 0 0 1 1 4 southwestern Alps 0 0 0 0 0 1 0 1 1 1 S Balaton Area 0 0 0 0 1 0 0 1 1 1 RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 85 1858, from the Rhaetian (Upper Triassic) of the Bavarian Alps. definition — A monotypic taxon including the species alpinum. diagnosis — Squamosal projecting far posteri- orly; upper temporal fossa relatively narrow; pos- terior (nasal) process of premaxilla enlarged and extending backward to reach the frontal, thereby separating the nasals from one another (conver- gent in Macroplacus); frontal narrowly entering the anterior margin of the pineal foramen; pineal foramen located in front of a distinct step in the skull roof; postfrontal small; the posterior palatine tooth plates elongate in adult (convergent in Ma- croplacus, and, to a lesser degree, in Cyamodus hildegardis). distribution — Norian to Rhaetian (Upper Tri- assic) of northern and southern Alps. Psephoderma alpinum Meyer, 1858 1858a Psephoderma Alpinum, Meyer, p. 246, PI. 29. 1858b Psephoderma Alpinum, Meyer, p. 646. 1863 Psephoderma alpinum, Curioni, p. 268. 1864 Psephoderma Alpinum, Meyer, p. 698. 1886 Psephoderma alpinum, Deecke, p. 195 // 1896 Psephoderma alpinum, Fraas, p. 14. 1902 Psephoderma alpinum, Huene, p. 33. 1902a Psephoderma alpinum, Jaekel, p. 16. 1903 Psephoderma alpinum, Freeh, p. 17. 1905 Psephoderma alpinum, Fraas, p. 367. 1907 Psephoderma alpinum, Jaekel, p. 52/, Figs. 36-37. 1915 Psephoderma alpinum, Drevermann, p. 404. 1921 Psephoderma alpinum, Broili, pp. 317. 1933 Psephoderma alpinum, Kuhn, p. 12/ 1936 Psephoderma alpinum Huene, pp. 134, 147. 1942 Psephoderma alpinum, Reiff, p. 39 ff. 1947 Placochenus n.f. Boni, Vialli, p. 120, Fig. 1. 1955 Psephoderma alpinum, Peyer and Kuhn- Schnyder, p. 479. 1956 Psephoderma alpinus, Huene, p. 374, Fig. 419. 1961 Psephoderma alpinum, Kuhn, p. 20/ 1968 Psephoderma alpina, Muller, Fig. 238. 1969 Psephoderma alpina, Kuhn, p. 16. 1975 Placochelyanus malanchinii, Pinna, p. 92, Fig. 1. 1975 Psephoderma alpina, Schubert-Klemp- nauer, p. 53. 1975 Psephoderma alpinum, Westphal, pp. 99/, 121/ Fig. 1. 1976a Placochelyanus malanchinii, Pinna, p. 11, 13, 41. 1976a Placochelyanus stoppanii, Pinna, p. \4ff., Figs. 2-7, Pis. 1-3. 1976a Psephoderma alpina, Pinna, p. 11. 1976b Psephoderma alpina, Pinna, p. 107. 1976b Placochelyanus malanchinii, Pinna, p. 108. 1976b Placochelyanus stoppanii, Pinna, p. 108., Figs. 1 a-e. 1976b Placochelys zitteli, Pinna, p. 107. 1976 Psephoderma alpinum, Westphal, p. 32, Fig. 1. 1977 Placochelys alpis sordidae, Jurcsak, p. 13. 1977 Placochelyanus stoppanii, Jurcsak, p. 14. 1977 Psephoderma alpina, Jurcsak, p. 14. 1978 Psephoderma alpinum, Pinna, p. 343 ff., Pis. 71-74. 1978 Placochelys malanchinii, Pinna, p. 349. 1978 Placochelyanus stoppanii, Pinna, p. 349. 1979 Psephoderma alpinum, Pinna, p. 195//, Fig. 1, PI. 9. 1979 Psephoderma alpinum, Pinna and Zucchi Stolfa, p. 310. 1980b Psephoderma alpinum, Pinna, p. 308//, Fig. 1, PI. 8. 1986 Psephoderma alpinum, Nosotti, p. 238//, Fig. 1. 1987 Psephoderma alpinum, Nosotti, p. 318 ff., Figs, la, 2a-b. 1989 Psephoderma alpinum, Mazin, p. 728, Fig. 4 A. 1989 Psephoderma alpinum, Nosotti and Pin- na, pp. 48, 83, Figs. 11, 12, PI. 11. 1989 Psephoderma alpinum, Pinna and Nosot- ti, p. \lff., Figs. 1-18, Pis. 25-33. 1989 Psephoderma alpinum, Pinna, pp. 150, 154, Fig. 6. 1990a Psephoderma alpinum, Pinna, pp. 146, 151, Fig. 1. 1990b Psephoderma alpinum, Pinna, p. 10, PI. 1. 1990c Psephoderma alpinum, Pinna, p. 138. 1992 Psephoderma alpinum, Alafont, p. 75, Figs. 3.14, 3.17. 1992 Psephoderma alpissordidae, Alafont, p. 75. 1992 Psephoderma alpinum, Pinna, p. 4 ff., Fig. 23b. 1993 Psephoderma alpinum, Dalla Vecchia, pp. 51, 53, 55, Fig. 5D. 86 FIELDIANA: GEOLOGY 1993 Psephoderma alpinum, Mazin and Pinna, p. 83 .# Figs. 2, 5, 6b. 1993b Psephoderma alpinum, Nosotti and Pin- na, p. 109 j£ Fig. 2C. 1993 Psephoderma alpinum. Pinna, pp. 117, 121, Fig. 12. 1993 Psephoderma alpinum. Pinna and Mazin, p. 126, Fig. 1. 1995 Psephoderma alpinum, Renesto and Tin- tori, p. 37 ff„ Figs. 2-7. 1996 Psephoderma alpinum, Nosotti and Pin- na, p. \9ff.. Fig. 24. 1997 Psephoderma alpinum, Rieppel and Zan- on, p. 215, Figs. IE, 2D, 3D. 1999 Psephoderma alpinum. Pinna, p. 35 ff. holotype — bsp AS I 8; carapace. stratum typicum — Kossen Schichten, Rhae- tian, Upper Triassic. locus typicus — Winkelmoos Alpe, Bavarian Alps, Germany. diagnosis — Same as for genus, of which this is the only known species. distribution — Same as for genus, of which this is the only known species. referred material — mbsn 1, carapace frag- ment, msnm V467, isolated tooth; V471, skull; V527, articulated skeleton, msnb S699, isolated tooth; 4614, carapace fragment; 4884, juvenile specimen without carapace; 5114, isolated tooth; 8358, carapace; 8359, tail; msnb uncatalogued, skull fragment and carapace fragment (original of Pinna, 1978, Pis. 71-73): ST 82003, articulated skeleton. comments — The species names of Placochelys alpis sordidae (Broili, 1921), Placochelys (Pla- cochelyanus) malanchinii (Boni, 1947 [1948]), Placochelys (Placochelyanus) stoppanii (Oss- wald, 1930), and Placochelys zitteli (Ammon, 1878) are all nomina dubia because the holotypes are either lost, or not diagnostic at the species level, or both. The material is, however, referable to the genus Psephoderma (see comments on Pla- cochelys placodonta above). The skull from Mte. Cornizzolo, identified as Placochelyanus malanchinii (Pinna, 1975), or Placochelyanus stoppanii (Pinna, 1976) is diag- nostic at the species level, however, and was cor- rectly synonymized with Psephoderma alpinum by Pinna (1978), as is indicated in the synonymy listing for the latter taxon. Meyer (1864, 1867) described Psephoderma anglicum on the basis of six osteoderms from the Holwell fissure fill (Rhaetian, Upper Triassic) of Holwell near Frome, Sumerset, Great Britain. Both the original material and numerous addition- al specimens are kept in the Bath Geological Mu- seum, and casts of the six osteoderms figured by Meyer (1867) are kept at The Natural History Museum, London (bmnh. R. 1511-1514). The material is not diagnostic (Storrs, 1994) and can- not be distinguished from Psephoderma alpinum (C. J. Duffin, personal communication). Placochelys and Potential Turtle Relationships of the Cyamodontoidea In his original description of Placochelys, Jae- kel (1902, 1907) postulated placodont, and spe- cifically cyamodontoid, relationships for turtles, which he based mainly on the presence of a dorsal and ventral dermal armor in some representatives of the Cyamodontoidea. Placodont relationships of turtles have been rejected (Gregory, 1 946) and papareptilian relationships have been suggested instead, either with pareiasaurs (Gregory, 1946; Lee, 1995, 1997) or with procolophonoids (Laurin & Reisz, 1995; Reisz & Laurin, 1991). Other analyses have recently pointed to sauropterygian affinities of turtles (Rieppel, 1994b; deBraga & Rieppel, 1997), which raises the question of the cyamodontoid relationships of turtles again. Both of these groups, cyamodontoids and turtles, are highly autapomorphic in their cranial anatomy, which renders comparison difficult. The general appearance of cyamodontoid skulls, in particular that of Placochelys, evokes general similarities with turtle skulls, especially those taxa that show some reduction of the dermal bones covering the cheek. Apart from similarities, however, there also exist many differences in the cranial anatomy of cyamodontoids and turtles. The geologically oldest and most plesiomorph- ic turtle is Proganochelys quenstedti, which was the focus of a recent monograph by Gaffney (1990). In his description of the skull of Progan- ochelys, Gaffney (1990) discussed epidermal scale areas that are often associated with projec- tions and bosses. Such are found on the prefrontal and postorbital bones, as well as along the pos- terior margin of the skull roof. Similar projections and bosses have been described above as dermal encrustations on the prefrontal, postfrontal, post- orbital, and parietal skull roof of Placochelys. Whether these dermal encrustations reflect areas of epidermal scales in cyamdontoids cannot be as- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 87 certained because of the lack of extant models. Nonetheless, Proganochelys, as well as all other turtles, lacks dermal tubercles secondarily fused to dermal bones of the temporal region of the skull. Several authors have assumed that a horny sheath, similar to the rhamphotheca of turtles, might have covered the margins of the edentulous rostrum in cyamodontoids. This character is high- ly conjectural and, if present, does not represent the plesiomorphic condition of cyamodontoids, which is characterized by a short and broad ros- trum formed by tooth-bearing premaxillae. There are differences in the composition of the dermatocranium of Proganochelys and cyamo- dontoids. Proganochelys lacks a postfrontal, which is present in cyamodontoids, yet the supra- temporal, absent in cyamodontoids, is present in Proganochelys. On the other hand, an ectoptery- goid is absent in both groups. All turtles except Proganochelys lack a lacrimal bone as well as a lacrimal duct; a lacrimal bone is present in Pro- ganochelys, and it alone encloses the lacrimal fo- ramen, which in one specimen appears to be paired (Gaffney, 1990: 41). A lacrimal is absent in all cyamodontoids, as indeed in all sauropter- ygians, and the lacrimal foramen is enclosed with- in the maxillary bone, with a possible contribution from the prefrontal in Placochelys. It appears, however, that in some cyamodontoids, the lacri- mal duct bifurcates in the anteroventral corner of the orbit (see discussion above), as is also the case in one specimen of Proganochelys. In all stem-group Sauropterygia, the dermal palate is fused to the basicranium and extends backward up to the occipital condyle, obscuring the basicranium in ventral view except at its pos- teriormost margin (Rieppel & Werneburg, 1998). This condition represents the basal morphology at the level of Sauropterygia. In placodonts in par- ticular, the palatines that carry large tooth plates are greatly enlarged within the dermal palate at the expense of the pterygoids. In contrast to other turtles, Proganochelys retains an open palatobasal articulation (Gaffney, 1983, 1990). In all other turtles the dermal palate is likewise fused to the basicranium, but the palatines are not enlarged at the expense of the pterygoids, and the basicra- nium remains exposed in ventral view. Palatal tooth plates are absent in all turtles, including Proganochelys. Fusion of the dermal palate to the basicranium results in an intracranial course of the internal ca- rotid both in cyamodontoids and in turtles. Pro- ganochelys itself retains the plesiomorphic con- dition, with a superficial course of the internal ca- rotid, which pierces the base of the basipterygoid process. In other turtles the internal carotid enters the foramen posterior canalis caroticis interni (Gaffney, 1972), the location of which is variable within the group (Gaffney, 1979; Rieppel, 1980). In derived cryptodires, the foramen posterior ca- nalis caroticis interni lies at the posterior end of the pterygoid. In cyamodontoids, the internal ca- rotid passes through a gap between the basioccip- ital tuber and a ventral flange of the opisthotic to enter the posterior part of the cranioquadrate pas- sage. From there it continues in a canal between the basicranium and the otic complex. Although trapped in an intracranial course in turtles (except Proganochelys) as well as in cyamodontoids, the passage of the internal carotid into the basicra- nium is topologically not equivalent in the two groups. As in other reptiles, the internal carotid bifur- cates near the dorsum sellae in cyamodontoids and in turtles. One branch, the cerebral carotid, passes through a foramen in the sella turcica into the cerebral cavity of the braincase. The other branch continues anteriorly as the palatine artery. In Ctenosaura, a lizard with an open palate, the palatine artery travels through the pyriform re- cess, across the transverse process of the ptery- goid, and continues on the dorsal surface of the palatine into the orbit (Oelrich, 1956). On its way, it supplies the soft palate with vessels, a major one piercing the inferior orbital membrane span- ning the inferior orbital foramen (sensu Oelrich, 1956; infraorbital foramen of other authors). In turtles, the palatine artery continues anteri- orly in the canalis caroticus lateralis within the pterygoid bone, from which it emerges through the foramen caroticum lateralis, which opens into the sulcus cavernosus directly lateral to the basi- sphenoid (Albrecht, 1976; Gaffney, 1972). Ac- cording to Albrecht (1976), a ventral branch of the palatine artery pierces a small foramen in the pterygoid to supply the soft palate in the area of the pterygoid-vomer suture. The inframaxillary artery, a branch of the infraorbital artery (which itself originates from the stapedial artery) passes through the foramen palatinum posterius (Rieppel, 1995c) in turtles and continues anteriorly along the margin of the triturating surface (Albrecht, 1976). As mentioned above, the palatine artery origi- nates from the internal carotid within a basicranial canal in cyamodontoids. It may have reached soft FIELDIANA: GEOLOGY palate tissues through small foramina in the pal- atine, which would be difficult to identify in fos- sils, or through the dental lamina foramina, par- ticularly the large posterior dental lamina fora- men. The homology of the inferior orbital foramen of diapsids (sensu Oelrich, 1956; infraorbital or suborbital foramen of other authors) and the fo- ramen palatinum posterius of turtles is a much debated subject. The different terminology im- plies an a priori assessment of nonhomology. DeBraga and Rieppel (1997) considered a subor- bital fenestra (i.e., an opening in the dermal palate below the orbit) that is bordered laterally by either the maxilla or the jugal to be an autapomorphy of Reptilia, with a reversal in Lanthanosuchoidea, within turtles, and in sauropterygians (the latter two considered sister-groups in this study). Exclu- sion of both the maxilla and the jugal from the lateral border of the suborbital fenestra occurs in Pareiasauria, Rhynchosauria, and in turtles. Tur- tles have a tendency to close the dermal palate to a degree that may result in the formation of a secondary palate. In the embryonic condition, when the dermal palatal elements have not yet fully ossified, the space that corresponds to the suborbital fenestra (foramen palatinum posterius of Gaffney, 1972) is bounded laterally by the maxilla and jugal (Rieppel, 1995b). As the bones continue to grow, the size of the foramen palatin- um posterius is reduced, and it may even be lost (Gaffney, 1979). In some fossil taxa (Plesioche- lys, Portlandemys; Gaffney, 1976), the maxilla enters the lateral margin of the foramen palatinum posterius. In Proganochelys, the foramen palatin- um posterius is located between the palatine and the pterygoid (Gaffney, 1990), which corresponds to the most frequently encountered position in both fossil and extant turtles (Gaffney, 1979). In cyamodontoid placodonts, a large (posterior) dental lamina foramen is located on the ptery- goid-palatine suture, behind the posterior palatine tooth plates. This foramen corresponds to the den- tal lamina foramen, which in Placodus is located lateral to the posterior palatine tooth plates (i.e., between the maxilla, ectopterygoid, and palatine). Sues (1987) was the first to suggest a homology of the posterior dental lamina foramen of Placo- dus with the suborbital foramen of other reptiles. The designation of these foramina as dental lam- ina foramina (Rieppel, 1995a) reflects the consid- eration that vertical tooth replacement necessitates the presence of dental lamina tissue in the replace- ment pit below the functional tooth, and vascular supply would reach the dental lamina tissue through the dental lamina foramen. Functional considerations should not bear on assessments of homology, and in cyamodontoids more so than in Placodus, the prominent posterior dental lamina foramina topologically correspond to the foramen palatinum posterius of turtles. In the second spec- imen of Protenodontosaurus (mfsn 1923GP), the posterior dental lamina foramina open not only to the ventral surface of the palate, but on the dorsal surface as well on both sides of the skull. Al- though an anomaly in comparison with other cy- amodontoids (including the holotype of Proteno- dontosaurus), this instance may still document the developmental potential for the posterior dental lamina foramen to pierce the palate in a topolog- ical position equivalent to the foramen palatinum posterius in turtles. However, in turtles it is the inframaxillary artery that passes through the fo- ramen palatinum posterius, whereas in cyamodon- toids it is the palatine artery that might have emerged from the posterior dental lamina fora- men. The epipterygoid is a prominent element in cryptodire turtles, as it is in cyamodontoids. The structure of the epipterygoid in Proganochelys re- mains incompletely known; the status of the epi- pterygoid in pleurodires cannot be critically eval- uated in the absence of embryological studies. In both cryptodires and cyamodontoids, the epipter- ygoid is an elongate (i.e., broad) element that con- tributes significantly to the closure of the second- ary lateral wall of the braincase. The proportional contributions to the secondary lateral wall of the braincase are different in turtles and cyamodon- toids, however. In turtles, the epipterygoid re- mains low, and the descensus parietalis is exten- sive (except in Proganochelys: Gaffney, 1990). In cyamodontoids, the epipterygoid is high, and the descensus parietalis relatively narrow. In both groups, the epipterygoid participates in the for- mation of a well-defined posterior margin for the foramen interorbitale. In turtles, the epipterygoid contacts the ventral flange of the parietal at the posterior margin of the interorbital foramen; the same is observed in Cyamodus, whereas in Pla- cochelys, the anterior dorsal margin of the epi- pterygoid meets the postorbital. The foramen interorbitale was defined by Gaff- ney (1972: 19) as "[t]he paired openings between the orbits, filled in life with cartilage" (which is pierced by the optic nerve and eye muscle nerves). Indeed, the cartilage filling the interor- bital foramen corresponds to the interorbital sep- RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 89 turn, which remains unossified in turtles and in cyamodontoids. But whereas the interorbital sep- tum (unpaired in turtles) represents an element of the endocranium (neurocranium) proper, the pos- terior margin of the interorbital foramen, defined by dermal (parietal and/or postorbital) and splanchnocranial (epipterygoid) elements, mor- phologically represents the secondary lateral wall of the braincase (located lateral to the cavum epiptericum), which is why the foramen is paired. A well-defined foramen interorbitale is absent in Placodus. Posteriorly, the epipterygoid forms the anterior margin of the trigeminal foramen in cyamodon- toids, which it may also do in some cryptodire turtles; in other cryptodire taxa, the epipterygoid may be excluded from the trigeminal foramen ei- ther by the pterygoid (Gaffney, 1979) or by the parietal (Rieppel, 1980). In cyamodontoids, the epipterygoid always meets the prootic at the dor- sal margin of the trigeminal foramen, a contact that may exceptionally occur in turtles (Gaffney, 1979: 98). The epipterygoid may contact the pro- otic at the ventral margin of the trigeminal fora- men in cyamodontoids, but it never does so in turtles. Some turtles show an internal subdivision of the trigeminal foramen by bony processes orig- inating from the surrounding bones (Gaffney, 1979: 127). Among the cranial material of cy- amodontoids, only the acid-prepared second spec- imen of Protenodontosaurus (mfsn 1923GP) shows a comparable internal subdivision of the trigeminal foramen. In turtles, the ventral contact of the epiptery- goid is with the crista pterygoidea of the ptery- goid. The crista pterygoidea forms the lateral mar- gin of the sulcus cavernosus, a trough on the dor- sal surface of the pterygoid extending between the crista pterygoidea and the basisphenoid rostrum, deep to the epipterygoid. The sulcus cavernosus essentially corresponds to the floor of the cavum epiptericum, which in turtles is incorporated into the cavum cranii by the development of a lateral wall of the braincase. Coming from behind, the vena capitis lateralis enters the cavum acustico- jugularis through the fenestra postotica. The ca- vum acustico-jugularis corresponds to the poste- rior part of the cranioquadrate passage of other reptiles. From it, the vena capitis lateralis reaches the sulcus cavernosus through the foramen cav- ernosum, located lateral to the posterior margin of the trigeminal foramen. The lateral head vein con- tinues its course along the sulcus cavernosus and emerges, together with the ophthalmic branch of the trigeminal nerve, from behind the posterior margin of the foramen interorbitale. In cyamodontoids, the anterior part of the epi- pterygoid rests on the palatine because of the sig- nificant enlargement of this element at the ex- pense of the pterygoid. Again, there is a gap be- tween the ventral margin of the epipterygoid and the rostrum basisphenoidale (exposed in Placo- chelys, mb.r. 1765), exposing the dorsal surface of the palatine, which here forms the floor of the cavum epiptericum. The cavum epiptericum is again incorporated into the cranial cavity through the formation of a secondary lateral wall of the braincase. The profundus branch of the trigeminal nerve will again have emerged from behind the posterior margin of the foramen interorbitale. The course of the lateral head vein remains unknown for cyamodontoids, however. In particular, the possible entry of the lateral head vein into the cavum epiptericum from behind cannot be ex- plained (see description above). The posterior part of the epipterygoid of cy- amodontoids shows a ventral margin of unfinished bone that overhangs the posterior part of the pal- atoquadrate recess, floored by the pterygoid. The palatoquadrate cartilage must have persisted in the adult, linking the epipterygoid with the quadrate. In Placodus, palatoquadrate cartilage persisted in the gap between the epipterygoid and the antero- medial flange of the quadrate, both overlapping the dorsal flange of the quadrate process of the pterygoid (Huene, 1931; see also the detailed de- scription above). Huene (1931; see also Kuhn- Schnyder, 1960, 1965a) referred to the persisting palatoquadrate cartilage in support of cynodont affinities of placodonts. However, cryptodire turtles also show the per- sistence (to a variable degree) of palatoquadrate cartilage in the adult (the epipterygoid and its re- lation to neighboring bones remains incompletely known in Proganochelys: Gaffney, 1990); the re- sult of a persisting palatoquadrate cartilage is the fossa cartilaginis epipterygoidei, located between the posterior process of the epipterygoid and the processus epipterygoideus of the quadrate and floored by the pterygoid. The processus epiptery- goideus of the quadrate of turtles is comparable to the anteromedial flange of the quadrate in cy- amodontoids, which broadly overlaps the ptery- goid and forms the posterolateral margin of the posterior part of the palatoquadrate recess. The persisting cartilage that connects the epipterygoid with the quadrate is thus located in topologically equivalent positions in cryptodire turtles and cy- 90 FIELDIANA: GEOLOGY amodontoids. The palatine remains restricted to an anterior position in turtles, however, and does not contact the quadrate (processus epipterygoideus) lateral to the fossa cartilaginis epipterygoidei. The fossa cartilaginis epipterygoidei is generally smaller in cryptodires than in cyamodontoids, and no anterior parts of the cartilaginous palatoquad- rate persist in turtles as they do in cyamodontoids. No ossifications in the primary lateral wall of the braincase have so far been reported for cy- amodontoids. This contrasts with Placodus, where Broili (1912) described a structure that he called an "alisphenoid bridge." Located deep to the dor- sal process of the epipterygoid, this ossification must represent the primary lateral wall of the braincase, although its exact nature remains de- batable (Rieppel, 1995a). Ossifications in the primary lateral wall of the braincase in front of the otic capsules are restrict- ed to the clinoid process in turtles. The clinoid process is located lateral to the dorsum sellae and represents the ossified basal portion of the embry- onic pila antotica. In Plesiochelys, the ossification of the pila antotica reaches up to the dorsal edges of the prootic (Gaffney, 1976). No part of the pila antotica is ossified in the cyamodontoid Placo- chelys. Other ossifications of the primary lateral wall of the braincase in the orbitotemporal region are absent in turtles with the exception of Progano- chelys (Gaffney, 1990), where an ossification is found in one skull only. This element most prob- ably corresponds to an ossification that incorpo- rates the pila antotica, pila metotica, and more dorsal elements of the primary lateral braincase wall (Gaffney, 1990, Fig. 48) and as such corre- sponds to the plesiomorphic sphenethmoid (de- Braga & Rieppel, 1997). Depending on the phy- logenetic position of turtles, the presence of a sphenethmoid will optimize as an autapomorphy of Pro ganochelys (Gaffney, 1990). The otico-occipital region of advanced turtles is uniquely derived by the differentiation of a ca- vum acustico-jugulare and "recessus scalae tym- pani" sensu Gaffney (1972), located between the quadrate laterally and the otic capsule medially, floored by the pterygoid and bounded dorsally by the paroccipital process (Gaffney, 1972). The structural relations of the cavum acustico-jugulare were reviewed in detail by Rieppel (1980, 1985). Turtles do not have a subdivided fissura metotica, as it is observed in lepidosaurs and archosaurs, and hence lack a true recessus scalae tympani (Rieppel, 1985), as do cyamodontoids (and sau- ropterygians in general). Lack of a subdivided fis- sura metotica is a plesiomorphic feature of am- niotes. The cavum acustico-jugulare of turtles cor- responds to the posterior part of the cranioquad- rate passage, within which a posteromedial pocket is partially isolated by what Gaffney (1972) de- scribed as a ventral process of the opisthotic (pro- cessus interfenestralis: Gaffney, 1972); the pos- teromedial pocket of the cavum acustico-jugulare was designated as "recessus scalae tympani" by Gaffney (1972), with the vagus nerve and the vena cerebralis posterior passing through it. In fact, however, the processus interfenestralis of Gaffney (1972) corresponds to the posterior wall of the otic capsule. It forms the posterior margin of the fenestra ovalis and in turtles is pierced by the glossopharyngeal nerve (an auta- pomorphy of Testudines). Behind the otic capsule, between it and the exoccipital, lies the metotic foramen (foramen metoticum of Rieppel, 1985; usually referred to as jugular foramen, but desig- nated as foramen jugulare anterius by Gaffney, 1972). The structure that creates the pocket re- ferred to as "recessus scalae tympani" by Gaff- ney (1972) is an elaboration of the exoccipitals and opisthotic, which provides a posterior wall to the cavum acustico-jugulare and shields the jug- ular foramen (foramen jugulare anterius of Gaff- ney, 1972) in the posterior view. As a conse- quence, this elaboration of the opisthotic and ex- occipital traps the vagus nerve in the pocket lo- cated between itself and the posterior wall of the otic capsule. The vagus nerve may leave this pocket posteriorly through a secondary foramen, the foramen jugulare posterius of Gaffney (1972), which is formed by the posterior elaboration of the exoccipital bordering on the opisthotic. The foramen jugulare posterius may be confluent with the posterior opening of the cavum acustico-jug- ulare (i.e., with the fenestra postotica), or it may be absent altogether (Gaffney, 1972). Proganochelys lacks a comparable cavum acus- tico-jugulare, largely as a result of a lesser devel- opment of the quadrate ramus of the pterygoid and lack of a posterior elaboration of exoccipital and opisthotic shielding the foramen jugulare an- terius in posterior view. This leaves the middle ear region open posteroventrally (Gaffney, 1990) and prevents the recognition of a "recessus scalae tympani" sensu Gaffney (1972, 1990). Hence, plesiomorphic relations obtain. The fenestra ves- tibuli opens into the cranioquadrate passage. The posterior margin of the fenestra vestibuli is formed by the opisthotic, which is traversed by RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 91 the glossopharyngeal nerve. Between the opis- thotic and the exoccipital lies the jugular foramen (foramen jugulare posterius sensu Gaffney, 1972). The two roots of the hypoglossal nerve leave the exoccipital through separate foramina. In plesiochelyids, the separation of a "recessus scalae tympanV sensu Gaffney remains incom- plete as well (Rieppel, 1980, 1985). There is no foramen jugulare posterius, and because of a less- er elaboration of the exoccipital and opisthotic, the foramen jugulare anterius is exposed in pos- terolateral view. The cavum acustico-jugulare as well as its posterior opening, the fenestra posto- tica, is better defined in plesiochelyids as com- pared with Proganochelys, because of increased development of the quadrate ramus of the ptery- goid flooring the posterior part of the cranioquad- rate passage. In all turtles, including Proganochelys, the in- ternal carotid enters the cranioquadrate posteriorly and gives off the stapedial artery within its pos- terior part or directly in front of the fenestra pos- totica where differentiated (Albrecht, 1976). The stapedial artery passes dorsal to the stapes (Al- brecht, 1976) on its way to the aditus canalis sta- pedio-temporalis (Gaffney, 1972). The canalis sta- pedio-temporalis is a canal that pierces the par- occipital process in its anterior part, and allows the stapedial artery to reach the temporal muscles through a foramen (the foramen stapedio-tempor- ale of Gaffney, 1972) located on the dorsal sur- face of the paroccipital process, most frequently between the quadrate and the prootic. In Progano- chelys, the opisthotic forms the posterior margin of the foramen stapedio-temporale, which is bounded anteriorly by the quadrate and medially by the prootic. With the posterior extension of the pterygoid in cyamodontoid placodonts, the posterior part of the cranioquadrate passage is closed ventrally, result- ing in the formation of a space located between the quadrate laterally and the otic capsule medi- ally, which is floored by the pterygoid and bound- ed dorsally by the paroccipital process. The jug- ular foramen lies outside this space, however, on the posterior surface of the occipital process, rath- er than in the posteromedial corner of this space, as in turtles. A ventral opisthotic flange, capturing the internal carotid between itself and the basi- occipital tuber as it enters the posterior opening of the cranioquadrate passage, is not known in turtles. In cyamodontoids, the stapedial artery may have passed below and in front of the stapes, but as in turtles, it pierces the paroccipital process on its way to the temporal muscles. The pteroc- cipital foramen of cyamodontoids is the functional equivalent of the stapedio-temporal foramen of turtles. The only difference is that the anterior margin of the pteroccipital foramen is formed by a neomorph otic process of the squamosal, where- as the stapedio-temporal foramen borders on the quadrate. Turtles have a long and slender stapes that con- tacts the tympanic membrane by means of a car- tilaginous extracolumella. In Proganochelys, however, a tympanic membrane appears to have been absent, and the distal end of a somewhat more massive stapes is received in a stapedial re- cess low on the medial surface of the quadrate in Proganochelys, just as in cyamodontoids. The splenial is absent in modern turtles, but where it is present in fossil turtles, it remains ex- cluded form the mandibular symphysis (Gaffney, 1979, 1990), unlike Placodus and cyamodontoids, where the splenial enters the mandibular symphy- sis. The retroarticular process is variably devel- oped in turtles, but where present it is usually rel- atively short and sloping, as in cyamodontoids. In summary, cyamodontoid placodonts and tur- tles share a number of striking convergences in their cranial as well as postcranial anatomy. Mor- phological similarities are often superficial, how- ever, and do not pass the test of similarity. The stapedial artery, for example, pierces the paroc- cipital process in both groups, but the bones sur- rounding the stapedial artery are different in cy- amodontoids and turtles. That turtles are, indeed, convergent on cyamodontoid placodonts was re- cently corroborated on the basis of cladistic anal- ysis (Rieppel & Reisz, 1999). Paleobiogeography and Paleoecology of Cyamodontoid Placodonts Analysis of the historical biogeography of cy- amodontoid placodonts remains incomplete, mainly because widespread occurrences of the clade in eastern Europe (Romania: Jurcsak, 1982; Huza et al., 1987) and along the northern Gond- wanan shelf (Brotzen, 1957; Haas, 1959, 1969, 1975; Gorce, 1960; Beltan et al., 1979) are sup- ported only by very fragmentary material that is difficult or impossible to place phylogenetically. On the basis of formally described taxa, the Pla- codontia remain restricted to the western Tethyan faunal province throughout their existence. Most 92 FIELDIANA: GEOLOGY recently, cyamodontoid placodonts have been col- lected, but not yet described, from the Middle and Upper Triassic of southern China, which is not surprising given a fairly strong signal for Asiatic (western Pacific) affinities for the sister group of the Placodontia, the Eosauropterygia (Rieppel, 1999), and its subclade, the pachypleurosaurs (Rieppel, 1998). Within the western Tethyan realm, the cyamo- dontoids from the Germanic Basin (Henodus and Cyamodus) form a clade, the monophyly of which is maintained even after the inclusion of an in- complete skull fragment from the Muschelkalk of Makhtesh Ramon, Israel, and of Cyamodus hilde- gardis. The taxa from the Germanic Basin group as a monophyletic clade to the exclusion of relat- ed taxa (other than Cyamodus hildegardis) from outside this basin (northern Alpine Triassic and the southern Alpine-Hungarian carbonate plat- form). This is a pattern of relationship and geo- graphic distribution that in essence is congruent with pachypleurosaur interrelationships and dis- tribution (Rieppel, 1998). The pattern suggests that there was a radiation of cyamodontoid placo- donts (and other sauropterygians) within the Ger- manic Basin that occurred independently from the Alpine Triassic. In spite of the incomplete knowledge of the cra- nial anatomy of Cyamodus hildegardis, inclusion of that taxon in the analysis shows it to group with the Cyamodus species from the Germanic Basin. The genus Cyamodus can thus be interpreted as a radiation of taxa within the Germanic Basin, which at the Anisian-Ladinian boundary gained access to the southern Alpine realm, there giving rise to a separate species. The same pattern of relationships and geographic distribution is again observed among Eosauropterygia, within pachy- pleurosaurs, and within the Nothosaurus-Lario- saurus clade. Taxonomic congruence thus pro- vides further support for a scenario that postulates a faunal exchange of sauropterygians, in this case of Cyamodus, between the Germanic and the southern Alpine Triassic through a southern gate- way, the Burgundy Gate (Rieppel & Hagdorn, 1997). Other than Cyamodus hildegardis, the cyamo- dontoids from the northern and southern Alpine Triassic are more closely related to each other than to the taxa from the Germanic Basin (with some rather weak indications that the Negev cy- amodontoid may be more closely related to the Alpine genera than to the taxa from the German Muschelkalk). The cyamodontoids collected from Middle and Upper Triassic deposits in the north- ern and southern Alps and in the Balaton area (Hungary: Placochelys) are records of a geo- graphic distribution across the Eurasian carbonate platform, which during Triassic times extended along the East European passive margin, between Sardinia and Moesia (Philip et al., 1995). During the Upper Triassic, a broad carbonate shelf dom- inated the southern Alpine region, extending northward into the Balaton area, the northern Alps, and the Carpathians (Marcoux et al., 1993; Marcoux & Baud, 1995). Among the cyamodontoids from outside the Germanic Basin, the best supported sister-group relationship is between Psephoderma, widespread in the southern Alpine Triassic, and Placochelys, from the Also Keuper of the Balaton area in Hun- gary. This sister-group relationship may reflect the distribution of these taxa on the southern Alps- Hungarian carbonate platform (Marcoux et al., 1993; Marcoux & Baud, 1995). The Bakony Mountains from where Placochelys comes have been identified as the southern end of the central Hungarian mountain range with Triassic outcrops of a typically southern Alpine-type facies (Der- court et al., 1984). This finding is corroborated by the close relationships between Placochelys and Psephoderma. Macroplacus is from the northern Alps, sepa- rated from the Hungarian platform by the Hallstatt trough, and Protenodontosaurus comes from the eastern part of the southern Alps (Tre Venezie area of northern Italy), which, with respect to eos- auropterygian components, shows closer faunal affinities with the northern Alpine and Germanic Triassic than with the more western parts of the southern Alpine Triassic (Rieppel & Dalla Vec- chia, 2001). The sister-group relationship of Pse- phoderma and Placochelys thus appears to sub- stantiate the biotic unity of the southern Alps- Hungarian platform, and the two taxa may have resulted from a vicariance event on this platform during the Upper Triassic. To test these paleobiogeographical patterns re- constructed on the Tethys map (Marcoux et al., 1993), the Brooks Parsimony Analysis procedure (Wiley, 1988; Brooks, 1990) was applied to cy- amodontoid phylogeny and geographic distribu- tion. Deleting Cyamodus hildegardis and the in- completely known taxon from the Negev from the analysis yielded a single area cladogram (Fig. 39), which shows the Germanic Basin in a basal di- chotomy with the areas of the Eurasian carbonate platform. Within the Eurasian carbonate platform, RIEPPEL: THE CRANIAL ANATOMY OF PLACOCHELYS PLACODONTA 93 a ex < < < 6 SO 03 < £ 1 d