Spiny (Paini) illuminate the history of the Himalayan region and Southeast Asia

Jing Chea,b, Wei-Wei Zhoua, Jian-Sheng Hua,c,FangYana, Theodore J. Papenfussb,DavidB.Wakeb,1, and Ya-Ping Zhanga,d,1

aState Key Laboratory of Genetic Resources and Evolution, Kunming Institute of Zoology, Chinese Academy of Sciences, Kunming 650223, People’s Republic of ; bDepartment of Integrative Biology, Museum of Vertebrate Zoology, University of California, Berkeley, CA 94720; and cCollege of Life Sciences and dLaboratory for Conservation and Utilization of Bioresources, Yunnan University, Kunming 650091, People’s Republic of China

Contributed by David B. Wake, June 15, 2010 (sent for review May 6, 2010) Asian frogs of the tribe Paini (Anura: ) range across frogs belonging to the tribe Paini, ranoid frogs endemic to Asia— several first-order tectono-morphological domains of the Cenozoic is not well understood, and its classification remains unsettled Indo-Asian collision that include the Tibetan Plateau, the Hima- (9–13). Dicroglossiine ranoids, including Paini, form a lineage layas, and Indochina. We show how the tectonic events induced by thought to have originated in present-day (14). Adults or the Indo-Asian collision affected the regional biota and, in turn, tadpoles of the tribe Paini (stone is the common name in how the geological history of the earth can be viewed from China) live mostly in swift boulder-strewn streams in the mountains a biological perspective. Our analysis of a concatenated dataset of South and Southeast Asia across the , Indochina, and comprising four nuclear gene sequences of Paini revealed two main southern China (15). The specific habitat requirements suggest that radiations, corresponding to the genera (I) and these species are poor overland dispersers and likely were affected (II). Five distinct clades are recognized: Tibetan plateau clade (I-1), by outcomes associated with large-scale crustal deformation (i.e., Himalaya clade (I-2), environs of Himalaya–Tibetan plateau clade >100 km). Their current distribution appears to be closely related to (I-3), South China clade (II-1), and Indochina clade (II-2). This pattern specific tectonomorphological features, including the Tibetan Pla- of relationships highlights the significance of geography in shaping teau, the Himalayas, the Hengduan Mountain Range, and Indo- evolutionary history. Building on our molecular dating, ancestral china, which include three biodiversity hotspots (16). These findings region reconstruction, and distributional patterns, we hypothesize raise questions of when, where, and how these frogs evolved. To a distinct geographic and climatic transition in Asia beginning in the answer these questions, a robust phylogeny is needed, together with Oligocene and intensifying in the Miocene; this stimulated rapid estimates of divergence times of living Paini, which also should diversification of Paini. Vicariance explains species formation provide clues concerning the effects of geological events and enable among major lineages within Nanorana. Dispersal, in contrast, testing geological hypotheses related to tectonic evolution in Asia. plays an important role among Quasipaa, with the southern Chi- Here, we present a comprehensive molecular phylogeny for nese taxa originating from Indochina. Our results support the tec- members of Paini based on four nuclear DNA loci. We used – tonic hypothesis that an uplift in the Himalaya Tibetan plateau several combinations of calibration points to obtain conservative region resulting from crustal thickening and lateral extrusion of timing estimates for the major early diversification events of this EVOLUTION Indochina occurred synchronously during the transition between group, combined with ancestral area reconstruction. As we will Oligocene and Miocene in reaction to the Indo-Asian collision. show, the tribe Paini is a good biological model system for gen- The phylogenetic history of Paini illuminates critical aspects of the erating and testing hypotheses of the spatiotemporal history and timing of geological events responsible for the current geography significance of tectonic events in Asia. of Southeast Asia. Results China | nuclear DNA | phylogeography | Tibet | tribe Paini Sequence Characteristics. DNA sequences were obtained from four nuclear fragments for 36 specimens: 521 bp of tyrosinase, 315 bp of he collision between India and Asia may be the largest active rhodopsin, 1,269 bp of Rag-1, and 1,143 bp of Rag-2 (Table S1). Torogenic event on earth. Following the initial collision in the Almost all PCR and sequence reactions were successful, with the early Cenozoic (ca. 50–55 Mya) or even earlier, about 70 Mya (1), exception of Rag-2 for Hoplobatrachus rugulosus. The alignment of many effects associated with the major tectonic episode continued Rag-2 revealed some gap regions. Fejervarya limnocharis and through the Oligocene and well into the Miocene (2). During these Fejervarya cancrivora both have a 3-bp deletion region. Nanorana times, associated geological processes occurred, ranging from the parkeri has a 15-bp deletion region, and Nanorana pleskei and uplift (thickening) of the Himalaya–Tibetan plateau to lateral ex- Nanorana ventripunctata both have a 9-bp deletion region. The trusion of the continental landmass (2, 3) (Fig. S1). However, final combined DNA alignment is fairly unambiguous; it contains details of spatiotemporal evolution related to creation of the 3,248 nucleotide sites, 885 of which were variable and 554 of which Himalayas and the Tibetan Plateau are still debated (1, 2, 4, 5). This were potentially phylogenetically informative. All sequences were first-order tectonic feature is of great interest to biologists and deposited in GenBank (Table S2). All sequences were translated geologists alike. The associated orogenic and environmental effects into amino acids without stop codons. (e.g., geomorphology, climate change) likely served as a major driving force for modifying and influencing genetic discontinuities, speciation, and evolution of organisms. The geological dynamics Author contributions: J.C. and Y.-P.Z. designed research; J.C., W.-W.Z., J.-S.H., F.Y., T.J.P., and D.B.W. performed research; J.C. and W.-W.Z. analyzed data; and J.C., D.B.W., and Y.-P.Z. both reduced physical barriers to range expansion and formed new wrote the paper. barriers that promoted variance. We expect the evolutionary his- The authors declare no conflict of interest. tory of organisms in this region to parallel geological and related Freely available online through the PNAS open access option. climate development, and even to suggest unrecorded earth Data deposition: The sequences reported in this paper have been deposited in the Gen- historical events. Bank database. For a list of accession numbers, see Table S2. are ideal organisms for inferring geological and 1 – To whom correspondence should be addressed. E-mail: [email protected] or environmental history (6 8). The evolutionary tree, [email protected]. largely based on molecular data, is being revealed at a rapid pace. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. However, the evolutionary history of one major group—spiny 1073/pnas.1008415107/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1008415107 PNAS | August 3, 2010 | vol. 107 | no. 31 | 13765–13770 Downloaded by guest on October 1, 2021 Phylogenetic Analysis. Initially, we intended to analyze each nu- virons of the Himalaya–Tibetan plateau clade (I-3) ≈23 Mya (95% clear gene separately, but the limited information content of each CI: 15–31 Mya). The split between the cladeincluding Himalaya taxa gene prevented us from obtaining robust results. Accordingly, we (I-2) and the Tibetan plateau clade (I-1) is dated at ≈19 Mya. The combined the four nuclear genes and used Bayesian inference (BI) diversification of group II occurred about 24 Mya (95% CI: 16–32 with a partitioned strategy. Bayesian analyses supported the mono- Mya). The second Indochinese clade (II-1-2) diverged from the phyly of the tribe Paini (Fig. 1A). Within the tribe, two major clade of southern Chinese taxa (II-1-1) about 18 Mya. fi clades (I and II) were recovered, with ve well-supported mono- Biogeographic reconstructions. The overall log-likelihood for our phyletic subclades (I-1, Tibetan plateau clade; I-2, Himalaya clade; Lagrange analysis was −74.9 with the use of values of λD = 0.006 I-3, environs of Himalaya–Tibetan plateau clade; II-1, South λ fi and E = 0.003. The most probable ancestral area and the cor- China clade; and II-2, Indochina clade), each de ned largely by responding likelihood estimate values from Lagrange analysis for geography. Within clade I, Tibetan plateau taxa (I-1) and Hima- major nodes are shown in Fig. 2, together with the most probable laya taxa (I-2) are sister lineages, which, together, are sisters to ancestral area and posterior probability values of ancestral area clade I-3. Within clade II, an Indochina group (II-2) is sister to all other species (II-1), including another Indochinese clade (II-2-2) reconstructions from Bayes dispersal-vicariance analysis (DIVA). and the taxa exclusively found in South China (II-2-1). For the Lagrange analyses, although some ambiguity and possible alternative resolutions exist, the highest likelihood estimates were Temporal and Spatial Diversification. Estimation of divergence times. consistent with the result of Bayes-DIVA. We considered it most The estimated divergence times for Paini are visualized in Fig. 2, and likely for the hypotheses here. the divergence time estimates for the in-group nodes (Fig. S2)are The Tribe Paini originated from the Indochinese region (Fig. 2; shown in Table S3. The initial divergences among Paini were about 0.72 and 0.98 for the likelihood estimate value from Lagrange 27 Mya [95% credibility interval (CI): 19–36 Mya]. Within group I, analysis and posterior probability value from Bayes-DIVA, re- the Himalaya–Tibetan plateau taxa (I-1 and I-2) diverged from en- spectively). The most recent common ancestor had likely already

Fig. 1. Phylogenetic hypothesis derived from analysis of the four combined nuclear sequences. (A) The 50% majority rule consensus from a BI analysis. Numbers near branches are Bayesian posterior probabilities (≥90). Vertical bars indicate clade designation with five different colors, corresponding to the designation in Fig. 3. (B) Two patches of spines on the breast. (C) Spines scattered on the ventral region. (D) N. parkeri and its still-water habitat. (E) Q boulengeri and its stream habitat. Branch colors reflect the ancestral state reconstructions, including habitat (stream and still water) and secondary sexual traits of male frogs. The blue bold line shows the still-water life history. The black bold line shows absence of spines on the breast or ventral region, fingers without spines, and no hypertrophied forelimbs during the breeding season for male frogs of those species.

13766 | www.pnas.org/cgi/doi/10.1073/pnas.1008415107 Che et al. Downloaded by guest on October 1, 2021 Fig. 2. Time tree of Paini. The tree topology, derived from BEAST analyses of all 36 samples, is consistent with BI, as shown in Fig. 1A. Branch lengths are proportional to divergence times. Calibration nodes are indicated by filled triangles (N1, N2, N3). Branch colors reflect biogeographic designations (for species at tips) and ancestral area reconstruction (for internal nodes). The upper circles indicate the results of Lagrange analysis. The pie diagrams reflect states with the highest likelihood, which is proportional to the probability at the node. The lower circles indicate the results of Bayes-DIVA. Pie charts represent the marginal probabilities for each alternative ancestral area derived by using DIVA while integrating over tree topologies using MCMC (Markov chain Monte Carlo). Note that the position of changes along branches could not be inferred; our midbranch designation for changes is arbitrary and was chosen for visual clarity. (A) Uplift of general west of China took place at about 27 Mya. (B) Extrusion of Indochina block and uplift of the Himalaya–Tibetan plateau region happened at almost the – same time, 23 24 Mya. (C) Geological movement in southern Tibet and the Himalayan region happened at about 19 Mya. (D) As a whole, Tibet reached at least EVOLUTION an altitude of 3,000 m about 8–9 Mya. B, Borneo; C, China; I, India; T, Tarim Basin.

spread in Indochina and in some western regions of China, al- Our molecular dating and ancestral area reconstruction anal- though support is low (0.49 and 0.35 for the likelihood estimate yses (Fig. 2) indicate that the tribe Paini originated from present- value from Lagrange analysis and posterior probability value from day Indochina. The common ancestors of the tribe likely had Bayes-DIVA, respectively). A subsequent vicariant event sepa- spread into adjacent western China by the start of the Oligocene rated two lineages of ancestral Paini, giving rise to Nanorana in the (≈27 Mya). The clade rapidly diversified through the Oligocene general area of western China and Quasipaa in Indochina. Within into the Miocene (Fig. 3) as Asia transitioned from a zonal pat- Quasipaa, an important dispersal event spread the clade from tern to a monsoon-dominated pattern, which was a significant Indochina to South China and, subsequently, a vicariant event climatic shift (17). At this time, the general western China group isolated clades II-1-1 and II-1-2 within the subgenus Quasipaa (Nanorana) and the Indochina group (Quasipaa) split, based on (clade II-1, 0.50 and 0.51 for the likelihood estimate value from our ancestral area inference. Modern taxa of Nanorana are Lagrange analysis and posterior probability value from Bayes- mostly distributed in West China at relatively higher elevations DIVA, respectively). Within Nanorana, a vicariant event would than the more lowland species of Quasipaa (Fig. 3B). explain the separation of the subgenus Chaparana (I-3) from the Nanorana. Vicariance played a significant role in the diversification subgenera Nanorana and Paa (I-1 + I-2) (0.44 and 0.58 for the of Nanorana. Dramatic crustal deformation induced by the Indo- likelihood estimate value from Lagrange analysis and posterior Asian collision contributed to the complicated landscapes in west- probability value from Bayes-DIVA, respectively). The common ern China. Massive mountains and deeply carved valleys existed, ancestor of the subgenera Paa and Nanorana occurred in the region which acted as barriers to distribution and resulted in species for- of the Himalaya–Tibetan plateau (class I terrain) (0.49 and 0.98 for mation, initially between lineages, that subsequently gave rise to the the likelihood estimate value from Lagrange analysis and posterior subgenera Nanorana, Paa,andChaparana at about 23 Mya (Fig. 2). probability value from Bayes-DIVA, respectively), and the recent Taxa from the Himalayas and Tibetan plateau (class I terrain) be- ancestor of the subgenus Chaparana mainly occurred outside of came isolated from peripheral regions (class II terrain). Later, the Himalaya–Tibetan plateau region (class II terrain). continued uplift of the Himalayan region (2) finally cut off genetic exchange between the Himalayan range and the interior Tibetan Discussion plateau, resulting in the split between the subgenera Nanorana Origin, Diversification, and Trait Evolution of the Tribe Paini. The and Paa at around 19 Mya. Further diversification produced addi- discovery of two major groups with five distinct lineages that tional species within the subgenus Nanorana in the Tibetan plateau have geographic identity (SI Text) was unexpected. Previous region at about 9 Mya, by which time the plateau had likely reached studies have been hampered by a complicated taxonomic history significant altitude, with extreme environments prevailing in the and unresolved relationships (9–11). interior (e.g., on the more or less flat Tibetan plateau with numer-

Che et al. PNAS | August 3, 2010 | vol. 107 | no. 31 | 13767 Downloaded by guest on October 1, 2021 Fig. 3. (A) Sampling sites of all in-group taxa used in this study. The site names are listed in Table S2. Taxa of the subgenus Nanorana (blue circle), subgenus Paa (purple circle), subgenus Chaparana (yellow circle), subgenus Quasipaa (green circle), and subgenus Eripaa (red circle). (B) Highest (▲) and lowest (△) elevations documented for each species were mapped on the simplified phylogenetic tree using parsimony analysis.

ous lakes but with an arid to semiarid climate and a cold atmo- Quasipaa. The southern Chinese clade, subgenus Quasipaa (II-1- sphere) (Fig. 1D). 1) and the Indochinese taxa of the subgenus Quasipaa (II-1-2) Compared with what we consider to be typical members of the were recovered as more closely related than the two Indochinese tribe Paini (i.e., stream-adapted, large body size; Fig. 1 B, C, and taxa clades (II-1-2 and II-2) are to each other. The southern E), the three species on the Tibetan plateau are characterized by Chinese taxa (II-1-1) (Fig. 1A) are deeply nested within those life in still water and relatively small body size (Fig. 1D), as well as Indochinese groups (II-2, II-1-2); this suggests that Indochina by having a reduced (or even absent) tympanum and columella was the place of origin for the southern Chinese taxa. Historical (implying some loss of auditory function) (13). Ancestral state biogeographic analysis supports this hypothesis (Fig. 2). The Quasipaa reconstruction (stream vs. still water) suggests that Paini origi- common ancestor of was likely distributed in present- day Indochina, and dispersal from Indochina to South China nated from a stream-adapted ancestor and then invaded still water likely occurred between 24 and 18 Mya. The split of the genus in response to drastic environment changes on the Tibetan pla- Quasipaa into the current subgenera Quasipaa (II-1) and Eripaa teau. Organ degeneration probably is in response to environ- (II-2), initiated about 24 Mya, is probably linked to orogenic mental extremes (e.g., low temperature, persistent wind, arid to movement of the Truong Son Mountain Range (Fig. 3) along the semiarid climate, hypoxia, food shortage with increasing high al- border between and , although relevant geological titude) (18). For example, low oxygen levels on the Tibetan Pla- studies of this region are few. The Truong Son Mountain Range teau are not suitable for many organisms; a high level of calling forms a distinct divide between the relatively dry Thai-Lao Pla- during courtship is metabolically demanding (19). Nontympanic teau (highland) of central Indochina and the lowlands of Viet- pathways of sound reception and ultrasonic signal system may be nam to the east. The complete isolation between the Indochinese used to compensate for anuran acoustic organ degeneration (20), (II-1-2) and exclusive southern Chinese (II-1-1) taxa occurred but frogs of the subgenus Nanorana need more study. after about 18 Mya. The South China Sea probably opened in the

13768 | www.pnas.org/cgi/doi/10.1073/pnas.1008415107 Che et al. Downloaded by guest on October 1, 2021 Miocene, and this event finally led to a barrier between South Mountain Range and the opening of the South China Sea. Our China and Indochina (21). Ductile movements of the Red River dating is broadly consistent with the geological hypothesis that zone caused the extrusion of Indochina and the offshore rift collision of India with Asia displaced Indochina southeastward between Indochina and South China; these events provide clues relative to South China along the Ailao Shan-Red River shear to explain the present distributional pattern of Quasipaa. The zone, which occurred mostly in the Oligocene and early Miocene presence of endemic species of clade II-1-2 adds a unique (3). The distribution of the genus Quasipaa (clade II), with iso- component to the central part of the Truong Son Mountain lation of the subgenus Quasipaa (South China block) from Eripaa Range, as evidenced also by some Insectivora (22). The close (Indochina block), may reflect this event. The relative importance relationship between these Indochinese taxa (clade II-1-2) and of crustal thickening vs. lateral extrusion attributable to the Indo- the southern Chinese taxa (clade II-1-1) suggests that the di- Asia collision has been intensely debated (4, 28). Our study vergence of clade II-1-2 offers important clues to explain the suggests that the geologically hypothesized uplift (thickening) and origin of the South China fauna. Central Vietnam likely served strike-slip extrusion must have occurred simultaneously to gen- as a refugium for forest specialists during geological movements erate the observed biotic distribution pattern at the Oligocene- and climatic oscillations. These findings highlight the conserva- Miocene boundary (≈23–24 Mya) (Fig. 2B). tion priorities and protected area designation of this region (23). The split of the subgenera Nanorana (clade I-1) and Paa (clade Most species of Paini breed in streams, and, characteristically, I-2) took place about 19 Mya, associated with geological events male frogs have clusters of spines, secondary sexual characters used that separated the Himalayan region and the Tibetan plateau during amplexus, on the chest or more or less scattered on the belly. (Fig. 2C). This result is compatible with other geological hy- Ancestral state reconstruction supports three independent sec- potheses that possible rapid uplift of southern Tibet and the ondary losses of spines on the belly (Fig. 1A). Within the subgenus Himalayas began about 20 Mya (2). The diversification of the Chaparana, this loss happened twice, first in Nanorana taihangnica subgenus Nanorana on the Tibetan plateau took place about and Nanorana quadranus and then later in Nanorana unculuanus. 9 Mya (Fig. 2), which suggests that species formation may be In the subgenus Quasipaa, this loss occurred only once (Quasipaa associated with active orogeny within Tibet; this hypothesis is yei). These four species also lack hypertrophied forelimbs and have compatible with the geological hypothesis that rapid uplift of fingers without spines during the breeding season, suggesting dif- Tibet took place about 8 Mya (2, 29, 30). All three species of the ferences in courtship and amplexus behavior compared with other subgenus Nanorana dwell at altitudes from nearly 3,000 m up to Paini species. Generally, our data support the hypothesis that 4,700 m (Fig. 3B). Members of this subgenus adapted pro- during the breeding season, nuptial spines always are accompanied gressively as the entire Tibetan plateau experienced significant by an enlarged forearm as well as spines on fingers (24). increases in altitude to at least the 3000-m level (Fig. 2D). In short, our data and analyses bring large-scale features of Implications for the Geological Evolution of Asia. Except for the Asian Tertiary history and topography into a framework consis- subgenus Nanorana, most species of the tribe Paini prefer to sit tent with the evolutionary history of the tribe Paini. Multistage on moss-covered rocks near cold mountain streams. This long- uplift events in different regions (Fig. 2) and a strike-slip extru- conserved life history trait, coupled with their present distribution sion event affecting Indochina (Fig. 2B) fit well with our con-

in Asia, makes this species group a potential model for examining ception of frog biogeography. Analyses of other taxa will provide EVOLUTION geological evolution and environmental changes in Asia during tests of our hypotheses. the Tertiary. Since the late Oligocene, uplift of the general western Chinese Materials and Methods area, including the present Himalayas, the Tibetan plateau, and Taxon Sampling. Species of Paini, genera Nanorana and Quasipaa, formed our their environs (class I and II terrain; Fig. S3), led to the high in-group, and species in the genera Hoplobatrachus, ,andFejer- western (class I + II) and low eastern (class III) terrains, which varya were used as out-group taxa. Thirty-six specimens in total were used, have subsequently characterized the topography of China. The among which 29 individuals representing at least 24 species formed the in- associated geological events (Fig. 2A) probably also triggered group. All sampledin-groupspeciesand locations are giveninTable S2and Fig.3. Our sampling included nearly all recognized species representing the five climate change, which, in turn, led to biotic reorganization in Asia. fl major clades recovered (10) using a bottom-up approach (31). We focused on Our molecular dating results re ect this event. The tribe Paini resolving the basal relationship with large numbers of characters using rel- began diversification during the Oligocene, and rapid divergences atively slowly evolving nuclear genes. of the major clades took place in the Miocene (Fig. 2), suggesting geographical and probably paleoclimatic change. This is consis- Extraction, Amplification, and Sequencing. Muscle or liver tissue samples were tent with inferences from mammalian fossil data, paleobotanical stored in 95% or 100% ethanol or were frozen at −80 °C. DNA was extracted evidence, and lithological and sedimentary data (17, 25, 26). The using the standard three-step phenol/chloroform extractions (32). Primers used paleobotanical and lithological evidence shows radical changes in in PCR and sequencing of the four nuclear protein-coding gene fragments are shown in Table S4. Amplification was performed in a 25- to 50-μL volume re- climate and vegetation from the Oligocene to the Miocene fi (transition between the Paleogene and Neogene) (26). Further- action using the same procedures as other studies (10, 33). Puri ed PCR products were directly sequenced with an automated DNA sequencer ABI 3730 more, since Oligocene time, mammalian faunal turnovers have fi (Applied Biosystems, Foster City, CA) in both directions for each species. They occurred in northwestern China, and signi cant reorganization of were submitted for a BLAST search in GenBank and also translated into amino faunas followed during the Miocene (25, 27). acids to ensure the target sequences had been amplified. The split between the subgenera Paa, Nanorana (clades I-1 + I-2), and Chaparana (clade I-3) (Fig. 2B; ≈23 Mya) suggests Data Analysis. All the nuclear sequence data derived from the genes for continuous uplift of the Himalayas and Tibetan plateau system, rhodopsin, tyrosinase, Rag-1, and Rag-2 were easily aligned with Clustal X which caused the separation between the Himalaya–Tibetan 1.81 (34) with default parameters and then verified by eye using Molecular plateau (class I terrain) and its environs (class II terrain). At Evolutionary Genetics Analysis (MEGA) 4 (35). Reconstructions of phyloge- about the same time (≈24 Mya), we infer the split of the sub- nies were primarily performed using BI (36) with a partitioned strategy. genera Quasipaa (clade II-2) and Eripaa (clade II-1) associated Details are given in SI Materials and Methods. with a major tectonic event along the Truong Son Mountain Temporal and Spatial Diversification. Divergence time estimates. We estimated Range, which served to divide the western highlands and eastern nodal ages within the tribe Paini and with 95% CIs from our nuclear DNA se- lowlands along the Ailao Shan-Red River shear zone. The shear quence data using the Bayesian molecular clock method implemented in the zone continued along the coast of Vietnam far to the south (Fig. software BEAST version 1.4.7 (37). We used two indirect estimates of di- S1), and its motion may have created the modern Truong Son vergence time as calibration points (38), which is similar to the method used in

Che et al. PNAS | August 3, 2010 | vol. 107 | no. 31 | 13769 Downloaded by guest on October 1, 2021 another study (39). The two studies (38, 39) included groups having more ex- Paini can be coded as stream dwellers, and only the three species of the tensive fossil records. Furthermore, we also used an additional calibration subgenus Nanorana belong to the still-water dwellers. Overall, parsimony point from biogeographic inference (40). Three calibration points were used in and likelihood gave the same results. total, and details are given in SI Materials and Methods. Ancestral area reconstructions. Biogeographic reconstructions were performed using Bayes-DIVA (41) and Lagrange 2.0 based on the stochastic model of geo- ACKNOWLEDGMENTS. We thank reviewers A. Yin and D. Hillis for constructive graphic range evolution (42, 43). Details are given in SI Materials and Methods. comments. We gratefully acknowledge the assistance and loans from many Ancestral state estimation. Ancestral values of the male secondary character- individuals and institutions (SI Acknowledgments).This work conducted in the laboratory of Y.P.-Z. was supported by grants from the National Basic Research istics were estimated using maximum likelihood and maximum parsimony in Program of China (973 Program, Grant 2007CB411600), the National Natural the program Mesquite (44). Most species can be clearly coded as having the Science Foundation of China, and the Bureau of Science and Technology of fi spines on the breast or ventral region, ngers with spines, and hypertro- Yunnan Province. General support to the laboratory of D.B.W. was received phied forelimbs during the breeding season, except for N. quadranus, from the US National Science Foundation AmphibiaTree of Life program N. taihangnica, N. unculuanus, and Q. yei (13). Furthermore, most species of (Grant EF-0339439).

1. Yin A, Harrison TM (2000) Geologic evolution of the Himalayan–Tibetan orogen. 24. Duellman WE, Trueb L (1986) Biology of Amphibia (McGraw–Hill, New York). Annu Rev Earth Planet Sci 28:211–280. 25. Qiu ZD, Li C (2004) The evolution of mammal fauna in China and the uplift of 2. Harrison TM, Copeland P, Kidd WSF, Yin A (1992) Raising Tibet. Science 255: Qinghai-Xizang Plateau. Sci China Ser D 34:845–854. 1663–1670. 26. Sun X, Wang P (2005) How old is the Asian monsoon system? Palaeobotanical records 3. Tapponnier P, et al. (1990) The Ailao Shan/Red River metamorphic belt: Tertiary left- from China. Palaeogeogr Palaeoclimatol Palaeoecol 222:181–222. lateral shear between Indochina and South China. Nature 343:431–437. 27. Meng J, Mckenna MC (1998) Faunal turnovers of Paleogene mammals from the 4. Tapponnier P, et al. (2001) Oblique stepwise rise and growth of the Tibet plateau. Mongolian Plateau. Nature 394:364–367. – Science 294:1671 1677. 28. Yang Y, Liu M (2009) Crustal thickening and lateral extrusion during the Indo-Asian fi 5. Royden LH, Burch el BC, van der Hilst RD (2008) The geological evolution of the collision: A 3D viscous flow model. Tectonophysics 465:128–135. – Tibetan Plateau. Science 321:1054 1058. 29. Zhisheng A, Kutzbach JE, Prell WL, Porter SC (2001) Evolution of Asian monsoons and 6. Zhang P, et al. (2006) Phylogeny, evolution, and biogeography of Asiatic Salamanders phased uplift of the Himalaya-Tibetan plateau since Late Miocene times. Nature 411: (Hynobiidae). Proc Natl Acad Sci USA 103:7360–7365. 62–66. 7. Heinicke MP, Duellman WE, Hedges SB (2007) Major Caribbean and Central American 30. Molnar P, England P, Martiod J (1993) Mantle dynamics, uplift of the Tibetan Plateau frog faunas originated by ancient oceanic dispersal. Proc Natl Acad Sci USA 104: and the Indian monsoon development. Rev Geophys 34:357–396. 10092–10097. 31. Wiens JJ, Fetzner JW, Jr, Parkinson CL, Reeder TW (2005) Hylid frog phylogeny and 8. Zeisset I, Beebee TJC (2008) Amphibian phylogeography: A model for understanding sampling strategies for speciose clades. Syst Biol 54:778–807. historical aspects of species distributions. Heredity 101:109–119. 32. Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Cloning: A Laboratory Manual 9. Jiang JP, et al. (2005) Phylogenetic relationships of the tribe Paini (Amphibia, Anura, Ranidae) based on partial sequences of mitochondrial 12s and 16s rRNA genes. (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY). Zoolog Sci 22:353–362. 33. Chiari Y, et al. (2004) New evidence for parallel evolution of colour patterns in – 10. Che J, et al. (2009) Phylogeny of the Asian spiny frog tribe Paini (Family Dicroglossidae) Malagasy poison frogs (Mantella). Mol Ecol 13:3763 3774. sensu Dubois. Mol Phylogenet Evol 50:59–73. 34. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG (1997) The 11. Ohler A, Dubois A (2006) Phylogenetic relationships and generic of the CLUSTAL_X windows interface: Flexible strategies for multiple sequence alignment tribe Paini (Amphibia, Anura, Ranidae, Dicroglossinae), with diagnoses of two new aided by quality analysis tools. Nucleic Acids Res 25:4876–4882. genera. Zoosystema 28:769–784. 35. Tamura K, Dudley J, Nei M, Kumar S (2007) MEGA4: Molecular Evolutionary Genetics 12. Frost DR, et al. (2006) The amphibian tree of life. Bull Am Mus Nat Hist 297:1–370. Analysis (MEGA) software version 4.0. Mol Biol Evol 24:1596–1599. 13. Fei L, Hu S, Ye C, Huang Y (2009) Fauna sinica. Amphibia, Anura Ranidae (Science 36. Ronquist FR, Huelsenbeck JP (2003) MRBAYES: Bayesian inference of phylogeny. Press, Beijing, China) Vol 3 (in Chinese). Bioinformatics 19:1572–1574. 14. Bossuyt F, Milinkovitch MC (2001) Amphibians as indicators of early tertiary “out-of- 37. Drummond AJ, Ho SYW, Phillips MJ, Rambaut A (2006) Relaxed phylogenetics and India” dispersal of vertebrates. Science 292:93–95. dating with confidence. PLoS Biol 4:e88. 15. Dubois A (1975) A new sub-genus (Paa) and three new species of the genus . 38. Bossuyt F, Brown RM, Hillis DM, Cannatella DC, Milinkovitch MC (2006) Phylogeny and Remarks on the phylogeny of Ranidae (Amphibia, Anura) (Translated from French). biogeography of a cosmopolitan frog radiation: Late cretaceous diversification ’ – Bulletin du Muséum National d Histoire Naturelle Zoologie 231:1093 1115. resulted in continent-scale endemism in the family ranidae. Syst Biol 55:579–594. 16. Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J (2000) Biodiversity 39. Roelants K, et al. (2007) Global patterns of diversification in the history of modern – hotspots for conservation priorities. Nature 403:853 858. amphibians. Proc Natl Acad Sci USA 104:887–892. fi 17. Guo ZT, et al. (2002) Onset of Asian deserti cation by 22 Myragoinferred from loess 40. Bernor RL, et al. (1987) A consideration of some major topics concerning Old World deposits in China. Nature 416:159–163. Miocene mammalian chronology, migrations and paleogeography. Geobios 20:431– 18. Liao JC, Liu NF (2008) Altitudinal variations of acoustic organs in anurans: A case study 439. from China. Ital J Zool (Modena) 75:125–134. 41. Nylander JAA, Olsson U, Alström P, Sanmartín I (2008) Accounting for phylogenetic 19. Taigen TL, Wells KD (1985) Energetics of vocalization by an anuran amphibian (Hyla uncertainty in biogeography: A Bayesian approach to dispersal-vicariance analysis of versicolor). J Comp Physiol B 155:163–170. the thrushes (Aves: Turdus). Syst Biol 57:257–268. 20. Feng AS, et al. (2006) Ultrasonic communication in frogs. Nature 440:333–336. 21. Hall R (1998) The plate tectonics of Cenozoic SE Asia and the distribution of land and 42. Ree RH, Smith SA (2008) Maximum likelihood inference of geographic range – sea. Biogeography and Geological Evolution of SE Asia, eds Hall R, Holloway JD evolution by dispersal, local extinction, and cladogenesis. Syst Biol 57:4 14. (Backhuys Publishers, Leiden, The Netherlands), pp 99–131. 43. Ree RH, Moore BR, Webb CO, Donoghue MJ, Crandall K (2005) A likelihood 22. Lunde DP, Musser GG, Son NT (2003) A survey of small mammals from Mt. Tay Con framework for inferring the evolution of geographic range on phylogenetic trees. Linh II, Vietnam, with the description of a new species of Chodsigoa (Insectivora: Evolution 59:2299–2311. Soricidae). Mammal Study 28:31–46. 44. Maddison WP, Maddison DR (2004) Mesquite: A modular system for evolutionary 23. Sterling EJ, Hurley MM (2005) Conserving biodiversity in Vietnam: Applying analysis. Version 1.05. Available at: http://mesquiteproject.org. Accessed February 10, biogeography to conservation research. Proc Calif Acad Sci 56:98–118. 2009.

13770 | www.pnas.org/cgi/doi/10.1073/pnas.1008415107 Che et al. Downloaded by guest on October 1, 2021