INTRODUCTION

The Cenomanian snake, Pachyrhachis problematicus, is known from two specimens from limestone quarries near 'Ein Yabrud, in the West Bank, north of Jerusalem. One specimen has a complete and articulated, albeit crushed, skull (Haas 1979, 1980a). In the second specimen, a disarticulated skull is crushed below the ribcage (Haas 1980b). Phylogenetic analysis by Caldwell and Lee (1997) places Pachyrhachis as the basal member or sister taxon of Serpentes. However, that phylogenetic hypothesis was tested and falsified by Zaher (1998), who concluded that Pachyrhachis was instead a basal macrostomatan allied with boids and pythonids. Disagreement regarding the phylogenetic position of Pachyrhachis persisted (Lee and Caldwell 1998; Caldwell 1999, 2000; Zaher and Rieppel 1999b, 2000; Rieppel and Zaher 2000a, 2000b), culminating in a seemingly intractable impasse (Rieppel and Kearney 2001). On the basis of a data set developed independently, Tchernov et al. (2000) found Pachyrhachis and a second taxon from ‘Ein Yabrud, Haasiophis, to be basal macrostomatans (see also Rieppel et al. 2003; Zaher and Rieppel 2002; Lee and Scanlon 2002a, 2002b). The same specimens and general observational techniques used by all workers led to conflicting interpretations of morphology. Cladistic analyses based on those interpretations subsequently yielded different phylogenetic hypotheses.

The source of differing interpretations of morphology derives from HUJ [The Hebrew University of Jerusalem]-PAL 3659. This specimen is difficult to study due to a number of taphonomic factors, further complicated by preparation techniques. George Haas (1979) of the Hebrew University of Jerusalem originally described Pachyrhachis using only HUJ-PAL 3659, which includes the crushed but beautifully articulated skull originally preserved on a limestone flag. After describing the dorsal surface of the skull (Haas 1979; see also Haas 1980a, 1980b), the specimen was embedded in resin and the ventral surface was acid prepared and described (Haas 1980a). Subsequent to Haas’s original description, the dorsal surface could only be viewed through the embedding resin. Additionally, the resin was covered with glass, which at some point was broken, further obscuring the view. The glass has since been removed, but the resin remains. The viability of optical techniques for examination of the morphology of HUJ-PAL 3659 is compromised due to taphonomic factors including crushing, breakage, overlapping elements, displacement during preservation, matrix residue from mechanical preparation on the dorsal surface of the skull, and to subsequent resin embedding of the dorsal surface for acid preparation of the ventral surface (Figure 1).

Figure 2 shows three renderings of the dorsal surface of the articulated skull of Pachyrhachis as taken from Haas (1979), Caldwell and Lee (1997), and Zaher and Rieppel (1999b). Given the complexity and history of HUJ-PAL 3659, it is not surprising that there are some differences in the interpretations as illustrated in Figure 2, differences that contributed to varying phylogenetic placement of Pachyrhachis. Although Haas recognized a number of diagnostic snake characters, he considered Pachyrhachis a highly derived platynotan or varanoid lizard. Caldwell and Lee (1997) recognized Pachyrhachis as a snake and the sister taxon to all other snakes, and that clade as the sister taxon to mosasauroids, resurrecting Cope’s (1869) concept of Pythonomorpha (see also Lee 1997a, 1997b; Lee and Caldwell 1998; Caldwell 2000). Zaher (1998) reinterpreted Pachyrhachis as an advanced snake, specifically a basal macrostomatan allied with boas and pythons (see also Zaher and Rieppel 1999a; Rieppel and Zaher 2000a, 2000b).

Snakes such as Pachyrhachis, Haasiophis, and Eupodophis, which possess all the bony elements of the hind limb (as opposed to the rudimentary limb and girdle elements of modern worm snakes, boas, and pythons), are primitive in that attribute. However, the issue of macrostomatan versus more primitive phylogenetic placement hinges in large part on conflicting morphological interpretations of skull elements (Rage and Escuillié 2000; Tchernov et al. 2000; Rieppel et al. 2003). The major differences among the three interpretations that are important here are: 1) the morphology and orientation of the quadrate; 2) the identification of the stapes or squamosal; 3) the number of mental foramina; 4) the identity of the bones in the circumorbital series; 5) the question of whether the exoccipitals contact above the foramen magnum; and 6) presence of a dorsal prefrontal process of the maxilla.

To test conflicting interpretations of morphology, the articulated skull of Pachyrhachis (HUJ-PAL 3659) was scanned using X-ray computed tomography (CT). Two- and three-dimensional reconstructions were computer generated to expose details from viewing perspectives that could not be attained using optical techniques. We also created a three-dimensional digital model of the skull utilizing specific measurements as boundaries. By removing distortion due to crushing, we were able to test the feasibility of the identification of the most problematic bones of the circumorbital series, the ectopterygoid, and the orientation and morphology of the quadrate. This model then provides a hypothesis of the appearance of the undistorted skull of Pachyrhachis. The value of digitally modeling morphology in this case is in providing a more precise identification of structures. It is fundamentally a taphonomic exercise to remove compaction, rotation, and distortion.

Agreement among investigators on primary observations of morphology is critical to the phylogenetic understanding of snakes with hind limbs. A morphological feature either exists, does not exist, or it remains ambiguous or unknown. We intend to elucidate and clarify what can or cannot be seen in the fossil using optical observation and CT derived digital data and provide the basis from which competing phylogenetic hypotheses can be tested and falsified. An independent phylogenetic analysis is beyond the scope of this study, which is intended only to test the data quality of previous primary observations and to discover new data.