Novtautesamerican MUSEUM PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

Home , Anhanguera (pterosaur), Cearadactylus, Dimorphodon, Rhamphorhynchus, Tapejara wellnhoferi

NovtautesAMERICAN MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3175, 34 pp., 23 figures June 26, 1996

Description of the Braincase of Two Early Cretaceous Pterosaurs (Pterodactyloidea) from Brazil

ALEXANDER WILHELM ARMIN KELLNER'

ABSTRACT The braincase of Tapejara wellnhoferi and An- sp. revealed the presence ofan interlaminar pneu- hanguera sp., two pterodactyloid taxa from the matic cavity with a complex system of trabeculae Early Cretaceous Santana Formation (Romualdo above the cranial cavity. Therefore, the actual Member) of the Araripe Basin, Brazil, are de- braincase is placed deeper inside the skull than scribed and compared with other archosaurian was previously supposed, raising questions about braincases. The presence of the orbitosphenoid, whether some pterosaur endocasts reported in the previously found in dinosaurs but absent in other literature express the true internal surface of the archosaurs, is reported in pterosaurs for the first braincase or just of the exocranial lamina. The time, suggesting that this bone is an ornithodiran braincase of Anhanguera sp. has some birdlike synapomorphy. Contrary to other archosaurs, there features (reduced olfactory bulbs and lateral place- is an ossification (the pseudomesethmoid) on the ment of the optic lobes) but its size indicates that anteroventral portion of those pterosaur brain- the pterosaur brain was more reptilian than was cases. A horizontal section through Anhanguera previously thought.

INTRODUCTION The braincase of pterosaurs is very poorly ten damaged during deposition and diagen- known mainly for two reasons. First, because esis. The skull, in particular, is vulnerable to oftheir fragility, pterosaurian remains are of- crushing and flattening, severely limiting de-

1 Student, Department of Vertebrate Paleontology, American Museum of Natural History and Department of Geological Sciences, Lamont Doherty Geological Observatory, Columbia University; Fellow, Conselho Nacional de Desenvolvimento Cientifico e Tecnol6gico-CNPq, Brasilia.

Copyright X American Museum of Natural History 1996 ISSN 0003-0082 / Price $4.20 2 AMERICAN MUSEUM NOVITATES NO. 3175 tailed anatomical studies of the posterior re- Terra-DNPM) in Rio de Janeiro, Brazil. gion. Second, specimens with three-dimen- Both were encased in calcareous nodules and sionally preserved bones are rare and valu- are preserved three dimensionally. able. Thus, detailed preparation ofthe brain- The most complete specimen (MCT 1500- case, which in many cases involves removing R) is referred to the tapejarid Tapejara welln- the posterior region ofthe skull or bones that hoferi. It comprises the skull and mandible are obstructing anatomical features, is often that were broken offand glued together (using discouraged. an epoxy compound) with the mandible and Very few studies of pterosaur braincases the anterior portion ofthe skull inverted. Un- are reported in the literature. Seeley (1870, fortunately, this and other artificial restora- 1871, 1901) was among the first researchers tions, is common practice among fossil col- to describe pterosaur cranial material in some lectors in the Araripe region. The second detail. Most of the specimens he studied, specimen (MCT 1501 -R) is referred to the however, were fragmentary, and therefore Anhangueridae and comprises only the dor- provided only limited information about the soposterior region of the skull. Both speci- braincase. Newton (1888) described the skull mens were prepared in detail, revealing sev- of Parapsicephalus purdoni from the Alum eral features previously unknown for ptero- Shale (Upper Lias), but focused primarily on saurs. the endocast. Edinger (1927, 1941) described specimens from Solnhofen and the Cam- bridge Greensand, but also centered her ob- ABBREVIATIONS servations on endocasts. More recently, bo basioccipital Wellnhofer (1985) and Campos and Kellner bs basisphenoid (1985) described three-dimensionally pre- clr columellar recess served pterosaurian skulls with preserved dep depression braincases from Brazil. Those specimens, eo exoccipital however, were not fully prepared and their f frontal observation were limited to the dermal skull fm foramen magnum roof and the occipital region. Bennett (1991) fo foramen described the braincase ofPteranodon in some gr groove detail, but the specimens are crushed, lim- ifl intermediate frontal lamina iting some observations. ios interorbital septum This paper describes two pterosaur brain- i jugal cases found in Cretaceous rocks of the Ar- la lacrimal aripe Basin, Northeast Brazil, and compares ls laterosphenoid them with braincases from other archosaurs. n nasal The material studied here comes from the oc occipital condyle Romualdo Member, upper stratigraphic unit op opisthotic of the Santana Formation (Beurlen, 1971). Os orbitosphenoid The Santana Formation is known worldwide p parietal for the quantity and exquisite preservation pcf postcranial fenestra of fossil remains (see Maisey, 1991, for a re- pfo pneumatic foramen view). The Romualdo Member essentially Po postorbital comprises shales of Albian age (Lima, 1978; pr prootic Pons et al., 1990), which contain calcareous psm pseudomesethmoid nodules. Fossils are found in the nodules and ptf posttemporal fenestra in the shales, but only the former are excep- q quadrate tionally well preserved. res resin The two pterosaur specimens described sac sagittal crest here belong to the Desiree Collection, which scp sclerotic plate is housed permanently in the Paleontology so supraoccipital Section of the Departamento Nacional da sq squamosal Produgcio Mineral (Museu de Ciencias da utf upper temporal fenestra 1 996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 3

(1 or r following abbreviations indicates left pared, allowing a complete view of the spec- or right side; roman numerals indicate fo- imen. Subsequently, a horizontal section was ramina for cranial nerves.) cut through the top of the skull (embedded in resin), allowing preparation of the pneu- MATERLALS AND METHODS matic cavity (or interlaminar pneumatic cav- The pterosaur braincases studied here were ity) present between the external (exocranial) prepared utilizing chemical techniques. Al- and internal (endocranial) laminae of the though the employment of acids in fossils is skull. a very old method (see Kellner, 1995a, for a Some material was removed mechanically review), the refinement of this technique for from the interlaminar pneumatic cavity of vertebrate fossils was done only in the 1950s the anhanguerid braincase and etched in a (Rixon, 1949; Toombs and Rixon, 1959). 3% formic acid solution. After drying, sam- The procedures adopted here were devel- ples were mounted on a conductive material oped particularly for three-dimensional tet- (carbon-based tape). Some were coated with rapods preserved in limestone (Kellner, 1991, a gold-palladium layer, while others were 1995a; Rutzky et al., 1994). This technique coated with a carbon layer. Both were exis carried out by immersing the specimen in amined by scanning electron microscope varying concentrations of formic acid solu- (SEM model: Zeiss DSM 950). The carbon- tions, which are saturated with calcium phos- coated samples were submitted to energy dis- phate to prevent bone deterioration. The persive spectroscopy for elemental charac- specimen is subsequently neutralized in plain terization in order to establish the compo- water and dried. The exposed parts of the sition of several parts of the specimen (e.g., bone are protected with different solutions of bone, matrix). a methacrylate resin (Paraloid B-72; Acriloid Comparisons were made with Recent spec- B-67) and the specimen is further immersed imens ofcrocodilians and birds to help iden- in the acid solution. This procedure is re- tify the cranial foramina. Previous interpre- peated until the desired preparation stage is tations of other fossil archosaur braincases reached. were also used, particularly the descriptions Due to the fragility of both specimens, a available forAllosaurus (Madsen, 1976), Sin- weak 3% formic acid solution was used. In raptor (Currie and Zhao, 1993a), Troodon some exceptionally delicate areas, a thin coat (Currie, 1985; Currie and Zhao, 1 993b), Syn- ofcyanoacrylate adhesive (super glue) diluted tarsus (Raath, 1985), Sphenosuchus (Walker, in acetone was applied to provide greater sta- 1990), Rhamphorhynchus (Wellnhofer, 1975), bility during the preparation process. Anhanguera (Wellnhofer, 1985; Campos and Several bones were covered with a reddish Kellner, 1985), and Pteranodon (Bennett, layer rich in iron, which was formed second- 1991). The anatomical terminology follows arily during diagenesis. This layer was not Baumel and Witmer (1993), using the English dissolved by acid, but had to be removed equivalents of the Latin terms. mechanically. The tapejarid specimen (MCT 1500-R) was SYSTEMATIC PALEONTOLOGY submitted to CT-Scan during an intermediate preparation stage. The results, however, To date, 17 different pterosaur species have were limited, apparently due to the small been named from the Romualdo Member (for density contrast between the bone and the a review see Campos and Kellner, 1985; Kell- calcareous matrix. ner, 1990; Wellnhofer, 1991 b; Dalla Vecchia, The anhanguerid braincase was divided 1993; Kellner and Campos, 1994). This rel- during the fossilization processes on a hori- atively high number of named taxa seems to zontal plane passing through the posttem- be inflated, especially since the earlier de- poral fenestrae. To prevent the separation of scribed species are based on incomplete ma- those parts during the preparation process, terial. Notwithstanding this, additional spec- the specimen was embedded in polyester res- imens have recently been recovered from the in. All lateral resin walls were removed after same deposits, which potentially represent the ventral and lateral portions were pre- new taxa (Kellner, 1995b). Thus, the ptero- 4 AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 1. Tapejara wellnhoferi (MCT 1 500-R). Stereophotographs of the left lateral view of the skull. Scale = 10 mm.

saurian fauna was diverse in the Araripe re- in the Romualdo Member (Campos and Kell- gion during the Albian. ner, 1985). The preserved portion of MCT The nasoantorbital fenestra of the speci- 1501-R is virtually indistinguishable from men MCT 1500-R is extremely elongated Anhanguera blittersdorffi, described by Cam- with an extended premaxillary crest above, pos and Kellner (1985), andAnhanguera san- which are shared derived features of the Ta- tanae, described by Wellnhofer (1985). The pejaridae (Kellner, 1995b). The absence of a latter was originally named Araripesaurus palatal ridge and the shape ofthe sagittal crest santanae, but afterwards regarded as a spe- allows its identification as Tapejara wellnho- cies ofAnhanguera (Kellner, 1990; Wellnho- feri. So far, only two specimens ofthis species fer, 1991a). Comparisons with other de- were known: the holotype, which consists of scribed cranial materials from the Romualdo an incomplete skull (Kellner, 1989) housed Member is not possible, since they lack the in the DNPM (Rio de Janeiro), and a skull posterior region of the skull (see Wellnhofer, and mandible (AMNH 24440), housed in the 1985; Leonardi and Borgomanero, 1985; American Museum of Natural History Kellner, 1990; and Dalla Vecchia, 1993). (Wellnhofer and Kellner, 1991). Specimen Considering the analysis above, the spec- MCT 1500-R differs from both by being imen MCT 1501 -R is provisionally classified slightly larger and by greater fusion of the as Anhanguera sp. bones (e.g., lacrimal). These differences are here regarded as ontogenetic, with MCT DESCRIPTION AND COMPARISONS 1500-R probably representing a comparatively older animal. The tapejarid specimen (MCT 1500-R: figs. The systematic allocation of the second 1-9) lacks the tip of the premaxillae, part of specimen (MCT 1501 -R) is problematic, since the middle portion of the skull, most of the it is formed only by the dorsoposterior region sagittal crest, and the anterior and posterior ofthe skull. Nevertheless its structure differs ends of the mandible. The description pre- from that found in Tapejaridae, and therefore sented here is limited to the parts that sur- precludes membership in this group. MCT round or are in direct contact with the brain- 1501-R differs from Tropeognathus mesem- case. A more detailed description ofall other brinus, that has a larger and more robust pa- bones, including the mandible, is presented rietal crest (Wellnhofer, 1987). The presence elsewhere. of a small and blunt sagittal crest, however, Many sutures in the tapejarid specimen suggests affinities of MCT 1501-R with the (MCT 1500-R) are not discernible as a result Anhangueridae, a pterosaur clade also found offusion ofthe bones. Some bones, however, 1 9961996KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 5

utf

Fig. 2. Tapejara welinhoferi (MCT 1 500-R). Left lateral view of the skull. Scale = 10 mm.

are unfused or only partially connected to There is no evidence ofa prefrontal or a post- surrounding elements, allowing the deter- frontal. mination of their shapes and limits. This On the dorsal side, the frontals fuse at the combination of unfused, partially fused, and midline, forming the base of a sagittal crest. completely fused elements suggest that this They are overlain by the premaxillae (not specimen represents a young animal in the preserved in this specimen, but see Welln- process of reaching the adult ontogenetic hofer and Kellner, 1991). The contact of the stage. frontal with the parietal is marked by a well- The specimen of Anhanguera sp. (MCT developed ridge, which extends from the lat- 1 501-R: figs. 10-23) comprises only the dor- eral margin above the orbit to the posterior soposterior part of the skull and is fully de- portion of the sagittal crest. In the middle scribed here. The estimated length ofthe skull part near the base ofthis crest, there is a small is about 580 mm, which is based on the pro- depressed area ending laterally in a foramen. portions of the posterior region of complete Inside the orbit, another foramen with a shal- anhanguerid skulls (Campos and Kellner, low groove leading to the pseudomesethmoid 1985; Wellnhofer, 1991a). could be observed. Almost all bones preserved in MCT 1501 -R The anteromedial contact ofthe frontal and are unfused and thus sutures are easily dis- the pseudomesethmoid is marked by a deep tinguished. This suggests that the specimen furrow that turns smoother posteriorly. represents a juvenile of younger ontogenetic The suture between the frontal and post- stage compared to the tapejarid material orbital is sigmoidal and runs dorsoventrally (MCT 1500-R). inside the posterior margin of the orbit. The limits with the laterosphenoid cannot be determined, although it is possible that the lat- FRONTAL ter also contacted the postorbital in this re- The frontal of Tapejara wellnhoferi (figs. 1, gion. 2) forms the skull roof above the orbit. It On the specimen referred to Anhanguera comprises two bony laminae (the medial is sp. the frontals are not completely fused dor- anteriorly very delicate and highly fenestrat- sally, and join at the midline in a straight ed) separated by bony struts (= trabeculae). suture. The anterior part of the frontal is not Anteriorly, it overlies the nasal and contacts preserved in this specimen. Laterally, the the lacrimal. A large piriform opening is pres- frontal contacts the postorbital (also not pre- ent between the nasal, frontal, and lacrimal. served) and posteriorly it is fused with the 6 AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 3. Tapejara wellnhoferi (MCT 1 500-R). Stereophotographs of the ventral view of the braincase. Scale = 10 mm.

parietal. The contact between frontal and pa- with the postorbital is almost straight, run- rietal is marked by a well-developed ridge ning dorsoventrally inside the upper tem- (similar to that of T. wellnhoferi), starting poral opening. above the dorsoposterior margin of the orbit Near the dorsolateral contact with the fron- and extending backward to the midline ofthe tal, the parietal displays a depressed area skull. The frontal, therefore, forms the major whose function is unknown. Dorsally, the pa- portion of a blunt sagittal crest present at the rietals fuse and form the basal portion of a posterior region ofthe skull (not figured here, sagittal crest, and are overlain by the poste- but see Campos and Kellner, 1985). The lim- rior extensions of the premaxillae (not pre- its with the laterosphenoid cannot be observed in this specimen, but see Wellnhofer served. and Kellner, 199 1). In most other pterosaurs, the frontals do The suture between the parietal and squa- not form a sagittal crest. The notable excep- mosal runs vertically to the anterodorsal tion is Pteranodon, where the extended sag- margin ofthe posttemporal fenestra; the con- ittal crest is mainly composed by those bones tact of the parietal with other bones (e.g., (Bennett, 1991). The frontal in Dsungarip- supraoccipital, laterosphenoid, and prootic) terus forms the basis for the cranial crest, as cannot be established in this specimen. Two in Tapejara, but differ by being more flat. small foramina are present inside the upper temporal opening: one in the anteroventral PARIETAL and another in the posterior part, the latter The parietal in Tapejara wellnhoferi (figs. well above the posttemporal fenestra. It is 1, 2) forms the anterior, medial, and posterior very likely that the laterosphenoid partici- margins of the upper temporal fenestra. An- pates in the anteroventral one. These foram- teriorly, the contact with the frontal is marked ina possibly served as passageways for blood by a well-developed ridge, while the suture vessels. 1 9961996KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 7

ios

PO utf cir

Ix-xII op

Fig. 4. Tapejara wellnhoferi (MCT 1500-R). Ventral view of the braincase. Scale = 10 mm.

In Anhanguera sp., the parietals are fused sion of the premaxillae above the parietal dorsally. Their border with the frontals is crest (Wellnhofer and Kellner, 1991), which marked by a developed ridge (described is lacking in Gallodactylus but could be pres- above). The extent to which the parietals par- ent in Dsungaripterus (personal obs.). Tu- ticipate in the sagittal crest is unknown, but puxuara follows the condition of Tapejara it is clearly not to the same degree as the but has a much larger cranial crest extended frontals. Posterolaterally, the parietals join posteriorly (Kellner and Campos, 1990, the frontals participating in the articular sur- 1992). face of the postorbital (not preserved). The The extension ofthe parietals in the occip- limits with the supraoccipital cannot be es- ital region can only be observed in a few ptero- tablished with confidence. Anteroventrally, saur specimens. According to Wellnhofer the parietal is overlain by a thin plate ex- (1975: fig. 4), the parietals in Rhamphorhyn- tending from the laterosphenoid, marked by chus extend posteriorly and are expanded in an irregular suture. Ventrally, the parietal has the occipital region. Diverging from this con- a notch to receive the prootic. dition, the parietal is excluded from the oc- The parietal region shows some variation cipital region in many pterodactyloids like among pterosaurs (see Wellnhofer, 1978, for Anhanguera (Wellnhofer, 1985; and MCT a general review). In most taxa, the parietals 1501-R), Tapejara (Wellnhofer and Kellner, are comparatively small bones, confined to 1991; and MCT 1 500-R), Dsungaripterus the posterior region ofthe skull, which do not (personal obs.), and Pteranodon (Bennett, form a cranial crest. Tropeognathus has a short 1991). In early pterodactyloids like Ptero- and blunt parietal crest, that overhangs the dactylus, Gnathosaurus, and Germanodac- posterior margin of the skull (Wellnhofer, tylus (see Wellnhofer, 1970), the condition is 1987). Gallodactylus and Dsungaripterus have not clear: although the parietals apparently the parietals extending backward, forming a participate in the occipital region, most of parietal crest (Fabre, 1976; Young, 1973). the specimens are too crushed for a positive Tapejara differs by having a posterior exten- identification. 8 AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 5. Tapejara wellnhoferi (MCT 1 500-R). Detail stereophotographs of the ventrolateral oblique view of the braincase, showing most cranial foramina. Scale = 10 mm.

In other archosaurs, the extension of the nestra, and the dorsal margin of the lower parietal varies. In recent crocodilians, the pa- temporal fenestra. Anteriorly, this bone con- rietals are small bones that occupy a pos- tacts the frontal and parietal, posteriorly the terodorsal position in the skull but are ex- squamosal, and ventrally the jugal. On the cluded from the occipital region (see lordan- lateral side of the skull, the suture between sky, 1973). In primitive crocodilomorphs, the postorbital and the squamosal runs in a however, the parietal reaches the occipital ventral direction inside the upper temporal surface as in Sphenosuchus (Walker, 1990). opening. On the medial side, this suture could In theropods, the parietal participates in the not be observed, and therefore it is not clear occipital region, although the extension var- to what extent the postorbital forms the lat- ies among taxa (see Madsen, 1976; Raath, eral wall of the upper temporal opening. In 1985; Currie and Zhao, 1993a, 1993b). Sim- this area, there is a well-developed concavity ilarly, a large ventral parietal flange overlaps on the medial side of this bony wall. the posterodorsal edge of the supraoccipital The contact with the jugal cannot be de- of Protoavis (Chatterjee, 1991), which is ab- termined in this specimen. Nevertheless, it sent in modem birds (Jollie, 1957; Chatter- seems that the ventral process of the post- jee, 1991). orbital is overlain laterally by the jugal, as observed in another specimen of this taxon (Wellnhofer and Kellner, 1991). POSTORBITAL No postorbitals were preserved in the spec- Only the left postorbital of Tapejara welln- imen of Anhanguera sp. hoferi is complete in MCT 1500-R (figs. 1- The postorbital is conservative in most 6). It is a triangular bony plate that forms the pterosaur taxa, forming a lateral triangular posterior margin of the orbit, participates in plate that participates in the same cranial the lateral margin of the upper temporal fe- openings (i.e., orbit, upper and lower tem- 1 996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 9

Fig. 6. Tapejara welinhoferi (MCT 1 500-R). Detail of the ventrolateral oblique view ofthe braincase, showing most cranial foramina. Scale = 10 mm. poral openings), although the shape and the poral fenestra, the posterodorsal margin of extension of the processes might vary (see the lower temporal fenestra, and the antero- Wellnhofer, 1978). According to Wellnhofer lateral margin of the posttemporal fenestra (1970), the postorbital in Pterodactylus, Gna- (figs. 1-8). It contacts the postorbital antero- thosaurus, and Germanodactylus is totally or laterally along a vertical suture, and overlaps partially excluded from the orbit by the pres- the ascending process of the quadrate later- ence of a postfrontal. ally and posteriorly. In the posterior region, In Recent and most fossil crocodilians, the the squamosal bears a process directed an- postorbitals are positioned dorsally and bear teroventrally. This bone also contacts the pa- a descending process that contacts the as- rietal and the supraoccipital (no suture is vis- cending process of the jugals (Iordansky, ible). 1973), differing from the condition in ptero- The contact of the squamosal with the op- saurs. In carnosaurs like Allosaurus (Madsen, isthotic is complex (figs. 7, 8). Beginning at 1976) and Sinraptor (Currie and Zhao, the middle part of the anteroventral margin 1993a), the postorbital is a heavy element, of the posttemporal fenestra, the suture be- that has a T-shaped lateral outline. In other tween these bones extends across this open- theropods, this bone is always more robust ing, terminating at its posterolateral margin. in comparison with that of pterosaurs (e.g., From this point it runs almost vertically for Currie, 1985). In modem birds, the postor- about 9 mm, abruptly changes direction to a bital is absent. 12 mm long diagonal line directed medially, and turns again horizontally ending at the lateral margin of the postcranial fenestra. SQUAMOSAL No squamosal is preserved in Anhanguera The squamosal of T. wellnhoferi partici- sp. In other anhanguerids, the squamosal pates in the lateral margin of the upper tem- forms the posterolateral region of the skull AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 7. Tapejara wellnhoferi (MCT 1 500-R). Stereophotographs of the occipital region. Scale = 10 mm. and overlaps the ascending process of the some theropods, particularly those closely re- quadrate (Wellnhofer, 1985; Campos and lated to birds like troodontids (Currie, 1985; Kellner, 1985). Currie and Zhao, 1993b), and in modem birds Among pterosaurs, the contact between the (Jollie, 1957). In those forms, the squamosal squamosal and quadrate varies. Contrary to is expanded on the occipital region (see Chat- that of most pterosaurs, the lateral quadrate- terjee, 1991), similar to the condition re- squamosal contact is very reduced in Eudi- ported in the pterosaur Rhamphorhynchus morphodon, Dimorphodon, and Rhampho- (Wellnhofer, 1975), but differing from the rhynchus (Wild, 1978; Wellnhofer, 1978). condition observed in pterodactyloids, as de- According to Wellnhofer (1975), the squa- scribed above. mosal in Rhamphorhynchus expands in the occipital region. This differs from the pterodactyloids where the occipital region can be QUADRATE observed as in anhanguerids (Wellnhofer, In Tapejara wellnhoferi, the quadrate ar- 1985; Campos and Kellner, 1985), tapejar- ticulates with the exoccipital/opisthotic and ids, Pteranodon (Bennett, 1991), and Dsun- the squamosal under the ventral margin of garipterus (personal obs.) where the squa- the posttemporal fenestra (figs. 5, 6). It is not mosal is limited to the posterior borders of clear in this specimen ifthe prootic also con- the skull. tacts the quadrate. On each side, the shaft of In Recent and many fossil crocodilians, the the quadrate is broken, preventing a detailed squamosal is a wide, horizontal bone that observation of this region. expands laterally and posteriorly, forming the No quadrate is preserved in the anhan- dorsolateral part of the occipital region (lor- guerid specimen (MCT 1501 -R) examined dansky, 1973). The same position of the here. squamosal is found in camosaurs like Allo- saurus (Madsen, 1976) and Sinraptor (Currie and Zhao, 1993a). This differs from the more SUPRAOCCIPITAL lateral position that the squamosal occupies The upper portion of the occiput in Ta- in the pterosaur skull. pejara wellnhoferi consists ofthe supraoccip- The squamosal is also laterally placed in ital (figs. 7,8), which extends dorsoposteriorly 1996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS I1I

q Fig. 8. Tapejara welinhoferi (MCT 1500-R). Occipital region. Scale = 10 mm.

Fig. 9. Tapejara wellnhoferi (MCT 1500-R). Stereophotographs of anterior view of skull, showing the intermediate frontal laminae. Scale = 10 mm. AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 10. Anhanguera sp. (MCT 1501 -R). Stereophotographs of lateral view of skull. Scale = 10 mm. and encloses the ventral part of the sagittal bones but not to the exoccipital and opis- crest (not preserved in this specimen, but see thotic, which have been slightly displaced Wellnhofer and Kellner, 1991). The supraoc- from their natural position. Ventrally the su- cipital also forms the dorsal margin of the praoccipital contacts the exoccipital along an foramen magnum and expands laterally to irregular horizontal surface that extends me- enclose the upper half of the posttemporal dially. fenestra. The contacts with other bones (pa- In more "primitive" pterosaurs like rietal, exoccipital, opisthotic, squamosal) are Rhamphorhynchus (Wellnhofer, 1975), the not clear. supraoccipital is comparatively small bone On the upper part, the supraoccipital is similar to that in other archosaurs (see Chat- pierced by two large pneumatic foramina, terjee, 1991). In the pterodactyloids where which are elongated in the dorsoventral di- the occiput can be observed (e.g., anhan- rection. Another small pneumatic foramen is guerids, tapejarids, Pteranodon, and Dsun- also preserved on the dorsal margin of the garipterus), the supraoccipital forms an ex- posttemporal fenestra. panded plate and occupies most ofthe upper In occipital view, the lateral sides of the portion ofthe occipital region, and is pierced supraoccipital are very rugose, particularly by two well-developed pneumatic foramina over the posttemporal fenestra. A small pro- (which are absent in Rhamphorhynchus, al- tuberance is present above the foramen mag- though it has several other foramina in this num, lateral to which several rugosities can area-see Wellnhofer, 1975). Similar to that also be found. Those features are presumably in most archosaurs (with the exception of for attachment of the neck musculature. some fossil and modem crocodilians-see The supraoccipital in Anhanguera sp. also Iordansky, 1973) this bone participates in the forms the upper portion of the occiput but foramen magnum. since this taxa does not have a well-devel- In all pterosaurs, the occiput has a post- oped sagittal crest on the back of the skull, it temporal fenestra, which is a primitive fea- does not extend as far posteriorly as it does ture for archosaurs, and becomes greatly re- in Tapejara. The supraoccipital bears a ridge duced or lost in theropods (Chatterjee, 199 1). on the midline above the foramen magnum According to Wellnhofer (1975), the supraoc- that extends upward. A well-developed pneu- cipital in Rhamphorhynchus does not take matic foramen is developed lateroventrally part in the posttemporal fenestra, thus dif- on each side of this ridge. fering from the condition observed in An- In this specimen (MCT 1501-R), the su- hanguera, Pteranodon, Dsungaripterus, and praoccipital is fused to most surrounding tapejarids. 1 996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 13

teroventral process of the squamosal (figs. 7, 8). Ventrally, a bony flange from the base of the opisthotic is directed ventromedially and joins the basioccipital, forming the passages for the cranial nerves IX-XII. The ventral surface inside the posttemporal fenestra is concave and possesses a few scattered foramina, the largest directed vertically near its lateral border (figs. 7, 8). In Anhanguera sp., the exoccipital is also fused to the opisthotic. The contact between the opisthotic and the prootic, however, is well demarcated by an irregular suture which trends ventrally (figs. 10, 11, 14, 15). In this Fig. 1 1. Anhanguera sp. (MCT 150 1-R). Left specimen, it can be seen that the opisthotic lateral view of skull. Scale = 10 mm. forms the ventrolateral margin of the columellar recess (see Cranial Nerves). The exoccipital-opisthotic are comparatively small in Rhamphorhynchus (see Welln- ExoccIPITAL AND OPISTHOTIC hofer, 1975) but generally form a well-de- The exoccipital in Tapejara wellnhoferi is veloped bony plate in pterodactyloids. The fused to the opisthotic laterally, forming a latter condition is also observed in Recent compact unit which obscures the contacts be- birds (e.g., Jollie, 1957; Chatterjee, 1991) and tween those bones (figs. 7, 8). Dorsally, the in Troodon (Currie, 1985). In modem croc- exoccipital-opisthotic plate is also fused to odilians (Iordansky, 1973), camosaurs (Mad- the supraoccipital and anteriorly it is fused sen, 1976; Currie and Zhao, 1993a), and in to the prootic. Posterolaterally it contacts the some other theropods like Syntarsus (Raath, squamosal and posteroventrally it contacts 1985), the lateral margin of the paroccipital the basioccipital. Laterally, the opisthotic processes is not as expanded. forms the paroccipital process, which is more BASIOCCIPITAL massive than other cranial bones. Anteriorly, near the medial margin of the The basioccipital in Tapejara welinhoferi posttemporal fenestra and inside the upper is situated below the foramen magnum and temporal opening, there is a depression be- comprises the occipital condyle (figs. 7, 8). tween the exoccipital-opisthotic, parietal, and The suture between the basioccipital and ex- possible the prootic (figs. 5, 6). Above the occipital originates at the ventral margin of anterior margin of this lateral depression, a the foramen magnum and extends to the dor- weak diagonal ridge runs dorsoposteriorly, sal margin of the opening. This opening is ending in a small foramen. further divided by a small splint of bone that In occipital view, the exoccipital forms the is pierced by several foramina, which are ap- lateral margin ofthe foramen magnum, join- parently related to the passages of cranial ing the basioccipital ventrally, without par- nerves (see Cranial Nerves). ticipating in the occipital condyle. The par- The occipital condyle is rounded, and its occipital processes are developed laterally and dorsal surface slightly flatter than the ventral expand gradually to the borders of the skull. surface. A medial vertically elongated rugose Dorsally, the paroccipital process is very thick area is present below the occipital condyle. and displaced posteriorly with respect to the Laterally, there is another rugose area on each supraoccipital. It encloses the lower portion side of the condyle (slightly less developed). of the posttemporal fenestra and ventrally Ventrally, the basioccipital and the basi- forms the dorsal margin of the postcranial sphenoid are fused, forming a laterally ex- fenestra. Lateroventrally the opisthotic forms panded plate. an articulation surface that receives the pos- The basioccipital ofAnhanguera sp. is sim- AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 12. Anhanguera sp. (MCT 1501-R). Stereophotographs of ventral view ofbraincase. Scale = 10 mm. ilar to that of T. wellnhoferi. The occipital sisphenoid is expanded. Below, this bone is condyle, however, is not preserved in the an- comparatively thin and forms the medial hanguerid specimen. border of the postcranial fenestra, up to the According to Bennett (1991), the lateral point where it contacts the basipterygoid. An- parts of the occipital condyle in Pteranodon teriorly, several bony struts connect the ba- are formed by the exoccipitals, differing from sisphenoid with the interorbital septum. This the condition in Tapejara. In Anhanguera connection does not extend to the basal por- santanae, the occipital condyle is apparently tion of the basisphenoid. only formed by the basioccipital (Wellnhofer, In the specimen ofAnhanguera sp. studied 1985), while the condition in Rhamphorhyn- here, only the upper portion of the basisphe- chus is not determined (Wellnhofer, 1975). noid (also fused with the basioccipital) is pre- In most archosaurs, the exoccipital partici- served, and thus provides no additional in- pates in the occipital condyle, although the formation (figs. 10, 1 1, 14, 1 5). There is, how- extent of this participation varies (e.g., see ever, one anhanguerid skull housed at the Chatterjee, 1991). Naturkunde Museum in Karlsruhe (Germa- ny) where a thin, bony lamina originating from the basisphenoid (and possibly also part BASISPHENOID of the parasphenoid) joins the interorbital The basisphenoid in Tapejara wellnhoferi septum. is fused with the basioccipital (figs. 7, 8). Con- The basisphenoid is very poorly known in sequently, the suture lying between them has pterosaurs. Wellnhofer (1975: 11, fig. 4a) re- been obliterated. The upper part of the ba- constructed this bone in Rhamphorhynchus 1 996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 15

pterosaur braincases studied in this paper. This bone forms the anteromedial surface of the orbit and the interorbital septum (figs. 1- 6), and is here called the pseudomesethmoid (see Discussion). In Tapejara wellnhoferi, the pseudomesethmoid is expanded anteriorly and contacts the posteromedial portion ofthe lacrimal. On the left side, a tiny cylindrical ossified structure is preserved lying ventral to the anterior process of the pseudomesethmoid; its function is unknown. Toward the posterior end, the pseudomesethmoid forms a plate that gradually narrows and is enclosed by a posterior bony expansion, that is possibly formed by the frontals. About 9 mm from its anterior margin, the pseudomesethmoid forms the interorbital septum, which is a bony plate directed ventrally. Posteriorly, the interorbital septum is connected with the basisphenoid by a series of bony struts. The anteroventral edge of the interorbital septum is very sharp but becomes more blunt toward the posterior region. The interorbital septum is connected with the orbitosphenoids and the laterosphenoids through bony processes. Fig. 13. Anhanguera sp. (MCT 1501-R). Ven- In Anhanguera sp., the limits of the pseu- tral view of braincase. Scale = 10 mm. domesethmoid with the surrounding bones, especially with the frontal, cannot be established because of fusion (figs. 10-15). Ante- riorly, the pseudomesethmoid plate is less as being essentially horizontal. In the rham- expanded than in Tapejara wellnhoferi. The phorhynchoid specimens that I examined, anterior extension of the interorbital septum however, the basisphenoid is elongated and is not fully preserved but its presence can be anteroventrally directed, as seen in other determined by a sulcus on the midline of the pterosaurs (e.g., Wellnhofer, 1985; Campos skull. Posteriorly, the interorbital septum is and Kellner, 1985; Bennett, 1991). formed by a very thin sagittal bony lamina, In other archosaurs, the basisphenoid is that is pierced by several openings, differing not as elongated as in pterosaurs despite the from the more robust bony plate found in fact that some have well-developed basipter- Tapejara welinhoferi. ygoid processes, as e.g., Syntarsus (Raath, The anteroventral region of the braincase 1985). In modem crocodilians, the main part can only be observed in a few pterosaur spec- ofthe basisphenoid is hidden by the quadrate imens. In one skull of Tapejara wellnhoferi, and the pterygoids (Iordansky, 1973). In birds, Wellnhofer and Kellner (1991) identified a the basipterygoid processes are reduced or bony plate at the anterior part of the brain- lost (Chatterjee, 1991), contrasting to the well- case (still unprepared) as a questionable lat- developed basipterygoid processes in ptero- erosphenoid. In light of the new information saurs. provided by the specimens studied here, this bone can be reidentified as the pseudomesethmoid. While redescribing Pteranodon, PSEUDOMESETHMOID Bennett (1991) attributed the anteroventral An ossified structure inserted between the ossification of the braincase as part of the medial part of the frontals was found in the laterosphenoid. The laterosphenoids, how- AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 14. Anhanguera sp. (MCT 1501-R). Detail of stereophotographs of ventrolateral oblique view of braincase, showing most cranial foramina. Scale = 10 mm. ever, are paired elements anterior to the InAnhanguera sp. the contacts between the prootic but still confined to the posterior re- laterosphenoid and the frontals are not dis- gion of the orbit (e.g., Currie, 1985; Giffin et cernible, but posteriorly, a suture with the al., 1987; Chatterjee, 1991; Clark et al., 1993; parietals can be partially observed. The lat- Currie and Zhao, 1993a). Therefore, this an- erosphenoid is expanded under the ventro- teroventral ossification of the braincase of lateral surface ofthe parietal and participates Pteranodon might also be pseudomeseth- or encloses a foramen, situated inside the up- moid. per temporal opening. Compared to the same foramen in Tapejara wellnhoferi, the one in Anhanguera sp. is very large. LATEROSPHENOID The laterosphenoid is not fused to the The laterosphenoid of Tapejara wellnho- prootic, which lies posteroventrally. The su- feri contacts the frontal and the pseudomes- ture between those bones is irregular; it starts ethmoid anteriorly, the orbitosphenoid me- medially between the openings for cranial dially, and the parietal and the prootic pos- nerves II and III, and runs posterolaterally teriorly; the limits between those bones, how- until it reaches the laterosphenoid-parietal ever, are very difficult to establish in this suture. specimen. A medially directed bony process In most archosaurs, the laterosphenoid has connects the laterosphenoid with the inter- a lateral process that joins the frontal and orbital septum. Posteriorly, this bone partic- forms the articulation surface for the post- ipates and is pierced by some openings for orbital. Whether pterosaurs have the same cranial nerves (see Cranial Nerves). lateral extension of the laterosphenoid can- 1 996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 17

res

Fig. 15. Anhanguera sp. (MCT 1501-R). Detail of ventrolateral oblique view of braincase, showing most cranial foramina. Scale = 10 mm. not be determined in the available speci- In the anhanguerid specimen, the orbito- mens. sphenoid is not fully preserved and was possibly destroyed before fossilization. Its presence, however, can be determined by the ru- ORBITOSPHENOID gose medial margins of the laterosphenoid. Two small and delicate ossifications lying Furthermore, there is a small, incomplete within a large vacuity of the ventral portion process on the anteromedial margin of the of the cranial cavity and ventrally to the lat- left laterosphenoid and another dorsal pro- erosphenoid are observed in the tapejarid cess ofthe interorbital septum; these are pos- braincase (MCT 1 500-R). These paired bones sibly part of the orbitosphenoid. fused at the midline, forming a rhomboid- The orbitosphenoid is absent in crocodil- shaped structure and are connected anteriorly ians (Iordansky, 1973) and apparently also in with the interorbital septum, laterally and several primitive archosauriforms (see Clark posteriorly with the laterosphenoid. Each of et al., 1993). The condition in birds is com- those bones, here interpreted as the orbito- plex, with some authors (e.g., Jollie, 1957) sphenoids, bears a small lateral foramen of regarding this area as composed of one bone unknown function. Posteriorly, they enclose (the orbitosphenoid), while others observed a larger, medial, rounded foramen, which two distinct elements-a smaller medially opens dorsally to the cranial cavity and ven- placed orbitosphenoid and a larger laterally trally to the interorbital fenestra, the latter placed pleurosphenoid (e.g., Goodrich, 1930), connecting the orbits. the latter regarded as the laterosphenoid by AMERICAN MUSEUM NOVITATES NO. 3175 most workers on fossil archosaunforms (Clark equally delicate horizontal bony surface that et al., 1993). A distinct orbitosphenoid has is formed by the dorsolateral part ofthe pseu- been reported in several theropods (Madsen, domesethmoid. 1976; Currie, 1985; Currie and Zhao, 1993a) Toward the anterior region of the skull, and ornithischians (e.g., Giffin et al., 1987; each intermediate frontal lamina develops a Homer, 1992). Its presence in pterosaurs sug- shallow, but long process that runs through gests that this bone has a wider distribution the ventral midline of the frontals. The con- within archosaurs, and is present at least at tact between those processes is marked by a the ornithodiran level. small groove, which is completely enclosed by them anteriorly. PROOTIC Anhanguera sp. lacks the above-described The prootic in Tapejara wellnhoferi is firm- structure. Instead, a septum formed by the ly fused with the neighboring bones, which frontals separates the two halves of the skull obscure the contacts between them (figs. 3- in front of the brain cavity. In the anterior 6). Nevertheless, this bone has a very com- region, this septum separates two oval chan- plex morphology and participates in several nels (7 x 5 mm) that are bordered dorsally openings for cranial nerves. by the frontals and ventrally by the pseudo- In Anhanguera sp., the prootic lies between mesethmoid. Those channels are connected the laterosphenoid, parietal, and opisthotic with the nasoantorbital fenestra (figs. 12, 13). (figs. 10-15). It forms the upper margin of The interior of those channels is wrinkled, the columellar recess and, as in Tapejara with several ridges on the lateroventral part wellnhoferi, also participates in some foram- (fig. 19). Those ridges run anteroposteriorly ina for the cranial nerves (see Cranial Nerves). toward the septum and some are pierced by According to Chatterjee (1992), the prootic lateral openings. was articulated with the quadrate in another The presence of the intermediate frontal arhanguerid skull. This observation, how- laminae has never been recorded before in ever, could not be confirmed in MCT 1501- archosaurs (see Discussion). R, where the prootic slightly overlaps the opisthotic but not enough to meet the quadrate. SKULL CAVITY In the tapejarid braincase, the lateral exten- The roof of the anhanguerid skull (MCT sion of the prootic cannot be determined. A 1501-R) was removed, revealing the exis- prootic articulation is found in crocodylo- tence of a well-developed pneumatic space morphs and ornithurine birds (Chatterjee, (interlaminar pneumatic cavity) between the 1992) and is absent in theropods. external (exocranial) and internal (endocranial) lamina of the bones that form the skull INTERMEDIATE FRONTAL LAMINAE roof (figs. 16, 17). After very delicate prep- An extremely delicate structure was ob- aration, a complex system of trabeculae in- served in a pneumatic space in front of the side the interlaminar pneumatic cavity was brain cavity of Tapejara wellnhoferi, poste- exposed, which lies over the actual braincase rior to the nasoantorbital fenestra (figs. 3, 4, (that contains the cranial cavity). 9). This structure is formed by two symmet- The trabecular system can be divided into rically displaced triangular bony plates (here two major areas. The anterior area is formed called intermediate frontal laminae), which by a central bony plate, which extends ver- apparently are part of the frontal bone. They tically from the midline between the frontals arise from the base of the medial portion of and partially divides the anterior part of the the frontals, and extend parallel to them. The interlaminar pneumatic cavity. On each side intermediate frontal laminae are connected ofthis central plate, there are four more plates, medially, but do not connect the frontal lat- which are displaced symmetrically with re- erally, leaving a passage between them. spect to the midline of the skull. All those The intermediate frontal laminae are ex- plates end on the anterior margin of the tremely thin and highly perforated, particu- braincase. Open spaces are found between larly in the center. Anteriorly, this structure them covered dorsally and ventrally by a is bordered on the ventral side by another meshwork of bony struts. Near the central 1 9961KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS plate, two channels bordered by the frontals matrix (not illustrated). Because of its fragil- connect the interlaminar pneumatic cavity ity, this bone was strengthened with layers of with the nasoantorbital fenestra. cyanoacrylate adhesive (super glue) diluted The posterior area ofthe trabecular system in acetone. This bone is incomplete and dis- is formed by a meshwork of bony struts, plays a very complex shape. which, although more irregular, tend to be The unbroken end ofthis bone forms a flat directed perpendicular to the sagittal plane surface, which is pierced on one side by two of the skull. foramina. Toward the broken end, three pro- Some parts ofthe trabeculae were removed cesses are visible. The proximal one has a and analyzed by Scanning Electron Micros- square outline; both remaining processes are copy (SEM). The cross section revealed that rodlike and are separated by an angle ofabout the trabeculae are formed by compact bone 42°. perforated by lacunae for osteocytes; the lat- According to the position of this bone in ter do not show any particular orientation the skull, it is probably part ofthe columella, (fig. 18). including the stapes. If this interpretation is Considering the topographic position ofthe correct, it differs from the columella observed bones forming the cranial cavity, the major in Recent reptiles (Baird, 1970) and birds part of the trabecular system is apparently (Starck, 1995), by being more flattened and formed by the frontal. Other bones that par- by having a more complex outline. Further ticipate in the braincase, such as the parietal preparation of this bone is necessary before and the supraoccipital (perhaps also the ex- a detailed description can be made. occipital/opisthotic), appear to participate in the trabecular system as well, although the FORAMINA FOR CRANLAL NERVES extent of this participation cannot be deter- The braincases described here are very mined. pneumatic and exhibit several foramina. The The braincase lies deep within the inter- interpretation ofwhich foramen might be the laminar pneumatic cavity. Its maximum passage for a particular cranial nerve is based length (parallel to the dorsal plane, excluding on comparisons with Recent bird and croc- portion for the olfactory bulbs) is 29 mm; odilian specimens, as well as with descrip- and its maximum width (between the lateral tions of other archosaur braincases. surface ofthe cerebral hemispheres) is 27 mm. There is no evidence ofthe pathway ofthe The most developed part of the braincase olfactory nerve (I) within the braincase of is formed by two comparatively large hemi- either specimen. This nerve possibly emerged spheres, each lying lateral to the sagittal plane through a large, medial opening positioned ofthe skull. Anteriorly, each bears a process, anterior to the laterosphenoid. possibly for the olfactory bulbs. Posterodor- In sal to them, there is a small centrally posi- Tapejara welinhoferi, the optic nerve (II) tioned pealike structure. Lateroventral to this passes through a large, single medial opening, structure, the dorsal surface of the braincase bordered by the frontals, laterosphenoids, and expands until it joins the lateral sides of the orbitosphenoids (figs. 3-6). The interorbital interlaminar pneumatic cavity. Posteriorly, septum is positioned ventral to this foramen the braincase surface extends almost verti- and does not divide it, which is the case in cally, enclosing a relatively large channel. some birds (Baumel and Witmer, 1993). Pos- Considering their positions and shapes, it teroventrally there is a smaller opening en- is very likely that in life those structures sur- closed anteriorly by the orbitosphenoid and rounded the cerebral hemispheres dorsally, laterally by the laterosphenoid, that might the cerebellum dorsoposteriorly, and the op- have served as the passage for the oculomotor tic nerve (III). lobes laterally. Anhanguera sp. (figs. 10-15) has a configuration for the passages of the optic (II) and ADDITIONAL SKULL ELEMENT oculomotor (III) nerves similar to that of Ta- On the left side of the anhanguerid speci- pejara wellnhoferi. The major difference is men, above the columellar recess, a delicate, the comparatively smaller size ofthe foramen flat bone was found loosely embedded in the for the oculomotor nerve (only preserved on 20 AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 16. Anhanguera sp. (MCT 1501-R). Stereophotographs of dorsal view, showing interlaminar pneumatic cavity after complete preparation. Scale = 10 mm. the left side, where a process-possibly part eral to the interorbital septum (figs. 5, 6). In ofthe orbitosphenoid-is connected with the this cavity, there are three foramina, which laterosphenoid). possibly represent independent passages for The foramen for the trochlear nerve (IV) the ophthalmic, maxillary, and mandibular in both braincases is much smaller than the rami of the trigeminal. The smallest one is previous ones and pierces the laterosphenoid. positioned anterolaterally with respect to the It is positioned anterolaterally to the foramen others. An elongated groove, directed toward for the oculomotor nerve and lateral to the the upper temporal opening, is present in front optic foramen (figs. 1-6, 10-15). In Tapejara ofthe smaller foramen. Ventromedial to this wellnhoferi (figs. 1-6) the passage for the foramen is another one, which is larger and trochlear nerve is situated closer to the optic has a similar but more marked groove. Both nerve than in Anhanguera sp. (figs. 10-15). grooves are separated anteriorly by a pro- As in other archosaurs, the exit for the tri- nounced ridge, which merges with the anter- geminal nerves is positioned between the lat- omedial margin of the upper temporal fe- erosphenoid and the prootic. The passages nestra. The largest of all three foramina is for the trigeminal nerve (VI, V2, V3) in Ta- positioned posteriorly and does not lead to pejara wellnhoferi are positioned inside a any groove. cavity bordered anteromedially by a curved, The configuration of the passages for the blunt ridge from the laterosphenoid which trigeminal nerve (V) in Anhanguera sp. dif- thickens medially; laterally, by a well-devel- fers from those described above. The cavity oped and sharp crest formed mainly by the in which this cranial nerve is positioned is prootic; and posteriorly by another crest deeper and closer to the passages for the optic formed by the laterosphenoid, that extends and oculomotor nerves. A single foramen for ventrally to a pointed process positioned lat- the trigeminal nerve is preserved. The lateral 1 996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS

Fig. 17. Anhanguera sp. (MCT 1501-R). Enlarged stereophotographs of dorsal view, showing interlaminar pneumatic cavity after complete preparation. Note position of braincase. Scale = 10 mm.

margin of the cavity containing the trigemi- contact region ofthe opisthotic and the prootic nal nerve consists oftwo blunt processes, the (figs. 5, 6). There are at least three foramina, anterior one formed by the laterosphenoid all leading into the braincase, which are pos- and the posterior one by the prootic. sibly related to this nerve. This region in An- The foramen for the abducens nerve (VI) hanguera sp. is very similar, but the foramina in Tapejara wellnhoferi is present at the limits are larger (figs. 14, 15). between the laterosphenoid and the basi- The columellar recess of Tapejara welln- sphenoid (figs. 5, 6). A shallow, ventrally di- hoferi is rounded and has an external diam- rected groove can be observed connected to eter of approximately 2 mm (figs. 5, 6). It this foramen on the left side. forms a tube that is directed lateroventrally. In Anhanguera sp., the foramen for the ab- The foramen ovale and the foramen for the ducens nerve (VI) is enclosed in a bony ring acoustic nerve (VIII) cannot be observed in that is located at the medial margin of the detail. laterosphenoid (figs. 14, 15). In Anhanguera sp., the columellar recess The facial nerve (VII) in Tapejara welln- (figs. 14, 15), although not complete in this hoferi is located in a cavity posterior to the specimen, is not as well developed as in T. one that houses the trigeminal nerve, in the wellnhoferi. 22 AMERICAN MUSEUM NOVITATES NO. 3175

The foramina for the cranial nerve could be observed in very few pterosaur braincases. Wellnhofer (1985) noticed a distinctive passage for the hypoglossal (XII) nerve above the occipital condyle, and interpreted the opening lateral to the occipital condyle as the passage for the cranial nerves IX, X, and XI. Such a position for the exit of XII has not been observed in the tapejarid braincase (MCT 1500-R). Pteranodon has a large foramen ventrolateral to the occipital condyle, which according to Bennett (1991) was the passage of cranial nerves IX to XII. Bennett (1991) also interpreted several other cranial foramina in Pteranodon, but which bones they relate to is not clear. The exit for the trigem- Fig. 18. Scanning electron micrograph of inal (V) was described as positioned between transverse section ofa trabecula from Anhanguera the prootic and the supraoccipital (Bennett, sp. (MCT 1501-R). Scale = 50,m. 1991: 53). This is a very unusual position for the trigeminal, and, if correct, differs from usual position of this cranial nerve in other In the occipital region of Tapejara welln- archosaurs and the pterosaur braincases de- hoferi, lateral to the occipital condyle and scribed here. bordered by the basioccipital and the exoccipital, a large opening encloses at least three foramina (four on the right side). Those fo- SoFr TISSUE ramina, which can also be seen from inside During the preparation of the anhanguerid the foramen magnum, are very likely related specimen (MCT 1501-R), some structures to the passages for the hypoglossal (XII) nerve. were found inside the brain cavity that were Whether other cranial nerves [glossopha- at least partially insoluble in the diluted for- ryngeal (IX), vagus (X), and accessorius (XI)] mic acid solutions employed (figs. 20-23). also pass through this opening cannot be de- Those structures consist ofspongy layers that termined (figs. 7, 8). mimic the bone surface ofthe trabeculae and The same region in Anhanguera sp. is cov- the braincase. ered with polyester resin, which hinders a Examination of those structures by SEM detailed examination. shows that they present a different texture

Fig. 19. Anhanguera sp. (MCT 1501-R). Stereophotographs of channels that connect interlaminar pneumatic cavity with nasoantorbital fenestra. Scale = 10 mm. 1 996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 23

not natural) produced during acid preparation. Alternatively, they could represent min- eralization of the soft tissue that was originally connected to the trabeculae. Neither these nor similar structures were found during acid preparation of other pterosaur material (personal obs.), including different parts ofthis same specimen (e.g., inside the orbital cavities). To my knowledge, none was reported in any other fossil vertebrate. Fur- thermore, the preservation of very delicate parts of soft tissues has been reported in several fossils from the Romualdo Member (the same unit where MCT 1501-R was found), which includes fishes (Martill, 1988), other pterosaurs (Martill and Unwin, 1989; Kell- ner, 1994b), dinosaurs (Kellner, 1996), and crocodilians (Kellner, 1994a). Although the observations above are not conclusive, it seems very likely that the structures preserved in the anhanguerid specimen represent part of the soft tissue that covered the trabeculae inside the cranial cavity ofthis pterosaur.

DISCUSSION The three-dimensionally preserved braincases from the Romualdo Member described here present new information that bears on different aspects ofpterosaur paleobiology and Fig. 20. Anhanguera sp. (MCT 1501-R). Pol- anatomy. For clarity, each one will be treated ished ventral section ofskull roofabove braincase. separately. Note trabeculae (dark), some of which are sur- rounded by a lighter material (preserved soft tis- HOMOLOGY OF THE PSEUDOMESETHMOID sue?). Scale = 10 mm. The pseudomesethmoid, as defined here, is an ossified structure that forms the anteromedial surface of the orbit and the inter- from the calcareous matrix (fig. 22A), and are orbital septum. Extant crocodilians have a directly associated with the trabeculae (figs. cartilaginous interorbital septum that ex- 20-22). They form concentric layers around tends forward and forms the cartilaginous in- cavities (figs. 22B, 23) which are intercon- ternasal septum. This cartilaginous structure nected, forming a complex network of open- further divides the nasal passages above the ings inside the brain cavity. None of those secondary palate (Iordansky, 1973), thus dif- (or similar structures) have been found inside fering from the bony structure observed in the actual braincase. the pterosaur braincases described in this pa- Energy dispersive spectroscopy of carbon- per. In no other archosaur has an ossified coated samples from those structures shows element closing the anteromedial surface of that they are essentially formed by calcite. the orbit been reported, except in birds, where Phosphate is occasionally present, but in this bone is called the mesethmoid and forms much lower quantities than in the bone. an interorbital septum. This raises the pos- There are two possible interpretations of sibility that the structure found in pterosaurs those structures. They could be artifacts (i.e., and in birds is homologous. 24 AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 21. Anhanguera sp. (MCT 1501 -R). A, Dorsal view showing interlaminar pneumatic cavity before complete preparation (2 hours in 3% diluted formic acid), scale = 10 mm; B, detail showing trabeculae with soft tissue, scale = 3 mm. 1 996196KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 25

A BB

Fig. 22. Scanning electron micrograph of sample removed from cranial cavity of Anhanguera sp. (MCT 1501-R). A, Note calcareous matrix (left), trabecula (center), and soft tissue (right); B, another section, showing connection between soft tissue and trabecula. Scale = 200 gm.

As pointed out several times in the liter- tral to any discussion about phylogeny (Pat- ature, homology is a fundamental concept of terson, 1982). Among the tests proposed by comparative biology (Bock, 1974) and cen- Patterson (1982) for homology is similarity 26 AMERICAN MUSEUM NOVITATES NO. 3175

Fig. 23. Scanning electron micrograph of sample removed from Anhanguera sp. (MCT 1501-R). Detail of soft tissue. Scale = 10 ,um.

(in fossils essentially restricted to the topog- logenetic hypothesis, pterosaurs are regarded raphy of the bones or elements and their re- as closely related to dinosaurs (Padian, 1984), lation to other structures), and congruence, being included in the Ornithodira (Gauthier, i.e., agreement ofa particular homology with 1986), but excluded from Dinosauromorpha other homologies. Some regard the latter as (Sereno, 1991). As pointed out before, no os- the only test in "the strict sense" (Pinna, sification of the anteromedial surface of the 1991). orbit has been observed in a dinosaur or in The avian mesethmoid tends to be very any other archosaur. Therefore, the ossified thick anteriorly, where it forms a triangular structure on this portion of the avian and anterior bony plate, unlike the more delicate pterosaur braincases very likely evolved in- structure found in pterosaurs. Furthermore, dependently. Otherwise, this structure would it is exposed on the dorsal surface ofthe skull be regarded as a synapomorphy uniting ptero- between the nasal processes of the premax- saurs and birds. Even in this case, because illae and frontals, and contacts the nasals there are several synapomorphies arguing ventrally (Jollie, 1957). The structure found against a sister-group relationship between in the pterosaur braincase does not contact pterosaurs and birds, the avian mesethmoid the nasals or premaxillae, and is not exposed fails the test of congruence (homology) with on the skull roof. Therefore, although it re- the pterosaurian structure (sensu Patterson, sembles the avian mesethmoid in its location 1982). and by forming an ossified interorbital sep- Since osteological terminology should, tum, the structure observed in pterosaurs whenever possible reflect homology (e.g., presents different relationships with other Clark et al., 1993), and in order to avoid any cranial bones, thus at least partly "failing" implication ofhomology with the avian mes- the test of similarity (sensu Patterson, 1982). ethmoid, a new term (pseudomesethmoid) is According to the current archosaurian phy- here adopted for the pterosaurian structure. 1 996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 27

PTEROSAURIAN ENDOCASTS SHAPE AND PROPORTIONS OF THE BRAIN The braincase ofpterosaurs is generally be- Based on the assumption that pterosaur en- lieved to enclose the brain very tightly, sim- docasts replicate the shape and proportions ilar to the condition observed in mammals ofthe actual brain, most authors regarded the and in birds, but unlike the condition present pterosaurian brain as birdlike (Seeley, 1870, in Recent reptiles (Hopson, 1979). Therefore, 1871, 1901; Newton, 1888; Edinger, 1927, it has been assumed that an endocast of the 1941; Langston, 1981; Wellnhofer, 1991b). braincase reflects the actual shape and struc- Opposing views that considered pterosaur ture of the brain (e.g., Seeley, 1871, 1901; brains to be more reptilian, are essentially Edinger, 1927, 1941; Jerison, 1973; Hopson, based on estimations of brain/body size re- 1979; Langston, 1981; Wellnhofer, 1991b). lationships (Jerison, 1973), although they ad- Edinger (1941) also pointed out that the out- mit the birdlike specialization of parts of the side surface of the pterosaur skull could in pterosaurian central nervous system (Hop- many cases indicate the actual shape of the son, 1979). brain. The three-dimensional pterosaur The anhanguerid specimen described here braincases described in this paper provide a clearly shows that the size of the actual cra- few caveats for those assumptions. nial cavity is much smaller than the cavity The anhanguerid specimen (MCT 1501 -R) enclosed by the exocranial lamina ofthe skull clearly shows that the actual cranial cavity, roof and the floor of the braincase. In other that comprises the brain, is positioned much words, the pterosaur brain in this taxon was deeper inside the skull than was previously proportionally smaller than expected, occu- supposed. The skull roofin this taxon is very pying maximally (if it completely filled the pneumatic, with the external (exocranial) cranial cavity) less than one-third of the vol- lamina separated from the internal (endo- ume inside the space between the exocranial cranial) lamina by a well-developed inter- laminae and the floor of the braincase. As a laminar pneumatic cavity. Although it is very comparison, the endocast recovered from likely that the bones surrounding the cranial Parapsicephalus purdoni (Upper Lias, Juras- cavity of this pterosaur might in fact enclose sic, England), that has been used as a general the brain very tightly, any endocast based on model for pterosaur brains (Hopson, 1979; the inner surface of the exocranial lamina of Wellnhofer, 199 lb), occupies almost the en- the skull roofwill not reflect the correct shape tire space between the exocranial laminae and and proportions of the brain of this taxon. the floor of the braincase (Newton, 1888; The pterosaur endocasts studied so far are Edinger, 1927; Hopson, 1979). based on specimens that have been found Despite the fact that the estimated dimen- isolated (Seeley, 1870; Lewy et al., 1992), un- sions of the anhanguerid cranial cavity are veiled by the removal of the external cranial larger than the endocast of Parapsicephalus bones (Newton, 1888), or artificially con- (which, according to Hopson, 1979, has the structed by casting the inner side ofthe skull following dimensions: length 25 mm; width roof, formed by the exocranial lamina (Edin- 17 mm; estimated total length of skull about ger, 1927). If the condition observed in the 190 mm), the relative size ofthe cranial cav- anhanguerid specimen is a general feature for ity is proportionally smaller in the Brazilian pterosaurs, questions arise regarding whether specimen since the skull is about three times the natural endocasts found so far reflect the longer. Although any generalization based on internal morphology of the cranial cavity or the estimated measurements above might be just the morphology of the interlaminar premature, it seems that either pterosaur cra- pneumatic cavity positioned above the cranial cavities tended not to grow very much nial cavity. It is also not clear if endocasts during their evolution, or that the endocast made directly from or associated with the preserved in Parapsicephalus that was ex- exocranial lamina ofthe skull roofwill reflect posed by the removal ofexternal cranial bones the inner surface ofthe braincase in the detail (Newton, 1888) might not represent the cor- that has been claimed. rect dimensions of the actual cranial cavity. AMERICAN MUSEUM NOVITATES NO. 3175

A summary of the current knowledge re- previously reported by Jerison (1973) and garding the shape, size, and relations between Hopson (1979). the different parts of the pterosaurian brain was provided by Hopson (1979), based essentially on the endocast preserved in Pa- INTERMEDIATE FRONTAL LAMINAE rapsicephalus. The olfactory bulbs are short, closely appressed, and not well developed, The intermediate frontal laminae are placed which has also been observed in many Recent anterior to the actual brain cavity in Tapejara birds (Bang and Cobb, 1968). The cerebral wellnhoferi. They appear to have been de- hemispheres are large, slightly longer than veloped in response to the rearrangement of wide. The optic tectum is the widest part of the cranial bones that resulted from the ex- the endocast and lies posterolateral to the pansion of the nasoantorbital fenestra in this cerebral hemispheres. It is overlain by a de- taxon. Similar development of the nasoant- veloped cerebellum which has large, poste- orbital fenestra is present in all other mem- riorly oriented floccular lobes. Edinger (1927) bers of the Tapejaridae and in Quetzalcoat- pointed out that those are much larger than lus, the only member of the Azhdarchidae in lobes in similar-sized birds. which the posterior part ofthe skull is known The shape of the braincase (formed by the (Kellner and Langston, 1994). Although more endocranial lamina of the skull roof) in the evidence is needed, the presence of those anhanguerid specimen shows some features structures might be a further shared-derived that are distinct from this general pattern of feature uniting tapejarids and azhdarchids, in the pterosaur brain. Anterior to the cerebral which the dorsal margin ofthe nasoantorbital hemispheres, there is a bony projection lo- fenestra is higher than the dorsal margin of cated above an empty space on the basal por- the orbit (Kellner, 1995b). tion ofthe skull (figs. 16, 17). This projection In Anhanguera sp., the intermediate fron- might have concealed the olfactory bulbs. If tal laminae are not observed. Instead, the area this interpretation is correct, then the olfac- anterior to the brain cavity has two channels, tory bulbs were comparatively small and po- divided by a septum, that extend inside the sitioned very deep in the braincase, similar, cranial cavity. Although tentative, those but longer than those observed in an endocast channels in the anhanguerid specimen might from the Cambridge Greensand (Cenoma- correspond to the lateral channels of the in- nian, England) described by Edinger (1927: termediate frontal laminae in T. wellnhoferi. 108). The cerebral hemispheres are well de- The function of the intermediate frontal veloped, divided along the sagittal plane, and laminae is unknown. It does not seem to be longer anteroposteriorly than mediolaterally. a supporting structure providing more The optic lobes are positioned posterolateral strength to the skull, since such structures to the cerebral hemispheres. The cerebellum never occur alone (e.g., Wellnhofer, 1985, and is small, rounded, and positioned on the pos- personal obs.) and tend to be formed by struts terior region of the cerebral hemispheres on of bones oriented along lines of forces. The the sagittal plane of the skull. The floccular intermediate frontal laminae differ from sup- lobes must have been positioned posteroven- porting struts in being more delicate and dis- tral to the cerebellum, and were not as ex- similar in arrangement. panded as had been pointed out before. According to the position of the interme- Considering the interpretations above, the diate frontal laminae, they resemble the crib- position of the optic lobes and the reduced riform plate of mammals, which is function- olfactory bulbs in the anhanguerid specimen ally associated with nasal turbinals. The in- are a condition paralleled in modern birds, termediate frontal laminae differ from the conforming to previous models of pterosaur mammalian structure by being more delicate brain structure (e.g., Edinger, 1927; Hopson, and not completely occupying the passage be- 1979). The relative size, however, suggests tween the braincase (frontals and nasals) and that the pterosaur brain (at least in this taxon) the nasal cavity (confluent with the antorbital was more reptilelike, even more so than was fenestra in derived pterosaurs). 1 9961996KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 29

CRANIAL TRABECULAE major caveat for this interpretation is related to the position ofthose structures (directly in The interlaminar pneumatic cavity with an front of the braincase, posterior to the nasal associated system of trabeculae present in- bones) and the absence of information on side the skull of the anhanguerid specimen whether structures inside the channels are also has never been described before in a ptero- present in the medioventral surface of the saur. nasals. The simplest and most compelling inter- Due to the lack of investigations of brain- pretation for this trabecular system in An- cases and the associated pneumatic system hanguera sp. is that it provides support for in other pterosaurs, there is no assurance that the cranial bones without adding much a developed trabecular system associated with weight. Analogous structures are present in the interlaminar pneumatic cavity ofthe skull birds, where they are called sandwich struc- roofis a common feature in this group. There tures (Biihler, 1972). Although widesptead is, however, some indirect evidence of its among birds, sandwich structures are not presence in other pterosaurs. Although no found in all taxa (Winkler, 1979) and can section through the braincase ofthe tapejarid vary during ontogeny (Warncke and Stork, specimen was made, it is clear from obser- 1977). These structures range from one-sto- vations through some cranial foramina that ried (e.g., domestic pigeon) to multi-storied the cranial cavity was housed deep within the (owls), comprising concentric perforated lay- skull in this taxon, and was very likely sep- ers ofbone separated from each other by small arated from the exocranial laminae ofthe skull trabeculae (Biihler, 1972, 1986). The trabec- roofby a system oftrabeculae. Newton (1888: ular system in the anhanguerid specimen 510-511), while describing the endocast of (MCT 1501-R) differs by not being oriented Parapsicephalus, mentioned that the brain horizontally, and by displaying several sym- cavity in another pterosaur skull (not figured metrical vertical plates. or further identified), that was sectioned lon- Other functions for sandwich structures in- gitudinally through the midline, was posi- clude thermal protection. This was demon- tioned "under a considerable thickness of strated in greenfinches, where the air present open cancellous bone." Seeley (1901: 54-55) in the bone lamellae in conjunction with the mentioned that he removed "the thick cov- skull feathers prevents heat loss (Wamcke and ering of the cellular bone which originally Stork, 1977). To what extent a thermoregu- extended" above the pterosaur endocast from latory function of the trabeculae structure the Cambridge Greensand (Cenomanian) he might have been possible for Anhanguera sp. described (but did not figure before prepa- remains unknown. ration was concluded). More recently, Ben- Since this trabecular system is directly nett (1991: 30) noted that the posterodorsal linked with the nasoantorbital fenestra, ar- region of the skull in Pteranodon (Late Cre- guments might arise that it also related to the taceous) was "underlain by sinuses above and respiratory system. The connection between around the orbit and the braincase." the cranial cavity and the nasoantorbital fe- Based on the observations above, it is very nestra is made by two well-developed chan- likely that the presence of pneumatic spaces nels that follow the midline of the skull. The including a trabecular system above the ac- surface inside those channels is wrinkled, tual braincase is a widespread feature among particularly near the lateroventral sides (fig. pterosaurs and not unique to anhanguerids. 19). They are further separated from each The presence of a pneumatic foramen pierc- other by a sagittal bony wall that also has a ing the supraoccipital, that is found in Pter- wrinkled surface, and which extends poste- anodon, anhanguerids, Dsungaripterus, and riorly inside the interlaminar pneumatic cav- tapejarids (and also possibly present in other ity (fig. 17). Some ofthe structures inside the pterodactyloids like azhdarchids), could be channels are reminiscent of avian nasal tur- indicative of such a pneumatic space inside binates (Bang and Wenzel, 1985), which are the skull roof. indicative ofendothermy (Ruben, 1995). The It also seems likely that all pterosaur groups 30 AMERICAN MUSEUM NOVITATES NO. 3175 might not have developed such a pneumatic and other primitive archosaurs (or archo- system above the cranial cavity, and that the sauriforms); it is found only in modem birds, trabecular system might vary as the sandwich where it is called the mesethmoid. As dis- structures vary in Recent birds. More de- cussed previously, it is unlikely that the avian tailed studies ofpterosaur braincases, includ- mesethmoid is homologous to the pterosau- ing sagittal sections through the skull, are rian pseudomesethmoid. needed to further understand the distribution Another common feature of pterosaur of those features within pterosaurs. braincases is the elongated and anteroventrally directed basisphenoid, that bears well- FEATURES OF THE developed basipterygoid processes. In most PTEROSAUR BRANCASES other archosaurs, including modem birds, the basisphenoid is more horizontal and propor- The classification ofpterosaurs within Ar- tionally shorter (see Chatterjee, 1991). chosauria presented by Romer (1956) is ac- The supraoccipital in the pterodactyloid cepted by most authors, although different pterosaurs (when observed) are expanded, proposals of pterosaurian relationships have forming most ofthe upper part ofthe occiput. been suggested (see Wild, 1978, 1983). De- This differs from most archosaurs but also spite discussion on which taxa should be in- from more "primitive" pterosaurs, as Rham- cluded in Archosauria (see Gauthier, 1986; phorhynchus (based on the reconstruction Benton and Clark, 1988), there is at least one presented by Wellnhofer, 1975). feature listed by Benton and Clark (1988) as Although both braincases studied here show an autapomorphy of archosaurs that can be a similar arrangement ofskull elements, there observed in the pterosaur braincase: the pres- are several differences between them. The ence of an ossified laterosphenoid. most significant is in the region between the Although most researchers regard ptero- cranial cavity and the nasoantorbital fenes- saurs as ornithodirans (Gauthier, 1986; Ser- tra. This area is expanded in tapejarids, prob- eno, 1991), no features ofthe braincase have ably in conjunction with the expansion ofthe been found to support this hypothesis. The nasoantorbital fenestra and the development pterosaur braincases studied here, however, of a sagittal crest on the posterior region of show the presence of two fragile bones that the skull. In anhanguerids, the posterior sag- are interpreted as the orbitosphenoids. Those ittal crest is very low and blunt., bones have been reported in several dinosaur Another difference lies in the configuration braincases and are absent in crocodilians and of the pseudomesethmoid. In Tapejara welln- apparently also in more primitive archosaurs hoferi, the anterior part of this bone is more (= archosauriforms, see Clark et al., 1993). expanded laterally than in Anhanguera sp. Although more detailed comparisons of ar- The interorbital septum (formed by the pseu- chosaur braincases are needed, it is possible domesethmoid) in Tapejara wellnhoferi is that the presence of ossified orbitosphenoids more massive. In the anhanguerid specimen, might be an autapomorphic character of the it is more delicate and perforated by more Omithodira. and comparatively larger openings that con- One of the most distinguishing features of nect the orbits. pterosaur braincases is the fragility of the The positions of the foramina for cranial bones, which is paralleled only in birds. Most are very in both pterosaur elements are very thin and exhibit a smooth nerves similar surface, contrasting to several bones found in braincases, except for the trochlear nerve (IV), other archosaurs like dinosaurs and croco- which is more close to the optic nerve (II) in dilians. the anhanguerid specimen, and the trigemi- Among the similar features shared by the nal nerve (V), which has distinct passages in pterosaur braincases studied here, the most the tapejarid specimen. outstanding is the presence of anteroventral ACKNOWLEDGMENTS ossification, that is here called the pseudomesethmoid. An ossification in this region of I am grateful foremost to John G. Maisey the skull is absent in dinosaurs, crocodilians, (AMNH) for providing incentive and for dis- 1996 KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 31 cussions during several stages of this study. Bleefeld (AMNH) for correcting the English I wish also to thank Karl F. Koopman, Bryn and Lorraine Meeker and Chester Tarka for J. Mader, George F. Barrowclough, Paul R. helping with the illustrations. I profited from Sweet, Darrel Frost, Eugene S. Gaffney, Mark discussions with Bryn J. Mader and Allison A. Norell, and John G. Maisey from the Andors (AMNH). I thank Max K. Hecht AMNH, and Diogenes de Almeida Campos (Queens College of CUNY); Lawrence M. (Departamento Nacional da Produqao Min- Witmer (Ohio University); and John G. eral, Rio de Janeiro) for access to specimens Maisey, Eugene S. Gaffney, and Diogenes A. under their care. Peling Fong Melville Campos for comments and suggestions that (AMNH) is thanked for the Scanning Elec- greatly improved this manuscript. tron Microscope analysis of the samples re- This paper represents part of my Ph.D. removed from the brain cavity of the Anhan- search at Columbia University/American guera sp. specimen (MCT 1501 -R), and Tim- Museum ofNatural History, which is funded othy Rowe (University of Texas) is thanked by the Conselho Nacional de Desenvolvi- for his help regarding preliminary CT-Scan mento Cientifico e Tecnologico (CNPq), Bra- of the tapejarid specimen (MCT 1 500-R). I zil. wish also to thank Ivy Rutzky and Ann R.

REFERENCES Baird, I. L. Bock, W. J. 1970. The anatomy of the reptilian ear. In C. 1974. Philosophical foundations of classical Gans (ed.), Biology ofthe Reptilia. Vol. evolutionary classifications. Syst. Zool. 2, Morphology B, pp. 193-275. Lon- 22: 375-392. don: Academic Press. Biihler, P. Bang, B. G., and S. Cobb 1972. Sandwichstrukturen der Schadelkap- 1968. The size of the olfactory bulb in 108 seln verschiedener V6gel-Zum Leicht- species of birds. Auk 85: 55-61. bauprinzip bei Organismen. I1 4-Mitt. Bang, B. G., and B. M. Wenzel Inst. leichte Flachentragwerke, pp. 39- 1985. Nasal cavity and olfactory system. In 50. A. S. King and J. McLelland (eds.), Form 1986. Das Vogelskelett-hochentwichelter and function in birds 3: 195-225. Lon- Knochen-Leichtbau. Arcus, Z. Archi- don: Academic Press. tektur Naturwiss. Rev. Achitecture Sci. Baumel, J. J., and L. M. Witmer 1986(5): 221-228. 1993. Osteologia. In J. J. Baumel, A. S. King, Campos, D. A., and A. W. A. Kellner J. E. Breazile, H. E. Evans, and J. C. 1985. Panorama ofthe flying reptiles study in Vanden Berge (eds.), Handbook of avi- Brazil and South America. An. Acad. an anatomy: nomina anatomica avium. Brasileira Cienc. 57(4): 453-466. Publ. Nuttall Ornithol. Club 23:45-132. Chatteijee, S. Bennett, S. C. 1991. Cranial anatomy and relationships of a 1991. Morphology of the Late Cretaceous new Triassic bird from Texas. Philos. Pterosaur Pteranodon and systematics Trans. R. Soc. London (B) 332: 277- of the Pterodactyloidea. 680 pp. Un- 346. published Dissertation, University Mi- 1992. A remarkable pterosaur skull from the crofilms International. Early Cretaceous of Brazil. J. Vertebr. Benton, M. J., and J. M. Clark Paleontol. 12(suppl. 3): 23A (abstr.). 1988. Archosaur phylogeny and the relation- Clark, J. M., J. Welman, J. A. Gauthier, and J. M. ships ofthe Crocodylia. In M. J. Benton Parrish (ed.), The phylogeny and classification 1993. The laterosphenoid bone of early ar- of the tetrapods 1: 289-332. Oxford: chosauriforms. J. Vertebr. Paleontol. Clarendon Press. 13(1): 48-57. Beurlen, K. Currie, P. J. 1971. As condig6es ecologicas e faciologicas 1985. Cranial anatomy of Stenonychosaurus da formacaio Santana na chapada do Ar- inequalis (Saurischia, Theropoda) and aripe (Nordeste do Brasil). An. Acad. its bearing on the origin of birds. Can. Brasileira Cienc. 43(supl.): 411-415. J. Earth Sci. 22: 1643-1658. 32 AMERICAN MUSEUM NOVITATES NO. 3175

Currie, P. J., and X. J. Zhao Jollie, M. T. 1993a. A new carnosaur (Dinosauria, Therop- 1957. The head skeleton of the chicken and oda) from the Jurassic ofXinjiang, Peo- remarks on the anatomy of this region ple's Republic of China. Can. J. Earth in other birds. J. Morphol. 100(3): 389- Sci. 30: 2037-2081. 436. 1993b. A new troodontid (Dinosauria, Therop- Kellner, A. W. A. oda) braincase from the Dinosaur Park 1989. A new edentate pterosaur of the Lower Formation (Campanian) of Alberta. Cretaceous from the Araripe Basin, Ibid. 30: 2231-2247. Northeast Brazil. An. Acad. Brasileira Dalla Vecchia, F. M. Cienc. 61(4): 439-446. 1993. Cearadactylus? ligabuei nov. sp., a new 1990. Os repteis voadores do Cretaceo Bra- early Cretaceous (Aptian) pterosaur from sileiro. Anu. Inst. Geocienc. 1989: 86- the Chapada do Araripe (Northeastern 106. Brazil). Boll. Soc. Paleontol. Italiana 1991. Tecnica de preparagcio para tetrapodes 32(3): 401-409. f6sseis preservados em rochas calcarias. Edinger, T. In Boletim de Resumos do 120 Congr. 1927. Das Gehirn der Pterosaurier. Z. Anat. Brasileiro Paleontol. 108. Sao Paulo Entwicklungsgesch. 83: 106-112. (abstr.). 1941. The brain of Pterodactylus. Am. J. Sci. 1994a. Remarks on pterosaur taphonomy and 239: 665-682. paleoecology. Acta Geol. Leopoldensia Fabre, J. A. 39(1): 175-189. 1976. Un nouveau Pterodactylidae sur le gise- 1994b. Pterosaur soft tissue from Brazil: to be ment Portlandien de Canjuers (Var): or not to be a wing membrane, that is Gallodactylus canjuersensis nov. gen., the question! J. Vertebr. Paleontol. 14(3, nov. sp. Ann. Paleontol. Vertebr. 62(1): suppl.): 3 1A (abstr.). 35-70. 1995a. Tecnica de preparagio para tetrapodes Gauthier, J. f6sseis preservados em rochas calc'arias. 1986. Saurischian monophyly and the origin A Terra, Rev. Tec. Inf. CPRM, 1: 24- of birds. California Acad. Sci. Mem. 8: 31. 1-55. 1995b. The relationships of the Tapejaridae Giffin, E. B., D. L. Gabriel, and R. E. Johnson (Pterodactyloidea) with comments on 1987. A new pachycephalosaurid skull (Or- pterosaur phylogeny. In A. Sun and Y. nithischia) from the Cretaceous Hell Wang (eds.), Sixth Symposium on Me- Creek Formation of Montana. J. Ver- sozoic Terrestrial Ecosystems and Bio- tebr. Paleontol. 7(4): 398-407. ta, Short Papers, pp. 73-77. Beijing: Goodrich, E. S. China Ocean Press. 1930. Studies on the structure and develop- 1996. Fossilized theropod soft tissue. Nature ment of vertebrates. 837 pp. Chicago: 379: 32. Univ. Chicago Press. [Reprinted 1986] Kellner, A. W. A., and D. A. Campos Hopson, J. A. 1990. Preliminary description of an unusual 1979. Paleoneurology. In C. Gans, R. C. pterosaur skull ofthe Lower Cretaceous Northcutt, and P. Ulinski (eds.), Biol- from the Araripe Basin. In D. A. Cam- ogy of the Reptilia. Vol. 9, Neurology pos, M. S. S. Viana, P. M. Brito, and G. A: 39-146. London: Academic Press. Beurlen (eds.), Atas do 10 Simposio sobre Horner, J. R. a bacia do Araripe e bacias interiores 1992. Cranial morphology of Prosaurolophus do Nordeste, pp. 401-405. Crato, Ceara. (Ornithischia: Hadrosauridae) with de- 1992. A new tapejarid from the Santana For- scriptions oftwo new hadrosaur species mation (Lower Cretaceous) from the and an evaluation of hadrosaurid phy- Araripe Basin, Northeast Brazil. J. Ver- logenetic relationships. Occas. Pap. Mus. tebr. Paleontol. 12(suppl. 3): 36A-37A. Rockies 2: 1-119. 1994. A new species of Tupuxuara (Pterosau- Iordansky, N. N. ria, Tapejaridae) from the Early Creta- 1973. The skull of the crocodilia. In C. Gans ceous of Brazil. An. Acad. Brasileira and T. S. Parsons (eds.), Biology of the Cienc. 66(4): 467-473. Reptilia. Vol. 4, Morphology D: 201- Kellner, A. W. A., and W. Langston, Jr. 262. London: Academic Press. 1994. The cranial anatomy of an azhdarchld Jerison, H. J. pterosaur from the Late Cretaceous of 1973. Evolution of the brain and intelligence. Texas. J. Vertebr. Paleontol. 14(suppl. New York: Academic Press, 482 pp. 3): 32A. 19961996KELLNER: BRAINCASE, CRETACEOUS PTEROSAURS 33

Langston, W., Jr. bien du bassin d'Araripe (N. E. du Bre- 1981. Pterosaurs. Sci. Am. 244(2): 122-136. sil). In D. A. Campos, M. S. S. Viana, Leonardi, G., and G. Borgomanero P. M. Brito, and G. Beurlen (eds.), Atas 1985. Cearadactylus atrox nov. gen., nov. sp., do 10 Simposio sobre a bacia do Araripe novo pterosauria (Pterodactyloidea) da e bacias interiores do Nordeste, pp. 241- Chapada do Araripe, Ceara, Brasil. Co- 252. Crato, Ceara. let. Trab. Paleontol. DNPM 27(2): 75- Raath, M. A. 80. 1985. The theropod Syntarsus and its bearing Lewy, Z., A. C. Milner, and C. Patterson on the origin of birds. In M. K. Hecht, 1992. Remarkably preserved natural endo- J. H. Ostrom, G. Viohl, and P. Welln- cranial casts ofpterosaurs and fish from hofer (eds.), The beginnings ofbirds, pp. the Late Cretaceous of Israel. GSI Curr. 219-227. Eichstatt: Freunde des Jura Res. 7: 31-35. Museums Eichstatt. Lima, M. R. Rixon, A. E. 1978. Palinologia da forma9cio Santana (Cre- 1949. The use of acetic and formic acids in taceo do Nordeste do Brasil). Ph.D. diss., the preparation of fossil vertebrates. Inst. de Geociencias, Univ. Sao Paulo. Mus. J. 49: 116-117. 335 pp. Romer, A. S. Madsen, J. H., Jr. 1956. Osteology ofthe reptiles. Chicago: Univ. 1976. Allosaurusfragilis: a revised osteology. Chicago Press, 772 pp. Utah Geol. Miner. Surv. Bull. 109: 1- Ruben, J. 163. 1995. The evolution of endothermy in mam- Maisey, J. G. (ed.) mals and birds: from physiology to fos- 1991. Santana fossils: an illustrated atlas. sils. Annu. Rev. Physiol. 57: 69-75. Neptune City NJ: T.F.H. Publ. 459 pp. Rutzky, I. S., W. B. Elvers, J. G. Maisey, and A. Martill, D. M. W. A. Kellner 1988. Preservation of fish in the Cretaceous 1994. Chemical preparation techniques. In P. Santana Formation of Brazil. Palaeon- Leiggi and P. May (eds.), Vertebrate pa- tology 31(1): 1-18. leontological techniques 1: 155-218. Martill, D. M., and D. M. Unwin New York: Cambridge Univ. Press. 1989. Exceptionally well preserved pterosaur wing membrane from the Cretaceous of Seeley, H. G. Brazil. Nature 340: 138-140. 1870. The Ornithosauria: an elementary study Newton, E. T. of the bones of Pterodactyles. Cam- On and auditory organ bridge, 135 pp. 1888. the skull, brain, 1871. Additional evidence of the structure of of a new species of pterosaurian (Sca- from the Upper the head in ornithosaurs from the Cam- phognathus purdoni), bridge Upper Greensand. Ann. Mag. Lias near Whitby, Yorkshire. Philos. Trans. R. Soc. London (B) 179: 503- Nat. Hist. 37: 20-36. 537. 1901. Dragons of the air. London: Methuen, Padian, K. 239 pp. 1984. The origin of pterosaurs. In W. E. Reif Sereno, P. C. and F. Westphal (eds.), Third Sympo- 1991. Basal archosaurs: phylogenetic relation- sium on Mesozoic Terrestrial Ecosys- ships and functional implications. J. tems, Short Papers, pp. 163-168. Tii- Vertebr. Paleontol. 11 (suppl. 4): 1-53. bingen: Attempto Verlag. Starck, J. M. Patterson, C. 1995. Comparative anatomy of the external 1982. Morphological characters and homol- and middle ear ofpalaeognathous birds. ogy. In K. A. Joysey and A. E. Friday Advances in anatomy, embryology, and (eds.), Problems of phylogenetic recon- cell biology 131: 137 pp. New York: struction, pp. 21-74. London/New Springer. York: Academic Press. Toombs, H. A., and A. E. Rixon Pinna, M. C. C. de 1959. The use of acids in the preparation of 1991. Concepts and tests of homology in the vertebrate fossils. Curator 2: 304-312. cladistic paradigm. Cladistics 7: 367- Walker, A. D. 394. 1990. A revision of Sphenosuchus acutus Pons, D., P. Y. Berthou, and D. A. Campos Haughton, a crocodylomorph reptile 1990. Quelques observations sur la palyno- from the Elliot Formation (late Triassic logie de l'Aptien Superieur et de l'Al- or early Jurassic) of South Africa. Phil- 34 AMERICAN MUSEUM NOVITATES NO. 3175

os. Trans. R. Soc. London (B) 330 (1256): 199 lb. The illustrated encyclopedia of ptero- 1-120. saurs. London: Salamander Books, 192 Warncke, G., and H. J. Stork PP. 1977. Biostatische und thermoregulatorische Wellnhofer, P., and A. W. A. Kellner Funktion der Sandwich-Strukturen in 1991. The skull of Tapejara wellnhoferi Kell- der Schadeldecke der Vogel. Zool. Anz. ner (Reptilia, Pterosauria) from the 199(3/4): 251-257. Lower Cretaceous Santana Formation Wellnhofer, P. ofthe Araripe Basin, Northeastern Bra- 1970. Die Pterodactyloidea (Pterosauria) der zil. Mitt. Bayer. Staatsslg. Paliiontol. Oberjura-Plattenkalke Siiddeutsch- Hist. Geol. 31: 89-106. lands. Abh. Bayer. Akad. Wiss., N. F. Wild, R. 141: 1-133. 1978. Die Flugsaurier (Reptilia, Pterosauria) 1975. Die Rhamphorhynchoidea (Pterosau- aus der Oberen Trias von Cene bei Ber- ria) der Oberjura-Plattenkalke Siid- gamo, Italien. Boll. Soc. Paleontol. Ita- deutschlands. I: allgemeine Skelettmor- liana 17(2): 176-256. phologie. Palaiontographica, A 148: 1- 1983. Uber den Ursprung der Flugsaurier. In 33. Erwin Rutte-Festschrift. 231-238. Kel- 1978. Pterosauria. Handb. Palaeoherpetol. 19: heim/Weltenburg: Weltenburger Aka- 82 pp. Stuttgart: Gustav Fischer Verlag. demie. 1985. Neue Pterosaurier aus der Santana-For- Winkler, R. mation (Apt) der Chapada do Araripe, 1979. Zur Pneumatisation des Schadeldachs Brasilien. Palaeontogr. 187: 105-182. der Vogel. Ornithol. Beob. 76(2/3): 49- 1987. New crested pterosaurs from the Lower 118. Cretaceous of Brazil. Mitt. Bayer. Young, C. C. Staatsslg. Paliontol. Hist. Geol. 27: 157- 1973. Reports of Paleontological Expeditions 186. to Sinkiang (II). Pterosaurian Fauna 1991a. Weitere Pterosaurierfunde aus der from Wuerho, Sinkiang. Acad. Sin. Inst. SantanaFormation (Apt) der Chapada Vertebr. Palaeontol. Paleoanthropol. do Araripe, Brasilien. Palaeontographi- Mem. 11: 18-35 (in Chinese). ca, A 215: 43-101.

Recent issues ofthe Novitates may be purchased from the Museum. Lists ofback issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024. TEL: (212) 769-5545. FAX: (212) 769-5009. E-MAIL: [email protected]

( This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).