RESULTS

The computed tomography and computer generated three-dimensional isosurface reconstruction techniques employed here yield results comparable to high fidelity casts enabling viewing of morphology in three axes of rotation. Additionally, serial sections illustrate topological relationships of individual elements, providing better understanding of those that are obscured by overlapping elements or matrix residue. Figure 5.1 is a low angle light photograph of the ventral surface of the specimen as preserved today. Figure 5.2 is the same view using the isosurface reconstruction of the CT data set. Comparison of Figure 5.1 and Figure 5.2 demonstrates the fidelity and therefore the viability of computer generated reconstruction of CT data for illustrative and quantitative purposes and provides a qualitative reference for subsequent discussion of the resin obscured dorsal surface of the skull. The isosurface reconstruction of the resin embedded dorsal surface is represented in Figure 5.4 as a stereo pair. Figure 5.3 and Figure 5.5 are interpretive line drawings derived from isosurface reconstructions, photographs, and microscopic study of the specimen. Elements that have been the subject of conflicting interpretation are indicated by ue1 through ue5 (unidentified elements) and are referenced in the following sections.

Orientation and Morphology of the Quadrates

The relationship of the quadrate to the supratemporal dorsally and the glenoid of the compound bone of the lower jaw ventrally is largely preserved on the right side, but the quadrate is rotated counterclockwise (Figure 3, Figure 6). Caldwell and Lee (1997) reconstructed Pachyrhachis with vertically oriented quadrates with broad lateral facing surfaces. To test this reconstruction, the length, orientation, and position of morphological landmarks were traced on the dorsal surface of the skull, including the maximum length of the mandibles as well as the center of the glenoid, maxillae, parietal-supraoccipital midline, snout midline, right supratemporal, and right quadrate (Figure 3.2). These tracings were then rotated and adjusted to parallel one another and align the anterior terminus of the premaxillae, maxillae, and mandibles (Figure 3.3).

The position of the right supratemporal was retained and taken to approximate the natural position for three reasons: 1) the primary force that crushed the specimen was orthogonal to the bedding plane as previously discussed; 2) this primary force also manifested in the leftward displacement of the left supratemporal and the leftward crushing of the sagittal crest (as the result of a leftward incline of the skull due to the right mandible underpinning the right side of the skull, which influenced left lateral but not anterior displacement); and 3) asymmetrical rotation and outward collapse of the quadrates was not likely to have contributed to anterior displacement as evidenced by the fact that the quadrates were crushed laterally, not dorsoventrally. This would most likely be the case if the quadrates splayed out prior to the remaining portion of the skull being crushed, and thus the resistive forces of the quadrates would be more medially oriented, not anteriorly.

The supratemporals in Pachyrhachis are long relative to other snakes. As preserved, the posterior free-ending extensions of the supratemporal comprise approximately one-third their total length. The anterior ends reach anterior to the anterior terminus of the prootic. In Recent snakes the supratemporal does not extend anterior to the anterior terminus of the prootic, and in that respect may justify a more posterior reconstruction. However, reconstruction of the supratemporals in a more posterior position would in fact exaggerate the macrostomatan condition of the distal extension of the supratemporal beyond the skull roof. The supratemporal and the parietal are developmentally part of the dermatocranium whereas the prootic is a part of the chondocranium and thus do not share obvious developmental relationships that would preclude the supratemporal anterior terminus exceeding that of the prootic. Given the taphonomic affects discussed and in absence of developmental constraints, the proportionality of sutured versus free-end of the supratemporal and the position of the anterior terminus relative to the anterior extent of the prootic is deemed reasonable.

Taking the length of the quadrate along its midline, the right quadrate requires 19 degrees of inclination to achieve articulation with both the mandibular glenoid posteroventrally and the supratemporal anterodorsally. The degree of inclination is of course controlled by both the position of the supratemporal and the position of the mandible. Therefore, the mandibular articulation surface of the quadrate was examined and compared with a number of alethinophidian and macrostomatan snakes to determine if the trochanter morphology could be used to determine orientation of the articulation. In specimens examined that possess a vertical or nearly vertical quadrate (e.g., Cylindrophis ruffus, Anilius scytale, Xenopeltis unicolor, Python regius ), the trochanter was symmetrical in lateral view. In specimens possessing a slightly posteroventrally inclined quadrate (e.g, Epicrates cenchria) the trochanter was asymmetrical with the majority of the trochanter restricted to the anterior portion. In Pachyrhachis, the trochanter is obscured on the right quadrate, but a well-defined trochanter is present on the anteroventral surface of the left quadrate (Figure 6.4), but not on the posteroventral surface (Figure 6.5). Thus, the articulating surface was predominantly restricted to the anterior portion of the distal quadrate, and therefore, the quadrate was not vertical in Pachyrhachis, but inclined to a moderate degree. This condition is consistent with the position and length of the supratemporal.

The quadrates in Pachyrhachis are preserved with a broad lateral profile due to crushing. The crushed state of the quadrate was apparently interpreted by Caldwell and Lee (1997) as the true morphology as addressed in their description and reconstruction. We disagree with this interpretation. During crushing, the distal portion of the quadrates was rotated relative to the cephalic condyles yielding the broad surface as preserved. In snakes the quadrate articulation with the supratemporal where present is in the para-sagittal plane and the mandibular condyle long axis is oriented orthogonal to the mandibular axis. Removal of distortion introduced by crushing was achieved by aligning the cephalic condyle with the sagittal plane and rotating the mandibular condyle to bring it roughly orthogonal to the sagittal plane, distributing the rotation linearly along the length of the quadrate (see Appendix 5 for details of reconstruction).

Stapes

The elements that Haas (1979) referred to as squamosals are the supratemporals as recognized by later authors. The tentative squamosal of Caldwell and Lee (1997) is the proximal portion of the right stapes as identified by Zaher and Rieppel (2002). The distal end of the stapes was correctly identified by Caldwell and Lee (1997) but incorrectly identified as the opisthotic paroccipital process by Zaher and Rieppel (2002). The element in question is sandwiched between the posterior pterygoid quadrate ramus, the distal supratemporal, and the quadrate (Figure 6.1-6.3). Apparently, the right stapes is broken or bent along its shaft. Our identification of this element as the stapes is supported by tracing it in serial sections 180 to 240 and 280 to 305 (Figure 7). The shafts representing both the proximal and distal portions of the right stapes are visible in the same plane. The stylohyal process of the quadrate is visible in Figure 6.3, 6.6. The reconstructed length of the stapes is consistent with the distance from the stylohyal process to the inferred position of the fenestra ovalis (Figure 4).

Comments. The supratemporal is extremely reduced or absent in scolecophidians (Zaher and Rieppel 2002) and the quadrate abuts the sidewall of the braincase. In alethinophidians examined (e.g., Cylindrophis ruffus, Anilius scytale), the supratemporal is relatively short and the quadrate is supported by both the supratemporal and the braincase and the stapes articulates with a posteriorly directed suprastapedial process of the quadrate via a separate intervening calcified cartilage (extracolumella or stylohyal) as is the case in Dinilysia (Rieppell 1979, figure 4 A, B; see also Caldwell and Albino 2002 for discussion of Dinilysia). In macrostomatans examined (e.g., Python regius, Python curtis, Epicrates cenchria), the supratemporals project posteriorly and provide sole support of the quadrate (see also Rieppell 1979, figure 4C), and the stapes articulates with the middle to distal medial quadrate shaft via a fused calcified cartilage, the stylohyal. In all macrostomatan specimens examined the quadrate shaft exhibits a medial protuberance at the point of articulation with the stapes, the stylohyal process. Pachyrhachis is found to lack a suprastapedial process and is shown to possess a stylohyal process on the distal medial quadrate shaft (Figure 6.3). Thus, evidence provided by the free-ending supratemporal, presence of a stylohyal process, correction of the orientation of the cephalic and distal condyles and resulting morphology of the quadrate, and the posteroventral inclination of the quadrate from the supratemporal to the articular agree best with the condition in macrostomatan snakes (see Appendix 5 for reconstruction of quadrate).

Number of Mental Foramina

A single mental foramen is visible in both isosurface and sectional views. The second foramen illustrated by Caldwell and Lee (1997, figure 1A; see also Figure 2.2 this paper) is an artifact of crushing into the void formed by the Meckel’s groove anterior to the anterior terminus of the surangular (Figure 8.1, 8.4-8.7, arrow 3; see also Appendix 2). Tracing through the slices, it is apparent the depression opens into the Meckel’s groove. Also, a third depression on the antrolateral surface of the left dentary superficially resembles a foramen (Figure 8.1, arrow 1). However, this structure lies within a crushed and distorted area and is merely an artifact of preservation. In life, the anterior tips of the dentaries curved sharply medially, approximating the dorsal outline of the maxillae. The main body of each dentary is preserved nearly perpendicular to its anatomical position causing the anterior medial portion to deform, twisting nearly 90 degrees, resulting in an area of crushing and fractures that can be seen in serial section (Figure 8.2-8.7). One of the foramina (Figure 8.1, arrow 2) can be traced into unfractured bone, has well- defined edges, and is interpreted here as a single and only true mental foramen. A single mental foramen is characteristic of Serpentes (Rieppel et al. 2003; Lee and Caldwell 1998).

Circumorbital Series

All authors agree on the identification of the postorbitals lying symmetrically on either side of the skull. Medial extensions (Figure 5, ue4 and ue5) of the postorbitals, overlying the anterior parietal as illustrated by Caldwell and Lee (1997), gives them a lizard-like appearance (Figure 2.2). However, Zaher and Rieppel (1999b) did not figure the medial structure as part of the postorbital but did indicate topographical relief in their interpretive drawing on the left side of the skull (Figure 2.3). It is unclear from the isosurface reconstruction (Figure 5.4, 5.5; see also Appendix 2) if the medial extensions were ever in contact with the undisputed portion of the postorbitals. However, it is clear that they are now topologically displaced and discontinuous with the postorbitals. The structures lie in a portion of the parietal table that is crushed and distorted. The structures themselves may be artifacts caused by crushing on top of the basipterygoid processes (Figure 5.4, 5.5). In any event the identity of these structures is ambiguous at best.

The most controversial elements in the skull of Pachyrhachis are those variably referred to as postfrontals by Haas (1979), as jugals by Caldwell and Lee (1997), and the broken anterior ectopterygoids by Zaher and Rieppel (1999b). They are shown here in Figure 5.5 (elements labeled ue1 and ue2). These elements are in close proximity to the distal terminus of the undisputed postorbitals lying somewhat symmetrically on the suborbital process of the maxillae. We are reticent about precisely identifying ue1 and ue2 as postfrontals, ectopterygoids, or jugals, because the evidence for each is the same, namely where they lie.

No recent author agrees with Haas’s (1979) identification of these elements as postfrontals. Given that the apparent force vector responsible for the crushing was dorsoventral, it is unlikely that the postfrontals would settle in the oblique position of these bones.

Supporting the argument of Zaher and Rieppel (1999b) that ue2 is a broken fragment of an ectopterygoid, is the fact that the right dentary is rotated under the skull, interrupting the natural position of the right ectopterygoid relative to the maxilla and pterygoid (Figure 5.1-5.5) and potentially causing breakage. The right pterygoid is displaced anteriorly and rotated to overlap slightly the parasphenoid (Figure 5.3). Thus it could have applied anteriorly directed force to the ectopterygoid and possibly caused breakage and displacement. Given the foreshortening of the left side of the skull and the leftward rotation of the snout, the ectopterygoid could have compressed in its long axis, retaining its natural anterior articulation with the maxilla, the broken posterior portion rotating clockwise and medially, coming to rest in its current position.

Arguing against identification as an ectopterygoid is Haas’s (1979) correct (in our opinion) identification of the ectopterygoid on the left side and the otherwise consistent symmetry of other elements of the skull. Element ue3 lies medial to the dentary and is identified here as anterior right ectopterygoid (Figure 5.5; see also Appendix 2 and Appendix 3 for illustration of topological and dorsoventral relationships). Additionally, element ue1 on the left side and ue2 on the right are more or less symmetrically displaced (Figure 5.4-5.5). However, the left mandible could not have applied the same amount and direction of force on the left ectopterygoid as on the right, because the left mandible now lies lateral to and does not interfere with the natural position of the left ectopterygoid. It is also unlikely that sufficient force could be applied to the ectopterygoids to cause the pattern of breakage and displacement preserved in ue1 and ue2. We reject the hypothesis that the elements are displaced broken ectopterygoids on the basis of identification of the intact ectopterygoids and taphonomic interpretations.

Caldwell and Lee‘s (1997) identification of the elements ue1 and ue2 as jugals is topologically plausible in part. However, their interpretation presents three issues. First, there is no clear and definitive justification for the presence of a jugal except topological position in relation to the maxillae, but, given the taphonomic affects identified herein, that same evidence is applicable and equally compelling with respect to the other alternatives. Second, the element’s relationship to the postorbitals and maxillae as preserved is problematic as there is no clear articulation facet or suture with either element (Appendix 3). Additionally, the position of the right and left elements relative to the maxillae are different. The medial portion of the left element is in contact with the maxilla whereas the lateral portion of the right element overlies the maxilla. Therefore, other than proximity to the postorbitals, there is no clear indication of articulation with the maxillae. Third, the morphology of the element is inconsistent with the morphology of the jugal in any known squamate in that it is merely a flat piece of bone that widens anteriorly and thus displays no morphology that clearly defines it as a jugal. Jugals have been reported as present in Dinylisia (Estes et al. 1970), but their presence was subsequently contradicted by Caldwell and Albino (2002). Jugals are unknown in all other snakes. We reject the hypothesis that Pachyrhachis had jugals on the basis of its morphology, topological relationships, and taphonomic considerations.

The identity of the elements ue1 and ue2 is most parsimoniously explained as the ventral portion of elongate postorbitals. The undisputed postorbitals are in fact the dorsal portion of the postorbitals and share a long articulation with the anterolateral parietal, and likely contacted the frontal as in some macrostomatans (Lee and Scanlon 2002b, figure 2 d,e,f). This is supported by multiple lines of evidence: 1) the anterior portions of both of the ectopterygoids are visible in dorsal view; 2) the ventral ends of both ue1 and ue2 are in contact with the anterior ends of the ectopterygoids; 3) the dorsal ends are in contact with the undisputed postorbitals; 4) the combined length of the undisputed portion of the postorbitals and their ventral counterparts is accommodated by the anterolaterally elongate parietal; and 5) taphonomic affects evidenced in the distortion of other morphology suggest predominantly dorsoventral forces acted upon the fossil and thus would reasonably produce the resultant configuration of the postorbitals, ue1, and ue2. Figure 4 illustrates the effects of crushing on HUJ-PAL 3659 (See also Appendix 4 for animation sequence of crushing).

Exoccipitals

The exoccipitals are compact and project only a short lateral distance (Figure 9.2). The area above the basioccipital preserves the supraoccipital but obscures the foramen magnum (Figure 9.1). There is evidence of' slight crushing parasagittally on the dorsal surface of the supraoccipital, indicating weak support from below and allowing the possibility of separation of the exoccipitals. However, this character remains inconclusive and must be treated as unknown in Pachyrhachis. Exclusion of the supraoccipital from the foramen magnum by contact of the exoccipitals has long been considered a synapymorphy of Serpentes (Underwood 1967; Rieppel 1979; Estes et al. 1988), but Zaher and Rieppel (2002) indicate this character appears to be variable within Macrostomata.

Dorsal Laminae of Maxillae

Haas (1979) mistook the left prefrontal for the frontal. Caldwell and Lee (1997) correctly identified prefrontals and frontals but reconstruct an ascending process of the maxillae, overlapping the prefrontal. Zaher and Rieppel (1999b) illustrate the structure as a portion of the prefrontal. The area in question is separated from the maxillae on both the right and left sides and is displaced ventrally relative to the preserved dorsal surface of the maxillae. Light photographs demonstrate artifacts of shadow and surface irregularity that obfuscate morphology when compared with surfaces derived from CT data as evidenced in Appendix 2. Comparing the light photographs and the CT renderings, it is clear these bones are separate from both the maxillae and the prefrontals, albeit not in a uniform way. The narial margin in Pachyrhachis can be traced up to and including the prefrontals, but unambiguous delineation of where the prefrontals end and the maxillae begin is difficult. However, the more posterior portion of the external nares curve medially as preserved and thus would indicate at least a moderate development of the dorsal laminae of the maxillae and that interpretation is accepted here. Some alethinophidians retain a modest dorsal lamina of the maxilla at the point of articulation with the prefrontal as in Cylindrophis (Lee and Scanlon 2002b, figure 4B). The presence of dorsal laminae does not exclude interpretation of Pachyrhachis at least at the level of Alethinophia.

Summary of Characters

Of the six characters examined all have been shown to: 1) have been misinterpreted; 2) possess the Serpentes, the alethinophidian, or the macrostomatan condition; or 3) be absent or ambiguous. Table 1 summarizes the interpretations of this study in comparison to previous studies. The quadrate lacks a suprastapedial process. A stylohyal process is present on the ventral medial shaft. The posteriorly projecting supratemporal provides sole support for the quadrate, which is inclined. Pachyrhachis is shown to possess an elongate and slender stapes and a single mental foramen. A jugal is absent, and the elements referred to herein as ue1 and ue2 are considered the ventral portion of the postorbital. The exoccipital contact above the foramen magnum is unknown. The maxillae possess a moderately developed ascending process.

The presence of a single mental foramen and loss of the jugal is diagnostic of Serpentes (Lee and Caldwell 1998; Tchernov et al. 2000). A dorsal lamina of the maxilla is present in alenthinophidians but is reduced in Recent macrostomatans. The dorsal laminae of Pachyrhachis are primitive. However, in Pachyrhachis the supratemporal provides sole support of the quadrate, the quadrate has a reduced suprastapedial process and a well-developed distomedial stylohyal process, and the stapes is long and slender. Thus, in respect to the quadrate suspensorium and morphology, Pachyrhachis displays the condition in macrotomatan snakes above the level of Loxemus and Xenopeltis (Tchernov et al. 2000). In the absence of characters eliminating Pachyrhachis from Serpentes and in the presence of skull characteristics otherwise diagnosing the macrostomatan condition as noted above, better support is found for the hypothesis placing Pachyrhachis problematicus as a basal macrostomatan (sensu Tchernov et al. 2000).  Figure 10 illustrates our revised reconstruction of the skull of Pahyrhachis problematicus.