Tupinambis

Harvey, Michael B., Ugueto, Gabriel N. & Gutberlet, Ronald L., 2012, 3459, Zootaxa 3459, pp. 1-156 : 69-71

Polyphyly of Tupinambis and Status of Salvator

Systematics of the tegus has received considerable interest in recent years. Four species recognized by earlier authors ( Peters & Donoso-Barros 1970: Tupinambis duseni , T. nigropunctatus , T. rufescens , and T. teguixin ) were reduced to two by Presch (1973) who reviewed older synonyms and recognized only Tupinambis rufescens and T. teguixin as valid. Although Presch (1973) referred most specimens of T. merianae to T. rufescens , he nonetheless placed Duméril and Bibron’s (1839) name in the synonymy of T. teguixin . Avila-Pires (1995) rejected Presch’s (1973) results, recognized the four species listed by Peters and Donoso-Barros (1970), and redescribed T. merianae . Her ( Avila-Pires 1995) careful review of older names, detailed morphological comparisons, and designation of various lectotypes resolved many persisting problems with the alpha taxonomy of Tupinambis .

After 1995, several new species were described: Tupinambis longilineus ( Avila-Pires 1995) , T. palustris ( Manzani & Abe 2002) , and T. quadrilineatus ( Manzani & Abe 1997) . Almost concurrently, Colli et al. (1998) described Tupinambis quadrilineatus under the junior synonym T. cerradensis . Although one study (Fitzgerald et al. 1991) found relatively little divergence between T. duseni and T. rufescens among mitochondrial DNA sequences, recent multivariate analysis of mensural and meristic data (Fitzgerald et al. 1991; Péres & Colli 2004) identified a suite of characters that effectively distinguish these superficially similar species.

Fitzgerald et al. (1991) identified two deeply divergent clades within Tupinambis : a “southern clade” including T. duseni , T. merianae , and T. rufescens and a “northern/Amazonian clade” of the remaining species. These authors did not explicitly test Tupinambis monophyly but used three species of Teiinae ( Ameiva ameiva , Cnemidophorus ocellifer , and Kentropyx viridistriga ) to root the tree. Noting the deep split within Tupinambis, Fitzgerald et al. (1991) acknowledged that a broader study including other macroteiid genera would be required to test its monophyly.

We show that Tupinambis is polyphyletic. The “southern clade” of Fitzgerald et al. (1991) is evidently closely related to Dracaena , whereas the northern/Amazonian clade is related to Crocodilurus . To resolve this problem of polyphyly, we here resurrect Salvator Duméril and Bibron for the southern clade.

Unlike Crocodilurus and Tupinambis sensu stricto (characters in parentheses), Dracaena and Salvator share character 5.0 round pupils (reniform); 26.2 two loreals (one); 28.3 a scale inserted between the first subocular, supralabials and posterior loreal (first subocular contacting supralabials or rarely separated from them by anterior expansion of second subocular); 32.1 a complete row of lateral supraocular granules (lateral supraocular granules absent); a high number of supraciliaries (23–28 vs. 14–19); 48.2 chinshields usually separated from infralabials by row of granular sublabials (usually in contact); 54.1/54.0 intertympanic sulcus incomplete medially or absent (complete); 68 a high number of longitudinal rows of ventrals (30–39 vs. 20–28); 83.1 keeled proximal subcaudals (smooth); and 108.0 macrohoneycomb on dorsals (loss of macrohoneycomb from dorsals and evolution of long aligned macroridges on supracaudal scales). Salvator and Dracaena are larger lizards, exceeding 400 mm SVL (usually less than 350 mm SVL) with strongly molariform teeth (at least in adults).

Some obvious differences in coloration also distinguish Salvator from Tupinambis . In Salvator , the dorsolateral stripe starts on the posterior supraciliary and extends to a paravertebral position on the body where it is mostly broken. In contrast, the dorsolateral stripe of Tupinambis starts at the same location on the supraciliary but remains in the dorsolateral position. As in Salvator , the stripe is difficult to see and broken in most species, but it is distinct in T. quadrilineatus . In Salvator , the upper lateral stripe is distinct, starts below the eye, and extends (often broken) to a point above the leg as in Kentropyx . In contrast, the upper lateral stripe of Tupinambis is very indistinct or absent, and it is not present on the flanks except in T. quadrilineatus . Nonetheless, even in T. quadrilineatus the stripe is broken and extends to the groin rather than above the leg as it does in Salvator . Finally, juvenile Salvator have white spots on the thighs and a bright green ( S. merianae ) or somewhat olivaceous ( S. rufescens ) color on the head, whereas juvenile Tupinambis lack white spots, basically resemble adults, and do not exhibit an ontogenetic change in dorsal head coloration.

Our conclusions regarding the relationships of Crocodilurus , Dracaena , Salvator , and Tupinambis are at odds with some other phylogenetic analyses ( Presch 1974a; Sullivan & Estes 1997; Teixeira 2003). Presch’s (1974a) analysis was phenetic, but Sullivan and Estes (1997) included his characters in a more modern analysis and found Crocodilurus to be sister to a clade containing Dracaena and the extinct genus Paradracaena . Tupinambis was placed outside this clade as sister to these three genera. But Crocodilurus and Dracaena shared only two synapomorphies: a robust quadrate and fewer than sixty caudal vertebrae. On the other hand, Sullivan and Estes’s character 5 (biconodont or triconodont premaxillary teeth present) contradicted this arrangement, evolving in Tupinambis and Crocodilurus under their hypothesis. Sullivan and Estes (1997, p. 111) state that a robust quadrate is “also present in Callopistes ,” although they did not map this character on the branch leading to Callopistes in their figure 7.3. A problem with the osteological studies of Presch (1974a) and Sullivan and Estes (1997) is that some specimens (e.g., BMNH 1853.3.7.27 and SDSNH 65496 and 66269) lack locality information and may be specimens of Salvator rather than Tupinambis . Nonetheless, SDSNH 64932 from Guyana and BMNH 1964.1825 from Trinidad or Tobago can only be T. teguixin .

Teixeira (2003) added high infralabial scale (her character 123) and fourth toe lamellae counts (her character 126) to the list of characters shared by Dracaena and Crocodilurus . However, these suggestions contradict the data presented in her Table 2 where Crocodilurus has the lowest mean fourth toe lamellae (29.68 ± 1.94) of all Tupinambinae (more than 30 in the other genera) and infralabial counts (12.00 ± 0.55) comparable to Tupinambis (11.80 ± 1.57) rather than Dracaena (20.43 ± 2.40). Data in her Table 2 reflects the same pattern that we observed. Uncertainty regarding the alpha taxonomy of tegus affected other early studies. Gorman (1970) described three distinctive karyotypes within the Tupinambinae . He reports that Dracaena and Tupinambis have an acrocentric chromosome 6 (vs. submetacentric in the other Tupinambinae ) that is intermediate in size between macro and microchromosomes (vs. large in Crocodilurus ). He did not provide catalogue numbers for his material. Preliminary work was performed on Tupinambis teguixin from Trinidad, however he also studied “ Tupinambis nigropunctatus from Brazil.” Unless this last specimen can be located, it is impossible to know if it is Tupinambis or Salvator ; this name has certainly been applied to S. merianae in the past. Examining trigeminal muscles, Rieppel (1980) reached similar conclusions, finding more similarities between Dracaena and Tupinambis . For example, these genera share (1) a broad based bodenaponeurosis (narrow-based in Callopistes , Crocodilurus , and the Teiinae ) attached to the caudomesial edge of the coronoid process and the dorsomesial edge of the surangular (vs. attached to coronoid process only), and (2) the pseudotemporalis profundus muscle spreading posteriorly below the mandibular ramus of the trigeminal nerve (vs. the muscle located entirely in front of the mandibular ramus in Callopistes , Crocodilurus , and Teiinae ). Rieppel (1980) evidently examined two species of tegus (identified as Tupinambis nigropunctatus and T. teguixin ) and reported no interspecific differences between them.

Some of our own characters contradict a close relationship between Dracaena and Salvator . The various species of Tupinambis and Salvator share 24.4/24.5 the nostril posterior or somewhat posterior to the nasal suture (slightly anterior to the suture in Dracaena , centered in Crocodilurus ), 31.2 fourth supraocular larger than first (smaller than), and 110.1 paired scale organs on supracaudals positioned to the sides of the keel (organs single and positioned atop keel). Finally, Crocodilurus and Dracaena are both aquatic genera and share several adaptations for swimming. Most notably, their tails are flattened and possess dorsolateral crests of high, heavily keeled scales. Reduction in the number of caudal vertebrae ( Sullivan & Estes 1997) also may be associated with swimming.

A posterior position of the nostril relative to the narial suture also occurs in Callopistes and is likely to be a symplesiomorphy of Salvator and Tupinambis , whereas character 31 is difficult to interpret since it is not applicable to Callopistes . Under our hypothesis, the aquatic genera would have evolved independently from relatively unspecialized terrestrial macroteiids like Callopistes , Salvator , and Tupinambis . Aquatic behavior would impose a strong selective pressure for convergent evolution of flattened tails with paired dorsolateral crests, reduced numbers of caudal vertebrae, and movement of lenticular scale organs to the tops of high keels. In these characters, the tails of aquatic macroteiids resemble tails of crocodilians and other aquatic lizards such as the gymnophthalmid Neusticurus , scincid Tropidophorus , and shinisaurid Shinisaurus . In the context of this discussion, some possible adaptations for aquatic habits have evolved in one of the aquatic macroteiids, but not the other. Like the other aquatic lizards mentioned above, Dracaena has heterogeneous dorsals, whereas the dorsals of Crocodilurus are homogeneous. Moreover, Dracaena lacks the long fifth toe of Crocodilurus , and Crocodilurus lacks the keeled ventrals of Dracaena . Both species have accessory dorsolateral crests at the base of the tail, although the arrangement of enlarged scales is different.

In summary, we recognize Salvator and Tupinambis as two highly divergent and clearly monophyletic genera. Our conclusions regarding relationships among the genera of Tupinambinae deserve further study. Limited surveys of osteology, chromosomes, and trigeminal jaw adductors suggest different relationships, but were hampered by the previously confused alpha taxonomy of tegus. Nonetheless, these studies identified potentially informative characters that should be assessed for more species of Salvator and Tupinambis .