Limestone jewel: A new colourful karst-dwelling pitviper (Serpentes: Viperidae: Trimeresurus ) from the poorly explored borderlands of southern peninsular Thailand

We describe a new species of pitvipers from Trang Province of Thailand, near the Thailand–Malaysian border, based on morphological and molecular (2427 bp from cyt b , ND4, and 16S rRNA mitochondrial DNA genes) lines of evidence. Morphologically, Trimeresurus ciliaris sp. nov. is distinguished from its congeners by the following combination of morphological characters: a long papillose hemipenis; first supralabial and nasal scale fused; three to four small supraocular scales; internasals not in contact; small scale between nasal and the scale formed by the fused second supralabial and loreal present; dorsal scales in 17–17–15 rows across the body; ventral scales 172–175 in males, 171 in female; subcaudal scales 59–63 in males, 61 in female, all paired; in life an emerald-green dorsum with reddish-brown bands; creamy-white venter lacking dark dots or stripes on the lateral sides of the ventrals; white vertebral spots present in both sexes on every two or three dorsal scales; dark brown spots forming discontinuous pattern present on 1–3 lateral dorsal scale rows; males with reddish-brown postocular stripe. The new species forms a distinct clade on the phylogenetic tree of the genus Trimeresurus and differs from the morphologically similar species T. venustus by a significant divergence in cytochrome b mitochondrial DNA gene sequences ( p = 12.5%). The new species is currently known from a small karstic area in the Nakawan Range spanning the border of Thailand and Malaysia, in particular in limestone forests in Trang and Satun provinces (Thailand); it likely also occurs in the adjacent parts of Perlis State (Malaysia). Our study also suggests that the taxonomy of T. kanburiensis species complex requires further studies; in particular our study suggests that the status of populations from Chumphon Province of Thailand and Pulau Langkawi Island of Malaysia should be re-assessed.


Introduction
The Asian pitviper genus Trimeresurus Lacépède, 1804 is a complex and species-rich group of venomous snakes that currently comprises 46 nominal species of terrestrial to arboreal species distributed across South and Southeast Asia (Mirza et al. 2023;Uetz et al. 2023).The genus is known to contain numerous cryptic species which are very similar morphologically and are often hardly distinguishable in the wild.Recently, molecular methods have helped to resolve some particularly problematic taxonomic issues among such complexes of cryptic pitviper species, with new taxa being described every year such as in T. albolabris complex (e.g., Chen et al. 2020Chen et al. , 2021;;Mirza et al. 2020), T. kanburiensis complex (e.g., Malhotra et al. 2011;Sumontha et al. 2021), T. popeiorum complex (e.g., Mirza et al. 2023), and T. stejnegeri complex (e.g., Rathee et al. 2022).
Recently, the peninsular Indian and Sri-Lankan taxon Peltopelor and the Southeast-Asian taxon Craspedocephalus were regarded as a distinct genus Craspedocephalus, following indirectly the conclusions of Malhotra and Thorpe (2004a), with two respective subgenera (e.g., Mallik et al. 2021).However, more recently Mirza et al. (2023), based on an extensive sampling and mitochondrial DNA (hereafter mtDNA)-based genealogy, suggested that these two taxa should be better treated as separate genera due to their deep divergence, along with biogeographic and morphological distinctiveness; this taxonomy was not followed by the Reptile Database (Uetz et al. 2023).Recognizing the limitations of the mtDNA-markers in reconstruction of phylogenetic relationships (see, e.g., Hillis et al. 2021;Gallego-Garcia et al. 2023), we herein treat Peltopelor and Craspedocephalus as subgenera of the genus Craspedocephalus, and recognize the genus Trimeresurus sensu lato with six subgenera pend-ing subsequent multilocus and genome-level studies on evolutionary relationships of this group.
The Trimeresurus kanburiensis species complex (previously also often referred to as 'T.venustus species complex', see David et al. 2004;Sumontha et al. 2021; however as T. kanburiensis represents the earliest described taxon of this group, the name 'T. kanburiensis species complex' is more appropriate) encompasses karst-dwelling pitvipers distributed across the Thai-Malay Peninsula and currently includes three nominal species: T. kanburiensis (northern Tenasserim mountains in Kanchanaburi Province of Thailand), T. kuiburi (central Tenasserim mountains in Prachuap Khiri Khan Province of Thailand), and T. venustus (currently known to occupy the southern part of Thai-Malay Penisula from the Isthmus of Kra in Chumphon Province of Thailand in the north to northernmost Peninsular Malaysia in the south, Sumontha et al. 2021) (see Fig. 1).For a long time T. kanburiensis and T. venustus were considered to be synonyms (Viravan et al. 1992;Warrell et al. 1992), until 2004, when both were recognized as distinct species (David et al. 2004;Malhotra and Thorpe 2004b).
Among other members of the T. kanburiensis species complex, the Beautiful Pitviper, T. venustus apparently has the widest distribution and is reported from Thailand (Surat Thani, Phang-Nga, Trang, Nakhon Si Thammarat, Krabi, and Chumphon provinces), and northern Peninsular Malaysia (Kedah State) (Grismer et al. 2006;Chanhome et al. 2011;Chan-ard et al. 2015;Sumontha et al. 2021; see Fig. 1 for details).However, recently Sumontha et al. (2021) demonstrated some distinctive features of the population from Chumphon and reported that this population differs in coloration and pattern from typical T. venustus s. str., and its taxonomic status requires further investigations.
During our recent field surveys in Trang Province of Thailand, we encountered several unusual pitviper specimens that were superficially similar to T. venustus in overall morphological habitus and body coloration.Closer morphological examination of scalation, pattern, color and morphometric characteristics of this population demonstrated clear morphological differences from T. venustus s. str.In the present paper we demonstrate that these differences are concordant with a significant divergence between mtDNA gene sequences among the two populations.Consequent phylogenetic analyses of three mtDNA genes (cyt b, ND4 and 16S) confirmed the placement of the Trang population within Trimeresurus s. str.where it forms a deeply divergent lineage, with sister relationships to a clade encompassing all other members of T. kanburiensis and T. macrops species complexes with exception of T. rubeus.Therefore, in the present paper we describe the Trimeresurus population from Trang Province as a new species and discuss the taxonomy of T. kanburiensis species complex.

Sampling
Fieldwork was carried out in Khao Ting Cave, Li Phang Subdistrict, Palian District, Trang Province, Thailand in June 2016 by P. Pawangkhanant and in January 2023 by P. Pawangkhanant, S. Idiiatullina, and T. Woranuch (Fig. 1, locality 10).Geographic coordinates and altitude were obtained using a Garmin GPSMAP 60CSx GPS receiver (USA) and recorded in datum WGS 84.Specimens were collected by means of snake hooks in the field, photographed in life, and euthanized using MS-222 solution within 24 h after capture following Hepps Keeney and Harrison (2022).Specimens were fixated in 4% buffered formalin for 24 h and later stored in 70% ethanol.Specimens were subsequently deposited in the herpetological collection of the School of Agriculture and Natural Resources, University of Phayao (AUP, Phayao, Thailand), the Queen Saovabha Memorial Institute (QSMI, Bangkok, Thailand), and the Zoological Museum of Moscow State University (ZMMU, Moscow, Russia).Tissues for genetic analyses were taken from liver or heart prior to preservation of specimens and stored in 96% ethanol.
Specimen collection and animal use protocols were approved by the Institutional Ethical Committee of Animal Experimentation of the University of Phayao, Phayao, Thailand (certificate number UP-AE64-02-04-005, issued to Chatmongkon Suwannapoom) and were strictly compliant with the ethical conditions specified in the Thailand Animal Welfare Act.Field work, including collection of animals in the field, was authorized by the Institute of Animals for Scientific Purpose Development (IAD), Bangkok, Thailand (permit numbers U1-01205-2558 and UP-AE59-01-04-0022, issued to Chatmongkon Suwannapoom).

DNA isolation and sequencing
Total genomic DNA was extracted from ethanol-preserved muscle or liver tissues using standard phenol-chloroform extraction procedures (Sambrook et al. 1989) followed with isopropanol precipitation.We used the polymerase chain reaction (PCR) to amplify three mtDNA fragments: complete sequences of cytochrome b (cyt b) and NADH dehydrogenase subunit 4 (ND4) genes and a fragment of 16S rRNA gene.Primers used for both PCR and sequencing are summarized in Table 1.
PCR conditions for cyt b gene followed the protocol of Dahn et al. (2018) with a touchdown, and included the following steps: (1) initial denaturation step at 94°C for 5 min; (2) 10 cycles of denaturation at 94°C for 1 min, annealing for 1 min with temperature decreasing from 50°C to 45°C with cool-down at 0.5°C per each cycle, and extension at 72°C for 1 min; (3) 24 cycles of denaturation at 94°C for 1 min, annealing at 45°C for 1 min and extension at 72°C for 1 min; (4) final extension at 72°C for 10 min; and (5) cooling step at 4°C for storage.For ND4 gene PCR conditions followed the protocol from Salvi et al. (2013): (1) initial denaturation step at 92°C for 3 minutes; (2) followed by 16 touchdown cycles with 30 seconds at 92°C, annealing temperature decreasing 0.5°C per cycle from 60°C to 52°C (30 seconds) and extension for 1 minute at 72°C; (3) followed with 20 more cycles similar to the latter but with annealing temperatures stable at 52°C; (4) a final extension at 72°C for 15 minutes; and (5) cooling step at 4°C for storage.For 16S rRNA, we used the PCR protocol that followed Green et al. (2010): (1) initial denaturation step at 94°C for 5 min; (2) 35 cycles of denaturation at 94°C for 1 min, annealing at 55°C for 1 min and extension at 72°C for 1 min; (3) final extension at 72°C for 10 min; and (4) cooling step at 4°C for storage.
All amplifications were run using an iCycler Thermal Cycler (Bio-Rad).PCR products were loaded onto 1% agarose gels in the presence of ethidium bromide and visualized in electrophoresis.The successful targeted PCR products were purified by Diatom DNA PCR Clean-Up kit and outsourced to Evrogen® (Moscow, Russia) for sequencing; sequence data collection and visualization was performed on an ABI 3730xl Automated Sequencer (Applied Biosystems).

Molecular phylogeny
To estimate the phylogenetic relationships of the genus Trimeresurus, we aligned the newly obtained cyt b, ND4 and 16S rRNA sequences together with all available previously published sequences of T. venustus, T. kanburiensis and T. kuiburi, as well as the sequences of 33 species of Trimeresurus species representing all major groups within the genus, five species of Craspedocephalus; we used the sequence of Azemiops feae to root the tree (Gen-Bank accession numbers, voucher and locality information are summarized in Table 2).
We initially aligned the nucleotide sequences in the MAFFT v.6 online webserver (Katoh et al. 2019) with Table 1.Primers used in this study.F -forward primer, R -reverse primer.default parameters, and subsequently checked them by eye in BioEdit 7.0.5.2 (Hall 1999) and slightly adjusted when required.The mean uncorrected genetic p-distances between sequences were calculated with MEGA 6.0.(Tamura et al. 2013) based on cyt b sequences of pitviper species of the subgenus Trimeresurus, and missing data or gaps were subjected to the pairwise deletion option.

Gene
The best-fit substitution models for the data set were selected for genes and codon positions in Partitionfinder 2.1.1 (Lanfear et al. 2012) using the Akaike information criterion (AIC).When the same model was proposed for different codon partitions of a given gene, they were treated as a single partition.Phylogenetic trees were estimated for the combined mitochondrial DNA fragments (cyt b, ND4 and 16S rRNA) data set.We inferred the matrilineal genealogy of Trimeresurus using Bayesian inference (BI) and maximum likelihood (ML) approaches.We used the IQ-TREE webserver (Nguyen et al. 2015) to generate the ML-tree and assessed the confidence in tree topology by 1000 ultrafast-bootstrap replications (UFBS).We conducted BI in MrBayes 3.1.2(Huelsenbeck and Ronquist 2001).Metropolis-coupled Markov chain Monte Carlo (MC-MCMC) analyses were run with one cold chain and three heated chains for one million generations and sampled every 1000 generations.The run was checked to ensure the effective sample sizes (ESS) were all above 200 by exploring the likelihood plots using TRACER v. 1.7 (Rambaut et al. 2018).We discarded the initial 1000 trees as burn-in.For BI-analysis we assessed the confidence in tree topology by the posterior probability (PP) (Huelsenbeck and Ronquist 2001).We a priori regarded the tree nodes with UFBS values of 95% or higher and PP values over 0.95 as strongly supported; UFBS values between 95% and 90% and PP values between 0.95 and 0.90 were regarded as moderate support, and lower values were regarded as a lack of node support (Huelsenbeck and Hillis 1993).

Morphological differentiation
Measurements and meristic counts followed Sumontha et al. (2021).Paired meristic characters were given in the left/right order.The following measurements were taken with a Mitutoyo digital caliper to the nearest 0.1 mm: ED, horizontal eye diameter; HD, maximum head depth; HL, head length (from the tip of rostral to the posterior end of the jaw); HW, maximum head width; SnL, snout length (from the tip of rostral to the anterior eye margin); SOL, supraocular length; SOW, supraocular width; SVL, snout-vent length; TaL, tail length; TL, total length; VED, vertical eye diameter.Examined meristic characters include: CP, cloacal plate(s) (divided or single); ASR, anterior number of dorsal scale rows (at one HL behind the head); IL, number of infralabial scales; IOS, interorbital scales, counted along a row between the middle of supraocular scales; MSR, number of dorsal scale rows at midbody (at the level of the ventral plate corresponding to the half of the total number of ventrals); PosOc, number of postocular scale(s); PSR, posterior number of dorsal scale rows (at one HL before the cloacal plate); PreV, number of preventrals (scales directly preceding the ventrals, unpaired, wider than long but not in contact on each side with the 1 st dorsal scale row); SC, number of subcaudal scales, not including the terminal pointed scute; SL, number of supralabial scales; SRR, dorsal scale rows reduction (counted following Dowling 1951a); VEN, number of ventral scales (counted following Dowling 1951b).We also counted the number of teeth on upper and lower jaws for each specimen.Sex was determined by examination of the hemipenes that were forcedly everted by using water injection in tail base prior to the preservation of the specimen.

Sequence characteristics
A total of 2,427 aligned base pairs were obtained from the three examined mtDNA fragments, including: 1,104 bp from cyt b, 803 bp from ND4 and 520 bp from 16S rRNA.Protein-coding sequences were translated into amino acids in order to confirm that no pseudogenes have been amplified.We deposited the newly obtained sequences in GenBank under the accession numbers OR470534-OR470580 and OR471621-OR471643 (see Table 2).
PartitionFinder 2.1.1 proposed the following partition schemes and substitution models: GTR+I+G model for 16S rRNA and the second codon position of ND4, GTR+G model for the first codon position of cyt b, HKY+I+G model for the second codon position of cyt b and the first and the third codon positions of ND4, and HKY+I model for the third codon position of cyt b, which resulted in four partitions in total.

Phylogenetic relationships
The ML and BI analyses recovered trees with very similar topologies, with minor topological differences associated with only a few deeper nodes with insufficient nodal support (Fig. 2).Our mtDNA-genealogy recovers the genus Trimeresurus as monophyletic (100/1.0,hereafter node support values are given for ML UFBS/BI PP, respectively) with respect to its sister genus Craspedocephalus, however, the genealogical relationships within Trimeresurus remain insufficiently resolved.Monophyly of the genus Craspedocephalus is not supported in our tree: it consists of two deeply-divergent mtDNA clades corresponding to the Southeast-Asian subgenus Craspedocephalus (95/1.0), and to the Indian and Sri-Lankan subgenus Peltopelor (100/1.0),however this topology lacks significant nodal support (56/0.74;see Fig. 2).Monophyly of the subgenus Trimeresurus is also not supported in our analysis with T. albolabris species group (100/1.0)recovered as genetically distinct in mtDNA se-quences from other members of the subgenus (93/0.98,Fig. 2).The latter include the members of T. macrops species complex and T. kanburiensis species complex which form a strongly supported clade (100/1.0)(Fig. 2).
The genealogical relationships within the macropskanburiensis clade are generally sufficiently resolved and suggest that Trimeresurus sp. from Trang Province forms a lineage with sister relationships to all other taxa of this group (this topology received significant support in ML analysis, but was not supported in BI analysis, 93/-; see Fig. 2).The phylogenetic position of T. rubeus from Vietnam remains sufficiently unresolved in our tree.The T. macrops and T. kanburiensis species complexes were recovered as not monophyletic, with their members grouped into two major clades.The first clade (Clade 1) includes T. venustus s. str.(from Nakhon Si Thammarat, Krabi, and Surat Thani provinces, 100/1.0)and T. cardamomensis (from eastern Thailand, Cambodia, and southern Vietnam); monophyly of this group got significant nodal support (92/1.0,Fig. 2).Clade 2 (100/1.0)includes specimens of T. kuiburi (from Prachuap Khiri Khan Province, 100/1.0),T. cf.venustus 1 (from Chumphon Province, 100/1.0),T. honsonensis, and T. macrops (Fig. 2), with T. kuiburi + T. cf.venustus 1 forming a monophyletic group which was not supported by ML analysis, but got a significant support in BI analysis (66/0.95).Trimeresurus kanburiensis was strongly suggested as a sister taxon to a group including Clades 1 and 2 (99/1.0,Fig. 2).In summary, our analyses demonstrate that Trimeresurus sp. from Trang Province (100/1.0)represents a new highly divergent and previously unknown lineage of pitvipers, which does not fall into radiation of T. venustus sensu lato, and is suggested as a sister lineage to all remaining members of the T. macrops and T. kanburiensis species complexes (Fig. 2).

Genetic distances
The uncorrected p-distances for the cyt b gene fragment among examined members of the subgenus Trimeresurus are presented in Table 3. Interspecific distances among the members of the subgenus Trimeresurus varied from p = 3.0% (between T. venustus and T. honsonensis) to p = 15.1% (between T. kuiburi, T. erythrurus and T. albolabris).The newly discovered Trimeresurus sp.lineage from Trang is highly divergent from other congeners with the minimal genetic distance of p = 7.7% (with T. honsonensis and T. cardamomensis) and the maximal distance of p = 17.3% (with T. albolabris) of sequence divergence in cyt b mtDNA gene.

Taxonomy
Our molecular results are further corroborated by the morphological analysis, which recovered a number of important diagnostic characters which allow distinguishing the population of Trimeresurus sp. from Trang from T. venustus s. str.and all other congeners (summarized Chresonymy.Cryptelytrops cf.venustus (partim): Chan et al. (2011: 258).
Paratypes (n = 4).QSMI 1538, adult female collected by P. Pawangkhanant on 18 June 2016 from the same locality as the holotype; AUP-02011, ZMMU Re-17662, ZMMU Re-17663, three adult males collected on 23 January 2023 from the same locality and by the same collectors as for the holotype.

Diagnosis.
A species of the genus Trimeresurus which is assigned to the subgenus Trimeresurus based on the following morphological attributes: a long papillose or calyculate hemipenis and partially fused first supralabial and nasal scales (Malhotra and Thorpe 2004a;David et al. 2011).The new species Trimeresurus ciliaris sp.nov.can be distinguished from all other congeneric species by the following combination of morphological characters: three or four small convex supraocular scales; internasals not in contact; small scale between nasal and the scale formed by the fused second supralabial and loreal present; dorsal scales in 17-17-15 rows; ventral scales 172-175 in males, 171 in a single female; subcaudal scales 59-63 in males, 52 in a single female, all paired; in life an emerald-green dorsum with reddish-brown crossbands; creamy-white venter lacking dark dots or stripes on the lateral sides of the ventrals; iris olive-green with faded-brown horizontal stripe; tail dark brown mottled with rusty spots; white vertebral spots present in both sexes located approximately every two or three dorsal scales; dark brown spots forming discontinuous pattern present on 1-3 lateral dorsal scale rows; males having a reddish-brown postocular stripe.
Description of the holotype.Adult male (Fig. 3), specimen in a good state of preservation.Body cylindrical, long and thin (SVL 345 mm, TaL 72 mm, TL 417 mm, TaL/TL 0.17) (Fig. 3A,B).Head triangular in dorsal view (Fig. 3C), elongate, clearly distinct from the neck (HL 16.5 mm, HW 11.4 mm, HW/HL 0.69).Snout elongate, flattened and rounded in dorsal view (Fig. 3C), rather rectangular in lateral view (Fig. 3E,F), with a very distinct and sharp canthus rostralis (SnL/HL 0.30, SnL/ED 1.92).Distance between eye and nostril 3.9 mm on both sides.Rostral slightly visible in dorsal aspect, triangular (Fig. 3C).Pupil vertically elliptical, loreal pit present, triangular in shape (Fig. 3E,F).Nostril completely enclosed in nasal scale; nasal scale partially fused with first supralabial (Fig. 3E).Shield bordering the anterior edge of loreal pit fused with second supralabial.Anterior subocular long, thin, and crescent-like, separated from the 4 th and 5 th supralabials by one row of scales; posterior subocular ovoid, separated from the 6 th supralabial by 2/1 scales (Fig. 3E, F).Three preoculars on each side of the head; two upper preoculars located above the loreal pit, elongated, in contact with the single loreal which separates them from the nasal; lower preocular forming the lower margin of the loreal pit, lower preocular in contact with third supralabial (Fig. 3E, F).A small scale between the nasal and the scale formed by the fused second supralabial and loreal; 2/2 postoculars; 9/9 supralabials, third the largest (Fig. 3E,F); 11/10 infralabials, those of the first pair in contact with each other behind the mental; the first three pairs of infralabials in contact with the single pair of chin shields (Fig. 3D).Six pairs of gulars aligned be-tween the chin shields and the first preventral (Fig. 3D).
One large pair of enlarged internasals, separated by three small scales.Three small slightly convex supraoculars, same size as or smaller than preoculars and postoculars (Fig. 3C).Scales on snout and in the interorbital region smooth, irregular, subimbricate; temporal and occipital scales moderately keeled (Fig. 3C,E,F).Dorsal scales in 17-17-15 rows.Dorsal scales all strongly keeled, except the first row, which is smooth (Fig. 3H).One preventral + 172 ventrals.Cloacal plate single; 61 subcaudals, all divided.Hemipenes long, papillose, and deeply forked with small, soft, basal spines (Fig. 3G).Each maxilla bearing a single large venom tooth; three teeth on palatine, 11 on pterygoid, and 12 on dentary.Figs 3-5).Dorsal surface of the head emerald-green with many scales partly or entirely dark red, especially on the snout, and on the interorbital, temporal and occipital regions (Fig. 3C).A reddish-brown postocular stripe extends from postocular scales to the neck (Fig. 3E,F).Black vertical pupil; iris olive-green with faded brown horizontal stripe (Fig. 3E,F).The background color of the dorsum emerald-green, similar to that of the head, with about 75 irregular, dark red cross-bands (Fig. 3A).These red bands are about two dorsal scales long mediodorsally, but getting narrower on the lower flanks where they are about 1-2 dorsal scale in length.At the level of the vertebral row, the red bands are generally separated by one dorsal scale.White vertebral spots present along the vertebral scale row, approximately every two or three scales (Fig. 3A).Ventrolateral stripe absent.Dark brown spots forming discontinuous pattern present on first to third lateral dorsal scale rows (Fig. 3H).Tail dark brown mottled with rusty spots.Ventral surface of the tail dark brown with white irregular bands, tail tip completely black.Infralabials greenish-white with faint brown spots (Fig. 3E,F); the ventral surfaces of the head creamy white with bluish tint (Fig. 3D); the ventral surfaces of body creamy white (Fig. 3B).

Coloration in life (
In preservative the background dorsal color faded to greyish-brown, with less contrasting dark marks on the head and bands on the dorsum; the ventral color became whitish grey. Variation.The main meristic and morphometric characters of the type series of Trimeresurus ciliaris sp.nov.are summarized in Table 4; color variation of the type series is presented in Figures 4, 5A, 6A-B, 7B-E.All members of the type series, similarly to the holotype, show a nasal scale partially fused with the first supralabial.Head White vertebral dots every 2-3 scales every 2-3 scales every 2-3 scales every 2-3 scales every 2-3 scales scalation of the paratypes generally agrees with that of the holotype, however male ZMMU Re-17662 has a suboculars fused in a single long crescent-shaped scale (Fig. 4B), while the two males AUP-02011 (Fig. 4C) and ZMMU Re-17663 (Fig. 4D) have four small supraoculars on each side of the head instead of three as in the holotype.Coloration of paratypes is very similar to that of the holotype; all specimens have a creamy-white belly and lack ventrolateral stripe.Female specimen QSMI 1538 has slightly lower number of subcaudals than in males (SC 52 vs. 59-63).The white vertebral spots are present both in males and in females; they are regularly spaced by two or three vertebral scales.The number of dark bands on dorsum varies from 70 to 81 among the members of the type series, without apparent sexual dimorphism.Coloration of female specimen QSMI 1538 largely faded due to preservation in formalin for seven years.
Distribution and natural history.Currently, Trimere surus ciliaris sp.nov. is known only from a narrow limestone area in the Nakawan Range spanning the border of Thailand and Malaysia, in particular in limestone forests in Trang (Palian District) and Satun (Tha Le Ban National Park) provinces, Thailand (Fig. 1).It is highly likely that the new species also inhabits the northernmost part of Perlis State of Peninsular Malaysia (Fig. 1, see Discussion for details).The new species occurs in lowland dipterocarp forest on limestone rocks and appears to be strongly associated with a karst landscape.The new species was recorded for the first time in limestone formations within the Tha Le Ban National Park, Satun Province of Thailand, in June 2016 (P.Pawangkhanant pers.obser.),where it was observed in a lowland dipterocarp forest with numerous large karst rocks.Almost all specimens of Trimeresurus ciliaris sp.nov.were collected during daytime (except ZMMU Re-17663 that was collected at night around 20h50).All specimens were found near large boulders of karst rocks covered with dense vegetation and leaf litter (Fig. 7A); one specimen was found after heavy rain around 18h00 while perching vertically on a limestone rock surface ca. 5 m above the ground.
In captivity, due to its small size, this species of pitviper mostly feeds on small geckos and microhylid frogs [Microhyla butleri Boulenger, M. cf.heymonsi Vogt, and Micryletta cf.lineata (Taylor)].Nothing is known about the diet of the new species in the wild, but in one specimen from Mueang District, Satun Province, caudal luring behavior was observed: the male specimen was wiggling its tail tip when an adult gecko Cnemaspis biocellata Grismer, Chan, Nasir & Sumontha approached its shelter.Other amphibian and reptile species recorded in syntopy with the new species include: Ingerophrynus parvus (Boulenger), Micryletta cf.lineata, Microhyla butleri, Kaloula latidisca (Chan, Grismer & Brown)  Etymology.The species name "ciliaris" is a Latin adjective in the nominative singular, masculine gender, derived from Latin word "cilium" meaning "an eyelash", and is given in reference to the characteristic small and distinct supraocular scales in the new species, which resemble eyelashes or eyebrows in lateral view.We suggest the following common names for the new species: "Ngu Hang Mhai Khao Hin Poon" (งู หางไหม้ เขาหิ นปู น) (in Thai), "Limestone Eyelash Pitviper" (in English), and "Resnitchataya ukrashennaya kufiya" (in Russian).
Comparisons.The new species is morphologically and phylogenetically placed within the subgenus Trime resurus (Malhotra and Thorpe 2004a;David et al. 2011) and morphologically is overall most similar to other limestone-dwelling pitviper species, including T. kanburien-sis, T. kuiburi, and T. venustus s. str., so the comparisons with these three species appear to be the most pertinent.Trimeresurus ciliaris sp.nov.can be easily distinguished from other congeners by having: 17 midbody scale rows, three or four small supraocular scales, and a small scale between the nasal and the scale formed by the fused second supralabial and loreal.The main diagnostic characters separating Trimeresurus ciliaris sp.nov.from these three species are summarized in Table 5.The comparison of body coloration and head scalation of these species is presented in Figures 5 and 6, respectively.From the superficially similar Trimeresurus truongsonensis Orlov, Ryabov, Vu & Ho, a member of the subgenus Virido vipera, the new species can be easily distinguished by having a long papillose and deeply forked hemipenis (vs.short spinose hemipenis) and partially fused first supralabial and nasal scales (vs.separate).

Discussion
In the present study we report on a previously unknown lineage of limestone-associated pitvipers of the subgenus Trimeresurus.According to our mtDNA-based genealo-gy, the newly discovered species Trimeresurus ciliaris sp.nov.forms a highly divergent lineage of pitvipers, which is suggested as the sister lineage to all other members of T. kanburiensis and T. macrops species complexes (see Fig. 2).The new species is also separated from its congeners by a significant divergence in cyt b gene sequences (with p = 7.7-17.3%),which is higher than the genetic distances between many other recognized species of the group (see Table 3).The mtDNA-based genealogy presented herein is insufficiently resolved in several deeper nodes; however in general it agrees well with the earlier genome-wide analysis by Mrinalini et al. (2015) which demonstrated the non-monophyly of T. kanburiensis and T. macrops species complexes suggesting complicated evolutionary history of this group.Though the phylogenetic position of the new species is not completely resolved in our phylogenetic tree, it is obvious that Trimeresurus ciliaris sp.nov.represents a deeply-divergent, previously unknown taxon which warrants taxonomic recognition.
Our study also provides some insights on phylogenetic relationships within the subgenus Trimeresurus.This subgenus was originally proposed by Malhotra and Thorpe (2004a)   and the fusion of the first supralabial and nasal scales).However, the monophyly of this group was not supported in the phylogenetic analysis of Malhotra and Thorpe (2004a).Later, David et al. (2011) demonstrated that the name Trimeresurus Lacépède, 1804 should be applied to this group (type species: Trimeresurus viridis Lacépède, 1804, nomen oblitum, a senior synonym of Trimeresurus albolabris insularis Kramer, 1977), however, these authors only discussed the nomenclature of the subgenus Trimeresurus and did not address its systematics.Our mtDNA-based genealogy suggested the significant differences between the members of T. albolabris and T. kanburiensis + T. macrops species groups included in this taxon.Though only based on limited data from three mtDNA genes, our study strongly suggests that the subgenus Trimeresurus sensu lato David et al. (2011) is likely non-monophyletic (Fig. 2).At the same time, the study by Malhotra et al. (2010) which included nuclear intron sequence data recovered the subgenus Trimeresurus as monophyletic.In summary, it is highly likely that the clade encompassing the T. kanburiensis + T. macrops species groups might warrant a taxonomic recognition as a separate subgenus of the genus Trimeresurus.Further comprehensive phylogenetic studies, including the analyses of multiple nuclear DNA genetic markers and additional morphological characters, are required to achieve a better resolution of evolutionary relationships within the subgenus Trimeresurus.
Our updated mtDNA-based genealogy also provides new data on phylogenetic relationships within the T. kan buriensis + T. macrops clade.Two morphologically defined species complexes were recognized within this clade: limestone-associated T. kanburiensis species group joining species with blotched coloration pattern on the dorsum (including T. kanburiensis, T. venustus, T. kuiburi, and also including Trimeresurus ciliaris sp.nov.) and lowland forest-associated T. macrops species group, which includes species with a uniform green coloration of dorsum (T.macrops, T. cardamomensis, and T. rubeus).Trimeresurus honsonensis has also a blotched dorsum coloration pattern and was reported to superficially resemble T. venustus, though this species is not associated with karst landscapes and occurs only in secondary forests with large granite boulders on Hon Son Island of southern Vietnam (Grismer et al. 2008).In our mtDNA-based genealogy the members of T. kanburiensis and T. macrops species complexes did not form reciprocally monophyletic clades, but were mixed with each other: T. cardamomensis is suggested as a sister species of T. venustus s. str., the phylogenetic position of T. rubeus remains unresolved, while T. macrops, T. honsonensis, T. kuiburi and T. cf.venustus 1 form a well-supported group (Fig. 2).Apparently, this result suggests that the transition between forest-dwelling and karst-dwelling lifestyle likely occurred multiple times independently within the subgenus Trimeresurus, often resulting in strikingly similar morphologies.It is noteworthy, that among other Trimeresurus subgenera, limestone-associated species often show coloration similar to that in T. kanburiensis group; for example, the karst-dwelling species Trimeresurus (Viridovipera) truongsonensis from central Vietnam and Laos also has a greenish dorsum with red blotches and superficially resembles T. venustus (Orlov et al. 2004).Further research is needed to investigate the consequences of lifestyle and habitat shifts in the evolution of Trimeresurus pitvipers (see discussion in Sanders et al. 2006).
Our work also provides new data on the diversity of limestone-associated Trimeresurus pitvipers in Thailand.In addition to the description of Trimeresurus cilia ris sp.nov., our study further corroborates the results of Sumontha et al. (2021), who suggested that the taxonomic status of T. venustus from Chumphon Province of Thailand required further study.Our molecular analyses demonstrated that despite its morphological similarity to T. venustus, the Chumphon population represents a distinct lineage of Trimeresurus, phylogenetically more closely related to T. kuiburi, T. macrops and T. honsonensis than to T. venustus sensu stricto.In the present paper, we tentatively refer to this population as T. cf.venustus 1, though we recognize this lineage as a distinct new yet undescribed species, the formal description of which is currently in progress and will be published elsewhere.
Further research is needed to clarify the taxonomic status of the T. kanburiensis species complex members occurring in Malaysia.A strange population of pitvipers superficially resembling T. venustus was recorded on Pulau Langkawi Island, Kedah State, Malaysia (Fig. 1, locality 13).This record published by Grismer et al. (2006: 65, fig. 18) is based on two specimens (LSUDP 974-75) originally identified as T. venustus.However, the coloration pattern of these specimens seem to be quite distinct as compared to other congeners, in particularly to Trimeresurus ciliaris sp.nov.and T. venustus s. str.As the genetic information for these specimens is still lacking, we tentatively refer to the Langkawi population as T. cf.venustus 2. Further integrative taxonomic analysis of additional materials from Malaysia, including morphological and molecular data, is required to clarify the status of this population which well may also represent a previously undescribed taxon.
Based on our observations, the newly described species Trimeresurus ciliaris sp.nov.inhabits not only Trang Province, but is also distributed in Satun Province of Thailand (recorded in Thale Ban National Park, Wang Prachan, Khuan Don District based on P. Pawangkhanant pers.obser.).Moreover, the new species likely also inhabits Perlis State of Malaysia (Fig. 1).Chan et al. (2011) reported on a strange population of T. cf.venustus from Perlis State Park, Northern Malaysia.This record was originally published under the name Cryptelytrops cf.venustus; based on examination of the available photo material it is conspecific with Trimeresurus ciliaris sp.nov.(Universiti Kebangsaan Malaysia Digital Photograph Collection: UKMDPC 1.0144).Additional records of a similarly-looking pitviper have been recently provided for Wang Kelian View Point, Kaki Bukit, Perlis State (Azrul Azrizal Azmy pers.obser.; see Fig. 6D,E).Morphologically these records are similar to Trimeresurus ciliaris sp.nov. in body habitus and coloration pattern; moreover the localities belong to the same limestone massif (Nakawan Range) continuing from Perlis State of Malaysia to Trang Province of Thailand and are located within approximately 90 km from the type locality of the new species.Located on the Malaysia-Thailand border, the Nakawan Range represents Malaysia's largest continuous limestone area and is known as an area with high herpetofaunal diversity (Chan et al. 2009;Grismer et al. 2014).Additional sampling efforts and along with molecular and morphological data on the Perlis population of Trimeresurus ciliaris sp.nov.are required; we suspect that this locality represents the southern limit of the new species distribution.Toxicological status of the new species remains unknown.In summary, further investigations are required to elucidate the distribution, diet, reproductive biology, population size, population trends and conservation status of the new species.
Our work raises the total number of species of the genus Trimeresurus to 47; Trimeresurus ciliaris sp.nov.represents the fifteenth species of Trimeresurus known for Thailand.Despite the recent significant progress, our knowledge of molecular phylogeny, classification and distribution of Asian pitvipers remains incomplete.Last but not least, the medical relevance of Trimeresurus pitvipers further underlines the necessity for future studies on the taxonomy of this genus.

Figure 2 .
Figure 2. Maximum Likelihood (ML) tree of Trimeresurus derived from the analysis of 2427 bp of cyt b, ND4 and 16S rRNA mitochondrial DNA gene sequences.For voucher specimen information and GenBank accession numbers see Table 2. Numbers at tree nodes correspond to ML UFBS/BI PP support values, respectively.Colors of clades and locality numbers correspond to those on the map in Figure 1.Photograph by P. Pawangkhanant.

Figure 3 .
Figure 3.The holotype of Trimeresurus ciliaris sp.nov. in life (ZMMU Re-17661, adult male) from Thum Khao Ting, Palian District, Trang Province, Thailand.A dorsolateral and B ventral view of body; C head in dorsal, D in ventral, E in lateral right, and F in lateral left aspects; G right hemipenis, partially everted; H close-up of dorsal scales.Photographs by P. Pawangkhanant.

(
360 mm in male, 333 mm in female vs. 412 mm in male, 572 mm in female); lower anterior number of dorsal scale rows (17 vs. 23, rarely 21 or 22); lower number of midbody scale rows (17 vs. 19); absence of dark lateral stripe on ventrals, Fig. 5A (vs.dark lateral stripe on ventrals always present, discontinuous, olive-brown, Fig. 5B); three or four small supraocular scales, Fig. 6B (vs.single large supraocular scale, Fig. 6D); body color in life (reddish-brown bands on an emerald-green background [Fig.5A] vs. dark olive-brown bands on olive-greyish background [Fig.5B]); white vertebral spots present in both sexes separated by 2-3 dorsal scales vs. white vertebral spots present only in males and are separated by 3-5 dorsal scales; iris color (olive-green with faded brown horizontal stripe [Fig.6A] vs. brown, slightly golden [Fig.6C]); and tail color (dark brown mottled with rusty spots vs. brownish-gray with olive-brown blotches).Trimeresurus ciliaris sp.nov.further differs from T. kuiburi by having: smaller maximal SVL in female (333 mm vs. 451 mm); internasals separated by two or three scales (vs.always in contact); lower anterior number of dorsal scale rows (17 vs. 21 rarely 23); lower number midbody scale rows (17 vs. 19); higher number of ventrals in females (172-175 vs. 164-166); ventral surfaces of body creamy-white (vs.pale-green); three or four small supraocular scales, Fig. 6A (vs.one large supraocular scale, Fig. 6F); small scale between nasal and second supralabial present vs. absent; body color in life (reddish-brown bands on emerald-green background [Fig.5A] vs. red to purple bands on bottle-green background [Fig.5C]); white vertebral spots present in both sexes separated by 2-3 dorsal scales vs. white vertebral dots present only in males and are separated by 5-6 dorsal scales; iris color (olive-green with faded brown horizontal stripe [Fig.6A] vs. copper [Fig.6E]); and tail color (dark brown mottled with rusty spots vs. red with some thin lighter bands).Trimeresurus ciliaris sp.nov.can be further differentiated from T. venustus s. str.by having: smaller maximal SVL (360 mm in male, 333 mm in female vs. 490 mm in male, 456 mm in female); lower anterior number of dorsal scale rows (17 vs. 23, rarely 21 or 25); lower number midbody scale rows (17 vs. 21, rarely 19); white vertebral spots present in both sexes separated by 2-3 dorsal scales (vs.absent); ventral surfaces creamy-white (vs.pale-green); lateral dark stripe on ventrals absent (vs.lateral dark stripe on ventrals always present, discontinuous, red); three or four small supraocular scales, Fig. 6A (vs.one large supraocular scale, Fig. 6H); small scale between nasal and second supralabial present vs. absent; body color in life (reddish-brown bands on emerald-green background [Fig.5A] vs. red to purple bands on dark bottle-green background [Fig.5D]); white vertebral spots present in both sexes separated by 2-3 dorsal scales vs. white vertebral dots absent in both sexes; iris color (olive-green with faded brown horizontal stripe [Fig.6A] vs. yellowish-brown to gold [Fig.6G]); and tail coloration (dark brown mottled with rusty spots vs. brown with dark purplish-brown crossbars).

Figure 6 .
Figure 6.Comparison of head coloration (left profile and dorsal view of the head) between males of A-B Trimeresurus ciliaris sp.nov.; C-D T. kanburiensis; E-F T. kuiburi; and G-H T. venustus s. str.Photographs by P. Pawangkhanant (A-D), A. Kaosung (E-F, H) and M. Naiduangchan (G).

Figure 7 .
Figure 7. Habitat of Trimeresurus ciliaris sp.nov.A microhabitat of the new species near the Thum Khao Ting Cave, Palian District, Trang Province, Thailand; B holotype ZMMU Re-17661, adult male, in life in situ; C paratype QSMI 1538, adult female, in life in situ; D-E male specimen from Wang Kelian, Perlis State, Malaysia (not collected).Photographs by P. Pawangkhanant (A-C) and Azrul Azrizal Azmy (D-E).
nov. from Perlis State, Malaysia.The study was completed within the frameworks of the research project E-1.2 of the Joint Vietnam-Russia Tropical Science and Technology Research Centre for 2023.We are especially thankful to Gernot Vogel (Germany) for sharing information related to Trimeresurus spp.N.A. Poyarkov gives thanks to the members of MSU HerpLab including Vladislav A. Gorin, Evgeniya N. Solovyeva, Andrei M. Bragin, Alexey V. Trofimets, Nikita S. Klyukin, and Dmitriy V. Akhipov for support and assistance.S. Idiiatullina is thankful to Sunutcha Suntrarachun, Nararat Laopichienpong, and Apinya Longya (QSMI) for providing some important DNA samples and for helping in the lab.We thank Christopher Joldnall for proofreading and linguistic help.We are very grateful to Patrick David (MNHN, France), Anita Malhotra (Bangor University, UK), Zeeshan Mirza (Max Planck Institute for Biology, Germany) and Uwe Fritz (Museum of Zoology, Senckenberg Dresden, Germany) for helpful comments which allowed us the oppurtunity to improve the previous version of the manuscript.This work was supported by the Russian Science Foundation to N.A. Poyarkov (Grant No. 22-14-00037, specimen collection and preservation, molecular and morphological analyses, data analysis), the Unit of Excellence 2023 on Biodiversity and Natural Resources Management, University of Phayao (UoE66001; data analysis), and in part by the Rufford Foundation (Grant No. 39897-1; data analysis).Specimen collection and animal use protocols were approved by the Institutional Ethical Committee of Animal Experimentation of the University of Phayao, Phayao, Thailand (certificate number UP-AE64-02-04-005, issued to C. Suwannapoom) and were strictly compliant with the ethical conditions of the Thailand Animal Welfare Act.Field work, including collection of animals in the field, was authorized by the Institute of Animals for Scientific Purpose Development (IAD), Bangkok, Thailand (permit numbers U1-01205-2558 and UP-AE59-01-04-0022, issued to C. Suwannapoom).

Table 2 .
Sequences and voucher specimens of the genus Trimeresurus and outgroup taxa used in this study.

Table 3 .
Uncorrected p-distances (percentage)between the sequences of cyt b mtDNA gene of species of the subgenus Trimeresurus included in the phylogenetic analyses.

Table 4 .
Morphological data on the type series of Trimeresurus ciliaris sp.nov.Notes: For abbreviations see Materials and methods section.

Table 5 .
Comparison of morphological characters of Trimeresurus ciliaris sp.nov.with T. kanburiensis, T. kuiburi, and T. venustus s. str.Diagnostic differences from the new species are marked in bold.