A Phylogenetic Analysis of Lepidosauromorpha Jacques Gauthier, Richard Estes, and Kevin de Queiroz INTRODUCTION OBJECTIVES OF THIS STUDY In this paper our primary objective is to analyze phylogenetic relationships within a subtaxon of Sauna, the Lepidosauromorpha (see Addendum, p. 94). Lepidosauromorpha includes the follow- ing saiunan diapsids: younginiforms, Paliguana whitei*, Palaeagama vielhaueri*, Saurosternon bai- nii*, kuehneosaurs, rhynchocephalians, and squamates, all of which are defined and diagnosed be- low according to our usage. For discussion of the use of the "*," see below. The principal prob- lem in discussing the taxa listed above is that the constitution of Rhynchocephalia and Squamata has not been stable, and that the early archosauromorphs and lepidosauromorphs have been lumped in paraphyletic "Eosuchia." As part of our analysis of lepidosauromorphs, we attempt to stabilize the concept of each lepidosauromorph taxon with as rigorous a diagnosis as the natiu-e of the mate- rial allows. Our analysis provides a relevant series of outgroups for evaluation of character polari- ty for phylogenetic analysis within Squamata (Estes et al., 1988). In order to give our phylogenetic analysis an explicit basis, we first constructed a minimum step cladogram by hand, using the character set given in Appendix I. The polarities are justified in Sections 2 and 3 by outgroup comparison. We then subjected the data matrix in Appendix II to analysis by the PHYSYS program written by J. S. Farris and installed in the California State Uni- versity CYBER system. The relatively minor differences between the hand cladogram and that of the computer are analyzed in Appendix III. We have tried to be as explicit as possible in our char- acter descriptions and to give as much discussion of variation as is feasible. Nevertheless, in a pa- per of this size, we have not been able to discuss the homoplasy in various taxa required by our conclusions in as much detail as we would have preferred. MONOPHYLETIC TAXA AND METATAXA As phylogenetic systematists our principal objective is to identify monophyletic taxa in the sense of Hennig (1966). Wiley (1981:2(X)) offered a set of criteria for phylogenetic taxonomy, and made the monophyly criterion his Rule 1. A monophyletic taxon is composed of an ancestor and all its descendants at any hierarchical level, from the least inclusive groups of organisms that share an exclusive common ancestry, to the taxon that includes all life. Such taxa are natural, in the sense that they are historical groups sensu Wiley (1981) or complete systems of common ancestry (de Queiroz, 1988). Monophyletic taxa are portions of life that have unique histories concerning such properties as their origin, diversification, and extinction. Wiley (1981:200) proposed that non-monophyletic groups could be included in a phylogenetic taxonomy "if they are clearly qualified as such." We reject the inclusion of known paraphyletic or polyphyletic taxa because their histories are largely a matter of definition. For example, the origin of the "Eosuchia" is the same as the origin of the Sauria as constituted here, and the diversity and temporal duration of the "Eosuchia" are determined less by evolutionary processes than by the point at which we arbitrarily separate such groups as squamates and archosaurs from the "eosuchians." In his sixth convention, Wiley (1981:213) identified as a problem area the treatment of known para- or polyphyletic groups. In our view, known para- or polyphyletic groups should be abandoned, and should have no place in a phylogenetic taxonomy. We recognize, however, that known paraphyletic groups may need to be mentioned in treatments of the history of taxonomy of 16 Phylogenetic Relationships of the Lizard Families certain taxa. For such cases, we recommend that Wiley's sixth convention be modified to limit use of quotation marks to known paraphyletic or polyphyletic taxa, because quotation marks are widely used in this context by many systematists ("Eosuchia" as discussed in this paper, is an ex- ample). We permit only one exception to the monophyly convention, the metataxon (Greek, meta, near), for taxa for which there is no character evidence supporting either monophyly or paraphyly. Metataxa that have previously been given formal names may be provisionally accepted in a phylogenetic taxonomy, but they must be abandoned if additional characters demonstrate non- monophyly. Because we have used quotation marks to denote known paraphyletic groups we have chosen the asterisk (*) to identify metataxa (e.g., Iguanidae*). The asterisk calls attention to the uncertain status of such taxa and distinguishes them from taxa characterized by synapomorphy as well as from known para- or polyphyletic groups. Metataxa are thus temporarily given the benefit of the doubt, but are formally accepted only until such time as characters are found that subdivide or clarify the relationships of the included organisms. We emphasize, however, that the recogni- tion of subdivisible unresolved groups is arbitrary and that in no case should such a group be new- ly named; the metataxon concept should only be used for unresolved taxa that have already been given formal names. Metaspecies (Donoghue, 1985) are metataxa at the least inclusive level, and are identified in the same way as any other metataxon, by the asterisk. MONOPHYLY OF DIAPSIDA Romer (1956), like Williston (1925) before him, questioned the monophyly of Diapsida (originally named by Osbom, 1903) because he was uncertain about the origin of the open lower temporal region of squamates; was it the result of emargination from below, or of fenestration and subsequent loss of the lower temporal bar? Nevertheless, Romer (1956) was influenced by workers who presented considerable evidence for the second alternative (e. g. Broom, 1925) and thus includ- ed squamates with all non-archosaurian diapsids in the Lepidosauria. The monophyly of Diapsida is now widely accepted (Reisz, 1977, 1981; Gaffney, 1980; Carroll, 1982). Gauthier (1984) pro- vided additional documentation of this monophyly, and divided diapsids into Araeoscelidia and Sau- ria; the latter taxon was redefined in the spirit of its original definition by McCartney (1802) to in- clude all Archosauromorpha (Huene, 1946, 1948, 1956; n. comb.) and Lepidosauromorpha, rather than applying it to the paraphyletic group "lizards." The Archosauromorpha includes Archosauria, a taxon restricted to the most recent common ancestor of birds and crocodiles and all its descendants (both living and fossil), as well as the following successively more remote outgroups of Archosau- ria (s. s.): Proterochampsidae, Erythrosuchidae, Proterosuchidae, and the even more remote Proto- rosauria, Champsosauridae (=Choristodera), Trilophosaurus, and Rhynchosauria. Although the precise relationships of the last four taxa within the archosauromorphs are not fully resolved, it is clear that they share some, if not all, of the synapomorphies that distinguish archosaurs from an- cestral lepidosaurs, among which are the following: 1. Enlarged premaxilla forming most of tip of snout. 2. Premaxilla with subnarial process extending dorsally up narial margin of maxilla to contact nasal, thereby excluding maxilla and septomaxilla from margin of external naris. 3. Parietal foramen small or absent. 4. Ventral process of squamosal half or less of height of lower temporal fenestra. 5. L-shaped quadratojugal. 6. Posterior process of jugal extends posterior to center of lower temporal fenestra. 7. Broad contact between ectopterygoid and jugal. LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier el al. 17 8. Bowed posterior margin of skull for passage of middle ear and support of tympanum, and a gracile, imperforate, stapes. 9. Tabulars absent. 10. Enlarged adductor chamber for temporal musculature, as indicated by tall quadrate that extends well below occipital condyle. 11. Parasphenoidal teeth absent. 12. Non-notochordal vertebrae in adults. 13. Transverse processes on trunk vertebrae moderately prominent. 14. Cleithrum absent. 15. Entepicondylar foramen in hum?rus absent. 16. Medial centrale in manus absent. 17. Complex concavo-convex astragalo-calcanear articulation, and a small, laterally direct- ed, calcanear tubercle (the archosauromorph duplex ankle joint sensu Thulbom, 1980). 18. Pedal centrale displaced laterally. 19. Hooked fifth metatarsal, and fifth distal tarsal fails to separate from anl?ge of fifth digit Contrary to long-standing opinion, rhynchosaurs, Trilophosaurus, champsosaurs, and proto- rosaurs are archosauromorphs, and have no close relationship to lepidosaurs (Gow, 1975; Carroll, 1977), thus demonstrating the paraphyly of "Eosuchia." It is possible that coelurosauravids (as constituted by Evans, 1982), thalattosaurs (sensu Romer, 1956; Merriam, 1905; Kuhn-Schnyder, 1952), plesiosaurs (Carroll, 1981), and ichthyosaurs (Tarsitano, 1983) are diapsids. However, coe- lurosauravids are too poorly known, while thalattosaurs and plesiosaurs, and to an even greater ex- tent ichthyosaurs, are too modified to contribute much to the resolution of the relationships among the basic taxa of this analysis. Indeed, it would be more appropriate to use the evidence presented here to test the possible diapsid affinities of these taxa. LEPBDOSAUROMORPHS: DIAGNOSES OF LEPIDOSAURS AND THEIR EXTINCT RELATIVES To facilitate discussion, we accept certain lower-level hypotheses that arise from preliminary analyses of the phylogenetic relationships within younginiforms (Currie, 1982), rhynchocephali- ans (see below), and squamates (Estes et al., 1988). In the discussion that follows, we emphasize that when we use the terms reptiles, diapsids, saurians, archosauromorphs, archosaurs, lepidosauro- morphs, younginiforms, lepidosauriforms, lepidosaurs, kuehneosaurs, rhynchocephalians, and squamates, we refer only to our own concept of these groups (see definitions and diagnoses below and the section Phylogenetic Taxonomy of Amniota); these may or may not be equivalent to con- cepts of these groups in other studies, but we have followed historical precedent whenever it is consistent with the results of our analysis. The younginiforms Younginiformes as constituted by Currie, 1982. Upper Permian to Lower Triassic. For many years, the concept of the "Eosuchia" has been that group of "primitive lepidosauri- ans, lacking the specializations of the more progressive members of the subclass" (Romer, 1956:519), or a "mixed bag of diapsid genera sharing only primitive characters" that is ancestral to both archosaurs and lepidosaurs (Evans, 1980:255). Broom (1914:1077) originally defined "Eosuchia" as a suborder of "thecodont reptiles" that retain postparietal and tabular bones, and a non-fenestrated antorbital region of the skull. Few have given further consideration to Broom's 18 Phylogenetic Relationships of the Lizard Families claim that Youngina is a "thecodont" (but see Gow, 1975) On the contrary, the persistent concept of the relationship of "eosuchians" to other diapsids is better expressed by Broom's (1914:1076) statement that "Youngina represents a type more primitive than any previously known, and one which is especially important in that it is very near to the ancestral form." To be included in the "Eosuchia," a species must possess the synapomorphies of the diapsids but lack those of archo- saurs and squamates. For the most part, the artificiality of this group has not escaped notice by later authors, and we underscore it here. Broom erected "Eosuchia" for Youngina alone, and this name could be applied to a monophy- letic taxon composed of Youngina and its close relatives. Nevertheless, "Eosuchia" in current us- age is paraphyletic, and we prefer the name Younginiformes of Romer (1945), as constituted by Currie (1982). We consider Younginiformes to include only Youngina capensis, Acerosodontosau- rus piveieaui, and the tangasaurs (Camp, 1945; including Tangasaurus mennelli, Hovasaurus bou- lei, Thadeosaurus colcanapi, and Kenyasaurus mariakaniensis as constituted by Currie, 1982). So delimited, the taxon is monophyletic. Diagnosis: Based primarily upon the works of Gow (1975) and Currie (1980, 1981a, 1981b, 1982), members of Younginiformes share the following combination of synapomorphies that dis- tinguishes them from all other Diapsida. 1. Relatively elongate and narrow snout (Fig. IB). 2. Reduction of postfrontal process of parietal and development of a posterior process on postfrontal that separates postorbital from parietal (Fig. IB). 3. Postorbital extends posterior to the end of the upper temporal fenestra (Fig. IB). 4. Loss of parasphenoidal teeth (Fig. 2A). 5. Olecranon process and sigmoidal notch of ulna poorly developed in adults (Currie, 1982). Currie (1980) also chaxiic\e.nzea Acerosodontosaurus piveteaui but did not determine the level of synapomorphy of the characters he discussed. At least two of the characters appear to be synap- omorphies of this taxon: the great ventromedial-dorsolateral width of the pubis and the twisted ap- pearance of the radius. Youngina has a uniquely modified iliac blade and a single row of middorsal osteoderms (also reported in ??eleosaurus; this combination of apomorphies has not been reported in any other diapsid (Gow, 1975). Thus, neither Acerosodontosaurus nor Youngina is likely to be ancestral to tangasaurs (Tangasauridae of Currie, 1982). The name Younginidae Broom (1914) is redundant because it says no more about the known pattern of synapomorphy than does the name Youngina capensis. Compared \.o Acerosodontosaurus, the skull of Youngina and the tangasaurs is relatively nar- rower across the anterior ends of the frontals (Currie, 1980); these two latter taxa also share spe- cialized articulations between the neural arches of the trunk vertebrae (Currie, 1981a), a strongly developed entepicondyle on the hum?rus in adults, and a radius that is longer than the shaft of the ulna (Currie, 1982). Accordingly, these taxa constitute the younginoids (= Younginoidea of Currie, 1982). Within younginoids, Currie (1982) has argued that tangasaurs may be distinguished from Youngina by the possession of the following synapomorphies: 1. Scapula low in lateral aspect, mainly a ventral element, and subequal to the coracoid in size. 2. Radius 50% - 65% of hum?rus length and 65% - 75% of tibia length in adults. 3. Fifth distal tarsal not present as a discrete element (if single specimens oi Kenyasau- rus and Hovasaurus have been correctly interpreted by, respectively, Harris and Car- FIGURE 1. Dorsal view of skull. A, Petrolacosaurus kansensis (araeoscelidan); B, Youngina ca- pensis (younginiform); C, Kuehneosaurus latus (kuehneosaur); D, Gephyrosaurus bridensis (rhynchoccphalian); E, Clevosaurus hudsoni (clevosaur); F, Planocephalosaurus robinsonae (clevosaur); G, Sapheosaurus thioUierei (sapheosaur); H, llomoeosaurus maximiliani (homoeosaur); I, Sphenodon punctaius (sphenodont); J, Sceloporus grammicus (iguanian); K, Elgaria mult tear ?nata (autarchoglossan). A after Reisz (1981); B after Carroli (1977); C, D, I after Evans (1980); E after Ro- binson (1973); F after Fraser (1982); G, H after Cocude-Michel (1963); J after Larsen and Tanner (1974); K after Rieppel (1980). FIGURE 2. Ventral view of skull. A, Youngina capensis (younginiform); B, Kuehneosaurus ?a- tus (kuehncosaur); C, Gephyrosaurus bridensis (rhynchocephalian); D, Planocephalosaurus robinsonae (clevosaur); E, Kallimodon cerinensis (sapheosaur); F, Sphenodon punctalus (sphenodont); G, Scel- oporus grammicus (iguanian); H, Elgaria muliicarinata (autarchoglossan). A after Carroll (1977); B, C, F after Evans (1980); D after Fraser (1982); E after Cocude-Michel (1963); G after Larsen and Tan- ner (1974); H after Rieppel (1980). LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier el al. 21 roll, 1977, and Currie, 1981b, then the fifth distal tarsal fuses to the fourth distal tar- sal in adults). 4. Hum?rus as long or longer than femur in adults. 5. Medial centrale contacts the fourth distal carpal, thus preventing the lateral centrale from contacting the third distal carpal (the available Youngina specimens are imma- ture, and Acerosondontosaurus is only partly preserved, so some of the characters Ust- ed above may later be found to specify more inclusive taxa within younginiforms). Fig. 11 is a cladogram depicting the phylogenetic relationships within younginiforms; the re- lationship of younginiforms within lepidosauromorphs appears in Fig. 13. Palaeagama vielhaueri* Broom (1926) Lower Triassic. Diagnosis: The unique type specimen is a fairly complete but very poorly preserved skull and postcranial skeleton. Based on the descriptions of Carroll (1975a, 1977), we are unable to identify synapomorphies for this taxon, and it is not known to possess any synapomorphies of taxa less inclusive than Lepidosauromorpha. Carroll grouped this taxon with Saurosternon* and Paliguana* (see below) but our analysis suggests that this is an overinterpretation. Palaeagama* is discussed below, following the section on Saurosternon bainii*, and again in Section 3. Paliguana whitei* Broom (1903) Upper Permian or Lower Triassic. Diagnosis: The unique type specimen is a partial skull. Based on the descriptions of Carroll (1975a; 1977), we are unable to identify synapomorphies for this taxon. It is discussed below fol- lowing the section on Saurosternon bainii* and again further below. Saurosternon bainii* Huxley (1868) Upper Permian. Diagnosis: The type is a relatively well-preserved and fairly complete postcranial skeleton. Based on the descriptions of Carroll (1975a; 1977), we are unable to identify synapomorphies for this taxon. An additional specimen, Albany Museum 4133, was said by Carroll (1975a:85) to have "few features to differentiate it" from the type specimen of Saurosternon* except in being about 20% larger. Carroll did not refer this specimen to Saurosternon*; because the specimen lacks data of any kind, we consider Saurosternon bainii* to apply only to the type specimen, British Mu- seum (Natural History) no. 1234. The three specimens listed above as Paliguana*, Palaeagama*, and Saurosternon* constitute Carroll's (1975a; 1977) family "Paliguanidae." Carroll was well aware that no synapomorphies unite these taxa. It is difficult to derive phylogenetically relevant information from comparison of a postcranial skeleton, a isolated skull, and a more complete although poorly preserved specimen; the specimens could either represent the same species or be only distantly related to each other. Nevertheless, Carroll has consistently treated "paliguanids" as if they were monophyletic. The tax- on "Paliguanidae" is rejected here and use of any name implying close relationship between Pali- guana*, Palaeagama* and Saurosternon* should be avoided. Because these taxa and others like them are based on plesiomorphy, it is not possible to refer other specimens to the taxon in ques- tion with any assurance. Such references may or may not be correct; lacking synapomorphies, a decision is impossible. 22 Phylogenetic Relationships of the Lizard Families Kuehneosaurs Kuehneosauridae and Eolacertilia of Robinson, 1962; non "Eolacertilia" of Carroll 1975a, 1977, Estes, 1983. Upper Triassic. Diagnosis: Based principally on Kuehneosaurtts latus and Icarosaurus siefkeri, and to a lesser extent on Kuehneosuchus ?atissimus, as described by Robinson (1962; 1967) and Colbert (1970). Seiffert (1973) and Estes (1983) have also considered Cteniogenys antiguas* to be a possible rela- tive of kuehneosaurs. These references, together with examination of the type of Icarosaurus and the analysis offered below, indicate that the following characters are synapomorphies of kuehneo- saurs (few if any of these are present in Cteniogenys aniiquus* and we are uncertain as to the rela- tionships of this poorly known lepidosauromorph). 1. External nares confluent on the midline owing to loss of intemarial process of pre- maxilla (Fig. IC). 2. Parietal foramen on frontoparietal suture (Fig. IC). 3. Absence (loss? fusion?) of supratemporal (Fig. IC). 4. Loss of ventral ramus of squamosal (Fig. 3B). 5. Loss of posterior ramus of jugal (Fig. 33). 6. Absence (loss? fusion?) of quadratojugal (Fig. 3B). 7. Quadrate foramen absent. 8. The distinctive shape and size of the transverse processes and ribs in the midtrunk re- gion, which are thought to have supported a gliding membrane (Colbert, 1970). 9. Non-notochordal trunk vertebrae lacking discrete intercentra. 10. Entepicondylar foramen absent. 11. Elongate gracile limbs. These synapomorphies indicate that the kuehneosaurs form a monophyletic group. The loss of the lower temporal bar and the presence of a lateral conch on the quadrate were the basis for Ro- binson's (1962; 1967) conclusion that kuehneosaurs were "lizards." As will be argued below, however, the form of the quadrate characterizes a more inclusive group than squamates alone, and the loss of the lower temf)oral bar is either convergence or, like the form of the quadrate, it is a sy- napomorphy of a more inclusive group. The name Eolacertilia, coined by Robinson (1962) as a higher level group name to include the kuehneosaurs, was made paraphyletic by inclusion of "paliguanids" (Carroll, 1975, 1977, Estes, 1983a). Eolacertilia (sensu Robinson, 1967) is a redundant taxon, carrying no more phylogenetic information than does Kuehneosauridae, and will not be needed until taxa that are closer to kuehne- osaurs than to lepidosaurs are discovered. Rhynchocephalians Gephyrosaurus bridensis of Evans, 1980 + Sphenodontida of Estes, 1983. Lower Triassic to Recent. Rhynchocephalians are represented by a single living species, Sphenodon punctatus. Like other small reptiles, rhynchocephalians are poorly represented in the fossil record, with fewer than three dozen species referred to this taxon. However, as the photographs in Cocude-Michel's (1963) monograph attest, some rhynchocephalians are among the most completely preserved fossil rep- tiles. Although t?ie reviews of Cocude-Michel (1963), Kuhn (1969) and Robinson (1973), are im- portant first steps, further analysis of rhynchocephalian anatomy and interrelationships is needed. Consequently, in order to identiiy rhynchocephalian synapomorphies, we briefly analyze here the /' \ ~ ^\_>-?-'i--A?[iJi_a_EiIi>tli^ FIGURE 3. Lateral view of skull. A, Youngina capensis (younginiform); B, Kuehneosaurus latus (kuehneosaur); C, Gephyrosaurus bridensis (rhynchocephalian); D, Planocephalosaurus robinso- nae (clcvosaur); E, Clevosaurus hudsoni (clevosaur); F, Sphenodon punciaius (sphenodonl); G, Scel- oporus grammicus (iguanian); H, Elgaria multicarinata (autarchoglossan); J, Sphenodon punctatus (sphcnodont). Lateral and medial view of mandible. I, L, Gephyrosaurus bridensis (rhynchocephalian); J, M, Sphenodon punctatus (sphenodont); K, N, Elgaria multicarinata. A after Carroll (1977); B after Robinson (1962); C, F, I, J, L, M after Evans (1980); D after Fraser (1982); E after Robinson (1973); G after Larsen and Tanner (1974); H, K, N after Rieppel (1980). 24 Phylogenetic Relationships of the Lizard Families phylogenetic relationships among the major rhynchocephalian groups. The following must be considered preliminary because (1) certain poorly known or equivocal taxa, such as Pleurosaurus, Palacrodon, Brachyrhinodon, and the eilenodonts (= Eilenodontinae of Rasmussen and Callison, 1981), are omitted from the analysis because of the difficulty of determining character states accu- rately; (2) our sample of rhynchocephalians (aside from what appears in the literature) is limited to five specimens of Recent Sphenodon, a cast of a species of the Upper Jurassic Homoeosaurus, and fragmentary remains of the Upper Triassic Clevosaurus and the Lower Jurassic Gephyrosaurus. We have relied heavily on the works of Cocude-Michel (1963), Robinson (1973, 1976), Evans (1980; 1981), and Fraser (1982) for supplementary information. The long-held belief that rhynchosaurs are rhynchocephalians is not justified. Carroll (1975b) showed that the synapomorphies traditionally thought to ally the rhynchosaurs with the rhynchoce- phalians, most notably the form of the teeth and snout, are erroneous. Indeed, rhynchosaurs are closer to archosaurs, and rhynchocephalians are closer to squamates (Gauthier 1984; Carroll, 1985). The rhynchocephalians are redefined here to reflect G?nther's (1867) original intent that this taxon should encompass Sphenodon and its nearest relatives. Diagnosis: Rhynchocephalia includes the Lower Jurassic Sf>ecies, Gephyrosaurus bridensis (Evans, 1980), and its sister taxon, the Lower Triassic to Recent Sphenodontida (see below for di- agnosis and included taxa). This conclusion differs from that of Evans (1980, 1981, 1985), who attempted to show that Gephyrosaurus and squamates were sister groups. Gephyrosaurus and sphenodontidans possess the following synapomorphies, however, that as a group are lacking in all other diapsids: 1. Postfrontal process of parietal reduced, and postfrontal develops a posterior process, thus excluding postorbital from the parietal (Fig. ID). 2. Postorbital overlaps the dorsal surface of the postfrontal distally, thus imparting a subtrapezoidal outline to the postfrontal when viewed from above (Fig. ID). 3. The lacrimal is quite reduced and barely exposed in lateral view (Fig. 3C). 4. Enlarged tooth row along maxillary side of palatine (Evans, 1980); palatine broa- dened laterally, restricting anterior portion of suborbital fenestra (Fig. 2C). 5. Pterygoid process of quadrate extends anterior to the level of the basioccipital tubera (Fig. 2C). 6. Dentary with long posterior process that extends more than half-way between the co- ronoid eminence and the articular condyle (Fig. 31). 7. Loss of splenial (Fig. 3L). 8. Strong anteroposterior ridge divides the dorsal surface of the mandibular condyle (Evans, 1980). 9. Length of mandibular condyle equal to width (see fig. 45 in Evans, 1980). The marginal dentition of Gephyrosaurus (Evans, 1980:236-239) is intermediate between the ancestral lepidosaur condition and that of sphenodontidans. Enlarged, conical, posterior teeth and the reduced amount of lateral attachment appear to be additional Gephyrosaurus-^cY?tno?on??zn (rhynchocephalian) synapomorphies. On the other hand, high tooth number, retention of tooth re- placement in the adult, and lack of extreme deposition of attachment bone around the teeth are an- cestral lepidosaurian features not seen in any sphenodontidan. Gephyrosaurus bridensis is, however, unlikely to be ancestral to any sphenodontidan because it has the following synapomorphies (modified from Evans, 1980): 1. Quadratojugal reduced, may be par?y fused to quadrate (Fig. 3C). 2. Loss of the quadratojugal-jugal contact owing to loss of the anterior process of the quadra- tojugal (Fig. 3C). LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 25 3. Frontals fused (Fig. ID). 4. Parietals fused (Fig. ID). 5. Dermal rugosities on skull roof (Evans, 1980, figs. 8,9). 6. Loss of supratemporal (Fig. ID). 7. Restriction of Meckelian fossa by dentary (Fig. 3L). This combination of synapomorphies is unique, although several of these characters have aris- en independently in other lepidosauromorphs (see below). Because there is only a single species with this combination of synapomorphies, there is no need at present for Evans' (1980) redundant higher level category, Gephyrosauridae. Sphenodontidans (= Sphenodontida of Estes, 1983): Sphenodontidans share the following combination of synapomorphies that are absent in Gephyrosaurus and other lepidosauromorphs: 1. Distinctive tooth form and regionalization (Robinson, 1976), including relatively fewer marginal teeth, each of which is relatively large and is fused to the dorsal mar- gin of the jaws, witii alternate-sized juvenile dentition anteriorly, and broad-based, somewhat laterally compressed, flanged teeth posteriorly. 2. Prominent coronoid eminence on mandible. 3. Premaxillary teeth replaced by dovmgrowths of premaxillae, forming paired chisels in all but juvenile individuals. 4. Four or five enlarged teeth at anterior end of palatine tooth row. 5. Palatine further enlarged laterally compared to Gephyrosaurus and suborbital fenestra consequentiy reduced. 6. Loss of lacrimal. 7. Jugal deeply overlaps quadratojugal laterally to extend well posterior to middle of lower temporal fenestra. 8. Jugal contacts squamosal at posteroventral margin of lower temporal fenestra. 9. Narrow quadrate without lateral conch. For purposes of this analysis, the following informal groups of sphenodontidans are recog- nized: clevosaurs, Homoeosaurus, sapheosaurs and sphenodonts. Synapomorphies for these groups, and the species included within them, are provided below. Clevosaurs: A group including Clevosaurus hudsoni (Robinson, 1973) and Planocephalosau- rus robinsonae (Fraser, 1982). Although quite different from one another, they possess the follow- ing synapomorphies: three or four enlarged, conical teeth in the posterior half of the maxillae, fol- lowed by a few very small teeth (Figs. 3D, E); loss of the anterior process of the quadratojugal (Figs. 3D, E); loss of contact between the posterior ramus of the jugal and the ventral ramus of the squamosal at the posteroventral comer of the lower temporal fenestra (Figs. 3D, E). Unfortunately, Planocephalosaurus presents difficulties with respect to the last two synapo- morphies. The referred remains are disarticulated and dissociated. Moreover, although most speci- mens have an incomplete lower temporal bar, at least some referred specimens display the ancestral sphenodontidan condition in maintaining the squamosal-jugal contact at the posteroventral margin of the lower temporal fenestra. Fraser and Walkden (1983) described Sigmala sigmala, another sphenodontid from the same fauna, and indicate that there are at least three others that are unde- scribed. Until more is known, we will consider tiie specimens that retain the ancestral condition to be the remains of Sigmala or one of the undescribed species. Thus, for purposes of this paper, Planocephalosaurus is considered to lack both the anterior process of the quadratojugal and the squamosal-jugal contact Homoeosaurus spp.: A genus most recently reviewed by Cocude-Michel (1963) and Kuhn (1969), the species o? Homoeosaurus are of interest owing to their lizard-like habitus. Compared 26 Phylogenetic Relationships of the Lizard Families to sphenodonts and sapheosaurs, Homoeosaurus is distinguished by relatively small size, gracile and elongate limbs, and a broad parietal table (Fig. IH). With the possible exception of the limb proportions, these characters are present in rhynchocephalians ancestrally (Fig. ID). These charac- ters are, however, confined to juveniles of the sphenodonts and sapheosaurs (e.g., the parietal forms a broad plate in juvenile Sphenodon; Howes and Swinnerton, 1901), suggesting that small size of Homoeosaurus, and the shape of the parietal table, are manifestations of the same phenome- non. \f Homoeosaurus is the sister species group of sapheosaurs, as some evidence suggests (see below), then the broad parietal table of Homoeosaurus may be a paedomorphic reversal rather than a retained ancestral feature. Sapheosaurs: A group including Leptosaurus neptunius, Kallimodon cerinensis, and Sapheo- saurus thiollierei. These taxa share a broader upper temporal arch (Fig. IG), an elongate and nar- rower upper temporal fenestra (Fig. IG), a postorbital region of the skull that exceeds the length of the preorbital region (Fig. IG), and laterally compressed caudal vertebral centra (Cocude-Michel, 1963). Sphenodonts: A group including Opisthias rarus and Sphenodon punctatus, which are unique among sphenodontidans in having successional, caniniform teeth at the anterior ends of the maxil- lary and dentary tooth rows. Sphenodon has been extensively studied, and it is not justifiable at present to diagnose the sphenodonts on the single character that unites the inadequately known Opisthias with Sphenodon. So far as they can be compared, Sphenodon possesses the following synapomorphies that are absent in Homoeosaurus, sapheosaurs, and clevosaurs. Future finds may determine which of these synapomorphies apply to sphenodonts generally, as opposed to Spheno- don alone. Enlarged, relatively broader nasal process of premaxilla (Fig. II); loss of jugal-squamosal con- tact below postorbital (Fig. 3F); diastema between maxillary and premaxillary tooth rows (Fig. 3F); elongate row of teeth on enlarged palatine, the latter nearly closing the suborbital fenestra (Fig. 2F); one vertebra added to presacral vertebral column (Hoffstetter and Gase, 1969); loss of two (sometimes one) sternal-rib connections, thus yielding eight cervicals, rather than seven, and fourteen poststemal vertebrae, ratiier than 12 (Hoffstetter and Gase, 1969); xiphistemum absent; very large, subrectangular supratemporal fenestra (Fig, 11); loss of posterior process on second sa- cral rib (Cocude-Michel, 1963). Homoeosaurus, sapheosaurs, and sphenodonts share the following combination of synapomor- phies that distinguishes them from clevosaurs and Gephyrosaurus: pattern of tooth wear facets, tootii ultrastructure, and anteroposterior length of the mandibular articulation, all of which indicate development of the propalinal masticatory movements to the degree seen in Recent Sphenodon (see Robinson, 1976; Rasmussen and Callison, 1981); parietal foramen nearer to frontoparietal suture (Fig. II); quadrate without lateral conch (Robinson, 1973); quadrate straight in lateral view; quad- rate reduced so that it does not extend dorsally more than half height of orbit (Fig. 3F); squamosal covers most of quadrate in lateral view (Fig. 3F); enlarged quadrate foramen (Robinson, 1973); loss of retroarticular process (Fig. 3M); short supratemporal process of parietal and long dorsal process of squamosal (Fig. IG); no discrete supratemporal (fused to squamosal?); greatiy enlarged postfron- tal extending far posteriorly on parietal (Fig. II); except for a few teeth on the vomer in juveniles and the palatine row, all other palatal teeth absent (Fig. 2E); enlarged palatine teeth set parallel to maxillary tooth row (Fig. 2E). Although the discussion above indicates that sphenodonts, sapheosaurs, and Homoeosaurus probably form a monophyletic group, phylogenetic relationships among these three taxa are un- clear. Sphenodonts and sapheosaurs share two apomorphic characters, a narrow parietal table (Fig. IG) and aliform outgrowths on the calcified distal segments of the first eleven poststemal ribs (Cocude-Michel, 1963). Although this character has not been reported in Homoeosaurus, the un- LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 27 likely preservation of calcified cartilage makes the level at which this character is a synapomorphy uncertain. On the other hand, sapheosaurs share two apparent synapomorphies with Homoeosau- rus: relatively large posterior flanges on the maxillary teeth and long ischial tubera. Because of this conflicting evidence and because the detailed relationships within sphenodontidans are not our primary concern here, we leave the relationships among sphenodonts, sapheosaurs, and Homoeo- saurus unresolved. The preceding analysis indicates that rhynchocephalians are readily categorized into a hierarchy of groups nested within groups. However, there is little to be gained from a new taxonomy until more thorough analyses of critical taxa, such as the early Triassic species, Palacrodon and Schars- chengia, and the late Triassic species, such as Polysphenodon, Clevosaurus, Planocephalosaurus, and Brachyrhinodon are made, and until more specimens of the late Jurassic eilenodonts and Opis- thias are available. At this point, we wish only to stress that as members of a monophyletic Rhynchocephalia, sphenodontidans and Gephyrosaurus are closer to one another than either of them is to any other lepidosauromorph. The phylogenetic conclusions of this preliminary analysis of rhynchocephalians are depicted in Fig. 12, and the relationships of that group to other lepidosauromorphs in Fig. 13. Squamales Squamata of Oppel, 1811. LateJurassic to Recent. "Lizards" are those squamates that lack the synapomorphies of snakes and of amphisbaenians. Therefore, we concern ourselves with squamates, rather than "lacertilians", in the remaining discus- sion. Because determination of squamate synapomorphies is a primary objective of this analysis, a detailed discussion of these characters appears below. MORE INCLUSIVE GROUPS OF LEPIDOSAUROMORPHA AND THEIR DIAGNOSTIC CHARACIERS The basic taxa of this analysis are the younginiforms, Palaeagama*, Paliguana*, Sauroster- non*, kuehneosaurs, rhynchocephalians, and squamates, as diagnosed above. These taxa share a combination of synapomorphies in the morphology of the forelimbs and girdles that will be dis- cussed below, and are referred to collectively here as Lepidosauromorpha. As defined here, Lepido- sauromorpha is the sister taxon of Archosauromcrpha (Huene, 1946, 1949, 1956); these two taxa together constitute Sauria (approximately sensu McCartney 1802, rather than as a synonym of "Lacertilia"). Sauria is itself the sister taxon of Araeoscelidia (Reisz et al., 1984) within Diapsida (Gauthier, 1984). In the list of characters below, the number of the character in the character list and data matrix (Appendices I and II) is placed in brackets, following the number in this listing. Archosauromor- pha and Araeoscelidia are used as first and second nearest outgroups, respectively, to determine character polarities in the following analysis. LEPIDOSAUROMORPHA This taxon includes Younginiformes, Palaeagama*, Saiirosiernon*, Paliguana*. Kuehneosau- ridae, Rhynchocephalia, and Squamata) Definition: Sphenodon and squamates and all saurians sharing a more recent common ancestor with them than they do with crocodiles and birds. Diagnosis: 1. [91] The lepidosauromorph sternum. According to Romer (1956), the ancestral condition of the diapsid sternum is retained by Crocodylia (see Romer, 1956, fig. 141). Anterior- ly, the cartilaginous sternum forms a broad plate lying between the coracoids on the ventral mid- 28 Phylogenetic Relationships of the Lizard Families line. Behind the coracoids, however, the sternum tapers abruptly to an elongate, rod-like structure supporting four or five ribs. The sternum bifurcates posteriorly to this rod-like structure to form the xiphistema, which support additional rib extensions (Gladstone and Wakely, 1932). The ster- num and its associated ventral rib extensions become calcified or ossified prior to the cessation of growth in living diapsids (see Romer, 1956; pers. obs.). Unfortunately, calcified cartilage is rarely preserved in fossils. Further, extensive calcification usually occurs late in development, thus mak- ing it even less likely to be preserved in subadults. More importantly, calcified cartilage is rarely preserved because it macerates quickly in water; in addition, compared to the sternum, the ventral rib extensions are considerably more lightly constructed and less calcified. Consequently, remnants of a calcified sternum are more likely to be found in an associated fossil than are ventral rib exten- sions. In some groups, such as sauropods and theropods, secondary ossification centers may arise in the anterior portion of the sternum lying between the pectoral girdles (Borsuk-Bialynicka, 1977). In birds other than Archaeopteryx*, as well as in pterosaurs, the sternum may become greatly enlarged, elaborated, and well ossified for the attachment of the flight musculature. Lepido- sauromorphs differ from other diapsids in that the rib-supporting posterior portion of the sternum is formed by two broad plates that fuse to one another prior to the cessation of growth (Broom, 1921; Carroll, 1975a, 1977; Currie, 1981b). This synapomorphy is present in younginiforms, Saurosternon*, rhynchocephalians, and squamates, but it is not preserved in Paliguana* or kuehne- osaurs. Carroll (1975a) de.scribed calcified remnants of the lepidosauromorph sternum in Palaeaga- ma*. In diapsids, the sternum and its associated ribs become calcified during postnatal develop- ment. However, sternal calcification appears to take place at earlier ontogenetic stages in rhyn- chocephalians and squamates than in crocodiles and birds (pers. obs.). This is probably the case in lepidosauromorphs generally, because calcification (ossification) apparently began relatively early in postnatal development in younginiforms (Currie, 1981b). 2. [114] Fully enclosed ectepicondylar foramen. An ectepicondylar groove is present on the distal end of the hum?rus in araeoscelidans (Reisz, 1981) and archosauromorphs ancestrally (Gauthier, 1984). Therefore, an ectepicondylar groove is considered to be present in saurians ances- trally. In younginiforms, however, the ectepicondylar groove becomes fully enclosed in bone dur- ing postembryonic ontogeny to form an ectepicondylar foramen (Currie, 1981b). The postcranial skeleton is unknown in Paliguana*, but an ectepicondylar foramen is present in Saurosternon* and Palaeagama* (Carroll, 1975a). This foramen is also present in kuehneosaurs (Colbert, 1970) and rhynchocephalians (Evans, 1980), and is generally present in squamates (Lecuru, 1969). Those squamates that lack an ectepicondylar foramen, however, also lack an ectepicondylar groove. The ectepicondylar groove becomes a foramen in the embryo in rhynchocephalians (Howes and Swin- nerton, 1901) and in squamates (pers. obs.). Thus, ontogenetic enclosure of the ectepicondylar groove is a synapomorphy of lepidosauromorphs. 3. [106] Medial centrale enlarged to nearly twice size of lateral centrale in manus. In diapsids ancestrally the medial and lateral centralia are subequal in size (Fig. 4A). This character is not pre- served in Paliguana*, Palaeagama*, and kuehneosaurs. However, in younginiforms (Currie, 1982), Saurosternon* (Carroll, 1975a), and all squamates and rhynchocephalians that retain an un- reduced manus, the medial centrale is as much as twice the size of the lateral centrale (Figs. 4B,C). LEPIDOSAURIFORMES, new taxon This taxon includes Paliguana*, Saurosternon*, Kuehneosauridae, Rhynchocephalia, and Squa- mata. Definition: Sphenodon and squamates and all organisms sharing a more recent common ances- tor with them than they do with younginiforms. LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 29 FIGURE 4. Dorsal view of left manus. A, Petrolacosaurus kansensis (araeoscelidan); B, Sphen- odon punctatus (sphenodont); C, Zonosaurus madagascariensis (autarchoglossan); D, Opetiosaurus bucchichi (autarchoglossan); E, CUdastes sp. (autarchoglossan). A after Reisz (1981); B after Carroll (1977) C modified from Carroll (1977); D after Komhuber (1901); E after Russell (1967). Paliguana*, Saurosternon*, kuehneosaurs, rhynchocephalians, and squamates possess synapo- morphies that are absent in younginiforms. Thus, younginiforms, archosauromorphs, and araeos- celidans will be used as successively more remote outgroups in the following analysis. Lepidosau- riformes is distinguished from its sister taxon, Younginiformes, by further modifications of the forelimbs and girdles, and the initial modifications associated with the development of a unique ankle joint. To these may be added the modifications of the quadrate and ear region associated with the development of an ear that is sensitive to airborne sound (Carroll, 1977). The form of the quadrate has long been considered a synapomorphy unique to squamates. More recently, however, this assumption has been questioned (Evans, 1980), and we argue below that this synapomorphy applies to all lepidosauromorphs except younginiforms, and not to squamates alone. In our view, the extinct members of the Lepidosauriformes, i.e., Paliguana*, Saurosternon*, and the kuehneosaurs, are too poorly known to be particularly informative. Small sample sizes and poor preservation conspire to leave too many unanswered questions. As is evident in the fol- lowing list of synapomorphies, their remains clearly indicate that these taxa are closer to rhynchocephalians and squamates than to any other reptiles. We use the name Lepidosauriformes for all lepidosauromorphs that shared a more recent common ancestor with lepidosaurs than they did with younginiforms. Because of the taxonomic instability resulting from new fossil finds, and the uncertainties associated with interpreting their morphology, we do not wish to restrict the 30 Phylogenetic Relationships of the Lizard Families name Lepidosauriformes to the above-listed groups alone; equally we do not wish to coin new names for all the more inclusive clades that might appear upon the discovery of new fossils. Therefore, the ultimate diagnosis of Lepidosauriformes will differ from that given below in that it will consist of only the first synapomorphy (ies) to arise in this clade. To emphasize the point that we have made earlier regarding the nature of metaspecies, note that although Paliguana* and Sawosternon* share synapomorphies with other taxa at this node, they are themselves undiagnosa- ble by synapomorphies. Diagnosis: 1. [19] Prominent lateral conch on quadrate for support of tympanum. There is no lateral conch on the quadrate of araeoscelidans (Reisz, 1981), archosauromorphs (including Pro- lacerta, see above), or younginiforms (Gow, 1975). A lateral conch is present on the quadrate of Paliguana* (Carroll, 1975a), kuehneosaurs (Fig. 3B), and all squamates that retain a tympanum (Evans, 1980). Sphenodon lacks both a tympanum and a lateral conch on the quadrate (Fig. 3F). So far as can be determined from the bones, this condition occurs in Homoeosaurus and sapheo- saurs as well (Figs. 1G,H). In the Triassic sphenodontidans such as Planocephalosaurus (Fig. 3D) and Clevosaurus (Fig. 3E), however, the quadrate has a poorly developed conch and the lateral mar- gin of the skull is clearly modified for support of a tympanum. As noted above, Gephyrosaurus is the sister taxon of all other rhynchocephalians, and this Early Jurassic species has a prominent lat- eral conch on the quadrate (Fig. 3C). Because rhynchocephalians and squamates are closer to one another than either is to Paliguana* or kuehneosaurs, a prominent lateral conch on the quadrate is considered to be a synapomorphy of lepidosauriforms. This requires the secondary loss of the quad- rate conch and tympanum within sphenodontidans, as has already been suggested by Evans (1980). Perhaps increased specialization for propalinal jaw movements in the sphenodont-homoeosaur- sapheosaur group (Robinson, 1976) may be functionally or developmentally related to the reduced sensitivity of the ear to airborne sound. The presence of a slender stapes is likely to be correlated with the quadrate conch, because a tympanum seems superfluous without a stapes slender enough to transmit its motion to the inner ear (Carroll, 1977, and references therein). Although there are lepidosaurs with a slender stapes but without a conch or tympanum (i.e., Sphenodon), the converse never seems to occur. Aside from the relatively stout stapes retained by younginiforms (Gow, 1975), and the slender stapes known in kuehneosaurs (Evans, 1980), this character is unknown in other extinct lepidosauromorphs. 2. [36] Quadrate bowed in lateral view. The quadrate is not modified for support of a middle and external ear in amniotes ancestrally, nor in reptiles, diapsids, or saurians (Gauthier, 1984). The ancestral condition is also retained by younginiforms (Fig. 3A). Lepidosauriforms differ in that the body of the quadrate is bowed anteriorly for support of the middle ear cavity and the lateral portion of the quadrate is likewise modified to form a conch that supports the tympanum. A bowed quad- rate is present in Paliguana* (Carroll, 1977), kuehneosaurs (Fig. 33), and in ancestral rhynchocephalians (Fig. 3C) and squamates (Figs. 3G-H). This character may be correlated with the ones preceding and following it; however, examples can be found in which one or more are ab- sent in the presence of the other (e.g., tall quadrate without lateral conch in archosauromorphs; Gauthier, 1984). We therefore treat them separately until the correlation is better documented. A straight quadrate has been developed secondarily in some rhynchocephalians (e.g., Sphenodon) and squamates (e.g., chamaeleons). 3. [17] Enlarged adductor chamber and quadrate extends well below occipital condyle. In ances- tral saurians the adductor chamber is small and the quadrate does not extend well below the level of the occipital condyle. This condition is seen in araeoscelidans (Reisz, 1981) as well as in amni- otes ancestrally (Panchen, 1972), and it is retained in younginiforms (Fig. 3A). In contrast, the adductor chamber is relatively large and the quadrate extends well below the occipital condyle in Paliguana* (Carroll, 1977), kuehneosaurs (Fig. 3B), and in ancestral rhynchocephalians (Fig. 3D) LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 31 I I FIGURES. Lateral view of scapulocoracoid. A, Petrolacosaurus kansensis (aTato&ccUdan); B, Gephyrosaurus bridensis (rhynchocephalian); C, Lepidophyma flavimaculalum (aularchoglossan). Ventral view of interclavicle. D, Youngina capensis (younginiform); E, Saurosternon bainii* (lepidosauriform); F, Lepidophyma smithi (aularchoglossan). A after Reisz (1981); B, E after Evans (1981); C after Lecuru (1968a); D after Gow (1975). and squamates (Figs. 3G-H). This character is not preserved in Saurosternon* and Palaeagama* (Carroll, 1977, figs. 2 and 3). 4. [72] Prominent retroarticular process of prearticular. A moderately developed retroarticular process is present in saurians ancestrally (Gauthier, 1984; and see Fig. 3A). The retroarticular pro- cess is, however, more prominently developed in all lepidosauromorphs that retain a tympanum (Figs. 3L). The synapomorphic condition is present in Paliguana* (Carroll, 1975a), kuehneosaurs (Robinson, 1962), and in rhynchocephalians (Fig. 31) and squamates (Fig. 3K) that retain a tympa- num. The mandibles are unknown in Palaeagama* and Saurosternon*. Although we are not sure of the level at which it arises within lepidosauromorphs, we note that the retroarticular process is formed entirely by the prearticular bone, rather than by the prearticular and articular as in archosau- romorphs. 5. [94] Lx)ss of cleithrum. Cleithra are present in diapsids ancestrally (Reisz, 1981). Cleithra are unknown among archosauromoiphs (Romer, 1956), but they are present in younginiforms an- cestrally (Currie, 1982). The shoulder girdles are not preserved in Paliguana* and Palaeagama*. No cleithra are present in Saurosternon*, kuehneosaurs, rhynchocephalians, or squamates. 6. [93] Gracile interclavicle. The interclavicle is a relatively broad, flat, T- or anchor-shaped element in diapsids ancestrally, judging from the form of the interclavicle in araeoscelidans (Reisz, 1981), the protorosaurian archosauromorph Prolacerta (Gow, 1975), and in younginiforms (Fig. 32 Phylogenetic Relationships of the Lizard Families 5D). The interclavicle is unknown in Paliguana* and Palaeagama*. Saurosternon* (Fig. 5E), rhynchocephalians (Evans, 1981), and squamates (Fig. 5F) differ from other lepidosauromorphs in that the interclavicle in general, and the lateral proces.ses of the interclavicle in particular, are rela- tively more slender and lightly constructed. Although the interclavicle is unknown in kuehneo- saurs, it is assumed to have been gracile, since a gracile interclavicle is present in Saurosternon*, which from the standpoint of the postcranial skeleton is the sister taxon of the kuehneosaur- rhynchocephalian-squamate group (see below). 7. [134] The lepidosauriform ankle joint. According to Brinkman (1980), Saurosternon* is like rhynchocephalians and squamates and unlike other diapsids in that it possesses a process on the ventromedial comer of the fourth distal tarsal that fits under the astragalus somewhat medial to the calcaneal-fourth distal tarsal articulation. This appears to be the initial event in the series of modifications that yields the specialized ankle of squamates (Brinkman, 1980; and see below).-The ankle is not preserved in Palaeagama* and Paliguana* (Carroll, 1975a), and it has not been de- scribed in kuehneosaurs (Evans, 1981). S. Evans (pers. comm.) informs us that on the basis of the dissociated astragalus and calcaneum elements in the British Museum (Natural History), the kuehneosaur ankle is not like that of squamates. UNNAMED TAXON INCLUDING KUHNEOSAURS. RHYNCHCXZEPHALIANS, AND SQUAMATES Kuehneosaurs and lepidosaurs possess synapomorphies that are apparently lacking in Paligua- na*, Palaeagama*, and Saurosternon*. These species, along with the younginiforms, archosauro- morphs, and araeoscelidans, are used as successively more remote outgroups in the following anal- ysis. Since much of the kuehneosaur material remains undescribed, it seems advisable not to name this taxon at present. It is clear nevertheless that kuehneosaurs, squamates, and rhynchocephalians possess synapomorphies in the manus and pelvis that distinguish them from Palaeagama* and Saurosternon*. Likewise, the members of this unnamed taxon possess synapomorphies in the skull that are lacking in Paliguana*. It is particularly at this node that the noncomplementary na- ture of the remains of Paliguana*, Palaeagama* and Saurosternon* are most keenly felt. So far as is known, kuehneosaurs, rhynchocephalians, and squamates share the following combination of sy- napomorphies that are absent in other lepidosauromorphs. Diagnosis: 1. [59] Anterolaterally oriented prefrontal-nasal suture. To judge by the condition in araeoscelidans (Fig. lA), younginiforms (Fig. IB) ana Paliguana* (Carroll, 1975a), the prefron- tal-nasal suture extends anteroposteriorly, parallel to the intemasal suture, in ancestral lepidosauro- morphs. This character is not preserved in Palaeagama* and Saurosternon*. In kuehneosaurs (Fig. IC) and rhynchocephalians (Fig. IH), however, the prefrontal-nasal suture is directed anterolateral- ly, diverging from the intemasal suture. The highly modified snout of Clevosaurus (Fig. IE) is considered secondary, because Planocephalosaurus is like other sphenodontidans in this regard (Fig. IF). Gephyrosaurus may also be secondarily modified in this respect (Fig. ID), because Evans (pers. comm. 1984) claims that the seemingly ancestral form of these sutures is accurate, and not an artifact of her reconstruction based on isolated elements. The snout is further transformed in squamates (Fig. IJ), being neither ancestral nor like that of kuehneosaurs and rhynchocephalians. In squamates, as discussed in greater detail below, the nasals are reduced and the facial process of the maxilla is correspondingly enlarged. Based on data discussed below, the snout of squamates is considered a further modification of the ancestral condition of lepidosauriforms. 2. [6] Loss of tabulars. Paired tabular bones are present in araeoscelidans (Reisz, 1981), but they have been lost in all archosauromorphs (see above). Younginiforms (Gow, 1975) and Pali- guana* (Carroll, 1975a), however, retain paired tabular bones. This character is not preserved in Saurosternon*, and the rear of the skull is too poorly preserved to discriminate between absence 4 LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 33 and non-preservation in Palaeagama* (Carroll, 1975). Thus, tabulars are considered to be present in lepidosauromorphs ancestrally. Tabulars are not reported in kuehneosaurs, and they are absent in all rhynchocephalians and squamates (Robinson, 1967). 3. [5] Loss of postparietals. A pair of small postparietal bones is present in araeoscelidans (Reisz, 1981), and a single postparietal, evidently representing the fusion of the originally paired elements, appears ancestral in archosauromorphs (Gauthier, 1984). The postparietals are retained by younginiforms (Gow, 1975) and Paliguana* (Carroll, 1977). This character is not preserved ei- ther in Saurosternon* or Palaeagama*. Accordingly, the bones are considered to have been paired in ancestral lepidosauromorphs. No postparietals have been reported in kuehneosaurs, and they are absent in all squamates and rhynchocephalians (Robinson, 1967). 4. [24] Loss of teeth on transverse flange of pterygoid. Teeth are present ancestrally on the transverse flange of the pterygoid in both archosauromorphs (Gauthier, 1984) and lepidosauro- morphs (Fig. 2A). Kuehneosaurs, squamates, and rhynchocephalians have lost the teeth on the transverse flange of the pterygoid (Fig. 2C). This character is not preserved in Paliguana*, Palaea- gama* and Saurosternon*, and the synapomoiphic condition may apply to a more inclusive group. 5. [33] Paroccipital process of opisthotic contacts quadrate. According to Reisz (1981), Gow (1975), and Carroll (1975a), the distal extremity of the opisthotic remains cartilaginous in diapsids ancestrally. Among lepidosauromorphs, the ancestral condition is retained by younginiforms and Paliguana* (Carroll, 1977). This character is not preserved in Saurosternon* and Palaeagama*. In contrast, the paroccipital process of the opisthotic is fully ossified and contacts the quadrate in kuehneosaurs (Robinson, 1967), and the apomorphic condition obtains throughout posthatching ontogeny in squamates and rhynchocephalians (pers. obs.). 6. [71] Angular reduced and lit?e exposed on lateral face of mandible. The angular is a prom- inent element in the mandible of araeoscelidans (Reisz, 1981), archosauromorphs (Gauthier, 1984), and younginiforms (Fig. 3A), for it extends more than one-third of the way up the lateral surface of the mandible. In contrast, the angular extends less than one-third of the way up the lateral surface of the mandible in kuehneosaurs, rhynchocephalians and squamates (e.g.. Fig. 3K). This character is not determinable in Palaeagama*, Paliguana*, and Saurosternon*. 7. [97] Hum?rus robust, but diameter of shaft reduced. Compared to the limbs of early synap- sids and captorhinids, amniotes like Paleothyris* and diapsids share the apomorphy of long and gracile limbs, and this description applies to the hum?rus as well. Although araeoscelidans have an exceptionally gracile hum?rus (Reisz, 1981), in saurians ancestially the element is more like that o? Paleothyris*; it is comparatively robust and thick-shafted relative to its length (Gauthier, 1984). This condition is retained in Saurosternon* (Carroll, 1977, fig. 10) and in a modified form in younginiforms (Currie, 1981b). The hum?rus is not preserved in Paliguana*. Compared to that of Saurosternon*, the diameter of the shaft of the hum?rus in kuehneosaurs (Colbert, 1970), rhynchocephalians (Evans, 1981), and especially squamates (see p. 55 below), is reduced relative to the length of that element. 8. [105] Metacarpal IV subequal or slightiy shorter than metacarpal ID. Because metacarpal IV is longer than III in araeoscelidans (Fig. 4A), archosauromorphs (Gauthier, 1984), and youngini- forms (Currie, 1981b), this appears to be the ancestral condition in archosauromorphs and lepido- sauromorphs. The manus is not preserved in Paliguana*, but Saurosternon* and Palaeagama* re- tain the ancestral condition because metacarpal IV is longer and stouter than III (Carroll, 1975a). To judge from the figures o? Icarosaurus (Colbert, 1970), the fourth metacarpal in kuehneosaurs is subequal to the third in length, thus approaching the condition seen in rhynchocephalians and squa- mates, in which the fourth metacarpal is shorter than the third (see below). 9. [121] Pelvic girdle fenestrate. As noted by Carroll (1977), the pelvis forms a solid plate in younginiforms, Saurosternon*, and Palaeagama*. This is the ancestral condition, because it oc- 34 Phylogenetic Relationships of the Lizard Families FIGURE 6. Lateral view of pelvic girdle. A, Petrolacosaurus kansensis (araeoscelidan); B, Pro- lacerta broomi (archosauromorph); C, Sphenodon punctatus (sphenodonl); D, Iguana iguana (iguanian). A after Reisz (1981); B after Gow (1975); C, D after Romer (1956). curs in araeoscelidans (Fig. 6A) and in archosauromorphs ancestrally (Fig. 6B). The pelvis is not preserved in Paliguana*. Kuehneosaurs (Colbert, 1970), rhynchocephalians (Fig. 6C), and squa- mates (Fig. 6D) are unique among lepidosauromorphs in possessing a fenestrate pelvic girdle. This character is not unique among diapsids, however, because fenestrate pelves have arisen inde- pendently in archosaurs &r\? Askeptosaurus (Evans, 1981). Nevertheless, the pelvic symphyses of other diapsids are readily distinguished from those of lepidosaurs by numerous morphological de- tails. For example, the pubis in squamates and rhynchocephalians is distinctive in the anteroven- tral curvature of the anterodorsal face of the element (Carroll, 1977). LEPIDOSAURIA This taxon indues rhynchocephalians and squamates; see Haeckel, 1866. Definition: The most recent common ancestor o? Sphenodon and squamates and all of its de- scendants. In this section, we argue that squamates and rhynchocephalians are closer to one another than either is to any other lepidosauromorph. Although Haeckel (1866) originally coined the term Le- pidosauria for what are here considered to be squamates, his subsequent writings made it clear that he considered Sphenodon a closely related form. To preserve his intent, and in order to append maximum information to a widely recognized taxon, we here restrict the name Lepidosauria to the least inclusive taxon of lepidosauromorphs that contains the conmion ancestor of Rhynchocephalia and Squamata. In the context of extant amniotes, we would emphasize that Sphenodon and squa- mates possess all the above-listed synapomorphies as well as those listed below; one might ques- tion whether or not a given fossil is part of this group, but not that lepidosaurs are most closely related among extant amniotes. To facilitate the following discussion, two primary clades of squamates are recognized, iguani- ans and autarchoglossans (as used in Gauthier, 1982, this name refers to all other noniguanian squamates, including gekkotans, snakes, and amphisbaenians). Diagnoses for squamate higher taxa are given in Estes et al. (1988). Based on the analysis presented here, the following characters are considered synapomorphies tliat distinguish lepidosaurs from other amniotes. LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 35 Diagnosis: 1. [160] Transverse cloacal slit The cloacal slit is oriented transversely in lepido- saurs, rather than anteroposteriorly as in other tetrapods (Oppel, 1811; G?nther, 1867). We be- lieve that the loss of the amniote penis corresponds to the transformtion of the cloaca in lepido- saurs. 2. [161] Sexual segment of kidney. All tetrapods other than lepidosaurs do not possess a sex- ual segment of the kidney (Fox, 1977). 3. [162] Tongue notched distally. Dibamus is an exception, but other data indicate that this taxon is an autarchoglossan squamate that must have lost the notched tongue (Estes et al., 1988). 4. [163] SmaU prey secured by tongue. Sphenodon and iguanians use the tongue to secure small prey (Gomiak, et al. 1982; pers. obs.), an attribute taken to an extreme in chamaeleons. No autarchoglossan is known to take small prey with the tongue. Minimizing evolutionary events cannot be used to resolve the level of synapomorphy in this case, because two steps are involved whether convergence between iguanians and Sphenodon is accepted or if the character is considered to apply to lepidosaurs generally and has become lost secondarily in autarchoglossans. We prefer the latter inteipretation because the form and function of the autarchoglossan tongue are in no other sense ancestral (Schwenk, 1988). Indeed, as discussed in Estes et al. (1988), correlated changes in the morphology of the tongue and Jacobson's organ suggest that the autarchoglossan tongue is modified to stress chemoreceptive over prey-capturing qualities, as suggested by Underwood (1971). 5. [164] Ciliary process of eye reduced or absent. At least some degree of development of a ciliary process is present in tetrapods ancestrally. According to Underwood (1970) lepidosaurs are unusual among amniotes in that the cihary process is reduced or absent. 6. [165] Tendon of nictitans attaches to orbital wall. In reptiles generally, the tendon of the nictitating membrane is attached to the m. pyramidalis, which takes its origin from the surface of the eyeball. In lepidosaurs, however, the tendon is transferred to the m. retractor bulbi and the in- terorbital septum (Underwood, 1970). 7. [167] Tenon in lower eyelid. A cartilaginous disc, the tenon, is present in lower eyelid of lepidosaurs alone among amniotes (Romer, 1956). 8. [166] Lepidosaur ecdysis. Ecdysis is irregular and piecemeal in amniotes ancestrally, but in lepidosaurs the skin is regularly shed in its entirety (Gans, 1978). 9. [168] Modified middorsal scale row. The middorsal row is considered "modified" if its gross appearance differs from that of the lateral scale rows. So defined, a modified middorsal scale row is present in Sphenodon, many iguanians, and in a single group of autarchoglossans, the mosasaurs (Williston, 1899). Etheridge and de Queiroz (1988) considered the modified scale row to be the an- cestral condition in iguanians. The detailed shape of the middorsal scale row varies markedly among iguanians, however, from the slightly larger and more projecting middorsal scales of Dipso- saurus to the oddly-shaped structures in some chamaeleons to the tail spines of Amblyrhynchus. There are also groups in which a middorsal scale row has been lost entirely, such as the scelopo- rines. In spite of this variation, presence of this character in rhynchocephalians, iguanians, and mosasaurs indicates that it represents the ancestral condition for lepidosaurs; loss of the modified middorsal scale row within autarchoglossans is considered secondary. The above synapomorphies, which would seldom be preserved in the fossil record, are less useful in determining the relationships of fossil species. Comparative osteology reveals that rhyn- chocephalians and squamates possess the following combination of synapomorphies that distin- guishes them from all amniotes, including kuehneosaurs, Saurosternon*, Paliguana*, Palaeaga- ma*, and younginiforms. 10. [8] Long anterior extent of squamosal, usually extending anterior to level of braincase, and 36 Phylogenetic Relationships of the Lizard Families approaching closely or contacting elongate postorbital ramus of jugal. In archosauromorphs ances- trally and in lepidosauromorphs, the squamosal extends anteriorly approximately halfway over the lower temporal fenestra and is separated from the jugal by a wide gap below the postorbital. The ancestral condition found in araeoscelidans (Reisz, 1981) and archosauromorphs (Gauthier, 1984), and it is retained in younginiforms (Fig. 3A), Paliguana* (Carroll, 1977), and kuehneosaurs (Fig. 3B). This character is not preserved in Palaeagama* and Saurosternon*. Variation among squa- mates and rhynchocephalians makes this character difficult to interpret. The synapomorphic condi- tion, in which the jugal and squamosal nearly or entirely exclude the postorbital from the lower temporal fenestra, is present in Gephyrosaurus (Fig. 3C) and in clevosaurs (Figs. 3D,E), although it is absent in Sphenodon (Fig. 3F). Similarly, squamates are variable in this regard. The synapo- morphic condition is present in iguanians (Fig. 3G). Except for xantusiids and cordylids, the squa- mosal is relatively elongate in autarchoglossans (Fig. 3H). However, this character is difficult to interpret in autarchoglossans that have modified the upper temporal arch, including the xantusiids and cordylids noted above. Although obscured by overlap of the postorbital, the synapomorphic condition is present in "aigialosaurs" and mosasaurs (Russell, 1967), and in gekkotans that retain the arches intact (Bavarisaurus). Perhaps related to changes in the shape of the postorbital, howev- er, the squamosal is more broadly separated from the jugal in all anguimorphs except xenosaurids, all lacertoids except teiids, and all scincoids except scincids (Estes et al., 1988). Other characters indicate that anguimorphs, lacertoids and scincoids are derived in this regard, and that teiids, xeno- saurids and scincids have regained the squamosal-jugal contact secondarily. This character cannot be interpreted in squamates that lack the temporal arches. 11. [1] Reduced lacrimal. Presence of a large lacrimal is ancestral for diapsids, although in most saurians this bone is less prominent than in araeoscelidans (Reisz, 1981). The ancestral con- dition is retained by younginiforms (Fig. 3A), Paliguana* (Carroll, 1975), and kuehneosaurs (Fig. 3B), but this character is not preserved in Palaeagama* and Saurosternon*. All lepidosaurs possess a synapomorphic condition in which the lacrimal is small and confined to the orbital rim, or it is absent (Figs. 3C,H). 12. [35] Imperforate stapes. The base of the robust stapes bears a foramen for the passage of the stapedial artery in araeoscelidans (Reisz, 1981) and younginiforms (Carroll, 1977). The stapes is more lightiy constructed in kuehneosaurs, but is still perforate (Evans, 1980). The stapes of rhynchocephalians is like that of kuehneosaurs in tiiat it is more lightiy constructed and elongate than in lepidosauromorphs ancestrally. However, the rhynchocephalian stapes differs in being im- perforate in postembryonic developmental stages (Gans, 1978). The stapes is relatively very thin (columelliform) in squamates (pers. obs.). The stapes is imperforate in squamates generally, ex- cept in dibamids (Greer, 1976) and some gekkotans (Underwood, 1957). During ontogeny, the po- sition of the stapedial artery changes from being enclosed in the stapes to passing posterior to the stapes in amphisbaenians (Gans and Wever, 1976). We are not certain if the artery lies within the stapes in embryonic Sphenodon (Estes et al., 1988, contra Gans, 1978), nor do we have informa- tion on this transformation in lepidosaurs other than amphisbaenians. Nevertheless, it appears safe to conclude that the loss of the stapedial foramen during ontogeny is a lepidosaur synapomorphy. Accepting this conclusion requires the secondary acquisition of the perforate stapes in dibamids on one hand, and within gekkotans on the other. The proposed reversals are in keeping with numer- ous morphological details that point to paedomorphosis in these two groups, which could account for the character discordance that lies at the base of the controversies that surround the relationships of dibamids and gekkotans (see Greer, 1984; Estes et al., 1988). 13. [31] Loss of teeth on the parasphenoid. Teeth are present on this element in diapsids an- cestrally, being present in araeoscelidans (Reisz, 1981). If Heleosaurus is an archosauromorph (Carroll, 1975b), then the latter group also retains teeth on tiie parasphenoid ancestrally, although LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 37 the element is edentulous in other archosauromorphs (Gow, 1975). Parasphenoidal teeth are absent in younginiforms (Fig. 2A), and their presence or absence is indeterminable in Paliguana*, Palaea- gama*, and Saurosternon*. Parasphenoidal teeth are present in kuehneosaurs (Evans, 1980), but they are absent in all rhynchocephalians and squamates (Figs. 2C-H). Lack of knowledge of the distribution of this character leaves the level of synapomorphy ambiguous. In this instance, we will treat the absence of this character in younginiforms as convergence, rather than assume that it applies to lepidosauromorphs generally, with kuehneosaurs subsequently regaining parabasisphe- noidal teeth. We accept the former hypothesis until this character can be determined in Paliguana*, Saurosternon* and Palaeagama*. 14. [29] Complete abducens canal and well-developed dorsum sellae. Detailed knowledge of the dorsal surface of the basisphenoid is available for relatively few early diapsids. Grooves in the dor- sal surface of the basisphenoid may be present in archosauromorphs such as Prolacerta (Gow, 1975). This character is not preserved in Paliguana*, Palaeagama*, and Saurosternon*. According to Evans (1980), however, in Kuehneosaurus latus the dorsum sellae is poorly developed, and nei- ther grooves nor canals mark the passage of the abducens nerves in the basisphenoid. In lepido- saurs, however, there are paired canals in the basisphenoid for the passage of the abducens nerves (Evans, 1980). The nerves enter the basisphenoid posterodorsally and emerge from the dorsum sel- lae anteriorly (Oelrich, 1956). In amphisbaenians, which lack the abducens and other nerves asso- ciated with the eye musculature, there are no abducens canals (Gans, 1978). Phylogenetic or func- tional reduction of the eyes in relation to fossorial habits could account for the absence of the abdu- cens canals in some ophidians as well. 15. [74] Mandibular condyle formed only by articular bone. In ancestral amniotes, the quad- rate articulates with the articular medially and the surangular laterally, such that one-fourth to one- third of the mandibular condyle is formed by the surangular bone (Gauthier, 1984). This condition occurs in synapsids (e.g., Sphenacodon), Testudines (e.g., Chelonia), and archosauromorphs (e.g., Champsosaurus). This region is not preserved, or it has not been reported in, extinct lepidosauro- morphs. In rhynchocephalians and squamates, however, the surangular forms only the outer rim of the articulation, with the condyle formed by the articular alone. Morphologists have long been mistaken in maintaining that a quadrate-articular attachment is ancestral for amniotes. This reflects the assumption, often perpetuated in textbooks, that "lizards" are relatively unmodified examples of the "primitive" reptilian type. Such treatments generally ignore character transformations within lineages not leading directly to mammals or birds. 16. [75] Teeth attached superficially to jaw. To judge by the condition seen in araeoscelidans I (Reisz, 1981), archosauromorphs (Gauthier, 1983), and younginiforms (Gow, 1975), the teeth are I set in shallow sockets in the jaw in ancestral lepidosauromorphs (subthecodont). The mode of tooth implantation is unknown in Paliguana*, Palaeagama*, and Saurosternon*. Robinson (1962) and Evans (1980), however, describe the teeth of kuehneosaurs as subpleurodont, being set in shal- low depressions. Our observations on Kuehneosaurus, however, indicate that the teeth are subthec- odont, i.e., lying in shallow depressions. The teeth of rhynchocephalians and squamates are at- tached superficially to the medial side of the jaw. Superficially attached teeth that undergo normal replacement and are not surrounded by extensive bone of attachment occur widely in squamates and are commonly referred to as pleurodont. Those of sphenodontidans, agamids*, and chamaeleontids I are further transformed in that they are generally not replaced and become extensively invested by attachment bone, blurring their individuality (acrodont). Acrodonty and pleurodonty are often de- fined by the position of the teeth relative to the jaws; acrodont teeth are said to attach apically, while pleurodont teeth attach to the medial surface of the jaws. Unfortunately, under these defini- f tions, the teeth of most agamids* are pleurodont, while those of Gephyrosaurus are closer to being acrodont. Because there seems to be general agreement that agamid* teeth are acrodont, and be- 38 Phylogenetic Relationships of the Lizard Families cause those of Gephyrosaurus have been described as pleurodont (Evans, 1980), the definition of acrodonty should be based on the general lack of replacement and the investment of the teeth by at- tachment bone, rather than on the position of the teeth with respect to the jaws. It is inaccurate to describe the teeth of lepidosaurs as either pleurodont or acrodont, as if both were unrelated transfor- mations of some more general mode of implantation, because acrodonty is simply a further modifi- cation of pleurodonty. Accepting superficial attachment as the ancestral condition in lepidosaurs requires the shallowly-socketed teeth of ophidians and the more deeply-socketed teeth of mosasaurs to be secondary modifications (Edmund, 1969). 17. [78] Accessory intervertebral articulations (zygosphenes and zygantra). There are no acces- sory intervertebral articulations in araeoscelidans (Reisz, 1981). Such articulations are also absent in archosauromorphs ancestrally, although accessory articulations have arisen separately in some rauisuchians on the one hand, and in saurischian dinosaurs on the other (Gauthier, 1984). Accessory intervertebral articulations are also absent in younginiforms ancestrally, although youn- ginoids have uniquely modified articulations between the neural spines (Currie, 1981a). Carroll (1975) described zygosphenial joints in Saurosternon*, although Evans (1981) was unable to con- firm his observation, and Carroll (1977:371, fig. 8) did not figure them. As Evans (1981) pointed out, it is difficult to see feebly developed zygosphenial joints in articulated skeletons such as those of Palaeagama* and Saurosternon*. The isolated vertebrae of kuehneosaurs lack zygosphenial joints (Colbert, 1970; Evans, 1981). Zygosphenial joints are present in Gephyrosaurus and Sphen- odon (Evans, 1981), and they are thus considered to have been present in the ancestral rhynchoce- phalian. Among squamates there is considerable variation in this character. Squamate zygospheni- al joints may vary from a simple, weakly developed condition, in which the prezygapophysial ar- ticular surfaces extend medially onto the neural arch above the spinal cord, to the separately encap- sulated and prominently developed structures characteristic of snakes (Hoffstetter and Gase, 1969; Winchester and Bellairs, 1977). As noted by R. Etheridge (pers. comm.), this character appears to vary with size among iguanians; with some exceptions in taxa that possess such articulations, the larger the lizard the more prominent the zygosphenial joints. This generalization also holds for several groups of autarchoglossans; lacertids, for example, appear to vary in a similar fashion. Compared to iguanians, the exceptions are much more conspicuous in autarchoglossans, because even the smallest snake or gymnophthalmid has prominently developed zygosphenial joints. They are absent in amphisbaenians and the overwhelming majority of anguimorphs, regardless of size (Hoffstetter and Gase, 1969). Variation in this character makes it difficult to determine the level at which it is a synapomorphy. Nevertheless, zygosphenial joints are present in rhynchocephalians, present or absent in iguanians, and present or absent in autarchoglossans. Accordingly, it is sim- pler to accept the hypothesis that zygosphenial joints are a lepidosaur synapomorphy, and that they have been lost secondarily in several squamate groups. 18. [80] Caudal autotomy septa present. There are no autotomy septa in the caudal vertebrae of araeoscelidans (Reisz, 1981), or those of archosauromorphs (Gauthier, 1984) and lepidosauro- morphs ancestrally. These structures are absent in younginiforms (Currie, 1981), Saurosternon* (Carroll, 1975a), and kuehneosaurs (Evans, 1981). In Paliguana* and Palaeagama* only the non- autotomic base of the tail is preserved. Except for Homoeosaurus, all other rhynchocephalians with the tail preserved display intravertebral fracture planes (Howes and Swinnerton, 1901; Cocude- Michel, 1963). Winchester and Bellairs (1977) have recently reviewed the development of this character in a few squamate species. To judge from their examples, it seems that the fragile, regen- erable tail that is characteristic of lepidosaurs forms late in development, resulting from a complex series of events yielding an intravertebral septum passing through the caudal ribs, dividing them into unequal portions. The distribution of this character has been reviewed by Etheridge (1967) and Hoffstetter and Gase (1969). R. Etheridge (pers. comm.) observed a crocodilian that regenerated its LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 39 tail after it had been severed through a vertebra! centrum. In most taxa (such as crocodilians) that lack autotomy septa and do not normally regenerate the tail, tail loss is accomplished by interverte- bral separation. Etheridge therefore suggested that the development of a caudal fracture plane may facilitate regeneration. The issue is complex, however, because amphisbaenians retain fracture planes but do not regenerate the tail (Gans, 1978). Accepting caudal autotomy septa as a synapo- morphy of lepidosaurs requires secondary losses among several squamate groups. In some instanc- es, the loss of a fragile tail is intelligible from a functional standpoint, such as in "aigialosaurs" and mosasaurs, in which the tail is used as a sculling organ. In other instances, however, there ap- pear to be no obvious functional explanation for its loss, as in the case of acrodont iguanians. Caudal autotomy septa have arisen independently in mesosaurs and in some captorhinids (Carroll, 1982). 19. [91] Sternum indistinguishably fused during embryogeny. As noted above, the sternal plates are paired for much of postnatal ontogeny in younginiforms, becoming fiised to one another only late in post-hatching ontogeny. This character is not preserved in Paliguana*, Palaeagama*, and kuehneosaurs. As Carroll (1977) pointed out, the sternal plates are paired in Saurosternon*, thus indicating the immaturity of the specimen. In squamates and rhynchocephalians the sternal plates are indistinguishably fused to one another in the embryo (Howes and Swinnerton, 1901). 20. [90] Xiphistemum greatly reduced or absent. As noted above, the amniote xiphistemum extends far posteriorly and includes at least five poststemal ribs ancestrally. However, Sphenodon lacks a xiphistemum (G?nther, 1867), and squamates have a very reduced structure that is usually composed of no more than two ribs (R. Etheridge, pers. comm.; Lecuni, 1968b). For reasons dis- cussed above, this character is seldom determinable in fossils. It seems likely that the reduced xi- phistemum is associated with the transformation of the sternum at the level of Lepidosauromor- pha; however, because this character is at present determinable only in Recent squamates and Sphenodon, we will be conservative in treating it as a synapomorphy of lepidosaurs alone. 21. [115] Ectepicondylar groove transforms into a foramen in embryo. As noted above, there is an ectepicondylar foramen in Palaeagama*, Saurosternon*, and kuehneosaurs, but in the absence of earlier ontogenetic stages of these taxa, it cannot yet be determined when enclosure takes place during ontogeny. This character may apply to a more inclusive group. Until more is known, however, it will be considered a synapomorphy of the lepidosaurs, in living members of which the I foramen is fully formed in the embryo (Howes and Swinnerton, 1901). 22. [104 & 105] Symmetrical metacarpals. In diapsids ancestrally, the metacarpals increase in size from first to fourth, with metacarpal five being subequal to the first in length (Fig. 4A). In addition, the second, third and fourth metacarpals are stoutest, with the first and fifth being consid- erably less robustly constructed. Paliguana* is unknown in this regard, but both Saurosternon* and Palaeagama* are ancestral in this respect (Carroll, 1975). In kuehneosaurs most of these rela- tions are maintained, except that metacarpal four is subequal to the third in length (see 3.3.8). In rhynchocephalians and squamates, however, metacarpals two and four are reduced, metacarpal four is shorter than the third, and the first and fifth metacarpals are more stoutly constructed (Fig. 4C); moreover, the first and fifth manal digits are set slightly below the other digits. The metacarpalia are, thus, more symmetrical, a condition that is considered a synapomorphy of Lepidosauria. This character is usually associated with the shortening of the fourth digit, which in most lepidosaurs is not much longer than the third. Archosaurs have an analogous condition, but their hands are other- wise different In groups that develop webbed hands, such as some tangasaurs and champsosaurs, the metacarpals may be symmetrical. In these groups, however, the entire manus is relatively broader and more symmetrical. 23. [123] Pelvis fused in adults. The ilium, ischium and pubis are fused to one another in adult araeoscelidans (Fig. 6A), as in amniotes ancestrally. However, the pelvic elements remain distinct in all archosauromorphs except for birds other than Archaeopteryx* (Fig. 6B and Romer, 40 Phylogenetic Relationships of the Lizard Families FIGURE 7. Lateral view of right knee (A and B) and distal view of left fibula (C and D). A, Caiman sclerops (archosaur); B, Sauromalus obesus (iguanian); C, Sphenodon punctatus (sphenodont); D, Tupinambis teguixin (autarchoglossan). 1956), as well as in younginiforms (Currie, 1981b). Accordingly, an unfused pelvis in the adult is considered a synapomorphy ~ via paedomorphosis ~ of Sauria. This character is not preserved in Paliguana*, but in both Palaeagama* and Saurosternon* the pelvic elements are separate. Unfortu- nately, the latter two taxa are represented by single, subadult specimens, thus rendering this charac- ter indeterminable. Kuehneosaurs evidently have the ancestral condition because fused pelves have not been reported in Kuehneosauras, which is represented by large samples including adults (Evans, 1981, and pers. comm.). In contrast, in all rhynchocephalians, and in all squamates with unreduced pelves, the ilium, ischium and pubis are indistinguishably coossified in fully mature in- dividuals (pers. obs.). In Gephyrosaurus, however, most specimens of pelvic bones are dissociat- ed, suggesting that it may retain the ancestral condition in this respect (S. Evans, pers. comm.). On the contrary, we would suggest that most are not fused because they are not yet fully mature. 24. [116, 120] Ilium forms less than 80-85% of surface area of acetabulum, and iliac blade narrow and more or less steeply inclined posterodorsally. In amniotes ancestrally the iliac blade is relatively small and is steeply inclined posterodorsally, and the ilium forms no more than 60-65% of the surface area of the acetabulum (Gauthier, 1984). This condition is retained in diapsids ances- trally (Reisz, 1981; see Fig. 6B). The ilium is transformed in saurians; it forms 80-85% of the surface area of the acetabulum, and the iliac blade is enlarged, roughly triangular in lateral view, extending to the level of the posterior end of the ischium (Fig. 6B). This condition is present in archosauromorphs (Gauthier, 1984), younginiforms (Currie, 1980), and kuehneosaurs (Evans, 1981), while the ilium is either unknown or too poorly preserved to interpret in Paliguana*, Pa- laeagama*, and Saurosternon*. In view of this distribution, it appears that an ilium like that seen in Fig. 6B is ancestral for saurians and lepidosauriforms. Accordingly, the relatively thinner (dorsoventrally) and more or less steeply inclined iliac blade, and the reduced contribution of the ili- um to the acetabulum that are present in squamates (Fig. 6D) and rhynchocephalians (Fig. 6C), ap- pear to be reversals to the condition seen in amniotes ancestrally. The ilium is relatively elongate posteriorly in lepidosaurs, however, as it is in other lepidosauriforms and saurians. 25. [117] Ilium with pubic flange. A pubic flange is absent on the ilium in saurians and le- pidosauromorphs ancestrally, as shown by its absence in archosauromorphs, younginiforms, and kuehneosaurs (see Evans, 1981:112, fig. 34). This character is not preserved in Paliguana*, Sau- LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 41 rosternon*, and Palaeagama* (Carroll, 1975a). Unlike other lepidosauromorphs, however, rhyn- chocephalians (Evans, 1981) and squamates (pers. obs.) share a short, flat, more or less pointed flange of the ilium that passes anterodorsally as a lappet over the acetabular junction of the pubis. 26. [119] Anteromedial portion of pubis outtumed dorsally. As discussed by Carroll (1977), squamates and rhynchocephalians are unlike other diapsids in sharing a distinctive anteroventral curvature of the anterodorsal face of the pubis. 27. [124] Lepidosaur knee joint. In amniotes ancestrally, the femur and fibula meet end-to-end (Fig. 7A). Rhynchocephalians and squamates share a unique form of the fibular attachment; the flattened and pointed proximal epiphysis of the fibula lies in a vertically oriented, recessed area on the dorsal side of the fibular condyle of the femur (Fig. 7B). In addition, the distal condyles of the femur are subequally developed in saurians ancestrally, whereas in lepidosaurs they are markedly asymmetrical in development, with the tibial condyle being the larger of the two. According to Rewcastle (1980), chamaeleons are unique among Recent squamates in their possession of more symmetrically-developed femoral condyles. Mosasaurs also appear to have more symmetrically- developed condyles. Based on other synapomorphies, and because of their specialized limbs, mosa- saurs and chamaeleons are considered to be secondarily derived in this regard. Because the osseous portion of the lepidosaur knee-joint is formed by the femoral and fibular epiphyses, and because the presence of epiphyses in extinct lepidosauromorphs is uncertain (see no. 33 below), it cannot yet be determined if this synapomorphy applies to a more inclusive group than lepidosaurs. 28. [127] Lateral centrale of pes fuses to astragalus in embryo. The lateral pedal centrale is a separate element in diapsids (Fig. 8A) and saurians ancestrally (Gauthier, 1984). This condition occurs in younginiforms (Currie, 1981b) and Saurosternon* (Fig. 8B), but this region is not pre- served in Paliguana*, Palaeagama*, or kuehneosaurs. In contrast, the centrale becomes fused to the astragalus in the embryos of squamates (Fig. 8D) and rhynchocephalians (Howes and Swinnerton, 1901; and see Fig. 8C). This synapomorphy may apply to a more inclusive group, but until more is known about this character it will be considered to apply to lepidosaurs only. As noted by Cruickshank (1979) and others, archosaurs also incorporate the centrale into the astragalus in em- FIGURE 8. Dorsal view of left pes. A, Petrolacosaurus kansensis (araeoscelidan); B, Sauroster- non bainii* (lepidosauriform); C, Sphenodon punctatus (sphenodont); D, Iguana iguana (iguanian). A after Reisz (1981); B, D after Carroll (1977); C after Romer (1956). 42 Phylogenetic Relationships of the Lizard Families bryos. Thus, the loss of a separate pedal centrale early in development is considered to have arisen convergently in archosaurs and lepidosaurs. 29. [126] Astragalus and calcaneum fused prior to attainment of maximum adult size. The astragalus and calcaneum are separate elements in diapsids ancestrally (Fig. 8A). They are separate elements in most archosauromorphs, although they may be fused in some archosaurs, such as pter- osaurs, and in birds and some other theropods (Gauthier, 1984). The astragalus and calcaneum are separate elements in younginiforms (Currie, 1982), Palaeagama*, and Saurosternon* (Carroll, 1975a), but this region is not preserved in Paliguana*. The elements are separate in juvenile lepid- osaurs, but they fuse to one another prior to the fusion between the scapula and coracoid (O. Riep- pel, pers. comm., cites an "adult" Varanus in which this fusion does not occur. If accurate as to assessment of age, we expect such variation to be rare). Evans (1981) stated that these elements are paired in kuehneosaurs, but provided no further evidence as to the stage of development repre- sented by the particular specimen observed; it could well be an immature individual in which the fusion had not yet taken place. Until more is known, this synapomorphy will be considered to ap- ply only to squamates and rhynchocephalians. Preliminary observations suggest that coossifica- tion between these elements may occur earlier in the ontogeny of squamates than in rh5nichoce- phalians. Mosasaurs appear to be paedomorphic in that fusion of the astragalus and calcaneum, along with other fusions associated with the cessation of growth, are unknown among the numer- ous fossil representatives of the group (Russell, 1967). 30. [128] Loss of first distal tarsal, thus bringing astragalocalcaneum into contact with first metatarsal. In araeoscelidans (Fig. 8A) and in saurians ancestrally (Gauthier, 1984) the first distal tarsal is ossified and separates the first metatarsal from the astragalus (Fig. 8B). This condition is seen in younginiforms and in Saurosternon* although this region is not preserved in Palaeagama*, Paliguana*, or kuehneosaurs. This character may apply to a more inclusive group, but until more is known, loss of the first distal tarsal is considered a lepidosaur synapomorphy. Among archosau- romorphs, the first distal tarsal also fails to ossify in archosaurs (Gauthier, 1984). Thus, the ab- sence of the first distal tarsal in archosaurs and lepidosaurs is considered convergent. 31. [130] Fifth distal tarsal absent. In diapsids a discrete fifth distal tarsal is present ancestral- ly (Fig. 8A). This element is unknown in all archosauromorphs (Gauthier, 1984). According to Harris and Carroll (1975) and Currie (1981b), the fifth distal tarsal fuses to the fourth during post- natal ontogeny in tangasaur younginiforms. Unfortunately, the ankles are unknown in Acerosodon- tosaurus, Youngina, Paliguana*, Palaeagama*, and kuehneosaurs. Saurosternon* does, however, retain a discrete fifth distal tarsal (Fig. 8B). The fifth distal tarsal is not a discrete element in rhyn- chocephalians (Fig. 8C) and squamates (Fig. 8D). It has been suggested that the fifth distal tarsal either has been lost, fused to the fourth distal tarsal, or fused to the fifth metatarsal. However, the available developmental evidence neither confirms nor denies these hypotheses (Robinson, 1975). Given the available evidence, the simplest resolution to this question is to posit that the "loss" of the fifth distal tarsal applies to saurians generally. The presence of this element in Saurosternon* would then be explained as either an evolutionary reversal or as further evidence that Saurosternon* is a subadult in which the suspected fusion has yet to occur. In view of the incomplete data, how- ever, we prefer to consider Saurosternon* to have retained the ancestral diapsid condition. This re- quires three separate "losses" of the fifth distal tarsal in diapsids: once in archosauromorphs, once in younginiforms, and once in lepidosaurs. Future finds in development and paleontology may ul- timately allow us to determine if "loss" means the same thing in each of these groups. 32. [132] Hooked fifth metatarsal. The fifth metatarsal is not hooked in diapsids ancestrally (Fig. 8A), but the element is apomorphic in saurians ancestrally in that it is short and broad-based (e.g.. Fig. 8B). In all archosauromorphs the fifth metatarsal is flattened in the plane of the pes and hooked (i.e., the element extends medially to contact the fourth distal tarsal). This is evidendy a case of convergence with the condition seen in lepidosaurs, because among lepidosauromorphs, LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 43 younginiforms, Saurosternon* (Fig. 8B), and kuehneosaurs (Evans, 1985) retain the ancestral sau- rian condition. This is perhaps why, with the exception of the "hooked" shape, the fifth metatar- sals of archosauromorphs and lepidosaurs are otherwise so different. For example, in contrast to the condition in archosauromorphs, the fifth metatarsal in lepidosaurs has medial and lateral plantar tubercles, is inflected about its long axis, and its proximal surface is angulated in a characteristic fashion (see Robinson, 1975, for a more thorough description). Palaeagama* and Paliguana* are unknown in this regard. 33. [137] Secondary ossification centers form on ends of all long bones and in several muscle attachments, such as those inserting on retroarticular process, neural arches, basioccipital tubera, posterior end of ilium and several other points on pelvis, as well as in knee and elbow joints. The presence of secondary ossification centers in lepidosaurs has been recognized for many years (e.g., Albrecht, 1883; Dollo, 1884). Moreover, their developmental and histological similarity to those of mammals is also widely recognized (Haines, 1969). Yet there is much speculation and little data relating to their developmental origin and function. Aside from the secondary center associated with the cnemial epiphysis of the tibia in po%t-Archaeopteryx* birds, no other diapsids possess this synapomorphy. As in the sternum, calcification within long bone epiphyses begins early in development (Haines, 1969). Subsequent replacement of the calcifications by endochondral ossifi- cation takes place in the adult (de Ricqies, 1976). According to Haines (1969), epiphyses are ab- sent in snakes. Because all other data clearly places snakes within the lepidosaurian squamates, those snakes are considered secondarily derived in this regard. The hypothesized reversal in snakes appears to have occurred in mosasaurs as well (Russell, 1967), although epiphyses are present in closely related taxa such as Pontosaurus lesinensis (Komhuber, 1873) and Opetiosaurus bucchichi (Komhuber, 1901). Gross examination may not allow one to discriminate between the long bones of fully adult lepidosaurs and those of other diapsids. To accomplish this would require histologi- cal examination, which provides a ready means of distinguishing between the two developmental mechanisms by which the same ends may be achieved in diapsid long bones (Enlow, 1969; Haines, 1969). The easiest and surest means of determining the presence of epiphyses in fossil di- apsids depends on finding a specimen with long bones at the stage of development in which endo- chondral ossification has spread from the metaphysis into the calcified epiphysis, but prior to the complete epiphysial-diaphysial coossification. A good example of this is the olecranon epiphysis that may be seen on the ulna of the sphenodontidan Kallimodon (Cocude-Michel, 1963). In this specimen, there is a metaphysial area, represented by a thin line, that separates the fully-formed I olecranon process from the diaphysis. In diapsids that do not have epiphyses, the olecranon pro- cess on the ulna is readily recognized to be of "normal" derivation when the distal extremity of the process is unfinished, indicating the presence of the usual growth cartilage (e.g., see the olecranon on the ulna of araeoscelidans; Reisz, 1981:44, fig. 20B). Ontogenetic series of archosauromorphs (i.e., Tanystropheus; Wild, 1973), and younginiforms (Currie, 1981b) show that epiphyses are cer- tainly lacking in archosauromorphs and lepidosauromorphs ancestrally. Unfortunately, pertinent lepidosauromorphs are not well enough known to be certain of the level at which this character di- agnoses a monophyletic group. Paliguana* lacks a postcranial skeleton, and Saurosternon* and Palaeagama* appear to be subadult specimens in which the epiphyses are not ossified. Carroll ( 1977) and Evans ( 1981 ) argued that the form and texture of the ends of the long bones of Sauros- ternon* and Palaeagama* indicate that epiphyses would have been present in adults of these taxa. Icarosaurus is represented by a subadult specimen. The form and texture of the diaphysial surfaces, together with the gaps between the long bones, again suggest that epiphyses may have been present in kuehneosaurs as well. Unfortunately, the known elements of Kuehneosaurus do not clarify the issue. In this taxon, the referred limb elements are from fully grown adults, so the mode of development is not determinable (S. Evans, pers. comm.). Thus, there appears to be 44 Phylogenetic Relationships of the Lizard Families some reason to suspect that epiphyses are more widely distributed among lepidosauromorphs than has been believed previously. In the absence of unequivocal evidence to the contrary, however, we consider this character a lepidosaurian synapomorphy. 34. [139] Many structures that normally ossify or remain cartilaginous in diapsids tend to cal- cify well before attainment of maximum adult size in squamates and rhynchocephalians. Among the structures that calcify are the epiphyses, trach?al rings, hyoid apparatus, epicoracoid, suprasca- pula, hypoischium, sesamoids, sternum, sternal ribs, xiphistemum, inscriptional ribs, and rem- nants of the interorbital cartilages. This complex is treated as a single character following the sug- gestion of Haines (1969) that the origin of the calcifications (not ossifications) probably has a common developmental basis. However, some calcifications, such as most sesamoids, may vary in number in lepidosaurs (Rewcastle, 1980). Our observations indicate that at least some of the variation relates to differences in the temporal sequence in which the calcifications appear during ontogeny. Previous workers may have been comparing individuals that differed in stage of devel- opment. Preliminary analysis indicates that at least some calcifications may not fit this category; nevertheless, the tendency of cartilages, tendons, ligaments, and sheets of connective tissue to cal- cify early in postnatal development is considered a synapomorphy of lepidosaurs. This synapo- morphy is usually indeterminable in fossils because calcified cartilage disintegrates in macerated specimens (see above). 35. [138] Most of skeleton composed of dense, lamellar, avascular bone of periosteal origin. As described by Enlow (1969) and de Ricql?s (1976), the bones of squamates and rhynchocephali- ans may be distinguished from those of other amniotes by the fact that the compact bone of perios- teal origin is virtually avascular. In addition, there is very little development of cancellous trabec- ulae in the mid-diaphyses of the long bones. Unfortunately, our knowledge of the distribution of this character is very incomplete. Because it appears to be correlated with small size and associated patterns of growth, we suspect that this character will be found to apply to a more inclusive group within lepidosauromorphs. Nevertheless, until evidence to the contrary is available, we consider it a lepidosaur synapomorphy. SQUAMATA Definition: The most recent common ancestor of Iguania and Autarchoglossa, and all of its de- scendants (see Estes et al., 1988). In this section, we discuss the synapomorphies of squamates, using rhynchocephalians, kueh- neosaurs, Paliguana*, Saurosternon*, Palaeagama*, and younginiforms as successively more re- mote outgroups. We recognize two principal groups of Squamata, Iguania (Cuvier, 1807) and Au- tarchoglossa (sensu Gauthier, 1982; includes Autarchoglossa of Camp, 1923, plus gekkotans and snakes; also equivalent to Scincogekkonomorpha of Sukhanov, 1961, 1976, plus snakes and am- phisbaenians). Iguania includes those taxa referred to Iguanidae*, Agamidae* and Chamaeleontidae. Iguania, Acrodonta (Agamidae* + Chamaeleontidae), and Chamaeleontidae are readily diagnosible; there is still no positive evidence for either the monophyly or paraphyly of Iguanidae* or Agami- dae*; they are therefore considered to be metataxa (see Estes, 1983; Estes et al., 1988; Etheridge and de Queiroz, 1988), and we will not concern ourselves further with iguanian subgroups. As here constituted, all other squamates (including snakes, dibamids, and amphisbaenians) are consid- ered autarchoglossans (see Estes et al., 1988, for subdivision of Autarchoglossa). "Eolacertilia" and "Lacertilia" are paraphyletic, and they will not be used in this discussion. Compared to extant mammals, turtles, archosaurs, and Sphenodon, squamates possess the fol- lowing synapomorphies of the soft anatomy. Diagnosis: 1. [169] Presence of discrete m. bursalis. Unlike other amniotes, squamates have LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 45 developed a discrete m. bursalis, a division of the m. retractor bulbi present generally in tetrapods (Underwood, 1970). 2. [144] Reduced cartilaginous component of anterior braincase and interorbital septum, with large fenestrae and significant membranous component. These regions are largely cartilaginous, with concomitandy smaller membranous components and fenestra, in other amniotes (Bellairs and Kamal, 1981). 3. [135] Interrupted third branchial arch. The third branchial arch is continuous in other am- niotes (Kluge, 1983). 4. [170] Femoral and preanal organs present. Femoral and/or inguinal epidermal holocrine follicular glands (terminology of Moody, pers. comm.) are unique to squamates among amniotes (Kluge, 1983). 5. [146] Pallets on ventral surface of tongue tip. Tongue pallets are absent in all amniotes except for squamates (Schwenk, 1988). 6. [159] Paired, evertible hemipenes in males. The amniote penis has been lost in lepido- saurs, and the hemipenes of squamates have long been recognized as diagnostic (e.g., Oppel, 1811). 7. [151] Lacrimal duct extends far anteriorly to become associated with duct of Jacobson's or- gan. The lacrimal duct is associated with the posterior end of choanal groove in Sphenodon, as in amniotes ancestrally (Bellairs and Boyd, 1950). 8. [147] Jacobson's organ completely separate from nasal capsule and develops a fungiform body. Jacobson's organ is typically a simple diverticulum of nasal capsule, and no other tetrapod possesses the squamate fungiform body (Parsons, 1970). 9. [148] Extensive development of sensory epithelium in Jacobson's organ. Squamates are distinctive in the degree of development of the sensory epithelium lining Jacobson's organ (Pratt, 1948). 10. [149] Rotation of Jacobson's organ and associated paraseptal cartilages. Jacobson's organ apparently rotated ninety degrees about its longitudinal axis, placing the paraseptal cartilage medi- ally with the duct of the organ opening ventrally into the oral cavity (Malan, 1946). In Spheno- don, the paraseptal cartilage forms a trough-like support for the floor of Jacobson's organ, the duct of which is positioned laterally and opens into the medial wall of the choana. 11. [150] Enlarged lateral nasal gland lodged in cavum conchale. In other reptiles the lateral nasal gland is small and not enclosed in cavum conchale (Malan, 1946; Pratt, 1948). 12. [145] Loss of caruncle. The presence of both an egg tooth and caruncle in amniotes ances- trally (Hill and deBeer, 1949), indicates that squamates, which possess only the former (Edmund, 1969), have lost the latter. 13. [154] Multiple interdigitations of mm. intermandibularis and mandibulohyoideus. These muscles are non-interdigitating in amniotes ancestrally (Camp, 1923; Rieppel, 1978). 14. [156] Complete, rather than partial, separation of m. depressor mandibulae from m. epis- temocleidomastoideus. These muscles are at least partly confluent in amniotes generally (Rieppel, 1978). 15. [155] Facialis nerve no longer participates in innervation of m. intermandibularis. The n. facialis participates in the innervation of this muscle in amniotes ancestrally (Rieppel, 1978). 16. [157] Fibers of m. clavodeltoideus extend to ventral surface of clavicles. This muscle is typically confined to the dorsal surface of the clavicles (Peterson, 1973). 17. [158] Meniscus in knee joint a single plate pierced by cruciate ligament The meniscus of the knee generally forms separate lateral and medial crescents on either side of cruciate ligament (Haines, 1942). 18. [143] Prominent perilymphatic sac. The perlymphatic duct does not form a prominent sack in the recessus scalae tympani in amniotes ancestrally (Baird, 1970). 46 Phylogenetic Relationships of the Lizard Families 19. [142] Postembryonic position of cochlear duct facing laterally. The cochlear duct faces ventrally in amniotes ancestrally and in the embryos of all squamates (Baird, 1970). The ancestral condition applies to some snakes as well, but this is considered a paedomorphic reversal. 20. [141] Loss of pars tuberalis of adenohypophysis. The pars tuberalis is present is amniotes ancestrally (Wingstrand, 1951). 21. [152] Interhyal either ligamentous or absent. The interhyal is cartilagenous in amniotes ancestrally (de Beer, 1937). 22. [153] Saccular ovaries. Solid ovaries are present in amniotes ancestrally (Porter, 1972). Most of the abovementioned synapomorphies are not determinable in fossils. The following combination of osteological synapomorphies will, however, separate squamates from rhynchoce- phalians and from all other lepidosauromorphs as well. The first two of these relate to the remarkable transformations of the septomaxilla and its as- sociated soft tissue (Malan, 1946). There is clear evidence that these transformations are intimate- ly related both functionally and developmentally (e.g., Bellairs and Boyd, 1950). Thus, as with the case of the tympanum and the lateral conch on the quadrate, these characters are at least partly deter- minable in fossils. 23. [38] Septomaxilla with posteroventral projection extending towards dorsal surface of vom- er to form posterior margin of duct of Jacobson's organ, and anterior and medial margins of duct formed by notch in vomer. These osteological characters are associated with fusion of the sides of the choanal passage to separate the opening of Jacobson's organ from the functional choana in the embryo (Fuchs, 1908). 24. [37] Septomaxilla invests enormously enlarged vestibule to roof Jacobson's organ dorsally and floor nasal passage ventrally. In diapsids the septomaxilla is limited to the posteroventral edge of the fenestra exonarina ancestrally (Gauthier, 1984). According to Malan (1946), the enormous size and posterior prolongation of the vestibule carried the septomaxilla to the interior of the nasal capsule, so that the septomaxilla forms a secondary roof above Jacobson's organ. 25. [62] Premaxillae fused in embryonic developmental stages. Because the premaxillae are paired in younginiforms (Fig. IB), kuehneosaurs (Fig. IC), and rhynchocephalians (Fig. ID), this condition is thought to be ancestral for lepidosaurs. This character cannot be determined in Pali- guana*, Palaeagama*, or Saurosternon*. In squamates, however, the premaxillae become fused to one another prior to hatching. Based on other data (Estes et al., 1988), the paired premaxillae present in most gekkotans and scincids are considered to have arisen secondarily and independently. 26. [2] Reduced nasals. In lepidosaurs ancestrally the greatest width of the nasals exceeds that of both nares (= fenestra exonarina). This condition occurs in kuehneosaurs (Fig. IC) and in rhyn- chocephalians (Fig. IF). Neither Saurosternon* nor Palaeagama* can be interpreted in this regard. Although the tip of the snout is missing in Paliguana*, broad nasals are indicated by the preserved impressions of the nasal bones (Carroll, 1977). Younginiforms and Gephyrosaurus may be excep- tions to this generalization. Youngina provides the best known example of the younginiform skull. As is evident from illustrations in Gow (1975), most specimens show varying degrees of postdepositional deformation. Perhaps this accounts for the nasals of Youngina being wider (Fig. IB), or narrower (Gow, 1975), depending on the reconstiiiction. Nevertheless, the available evi- dence indicates that younginiforms in general seem to have elongate snouts with concomitantly el- ongate and narrow nasals. By comparison, araeoscelidans (Fig. 1 A), archosauromorphs ancestrally (Gauthier, 1984), and most lepidosauromorphs are relatively short-snouted. Thus, we have inter- preted elongate snouts as a younginiform synapomorphy (see diagnosis above). The problem of accurate reconstruction is more acute for Gephyrosaurus because its remains are dissociated and dis- articulated. In tiie reconstruction of Gephyrosaurus (Fig. ID) given by Evans (1980), the nasals are relatively narrow compared to those of other rhynchocephalians. This appears to be a unique LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 47 attribute of Gephyrosaurus, because S. Evans (pers. comm.) assures us that the reconstruction leaves no margin for error in this case. Modification of the snout in most squamates leaves them with relatively smaller and narrower nasals than occur in lepidosauromorphs generally (Fig. IJ). It appears that in squamates the facial process of the maxilla is more prominently developed, thus taking over some of the role of the nasal in roofing the snout. There are, however, several squa- mates in which this is not the case. Some iguanines among iguanians and polyglyphanodontine teiids among autarchoglossans have relatively broad nasals compared to those of their close rela- tives; both groups are composed of relatively large lizards. Another class of exceptions to the gen- eral case is provided by fossorial autarchoglossan squamates. In these groups, the snout plays an important role in entering the substrate and, like the rest of the skull, it is compact, streamlined and robustly constructed. Although fossorial squamates may have broad nasals, it seems unlikely that this results from simple retention of an ancestral state. Rather, the broad nasals, like virtually every other bone in the skull, are more likely to reflect the constraints imposed by small size and subterranean life. In spite of the cautions noted above, relatively reduced nasals that are not as broad as the distance across the nares are considered a synapomorphy of Squamata. 27. [3] Frontoparietal suture more or less transverse in dorsal view and broader than nasofron- tal suture. In lepidosaurs ancestrally the widths of the nasofrontal and frontoparietal sutures are subequal. In addition, the frontoparietal suture is roughly W-shaped, or shaped like an inverted U. The ancestral condition occurs in younginiforms (Fig. IB), Paliguana* (Carroll, 1975a), kuehneo- saurs (Fig. IC), and rhynchocephalians (Fig. IE). This region is not adequately preserved in Pa- laeagama*, and the skull is absent in Saurosternon*. Squamates have long been recognized as unique among reptiles in the shape and width of the frontoparietal suture. Most authors, however, apparently have observed this suture only in articulated skulls, because disarticulated skulls reveal that the frontoparietal suture is seldom straight, particularly ventrally. Nevertheless, the shape of this suture in dorsal view is striking, and this observation has engendered much speculation about its possible role in intracranial mobility. The available experimental evidence speaks against many of the previous interpretations (Smith, 1980), however, and points to the difficulties inher- ent in determining function from structure alone. Some squamates, such as gymnophthalmids, la- certids, some cordylids, and most mosasaurs, lack straight frontoparietal sutures, although there are exceptions in each group. The most conspicuous exceptions to the general case, however, are the amphisbaenians and ophidians. Although the frontoparietal sutures of ophidians and amphisbaeni- ans are not straight, neither are they ancestral. In the former group, the frontoparietal suture is more nearly U-shaped when viewed from behind (see Estes et al., 1970), and in the latter group, the bones are deeply interdigitated (Gans, 1978). Neither of these conditions can be said to be an- cestral (e.g., Gephyrosaurus, Fig. ID; Kuehneosaurus, Fig. IC). More importantly, the evidence presented by Estes et al., 1988) indicates that snakes and amphisbaenians are autarchoglossans, most of which have straight frontoparietal sutures. Because iguanians (Fig. 1J), the sister taxon of autarchoglossans (Fig. IK), also have straight frontoparietal sutures, this condition is considered synapomorphic for squamates. 28. [63] Parietals fused in embryo. With the exceptions of Gephyrosaurus (Fig. ID) and Pla- nocephalosaurus (Fig. IF), no other lepidosauromorphs save for squamates have fused parietals. Evans (1980) reported finding fused parietals in a specimen of Sphenodon, but we are unable to confirm this observation either in the five adults we examined, or in any of those figured or dis- cussed in the literature. Clevosaurus (Fig. IE), the sister taxon of Planocephalosaurus, has paired parietals, as do all other extinct sphenodontidans (e.g.. Fig. IH). Paired parietals are also present in all possible outgroups within lepidosauromorphs. Thus, the fused parietals of Gephyrosaurus on the one hand, and Planocephalosaurus on the other, are considered to have been separately derived. There is, of course, no information about when parietal fusion takes place during ontogeny in 48 Phylogenetic Relationships of the Lizard Families these taxa. The parietal bones may be fused or not in gekkotans and xantusiids. This fact, togeth- er with the fused parietals in all other autarchoglossans and iguanians, indicates that fused parietals are a synapomorphy of squamates (e.g., Figs. IJ, K). Although parietal fusion may take place at different times in the ontogeny of some gekkotans and xantusiids, in all other squamates it occurs in the embryo. 29. [14] Supratemporal displaced to a deep position, wrapping around ventral surface, and prominently developed on anterior face, of supratemporal process of parietal. As pointed out by Robinson (1967), in diapsids ancestrally the supratemporal lies in a superficial position posterolat- eral to the supratemporal process of the parietal (Fig. 1 A). Supratemporals are present in youngini- forms (Fig. IB) eaidPaliguana* (Carroll, 1977), but their presence cannot be determined in Palaea- gama* and Saurosternon*. Supratemporals are said to be absent in kuehneosaurs, Gephyrosawus, and all sphenodontidans save for Clevosaurus. They are absent in Sphenodon and their absence in sapheosaurs and Homoeosaurus is probable in view of the large sample of articulated skulls de- scribed in the literature and seen by us. Likewise, they are certainly present in the articulated skulls referred to Clevosaurus. Evans (1980) and Fraser (1982) reported that supratemporals are ab- sent in kuehneosaurs, Gephyrosaurus (the sister taxon of sphenodontidans), and Planocephalosau- rus (the sister taxon of Clevosaurus). S. Evans (pers. comm.) states that in the very well pre- served material of both these genera, no facet for a supratemporal is visible and that a single Ge- phyrosaurus specimen of an associated parietal and squamosal lacks an intervening supratemporal. Supratemporals are poor candidates for preservation because they are small bones in lepidosauro- morphs and because they are only loosely attached to the remainder of the skull. The history of ideas concerning the presence or absence in Youngina of these and other reduced bones (such as tab- ulars and postparietals) demonstrates that even in the case of articulated skulls these elements may not be preserved. Whether or not these bones are in fact lost in the above groups, it is clear that the size and position of the supratemporal is a squamate synapomorphy. The chief variations seen among squamates are the pronounced development of this element onto the braincase in mosasaurs (Russell, 1967), its posterior displacement on the supratemporal process of the parietal in iguanine iguanians (Etheridge and de Queiroz, 1988), and its elongation and singular role in quadrate suspen- sion in most snakes. 30. [15] Loss of ventral ramus of squamosal. With the exception of Clevosaurus (Fig. 3E), kuehneosaurs (Fig. 3B), and squamates (Figs. 3G,H), in all other lepidosauromorphs the squamo- sal has a prominent descending process that extends to the quadratojugal ventrally (Figs. 1 A,F). This character is not preserved in Palaeagama* or Saurosternon*. There is a descending squamosal process in Paliguana*, but poor preservation precludes a conclusion about its ventral extent (see the conflicting reconstructions of Broom, 1903, 1925, and Carroll, 1977). The ventral process of the squamosal is reduced in Clevosaurus, but remains well developed in all other rhynchocephali- ans. The condition in Clevosaurus is thus considered convergent, rather than a transitional character bridging the gap between the ancestral condition and that seen in either kuehneosaurs or squamates. Since Robinson (1962), it has been generally agreed that the absence of the descending process of the squamosal in kuehneosaurs is a synapomorphy uniting them with squamates. As argued above, however, rhynchocephalians, not kuehneosaurs, are closest to squamates among lepidosau- romorphs. Therefore, the loss of the ventral ramus of the squamosal is considered to have occurred independently in squamates and kuehneosaurs. 31. [16] Quadrate notched or fenestrate above for reception of posteroventral peg-like process of squamosal. As pointed out by Robinson (1967), in diapsids ancestrally the squamosal caps the quadrate dorsally. This condition is also present in younginiforms (Fig. lA), Paliguana* (Carroll, 1977), kuehneosaurs (Fig. IB), and rhynchocephalians (Fig. IC). This region is not preserved in Palaeagama* and Saurosternon*. The squamosal-quadrate articulation is uniquely modified in squa- LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 49 mates. Rather than capping the quadrate, the quadrate process of the squamosal is reduced to a peg that abuts the head of the quadrate, usually fitting into a fossa, fenestra, or notch (Robinson, 1967). 32. [12; 21] Quadratojugal and quadrate foramen absent. A quadrate foramen, bound by the quadratojugal laterally and the quadrate medially, is present in lepidosauromorphs ancestrally. This condition occurs in araeoscelidans (Reisz, 1981), archosauromorphs (Gow, 1975), and youngini- forms (Gow, 1975). This character cannot be determined in either Palaeagama* or Saurosternon*. Paliguana* is problematic in this regard. Broom (1903; 1925) concluded that both the quadratoju- gal and quadrate foramen are present. Carroll (1977), however, considered the quadratojugal to be absent and concluded that the Broom's "quadrate foramen" is an artifact of preparation. R. Reisz (pers. comm. 1987) informs us that the quadrate foramen is in fact present. Kuehneosaurs have lost both the quadratojugal and quadrate foramen, and Robinson (1962) considered this a synapo- morphic resemblance to squamates. It is not possible to determine at present if these characters were lost in the common ancestor of the kuehneosaur-lepidosaur group, and reevolved in the rhyn- chocephalians, or if they were lost independently in kuehneosaurs on the one hand, and in squa- mates on the other. Until more evidence is available, we posit the latter to have been the case. The anterior process of the quadratojugal forms the posterior part of the lower temporal arch, the loss of which is discussed below. 33. [11] Posterior process of jugal much reduced or absent. As noted in the discussion of rhynchocephalian relationships above, ancestral lepidosauromorphs have a jugal with a prominent process extending posteriorly to contact the quadratojugal below the midline of the lower temporal fenestra. This character is also found in araeoscelidans (Reisz, 1981), archosauromorphs (Gauthier, 1984), younginiforms (Fig. lA), and in a modified form in rhynchocephalians (Fig. IF). This character cannot be determined in Saurosternon* (Carroll, 1975a). Carroll (1975a) considered the posterior process of the jugal to be absent in Paliguana* and Palaeagama*, but Broom (1925) con- sidered it to be present. Because of this discrepancy, we consider this character to be indetermina- ble in both these taxa. Like the preceding character, this one cannot be optimized by minimizing evolutionary events, because a posterior process of the jugal is absent in kuehneosaurs (Fig. IB), present in rhynchocephalians, and absent in squamates (Fig. IH). In keeping with our interpreta- tion of the previous character, we thus consider the posterior process of the jugal to have been lost separately in kuehneosaurs and in squamates, rather than in their common ancestor. A parsimoni- ous resolution of this transformation sequence requires finding taxa that are closer to lepidosaurs than are the kuehneosaurs. In araeoscelidans (Reisz, 1981), archosauromorphs (Gauthier, 1984), and lepidosauromorphs an- cestrally (Fig. 3 A), the posterior process of the jugal met the anterior process of the quadratojugal to form a lower temporal bar. The bar is retained in younginiforms, but as noted above, this char- acter cannot be determined unequivocally in Palaeagama*, Paliguana*, or Saurosternon*. Among the remaining lepidosauromorphs, however, only members (^ the sphenodont-sapheosaur- Homoeosaurus clade have a complete temporal bar; it is absent in kuehneosaurs and squamates and is interrupted posteriorly in clevosaurs and Gephyrosaurus. Acceptance of our proposed hypothesis of relationships necessitates reinterpretation of the evo- lution of the lower temporal bar in lepidosauromorphs. The absence of this feature has been used to place kuehneosaurs (Robinson, 1967), "paliguanids" (Carroll, 1975a, 1977), dn? Gephyrosaurus (Evans, 1980) closer to squamates than to sphenodontidans. If rhynchocephalians, many of which possess a complete lower temporal bar, are the sister group of squamates, then the bar has either (1) been lost several times convergently within lepidosauromorphs (as suggested by e.g., Evans, 1980) or (2) been lost only once in this group but reevolved within sphenodontidans. We believe that the second hypothesis may be more reasonable. If the lower temporal bar were interrupted only once, only a single reversal is required in the common ancestor of sphenodonts, sapheosaurs, I 50 Phylogenetic Relationships of the Lizard Families and Homoeosaurus. Furthermore, the relationships within Lepidosauria are congruent with an hy- pothetical transformation series in which a complete lower temporal bar might have reevolved. Squamates would retain the ancestral lepidosaurian condition (based on outgroup comparison with kuehneosaurs) in which the lower temporal bar is absent; Gephyrosaurus and clevosaurs, with an enlarged posterior process of the jugal but an incomplete lower temporal bar, would represent inter- mediate stages in the transformation series. Finally, sphenodonts, sapheosaurs, and Homoeosaurus would exhibit the culmination of the transformation series in the reevolution of a complete lower temporal bar. 34. [22] Loss of vomerine teeth. Numerous small teeth are present on the vomers of youn- giniforms (Fig. 2A) and rhynchocephalians ancestrally (Fig. 2C). This character is not preserved in Paliguana.*, Palaeagama*, and Saurosternon* (Carroll, 1975a), or in Kuehneosaurus (Robinson, 1962). However, what can be seen of the palate in Icarosaurus (Colbert, 1970) suggests that the full complement of palatal teeth is present in kuehneosaurs. The presence of vomerine teeth in rhynchocephalians provides further corroboration for the view that vomerine teeth were present in lepidosaurs ancestrally. Aside from the anguine Pseudopus and the glyptosaurine sister taxa Arpad- osaurus and Melanosaurus among the anguids, no other squamates have vomerine teeth. All an- guine relatives of Pseudopus and all glyptosaurine relatives o? Melanosaurus and Arpadosaurus are like other anguids in lacking vomerine teeth. Accordingly, this character is considered apomorphic within both anguines and glyptosaurines (Gauthier, 1982). 35. [26] Pterygoids separated from one another and from vomers by apposition of palatines medially. In lepidosauromorphs ancestrally the pterygoids meet anteriorly and articulate with the vomers, thus separating the palatines on the midline. This condition is found in araeoscelidans (Reisz, 1981), archosauromorphs (Gauthier, 1984), and younginiforms (Fig. 2A), but is not pre- served in Paliguana*, Palaeagama* or Saurosternon* (Carroll, 1975a). What is known of the pa- late in kuehneosaurs indicates that they retained a pterygoid-vomer contact (Fig. 2B), yet their state of preservation does not allow a definite conclusion on this point. Retention of a pterygoid-vomer articulation in rhynchocephalians indictes that this contact was present in lepidosaurs ancestrally. Squamates have lost the pterygoid-vomer contact by interposition of the palatines on the midline. There are a few exceptions to this generalization. The pterygoids come near to or contact the vom- ers in polyglyphanodontine teiids (= Adamisauridae, Polyglyphanodontidae and Macrocephalosauri- dae of Sulimski, 1975; Estes, 1983). In addition, this character may be found in Shinisaurus (Xenosauridae), and in occasional specimens of Teius (Teiidae) and Uromastyx (Agamidae*) (Estes, 1983; pers. obs.). Our knowledge of the position of these taxa among other squamates indicates that these are cases of independent character reversal. 36. [39] Palatine reduced posteromedially, and pterygoid broadly exposed in suborbital fenes- tra. In lepidosauromorphs ancestrally the palatine has an extensive posteromedial component that closely approaches or contacts the ectopterygoid to exclude, or nearly exclude, the pterygoid from the suborbital fenestra. This condition occurs in araeoscelidans (Reisz, 1981), archosauromorphs (Gauthier, 1984), and younginiforms (Fig. 2A), but cannot be determined in Paliguana*, Palaeaga- ma* or Saurosternon*. The shape of the palatine is plesiomorphic in kuehneosaurs, but its rela- tion with the ectopterygoid is unknown (Fig. 2B). Rhynchocephalians are plesiomorphic (Figs. 2C-F), but squamates are apomorphic in that the palatine is usually broadly separated from the ec- topterygoid medially, and the pterygoid is consequently broadly exposed in the suborbital fenestra (Figs. 2G,H). Some squamates, such as some large iguanine iguanians and several autarchoglos- sans in which bones about the fenestra are modified (e.g., varanoids), appear to have reversed this- character. O. Rieppel (pers. comm.) notes that reversals may occur in some small squamates, par- ticularly fossorial forms with reduced eyes and concomitandy reduced suborbital fenestrae. 37. [60] Choanal fossa on ventral surface of palatine. The anterior margin of the palatine is LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 51 emarginate for the passage of the internal choana in lepidosauromorphs an?strally. This condition occurs in younginiforms (Fig. 2A), kuehneosaurs (Fig. 2B), and rhynchocephalians (Fig. 2C). The palate is not preserved in Paliguana*, Palaeagama*, and Saurosternon* (Carroll, 1975a). Squa- mates are further derived in having a relatively prominent fold in the body of the palatine that forms a fossa overlying the internal nares (Fig. 2H). 38. [20] Reduction of quadrate ramus of pterygoid and pterygoid ramus of quadrate, yielding a loose pterygoid-quadrate attachment formed by fibrous connective tissue. The quadrate and ptery- goid overlap one another to form a firm osseous union in lepidosauromorphs ancestrally. This con- dition is present in younginiforms (Fig. 2B), Paliguana* (Carroll, 1975a), kuehneosaurs (Fig. 2B), and rhynchocephalians (Fig. 2B), but cannot be determined in Palaeagama* and Saurosternon* (Carroll, 1975a). As noted above, rhynchocephalians are derived in having a deeply overlapping pterygoid-quadrate articulation (Fig. 2C). Squamates differ from all other lepidosauromorphs in that the osseous pterygoid-quadrate attachment is replaced by fibrous connective tissue (Robinson, 1967). Although this joint is still formed primarily by fibrous connective tissue, a secondarily de- veloped pterygoid process is present on the quadrate of lacertoids and Heloderma (Estes et al., 1988). 39. [34] Paroccipital process expanded distally and takes part in support of quadrate dorsally. The paroccipital process contacts the quadrate in ancestral lepidosaurs, the contact being present in kuehneosaurs, rhynchocephalians, and squamates (see 3.3.5). However, unlike the condition seen in other lepidosaurs, in squamates the paroccipital is expanded distally to play a larger role in sup- porting the quadrate (Romer, 1956). 40. [35] Stapes very slender. As argued above, a slender stapes is present in lepidosaurs ances- trally. Compared to that of Sphenodon, however, the stapes of squamates is even more slender (Romer, 1956). As noted above, stapes are known only in younginiforms and kuehneosaurs among extinct lepidosauromorphs. Many squamates in which the tympanum has been covered, re- duced, or lost have enlarged the stapes secondarily. If Sphenodon has similarly modified the tym- panic region (see p. 30) then it is possible that this synapomorphy applies to Lepidosauria. 41. [28] Columelliform epipterygoid with narrow base that does not contact quadrate. In le- pidosauromorphs ancestrally the epipterygoid is broad-based and extends posteriorly to contact the pterygoid process of the quadrate (Romer, 1956). This condition is found in younginiforms (Gow, 1975) and rhynchocephalians (Evans, 1980) but is not preserved in Paliguana*, Palaeagama*, Sau- rosternon*, and kuehneosaurs (Carroll, 1975; Evans, 1980). Squamates differ from other lepido- sauromorphs in that the quadrate-epipterygoid contact has been lost, and the epipterygoid is colu- melliform. Broom (1914:1076) studied the development of the pterygo-quadrate bar in lepidosaurs, and found that " [in squamates] the lower end of the quadrate is fixed to the lower end of the epip- terygoid by a small bar of cartilage almost exactly as in Sphenodon." 42. [32] Subdivision of embryonic metotic fissure to form recessus scalae tympani anteriorly and jugular foramen posteriorly; lateral aperture of recessus scalae tympani constitutes an analog of mammalian "fenestra rotunda". In amniotes and lepidosauromorphs ancestrally, there is no subdi- vision of the fissura metotica and consequently no "fenestra rotunda." The absence of a "fenestra rotunda" is, however, not the same as the absence of a subdivided metotic fissure, because the "fenestra rotunda" may be absent even though the metotic fissure is subdivided in a few squamates (0. Rieppel, pers. comm.). The ancestral condition occurs in araeoscelidans (Reisz, 1981), archo- sauromorphs (Gauthier, 1984), younginiforms (Gow, 1975), kuehneosaurs (Robinson, 1962; 1967), and rhynchocephalians (Romer, 1956). This character cannot be determined in Paliguana*, Palaeagama*, and Saurosternon* (Carroll, 1975a). Only squamates among diapsids possess this form of "fenestra rotunda," in which a hypertrophied portion of the ampuUary region of the otic capsule contacts the basal plate in late embryos, thus subdividing the anterior end of the metotic 52 Phylogenetic Relationships of the Lizard Families fissure (Kamal, 1971; who also notes that the process of subdivision may differ slightly in some snakes). Greer (1984) has recently suggested that because the "fenestra rotunda" is absent in di- bamids, they might be the sister taxon of all other squamates. However, serially-sectioned ?>t??a- mus skulls indicate that dibamids have a subdivided metotic fissure (O. Rieppel, pers. comm.). In any case, dibamids are here considered autarchoglossan squamates (Estes et al., 1988). Because a "fenestra rotunda" of this type is present in all other autarchoglossans, as it is in iguanians, modi- fications of this region in dibamids must be considered secondary. 43. [30] Vidian canal fully enclosed posterolaterally. According to de Beer (1937) and Evans (1980), the carotid artery and facial nerve of diapsids lie in an open channel along the lateral surface of the basisphenoid or parabasisphenoid before they enter the bone. Completion of the lateral wall of this channel by unossified connective tissue represents the ancestral condition. This condition is present in younginiforms (Gow, 1975), kuehneosaurs (Robinson, 1962), and rhynchocephalians (Save-Soderbergh, 1947; Evans, 1980), but is not determinable in Paliguana*, Palaeagama*, and Saurosternon* (Carroll, 1975a). Squamates differ from other lepidosauromorphs in that the lateral wall of the channel ossifies late in embryonic development to form a fully enclosed Vidian canal (Save-Soderbergh, 1947). Rieppel (1979) reported that the highly modified braincase of scoleco- phidian snakes shows some variation in this regard. Snakes lacking a fully enclosed Vidian canal are here considered to have acquired this character secondarily. 44. [64] Exoccipital fused to opisthotic prior to hatching. In lepidosauromorphs ancestrally the exoccipital fuses to the opisthotic relatively late in development, usually well after sexual ma- turity. This character relates to the timing of the fusion event in development and is not easily de- termined in fossils. Without additional data, the presence of a fused exoccipital-opisthotic in an isolated skull shows only that the event has taken place, but not when it occurred. In this in- stance, however, the converse case may be informative. For example, if a skull is found in which these bones remain separate, then this taxon is plesiomorphic compared with squamates. Thus, younginiforms (Gow, 1975), Paliguana* (Carroll, 1975a), kuehneosaurs (Evans, 1980), and rhyn- chocephalians (pers. obs.) are plesiomorphic, because at least some postembryonic specimens re- ferred to these taxa have separate exoccipitals. This character is indeterminable in Palaeagama* and Saurosternon* (Carroll, 1975a). In some squamates that we believe for other reasons to be paedo- morphic (e.g., Klauberina and Dibamus), the exoccipital may remain suturally distinct in newly hatched individuals. We have also observed sutural separation of these bones in some late embryos of the skink Tiliqua, and O. Rieppel (pers. comm.) has seen it in the lacertid Podareis. Dibamus may retain a separate exoccipital until very late in post-hatching development (Greer, 1985). How- ever, because exoccipital-opisthotic fusion takes place in late embryos (rarely in newly hatched in- dividuals) in all other autarchoglossans, and in late embryos in all iguanians, Dibamus is consid- ered paedomorphic in this character. 45. [70] Angular reduced, not reaching mandibular condyle. As noted above, the angular is a prominent bone in the mandible of lepidosauromorphs ancestrally. The element extends posterior- ly to the level of the mandibular condyle in younginiforms (Fig. 3A) and rhynchocephalians (Fig. 31). Its posterior extent is indeterminable in Paliguana*, Palaeagama*, Saurosternon*, and kuehneosaurs (Carroll, 1975a; Robinson, 1962). Except for some iguanine iguanians and some teiids, the angular is much reduced in squamates, and it never reaches below the mandibular condyle (Fig. 3K). 46. [69] Coronoid eminence prominent and formed only by uniquely modified coronoid bone. The coronoid eminence is relatively feebly developed in lepidosauromorphs ancestrally (Evans, 1980). Sphenodontidans are distinguished from other rhynchocephalians (Gephyrosaurus) by pro- nounced development of the coronoid eminence (compare Figs. 31, J). In sphenodontidans, however, the ancestral relations of the elements are maintained; the coronoid eminence is thus LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 53 formed by the surangular laterally and the coronoid medially, although the latter is enlarged and projects slightly above the former bone. The ancestral condition, with a low coronoid eminence and a small, subtriangular, medially-placed coronoid bone, is present in araeoscelidans (Reisz, 1981), archosauromorphs (Gauthier, 1984), younginiforms (Gow, 1975), and rhynchocephalians 'ancestrally (Fig. 3L). This character is indeterminable in Paliguana*, Palaeagama*, and Sauroster- non* (Carroll, 1975a). Kuehneosaurs are problematic, for although they have a feebly developed coronoid eminence on the mandible, the coronoid bone itself is not present. Since the coronoid bone is small, and it is attached superficially to the mandible in non-squamate lepidosauromorphs, I we consider it more likely that the element was not preserved, rather than interpreting the coronoid I to have been lost phylogenetically. Squamates are unique in having a very prominent coronoid I eminence formed by the coronoid bone alone. The coronoid bone is uniquely modified in that it [sits on top of the mandible, often extending laterally to overlap the dentary (Fig. 3K). Further, the coronoid has a prominent posteromedial process that extends down the prearticular to form the an- teromedial margin of the mandibular fossa (Fig. 3N). The coronoid has been modified in some squamate groups. In mosasaurs the prominent posteromedial process of the coronoid mentioned above has been lost. Most snakes have greatly reduced or lost the coronoid bone, although Cylin- I drophis (O. Rieppel, pers. comm.), various booids, and to a lesser extent scolecophidians, may re- ' tain the element in a less modified form (McDowell and Bogert, 1954). The coronoid sits on top as well as on the medial side of the mandible in snakes and mosasaurs ancestrally, but has no later- al development onto the dentary. Of course, in order to develop the intramandibular mobility char- acteristic of these groups, the dentary-coronoid overlap would have to be greatly reduced or absent. In several other squamate groups in which the dentary extends far posteriorly, the lateral process of the coronoid may be reduced (e.g., agamids*) or overlapped by the dentary (e.g., xantusiids). Am- phisbaenians are like fossorial squamates in general in that the dentary characteristically overrides the coronoid, and some members of this group resemble alethinophidian snakes in the contribution of the surangular to the coronoid eminence. Thus, at least some amphisbaenians and snakes pos- sess mandibles that are anomalously plesiomorphic in this regard. 47. [86] Anterior cervical and posterior trunk ribs single-headed. Ribs from the midtrunk re- gion are single-headed in diapsids ancestrally, with two-headed ribs confined to the anterior cervi- cals and the most posterior trunk vertebrae (Reisz, 1981). With development of prominent trans- verse processes in the trunk vertebrae within archosauromorphs, the two-headed structure of the ribs becomes increasingly prominent (Gauthier, 1984). Lepidosauromorphs, however, retain feebly-developed transverse processes and most of their ribs are single-headed. Two-headed cervical ribs are retained by younginiforms (Gow, 1975), kuehneosaurs (which may have a third process; Robinson, 1962), and rhynchocephalians (Evans, 1980). These elements are unknown in Paligua- na*, Palaeagama*, and Saurosternon* (Carroll, 1975a). As pointed out by Hoffstetter and Gase (1969), all ribs are single-headed in squamates. 48. [79] Cervical intercentra form prominent hypapophyses. The cervical intercentra may be slightly enlarged compared to others in the column (e.g., Sphenodon), but they never form the prominent, blade-like, hypapophyses seen in squamates alone among amniotes (Hoffstetter and Gase, 1969). The ancestral condition is retained in younginiforms (Gow, 1975), rhynchocephali- ans (pers. obs.), and (apparendy) kuehneosaurs (Evans, 1980). The character is indeterminable in Paliguana*, Palaeagama*, and Saurosternon* (Carroll, 1975a). Heloderma appears exceptional among squamates in having reduced these elements secondarily. 49. [171] One vertebra added to cervical series by loss of contact between the rib of the eighth vertebra and the sternum. The more posterior cervicals are not morphologically distinguishable from the anterior trunk vertebrae in many diapsids. Hoffstetter and Gase (1969) reviewed the vari- ous criteria that have been used to determine the number of cervical vertebrae. We agree with their 54 Phylogenetic Relationships of the Lizard Families view that the criterion of sternal rib attachments is the only one that gives uniform results in an objective fashion. Unfortunately, this criterion is useless for the overwhelming majority of fossil specimens. Unless preserved in exquisite detail, such as in some of the Solnhofen Homoeosaurus and sapheosaurs (Cocude-Michel, 1963), the connections between the calcified ventral ribs and ster- num can rarely be observed in a fossil. By this criterion, living squamates (except chamaeleons) have at least eight cervicals. Because Sphenodon also has eight cervicals, we consider this number to have been achieved independently in these two taxa, following the usual interpretation that sev- en cervicals are present in lepidosaurs and rhynchocephalians ancestrally (Hoffstetter and Gase, 1969). We must emphasize, however, that in the case of fossils this count can be reliably deter- mined only in some Homoeosaurus and sapheosaurs. 50. [87] Sacral and caudal ribs fuse to their respective centra in embryo. The sacral and caudal ribs fuse to their respective centra prior to the attainment of maximum adult size in ancestral sauri- ans (Gauthier, 1984). According to most authors, these elements remain separate in juvenile forms, usually fusing at about the time when the scapulocoracoid coossifies (Currie, 1981b). This is the condition in younginiforms (Currie, 1981b), Palaeagama* (Carroll, 1975a), and rhynchoce- phalians (Howes and Swinnerton, 1901). This character is indeterminable in Paliguana* (Carroll, 1975a). The fusions have already taken place in the available specimens of kuehneosaurs (Colbert, 1970) and Saurosternon* (Carroll, 1975a), so there is as yet no way to determine when in develop- ment they occurred. Squamates are distinguished from other lepidosauromorphs by the fusion of the sacral and caudal ribs to their respective centra prior to hatching. 51. [77] Fusion of neural arches to their respective centra in embryo. The neural arches re- main separate from their respective centra until late in development in lepidosauromorphs ances- trally. The ancestral condition has the same distribution among lepidosauromorphs as does the preceding character. Thus, in a specimen in which the scapula and coracoid are separate, the neural arches, and sacral and caudal ribs, are separate from their respective centra. By maximum adult size, however, these suturally united elements fuse to one another. Indeed, the fusions may be cor- related events in the development of lepidosauromorphs. In crocodilians, however, caudal rib fu- sion precedes that between neural arches and centra in the trunk region. Thus, until more is known, we will treat them as different characters. Winchester and Bellairs (1977) have shown that the neuro-central suture is covered by periosteal bone until very late in prejuvenile development in squamates. Xantusiids (e.g., Lepidophyma, Klauberina) and gekkotans (e.g., Coleonyx brevis) are to our knowledge the only squamates in which hatchlings retain a neuro-central suture. Both groups are distinctive among squamates in the degree of paedomorphosis exhibited in their verte- bral development. 52. [83] Trunk vertebrae without discrete intercentra. Intercentra are present throughout the vertebral column in lepidosauromorphs ancestrally. Intercentra are present in younginiforms (Gow, 1975), Palaeagama* and Saurosternon* (Carroll, 1975a), and in rhynchocephalians aside from Homoeosaurus (Evans, 1981). This character is indeterminable in Paliguana* (Carroll, 1975a). Kuehneosaurs evidently lack discrete intercentra (Evans, 1981). Although intercentra per- sist in the cervical and caudal regions in all squamates, they are commonly absent in the trunk re- gion (Hoffstetter and Gase, 1969). Among squamates, discrete intercentra in the trunk region are known only in xantusiids and gekkotans. Contrary to most accounts, we have been unable to find persistent intercentra in the trunk region of xantusiids, with the exception of some juvenile Xantu- sia and Lepidophyma. Likewise, trunk intercentra may be present or absent among gekkotans (Kluge, 1983). Because no other autarchoglossans have trunk intercentra, and they are lacking in iguanians, their appearance in some gekkotans and xantusiids is considered another example of pae- domorphosis in these groups. 53. [84] Procoelous, non-notochordal vertebrae, with centrum forming conical frustrum. This LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 55 character, like the preceding, has been the subject of considerable debate (see review in Hoffstetter and Gase, 1969; Kluge, 1987). Spool-shaped, notochordal, amphicoelous vertebrae are present in lepidosauromorphs ancestrally. This character is not preserved in Paliguana* (Carroll, 1975a). The ancestral condition occurs in younginiforms (Currie, 1981b), Palaeagama*, and Saurosternon* (Carroll, 1975a), and in rhynchocephalians (Evans, 1981). In kuehneosaurs, the vertebrae are platy- coelous and non-notochordal (Robinson, 1962). Some gekkotans appear to possess the ancestral condition and others do not (Moffat, 1973). Some juvenile xantusiids, like some gekkotans, have an "intermediate" condition. In these taxa, spool-shaped centra are retained, but the notochord is obliterated, and there is a poorly-developed procoelous condyle. During posthatching ontogeny, xantusiid vertebral centra transform from being spool-shaped with poorly developed condyles to a conical shape with more prominent condyles (pers. obs.). Once again, except for some xantusiids and gekkotans, all other autarchoglossans are like iguanians in possessing the synapomorphic ver- tebral form. Accordingly, procoelous vertebrae with conical centra and no notochordal remnants are considered synapomorphic for squamates. Thus, xantusiids and gekkotans display varying de- grees of paedomorphosis in their vertebral development (Underwood, 1954; Kluge, 1983). 54. [92] Anterior margin of scapulocoracoid with two emarginations, one in coracoid and one between scapula and coracoid. The scapula and coracoid are not emarginate in araeoscelidans (Fig. 5A), nor in lepidosauromorphs ancestrally (Fig. 5B). The ancestral condition is found in youngin- iforms (Gow, 1975), kuehneosaurs (Colbert, 1970), and rhynchocephalians (Fig. 5B), and the char- acter is not determinable in Paliguana* and Palaeagama* (Carroll, 1975a). Contrary to Carroll (1975a; 1977), the scapulocoracoid of Saurosternon* is not emarginate anteriorly. In squamates scapulocoracoid emarginations deeply encroach into the body of the element (Fig. 5C). Carroll's specimen drawings (1975a) clearly show that the anterior margin of the scapulocoracoid is straight in Saurosternon*. In addition, the element Carroll (1977) identifies as a "procoracoid bar," is prob- ably a cervical rib (Estes, 1983). We follow Evans (1981) in considering the scapulocoracoid of Saurosternon* to be plesiomorphic. With few exceptions, all squamates with unreduced forelimbs have anteriorly emarginate scapulocoracoids (Lecuru, 1968a); the number varies from one (e.g., lleloderma) to four (e.g.. Iguana). Accepting the relationships argued in Estes et al. (1988) sug- gests that the scapulocoracoid and anterior coracoid emarginations are ancestral for squamates. Thus, more or fewer fenestrae are considered apomorphic within squamates. 55. [95] Clavicle contacts suprascapula. In lepidosauromorphs ancestrally the dorsal end of the clavicle articulates with the scapula. This condition is found in younginiforms (Gow, 1975) and rhynchocephalians (Fig. 5B), but is not determinable in Paliguana*, Palaeagama*, Sauroster- non*, and kuehneosaurs (Evans, 1981). With the exception of most agamids*, a few iguanids* (Polychrus, Corythophanes), and the gekkotan Uroplatus, all other squamates with unreduced fore- limbs are unique among diapsids in that the clavicle extends dorsally to contact the suprascapular cartilage (Lecuru, 1968b; Fig. 5C). 56. [97] Elongate, gracile limbs. In comparison to rhynchocephalians of equal snout-vent length, squamates have relatively more elongate and lightly constructed limbs. Some squamates, such as Heloderma or Iguana, may have equally robust limbs, but such animals are invariably much larger than other, similarly proportioned lepidosauromorphs. In addition, the carp?is and tar- sals are relatively smaller, and the metacarpals and metatarsals are more broadly overlapping and tightly packed proxLmally. The tibia and fibula approach one another more closely at their respec- tive articulations with the proximal tarsals. The hand is relatively smaller and the radius and ulna are more closely set at their proximal articulations with the hum?rus and their distal articulations with the proximal carp?is. The entire hum?rus is gracile compared to that of other lepidosauro- morphs. Owing to preservation, one can seldom see all these characters in extinct lepidosauro- morphs. So far as they are preserved, however, their limbs conform to the proportions and rela- 56 Phylogenetic Relationships of the Lizard Families tions found in Recent Sphenodon (pers. obs.). Thus, one usually finds the general proportions of extinct lepidosauromorphs described as being Sphenodon-like (e.g., Carroll, 1977). A conspicuous exception is the more squamate-like limb proportions of some species o? Homoeosaurus (see Rhynchocephalia above) and the kuehneosaurs. The synapomorphies listed above derive from comparison of Sphenodon and squamates. So many differences exist that it seems likely that fu- ture finds of more completely preserved lepidosauromorphs will allow further hierarchical division of these data. 57. [98] Loss of entepicondylar foramen in hum?rus. An entepicondylar foramen is present in diapsids ancestrally (Reisz, 1981). Among lepidosauromorphs, only kuehneosaurs and squamates lack this foramen. The presence of an entepicondylar foramen in rhynchocephalians (Evans, 1981) means that this character cannot be evaluated unambiguously. The foramen could have been lost in the kuehneosaur-lepidosaur ancestor, and reevolved in rhynchocephalians. Alternatively, the en- tepicondylar foramen could have been lost independently in kuehneosaurs on the one hand and squa- mates on the other. In either case, two evolutionary steps are required, and the decision is thus equivocal. Until more is known, we will consider kuehneosaurs and squamates to have lost this foramen independently. 58. [100] Squamate ulna-ulnare joint. An ulna with a convex distal extremity and a ball-in- socket ulna-ulnare joint are present in diapsids ancestrally (Reisz, 1981). This condition is found in younginiforms (Currie, 1981b) and rhynchocephalians (Fig. 9A). This character cannot be de- termined in most extinct lepidosauromorphs, either because the region is not preserved (i.e., Pali- guana*) or the specimens are apparently not adults (i.e., Saurosternon*, Palaeagama*, Icarosaurus). Squamates are unique among diapsids in possessing an enlarged distal epiphysis that is nearly hem- ispherical in profile and fits into a concomitantly enlarged depression in the ulnare (Fig. 9B). 59. [99] Specialized radius-radiale joint. The distal epiphysis on the radius forms a transverse- ly oriented concavity that articulates with a convex, roller-like surface on the radiale in lepidosauro- morphs ancestrally. This condition occurs in younginiforms (Currie, 1981b) and rhynchocephali- FIGURE 9. Posterior view of distal end of right ulna (A and B), radius (C, D above), and radiale (D below). A, C, Sphenodon punctalus (sphenodont); B, Tupinambis teguixin (autarchoglossan); D, Amblyrhynchus cristatus (iguanian). LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 57 ans (Fig. 9C); it is not determinable in Paliguana*, Palaeagama*, Saurosternon*, and kuehneo- saurs. Squamates are unique in that they have developed a "styloid" process on the radius; the pos- teromedial margin of the distal epiphysis of the radius is produced into a bluntly-pointed process lying in a concave depression on the posteromedial surface of the proximal end of the radiale (Fig. 9D). 60. [101] Intermedium reduced or absent in hand. The intermedium is a large element that ar- ticulates with the ulna in lepidosauromorphs ancestrally. This is the condition in araeoscelidans (Fig. 4A), younginiforins (Currie, 1981b) and rhynchocephaUans (Fig. 4B). As with the previous two characters, this one cannot be determined in many extinct lepidosauromorphs. Among them, only Saurosternon* has some of the carp?is preserved, and Carroll (1975a: 1977) considers its in- termedium to be reduced, thus approaching the squamate condition. Since rhynchocephalians also retain the ancestral condition, Carroll's conclusion that Saurosternon* possesses a reduced interme- dium would most reasonably be interpreted as convergence. Unfortunately, the carp?is are dis- placed to some extent and, in keeping with the subadult stage of this fossil, they are poorly ossi- fied. Thus, the position, size, and shape of the "intermedium" cannot be interpreted unambiguous- ly. This apparent case of convergence could as well reflect that the "intermedium" has been misi- dentified, or that it has not reached its adult size and shape. In view of these ambiguities, we con- sider this character to be indeterminable in Saurosternon*. Casta?eda and Alvarez (1968, repeated in Renous-Lecuru, 1973) reported a large intermedium in Bipes, the only limbed amphisbaenian. Greer and Gans (1984) indicated that this bone is a lateral centrale rather than an intermedium, which is absent. Openosaurus (Komhuber, 1901) and mosasaurs (Russell, 1967) are the only oth- er squamates in which a large intermedium has been identified (Russell, 1967). However, the so- called intermedium in this group has the articulations of a lateral centrale distally and an intermedi- um proximally, and it may thus represent a compound element (see Figs. 4D,E). Until more is known, a greatly reduced or absent intermedium is considered a synapomorphy of squamates (see Fig. 4C). 61. [102] Lateral centrale in hand contacts second distal carpal, thus interrupting contact be- tween medial centrale and third distal carpal. In lepidosauromorphs ancestrally the medial centrale contacts the third distal carpal, thus excluding the lateral centrale from contacting the second distal carpal. This condition occurs in araeoscelidans (Fig. 4A), younginiforms (Currie, 1981b) and rhynchocephalians (Fig. 4B); it is indeterminable in Paliguana*, Palaeagama*, Saurosternon*, and kuehneosaurs. Squamates are unique among lepidosauromorphs in that, relative to the ancestral condition, the lateral centrale is slightly enlarged (or the medial centrale is slightly reduced?) and it contacts the second distal carpal to interrupt the contact between the medial centrale and third distal carpal (Fig. 4C). 62. [103] Modified joint between first metacarpal and wrist. In diapsids ancestrally the first metacarpal contacts the fu^st distal carpal, which in turn contacts the medial centrale and second dis- tal carpal (Fig. 4 A). As seen in crocodilians and Sphenodon, all metacarpals except the fifth over- lap one another proximally in saurians ancestrally (pers. obs.). The ancestral relations of the first metacarpal and first distal carpal are retained in younginiforms (Currie, 1981b) and rhynchocephali- ans (Fig. 4B). This character is not determinable in Paliguana*, Palaeagama*, and kuehneosaurs. The first metacarpal and first distal carpal are preserved in nearly natural association in Sauroster- non* (Carroll, 1975a). As noted above, however, displacement of the more proximal carp?is does not allow unambiguous interpretation of the relations of these elements to the remainder of those in the wrisL Following Carroll (1977), we consider the element usually thought of as distal carpal one to be the medial centrale in squamates, and the proximal epiphysis of metacarpal one to be the true first distal carpal. The proximal end of the fu"St metacarpal is thus uniquely modified in squa- mates; it extends into the row of distal carp?is to contact the medial centrale, and the base of the 58 Phylogenelic Relationships of the Lizard Families B \ FIGURE 10. Posterior view of left tibio-astragalar joint. Corucia zebrata (aularchoglossan). A, Leiolepis belUana (iguanian); B, element is expanded laterally to contact the second distal carpal (Fig. AC). The form of the joint between the first metacarpal and the wrist allows for some independence of motion between the first and other digits of the hand. At rest, the first manual digit projects medially, and as in sauri- ans generally, the fifth manual digit projects laterally. As a consequence, the outer digits of the squamate hand can extend in opposite directions and converge upon one another when contracted. Presumably, the modifications of the first metacarpal reflect development of some grasping ability in the squamate manus. 63. [121] Pubes in relatively narrow contact at symphysis, and pelvic fenestra consequently enlarged. As noted above, the pelvis is fenestrate in lepidosaurs ancestrally (e.g., kuehneosaurs, Robinson, 1962, 1967; rhynchocephalians, Fig. 6C; squamates. Fig. 6D). Squamates differ from rhynchocephalians and kuehneosaurs, however, in that the pubis is a more lightly constructed ele- ment. In particular, the pubes become more gracile distaliy, so that the pubic symphysis is rela- tively narrow. Primarily as a result of the modification of the pubis, the pelvic fenestra is relative- ly larger in squamates than it is in the other members of the kuehneosaur-lepidosaur group. 64. [133] Modification of locked tibio-astragalar joint by loss of ridge and trough articulation. As pointed out by Reisz (1981), the distal extremity of the tibia forms a ridge that fits into a com- plementary-shaped trough formed between two raised ridges on the proximal surface of the astragal- us. Reisz (1981) interpreted this modification as representing a locked joint between the tibia and astragalus, a diapsid synapomorphy. Identification of this character requires fully adult individuals with well-preserved ankles that may be prepared in such a way that this character can be examined. Most of the pertinent fossils cannot be interpreted in this regard. Reisz (1981) noted that this joint is present in araeoscelidans and younginiforms, and we have observed it in Sphenodon. Among squamates, however, the tibio-astragalar joint has been modified. No squamate retains the ancestral tibio-astragalar joint intact. Aside from the suppression of the ridge and trough connection, how- ever, that of iguanians is least modified (Fig. lOA). The autarchoglossan tibio-astragalar joint is very different (Fig. lOB; and see Estes et al., 1988). Accordingly, the form of the tibio-astragalar joint retained by iguanians is considered to represent the ancestral squamate condition. 65. [125] Fibular-astragalocalcanear joint involves most of distal end of fibula. In lepidosau- LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 59 romorphs ancestrally the fibular-astragalocalcanear joint is confined to a small portion of the distal end of the fibula. This condition is found in araeoscelidans (Reisz, 1981), archosauromorphs (e.g., Sigogneau-Russell and Russell, 1978), younginiforms (Currie, 1981b), and rhynchocephalians (Fig. 7C). The shape of this joint is unknown in Paliguana*, Palaeagama*, Saurosternon*, and kuehneosaurs. The distal epiphysis of the fibula is uniquely modified in squamates. Its shape is such that the fibular-astragalocalcanear joint involves most of the distal end of the fibula (Fig. 7D). The fibular articular surface on the astragalocalcaneum is modified in a complementary fash- ion, forming a broad, steeply-inclined band on the lateral face of the element. 66. [134] Squamate ankle joint. Synapomorphies in lepidosauromorph ankle morphology have been described above. Brinkman (1980) argued that the squamate ankle differs from the condi- tion in lepidosaurs ancestrally by having developed a complex tongue-in-groove structure at the joint between the astragalocalcaneum and fourth distal tarsal. One of the specializations of the squamate ankle joint is a prominent, dorsally-directed flange on the calcaneum. Another is the ex- tension of the articular surface of the astragalocalcaneum onto the dorsal aspect of the fourth distal tarsal. According to Brinkman (1980), both these modifications are lacking in Saurosternon*. Based on Sphenodon (pers. obs.) and Gephyrosaurus (Evans, 1981), it is evident that they are lack- ing in rhynchocephalians as well. Ankle joints are unknown in Paliguana* and Palaeagama*. S. Evans (pers. comm.) indicates that dissociated astragalus and calcaneum elements of kuehneosaurs are plesiomorphic compared with those of squamates. 67. [132] Squamate hooked fifth metatarsal. As described above, lepidosaurs share a number of modifications of the shape of the fifth metatarsal that are lacking in ancestral lepidosauro- morphs. This collection of synapomorphies has been described by Robinson (1975), and is usual- ly summarized by the term "hooked fifth metatarsal". As described by Robinson (1975), however, the squamate fifth metatarsal is further specialized, because of the angulation of the proximal head of the element, the prominence of the medial plantar tubercle, and the lateral displacement of the lateral plantar tubercle. 68. [129] Second distal tarsal absent. This bone is a separate element in lepidosauromorphs ancestrally, being present in younginiforms (Currie, 1981b), Saurosternon* (Fig. 8B), and rhyn- chocephalians (Fig. 8C). The character is not preserved in Paliguana*, Palaeagama*, and kuehneo- saurs. In all squamates, however, the second distal tarsal is absent (Fig. 8D). 69. [136] Gastralia absent. Gastralia are present in diapsids ancestrally, being retained by arae- oscelidans (Reisz, 1981), younginiforms (Currie, 1981b), Palaeagama* ana Saurosternon* (Carroll, 1975a), kuehneosaurs (Evans, 1981), and rhynchocephalians (Romer, 1956). Gastralia are retained by archosauromorphs as well, although they have been lost independently within sau- ropods and birds, and in all omithischians (Gauthier, 1984). Among lepidosauromorphs, only squamates lack gastralia. TAXONOMY CRITERIA USED IN THIS STUDY In this paper we have documented the successive levels of inclusion for taxa within Lepidosau- romorpha. This hierarchy of groups as determined by their nested synapomorphies, and the phylog- enetic relationships that may be hypothesized from them, are the principal articles of interest to us here. Although the cladograms (Figs. 11-13) adequately summarize our conclusions, some sys- tematists may prefer a listed taxonomy as well. We provide one here based on criteria given be- low. These criteria in many ways conflict with those of most traditional taxonomies, as well as the conventions adopted by Wiley (1981), principally because he sought to reconcile the problems of ranking in phylogenetic taxonomy and we explicitly reject the use of ranks for reasons specified below. 60 Phylogenetic Relationships of the Lizard Families 1. No formal categorical ranks are recognized. Wiley's first convention (1981:200-201) advo- cated a Linnean hierarchy coupled with some additional terms, although he recognized that there was no biological necessity to do so. Most workers apply formal categorical ranks to taxa, but for reasons discussed below we have provided only an indented taxonomy of Lepidosauromorpha. Wi- ley offered some practical objections to the use of an indented taxonomy, such as the difficulty of following the sequence if the list spans more than one page. We grant that this may be the case in some instances, but we believe that the points made below override such objections. Initially, we explored the possibility of providing a ranked taxonomy that would avoid prolife- ration of names and categorical levels, as suggested by Wiley (1981). However, if we retained the rank of Class for Reptilia as constituted here, the rank then available for Aves approaches the ge- neric level, even with the use of the additional categories that we wanted to avoid. The phyletic se- quencing convention discussed by Wiley (1981:206, convention 3) avoids proliferation of names and categories, but leaves many taxa unnamed. Nevertheless, any categorical levels that we used would be subject to considerable revision - and additional proliferation of names - as similar studies of other groups of vertebrates are made. Because extensive revision of vertebrate taxonomy is not a goal of our paper, we have had to look for alternatives. Beyond these practical matters, there are theoretical reasons for discarding categorical ranks. Most systematists are aware that formal ranks such as Class, Order, or Family are not equivalent across all groups. For example, aside from being monophyletic, it is difficult to discern ways in which the orders Rhynchocephalia and Cole?ptera, or the families Felidae and Euphorbiaceae, are equivalent. They differ profoundly in taxonomic and morphologic diversity, as well as in their times of origin. In addition, from the review of Mishler and Donoghue (1982), it is doubtful that currently recognized species and species concepts are equivalent across groups. The lack of equiva- lency of taxa assigned equal ranks has not prevented some biologists from attributing phylogenetic relevancy to "patterns" in, for example, ordinal or familial diversity through time. Such endeavors probably reveal more about the ill-defined concepts of the ranks Order and Family than about the structure of the biological world. One could make taxonomic ranks equivalent on the basis of absolute time of origin as sug- gested by Hennig (1966). This would provide a particularly useful basis for comparison, especial- ly for those interested in examining rate-related processes or biogeographic patterns. Unfortunately, the fossil record provides only minimum estimates of times of origin, and then only for a fraction of the extant biota. One could also rank taxa on the basis of relative divergence times. However, only sister groups would be equivalent; this equivalence would not extend to taxa given the same rank that are not sister groups. For example, in terms of categorical rank in this sense, Gephyro- saurus is equivalent to Sphenodontida, just as Iguania is equivalent to Autarchoglossa. However, even if Gephyrosaurus and Autarchoglossa were given the same rank, this would not imply equiva- lence in any phylogenetically meaningful properties. Discarding categorical ranks does not hamper discussions of the phylogenetic properties of monophyletic taxa. Moreover, no changes in nomenclature are necessary because we preserve rank-associated suffixes and Linnean binomials for the sake of stability. We believe that this ap- proach will have a positive influence on studies of morphologic and taxonomic diversity, and we hope that workers interested in such questions will follow Vrba (1980) by pursuing them in the context of sister taxa rather than taxonomic ranks. 2. No redundant names are recognized. In making this decision, we differ from Wiley's sec- ond convention (1981:200,205). Wiley recommended that redundancy be avoided except in the case of the five required Linnean higher categories (Phylum, Class, Order, Family, Genus). Be- cause we reject current concepts of rank, there is no longer a need to retain any redundant names. As noted above, we are mainly interested in nested patterns of co-occurring synapomorphies and the phylogenetic relationships among taxa that may be hypothesized from them. In this con- LEPIDOSAUROMORPHAN PHYLOGENY - Gauthier et al. 61 ./ .e. J?> J> .<& Jty < ^^ fi> "> .> s^' .^^ .'b- s> -?> .?" o .tr ^^" .'^ .'b- J2> G> e. , 7 7 7 1 0 0 0 0 0 0 19 0 0 0 1 1 0 0 0 0 1 1 20 0 0 -> 0 7 7 0 0 0 0 0 0 1 21 0 0 ? 7 7 1 0 0 0 0 0 1 22 0 0 ?> 7 7 1 1 1 0 0 1 23 0 0 7 7 0 4 3 3 2 1 0 24 0 ? ? 7 1 1 1 1 1 1 1 25 0 0 7 7 0 1 1 1 0 0 0 0 26 0 0 7 7 7 0 0 0 0 0 1 27 0 0 7 0 7 7 0 1 1 1 1 1 0 28 0 0 7 7 7 7 7 0 0 0 7 0 1 29 ? 0 7 7 7 7 0 1 7 7 1 1 1 30 o 0 7 7 7 7 0 0 7 ? 0 0 1 LEPIDOSAUROMORPH PHYLOGENY - Gauthier et al. 83 Tan You Ace Pig Pag Sau Kue Sph Horn Sap Cle Gep Squ 31 1 1 ? ? ? 7 0 1 1 1 1 1 1 32 0 7 0 ? ? 0 0 0 0 0 0 0 1 33 0 0 7 0 7 7 1 1 1 1 1 1 1 34 0 0 7 0 7 7 0 0 0 0 0 0 1 35 0 0 7 7 7 7 0 1 7 7 7 7 2 36 0 0 0 1 7 7 1 0 0 0 1 1 1 37 7 7 7 7 7 7 7 0 0 0 0 ? 1 38 ?j 7 7 7 ? ? 7 0 7 7 7 7 1 39 0 0 7 ? 7 7 0 0 0 0 0 0 1 40 0 0 0 0 0 7 0 1 1 1 0 0 0 41 1 1 1 7 7 7 0 1 1 1 1 1 0/1 42 1 1 1 0 7 7 0 0 0 0 0 0 0 43 0 0 7 0 7 7 0 0 0 1 0 0 0 44 0 0 7 7 7 7 0 0 0 1 0 0 0 45 0 0 0 0 7 7 0 0 0 1 0 0 0 46 0 0 0 0 7 7 0 1 1 1 0 0 0 47 0 0 0 0 0 ? 0 1 0 1 0 0 0 48 0 0 0 0 0 7 0 1 1 1 0 0 0 49 0 0 7 7 7 7 0 3 2 2 2 1 0 50 0 0 7 ? 7 ? 0 1 0 0 0 0 0 51 0 0 7 7 7 7 0 1 0 0 0 ~?" 0 52 0 0 7 ? 7 7 0 1 7 0 0 0 0 53 0 0 ? 0 7 7 0 1 0 0 0 0 0 54 0 0 7 7 7 7 N 1 7 7 0 0 N 55 0 0 7 7 7 7 0 1 1 1 1 0 0 56 0 0 0 0 0 7 0 1 2 2 1 0 0 57 0 0 0 7 7 7 0 0 0 0 1 0 0 58 0 0 0 7 7 7 0 1 1 1 0 0 0 59 0 0 7 0 *? 7 1 1 1 1 1 1 1 60 0 0 7 7 7 ? 0 0 0 0 0 0 1 61 1 1 7 0 0 7 0 0 0 0 0 0 0 62 7 0 7 7 7 7 0 0 0 0 0 0 1/0 63 0 0 0 0 0 7 0 0 0 0 0/1 1 1/0 64 0 0 ? 0 7 7 0 0 7 7 0 0 1/0 65 0 0 0 0 0 7 0 0 0 0 0/1 1 0/1 66 0 0 7 0 7 7 0 1 1 1 1 1 0/1 67 0 0 0 7 7 7 0 1 1 1 1 1 0/1 68 ?> 7 7 ? 7 7 0 0 0 0 0 1 0/1 69 0 0 0 7 7 7 7 0 0 0 0 0 1 70 0 0 0 7 7 7 0 0 0 0 0 0 1 71 0 0 7 ? 7 7 1 1 1 1 1 1 1 72 0 0 0 1 r? 7 1 0 0 0 1 1 1 73 0 0 0 0 7 ? 0 1 1 1 0 0 0 74 7 0 7 7 7 7 7 1 7 ? 7 1 1 75 0 0 0 7 *? 7 0 2 2 2 2 1 1/2 84 Phylogenetic Relationships of the Lizard Families Tan You Ace Pig Pag Sau Kue Sph Horn Sap Cle Gep Squ 76 ?? 7 0 7 7 7 7 2 2 2 7 1 0 77 0 0 0 7 '? 7 7 0 o 7 7 0? 1 78 0 0 0 7 7 7 7 1 7 7 7 1 1/0 79 0 0 7 7 7 7 0 0 0 0 7 0 1 80 0 0 7 7 7 0 0 1 1 1 7 1 1/0 81 1 1 0 ? 0 0 0 0 0 0 7 0 0 82 0/1 0 0 7 0 0 0 0 0 1 7 0 0 83 0 0 0 7 0 0 1 0 0 0 7 0 1/0 84 0 0 0 7 0 0 0 0 0 0 7 0 1/0 85 0 0 0 7 0 0 1 0 0 0 7 0 0 86 0 0 7 7 7 7 0 0 0 0 0 0 1 87 0 0 0 7 0 7 7 0 7 7 7 7 1 88 0/1 0 7 7 7 7 1 1 0 0 7 0 0/1 89 7 7 7 7 7 7 7 1 0 1 7 7 0 90 ? 7 7 7 7 7 7 2 7 ?> 7 7 1 91 1 7 7 7 7 1 7 2 2 2 7 7 2 92 0 0 7 7 7 0 0 0 0 0 7 0 1 93 0 0 7 7 7 1 7 1 1 1 7 1 1 aq 0 0 7 7 7 1 7 1 1 1 7 1 1 95 0 0 7 7 7 0 7 0 0 0 7 0 1/0 96 0 0 1 7 0? 0 0 0 0 0 7 7 G 97 0 0 0 7 0 0 2 1 2 1 7 1 2 98 0 0 0 7 7 0 1 0 0 0 7 0 1 99 0 0 0 7 7 o o 0 0 0 0 '> 1 100 0 0 0 7 9 7 7 0 0 0 7 7 1 101 0 0 0 7 7 1? 7 0 0 0 7 7 1 102 0 7 0 7 ?:> 7 7 0 0 0 7 7 1 103 0 0 0 7 7 0 7 0 0 0 7 7 1 104 0 0 0 7 0 0 0 1 1 1 7 7 1 105 0/2 0 0 7 0 0 1 2 2 2 7 7 2 106 1 1 1 ? 7 1 7 1 1 1 7 7 1 107 1 1 1 7 7 7 0 0 0 0 7 0 0 108 1 1 0 7 0 0 0 0 0 0 7 9 0 109 1 1 0 -:> 0 0 0 0 0 0 7 0 0 110 1 0 0 7 0 0 0 0 0 0 7 0 0 111 1 7 0 7 7 7 7 0 0 0 7 7 0 112 1 0 7 7 7 0 0 0 0 0 7 0 0 113 1 0 0 7 0 0 0 0 0 0 7 7 0 114 1 1? 7 ? 1 1 1 1 1 1 7 1 1/N 115 0 0 7 7 7 7 7 1 7 7 7 7 1 116 0 0 0 7 7 7 0 1 1 1 7 1 1 117 0 0 0 7 7 7 0 1 7 7 7 1 1 118 0 1 0 7 7 7 0 0 0 0 7 0 0 119 0 0 0 7 7 7 7 1 1 1 7 1 1 120 0 0 0 7 7 7 0 1 1 1 7 1 1 LEPIDOSAUROMORPH PHYLOGENY - Gauthier et al. 85 Tan You Ace Pig Pag Sau Kue Sph Horn Sap Cle Gep Squ 121 0 0 0 7 0 0 1 1 1 1 7 1 2 122 0 0 0 7 0 0 0 0 1 1 7 0 0 123 Q 0 7 7 7 7 0? 1 1 1 7 1 1 124 0 0 0 7 7 7 7 1 1 1 ? 1 1 125 0 7 7 7 7 7 7 0 0 0 7 7 1 126 0 7 7 7 7 0 7 1 1 1 7 1 1 127 0 7 7 7 7 0 7 1 1 1 7 1 1 128 0 7 7 7 7 0 7 1 1 1 7 7 1 129 0 7 7 7 7 0 7 0 0 0 7 7 1 130 0/1 7 7 7 7 0 ? 1 1 1 7 1 1 131 1 7 7 7 7 0 7 ?p 7 ?p 7 7 ? 132 0 0 7 7 7 0 0 1 1 1 7 1 2 133 0 7 7 7 7 7 7 0 7 7 7 7 1 134 0 0 7 7 7 1 7 1 1 1 7 1 2 135 ? 7 7 ? 7 7 0 7 7 7 7 1/0 136 0 0 0 7 0 0 0 0 0 0 ? 0 1 137 0 0 0 ? 7 7 1 1 1 7 1 1 138 ? 7 7 ? 7 ? 1 7 7 ? 7 1 139 ?> 7 7 7 7 7 1 7 7 7 7 1 140 0 1 0 0? 0 0 0 0 0 7 7 0 141 7 7 7 7 7 7 0 7 7 7 7 1 142 7 7 ? 7 7 7 0 7 7 7 7 1 143 ? 7 7 7 7 7 0 7 7 7 7 1 144 7 7 7 ? ? ? 0 7 7 7 7 1 145 7 ? 7 7 7 7 0 7 7 7 7 1 146 ? 7 7 7 7 7 0 7 7 7 7 1 147 9 7 7 7 7 7 0 7 7 7 7 1 148 ?7 7 ? 7 7 7 7 0 ? 7 7 7 1 149 7 7 7 7 7 7 7 0 7 7 7 7 1 150 7 7 7 7 7 7 7 0 7 7 7 7 1 151 7 7 7 7 7 7 7 0 7 7 7 7 1 152 7 7 7 7 7 7 7 0 7 7 7 7 1 153 ? 7 7 7 7 7 7 0 7 7 7 7 1 154 7 7 7 7 7 7 7 0 7 7 7 7 1 155 7 7 7 7 7 7 7 0 7 7 7 7 1 156 7 7 ? 7 ? 7 7 0 ? 7 7 7 1 157 ? 7 7 7 7 7 0 7 7 7 7 1 158 7 7 7 7 7 7 7 0 7 7 7 7 1 159 7 7 7 7 7 7 0 7 7 7 7 1 160 7 7 7 7 7 7 1 7 7 7 7 1 161 7 7 7 7 7 7 7 1 7 7 7 7 1 162 7 7 7 7 7 7 1 7 7 7 7 1 163 7 7 7 ? 7 ? 1 7 7 7 7 1 164 7 7 7 7 9 7 1 7 7 ? 7 1 165 7 7 7 7 7 7 1 7 7 7 7 1 86 Phylogenetic Relationships of the Lizard Families Tan You Ace Pig Pag Sau Kue Sph Horn Sap Cle Gep Squ 166 ? 7 7 7 ? 7 ? 1 7 7 7 ? 1 167 ? 7 7 ? ? 7 ? 1 7 7 7 7 1 168 7 ? 7 7 7 7 7 1 7 7 7 7 1 169 7 7 7 7 7 7 7 ? 7 7 7 7 1 170 7 7 7 7 7 7 7 0 7 7 7 7 1/0 171 7 7 7 7 7 7 7 1 0? 0? 7 7 1 age I APPENDIX HI Results of a Computer-Based Phylogenetic Analysis of the Data in Appendices I and II We analyzed the data in Appendices I and II using the Wagner program in the PHYSYS packa for computer assessment of phylogenetic relationships written by J. S. Farris and installed in the California State University CYBER system. Both "?" and "N" character state scores were entered as missing data. Characters that varied within basic taxa were assigned the first of the two or more states listed in Appendix II. Numbers here follow those in Appendices I and II. Characters of soft anatomy (numbers 141-171) were not included in this analysis. These characters serve to support the monophyly of both Lepidosauria and Squamata among other living organisms, but they are of little use in placing many of the fossils we discuss. For example, extant archosauromorphs enable us to infer that the apomorphic conditions were not present in Sauria ancestrally. It is equally parsi- monious to accept that the apomorphic resemblances in soft anatomical characters shared by Recent Sphenodon and squamates were present in their most recent common ancestor. Because of missing data, however, we cannot determine the level at which these characters arose within lepidosauro- morphs prior to the origin of lepidosaurs. In such cases we are inclined to place the characters at the level at which they can be observed (i.e., Lepidosauria), although we recognize that they may ap- ply to more inclusive groups of lepidosauromorphs. Wagner analysis yielded twenty-eight cladograms of equal length. They differed from one anoth- er and from our proposed hypothesis (see text and Fig. 13) only in the relative positions of the three metataxa, Palaeagama*, Paliguana*, and Saurosternon*, and in the relationships among sphen- odonts, sapheosaurs, and Homoeosaurus. Character discordance accounts for the unresolved relation- ships among Homoeosaurus, sapheosaurs, and sphenodonts, but missing data in the metataxa are re- sponsible for most of the cladograms. In all twenty-eight cladograms, Paliguana* and Saurosternon* formed a monophyletic group with kuehneosaurs plus lepidosaurs, although their precise relationships to the kuehneosaur- lepidosaur clade were variable. Palaeagama* was more variable in its position, either forming one branch of an unresolved trichotomy with younginoids and Acerosodontosaurus, occuring at an unre- solved position at the basal node of the cladogram, or taking various positions within a clade con- sisting of itself, Paliguana*, Saurosternon*, and kuehneosaurs plus lepidosaurs, although never with- in the last group. Sphenodonts, sapheosaurs, and Homoeosaurus consistently form a clade, but sa- pheosaurs are sometimes the sister group of sphenodonts, and other times the sister group of Homo- eosaurus. The twenty-eight alternative cladograms were distilled to form a single consensus cladogram (Appendix Fig. 1) using the Adams program in PHYSYS (see Adams, 1972). The following list gives the synapomorphies of the terminal and subterminal nodes (hypothesized clades except in the case of metataxa indicated by *) of the Adams consensus cladogram as determined by the Diagnose program in PHYSYS. Numbers of nodes in this list correspond with those on the consensus clado- gram (Appendix Fig. 1). In addition to listing synapomorphies, we also indicate all required conver- gences and character reversals. The distribution of some characters is such that parsimony will not allow an unambiguous historical interpretation. We discuss briefly those cases in which our inter- pretations of character optimization differ from those of PHYSYS (viz., Ferris optimization), al- though we are aware that such decisions are arbitrary. ' LEPIDOSAUROMORPH PHYLOGENY - Gauthier et al. 87 ^ ^ 05 f-^ ra E c * ?O 0) CO O? c 03 (/) fl? (D o 13 ca OJ 3 CO ra CO ?. CO CL -a tf? o o c 0) O) ^ J= Q a tD \ \ . CO \ o 3 Ol h- CM 3 ca U) ca o M >? ? XI