﻿Two new karst-adapted species in the Cyrtodactyluspulchellus group (Reptilia, Gekkonidae) from southern Thailand

﻿Abstract The exploration of unsurveyed areas in southern Thailand discovered two new karst-adapted species, Cyrtodactylussungaiupesp. nov. and Cyrtodactyluswangkhramensissp. nov., from Thung Wa and La-ngu Districts, Satun Province, respectively. These new species are members of the C.pulchellus group that occur along the Thai-Malay Peninsula. The new species can be distinguished from all other congeners by their key morphological characters and genetic divergence. Morphologically, Cyrtodactylussungaiupesp. nov. and Cyrtodactyluswangkhramensissp. nov. can be diagnosed from other members by having a combination of differences in body size; degree of dorsal tuberculation; absence of tubercles on ventral surfaces; number of ventral scales, paravertebral tubercles and femoroprecloacal pores in males only; deep precloacal groove only in males; absence of a scattered pattern of white dorsal tubercles; number of dark body bands; and the extent of caudal tubercles on an original tail. Although the two species are sister taxa and have nearly identical morphologies, they are considered to be different species, based on a relatively high uncorrected pairwise genetic divergence of the mitochondrial ND2 gene (6.59–6.89%), statistically significant univariate and multivariate morphological differences (PERMANOVA and ANOVA) and diagnostic characteristics of caudal tuberculation on the original tail. Moreover, Cyrtodactylussungaiupesp. nov. and Cyrtodactyluswangkhramensissp. nov. are currently restricted to their karstic type localities which may serve as a geographic barrier to dispersal and gene flow.

During a herpetological survey in Satun Province, southern Thailand, two populations (Thung Wa and La-ngu Districts) of the C. pulchellus group were collected from karst formations.Molecular and morphological studies of these two populations revealed that they differ from currently recognised species and from each other.Herein, the two populations are considered to be new species and are described below.

Sampling
Field surveys were conducted in karst formations in Thung Wa and La-ngu Districts, Satun Province, southern Thailand (Fig. 1; Suppl.material 1).The specimens of the C. pulchellus group were generally observed at night (1900-2100 h) between October 2016 to April 2022.Geographical coordinates and elevations of each locality were recorded using a Garmin GPSMAP 64s.Ambient temperature and relative humidity were recorded using a Kestrel 400 Weather Meter.Ecological data at the time of capture (e.g.time, microhabitat and substrate) were recorded for each specimen.Specimens were euthanised by injecting tricaine methanesulphonate .Liver tissue was removed and preserved in 95% ethyl alcohol and stored at -20 °C for genetic study.Prior to permanent storage in 70% ethyl alcohol, voucher specimens were initially fixed in 10% formalin for morphological study.Voucher specimens were deposited in the herpetological collections of the Zoological Museum, Kasetsart University, Thailand (ZMKU).

Molecular analyses
Total genomic DNA of 14 newly-collected specimens was extracted from ethanol-preserved liver tissues using a NucleoSpin Tissue Kit (Macherey-Nagel GmbH & Co. KG, Germany) with standard manufacturer's protocols.A partial sequence of the mitochondrial NADH dehydrogenase subunit 2 (ND2) gene and flanking tRNAs was amplified using polymerase chain reaction (PCR) under the following conditions: initial denaturation at 94 °C for 4 min, followed by 33-35 cycles of denaturation at 94 °C for 30 sec, annealing at 48-52 °C for 30 sec, extension at 72 °C for 90 sec and final extension at 72 °C for 7 min using the Metf6 (5'-AAGCTTTCGGGCCCATACC-3') and COIH (5'-AGRGTGCCAATGTCTTTGT-GRTT-3') primer pairs, following Macey et al. (1997).All PCR products were purified and sequenced for forward and reverse strands using amplifying primers on ABI 3730XL DNA Sequencer by Sangon Biotech Inc. (Shanghai, China) using BigDye version 3 chemistry (Applied Biosystems, CA, USA).Sequences were edited and manually checked in Geneious R11 (Biomatters Ltd., Auckland, New Zealand).All newly-generated sequences were deposited in GenBank under the accession numbers OR346980-OR346993 (Suppl.material 1).
Maximum Likelihood (ML) and Bayesian Inference (BI) analyses were used to estimate phylogenetic trees.The ND2 dataset was partitioned by codon position for the protein-coding region and the tRNAs were treated as a separate partition.The evolutionary model selected for each partition was determined using ModelFinder with the Bayesian Information Criterion (BIC) as implemented in IQ-TREE (Kalyaanamoorthy et al. 2017).The selected models for ML and BI are listed in Table 1.The ML analysis was performed in the IQ-TREE webserver (Trifinopoulos et al. 2016).The ultrafast bootstrap approximation (UFB; Hoang et al. [2018]) was used to construct a final consensus tree via 1,000 bootstrap replicates.Nodes with UFB values of ≥ 95 were considered to be strongly supported (Minh et al. 2013).The BI analysis was run with MrBayes 3.2.6 on XSEDE (Ronquist et al. 2012) implemented in the CIPRES Science Gateway v.3.3 (Miller et al. 2010) with the models of evolution most closely approximating those calculated for the ML analysis.Two independent runs were performed with four chains per run, three hot and one cold.The Markov Chain Monte Carlo (MCMC) chains were run for 10 million generations and sampled every 1,000 generations, with 25% of each run discarded as burn-in.Stationarity and the effective sample sizes (ESS) for all parameters were assessed in Tracer v. 1.7.1. (Rambaut et al. 2018).Nodes with Bayesian posterior probabilities of ≥ 0.95 were considered strongly supported (Huelsenbeck and Ronquist 2001;Wilcox et al. 2002).Uncorrected pairwise genetic divergences (p-distance) were calculated for both intra-and interspecific using the default settings in MEGA 11 (Tamura et al. 2021).

Morphology
The morphological examination was conducted on preserved specimens of the Thung Wa population, the La-ngu population, C. astrum, C. lekaguli and C. stellatus (Appendix 1).For C. lekaguli, only type series and specimens from the type locality (FMNH, THNHM and ZMKU) were examined.The institutional abbreviations of examined specimens follow Sabaj (2020).All specimens were examined by the first author, except type specimens of C. lekaguli, which was examined by Attapol Rujirawan.Characters and abbreviations were modified from previous studies of the C. pulchellus group (Grismer et al. 2016;Quah et al. 2019;Wood et al. 2020;Termprayoon et al. 2021a), the C. peguensis group (Grismer et al. 2018) and the C. oldhami group (Grismer et al. 2018;Yodthong et al. 2022).Morphometrics were measured using digital calipers to the nearest 0.1 mm on the left side of the body.Sixteen measurements were as follows: SVL Snout-vent length, taken from the tip of snout to the vent; TW Tail width, taken at the base of the tail immediately posterior to the postcloacal swelling; TL Tail length, taken from vent to the tip of the tail, original or regenerated; FL Forearm length, taken from the posterior margin of the elbow while flexed 90° to the inflection of the flexed wrist; TBL Tibia length, taken from the posterior surface of the knee while flexed 90° to the base of the heel; AG Axilla to groin length, taken from the posterior margin of the forelimb at its insertion point on the body to the anterior margin of the hind limb at its insertion point on the body; HL Head length, the distance from the posterior margin of the retroarticular process of the lower jaw to the tip of the snout; HW Head width, measured at the angle of the jaws; HD Head depth, the maximum height of head from the occiput to the throat; ED Eye diameter, the greatest horizontal diameter of the eye-ball; EE Eye to ear distance, measured from the anterior edge of the ear opening to the posterior edge of the eye-ball;

ES
Eye to snout distance, measured from anterior-most margin of the eyeball to the tip of snout; EN Eye to nostril distance, measured from the anterior margin of the eye-ball to the posterior margin of the external nares; IO Inter orbital distance, measured between the anterior edges of the orbit; EL Ear length, the greatest vertical distance of the ear opening;

IN
Internarial distance, measured between the nares across the rostrum.
Meristic characters were evaluated under a Nikon SMZ745 dissecting microscope on both left (L) and right (R) sides, when possible, for the following eighteen characters:

SL
Supralabial scales, counted from the largest scale immediately posterior to the dorsal inflection of the posterior portion of the upper jaw to the rostral scale;

SL-mid-eye
The numbers of supralabial scales, counted from the largest scale immediately below the middle of the eye-ball to the rostral scales; IL Infralabial scales, counted from the largest scale immediately posterior to the dorsal inflection of the posterior portion of the upper jaw to the mental scale; IL-mid-eye The numbers of infralabial scales, counted from the largest scale immediately below the middle of the eye-ball to the mental scales;

PVT
The number of paravertebral tubercles between limb insertions, counted in a straight line immediately left or right of the vertebral column;

LRT
The number of longitudinal rows of body tubercles, counted transversely across the centre of the dorsum from one ventrolateral fold to the other;

VS
The number of longitudinal rows of ventral scales, counted transversely across the centre of the abdomen from one ventrolateral fold to the other;

4FLU
The number of small, unmodified subdigital lamellae distal to the digital inflection on the fourth finger, counted from the digital inflection to the claw;

4FLE
The number of expanded subdigital lamellae proximal to the digital inflection on the fourth finger, counted from the base of the first phalanx where it contacts the body of the hand to the largest scale on the digital inflection;

4FL
The total number of subdigital lamellae beneath the fourth finger;

4TLU
The number of small, unmodified subdigital lamellae distal to the digital inflection on the fourth toe, counted from the digital inflection to the claw;

4TLE
The number of expanded subdigital lamellae proximal to the digital inflection on the fourth toe, counted from the base of the first phalanx where it contacts the body of the foot to the largest scale on the digital inflection;

4TL
The total number of subdigital lamellae beneath the fourth toe, counted from the base of the first phalanx to the claw;

FPP
The total number of precloacal and femoral pores in male (i.e. the sum of the number of femoral and precloacal scales bearing pores combined as a single meristic referred to as the femoroprecloacal pores);

PCT
The number of rows and total number of postcloacal (hemipenial) tubercles in adult male; BB The number of dark body bands between limb insertions;

LCB
The number of light caudal bands on the original tail;

DCB
The number of dark caudal bands on the original tail.
Additional non-meristic characters evaluated were the degree of body tuberculation, weak tuberculation refers to dorsal body tubercles that are low and rounded, whereas prominent tuberculation refers to tubercles that are raise and keeled; the presence or absence of tubercles on the dorsal and ventral surfaces of the forearms; the presence or absence of tubercles in the gular region, throat and ventrolateral body folds; body tubercles extending past the base of the tail or not; the width of the dark body bands relative to the width of the interspace between the bands; the presence or absence of dark pigmentation infused in the white caudal bands of adults; the presence of caudal tubercles; the presence or absence of a precloacal depression or groove; femoroprecloacal pore continuous or not; the presence or absence of scattered white/yellow tubercles on the dorsum; and the presence or absence of white tail tip on the posterior portion of the original tail in hatchlings and juveniles.Colour pattern was taken on dorsal, ventral, lateral image of the body in both sexes and of all possible age classes prior to preservation.
Additional morphological data for analyses and comparisons were obtained from the original descriptions of other species in the C. pulchellus group (Grismer et al. 2012;Quah et al. 2019;Wood et al. 2020;Termprayoon et al. 2021a).
Statistical analyses were conducted using the software R (R Core Team 2019).Tail characters (TL and TW) were excluded due to their different conditions (e.g.original, regenerated or broken).Morphometric characters (except SVL) were size-adjusted in order to eliminate bias of allometry by allometric equation: X adj = log[X ± β(SVL ± SVL mean )], where X adj = adjusted value; X = measured value; β = unstandardised regression coefficient for each species; SVL = measured snout-vent length; SVL mean = overall average SVL of each allometry species (Thorpe 1975(Thorpe , 1983;;Turan 1999;Lleonart et al. 2000)-implemented through the R package GroupStruct (Chan and Grismer 2021).The adjustments are automatically conducted separately on each species and then concatenated into a single data frame to ensure there is no interspecific conflation of variation (Reist 1985;McCoy et al. 2006).
Fourteen morphometrics (SVL, FL adj , TBL adj , AG adj , HL adj , HW adj , HD adj , ED adj , EE adj , ES adj , EN adj , IO adj , EL adj and IN adj ) and seven meristics (SL, IL, PVT, LRT, VS, 4TL and BB) were concatenated to a single dataset.Femoroprecloacal pore counts were excluded from analyses due to their presence in only males.Some meristic characters (SL-mid-eye, IL-mid-eye, 4FLU, 4FLE, 4FL, 4TLU and 4TLE) were omitted due to inadequate data of some species.The dataset was analysed using principal components analysis (PCA) to reduce noise in the dataset and explore interspecific differences in morphospace amongst species using the FactorMineR package (Lê et al. 2008).The first two PC plots were visualised using ggplot2 package (Wickham 2016).The significant differences in centroid locations and group clustering amongst species were determined using a non-parametric permutation-multivariate analysis of variance (PERMANOVA) from the vegan package in R (Oksanen et al. 2020).A Euclidean (dis)similarity matrix using 50,000 permutations was used in the analysis, based on the loadings of the first four dimensions recovered from the PCA.The PCAtest (Camargo 2022) was used to evaluate the overall significance of the PCA dataset, the significance of the PC axes, based on their eigenvalues (i.e.determining which axes contained statistically significant signal) and the contribution of each of the observed variables of the significant axes across the dataset.This removes the partially subjective and ad hoc thresholds used to describe non-trivial PC axes.The function runs 1,000 random permutations and bootstrap replicates of the empirical data.Based on the bootstrap resampling and permutation, 95%-confidence inter- vals around mean values were calculated.Significant p-values imply there is non-random correlational structure in the overall dataset and that the PCA is biologically meaningful.Statistically-significant eigenvalues indicate their respective PC axes reflect non-random correlations amongst variables and statistically significant loadings indicate their respective variables have a larger contribution in the PC score beyond random noise.This test delivers statistically defensible quantitative results by delimiting which PC axes contain signal and opposed to noise and, in so doing, illuminates the most significant PC axes influencing the results of the PERMANOVA.
For statistical comparison, the small sample size of C. dayangbuntingensis (n = 2) was excluded from the univariate analysis.Data of each species were tested for normality using the Shapiro-Wilk test (p ≥ 0.05).Normally distributed data were tested for homogeneity of variances using Levene's test (p ≥ 0.05).Differences amongst species were compared using Analysis of Variance (ANOVA) and Tukey post hoc test (Tukey's test) for normalised and equal variance data or Welch's F-test and Games-Howell post hoc test for unequal variance data.Non-normalised characters were compared using non-parametric Kruskal-Wallis test and followed by a post hoc Dunn's multiple comparison (Dunn's test).

Phylogenetic relationships
The aligned dataset of the partial ND2 gene and flanking tRNAs contained 1,444 bp of 100 individuals.The Maximum Likelihood value of the best ML tree was lnL = -15,495.881.The standard deviation of split frequencies was 0.002877 between the two simultaneous BI runs and the ESS values were ≥ 7594.3 for all parameters.The results of ML and BI phylogenetic analyses revealed two largely concordant topologies across the major nodes, with minor differences in support values.Only the ML tree is represented in this study (Fig. 2).Both analyses recovered the Thung Wa and La-ngu populations as members of the C. pulchellus group and each was embedded within Clade A, comprised of C. astrum, C. dayangbuntingensis, C. langkawiensis, C. lekaguli and C. stellatus.Each population formed a well-supported monophyletic lineage (100 UFB, 1.00 BPP), referred to Thung Wa lineage (Cyrtodactylus sungaiupe sp.nov.[see below]) and La-ngu lineage (Cyrtodactylus wangkhramensis sp.nov.[see below]).These two lineages were strongly supported (98 UFB, 1.00 BPP) as sister taxa.

Morphology
The PCA analysis of the seven lineages in Clade A recovered morphological differences along the ordination of the first two PC axes (Fig. 3).The PC plot showed that Cyrtodactylus sungaiupe sp.nov.clustered separately from C. astrum, C. dayangbuntingensis, C. langkawiensis and Cyrtodactylus wangkhramensis sp.nov.and only slightly overlapped with C. lekaguli.Cyrtodactylus wangkhramensis sp.nov.also clustered separately from all other species, except C. stellatus (partially overlapped).The first three components with eigenvalues > 1.00 (PC1-PC3) accounted for 53.30% of all variables and mostly loaded on morphometric characters (Table 3).PC1 explained 28.59% of the variation and was heavily loaded on FL adj , TBL adj , HW adj , HD adj , EE adj , ES adj and ENadj .PC2 accounted for 14.97% of the variation and was heavily loaded on AG adj and IN adj .PC3 accounted for 9.74% of the variation and was heavily loaded on EL adj .The PERMANOVA and PCAtest analyses indicated that the PCA contained a highly significant diagnostic signal amongst the lineages.PERMANOVA recovered the morphological clustering and centroid placement of Cyrtodactylus sungaiupe sp.nov.and Cyrtodactylus wangkhramensis sp.nov.as significantly different (p < 0.001-0.036;Table 4) from each other and all other species in Clade A. PCAtest recovered PC1-3 as containing statistically different signal from the other 18 axes (p = 0.000 for all axes).Along PC1, all characters, except AG adj , EL adj , IN adj and IL had statistically significant different loadings as opposed to the loadings of those same characters on the other axes (Fig. 4; Table 3).Statistically significant loadings for characters AG adj , IO adj , IN adj , PVT and VS were found along PC2 and along PC3, SL, PVT and LRT were statistically different from their respective loads on the other PC axes (Fig. 4; Table 3).The significant differences recovered by the ANOVA (or Kruskal-Wallis test) amongst seven lineages in Clade A were found in most characters (p < 0.001-0.008),except 4TL and BB (p = 0.055-0.126)(Table 5; Suppl.material 2).The pair-

Taxonomic hypotheses
According to the concordant results of the phylogenetic analyses, PCA, ANOVA, PERMANOVA, PCAtest and diagnostic morphological characters (see "Comparison"), the populations from Thung Wa and La-ngu Districts, Satun Province are distinctly separated from other species of the C. pulchellus group and each other.We, therefore, hypothesise that they represented distinct unnamed species and are described below.Description of holotype.Adult male SVL; 89.7 mm; head moderate in length (HL/SVL 0.29) and wide (HW/HL 0.65), flattened (HD/HL 0.37), distinct from neck and triangular in dorsal profile; lores concave anteriorly, inflated posteriorly; frontal and prefrontal regions concave; canthus rostralis rounded anteriorly; snout elongated (ES/HL 0.40), rounded in dorsal profile, laterally constricted; eye large (ED/HL 0.22); ear opening elliptical, moderate in size (EL/HL 0.07), obliquely orientated; eye to ear distance slightly greater than diameter of eye; rostral rectangular, divided dorsally by an inverted Y-shaped furrow, bordered posteriorly by left and right supranasals and internasal, bordered laterally by first supralabials; external nares bordered anteriorly by rostral, dorsally by a large anterior supranasal, posteriorly by two postnasals, ventrally by first supralabial; 9/10 (left/right) rectangular supralabials extending to below mid-point of eye, 13/15 to below the posterior margin of the eye-ball, decreasing abruptly just posterior to mid-point of eye; 8/7 infralabials extending to below mid-point of eye, 13/11 to upturn the labial margin, decreasing gradually in size posteriorly; scales of rostrum and lores slightly raised, larger than granular scales on top of head and occiput, those on posterior portion of canthus rostralis slightly larger; scales on top of head and occiput intermixed with rounded, small tubercles; dorsal superciliaries elongate, smooth, largest anteriorly; mental triangular, 2.5 mm in width, 3.3 mm in length, bordered laterally by first infralabials and posteriorly by left and right, trapezoidal postmentals which contact medially for approximately 50% of their length; one row of slightly enlarged, elongate sublabials extending posteriorly to the sixth (left) and seventh (right) infralabials; small, granular, gular scales grading posteriorly into larger, flat, smooth, imbricate, pectoral and ventral scales.
Body relatively short (AG/SVL 0.47) with well-defined, non-tuberculate, ventrolateral folds; dorsal scales small, granular, interspersed with low, regularly arranged, weakly-keeled tubercles, smaller intervening tubercles occasionally present; tubercles extend from occiput to base of tail, but not further than 1/10 of tail; tubercles on occiput and nape relatively small, those on body largest; approximately 19 longitudinal rows of tubercles at mid-body; 36 paravertebral tubercles; 37 flat imbricate ventral scales between ventrolateral body folds; ventral scales larger than dorsal scales; precloacal scales large, smooth; deep precloacal groove.
Forelimbs moderately slender, relatively short (FL/SVL 0.17); dorsal scales on forelimbs raised, granular, larger than those on body; dorsal scales on forearm intermixed with enlarged, subconical and weakly-keeled tubercles, brachium without tubercles; scales of ventral surface of forearm flat, subimbricate, tubercles absent; palmar scales small, weakly rounded; digits well-developed, inflected at basal, interphalangeal joints; 19/19 (left/right) subdigital lamellae on the fourth finger, 6/6 proximal subdigital lamellae rectangular, broadly expanded proximal to joint inflection, 13/13 distal subdigital lamellae slightly expanded distal to inflection becoming gradually more expanded near the claw; claws well-developed, sheathed by a dorsal and ventral scale; hind limbs more robust than forelimbs, moderate in length (TBL/SVL 0.20), enlarged, subconical, weakly-keeled tubercles on dorsal surface of legs separated by smaller juxtaposed scales; ventral scales of thigh flat, smooth, imbricate, larger than dorsal granular scales; ventral, tibial scales flat, smooth, imbricate; a single row of 35 enlarged femoroprecloacal scales extending nearly from knee to knee through precloacal region where they are continuous with enlarged, pore-bearing precloacal scales; 33 contiguous pore-bearing femoroprecloacal scales, forming an inverted T bearing a deep, precloacal groove (Fig. 6D); six pore-bearing scales bordering groove (three on each side); postfemoral scales immediately posterior to enlarged scale row small, nearly granular, forming an abrupt union with postfemoral scales on posteroventral margin of thigh; plantar scales weakly rounded to flat; 20/20 (left/right) subdigital lamellae on fourth toe, 7/7 proximal subdigital lamellae rectangular, broadly expanded proximal to joint inflection, 13/13 distal subdigital lamellae slightly expanded distal to inflection becoming gradually more expanded near the claw; claws well-developed, sheathed by a dorsal and ventral scale.
Original tail 122.5 mm in length, slightly longer than SVL (TL/SVL = 1.37), 6.4 mm in width at base, tapering to a point; dorsal scales of tail flat, squarish; original portion segmented, approximately 7-8 transverse scales rows per segment; one transverse row of four dorsal tubercles on posterior margin of 1 st segment and one tubercle on 2 nd segment; caudal tubercles extended 1/10 of anterior portion of original tail (Fig. 6E); subcaudal region bearing large median row of transverse scales; shallow dorsal and lateral caudal furrow; base of tail bearing hemipenial swellings; two rows of (1+4)L/(2+3)R medium-sized postcloacal tubercles on each hemipenial swelling; postcloacal scales smooth, flat, large, imbricate.
Colouration in life (Fig. 5).Ground colour of head, body, and limbs light-brown; superciliaries yellow anteriorly and posteriorly; supralabial and infralabial scales light-brown with off-white markings posteriorly; wide, dark-brown nuchal band edged anteriorly and posteriorly by thin, yellowish lines bearing tubercles extending from posterior margin of one eye to posterior margin of another eye; four similar dark-brown body bands between nuchal loop and hind limb insertions edged anteriorly and posteriorly by broken, thin, creamy-white to yellow lines bearing tubercles, first band terminating at shoulders, second and third bands terminating just dorsal to ventrolateral folds, the fourth band terminating at femurs; dark body bands slightly larger than light-coloured interspaces; one additional dark-brown band posterior to hind limbs (postsacral band); ventral surfaces of head, abdomen and limbs greyish-white; creamy pale yellow postcloacal tubercles; tail bearing nine dark bands separated by ten light-brown (anteriorly) to white (posteriorly) bands, white caudal band infused with dark pigmentation; subcaudal region off-white anteriorly, becoming darker posteriorly.
Colouration in preservative (Fig. 6).The overall colour pattern of head, body, limbs and tail similar to that in life with some fading.Ground colour of head, body, limbs and dorsum tan; dark bands on dorsum and tail brown; yellow-coloured tuberculation on dorsum fading to off-white or white; light-beige coloured on the ventral surface.
Variation.Meristic and morphometric data for the type series and referred specimens of Cyrtodactylus sungaiupe sp.nov.are given in Tables 6, 7 and Suppl.material 3.All paratypes and referred specimens resemble the holotype in general aspects of morphology with variations in colouration and banding pattern.All specimens have continuous dark body bands, except one specimen (ZMKU R 01010) where the second to fourth dorsal bands do not connect on the mid-line and are offset (Fig. 7A).Dorsal scale on brachium of two specimens (ZMKU R 01010 and ZMKU R 01012) intermixed with tubercles, whereas absent in the holotype and other specimens.ZMKU R 01005 has dark marking on left femur.Three male paratypes have discontinuous pore-bearing femoroprecloacal scales; ZMKU R 01004 has three poreless scales (one on the left and two on the right), ZMKU R 01005 has one enlarged and one small poreless scales on the left and ZMKU R 01010 has two poreless scales (one on each side).Last row of caudal tubercles of paratypes (ZMKU R 01006 and ZMKU R 01008) extending to 4 th -5 th segment of original portion (to 2 nd light caudal band), approximately 1/8 of anterior portion of original tail.Juveniles (ZMKU R 01013-01014) have body pattern similar to adults, but less prominent tuberculation, light-yellow ground colour of body, edged anteriorly and posteriorly by yellowish lines bearing tubercles, the original tail has approximately six or ten dark caudal bands, the posterior portion of tail is white (Fig. 7B).Table 6.Descriptive measurements (millimetres), meristic (left/right) and non-meristic characters of the type series of Cyrtodactylus sungaiupe sp.nov.Key: H = holotype, P = paratype, M = male, F = female; Or = original tail, Br = broken, Re = regenerated; / = data unavailable or inapplicable; L = left, R = right.Abbreviations are defined in Materials and methods.to two rows of postcloacal tubercles (vs. up to three rows).Additionally, PCA showed that Cyrtodactylus sungaiupe sp.nov. is clearly separated in morphospace from C. dayangbuntingensis (Fig. 3).Cyrtodactylus sungaiupe sp.nov.morphologically differ from C. langkawiensis by having larger maximum SVL of 104.6 mm (vs.99.8 mm); and caudal tubercles extending between 1/8 and 1/10 of anterior portion of tail (vs.≥ 1/3 of the tail); and having statistically significant different mean values of mensural characters of AG adj , HL adj , HW adj and IN adj (p < 0.001-0.033;Table 5).It can be further separated from C. langkawiensis in having statistically significant different mean values of meristic characters of SL, IL, PVT, LRT and VS (p < 0.001-0.046;Table 5).
Cyrtodactylus sungaiupe sp.nov.differ from C. lekaguli by having smaller maximum SVL of 104.6 mm (vs.108.5 mm); and caudal tubercles extending between 1/8 and 1/10 of anterior portion of tail (vs.≥ 1/3 of the tail); and having statistically significant different mean values of mensural characters of HW adj , HD adj , EE adj , IO adj and EL adj (p < 0.001-0.008;Table 5).It can be further separated from C. lekaguli in having statistically significant different mean values of meristic characters of SL, IL and LRT (p < 0.001-0.006;Table 5).
Cyrtodactylus sungaiupe sp.nov.differ from C. stellatus by having larger maximum SVL of 104.6 mm (vs.96.1 mm); absence of scattered pattern of white tubercles on dorsum (vs.present); absence of precloacal in female (vs.present); and having statistically significant different mean values of mensural characters of SVL, FL adj , TBL adj , AG adj , HL adj , HW adj , HD adj , EE adj , ES adj and EN adj (p < 0.001-0.035;Table 5).It can be further separated from C. stellatus in having statistically significant different mean values of meristic characters of PVT (p = 0.003; Table 5).Description of holotype.Adult male SVL; 94.2 mm; head moderate in length (HL/SVL 0.28) and wide (HW/HL 0.63), flattened (HD/HL 0.38), distinct from neck and triangular in dorsal profile; lores concave anteriorly, inflated posteriorly; frontal and prefrontal regions deeply concave; canthus rostralis rounded anteriorly; snout elongate (ES/HL 0.41), rounded in dorsal profile, laterally constricted; eye large (ED/HL 0.23); ear opening elliptical, moderate in size (EL/HL 0.08), obliquely orientated; eye to ear distance slightly greater than diameter of eye; rostral rectangular, divided dorsally by an inverted U-shaped furrow, bordered posteriorly by left and right supranasals and large hexagonal internasal, bordered laterally by first supralabials; external nares bordered anteriorly by rostral, dorsally by a large anterior supranasal, posteriorly by two postnasals, ventrally by first supralabial; 8/8 (left/right) rectangular supralabials extending to below mid-point of eye, 12/12 to below the posterior margin of the eye-ball, decreasing abruptly just posterior to mid-point of eye; 7/7 infralabials extending to below mid-point of eye, 11/11 to upturn the labial margin, decreasing gradually in size posteriorly; scales of rostrum and lores slightly raised, larger than granular scales on top of head and occiput, those on posterior portion of canthus rostralis slightly larger; scales on of head and occiput intermixed with rounded, small tubercles; dorsal superciliaries elongate, smooth, largest anteriorly; mental triangular, 3.60 mm in width, 2.38 mm in length, bordered laterally by first infralabials and posteriorly by left and right, trapezoidal postmentals which contact medially for approximately 45% of their length; one row of slightly enlarged, elongate sublabials extending posteriorly to the seventh (left/ right) infralabials; small, granular, gular scales grading posteriorly into larger, flat, smooth, imbricate, pectoral and ventral scales.

Cyrtodactylus wangkhramensis
Body relatively short (AG/SVL 0.47) with well-defined, non-tuberculate, ventrolateral folds; dorsal scales small, granular, interspersed with low, regularly arranged, weakly-keeled tubercles, smaller intervening tubercles occasionally present; tubercles extending from occiput to base of tail, but not further than 1/5 of tail; tubercles on occiput and nape relatively small, those on body largest; approximately 20 longitudinal rows of tubercles at mid-body; 32 paravertebral tubercles; 35 flat imbricate ventral scales between ventrolateral body folds; ventral scales larger than dorsal scales; precloacal scales large, smooth; deep precloacal groove.
Forelimbs moderately slender, relatively short (FL/SVL 0.17); dorsal scales on forelimbs slightly raised, granular, slightly larger than those on body intermixed with enlarged, subconical and weakly-keeled tubercles; scales of ventral surface of forearm flat, subimbricate, tubercles absent; palmar scales small, weakly rounded; digits well-developed, inflected at basal, interphalangeal joints; 17/17 (left/right) subdigital lamellae on the fourth finger, 6/6 proximal subdigital lamellae rectangular, broadly expanded proximal to joint inflection, 11/11 distal subdigital lamellae slightly expanded distal to inflection becoming gradually more expanded near the claw; claws well-developed, sheathed by a dorsal and ventral scale; hind limbs more robust than forelimbs, moderate in length (TBL/SVL 0.20), enlarged, subconical, weakly-keeled tubercles on dorsal surface of legs separated by smaller juxtaposed scales; ventral scales of thigh flat, smooth, imbricate, larger than dorsal granular scales; ventral, tibial scales flat, smooth, imbricate; a single row of 34 enlarged femoroprecloacal scales extending nearly from knee to knee through precloacal region where they are continuous with enlarged, pore-bearing precloacal scales; 32 contiguous pore-bearing femoroprecloacal scales (Fig. 10D), forming an inverted T bearing a deep, precloacal groove; seven pore-bearing scales bordering groove (four on left and three on right); postfemoral scales immediately posterior to enlarged scale row small, nearly granular, forming an abrupt union with postfemoral scales on posteroventral margin of thigh; plantar scales weakly rounded to flat; 18/18 subdigital lamellae on fourth toe, 7/7 proximal subdigital lamellae rectangular, broadly expanded proximal to joint inflection, 11/11 distal subdigital lamellae slightly expanded distal to inflection becoming gradually more expanded near the claw; claws well-developed, sheathed by dorsal and ventral scales.
Regenerated tail 120.3 mm in length, slightly longer than SVL (TL/SVL = 1.28), 5.4 mm in width at base, tapering to a point; dorsal scales of tail flat, squarish; original portion segmented, 7-9 transverse scales rows per segment; seven transverse rows of 2-6 dorsal tubercles on posterior margin, fewer posteriorly; subcaudal region bearing large median row of transverse scales; regenerated portion of tail covered with small, smooth, rectangular scales dorsally and ventrally; shallow dorsal and lateral caudal furrow extending entire length of original tail; base of tail bearing hemipenial swellings; two rows of (3+2)L/ (1+3)R medium-sized postcloacal tubercles on each hemipenial swelling; postcloacal scales smooth, flat, large, imbricate.
Colouration in life (Fig. 9).Ground colour of body, limbs, light-brown, those on head lighter; faded, incomplete V-shaped rostral chevron; superciliaries yellow; supralabial and infralabial scales light-brown anteriorly, with off-white-coloured medially to posteriorly; wide, dark-brown nuchal band edged anteriorly and pos- teriorly by thin, white lines bearing tubercles extending from posterior margin of one eye to posterior margin of another eye; four dark-brown body bands between nuchal loop and hind limb insertions edged anteriorly and posteriorly by broken, thin, white lines formed by a single row of white tubercles, first band terminating at shoulders, second and third bands terminating just dorsal to ventrolater-al folds, the fourth band terminating at femurs; dark body bands slightly larger than light-coloured interspaces; creamy pale yellow tubercles on dorsal surfaces of limbs, those on body darker, similar to ground colour; one additional darkbrown band posterior to hind limbs (postsacral band); ventral surfaces of head, abdomen and limbs greyish-white; creamy pale yellow postcloacal tubercles; tail bearing approximately six dark bands separated by six light-brown (anteriorly) to white (posteriorly) bands, white caudal band heavily infused with dark-brown pigmentation on original portion; regenerated portion and subcaudal region tan.
Colouration in preservative (Fig. 10).The overall colour pattern of head, body, limbs and tail similar to that in life with some fading.Ground colour of head, body, limbs and dorsum dark tan; dark body bands darker; creamy pale yellow-coloured tuberculation on dorsum fading to off-white; beige coloured on the ventral surface.
Variation.Meristic and morphometric data for the type series of Cyrtodactylus wangkhramensis sp.nov.are given in Table 8 and Suppl.material 3.All paratypes resemble the holotype in general aspects of morphology with variations in colouration and banding pattern.Eight specimens have four dark body bands and three specimens (ZMKU R 01016-01017 and ZMKU R 01021) have five dark bands on dorsum (Fig. 11A).Female paratype (ZMKU R 01025) has four dark body bands with an irregular pattern on the third (Fig. 11B).Original tails (ZMKU R 01019-01021) have 11 dark caudal bands and 10-11 light caudal bands.Last row of caudal tubercles of paratypes (ZMKU R 01019-01021) extending to 4 th -7 th segment of original portion, approximately 1/5-1/7 of anterior portion of original tail.
Natural history.Type series of Cyrtodactylus wangkhramensis sp.nov.was collected from a karst formation at night (1900-2100 h) during March 2019.Eleven specimens were captured from substrates surrounding the karstic area (karst wall, crevices, karst boulders, plantation and concrete poles).Ambient temperature was 25.7 °C and relative humidity was 90.7%.The male holotype (ZMKU R 01018) was found on a termite mound adjacent to a karst outcrop.Five individuals (ZMKU R 01016-01017, ZMKU R 01020-01021 and ZMKU R 01023) were found on a karst wall.One specimen ZMKU R 01015 was found perched on a vine near a karst wall.ZMKU R 01019 was collected from within a karst crevice and ZMKU R 01022 was found on a karst boulder.Two specimens (ZMKU R 01024 and ZMKU R 01026) were perched on concrete poles.
One individual (ZMKU R 01025) was found on a vine in a cave, approximately 10 m from the entrance, where the temperature was 26.1 °C and the relative humidity was 97.9%.Cyrtodactylus wangkhramensis sp.nov.appears to be sympatric with Gehyra mutilata (Wiegmann, 1834).
Etymology.5).It can be differentiated from C. lekaguli in having statistically significant different mean values of meristic characters of SL, PVT and LRT (p = 0.001-0.036;Table 5).Cyrtodactylus wangkhramensis sp.nov.differs from C. stellatus by having 32-36 femoroprecloacal pores (vs.24-29); absence of precloacal pores in females (vs.present); absence of scattered pattern of white tubercles on dorsum (vs.present) and having statistically significant different mean values of mensural characters of AG adj , HL adj , HW adj and HD adj (p < 0.001-0.048;Table 5).It can be further separated from C. stellatus in having statistically significant different mean values of meristic characters of PVT (p < 0.001; Table 5).

Discussion
This study discovered two unrecognised Cyrtodactylus species described here as Cyrtodactylus sungaiupe sp.nov.and Cyrtodactylus wangkhramensis sp.nov.from karst areas in southern Thailand.It is noteworthy that Cyrtodactylus sungaiupe sp.nov.and Cyrtodactylus wangkhramensis sp.nov.are closely related to insular sister species C. dayangbuntingensis and C. langkawiensis from the nearby offshore landbridge Dayang Bunting and Langkawi Islands, respectively.The divergence of these sister clades (two new species and insular species) could have occurred during the Last Gracial Maximum when sea levels were lower than present-day exposing the land-mass connecting mainland and offshore islands (Sathiamurthy and Voris 2006).The higher post-glacial sea level of present day creates a geographic barrier separating these sister clades.Curiously, based on geographic proximity and habitat continuity, one would expect Cyrtodactylus sungaiupe sp.nov.and Cyrtodactylus wangkhramensis sp.nov. to be more closely related to C. stellatus from Tarutao Island or C. astrum from the Thai-Malay border (Fig. 1).The short internodal branch lengths separating all the species of Clade A (Fig. 2) attests to the near simultaneous and somewhat stochastic divergence of these species from one another.This similar phylogeographic pattern is also found in other taxa, such as Cnemaspis in Southeast Asia (Grismer et al. 2014b;Wood et al. 2017;Ampai et al. 2022).
The two newly-recognised species, Cyrtodactylus sungaiupe sp.nov.and Cyrtodactylus wangkhramensis sp.nov.resemble each other in morphology owing to their late and rapid divergence from one another and, thus, share a number of putative ancestral traits (e.g.Sheridan and Stuart [2018]).Moreover, they occur in the same karstic habitat where the same environmental pressures strongly influence their morphology and, thus, selection for change is presumably low (Grismer et al. 2020).Karsts are unique and usually occur as isolated landscapes which provided numerous suitable microhabitats that can be drivers of speciation (Grismer et al. 2021b) and, moreover, act as geographic barriers limiting the distributional range of organisms and, thus, usually bear high levels of endemism (Clements et al. 2006;Grismer et al. 2021b).
The numerous studies and explorations of unsurveyed areas, including offshore islands, mountain forests and karst outcrops and the use of integrative approaches in taxonomy have influenced the recent rise in gekkonid diversity in Thailand in the last decade (e.g.Ampai et al. (2022); Chomdej et al. (2022); Grismer et al. (2022a); Rujirawan et al. (2022); Yodthong et al. [2022]).This is strikingly clear for the C. pulchellus group.The two additional karst-associated species described here brings the number of species in the C. pulchellus species group from 17 to 19, of which six have been recognised in the Thai portion of these complexes' distribution.Furthermore, it brings the number of Cyrtodactylus species in Thailand to 47 (Termprayoon et al. 2021a;Chomdej et al. 2022;Grismer et al. 2022aGrismer et al. , 2023;;Yodthong et al. 2022).The fact that the species diversity of the C. pulchellus group was underestimated by 19 fold as late as 2008 is a testament to the necessity of statistically and genetic-based integrative analyses.These analyses, coupled with descriptions of each species microhabitat will engender useful species-specific conservation management plans.
Currently, there are still numerous nearby karstic regions awaiting surveys which are needed to delimit the geographic ranges of any new and currently-recognised species.Further taxonomic studies using multi-locus genetic data are also needed to delimit species boundaries within other species groups throughout Thailand.

Figure 1 .
Figure 1.Map showing the type localities of Cyrtodactylus sungaiupe sp.nov.(red star) in Thung Wa District and Cyrtodactylus wangkhramensis sp.nov.(blue star) in La-ngu District, Satun Province, Thailand and the type localities of closely-related species, C. astrum, C. dayangbuntingensis, C. langkawiensis, C. lekaguli and C. stellatus.Circles represent additional localities of specimens used in molecular analyses.

Figure 2 .
Figure 2. Maximum Likelihood phylogenetic tree of Cyrtodactylus sungaiupe sp.nov.and Cyrtodactylus wangkhramensis sp.nov.within the C. pulchellus group reconstructed from 1,444 bp of ND2 and flanking tRNAs.Nodes show ultrafast bootstrap support (UFB) and Bayesian posterior probabilities (BPP) values.

Figure 3 .
Figure 3. Principal Component Analysis (PCA) of Cyrtodactylus sungaiupe sp.nov., Cyrtodactylus wangkhramensis sp.nov.and the closely-related species in Clade A of the C. pulchellus group.The letters in the scatter plots refer to holotype (= H), paratype (= P) and topotype (= T).

Figure 4 .
Figure 4. Statistically significant variation of the first three PCs based the permutation (grey) and bootstrap (red) analyses conducted in PCAtest.

Figure 5 .
Figure 5. Adult male holotype of Cyrtodactylus sungaiupe sp.nov.(ZMKU R 01009) from the type locality in Thung Wa Subdistrict, Thung Wa District, Satun Province, Thailand A dorsolateral view of specimen in life B dorsal and ventral views immediately after euthanasia.

Figure 6 .
Figure 6.Adult male holotype of Cyrtodactylus sungaiupe sp.nov.(ZMKU R 01009) in preservation A dorsal and ventral views of body B head dimensions showing lateral, dorsal and ventral views C ventrolateral fold on the left side D precloacal depression and pore-bearing femoroprecloacal scales E dorsal view of tail showing last row of tubercles.

Figure 7 .
Figure 7. Variation in colouration and banding pattern of Cyrtodactylus sungaiupe sp.nov.immediately after euthanasia A male paratype (ZMKU R 01010) showing disconnected bands on dorsal B juvenile (ZMKU R 01013) having light-yellow on the body and bearing white tail tip.

Figure 8 .
Figure 8. Habitat of Cyrtodactylus sungaiupe sp.nov. at the type locality in Thung Wa Subdistrict, Thung Wa District, Satun Province, Thailand A landscape view of karst tower B vegetation structure and C karst boulders surrounding karst body D immature female (not collected) on karst boulder.

Figure 9 .
Figure 9. Adult male holotype of Cyrtodactylus wangkhramensis sp.nov.(ZMKU R 01018) from the type locality at Tham Wangkhram, La-ngu District, Satun Province, Thailand A dorsolateral view of specimen in life B dorsal and ventral views immediately after euthanasia.

Figure 10 .
Figure 10.Adult male holotype of Cyrtodactylus wangkhramensis sp.nov.(ZMKU R 01018) in preservation A dorsal and ventral views of body B head dimensions showing lateral, dorsal and ventral views C ventrolateral fold on the left side D precloacal depression and pore-bearing femoroprecloacal scales E dorsal view of tail showing last row tubercles.

Figure 11 .
Figure 11.Variation in banding pattern of Cyrtodactylus wangkhramensis sp.nov.immediately after euthanasia A adult male (ZMKU R 01016) having five dark body bands on dorsum B adult male (ZMKU R 01026) showing irregular pattern on the 3 rd body band.

Table 1 .
The evolutionary model selected of partitioning ND2 gene and tRNAs estimated by BIC implemented in IQ-TREE.Selected models were applied for the Maximum Likelihood (ML) and Bayesian Inference (BI) analyses.

Table 2 .
Percentage uncorrected pairwise sequence divergences (p-distances) of Cyrtodactylus sungaiupe sp.nov., Cyrtodactylus wangkhramensis sp.nov.and closely-related species (Clade A), based on 1,444 base pairs of mitochondrial ND2 gene and flanking tRNAs.The p-distance values are given as mean and ranges in parentheses.Intraspecific distances are in bold font.

Table 3 .
Summary statistics and factor loadings of the first three principal components (PC1-3) of morphometric and meristic characters of Cyrtodactylus sungaiupe sp.nov., Cyrtodactylus wangkhramensis sp.nov.and closely-related species including C. astrum, C. dayangbuntingensis, C. langkawiensis, C. lekaguli and C. stellatus.Bold fonts indicate high loadings.Abbreviations are listed in Materials and methods.Cyrtodactylus wangkhramensis sp.nov. in four mensural characters of AG adj , HW adj , EE adj and EN adj (p < 0.001-0.009)and a meristic character of SL (p = 0.045).The mean values of significant characters were greater in the Cyrtodactylus sungaiupe sp.nov.than Cyrtodactylus wangkhramensis sp.nov.(Table 5; Suppl.material 2).The summarised pairwise differences between Cyrtodactylus sungaiupe sp.nov., Cyrtodactylus wangkhramensis sp.nov.and closely-related species in Clade A are shown in Table 5 and Suppl.material 2.
wise comparison post hoc tests showed that Cyrtodactylus sungaiupe sp.nov.was significantly different from

Table 4 .
Summary results of the PERMANOVA analysis of Cyrtodactylus sungaiupe sp.nov., Cyrtodactylus wangkhramensis sp.nov.and closely-related species in Clade A. Bold fonts indicate significant differences.

Table 5 .
Summary pairwise differences of statistically significant characters (Tukey's test; p < 0.05) from morphometric and meristic characters of Cyrtodactylus sungaiupe sp.nov., Cyrtodactylus wangkhramensis sp.nov.and closely-related species (Clade A).Abbreviations are listed in Materials and Methods.Key: * tested by Games-Howell test; ** tested by Dunn's test.No significantly different characters of 4TL and BB were excluded.

Table 7 .
Descriptive meristic (left/right) and non-meristic characters of referred specimens of Cyrtodactylus sungaiupe sp.nov.Key: RF = referred specimens, IM-M = immature male, J = juvenile; / = data unavailable or inapplicable.Abbreviations are defined in Materials and methods.