Vicariant Patterns of Fragmentation Among Gekkonid Lizards of The

Home , Teratoscincus, Teratoscincus scincus

Molecular Phylogenetics and Evolution Vol. 12, No. 3, August, pp. 320–332, 1999 Article ID mpev.1999.0641, available online at http://www.idealibrary.com on

Vicariant Patterns of Fragmentation among Gekkonid Lizards of the Genus Teratoscincus Produced by the Indian Collision: A Molecular Phylogenetic Perspective and an Area Cladogram for Central Asia J. Robert Macey,*,1 Yuezhao Wang,† Natalia B. Ananjeva,‡ Allan Larson,* and Theodore J. Papenfuss§ *Department of Biology, Box 1137, Washington University, St. Louis, Missouri 63130; †Chengdu Institute of Biology, Chengdu, Sichuan, China; ‡Zoological Institute, Russian Academy of Sciences, St. Petersburg, Russia; and §Museum of Vertebrate Zoology, University of California, Berkeley, California 94720

Received October 21, 1998; revised December 29, 1998

Key Words: Reptilia; Sauria; Gekkota; Gekkonidae; bio- A well-supported phylogenetic hypothesis is pre- geography; Indian collision; mitochondrial DNA; replica- sented for gekkonid lizards of the genus Teratoscincus. tion; phylogenetics; Asia; China; Kazakhstan; Pakistan. Phylogenetic relationships of four of the five species are investigated using 1733 aligned bases of mitochondrial DNA sequence from the genes encoding ND1 (subunit one The gekkonid lizard genus Teratoscincus is endemic of NADH dehydrogenase), tRNAIle, tRNAGln, tRNAMet, ND2, to the desert regions of central and southwest Asia. tRNATrp, tRNAAla, tRNAAsn, tRNACys, tRNATyr, and COI Five species of Teratoscincus are currently recognized (subunit I of cytochrome c oxidase). A single most parsi- (Macey et al., 1997a). Two species, T. bedriagai and T. monious tree depicts T. przewalskii and T. roborowskii as microlepis, are restricted to desert regions south of the a monophyletic group, with T. scincus as their sister taxon Hindu Kush in Iran, Afghanistan, and Pakistan (Fig. and T. microlepis as the sister taxon to the clade contain- 1). Teratoscincus scincus occurs both in the region of ing the first three species. The aligned sequences contain southwest Asia where T. bedriagai and T. microlepis 341 phylogenetically informative characters. Each node is supported by a bootstrap value of 100% and the shortest are found (Anderson, 1993), as well as to the north in suboptimal tree requires 29 additional steps. Allozymic the Central Asian republics of the former USSR (Kaza- variation is presented for proteins encoded by 19 loci but khstan, Kyrgyzistan, Tadjikistan, Turkmenistan, and these data are largely uninformative phylogenetically. Uzbekistan; Szczerbak and Golubev, 1996). The two Teratoscincus species occur on tectonic plates of Gond- remaining species occur in China and Mongolia. Tera- wanan origin that were compressed by the impinging toscincus przewalskii is found in desert regions of the Indian Subcontinent, resulting in massive montane uplift- Taklimakan, Hami Depression, and low-elevation Gobi. ing along plate boundaries. Taxa occurring in China Teratoscincus roborowskii is restricted to the Turpan De- (Tarim Block) form a monophyletic group showing vicari- pression in China, which is the second lowest depression in ant separation from taxa in former Soviet Central Asia the world. These two species do not overlap in distribution and northern Afghanistan (Farah Block); alternative bio- and are geographically separated from the remaining geographic hypotheses are statistically rejected. This vi- three species by the Tien Shan and Pamir (Macey et al., cariant event involved the rise of the Tien Shan-Pamir 1997a). Prior to the taxonomic revision of Macey et al. and is well dated to 10 million years before present. Using (1997a), T. roborowskii was considered conspecific with T. this date for separation of taxa occurring on opposite scincus, suggesting that populations occurring east of the sides of the Tien Shan-Pamir, an evolutionary rate of Tien Shan-Pamir do not form a monophyletic group (see, 0.57% divergence per lineage per million years is calcu- Szczerbak and Golubev, 1996; Zhao and Adler, 1993). lated. This rate is similar to estimates derived from fish, Phylogenetic patterns of fragmentation are investigated bufonid frogs, and agamid lizards for the same region of in taxa occurring on different tectonic plates of Gondwanan D the mitochondrial genome ( 0.65% divergence per lin- origin now separated by mountain uplifting that resulted eage per million years). Evolutionary divergence of the from the Indian collision. We predict that phylogenetic mitochondrial genome has a surprisingly stable rate relationships in Teratoscincus reflect the historical forma- across vertebrates. tion of faunal barriers by the uplift of the Karakorum, ௠ 1999 Academic Press Pamir, Tien Shan, and Hindu Kush mountains (Fig. 1). Data are reported for nuclear-encoded allozymic 1 To whom correspondence should be addressed. Fax: (314) 935-4432. variation representing 19 loci and mitochondrial DNA

320 1055-7903/99 $30.00 Copyright ௠ 1999 by Academic Press All rights of reproduction in any form reserved. VICARIANT PATTERNS IN ASIAN Teratoscincus LIZARDS 321

FIG. 1. Map of central and southwest Asia illustrating the approximate distribution of Teratoscincus species relative to major mountain belts and basins. Montane regions that separate Teratoscincus species are depicted with stippling. Note that the mountain chain consisting of the Hindu Kush, Karakoram, Himalaya, Pamir, and Tien Shan all connect, causing a dramatic barrier to the distribution of Teratoscincus species. In addition, the Kopet-Dagh is connected to the Hindu Kush by the Badkyz Plateau (not illustrated) causing a barrier between the Central Asian republics of the former USSR (Kazakhstan, Kyrgyzistan, Tadjikistan, Turkmenistan, and Uzbekistan) and Southwest Asia. The Pamir-Tien Shan is particularly noteworthy because it separates T. przewalskii and T. roborowskii from T. scincus. sequences from the genes encoding ND1 (subunit one of Asia south of the Hindu Kush. Gekko gecko serves as an NADH dehydrogenase), tRNAIle, tRNAGln, tRNAMet, outgroup for both data sets. In the allozymic data, an ND2, tRNATrp, tRNAAla, tRNAAsn, tRNACys, tRNATyr, additional gekkonine outgroup, Cyrtodactylus tibeta- and COI (subunit I of cytochrome c oxidase). Both data nus, is used. An eublepharine species, Eublepharus sets include taxa that occur on each side of the Tien turkmenicus, serves as an additional outgroup for the Shan and Pamir (T. scincus to the west, and T. przewal- mitochondrial DNA sequence data. skii and T. roborowskii to the east); the phylogenetic relationships of these taxa are a major focus of this study. Unfortunately, T. bedriagai, known only from MATERIALS AND METHODS extreme eastern Iran and adjacent Afghanistan, is not available for either data set. The sequence data include Specimen Information representatives of the other four Teratoscincus species Museum numbers and localities for voucher speci- (T. microlepis, T. przewalskii, T. roborowskii, and mens from which DNA was extracted and GenBank T. scincus). Allozymic data were not collected for accession numbers are presented below. Acronyms are T. microlepis, which occurs exclusively in southwest CAS for California Academy of Sciences, San Francisco 322 MACEY ET AL. and MVZ for Museum of Vertebrate Zoology, University TABLE 1 of California at Berkeley. The acronym followed by a dash RM represents a field number of the first author The 19 Allozymic Systems Scored and the Five Electrophoretic Conditions within Which for an uncatalogued specimen being deposited at the They Were Resolved Museum of Vertebrate Zoology. The previously reported sequence for T. przewalskii (Macey et al., 1997b) has Electrophoretic been extended by 19 bases to include 6 additional Enzyme Abbreviation E. C. No. conditionsa amino acid positions and a third codon position of the 1. Aconitase hydratase ACOH-1 4.2.1.3 1 ND1 gene; the GenBank accession has been updated 2. Aconitase hydratase ACOH-2 4.2.1.3 1 accordingly. 3. Aspartate amino- Eublepharus turkmenicus: sequence CAS 184771, transferase AAT 2.6.1.1 3 AF114248, Elev. 300 m, 39° 06Ј N 55° 08Ј E, vicinity of 4. Carboxylic ester b Temen Spring, 2.5 km west of Danata (39° 07Ј N 55° 08Ј hydrolase EST-D 3.1.1.– 3 5. Fructose-bisphos- E) on paved Rd. from Danata to the Ashgabad (Ash- phate aldolase FBA 4.1.2.13 4 kabad) to Krasnovodsk Rd., then 5.4 km south on Dirt 6. Glucose-6-phosphate Rd., Krasnovodsk Region, Turkmenistan. Cyrtodacty- isomerase GPI 5.3.1.9 2 lus tibetanus: allozymes CAS 171751–171760, Elev. 7. Glycerol-3-phosphate 3700 m, at base of mountains, approx 3 km WNW dehydrogenase G3PDH 1.1.1.8 1 8. L-Iditol dehydroge- (airline) of the Potala Palace, Lhasa (29° 39Ј N 91° 06Ј nase IDDH 1.1.1.14 4 E), Lhasa Municipality, Xizang (Tibet) Autonomous 9. Isocitrate dehydroge- Region, China. Gekko gecko: allozymes MVZ 215269, nase IDH-1 1.1.1.42 1 215312, 215314, 215356, 215358–215360, sequence 10. Isocitrate dehydrogenase IDH-2 1.1.1.42 1 MVZ 215314, AF114249, Patong Beach, Kathu District, 11. L-Lactate dehydroge- Phuket Island, Phuket Province, Thailand. Teratoscin- nase LDH-1 1.1.1.27 3 cus microlepis: sequence MVZ-RM10464, AF114250, 12. L-Lactate dehydroge- captive bred from wild-caught specimens probably origi- nase LDH-2 1.1.1.27 3 nating from Pakistan. Teratoscincus scincus: allozymes 13. Malate dehydrogenase MDH-1 1.1.1.37 1 MVZ 216056–216065, sequence MVZ 216056, 14. Malate dehydroge- AF114251, near Alma-Ata (43° 15Ј N 76° 57Ј E), Alma- nase MDH-2 1.1.1.37 1 Ata Region, Kazakhstan. Teratoscincus przewalskii: 15. Peptidase B PEP-B 3.4.11.4 4 allozymes CAS 171010–171019, sequence CAS 171010, 16. Peptidase D PEP-D 3.4.13.9 5 U71326 (Macey et al., 1997b,c), Elev. 1000 m, 19.5 km 17. Phosphogluconate dehydrogenase PGDH 1.1.1.44 1 east of the Uygur girl-Hami Mellon monument in the 18. Purine-nucleoside center of Hami (42° 48Ј N 93° 27Ј E), also at km 177.9 phosphorylase PNP 2.4.2.1 5 from the Gansu Province line on the Lanzhou-Urumqi 19. Superoxide dismu- Rd., then 9.0 km NE on dirt road to Mirowlu, Hami tase SOD 1.15.1.1 2 (Kumul) Prefecture, Xinjiang Uygur Autonomous Re- a Electrophoretic conditions: (1) Amine–citrate (morpholine) pH gion, China. Teratoscincus roborowskii: allozymes CAS 6.0,250vfor6hor300vfor5h(ClaytonandTretiak, 1972); (2) 171203–171212, sequence CAS 171203, AF114252, Elev. Histidine–citrate pH 7.8, 150 v for 8 h (Harris and Hopkinson, 1976); 470 m, 1.7 km south of the Lanzhou-Urumqi Rd. on (3) Lithium–borate/Tris–citrate pH 8.2, 250 v for6hor300vfor5h; Shanshan main street, then 13.3 km east of Shanshan (4) Phosphate–citrate pH 7.0, 120 v for 7 h; (5) Tris–HCL pH 8.5, 250 v for 4 1/2 h (all, Selander et al., 1971). main street on Ma Chang Dadu Rd., Ma Chang Dadu b EST-D, dimeric esterase. District, Shanshan (42° 52Ј N 90° 10Ј E), Turpan Prefecture, Xinjiang Uygur Autonomous Region, China. ylic ester hydrolase (Dimeric Esterase) was resolved Laboratory Protocols using 4-methylumbelliferyl acetate as the substrate; Tissues were taken in the field and immediately Peptidase B (PEP-B) was resolved using L-leucyl-L- frozen in liquid nitrogen and later transferred to an alanine as the substrate, and Peptidase D (PEP-D) ultracold freezer and maintained at Ϫ80°C. For analy- with the use of L-phenylalanyl-L-proline as the sub- sis of allozymic variation, liver tissue was homogenized strate. The isozymes, and loci if more than one, were separately from other tissue. Horizontal starch-gel labeled according to their migration from anode to electrophoresis was employed to differentiate variation cathode. in 19 presumptive loci. The 19 loci and five buffer Genomic DNA was extracted from liver using the conditions utilized to resolve them are displayed in Qiagen QIAamp tissue kit (Qiagen Inc., Hilden, Ger- Table 1. Allozymes were stained using standard meth- many). Amplification of genomic DNA was conducted ods (Harris and Hopkinson, 1976; Murphy et al., 1990; using a denaturation at 94°C for 35 s, annealing at Richardson et al., 1986; Selander et al., 1971). Carbox- 50°C for 35 s, and extension at 70°C for 150 s with 4 s VICARIANT PATTERNS IN ASIAN Teratoscincus LIZARDS 323 added to the extension per cycle, for 30 cycles. Negative one of NADH dehydrogenase), tRNAIle, tRNAGln, controls were run for all amplifications. Amplified tRNAMet, ND2, tRNATrp, tRNAAla, tRNAAsn, tRNACys, products were purified on 2.5% Nusieve GTG agarose tRNATyr, and COI (subunit I of cytochrome c oxidase). gels (FMC Bioproducts, Rockland, ME) and reamplified Protein-coding sequences were translated to amino under similar conditions. Reamplified double-stranded acids using MacClade (Maddison and Maddison, 1992) products were purified on 2.5% acrylamide gels (Mani- for alignment. Transfer-RNA secondary structure was atis et al., 1982). Template DNA was eluted from determined manually using the criteria of Kumazawa acrylamide passively over 3 days with Maniatis elution and Nishida (1993) to ensure proper alignment (Macey buffer (Maniatis et al., 1982). Cycle-sequencing reac- and Verma, 1997). tions were run using the Promega fmol DNA-sequencing system (Fisher Scientific, Pittsburgh, PA) with a Phylogenetic Analysis denaturation at 95°C for 35 s, annealing at 45–60°C for Phylogenetic trees were estimated using PAUP (Swof- 35 s, and extension at 70°C for 1 min for 30 cycles. ford, 1993) with exhaustive searches. Bootstrap resam- Sequencing reactions were run on Long Ranger sequenc- pling was applied to assess support for individual nodes ing gels (FMC Bioproducts, Rockland, ME) for 5–12 h at with 1000 bootstrap replicates using branch-and-bound 38–40°C. searches. Decay indices (ϭ‘‘branch support’’ of Bremer, Amplifications were done using either primers L3881, 1994) were calculated for all internal branches of the L3887, L4160, L4178a, or L4178b in combination with tree. Branch-and-bound searches retained suboptimal primer H5934. In addition, some taxa were amplified trees. The decay index for a particular node was tabulated with primers L3002 and H4419. Both strands were as the difference in length between the overall shortest tree sequenced using the primers in Table 2. Primer num- and the shortest tree lacking that node. bers refer to the 3Ј end on the human mitochondrial Wilcoxon signed-ranks tests (Felsenstein, 1985; genome (Anderson et al., 1981), where L and H corre- Templeton, 1983) were applied to examine statistical spond to light and heavy strands, respectively. significance of the shortest tree relative to alternative Sequence Alignment hypotheses. This test asks whether the most parsimonious tree is significantly shorter than an alternative or Reported sequences are presented in Fig. 2 and whether their differences in length can be attributed to correspond to positions 4161 to 5936 on the human chance alone (Larson, 1998). Wilcoxon signed-ranks mitochondrial genome (Anderson et al., 1981). This tests were conducted as two-tailed tests. Felsenstein sequence contains the genes encoding ND1 (subunit (1985) showed that the two-tailed test is conservative. The test statistic Ts was compared with critical values for the Wilcoxon rank sum in table B.11 of Zar (1984). TABLE 2 Alternative phylogenetic hypotheses were tested us- Primers Used in This Studya ing the most parsimonious phylogenetic topologies com- patible with them. Each branch in the overall shortest Positionb Gene Primer sequencec tree was tested statistically by comparing the overall L3002 16S rRNA 5Ј-TACGACCTCGATGTTGGATCAGG-3Ј shortest tree to the shortest tree(s) that did not contain L3887 ND1 5Ј-GACCTAACAGAAGGAGAATCAGA-3Ј the node being examined. Alternative tree(s) were L4160 ND1 5Ј-CGATTCCGATATGACCARCT-3Ј found with branch-and-bound searches that retained L4178a ND1 5Ј-CARCTWATACACYTACTATGAAA-3Ј suboptimal trees using PAUP (Swofford, 1993). Statisti- Ј Ј L4178b ND1 5 -CAACTAATACACCTACTATGAAA-3 cal tests were conducted using the ‘‘compare trees’’ L4221 tRNAIle 5Ј-AAGGATTACTTTGATAGAGT-3Ј H4419 tRNAMet 5Ј-GGTATGAGCCCAATTGCTT-3Ј option of MacClade (Maddison and Maddison, 1992). L4437 tRNAMet 5Ј-AAGCTTTCGGGCCCATACC-3Ј H4645 ND2 5Ј-ACAGAAGCCGCAACAAAATA-3Ј Cladistic Analyses of Allozymic Data L5002 ND2 5Ј-AACCAAACCCAACTACGAAAAAT-3Ј Allozymic data were coded in two ways for cladistic H5540 tRNATrp 5Ј-TTTAGGGCTTTGAAGGC-3Ј L5556 tRNATrp 5Ј-GCCTTCAAAGCCCTAAA-3Ј phylogenetic analysis. While presence–absence coding L5638 tRNAAla 5Ј-CTGAATGCAACTCAGACACTTT-3Ј of alleles has received considerable criticism for a lack H5692 tRNAAsn 5Ј-TTGGGTGTTTAGCTGTTAA-3Ј of independence of alleles and the possibility of no allele H5934 COI 5Ј-AGRGTGCCAATGTCTTTGTGRTT-3Ј being reconstructed for an ancestral node (Swofford Ј Ј H5937 COI 5 -GTGCCAATGTCTTTGTG-3 and Olsen, 1990), it remains the method that provides a All primers are from Macey et al. (1997b) except L4160 which is the greatest amount of resolution. Alternatively, combi- from Kumazawa and Nishida (1993). nations of alleles for a particular locus may be coded as b Primers are designated by their 3Ј ends which correspond to the discrete character states (Buth, 1984). If step matrices position in the human mitochondrial genome (Anderson et al., 1981) are used to connect character states, a greater amount by convention. H and L designate heavy-strand and light-strand primers, respectively. of information can be retained (Mabee and Humphries, c Positions with mixed bases are labeled with the standard one- 1993). In our analysis, step matrices were constructed letter code: R ϭ GorA,Wϭ AorT,Yϭ CorT. on the basis of gains and losses of alleles. For example, a 324 MACEY ET AL. VICARIANT PATTERNS IN ASIAN Teratoscincus LIZARDS 325

Cys FIG. 3. Overlap between the tRNA gene and a presumptive nonfunctional OL. Pseudo-OL is depicted as template heavy-strand sequence and the tRNACys gene, which is encoded on the light strand, is depicted as the presumptive transcribed tRNA. Boldface bases are used in the function of light-strand replication and also tRNACys. Three bases of the tRNACys gene, two of which encode the AA-stem, also would be used in Cys the OL stem. Another three bases of the tRNA gene encoding the AA-stem also would be used in initiation of light-strand replication (required for in vitro replication in humans; Hixson et al., 1986). Hence, ﬁve out of seven bases encoding one side of the tRNACys AA-stem also would be used in initiation of light-strand replication. The heavy-strand template sequence identiﬁed as the point of light-strand elongation in mouse 3Ј-GCC-5Ј (Brennicke and Clayton, 1981) is not present. The heavy-strand sequence conserved in lizards as 3Ј-GBCCB-5Ј (Macey et al., 1997b) related to the 3Ј-GGCCG-5Ј sequence found to be required for in vitro replication in humans (Hixson et al., 1986) also is missing (underlined with arrows). This missing sequence has been hypothesized to be the result of an insertion to the D-loop/stem region of the Cys tRNA gene during replication. This insertion resulted in a realigned AA-stem on the opposite side, causing compensatory mutation to the OL functional regions and disrupting the initiation of light-strand replication (Macey et al., 1997c,d).

ﬁxed difference between two alleles was counted as two coding genes have no premature stop codons, suggest- steps, one allele lost and another gained. In the case of a ing that these sequences represent functional copies two-allele polymorphism in one population with one allele that encode a protein. Transfer-RNA genes appear to shared with another monomorphic population, a single encode tRNAs with stable secondary structures, indicat- gain or loss was counted as a single step. Additional ing functional genes. Strand bias further supports our polymorphisms were counted in the same manner. conclusion that the six DNA sequences analyzed here are from the mitochondrial genome. The strong bias RESULTS against guanine on the light strand (G ϭ 11.0–13.8%, A ϭ 29.8–34.3%, T ϭ 22.8–27.3%, and C ϭ 27.4–33.0%) Sequences ranging in size from 1763 to 1777 bases of is characteristic of the mitochondrial genome but not mitochondrial DNA for six taxa of gekkonid lizards are the nuclear genome. See Macey et al. (1997b,d) for presented as 1815 aligned positions in Fig. 2. similar strand bias across most squamate-reptile fami- Authentic Mitochondrial DNA lies for the same region of the mitochondrial genome. Cys Several observations suggest that the DNA se- Origin for Light-Strand Replication and tRNA quences analyzed here are from the mitochondrial Two structural features of the mitochondrial genome genome and not nuclear-integrated copies of mitochon- (origin for light-strand replication and secondary struc- drial genes (see Zhang and Hewitt, 1996). Protein- ture of tRNACys) show derived states in Gekko gecko

FIG. 2. Length-variable regions among the 1815 aligned mitochondrial DNA sequences as used in the phylogenetic analysis. Six regions totaling 82 positions are excluded from the analysis and are denoted by an ‘‘x’’ above the sequence. Sequences in the tRNACys gene of Gekko gecko treated as missing are doubly underlined. The peudo-start codon of ACG for COI in Gekko gecko also is doubly underlined. The codon directly following these three nucleotides is ATG, the typical start codon for most mitochondrial proteins. Positions 401–1300 from the ND2 gene are not shown because this region has no length variation. Sequences are presented as light-strand sequence and tRNA secondary structure is designated above the sequence. Stems are indicated by arrows in the direction encoded: AA, amino acid-acceptor stem; D, dihydrouridine stem; AC, anticodon stem; T, T⌿C stem. The tRNA anticodons are designated COD. Asterisks indicate the unpaired 3Ј tRNA position 73. Periods represent bases located outside stem regions; 1 depicts the ﬁrst codon position of protein-coding sequences. STP represents stop codons. 326 MACEY ET AL.

(Fig. 3). Most vertebrates have a mitochondrial gene Genic Variation Ile Gln Met Trp order of ND1, tRNA , tRNA , tRNA , ND2, tRNA , Different levels of variation are observed among the 3 Ala Asn tRNA , tRNA ,OL (origin for light-strand replica- protein-coding genes, 8 tRNA-coding genes, and 2 tion), tRNACys, tRNATyr, and COI. In Gekko gecko, a noncoding regions (Table 4). All 11 genes sequenced stem-and-loop structure is present in the typical verte- contain phylogenetically informative characters. The 8 Asn Cys brate position for OL between the tRNA and tRNA tRNA genes each have phylogenetically informative genes. This sequence does not have the functional sites in stem and nonstem regions except nonstem Trp characteristics of OL identified in studies of mamma- regions of the tRNA gene. Each of the three protein- lian mitochondrial replication (Brennicke and Clayton, coding genes contains phylogenetic information in first, 1981; Hixson et al., 1986). We therefore interpret these second, and third codon positions. Most of the variation sequences as nonfunctional. In addition, the tRNACys and phylogenetically informative sites are from protein- gene of Gekko gecko encodes a tRNA that lacks a coding regions. Only 23% (194 sites) of the variable and D-stem and instead contains a D-arm replacement loop 19% (65 sites) of the phylogenetically informative sites (Fig. 3). The sequences reported here provide further are from tRNA genes and noncoding regions. Of the 276 evidence that loss of a recognizable origin for light-strand phylogenetically informative characters from protein- replication between the tRNAAsn and tRNACys genes and coding regions, 159 are from third positions of codons. changes in secondary structure of tRNACys may be evolu- Third-position sites account for nearly half of the tionarily coupled (Macey et al., 1997b,c,d, 1998c). phylogenetically informative sites in the total data set. Only 12% (40 sites) of the phylogenetically informative Assessment of Homology and Sequence Alignment sites occur in regions encoding stems of tRNAs, suggesting that compensatory substitutions do not compromise Among protein-coding genes, a length-variable re- the phylogenetic analysis. gion encoding the C-terminus of ND1 extending beyond that of Eublepharus is excluded from analyses (posi- Phylogenetic Relationships tions 101–118). As aligned in Fig. 2, this region contains The allozymic data are unable to resolve relation- no phylogenetically informative sites. Teratoscincus ships among Teratoscincus species (Table 3). When the scincus has two extra amino acids in the ND2 gene and data are coded as combinations of alleles for a particu- gaps are placed in the other taxa at positions 1330– lar locus and used as discrete character states, no 1335. Gekko gecko appears to have a pseudo-start codon informative characters are found. When step matrices of ACG for COI in the position homologous with the start are used to connect these character states, three equally codon of other gekkonid lizards. The codon directly follow- most parsimonious trees are produced with a length of ing these three nucleotides is ATG, the typical start codon 99 steps. A strict consensus tree of these three trees for most mitochondrial proteins (Fig. 2). does not contain a monophyletic Teratoscincus; instead, In two tRNA genes, both the dihydrouridine (D) and the grouping of T. przewalskii and Cyrtodactylus is the T⌿C (T) loops are excluded from analyses because found (results not shown). This grouping collapses in a of questionable alignment. In the tRNATrp gene, the single step and therefore is not a significant result. positions excluded are 1398–1409 (D-loop) and 1441– When these allozymic data are coded as the presence 1448 (T-loop); in the tRNACys gene, the positions ex- or absence of alleles, 13 informative characters identify cluded are 1658–1665 (T-loop) and 1697–1699 (D-loop). two equally most parsimonious trees of 78 steps (consistency index 0.987). A strict consensus of these two trees In addition, Gekko gecko has a tRNACys gene encoding a depicts a monophyletic Teratoscincus, which is sup- tRNA that lacks a D-stem and instead contains a D-arm ported by a bootstrap value of 100% and a decay index replacement loop (Macey et al., 1997c). Transfer-RNA of 11 (results not shown). One of the shortest trees gene sequences that encode tRNAs with D-arm replace- groups T. scincus and T. przewalskii as sister species ment loops are subject to stem realignments and shifts and the other tree shows T. scincus and T. roborowskii in bases from one stem to another, making homology of as sister species. The third possibility of a sister-taxon bases difficult to assess (Macey et al., 1997c). Therefore, relationship between T. przewalskii and T. roborowskii sequences in the AA-stem and adjacent bases of the requires only a single extra step. Cys tRNA gene from Gekko gecko are coded as missing in In contrast to the allozymic data, the DNA sequence phylogenetic analyses (positions 1645–1657, 1704– data provide exceptionally good phylogenetic resolu- 1712). In addition, the origin for light-strand replication. The shortest estimate of phylogeny derived from tion (OL) is unalignable between Gekko gecko and the the 1733 aligned positions (341 informative) is stable other taxa and therefore excluded from phylogenetic on all branches until trees with lengths of 29 steps analyses (positions 1612–1644). longer are retained in a parsimony search (Fig. 4). In Of the 1815 aligned positions, 1733 are used in the addition, the bootstrap analysis finds no other trees; phylogenetic analysis. The excluded positions are less thus, all branches receive a 100% bootstrap value. than 5% of the aligned positions. The sister-group relationship of the Chinese T. przew- VICARIANT PATTERNS IN ASIAN Teratoscincus LIZARDS 327

TABLE 3

Distribution of Allelic Variation in Cyrtodactylus, Gekko, and Teratoscincus Species

1 2 3 4 5 6 7 8 9 10 N ACOH1 ACOH2 AAT EST-D FBA GPI G3PDH IDDH IDH1 IDH2

Cyrtodactylus tibetanus 10 g b a a e b c e b e Gekko gecko 7 g a b e d a d d 0.50 c d f 0.50 Teratoscincus scincus 10 c 0.15 c 0.20 b b 0.10 b a 0.20 a 0.20 a 0.35 c 0.90 a 0.05 d 0.05 d 0.70 c 0.90 c 0.80 b 0.80 c 0.65 d 0.10 c 0.05 e 0.75 e 0.10 f 0.90 f 0.05 Teratoscincus przewalskii 10 a d b c c a b c a 0.05 c c 0.90 e 0.05 Teratoscincus roborowskii 10 b f b c 0.60 a a b b 0.05 c b 0.10 d 0.40 c 0.95 c 0.90

11 12 13 14 15 16 17 18 19 N LDH1 LDH2 MDH1 MDH2 PEP-B PEP-D PGDH PNP SOD

Cyrtodactylus tibetanus 10 b a b c c d c c c Gekko gecko 7a b a a a a0.14a a b b 0.86 Teratoscincus scincus 10 a c a b b c 0.95 b b a f 0.05 Teratoscincus przewalskii 10 a c a b b e b b a Teratoscincus roborowskii 10 a c a b b c b b a

TABLE 4

Distribution of Phylogenetically Informative and Variable Positions

ND1 Codon positions tRNAIle tRNAGln tRNAMet

1st 2nd 3rd Stem Nonstem Stem Nonstem Stem Nonstem

Informative sites 7 4 13 638311 Variable sites 19 8 30 10 12 22 15 8 9 ND2 Codon positions tRNATrp a tRNAAla tRNAAsn Noncodingb 1st 2nd 3rd Stem Nonstem Stem Nonstemregion 1 Stem Nonstem

Informative sites 61 40 143 6 — 5 3 — 6 6 Variable sites 185 112 286 17 2 17 7 1 19 8 COI tRNACys a tRNATyr Codon positions Noncodingb Stem Nonstem Stem Nonstemregion 2 1st 2nd 3rd

Informative sites 6 2 2 6 1 4 1 3 Variable sites 11 5 18 12 1 5 3 9 Protein coding codon positions tRNA Noncoding All aligned Total 1st 2nd 3rd Stem Nonstem regions sequence

Informative sites 72 45 159 40 24 1 341 Variable sites 209 123 325 122 70 2 851

a Not including D- and T-loops, which are excluded from the analyses. b Noncoding region 1 is between the tRNAAla and the tRNAAsn genes. Noncoding region 2 is between the tRNATyr and the COI genes. 328 MACEY ET AL. alskii and T. roborowskii is the weakest branch DISCUSSION with a decay index of 29. This clade is the sister taxon to T. scincus from the former USSR and southwest The DNA sequences recover phylogenetic relation- Asia; monophyly of the group containing T. przewal- ships on all branches of the tree with remarkable skii, T. roborowskii, and T. scincus is supported support (Fig. 4). Phylogenetic analysis is most condu- by a decay index of 30. Hence, T. microlepis from cive to reconstructing branching events that are well the Helmand Basin and adjacent regions in south- spaced in time, and the geologic history of Asia suggests west Asia is the basal taxon among Teratoscin- that divergence events among Teratoscincus species are cus species sampled. The monophyly of Teratoscin- separated by sizable intervals. cus is extremely well supported with a decay index The phylogenetic estimate recovered is consistent of 62. with geological predictions. A sequential uplifting of The phylogenetic results provide an area cladogram mountains resulted from compression of Gondwanan for Central Asia. To conﬁrm these results and to test plates along the southern margin of Eurasia as the support for the origins of clades found in separate impinging Indian subcontinent moved northward. Tera- historical regions, the Wilcoxon signed-ranks test toscincus species may be assigned to tectonic plates of (Felsenstein, 1985; Templeton, 1983) is applied. Gondwanan origin that are now in Asia. While the evolutionary history of Teratoscincus postdates the When the overall shortest tree (Fig. 4) is compared Indian collision, which occurred in the middle Eocene independently to the shortest alternative trees (Appen- [Dewey et al., 1989 (45 MYBP); Windley, 1988 (50 dix 1) not showing each of the nodes in the over- MYBP)], the pre-Eocene events produced the geological all shortest tree, these alternative trees are rejected conditions necessary for the Eocene and later events in favor of the overall shortest tree (see Appendix 1; that are directly related to the phylogenetic history A-1 and 2, n ϭ 53, T ϭ 324; B-1, n ϭ 58, T ϭ 413; s s under investigation (Fig. 5). C-1, n ϭ 128, T ϭ 2128.5; all P Ͻ 0.001). All subopti- s Approximately 300 MYBP, the Tarim Plate (Taklima- mal trees are rejected in favor of the overall shortest kan Desert) collided with the Siberian and Kazakhstan tree. blocks of Laurasia (Feng et al., 1989; Kwon et al., 1989) forming the paleo-Tien Shan (mountains) in the vicinity of what is now the western Chinese–former Soviet border (Fig. 5). A complex series of events followed with the successive accretion of blocks to the southern margin of Eurasia. The blocks termed the Cimmerian Continent broke from Gondwanaland in the late Per- mian (250 MYBP), migrated northward across the Tethys Sea, and subsequently broke apart along the way. The individual blocks collided with Eurasia, and most completed suturing from late Triassic to middle Jurassic (225–175 MYBP; for details see Sengo¨r, 1984; Sengo¨r et al., 1988). These blocks now are situated from Turkey through Iran, Afghanistan, and Tibet. They include the North Tibet Block (Qiangtang and Songban- Ganzi Terranes), South Tibet Block (Lhasa Terrane), Farah Block (northern Afghanistan), and Helmand Block (southern Afghanistan). Figure 5 shows the current positions of these blocks. Paleo-sutures of these blocks were reactivated as strike-slip faults by the Indian collision, and dramatic mountain building occurred along these faults, separat- ing Teratoscincus species. A sequential uplifting of mountain ranges is observed, with the ordering largely dependent on geographic distance from the contact of the collision. The Indian collision ﬁrst caused the trans-Himalayan uplift (Hindu Kush, Karakorum, and Himalaya), which was followed by the Pamir-Tien Shan FIG. 4. Single most parsimonious tree produced from analysis of uplift. the 1733 aligned (341 phylogenetically informative) positions from mitochondrial DNA sequences. The tree has a length of 1268 steps Teratoscincus microlepis can be assigned unambigu- and a consistency index of 0.864. Bootstrap values are presented ously to the Helmand Block south of the Hindu Kush. above branches and decay indices below branches. Teratoscincus przewalskii and T. roborowskii can be VICARIANT PATTERNS IN ASIAN Teratoscincus LIZARDS 329

FIG. 5. Movements of tectonic plates. During the middle Eocene, 50–45 MYBP, India first contacted Eurasia. Since that time, India and Laurasian plates have converged 2365 km in the west, 2475 in the center, and 2750 km in the east (Dewey et al., 1989; Molnar et al., 1987). The high altitudes now present in the Hindu Kush, Karakoram, Tien Shan, and Pamir are attributed to the Indian collision (Dewey et al., 1988, 1989). The Hindu Kush between the Helmand and Farah blocks is associated with the uplift of the trans-Himalaya, which includes the Karakoram and is one of the earlier uplifting events. The Tien Shan and Pamir, which now separate the Taklimakan Desert (Tarim Plate) from the former Soviet deserts and the Farah Block, were formed approximately 10 MYBP (Abdrakhmatov et al., 1996; Tapponier et al., 1981). Crustal shortening and deformation rates are from Dewey et al. (1989). The map is modified from Tapponier et al. (1981). assigned unambiguously to the Tarim Block east of the formation of the Hindu Kush, which predates the Tien Tien Shan and Pamir. Teratoscincus scincus can be Shan-Pamir uplift, would have separated T. microlepis assigned either to the Farah or the Helmand blocks, on the Helmand Block from T. scincus on the Farah north or south, respectively, of the Hindu Kush. Note Block. that Laurasian plates north of the Farah Block under- Using the date of 10 MYBP for the formation of the went periods of flooding by the Paratethys Sea until 3.5 Tien Shan-Pamir uplift, which separates T. scincus on MYBP (Dercourt et al., 1986; Steininger and Rogl, the Farah Block from T. przewalskii and T. roborowskii 1984) and therefore are unlikely to be a region of on the Tarim Block, a rate of evolutionary divergence of historical endemism for T. scincus. The Tien Shan- 0.57% per lineage per million years is estimated. This Pamir is well dated at 10 MYBP (Abdrakhmatov et al., estimate approximates the rate of 0.65% divergence per 1996; Tapponier et al., 1981). The separation between lineage per million years suggested for the same seg- T. scincus from the ancestor of T. przewalskii and ment of the mitochondrial genome from fish, bufonid T. roborowskii is expected to be coincident with the frogs, and agamid lizards (Bermingham et al., 1997; uplift of the Tien Shan-Pamir (10 MYBP). If this Macey et al., 1998a,b). If our interpretation of T. scincus hypothesis is correct, T. scincus would be assigned to being from the Farah Block is correct, then the forma- the Farah Block north of the Hindu Kush. Hence, the tion of the Hindu Kush would separate T. scincus on the 330 MACEY ET AL.

TABLE 5 second, the formation of the Tien Shan-Pamir occurred between the Farah and the Tarim blocks at 10 MYBP. Pairwise Comparisons of DNA Sequences Between Eublepharus, Gekko, and Teratoscincus Speciesa ACKNOWLEDGMENTS 123456

1. Eublepharus — 35.5% 31.0% 30.8% 30.3% 30.6% This work was supported by grants from the National Science 2. Gekko 591 — 30.7% 32.1% 32.0% 32.5% Foundation (predoctoral fellowship to J.R.M.; DEB-9726064 to A.L., 3. T. microlepis 527 510 — 16.0% 15.9% 16.6% J.R.M. and T.J.P.), National Geographic Society (4110-89 and 4872-93 4. T. scincus 525 534 272 — 11.1% 11.6% to T.J.P. and J.R.M.), Russian Foundation of Basic Research (N 5. T. przewalskii 516 533 271 190 — 6.5% 97-04-50093 to N.B.A.), and the California Academy of Sciences. We 6. T. roborowskii 521 541 282 198 112 — thank Zhili Fang, Tatjana Dujsebayeza, and Ermi Zhao for ﬁeld assistance, and Alok Verma for assistance in the laboratory. The ﬁrst a Percentage sequence divergence is shown above the diagonal and author thanks David B. Wake and Margaret F. Smith for the the number of base substitutions between sequences is shown below opportunity to collect allozymic data at the Museum of Vertebrate the diagonal. Taxa are abbreviated with T. representing Teratoscin- Zoology. cus. APPENDIX 1 Farah Block from T. microlepis on the Helmand Block. Alternative hypotheses used in Wilcoxon signed- This event predates the formation of the Tien Shan- ranks tests (Felsenstein, 1985; Templeton, 1983). Pamir uplift and these taxa should show more than 10 Lengths of trees and consistency indices (CI) (Swofford, million years of sequence divergence. The average 1993) are given in parentheses. pairwise sequence divergence across the Tien Shan- The most parsimonious trees that do not contain a Pamir is 11.4% and that between T. microlepis and monophyletic grouping of T. przewalskii and T. ro- T. scincus, presumably across the Hindu Kush, is 15.9% borowskii (length of 1297 steps and a CI of 0.845): A-1 ϭ (Table 5). Because saturation of mitochondrial se- (Eublepharus, (Gekko, (T. microlepis, ((T. scincus, quences occurs past 10 MYBP (Moritz et al., 1987), this T. roborowskii), T. przewalskii)))). A-2 ϭ (Eublepharus, latter split is probably much older than 10 MYBP. (Gekko, (T. microlepis, ((T. scincus, T. przewalskii), Hence, the DNA sequence data are consistent with T. roborowskii)))). our interpretation of two vicariant events across paleo- The most parsimonious tree that does not contain a sutures of accreted Gondwanan plates (Fig. 6). First, monophyletic grouping of T. scincus, T. przewalskii, and the formation of the Hindu Kush between the Helmand T. roborowskii (length of 1298 steps and a CI of 0.844): and the Farah blocks occurred before 10 MYBP and, B-1 ϭ (Eublepharus, (Gekko, ((T. microlepis, (T. przewalskii, T. roborowskii)), T. scincus))). The most parsimonious tree that does not contain a monophyletic grouping of all Teratoscincus species (length of 1330 steps and a CI of 0.824): C-1 ϭ (Eublepharus, ((Gekko, T. microlepis), (T. scincus, (T. przewalskii, T. roborowskii)))).

REFERENCES

Abdrakhmatov, K. Ye., Aldazhanov, S. A., Hager, B. H., Hamburger, M. W., Herring, T. A., Kalabaev, K. B., Makarov, V. I., Molnar, P., Panasyuk, S. V., Prilepin, M. T., Reilinger, R. E., Sadybakasov, I. S., Souter, B. J., Trapeznikov, Yu. A., Tsurkov, V. Ye., and Zubovich, A. V. (1996). Relatively recent construction of the Tien Shan inferred from GPS measurements of present-day crustal deformation rates. Nature 384: 450–453. Anderson, S., Bankier, A. T., Barrell, B. G., de Bruijn, M. H. L., FIG. 6. Area cladogram for the tectonic plates derived from Coulson, A. R., Drouin, J., Eperon, I. C., Nierlich, D. P., Roe, B. A., Gondwanaland based on fragmentation patterns among Teratoscin- Sanger, F., Schreier, P. H., Smith, A. J. H., Staden, R., and Young, cus species. Approximate dates for branching events are shown. As I. G. (1981). Sequence and organization of the human mitochon- the Indian subcontinent impinged into Eurasia following collision, drial genome. Nature 290: 457–465. paleo-sutures were reactivated as strike-slip faults that formed mountain chains (Fig. 5). Two events are suggested to have caused Anderson, S. C. (1993). A note on the syntopic occurrence of three vicariant speciation among Teratoscincus species. First, the forma- species of Teratoscincus in eastern Iran. Dactylus 1(4): 8–10. tion of the Hindu Kush between the Helmand and the Farah blocks Bermingham, E., McCafferty, S. S., and Martin, A. P. (1997). Fish occurred prior to 10 MYBP and, second, the formation of the Tien biogeography and molecular clocks: Perspectives from the Panama- Shan-Pamir occurred between the Farah and the Tarim blocks at 10 nian Isthmus. In ‘‘Molecular Systematics of Fishes’’ (T. D. Kocher MYBP. The biological and geological data are in complete agreement. and C. A. Stepien, Eds.), pp. 113–128. Academic Press, San Diego. VICARIANT PATTERNS IN ASIAN Teratoscincus LIZARDS 331

Bremer, K. (1994). Branch support and tree stability. Cladistics 10: caucasia species group: Testing hypotheses of biogeographic frag- 295–304. mentation and an area cladogram for the Iranian Plateau. Mol. Brennicke, A., and Clayton, D. A. (1981). Nucleotide assignment of Phylogenet. Evol. 10: 118–131. alkali-sensitive sites in mouse mitochondrial DNA. J. Biol. Chem. Macey, J. R., Schulte, J. A., II, Larson, A., Fang, Z., Wang, Y., Tuniyev, 256: 10613–10617. B. S., and Papenfuss, T. J. (1998b). Phylogenetic relationships of Buth, D. G. (1984). The application of electrophoretic data in system- toads in the Bufo bufo species group from the eastern escarpment atic studies. Annu. Rev. Ecol. Syst. 15: 501–522. of the Tibetan Plateau: A case of vicariance and dispersal. Mol. Phylogenet. Evol. 9: 80–87. Clayton, J. W., and Tretiak, D. N. (1972). Amine-citrate buffers for pH control in starch gel electrophoresis. J. Fish. Res. Board Canada Macey, J. R., Schulte, J. A., II, Larson, A., and Papenfuss, T. J. 29: 1169–1172. (1998c). Tandem duplication via light-strand synthesis may provide a precursor for mitochondrial genomic rearrangement. Mol. Dercourt, J., Zonenshain, L. P., Ricou, L.-E., Kazmin, V. G., Le Biol. Evol. 15: 71–75. Pichon, X., Knipper, A. L., Grandjacquet, C., Sbortshikov, I. M., Geyssant, J., Lepvrier, C., Pechersky, D. H., Boulin, J., Sibuet, Macey, J. R., and Verma, A. (1997). Homology in phylogenetic J.-C., Savostin, L. A., Sorokhtin, O., Westphal, M., Bazhenov, M. L., analysis: Alignment of transfer RNA genes and the phylogenetic Lauer, J. P., and Biju-Duval, B. (1986). Geological evolution of the position of snakes. Mol. Phylogenet. Evol. 7: 272–279. tethys belt from the Atlantic to the Pamirs since the Lias. Tectono- Maddison, W. P., and Maddison, D. R. (1992). ‘‘MacClade, analysis of physics 123: 241–315 and maps (plates I–X). phylogeny and character evolution, version 3.0,’’ Sinauer, Sunder- Dewey, J. F., Shackleton, R. M., Chang, C., and Sun, Y. (1988). The land, MA. tectonic evolution of the Tibetan Plateau. Phil. Trans. R. Soc. Lond. Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982). ‘‘Molecular A 327: 3–16. Cloning: A Laboratory Manual,’’ Cold Spring Harbor Laboratory Dewey, J. F., Cande, S., and Pitman, W. C., III. (1989). Tectonic Press, Cold Spring Harbor, NY. evolution of the India/Eurasia collision zone. Eclogae Geol. Helv. Molnar, P., Burchfiel, B. C., Liang, K., and Zhao, Z. (1987). Geomor- 82: 717–734. phic evidence for active faulting in the Altyn Tagh and northern Felsenstein, J. (1985). Confidence limits on phylogenies with a Tibet and qualitative estimates of its contribution to the conver- molecular clock. Syst. Zool. 34: 152–161. gence of India and Eurasia. Geology 15: 249–253. Feng, Y., Coleman, R. G., Tilton, G., and Xiao, X. (1989). Tectonic Moritz, C., Dowling, T. E., and Brown, W. M. (1987). Evolution of evolution of the west Junggar region, Xinjiang, China. Tectonics 8: animal mitochondrial DNA: Relevance for population biology and 729–752. systematics. Annu. Rev. Ecol. Syst. 18: 269–292. Harris, H., and Hopkinson, D. A. (1976 et seq.). ‘‘Handbook of Enzyme Murphy, R. W., Sites, J. W., Jr., Buth, D. G., and Haufler, C. H. (1990). Electrophoresis in Human Genetics,’’ Oxford, North-Holland, Am- Proteins I: Isozyme electrophoresis. In ‘‘Molecular Systematics’’ sterdam. [Loose leaf with supplements in 1977 and 1978] (D. M. Hillis and C. Moritz, Eds.), pp. 45–126. Sinauer, Sunder- Hixson, J. E., Wong, T. W., and Clayton, D. A. (1986). Both the land, MA. conserved and divergent 5Ј-flanking sequences are required for Richardson, B. J., Baverstock, P. R., and Adams, M. (1986). ‘‘Allozyme initiation at the human mitochondrial origin of light strand Electrophoresis. A Handbook for Animal Systematics and Popula- replication. J. Biol. Chem. 261: 2384–2390. tion Studies,’’Academic Press, Sydney. Kumazawa, Y., and Nishida, M. (1993). Sequence evolution of mito- Selander, R. K., Smith, M. H., Yang, S. Y., Johnson, W. E., and Gentry, chondrial tRNA genes and deep-branch animal phylogenetics. J. J. R. (1971). Biochemical polymorphism and systematics in the Mol. Evol. 37: 380–398. genus Peromyscus. I. Variation in the old-field mouse (Peromyscus Kwon, S.-T., Tilton, G. R., Coleman, R. G., and Feng, Y. (1989). polionotus). Studies in Genetics VI. Univ. Texas Publ. 7103: 49–90. Isotopic studies bearing on the tectonics of the west Junggar Sengo¨r, A. M. C. (1984). The Cimmeride orogenic system and the region, Xinjiang, China. Tectonics 8: 719–727. tectonics of Eurasia. Geol. Soc. Am. Spec. Paper No. 195: 1–82. Larson, A. (1998). The comparison of morphological and molecular Sengo¨r, A. M. C., Altiner, D., Cin, A., Ustaomer, T., and Hsu, K. J. data in phylogenetic systematics. In ‘‘Molecular Approaches to (1988). Origin and assembly of the Tethyside orogenic collage at the Ecology and Evolution’’ (R. DeSalle and B. Schierwater, Eds.), pp expense of Gondwana Land. In ‘‘Gondwana and the Tethys’’ (M. G. 275–296. Birkhaüser, Basel. Audley-Charles and A. Hallam, Eds.), pp. 119–181. Geol. Soc. Spec. Mabee, P. M., and Humphries, J. (1993). Coding polymorphic data: Publ. No. 37. Oxford Univ. Press, Oxford. Examples from allozymes and ontogeny. Syst. Biol. 42: 166–181. Steininger, F. F., and Rogl, F. (1984). Paleogeography and palinspas- Macey, J. R., Ananjeva, N. B., Wang, Y., and Papenfuss, T. J. (1997a). tic reconstruction of the Neogene of the Mediterranean and Parate- A taxonomic reevaluation of the gekkonid lizard genus Teratoscin- thys. In ‘‘The Geological Evolution of the Eastern Mediterranean’’ cus in China. Russian J. Herpetol. 4: 8–16. (J. E. Dixon and A. H. F. Robertson, Eds.), pp. 659–668. Blackwell Sci., Oxford. Macey, J. R., Larson, A., Ananjeva, N. B., Fang, Z., and Papenfuss, T. J. (1997b). Two novel gene orders and the role of light-strand Swofford, D. L. (1993). ‘‘PAUP: Phylogenetic analysis using parsi- replication in rearrangement of the vertebrate mitochondrial ge- mony, version 3.1.1,’’ Illinois Natur. Hist. Survey., Urbana, IL. nome. Mol. Biol. Evol. 14: 91–104. Swofford, D. L., and Olsen, G. J. (1990). Phylogeny reconstruction. In Macey, J. R., Larson, A., Ananjeva, N. B., and Papenfuss, T. J. (1997c). ‘‘Molecular Systematics’’ (D. M. Hillis and C. Moritz, Eds.), pp. Replication slippage may cause parallel evolution in the secondary 411–501. Sinauer, Sunderland, MA. structures of mitochondrial transfer RNAs. Mol. Biol. Evol. 14: Szczerbak, N. N., and Golubev, M. L. (1996). Gecko Fauna of the 31–39. USSR and Contiguous Regions. Soc. Study Amphib. Reptiles Contr. Macey, J. R., Larson, A., Ananjeva, N. B., and Papenfuss, T. J. Herpetol. 13: 1–233. (1997d). Evolutionary shifts in three major structural features of Tapponier, P., Mattauer, M., Proust, F., and Cassaigneau, C. (1981). the mitochondrial genome among iguanian lizards. J. Mol. Evol. Mesozoic ophiolites, sutures, and large-scale tectonic movements 44: 660–674. in Afghanistan. Earth Planet. Sci. Lett. 52: 355–371. Macey, J. R., Schulte, J. A., II, Ananjeva, N. B., Larson, A., Rastegar- Templeton, A. R. (1983). Phylogenetic inference from restriction Pouyani, N., Shammakov, S. M., and Papenfuss, T. J. (1998a). endonuclease cleavage site maps with particular reference to the Phylogenetic relationships among agamid lizards of the Laudakia evolution of humans and the apes. Evolution 37: 221–244. 332 MACEY ET AL.

Windley, B. F. (1988). Tectonic framework of the Himalaya, Karako- Zhang, D.-X., and Hewitt, G. M. (1996). Nuclear integrations: Chal- rum and Tibet, and problems of their evolution. Phil. Trans. R. Soc. lenges for mitochondrial DNA markers. Trends Ecol. Evol. 11: Lond. A 326: 3–16. 247–251. Zar, J. H. (1984). ‘‘Biostatistical Analysis,’’ 2nd ed., Prentice Hall Zhao, E.-M., and Adler, K. (1993). Herpetology of China. Soc. Study International, Englewood Cliffs, NJ. Amphib. Reptiles Contr. Herpetol. 10: 1–522.