SKULL

The skull of IGM 3/196 is dorsoventrally flattened with little compression. The outline of the skull features an elongate round snout, and laterally expanded orbits, which is seen in all monstersaurs except Gobiderma pulchrum. The premaxilla, maxillae, jugals, and septomaxillae are preserved in articulation. Posterior to the frontal, the skull roof and entire braincase are eroded away. The skull roofing bones have no fused osteoderms, but show shallow surface striations like those in the holotype (Norell et al., 1992). Most of the palatal bones are preserved, with only the posterior part of basisphenoid missing (fig. 3A).

PREMAXILLA: The premaxilla consists of a horizontal maxillary process and a nasal process rising about 30° from the level of the maxillary process. The maxillary process bears nine tooth positions as reported for the holotype (IGM 3/14, Norell et al., 1992). A premaxilla-maxilla aperture is not present. In ventral view, a short bipartite vomerine process is present in the middle of the maxillary process. The nasal process of the premaxilla is slender; posteriorly it tapers and inserts between the nasals. Two small projections are present at the base of the nasal process where it branches off from the maxillary process. The lateral projections abut the premaxilla-maxilla suture, forming two small openings (fig. 3B, dorsal view). The presence of the premaxilla projections is an autapomorphy for Estesia mongoliensis. Heloderma and Gobiderma pulchrum do not possess similar structures; the premaxilla is not preserved for Heloderma texana, Lowesaurus matthewi, and Eurheloderma gallicum. The function of these projections is unclear, but the size and position of the apertures formed by the projections is comparable to the ethmoidal foramina of extant Heloderma, Lanthanotus borneensis, Varanus, and Shinisaurus crocodilurus. In extant anguimorph lizards, similar lateral projections are known only in Diploglossus (Rieppel, 1980: fig. 18), but the projections are not as extensive as those in Estesia mongoliensis, and not in contact with the maxilla.

Major anguimorph clades demonstrate considerable variation in the contacts between the premaxilla and nasal. In the Monstersauria (and squamates in general), the premaxilla lies on the surface of the nasals with no additional ventral process extending beneath the nasals (fig. 4A, B, D, E). Varanus, however, demonstrates a derived condition where the premaxillary process of the nasal embraces the premaxilla along its lateral margins (fig. 4C). Correspondingly the nasal process of the premaxilla divides into dorsal and ventral branches to interlock with the nasal. This is also seen in some fossils allied to varanids (sensu Conrad et al., 2011a), e.g., Aiolosaurus oriens (Gao and Norell, 2000: fig. 34A). Mosasaurs show an additional hyperderived condition where the nasals are greatly reduced or lost (Russell, 1967). IGM 3/196 shows that Estesia mongoliensis has a horizontal articulation surface between the premaxilla and nasal, resembling the condition in Heloderma and other monstersaurs.

NASAL: The nasals are paired, thin, and elongate, contacting the premaxilla anteriorly and the frontal posteriorly. Another apomorphy of Estesia mongoliensis shown in IGM 3/196 is that the nasals taper anteriorly and suture beneath the premaxilla. In other monstersaurs with nasals preserved (e.g., Heloderma, Gobiderma pulchrum), the anterior tips of left and right nasals remain separate and do not completely suture along the midline. Posteriorly, the nasals contact the frontals in a straight suture, although details of this suture are obscured by erosion.

MAXILLA: The maxilla forms most of the lateral margin of the rostrum. It is roughly triangular in lateral view, with a long and low premaxillary process and a relatively short posterior process ending abruptly anterior to the orbit (fig. 2). The premaxillary process is dorsoventrally flat. In dorsal view, the premaxillary process curves medially to give the snout a blunt shape. The medial curving begins at the level of the fourth posterior tooth position. The premaxillary process notches anteriorly, articulating with the premaxilla and the septomaxilla (fig. 3A, B). The posterior (nasal) process gradually rises from the premaxillary process. In lateral view, the margin of the bony naris is weakly inclined posteriorly. Most monstersaurs, e.g., Heloderma and Eurheloderma gallicum, have steeply inclined nasal processes. In comparison, varanid lizards have weakly inclined nasal processes developed in conjunction with a posterior retraction of the nares. “Retracted nares” is a common character used in anguimorph matrices (Estes et al., 1988; Norell et al., 2007; Conrad et al., 2011b). In general, true retraction is thought to only appear in varanids. Correlated features that develop with naris retraction include weakly inclined lateral naris margins, a fused nasal that tapers posteriorly, and anterior processes of the frontal contributing to the naris margins. Nares of extant Heloderma extend more posteriorly than the primitive anguimorph condition, but are generally treated as not retracted, due to the extensive connective tissues between the nasal and prefrontal (see discussion in Estes et al., 1988: char. 2). Discoveries of fossil monstersaurs have revealed variation in the extent of narial posterior elongation (Hoffstetter, 1957; Norell et al., 1992; Conrad et al., 2011b), and so there is a call for a more precise definition of “retracted nares.” In this study, “retracted nares” is used when the posterior margins of the nares not only invade the space between the nasal and maxilla, but also incorporate the anterior margins of the frontal. Note that “retraction” cannot be assumed only by the weak inclination of the nasal process of the maxilla. In this context, Estesia mongoliensis has a varanidlike, low and posteriorly inclined, maxilla, but does not develop full naris retraction because the frontals do not contribute to the narial margin.

FIGURE 2.

Lateral view of IGM 3/196. Scale bar = 10 mm.

In ventral view, the maxilla contacts the premaxilla and vomer anteriorly. Posteriorly, it is excluded from the infraorbital fenestra by the palatine and ectopterygoid. Eleven tooth positions are observed on each maxilla.

SEPTOMAXILLA: The septomaxilla is subtriangular in dorsal view. It is concave medially with a low ridge, resembling the shape of the septomaxilla in other monstersaurs. In contrast, the septomaxilla of Varanus (fig. 4C) is convex in the middle. The septomaxilla of IGM 3/196 lies on top of the maxilla and vomer, and housed Jacobsons organ within it. Like in Heloderma and Gobiderma pulchrum (Conrad et al., 2011b), the low medial flange of the septomaxilla does not contact the premaxilla or nasal dorsally. Due to preservation, the septomaxilla is unknown in Eurheloderma gallicum, Primaderma nessovi, and Paraderma bogerti. The septomaxilla can be provisionally described in Heloderma texana (Pregill et al., 1986) with further preparation on reported materials; unfortunately, the only CT-scanned specimen (TMM 40635-230) seems to lack this element.

PREFRONTAL: The prefrontal has a dorsal process extending along the orbit anteroposteriorly, and a ventral process forming part of the anterior wall of the orbit. The dorsal processes on both sides are badly weathered. The ventral process of the prefrontal surrounds a single lacrimal foramen with the lacrimal and palatine. Like the holotype of Estesia mongoliensis (Norell et al., 1992: fig. 10A), the prefrontal, lacrimal, and jugal of IGM 3/196 meet at a single point on the ventral margin of lacrimal foramen. In anguimorph lizards, the plesiomorphic morphology lacks contact between the prefrontal and jugal. In derived monstersaurs (extant Heloderma), the jugal extensively contacts with the prefrontal (Norell et al., 1992; fig. 10C). Estesia mongoliensis demonstrates an intermediate morphology between the derived and primitive conditions.

LACRIMAL: The lacrimal is triangular in lateral view and forms the lateral margin of the orbit. IGM 3/196 has a single lacrimal foramen, surrounded by the lacrimal, prefrontal, and jugal. On the holotype of Estesia mongoliensis, the lacrimal foramen appears as a small slit due to lateral compression, but IGM 3/196 shows that the lacrimal foramen is clearly larger than the palatine foramen. The presence of a single posterior lacrimal foreman is plesiomorphic in the Anguimorpha and squamates in general (Estes, et al., 1988). In anguimorph lizards, only Lanthanotus borneensis and Varanus demonstrate a derived condition with double posterior lacrimal foramina. Heloderma and Gobiderma pulchrum (Conrad et al., 2011b) both have one posterior lacrimal foramen, and the same condition is found in Estesia mongoliensis.

JUGAL: The jugal is angular with a short suborbital process and a long tapering postorbital process. The suborbital process extends along the medial side of the maxilla, excluding the maxilla from the orbital rim. Dorsally, the postorbital process of the jugal is smooth and rodlike, and tapers towards the postorbitofrontal. No posterior process develops at the base of postorbital process. The postorbital process contacts the postorbitofrontal to form a complete orbital rim. In anguimorph lizards, a greatly reduced jugal (that does not contact the postorbitofrontal) is a synapomorphy of Varanus. A complete orbital rim is plesiomorphic and shared by major anguimorph clades and possibly some fossils allied to varanids (e.g., Saniwa ensidens). The medial side of the jugal bears a distinct fossa for muscle attachment. This fossa exists in most anguimorphs, generally projecting medially. In Estesia mongoliensis and extant Heloderma, this distinct fossa faces posteriorly, largely due to the unique lateral expansion of the skull posterior to the rostrum. No dermal rugosities are preserved on the jugal.

VOMER: Both vomers, elongate and rodlike, are preserved in IGM 3/196. The vomer is more than twice the length of the palatine, in keeping with elongation of the snout. Vomer elongation is also present in Heloderma, Varanus, Lanthanotus, and allied fossil taxa including Gobiderma (the Deep Scaly Project, 2011a; digital morphology accessed through  http://www.digimorph.org), Ovoo gurvel (Norell et al., 2007), and Paravaranus angustifrons (Borsuk-Bialynicka, 1984: pl. 1). All mosasaurs have an elongate snout, but the proportion between vomer and palatine remains unclear for many species. In other anguimorph clades, the vomer is shorter than the palatine.

In IGM 3/196, the vomer contacts the premaxilla and maxilla at its anterior tip, but has no contact with the maxilla along its lateral edge. In Varanus the vomer contacts the maxilla anterolaterally to form a complete floor for the chamber enclosing Jacobsons organ (Rieppel et al., 2008). This is termed the neochoanate condition in Lakjer (1927). In Heloderma and Lanthanotus borneensis the vomer does not contact the lingual process of the maxilla, forming the incomplete neochoanate condition (Lakjer, 1927). The holotype of Estesia mongoliensis appears to show a “neochoanate condition” (Norell et al., 1992: fig. 4), but the less distorted skull of IGM 3/196 indicates that the contact in the holotype is due to lateral deformation of the skull and Estesia mongoliensis actually has an incomplete neochoanate condition.

On the ventral surface of the anterior tip of the vomer are two lacrimal grooves (sensu Oelrich, 1956) surrounding a lacrimal duct. In anguimorph lizards, the plesiomorphic morphology is that two anteroposterior ridges border the lacrimal groove (fig. 5C). In extant Heloderma, the ridges are reduced to small knobs, and the lacrimal groove is not prominent (fig. 5B). In fossil monstersaurs, Estesia mongoliensis and Goboderma pulchrum retained the plesiomorphic condition (fig. 5A). The vomers in other fossil members remain unknown.

PALATINE: Both palatines are preserved in IGM 3/196; they are wider than long with the vomerine branch shorter than the maxillary branch. The palatine is posteriorly located; the vomeropalatine suture extends to the level of the posteriormost tooth position in the maxilla. In Heloderma the vomeropalatine suture extends posterior to the last maxillary tooth. In Varanus, the vomerine process of the palatine generally extends more forward compared to that in Estesia mongoliensis and Heloderma. No obvious tooth patch is preserved on the palatine of IGM 3/196. One single tooth can be observed on the right palatine near its suture with the pterygoid, but it is unclear whether the palatine teeth form a single row, multiple rows, or a patch. The holotype of Estesia mongoliensis seems to bear a tooth patch on the palatine (Norell et al., 1992), but weathering of its ventral surface precludes conclusive descriptions of this character.

FIGURE 3.

A. Skull of IGM 3/196 in dorsal and ventral view (scale bar = 10 mm). B. Line drawings of the skull of IGM 3/196 in dorsal and ventral view (scale bar =10 mm). Abbreviations are listed in appendix 6.

FIGURE 4.

Comparisons of snout morphologies in the Anguimorpha, with the septomaxilla highlighted in white. A. Estesia mongoliensis (IGM 3/196); B. Heloderma suspectant (AMNH R-147173); C. Varanus salvadorii (AMNH R-59873); D. Lanthanotus borneensis (the Deep Scaly Project, 2011b); E. Shinisaurus crocodilurus (Bever et al., 2005). Not to scale.

PTERYGOID: Both pterygoids are preserved on IGM 3/196, although most of their posterior processes (quadrate processes) are weathered away. The pterygoid is triradiate, with the palatine and ectopterygoid processes projecting anteriorly and the quadrate process projecting posteriorly. The palatine process elevates to form a flat platform dorsal to its contact with the ectopterygoid process. This platform is covered by a patch of teeth, resembling the pterygoid tooth patch in Gobiderma pulchrum. The presence of palatal teeth is variable among individuals in Heloderma suspectum, but when present, the teeth are arranged in a single row. Eurheloderma gallicum possess a single row of teeth on the pterygoid (Hoffstetter, 1957). The pterygoids of other fossil monstersaurs remains unknown. The dorsal surface of the pterygoid has a distinct groove running along the ectopterygoid process to the terminus of the quadrate process, as seen in Heloderma and Lanthanotus borneensis, but this groove is absent in Varanus.

ECTOPTERYGOID: The ectopterygoid consists of a short anterior process and two posterior processes. The anterior process contacts the palatine medially, excluding the maxilla from participation in the margin of the suborbital fenestra. The contact between the ectopterygoid and palatine is less extensive in Estesia mongoliensis compared to Heloderma. The ectopterygoid diverges posteriorly into dorsal and ventral processes, embracing the anterolateral process of the pterygoid. The ventral branch is short and sutures with the pterygoid nearly vertically. The dorsal branch resembles that of Heloderma in that it extends about two thirds of the length of the ectopterygoid process of the pterygoid, which is visible in lateral view (fig. 2). In dorsal view, the ectopterygoid fits into a distinct groove on the dorsal surface of the pterygoid that terminates anterior to the epipterygoid.

MANDIBLE

Both mandibular rami of IGM 3/196 are preserved. The left ramus is nearly complete, curving ventrally and medially. Articulating the left mandibular ramus to the skull shows that the dentary and coronoid have kept their original life position, whereas the postdentary elements have shifted anteriorly (fig. 6). The right dentary and coronoid are complete, but their medial surface is hidden by postdentary elements that shifted ventral to the dentary. Neither the splenial or angular is preserved on either side. The mandibular fossa of IGM 3/196 is deep and elongate in medial view. Extant Heloderma and Varanus have an elongate mandibular fossa; in Lanthanotus borneensis, the mandibular fossa is restricted and nearly covered by a ventral lappet of the surangular in the lingual surface.

FIGURE 5.

Comparisons of anterior palate of: (A) Estesia mongoliensis (IGM 3/196); (B) Heloderma suspectum (AMNH R-147173); and (C) Varanus salvadorii (AMNH R-59873). Not to scale.

DENTARY: The dentary bulges outward laterally. In dorsal view, the symphyseal facet for the opposing mandibular ramus is nearly vertical to the long axis of the rest of the dentary. The medial inflection of the dentary begins at the second tooth position anteriorly. Medially, Meckel's groove is open at least in the anterior half of the dentary, according to the position of the articulation facet for the surangular below the last dentary tooth. The opening of the intramandibular septum lies posterior to the last dentary tooth below the descending subdental shelf, or suprameckelian lip (Bhullar and Smith, 2008). The posterior margin of the intramandibular septum of the dentary is U-shaped and smooth, without the eminence reported in some Miocene helodermatids (Bhullar and Smith, 2008). Both dentaries bear eight tooth positions. The mandibular teeth are loosely arranged and modified pleurodont with basally infolded plicidentine. In lateral view, six alveolar foramina pierce the lateral wall of the dentary, with the last foramen lying posterior to the level of the last tooth. The dentary of Estesia mongoliensis shares with other monstersaurs the following characters: extensive height (measuring about half as long as the length) and a distinct medial inflection at the symphysis that gives the snout a round outline (not distinct in Gobiderma pulchrum). Compared to other monstersaurs, Estesia mongoliensis is unique in having a long edentulous posterior rim in the dentary. The tooth row occupies only about two thirds the length of the dentary (figs. 2, 7), which is much shorter than that in Heloderma and Gobiderma pulchrum (see IGM 3/55).

FIGURE 6.

The mandible of IGM 3/196 in dorsal and ventral views. Scale bar = 10 mm; abbreviations are listed in appendix 6.

CORONOID: The coronoid is fragmentary on the left side; the right coronoid is obscured by other overlying bones. The labial branch of the coronoid is short, overlapping the dorsal part of the dentary-surangular suture. The lingual process is long, as shown in the right coronoid. The angular process of the coronoid is not well preserved on either side, but shows a general shape similar to that of the holotype (Norell et al., 1992).

SURANGULAR: The surangular is exposed on the left mandibular ramus, although it shifted anteriorly postmortem. Consequently, presence of an anterior surangular foramen cannot be confirmed. A posterior surangular foramen is located near the surangular-prearticular suture. In lingual view, the surangular forms most of the dorsal margin of the mandibular fossa, whereas the angular forms the ventral margin.

PREARTICULAR: The prearticular extends from the posterior margin of the dentary to the posterior end of the mandible. In ventral view, a prominent triangular medial process of the prearticular coextends with the glenoid process of the mandible, as seen in Heloderma, Lanthanotus borneensis, and Varanus. The retro articular process is equal in both length and width, and is much shorter than in extant Heloderma. Piercing the medial surface of the retro articular process is a foramen that transmits the chondra tympani nerve in some extant lizards (Oelrich, 1956).

ARTICULAR: The articular is well ossified, forming the anterior wall of the glenoid fossa. Its exposed surface is rough, suggesting the attachment of cartilage in life. The purported loss of the articular in Estesia mongoliensis reported in previous studies (Norell et al., 1992; Norell and Gao, 1997) is probably due to incomplete ossification or preservation.

FIGURE 7.

Tooth specializations (grooves and serrations) in the Anguimorpha. A. Estesia mongoliensis (IGM 3/196), with arrows pointing to tooth grooves and flanges on the anterior and posterior carinae of a dentary tooth. B. Dentary tooth of Heloderma suspectum (AMNH R-71082). C. Dentary tooth of Tylosaurus proriger (AMNH FR 1543). D. Dentary tooth of Varanus komodoensis (AMNH R-37909). Not to scale.

DENTITION

Similar to Heloderma, Estesia mongoliensis has larger teeth lying in the middle of the tooth row. The premaxillary teeth are distinctly smaller than the maxillary teeth. All teeth are recurved. They are neither as elongate nor as cylindrical as those of extant Heloderma. IGM 3/196 reveals a new character shared by Estesia mongoliensis and extant Heloderma: distinct flanges (“trenchant margins delimited from main tooth shaft,” as mentioned in Pregill et al., 1986) on the rostral and caudal carinae of the dentary teeth, forming sharp cutting edges (fig. 7B). Such flanges are not known in other anguimorphs. When describing tooth morphology in this study, “rostral” refers to the direction toward the mandibular symphysis, whereas “caudal” (“distal” in Bhullar and Smith, 2008) refers to the direction toward the retroarticular process in the mandible. Extant Heloderma lizards bear the most extensive flanges that are expanded at the tooth tip. These flanges are not the same structures as the venom grooves. Although venom grooves in Heloderma develop associated with the flanges, the flanges can also be observed in taxa that do not bear venom grooves (Primaderma nessovi; Nydam, 2000). Estesia mongoliensis bears flanges and grooves in the marginal teeth, and these structures are most prominent in large teeth. IGM 3/196 has several teeth that clearly show flanges that delimit the main tooth shaft and extend from the tooth root to the tip (fig. 7A). Tooth flanges in Estesia mongoliensis are not as wide as those in extant Heloderma, indicating the primitive condition of this structure. Similar flanges are also observed in Palaeosaniwa canadensis (Pregill et al., 1986; Balsai, 2001) and UF 206579 (Bhullar and Smith, 2008). The presence of tooth flanges in other monstersaurs remains uncertain, although they are probably not present in Gobiderma pulchrum. Stevens (1977) reported venom grooves of Heloderma texana, but it is impossible to tell whether a flange develops with the groove because the reported teeth were broken and no crowns are preserved. In Hoffstetter's (1957) description of teeth in Eurheloderma gallicum, a flange develops on the rostral carinae, but the caudal flange is not mentioned.

Compared to the tooth flanges, venom grooves are better documented in the Monstersauria, and have been summarized in various studies (Pregill et al., 1996; Nydam, 2000). IGM 3/196 provides new information of the venom grooves of Estesia mongoliensis. On large dentary teeth of IGM 3/196, grooves extending from the tooth root to the tip can be observed on both the rostral and caudal carinae (fig. 7A), with the caudal one much deeper than the rostral. In extant Heloderma, however, the deep venom groove is present only on the rostral carina. The caudal groove is sometimes observed (especially in Heloderma horridum), but the caudal groove is always shallower than the rostral when present. All other monstersaurs with venom grooves have deep root-to-tip tooth grooves in the rostral carina. These taxa include Lowesaurus matthewi (Gilmore, 1928; Pregill et al., 1986), Eurheloderma gallicum (Hoffstetter, 1957), Heloderma texana (Stevens, 1977), Paraderma bogerti (Gao and Fox, 1996) and the unnamed specimen UF 206579 (Bhullar and Smith, 2008).

AXIAL SKELETON

VERTEBRAE: IGM 3/760 (fig. 8) preserves 10 dorsal vertebrae associated with ribs, and a fragmentary right forelimb. Although Estesia mongoliensis lacks diagnostic characters in the postcrania, the most obvious feature is the large size of the skeleton. IGM 3/760 is referred to Estesia mongoliensis for its large size and constricted articular condyles of the vertebrae resembling other reported postcrania of this species (Norell and Gao, 1997). The centra of the dorsal vertebrae are constricted posteriorly, and bear distinct articular condyles. The articulation surfaces between the vertebrae are oblique, so in lateral view the condyle of the proceeding vertebra lies below the articulation surface of the more posterior element. The parapophysis and hypopophysis are fused for articulation with single headed ribs. All neural arches are broken; however, the preserved bases indicate that the neural arches are narrow dorsally. Ribs are single headed, and slightly curved back.

FORELIMB: IGM 3/760 provides information on limb elements of Estesia mongoliensis for the first time. The manus of the right forelimb is preserved with the vertebral column. The carpals and metacarpals are well ossified, and are not particularly reduced or elongate in proportion to the body trunk, indicating that Estesia mongoliensis had limbs at least as powerful as extant monstersaurs or varanids. The suture between the carpals also indicates this individual is an adult, as these elements fuse at somatic maturity. The exact count of wrist elements cannot be determined due to preservation. Four carpals are preserved; they are the medial centrale and distal carpals 2–4 based on shape comparison to extant anguimorph lizards.

ESTESIA MONGOLIENSIS As A BASAL MONSTERSAUR

As mentioned before, previous studies report varying phylogenetic positions for Estesia mongoliensis (Norell et al., 1992; Norell and Gao, 1997; Conrad et al., 2011a; Yi and Norell, 2010), but these studies differ in taxon and character sampling. In a preliminary analysis (Yi and Norell, 2010), we suggested that Estesia mongoliensis is the most basal member of the Varaniformes (sensu Conrad, 2008), but our subsequent work shows that this was probably due to inadequate taxon sampling. In comparison to the matrix used in Yi and Norell (2010), the current matrix is updated in several aspects: first, we expanded our taxon sampling to genus level in the Anguimorpha, including broad sampling in the Mosasauria. Although Estesia mongoliensis never grouped with mosasaurs, adding mosasaurs in the matrix changed the character optimization of Varanus-like characters (e.g., the fusion of nasals). In the updated data matrix, Estesia mongoliensis “shifted back” to the Monstersauria, possibly due to the extra steps of reversals required if Estesia mongoliensis remained within the Varaniformes.

In addition to expanded taxon sampling, we added eight new morphological characters based on the observations in this study; we also updated the characters codings for Gobiderma pulchrum based on the recent work by Conrad et al. (2011b). Gobiderma pulchrum has been assigned to different positions in the anguimorph phylogeny by different authors. When first reported by Borsuk-Bialynicka (1984), Gobiderma pulchrum was considered as a “necrosaurian grade” anguimorph lizard, convergent with Helodermatidae. Pregill et al. (1986) argued that it should be more closely related to Xenosaurus. Conrad et al. (2011b) provided detailed descriptions on the morphology of Gobiderma pulchrum based on new specimens, supporting the monstersaurian affinity of this taxon. Updating character codings for Gobiderma pulchrum not only augmented information for this particular taxon, but also changed optimizations of “Heloderma-like” characters in the anguimorph tree. Finally, the current data matrix included molecular characters to further explore the possible phylogenetic positions for Estesia mongoliensis within a total evidence framework.

The morphology-only and combined analysis supported Estesia mongoliensis as a basal monstersaur. The strict consensus tree of the combined analysis recovered 34 unambiguous morphological synapomorphies for the Monstersauria (appendix 5).

INGROUP RELATIONSHIPS OF THE MONSTERSAURIA

The strict consensus trees of the morphology-only (fig. 10) and combined analysis (fig. 9) both support Heloderma texana as the sister group to the two extant Heloderma species. The morphological tree further recovers a monophyletic group consisting of Eurheloderma gallicum, Lowesaurus matthewi, Paraderma bogerti, and UF 206579. The ingroup relationships of the Monstersauria, shown in both strict consensus trees, suggest a complex biogeographic history of the group. First, in the Late Cretaceous, the two Asian species (Gobiderma pulchrum and Estesia mongoliensis) do not form a monophyletic clade; neither do the three North American species (Primaderma nessovi, Paraderma bogerti, and Palaeosaniwa canadensis). Second, among the Cenozoic North American taxa, only Heloderma texana is closely related to the extant species; the other two taxa (Lowesaurus matthewi and UF 206579) are probably more closely related to the European taxon Eurheloderma gallicum and the Late Cretaceous North American taxon Paraderma bogerti, according to the morphology-only analysis. Inferring from the morphological consensus tree, extant monstersaurs are most closely related to North America fossil taxa, and the European taxon Eurheloderma gallicum likely dispersed from North America. Tracing back the evolutionary history to the Mesozoic, the geographic region for the origin of the Monstersauria remains unclear due to the uncertainty in the phylogenetic relationships between basal taxa. However, monstersaurs clearly had a cosmopolitan distribution across Laurasia in the Cretaceous. The available fossil record shows that Asian monstersaurs became extinct after the Cretaceous, and the current phylogeny indicates no dispersal event out of Asia.

CHARACTER EVOLUTION OF VENOM-RELATED TOOTH SPECIALIZATIONS IN THE ANGUIMORPHA

Lizards are traditionally considered as nonvenomous except for extant Heloderma. Discoveries of fossil monstersaurs reveal that the origin of the lizard venom system possibly dates back to the Cretaceous (Norell et al., 1992; Norell and Gao, 1997; Balsai, 2001). Recent biochemical studies (Fry et al., 2006, 2012) show intriguing results that Varanus and some iguanids share several toxin proteins with snakes. This suggests that the evolution of a venom system in lizards has a longer and more complex evolution history than previously considered, and fossils are essential for illustrating this history. Since venom glands and toxin proteins are rarely preserved in fossils, morphological characters of venom delivery (tooth characters), when used with caution, are a reliable source of data for inferring whether a fossil was venomous. We suggest the incorporation of phylogenetic information when inferring venom use in fossils. A morphological character can be used as an indicator for venom use in fossils when all the following criteria are met: (1) the fossil and extant species are included in a monophyletic group; (2) the venom-related character is a synapomorphy of the monophyletic group including the fossil and extant taxa; (3) in the monophyletic group, the venom-related morphology is associated with venom use in all extant taxa that are venomous.

Two tooth characters have been used to indicate venom use in anguimorph lizards: deep root-to-tip venom grooves and serrations on the tooth carina. In the Anguimorpha, venom grooves are present only in monstersaurs and are connected with the specialized venom gland in extant species. Serrated teeth have recently been suggested as an aid for venom delivery in Varanus komodoensis, and are used as a morphological evidence for venom use in the fossil Varanus priscus, the largest known terrestrial lizard (Fry et al., 2012). However, the distribution of tooth serrations in the Anguimorpha is yet to be fully sampled. We included the above two characters in our analyses of anguimorph phylogeny (fig. 11). Optimization of the two characters on the strict consensus tree of the combined analysis shows that base-to-tip groove originated once in the Monstersauria, whereas tooth serrations originated twice, independently in Varanus and the Mosasauria.

FIGURE 11.

Character evolution of tooth specializations in the Anguimorpha. The labels on the branches indicate the origination of derived states as optimized on the tree (ACCTRAN optimization). Clades in black are those with specializations (derived states) on the marginal teeth. Clades in gray are those that show the plesiomorphic state of the corresponding characters: Character 429 — root-to-tip grooves in marginal teeth: (0) absent; (1) present, at anterior carina; (2) present, at anterior and posterior carinae. Character 430 — serrations on marginal teeth: (0) absent; (1) present.

In the anguimorph phylogeny (fig. 11), the Monstersauria is monophyletic and Estesia mongoliensis falls within a clade including Heloderma. Root-to-tip tooth grooves are clearly used for venom delivery in extant Heloderma. As a result, we propose that Estesia mongoliensis has a venom system comparable to extant taxa.

Presence of serrated teeth in Varanus is a weakly supported indication for venom use in fossil taxa. Fry et al. (2006) suggested the Komodo dragon (Varanus komodoensis) is venomous, with oral glands secreting “venom proteins” that are also found in some snakes. A following paper by Fry et al. (2009) tied the shearing effect of tooth serrations (fig. 7D) to venom distribution during predation in Varanus komodoensis, and further suggested the Pleistocene fossil Varanus priscus (with serrated teeth) was venomous. It is noteworthy, however, although Varanus priscus and Varanus komodoensis both fall within a Varanus subclade in our anguimorph phylogeny (fig. 11), not every extant species of this clade bears serrated teeth. Varanus gouldii is a medium-sized carnivorous lizard lacking serrated teeth, as does Varanus exanthematicus. The genus Varanus has more than 80 extant species, only four of which are documented to have protein-secreting oral glands (“venom glands” of Fry et al., 2012). Whether protein-secreting oral glands are present in the Varanus species without serrated teeth is yet to be studied. In addition, the anguimorph phylogeny (fig. 11) shows that serrated teeth originated twice in the Anguimorpha, once in Varanus and once in the Mosasauria (fig. 7C). The Mosasauria is an extinct anguimorph lineage with no other morphological indication for venom use. Finally, serrated teeth are widely distributed in nonvenomous reptiles, especially among large predators (crocodylians and dinosaurs), casting doubt on the direct relationship between serrated teeth and venom use. In summary, the information on the distribution of serrated tooth and protein-secreting oral glands in the Squamata is currently so patchy that using serrated teeth to indicate venom use in squamates remains tentative at best.