Relationships of the Salamandrid Genera Paramesotriton, Pachytriton, and Cynops Based on Mitochondrial DNA Sequences

Copeia, 2001(4), pp. 997–1009

LAUREN M. CHAN,KELLY R. ZAMUDIO, AND DAVID B. WAKE

We compared 786 base pairs of cytochrome b mitochondrial DNA sequence to examine the evolutionary relationships among seven species belonging to three genera of Asian newts: Paramesotriton, Pachytriton, and Cynops. We ﬁnd strong evidence supporting recognition of a clade for these genera. Although bootstrap support values are relatively low, both parsimony and likelihood analyses suggest that the species of Paramesotriton sampled form a monophyletic group with Paramesotriton caudopunctatus basal to the other three species. Cynops appears to be paraphyletic, with Pachytriton and Paramesotriton being more closely related to Cynops pyrrhogaster than to Cynops cyanurus. Pachytriton and Paramesotriton exhibit some morphological similarities and have more specialized breeding habits and environmental require- ments than Cynops, suggesting that they shared an evolutionary history before di- verging. Our morphological investigations corroborate previous studies that suggested Cynops is the most generalized representative of the clade and that it retains several ancestral character states.

SIAN newts of the genera Paramesotriton, the molecular analyses, thereby offering limited A Pachytriton, and Cynops comprise 15 cur- resolution on relationships among species. rently recognized species and several undescri- Nonetheless, in the combined molecular and bed species that are widely distributed in south- morphological analyses Cynops, Paramesotriton, eastern Asia, including Japan, China, and north- and Pachytriton form a well-supported polytomy ern Vietnam (Zhao, 1999; Thorn and Raffae¨lli, establishing the monophyly of this group. 2001). Phylogenetic studies of morphology Studies including multiple species from each (Wake and O¨ zeti, 1969) and molecular charac- genus have also had difficulty resolving relation- ters (Titus and Larson, 1995) for the family Sal- ships within this clade of Asian newts. An allo- amandridae have linked these three genera in zyme study of salamandrids including two spe- a trichotomy with little or no resolution. Initial cies of Cynops (pyrrhogaster and ensicauda) and surveys focusing on variation in behavior, repro- one species of Paramesotriton (hongkongensis) but ductive pattern, external morphology, and no species of Pachytriton suggested that Cynops skull/hyobranchial characters found that the re- may be paraphyletic with respect to Paramesotri- lationships among the three genera differed de- ton (Hayashi and Matsui, 1989). Finally, a phe- pending on the characters used for analyses netic investigation of the relationships within (Wake and O¨ zeti, 1969). These authors con- Paramesotriton found evident differences in hy- cluded that Pachytriton is most closely related to oid apparatus and skull characters among five Cynops wolterstorffi (at the time placed in the species in this genus (Pang et al., 1992). Be- monotypic genus Hypselotriton) based on overall cause no outgroups from other genera were similarity in morphology and aquatic feeding used for comparison, it is difficult to determine mechanisms and argued that these two might how these results extend to further relation- have arisen from an ancestral stock close to that ships among all Asian newts. The general lack which gave rise to Cynops (only Cynops pyrrhogas- of resolution in these systematic studies stems ter was available to them). However, they were from the conservative nature of morphology in uncertain concerning the phylogenetic relation- this lineage, with numerous plesiomorphic ships of Paramesotriton, which had been consid- characters retained in species having an overall ered a close relative of Cynops and Pachytriton ‘‘generalized’’ form (O¨ zeti and Wake, 1969). (Freytag and Petzold, 1961; Freytag, 1962). Here we compare mitochondrial DNA A more recent investigation combining mo- (mtDNA) gene sequences from representatives lecular (mitochondrial DNA) and morphologi- of these three genera to examine the phyloge- cal characters was also unable to resolve these netic relationships among them. Specifically, we relationships (Titus and Larson, 1995). Howev- use molecular data to address three questions er, this study focused on higher level relation- about relationships within this group using mo- ships among salamandrids and included only lecular data. First, we examine higher level re- one species of each of our three focal genera in lationships among the genera Paramesotriton,

᭧ 2001 by the American Society of Ichthyologists and Herpetologists 998 COPEIA, 2001, NO. 4

Pachytriton, and Cynops. Second, we focus within eye and was straightforward because no inser- each genus to look at relationships among spe- tions or deletions were present. Amino acid cies. And third, we address the possibility of the translations of our sequences were compared paraphyly of Cynops. In addition, we compare with that of Xenopus (Roe et al., 1985) to ensure morphological data for 14 of the 15 described that there were no nonsense mutations or species and one undescribed species to identify frameshifts. We sequenced 19 individuals of diagnostic characters that may further clarify re- which 15 were unique haplotypes used for phy- lationships within this clade. logenetic analysis (submitted to GenBank under accession numbers AF295671–AF295685). In MATERIALS AND METHODS addition, we included in our analyses the partial cytochrome b sequences for two species of Tri- Laboratory protocols.—Included in our study were turus (vulgaris and carnifex) obtained from 16 individuals representing four of the six spe- GenBank. cies of Paramesotriton, one of the two species of All phylogenetic analyses of the cytochrome b Pachytriton, and two of the seven species of Cy- sequences were conducted using the program nops. Two species of Triturus (vulgaris and car- PAUP*4.0beta2 (D. L. Swofford, Sinaner Assoc., nifex) from Europe, Taricha granulosa from the Inc., Sunderland, MA, 1999, unpubl.). We as- United States, and two species of Tylototriton (tal- signed the two species of Tylototriton as out- iangensis and verrucosus) from southeast Asia groups for all analyses. Pairwise sequence diver- were selected as sequential outgroups to our gences and Kimura two-parameter (K2p) cor- clade (Titus and Larson, 1995; Appendix 1). rected divergences were estimated among all Partial cytochrome b mtDNA sequences for Tri- pairs of sequences. We assessed levels of satu- turus vulgaris and Triturus carnifex were obtained ration for base substitutions by plotting percent from GenBank (Caccone et al., 1997; accession sequence divergences against K2p distances for numbers U55498 and U55499). For all other in- transitions and transversions at each codon po- dividuals, genomic DNA was isolated from fro- sition. Kimura two-parameter values higher than zen tissues or from samples preserved in EtOH corresponding uncorrected percent sequence by standard proteinase K digestion followed by divergence suggest that transitions and transver- either salt or phenol-chloroform purification. sions at the third codon position may be satu- We used the polymerase chain reaction (PCR) rated (Fig. 1). Therefore, to determine the ef- to amplify approximately 690 base pairs of the fect that saturation may have on topology, we cytochrome b region of the mtDNA with the primers MVZ16 (5Ј-AAA TAG GAA RTA TCA analyzed our data using both equal weighting YTC TGG TTT RAT-3Ј) and either MVZ15 (5Ј- and with third position changes downweighted GAA CTA ATG GCC CAC ACW WTA CGN AA- to both 25% and 50% of first and second posi- 3Ј) or Triton-cytb-Fl (5Ј-CAA CGC CAT CAA tion changes. Other than the weighting option, ACA TCT CA-3Ј). PCR amplification reactions all other assumptions and parameters were were performed in total volumes of 25 ␮l with identical in phylogenetic reconstruction. containing 100 ng of DNA template, 1X Taq Maximum parsimony (MP) analyses consisted buffer, 1.0 ␮M of each primer, 0.75 mM dNTPs, of branch-and-bound searches using initial up- per bound computed via stepwise addition, 1.5 mM MgCl2, and 0.625 units Taq polymerase. Amplification consisted of initial denaturation ‘‘furthest’’ addition sequence, and ‘‘MulTrees’’ at 94 C for 5 min followed by 35 cycles of de- options in effect. We also performed MP boot- naturation for 1 min at 94 C, annealing for 1 strap analysis, with 1000 replicates, as a measure min at 45–47 C, and extension for 1.25 min at of internal support; the settings for bootstrap 72 C. PCR amplifications were terminated with analyses were the same as those for the original a final extension period of 5 min at 72 C. We branch-and-bound search. used ABI fluorescent dye terminator chemistry Maximum likelihood (ML) analyses included to cycle sequence fragments in both directions heuristic searches with 100 replicates of random with the same primers used in amplification. addition of sequences and one tree held at each Products were electrophoresed on a 4.75% ac- step. For ML analyses, we selected TBR branch rylammide gel on an ABI 377 automated se- swapping, the ‘‘MulTrees’’ option in effect, and quencer (Applied Biosystems, Costa Mesa, CA). ‘‘steepest descent’’ option not in effect. We chose the HKY model (Hasegawa et al., 1985), Phylogenetic analyses.—MtDNA sequences were with starting branch lengths obtained using aligned to each other and to the cytochrome b Rogers-Swofford approximation and no en- sequence of Xenopus laevis in the program Se- forcement of a molecular clock. ML bootstrap quencher version 3.1. Alignment was done by used the same settings, with the exception that CHAN ET AL.—ASIAN NEWT SYSTEMATICS 999

Fig. 1. Pairwise divergence plots for all individuals used in molecular analyses. Kimura two-parameter distances (x-axis) are plotted against uncorrected percent sequence divergences (y-axis) for transitions and transversions at each codon position. we used only 10 random sequence addition rep- vious studies that had identified them as useful licates to decrease total analysis time. in discerning among salamander genera and We estimated decay indices for all branches species (e.g., Chang and Boring, 1939; Wake on our parsimony tree using the program and O¨ zeti, 1969; Zhao and Hu, 1988). Our goal AutoDecay 4.0.2 (T. Eriksson, Stockholm Univ., was to examine whether these characters were 1999, unpubl.). We tested our preferred topol- diagnostic among species. Thus, we examined ogy against alternate hypothetical trees using intraspecific variation and compared it to vari- tree comparison tests. Tests were conducted ation previously reported among species. We with the most parsimonious tree under parsi- made cranial measurements, including total mony criteria using the Wilcoxon signed-ranks skull length, face skull length, neural skull test (Templeton, 1983) and with the tree ob- length, skull breadth, and length of the maxil- tained from likelihood searches under likeli- laries. We noted cranial characteristics, includ- hood criteria with the KH test (Kishino and ing the state of the fronto-squamosal arch, po- Hasegawa, 1989). sition of the maxilla relative to the pterygoid, length of the frontal processes and the degree Diagnostic morphological characters.—We gathered of contact of the nasals. In addition, we deter- morphological data for all ingroup taxa includ- mined the tarsal and carpal patterns of each ed in our molecular analysis and most of the limb and the number of trunk, caudo-sacral, remaining species of Paramesotriton, Pachytriton, and caudal vertebrae. Given the limited number and Cynops. These included all described spe- of morphological characters, we did not code cies of Paramesotriton (caudopunctatus, deloustali, character states and subject them to phyloge- fuzhongensis, hongkongensis, guangxiensis, and chi- netic analyses. Our objective was only to identify nensis), both described (labiatus, brevipes) and diagnostic characters and to possibly emphasize one undescribed species of Pachytriton, and six those that would be useful in further phyloge- of the seven described species of Cynops (orien- netic studies. talis, orphicus, cyanurus, pyrrhogaster, wolterstorffi, and ensicauda). Outgroups were excluded from RESULTS morphological analyses because previous studies (e.g., Titus and Larson, 1995) and our own We collected 19 cytochrome b sequences rep- molecular analyses confirm the monophyly of resenting seven species from the three genera the ingroup taxa. Data were scored from and three successively more distant outgroup cleared-and-stained specimens and X-rays of al- species. We also included in our analyses previ- cohol-preserved specimens. We chose morpho- ously published GenBank sequences from two logical characters for examination based on pre- Triturus species as additional outgroups (Cac- 1000 COPEIA, 2001, NO. 4

clade. However, the support values for the branch resulting in a paraphyletic Triturus are consistently low (bootstraps range from 14– 52%). Thus, our topology does not necessarily support the paraphyly of Triturus. In all phylogenetic analyses, the monophyly of the Cynops- Pachytriton-Paramesotriton clade is well supported (for measures of nodal support, see Fig. 2). Within the ingroup clade, measures of nodal support for unweighted analyses were compa- rable to those with third position changes downweighted and some nodes are well supported by bootstrap and decay indices. We find strong support for the monophyly of each of the species for which two or more pop- ulations were sampled. We record a bootstrap value of 100 for the monophyly of the following species (Fig. 2): Paramesotriton deloustali (16 decay), Paramesotriton guangxiensis (27 decay), Par- amesotriton hongkongensis (19 decay), and Pachy- triton labiatus (17 decay). We also see significant Fig. 2. Maximum likelihood phylogram for the phylogenetic structure within Paramesotriton taxa included in this study. Numbers on branches are with unweighted MP bootstrap values of 90 or various measures of nodal support. From top to bot- greater for the clade including P. deloustali and tom: decay indices, unweighted MP bootstraps, P. guangxiensis (90 bootstrap, 7 decay), and an- weighted MP bootstraps (50%), and unweighted ML other clade including P. deloustali, P. guangxien- ϭ bootstraps. Generic abbreviations are as follows: Pr. sis, and P. hongkongensis (93 bootstrap, 8 decay). Paramesotriton, Pc. ϭ Pachytriton, C. ϭ Cynops, Tr. ϭ Triturus, Ta. ϭ Taricha, Ty. ϭ Tylototriton. Little support is found for adding Paramesotriton caudopunctatus to the clade (32 bootstrap, 1 decay). cone et al., 1997). Two Cynops cyanurus had The two species of Cynops do not form a identical sequences as did four of the five Par- clade. Instead our analyses recover a clade in- amesotriton deloustali; therefore, we used 17 cluding Cynops pyrrhogaster and all Paramesotriton unique sequences in our phylogenetic analyses. and Pachytriton (65 bootstrap, 4 decay), suggest- Of the 786 base pairs, 488 were constant, 64 ing that Cynops is paraphyletic with respect to were variable, and 234 were parsimony infor- Paramesotriton and Pachytriton. mative. Among species within ingroup genera, Tree comparison tests in MP (Templeton the largest sequence divergences were 15.8% tests) cannot reject the hypothesis that Cynops is for Paramesotriton and 13.2% for Cynops. For the monophyletic. Constraining the two species of three Pachytriton labiatus, the highest sequence Cynops to be monophyletic results in a tree only divergence (K2p) was 3.5%. Sequence diver- seven steps longer (L ϭ 707, Wilcoxon signed gences among members of the three ingroup ranks comparison to the most parsimonious genera varied from 10.9% between Pachytriton tree P ϭ 0.17). Maximum likelihood (Kishino- labiatus and Paramesotriton caudopunctatus, to Hasegawa) tests are also unable to reject this 18.9% between Cynops cyanurus and Paramesotri- hypothesis. Tree comparison tests result in a ton guangxiensis (Appendix 2). tree only slightly longer (-ln L ϭ 4256.64, Kish- Maximum parsimony analysis yielded a single ino-Hasegawa tree comparison with the most most parsimonious topology (L ϭ 700; CI ϭ likely tree P ϭ 0.17). Nonetheless, phylogenetic 0.599; HI ϭ 0.401). Maximum likelihood yields reconstructions under both optimality criteria one most likely tree with a score of -ln L ϭ yield a topology with the basal paraphyly of Cy- 4248.81. MP and ML trees were identical in to- nops and with moderate values of support for pology with respect to all ingroup taxa and ML bootstrap and higher levels for MP boot- three of the four outgroup taxa (Taricha granu- strap. losa and two species of Tylototriton; Fig. 2). They Morphological data were collected from 14 differed only in relationships among the two described and one undescribed species of Par- Triturus species; in MP the Triturus form a amesotriton, Pachytriton, and Cynops. Members of monophyletic group and in ML they are se- these genera are similar morphologically, shar- quential branches at the base of the ingroup ing the same carpal and tarsal patterns (inter- CHAN ET AL.—ASIAN NEWT SYSTEMATICS 1001 medium and ulnare fused in the manus and dis- caudopunctatus is less robust, and its skull is lon- tal tarsals 4 and 5 fused in the pes) as well as ger and narrower (ratio of skull width to skull other cranial and skeletal characters, such as length of P. caudopunctatus ϭ 0.70 Ϯ 0.01) com- the boxlike nature of the skull with its flattened pared to the broader skulls of the other four dorsal surface and grooved surface of the pari- species of Paramesotriton (width to length of Par- etals behind the fronto-squamosal arches (Fig. amesotriton excluding P. caudopunctatus ϭ 0.85 Ϯ 3). However, species in these three genera are 0.11). The moderately stout and bony epibran- not identical and variation in cranial and exter- chials of P. caudopunctatus are flared dorsolat- nal morphology, especially coloration, is useful erally, similar to the epibranchials of Pachytriton, in distinguishing among them. whereas the other species of Paramesotriton have Of the species examined, those in the genus a hyobranchial apparatus like the one described Pachytriton are the most distinct in terms of mor- for P. hongkongensis with relatively slender and phology. All species of Pachytriton have smooth nearly straight epibranchials (O¨ zeti and Wake, skin, a relatively slender body lacking a vertebral 1969). ridge and a tail that is compressed laterally to We examined six of the seven described spe- varying degrees (see plate 6B Zhao and Hu, cies of Cynops, and whereas we found osteolog- 1988; plates 4B and 4C Zhao and Adler, 1993). ical variation, there is overall morphological The skull of Pachytriton is long and narrow; the similarity among species. All Cynops are smaller- average skull breadth to skull length ratio for bodied than either Pachytriton or Paramesotriton all adults examined is 0.66 Ϯ 0.06. Where the (Fig. 3). They have a vertebral ridge, although relatively long maxillary bones approach the it is not always prominent, and almost all indi- pterygoids, the elements form approximately viduals lack lateral ridges. The tail is laterally straight lines. The fronto-squamosal arch is rare- compressed, and except for C. wolterstorffi, the ly complete and attenuate if formed. In all spec- skin is granular. Some individuals have parietal imens of Pachytriton, the frontal process of the ridges, although they are generally not as prom- premaxilla, which is both long and broad, sep- inent as those of Paramesotriton. In all Cynops, arates the nasals. The hyobranchial apparatus of the relationship of the maxillary bone to the Pachytriton is unique with the stout, bony epi- pterygoid is similar to that of Paramesotriton, branchials flaring dorsolaterally and wrapping with the tip of the maxilla outside and anterior around the neck (O¨ zeti and Wake, 1969). One to the pterygoid and with no contact between specimen of Pachytriton had 13 trunk vertebrae, these elements. The fronto-squamosal arch was whereas all others had 12. Caudo-sacral verte- complete in all individuals, but it is somewhat bral counts were more variable; individuals had attenuated in some individuals. The hyobran- either two or three caudo-sacral vertebrae with chial apparatus of most Cynops was similar to no species-specific pattern emerging. that described for C. pyrrhogaster by O¨ zeti and All Paramesotriton have rough skin and a Wake (1969) with straight epibranchials. How- prominent vertebral ridge, often with a lateral ever, the epibranchials in some C. cyanurus are ridge along each side of their back. The parietal relatively straight, whereas others are moderate- ridges of the skull are prominent as well and ly curved and those of C. wolterstorffi moderately the tail is high and laterally compressed with to strongly curved (O¨ zeti and Wake, 1969). Ex- bony apophyses extending dorsally and ventral- cept for one individual with 14 trunk vertebrae, ly from the caudal vertebrae (Fig. 3). The tips all Cynops had 13 trunk vertebrae and like Par- of the maxillary bones do not contact the pter- amesotriton and Pachytriton, two or three caudo- ygoid as in Pachytriton; they instead lie outside sacral vertebrae. and anterior to the pterygoid, thus forming an There are osteological differences among the angle rather than a straight line. The fronto- species of Cynops, which can be divided into squamosal arch is complete in all specimens ex- four groups based on two main characters: the amined and relatively stout in all species except length of the frontal process of the premaxilla, P. caudopunctatus. The nasals are well separated, and the degree of contact of the nasals (Fig. 4). and there is a long frontal process of the pre- Cynops orphicus is the only species in our sample maxilla. As in Pachytriton, most Paramesotriton with a long frontal process of the premaxilla have 12 trunk vertebrae (two individuals had and with nasals widely separated as in Parame- 11), and the number of caudo-sacral vertebrae sotriton and Pachytriton. In Cynops cyanurus and varies from two to three. Cynops wolterstorffi, the frontal process of the Several morphological characters distinguish premaxilla is long and the nasals almost or nar- P. caudopunctatus from the other species of Par- rowly contact one another. The nasals of Cynops amesotriton in this study (P. guangxiensis, hongkon- orientalis almost or narrowly contact one anoth- gensis, deloustali, and chinensis). Paramesotriton er as well, but the frontal process of the pre- 1002 COPEIA, 2001, NO. 4

Fig. 3. X-rays of the three salamandrid genera used in this study. (A) Paramesotriton hongkongensis (MVZ 230370), (B) Pachytriton labiatus (MVZ 230358), (C) Cynops pyrrhogaster (MVZ 191972), (D) Cynops cyanurus (MVZ 219758). Scale bar under each individual equals one centimeter. CHAN ET AL.—ASIAN NEWT SYSTEMATICS 1003

maxilla is short rather than long (Zhao and Hu, 1988). In Cynops pyrrhogaster and Cynops ensicauda, the frontal process of the premaxilla is short, but the nasals broadly contact one another.

DISCUSSION Morphological evidence for the monophyly of Cynops, Paramesotriton, and Pachytriton has been weak, and authors have had varied inter- pretations. Some studies based mainly on general morphology (e.g., Freytag, 1962) considered the three genera to be close relatives, whereas a character-based analysis of feeding morphology and other features (Wake and O¨ z- eti, 1969) failed to find strong evidence of monophyly. More recently, evidence concerning the relationships has come from molecular studies. Hayashi and Matsui (1989) made the first at- tempt at discerning relationships among salamandrid genera with molecular data, examin- ing allozymic variation at 17 loci in 11 species. Unfortunately, their sample did not include any Pachytriton, so we cannot infer from their clad- ogram the position of that genus. Nonetheless, they recovered a well-supported clade that included Cynops and Paramesotriton, with little resolution. Titus and Larson (1995) used mitochondrial sequences of 12S and 16S mtDNA and morphological characters to examine evolutionary relationships within the family. Their best-supported combined tree reveals strong support for the monophyly of the Cynops, Para- mesotriton, and Pachytriton clade (100 bootstrap) but with no resolution among the species within it. In their topology, these genera are represented as a trichotomy, with one of the Triturus species as the sister taxon to this group. Despite differences in sampling and methodology, col- lectively these studies suggest the three genera are probably each others’ closest relatives. The seven species of Cynops cluster into three well-defined species groups (Zhao and Hu, 1988): the pyrrhogaster group (including pyrrhogaster and ensicauda), the monotypic orientalis group, and the wolterstorffi group (including both wolterstorffi and cyanurus). Zhao and Hu Fig. 4. Dorsal view of the skulls of representatives (1988) considered the pyrrhogaster species group from three of the four Cynops species groups. (A) Cy- to be the most basal of the three based on mor- nops orphicus (MVZ 22465), (B) Cynops pyrrhogaster phological and behavioral characters. In the (MVZ 185212), and (C) Cynops wolterstorffi (AMNH 5455). molecular portion of this study, we included representatives from the pyrrhogaster and wolterstorffi species groups. Our topology supports the paraphyly of this genus and suggests that C. pyrrhogaster may be more closely related to Para- mesotriton and Pachytriton than to C. cyanurus. 1004 COPEIA, 2001, NO. 4

Wake and O¨ zeti (1969) proposed that Cynops that of other salamandrid species (O¨ zeti and had a more ‘‘generalized’’ morphology, and Wake, 1969). one might expect that more specialized or de- The support for a Paramesotriton clade includ- rived morphologies would evolve from within a ing guangxiensis, deloustali, and hongkongensis is group such as this one. relatively high (96 bootstrap, 7 decay), but sup- We cannot say with certainty that the two port for a clade including the fourth Parameso- members of the genus Cynops are a paraphyletic triton, P. caudopunctatus, is low (31 bootstrap, 1 assemblage from the molecular data alone be- decay); thus, we cannot say with certainty that cause tree comparison tests do not reject the Paramesotriton is monophyletic. In contrast to possibility that this genus is monophyletic. How- other species of Paramesotriton, P. caudopunctatus ever, distinct cranial morphologies among Cy- is easily diagnosed on the basis of external mor- nops species groups underscore the possibility phology and coloration. Thus, we are confident that Cynops may be paraphyletic and that both that our single sample from a commercial spec- Pachytriton and Paramesotriton may have evolved imen was correctly identified (T. Titus, pers. from within Cynops as currently recognized. Cy- comm.). Nonetheless, given the observed ge- nops can be divided into four groups on the ba- netic distances and the interesting position of sis of two cranial characters. Broad versus nar- this taxon on our tree future systematic studies row contact of the nasals and long versus short should confirm our findings. Although not an- frontal process of the premaxilla together dis- alyzed in a cladistic framework, morphological tinguish the pyrrhogaster group, wolterstorffi data are in agreement with our topology, with group, orientalis group, and C. orphicus. The first P. guangxiensis, deloustali, and hongkongensis three species groups were previously defined by more similar to one another than to P. caudo- Zhao and Hu (1988), and we suggest that based punctatus. Like other Paramesotriton, P. caudo- on cranial variation alone, C. orphicus could be punctatus has granular skin, a Paramesotriton-like considered a separate monotypic orphicus skull (e.g., complete fronto-squamosal arch, group. We were unable to examine any speci- maxillaries at an angle to pterygoid, nasals sep- men of C. chenggongensis, but assignment of this arated), and prominent vertebral, lateral, and parietal ridges, which suggest that it is allied species to one of these morphological groups with other Paramesotriton. However, the skull of should be possible with the examination of P. caudopunctatus is long and narrow, and the these characters. epibranchials of the hyobranchial apparatus are Although not analyzed in a cladistic frame- flared like those of Pachytriton. Thus, the mor- work, the data presented by Zhao and Hu phological data exhibit a combination of Para- (1988) showed that Paramesotriton and Pachytri- mesotriton-like and Pachytriton-like characters. ton share many derived character states for os- This may be a result of the basal position of the teological and hyoid apparatus characters rela- species within the genus or it might also reflect tive to Cynops. Our topology supports this rela- the ecology of this species, which is stream tionship; the Paramesotriton and Pachytriton in dwelling compared to the other members of the our study represent a monophyletic assemblage genus which are predominantly pond dwellers relatively well supported by our analyses (boot- (Bischoff and Bo¨hme, 1980). Some behavioral strap values of 57–62%). characteristics of P. caudopunctatus are also The Pachytriton clade in our topology is well Pachytriton-like; their courtship pattern resem- supported (100 bootstrap, 16 decay). We re- bles that of other Paramesotriton, whereas their corded relatively large amounts of divergence egg-laying and feeding behaviors are similar to (K2p of 0.3–3.5%) within what is currently con- those of Pachytriton (Sparreboom, 1983; Reha´k, sidered to be a single species, P. labiatus, and 1984). our results suggest that more than a single spe- Relationships within Paramesotriton have not cies may be represented. Thiesmeier and Horn- been examined in a detailed cladistic frame- berg (1997) discussed two undescribed species, work. Several species of Paramesotriton have been and a complete revision of this genus is in or- described only recently (Liu and Hu, 1973; der. Pachytriton appears to have a more derived Huang et al., 1983; Wen, 1989), and more forms morphology, possibly resulting from adaptation likely will become known with further collection to an almost completely aquatic life in fast-mov- efforts in southeast Asia. In the original descrip- ing streams (Pope and Boring, 1940); they have tions, authors noted gross morphological simi- smooth skin, a narrow skull, and an uncom- larities between pairs of taxa and suggested pressed tail. Additionally, the hyobranchial ap- their close relationship (e.g., Wen, 1989). A paratus of Pachytriton is highly specialized for phenetic comparison of the morphology of five aquatic ‘‘gape and suck’’ feeding, more than of the six species showed that P. fuzhongensis and CHAN ET AL.—ASIAN NEWT SYSTEMATICS 1005 chinensis are very similar (Pang et al., 1992). Cynops pyrrhogaster.—MVZ 22656–22657, In addition, Pang et al. concluded that P. 185212, 191972–191973, 198773–198774. guangxiensis and hongkongensis were close rela- Cynops wolterstorffi.—AMNH 5453–5455, CAS tives and basal to fuzhongensis, chinensis, and cau- 6664, 54852, 54908–54909, MCZ 7170, 7173, dopunctatus. Analysis of our molecular data finds 8154–8157, 8751, 9621. P. caudopunctatus to be the most basal of the spe- Pachytriton brevipes.—MVZ 204297, 204299– cies of Paramesotriton. Tissue samples of chinensis 204300, 206174. and fuzhongensis were not available to us; how- Pachytriton labiatus.—MVZ 230147, 230354– ever, external, cranial, and hyobranchial char- 230359, MVZ 230720. acteristics show these three species to be more Pachytriton sp. nov. B.—MVZ 206173. similar to guangxiensis, deloustali, and hongkon- Paramesotriton caudopunctatus.—MVZ 204295– gensis than to caudopunctatus. Based on morpho- 204296. logical evidence for all Paramesotriton examined Paramesotriton chinensis.—CAS 6380, MVZ and molecular evidence for four or the six spe- 230360. cies of Paramesotriton, we suggest that chinensis Paramesotriton deloustali.—MVZ 206310–206312, and fuzhongensis are more recently diverged 222122, 223627–223629, 225135, 226269– than P. caudopunctatus. 226270. Our molecular data show that C. cyanurus is Paramesotriton fuzhongensis.—MVZ 230361– deeply differentiated from all other samples 230364. (lowest K2p is 13.2% to C. pyrrhogaster) and that Paramesotriton guangxiensis.—MVZ 220905– it is basal to the remainder of the Asian taxa 220906. examined here. We were not able to secure mo- Paramesotriton hongkongensis.—MVZ 110576– lecular sequences for C. wolterstorffi (which is 110578, 184859–184860, 198499, 198697– likely extinct, E. Zhao and D. Yang, pers. 198698, 198700–198704, 219766, 230370. comm.), but morphological features of the species ally it with C. cyanurus; both species are ACKNOWLEDGMENTS highland forms that live in the same general region. Should the paraphyly of Cynops be con- We thank T. Titus, T. Papenfuss, and R. Macey firmed, an appropriate taxonomic resolution for tissues and J. Vindum (CAS), J. Rosado would be to recognize the genus Hypselotriton (MCZ), L. Ford and D. Frost (AMNH), and A. (Wolterstorff, 1934) as a valid taxon containing Resatar and H. Voris (FMNH) for loaning us at least cyanurus and wolterstorffi. specimens in their care. We also thank K. Klitz Although this study contributes to our under- for preparing Figure 4. Three anonymous re- standing of the relationships among species in viewers commented on the original version of this Asian salamander radiation, there are still this manuscript and improved the final product. many questions to be addressed. In general, This study was initiated during a research trip studies of these taxa have been plagued by re- funded by the National Geographic Society, and duced number of characters available because laboratory investigations were funded by a Na- of conserved morphology in this group or lim- tional Science Foundation Minority Postdoctor- ited availability of multiple, reliable samples of al Fellowship to KZ and by the Museum of Ver- all taxa. Molecular studies have clarified some tebrate Zoology. We thank the staff and students of the relationships within this group (Hayashi of the MVZ Evolutionary Genetics Laboratory and Matsui, 1989; Titus and Larson, 1995) but and Cornell’s Evolutionary Genetics Core Facil- with only limited success. Given these limita- ity for their help with molecular data collection. tions and the apparent paraphyletic nature of some currently recognized genera, we predict LITERATURE CITED that future studies combining morphology and independent mitochondrial and nuclear mark- BISCHOFF,W.,AND W. BO¨ HME. 1980. Zur Kenntnis von ers will be most successful in elucidating rela- Paramesotriton caudopunctatus (Hu, Djao and Liu, tionships within this clade. 1973) n. comb. (Amphibia: Caudata: Salamandri- dae). Salamandra 16:137–148. CACCONE, A., M. C. MILINKOVITCH,V.SBORDONI, AND MORPHOLOGICAL MATERIAL EXAMINED J. R. POWELL. 1997. Mitochondrial DNA rates and biogeography in European newts (genus Euproctus). Cynops cyanurus. —MVZ 219757–219760. Syst. Biol. 46:126–144. Cynops ensicauda.—CAS 22598, MVZ 57903– CHANG, T.-K., AND A. M. BORING. 1935. Studies in var- 57904. iation among the Chinese amphibia. I. Salamandri- Cynops orientalis.—MVZ 204305–204308. dae. Peking Nat. Hist. Bull. 9:327–361. Cynops orphicus.—MVZ 22460, 22468, 22472. FREYTAG, G. E. 1962. U¨ ber die Wassermolchgattungen 1006 COPEIA, 2001, NO. 4

Paramesotriton Chang 1935, Pingia Chang 1935 und Xenopus laevis mitochondrial DNA genome. J. Biol. Hypselotriton Wolterstorff 1934 (Salamandridae). Chem. 260:9759–9774. Mitt. Zool. Mus. Berl. 38:451–459. SPARREBOOM, M. 1983. On the sexual behaviour of ———, AND H-G. PETZOLD. 1961. Beitra¨ge zur O¨ ko- Paramesotriton caudopunctatus (Liu and Hu) (Am- logie und Anatomie von Paramesotriton deloustali phibia: Caudata: Salamandridae). Amphib.-Reptilia (Bourret 1934) nebst Bemerkungen u¨ber Pachytri- 4:25–33. ton brevipes (Sauvage 1875). Sitz. Ges. Naturf. Freun- TEMPLETON, A. R. 1983. Phylogenetic inference from de Berlin (N.F.) 1:143–162. restriction endonuclease cleavage site maps with particular reference to the humans and apes. Evo- HASEGAWA, M., H. KISHINO, AND T. YANO. 1985. Dating lution 37:221–244. of the human-ape splitting by a molecular clock of THIESMEIER, B., AND C. HORNBERG. 1997. Paarung, mitochondrial DNA. J. Mol. Evol. 22:160–174. Fortpﬂanzung and Larvalentwicklung von Pachytri- HAYASHI,T.,AND M. MATSUI. 1989. Preliminary study ton sp. (Pachytriton A) nebst Bemerkungen zur Tax- of phylogeny in the family Salamandridae: allozyme onomie der Gattung. Salamandra 33:97–110. data, p. 157–167. In: Current herpetology in East THORN, R., AND J. RAFFAE¨LLI. 2001. Les Salamandres Asia. M. Matsui, T. Hikida, and R. C. Goris (eds.). de L’Ancien Monde. Societe Nouvelle des Editions Herpetological Society of Japan, Kyoto. Boubee, Paris. HUANG, Z., Z. TANG, AND Z. TANG. 1983. A new species TITUS,T.A.,AND A. LARSON. 1995. A molecular phy- of the genus Trituroides from Guangxi, China. Acta logenetic perspective on the evolutionary radiation Herpetol. Sinica 2:37–39. of the salamander family Salamandridae. Syst. Biol. KISHINO, H., AND M. HASEGAWA. 1989. Evaluation of 44:125–151. the maximum likelihood estimate of the evolution- WAKE, D. B., AND N. O¨ ZETI. 1969. Evolutionary rela- ary tree topologies from DNA sequence data, and tionships in the family Salamandridae. Copeia the branching order in Hominoidea. J. Mol. Evol. 1969:124–137. 29:170–179. WEN, Y. 1989. A new species of the genus Paramesotri- LEVITON, A. E., R. H. GIBBS JR., E. HEAL, AND C. E. ton (Amphibia: Caudata) from Guangxi and a com- DAWSON. 1985. Standards in herpetology and ich- parison with P. guangxiensis. Chinese Herpetol. Res. thyology. Part I. Standard symbolic codes for insti- 2:15–20. WOLTERSTORFF, W. 1934. U¨ ber die Gattung Hypselotri- tutional resource collections in herpetology and ton. Zool. Anz. 108:257–263. ichthyology. Copeia 1985:802–832. ZHAO, E. 1999. Distribution patterns of amphibians in LIU, C., AND S. HU. 1973. Description of Trituroides temperate eastern Asia, p. 421–443. In: Patterns of caudopunctatus. In: A survey of amphibians and rep- distribution of amphibians, a global perspective. W. tiles in Kweichow Province, including a herpeto- E. Duellman (ed.). John Hopkins Univ. Press, Bal- faunal analysis. S. Hu, E. Dhao, and C. Liu (eds.). timore, MD. Acta Zool. Sinica 19:176–178. ZHAO, E., AND K. ADLER. 1993. Herpetology of China. O¨ ZETI, N., AND D. B. WAKE. 1969. The morphology SSAR. Contributions in Herpetology, Ithaca, NY. and evolution of the tongue and associated struc- ZHAO, E., AND Q. HU. 1988. The classiﬁcation and evo- tures in salamanders and newts (family Salaman- lution of the Chinese Salamandridae, p. 12–26. In: dridae). Copeia 1969:91–123. Studies on Chinese salamanders. E. Zhao, Q. Hu, PANG, J., Y. JIANG, AND Q. HU. 1992. Systematic study Y. Jiang, and Y. Yang (eds.). Contributions in Her- on Chinese newt, genus Paramesotriton (Caudata: petology, SSAR, Ithaca, NY. Salamandridae), p. 89–100. In: Collected papers in herpetology. Y. Jiang (ed.). Sichuan Publishing MUSEUM OF VERTEBRATE ZOOLOGY, 3101 VALLEY House of Science and Technology, Chengdu, Peo- LIFE SCIENCES BUILDING,UNIVERSITY OF CALI- ple’s Republic of China. FORNIA,BERKELEY,CALIFORNIA 94720–3160. POPE,C.H.,AND A. M. BORING. 1940. A survey of PRESENT ADDRESS: (LMC, KRZ) DEPARTMENT Chinese Amphibia. Peking Nat. Hist. Bull. 15:13– OF ECOLOGY AND EVOLUTIONARY BIOLOGY, 86. CORNELL UNIVERSITY, E145 CORSON HALL, REHA´ K, I. 1984. A study on Paramesotriton deloustali in ITHACA,NEW YORK 14853–2701. E-mail: captivity, with descriptions of the egg, the larva, the (LMC) [email protected]; (KRZ) krz2@ juvenile and the adult (Amphibia: Caudata: Sala- cornell.edu; and (DBW) wakelab@uclink4. mandridae). Vest. Cs. Spolec. Zool. 48:118–131. berkeley.edu. Send reprint requests to LMC. ROE, B. A., D. P. MA,R.K.WILSON, AND J. F. WONG. Submitted: 5 Sept. 2000. Accepted: 18 April 1985. The complete nucleotide sequence of the 2001. Section editor: J. D. McEachran. CHAN ET AL.—ASIAN NEWT SYSTEMATICS 1007 Tam Dao, Vinh PhuTam Prov., Dao, Vietnam Vinh PhuTam Prov., Dao, Vietnam Vinh PhuTam Prov., Dao, Vietnam Vinh PhuTam Prov., Dao, Vietnam Vinh PhuLinming Prov., Vietnam Co., Guang XiLinming Prov., Co., China Guang XiHo Prov., Chung China Valley, Hong KongHong Island, Kong China Island, China commercial specimen commercial specimen Jiaxing Prefecture, Zhejiang Prov.,commercial China specimen commercial specimen Chuxiong, Yunnan Prov., China Chuxiong, Yunnan Prov., China Camp Kilowan, Polk Co.,Autonomous Oregon Prefecture, Sichuan Prov.,Jingdong, China Yunnan Prov., China . The full taxon name, voucher number, and collection locality are listed for ingroup MVZ 222122 MVZ 223628 MVZ 223629 MVZ 223627 UTA 40127 MVZ 220905 MVZ 220906 MVZ 230366 MVZ 230369 no voucher MVZ 230720 CAS 194298 MVZ 230147 KU 219723 MVZ 219757 MVZ 219758 KU 219725 CAS 195126 MVZ 219761 NALYSIS A OLECULAR M EQUENCED FOR S and outgroup taxa. Institutional abbreviations are as listed in Leviton et al. (1985). ALAMANDRIDAE S Paramesotriton deloustali Paramesotriton deloustali Paramesotriton deloustali Paramesotriton deloustali Paramesotriton deloustali Paramesotriton guangxiensis Paramesotriton guangxiensis Paramesotriton hongkongensis Paramesotriton hongkongensis Paramesotriton caudopunctatus Pachytriton labiatus Pachytriton labiatus Pachytriton labiatus Cynops pyrrhogaster Cynops cyanurus Cynops cyanurus Taricha granulosa Tylototriton taliangensis Tylototriton verrucosus PECIMENS OF S LL 1. A PPENDIX A AbbreviationPdel1 Pdel2 Pdel3 Pdel4 Pdel5 Pgua1 Pgua2 Phon4 Phon7 Pcau1 TaxonPlab1 Plab2 Plab3 Cpyr1 Ccya1 Ccya2 Tgra1 Ttal1 Tver1 Specimen # Locality 1008 COPEIA, 2001, NO. 4 2 0.128 0.109 0.133 0.135 0.133 0.129 0.133 74 91 23 — 117 113 159 152 144 0.114 0.128 0.130 0.158 0.154 0.113 69 94 69 66 71 — 115 113 148 149 139 0.101 0.103 0.118 0.113 0.001 88 69 80 80 87 — 101 115 109 153 146 140 . Above diagonal: K2p corrected divergences. Below NALYSES A 1 0.104 0.106 0.121 0.115 89 71 84 82 89 — 105 118 113 153 151 141 OLECULAR M 0.081 0.083 0.002 82 67 85 76 76 86 67 — 100 110 106 149 143 141 NCLUDED IN THE I NDIVIDUALS 1 0.087 0.088 85 69 87 77 79 88 70 — 102 112 108 148 144 142 17 I diagonal: absolute number of changes (bp, total length 786). 0.002 83 93 62 74 80 81 84 53 51 63 — 114 104 147 147 145 IVERGENCE AMONG THE D 1 12345678 82 92 61 73 79 80 83 52 50 62 — 114 104 146 146 144 EQUENCE S AIRWISE Cpyr1 Ccya2 Tvul1 Tcar1 Tgra1 Ttal1 Tver1 Phon7 Pcau1 Plab1 Plab2 Plab3 Pdel2 Pdel3 Pgua1 Pgua2 Phon4 2. P 6 7 8 9 1 2 3 4 5 11 12 13 14 15 16 17 10 PPENDIX A CHAN ET AL.—ASIAN NEWT SYSTEMATICS 1009 — 0.231 0.246 0.210 0.231 0.195 0.095 0.248 0.245 0.256 0.256 0.256 0.271 0.273 0.270 0.267 0.248 — 0.248 0.279 0.232 0.223 0.233 0.263 0.260 0.268 0.269 0.273 0.278 0.281 0.277 0.275 0.264 59 — 0.249 0.277 0.189 0.198 0.267 0.255 0.267 0.271 0.257 0.263 0.266 0.278 0.273 0.262 136 115 — 0.221 0.218 0.188 0.219 0.214 0.216 0.229 0.232 0.221 0.221 0.233 0.229 0.219 105 116 116 — 0.209 0.245 0.235 0.219 0.226 0.232 0.243 0.248 0.248 0.246 0.241 0.232 107 101 119 107 XTENDED — 2. E 0.132 0.175 0.155 0.149 0.146 0.156 0.167 0.169 0.189 0.186 0.177 124 113 156 157 139 PPENDIX A 0.144 0.108 0.115 0.110 0.125 0.141 0.143 0.148 0.142 0.142 82 — 111 116 149 144 133 0.147 0.114 0.035 0.031 0.148 0.150 0.159 0.155 0.147 77 94 — 122 118 147 153 143 9 1011121314151617 0.130 0.107 0.003 0.137 0.139 0.137 0.131 0.133 69 87 20 — 115 114 154 149 144 Cpyr1 Ccya2 Tvul1 Tcar1 Tgra1 Ttal1 Tver1 Phon7 Pcau1 Plab1 Plab2 Plab3 Pdel2 Pdel3 Pgua1 Pgua2 Phon4 6 7 8 9 1 2 3 4 5 11 12 13 14 15 16 17 10