Teratosaurus suevicus Meyer, 1861

Brusatte, Stephen L., Butler, Richard J., Sulej, Tomasz & Niedźwiedzki, Grzegorz, 2009, The taxonomy and anatomy of rauisuchian archosaurs from the Late Triassic of Germany and Poland, Acta Palaeontologica Polonica 54 (2), pp. 221-230 : 222-228

Teratosaurus suevicus Meyer, 1861

Holotype: NHM 38646, right maxilla.

Type horizon: Mittlerer Stubensandstein, Mittelkeuper (Late Triassic:?middle Norian).

Type locality: Heslach , near Stuttgart, Germany, approximately 48.75 ° N 9.14 ° E (for further details, see Benton 1986) .

Emended diagnosis.—Rauisuchian archosaur distinguished from all other rauisuchians by the possession of an autapomorphic prominent and sharp groove for the dental lamina on the medial surface of the maxilla, below which is a distinctive step separating the medial surface of the main body and the interdental plates. Differentiated from “ T.” silesiacus by: a groove for the dental lamina (and associated tooth replacement foramina) deflected rostroventrally between alveoli 1 and 2; first maxillary alveolus smaller than subsequent alveoli; ventral margin of the maxilla convex anteriorly and straight posteriorly; the caudoventral groove on medial surface of the palatal process long and deep; absence of a prominent and sharply defined caudodorsally arched ridge extending from the base of the palatal process; absence of a deep fossa present dorsal to the palatal process.

Remarks.— Galton (1985) referred multiple teeth ( NHM 38647, 38648; SMNS 5212, 9918, 53535, 53536) from the Mittlerer Stubensandstein of Germany to Teratosaurus , but we conservatively consider these Archosauria indet., as their fragmentary nature means that they do not possess clear diagnostic features of either the genus or species, and could conceivably belong to large theropod dinosaurs or any carnivorous archosauriform clade. Galton (1985: 9) suggested that a referred ilium, SMNS 52972, “may” have come from the same quarry at Heslach as the holotype, and could even represent the same individual. However, this cannot be confirmed from available locality data, and so we remove this specimen from the hypodigm of T. suevicus , pending the discovery of associated rauisuchian material from the Mittlerer Stubensandstein. We conservatively consider the ilium as Rauisuchia indet.: it possesses a dorsally−trending ridge above the acetabulum, as is characteristic of rauisuchians ( Gower 2000; Brusatte et al. 2008, in press) and is extremely similar to the ilia of large rauisuchians in overall morphology ( Huene 1942; Krebs 1965; Sill 1974; Chatterjee 1985; Sen 2005; Gower and Schoch 2009).

Several authors, beginning with Galton (1985), suggested that SMNS 52972 belongs to a “poposauroid” rauisuchian— a term that has long had a vague meaning due to lack of well−constrained phylogenies, a paucity of phylogenetic definitions for discrete clades, and the long−time chimaeric nature of Postosuchus (which was originally described based on the material of three taxa: one “rauisuchid” and two “poposauroids,” clade terminology sensu Brusatte et al. 2008; Long and Murry 1995). Recent phylogenetic analyses have recovered a discrete clade centered on Poposaurus , which includes taxa such as Arizonasaurus , Effigia , Lotosaurus , and Shuvosaurus to the exclusion of the large, quadrupedal, predatory “rauisuchids” (e.g., Batrachotomus , Postosuchus , Prestosuchus , Saurosuchus , Teratosaurus ) ( Nesbitt 2007; Weinbaum and Hungerbühler 2007; Brusatte et al. 2008, in press). This clade has been defined as Poposauroidea ( Weinbaum and Hungerbühler 2007). The ilium SMNS 52972 lacks three characters diagnostic of Poposauroidea or slightly less inclusive ingroup clades in the analysis of Brusatte et al. (2008, in press): a crest above the acetabulum that trends anterodorsally, a deep fossa anterior to the crest, and an articulation for the first sacral rib reaching the anterior margin of the preacetabular process medially. Thus, SMNS 52972 cannot be referred to Poposauroidea based on current concepts and definitions of this group, and instead is much more similar to the ilia of large “rauisuchids” such as Postosuchus .

A number of characters previously proposed as diagnostic of Teratosaurus ( Galton 1985; Sulej 2005) can no longer be considered valid. Several of these characters refer to elements not preserved in the holotype of the genus (e.g., the premaxilla, nasals, prefrontals, ilium, osteoderms, etc.). Three characters of the maxilla have previously been proposed as diagnostic: (1) 13 maxillary alveoli ( Galton 1985; Sulej 2005); (2) palatal process is large ( Sulej 2005); (3) suture between the maxilla and lacrimal is V−shaped ( Galton 1985; Sulej 2005). Thirteen maxillary teeth and a large palatal process are both features also present in other rauisuchians, most notably Postosuchus kirkpatricki ( Chatterjee 1985) . The nature of the suture between the maxilla and the lacrimal cannot be assessed in the type specimen of T. suevicus ( NHM 38646), because this area of the maxilla is broken and not fully prepared (see below). Similarly, in T. silesiacus this part of the maxilla is broken ( Sulej 2005: fig. 1A, D), and the exact morphology of the suture cannot be determined.

Redescription.—First described and figured by Meyer (1861), the holotype maxilla of Teratosaurus suevicus was re−described briefly by Benton (1986) and Galton (1985). The best published photographs of the specimen were provided by Galton (1985: pl. 1) and an accurate line−drawing was provided by Benton (1986: fig. 1). Galton’s (1985: fig. 3A) line−drawing is inaccurate in both its proportions and in the topographical relationships of several anatomical features; for example, his figure shows the ‘infraorbital foramen’ positioned ventral to the sutural surface for the palatine, when this foramen is actually positioned dorsal to the sutural surface.

The specimen is a right maxilla and is preserved in two pieces—most of the element is preserved as a single piece (and was figured as such by Benton 1986: fig. 1), whereas the rostrolateral corner of the maxilla is preserved separately (the two pieces were figured in articulation by Galton 1985: pl. 1: 1). Here we figure the maxilla both without the second smaller piece ( Fig. 1A, C ) and with that piece in place ( Figs. 1B , 2 ). The maxilla is composed of a sub−rectangular main body which is broken at its caudal end, as well as an ascending process. The caudodorsal portion of the ascending process is partly unprepared, as noted by Benton (1986: caption to fig. 1), and consists of whitish−grey sandstone and glue ( Fig. 1 ). Galton (1985) described this area as “well−preserved” and suggested that a V−shaped caudodorsal edge of the ascending process might be a diagnostic character for the species. Sulej (2005) also noted a supposed V−shaped notch in the maxilla of T. silesiacus , and suggested that it might be diagnostic for the genus (but see above). However, because this area is not fully prepared in the holotype, it cannot be determined if the observed V−shape is natural, or an artefact of taphonomic processes, breakage, or reconstruction. Much of the lateral surface of the maxilla is damaged, with the external bone surface missing ( Fig. 1A ). As noted by Benton (1986), the specimen is heavily cracked and has been repaired with glue and thickly coated with preservative in places.

The main body of the maxilla is deep and transversely compressed. The dorsal and ventral margins are approximately subparallel across their lengths, resulting in a rectangular−shaped main body that does not taper in depth posteriorly in the region underneath the antorbital fenestra. In lateral view, the ventral margin of the maxilla is nearly straight above alveoli 7–13, and gently convex above alveoli 1–6. Most of the lateral surface of the main body is damaged, and nutrient foramina can only be identified on the separate small rostrolateral fragment: three (possibly four) foramina are present in an irregularly spaced row above alveoli 1–3. Because of the damage to the lateral surface of the maxilla, a distinct border to the antorbital fossa cannot be recognised. As in many rauisuchians, the antorbital fenestra is triangular in shape, and tapers in dorsoventral depth anteriorly ( Fig. 1A, B ). The anterior margin of the fenestra is smoothly rounded in Teratosaurus , similar to the condition in Arganasuchus ( Jalil and Peyer 2007) , Batrachotomus ( Gower 1999) , Fasolasuchus ( Bonaparte 1981) , and Postosuchus kirkpatricki ( Chatterjee 1985) . Prestosuchus ( Barberena 1978) and Saurosuchus ( Alcober 2000) exhibit a slightly different morphology, in which the fenestra tapers more sharply anteriorly, due to an ascending ramus oriented closer to horizontal than in the taxa listed above.

The ascending process projects caudodorsally at an angle of approximately 35 ° to the long axis of the main body. Few details of the morphology of this process can be determined due to its poor preservation and incomplete preparation (see above), although its rostral margin does appear to be slightly thickened. It is clear that the maxilla does not possess a separate rostral ramus, which in some rauisuchians is present as a discrete projection separated from the ascending ramus by a concave step; e.g., Arganasuchus ( Jalil and Peyer 2007) , Arizonasaurus ( Nesbitt 2005) , Batrachotomus ( Gower 1999) , Effigia ( Nesbitt 2007) , Fasolasuchus ( Bonaparte 1981) , Ticinosuchus ( Krebs 1965) , Yarasuchus ( Sen 2005) . Instead, Teratosaurus and other rauisuchians without a rostral ramus (e.g., Postosuchus kirkpatricki , Saurosuchus ) exhibit a smoothly continuous and convex rostral margin of the ascending ramus.

At the rostral end of the main body, the maxilla possesses a bevelled, rostromedially facing surface (parts of which are present on both pieces of the maxilla) that would have laterally overlapped the premaxilla ( Figs. 1 , 2 ). This surface is concave both rostrocaudally and dorsoventrally. A similar articular surface has been described for Batrachotomus ( Gower 1999: “a.pmx” on fig. 4) but in that taxon the surface is proportionally shorter dorsoventrally and forms a more distinct lobe−like process, restricted to the rostroventral corner of the main body. Immediately dorsal and lateral to this articular surface, a foramen is present that opens into a caudodorsally extending canal ( Fig. 2B ). As mentioned briefly by Galton (1985), the breakage of the lateral surface of the maxilla has exposed the route of this canal, which is represented by sediment infilling. The canal arches caudodorsally, reaching its maximum dorsal position above alveolus 3; it then arches caudoventrally and expands in dorsoventral depth, and is visible until a point level with the caudal margin of alveolus 4. This foramen and canal was identified by Galton (1985: “fo” on fig. 1E, “f”, “c” on pl. 1: 1) as possibly housing a branch of the maxillary artery and the inferior orbital nerve; we use the term rostrolateral foramen for this opening ( Fig. 2 ; Sulej 2005: fig. 4A–C).

Galton (1985) suggested that a subnarial fenestra was present between the premaxilla and maxilla in the holotype of Teratosaurus . This opening is common in rauisuchians, and is usually expressed as a dorsoventrally elongate slit that is at least three times as deep as long anteroposteriorly; e.g., Postosuchus kirkpatricki ( Chatterjee 1985) , Saurosuchus ( Alcober 2000) , Yarasuchus ( Sen 2005) . However, some taxa possess a smaller and more circular opening in this region, which is likely a homologous feature (see discussion in Gower 2000; e.g., Batrachotomus ). Contra Galton (1985), the presence and morphology of this opening cannot be reliably assessed in the holotype of Teratosaurus , since the premaxilla is not preserved. Regardless, features of the maxilla suggest that this opening may have been small or entirely absent. Most importantly, the rostral surface of the maxilla is nearly vertical and lacks a distinct notch, which is often clearly visible in taxa that possess a distinctive subnarial opening ( Chatterjee 1985: fig. 3; Alcober 2000: fig. 11).

On the medial surface of the maxilla, immediately caudodorsal to the premaxillary articular surface, there is a palatal process (anteromedial process of Galton 1985), which is directed rostroventrally, but is broken and incomplete at its rostral margin ( Figs. 1 , 2 ). The rostrolateral portion of this process is gently bevelled, probably for articulation with the posteriorly projecting maxillary process of the premaxilla. The preserved medial surface of the palatal process is generally convex along a rostrodorsal to caudoventral axis; however, at the caudoventral corner of the process this convexity is broken by two subparallel rostroventral−to−caudodorsally extending grooves. Of these two grooves, the more rostral in position is shallow and subtly developed, whereas the more caudal groove is deeply incised and at its posterior end merges smoothly with the medial surface of the maxilla. These grooves are likely for articulation with the opposing maxilla, and possibly the vomer. The palatal process overhangs a large and deep foramen that opens into a caudodorsally extending canal—this is the rostromedial foramen ( Fig. 2 ; Sulej 2005) and is approximately twice the size of the rostrolateral foramen described above.

The medial surface of the maxilla is generally well−preserved ( Fig. 1B ). The most prominent feature of the medial surface is a deep and sharp groove, which separates the main body of the maxilla from the interdental plates. This groove, identified by Galton (1985) as the “dental groove”, is present in archosaurs generally, and likely housed the soft tissue dental lamina ( Edmund 1960; Brusatte and Sereno 2007). However, in T. suevicus the groove is especially prominent and sharp, and demarcates a distinct step between the main body and the interdental plates. The groove is not as sharp or distinct in T. silesiacus ( ZPAL AbIII/563), Postosuchus kirkpatricki ( TTUP 9000), and other rauisuchians in general, and its morphology is thus autapomorphic for T. suevicus . Ventral to the groove, the interdental plates are fused into a continuous lamina, and even the much smaller posterior plates are heavily fused. Fusion of the interdental plates is uncommon in crurotarsans, but is also present in Postosuchus kirkpatricki ( TTUP 9000) and Fasolasuchus ( Bonaparte 1981) , and may be phylogenetically informative. The plates are unfused in T. silesiacus ( ZPAL AbIII/563). Despite the fusion, it is apparent that individual plates of T. suevicus are large and nearly square shaped anteriorly, the largest being above alveoli 2–4, but taper in depth and become more triangular posteriorly. Their medial surfaces are covered in a series of fine pits, which results in a roughened surface texture. Dorsally, the plates are separated by small replacement foramina, one for each tooth, which run along the sharp groove for the dental lamina. Ventrally, adjacent plates are separated by small triangular notches.

A prominent elliptical depression on the medial surface, above alveolus 7, was identified as the infraorbital foramen by Galton (1985). This feature is also present in Arganasuchus ( Jalil and Peyer 2007) , Arizonasaurus ( Nesbitt 2005) , Batrachotomus ( Gower 1999) , and Postosuchus kirkpatricki ( Chatterjee 1985) , and is probably a more general archosaur character. Other depressions on the medial surface may also represent foramina (e.g., Benton 1986: fig. 1), but are incompletely preserved and cannot be distinguished unambiguously from damage. Below the “infraorbital foramen” (not above it, as shown by Galton 1985: fig. 3A), and above alveoli 6–10, there are three rostrocaudally extending ridges that represent the sutural surface for the palatine. These ridges face almost entirely medially.

In total, parts of eight teeth can be identified, and there are 13 alveoli ( Fig. 1 ). Alveolus 1 is considerably smaller in circumference than subsequent alveoli, and it contains an unerupted replacement tooth that is visible on the medial surface of the small rostrolateral fragment. This crown is considerably smaller than subsequent crowns, and is finely serrated on rostrolateral and caudomedial surfaces. Alveolus 2 contains a large and incompletely erupted crown; similarly incompletely erupted crowns are present in alveoli 4 and 6. Fully erupted crowns are present in alveoli 3 and 5. Galton (1985:7) also suggested that a fully erupted crown was present in alveolus 7, but has “most of [the] crown missing”, whereas Benton (1986: 295) suggested that the tooth in alveolus 7 was “small”. However, only fragments of this crown are visible, and only in lateral view where the lateral wall of the maxilla has broken away. In ventral view, the alveolus itself is infilled with matrix, such that it is not possible to determine the degree of eruption. Although Benton (1986) suggested that the posterior six sockets lack teeth, part of a crown is visible within alveolus 10, through the replacement foramina. The teeth are slightly recurved with oval cross−sections and finely serrated mesial and distal carinae. Erosion makes it difficult to confirm the presence of transverse enamel wrinkles, which are actually present on most rauisuchians (e.g., Batrachotomus : SMNS 52970) despite previously being considered a feature of derived theropod dinosaurs (Brusatte et al. 2007). Comparison with rauisuchian material from Poland.— Sulej

(2005) described a partial skull and postcranial skeleton of a rauisuchian from the Carnian of Poland ( ZPAL AbIII/563) as a new species of Teratosaurus , T. silesiacus . The entire preserved skeleton of the holotype specimen of Teratosaurus silesiacus , including the maxilla ( Fig. 3 ), was described in detail by Sulej (2005) and does not warrant redescription here. Overall, the maxilla of T. silesiacus is very similar to that of T. suevicus , but several differences are apparent.

Sulej (2005) proposed a number of differences between the two species of Teratosaurus , most importantly several characters apparently present in T. silesiacus but not in T. suevicus : (1) rostromedial foramen of the maxilla on the medial surface; (2) ascending process of maxilla is strongly oblique; (3) foramina for replacement teeth are not connected together by a dental groove; (4) foramina for replacement teeth are set in a straight line. All of these characters were considered either inaccurate or of questionable significance by Lucas et al. (2007). In order to reassess the taxonomic status of Teratosaurus silesiacus , we begin by reassessing each of these proposed differences. Subsequently, we note a number of other differences based upon our re−examinations of both specimens.

(1) Rostromedial foramen orientation: Sulej (2005) suggested that the rostromedial foramen of T. suevicus opens rostroventrally and is visible in rostral view; by contrast, the equivalent foramen of T. silesiacus opens medioventrally and is not clearly visible in rostral view. Lucas et al. (2007) accepted this difference as genuine, but suggested that it could result from taphonomic processes. During our re−examination of the maxillae of both T. suevicus and T. silesiacus we have been unable to recognise a significant difference in the orientation of the opening of this foramen. Importantly, the foramen of T. suevicus is visible in rostral view largely as a result of the breakage of the rostral end of the palatal process. Therefore, we do not consider this feature to represent a genuine difference.

(2) Orientation of the ascending process: In order to assess the obliquity of the ascending process in both species, we measured the angle between a line drawn through the base of the palatal process and the midpoint of the caudodorsal margin of the ascending process, and a line drawn between the base of the palatal process and the midpoint of the caudal end of the main body. In both taxa this angle was approximately 35 °. Therefore, there does not appear to be a significant difference in the obliquity of the ascending process in the two taxa.

(3) Dental groove on medial surface: Lucas et al. (2007) suggested that a distinct dental groove connecting replacement foramina on the medial surface of the maxilla, a distinctive feature of T. suevicus , is also present in T. silesiacus . As outlined above, we consider the distinct and sharp morphology of this groove to be an autapomorphy of T. suevicus . There is a corresponding groove present in T. silesiacus ( Fig. 3B ), as occurs in archosaurs generally. However, this groove is much subtler than that of T. suevicus , and in places even appears to be entirely absent (most clearly dorsal to the interdental plate overlying alveolus 3). Therefore, we consider this difference to be a valid distinction between T. silesiacus and T. suevicus ( Sulej 2005; contra Lucas et al. 2007), and the sharp groove as the sole autapomorphy diagnosing T. suevicus .

(4) Orientation of tooth replacement foramina: Lucas et al. (2007) suggested that the alignment of the foramina for replacement teeth was no straighter in T. silesiacus than in T. suevicus , and dismissed this proposed difference. In fact, in T. suevicus the replacement foramina are positioned in a straight line along most of the tooth row ( Fig 1B ). However, at the rostral end the first two foramina are positioned more ventrally, with the first foramen being nearly adjacent to the alveolar margin ( Fig. 2A ). As a result, the dental groove connecting these replacement foramina curves ventrally at its rostral end. This is presumably due to the reduced size of the first maxillary crown in T. suevicus (discussed in greater detail below). In contrast, in T. silesiacus the replacement foramina are set in a completely straight line, and do not arch ventrally towards the alveolar margin at the rostral end ( Fig. 3B ). Thus, we consider this difference to be a valid distinction between T. silesiacus and T. suevicus ( Sulej 2005; contra Lucas et al. 2007).

A number of other differences between the maxillae of T. silesiacus and T. suevicus exist, which have not been previously noted or discussed. As noted above, the first (most rostrally positioned) alveolus in T. suevicus is substantially smaller than subsequent alveoli: its rostrocaudal length is less than a third of the length of alveolus 3 ( Fig. 1C ). The unerupted crown which occupied this alveolus is likewise much smaller than subsequent crowns. No such size difference between the most rostral alveolus and subsequent alveoli is observed in T. silesiacus : the first alveolus is only very slightly smaller in size than alveoli 2–4 ( Fig. 3C ). This difference cannot simply be a result of differences in the degree of eruption, because the first maxillary tooth is unerupted in the holotype specimens of both taxa ( Sulej 2005). Furthermore, a very small first alveolus is seen in some ( Batrachotomus [ Gower 1999], Postosuchus kirkpatricki [ TTUP 9000]) but not all ( Arizonasaurus [ Nesbitt 2005], Prestosuchus [ Barberena 1978]; possibly Tikisuchus [ Chatterjee and Majumdar 1987]) rauisuchians, and may be a character of phylogenetic significance for rauisuchian interrelationships.

The ventral margin of the maxilla in T. suevicus is straight along its caudal half (alveoli 7–13) and gently convex rostrally. By contrast, in T. silesiacus the ventral margin has a sinuous outline in lateral view: from the caudal end it arches upwards (reaching its highest point below alveolus 8) and then is strongly convex at the rostral end (maximum convexity is below alveolus 4) ( Fig. 3A, B ).

The morphology of the palatal process also differs between the two species. In both species a pair of rostroventrally−to−caudodorsally extending grooves is present on the caudoventral corner of the palatal process, for articulation with the opposing maxilla and possibly the vomer. In T. silesiacus the more caudal of these grooves is notably short in length and lacks a well−defined ventral margin; moreover, at its posterior end it is set above the level of the medial surface of the maxilla ( Fig. 3B ). In contrast, in T. suevicus this groove is proportionally much longer and merges smoothly with the medial surface of the maxilla (see above) ( Fig. 1B ). In addition, in T. silesiacus there is a prominent and sharply defined caudodorsally arched ridge that extends from the base of the palatal process and fades out above alveolus 3; no such ridge is present in T. suevicus despite the good preservation of this area. In T. silesiacus there is a prominent large fossa (possibly pneumatic and associated with the antorbital sinus of archosaurs: Witmer 1997) dorsal to the palatal process and the caudodorsally arched ridge. Although the equivalent area is poorly preserved and somewhat crushed in T. suevicus , an equivalent prominent fossa does not appear to have been present.

In addition to these anatomical differences, we also note significant differences in both size and stratigraphic position between T. silesiacus and T. suevicus . T. suevicus is significantly larger than T. silesiacus (maximum rostrocaudal length of the maxilla is 165 mm in T. silesiacus and 245 mm in T. suevicus ). It is possible that T. silesiacus could represent an earlier ontogenetic stage of T. suevicus —the limited available material makes assessing the ontogenetic stage of either taxon difficult at present, particularly given the lack of data in general on ontogeny in basal archosaurs ( Irmis 2007). However, the neurocentral sutures are fused in all preserved vertebrae in T. silesiacus (the axis, third cervical, caudals) which is consistent with a mature ontogenetic stage in archosaurs ( Irmis 2007). The stratigraphic incongruity (with T. silesiacus known from the late Carnian, and T. suevicus from the middle Norian) cannot by itself be used to support specific level distinction; however, in combination with anatomical and size differences, it may be informative.

In summary, we consider two of the four differences between T. silesiacus and T. suevicus proposed by Sulej (2005) to be valid. We also recognise an additional four anatomical characters that differ between these taxa. Combined with differences in size and stratigraphic position, these anatomical differences provide strong evidence to support the distinctiveness of T. silesiacus . We therefore reject the proposal of Lucas et al. (2007) that T. silesiacus is a junior synonym of T. suevicus .

Taxonomy of the Polish rauisuchian material.—Clearly, the German and Polish rauisuchian specimens are distinct and belong to different taxa. With this established, it must now be determined whether these taxa can be united as sister taxa relative to other rauisuchians, the minimum requirement needed to retain the Polish material ( ZPAL AbIII/563) in the genus Teratosaurus . If there is no such evidence, then the Polish taxon must be given a new generic name.

Ideally, in these cases it is best to use a phylogenetic analysis to test the validity of a genus, which in actuality is simply a monophyletic clade exclusive to all other taxa considered as genera in a given taxonomy. In this case, the fundamental question is whether the two species ( T. suevicus and T. silesiacus ) comprise a clade relative to Postosuchus and other rauisuchians. However, this is extremely problematic, as T. suevicus is only known from a small part of the skeleton (the maxilla) and including it in an analysis is likely to drastically increase the number of most parsimonious trees and may give a spurious result due to missing data alone. Furthermore, widespread missing data renders it likely that T. suevicus would be removed from the phylogeny altogether by certain consensus techniques that are useful for large analyses with extensive missing data and homoplasy (e.g., reduced consensus: Wilkinson 1994), a reality for higher−level archosaur analyses ( Gower and Wilkinson 1996; Brusatte et al. 2008). Thus, we are hesitant to rely on phylogenetic analysis to test the validity of Teratosaurus based on data in hand, and instead focus our attention on diagnostic characters, both autapomorphies and unique combinations of characters.

As the holotype of the type species of Teratosaurus , T. suevicus Meyer, 1861 , is limited to a single maxilla ( NHM 38646) and no other specimens can be definitely referred to this species (see above), any diagnostic features of the genus must be identified only on this element. The maxilla of Teratosaurus is extremely similar to that of Postosuchus kirkpatricki , and the two may be closely related as suggested by Sulej (2005). Our comparisons indicate that there are no unique synapomorphies shared by the German ( NHM 38646) and Polish ( ZPAL AbIII/563) maxillae that unambiguously differentiate them from Postosuchus kirkpatricki and other rauisuchians. In addition, we cannot even identify a unique combination of maxillary characters shared by the German and Polish material to the exclusion of Postosuchus kirkpatricki .

One possible feature that appears to be different in Postosuchus kirkpatricki and the German + Polish material is the size of the “subnarial” opening between the maxilla and premaxilla. This opening has been figured as large and slitlike in Postosuchus kirkpatricki ( Chatterjee 1985; Long and Murry 1995) and as small or nonexistent in the German ( Benton 1986) and Polish ( Sulej 2005) material. However, the large size of the opening in Postosuchus kirkpatricki has been exaggerated in published figures, and the true size of this fenestra is similar to that in both the German and Polish material (Sterling Nesbitt, personal communication 2008).

Thus, as (1) the type species of Teratosaurus , T. suevicus , is valid; (2) T. suevicus and Polish material referred to T. silesiacus are distinct; and (3) there are no autapomorphies or a combination of characters that unite T. suevicus and the Polish material relative to Postosuchus kirkpatricki and other rauisuchians, a new generic name must be erected for the Polish material.

Figures

Fig. 1. Right maxilla of rauisuchian Teratosaurus suevicus Meyer, 1861 from Mittlerer Stubensandstein, Upper Triassic of Heslach, Germany, NHM 38646, holotype. Photographs in lateral (A), medial (B), and ventral (C) views. Designation “m” refers to maxillary tooth position.

Fig. 2. Right maxilla of rauisuchian Teratosaurus suevicus Meyer, 1861 from Mittlerer Stubensandstein, Upper Triassic of Heslach, Germany, NHM 38646, holotype. Photographs in medial (A) and anterior (B) views. A is a close−up of the medial surface of the anterior end of the maxilla. Designation “m” refers to maxillary tooth position.

Fig. 3. Left maxilla of rauisuchian Polonosuchus silesiacus (Sulej, 2005) from Late Carnian, Upper Triassic of Krasiejów Claypit, Poland, ZPAL AbIII/563, part of holotype. Photographs in lateral (A), medial (B), and ventral (C) views. Designation “m” refers to maxillary tooth position.