Redescription of Calyptosuchus (Stagonolepis) wellesi (Archosauria: Pseudosuchia: Aetosauria) from the Late Triassic of the Southwestern United States with a discussion of genera in vertebrate paleontology

Atlas/axis

There are many axes in the collection from the Placerias Quarry. Case (1922) describes the ventral surface of the axis in D. spurensis as flat, and most of the specimens in the collection possess flat ventral surfaces. However, UCMP 139803 (from CF1) has a distinct ventral keel (Fig. 5A) and therefore mostly likely is referable to C. wellesi which has keeled cervical vertebrae (e.g., UCMP 27225; Murry & Long, 1989) rather than D. spurensis which has cervicals with a smooth ventral surface (e.g., UMMP 7476; MNA V9300). The upper portion of the neural arch, including the zygapophyses, is broken (Figs. 5B–5D). The atlantal neural arches are also broken. The centrum of the axis has distinct concave sides that are overhung by a thickened ridge, which bears the diapophyses (Fig. 5D). The parapophyses are situated anteroventrolaterally on the centrum and are connected ventrally by a thickened crescentic ridge that forms the anterior portion of the atlas intercentrum (Fig. 5A). The suture between the atlas intercentrum and the axis centrum is visible in ventral view.

The parapophyseal facets are round and directed ventrolaterally and slightly posterior. The odontoid process is attached (Figs. 5A, 5B, 5D and 5E); its sutures with the centrum are still visible so the fusion is not complete. The dorsal surface of the odontoid process forms a slightly concave trough that opens posteriorly into the neural canal (Fig. 5E). The canal is large, about one-half the diameter of the posterior articular face of the centrum. In posterior view, the articular face of the centrum has a flat (horizontal) dorsal margin. The face is concave with well-developed rims. The length of the atlas/axis including the odontoid process is 48.7 mm. The axis centrum has a width of 30.6 mm and a height of 25.4 mm.

Postaxial cervical vertebrae

Numerous vertebrae were recovered in grid square CF1, where the atlas/axis (UCMP 139803) was recovered, including several cervical vertebrae. These centra possess cervical keels and therefore cannot be referred to Desmatosuchus (Murry & Long, 1989; Long & Murry, 1995) and are assigned to Calyptosuchus. The presence of ventral keels on the cervical centra of Calyptosuchus is verified by specimen UCMP 27225. Long & Murry (1995: fig. 74) figured what presumably they thought to represent a cervical series of Calyptosuchus, but unfortunately did not provide explicit specimen numbers to identify the specimen further and it could not be located for the current study.

The cervical vertebrae of Calyptosuchus are platycoelous, the anterior face being anteriorly concave and the posterior face nearly flat. Both faces are oval and taller than wide. On the anterior cervicals (e.g., UCMP 139793, 139794) the small, subrounded parapophysis is situated at the base of the centrum (Figs. 5F–5I). On more posterior centra (e.g., UCMP 139813) the parapophysis is located closer to the top of the centrum, below the neurocentral suture (Figs. 5J–5M). Anterior cervicals are also anteroposteriorly shorter than the posterior cervicals (Figs. 5H and 5K). The ventral keel is well-developed and in some specimens (e.g., UCMP 78714) the keel is expanded posteriorly into a small tab (Figs. 5N and 5O). UCMP 78714 also preserves a portion of the neural arch. Although crushed and distorted it shows that the zygapophyses were elongate (Fig. 5P). Prezygadiapophyseal and postzygadiapophyseal laminae (sensu Wilson, 1999) are present.

Trunk vertebrae

The trunk vertebrae of Calyptosuchus are more difficult to identify than the cervical vertebrae from the mixed collection of material from the Placerias Quarry; however, there are vertebrae with more elongate neural spines that also lack typical accessory articulations (hyposphenes–hypantra) on the neural arch. This readily distinguishes them from the trunk vertebrae of D. spurensis which possess much shorter (dorsoventrally) neural spines and exhibit hyposphenes and hypantra (Parker, 2008b; Stefanic, 2017). The trunk centra of Calyptosuchus lack the lateral fossae present in A. scagliai (Desojo & Ezcurra, 2011). There are also posterior trunk vertebrae preserved in the holotype (UMMP 13950; Case, 1932).

UCMP 139694 is most likely the 10th presacral (first trunk) vertebra as it is transitional in the position of the parapophysis between the cervical and trunk series (Figs. 6A and 6B). The parapophysis is situated on the anterodorsal surface of the centrum and it is confluent with the transverse process and connected by a well-developed anterior centrodiapophyseal lamina (acdl; sensu Wilson, 1999). In D. spurensis this specific placement of the parapopohysis occurs in the 10th presacral position and in the following vertebra (11th presacral) the parapophysis moves onto the transverse process (Case, 1922; Parker, 2008b). The neural arch of UCMP 139694 also bears a posterior centrodiapophyseal lamina (pcdl) but it is not as well-developed as the anterior centrodiapophyseal lamina (acdl). The joining of these two laminae forms a ventrolaterally opened shallow triangular fossa situated ventral to the transverse process. A postzygadiapophyseal lamina (podl; Wilson, 1999) is present as a well-developed thin ridge of bone connecting the transverse process and the postzygapophysis. The centrum is spool-shaped, platycoelous, ventrally smooth, and measures 37.9 mm in length (Fig. 6B). The centrum also has a height of 31.8 mm and a width of 31.6 mm.

UCMP 139796 from CF1 (Figs. 6C–6H) has the typical platycoelous, spool-shape found in aetosaurs and represents a mid-trunk vertebra. The centrum measures 43.4 mm in length, with a height of 35.4 mm and a width of 32.4 mm; thus the lengths of the centra increase along the trunk portion of the vertebral column similar to D. spurensis (Parker, 2008b). The articular faces of the centrum are nearly flat, but the anterior face is still slightly concave, with expanded rims (Figs. 6C and 6D). The neural arch is taller than the centrum articular faces and the oval neural canal is large (19.4 mm high) (Fig. 6E). In right lateral view the transverse process is mostly broken away (Fig. 6D), but a thick strut originates on the posterolateral corner of the neural arch and terminates on the ventral surface of what is left of the transverse process. This strut represents the posterior centrodiapophyseal lamina (pcdl). A postzygadiapophyseal lamina (podl) forms a shelf from the posterior edge of the transverse process to the right postzygapophysis. A shallow postzygapophyseal centrodiapophyseal fossa (sensu Wilson et al., 2011) opens posterolaterally, formed by the junction of these two laminae (Fig. 6D). Although the posterior portion of the neural arch is broken it is clear that there is no deep hyposphene between the postzygapophyses as in Desmatosuchus (MNA V9300). The postzygapophyses are not steeply inclined, instead projecting at about 30° above horizontal. The postzygapophyses project well posterior to the posterior face of the centrum (Fig. 6F). Anteriorly on the neural arch there is a deep round fossa between the prezygapophyses and the neural spine, the spinoprezygapophyseal fossa (sprf, Wilson et al., 2011; Fig. 6D). The neural spine is not anteroposteriorly elongate measuring only about 27 mm at the base and the spinal laminae are present but weakly developed.

Another trunk vertebra from CF1 (UCMP 139702) preserves a few more details. In front of the anterior fossa (sprf) described for UCMP 139796, the prezygapophyses meet to form a broad shelf or ventral bar (Fig. 6G) as in S. robertsoni (Walker, 1961: fig. 7j). There is no hypantrum. The right transverse process is nearly complete. It is broad, about 26.7 mm in width, compared to the centrum, which has a width of 25.7 mm. The upper surface of the transverse process is flat and the ventral surface thickened with the strut described for UCMP 139796, which continues onto the base of the neural arch. The parapophysis is positioned 29.3 mm laterally from the origin of the transverse process. The distal end of the transverse process, the diapophysis, is not preserved but even incomplete, the process has a length of 44.4 mm. The zygapophyses are inclined at close to 45° to the horizontal. The centrum length is 39.9 mm long and 28.3 mm high.

A third trunk vertebra from CF1 (UCMP 139795) preserves the postzygapophyseal region extremely well. As with the other trunk vertebrae there are no accessory processes (hyposphene). Instead at the base of the medial union of the postzygapophyses there is a small posteriorly pointed projection that would rest on top of the ventral bar formed by the joined prezygapophyses of the subsequent vertebra. This pointed projection also occurs in S. deltatylus (PEFO 34045). The ventral bar and posterior projection in the trunk vertebrae is also shared with some phytosaurians (e.g., Smilosuchus, TMM 43685-206).

Two other well-preserved trunk vertebrae (Figs. 7A–7D) referable to C. wellesi are from UMMP 7470, which includes a partial sacrum and the two trunk vertebrae, as well as two paramedian osteoderms. The best preserved vertebra is a nearly complete anterior mid-trunk vertebra (Case, 1932: figs. 2–4). The centrum is laterally compressed and ventrally concave because of the flaring articular rims. It has a length of 48.7 mm, and width of 42.3 mm, and a height of 42.8 mm. The neural arch and spine are tall, twice the height of the centrum at 78.8 mm, with 55.2 mm for the neural spine height. The neural spine is mediolaterally thin, expanded anteroposteriorly (34.2 mm long) and terminates with a pronounced lateral expansion (spine table). The postzygapophyses extend posteriorly beyond the posterior articular face of the centrum and are oriented at 45° above horizontal. The prezygapophyses form a flat plate almost indistinguishable from the transverse processes (Figs. 7A and 7C). The transverse processes are broad with a flat dorsal surface, and nearly twice the width of the centrum (82.3 mm). The processes are of the typical aetosaurian arrangement with both rib articulations situated on the transverse processes (Figs. 7A and 7C). Transverse processes and postzygapophyses are connected by a thin sharp postzygapophyseal lamina (podl), which forms the deep spinopostzygapophyseal fossa (spof) just anterior to the postzygapophyses (Figs. 7B and 7D).

Long & Murry (1995: fig. 75a) considered the transverse processes of the dorsal series extremely elongate throughout the entire column. However, they figured posterior trunk vertebrae of UMMP 13950 as an example, which have the ribs fused to the transverse processes, giving the appearance of greatly elongate processes (as noted by Case, 1932). This fusion of transverse process and rib is also found in S. deltatylus (PEFO 34045) as well as D. spurensis (MNA V9300; Parker, 2008b). However, the processes in C. wellesi differ from those two taxa in that they are flat dorsoventrally and anteroposteriorly broad (Case, 1932: pl. 4, fig. 1). The centra of the posterior most trunk vertebrae are anteroposteriorly short in comparison with those of the mid-trunk vertebrae, with large flaring articular rims.

Sacral vertebrae

The best preserved sacral vertebrae are in the holotype (UMMP 13950) as well as in the partial pelvis (UMMP 7470) and were well-described and figured by Case (1922, 1929, 1932). There are two vertebrae in the series, which differ from those of desmatosuchine aetosaurs in that they are not fused to each other (Parker, 2008b; Griffin et al., 2017) although Case (1932) noted that the zygapophyses between the two sacral vertebrae were reduced in size. The articular faces of the centra are round. The neural arches are robust and bear the heavy, expanded sacral ribs, and the neural spines are also robust and taller than the centra. The neural spines possess expanded apices or “spine tables.”

An isolated specimen (UCMP 139785) from grid block C78W in the Placerias Quarry is most likely referable to C. wellesi as it does not show fusion to the other sacral as do others in the collection (e.g., UCMP 139787). The vertebra is very massive with the proximal portions of the sacral ribs firmly sutured to the neural arch (Figs. 8A–8D). The upper surface of the ribs is swept posteriorly (Fig. 8B). The centrum faces are roughly “heart-shaped” and the ventral surface lacks a keel (Figs. 8C and 8D). The neural spine is broken off, but was obviously robust (thick and elongate) as in UMMP 7470. There is a distinct spinoprezygapophyseal fossa (Fig. 7A) under the prezygapophyses.

Caudal vertebrae

The Placerias Quarry collection contains dozens of aetosaur caudal centra with broken neural arches; however, at this time it is not possible to assign these elements to particular taxa. However, the first seventeen vertebrae of the caudal series of C. wellesi are well-preserved in articulation in the holotype (UCMP 13950) and were described by Case (1932). The most notable feature of the caudal series of UCMP 13950 is the height of the neural spines, which is greater than the height of the centrum. This differs from aetosaurs such as D. spurensis (MNA V9300) and Paratypothorax (PEFO 3004) where the height of the neural spine is equal to or less than the height of the centrum. It is similar to the condition in A. scagliai (PVL 2073) and S. robertsoni (Walker, 1961: fig. 10).

Long & Murry (1995: 83) state that the ventral grooves of the caudal centra in C. wellesi are narrower than those of D. spurensis and “bear faint, longitudinal ridges.” However, they provide no basis for their taxonomic referrals nor any specimen numbers, so this claim cannot be verified. The caudal ribs or transverse processes of paratypothoracins originate close to the base of the centrum (e.g., PEFO 3004). No centra with low caudal ribs are currently known from the Placerias Quarry, and thus all of the preserved centra presumably belong to C. wellesi or D. spurensis although they cannot be distinguished between those taxa.

Scapulocoracoid

No bones of the pectoral girdle are preserved in the holotype of C. wellesi (UMMP 13950). Long & Murry (1995) assign several scapulocoracoids (UCMP 78698, UCMP 32196, UCMP 27976) from the Placerias Quarry to C. wellesi; however these elements were recovered from areas CD and CE which provided many osteoderms of D. spurensis and none referable to C. wellesi (Fig. 2). Furthermore, coracoids assigned to C. wellesi (UCMP 32196, UCMP 27976; Long & Murry, 1995) are from C8 and C75W, and also from areas that provided predominantly material of Desmatosuchus (Fig. 2). Thus, none of the Placerias Quarry material can be unambiguously assigned to C. wellesi. Differences between the coracoids of D. smalli (TTU P-9023) and S. robertsoni (Walker, 1961) pertain to the development of the subglenoid buttress. Unfortunately this area is not preserved in any of the Placerias Quarry specimens.

Forelimb

As with the shoulder girdle, no forelimb elements are present in the holotype of C. wellesi (UCMP 13950). Moreover, Long & Murry (1995) did not assign any forelimb material to C. wellesi. The UCMP Placerias Quarry collection contains numerous aetosaur humeri, but none can be clearly referred to C. wellesi.

Pelvic girdle

Several pelvic girdles have been referred to C. wellesi including the holotype (UCMP 13950; Fig. 9), a specimen from the Dockum Group of Texas (UMMP 7470), and elements from the Placerias Quarry (Case, 1929, 1932; Long & Murry, 1995). The Placerias Quarry elements include a left ilium (UCMP 32422) and a corresponding left ischium (UCMP 32148), both from grid CF1 (Figs. 10A and 10B), and figured by Long & Murry (1995: figs. 79–80). The collection from CF1 also contains a crushed, but complete right ilium (UCMP 25941) and a right ischium (UCMP 32153) (Fig. 10C). These elements match the two figured by Long & Murry (1995) perfectly and all four elements probably belong to the same individual (Long & Murry, 1995). The difference in color between these elements in Fig. 10 is a photographic lighting artifact. Grid CF1 contains a fair amount of material referable to Calyptosuchus, mainly cervical vertebrae, including some paramedian osteoderms, so referral of these pelvic elements to C. wellesi is supported.

The problem with assigning isolated ilia from the quarry to specific taxa is that the morphology of the ilium of Desmatosuchus is poorly understood. The holotype of D. spurensis (UMMP 7476) preserves only a fragmentary left ilium that is missing almost the entire posterior portion of the iliac blade. A referred specimen of D. spurensis (MNA V9300) as well a specimen of D. smalli (TTU P-9172) preserve nearly complete sacra; however, the anatomy of the ilia is difficult to interpret on these specimens because they are highly distorted, in part because of the complete fusion of the sacral ribs to the ilia (see Parker, 2008b). Long & Murry (1995: figs. 91–92) assigned an isolated right ilium from Crosby County Texas (UMMP 7322) to D. spurensis. This specimen possesses an acute angle between the anterior portion of the iliac blade and the anterior edge of the iliac body as well as a triangular (in lateral view) posterior iliac blade. The holotype ilium (UMMP 7476) as preserved is consistent with this although much of the anterior portion of the iliac blade is damaged. If UMMP 7322 is indeed referable to D. spurensis UCMP 32422 differs from it mainly in that the posterior iliac blade is squared off and not pointed as in UMMP 7322. This is the character Long & Murry (1995) used to assign ilia to C. wellesi and this referral is followed here.

Ilium

The ilia in C. wellesi have ventrally directed acetabula; however, to make the following description easier to follow the element is described as if it is oriented vertically, thus the iliac blade is dorsal and the acetabulum ventral and lateral. The preacetabular process of the iliac blade in UCMP 25941 is short and does not extend far anterior of the pubic peduncle (Figs. 10A and 10B). It is 50 mm long, mediolaterally thick and triangular in lateral view with a ventrally curved tip. The postacetabular portion of the iliac blade extends well beyond the posterior edge of the pubic peduncle and is thickened very close to its proximal end. The entire iliac blade is 180 mm long, and 52 mm high above the acetabulum. The dorsal surface is highly rugose, marked with scars for the attachment of the M. iliotibialis 1–3 (Schachner, Manning & Dodson, 2011). The acetabular area is roughly diamond-shaped in lateral view and delineated dorsally by a well-developed supraacetabular rim (Fig. 10A). The main iliac body is slightly concave dorsal to the acetabulum, lacking the deep recess found between the supraacetabular rim and the posterior portion of the iliac blade in S. deltatylus.

The pubic and iliac peduncles are thickened anteriorly and posteriorly respectively, and both are comma-shaped in ventral views. The two peduncles meet at a ventrally directed point ventral to the iliac portion of the acetabulum. Medially, there are scars for the two sacral ribs, which cover not only the iliac neck but also a large portion of the ilium ventral to the iliac blade and medial to the acetabulum (Fig. 10B). This is a result of the ventrally directed acetabula as in A. scagliai (PVL 2073) and Typothorax coccinarum (PEFO 33967). The iliac blade thins dorsally from the sacral rib scars. Overall the ilium of C. wellesi is very similar to that of A. scagliai (PVL 2073) and Ebrachosaurus singularis (Kuhn, 1936). It differs from N. engaeus (PVL 3525) in having a much more robust anterior process of the iliac blade. It differs significantly from the ilium of T. coccinarum (UCMP 122683) which has a taller, but anteroposteriorly shorter iliac blade, as well as a more gracile, and “hooked” anterior process which does not extend anteriorly beyond the pubic peduncle (Long & Murry, 1995: figs. 106–107). The right ilium is well-preserved in the referred specimen UMMP 7470 (Case, 1922: fig. 28b). It is nearly identical to UCMP 25941 with the thickened, short, recurved anterior iliac blade. Both ilia are present in the holotype (UMMP 13950) but both are incomplete, crushed, and presently badly broken (Fig. 9; Case, 1932, pl. II). Note that the photo of the pelvic girdle and vertebral column in Plate II in Case (1932) is reversed.

Ischium

The left ischium (UCMP 32148) associated with the UCMP ilium described above is nearly complete (Fig. 10A). It is anteroposteriorly short, not much longer than tall, with a length of 110 mm and a height of 97 mm. This differs from the ischia of A. scagliai (PVL 2073), S. robertsoni (Walker, 1961), and Aetosaurus ferratus (Schoch, 2007), where the posterior process is more elongate. The pubic peduncle is comma-shaped in dorsal view and contacts the corresponding peduncle of the ilium. The oval acetabular surface is deeply concave and bordered posteriorly and ventrally by a strongly raised, curved rim. The main body of the ischium is essentially a thickened “rod” that curves posteriorly and dorsally. A mediolaterally thin flange of bone extends ventrally for the entire length of the “rod” (Fig. 10A). The ventral margin is straight. The lateral surface of the thin flange is rugose presumably for attachment of the third head of the M. puboischiofemoralis externus (Schachner, Manning & Dodson, 2011). Medially there is an elongate suture for the opposing ischium. The anterior margin bears a distinct notch. This notch is also present on the right ischium of UMMP 7470. The posterior process of UMMP 7470 is more elongate than that of UCMP 32148, but still not as elongate as in Walker’s (1961) reconstruction of S. robertsoni. The ischia are also present in UMMP 13950 but are poorly preserved (Fig. 9). Case (1932: pl. III) restores the ischium as dorsoventrally deep and anteroposteriorly short, consistent with UCMP 32148.

Pubis

The best preserved pubis from the Placerias Quarry material is a left element (UCMP 32150) from grid CF2 (Figs. 10D–10G). It shares the same preservation, color, and size with the ilium and ischium described above, but does not quite articulate and thus may not belong to the same individual. The pubic rod is slender and its distal end is broken away (Figs. 10D and 10E). The concave acetabular surface is reduced compared to the area on the ischium and there is a groove just ventral to this surface. The articular surface for the ilium is comma-shaped in dorsal view (Fig. 10F). The obturator flange is broken away (Fig. 10G) so the number of openings in this element cannot be determined. Walker (1961) restored the pubis of S. robertsoni with pubic foraminae and a pubis of S. deltatylus (PEFO 31217) also has two openings. Only a single foramina is present in the pubis of D. spurensis (MNA V9300) and the number of foraminae is unknown in A. ferratus (Schoch, 2007).

The proximal portion of the right pubis is present in UMMP 7470 (Case, 1922: fig. 28b). The posterior margin as preserved shows the anterior border of an obturator foramen but the element is not complete enough to determine if there was a second opening. The proximal head of UMMP 7470 bears a deep lateral groove that originates at the acetabular rim and extends parallel to the anterior margin of the pubis. The distal end of the element is broken away so that the extent of the groove cannot be determined. This groove is only weakly developed in UCMP 32150, which is also missing its distal end. UMMP 13950 preserves the distal end of the pubis, which expands into the broad pubic “apron” typical for suchians (Case, 1932). Case (1932: pl. III) reconstructs the girdle as dorsoventrally shallow with the distal margin of the pubis at the same horizontal level as the ventral margin of the ischium. This differs greatly from the condition in D. spurensis (MNA V9300) where the pubis extends well below the level of the ischium, but is similar to the short pubes of T. coccinarum (Long & Murry, 1995).

The distal end of the pubic rod extends slightly beyond the ventral margin of the pubic apron, as is typical for aetosaurs. This end is slightly swollen as in S. robertsoni (Walker, 1961), but does not form the distinct knobby pubic boot found in D. spurensis (MNA V9300).

Femur

The best preserved femur that can be referred to C. wellesi is UCMP 25918, which is a left side element from CF1 (Figs. 11A–11D; Long & Murry, 1995: figs. 81, 83). It is of similar preservation and the right size to match the pelvic elements described above so it is very possible that all of these elements belong to a single individual. Long & Murry (1995) describe it as “more gracile” than femora from the quarry that they assign to D. spurensis. Overall it is less sigmoidal than the femur of phytosaurs, as is characteristic of aetosaurs (Figs. 11A–11C). It has a total length of 329 mm. The proximal head is badly eroded (Figs. 11A and 11B). The fourth trochanter is a pronounced crescent-shaped ridge located about 120 mm ventral to the proximal end (Fig. 11A). The distal femoral condyles are well-preserved (Fig. 11D). The medial condyle has a posteromedial corner with an angle of 90° and a rounded anteromedial corner. The lateral condyle is larger than the medial and anterolaterally bears a distinct crista tibiofibularis. The angle between the crista tibiofibularis and the lateral condyle is obtuse. The posterolateral corner of the lateral condyle is rounded and expanded posteriorly.

Tibia

UCMP 25887 from C64M occurs within a cluster of osteoderms of C. wellesi, but material referable to D. spurensis occurs in that grid as well. Nonetheless, this left tibia is much more gracile than others found in the quarry (e.g., UCMP 25877), which probably belong to Desmatosuchus (Fig. 12; Long & Murry, 1995). UCMP 25887 (Figs. 13A–13D) has a length of 186 mm, shorter than the femur as is typical for aetosaurs. The proximal head is oval in proximal view with a width of 73 mm, a length of 52 mm and is divided into two distinct sections by a nearly central ridge. The medial surface has slightly more area than the lateral surface and it is concave, whereas the lateral surface is convex. A cnemial crest is absent (Nesbitt, 2011), and there is a distinct “lip” posteriorly on the lateral portion of the head. The posterior portion of the distal end possesses a dorsoventrally oriented groove (Nesbitt, 2011: char. 337–1) for articulation with the astragalus. There is some damage to the medial condyle of the distal end in UCMP 25887. Overall there are few noticeable differences in the distal ends of UCMP 25887 and UCMP 25877 other than size. However, the proximal end in UCMP 25877 is much more expanded medially and has a distinct dorsal notch on the dorsolateral surface. There are two other gracile tibiae in the Placerias Quarry collection; UCMP 25896 (Figs. 13E–13G) is a left tibia from grid CH1, and UCMP 25894 is a left tibia from grid CH2 that was figured by Long & Murry (1995: fig. 84).

Fibula

UCMP 25802 from grid C67M is gracile compared to other fibulae in the Placerias Quarry collection and, as preserved, matches much of the material of C. wellesi. Long & Murry (1995) also assigned this element to C. wellesi. The specimen represents the proximal end of a left fibula. The iliofibularis trochanter is broken off. There is a small tubercle on the medial side of the shaft. Long & Murry (1995: 84) state that “the diagonal ridge, so prominently exhibited along the medial fibular shaft of Desmatosuchus [spurensis], may not have been present in [Calyptosuchus] wellesi.” However, UCMP 25802 is not complete enough to evaluate this claim.

Astragalus

There are many astragali in the Placerias Quarry collection, but none fits the gracile tibiae in the collection that probably represent C. wellesi. Long & Murry (1995) figured and assigned a right astragalus from grid CF2 to C. wellesi (UCMP 34485); however, this specimen is currently on loan to another researcher and I was unable to examine it. Nonetheless, Long & Murry (1995) stated that they were unable to differentiate between the astragali of Desmatosuchus and Calyptosuchus and thus it is unclear how this assignment was originally made. Thus, neither the type nor referred specimens of C. wellesi preserve the astragalus.

Calcaneum

As with the astragali there are lots of aetosaur calcanea in the collections as well, but as the calcaneum of Desmatosuchus is unknown, they cannot be differentiated. Long & Murry (1995: fig. 82) figured a left calcaneum (UCMP 34481) from CG1 as pertaining to C. wellesi. It is not clear what characters they used to make this assignment. UCMP 34481 is very similar to the calcaneum of A. scagliai (PVL 2073) with a dorsoventrally flattened, mediolaterally expanded posterior tuber, and a deep concavity on the ventral surface of the anterior portion of the tuber. This deep concavity is sharply rimmed and also prominent in T. coccinarum (AMNH FR 2713).

Osteoderms

The holotype of C. wellesi (UMMP 13950) preserves an articulated set of osteoderms starting with the posterior dorsal trunk series and extending back through much of the tail (Fig. 14). These include trunk, lateral, and appendicular osteoderms and, importantly, they are associated with a vertebral column to aid with placement of specific rows. A significant landmark is the neural spine pushed up through the dorsal carapace, which is that of the first caudal vertebra (Case, 1929). Accordingly I have placed it between the first and second caudal paramedians where it pushed the first paramedian anteriorly and displaced the second paramedian posteriorly (Figs. 14 and 15). UMMP 13950 was thoroughly described by Case (1932) and is not in need of a full redescription.

Referred specimens from the St. Johns, Arizona area (Blue Hills, Placerias Quarry) provide more details regarding the mid-dorsal region as well as the ventral trunk osteoderms. Cervical osteoderms are currently unknown for C. wellesi. The cervical lateral plates assigned by Long & Ballew (1985) to C. wellesi that were reportedly characteristic of the genus (Long & Murry, 1995) actually belong to a paratypothoracin aetosaur, most likely Tecovasuchus (Parker, 2005; Heckert et al., 2007).

Paramedian osteoderms

Lateral osteoderms

The lateral osteoderms from the ninth dorsal trunk row (of 16 total) through the 16th caudal rows (of approximately 40 according to Schoch (2007) for A. ferratus) are present and well-preserved in the holotype (UMMP 13950). Thus, the positions of isolated lateral osteoderms with matching anatomy can be placed with confidence. Aetosaurian lateral osteoderms are roughly square to rectangular with a pronounced dorsal eminence or boss (Heckert & Lucas, 2000). Typically the osteoderms are flexed to some degree, divided into two “flanges” (dorsal and lateral or ventral) by the eminence (Long & Ballew, 1985; Parker, 2007). Importantly, all of the lateral osteoderms in UMMP 13950 have more rectangular dorsal flanges, however, lateral osteoderms with strongly triangular dorsal flanges are present in the referred material of C. wellesi. These osteoderms must be from positions anterior to the ninth dorsal row. All of the lateral osteoderms have prominent anterior bars, pyramidal dorsal eminences, and a surface ornamentation of grooves and ridges radiating from the eminence.

The anteriormost lateral osteoderms of the trunk series are well-represented in specimen UCMP 27225, a partial skeleton represented by osteoderms and vertebrae and collected by Charles Camp near St. Johns in 1926. They are quadrilateral in dorsal view with distinct dorsal and lateral flanges separately by an elongate keeled dorsal eminence with a pyramidal terminal end that projects just slightly beyond the posterior osteoderm margin (Figs. 18A–18D). The dorsal flange is distinctly triangular in dorsal view and is reduced in size compared to the lateral flange. The lateral flange appears to increase in width in more posteriorly situated osteoderms. The medial edge of the dorsal flange is strongly sigmoidal and the anterior bar is indented where the anterolateral projection of the adjacent paramedian osteoderm overlies it.

In the next positions, but still anterior to the ninth dorsal trunk row, the dorsal flanges retain their sigmoidal lateral edge, but become more quadrilateral in dorsal view (Figs. 18E and 18F). The lateral flanges are very wide and rectangular. They are still significantly larger than the dorsal flange. The next form of lateral osteoderm occur in the 9th–12th dorsal trunk positions based on comparison with the holotype (UMMP 13950) and are best represented in the Placerias Quarry material by left and right osteoderms (UCMP 136744; Figs. 18G–18J).

The dorsal eminence is larger and very hook-like. The dorsal flange is quadrilateral in dorsal view and maintains the strongly sigmoidal medial margin. The lateral flanges are still much wider than the dorsal flanges but are no longer rectangular. Instead they are strongly quadrilateral with a distinct mediolateral slant so that the anterior margin is much wider than the posterior margin. This forms a distinct anterolateral “wing” that characterizes the osteoderms from this portion of the carapace. In posterior view the angle between the flanges is approaching 90°, much more flexed than the preceding lateral osteoderms.

The sacral and anteriormost caudal lateral osteoderms are represented by a right (UCMP 78751) and two left (UCMP 136744, MNA V3744) osteoderms (Figs. 18K–18N). These osteoderms are reduced in overall width, the lateral flange remains larger than the dorsal flange, but only slightly and anterolateral “wing” is no longer prominent. The dorsal eminence is still strong, but not as hook-like as the previous osteoderms.

At about the third caudal row the dorsal eminence of the lateral osteoderms becomes very rectangular, and the dorsal and lateral flanges are more equal in size. Overall the osteoderms are lengthening anteroposteriorly, corresponding with the increasing length of the caudal vertebrae. These positions are represented by two right osteoderms, UCMP 27048 from the Blue Hills area of St. Johns, and UCMP 136744 from the Placerias Quarry (Figs. 18O–18Q). The dorsal eminence is taller but blunter, not hook-like. The angle of flexion between the dorsal and lateral flanges is a strong 90° in these osteoderms.

Ventral osteoderms

Ventral trunk osteoderms are best represented in UCMP 27225 (Fig. 19). They are square to broadly rectangular with a strong, but narrow anterior bar. The external surface ornamentation consists of a fine pattern of grooves and ridges radiating from a central, unraised area on the osteoderm.

Appendicular osteoderms

Numerous appendicular osteoderms are preserved close to life position in the holotype (UMMP 13950: Fig. 13). They consist of small rounded to oval osteoderms with faint surface pitting. They would have been situated manly along the upper portion of the individual limbs.