Top Banner
Zoological Journal of the Linnean Society (1999), 125: 115–147. With 7 figures Article ID: zjls 1997.0144, available online at http://www.idealibrary.com on Squamate phylogeny and the relationships of snakes and mosasauroids MICHAEL W. CALDWELL* Department of Geology, The Field Museum, Roosevelt Road at Lakeshore Drive, Chicago, IL 60605, U.S.A. Received March 1996; accepted for publication July 1997 Cladistic analysis of extant and fossil squamates (95 characters, 26 taxa) finds the fossil squamate, Coniasaurus Owen, 1850, to be the sister-group of the Mosasauroidea (mosasaurs and aigialosaurs). This clade is supported in all 18 shortest cladograms (464 steps; CI 0.677; HI 0.772) by nine characters of the dermatocranium, maxilla, and mandible. A Strict Consensus Tree of the 18 shortest trees collapses to a basal polytomy for most major squamate clades including the clade (Coniasaurus, Mosasauroidea). A Majority Rule Consensus Tree shows that, in 12 of 18 shortest cladograms, the clade ConiasaurusMosasauroidea is the sister-group to snakes (Scolecophidia (Alethinophidia, Dinilysia); this entire clade, referred to as the Pythonomorpha ([[Scolecophidia [Alethinophidia, Dinilysia]], [Coniasaurus, Mosasauroidea]]) is the sister-group to all other scleroglossans. Pythonomorpha is supported in these 12 cladograms by nine characters related to the lower jaw and cranial kinesis. In 6 of 18 shortest cladograms, snakes are the sister-group to the clade (Amphisbaenia (Dibamidae (Gekkonoidea, Eublepharidae))). None of the cladograms support the hypothesis that coniasaurs and mosasauroids are derived varanoid anguimorphs. Two additional analyses were conducted: (1) manipulation and movement of problematic squamate clades while constraining ‘accepted’ relationships; (2) additional cladistic analyses beginning with extant taxa, and sequentially adding fossil taxa. From Test I, at 467 steps, Pythonomorpha can be the sister-group to the Anguimorpha, Scincomorpha, ‘scinco-gekkonomorpha’ [scincomorphs, gekkotans, and amphibaenids-dibamids]. At 471 steps Pythonomorpha can be placed within Varanoidea. Treating only mosasauroids and coniasaurs as a monophyletic group: 469 steps, mosasauroids and coniasaurs as sister- group to Anguimorpha; 479 steps, mosasauroids and coniasaurs nested within Varanoidea. Test II finds snakes to nest within Anguimorpha in a data set of only Mosasauroidea+Extant Squamates; the sistergroup to snakes+anugimorphs is (Amphisbaenia (Dibamidae (Gek- konoidea, Eublepharidae))). No one particular taxon is identified as a keystone taxon in this analysis, though it appears true that fossil taxa significantly alter the structure of squamate phylogenetic trees. 1999 The Linnean Society of London ADDITIONAL KEY WORDS:—coniasaurs – mosasauroids – phylogeny – snakes – squamates. * Present address: Paleobiology, Research Division, Canadian Museum of Nature, P.O. Box 3443, Stn. D, Ottawa, Ontario, Canada, K1P 6P4. Email: [email protected] 115 0024–4082/99/010115+33 $30.00/0 1999 The Linnean Society of London
33

Caldwell, 1999a

Mar 06, 2016

Download

Documents

Felipe Elias

ADDITIONAL KEY WORDS:—coniasaurs – mosasauroids – phylogeny – snakes – squamates. MICHAEL W. CALDWELL∗ ∗ Present address: Paleobiology, Research Division, Canadian Museum of Nature, P.O. Box 3443, Stn. D, Ottawa, Ontario, Canada, K1P 6P4. Email: [email protected] © 1999 The Linnean Society of London Received March 1996; accepted for publication July 1997 Article ID: zjls 1997.0144, available online at http://www.idealibrary.com on INTRODUCTION CONTENTS
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Caldwell, 1999a

Zoological Journal of the Linnean Society (1999), 125: 115–147. With 7 figures

Article ID: zjls 1997.0144, available online at http://www.idealibrary.com on

Squamate phylogeny and the relationships ofsnakes and mosasauroids

MICHAEL W. CALDWELL∗

Department of Geology, The Field Museum, Roosevelt Road at Lakeshore Drive,Chicago, IL 60605, U.S.A.

Received March 1996; accepted for publication July 1997

Cladistic analysis of extant and fossil squamates (95 characters, 26 taxa) finds the fossilsquamate, Coniasaurus Owen, 1850, to be the sister-group of the Mosasauroidea (mosasaursand aigialosaurs). This clade is supported in all 18 shortest cladograms (464 steps; CI0.677; HI 0.772) by nine characters of the dermatocranium, maxilla, and mandible. AStrict Consensus Tree of the 18 shortest trees collapses to a basal polytomy for mostmajor squamate clades including the clade (Coniasaurus, Mosasauroidea). A Majority RuleConsensus Tree shows that, in 12 of 18 shortest cladograms, the clade Coniasaurus–Mosasauroidea is the sister-group to snakes (Scolecophidia (Alethinophidia, Dinilysia); thisentire clade, referred to as the Pythonomorpha ([[Scolecophidia [Alethinophidia, Dinilysia]],[Coniasaurus, Mosasauroidea]]) is the sister-group to all other scleroglossans. Pythonomorphais supported in these 12 cladograms by nine characters related to the lower jaw andcranial kinesis. In 6 of 18 shortest cladograms, snakes are the sister-group to the clade(Amphisbaenia (Dibamidae (Gekkonoidea, Eublepharidae))). None of the cladogramssupport the hypothesis that coniasaurs and mosasauroids are derived varanoid anguimorphs.Two additional analyses were conducted: (1) manipulation and movement of problematicsquamate clades while constraining ‘accepted’ relationships; (2) additional cladistic analysesbeginning with extant taxa, and sequentially adding fossil taxa. From Test I, at 467steps, Pythonomorpha can be the sister-group to the Anguimorpha, Scincomorpha,‘scinco-gekkonomorpha’ [scincomorphs, gekkotans, and amphibaenids-dibamids]. At 471steps Pythonomorpha can be placed within Varanoidea. Treating only mosasauroids andconiasaurs as a monophyletic group: 469 steps, mosasauroids and coniasaurs as sister-group to Anguimorpha; 479 steps, mosasauroids and coniasaurs nested within Varanoidea.Test II finds snakes to nest within Anguimorpha in a data set of only Mosasauroidea+ExtantSquamates; the sistergroup to snakes+anugimorphs is (Amphisbaenia (Dibamidae (Gek-konoidea, Eublepharidae))). No one particular taxon is identified as a keystone taxon inthis analysis, though it appears true that fossil taxa significantly alter the structure ofsquamate phylogenetic trees.

1999 The Linnean Society of London

ADDITIONAL KEY WORDS:—coniasaurs – mosasauroids – phylogeny – snakes –squamates.

∗ Present address: Paleobiology, Research Division, Canadian Museum of Nature, P.O. Box 3443,Stn. D, Ottawa, Ontario, Canada, K1P 6P4. Email: [email protected]

1150024–4082/99/010115+33 $30.00/0 1999 The Linnean Society of London

Page 2: Caldwell, 1999a

M. W. CALDWELL116

CONTENTS

Introduction . . . . . . . . . . . . . . . . . . . . . . . 116Methods . . . . . . . . . . . . . . . . . . . . . . . . 117

Matrix structure and analysis . . . . . . . . . . . . . . . . 117Characters and character states . . . . . . . . . . . . . . . 117Extant taxa . . . . . . . . . . . . . . . . . . . . . 118Fossil taxa . . . . . . . . . . . . . . . . . . . . . . 118Polarity, rooting and outgroups . . . . . . . . . . . . . . . 119Polymorphic characters . . . . . . . . . . . . . . . . . 120

Phylogenetic analysis . . . . . . . . . . . . . . . . . . . . 120Results . . . . . . . . . . . . . . . . . . . . . . . 120Relationships of Coniasaurus, mosasauroids and Serpentes . . . . . . . 121Relationships of Coniasaurus . . . . . . . . . . . . . . . . 122(Serpentes (Mosasauroidea, Coniasaurus)) . . . . . . . . . . . . 122

Global squamate relationships . . . . . . . . . . . . . . . . . 125Mosasauroids+Coniasaurs as anguimorphs: additional tests . . . . . . . 128

Test number I . . . . . . . . . . . . . . . . . . . . 128Test number II: sequential addition of fossil taxa . . . . . . . . . 131Inter-subjective consensus . . . . . . . . . . . . . . . . . 135

Conclusions . . . . . . . . . . . . . . . . . . . . . . . 137Snakes, Mosasauroidea, Coniasaurus; Pythonomorpha . . . . . . . . 137Squamata . . . . . . . . . . . . . . . . . . . . . . 138Snake origins . . . . . . . . . . . . . . . . . . . . . 139

Acknowledgements . . . . . . . . . . . . . . . . . . . . 140References . . . . . . . . . . . . . . . . . . . . . . . 140Appendix 1 . . . . . . . . . . . . . . . . . . . . . . . 143Appendix 2 . . . . . . . . . . . . . . . . . . . . . . . 146Appendix 3 . . . . . . . . . . . . . . . . . . . . . . . 147

INTRODUCTION

Recent cladistic analysis of morphological characters from mosasaurs, aigialosaurs,and fossil and extant varanoids (Caldwell, Carroll, & Kaiser, 1995), found that theaccepted phylogenetic scheme finding mosasaurs to be derived varanoids, and morespecifically, the sister-taxon to varanids (McDowell & Bogert, 1954; Carroll &deBraga, 1992; deBraga & Carroll, 1993), was potentially inaccurate. Amongvaranoids, Varanus has usually been identified as the sister-group of mosasauroids(Carroll & deBraga, 1992; deBraga & Carroll, 1993), though Lee (1997) suggestedthat a mosasauroid–snake clade might be the sister-taxon of a varanoid cladecomposed of varanids, lanthanotids, and related fossil forms.

The retracted narial opening and elongate head, common to both mosasauroidsand varanids, but absent or less exaggerated in Heloderma and Lanthanotus, havetraditionally been important characters supporting ‘ancestor-descendant’ relationsbetween mosasauroids and varanid lizards. The possible polyphyly of a mosasauroid-varanid grouping, identified by Caldwell et al. (1995), was based on the re-charac-terization of the structure of the bony nares and palate of mosasauroids. Caldwellet al. (1995) argued that the bony elements of the elongate snout and retracted nareswere not similar in form or topological relation between varanids and mosasauroids,and when tested for congruence in a cladistic analysis of all characters, failed thattest of congruence (Patterson, 1982). Synapomorphies were not found, homologywas not inferred, and monophyly of a taxon describing mosasauroids in some closerelationship to varanids was not supported.

Page 3: Caldwell, 1999a

SQUAMATE PHYLOGENY 117

Based on these findings, a research program designed to investigate ingroup andoutgroup relations of mosasauroids was initiated. The first step was considerationof the composition of Mosasauroidea. Previous schemes have included the Ai-gialosauridae and Mosasauridae. DeBraga & Carroll (1993) consider the taxonmonophyletic despite the fact that they did not test aigialosaur monophyly inthe context of all Mosasauroidea. In contrast, Bell (1993) concluded that theAigialosauridae was likely paraphyletic. Caldwell (1996) tested Bell’s conclusionsusing data from various aigialosaurs and mosasaurs and could find no support foraigialosaurian monophyly within Mosasauroidea (in a strict consensus of all trees,aigialosaurian taxa collapsed to a polytomous node nested within the clade Mo-sasauridae). Paraphyly of Aigialosauridae was further supported by reference to amajority-rule consensus tree (see fig. 2B, Caldwell [1996]). The consensus derivedfrom these studies is that aigialosaurs are paraphyletic.

The search for the monophyletic squamate group that includes mosasauroids,and to resolve mosasauroid ingroup relationships, has since led to the re-examinationof other putative fossil varanoids that have been considered ‘related’ to mosasaurs,i.e. dolichosaurs, coniasaurs (Caldwell & Cooper, 1998), and other aigialosaurs. Ipresent the results of a cladistic analysis of fossil and extant squamates and discussthem in terms of the sister-group relationships of coniasaurs, mosasauroids, andsnakes. The results presented here are preliminary in the sense that the data set iscontinually being expanded by the addition of new data. This data is derived fromthe ongoing descriptions of new and poorly known coniasaur-like squamates.

METHODS

Matrix structure and analysis

A data matrix composed of 25 taxa of fossil and extant squamate reptiles, and95 characters (Appendices 1, 2), was analysed using Heuristic algorithms in thesoftware program PAUP Version 3.1.1 for the Macintosh (Swofford, 1993). SpecificHeuristic Search Options were: Random Addition Sequence (100 Replications);Tree-Bisection-Reconnection (branch swapping); and Steepest Descent Off. It isimportant to recognize that Heuristic Searches do not perform maximum parsimonyanalyses and that the ‘shortest’ trees produced by any one analysis are the mostlocally Optimal Trees; it also is not known if these resultant cladograms are theGlobal Optimum or Local Optimum type (Swofford, 1993). Therefore, cladogramsdiscussed and consensus trees presented will not be referred to as Most ParsimoniousTrees (MPTs), or the product of MPTs, but rather as ‘shortest trees’ found.

Multistate characters were unordered and character distributions optimized byACCTRAN. ACCTRAN optimizes characters by accelerating character trans-formations. The effect is that changes appear lower in the tree, thereby identifyingsynapomorphies (unequivocal and equivocal characters) for more inclusive clades,rather than being optimized as apomorphies of more exclusive clades or eventerminal taxa (DELTRAN).

Characters and character states

Eighty-nine of the 95 characters described (Appendix 1) and coded in the datamatrix (Appendix 2) were derived from Estes et al. (1988); in early stages of this

Page 4: Caldwell, 1999a

M. W. CALDWELL118

study, the 89 characters were coded as given by Estes et al. During my ownexamination of squamate osteological materials (see below, and Appendix 3), sixnew characters were created, and more than half of the state codings for the 89characters of Estes et al. (1988) were modified.

The character set created by Estes et al. (1988) described 130 osteological and 18soft tissue characters for 19 extant squamate taxa. Soft tissue characters wereexcluded from this study because of their absence in fossil material, the small numberincluded in the data set of Estes et al. (1988), and the colossal size of the task requiredto rescore old and new characters for so many extant taxa. Another 41 of thecharacters described by Estes et al. (1988) either were not scored or were subsumedin the character descriptions of the 89 characters I included in my own data matrix(Appendices 1, 2). These excluded characters were judged redundant, clearlydependent on other characters, or did not characterize the intended homologue asI interpreted it.

Extant taxa

The squamate taxa included in this study as terminal taxa are derived in partfrom Estes et al. (1988). Changes and additions were made to the composition ofthe ingroup by the inclusion and distinction of Shinisaurus, Xenosaurus, Eublepharidae,Gekkonoidea, alethinophidian and scolecophidian snakes, and ‘other iguanids’ [seebelow].

Sub-division of the Xenosauridae is based on the recognition of poor support forxenosaurid monophyly (Rieppel, 1980a; Caldwell, pers. observ.). Character statesfor Shinisaurus were taken from Hecht & Costelli (1969), Costelli & Hecht (1971),Rieppel (1980a) and from observation of skeletonized materials (Caldwell, pers.observ.).

Division of the Gekkota into Eublepharidae and Gekkonoidea is derived fromKluge (1987); states for eublepharids and other gekkonoids are from Kluge (1967,1987). Character states of the Pygopodidae (Kluge, 1976) are included in the statecodings for Gekkonoidea.

Division of Serpentes into Alethinophidia and Scolecophidia is based on Rieppel(1988) and Kluge (1991), and was done in order to avoid creating a paraphyleticSerpentes by the subsequent inclusion of the fossil taxon Dinilysia (see below).

Iguanian codings were taken from Etheridge & deQueiroz (1988), Estes et al.(1988), and Frost & Etheridge (1989). The taxon ‘other iguanids’ contains Frost &Etheridge’s (1989) families Corytophanidae, Crotaphytidae, Hoplocercidae,Iguanidae, Opluridae, Phrynosomatidae, Polychridae and Tropiduridae. These taxawere excluded as independent terminal units due to the required memory increaseto analyse more than 34 taxa (as per Swofford [1993]), and because the emphasisin this analysis is on relationships in squamate clades other than Iguania sensu Frost& Etheridge (1989).

Fossil taxa

The phylogenetic relationships of mosasauroids, dolichosaurs and coniasaurs area central theme of the investigation presented here. The hypothesis that some or

Page 5: Caldwell, 1999a

SQUAMATE PHYLOGENY 119

all of these taxa might be derived varanoid lizards, and that the characters offossil varanoids might be essential to place mosasauroids, etc., within Varanoidea,prompted the inclusion of the Cretaceous fossil varanoid Estesia mongoliensis; characterstates for Estesia are from Norell et al. (1992).

Other fossil taxa include Dinilysia patagonica (Estes, Frazetta & Williams, 1970),which was included in the analysis to test characters from a primitive fossil snakerelative to mosasauroids, dolichosaurs, coniasaurs and living snakes (alethinophidiansand scolecophidians). States for the Cretaceous fossil snake Dinilysia patagonica arefrom Estes et al. (1970), Hecht (1982), and Rage & Albino (1989).

Characters and character states for the Mosasauridae are derived from examinationof clidastine, platecarpine and halisaurine mosasaurs (Bell, 1993; Caldwell, 1996),and from various aigialosaur taxa (Carroll & deBraga, 1992; deBraga & Carroll,1993; Caldwell et al., 1995). The probable paraphyly of Aigialosauridae is recognizedhere (Bell, 1993), but for the purposes of this analysis the terms aigialosaur(s)/aigialosaurid(s) will be used in reference to mosasauroids that still retain limbs thatare not specialized as paddle-like appendages (see Caldwell, 1996). States forConiasaurus were obtained from a new and undescribed species of Coniasaurus(manuscript submitted), and from a redescription of the type species, Coniasauruscrassidens (Caldwell & Cooper, 1998).

Polarity, rooting and outgroups

The extant lepidosaur Sphenodon was used as the outgroup in this analysis forrooting the tree and establishing character polarities. In comparison, Estes et al.(1988) determined polarity using an outgroup code (the Outgroup ComparisonMethod of Maddison, Donoughue & Maddison [1984]) derived from eight taxa(Sphenodon, Rhyncocephalia, Archosauromorpha, Younginiformes, Palaeagama, Sauro-sternon, Paliguana and Kuehneosaurus). The rationale for limiting the number of outgrouptaxa to only Sphenodon is based on a series of analyses using three different outgroupsor outgroup codes.

Three tests of polarity and rooting were constructed to determine the most viableand informative outgroup:(1) using the outgroup code of Estes et al. (1988); (2) usingall eight taxa as independent terminals in a global analysis of the ingroup andoutgroup taxa; (3) using only Sphenodon. The intention of these three tests of outgroupstructure was to determine the necessity and importance of using more than oneoutgroup taxon (see Nixon & Carpenter, 1993)

In all three tests the ingroup tree topologies were identical. Using all eight taxaas independent terminals simply produced a large number of trees, of greater lengththan the trees found using only Sphenodon (the relationship of outgroup taxa tothemselves and to Squamata was not resolved). The only uncertainty introduced byusing only Sphenodon, was that some character polarities could not be determinedbecause the character was missing or unidentified in that taxon.

In conclusion, using only Sphenodon produced the same ingroup topologies as didthe use of all eight taxa, and the use of the Estes et al. (1988) outgroup code.Exclusive use of Sphenodon produced ‘better’ polarity determination for almost allstate changes because Sphenodon is an extant taxon and is therefore more readilyobserved in all its detail. Exclusive use of Sphenodon also reduced the number ofshortest trees without altering the topology of ingroup relations.

Page 6: Caldwell, 1999a

M. W. CALDWELL120

AgamidaeChamaeleonidaeOther IguaniansAnguidaeShinisaurusXenosaurusHelodermaLanthanotusEstesiaVaranusCordylidaeScincidaeGymnophthalmidaeTeiidaeLacertidaeXantusiidaeDibamidaeGekkonoideaEublepharinaeAmphisbaeniaAlethinophidiaDinilysiaScolecophidiaMosasauroideaConiasaurusSphenodon

A AgamidaeChamaeleonidaeOther IguaniansAnguidaeShinisaurusXenosaurusHelodermaLanthanotusEstesiaVaranusCordylidaeScincidaeGymnophthalmidaeTeiidaeLacertidaeXantusiidaeDibamidaeGekkonoideaEublepharinaeAmphisbaeniaAlethinophidiaDinilysiaScolecophidiaMosasauroideaConiasaurusSphenodon

B33

100

100

100

100

100100

100

100

100100

100100

67

83

67100

50 100

100

100

100

67

Figure 1. Consensus trees of 18 cladograms (464 steps) showing ingroup relationships of 21 extant andfossil squamate taxa using morphological data (95 osteological characters). A, Strict Consensus Tree.B, Majority Rule Consensus Tree.

Polymorphic characters

At higher taxonomic levels of investigation, in any diverse group, large numbersof polymorphic states are observed in terminal taxa. However difficult to study,polymorphisms are data, and are therefore included in this analysis. Admittedly,many polymorphisms may exist that have not been coded in this matrix; in mostcases, this is omission through absence of information rather than exclusion by choice.However, some polymorphisms were excluded by choice. Where a polymorphism wasnoted in only one individual, or one species of a diverse family-level terminal taxon,it was not coded. For such states it was assumed that the polymorphism is‘autapomorphic’ to that individual or species, and not informative in the sense ofdetermining cladistic branching order.

PHYLOGENETIC ANALYSIS

Results

Cladistic analysis of the data matrix (Appendix 2) found 18 shortest cladograms(464 steps) with a Consistency Index (CI) of 0.677, a Homoplasy Index (HI) of0.772, and a Retention Index (RI) of 0.604. The Strict Consensus Tree (Fig. 1A)

Page 7: Caldwell, 1999a

SQUAMATE PHYLOGENY 121

shows Coniasaurus in the sister-group position to the Mosasauridea (Mosasauridaeand the paraphyletic aigialosaurs). The topology supports previous hypotheses ofrelationship for most squamate crown groups (Estes et al., 1988), but provides noresolution on more problematic relationships at internal nodes. Two major groupsof squamates are identified: ((Other Iguanians, Chamaeleonidae, Agamidae) + (allother taxa)). Relationships between non-iguanian squamates collapse to a singleinternal node that defines the Scleroglossa. However, a number of clades are resolvedeven though the sister-group relationships of those clades at internal nodes are not(Fig. 1A).

The Majority Rule Consensus Tree (Fig. 1B) shows that in 67% of the cladograms(12 of 18) Coniasaurus and Mosasauroidea are the sister-group to ophidians (Sco-lecophidia, (Dinilysia, Alethinophidia)); this clade lies outside Scleroglossa (all othersquamates excluding iguanians). Coniasaurus and the taxon Mosasauroidea are nevernested within Varanoidea. The Majority Rule Consensus also shows at least 67%support for the relationships shown in (Fig. 1B).

In six of 18 cladograms a monophyletic Serpentes is found to be the sister-groupto an incompletely resolved clade composed of Dibamidae, Amphisbaenia, andgekkotans (in all six cladograms gekkotan monophyly is maintained). The uncertaintywith the Dibamidae-Amphisbaenia–Gekkotan clade results from the instability ofdibamids and amphisbaenids, and whether or not these two taxa form a clade, orare successive sister-groups along the branch leading to gekkotans. In all six of thesecladograms the basal scleroglossan position of the Mosasuroidea+Coniasaurus cladeis maintained.

In 15 of 18 cladograms, the clade containing dibamids, amphisbaenids andgekkotans is found to be monophyletic. In 9 of 18 cladograms the relationships ofthis group are resolved as follows: ((Amphisbaenia) + (Dibamidae (Gekkonoidea,Eublepharidae))). The clade (Amphisbaenia (Dibamidae (Gekkonidae, Euble-pharidae))) is the sister to the conventional Scincomorpha (Xantusiidae (Lacertidae(Teiidae, Gymnophthalmidae))) + (Cordylidae, Scincidae)))).

The conventional structure of the clade Anguimorpha, i.e. Anguoidea andVaranoidea, is found in all cladograms and consensus trees (Fig. 1A, B). However,the relationships of several crown-group taxa differ compared to conventionalphylogenies. The Xenosauridae of McDowell & Bogert (1954) and Estes et al. (1988)is found to be paraphyletic: Shinisaurus and Xenosaurus are successive sister-taxa toAnguidae. Within Varanoidea, Varanus is found as the sister-taxon to all othervaranoids. Estesia is the sister to Lanthanotus and Heloderma. The clade (Estesia(Lanthanotus, Heloderma))) is informally referred to here as the ‘estesioids’.

In only six of 18 trees are the iguanian crown group taxa conventionally resolved(Other Iguanians (Chamaeleonidae, Agamidae)).

RELATIONSHIPS OF CONIASAURUS, MOSASAUROIDS AND SERPENTES

The following discussion examines the relationships of Serpentes as reconstructedin 12 of the 18 shortest cladograms (Fig. 1B). Character state distributions are takenfrom the cladogram with a topology identical to that of the Majority Rule ConsensusTree (Fig. 1B). The characters and characters states as optimized on the selectedtree serve as the template for the discussion of characters and synapomorphiesdiagnosing clades presented in the remainder of this paper.

Page 8: Caldwell, 1999a

M. W. CALDWELL122

Of the possible relationships of snakes, mosasauroids and Coniasaurus, only twovariations are found among 18 shortest cladograms: (1) (Serpentes (Mosasauroidea,Coniasaurus)) as the sister-group to all other scleroglossans (Fig. 1B); (2) Serpentes asthe sister-group to a clade composed of amphisbaenids, dibamids and gekkotans,with (Mosasauroidea, Coniasaurus) as the sister-group to all scleroglossans; this secondconsensus topology (snake-amphisbaenid relationships) was found in six of 18cladograms.

Because the relationship and characters of dibamids, amphisbaenids and snakeshave been examined recently (Wu, Brinkman & Russell, 1996), the phylogenetichypothesis as reconstructed in Figure 1B is explored more fully here. This is not tosuggest that a snake–amphisbaenid–dibamid clade is poorly supported, but ratherthat the hypothesis of a snake-mosasuroid clade has never been so well supported(12 of 18 cladograms, and a large number of characters). Therefore, a detailedexamination of this hypothesis is justified by the paucity of such investigations.

Relationships of Coniasaurus

The clade (Mosasauroidea, Coniasaurus) is supported by nine characters. Thisfinding is in agreement with earlier results presented by Polcyn and Bell (1994) andwith untested suggestions made by Caldwell & Cooper (in press). Six characters areoptimized unequivocally: 4[0–1], frontals fused; 9[0–1], frontal tabs project overparietal; 17[0–1], parietal tabs are present; 20[0–1], maxilla does not extendposteriorly below orbit; 47[0–1], subdivision of intramandibular septum of Meckel’scanal occurs near posterior end of tooth row with well developed septum; 55[0–1],elongate anterior extension of coronoid. Three characters are optimized equivocally:2[0–1], no contact between nasals and prefrontals; 7[1–0], weakly developeddescending process of frontals, prefrontal participates widely in orbitonasal fenestra;44[0–1], Vidian canal posterior opening at basisphenoid prootic suture.

It is significant that a number of characters shared by mosasauroids and coniasaurs(see Caldwell & Cooper, in press) were not included in the character descriptionsor data matrix (Appendices 1, 2). This omission was intentional. The goal was toavoid biasing the result of a global analysis of squamates to insure that the ‘expected’result would be obtained. Previous authors (Owen, 1850; Nopcsa, 1923; McDowell& Bogert, 1954; Bell et al., 1982) had provided such a range of possible options forthe relationships of Coniasaurus that it was clear that a more global test needed tobe procedurally sound.

(Serpentes (Mosasauroidea, Coniasaurus))

The clade (Serpentes (Mosasauroidea, Coniasaurus)) is supported by nine characters:five unequivocal, four equivocal. Most of these characters reflect on specializationsof the lower jaw and postcranial skeleton, i.e. feeding and locomotory specializations.Because the debate on the phylogenetic relationships of mosasaurs and snakes asindependent lineages, and as a possible clade (Cope, 1869), has been so extensiveand long-lived, the characters synapomorphic for (Serpentes (Mosasauroidea, Con-iasaurus)) are discussed in detail.

Page 9: Caldwell, 1999a

SQUAMATE PHYLOGENY 123

Character 33[0–1], long posterior processes of the septomaxilla. This character is difficult tocode in many taxa due to a lack of disarticulated skulls. However, in mosasaurs,where this character can be observed in Plotosaurus (Camp, 1942), the medial flangesof the septomaxillae extend posteriorly to contact, or come very close to contactingthe medial wall of the prefrontal. In Coniasaurus this feature cannot be scored. Theseptomaxilla of mosasaurs is autapomorphic as compared to other squamates. Theexception, in the context of the posteriorly directed medial processes, is withinsnakes. The condition observed in Cylindrophis, Anilius and Boa shows that theseptomaxilla extends posteriorly to a point below the nasofrontal suture and maycontact the prefrontal or decensus frontalis (pers. observ.).

Character 52[0–1], vertical articulation between the splenial and angular. (Fig. 2A–D). Thevertical articulation of the angular and splenial in mosasauroids, coniasaurs, andsnakes is unique among squamates. McDowell & Bogert (1954:60) pointed out thesimilarities of the angular-splenial articulation between these taxa referring to thejoint as the ‘aigialosaurian hinge’ (mosasaurs and aigialosaurs). They also consideredLanthanotus, a rare varanoid from Borneo, to share this condition and intended theircharacter complex to support Lanthanotus as a ‘structural ancestor’ to snakes (theyalso regarded Lanthanotus as a possible extant aigialosaur). They also noted thatabsence of a lateral exposure of the angular was a shared character shared bysnakes, mosasauroids, and Lanthanotus.

Even though a vertical articulation of the angular and splenial is common toLanthanotus, Mosasauroidea + Coniasaurus, and snakes, differences exist in the waythe joint is constructed. Close inspection reveals that Lanthanotus does not show thesame structure as snakes and Mosasauroidea+ Coniasaurus, but rather shares two ofthree characters with other varanoids. In Lanthanotus, Rieppel (1983) observed thatthere is a lateral, internal flange of the angular that crosses the intramandibularhinge thereby overlapping the splenial laterally. Examination of varanids andhelodermatids shows a lateral and internal overlapping process of the angularextending between the splenial and dentary, and a medial process of the splenialthat overlaps the angular ventromedially. However, Varanus and Heloderma do nothave a medially vertical contact of the angular-splenial.

A functional intramandibular hinge is present in a large number of squamates(Estes et al., 1988). As such, intramandibular kinesis is a functional characteristic,like flight, and has been achieved in numerous diapsids employing a varietyof different morphologies between the dentary, splenial, and postdentary bones.Congruence of function is meaningless unless characterization of the primaryhomologues, the elements performing the function, i.e. the bones at that joint, alsopass the tests of similarity and topological relationship (Patterson, 1982).

Character 53[0–1], reduced overlap of the postdentary-dentary bones. This character is sharedwith varanoids, in which it is an unequivocal apomorphy of that clade. The principaldifficulty is to decide during the primary analysis of homologues if the joint is thehomologue for which congruence must be tested, or if specific elements associatedwith that joint are of relevance to discovery of a pattern of relationship. The latterperspective is taken here. As such, mobility in the middle of the lower jaw is seenas convergent.

Character 54[0–1], reduced splenial-dentary suture. The distribution of this character isidentical to the previous character. Unlike the previous two characters, the nature

Page 10: Caldwell, 1999a

M. W. CALDWELL124

Figure 2. Details of the angular and splenial. A, angular-splenial of the holotype specimen of Dolichosauruslongicollis, BMNH R49002, in lateral view; B, mandible and angular-splenial of Coniasaurus cf. C.crassidens, BMNH R3421, in medial view; C, reconstruction of the mandible of an aigialosaur (deBraga& Carroll, 1993); D, detail of the mandible and angular-splenial of Elaphe obsoletus, RTMP 90.7.190,in medial view. Abbreviations: A, angular; C, coronoid; Comp, compound bone of snakes; D, dentary;P, prearticular; Sa, surangular; S, splenial.

of this feature, a reduced splenial, leading to a very open Meckel’s groove, need notbe directly related to an intramandibular joint. Exposure of the groove and areduction of the splenial to a thin splint of bone covering the lower portion of the

Page 11: Caldwell, 1999a

SQUAMATE PHYLOGENY 125

groove is not obviously related to the hinge/joint area. Of primary concern to themechanics of the joint is the nature of the splenial at the hinge, and the degree towhich it extends over the angular, prearticular, and coronoid. A more likely scenariois that reduction of the splenial is paedomorphic and is linked to exposure ofMeckel’s groove and cartilage. Whether or not this character is synapomorphic forsnakes and (Mosasauroidea, Coniasaurus), to the exclusion of varanoids, or somecombination of this three-taxon problem, can only be solved by congruence withother characters. The current hypothesis predicts convergence between py-thonomorphs and varanoids; this character could also be found apomorphic for anycombination, or in fact for all three, in the context of global squamate phylogeny.

Character 56[0–1], anterior end of coronoid meets dentary directly. A similar contact of thecoronoid with the dentary is observed in some iguanians, amphisbaenids, andLanthanotus. This character shows a possible reversal within the type species ofConiasaurus where the coronoid is immediately opposite to, but does not contact, thedentary.

Character 62[0–1], mandibular symphysis absent. Absence of an intermandibular sym-physis (Character 67), i.e. lack of a bony symphyseal surface rigidly connecting theright and left mandibular rami. This character is only observed among py-thonomorphs. In association with other characters linked to mobility and gape inthe lower jaw, the absence of a symphysis has promoted feeding strategies that havebeen widely adaptable throughout pythonomorph evolution. Snakes have continuedthe evolutionary trend of increasing jaw mobility to its extremes. Mosasauroidsappear to have limited their experiments with jaw mobility to the elements of thelower jaw.

Character 73[1–0], zygosphene-zygantra present. Zygosphenes and zygantra are presentin the known dorsal vertebrae of Coniasaurus (Caldwell & Cooper, 1998; see above).They are known to be present throughout the vertebral column in aigialosaurs(Carroll & deBraga, 1992; pers. observ.) and in anterior vertebrae in mosasaurs(Russell, 1967). Snakes uniformly possess zygosphenes and zygantra throughout thevertebral column. The principal difference between the snake and mosasauroid/Coniasaurus condition is the shape of the neural arch lamina separating the zygo-sphenes. In mosasauroids and Coniasaurus, and other limbed squamates in whichzygosphenes and zygantra are found, the laminar arch is notched (lacertids, teiids,gymnophthalmids, some large iguanids, and some cordyline cordylids [(Estes et al.,1988]). This notch is not present in snakes, though a notch has been reported inthe putative oldest snake (Rage & Richter, 1994).

Character 64[1–0], presence of pterygoid teeth (equivocal). Pterygoid teeth are found inHeloderma, Lanthanotus, mosasauroids, Coniasaurus (at least one species), Dinilysia,Shinisaurus, and in some iguanians, anguids, cordylids, scincids, gymnophtalmids,teiids, lacertids and alethinophidian snakes. Pterygoid teeth are not present inscolecophidians, varanids, Estesia, agamids, or chamaeleonids. The distribution ofthis character is so plastic that as a presence/absence character it is uninformative.

GLOBAL SQUAMATE RELATIONSHIPS

A number of important and intriguing alternative relationships for other squamategroups were also discovered in this analysis. They merit some discussion of supporting

Page 12: Caldwell, 1999a

M. W. CALDWELL126

characters, and in some cases indicate that the included taxa require furthersystematic examination. Characters that are optimized unequivocally are noted bybold font (e.g. 4[0–1]) and are given before those characters optimized equivocally;numbers in parantheses, following the character number, indicate the direction ofcharacter state transformation.

Iguanians. Six characters support the monophyly of Iguania (Other Iguanids(Chamaeleonidae, Agamidae)): 4[0–1], frontals fused; 6[0–1], frontal shelf broad;14[0–1], jugal-squamosal contact on supratemporal arch; 5[0–1], frontals constrictedbetween orbits; 11[0–1], postfrontal forked; 39[0–1], epipterygoid absent.

Scleroglossa, exclusive of (Serpentes (Mosasauroidea, Coniasaurus)). Six characters supportthis clade: 25[0–1], dorsal process of squamosal absent; 85[0–1], clavicle angledand curved anteriorly; 89[0–1], interclavicle cruciform with large anterior process;95[0–1], pubis long with narrow symphysis, ventrally directed, pubic tubercleanteroventral; 50[0–1], dorsal extension of dentary coronoid process contacts co-ronoid; 83[1–0], epicoracoid cartilage does not contact meso- and suprascapularcartilages.

(Scleroglossa (Serpentes (Mosasauroidea, Coniasaurus))). Fifteen characters support themonophyly of this clade: 12[0–1], postorbital absent as a separate element; 30[0–1],vomer longer than half of the maxillary tooth row; 31[0–1], septomaxillae meet ornearly meet on midline in raised crest; 32[0–1], septomaxillae expanded and convex;36[0–1], choanal fossae of palatines large; 40[0–1], alar process of prootic elongate;75[0–1], cervical intercentra underly posterior part of preceding centra; 79[0–1],number of presacrals greater than 26; 7[0–1], descending processes of frontals welldeveloped and exclude or nearly exclude prefrontals from margins of orbitonasalfenestra; 13[0–1], postorbital less than half of orbit length and has reduced ventralprocess; 49[0–1], subdental shelf large; 59[0–1], retroarticular process absent;60[0–1], retroarticular process twisted; 66[0–1], marginal tooth replacement post-erolingual, small pits; 94[0–1], epiphyses fusion of diaphysis at same time as braincaseelements.

Anguoidea and a paraphyletic Xenosauridae. Three characters support the Anguoidea:4[0–1], frontals fused; 5[0–1], frontals constricted between orbits; 14[0–1], jugal-squamosal contact on supratemporal arch.

Four characters support the clade Shinisaurus-Anguidae as distinct from Xenosaurus:27[0–1], palpebral ossifications present; 44[0–1], posterior opening of vidian canalat basisphenoid-prootic suture; 64[1–0], pterygoid teeth present; 78[1–0], autotomysepta in caudal vertebrae present.

Varanoids and estesioids. Thirteen characters support the Varanoidea: 20[0–1], pos-terior extent of maxilla just anterior to orbit; 34[0–1], neochoanate; 37[0–1],ectopterygoid contacts palatine; 53[0–1], reduced articulation of dentary-postdentarybones; 54[0–1], reduced bone-bone contact at splenial-dentary suture; 66[1–2],marginal tooth replacement posterolingual, no pits; 67[0–1], plicidentine present;71[0–1], condyle-cotyle orientation strongly oblique; 38[0–1], ectopterygoid ex-panded and fenestra restricted; 50[1–0], dorsal extension of dentary coronoid processabsent; 51[0–1], surangular expanded anterodorsally and nearly vertical at anteriormargin; 72[0–1], centrum constricted anterior to condyles; 83[0–1], no contact ofepicoracoid cartilage with meso- or suprascapula.

Page 13: Caldwell, 1999a

SQUAMATE PHYLOGENY 127

Four characters support the clade referred to here as ‘estesioids’: 3[0–1], prefrontalcontacts postorbital, or postorbitofrontal, above orbit; 19[0–1], pineal foramenabsent; 44[0–1], opening of vidian canal at basisphenoid-prootic suture; 51[1–2],anterior end of surangular terminates closer to coronoid eminence on surangular.

Four characters support the clade Lanthanotus–Heloderma: 10[1–0], postfrontalpresent as separate element; 15[0–1], supratemporal fenestra open and no arch;64[1–0] pterygoid teeth present; 63[1–0], palatine teeth present.

Amphisbaenids, dibamids and gekkotans (A-D-G). Eight characters support the clade(Amphisbaenia (Dibamidae (Gekkonoidea, Eublepharidae))): 2[0–1], no contact ofnasal and prefrontal bones; 26[0–1], supratemporal absent; 8[0–1], contact ofdescending processes of frontals; 23[0–1], jugal reduced or absent, postorbitalbar incomplete; 37[0–1], ectopterygoid contacts palatine; 46[0–2], Meckel’s canalenclosed, bone fused; 57[0–1], angular absent; 80[0–3], two or fewer rib attachmentson sternum.

Two characters support the clade (Dibamidae (Gekkonoidea, Eublepharidae)):75[2–0], cervical intercentra under posterior part of preceding centrum; 82[0–1],scapula emarginated.

Gekkota is supported by 12 characters: 10[1–0], postfrontal absent as separateelement; 45[0–1], jaw adductors originate on ventral surface of parietal; 55[0–1],elongate anterior extension of coronoid; 66[1–0], lingual tooth replacement withlarge resorption pits; 4[0–1], frontals fused; 17[1–0], parietal tabs absent; 43[0–1],lateral head vein enclosed in bony canal formed by crista prootica; 50[1–0], dentarycoronoid process no contact with coronoid; 69[1–0], more than 14 scleral ossicles;70[1–0], second ceratobranchials present; 80[3–1], four rib attachments on sternum;92[0–1], postcloacal bones present.

Scincomorphs+(Amphisbaenia (Dibamidae (Gekkonoidea, Eublepharidae))). Four characterssupport this clade: 19[0–1], pineal foramen absent; 78[1–0], autotomic septa present;15[0–1], supratemporal fenestra open and no arch; 17[0–1], parietal tabs present.

Scincomorphs. Seven characters support this clade: 18[0–1], parietal downgrowthspresent and extend to, or almost to, epipterygoids; 45[0–1], jaw adductors originateon ventral surface of parietal; 66[1–0], lingual tooth replacement, large resorptionpits; 95[1–2], pubis long, narrow, ventrally directed, symphysial process elongate;15[1–2], supratemporal arch present, fenestra closed; 73[1–0], zygosphenes andzygantra present; 75[2–1], cervical intercentra under anterior part of precedingcentrum.

Scincoidea. Three characters support this clade: 10[1–0], postfrontal present asseparate element; 93[0–1], osteoderms present; 12[1–0], postorbital present asseparate element.

Lacertoidea. Seven characters support this clade: 58[0–1], prearticular crest present;59[1–0], sulcus/pit on retroarticular process absent; 60[1–0], retroarticular processnot twisted; 70[1–0], second ceratobranchials present; 77[0–1], two pairs of divergingtransverse processes of caudal vertebrae; 79[1–0], more than 26 presacrals; 90[0–1],sternal fontanelle present.

Serpentes. Serpentes is supported by ten characters: 8[0–1], median contact ofdescending process of frontals; 15[0–1], supratemporal fenestra open and no arch;19[0–1], pineal foramen absent; 22[1–0], anteroventral border of orbit formed by

Page 14: Caldwell, 1999a

M. W. CALDWELL128

maxilla with jugal confined to medial surface of maxilla; 23[0–1], jugal reduced orabsent, postorbital bar incomplete; 24[0–1], squamosal absent; 34[0–2], posteriorborder of opening for Jacobson’s organ closed by contact of vomer and septomaxilla;39[0–1], epipterygoid absent; 66[1–3], marginal tooth replacement posterolingual,no pit, tooth rotates into position; 13[0–1], Postorbital less than one half of posteriorborder of orbit.

The clade Alethinophidia–Dinilysia is supported by two characters: 12[1–0],postorbital present as separate element; 63[1–0], palatine teeth present.

MOSASAUROIDS+CONIASAURS AS ANGUIMORPHS: ADDITIONAL TESTS

Commonly accepted phylogenetic relationships for mosasauroids and coniasaursfind them to be either ingroup anguimorphs, closely related to Varanus or Lanthanotus,or the sister-taxon of Anguimorpha (McDowell & Bogert, 1954; deBraga & Carroll,1993; Lee, 1997). Similar hypotheses have been made that find snakes as the sister-group to anguimorphs, or more specifically one of the extant varanoid genera (forreviews see Rieppel, 1988; Lee, 1997).

As stated earlier, Caldwell et al. (1995) found no support for mosasauroids asderived varanid lizards, a conclusion supported in this analysis (Fig. 1A, B). It isalso clear from this analysis that any classification placing mosasauroids withinanguimorphs as either derived varanoids or the sister-group to varanoids, would bepolyphyletic.

Therefore, in the context of this data set, it is important to further test thepossibility that (Mosasauroidea, Coniasaurus) are nested within Anguimorpha eitherwith or without close relationship to snakes. Two different tests were constructed:(1) using MacClade 3.1 (Maddison & Maddison, 1992), the manipulation variousclades while constraining other ‘accepted’ squamate relationships, and noting thenumber of steps required to produce alternate phylogenies; (2) additional PAUPanalyses using a restricted ingroup set, beginning with only extant taxa, and addingsequentially, the fossil taxa listed in Appendix 2.

Test number I

The topologies and lengths of cladograms n-steps removed from the shortest trees(464+n) were examined. Results are tabulated for tree lengths found when (Serpentes(Mosasauroidea, Coniasaurus)) was moved into various sister-group relations withother squamate taxa and clades (Table 1), when the clade (Serpentes (Mosasauroidea,Coniasaurus)) was broken apart and Serpentes moved into various sister-group positions(Table 2), and when Serpentes was placed in the sister-group position to Anguimorphaand (Mosasauroidea, Coniasaurus) moved into various sister-group positions withinAnguimorpha (Table 3).

(Serpentes (Mosasauroidea, Coniasaurus))Monophyletic (Table 1).The next-shortest alternative topology to Figure 1B has a tree length (TL) of 467steps, and places (Serpentes (Mosasauroidea, Coniasaurus)) in the sister-group position

Page 15: Caldwell, 1999a

SQUAMATE PHYLOGENY 129

T 1: Other possible sister-group relationships of the Pythono-morpha+Serpentes (Fig. 15) and the number of steps from the shortest tree

required to demonstrate them

Number of steps Pythonomorpha as sister to:

464 scleroglossan clades467 scincogekkonomorphs, Varanoidea, Scinomorpha468 Anguimorpha, Amphisbaenia460 dibamogekkota471 Varanus472 estesioids (Estesia + Lanthanotus + Heloderma)473 Dibamidae474 Lanthanotus, Heloderma, Lanthanotus+Heloderma475 Estesia469–472 Anguoidea, or with various anguoid taxa471–473 Scincoidea, or with various scincoid taxa471–476 Iguania, or with various iguanian taxa471–481 Lacertoidea, or with various lacertoid taxa473–476 Gekkota, or with various gekkotan taxa471 Sister to Squamata

T 2: Other possible sister-group relationships of Serpentes (Alethinophidia+Dinilysia+Scolecophidia) and the number of steps from the shortest tree required to examine them.The clade Mosasauroidea+Coniasaurus is fixed outside Scleroglossa and the relationships ofsnakes are tested against this topology by comparison to all other squamates. Emphasis isplaced on the number of steps required to find sister-group relationships of Serpentes withAmphisbaenia+Dibamidae+Gekkota, and Anguimorpha, or with various anguimorph taxa

Number of steps Snakes as sister-group to:

464 Mosasauridae+Coniasaurus; Amphisbaenia+Dibamidae+Gekkota465 scincogekkonomorphs466 Amphisbaenia, Scincomorpha, dibamogekkota468 Anguimorpha, Scleroglossa469 Dibamus470 Gekkonoidea, Eublepharidae470 Varanoidea473 Varanus473 estesioids (Estesia+Lanthanotus+Heloderma)476 Lanthanotus, Heloderma, or Estesia469–480 Lacertoidea, or with various lacertoid taxa469–471 Scincoidea, or with various scincoid taxa472–477 Anguoidea or with various anguoid taxa473–480 Iguania, or with various iguanian taxa∗473 Sister to Squamata

to either Varanoidea, scincogekkonomorphs, or Scincomorpha. At 468 steps (Ser-pentes (Mosasauroidea, Coniasaurus)) is the sister-group to either Anguimorpha,Amphisbaenia, or Amphisbaenia+Dibamids+gekkotans. Finding (Serpentes (Mo-sasauroidea, Coniasaurus)) to be within Varanoidea requires an increase of 7 steps(TL=471) to be the sister-taxon to Varanus, and between 8 to 10 steps to be thesister to any of the estesioid taxa.

While tempting to argue that three steps (as the sister-taxon to Varanoidea) isminimal, it is noted that this same number of steps finds (Serpentes (Mosasauroidea,Coniasaurus)) to be the sister-group to scincomorphs or amphisbaenids, dibamids,

Page 16: Caldwell, 1999a

M. W. CALDWELL130

T 3: Number of steps required to demonstrate other possible sister-group relationships ofMosasauroidea+Coniasaurus. The monophyly of other major squamate clades is constrained as foundin the 12 shortest trees, and Serpentes (Alethinophidia+Dinilysia+ Scolecophidia) is fixed in the sister-group position to Anguimorpha. Mosasauroidea+Coniasaurus is examined within Anguimorpha. This

test examines the view that mosasauroids and coniasaurs are derived varanoid squamates

Number of steps Serpentes outside Anguimorpha and M+C as sister to:

470 Anguimorpha471 Serpentes+Anguimorpha, Varanoidea472 Anguoidea473 Varanus, Shinisaurus474 Xenosaurus, Shinisaurus+Anguidae475 Anguidae476 estesioids478 Heloderma+Lanthanotus479 Lanthanotus, Estesia, Heloderma470–478 Serpentes outside scleroglossans and M+C as sister-group to various anguimorphs

and/or gekkotans. Nesting (Serpentes (Mosasauroidea, Coniasaurus)) within varanoidsis so many steps removed from the shortest tree that any relationship withinSquamata is equally well supported (Table 1). The most important alternativerelationship to that indicated in the Majority Rule Consensus Tree (Fig. 1B) is thesister-group relationship to amphisbaenids and dibamids.

(Serpentes (Mosasauroidea, Coniasaurus))Polyphyletic: I (Table 2)

In this test, the clade including Serpentes + (Mosasauroidea, Coniasaurus) isconsidered to be a polyphyletic assemblage, Serpentes is removed, (Fig. 1B), and(Mosasauroidea+ Coniasaurus) is constrained to the node Scleroglossa. Constrainingrelationships in this manner finds Serpentes to be the sister to Anguimorpha at 468steps. Placing Serpentes in the sister-group position to any or all varanoid taxa is 4to 11 steps longer than the shortest tree. As noted above, there are a great numberof equally parsimonious topologies supporting relationships with other squamatetaxa at this number of steps removed from the shortest tree.

However, an alternative and potentially important sister-group relationship forSerpentes (assuming a polyphyletic (Serpentes (Mosasauroidea, Coniasaurus))) existsat 464 steps: Serpentes (Amphisbaenia (Dibamidae (Gekkonoidea, Eublepharidae).This topology was discussed above and contributes to the polytomy resolved in theStrict Consensus Tree (Fig. 1A). At 465 steps, Serpentes can be reconstructed inthe sister-group position to scincogekkonomorphs. At 466 steps, Serpentes can bereconstructed as the sister-group to Amphisbaenia, Scincomorpha, or dibamo-gekkotans.

(Serpentes (Mosasauroidea, Coniasaurus))Polyphyletic: II (Table 3)

Serpentes is fixed in the sister-group position to Anguimorpha (Estes et al. 1988;Schwenk, 1988, 1994) and (Mosasauroidea, Coniasaurus) tested relative to this clade.At 470 steps (Mosasauroidea, Coniasaurus) is the sister-group to Anguimorpha withSerpentes as the sister-taxon to both. At 471 steps (Mosasauroidea, Coniasaurus) is

Page 17: Caldwell, 1999a

SQUAMATE PHYLOGENY 131

the sister-group to Serpentes+Anguimorpha. For (Mosasauroidea, Coniasaurus) tobe a sister-group to any other anguimorph taxon, tree lengths must increase by 7to 15 steps beyond the tree length (TL=464) of the 18 shortest trees (Fig. 1A, B).

Hypotheses of varanoid affinities of (Mosasauroidea, Coniasaurus) requires thelongest tree lengths of any of the alternative trees tested. If Serpentes is fixed outsideof Scleroglossa or Squamata, consistent with suggestions by Underwood (1970) andRieppel (1988), than finding (Mosasauroidea, Coniasaurus) to be within Anguimorpharequires tree lengths of 470–478 steps.

Test number II: sequential analysis of fossil taxa

It has been argued that certain taxa, fossil and/or extant, act as ‘keystone’ taxathat significantly affect tree topology by altering internal nodes (Gauthier et al., 1988;Donoghue et al., 1989). Data sets excluding fossil taxa are likely to be inadequateas many fossil taxa demonstrate characters that are informative regarding ambiguousapomorphies present in crown-groups (i.e. extant taxa only). The question askedhere, in terms of this data set, is whether any or all of the fossil taxa included inthis study act as keystone taxa in the cladistic analysis (Fig. 1A, B), i.e. whether anytaxa are essential to the tree topology as reconstructed. To test this question, andto further examine the problem of (Mosasauroidea, Coniasaurus) as varanoids oranugimorphs, fossil taxa were added to a data matrix composed only of extantsquamate taxa. Each new and larger matrix was then analysed cladistically usingthe same protocol as described in Methods for the entire matrix (Appendix 2).

Extant taxa: the baseline (Fig. 3 A, B). Analysis found 160 cladograms of 441 stepseach (CI 0.707, HI 0.760, RI 0.589). A Strict Consensus Tree of these cladogramsfinds little resolution of squamate relationships that is comparable to previousphylogenies (Estes et al., 1988; Schwenk, 1988, 1994). Iguanian monophyly is notsupported, though there is support for the distinction between Iguanian taxaand all other squamates. The monophyly of Scleroglossa is supported. However,Anguimorpha and Anguoidea (Anguidae, Shinisaurus, Xenosaurus) are found to beparaphyletic. Anguidae is found to be the sister to all non-anguid scleroglossans,while Shinisaurus and Xenosaurus are successive outgroups to Anguidae + all otherscleroglossans. Varanoid monophyly is supported though ingroup relationshipsare unresolved. Scincomorpha is not recognized: Scincidae is unresolved withinScleroglossa, though Cordylidae + Lacertoidea (xantusiids, lacertids, gym-nophthalmids, teiids) is found to be a monophyletic group. Gekkotan monophylyis supported. Amphisbaenids and dibamids are unresolved within Scleroglossa.Alethinophidians and scolecophidians form a monophyletic ophidian clade that isalso unresolved within scleroglossans.

The Majority Rule Consensus Tree (Fig. 3B) supports, in more than 50% of the160 cladograms, most of the previously hypothesized squamate crown groups Camp,1923; Estes et al., 1988). The major exceptions are a polyphyletic Anguimorpha, aparaphyletic Anguoidea, and a paraphyletic Scincomorpha. A new clade is foundthat places varanoids in the sister-group position to a gekkotan–snake–amphisbaenid–dibamid clade. This entire clade shares a common ancestorwith the cordylid–lacertoid clade.

None of the 160 cladograms of extant squamates has a similar topology to any

Page 18: Caldwell, 1999a

M. W. CALDWELL132

Agamidae

Chamaeleonidae

Other Iguanians

Anguidae

Cordylidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Amphisbaenia

Alethinophidia

Scolecophidia

Gekkonoidea

Eublepharinae

Heloderma

Lanthanotus

Varanus

Scincidae

Xenosaurus

Shinisaurus

Sphenodon

BAgamidae

Anguidae

Cordylidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Gekkonoidea

Eublepharinae

Heloderma

Lanthanotus

Varanus

Scincidae

Amphisbaenia

Alethinophidia

Scolecophidia

Xenosaurus

Shinisaurus

Chamaeleonidae

Other Iguanians

Sphenodon

A51

51

100

100

78

46

100

100

100100

66

8076

10066

100100

100

98

Figure 3. Consensus trees derived from 160 cladograms (441 steps; CI 0.707; HI 0.760) showingingroup relationships of 21 extant squamate taxa using morphological data (95 osteological characters).A, Strict Consensus Tree. B, Majority Rule Consensus Tree.

of the 18 cladograms found using the complete data set. Distinction of the nodedifferentiating between scleroglossans and iguanians appears to be a constant at anylevel of observation (Camp, 1923; Estes et al., 1988; this study in its several parts).Apart from this well-supported dichotomy, previous hypotheses of monophyly forLacertoidea, Varanoidea, Ophidia, Gekkota, and Iguania are supported in both theanalysis of fossil and extant squamates (Fig. 1A, B) and by the analysis of extantsquamates only (Fig. 3A, B). The basal relationships of these and other taxa areuncertain in both studies but are clearly better resolved in the analysis includingfossil taxa.

Sequential addition of fossil taxa(1) Extant taxa + Dinilysia (Fig. 4A, B). Analysis of this matrix finds 8 trees of 443steps each (CI 0.704, HI 0.761, RI 0.602) in which Dinilysia forms a monophyleticgroup with Alethinophidia; Scolecophidia is the sister to that clade. Characters ofDinilysia are most congruent with those of alethinophidians, i.e. as crown-groupSerpentes.

The Strict Consensus Tree (Fig. 4A) shows that the addition of Dinilysia has animportant effect on the resolution of more basal scleroglossan relationships; where

Page 19: Caldwell, 1999a

SQUAMATE PHYLOGENY 133

Agamidae

Anguidae

Cordylidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Amphisbaenia

Gekkonoidea

Eublepharinae

Heloderma

Lanthanotus

Varanus

Alethinophidia

Dinilysia

Scolecophidia

Scincidae

Xenosaurus

Shinisaurus

Chamaeleonidae

Other Iguanians

Sphenodon

BAgamidae

Anguidae

Cordylidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Gekkonoidea

Eublepharinae

Heloderma

Lanthanotus

Varanus

Scincidae

Amphisbaenia

Alethinophidia

Dinilysia

Scolecophidia

Xenosaurus

Shinisaurus

Chamaeleonidae

Other Iguanians

Sphenodon

A

100100

100

100

71

9371

100

50100

93

6410079

4350100

100

100100

Figure 4. Consensus trees derived from 8 cladograms (443 steps; CI 0.704; HI 0.761) of 21 extantsquamates and the fossil taxon Dinilysia using morphological data (95 osteological characters). A, StrictConsensus Tree. B, Majority Rule Consensus Tree.

derived characters of snakes are unresolved with respect to amphisbaenids anddibamids versus varanoids, dinilysid characters resolve these problems. Addition ofDinilysia therefore influences the number of trees found (8 versus 160 for extant taxaonly). This results from the redistribution of characters found apomorphic for thesnake–dibamid–amphisbaenid clade (from the analysis of extant taxa) that no longersupport that grouping to the exclusion of a snake–varanoid clade.

(2) Extant taxa + Dinilysia + Estesia (Fig. 5 A, B). Analysis found 8 trees of 445steps each (CI 0.701, HI 0.762, RI 0.609). The addition of Estesia is important tothe resolution of characters within Varanoidea as constituted by this data set.Resolving character incongruities within varanoids, in association with congruenceadded by characters of Dinilysia, reduces the number of cladograms for extant taxaonly (i.e. 160) to only 8. This small number of trees is associated with only a 4 stepincrease in treelength (441 to 445).

Varanoids are found to be monophyletic, though with some restructuring ofrelationships within the crown-groups. The clade Heloderma–Lanthanotus is differ-entiated within Varanoidea, as are the ‘estesioids’, when Estesia is included.

Snakes are resolved in a clade with dibamids and amphisbaenids, but cannot beresolved as the sister-taxon to varanoids specifically, nor anguimorphs generally.The monophyly of Serpentes is not altered and Dinilysia remains as the sister to

Page 20: Caldwell, 1999a

M. W. CALDWELL134

Agamidae

Chamaeleonidae

Other Iguanians

Anguidae

Shinisaurus

Xenosaurus

Heloderma

Lanthanotus

Estesia

Varanus

Cordylidae

Scincidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Amphisbaenia

Gekkonoidea

Eublepharinae

Alethinophidia

Dinilysia

Scolecophidia

Sphenodon

Agamidae

Chamaeleonidae

Other Iguanians

Anguidae

Shinisaurus

Xenosaurus

Heloderma

Lanthanotus

Estesia

Varanus

Cordylidae

Scincidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Amphisbaenia

Gekkonoidea

Eublepharinae

Alethinophidia

Dinilysia

Scolecophidia

Sphenodon

BA33

100

100

100

100100 100

100

100100

100100

100

100100

100

100

100 100100

100

Figure 5. Consensus trees derived from 8 cladograms (445 steps; CI 0.701; HI 0.762) of 21 extantsquamates and the fossil taxa Dinilysia and Estesia using morphological data (95 osteological characters).A, Strict Consensus Tree. B, Majority Rule Consensus Tree.

Alethinophidia (Fig. 5A, B), while the sister-group relationship of dibamids, gekkotans,and amphisbaenids are with snakes. Scincomorpha is monophyletic, and iguanianinterrelationships remain unresolved though this clade appears to be monophyletic(Fig. 5A, B).

(3) Extant taxa + Dinilysia + Estesia + Coniasaurus (Fig. 6A, B). Analysis found 8trees of 453 steps each (CI 0.689; HI 0.766; RI 0.599). Based on this data set,Coniasaurus is found to be outside Anguimorpha, with snakes as the sister-groupto amphisbaenid–dibamid clade. Scleroglossans are distinct from iguanians, andAnguimorpha is paraphyletic.

The addition of Coniasaurus to the analysis introduces character states that identifyunsuspected incongruence within the data set; significant changes seem to consistentlyinfluence the monophyly of the Anguimorpha by destabilizing characters thatnormally support anguoid monophyly; a similar set of incongruent characters affectthe scincomorpha and ingroup structure of varanoids.

Other clades for which robust apomorphies appear to be absent are scincids. Itis also clear that characters of Coniasaurus are congruent with anguimorph stateswhen no other mosasauroids are included. The sister-group relationships of Serpentesindicate that Serpentes and dibamids–amphisbaenids are synapomorphic for a

Page 21: Caldwell, 1999a

SQUAMATE PHYLOGENY 135

Agamidae

Chamaeleonidae

Other Iguanians

Anguidae

Xenosaurus

Cordylidae

Scincidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Gekkonoidea

Eublepharinae

Amphisbaenia

Alethinophidia

Dinilysia

Scolecophidia

Heloderma

Lanthanotus

Estesia

Varanus

Shinisaurus

Coniasaurus

Sphenodon

BAgamidae

Chamaeleonidae

Other Iguanians

Anguidae

Xenosaurus

Cordylidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Scincidae

Dibamidae

Gekkonoidea

Eublepharinae

Amphisbaenia

Alethinophidia

Dinilysia

Scolecophidia

Heloderma

Lanthanotus

Estesia

Varanus

Shinisaurus

Coniasaurus

Sphenodon

A

100

100

100

100

50

100

100100

50100

100

100

100100

100100

100

100

100

100

100100

Figure 6. Consensus trees derived from 8 cladograms (453 steps; CI 0.689; HI 0.766) of 21 extantsquamates and the fossil taxa Dinilysia, Estesia, and Coniasaurus, using morphological data (95 osteologicalcharacters). A, Strict Consensus Tree. B, Majority Rule Consensus Tree.

number of character states. In all cladograms an Amphisbaenia+ ‘dibamogekkota’clade is the sistergroup to Serpentes. This clade is supported by characters oflimblessness.

(4) Extant taxa + Mosasauroidea (Fig. 7A, B). Analysis found 4 trees of 461 stepseach (CI 0.681; HI 0.770; RI 0.568). Clade structure in these four cladograms ismost similar to the that found in Test II, Number 3, above. Coniasaurus might havethe same effect if its morphology was as completely known as that of mosasauroids(e.g. Russell, 1967; Bell, 1993).

Inter-subjective consensus

Test II indicates the difficulty of congruence and consensus between subjective/qualitative phylogenetic analyses. What taxa should be included? What charactersbecome redundant or uninformative in one analysis compared to a second or thirdstudy where such characters are informative? And finally, how do we compare theresults of such studies when they can vary so widely between analyses that vary byonly a single taxon?

Page 22: Caldwell, 1999a

M. W. CALDWELL136

Agamidae

Anguidae

Xenosaurus

Cordylidae

Scincidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Gekkonoidea

Eublepharinae

Heloderma

Lanthanotus

Varanus

Alethinophidia

Scolecophidia

Mosasauroidea

Amphisbaenia

Shinisaurus

Chamaeleonidae

Other Iguanians

Sphenodon

BAgamidae

Anguidae

Cordylidae

Scincidae

Gymnophthalmidae

Teiidae

Lacertidae

Xantusiidae

Dibamidae

Gekkonoidea

Eublepharinae

Heloderma

Lanthanotus

Varanus

Alethinophidia

Scolecophidia

Mosasauroidea

Amphisbaenia

Xenosaurus

Shinisaurus

Chamaeleonidae

Other Iguanians

Sphenodon

A

100

100100

100100

100

67

100

100

100

100

100

50 50

100

100100

100

100

100

Figure 7. Consensus trees derived from 4 cladograms (461 steps; CI 0.681; HI 0.770) of 21 extantsquamates and the fossil taxon Mosasauroidea using morphological data (95 osteological characters).A, Strict Consensus Tree. B, Majority Rule Consensus Tree.

The discovery of monophyletic groups is the goal of cladistic analysis. Thetechnique used in such heuristic processes is intended to recover the order ofbranching/cladogenic events. The data set supporting this technique is derived fromthe primary analysis of homologues and the secondary discovery of synapomorphies/homology by congruence of characters. In the context of any one character matrix,the monophyletic ‘signal’, i.e. the strength of support for a clade through thecongruence of characters, is the strongest signal of relationship that persists in anycharacter matrix. This is evidenced by the stability of several clades found in thisstudy (e.g. Serpentes, Varanoidea, Iguania) throughout the various series of matricesanalysed and discussed above. In other words, these groups are ‘real’ and the datasets will continually support their reconstruction.

This does not mean that monophyletic groups persist despite our best efforts toobscure them. On the contrary, monophyletic groups can easily be made paraphyleticor polyphyletic by experimental inaccuracies. Tests of similarity, topological relation,and congruence will present synapomorphies, and by phylogenetic interpretation,homologies, only when total evidence in both characters and taxa is considered, andonly when the foundation of phylogenetic analysis rests on rigorous characterization ofprimary homologues. On this last point the subjectivity of analysis can be minimizedbut not extinguished; errors in characterizing primary homologues will continue. It

Page 23: Caldwell, 1999a

SQUAMATE PHYLOGENY 137

is unrealistic to expect that characters and their states can be rigorously defined thefirst, second, or tenth time that they are characterized for cladistic analysis. In otherwords, satisfactory means must be explored to find inter-subjective consensus betweendisparate data sets and disparate phylogenetic hypotheses.

CONCLUSIONS

Snakes, Mosasauroidea, Coniasaurus; Pythonomorpha

The prevalent opinion on mosasaurian relationships, held since Cuvier (1802),was that mosasaurs and related forms were aquatic varanoid lizards. The controversyand debate on mosasaurian phylogeny began when Cope (1869) identified char-acteristics of mosasaurs that he felt merited recognition of a closer relationshipbetween mosasaurs and snakes, than between mosasaurs and any other lizards.Further, Cope’s hypothesis did not recognize an any close relationship betweenmosasaurs and varanoid lizards.

A series of papers and responses, arguing points of morphology and phylogeny,were exchanged between Cope and Baur (see Cope 1869, 1895a, b, 1896a, b, andBaur, 1895, 1896). Baur took the position that mosasaurs were derived, aquaticvaranoid lizards, and that there were no important similarities between ophidiansand mosasaurs. Owen (1877), Boulenger (1891), and Osborn (1899) also extendedopinions on this question of squamate phylogeny; Owen criticized Cope, whileBoulenger and Osborn recognized some merit in Cope’s arguments. The debatehas been long-lived as successive generations of systematists have continued toexamine the relationships of aigialosaurs, mosasauroids, snakes and dolichosaurs(Fejervary, 1918; Nopsca, 1903, 1908, 1923; McDowell & Bogert, 1954; Rieppel,1988; Carroll & deBraga, 1992; deBraga & Carroll 1993; Caldwell et al., 1995;Caldwell, 1996; Lee, 1997; Caldwell & Lee, 1997).

The hypothesis of squamate phylogeny presented here (Fig. 1A, B) is derivedfrom a research program that originated with the re-characterization of the externalbony nares of varanoids and mosasauroids (Caldwell et al., 1995). The ‘posteriorlyretracted’ nares of Varanus, and to a far lesser degree Lanthanotus, had for a very longtime been ‘homologized’ with the ‘retracted nares’ of mosasauroids (aigialosaurs andmosasaurs). Caldwell et al. (1995) showed that the only ‘homologue’ shared byvaranids and mosasauroids is the presence of a large, empty space on the dorsalsurface of the muzzle (the structure of the soft tissues in mosasaurs is unknown);nine other characters, derived from the bony elements framing this space, werefound to have different states that were not synapomorphic for mosasauroids andvaranoids. Therefore, if mosasauroids and Coniasaurus were indeed derived varanoids,or anguimorphs/platynotans of some sort, it seems logical that this relationshipwould have been supported in at least some of the shortest trees. The introductionof the Cretaceous lizard Estesius into the data matrix found this taxon to be avaranoid. If mosasauroids and Coniasaurus were varanoids they too would have beennested within that clade. This was not found to be the case. Other characters ofmosasaurs, considered homologous with varanoids, were found by Caldwell et al.(1995) to be plesiomorphic for all included anguimorphs. This paper details a similarsuite of characters that are synapomorphic for mosasaurs and other squamates atthe level of Scleroglossa, not Anguimorpha, and certainly not Varanoidea.

Page 24: Caldwell, 1999a

M. W. CALDWELL138

Historical limitations on ascertaining the relationships of mosasaurs, and similarlysnakes, have been due to a priori assumptions of crown-group relationship with othersquamates. Where the ingroup examined by Caldwell et al. (1995) and Lee (1997)was limited to varanoids and some anguids, the ingroup tested here is not. Wherethe ingroup examined by Estes et al. (1988) was limited to extant squamates, theingroup tested here is not. The taxonomic scope of this study is the logical test ofhypotheses generated from all previous studies and therefore relies heavily on thedata and hypotheses they produced. As phylogeny is historically contingent, so isthis analysis.

Despite awareness of Cope’s (1869, 1895a, b, 1896a, b) hypothesis of a closerelationship of mosasaurs with snakes, the initial intention of this study was toexamine the phylogenetic relationships of coniasaurs and dolichosaurs relative tomosasaurs and aigialosaurs, and not to address snake-lizard relationships, nor toinvestigate Pythonomorpha sensu stricto Cope (1869). However, the synapomorphiesfound supporting a clade composed of Serpentes, Mosasauroidea, and Coniasaurusmerit serious consideration of the included taxa as a ‘real’ clade. Detailed charac-terization of the morphology of more poorly known, but putatively closely relatedsquamates such as adriosaurs, acteosaurs, dolichosaurs, and aigialosaurs, no matterhow incomplete the material may be, might alter the understanding of theirphylogeny, and by extension, squamate phylogeny. To this end, redescription of theholotype and referred material of Coniasaurus crassidens has been completed (Caldwell& Cooper, 1998), as has the description of a new species of Coniasaurus (Caldwell,in press).

Squamata

As indicated by previous studies (Gauthier et al., 1988; Donoghue et al., 1989),and as discussed above, this study shows that the addition of fossil taxa to phylogeneticanalyses can significantly alter our understanding of the interrelationships of or-ganisms. Fossil taxa, despite their acknowledged incompleteness, offer data that iscritical to the effective application of the principle of total evidence in phylogenyreconstruction (Kluge, 1989). Limiting data to specific subsets (e.g. extant vs. fossil)is logically inconsistent not only for characters but also for taxa. It is also evidentthat fossils need not be treated in any particularly unique manner relative to extanttaxa, nor in relation to missing data (Kluge, 1990; Rieppel, 1994).

Squamata as a whole appears to be a robust group, though admittedly, thisanalysis did not specifically address this question. Two major clades of squamatesare recognized: one clade composed of all iguanian taxa, and the other composedof all non-iguanian squamates including pythonomorphs. Scleroglossa requires nomodification of taxic composition (Estes et al., 1988) and is retained to define thenode differentiating pythonomorphs and all other non-iguanians. Assertions thatLacertilia or ‘lizards’ are paraphyletic (Estes et al., 1988) and that snakes were derivedfrom within “lizards” (Camp, 1923), are supported by the re-structuring of Squamataregardless of the presence or absence of fossil taxa.

The basic composition of Anguimorpha (Anguoidea + Varanoidea) and Scin-comorpha (Scincoidea + Lacertoidea) is by no means well supported using onlyosteological data (this data set). At this level among crown-groups, clade structureis re-organized for a number of taxa. The lack of resolution for Anguoidea found

Page 25: Caldwell, 1999a

SQUAMATE PHYLOGENY 139

in the analysis of only extant taxa (Fig. 3) is resolved and paraphyly of extantXenosauridae is recognized in the full data set (Fig. 1A, B). Addition of the fossilEstesia mongoliensis (Norell et al., 1992) alters varanoid ingroup structure by findingLanthanotus be the sister-group to Heloderma, and Varanus the sister to all three (Figs1 and 4).

An important and well supported clade are the ‘Dibamogekkota’ (Fig. 1). Dibamidsshare eight characters with gekkotans. Amphisbaenians share nine characters with‘Scincogekkonomorphs’ and are the sister-group to ‘Dibamogekkota’. Only onecharacter supporting relationships for either taxon concerns features of limbs orgirdles. Based on these results, new relationships are found that support previoushypotheses of relationship between ‘scincogekkonomorphs’ and ‘dibamogekkotans’(Rieppel, 1981, Rieppel, 1984; Greer, 1985). Following suggestions by Rieppel(1984) and Greer (1985), subsequent investigations of the relationships proposedhere would benefit by closer examination of dibamids, amphisbaenids, Anelytropsis,feylinine and acontine scincomorphs, and gekkonids.

Snake origins

The problem of snake origins is now the focus of an extended examination ofother taxa including the very poorly known ‘ophiomorphs’ described by Kornhuber(1873) and Haas (1979, 1980a, b; see also Calligaris [1988]). Though no predictionis possible regarding the information and resolution that new data will provide, itis already evident that the phylogenetic relationships of snakes as reconstructed heredrastically affect hypotheses of snake origins and hypotheses on the sister-grouprelationships of snakes (Caldwell & Lee, 1997; Lee & Caldwell, 1998).

Problems of snake relationship within Squamata affect adaptationist hypothesesof a fossorial versus marine origin for snakes (Bellairs & Underwood, 1951; Under-wood, 1967; Bellairs, 1972; for a review see Rieppel, 1988). If the sister-group ofsnakes is a group of aquatic squamates that show a marked evolution towardshighly modified limbs (Caldwell, 1996), then a reasonable alternative exists for theconventional view of a fossorial origin for snakes and snake characters. Mosasaurlimb and girdle structure suggests that these animals were obligatorily aquatic(Russell, 1967; Caldwell, 1996), though this is not so readily apparent for aigialosaurs(Caldwell et al., 1995), dolichosaurs and coniasaurs (Owen, 1850; Bell et al., 1982;Polcyn & Bell, 1994; Caldwell & Cooper, 1998).

There is no doubt that subsequent snake evolution has involved adaptationsamong various taxa for fossorial habits. However, the question remains regardingthe origin of the basic body plan of snakes: elongate, limbless, squamates with smallheads and highly mobile skull bones and a bony cranium. All of these characters,with the exception of complete limblessness, are present in Pachyrhachis problematicus(Caldwell & Lee, 1997), the earliest and most complete snake known. This earlysnake shows numerous adaptations to marine environments (Lee & Caldwell, 1998)that could have been co-opted in later snake evolution for adaptations to terrestrial,and specifically, fossorial habits.

Later snake evolution towards burrowing habits has certainly produced numerouscranial specializations that characterize the highly fossorial scolecophidians versusthe more adaptively diverse alethinophidians (i.e. fossorial, aquatic, arboreal, ter-restrial). If fossoriality is primitive for snakes, as judged against the extreme adaptations

Page 26: Caldwell, 1999a

M. W. CALDWELL140

of scolecophidians, then numerous features of alethinophidians and Pachyrhachis mustbe explained as very complex reversals.

Invoking complex processes to defend untested polarities for multiple charactersis unwarranted, especially if the common ancestor of Serpentes was less fossorialthan either descendant lineage. As hypothesized in the analysis given in this paper,snakes and mosasauroids/coniasaurs share a limbed common ancestor that waslikely aquatic, not fossorial (Fig. 1A, B). Therefore, a ‘propensity’ for altering limbdevelopment was common to both descendent lineages. Serpentes evolved completeloss of the limbs. Coniasaurs and dolichosaurs reduced the overall size of their limbs,making the front limb smaller than the rear limb. Mosasaurs on the other hand,evolved the most highly modified limbs of the entire clade, producing paddle-likelimbs (Caldwell, 1996). Among pythonomorphs, one clade retained limbs andadopted aquatic habits while the other lost its limbs and adapted to a wide rangeof terrestrial habitats.

ACKNOWLEDGEMENTS

For assistance while gathering data, I thank S. Chapman, J. Cooper, A. Currant,K. deQuieroz, J. Evans, D. Frost, C. Price, and V. Sowiak. I thank M. Wilson andJ. Clark for use of lab space and equipment. I thank J. Clark and O. Rieppel fordiscussion and criticism of both the manuscript and my ideas. Research was supportedby a Natural Sciences and Engineering Research Council of Canada (NSERC)Postdoctoral Fellowship.

REFERENCES

Baur G. 1895. Cope on the temporal part of the skull, and on the systematic position of theMosasauridae – A reply. American Naturalist 29: 998–1002.

Baur G. 1896. The paroccipital of the Squamata and the affinities of the Mosasauridae once more.A rejoinder to Professor E.D. Cope. American Naturalist 30: 143–152.

Bell GL. 1993. A phylogenetic revision of Mosasauroidea (Squamata). Ph.D. dissertation, Universityof Texas, Austin, Texas, 293 pp..

Bell BA, Murry PA, Osten LW. 1982. Coniasaurus Owen, 1850 from North America. Journal ofPaleontology 56: 520–524.

Bellairs A d’A, Underwood G. 1951. The origin of snakes. Biological Reviews of the CambridgePhilosophical Society 26: 193–237.

Bellairs A d’A. 1972. Comments on the evolution and affinities of snakes. In: Joysey K, Kemp T,eds. Studies in Vertebrate Evolution Edinburgh: Oliver and Boyd, 157–172.

Boulenger G. 1891. Notes on the osteology of Heloderma horridum and H. suspectum with remarks onthe systematic position of the Helodermatidae and on the vertebrae of the Lacertilia. Proceedings ofthe Zoological Society of London 1891: 109–118.

Caldwell MW. 1996. Ontogeny and phylogeny of the mesopodial skeleton in mosasauroid reptiles.Zoological Journal of the Linnean Society 116: 407–436.

Caldwell MW, Carroll RL, Kaiser H. 1995. The pectoral girdle and forelimb of Carsosaurusmarchesetti (Aigialosauridae), with a preliminary phylogenetic analysis of mosasauroids and varanoids.Journal of Vertebrate Paleontology 15: 516–531.

Caldwell MW, Cooper J. (1998). Redescription, Palaeobiogeography, and Palaeoecology of Con-iasaurus crassidens Owen, 1850 (Squamata) from the English Chalk (Cretaceous; Cenomanian).Zoological Journal of the Linnean Society (in press).

Page 27: Caldwell, 1999a

SQUAMATE PHYLOGENY 141

Caldwell MW, Lee MSY. 1997. A snake with legs from the marine Cretaceous of the Middle East.Nature 386: 705–709.

Calligaris R. 1988. I rettili fossili degli “Strati calcariei ittilitici di comeno” e dell’isola di Lesina. Attidel Museo Civico di Storia naturale Trieste 41: 85–125.

Camp CL. 1923. Classification of the Lizards. Bulletin of the American Museum of Natural History 48:289–481.

Camp CL. 1942. California Mosasaurs. University of California Memoirs 13: 1–68.Carroll RL, DeBraga M. 1992. Aigialosaurs: Mid-Cretaceous varanoid lizards. Journal of Vertebrate

Paleontology 12: 66–86.Cope ED. 1869. On the reptilian order Pythonomorpha and Streptosauria Proceedings of the Boston

Society of Natural History 12: 250–261.Cope ED. 1895a. Baur on the temporal part of the skull, and on the morphology of the skull in the

Mosasauridae. American Naturalist 29: 855–859.Cope ED. 1895b. Reply to Dr. Bauer’s critique of my paper on the paroccipital bone of the scaled

reptiles and the systematic position of the Pythonomorpha. American Naturalist 29: 1003–1005.Cope ED. 1896a. Criticism of Dr. Bauer’s rejoinder on the homologies of the paroccipital bone, etc.

American Naturalist 30: 147–149.Cope ED. 1896b. Boulenger on the difference between Lacertilia and Ophidia; and on the Apoda.

American Naturalist 30: 149–152.Costelli JJ, Hecht MK. 1971. The postcranial osteology of the lizard Shinisaurus: The appendicular

skeleton. Herpetologica 27: 87–98.Cundall D, Rossman DA. 1993. Cephalic anatomy of the rare Indonesian snake Anomochilus weberi.

Zoological Journal of the Linnean Society 109: 235–273.Cundall D, Wallach V, Rossman DA. 1993. The systematic relationships of the snake genus

Anomochilus. Zoological Journal of the Linnean Society 109: 275–299.DeBraga M, Carroll RL. 1993. The origin of mosasaurs as a model of macroevolutionary patterns

and processes. Evolutionary Biology 27: 245–322.Donoghue MJ, Doyle JA, Gauthier J, Kluge AG, Rowe T. 1989. The importance of fossils in

phylogeny reconstruction. Annual Reviews of Ecology and Systematics 20: 431–460.Estes R, deQuieroz K, Gauthier J. 1988. Phylogenetic relationships within Squamata. In: Estes

R, Pregill G, eds. Phylogenetic Relationships of the Lizard Families. Stanford: Stanford University Press,119–281.

Estes R, Frazetta TH, Williams EE. 1970. Studies on the Fossil Snake Dinilysia patagonica Woodward:Part 1. Cranial Morphology. Bulletin of the Museum of Comparative Zoology 140: 25–74.

Etheridge R, deQuieroz K. 1988. A phylogeny of Iguanidae. In: Estes R, Pregill G, eds. PhylogeneticRelationships of the Lizard Families. Stanford: Stanford University Press, 283–367.

Fejervary GJ. de. 1918. Contributions to a monography on fossil Varanidae and Megalanidae.Budapesch, Annales Historico-Naturales Musei Nationalis Hungarici 16: 341–467.

Frost DR, Etheridge R. 1989. A phylogenetic analysis and taxonomy of Iguanian lizards (Reptilia:Squamata). The University of Kansas Museum of Natural History, Miscellaneous Publications 81: 1–65.

Gauthier J, Kluge AG, Rowe T. 1988. Amniote phylogeny and the importance of fossils. Cladistics4: 105–209.

Greer A. 1985. The relationships of the lizard genera Anelytropsis and Dibamus. Journal of Herpetology19: 116–156.

Haas G. 1979. On a new snakelike reptile from the Lower Cenomanian of Ein jabrud, near Jerusalem.Bulletin du la Museum Nationale d’Histoire Naturelle, Paris, Series 4, 4: 51–64.

Haas G. 1980a. Pachyrhachis problematicus Haas, a snakelike reptile from the Lower Cenomanian:ventral view of the skull. Bulletin du la Museum Nationale d’Histoire Naturelle, Paris, Series 4, 2: 87–104.

Haas G. 1980b. Remarks on a new ophiomorph reptile from the Lower Cenomanian of Ein Jabrud,Israel. In: Jacobs LL, ed. Aspects of Vertebrate History, in Honor of E.H. Colbert. Flagstaff: Museum ofNorthern Arizona Press, 177–102.

Hecht MK. 1982. The vertebral morphology of the Cretaceous snake, Dinilysia patagonica Woodward.Neues Jahrbuch fur Geologie und Palaontologie, Monatshefte 1982: 523–532.

Hecht MK, Costelli J, Jr. 1969. The postcranial osteology of the lizard Shinisaurus 1. The vertebralColumn. American Museum Novitates 2378: 1–21.

Kluge AG. 1967. Higher taxonomic categories of Gekkonid lizards and their evolution. Bulletin of theAmerican Museum of Natural History 135: 1–59.

Kluge AG. 1976. Phylogenetic relationships in the lizard family Pygopodidae: an evaluation of theory,methods, and data. Miscellaneous Publications Museum of Zoology, University of Michigan 152: 1–72.

Page 28: Caldwell, 1999a

M. W. CALDWELL142

Kluge AG. 1987. Cladistic relationships in the Gekkonoidea (Squamata, Sauria). Miscellaneous PublicationsMuseum of Zoology, University of Michigan 173: 1–54.

Kluge AG. 1989. A concern for evidence and a phylogenetic hypothesis of relationships amongepicrates (Boidae, Serpentes). Systematic Zoology 38: 7–25.

Kluge AG. 1990. On the special treatment of fossils and taxonomic burden: A response to Loconte.Cladistics 6: 191–193.

Kluge AG. 1991. Boine snake phylogeny and research cycles. Miscellaneous Publications Museum ofZoology, University of Michigan 178: 1–58.

Kornhuber A. 1873. Uber einen neuen fossilen Saurier aus Lesina. Herausgegeben von der k. k. geologischenReichsanstalt, Wien 5: 75–90.

Lee, MSY. 1997. The phylogeny of varanoid lizards and the affinities of snakes. Philosophical Transactionsof the Royal Society of London B 352: 53–91.

Lee MSY, Caldwell MW. 1998. Anatomy and relationships of Pachyrhachis problematicus, a primitivesnake with hindlimbs. Philosophical Transactions of the Royal Society of London B (in press).

Maddison WP, Maddison DR. 1992. MacClade; Analysis of phylogeny and character evoluation. Version 3.Sinauer Associates, Sunderland, Massachusetts.

Maddison WP, Donoghue MJ, Maddison DR. 1984. Outgroup analysis and parsimony. SystematicZoology 33: 83–103.

McDowell SB, Bogert CM. 1954. The systematic position of Lanthanotus and the affinities of theanguimorph lizards. Bulletin of the American Museum of Natural History 105: 1–142.

Nixon KC, Carpenter JW. 1993. On outgroups. Cladistics 9: 413–426.Nopcsa F. 1903. Uber die Varanus antigen Lacerten Istriens. Beitrage zur Palaontologie und Geologie

Osterreich-Ungarns und des Orients 15: 31–42.Nopcsa F. 1908. Zur Kenntnis der fossilen Eidechsen. Beitrage zur Palaontologie und Geologie Osterreich-

Ungarns und des Orients 21: 33–62.Nopcsa F. 1923. Eidolosaurus und Pachyophis. Zwei neue Neocom-Reptilien. Palaeontographica 65: 97–154.Norell MA, McKenna MC, Novacek MJ. 1992. Estesia mongoliensis, a new fossil varanoid from the

Late Cretaceous of the Barun Goyot Formation of Mongolia. American Museum Novitates 3045: 1–24.Osborn HF. 1899. A complete mosasaur skeleton, osseous and cartilaginous. Memoirs of the American

Museum of Natural History 1: 167–188.Owen R. 1850. Description of the Fossil Reptiles of the Chalk Formation. In: Dixon F, ed. The

Geology and Fossils of the Tertiary and Cretaceous Formations of Sussex London: Longman, Brown, Greenand Longman, 378–404.

Owen R. 1877. On the rank and affinities in the reptilian class of Mosasauridae, Gervais. QuarterlyJournal of the Geological Society of London 33: 682–715.

Patterson C. 1982. Morphological characters and homology. In: Joysey KA, Friday AE, eds. Problemsof phylogenetic reconstruction Vol. 21. New York: Academic Press, 21–74.

Polcyn M, Bell GL. 1994. Coniasaurus crassidens and its bearing on varanoid-mosasauroid relationships.Journal of Vertebrate Paleontology, Supplemental 14: 42A.

Rage J-C, Albino AM. 1989. Dinilysia patagonica (Reptilia, Serpentes): materiel vertebral additionneldu Cretace superieur d’Argentine. Etude complementaire des vertebres, variations intraspecifiqueet intracolumnaires. Neues Jahrbuch fur Geologie und Palaontologie, Monatshefte 1989: 433–447.

Rage J-C, Richter A. 1994. A snake from the Lower Cretaceous (Barremian) of Spain: The oldestknown snake. Neues Jahrbuch fur Geologie und Palaontologie, Monatschefte 1994: 561–565.

Rieppel O. 1978. The evolution of the naso-frontal joint in snakes and its bearing on snake origins.Zeitschrift fur Zoologie, Systematik und Evolutionforschung 16: 14–27.

Rieppel O. 1979. A cladistic classification of primitive snakes based on skull structure. Zeitschrift furZoologie, Systematik und Evolutionforschung 17: 140–150.

Rieppel O. 1980a. The phylogeny of anguinomorph lizards. Basel: Birkhauser Verlag.Rieppel O. 1980b. The sound-transmitting apparatus in primitive snakes and its phylogenetic

significance. Zoomorphology 96: 45–62.Rieppel O. 1981. The skull and adductor musculature in some burrowing scincomorph lizards of

the genus Acontias, Typhlosaurus, and Feylinia. Journal of Zoology, London. 195: 493–528.Rieppel O. 1983. A comparison of the skull of Lanthanotus borneensis (Reptilia: Varanoidea) with the

skull of primitive snakes. Zeitschrift fur Zoologie, Systematik und Evolutionforschung 21: 142–153.Rieppel O. 1984. The cranial morphology of the fossorial lizard genus Dibamus with a consideration

of its phylogenetic relationships. Journal of Zoology, London 204: 289–327.Rieppel O. 1988. A review of the origin of snakes. Evolutionary Biology 22: 37–130.

Page 29: Caldwell, 1999a

SQUAMATE PHYLOGENY 143

Rieppel O. 1994. The role of palaeontological data in testing homology by congruence. ActaPalaeontologica Polonica 38: 295–302.

Russell DA. 1967. Systematics and morphology of American mosasaurs. Peabody Museum of NaturalHistory, Yale University Bulletin 23: 1–241.

Schwenk K. 1988. Comparative morphology of the lepidosaur tongue and its relevance to squamatephylogeny. In: Estes R, Pregill G, eds. Phylogenetic Relationships of the Lizard Families. Stanford: StanfordUniversity Press, 569–598.

Schwenk K. 1994. Why snakes have forked tongues. Science 263: 1573–1577.Swofford DL. 1993. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.1.1.. Laboratory of Molecular

Systematics, Smithsonian Institution, Washington D.C.Underwood G. 1957. Lanthanotus and the anguimorphan lizards: A critical review. Copeia 1957:

20–30.Underwood G. 1967. A contribution to the classification of snakes. Publication of the British Museum

(Natural History) 653: 1–179.Wu X-C, Brinkman DB, Russell AP. 1996. Sineoamphisbaena hexatabularis, an amphisbaenian

(Diapsida: Squamata) from the Upper Cretaceous redbeds at Bayan Mandahu (Inner Mongolia,People’s Republic of China), and comments on the phylogenetic relationships of the Amphisbaenia.Canadian Journal of Earth Sciences 33:, 541–577.

APPENDIX 1

Character and character state descriptions

Numbers in parentheses indicate the character number of Estes et al. (1988), from which thecharacter was derived. Numbers in boldface indicate new characters, or highly altered and rewrittencharacters taken from Estes et al. (1988).

1 (3). Nasals: paired (0); fused (1).2 (4). Nasal and prefrontal bones contact: present (0); absent (1).3 (5). Prefrontal contact with posterior orbital bones above orbit: no contact (0); contact (1).4 (6). Frontals: paired (0); fused (1).5 (7). Lateral borders of frontals: more or less parallel (0); constricted between orbits (1).6 (8). Frontal Shelf: lacking broad shelf below nasals (0); broad shelf present (1).7 (9). Descending processes of frontals, participation in orbitonasal fenestra: weakly developed andprefrontal participating in wide orbitonasal fenestra (0); prominently developed and prefrontals narrowlyor not at all in margins of narrow orbitonasal fenestra (1).8 (10). Median contact of descending process of frontals: not in contact (0); in contact (1).9 (11). Frontal tabs projection posteriorly over parietal: absent (0); present (1).10 (12). Postfrontal: present as separate element (0); absent as separate element (1).11 (13). Postfrontal forking: sub-triangular, not forked medially (0); semilunate, forked medially, claspingfrontoparietal suture (1).12 (16). Postorbital: present as separate element (0); absent as separate element (1).13 (17). Postorbital contribution to posterior border of orbit: one half of posterior orbital border andhas strong ventral process (0); less than one half of orbit and is temporal bone with reduced ventralprocess (1).14 (18). Jugal-squamosal contact on supratemporal arch: no contact (0); contact (1).15. Supratemporal fenestra: widely open and supratemporal arch present (0); open and no arch (1);closed, arch present (2).16 (21). Parietals: paired (0); fused (1).17 (22). Parietal tabs: absent (0); present (1).18 (23). Parietal down growths anterior to epipterygoid: absent (0); present, extending to, or almost tothe epipterygoids (1).19 (26). Pineal foramen: present (0); absent (1).20 (27). Posterior extent of maxilla: beneath orbits (0); just beyond anterior extent of orbits (1).21 (30). Lacrimal foramen number: single (0); double (1).22 (31). Anteroventral border of orbit: formed by maxilla with jugal confined to medial surface ofmaxilla (0); formed by jugal (1).

Page 30: Caldwell, 1999a

M. W. CALDWELL144

23 (32). Jugal-postorbital bar: jugal large, postorbital bar complete (0); jugal reduced or absent,postorbital bar incomplete (1).24 (33). Squamosal: present (0); absent (1).25 (34). Dorsal process of squamosal: present (0); absent (1).26 (35). Supratemporal: present (0); absent (1).27 (36). Palpebral ossifications: absent (0); absent (1).28 (37). Pterygoid lappet of quadrate: present (0); absent (1).29 (38). Vomer fusion: absent (0); present (1).30 (39). Vomer size: short, less than one half length maxillary tooth row (0); elongate, more than halfof maxillary tooth row (1).31 (40). Median contact of septomaxillae: widely separated (0); meet or nearly meet on midline inraised crest (1).32 (41). Septomaxillae: flat or concave (0); expanded and convex (1).33. Septomaxilla posterior extension: short posterior processes (0); long posterior processes contactingor close to prefrontals (1).34 (42). Posterior border of opening for Jacobsen’s organ: not closed by bone (Palaeochoanate) (0);closed by contact of maxilla and vomer (neochoanate) (1); closed by contact of vomer and septomaxilla(2).35 (43). Medial extensions of palatine forming air passages for bony secondary palate: absent (0);present (1).36 (44). Choanal fossae of palatines: small in relation to palatine size (0); large (1).37 (45). Ectopterygoid contact with palatine: no contact (0); contact (1).38 (46). Ectopterygoid size and suborbital fenestra: slender and fenestra wide (0); expanded and fenestrarestricted (1).39 (47). Epipterygoid: present (0); absent (1).40 (49). Alar process of prootic: short a(0); elongate (1).41 (50). Supratrigeminal process of prootic: absent or weak (0); finger-like above notch (1).42 (51). Opisthotic-exoccipital fusion: bone separate (0); fused (1).43 (52). Enclosure of lateral head vein in bony canal formed by crista prootica: no enclosure (0);enclosure (1).44 (53). Posterior opening of vidian canal: within basisphenoid (0); at basisphenoid-prootic suture (1);within prootic (2).45 (54). Origin of jaw adductor musculature: dorsal surface of parietal (0); ventral surface of parietal(1).46 (55). Meckel’s canal enclosure: dentary forms open groove (0); bony dentary tube formed by unionof upper and lower borders of canal (1); dentary tube closed and fused (2).47 (56). Intramandibular septum of Meckel’s canal: subdivision anterior to posterior end of tooth row,intramandibular septum poorly developed (0); subdivision occurs near posterior end of tooth row withwell developed septum (1).48 (57). Meckel’s canal exposure: medial (0); ventral (1).49 (58). Sub-dental shelf size: small or absent (0); large (1).50 (60). Dorsal extension of dentary coronoid process contact with coronoid: absent (0); present (1).51 (61). Lateral view of disarticulated surangular: tapers anteriorly, pointed distally (0); expandedanterodorsally and nearly vertical at anterior margin (1); similar to (1) but anterior end of surangularterminates closer to level of coronoid eminence on surangular (2).52. Contact between angular and splenial: overlapping-interdigitating (0); no overlap, vertical articularfaces (1); vertical faces, with lateral lappet from angular behind splenial [Lanthanotus condition] (2).53 (64). Dentary-Postdentary articulation, in medial view: extensive overlap, tongue and groove (0):reduced overlap (1).54 (67). Splenial-dentary suture: extensive bone-bone contact (0); reduced bone-bone contact (1).55 (69). Anterior extension of coronoid: short (0); elongate (1).56 (70). Anterior end of coronoid: clasps dentary laterally and medially (0); meets dentary directly, nooverlap (1); only clasps dentary medially (2).57 (72). Angular: present (0); absent (1).58 (73). Prearticular crest: absent (0); present (1).59 (74). Retroarticular process: sulcus or pit present (0); absent (1).60 (79). Retroarticular process torsion: not twisted (0); twisted (1).61 (81). Adductor fossa size: small or moderate (0); expanded, inflated, widely open (1).

Page 31: Caldwell, 1999a

SQUAMATE PHYLOGENY 145

62. Mandibular symphysis: bony joint formed (0); no bony joint (1).63 (82). Palatine teeth: present (0); absent (1).64 (83). Pterygoid teeth: present (0); absent (1).65 (84). Marginal tooth implantation: pleurodont (0); acrodont (1).66 (85). Marginal tooth replacement: lingual, large resorption pits (iguanid type)(0); posterolingual,small pits (1); posterolingual no pit (2); posterolingual, no pit, tooth rotates from horizontal to vertical(3).67 (86). Basal infolding of marginal teeth: non-plicidentine (0); plicidentine (1).68 (87). Step or offset in tooth margin of maxilla: absent (0); present (1).69 (88). Scleral ossicle number: more than 14 (0); less than 14 (1).70 (91). Second ceratobranchials: present (0); absent (1).71 (92). Condyle-cotyle orientation: little or slight obliquity (0); strongly oblique (1).72 (94). Centrum constriction anterior to condyles: not constricted (0); constricted (1).73 (96). Zygosphenes and zygantra: present (0); absent (1).74. Number of cervical vertebrae: 7–9 (0); 6 or less (1).75 (97). Cervical intercentra position: intervertebral (0); under anterior part of following centrum (1);under posterior part of preceding centrum (2).76 (99). Posterior trunk intercentra: present (0); absent (1).77 (100). Transverse processes of caudal vertebrae: single pair or two pairs converging (0); two pairsdiverging transverse processes; anterior pair of transverse processes absent (1); anterior pair of transverseprocesses absent (2).78 (103). Autotomy septa in caudal vertebrae: present (0); absent (1).79 (106). Number of presacrals: less than 26 (0); greater than 26 (1).80 (109). Rib attachments on sternum: five (0); four (1); three (2); two or less (3).81 (110). Postxiphisternal ribs: none continuous (0); some continuous (1).82 (111). Scapular emargination: absent (0); present (1).83 (114). Epicoracoid cartilage extent: contacts mesoscapula and suprascapula (0); no contact (1).84 (115). Clavicle: present (0); absent (1).85 (116). Clavicle shape: simple curved rod (0); angulated curving anteriorly (1).86 (117). Dorsal articulation of clavicle with scapula: present (0); absent (1).87 (118). Interclavicle: present (0); absent (1).88 (11). Interclavicle lateral process: present (0); absent (1).89 (120). Interclavicle shape and size of anterior process: t or anchor-shaped, anterior process smallor absent (0); cruciform, large anterior process (1).90 (121). Sternal fontanelle: absent (0); present (1).91 (122). Ectepicondylar foramen: present (0); absent (1).92 (125). Postcloacal bones: absent (0); present (1).93. Osteoderms: absent (0); present (1).94 (130). Long bone epiphyses: present (0); absent (1).95. Pubis: short, symphysial process short, ventrally directed, pubic tubercle posterodorsally placed (0);long, symphysis narrower, ventrally directed, pubic tubercle anteroventral (1); as 1 but symphsialprocess elongate and anteriorly directed.

Page 32: Caldwell, 1999a

M. W. CALDWELL146

APPENDIX 2

Question marks indicate missing character states (usually for fossil taxa). Hyphens indicate characterstates where the character is not applicable. For polymorphic characters, letters are used in the datamatrix to represent the various state combinations: A=(0,1); B=(0,1,2); C=(0,2); D=(1,2); E=(1,3);F=(0,1,2,3)

5 10 15 20 25 30 35 40

Agamidae 00011 10001 -0010 1A000 0A000 A0100 00000 000A0Anguidae 0AA1A 01A00 10100 1000A 01AA1 A1101 11000 10001Chamaeleonidae 10A11 100A1 -1010 1A0A0 0A000 00110 00000 00010Cordylidae 0A010 01000 10102 1A1A0 0A001 0A101 11000 10001Dibamidae 01000 01001 11--1 11011 --1A1 10101 11021 110A0Gekkonoidea 0AAA0 01A00 11101 00010 001A1 101A1 11000 1A001Eublepharinae 01010 01100 11101 10010 -1101 A0101 11000 11001Gymnophthalmidae 0A01A 01A10 10100 11110 01001 000A1 11000 10001Heloderma 00100 01100 11101 10011 01001 00001 11000 11101Other Iguanians 0001A 1A000 00010 1A0A0 01000 00A00 00000 000AALacertidae 0100A 01001 11102 110A0 0A001 01001 11000 10001Lanthanotus 10100 01000 11101 10011 11001 00101 11010 11101Scincidae 00A00 01000 1A112 1AAA0 0AA01 0A1A1 11001 10001Teiidae 0A010 00000 00110 111A0 0100A 00001 11000 11101Varanus 11000 01101 11100 10001 11101 01101 11010 11A01Estesia 00100 01001 1?100 10011 01001 0??01 11010 111??Xantusiidae 00000 01A01 11102 A11A1 00001 00111 11000 11101Xenosaurus 00011 01001 11110 10000 01000 00101 11000 10001Shinisaurus 00011 00001 11100 10000 01001 01100 10000 10001Amphisbaenia 01100 01101 11--1 1?0A0 01A01 10101 11010 1A0A1Alethinophidia 00A00 01101 -00-1 10010 0011- 00101 11120 1001-Scolecophidia A0000 01101 -1--1 10010 -011- 10101 11120 1--11-Dinilysia 00000 01100 10001 1??10 0011- 00101 11??0 1001-Mosasauroidea 01010 00011 11100 11001 01000 00101 11100 10001Coniasaurus ??010 0001? ????0 11??1 ???0? ???01 ???0? ?????Sphenodon 00000 00000 10000 00000 00000 10000 00000 00000

45 50 55 60 65 70 75 80

Agamidae A100A 00000 00000 00000 00111 -001A 0A101 1010BAnguidae 0101A 0110A 00000 00011 00AA0 10011 001A2 10A11Chamaeleonidae 0100A 0AA00 000-0 000-0 00111 -0011 00110 10103Cordylidae 010A1 A0011 00000 -00A1 001A0 0001A 00A0A 1A0A0Dibamidae 00020 2?-11 000-0 -1011 00110 10011 00110 10013Gekkonoidea 01A01 2?-10 00001 0101A 00110 0000A 0010C 100AEEublepharinae 01101 20-10 00001 0A011 00110 000A0 00100 100A1Gymnophthalmidae 01A0A C00A0 00001 002A0 101A0 A011A 00001 110A1Heloderma 01011 01100 20110 00011 00A00 21011 10102 10111Other Iguanians A100A B0000 00000 AA0A0 00AA0 00A1A 00A0A 1AA0FLacertidae 01011 00010 00000 00100 101A0 00110 0000A 110A0Lanthanotus 01010 01100 22111 10011 00000 21011 11102 10113Scincidae 010A1 B001A 00000 -A011 A01A0 A0011 000A2 10A11Teiidae 01000 A0010 00000 00200 101A0 A011A 0A001 110A1Varanus 0100A 01100 10111 00011 00110 21001 11102 10112Estesia ????0 ???00 2?10 000?1 00110 210?? ????? ?????Xantusiidae 011A1 20-11 000-0 -1100 00110 00010 00101 AA001Xenosaurus 01001 01101 00000 00011 00110 10011 00102 10111Shinisaurus 01010 01101 00000 00010 00100 10011 00102 10000Amphisbaenia 010?0 B?-0A 00000 AA011 0011A D001A 001-2 10A13Alethinophidia 01000 00010 01110 1A011 010A0 30011 0A0-2 1011-Scolecophidia 01000 00010 01110 1A011 01110 30011 000-2 1011-Dinilysia 01?00 00010 0111? ?00?? 01000 ?0??? 000?? ????-Mosasauroidea 01010 01010 01111 10011 01100 1001? 00002 10110Coniasaurus ????? 01000 01111 00?1? 0??00 ?00?? 000?? ?????Sphenodon 00000 0?000 000-0 000-- 00011 -0000 --000 00001

Page 33: Caldwell, 1999a

SQUAMATE PHYLOGENY 147

APPENDIX 2 contd

85 90 95

Agamidae 00A00 A00AA 00000Anguidae 0A001 1A010 00112Chamaeleonidae 1111 -1--A 10000Cordylidae A000A 10010 00112Dibamidae ----- ----- -00--Gekkonoidea AAA01 AAA1A 0AA11Eublepharinae A1001 10010 0AA11Gymnophthalmidae A0001 10A11 10012Heloderma 00101 101-0 00111Other Iguanians AAA00 A000A 00A00Lacertidae A0101 1001A 00112Lanthanotus 00101 10010 00111Scincidae AA001 1A01A 00112Teiidae 0A001 10011 10011Varanus 0010A 100AA 00A10Estesia ????? ????? ??1??Xantusiidae 00001 1001A 00012Xenosaurus 00001 10000 00111Shinisaurus 00001 10010 00111Amphisbaenia 001A- 11--0 1001-Alethinophidia ----- ----- -00--Scolecophidia ----- ----- -00--Dinilysia ????? ????? ?????Mosasauroidea 0A1A0 ?0000 000?0Coniasaurus ?0??? ????? ??0??Sphenodon 00000 00000 00000

APPENDIX 3

Osteological characters of squamate taxa were obtained by examination of specimens and fromavailable literature. Museum abbreviations: (AMNH) American Museum of Natural History, NewYork, New York; (BMB) Booth Museum of Natural History, Brighton, Sussex, England; (BMNH R),The Natural History Museum (British Museum), London, England; (FMNH) Field Museum of NaturalHistory, Chicago, Illinois; (USNM) National Museum of Natural History, Smithsonian Institution,Washington, D.C.

S: typhlopids (Typhlops AMNH 3001, 11633) (Rieppel, 1978; 1979; 1980b; Estes etal., 1988); Anomochilus (Cundall & Rossman, 1993; Cundall, Wallach & Rossman, 1993); Dinilysia (Esteset al., 1970). A: acrochordids: (Acrochordus USNM 347549); anilioids (Anilius scytaleFMNH 11175); (Cylindrophis rufus FMNH 13100; Cylindrophis USNM 297456), colubrids (Elaphe obsoletaRTMP 90.7.190); hydrophiids, (Aypisurus laevis AMNH 86176, Laticauda colubrina AMNH 81880), booids(Python sp., USNM 220308; Boa sp. AMNH 57476; USNM 220299; 220300). A: Elgariamulticarinatus (USNM 11298; 292548); Ophisaurus apodus (FMNH 22088); Shinisaurus crocodilurus (AMNH44928); Xenosaurus grandis (USNM 111531). V: Varanus komodoensis (NMNH 228163); Varanussalvator (FMNH 31358); Varanus rudicollis FMNH 145710; Heloderma suspectum (FMNH 218077; NMNH228171); Lanthanotus borneensis (FMNH 134711). T: Dracaena guianensis (FMNH 207657;22365);Tupinambis tequixin (FMNH 98759). G: Cnemidophorus sexlineatus (FMNH 98505–98507).C: Cordylus sp. (RTMP T–20); Gerrhosaurus flavigularis (USNM KdQ 134). S: Eumecesobsoletus (USNM 313463); Corucia zebrata (USNM 120164). L: Lacerta lepida (USNM 279861).X: Xantusia riversiana (USNM 313463). O : Anolis carolensis (RTMP T160);Dipsosaurus dorsalis (RTMP, T343); Phrynosoma douglassi (RTMP T17); Ctenosaura similis (FMNH 211849);Conolophus subcristatus (FMNH 22406); Amblyrhynchus cristatus (FMNH 15072). A: Uromastyx aegypticus(FMNH 63961); Physignathus draconoides (RTMP 90.7.347). C: Chamaeleo oweni (FMNH25408); Chamaeleo jacksoni (FMNH 206753). G: Gekko gekko (FMNH 14448); Hemitheconyx sp.(RTMP T352); Pygopus nigriceps (USNM 292076). D: Dibamus novaeguineae (USNM 305914).A: Amphisbaenia caeca (USNM 129269); Rhineura floridana (USNM 220289).