A new species of the genus Theloderma Tschudi, 1838 (Amphibia: Anura: Rhacophoridae) from Tay Nguyen Plateau, central Vietnam

A new species of small tree frog from a primary montane tropical forest of central Vietnam, Tay Nguyen Plateau, is described based on morphological, molecular, and acoustic evidence. The Golden Bug-Eyed Frog, Theloderma auratum sp. nov., is distinguishable from its congeners and other small rhacophorid species based on a combination of the following morphological attributes: (1) bony ridges on head absent; (2) smooth skin completely lacking calcified warts or asperities; (3) pointed elongated tapering snout; (4) vocal opening in males absent; (5) vomerine teeth absent; (6) males of small body size (SVL 21.8–26.4 mm); (7) head longer than wide; ED/SVL ratio 13%–15%; ESL/SVL ratio 16%–20%; (8) small tympanum (TD/EL ratio 50%–60%) with few tiny tubercles; (9) supratympanic fold absent; (10) ventral surfaces completely smooth; (11) webbing between fingers absent; (12) outer and inner metacarpal tubercles present, supernumerary metacarpal tubercle single, medial, oval in shape; (13) toes half-webbed: I 2–2¼ II 1½–2¾ III 2–3¼ IV 3–1½ V; (14) inner metatarsal tubercle present, oval; outer metatarsal tubercle absent; (15) iris bicolored; (16) dorsal surfaces golden-yellow with sparse golden-orange speckling or reticulations and few small dark-brown spots; (17) lateral sides of head and body with wide dark reddish-brown to black lateral stripes, clearly separated from lighter dorsal coloration by straight contrasting edge; (18) ventral surfaces of body, throat, and chest greyish-blue with indistinct brown confluent blotches; (19) upper eyelids with few (3–5) very small flat reddish superciliary tubercles; (20) limbs dorsally reddish-brown, ventrally brown with small bluish-white speckles. The new species is also distinct from all congeners in 12S rRNA to 16S rRNA mitochondrial DNA fragment sequences (uncorrected genetic distance P>8.9%). Advertisement call and tadpole morphology of the new species are described. Our molecular data showed Theloderma auratum sp. nov. to be a sister species of Th. palliatum from Langbian Plateau in southern Vietnam.


INTRODUCTION
Rhacophoridae is a diverse family of largely arboreal tree frogs consisting of 414 currently recognized species in 18 genera (AmphibiaWeb, 2018;Frost, 2018). Rhacophorids are distributed throughout Sub-Saharan Africa, China, Southern and Southeast Asia, Japan, Taiwan, the Philippines, and the Greater Sunda Islands (Frost, 2018). Species diagnostics and generic allocation within Rhacophoridae can be difficult, and molecular phylogenetic analyses are essential for correct taxonomic decisions Li et al., 2008Li et al., , 2009Orlov et al., 2012;Poyarkov et al., 2015;Rowley et al., 2011;Yu et al., 2009).
Frogs of the genus Theloderma Tschudi, 1838 are highly arboreal small-to large-sized rhacophorid frogs, which were traditionally grouped in one genus based on the presence of dorsal asperities or tuberculate dorsal skin (Liem, 1970;Rowley et al., 2011). However, recent studies have reported on several smooth species of rhacophorids that almost completely or totally lack dorsal asperities but are still assigned to Theloderma based on molecular data and reproductive biology (Nguyen et al., 2014;Orlov et al., 2012;Poyarkov et al., 2015). Thus, the taxonomy of Theloderma is in a current state of flux. Several studies have indicated a sister-clade relationship between Nyctixalus Boulenger, 1882 and Theloderma, with monophyly of the latter being poorly supported or not recovered (Dever et al., 2015;Dever, 2017;Rowley et al., 2011). Thus, Poyarkov et al. (2015) recognized Nyctixalus as a subgenus of Theloderma and assigned the Th. stellatum Taylor, 1962-Th. horridum (Boulenger, 1903) species group to the subgenus Stelladerma Poyarkov, Orlov, Moiseeva, Pawangkhanant, Ruangsuwan, Vassilieva, Galoyan, Nguyen & Gogoleva, 2015. Subsequent studies with larger taxon and gene sampling Sivongxay et al., 2016) supported Nyctixalus as a sister clade with respect to Theloderma as well as monophyly of Theloderma and its subdivision into two highly divergent clades (corresponding to subgenera Theloderma and Stelladerma sensu Poyarkov et al., 2015). Theloderma moloch (Annandale, 1912), an enigmatic taxon whose phylogenetic position and assignment to Theloderma became uncertain due to sequences published by Li et al. (2009), was simultaneously rediscovered in Chinese and Indian regions of the Himalayas and shown to be a member of Theloderma s. str. (Hou et al., 2017;Lalronunga & Lalrinchhana, 2017;Li et al., 2016). In addition,  and Biju et al. (2016) concurrently demonstrated that sequences previously reported as "Th. moloch" by Li et al. (2009) actually corresponded to a distinct lineage of Rhacophoridae, representing new genus Nasutixalus Jiang, Yan, . The taxonomic position of Th. andersoni (Ahl, 1927) is also relatively unclear, with Hou et al. (2017) suggesting that this species might be a member of the genus Raorchestes Biju, Shouche, Dubois, Dutta & Bossuyt, 2010. Taking these reassignments and new species descriptions into consideration, a total of 25 species of Theloderma are recognized and distributed throughout Southeast Asia, from Assam in northeastern India to Myanmar, southern China and Indochina, the Malay Peninsula, and Sumatra and Borneo of the Greater Sunda Islands. New species in the genus continue to be discovered, with 13 described in the last 15 years alone (Frost, 2018).
During recent fieldwork in the montane forests of Tay Nguyen Plateau in central Vietnam, we encountered a small-sized rhacophorid species with a smooth dorsum. This tree frog was identified morphologically as a probable member of the genus Theloderma based on the presence of an intercalary cartilage between the terminal and penultimate phalanges of digits, a distinct tympanum, tips of the digits expanded into large disks bearing circummarginal grooves, absence of vomerine teeth, horizontal pupils, rounded canthus rostralis, terminal phalanx with a Y-shaped distal end, and the skin of the head not co-ossified to the skull (Liem, 1970;McLeod & Ahmad, 2007;Rowley et al., 2011). This species of frog was earlier reported by Orlov & Ananjeva (2007) and Orlov et al. (2012) as Theloderma laeve (Smith, 1924), a taxon described from the Langbian Plateau of southern Vietnam and originally assigned to the genus Philautus due to the presence of smooth skin on the dorsum and the lack of warts, tubercles, and asperities (Smith, 1924). Following examination of the holotype of Philautus laevis Smith, 1924, Poyarkov et al. (2015 demonstrated it to be conspecific to Theloderma bambusicolum Orlov, Poyarkov, Vassilieva, Ananjeva, Nguyen, Nguyen & Geissler, 2012 and confirmed its assignment to the genus Theloderma as Theloderma laeve (Smith, 1924). Consequently, the taxon previously identified as Th. laeve by Orlov et al. (2012) is not assigned to any currently recognized species of Theloderma. Close examination showed that this taxon can be easily distinguished from other known members of the genus Theloderma by a combination of several adult morphological characteristics. Molecular phylogenetic analysis of a 2 518-bp mtDNA fragment further indicated that this taxon is nested within the genus Theloderma with high support values and represents a sister species to Th. palliatum Rowley, Le, Hoang, Dau & Cao, 2011. In the present paper we discuss the phylogenetic position and taxonomic affiliation of this frog and describe it as a new species.  Figure 1). Geographic coordinates and altitude were obtained using a Garmin GPSMAP 60CSx GPS receiver (USA) and recorded in datum WGS 84. Specimens were euthanized by 20% benzocaine and tissue samples (femoral muscles) for genetic analysis were collected and stored in 96% ethanol prior to specimen preservation. Specimens were subsequently preserved in 70% ethanol and deposited in the herpetological collection of the Zoological Museum of Moscow State University (ZMMU) in Moscow, Russia. Comparative materials examined are stored in the herpetological collections of ZMMU and the Zoological Institute R.A.S. (ZISP) in St. Petersburg, Russia. A list of examined specimens is given in Appendix I.

Morphological description
Specimens of Theloderma sp. were photographed in life and after preservation. Sex was determined by direct observation of calls and by examination of the nuptial pads in collected males. Measurements were taken using a digital caliper under a light dissecting microscope to the nearest 0.01 mm, subsequently rounded to 0.1 mm. Statistical analyses were performed with Statistica 8.0 (StatSoft, Inc., 2007).

palliatum in Vietnam
Star denotes type locality of new species; dot in circle denotes type locality for previously described species.

DNA isolation, PCR, and sequencing
Total genomic DNA was extracted from the ethanol-preserved muscle tissues using standard phenol-chloroform extraction (Hillis et al., 1996). Total DNA concentration was estimated in 1 µL using a NanoDrop 2000 (Thermo Scientific, USA), and consequently adjusted to 100 ng DNA/µL.
We amplified the mtDNA fragment consisting of partial sequences of 12S rRNA, tRNA val , and 16S rRNA mtDNA genes. These markers were chosen due to their successful use in studies of Rhacophorid taxonomic diversity (Bain et al., 2009;Dever et al., 2015;Li et al., 2008Li et al., , 2009Li et al., , 2013Meegaskumbura et al., 2015;Nguyen et al., 2014Nguyen et al., , 2015Poyarkov et al., 2015;Rowley et al., 2011;Wilkinson & Drewes, 2000;Wilkinson et al., 2002;Yu et al., 2007Yu et al., , 2008and references therein). PCR was performed in 20-µL reactions using 50 ng of genomic DNA, 10 nmol of each primer, 15 nmol of each dNTP, 50 nmol additional MgCl 2 , Taq PCR buffer (10 mmol/L Tris-HCl, pH 8.3, 50 mmol/L KCl, 1.1 mmol/L MgCl 2 , and 0.01% gelatin), and 1 U of Taq DNA polymerase. Primers used in PCR and sequencing were obtained from previous studies (Hedges, 1994;Li et al., 2008Li et al., , 2009Wilkinson et al., 2002) and are summarized in Table 1. The PCR conditions for amplification of the 12S rRNA and 16S rRNA gene fragments followed those described in Poyarkov et al. (2015). PCR products were visualized by agarose electrophoresis in the presence of ethidium bromide and consequently purified using 2 µL from a 1:4 dilution of ExoSapIt (Amersham, UK) per 5 µL of PCR product prior to cycle sequencing. Sequence data collection and visualization were performed on an ABI 3730xl automated sequencer (Applied Biosystems, USA) in Evrogen Inc., Moscow. The obtained sequences were aligned and deposited in GenBank under the accession numbers MG917762-MG917772 (Table 2).

Phylogenetic analyses
Sequences of the 12S rRNA, tRNA val , and 16S rRNA mtDNA fragments from 135 Rhacophoridae specimens, including 122 representatives of Theloderma (ca. 27 species), and 13 sequences of outgroup members of Rhacophoridae (genera Rhacophorus, Nasutixalus, and Nyctixalus) were included in the final alignment with a total length of 2 518 bp.
Details on voucher specimens and GenBank accession Nos. used in phylogenetic analyses are summarized in Table 2. Nucleotide sequences were initially aligned using ClustalX 1.81 The dataset was divided into three partitions, 12S rRNA, tRNA val , and 16S rRNA, with the optimal evolutionary models then estimated using MODELTEST v.3.06 (Posada & Crandall, 1998). For the 12S and 16S rRNA partitions, the best-fitting model according to the Akaike information criterion (AIC) was the HKY+I+G model; whereas for the tRNA val partition, the Kimura 2-parameter model (+G+I) was selected as the one of best fit. Mean uncorrected genetic distances (P-distances) between sequences were determined with MEGA 7.0 (Kumar et al., 2016).
Matrilineal genealogy was inferred using Bayesian inference (BI) and maximum likelihood (ML) algorithms. The BI analyses were conducted in MrBayes 3.1.2 (Huelsenbeck & Ronquist, 2001;Ronquist & Huelsenbeck, 2003). Metropolis-coupled Markov chain Monte Carlo (MCMCMC) analyses were run with one cold chain and three heated chains for ten million generations and sampled every 1 000 generations. Five independent MCMCMC runs were performed and 1 000 trees were discarded as burn-in. Confidence in tree topology was tested by posterior probability (PP) for Bayesian inference (BI) trees (Huelsenbeck & Ronquist, 2001). Nodes with posterior probability values over 0.95 were a priori regarded as sufficiently resolved, whereas those between 0.95 and 0.90 were regarded as tendencies, and those below 0.90 were considered as unsupported.
The ML analyses were conducted using Treefinder (Jobb et al., 2004) and confidence in node topology was tested by non-parametric bootstrapping with 1 000 replicates (ML BS, see Felsenstein, 1985). We a priori regarded tree nodes with bootstrap (ML BS) values of 70% or greater and Bayesian posterior probabilities (BI PP) values over 0.95 as sufficiently resolved; ML BS values between 70% and 50% (BI PP between 0.95 and 0.90) were treated as tendencies and nodes with ML BS values below 50% (BI PP below 0.90) were regarded as unresolved (Felsenstein, 2004;Huelsenbeck & Hillis, 1993).

Acoustic analyses
Advertisement calls of the Theloderma sp. were recorded in habitat in Kon Chu Rang Nature Reserve, Gia Lai Province, Tay Nguyen Plateau, Vietnam (N14°30 19.7 , E108°32 30.0 ; 950 m a.s.l.) on 26 May 2016 from 0224 h to 0300 h at an ambient temperature of 21.5°C (temperature was measured at the calling site immediately after recording with a digital thermometer KTJ TA218A Digital LCD Thermometer-Hydrometer) using a portable digital audio recorder Zoom h5 (ZOOM Corporation, Tokyo, Japan) in stereo mode with 48 kHz sampling frequency and 16-bit precision. In total, two recordings from two males (ZMMU A-5828 and A-5829) were made.
Calls were analyzed using Avisoft SASLab Pro software v.5.2.05 (Avisoft Bioacoustics, Germany). Before analysis, we reduced background noise using low-pass (up to 500 Hz) and high-pass filters (down to 7 kHz). All parameters were measured in the spectrogram window of Avisoft. Spectrograms were created under Hamming window, FFT-length 1 024 points, frame 75%, and overlap 87.5%. For figure spectrograms, we lowered the sampling rate to 22.05 kHz. Figure spectrograms were created under Hamming window, FFT-length 512 points, frame 100%, and overlap 50%. In total, we measured 214 Theloderma sp. calls.
We measured four temporal parameters (duration of each call and series and interval between successive calls and series) and three frequency parameters (initial and final fundamental frequency and frequency of maximum amplitude). We also calculated the call repetition rate (calls/s) by counting the number of calls within each series, minus one, and dividing that number by the series duration. All numeral parameters are given as means±SE, and the minimum and maximum values are given in parentheses (min-max).

Phylogenetic analyses Sequences and statistics
The final alignment contained 2 518 aligned characters, with 1 090 conserved sites and 1 395 variable sites, of which 1 170 were found to be parsimony-informative. The transition-transversion bias (R) was 2.372 (data given for ingroup only).

Position of Theloderma sp. in matrilineal genealogy
The phylogenetic analyses results are presented in Figure 2. The ML and BI phylogenetic analyses resulted in essentially similar topologies. In general, the BI cladogram topology was consistent with results reported in previous work Sivongxay et al., 2016), suggesting monophyly of the clade joining Nyctixalus and Theloderma (node support values 1.0/100, hereafter given for BI PP/ML BS, respectively) and monophyly of Theloderma, though only with moderate support (0.90/85).   (Figure 2). Clade II was further subdivided into subclade II-A (Sundaland, Indochina, eastern Himalayas, southern China; 1.0/96) and subclade II-B (Indochinese species; 1.0/85).
Phylogenetic relationships within subclade II-B were essentially unresolved (Figure 2), though the following species groups were supported: Th.
The Theloderma sp. from Tay Nguyen Plateau was recovered as a sister species of Th. palliatum, which inhabits the montane forests of Langbian Plateau, with strong node support (1.0/100) ( Figure 2); however, no clear structuring was observed within the Theloderma sp. clade.

Sequence divergence
The uncorrected P-distances among and within the studied 2 518-bp mtDNA fragments for the examined Theloderma species are shown in Table 3 (data given for the ingroup only). The interspecific uncorrected genetic P-distances between the Theloderma sp. from Tay Nguyen Plateau and other congeners varied from 8.9% (between Theloderma sp. and sister species Th. palliatum) to 17.9% (between Theloderma sp. and Th. laeve) (Table 3). This degree of pairwise divergence was high, notably greater than the genetic divergence thresholds representing species level differentiation in frogs (Vences et al., 2005a(Vences et al., , 2005bVieites et al., 2009).

Taxonomy
Based on the phylogenetic analyses of the 2 518-bp length 12S rRNA and 16S rRNA mtDNA fragment sequences, the examined specimens of Theloderma sp. from Tay Nguyen Plateau in central Vietnam represented a highly divergent mtDNA lineage, clearly distinct from all other Theloderma species for which comparable mtDNA sequences are available (Figure 2), and a sister species of Th. palliatum from Langbian Plateau of southern Vietnam. The observed differences in mtDNA sequences were congruent with evidence from the diagnostic morphological characters (see "Comparisons"). These results support our hypothesis that the small-sized "smooth" Theloderma sp. from Tay Nguyen Plateau represents a previously unknown species, which we describe herein.
The new species is also markedly distinct from all congeners for which comparable sequences of the 2 518-bp length 12S rRNA to 16S rRNA mitochondrial DNA fragments are available (uncorrected genetic distance P>8.9%).

Description of holotype:
Small-sized rhacophorid frog specimen in a good state of preservation; body slender, dorsoventrally compressed (Figure 3). Skin on ventral surface of left femur of holotype medially dissected 5 mm, with femoral muscles (partial) removed for molecular genetic analyses. Measurements of holotype are given in Table 4 (in mm).
Skin texture and skin glands: Dorsal skin smooth, with numerous small flat tubercles irregularly scattered on dorsal surfaces of head and body, forming weak reticulate pattern in sacral area ( Figure 3A), lateral surfaces of head and body completely smooth; tympanum with few tiny evenly scattered tubercles; upper eyelids with few (3-5) very small flat reddish superciliary tubercles ( Figure 3D); calcified asperities and warts on dorsum absent; area above insertion of forelimbs, and lateral sides of belly smooth; supratympanic and dorsolateral folds absent; dorsal surface of limbs weakly shagreened, ventral sides of limbs smooth; throat and chest smooth, belly and ventral surface of thigh smooth; dermal appendage at vent, dermal fringes and dermal tibiotarsal projections absent.

Color of holotype in life:
Background of dorsal surface of head, body, forearms, thighs, and shanks uniform golden-yellow with numerous tiny flat golden-orange tubercles on dorsal surfaces and supraciliary tubercles, getting somewhat denser posteriorly and forming weak reticulations in sacral area ( Figure 3A; Figure  4). Small dark-brown spots of irregular shape located on dorsal surface of snout between nostrils, upper eyelids (two small spots on left and four spots on right upper eyelid), two spots on orbit margins between eyes, few small brown flecks in scapular area, two larger brown spots on mid-dorsal line (one in middle of dorsum, one in sacral area) and a single small spot above cloaca ( Figure 3A). Elbows dorsally dark orange-brown. Four dark transverse dark-brown lines on dorsal surface of each thigh, orange-brown spot on each knee; on shanks a single short brown line at proximal end, larger light-brown blotch medially and smaller brown spot at distal end near tibiotarsal joint ( Figure 3A). Lateral sides of head and body with wide dark reddish-brown to black lateral stripes, clearly separated from lighter dorsal coloration by straight contrast edge, running from snout tip along canthus rostralis to anterior eye corner, then from posterior eye corner above tympanum and posteriorly to groin; dorsal edge of wide dark lateral stripe sharp and straight until groin, which shows very shallow dark inguinal loop with rounded edges. Tympanum uniformly dark-brown with no markings; tympanal area ventrally reddish-brown, dorsally black, sharply edged from light golden-yellow supratympanic area ( Figure 3D).
Background of dorsal surfaces of arms, hands, and feet reddish-brown, with dense bluish-white to turquoise speckles, which spread to fingers and toes ( Figure 3A); disks on digits dorsally brownish with dense whitish marbling, ventrally grey. Ventral surfaces of body, including belly and chest greyish-blue with indistinct brown confluent blotches, getting smaller and denser anteriorly, forming dense brownish pattern on throat ( Figure 3B). Two oblique brownish spots in chest area. Ventral surfaces of forelimbs grey with whitish speckling (Figure 3B, C). Ventral surfaces of hindlimbs reddish-brown with large bluish to turquoise blotches on groin, femur, and medial part of shank ( Figure 3B). Throat, chest, and ventral surfaces of thighs with small bluish-white speckles.

Color of holotype in preservative:
After preservation in ethanol for two years, coloration pattern of holotype resembles that observed in life; however, yellowish and reddish tints faded completely, turning beige-grey, brownish-black patterns on lateral surfaces appeared brownish-grey; ventral coloration lost purple and bluish tints and looked beige-grey with brown spots.
Variation: All individuals in type series were very similar in morphology, body proportions, and body coloration; measurements of type series are shown in Table 4 and representative photographs showing variation in dorsal and ventral coloration of four male paratypes in life are given in Figure 5 and Figure 6. All specimens show certain variation in number and position of small dark-brown spots on dorsum ( Figure 5A, Figure 6) and ventral pattern, including size of two oblique brownish spots in chest area ( Figure 5B). Coloration of Theloderma auratum sp. nov. showed slight variation in response to diel period and microhabitat conditions. In life, coloration of dorsum was somewhat lighter nocturnally than during daytime, with dorsal surfaces looking light-beige to cream. Variation in color related to diel period, stress, and other conditions has been reported for other species of Theloderma (McLeod & Ahmad, 2007;Rowley et al., 2011).
Tadpole description: Description of larval morphology was based on five tadpoles (Gosner stages 29-36) (ZISP 13422-13426) (see Referred specimens for details). Identification of tadpoles was confirmed by 16S rRNA partial sequencing (see Table 2). The main morphometric parameters of the tadpoles are given in Table 5. Ontogenetic stages of Theloderma auratum sp. nov. are shown in Figure 7. Details of tadpole morphology are presented in Figure 8.    External morphology: Body oval, wider than high, body height 71%-78% of body width. Body longer than wide: body width 66%-74% of body length. Snout blunt, rounded. Tail more than two times longer than body, body length 55%-65% of tail length ( Figure 7A, B); 23-28 myotomes discernable in lateral view. Nostrils rounded, oriented anterodorsally, located almost at same distance to snout as to eye (RN/NP ratio 71%-105%).
Internarial distance (IND) 56%-77% of interorbital distance (IOD). Eyes with dorsal orientation. Eye diameter (ED) 9%-18% of body length (BL). Eye-nostril distance (END) larger than eye diameter (ED). Spiracle single, sinistral, located closer to posterior portion of body, directed laterally. Distance from tip of snout to spiracle opening 67%-75% times body length. Vent tube in medial position, with aperture located on same line as margin of ventral tail fin. Height of tail musculature at highest portion 58%-74% of tail height and 50%-57% of maximum body height. Maximum height of dorsal tail fin 34%-42% of maximum tail height. Ventral tail fin same height as dorsal tail fin; maximum height of lower tail fin 35%-46% of maximum tail height. Ventral and dorsal tail fins start roughly at level of vent. Dorsal and ventral tail fins slightly higher in posterior one third of tail length; tail tip gently rounded ( Figure 7C; Figure 8A).

Coloration in life:
Body strongly pigmented: dorsally and ventrally uniformly dark-grey ( Figure 7C, D). In dorsal view, eyes not totally visible on strongly pigmentated background ( Figure 7C). Iris black. Spiracle slightly pigmented. Dorsal and ventral tail fins translucent at edges with greyish coloration, much lighter than on body.

Coloration in preservative:
In ethanol, all parts of tadpoles were less intensively pigmented and turned greyish.
In preservation, myotomes of tail muscles were more discernable than in life.
Advertisement call: Advertisement calls of Theloderma auratum sp. nov. were uttered in a series of 14.27±1.31 (7-21, n=15) tonal calls (Figure 9), resembling an orthopteran call to the human ear. Series duration varied from 3.04 s to 11.34 s (7.88±0.62 s, n=15) and the interval between successive series comprised 21.61±2.57 s (12.16-44.4 s, n=13). Call duration varied from 30 ms to 75 ms (61±0.5 ms, n=214) and inter-call interval varied from 411 ms to 842 ms (529±6.76 ms, n=199) usually decreasing gradually from the beginning to end of the series (Figure 9). Call repetition rate within the series was 1.68±0.03 calls/s (1.53-1.98 calls/s, n=15). Initial fundamental frequency was 2 777±3 Hz (2 570-2 900 Hz, n=214) and final fundamental frequency was 2 856±4 Hz (2 710-3 090 Hz, n=214). The initial and final fundamental frequencies also represented the minimum and maximum fundamental frequencies, respectively. Frequency modulation was usually expressed in the weak lift of fundamental frequency during the whole call. The presence of harmonics varied between/within recordings and mostly depended on recording quality (e.g., sensitivity of recording equipment, distance from vocalizing animal, signal volume, and background noise). Calls from the highest quality recording contained poorly visible second harmonics (Figure 9).
The maximum amplitude frequency varied from 2 760 to 2 950 Hz (2 829±2 Hz, n=214) and coincided with the fundamental frequency. Position in mtDNA genealogy and sequence divergence: According to our mtDNA data, Theloderma auratum sp. nov. belongs to the subgenus Theloderma s. str. (Poyarkov et al., 2015), and is reconstructed as a sister species of Th. palliatum (Figure 2, clade II-B). Uncorrected genetic P-distances between Theloderma auratum sp. nov. 12S rRNA and 16S rRNA sequences and all homologous sequences of congeners included in our analyses varied from 8.9% (with sister species Th. palliatum) to 17.9% (with Th. laeve) (see Table 3). This value is higher than that observed between several currently recognized species of Theloderma (Table 3).  Theloderma auratum sp. nov. is a very secretive species with apparently nocturnal activity.
Animals were usually encountered at night (between 1900 h and 0300 h) after or during heavy rain; males were recorded when calling from shrubs or small trees ca. 20-90 cm above the ground. Diet of the new species remains unknown. In Kon Plong District, Kon Tum Province, both males and females were observed at an elevation of 1 000 m a.s.l. when sitting on leaves of shrubs at 40-50 cm from the ground; animals were active at an air temperature of 20-22°C and 100% humidity.
Reproductive biology: Reproduction of the new species was observed under laboratory conditions.
In terrarium, clutches of Theloderma auratum sp. nov. were found on the under-surface of snags ( Figure 7) and were usually deposited 2-4 cm above the surface of the water. A clutch consisted of one or two, rarely three eggs, enveloped in slimy transparent external egg-capsules (see Figure 7A, B). Duration of embryonic development comprised 10-12 d at ambient air temperatures of 22-23°C. Tadpoles were fed on dry fish food and started active feeding 2-3 d after hatching. Duration of larval development from hatching to the end of metamorphosis was about 2.5 months. At metamorphosis, the main aspects of coloration and dorsal pattern corresponded to that of the adults, except the dorsal coloration pattern: metamorphs had a more intensive orange dorsum with orange reticulate pattern ( Figure  7E), which disappeared with age.

Comparisons: Theloderma auratum sp.
nov. is easily distinguished from most species of the genus Theloderma by a combination of the following morphological attributes: (1) small to medium body size  (2) absence of vomerine teeth; (3) smooth skin on dorsum with dorsal warts or asperities absent, (4) long obtuse snout with pointed snout tip, ESL/SVL ratio 16%-20%; (5) uniform golden-yellow dorsal coloration with weak golden-orange speckling and reticulations; (6) dark reddish-brown to black lateral stripes, separated from dorsal light coloration by sharp edge, forming very weak and shallow inguinal loop; and (7) bicolored iris, golden-orange dorsally, copper-red with thick black intervening ventrally, with a wide black longitudinal medial stripe.
The small body size, absence of vomerine teeth, and smooth dorsal skin lacking asperities or warts distinguishes Theloderma auratum sp.
The Malayan species Th. licin and Th. asperum, Indochinese species Th. albopunctatum, and poorly known Th. baibungense (Jiang, Fei & Huang, 2009) (Liu & Hu, 1962) and Th. lateriticum Bain, Nguyen & Doan, 2009 -by golden-yellow dorsum with golden-orange reticulations and few dark spots (vs. tea brown to dark grey dorsum in Th. rhododiscus and dorsum with deep brick-red background color with distinct black middorsal spot or blotches in Th. lateriticum), greyish belly or whitish belly with brown reticulations (vs. belly brownish-black scattered with grey-white network in both taxa), ventral surfaces of disks grey (vs. orange or red in Th. rhododiscus), completely smooth dorsum without asperities or warts (vs. distinctly white dorsal asperities present in both taxa), and distinctly bicolored golden/black iris (vs. uniformly dark reddish brown iris in both taxa). Theloderma lacustrinum Sivongxay, Davankham, Phimmachak, Phoumixay & Stuart, 2016 from Laos can be differentiated from the new species by its uniform red iris (vs. distinctly bicolored golden/black iris), beige-brown upper jaw (vs. uniformly dark reddish-brown upper jaw), and presence of large black blotches in inguinal region (vs. dark inguinal blotches absent).
Theloderma truongsonense ( Figure 11A) from the central and southern Annamite (Truong Son) Mountains can be distinguished from Theloderma auratum sp. nov. by the following morphological attributes: dorsal skin shagreened with numerous small asperities scattered on dorsum and dorsal surfaces of head including upper eyelids (vs. completely smooth dorsum with only few (3-5) flat superciliary tubercles), comparatively shorter snout with rounded snout tip, ESL/SVL ratio below 16% (vs. tapering snout with pointed tip, comparatively longer, ESL/SVL ratio above 17%), dorsum brown with darker patterning (vs. golden-yellow dorsum with golden-orange reticulations and few dark spots), dorsal edge of dark lateral stripe irregular, deep inguinal loop or large black inguinal blotch surrounded by bluish to turquoise edging always present (vs. dark lateral stripe with smooth sharp edge, inguinal loop absent or shallow), comparatively wider and shorter head, HW/HL ratio above 85% (vs. comparatively narrower and longer head, HW/HL ratio below 84%), and comparatively shorter tibia, TL/SVL ratio below 52% (vs. comparatively longer tibia, TL/SVL ratio above 53%).
Theloderma laeve ( Figure 11C) from low to mid-elevations of Langbian Plateau, southern Vietnam, can be diagnosed from Theloderma auratum sp. nov. by the following combination of morphological characters: dorsum beige with brownish patterning and thin light middorsal stripe (vs. golden-yellow dorsum with golden-orange reticulations and few dark spots), lateral stripe brownish-violet (vs. black to dark brown lateral stripe), ventral surfaces uniform violet-grey with no dark patterning (vs.

Etymology:
The specific name "auratum" is a Latin adjective in the nominative singular (neutral gender), derived from Latin "aurum" for "gold", referring to the golden-yellowish dorsal coloration of the new species.

Recommended vernacular name:
We recommend the following common name in English: Golden Bug-Eyed Frog. Recommended vernacular name in Vietnamese:Ếch Cây Sần Vàng.
Conservation status: Theloderma auratum sp. nov. is, to date, known from five localities in the Tay Nguyen Plateau of central Vietnam. Further research is required to estimate its actual distribution, population trends, and possible threats. It appears that the new species is associated with primary undisturbed montane forests and may be affected by growing anthropogenic pressure and forest destruction, as observed in different areas of central Vietnam. Given the available information, we suggest Theloderma auratum sp. nov. be tentatively considered as a Data Deficient species following IUCN's Red List categories (IUCN, 2001).

DISCUSSION
Our phylogenetic data largely confirmed the phylogeny of Theloderma as given by Poyarkov et al. (2015), Nguyen et al. (2015), and Sivongxay et al. (2016), but included several taxa not analyzed in these studies. Our data supported subdivision of Theloderma into two major clades, corresponding to the subgenera Stelladerma (Clade I) and Theloderma s. str. (Clade II). Furthermore, Th. annae was reconstructed as a sister species of Th. nebulosum with high node support, and Th. moloch, Th. phrynoderma, and Th. ryabovi were supported as a monophyletic sister group to the Th. asperum complex.
Our research clearly demonstrated that current understanding of the diversity of Theloderma is far from complete. Within Th. asperum, our data strongly suggested paraphyly of Th. pyaukkya from Myanmar with respect to Th. baibungense from the eastern Himalayas. In addition, Th. pyaukkya could be subdivided into two highly divergent non-monophyletic lineages from northern and central Myanmar (P-distance 3.9%). Dever (2017) described Th. pyaukkya without including Th. baibungense in phylogenetic analysis and without examination of Th. baibungense specimens; comparisons of Th. pyaukkya in this paper do not allow the discrimination of Th. pyaukkya from Th. baibungense. Our data indicated that the taxonomic status of Th. pyaukkya lineages requires careful reconsideration: synonymy of this species with Th. baibungense can be assumed; however, high genetic differentiation between its lineages suggests that taxonomic reassessment of this group is necessary. Other examples of high intraspecific genetic distances included Indochinese species Th. lateriticum (P=3.6%), Th. truongsonense (P=3.8%), and Th. albopunctatum (P=2.5%); these results suggest the possible presence of cryptic species in these complexes and incomplete taxonomy of the group. Finally, Theloderma sp. from Gia Lai Province, previously identified as Th. laeve by Nguyen et al. (2015), clearly represented a deeply divergent mtDNA lineage sister to Th. laeve s. str. (P=9.0%) and likely corresponds to a yet undescribed species.