The skull, including mandible, is essentially complete. The relationships among the elements are little disturbed, and the topographical features (sulci, ridges, and eminences) as well as most of the sutures are clear. The skull table (Fig. 1), left circumorbital region, and left side of the snout (Fig. 2) exhibit minor distortion. The left postorbital region has been folded toward the midline and is slightly disarticulated, but can be interpreted with the help of the less distorted right cheek (Fig. 3). A preservation break has allowed the right cheek and right lateral portion of the snout to be removed, permitting preparation of the ventral aspect of the braincase, palate, and medial surface of the cheek (Fig. 4). Although the left lower jaw is preserved in place (Fig. 2), most of the posterior portion of the surangular is missing and the angular and articular are slightly displaced. The lateral and ventral aspects of the ramus are well exposed, but most of the medial aspect is obscured. The right ramus has been displaced to the left and rotated 180 degrees about its long axis, and rests upside down against the medial surface of the left ramus, exposing much of its medial surface (Fig. 3).

As most of the skull has sustained only minor deformation, the shape and contours of the skull can be reconstructed with confidence (Fig. 5 and Fig. 6). Measured along the midline, the preorbital region is slightly longer than the postorbital region (Fig. 5), giving the skull distinctly different proportions compared to those of B. texensis (Williston, 1914, fig. 1; Table 1), but is more closely comparable to those of B. brevis (Carroll, 1964), B. olsoni (DeMar, 1967 and Schoch, 2012), and ‘B.’ novomexicanus (Langston, 1953 and Schoch, 2012). The snout is narrow anteriorly, conferring a sub-triangular outline to the skull in dorsal view. This is distinct from the parabolic outline typical of B. texensis and B. brevis, but resembles more closely that of B. olsoni (Schoch, 2012). Even allowing for diagenetic distortion, the skull is much taller than that of B. texensis (Fig. 6; Table 1) or Dissorophus (Schoch, 2012, fig. 4C), with nearly vertical cheeks and snout sides. In lateral profile, the snout rises rapidly over the external nares, and at a point midway between the naris and the anterior margin of the orbit, the skull is nearly as tall as it is over the occiput. Therefore, with the exception of a small central bowl-shaped depression between the external nares, the snout is strongly arched dorsally. This is distinctly unlike the gradually sloping, wedge-shaped profile exhibited by most dissorophoids (Schoch, 2012, fig. 4) or the dorsally concave lateral profile of the snout seen in Cacops (Fröbisch and Reisz, 2012 and Reisz et al., 2009), but is much more comparable to that of B. olsoni (Schoch, 2012, fig. 4D). The orbits and external nares are relatively larger than those described for Dissorophus or B. texensis, but more comparable in size to those in B. olsoni, ‘B.’ novomexicanus, and B. brevis (Schoch, 2012, fig. 1). The jaw articulation is slightly posterior to the occipital plate.

As in Broiliellus and Cacops, the skull roof bears a bilaterally symmetrical series of low coarsely ornamented ridge-like swellings. The most conspicuous of these are borne by the prefrontal, lateral edge of the frontal, postorbital, lateral edges of the supratemporal and tabular, the central portion of the parietal and along the posterior edge of the postparietal. The frontals and nasals may also have borne similar ridges, but this portion of the skull has been too disturbed to confirm this. The ventral edge of the quadratojugal bears a rugose, laterally projecting shelf. Immediately above the shelf, the external surface of the quadratojugal is devoid of sculpturing. Low, rounded swellings also project from the anteroventral portion of the jugal and dorsal lamina of the maxilla. As in Broiliellus, Cacops, and Dissorophus, these ridges are accentuated by adjacent shallow depressions on the bone surface. Median dorsal depressions are located between the external nares, between the orbits near their anterior margins, and on the skull table between the parietal foramen and occipital plate. A bilateral series comprises, on each side, a large depression on the dorsolateral surface of the snout that occupies most of the surface between the orbit and naris, a smaller, horizontally elongate depression immediately below the previous, a circumorbital series including a deeply dished depression on the lateral exposure of the palatine, on the prefrontal anteromedial to the prefrontal depression, on the frontal and parietal medial to the frontal depression, on the postfrontal/postorbital medial to the postorbital protuberance and the postorbital/jugal ventral to the postorbital depression.

The premaxilla forms the anterior margin of the large external naris. The posterodorsally directed alary process is modestly developed. The dorsal surface of the snout between the nares is slightly compressed toward the midline and distorted, but there is no trace of an interpremaxillary fontanelle. Although considerable matrix was removed from the external naris, a septomaxilla could not be found.

The dorsal surface of the snout is disrupted, making it impossible to trace some of the sutures. The nasal in CM 41705 is unlike that in most dissorophoids in not being rectangular (Bourget and Anderson, 2011), or widest at their mid-length (Fröbisch and Reisz, 2008 and Schoch and Rubidge, 2005), but rather is much narrower anteriorly, at most half as wide at its suture with the premaxilla than posteriorly at its suture with the frontal. This is correlated with the distinctly narrow anterior portion of the snout.

The prefrontal forms the anterodorsal margin of the orbit. It does not contact the postfrontal, permitting the frontal to contribute to the dorsal orbital rim. A slender, ventral process of the prefrontal extends ventrally along the anterior orbital wall medial to the lacrimal to contact the dorsal surface of the palatine.

The parietals are narrowest anteriorly and widest at mid-length, where each wedges between the postfrontal and supratemporal. The maximum length of each parietal does not exceed twice its maximum width. Their common median suture is interrupted by a large, circular parietal foramen. The supratemporals are large and pentagonal in outline. Laterally, each forms the anterior part of the supratympanic shelf and dorsal rim of the otic embayment (Fig. 6). The dorsoventrally thickened tabular forms the posterior portion of the supratympanic shelf and posterodorsal part of the otic embayment, whereas ventrally it forms a small portion of the supratympanic flange. Posteriorly, the tabular is drawn out into a large, blunt ‘horn’ that curves ventrally a short distance toward the dorsal process of the quadrate. At its posteroventral apex, it bears a short ventrally directed process that is clearly too short to have reached the dorsal process of the quadrate, leaving the notch open posteriorly. The postparietals have considerable dorsal exposure on the skull table. The anteroposterior length of each postparietal is approximately half its transverse width. The more or less transverse parietal-postparietal suture is clearly visible, but the precise relationships of the postparietal with the supratemporal and tabular are more difficult to determine. The postparietals bear the raised, rugose posterior rim of the skull table. Pits cover the posterior slope of this rim, but sculpturing does not extend onto their occipital flanges (discussed below under ‘Occiput’).

The maxilla forms the ventral and posteroventral portions of the external naris, and the slightly convex ventral margin of the upper jaw (Fig. 6). Its dorsal lamina is well developed anteriorly, whereas posteriorly it is abruptly narrowed by a large, lateral exposure of the palatine (lep) and suborbital process of the jugal. Together, the latter two bones form a narrow suborbital bar. Neither maxilla is complete posteriorly, and the skull is poorly preserved in this region (Fig. 2 and Fig. 3); as a result, the mutual relationships of the quadratojugal, jugal, and maxilla are unclear. However, as in other dissorphoids (Schoch, 2012), the maxilla probably contacted the quadratojugal posteriorly, preventing the jugal from forming any part of the ventral margin of the skull (Fig. 6). The lacrimal contributes a short, ventrally directed, wedge-shaped process to the anterior orbital wall lateral to the ventral palatine process of the prefrontal. A lacrimal duct could not be recognized on the anterior wall of either orbit. Anterior to the orbit, the lacrimal expands dorsally and ventrally to its maximum height at about one-third the distance to the external naris. Anterior to this level, it is constricted strongly by a convex ventral lamina of the nasal and dorsal lamina of the maxilla, and ends as a small contribution to the posterior margin of the external naris.

The large lep extends from the midpoint of the orbit to well anterior to the anterior border of the orbit. On the left side, a lateral exposure of the ectopterygoid (lee) is clearly visible (Fig. 2), whereas on the right side (Fig. 3), it has fused indistinguishably with the anterior end of the jugal. A separate lee has been described in some trematopids (Dilkes, 1990) and in juvenile Cacops (Reisz et al., 2009), but its occurrence in the latter suggests that it might be an ontogenetic feature (Reisz et al., 2009). The large postorbital tapers to a narrow triangular process as it extends far ventrally along the posterior orbital wall to its posteroventral corner. A very similar configuration is seen in B. olsoni (Bolt, 1974b, fig. 4) and in at least some specimens of Cacops (Reisz et al., 2009), but is otherwise unreported in other dissorophoids. The broadly rounded posterior portion of the postorbital narrows as it inserts between the postfrontal and squamosal to contact the supratemporal. The lateral edge of its modest contribution to the anterolateral corner of the skull table bears a ridge-like eminence that separates a distinct dorsally facing depression and a more poorly developed lateral depression. A similar morphology is seen in Cacops (Williston, 1910), although the ridge and depressions are better developed and occupy a slightly different position. The jugal and postorbital form most of the cheek. The superficial portion of the squamosal (i.e., that bearing dermal sculpturing) is restricted to a narrow crescent that forms the anterior margin of the otic embayment. The extensive, inset surface of the embayment is formed almost exclusively by a non-sculptured portion of the squamosal. Dorsal exposure of the squamosal on the lateral margin of the skull table is limited to a small triangular area at the postorbital-supratemporal suture (Fig. 5). The quadratojugal exhibits a low profile in lateral view, with its anterior portion beneath the jugal narrowing to a process-like extension that narrowly contacts the posterior end of the maxilla (Fig. 3).

Both quadrates are preserved, but the left is more complete. Each bears a short, stout dorsal process that flares slightly at its dorsal end. It terminates in a facet of unfinished bone (Fig. 2), suggesting that it had not finished growing, or perhaps supported a ligament that extended up to the tabular.

The otic embayment is large and nearly circular in outline (Fig. 6) rather than horizontally oval as in B. texensis and B. brevis. The shape of the embayment in the latter taxa is a consequence of the low profile of their skulls, but may have been exaggerated by postmortem compression. The squamosal includes extensive, non-sculptured, slightly inset ventral portion that borders the otic embayment. Anterodorsally, the embayment becomes increasingly depressed below the superficial, sculptured surface of the squamosal. Together, the supratemporal and tabular form an extremely narrow, laterally projecting supratympanic shelf that borders the dorsal margin of the embayment. Its lateral edge bears a continuation of the dermal ornamentation of the skull roof, whereas its ventral surface is smooth. A large supratympanic flange forms the inset dorsal surface of the otic embayment. It is unusual in that, unlike other dissorophids, the supratemporal does not appear to contribute a semilunar flange (Bolt, 1974a); rather, the supratympanic flange appears to be almost completely formed by the squamosal. Elongate, horizontal striations on the external surface of the flange suggest rapid posterior growth of the squamosal rather than ventral-ward growth from the ventral surface of the supratemporal. The tabular contribution to the flange is restricted to a small portion of its posterodorsal corner (Fig. 6). The otic notch is small, especially anteriorly, where it is restricted to a narrow, anterodorsally oriented slot, much as in B. olsoni (Bolt, 1974a, fig. 4) and B. brevis (Carroll, 1964, fig. 9B), as well as in some trematopids.

The orbits of CM 41705 (Fig. 2 and Fig. 3; Table 1) are relatively much larger than in Dissorophus and B. texensis (Schoch, 2012, fig. 1C, E), but are comparable to those of ‘B.’ novomexicanus, B. olsoni, and B. brevis (Schoch, 2012, fig. 1B, D, F). The orbital margins of most of the circumorbital bones are deflected inward, producing a distinctly thickened orbital wall, especially anteriorly. Between the orbit and naris, the medial surfaces of the prefrontal, nasal, and possibly the lacrimal contribute to a sheet of bone that projects ventromedially from their ventral surfaces to form the dorsal wall of the bony nasal capsule (Fig. 7). This appears to be homologous to the narial flange described by Bolt (1974b) and Dilkes (1993). Several small rectangular plates of bone are visible along the ventral edge of the right orbit (Fig. 3). They appear to be associated with the medial margins of the palatine. Similar ossifications have been described in Cacops (Reisz et al., 2009, fig. 3) and may represent ossifications of soft palate that spans the interpterygoid vacuity.

The vomers, palatines, ectopterygoids, and pterygoids all bear a shagreen of small, closely spaced denticles, but conspicuous longitudinal denticle-bearing ridges are absent. In addition, small bony plates bearing denticles overlie, but do not attach to, the ventral surface of the vomers. Similar ossicles have been reported in amphibamids, dissorophids, and trematopids (Fröbisch and Reisz, 2008 and Schoch and Rubidge, 2005). The vomers narrow rapidly anteriorly to about half their maximum posterior width as they extend between the internal nares. In contrast to the vomers in Broiliellus and Dissorophus, in which they form a broad, flat plate, the vomers of CM 41705 are narrowly separated at the midline along nearly their entire length. Here they turn sharply dorsally as vertical laminae to form a median structure referred to as the internasal septum by Bolt (1974c), each forming the medial wall of a bony nasal capsule (Fig. 4 and Fig. 7). Together, they define a deep, narrow, median internarial furrow (Fig. 5B), the external surface of which is covered by denticles almost to its dorsal margin (Fig. 7). Dorsally, the laminae converge, each approaching, and possibly articulating with, the ventromedially projecting narial flange formed by the prefrontal and nasal (Fig. 7; see ‘Orbit’ above). A similar morphology has been described in trematopids (Dilkes, 1990, Dilkes, 1993 and Polley and Reisz, 2011), and possibly in B. brevis (Schoch, 2012, fig. 2F), although Carroll (1964) does not mention this feature in his description of the latter species. Each vomer bears a fang and companion replacement pit that flank the anterior end of the internarial furrow. Preservation is too poor to establish whether an intervomerine fontanelle was present. Laterally, each vomer forms the long, dorsally inflected medial margin of a choana.

The palatine forms the anterolateral margin of an interpterygoid vacuity, widely separating the vomer and pterygoid. The palatine extends anteriorly as a long, triangular process along the medial surface of the maxilla, partially constricting the lateral posterior third of the choana (Fig. 4 and Fig. 5B). A pair of stout fangs occupies the posterior base of the process. The lateral margin of the choana is formed by a medially projecting alveolar shelf of the maxilla. Posteriorly, the palatine articulates with the pterygoid medially, and the small ectopterygoid laterally. A pair of small ectopterygoid fangs are located a short distance posterior to the palatine-ectopterygoid suture. The anterior palatal process of the pterygoid is short and borders the posterior half of the interpterygoid vacuity. A modest transverse flange projects into the ventral opening of the adductor chamber. The basal process of the pterygoid is stout and dorsoventrally expanded distally (Fig. 4 and Fig. 7). Its articulation with the basipterygoid process of the basisphenoid cannot be identified on either side, suggesting that the basicranial joint had co-ossified, but the bone is broken in this area, and preservation is too poor to be sure.

Because the elements of the occiput are slightly disarticulated, and the atlas-axis complex remains in articulation with the occipital condyle (Fig. 1 and Fig. 4), some details of this portion of the braincase are obscured. Nevertheless, it is clear that the postparietals have extensive, smooth occipital flanges (Fig. 6B). Although their common median suture cannot be traced, their sutures with the much smaller occipital flanges of the tabulars are clearly visible. The paroccipital processes of the opisthotics are poorly preserved, and in neither case is the sutural contact with the prominent, ventromedial occipital flange of the tabular that forms the ventral border of the posttemporal fossa in other temnospondyls evident. Nevertheless, a deep sulcus on the dorsal portion of the posterior surface of the right opisthotic just ventral to its suture with the lateral edge of the postparietal probably represents a small posttemporal fossa. However, dorsoventral compression has forced the opisthotic and postparietal together, making it impossible to determine whether the fossa leads to a fenestra opening anteriorly into the adductor chamber. The right stapes is preserved in approximately its correct position. It comprises a modestly expanded footplate and a straight, distally narrowing shaft. A stapedial foramen is not evident. The slightly displaced right prootic is preserved immediately anterior to the stapedial footplate. The exoccipital and basioccipital appear to form a strongly co-ossified complex, with the exoccipitals being fused indistinguishably above the foramen magnum, which may simply indicate advanced maturity (Maddin et al., 2010). The exoccipital-basioccipital complex is slightly separated from the skull table, and the ventral edges of the occipital flanges of the postparietals with which it articulated are consequently broken into displaced fragments. The atlas is firmly articulated with the concave occipital cotyle (= ‘condyle’), but it is clear that the cotyle is not ‘figure-8’ shaped, as it is in most dissorophoids. Rather, the basioccipital forms the greater portion of the cotyle, giving it a dorsoventrally compressed oval or ‘kidney bean’-shaped outline with a gently convex ventral margin.

The parasphenoid comprises a broad, posteriorly widening basal plate approximately twice as wide (measured across the basal tubers), as long (measured at the midline), and an anteriorly projecting cultriform process. The latter is narrow in ventral view, and approximately circular in cross-section. It is broken and poorly preserved anteriorly. Foramina for the carotid arteries could not be identified. The basal plate lacks denticles, but small, scattered plates bearing denticles are preserved in this area. Several small plates lacking denticles and similar to those preserved associated with the right palatine (see ‘Orbit’ above) are preserved in close proximity to the cultriform process. In Dissorophus and other species of Broiliellus, the basal plate bears deep pockets for attachment of axial musculature (Schoch, 2012, character 65). The presence of this character cannot be established in CM 41705 as a result of postmortem disruption of the bone surface and presence of the overlying bony plates.

The sphenethmoid (Fig. 4 and Fig. 5) is large and quasi-barrel-shaped. It appears to be co-ossified with the ventral surface of the skull roof. The posterior portion of the cultriform process of the parasphenoid remains attached to its midventral surface. The sphenethmoid is separated from the otic region by a large gap, presumably for passage of branches of the fifth, sixth and seventh cranial nerves. A pair of foramina at the posterior end of the sphenethmoid probably allowed anterior passage for the optic nerves.

The mandible (Fig. 2 and Fig. 3) is not completely exposed or preserved, but most of its anatomy can be reconstructed (Fig. 8) by referring to both rami. The left dentary, better exposed (Fig. 2), reveals a broken parasymphysial tusk and companion replacement alveolus adjacent to the symphysis. As in most other dissorophoids, there is no evidence for a separate adsymphyseal ossification, but preservation is poor in this area. The anterolateral surface of the dentary bears the typical temnospondyl ‘pit and ridge’ ornamentation. Posteriorly, this ornamentation rapidly transitions into elongated horizontal grooves that become progressively shallower. The posterior end of the dentary is incomplete, making its relationship with the surangular unclear. The ventral edge of the ramus is not well preserved, but lateral exposure of the splenial and postsplenial appears to have been extremely marginal. The lateral surface of the angular is strongly ornamented, and at its posteroventral corner is drawn out into a distinct, distally swollen keel. Its anterior tip, which appears to be complete, does not extend as far forward as it does in most temnospondyls.

The medial surface is better preserved on the right ramus. All three coronoids are completely covered with small denticles, but there is no evidence for coronoid fangs. The large anterior coronoid forms nearly the entire anterior half of the coronoid series, restricting the exposures of the dentary and splenial to the symphysis and ventral edge of the ramus, respectively. This morphology is confirmed on the left ramus, which also shows that the splenial extends almost to the symphysis. The left ramus preserves the sutural contact between the anterior and middle coronoids. Most of the posterior coronoid is preserved, but its suture with the middle coronoid could not be identified. The medial exposure of the postsplenial extends anteriorly as a narrow triangular wedge beneath the middle coronoid and posterior end of the anterior coronoid, ending in a very narrow contact with the posterior end of the anterior splenial. Posteriorly, the contact between the postsplenial and angular is preserved. The dorsal and posterior rim of a modest-sized Meckelian fenestra is preserved. The posterior portion of the left ramus is disrupted, but essentially complete. The dorsal rim of the posterior coronoid forms a prominent, rounded crest. Coarse dermal sculpturing extends onto the ventromedial surface of the angular. The prearticular has been broken into several pieces, and displaced posteriorly to expose a portion of the articular. It exhibits no evidence of a medial inflection along the rim of the adductor chamber, as seen in trematopids (Dilkes, 1990).

The marginal dentition is not well preserved, and about half of the teeth being missing. The preserved teeth are small, and most have lost the tips of their crowns, but the most complete form simple, blunt, conical pegs. They show little recurvature, and there is no evidence of accessory cusps, pedicellate structure, or labyrinthine infolding. Neither premaxilla is complete ventrally, but it probably supported seven or eight teeth. A few of the more posterior premaxillary teeth are relatively well preserved, and appear to be the largest of the upper tooth row. The anterior-most two or three maxillary teeth are slightly larger than the others, but there is no significant development of caniniform dentition. Each maxilla would have had spaces for approximately 40 tooth positions. DeMar (1967, p. 126) estimated 47 or 48 teeth in the “upper jaw” of B. olsoni, but did not specify whether this includes the premaxillary dentition. ‘B.’ novomexicanus has “at least 56” (Langston, 1953, p. 381). Eighteen small, straight-sided conical teeth, most alternating with replacement pits, are preserved in the left dentary; the total count is estimated at 45 tooth positions. Tooth size increases gradually toward the posterior end of the dentary, although in the last several teeth of the series, this trend is reversed.

The left orbit is almost completely filled with large, irregular-shaped palpebral ossifications that bear muted dermal ornamentation on their external surfaces similar to that on the skull roof (Fig. 2).

The right scapulocoracoid, cleithrum and clavicle are preserved in articulation (Fig. 9). The cleithrum expands slightly at its dorsal termination, but does not extend posteriorly over the dorsal edge of the scapula. The interclavicle is partly covered by the ventral, horizontal plate of the clavicle. Too little of the former is exposed to estimate the shape of the bone, but it does not appear to be very large. The right humerus is preserved in articulation with the glenoid of the scapula. Most of the proximal head and surrounding margin of the glenoid are badly eroded. The central portion of the shaft was lost at some point after collection, but its impression is preserved in the surrounding matrix. The distal head and enough of the distal portion of the shaft are preserved to confirm that neither a supinator process nor entepicondylar foramen was present.

Posterior to the scapular blade and dorsal to the humerus are three short, curved, blade-like ribs (Fig. 9).

Ten rhachitomous presacral vertebrae are preserved in articulation. Four successive vertebrae in the middle of this series lack the dorsal portions of their neural spines that appear to have been lost during collection. The other vertebrae (Fig. 10) retain their neural spines and associated dorsal osteoderms, although some of the latter are slightly displaced. The osteoderms are organized into an internal and an external series. Each osteoderm of the internal series is closely associated with, if not fused to, the dorsal edge of a neural spine. There is no evidence of ventrally projecting flanges inserting between successive neural spines as described in Dissorophus or Broiliellus (DeMar, 1966a). Each osteoderm of the external series is intersegmental in position, apparently contacting dorsally facing facets on the anterior and posterior margins of successive internal osteoderms. A sculpturing of sub-circular pits, resembling that on the skull roof, covers the external surfaces of all osteoderms. This pattern is quite distinct from that exhibited by the osteoderms of the type of B. texensis (Williston, 1914, fig. 3), which comprises mediolaterally elongate grooves. None of the osteoderms extend far laterally, being at most half as wide as those in the type of B. texensis (Williston, 1914, fig. 3) despite the fact that the skull of CM 41705 is approximately 10% larger (Table 1). They also appear to be much smaller than the osteoderms described in ‘B.’ novomexicanus (Williston, 1911, plate XXXVIII). Of all known dissorophoids, the osteoderms of CM 41705 most closely resemble those of Cacops (DeMar, 1966a and Dilkes and Brown, 2007) in relative size, shape, number per segment, arrangement, and relationship to the neural spines.

Our phylogenetic analysis nests CM 41705 securely within the genus Broiliellus. However, it exhibits a number of autapomorphies that clearly establish it as a distinct species, B. reiszi. The nasals, rather than being anteroposteriorly elongate rectangles, are very narrow anteriorly, no more than half as wide anteriorly as they are posteriorly. Similarly, the vomers are half as wide anteriorly as posteriorly. The vomers in other dissorophoids are only slightly narrower anteriorly. These unusual proportions are the consequence of the triangular, anteriorly narrowing snout. The width of the posterior third of the choana is somewhat narrowed as the result of an anterior projection of the palatine. The parasphenoid basal plate is greatly flared posteriorly to a width nearly twice that measured at the level of the basal articulation; only the amphibamid Doleserpeton shares this feature.

The suture pattern on the supratympanic flange is unusual. Typically in dissorophoids, the supratemporal, tabular, and squamosal intersect in a roughly ‘Y’-shaped pattern, on or just posterior to the semilunar flange of the supratympanic flange (Bolt, 1974a, fig. 2). In dissorophoids, the supratemporal typically forms a portion of the flange, but a ventral contact between the squamosal and tabular precludes the supratemporal from forming any part of its ventral margin. In B. reiszi, the supratemporal does not seem to form any part of the flange, the latter being formed almost entirely by the squamosal, with only a small contribution from the tabular.

The basioccipital is large and contributes significantly to the ventral portion of the occipital condyle. As a result, the condyle is horizontally oval or kidney-shaped in posterior outline rather than the ‘figure-8’-shape typical for other dissorophoids. The medial edges of the vomers are deflected dorsally, forming an ‘internarial septum’ (character 44 of Schoch, 2012). However, this feature might have a wider distributed than is currently appreciated. It has been reported in trematopids (Dilkes, 1990, Dilkes, 1993 and Polley and Reisz, 2011) and possibly in B. brevis (Schoch, 2012, fig. 2F). CT scans have revealed its presence in Doleserpeton and numerous other taxa (J.S.A., per. obs.) to be described elsewhere. As such, the phylogenetic significance of this feature is unclear. Finally, the angular has a prominent swollen keel on its posteroventral corner. More typical for dissorophoids is the presence of a sharp keel, as seen in Cacops (Reisz et al., 2009 and Williston, 1910).

Our majority rule consensus tree hypothesizes a sister group relationship between B. reiszi and all other species of Broiliellus + Dissorophus (Fig. 11) despite the numerous differences seen in cranial and osteoderm morphology amongst these taxa. Of all Broiliellus species, B. reiszi bears the least resemblance, in terms of its ‘Gestalt’, to B. texensis. In the latter, the preorbital region is distinctly shorter than the postorbital region. Its snout is more broadly parabolic in dorsal aspect, and lacks any trace of a facial constriction. The skull is low and wedge-shaped in lateral profile, with the dorsal surface rising gradually from above the external nares to the skull table. The cheeks are not as tall, resulting in an anterodorsally oriented oval otic embayment that is distinct from the circular embayment in B. reiszi. The nares and orbits are distinctly smaller. The vomerine plate is essentially flat, with no vaulting or other indication that it possessed an internarial septum.

These many differences, in particular between B. reiszi and B. texensis despite the close relationships hypothesized in our analysis, highlights the importance of avoiding establishing ‘key features’ to define taxa, because the test of congruence of all features can lead to contradictory topologies. For example, the arrangement and structure of osteoderms is most similar to the more distantly related Cacops. However, as Schoch (2012) noted, the monophyly of Dissorophoidea does not rest on the presence, composition and structure of osteoderms (many other non-related characters supports this). In any case, there is considerable variation in osteoderm morphology within at least some dissorophid taxa, as well as overlap in morphology between taxa (Berman and Lucas, 2003), indicating that the various patterns may have evolved convergently from the simple osteoderms of a basal dissorophid such as Aspidosaurus. If so, osteoderm morphology is of doubtful utility in dissorophid systematics.

On the other hand, B. reiszi shares a number of obvious similarities with B. olsoni despite the more remote relationship hypothesized by our majority rule consensus tree. As in B. olsoni, the skull is triangular in dorsal profile (DeMar, 1967 and Schoch, 2012). This form is distinct from the more parabolic shape seen in most dissorophids, including other Broiliellus species (but not Dissorophus, interestingly; Schoch, 2012, fig. 1C). Most dissorophids except B. olsoni and B. reiszi have a straight ventral maxillary and premaxillary margin. With the exception of B. olsoni and ‘B.’ novomexicanus (‘Rio Arriba Taxon’ of Schoch, 2012), the external nares and orbits are relatively large compared with other Broiliellus species and dissorophids in general except juvenile individuals. The nares are also circular, unlike the ventrally compressed ellipses present in other Broiliellus species.

B. reiszi has a high skull, with nearly vertical cheeks, a dorsal profile that quickly rises posteriorly over the external nares, and by the midpoint between the naris and the anterior margin of the orbit, is nearly as high as it is over the occiput. This profile is very similar to that in B. olsoni, but distinct from that seen in other Broiliellus species, which is distinctly wedge-shaped in lateral view. This high-sided skull results in a nearly vertical anterior margin of the otic embayment, much as described for B. olsoni by DeMar (1967). The slit-like extension of the anterior portion of the otic notch is very similar to that of B. olsoni (Bolt, 1974a, fig. 4). B. reiszi also shares with B. olsoni the unusually large postorbital that extends far ventrally (Bolt, 1974a, fig. 4). Among dissorophoids, this feature has been otherwise described only in Cacops (Reisz et al., 2009). Finally, the tooth count is similar between these two taxa (47 or 48 teeth in the upper jaw in B. olsoni compared with an estimated 40 maxillary teeth and seven or eight premaxillary teeth in B. reiszi).

One other species of Broiliellus has been described from New Mexico. Originally named ‘Aspidosaurus’ novomexicanus (Williston, 1911) on the basis of an incomplete skull and partial postcranial skeleton, probably from El Cobre Canyon, Langston (1953) reassigned the material to Broiliellus based on a slightly better skull from the Arroyo de Agua region (about 15 miles west of El Cobre Canyon). Schoch (2012), referring to it simply as the ‘Rio Arriba Taxon’, provided a reconstruction of the skull and included it in a phylogenetic analysis. In his analysis, it was placed as the sister taxon to the cacopines (Cacops + (Kamacops + Zygosaurus)). It retains this position in our majority rule consensus (Fig. 11), although now forms an unresolved polytomy with cacopines, Conjunctio, Aspidosaurus, and the ‘Admiral Taxon’ of Schoch (2012). This specimen is currently under study by R. Schoch, and we look forward to a resolution of this taxon's relationship to Broiliellus.

Although the general proportions of ‘B.’ novomexicanus are similar to those of B. reiszi (relative length of snout, large orbits and nares, etc.), the snout is blunter, and the quadrate is placed further rostrally—distinctly anterior to the occiput (Schoch, 2012). In B. reiszi, the quadrate is slightly posterior to the occiput. Other features that distinguish ‘B.’ novomexicanus from B. reiszi include: the presence of an interpremaxillary fontanelle, much narrower skull table, relatively narrow interorbital distance (Schoch, 2012), a higher maxillary tooth count (56 compared with approximately 40, Langston, 1953), an unusually large postfrontal that inserts posteriorly between the postorbital and supratemporal, preventing the latter two bones from making contact (character 55 of Schoch, 2012), and considerably larger dorsal osteoderms (Williston, 1911, plate XXXVIII).

B. reiszi is similar to Cacops in a number of features. Other than the long ventral extension of the postorbital and the lateral extent of the osteoderms, it shares with Cacops the general distribution of cranial eminences and ridges on the skull. Additionally, this new species preserves a lee on the left (but not on the right) side of the skull. Some juvenile specimens of Cacops also exhibit a separate lee, suggesting the possibility that the lee is more widespread among dissorophoids but has gone unnoticed because of its fusion with the jugal relatively early in development. However, the holotype of B. reiszi appears to pertain to a relatively mature individual. Both dermal and endochondral skull elements are well ossified, co-ossification of sutures are generally well advanced, dermal sculpturing is coarse, vertebral elements are well ossified, and limb bones bear well-formed articular surfaces. As such, it is unclear whether the retention of an open suture between the jugal and lee in B. reiszi represents individual variation, or a heterochronic shift. Further study of dissorophoids using CT could dissect this area and more clearly delineate any connection between the jugal and ectopterygoid in other taxa.

The morphological disparity exhibited by the species of Broiliellus is striking. The variation in much of the morphology (e.g., the proportions of the snout, the very low and wedge-shaped skull in B. texensis but tall and convex in B. reiszi and B. olsoni) is more conspicuous than expected in closely related taxa. Species within a genus, or genera within a family tend to conform more closely to a standard morphology. Furthermore, some species within the genus Broiliellus share features with ‘B.’ novomexicanus and Cacops despite their apparently more remote relationships (Fig. 11). Whether this is a consequence of historical taxonomic practice, a failure of the current character by taxon matrix to capture existing variation, poor taxonomic sampling, or a real biological phenomenon, remains to be discovered.