HERPETOLOGICAL JOURNAL, Vol. 9, pp. 147-151 (1999) SQUAMATE RELATIONSHIPS BASED ON C-MOS NUCLEAR DNA SEQUENCES 1 D. J. HARRIS1 , E. A. S!NCLAIR1 , N. L. MER CADER2, J. C. MAR SHALL AND K. A. CRANDALL3 1 Department of Zoology, Brigham Young University, 574 Widtsoe Building, Provo, Utah, 84602. USA 2 Departament de Biologia Animal, Universitat de Barcelona, Avinguda Diagonal 645. Barcelona 08028, Sp ain 3 Department of Zoologyand Monte L. Bean Museum. Brigham Young University. 574 Widtsoe Building. Provo. Utah. 84602. USA Relationships among squamate families have classically been difficult to establish, with morphological characters being interpreted to give many diffe rent topologies. Here we combine new C-mos nuclear DNA sequence data with those already published to assess relationships of 19 families within the Squamata. Monophyly of all the families examined is upheld. Many relationships between families are estimated, although it appears there may have been rapid cladogenesis associated with the origins of the Squamata. Key words : Phylogeny, squamates, C-mos INTRODUCTION METHODS Squamate relationships have remained contentious The additional species examined were: F. since Camp's (1923) "Classification of the Lizards". Cordylidae: Cordy/us cordylus; F. Gekkonidae: Despite extensive analyses based on morphological Bunopus tuberculatus, Stenodactylus doriae; F. characters many relationships remain unknown. Most Jguanidae: Dipsosaurus dorsalis, Iguana iguana; F. widely accepted are the relationships suggested by Lacertidae: Acanthodactylus scute!latus, Lacer/a Estes et al. (! 988), although the analysis has been criti­ kulzeri, Podarcis hispanica; F. Trogonophidae; cized (Kluge, 1989), and alternative suggestions for Diplometophon zarudnyi; F. Xantusidae: Lepidophyma relationships have been made using different morpho­ gaigae, Xantusia vigilis. These were selected to cover logical characters (Presch, 1988). Surprisingly, the five families not included by Saint et al. ( 1998), and to advent of DNA sequence data has had little impact on extend the number of the family Gekkonidae examined our understanding of squamate relationships. Although from one to three. many studies have examined inter-familial relation­ Total genomic DNA was extracted from small (I or 3 ships (e.g. Hedges et al., 1991; Harris et al., 1998), 2 mm ) pieces of tail tissue. The material was fi nely these have been limited due to their use of mitochon­ diced and digested with proteinase K (Kocher et al., drial DNA sequences, which are typically saturated 1989). Purification was by phenol/chloroform extrac­ before the divergence times necessary to estimate rela­ tions (Sambrook et al., 1989), followed by centrifugal tionships across squamates. dialysis through a Centricon 30000 MW membrane Recently Saint et al. (! 998) used a fragment of the (Amicon). Polymerase Chain Reaction (PCR) primers nuclear gene C-mos to investigate relationships of Aus­ used in both the amplification and the sequencing were tralian reptiles relative to their overseas relatives. They G73 and G74 (Saint et al., 1998). PCR conditions were showed that C-mos was likely to be a single copy gene the same as those used by Saint et al. ( 1998). Successful in squamates, had no introns, and that a fragment of PCR products were purified using a Qiaex II kit about 400 base pairs could be amplified across many (Qiagen), and sequenced from both strands on an Ap­ squamate families. Graybeal (1994) had already shown plied Biosystems DNA Sequencing System. that C-mos might be phylogenetically informative among taxa that had diverged up to 400 mya. To esti­ SEQUENCE ANALYSIS mate relationships across squamates, we have extended Genbank accession numbers are AF1 48702 to the number of families included, and compared the esti­ AF148712. The sequences were aligned by eye to the mates of phylogeny produced from Maximum previously published sequences (Genbank AF039462 Parsimony (MP) and Maximum Likelihood (ML) with to AF039482) of Saint et al. (1998). The aligned se­ those previously derived from morphological charac­ quences were 375bp long. The codon reading frame ters. was infered by comparison with the published se­ quences. Of the new sequences, all the lacertids and the two geckos had a deletion of seven codons, and the Correspondence: D. J. Harris, Campus Agrario de Vairao,R. Monte-Crasto, 4480 Vila do Conde, Portugal. E-mail: Diplometophon had an eight codon deletion. These [email protected] were in the same region (bp 727-768 of human C-mos 148 D. J. HARRIS ET AL. ..----Sphenodon punctalus ] Lacertinae Lacertidae Acan thodactylus scut•llatus J Eremiainae Amphisbaenidae Diplome tophon mrudnyi Trogonophidae ..--_,1 �00_ _, Xantusia vigilis 100 J Xantusidae Lepidophyma gaigae ] T1 lrquascm caidts Egerma lucluosa Scinadae L1pm1JJ noctua Sphenonwrphus leptofascUJtus Ablepharus pannon1cus 78 Cordy/us cordylus 73 Cordylidae Iguana ig uana Dipsosaurus dorsalis ] Iguanidae Sceloporus grammicus Phrynosomatidae Ctenophorus decresii J Agamidae Physignathus coeincinus '------- Chamaeleosp. Chamaeleonidae Elgaria multicarinata Anguidae 55 '------ Varanus rosenbergi Varanidae Python reticu latus Pythonidae Candoia carinata Boidae J Boo idea �-- Rmnphotyphlops auslralis Typhlopidae 100 J Gekkonidae 84 Eublepharidae J Pygopodidae Cnemidophorus tigris Te iidae Chelodina rugosa Elseya denlala �---- Crocodylus poros us FIG. 1. Single most parsimonious tree derived from an analysis of C-mos nucleotide sequence. Numbers above branches indicate bootstrap support (1000 replicates). Numbers below branches indicate boostrap support from an MP analysis based on the amino acid sequence, with all changes weighted equally. See text for details. The tree was rooted using the Crocodylus, Chelodina and Elseya sequences. sequence), and overlapped deletions also found in the neity exists within the data set; and (5) there is no sig­ skink Lipinia noctua and the teiid Cnemidophorus nificant proportion of invariable sites. Once a model of tigris. They were therefore treated as missing data in evolution was chosen, it was used to estimate a tree us­ the analyses. ing maximum likelihood (Felsenstein, 1981) , using The data were analysed using PAUP* (Swofford, random sequence addition and a heuristic search with 1998). When estimating phylogenetic relationships 10 replicates. Also an MP analysis was performed.Two among sequences, one assumes a model of evolution hundred and nine of the 375 characters were parsi­ regardless of the optimality criteria employed. Deter­ mony-informative. A 10 replicate heuristic search was mining which model to use given the data is a statistical carried out, and support fornodes was estimated using problem (Goldman, 1993). We used the approach out­ the bootstrap (Felsenstein, 1985) technique, with 1 OOO lined by Huelsenbeck & Crandall (1997) to test replicates. A furtherMP analysis was carried outon the alternativemodels of evolution, employing PAUP* and translated amino acid sequences. All changes were Modeltest (Posada & Crandall, 1998). A starting tree weighted equally. was obtained using neighbour-joining. With this tree, likelihood scores were calculated forvarious models of RESULTS evolution and then compared statistically using a chi­ Using MP, 209 of the 375 characters were parsi­ square test with degrees of freedom equal to the mony-informative. A 10 replicate heuristic search difference in free parameters between the models being found one MP tree with 892 steps. (Cl= 0.46, HI= tested. The null hypotheses tested in this way included: 0.54). Support for nodes was estimated using the (1) nucleotide frequenciesare equal; (2) transition rates bootstrap (Felsenstein, 1985) technique, with 1000 rep­ are equal to transversion rates; (3) transition rates are licates (Fig. 1). In the translated amino acid sequence, equal and transversion rates are equal; (4) rate homoge- 60 characters were informative. A ten replicate heuris- PHYLOGENY OF SQUAMATES 149 TABLE I. Tests of hypotheses relating to the model of evolution appropriate forphylogeny reconstruction (Huelsenbeck and Crandall, 1997). ?-values were obtained using the computer program Modeltest (Posada & Crandall, 1998). Due to the perfo�ance of multiple tests, the significance level of rejection of the null hypothesis should be adjusted via the Bonferroni correction to a= 0.01. Null Hypothesis Models Compared -lnL0 -lnL1 df p Equal base frequencies H0: JC69, H1 : F8 1 5010. l 5005.5 3 0.012 Equal ti/tv rates H0: JC69, H, : K80 5010.1 4763.8 1 <0.001 Equal ti and equal tv rates H0: K80, H1 : GTR 4763.8 4761.6 3 0.340 Equal rates among sites H0: K80, H1 : K80+G 4763.8 4582.3 I <0.001 Proportion of invariable sites H0: K80+G, H1 : K80+G+invar 4582.3 4568.9 <0.001 Molecular clock H0: no rate heterogeneity, 4635.8 4568.9 34 <0.00 1 H1 : rate heterogeneity Sphenodon punctatus Podarcishispanica Acanthodactylus sc�tellatus Lacertidae Lacertkulzeri J Bipesbiporus Amphisbaenidae Diplomdophon zarudnyi Trogonophidae J Amphisb aenia �---1 Xantusiauigilis Ltpidophymagaigae J Xantusidae Tiliquascincuides Egem ialu ct11 osa Lipinianoctua Scincidae Sphenomorphus leptofasciatus J Ablepharus pannonicus Cordy/us cordylus Cordylidae lgu1111a iguana ] Dipsosaurus dorsalis Iguanidae Sceloporusgrammic us Phrynosomatidae Ctenophorus decresii ] Physignathus cocincinus Agamidae Chamaeleo sp. Chamaeleonidae Python reticu/atus Pythonidae Candoiacarinata Boidae J Booide a Ramphotyphus australis Ty phlopidae Elgaria luctuosa Anguidae . J Angwmorpha Varanus rosenbergi Vararudae