1.1. Early collections

Early field collecting trips to Upper Triassic rocks in New Mexico and Texas in the 1870s and 1880s by Edward Cope and his field collector David Baldwin resulted in the collection of numerous fossils, including two phytosaurs (e.g. AMNH FR 2318, type specimen of Machaeroprosopus buceros; unnumbered type specimen of Belodon scolopax, possibly lost) and much early dinosauriform material, including the original type materials of Coelophysis bauri (Cope Reference Cope1887b), C. longicollis (Cope Reference Cope1887b) and C. willistoni (Cope Reference Cope1887a), as well as material recently recognised as belonging to the dinosauromorph Dromomeron romeri (Nesbitt et al. Reference Nesbitt, Irmis, Parker, Smith, Turner and Rowe2009). In addition, materials were collected that formed the basis for two new archosaurian taxa. The first of these was Typothorax coccinarum, named by Cope (Reference Cope1875a) for osteoderms fragments, vertebral fragments and coprolites (USNM 2585, USNM 2586) from the Gallina area in the Chama Basin of north-central New Mexico. Cope (Reference Cope1887a) subsequently assigned better preserved material collected by Baldwin from the same locality in 1881 to T. coccinarum, including two femora and more complete osteoderms (AMNH FR 2709, AMNH FR 2710). Baldwin's collection contained other osteoderms, vertebrae and limb bones (AMNH FR 2713) that Cope utilised to erect a second new taxon, Episcoposaurus horridus. At some point, presumably in the late 1800s, a small collection of material of putative syntype material of Typothorax coccinarum was made and reposited at the Academy of Natural Sciences in Philadelphia (ANSP 14695; Spamer & Daeschler Reference Spamer and Daeschler1995; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002).

Cope (Reference Cope1892) erected a third taxon, Episcoposaurus haplocerus, for another collection made by Baldwin, this time from the panhandle of Texas, which included osteoderms and vertebrae (ANSP 14688). Apparently, Cope personally returned to the Texas site after Baldwin's initial visit and collected additional materials from the site, including a complete ilium (currently part of ANSP 14688) and more vertebrae and osteoderms (TMM 40041-3, TMM 40041-5); however, these materials were not included in the original description (Wilson Reference Wilson1950; Gregory Reference Gregory1953; Parker Reference Parker2008a).

Case (Reference Case1920) erected the name Desmatosuchus spurensis for a skeleton (UMMP 7476) of another purported phytosaur from the panhandle of Texas, not far from where Baldwin had made his collection, and subsequently provided more thorough descriptions of that material (Case Reference Case1921, Reference Case1922). Desmatosuchus was diagnosed by the presence of enlarged recurved horns on the armour of the cervical region, a character also found in Episcoposaurus haplocerus, which led some researchers to suggest that the two were synonymous (e.g. Sawin Reference Sawin1947). That specimen (UMMP 7476) was also the first North American form to preserve the skull, which demonstrated that the cranium in these specimens was short-snouted and very different from typical phytosaurs (Case Reference Case1920, Reference Case1922).

1.2. Subsequent reviews

Huene (Reference Huene1915) redescribed much of Cope's New Mexican material, including the referred material of T. coccinarum (mistakenly referred to by Huene as the type specimen), and the syntype material of E. horridus. Huene (Reference Huene1915) correctly recognised that the E. horridus material belonged to more than one individual, and removed elements from the syntype material. In particular, he removed the forelimb materials, which he assigned to a small ‘mystriosuchid’ phytosaur; however, this was mostly done based on element size (in comparison to the massive hind limb elements in the material), rather than by utilising specific morphological characters.

The first comprehensive review of these materials was completed by Gregory (Reference Gregory1953) in his study of North American phytosaurs. Gregory (Reference Gregory1953) argued that Huene's (Reference Huene1915) removal of the forelimb material from the type of E. horridus (AMNH FR 2713) was premature and that, in comparison to the skeleton of the recently described “Typothorax” (=Longosuchus) meadei (Sawin Reference Sawin1947), the hindlimb and forelimb material could belong to the same animal. Gregory (Reference Gregory1953) also provided the first detailed description and figures of the type material of Episcoposaurus haplocerus (ANSP 14688). That author argued convincingly for the synonymy of Desmatosuchus spurensis and Episcoposaurus haplocerus, as well as that of Episcoposaurus horridus and Typothorax coccinarum. Because Episcoposaurus was now a junior synonym of Typothorax, the E. haplocerus material was assigned to Desmatosuchus as D. haplocerus. As noted above, Gregory recognised that the aetosaur material was mixed with some phytosaur material and removed some elements (e.g. ilium from ANSP 14688), but considered an attempt to completely separate the specimens “hopeless”.

Heckert & Lucas (Reference Heckert, Lucas, Heckert and Lucas2002) figured and provided brief redescriptions of some of this material (USNM 2585, USNM 2586, ANSP 14688, ANSP 14865) and provided lectotype designations. Lucas et al. (Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007) briefly discussed and figured the referred specimens of Typothorax coccinarum (AMNH FR 2709, AMNH FR 2710) and the syntypes of Episcoposaurus horridus (AMNH FR 2713). They also proposed a lectotype for E. horridus, but unfortunately they mistakenly considered much of AMNH FR 2709 and AMNH FR 2710 to be part of the syntype series (Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, p. 242).

4.1. Holotype specimens

USNM 2585 and USNM 2586 consist of a total of eleven bone fragments and three coprolites. These specimens were described originally by Cope (Reference Cope1875 a, Reference Copeb) and figured by Cope (Reference Cope1877). They were later redescribed by Heckert & Lucas (Reference Heckert, Lucas, Heckert and Lucas2002). The figure numbering from Cope's Reference Cope1877 publication is still apparent on each specimen, making correlation of elements with Cope's description simple. Moreover, although all of the elements are mentioned here, Cope (Reference Cope1887a) emended the type specimens to only include the osteoderm fragments. Heckert & Lucas (Reference Heckert, Lucas, Heckert and Lucas2002, fig. 4g–h, k–l) also figured three fragments as part of USNM 2585, but these are actually NMMNH specimens collected in the 1980s (NMMNH P-18197, NMMNH P-18199, NMMNH P-18201) purportedly from the type locality, and are therefore nothing more than possible topotypes (Lucas & Hunt Reference Lucas and Hunt1992).

The jaw fragment from USNM 2585 (Fig. 1a; Cope Reference Cope1877, pl. 12, fig. 1a–c) is a robust element about 10 cm in length, with a flat medial symphyseal surface and at least four alveoli on the dorsal surface. The morphology of this element is consistent with that of a partial fragment of the left dentary of a phytosaur, a conclusion reached by Cope (Reference Cope1887a), who explicitly removed this element from the syntypes of Typothorax coccinarum.

The partial vertebral centrum from USNM 2585 (Fig. 1b; Cope Reference Cope1877, pl. 12, fig. 2) is fragmentary and poorly preserved. It is mediolaterally compressed, with flared rims consistent with an archosauriform caudal centrum; however, no chevron facets are preserved on the ventral surface, suggesting that if this identification is correct it is the anterior portion of the centrum that is preserved. Given the presence of a partial phytosaur dentary with the material, its referral to Typothorax cannot be confirmed. Cope (Reference Cope1887a) also removed this specimen from the syntype materials.

The largest osteoderm fragment in USNM 2585 (Fig. 1c; Cope Reference Cope1877, pl. 12, fig. 4; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 4c–d) is a weathered portion of a paramedian osteoderm. The dorsal surface is ornamented with densely packed small oval pits typical on portions of the paramedian osteoderms of several aetosaur taxa, including various paratypothoracisines (e.g. PEFO 26690, PEFO 26725) and T. coccinarum. The ventral surface is flat and smooth.

A small (>3 cm length) osteoderm fragment in USNM 2585 (Fig. 1d; Cope Reference Cope1877, pl. 12, fig. 5; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 4e–f) probably also originates from a paramedian osteoderm. Heckert & Lucas (Reference Heckert, Lucas, Heckert and Lucas2002, p. 198) described this as a left caudal lateral osteoderm based on the presence of a “pronounced spike” and presumably deeper pitting. However, the ornamentation is no deeper than osteoderms from the dorsal series and the ‘spike’ is actually the osteoderm broken into a jagged fragment.

USNM 2586 contains a larger (∼4 cm long) paramedian osteoderm fragment (Fig. 1e; Cope Reference Cope1877, pl. 12, fig. 9; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 4a–b). This fragment possesses an dorsal surface ornamentation of ovate pits separated by a reticulate pattern of ridges, consistent with that found in Typothorax, but which also occurs in portions of paratypothoracisin osteoderms (e.g. PEFO 26690, PEFO 26725). Heckert & Lucas (Reference Heckert, Lucas, Heckert and Lucas2002) stated that this fragment is referable to Typothorax by the presence of a transverse ventral strut, but this feature is also found in some paratypothoracisins (e.g. Paratypothorax, PEFO 3004).

USNM 2586 also contains two unidentifiable bone fragments (Cope Reference Cope1877, pl. 12, figs 7–8), a phytosaur osteoderm fragment (Cope Reference Cope1877, pl. 12, fig. 5), a fragmentary tooth crown (Cope Reference Cope1877, pl. 12, fig. 13), and three coprolites (Cope Reference Cope1877, pl. 12, figs 10–12). The three bone fragments are morphologically and taxonomically indeterminate and were removed from the type materials by Cope (Reference Cope1887a). Cope (Reference Cope1875a, Reference Copeb) originally referred the tooth crown to Typothorax, but was less certain of this referral in his 1887 paper and removed it accordingly. The tooth is not aetosaurian and instead belongs to a phytosaur, because it is elongate, D-shaped in apical view and possesses distinct mesial and distal carinae with fine serrations (Hungerbühler Reference Hungerbühler2000). Cope (Reference Cope1875a, Reference Copeb) also described the coprolites as full of fish scales, but argued that they were too small to belong to the associated Typothorax specimen.

In summary, Cope (Reference Cope1887a) removed from his type specimens all of the materials that were not osteoderm fragments with the characteristic pitted sculpturing. That emendation of the type materials is supported here because none of the removed specimens possess apomorphies of Typothorax. However, assignment of the remainder of the material to Typothorax is also problematic because the fragments are very small, and the characteristics possessed by the fragments that are used to assign them to Typothorax coccinarum (e.g. ovate pitting, reticular sculpture, ventral strut) are found also in portions of paratypothoracisin osteoderms. The possible taxonomic implications of this will be discussed in more detail below.

4.2. ANSP specimens

ANSP 14695 consists of a paramedian osteoderm fragment, a dorsal vertebral centrum, four bone fragments, and a coprolite which are purported to represent syntypes of Typothorax coccinarum (Spamer & Daeschler Reference Spamer and Daeschler1995, fig. 15a–g; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 5a–g). Of this material, only the osteoderm fragment is referable to Typothorax; the centrum (Fig. 2a) most likely belongs to a phytosaur (Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002), and the other fragments are taxonomically indeterminate. The osteoderm fragment (Fig. 2b) possesses a smooth raised anterior bar as well as the dorsal surface pitting that is characteristic of the genus. The element is too incomplete to determine if a dorsal eminence was present; however, the ventral surface possesses a thickened ventral strut (Fig. 2c).

The provenance of these specimens is problematic. There is a field note (Fig. 2d) with the material in Cope's hand which reads “Plate 7–18, spine, Typothorax? coccinarum, 547”. There are two more words written on the tag which Spamer & Daeshler (Reference Spamer and Daeschler1995) interpreted as “Texas Trias”; the word Trias is written clearly, however, the word “Texas” is not. If this specimen is indeed from Texas, it cannot be part of the syntype series of Typothorax coccinarum, which was collected from New Mexico. Because they are reposited in the ANSP with the specimens of Episcoposaurus haplocerus, which are from Texas, and because Cope's field note shows uncertainty in the taxonomic assignment by adding the question mark it would appear that these specimens are also probably from Texas and not part of the type series.

4.3. Original referred specimens (AMNH FR 2709, AMNH FR 2710)

4.3.5. Other elements from AMNH FR 2710

Other material from AMNH FR 2710 and considered to be part of the referred materials by Huene (Reference Huene1915) include a proximal end of a right tibia (Fig. 6a; Huene Reference Huene1915, fig. 2; Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 5b) and the proximal end of a left ischium (Fig. 6b; distal scapula of Huene Reference Huene1915, fig. 3; Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 5f–g). These specimens were also considered to be topotypes of T. coccinarum by Lucas et al. (Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007); however, they both belong to a phytosaur and not to any aetosaur.

Figure 6

AMNH FR 2710: (a) proximal end of right phytosaur tibia in medial view; (b) proximal end of left phytosaur ischium in lateral view; (c) proximal end of phytosaur cervical rib in oblique view. Scale bar=5 cm. Arrow shows anterior direction. Abbreviations: a.il=articular surface with the ilium; ac=acetabulum; bf=bone fragment; rh=rib head.

Lucas et al. (Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 2j) identified an element from AMNH FR 2710 (Huene Reference Huene1915, fig. 4) as a right (?) lateral osteoderm; however, the morphology of this element is not consistent with any osteoderm, and the presence of a three-pronged proximal surface (Fig. 6c) is strongly suggestive of cervical ribs from a phytosaur and consistent with Huene's (Reference Huene1915) original identification.

4.4. Summary

Reexamination of the original type and referred materials (USNM 2585, USNM 2586, ANSP 14695, AMNH FR 2709, AMNH FR 2710) demonstrates that only the following elements are most likely referable to Typothorax coccinarum. In USNM 2865, only three osteoderm fragments possess characters considered to be diagnostic of Typothorax, specifically the random pattern of circular round pits (Long & Ballew Reference Long and Ballew1985). This is also the case for the single osteoderm fragment from the purported syntype specimen ANSP 14695. Furthermore, this latter element possesses the ventral thickening also considered typical for the genus. The referred material (AMNH FR 2709, AMNH FR 2710) mostly is assignable to T. coccinarum based on the ornamentation and morphology of the osteoderms, as well as on the association of the partial left femur and morphological correspondence of the right femur. Moreover, these femora possess characteristics found only in aetosaurs and not in phytosaurs supporting their referral. The other material in these samples is either non-diagnostic or referable to a phytosaur.

5.1. Dorsal region osteoderms

AMNH FR 2713 contains two dorsal lateral osteoderms, one from the left side and the other from the right. The right side osteoderm (Fig. 7a–c; Huene Reference Huene1915, fig. 23a–b; Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 3c–d) is the smaller of the two, with a length of 84 mm, and consists of distinct dorsal and lateral flanges at about 90° to each other and separated by a distinct sharp anteroposteriorly aligned ridge (dorsal eminence). In medial view, the ridge rises gently to an apex situated near the posterior osteoderm margin, and posterior to this apex plunges sharply posteroventrally terminating at the osteoderm margin (Fig. 7a). The dorsal flange is reduced greatly compared to the lateral flange, as is typical for typothoracisin aetosaurs, and it has a gently rounded medial edge that articulates with a corresponding paramedian osteoderm (Fig. 7b–c). The lateral flange is dorsoventrally deep, and possesses a gently rounded ventral margin (Fig. 7c). The dorsal portion of the anterolateral margin bears a raised anterior bar that extends slightly onto the dorsal flange. The posterior margin of the anterior bar is straight and slopes dorsolaterally and anteriorly. In lateral view, the posteroventral margin of the lateral flange slopes posteroventrally and medially, forming a 90° angle with the anterior bar. The dorsal surface ornamentation appears to have been obliterated during removal of the adhering matrix and large patches of matrix remain. Comparison with known articulated Typothorax specimens (e.g. NMMNH P-56299; Heckert et al. Reference Heckert, Lucas, Rinehart, Celeskey, Spielmann and Hunt2010) suggests that this is an anterior dorsal lateral osteoderm and was originally situated anterior to the tenth dorsal position.

The left side osteoderm (Fig. 7d–e; Huene Reference Huene1915, fig. 22a–b; Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 2i) is larger (length of 94 mm) and therefore from further back in the carapace. This osteoderm is very similar to the one previously described, except that the angle between the flanges is more acute and the dorsal flange is larger in relation to the lateral flange, nearly two-thirds the size (Table 3). As with the right side, osteoderm the dorsal ornamentation is also obliterated or covered with matrix on this specimen; however, the raised smooth anterior bar is visible, despite the anteroventral portion being broken away. The inner surface of the osteoderm is still buried in matrix, giving it the triangular cross section depicted by Huene (Reference Huene1915, fig. 22b) and a small bone fragment of another element is adhered to the posterodorsal surface.

5.2. Caudal region osteoderms

Caudal paramedian osteoderm (Fig. 7f; Huene Reference Huene1915, fig. 24a, b; Gregory Reference Gregory1953, fig. 17a, b; Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 3a). The dorsal surface of this osteoderm is well preserved, but the ventral surface is still covered in matrix. Comparison with MCZ 1488 shows that this is a left dorsal caudal osteoderm from the distal portion of the middle of the tail (Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007). In dorsal view, the osteoderm is suboval in shape, with a straight medial edge and rounded anterior, lateral and posterior edges. The anterior portion possesses an anterior bar, which is flush with, rather than raised above, the rest of the dorsal surface sculpturing. There is a large pyramidal dorsal eminence in the posterior portion of the osteoderm, which does not contact the posterior plate margin. This eminence is situated slightly medial of centre and from its apex a thin ridge extends obliquely (anteromedially) to the anteromedial corner of the osteoderm, effectively dividing it into dorsal and lateral flanges. The area of the dorsal flange is approximately half that of the lateral flange. An ornamentation of elongate grooves and flat ridges radiates from the dorsal eminence.

5.3. Caudal vertebra

A single elongate amphicoelous vertebra is present in the material (Fig. 7g–h; Huene Reference Huene1915, fig. 19; Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 2a–c). The specimen is much longer than wide and hour-glass shaped in ventral view (Fig. 7g). Ventrally there are two chevron facets and a broad, yet weak ventral keel. Laterally, near the base of the centrum, there are horizontal keels that extend between the mid-points of the circular articular faces (Fig. 7h). A portion of the neural arch is present. The neural spine is broken, but must have been elongate and low. The postzygapophyses are broken away, whereas the prezygopophyses are mostly present but encased in matrix. The morphology of this element is consistent with that of an aetosaur and the size of the element is predicted by the size of the lectotype femur and the paralectotype caudal osteoderm. Thus, this element is retained in the lectotype material.

The anterior caudal vertebra (Huene Reference Huene1915, fig. 18) could not be located in the material.

5.4. Humeri

There are two right humeri under the catalog number AMNH FR 2713 (Fig. 8a–b). Because only a single humerus (Fig. 8a) was mentioned by Cope (Reference Cope1887a) and Huene (Reference Huene1915), the provenance of the second specimen (Fig. 8b) is not certain; however, both possess straight lateral margins and belong to phytosaurs (Nesbitt Reference Nesbitt2011). Furthermore, the area between the distal articular surfaces is flat, without the distinct deep sulcus found in aetosaurs (e.g. Martz Reference Martz2002, fig. 4.10). The larger of the two measures 220 mm in length and was described as part of the syntype material by Cope (Reference Cope1887a) and Huene (Reference Huene1915, fig. 25). Huene (Reference Huene1915) argued that the forelimb bones were too reduced in comparison to the hindlimb bones to belong to the same animal, and removed them from the syntype series and subsequently assigned them to a ‘mystriosuchid’-type phytosaur. In contrast, Gregory (Reference Gregory1953) argued that on the basis of comparison with the skeleton of Longosuchus meadei (Sawin, Reference Sawin1947), the limb proportions were acceptable and thus reinstated the forelimb elements as part of the syntype series. Unfortunately, he was not able to recognise that these elements were phytosaurian and therefore not referable to Typothorax or E. horridus. Likewise, Lucas et al. (Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 4a) also did not recognise the phytosaurian identity of these elements. Furthermore, they (p. 242) noted that the humerus was similar in morphology to a humerus (UCMP V2816/34240) figured by Long & Murry (Reference Long and Murry1995, fig. 105d–e) as Typothorax coccinarum. Unfortunately, this humerus was also misidentified and is phytosaurian as well (Martz Reference Martz2002). Therefore, the humerus of E. horridus is unknown.

5.5. Ulnae

AMNH FR 2713 contains portions of four ulnae, two complete (left and right; Fig. 8d–f) and two proximal ends (both rights; Fig. 8c, g). The left ulna (Fig. 8d) and a proximal end of a right (Fig. 8c) are about the same size and may represent a matching pair. Huene (Reference Huene1915, fig. 27) figured the left element and assigned it to a small phytosaur, whereas Gregory (Reference Gregory1953) and Lucas et al. (Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007) considered it to be aetosaurian. Lucas et al. (Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 4d–e) also figured a 5th specimen, which the present author was unable to locate in the materials.

Phytosaurian and aetosaurian ulnae are superficially similar in their overall morphology except for one key detail; in aetosaurs the glenoid surface is much expanded and possesses a well-defined rim and lateral tuber (Nesbitt Reference Nesbitt2011; Fig. 8h). Thus the left ulna figured by Huene (Reference Huene1915) and Lucas et al. (Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007) (Fig. 8d) belongs to a phytosaur, as does its possible counterpart (Fig. 8c) and the complete right ulna (Fig. 8e, f, i). However, one of the right proximal ends (Fig. 8g–h) possesses aetosaurian characteristics. In proximal view, the concave glenoid fossa is subrectangular, unlike the semilunate surface in phytosaurs (compare Figs 8h and 8i). The olecranon process is a raised and thickened knob on the posterior portion of the glenoid fossa. It does not taper posteriorly as described for T. antiquum (Lucas et al. Reference Lucas, Heckert, Hunt, Heckert and Lucas2002). The lateral portion of the glenoid fossa is a thickened rim that is expanded in relation to the lateral surface of the bone shaft, and ends in a distinct lateral (radial) tuber. The medial surface of the proximal head is flattened. The anteromedial corner of the glenoid fossa has a distinct anteromedially projecting lip that slightly overhangs the flattened medial surface of the head and is the medial extension of the radial tuber. The midshaft and distal end of this element is not preserved.

5.6. Radii

The proximal ends of two radii are present in the material (Fig. 9a–d). Huene (Reference Huene1915, fig. 26) figured one of them (Fig. 9a–b), although the more distal portion now appears to be missing (see Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 4F). The proximal surfaces are convex, ovate in proximal view and possess a flaring rim. The shaft is slightly ovate in shape. Among phytosaurs and aetosaurs these specimens most closely resemble the proximal radii of Typothorax antiquum (Lucas et al. Reference Lucas, Heckert, Hunt, Heckert and Lucas2002, fig. 7c), although assignment to a phytosaur cannot be completely ruled out because they are also similar. A distal end is also present with a rounded rather than compressed cross-section that corresponds nicely to an aetosaur rather than to a phytosaur (Fig. 9e–f; Nesbitt Reference Nesbitt2011); however, it is not possible to determine to which proximal head it belongs.

5.7. Femur

A very large and mostly complete right femur is the most striking bone of the collection, predominantly because of its size (Fig. 10a–f). Cope (Reference Cope1887a) considered this element to be characteristic for Episcoposaurus in that it was straighter and more robust than the femur of Typothorax (AMNH FR 2710), which prompted Huene (Reference Huene1915, fig. 12) to designate it as the lectotype. The specimen is separated into two pieces: (1) the proximal end and much of the shaft, which is 29·5·cm in length (Fig. 10a–b); and (2) the distal condyles, which are 9·8 cm in length (Fig. 10c–d). The complete element would have measured over 40 cm in length (Huene Reference Huene1915). The shaft is robust and straight, different from the more sigmoidal shaft in phytosaurs (e.g. PEFO 31219). The fourth trochanter is situated approximately 14 cm distal from the proximal end of the element. The trochanter is damaged, but is moundlike and rounded (sensu Nesbitt Reference Nesbitt2011, char. 316:0), not a sharp crescentic ridge as in AMNH FR 2710, and is located on the posterior face of the shaft (Fig. 10b). The shaft is oval in cross section and mediolaterally narrows slightly distal to the position of the 4th trochanter before expanding again more distally.

The femoral head is also badly abraded and much of the anteromedial tuber has been broken or worn away (Fig. 10e). Nonetheless, the three tubers are still apparent distinguishing this femur from that of phytosaurs (Nesbitt Reference Nesbitt2011). The proximal surface is marked by a deep, but narrow mediolaterally oriented sulcus, similar to what is present in other archosauriforms such as Effigia (Nesbitt Reference Nesbitt2007). Similar sulci are randomly present in other aetosaur and phytosaur specimens and appear to represent incomplete ossification of the proximal articular surface. The section with the distal condyles is poorly preserved, not completely prepared and relatively unremarkable, except for the presence of a broken, but prominent crista tibiofibularis (Fig. 10c–d, f).

5.8. Tibia

Distal and proximal portions of a large right tibia are present in AMNH FR 2713 (Huene Reference Huene1915, figs 13–14; Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 5c–e). Overall it is much more similar to TTU P-9214, than to the Canjilon Quarry material (UCMP V2816/122678; Martz Reference Martz2002), as it lacks the profound proximal expansion found in the UCMP specimens (Long & Murry Reference Long and Murry1995, fig. 111). The proximal end is poorly preserved and still marked with numerous tool marks from mechanical preparation (Fig. 10g). The posterior face possesses a distinct concave surface (Fig. 10h), whereas the anterior surface is convex. The lateral edge is thinner anteroposteriorly than the medial edge, broadening proximally into a distinct but broken posterolaterally directed tuber. Although the majority of the shaft is missing, the proximal portion that is preserved is smooth and slightly anteroposteriorly compressed. The distal end of the tibia (Fig. 10i–j) is crescent-shaped in ventral view, with a dorsoventrally oriented groove that provides the specimen with the typical “screw-joint” morphology (sensu Gauthier Reference Gauthier1986) found in suchians and not in phytosaurs (Nesbitt Reference Nesbitt2011).

5.9. Calcaneum

AMNH FR 2713 includes a very large and well preserved left calcaneum (Fig. 10k–n; Huene Reference Huene1915, fig. 15; Long & Murry Reference Long and Murry1995, fig. 112a–d; Lucas et al. Reference Lucas, Spielmann, Heckert, Hunt, Lucas and Spielmann2007, fig. 5k–m). This element is dorsoventrally compressed and very broad. The facet for the tibia on the dorsal surface is well-developed, convex and hemi-cylindrical in shape, typical for suchians (Fig. 10k; Sereno Reference Sereno1991; Nesbitt Reference Nesbitt2011). Posterior to this is a deep transverse trough which is confluent with the dorsal surface of the medial projection or lip that outlines the concave facet that is the articulation point for the astragalus (Brochu Reference Brochu1992). This trough (the dorsal portion of the calcaneal shaft sensu Nesbitt Reference Nesbitt2011) effectively separates the tibial facet from the posterior portion of the posteriorly directed calcaneal tuber, which is delimitated by a dorsoventrally expanded lip. Lateral to the tibial facet and dorsal trough is an ovate strongly concave depression that opens dorsolaterally, and probably articulated in part with the 5th metatarsal (Fig. 10l; Walker Reference Walker1961). The concave articulation point for the astragalus is shallow. The lip or spout for this articulation extends medially just past the medial expansion of the posterior tuber. The medial surface of the posterior tuber between the tuber distal expansion and the lip or spout is strongly concave (Fig. 10m). Ventrally, the articular surface for the 4th distal tarsal is flat. Posterior to this surface, the main body of the posterior tuber is deeply excavated into a mediolaterally broad rectangular fossa (Fig. 10n; the “clear gap” of Nesbitt Reference Nesbitt2011). A similarly deep depression is found in all aetosaurs and varies in form among taxa, but appears to be much less developed in the holotype specimen of Typothorax “antiquum” (NMMNH P-36075; Lucas et al. Reference Lucas, Heckert, Hunt, Heckert and Lucas2002, fig. 9a). The posterior surface of the calcaneal tuber in AMNH FR 2713 is only slightly concave, lacking the deep groove found in Alligator (Brochu Reference Brochu1992) and other paracrocodylomorphs (Nesbitt Reference Nesbitt2011).

5.10. Other material

5.11. Summary

Material from AMNH FR 2713 referable to Typothorax coccinarum includes the three osteoderms, a proximal end of a right ulna, the distal radius, the right femur, the left calcaneum, the tibia fragments and, probably, the proximal ends of the two radii. These elements can be grouped by size, with the caudal paramedian osteoderm, femur, tibia and calcaneum belonging to a larger specimen of Typothorax. A right fibula (AMNH FR 19338) from the same locality is also aetosaurian and most likely belongs to the smaller Typothorax specimen. The two humeri and the one complete and two partial ulnae belong to phytosaurs. Finally, a fragment of the proximal end of a humerus is consistent in its morphology to a silesaurid dinosauriform.

6.1. Cervical and anterior dorsal osteoderms

Following Long & Ballew (Reference Long and Ballew1985), aetosaur osteoderms are assigned to different body regions based upon the type of vertebrae they cover (e.g. cervical, dorsal, caudal). However, in many forms, including Desmatosuchus, although the vertebral transition is somewhat clear (Case Reference Case1922; Parker Reference Parker2008a), paramedian osteoderms from the posterior cervical and anterior dorsal regions can be difficult to distinguish without the presence of associated lateral osteoderms (Parker Reference Parker2008a; contra Long & Ballew Reference Long and Ballew1985).

The most prominent osteoderm is a large, recurved lateral of the right (not left as stated by Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002) side (Fig. 12a–b; Gregory Reference Gregory1953, figs 1–3; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 1a–e). Although broken, the spine measures 16.3 cm in length from the base of the lateral flange to the broken spine tip. When complete, it probably had a length of around 20 cm. The length and recurvature of the spine suggest that this osteoderm is from the fourth or fifth cervical series. The spine is dorsoventrally flattened, with a tapering anterior edge and a more rounded posterior surface. The lateral flange is nearly twice as long as the dorsal flange, a character diagnostic of the cervical series in Desmatosuchus (Parker Reference Parker2008a). This osteoderm differs slightly in that the spine is directed more strongly dorsally and is less recurved than in D. spurensis (MNA V9300; UMMP 7476), and is more like those in the holotype of D. smalli (TTU P-9024; Gregory Reference Gregory1953, p. 21; Parker Reference Parker2008a).

A cervical paramedian osteoderm articulates perfectly with this lateral and thus was almost certainly associated in life (Fig. 12b; Gregory Reference Gregory1953, figs 1, 3; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, figs 1a, 2h, k). It is roughly quadrangular in dimension (Table 5). The osteoderm is dorsoventrally thickened (3 cm) with well developed tongue-and-groove (sensu Long & Ballew Reference Long and Ballew1985) articular surfaces, which are typical for desmatosuchines (Heckert & Lucas Reference Heckert and Lucas2000; Parker Reference Parker2007). A prominent dorsal boss rises from the centre of the osteoderm and the ventral surface is smooth and strongly mediolaterally concave. These dimensions and characteristics are consistent with placement as a posterior cervical paramedian based on more complete specimens (UMMP 7476).

A slightly smaller cervical paramedian from the left side (Fig. 12c; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 2e, g, j) has similar dimensions with the right side paramedian just described (Table 5), suggesting that they may correspond in position or that this osteoderm could fall into the position immediately anterior. Alternatively, Gregory (Reference Gregory1953, figs 9–10) figured it articulating with an anterior dorsal osteoderm series from the right side (described below). This is equally plausible given the similarity in the overall morphology of the posterior cervical and anterior dorsal, paramedians. The anterior lamina is broken away, but the rest of the osteoderm is complete. There is a prominent dorsal eminence that is situated centrally on the plate and oriented obliquely from posterolateral to anteromedial. The anterior and lateral portions of the dorsal surface of the osteoderm possess the faint pattern of pits and ridges characteristic of Desmatosuchus (Long & Ballew Reference Long and Ballew1985).

Conjoined paramedian and lateral osteoderms of the right (not left as identified by Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002) side (Fig. 12d–f; Gregory Reference Gregory1953, figs 9–11; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 2c) are from the anterior dorsal series (Gregory Reference Gregory1953), probably from the first or second series. Both osteoderms are incomplete, the lateral flange is missing from the lateral osteoderm and the anterior laminae are broken off from both. The dorsal flange of the lateral osteoderm is 5 cm wide and at least 9.2 cm long. The associated paramedian is similar to the others described previously in possessing a well developed dorsal boss and having a smooth strongly convex ventral surface. The paramedian is strongly sutured to the associated lateral osteoderm (Fig. 12e). The lateral plate is flexed obtusely at approximately 130 degrees from the horizontal (Fig. 12f). The lateral flange is missing, so its dimensions cannot be compared to the dorsal flange. Gregory (Reference Gregory1953) considered the dorsal eminence to be a broken off spine, yet reduced in comparison with spines in the cervical series. However, comparison of this osteoderm with those of other Desmatosuchus specimens (TMM 31173-54, TMM 31213-1, MNA V9300; Fig. 13) demonstrates that not much of the eminence is missing and that, overall, the morphology of the lateral osteoderm is quite consistent between these specimens. Interestingly, the paramedian osteoderm eminence in ANSP 14688 and TMM 31213-1 (both from the Dockum Group of Texas) is much more pronounced than in MNA V9300 (from the Chinle Formation of Arizona); however, an intermediate condition in TMM 31173-54 (also from the Dockum of Texas) suggests that this is simply a variable character and presently not taxonomically relevant.

6.2. More posterior osteoderms

Another lateral osteoderm of the right side in ANSP 14688 has a well-developed, but broken, spine (Fig. 14a–d; Gregory Reference Gregory1953, figs 12–13; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 1f, g). The dorsal flange is broad, but unfortunately the lateral flange is broken away. Heckert & Lucas (Reference Heckert, Lucas, Heckert and Lucas2002) considered the dorsal flange to represent a portion of a conjoined paramedian osteoderm; however, the proposed articulation between the elements is a large crack and only a single element is present. A broad dorsal flange and the development of the dorsal eminence into a well-developed spine suggest a more posterior placement in the dorsal carapace for this specimen (also see Gregory Reference Gregory1953). Parker (Reference Parker2008a) noted that the presence of well-developed spines in the anterior dorsal region of ANSP 14688 would be an autapomorphy for D. haplocerus as the dorsal lateral osteoderms of Desmatosuchus spurensis (e.g. MNA V9300) do not develop a strong spine until about the 10th row. However, a more posterior placement is further supported by articulation of one of these lateral osteoderms with a partial paramedian osteoderm (Fig. 14c–d). This paramedian is wider than long, so it belongs to at least the mid to posterior dorsal region (Parker Reference Parker2008a). Still, this placement is not unambiguous, because the dorsal flange of the lateral plate is thicker and shorter than mid to posterior dorsal laterals of MNA V9300. Indeed, these dimensions, as well as the robustness of the spine base, are more similar to the pelvic or anterior and mid caudal laterals of MNA V9300, and therefore these osteoderms are tentatively assigned to that region. Heckert & Lucas (Reference Heckert, Lucas, Heckert and Lucas2002) considered these to be cervical osteoderms, but the morphology of the articulating paramedian is not consistent with that assignment, as it is too thin and wide. Furthermore, the dorsal flange is too broad for osteoderms of the cervical lateral series.

Figure 14

“Episcoposaurus” haplocerus dorsal and caudal osteoderms (ANSP 14688): (a–b) right anterior caudal lateral osteoderm in dorsal (a) and anterior (b) views; (c–d) same right anterior caudal lateral osteoderm articulated with a paramedian osteoderm in posterior (c) and dorsal (d) views; (e–h) paramedian osteoderms from either the dorsal or caudal regions. Scale bar=5 cm. Abbreviations: al=anterior lamina; de=dorsal eminence; df=dorsal flange; lf=lateral flange.

Spines of the dorsal lateral series in D. smalli (TTU P-9024) are more gracile and elongate, and possess ventrally recurved tips, which differ from the shorter, more robust and straight spines of D. spurensis (UMMP 7476; MNA V9300). Unfortunately, the preserved spines in ANSP 14688 are too fragmentary to determine which morphotype they possess. This is the main reason why Parker (Reference Parker2008a) considered D. haplocerus to be a nomen dubium.

The majority of the paramedian osteoderm fragments in ANSP 14688 are thinner and consistent with the dorsal paramedian region. Several fairly complete and partial dorsal paramedians have been reconstructed from fragments (Fig. 14e–h; Gregory Reference Gregory1953, fig. 14; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 2a–d, i). These osteoderms are wider than long, possess a randomly arranged dorsal surface ornamentation of pits and grooves, a thin anterior lamina and a centered, raised dorsal eminence. The surface ornament is more “honeycombed” and thus more similar to D. smalli (TTU P-9024) than D. spurensis (UMMP 7476) (Parker Reference Parker2008a). In addition, as previously mentioned, there are boxes of osteoderm fragments in ANSP 14688, which with formal preparation may form more elements.

6.3. Other material

The dorsal vertebra (Fig. 15a) mentioned by Gregory (Reference Gregory1953, p. 17) belongs to a phytosaur and should be excluded from the lectotype materials.

Figure 15

Non-osteoderm, postcranial material from ANSP 14688: (a) phytosaur dorsal vertebral centrum in lateral view; (b–c) phytosaur sacral vertebra in ventral (b) and posterior or anterior (c) views; (d) aetosaur mid-caudal vertebral centrum in ventral view; (e) left phytosaur ilium in lateral view; (f–g) proximal portion of a right aetosaur scapula in lateral (f) and medial (g) views. Scale bar=5 cm. Abbreviations: gl=glenoid; sr=sacral rib; t=tuber; trt=triceps tubercle.

An isolated sacral vertebra (Fig. 15b–c; Gregory Reference Gregory1953, figs 6–8; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 3c–d) is 5.8 cm high, 7 cm wide and 6.5 cm long. The neural arches are broken away and the proximal portion of the right sacral rib is fused to the centrum. The ventral surface is convex, not flattened as in aetosaurs. Overall, the morphology compares more favourably with a phytosaur and thus this specimen should be excluded from the lectotype (Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002).

The caudal centrum (Fig. 15d) described by Gregory (Reference Gregory1953, p. 17) is a badly weathered anterior caudal centrum that is 5.6 cm high, 5 cm wide and 5 cm long. The neural arch is broken away, as are the apparently very short transverse processes. The ventral surface is grooved, with the anterior portion separating the ventrally extending articular surfaces for the chevron bones. This morphology is consistent with the anterior caudal vertebrae (∼8th position, Parker Reference Parker2008a) of aetosaurs and, therefore, probably belongs with the specimen.

A nearly complete left ilium is also present in ANSP 14688; however, it was not mentioned by Cope (Reference Cope1892) and a field tag with the specimen states that it was collected from near the same site as the armour, but on a return visit (Gregory Reference Gregory1953). Spamer & Daeschler (Reference Spamer and Daeschler1995, fig. 21c) and Heckert & Lucas (Reference Heckert, Lucas, Heckert and Lucas2002, fig. 3e, f) figured only the anterior portion of the element, but the posterior portion is present as well (see Fig. 15e; Gregory Reference Gregory1953, fig. 15). This specimen differs significantly from the ilia of aetosaurs (e.g. UMMP 7322) in having an extremely short anterior iliac blade and a low iliac blade with a nearly horizontal dorsal margin, belongs to a phytosaur, and should be removed from the lectotype (Gregory Reference Gregory1953, p. 23; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, pp. 200, 203; Parker Reference Parker2008a, p. 34; contra Spamer & Daeschler Reference Spamer and Daeschler1995, fig. 21c; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, p. 194, fig. 3e, f).

The proximal portion of a right (not left as identified by Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002) scapula is present (Fig. 15f–g; Gregory Reference Gregory1953, figs 4, 5; Heckert & Lucas Reference Heckert, Lucas, Heckert and Lucas2002, fig. 3 a, b). It measures 18 cm in length and is anteroposteriorly broad. The acromion ridge is concave, and not strongly convex as in phytosaurs (UCMP 26699; Camp Reference Camp1930; McGregor Reference McGregor1906, fig. 20). The supraglenoid area extends strongly posteriorly, and is kinked medially. There is a well developed triceps tubercle on the posterior surface approximately 8 cm above the glenoid lip (Fig. 15f–g). Overall, the element compares well with that of aetosaurs, including Desmatosuchus (TTU P-9024; Small Reference Small1985, fig. 7). A teardrop-shaped tuber is present on the posterior edge just dorsal to the glenoid fossa, which is consistent with this identification (Fig. 15f; Gregory Reference Gregory1953, fig. 4; Nesbitt Reference Nesbitt2011, character 219).

6.4. Topotypes

Parker (Reference Parker2008a, p. 34) provided a discussion of the status of topotypes assigned to E. haplocerus (also see Wilson Reference Wilson1950). Properly referred material includes two lateral osteoderms from TMM 40041-3 and 39 cervical and anterior dorsal plate fragments (TMM 40041-5), supposedly collected from the type locality by Cope in Reference Cope1892. These specimens share diagnostic characters with E. haplocerus and therefore are referable to that taxon (Parker Reference Parker2008a).

6.5. Summary

Reexamination of the type materials of Episcoposaurus haplocerus reveals that the majority of the material belongs to a single individual, including the osteoderms, caudal centrum and scapula. The ilium and dorsal vertebrae belong to phytosaurs and should be removed from the type materials. McGregor (Reference McGregor1906) states that the type of E. haplocerus was found associated with remains of the phytosaur “Phytosaurus superciliosus” (TMM 18562-1), which consists chiefly of disarticulated skull material. Thus, it is possible that these elements belong with that specimen. The sacral centrum is also more consistent with that of phytosaurs and should also be removed.