Phylogenetic and multivariate analyses of Gekko smithii Gray, 1842 recover a new species from Peninsular Malay sia and support the resurrection of G. albo

An integrative taxonomic analysis of Sundaic populations of Gekko smithii from the Thai-Malaya Peninsula, Sumatra, and Borneo recovered four deeply divergent mitochondrial lineages that are separated by major geographic barriers (mountains and seaways). Furthermore, they bear a number of concordant statistically significant differences in meristic and morphometric features, morphospatial separation in multivariate space, and discrete differences in color pattern. Gekko smithii sensu stricto is restricted to southern Thailand south of the Isthmus of Kra and Peninsular Malaysia west of the Banjaran (mountain range) Titiwangsa, being that the type locality is on Penang Island, Penang. Gekko hulk sp. nov. is a new species from extreme southern Thailand and Peninsular Malaysia east of the Banjaran Titiwangsa and five east coast islands—the type locality being Pulau (island) Tioman, Pahang. Gekko cf. albofasciolatus is tentatively used to include Bornean populations west of the Iran Mountains in Sabah and Sarawak which, in the absence of molecular data, cannot unequivocally be separated morphologically from G. albofasciolatus from the type locality at Banjarmasin, Kalimantan, Indonesia east of the Iran Mountains. In the absence of molecular data, G. albomaculatus is resurrected to include mainland Sumatran, Nias Island, and Banyak Islands populations which, based on their morphology, cannot be separated from descriptions of G. albomaculatus from the type locality of Bangka Island, 15 km off the southeast coast of mainland Sumatra. Further integrative analyses of all Sumatran and Bornean populations are currently underway as well as the enigmatic Wallacean populations from Sulawesi. Data are presented that strongly suggest all references to G. smithii from Java stem from a 151 year-old misidentification of a specimen of G. gecko of unknown provenance. Additionally, there are no vouchered records of G. smithii from Myanmar. The phylogeographic patterns of Sundaic populations of the G. smithii complex are concordant with those of a plethora of other Sundaic lineages. Vertebrate Zoology 72, 2022, 47–80 | DOI 10.3897/vz.72.e77702 Copyright L. Lee Grismer et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. L. Lee Grismer et al.: Phylogenetic taxonomy of the Gekko smithii complex 48


Introduction
The genus Gekko Laurenti, 1768 is an ecologically and morphologically diverse radiation of scansorial, nocturnal lizards comprising at least 82 species that collectively range throughout Southeast and East Asia to western Melanesia (Uetz et al. 2021;Wood et al. 2020a). Throughout its vast distribution, Gekko manifests a broad array of adaptive types ranging from large human commensals and granite boulder-adapted specialists, to highly cryptic arboreal species and parachuters Wood et al. 2020a). Based on an extensive integrative analysis, Rösler et al. (2011) constructed six phenotypic species groups within the genus Gekko. Two of these groups, G. gecko and G. japonicus, were recognized by Wood et al. (2020a) in a phylogenomic taxonomy that constructed seven subgenera and necessitated the synonymy of Luperosaurus Gray, 1845 and Ptychozoon Kuhl and van Hasselt, 1822 with Gekko. Included among the seven species of the subgenus Gekko, is G.
smithii Gray, 1942, a large widespread, aggressive, scansorial, lowland, forest-adapted species and semi-human commensal found throughout much of Sundaland from southern Thailand south of the Isthmus of Kra through Peninsular Malaysia and Singapore to Sumatra, Borneo, and Sulawesi (Koch et al. 2009;Grismer 2011a,b) (Fig. 1). Chandramouli et al. (2021) indicated that G. smithii from Great Nicobar Island (erroneously reported as G. gecko by Biswas and Sanyal (1977)) was morphologically distinct and based on statistically significant differences, described it as G. stoliczkai Chandramouli, Gokulakrishnan, Sivaperuman, and Grismer, 2021 which we include here to be the eighth species of G. (Gekko). Reports of G. smithii from Java and Myanmar are likely erroneous (see below).
Integrative taxonomic analyses of many other species of Sundaic amphibians and reptiles were recovered as species complexes and whose pronounced phylogeo- Figure 1. Distribution of the species of the Gekko smithii complex from throughout their respective Sundaic land masses. Circles denote the localities of specimens that were both examined first hand and are represented in the phylogeny (Fig. 2). Stars denote type localities. Squares denote localities of specimens or photographs examined here or verified from other publications as well as vouchered samples in the literature. Localities from non-peer reviewed literature or unverifiable online data were not included. graphic sub-structuring across well-defined geographic features necessitated their taxonomic partitioning into multiple species (e.g., Grismer et al. 2013Grismer et al. , 2014aGrismer et al. ,b, 2015Grismer et al. , 2018aGrismer et al. ,b, 2019Loredo et al. 2013;Matsui et al. 2014;Matsui 2019;Chan et al. 2016Chan et al. , 2017Chan et al. , 2018Chan et al. , 2020Harvey et al. 2016;Grismer and Davis 2018;O'Connell et al. 2018O'Connell et al. , 2019Quah et al. 2019Quah et al. , 2020Quah et al. , 2021aWood et al. 2020b). Therefore, we initiated the first integrative analysis of G. smithii based on morphology, color pattern, and genetic data derived from the mitochondrial gene NADH dehydrogenase subunit 2 (ND2). The analysis was focused primarily on the phylogeographic structure of populations on the Thai-Malay Peninsula and their associated islands-including the type locality on the west coast island of Penang that has remained unsampled since Gray's description in 1842. Although sampling was dense on the Thai-Malay Peninsula, we sequenced available tissue samples and examined representative specimens from Borneo and Sumatra for comparison. The phylogenetic analyses recovered significant mitochondrial divergences across well-established biogeographic boundaries within Peninsular Malaysia and between Peninsular Malaysia, Sumatra, and Borneo that were corroborated by statistically significant differences in body shape, scalation, color pattern, and morphospatial arrangement in multivariate space. As such, we consider G. smithii to be a species complex and adjust the taxonomy accordingly with the description of a new species from Peninsular Malaysia and the resurrection of one other from Sumatra.

Species delimitation
The general lineage concept (GLC: de Queiroz 2007) adopted herein proposes that a species constitutes a population of organisms evolving independently from other such populations owing to a lack of gene flow. By "independently," it is meant that new mutations arising in one species cannot spread readily into another species (Barraclough et al. 2003;de Queiroz 2007). Integrative studies on the nature and origins of species are using an increasingly wider range of empirical data to delimit species boundaries (Coyne and Orr 1998;Fontaneto et al. 2007;Knowles and Carstens 2007;Leaché et al. 2009), rather than relying solely on morphology and traditional taxonomic methods. Under the GLC implemented herein, molecular phylogenies were used to recover monophyletic mitochondrial lineages of individuals (populations) in order to develop initial species-level hypotheses. Univariate and multivariate analyses of meristic and morphometric data were then used to search for statistically significant character differences and morphospatial patterns consistent with the phylogenetic delimitations of the species-level hypotheses-thus not conflating species delimitation with species diagnosis.
We used both maximum likelihood (ML) and Bayesian inference (BI) to estimate the phylogenetic relationships among the sampled geckos in our sequence alignment. A ML phylogeny was estimated using the IQ-TREE webserver (Nguyen et al. 2015;Trifinopoulos et al. 2016) preceded by the selection of substitution models using the Bayesian Information Criterion (BIC) in ModelFinder (Kalyaanamoorthy et al. 2017), which supported TPM2+F+G4 as the best fit model for the tRNAs, HKY+F+I for ND2 codon position one, TPM2u+F+G4 for position 2, and KKY+F for position 3. One-thousand bootstrap pseudoreplicates via the ultrafast bootstrap (UFB; Hoang et al. 2018) approximation algorithm were employed and nodes having UFB values of 95 and above were considered highly supported (Minh et al. 2013). A Bayesian inference (BI) analysis was carried out in MrBayes 3.2.3. (Ronquist et al. 2012) on XSEDE using the CIPRES Science Gateway (Cyberinfrastructure for Phylogenetic Research; Miller et al. 2010) employing default priors and models of evolution that most closely approximated those selected by the BIC and used in the ML analysis. Two independent Markov chain Monte Carlo (MCMC) analyses were performed, each with four chains, three hot and one cold. The MCMC simulation ran for 40 million generations, was sampled every 4000 generations, and the first 10% of each run were discarded as burn-in. Convergence and stationarity of all parameters from both runs were checked in Tracer v1.6 (Rambaut et al. 2014) to ensure effective sample sizes (ESS) were well above 200. Post-burn-in sampled trees from both runs were combined and a 50% majority-rule consensus tree was constructed. Nodes with Bayesian poste- Table 1. Taxon sampling, locality data, and ND2 GenBank accession numbers from specimens used in the phylogenetic analyses. rior probabilities (BPP) of 0.95 and above were considered highly supported (Huelsenbeck et al. 2001;Wilcox et al. 2002). After removing the outgroup, MEGA7 (Kumar et al. 2016) was used to calculate uncorrected pairwise sequence divergence among and within species using the complete deletion option which removes missing data and gaps.

Morphological analyses
Morphometric data were taken with Mitutoyo dial calipers to the nearest 0.1 mm under a Nikon SMZ 1500 dissecting microscope on the left side of the body where appropriate. Data taken were snout-vent length (SVL), taken from the tip of the snout to the vent; tail width (TW), measured at the base of the tail; axillia-groin length (AG), 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; head length (HL), measured from the posterior margin of the retroarticular process of the lower jaw to the tip of the snout; head width (HW), measured at the angle of the jaws; head depth (HD), measured from the top of the head posterior to the eyes to the bottom of the lower jaw; internarial distance (IN), measured across the snout between the dorsal margins of the external nares; orbital diameter (OD), the horizontal diameter of the bony orbit; eye to ear distance (EE), measured from the anterior margin of the auricular opening to the posterior margin of the eyeball; snout-eye length (SE), measured from anteriormost margin of the eyeball to the tip of snout; nares-eye length (NE), measured from the anterior margin of the eyeball to the posterior margin of the external nares; auricular opening (TD), measured as the horizontal distance of the ear opening; interorbital distance (IO), measured across the narrowest part of the frontal bone between the orbits; forearm length (FL), measured from the distal edge of the elbow when flexed 90° to the wrist; and crus length (CL), measured from the distal edge of the knee when flexed 90° to the ankle. The tail width (TW) measured at the base of the tail where it contacts the body.
Meristic characters evaluated were scales across frontal bone (FS), counted as the number of scales across the frontal bone at the midline of the orbits; supralabials (SL) and infralabials (IL) counted from the angle of the jaw to the rostral and mental scales, respectively; chin scales (CS), counted as the number of enlarged scales medial to and contacting the infralabials; ventral scales counted across the belly between the ventrolateral body folds midway between the limb insertions (VS); midbody scales counted across the dorsum between the ventrolateral body folds (MB); paravertebral tubercles (PVT) counted as the number of longitudinally arranged tubercles between an imaginary line between the middle of the limb insertions; longitudinal rows of tubercles (LRT) counted across the dorsum between the ventrolateral body folds midway between the limb insertions; subdigital lamellae on the first (TL1) and fourth (TL4) toes; precloacal pore-bearing scales in males (PP); and cloacal spurs (CSP). Codes for natural history collections follow Sabaj (2020). Specimens examined and the raw data used in the analyses are in Table S1.

Species
Cat no. Locality Accession no.

Statistical analyses
Seven of the 86 preserved specimens of Gekko smithii from the Thai-Malay Peninsula used in this study were not represented in the molecular phylogenetic analyses. These speciemens were place in a specific mitochondrial lineage based on their geographic proximity to that lineage. Following this, all individuals were subjected to a Discriminant Function Analysis (DFA) using the MASS Package in R (Ripley et al. 2018) based on a concatenated data set of eight meristic and 14 adjusted (see below) morphometric characters to assess the probability of the placement of each individual into a particular mitochondrial lineage based on geography. DFA uses linear combinations of untransformed data to characterize and separate predefined groups and explicitly attempts to model the differences between them. The predict() command was used to calculate the posterior probability for lineage membership of each individual. A second discriminant analysis was performed using the principal component loadings from the principal component analysis (see below) as a cross-validation for each individual's lineage membership.
All characters conformed to parametric test assumptions of normality (Shapiro-Wilk test, p < 0.05) and homogeneity of variances (F-test, p > 0.05). One-way analyses of variance (ANOVA) were conducted on characters to search for the presence of statistically significant mean differences (p < 0.05) across the data set. Characters bearing statistical differences were subjected to a TukeyHSD post hoc test to ascertain which population pairs differed significantly from each other for those particular characters. Boxplots were generated for discrete meristic characters in order to visualize the range, mean, median, and degree of differences between species pairs bearing statistically different mean values and violin plots with embedded boxplots were generated for continuous morphometric characters to visualize the same plus the distribution frequency of the data. All statistical analyses were performed in R [v3.4.3].
The morphospatial relationships of each species relative to one another and the clustering of the sampled individuals were visualized using multivariate ordination analyses on the appropriate combinations of data (meristic and morphometric). These analyses were employed as a multivariate assessment to determine if the respective data sets were morphospatially consistent with one another and the putative species boundaries delimited by the molecular phylogenetic analyses and diagnosed and defined by the univariate analyses (see below). It should be made clear these are not species delimitation analyses. Principal component analysis (PCA) implemented by the prcomp() command in R was employed to analyze the morphometric data. PCA is a dimension reducing analysis that decreases the complexity of a data set by finding a subset of input variables that contain the most relevant information (i.e. the greatest variance in the data) while de-emphasizing those characters that do not, thus increasing the overall accuracy of the model by eliminating noise and the potential of overfitting (Agarwal et al. 2007). This unsupervised analysis (i.e. data points (specimens) are not a priori assigned to species) that recovers morphospatial relationships among the sampled individuals (i.e. data points) and how well they form clusters that may or may not be aligned with the putative species boundaries delimited by the phylogenetic analyses. To ensure that allometric biases in the raw data were appropriately removed prior to analysis, hatchlings were omitted from the data set and the raw data were adjusted using the following equation: X adj =log(X)-β[log(SVL)-log(SVL mean )], where X adj =adjusted value; X=measured value; β=unstandardized regression coefficient for each population; and SVL mean =overall average SVL of all populations (Thorpe 1975(Thorpe , 1983Turan 1999;Lleonart et al. 2000)-accessible in the R package GroupStruct (available at https://github.com/chankinonn/ GroupStruct). The morphometrics of each species were adjusted separately and then concatenated prior to analysis so as not to conflate intra-with interspecific variation (Reist 1985;McCoy et al. 2006).
Discriminant analysis of principal components (DAPC) from the ADEGENET package in R (Jombart et al. 2015) was performed on the morphometric data set. The DAPC is a supervised analysis that places individuals from each predefined population into separate clusters (i.e. plots of points) bearing the smallest within-group variance that produce linear combinations of centroids having the greatest between-group variance (i.e. linear distance; Jombart et al. 2010). DAPC relies on scaled data from an internal PCA as a prior step to ensure that variables analyzed are not correlated and number fewer than the sample size. Dimension reduction of the DAPC prior to plotting is accomplished by retaining the first set of PCs that account for approximately 90% of the variation in the data set (Jombart and Collins 2015) as determined from a scree plot generated as part of the analysis. Retaining too many PCs forces false structure to appear in the data while retaining too few, runs the risk of missing true structure.
A principal coordinate analysis (PCoA)-conceptually similar to a PCA-using a Gower (dis)similarity index was employed on a concatenated meristic and adjusted morphometric data set. The dissimilarity matrix is used as the input to the analysis, not the original variables themselves. Therefore, information concerning the original variables cannot be recovered. Because this unsupervised multivariate analysis is based on a (dis)similarity index constructed from Euclidean distances between data points, it is appropriate for data sets containing discrete characters (scale counts) because it does not require the data to fulfill the assumptions of linearity (as does PCA) or unimodality (Marhold 2011;Paliy and Shankar 2016), thus allowing more flexible handling of mixed data sets. Furthermore, it is insensitive to null values in the data frame (i.e. missing data) so informative characters such as precloacal pores, which occur only in males, may not have to be removed from the analysis.
Non-parametric permutation multivariate analyses of variance (PERMANOVA) from the VEGAN package in R (Oksanen et al. 2018) were used a priori to determine if the centroid locations of each species in the PCoA and PCA data sets were statistically different (Skalski et al.

2018
). The analyses are based on the prior calculation of the distance between any two data points in a Gower (dis) similarity matrix from the PCoA data set and a Euclidean (dis)similarity matrix from the PCA data set, using 5000 permutations. The analyses are not based on the output of the PCoA or PCA and are thus independent of them. A pairwise post hoc test calculates the differences between all combinations of species pairs, generating a Bonferroni-adjusted p-value and an F ratio. p < 0.05 is considered significant and larger F-ratios indicate more pronounced group separation. A rejection of the null hypothesis (i.e. centroid positions and/or the spread of the data points are no different from random) signifies a significant difference among groups.

Results
The ML and BI analyses recovered trees with identical strongly supported topologies delimiting four major lineages separated by relatively long branch lengths that correspond to distinct geographic regions (Fig. 2). Surprisingly, individuals from the Thai-Malay Peninsula do not form a monophyletic group but are separated into west and east peninsular lineages on opposite sides of the Banjaran Titiwangsa that divide the northern three-quarters of Peninsular Malaysia into eastern and western halves and to a great extent, separate Peninsular Malaysia from southern Thailand (Fig. 1). The east peninsular lineage occurs on at least five east coast islands, the adjacent continental lowlands, southern Peninsular Malaysia, and Singapore (Grismer 2011a,b), and is the sister lineage to a clade of individuals from islands off the west coast of Sumatra including Nias (a deep-water island) and Tuangku (of the Banyak Islands) (Fig. 1), and presumably the rest of Sumatra (see below). The west peninsular lineage is the sister lineage to the eastern peninsular lineage plus the Sumatran lineage and occurs on at least two islands off the west coast of the Malay Peninsula ) and ranges northward through southern Thailand to the Isthmus of Kra (Fig. 1). Individuals from Borneo comprise the sister lineage to all other included G. smithii populations. Uncorrected pairwise sequence divergences among all lineages ranges from 2.5% between the Sumatran and east peninsular lineage to 13.3% between both the Sumatran and west peninsular and Bornean lineages. The east and west peninsular lineages are 4.8% divergent ( Table  3). The phylogenies recovered here corroborate the limited mito-nuclear phylogeny of Rösler et al. (2011)   to sister lineages from eastern (Pahang and Johor states) and western (Selangor state) Peninsular Malaysia. The DFA and DAPC placed nearly all Thai individuals for which there were no sequence data with all the sequenced individuals from the west peninsular lineage with a 99-100% posterior probability (PsP). The only exception was THNHM 01841 from the Hala-Bala Wildlife Sanctuary, Narathiwat in southeastern Thailand which  was placed in the Sumatran lineage with a 79.5% PsP and in the west peninsular lineage with a 20.2% PsP. The DAPC placed it in the west peninsular lineage with a 100% PsP. All other individuals from the east peninsular lineage, Sumatra, and Borneo grouped together with a 98.9-99.9%, 83.0-99.6%, and 88.2-99.1% PsP, respectively. These data were taken as statistical justification for downstream statistical comparisons among the four major lineages. Given the results of both analyses, we consider the placement of THNHM 01841 in the Sumatran lineage as an anomaly. For downstream analyses, it was considered part of the west peninsular lineage and its placement therein is discussed below. The PCA and DAPC ( Fig. 3) were congruent with the PCoA (Fig. 4) in showing separation among the lineages although PERMANOVA analyses revealed the PCA was slightly more discriminating (Table 5). All three analyses recovered complete separation between the east and west peninsular lineages along the collective ordination of the first two axes and all recovered reasonably complete separation of the Bornean and Sumatran lineages from the west peninsular lineage. Principle component (PC) 1 of the PCA accounted for 51.4% of the variation in the data set and loaded most heavily for head size (HH, HL, HW, EE, NE) and crus length (CL) ( Table 4; Fig. 3). PC2 accounted for an additional 10.7% of the variation and loaded most heavily for interorbital distance (IO) and orbital diameter (OD). The PERMANOVA analyses indicated that all the centroid locations in the PCA are statistically significantly different from one another ( Table  5). The only centroid locations not significantly different in the PCoA were between the Bornean and east peninsular lineages and the Bornean and Sumatran lineages (p-adjusted = 0.25 and 1.00, respectively). Although significant, the centroid placement in the PCA of the Bornean and Sumatran lineages was far more similar (F = 3.39; p-adjusted = 0.02) than that between the east and west peninsular lineages (F = 102.02; p-adjusted = 0.0001). This was mirrored in the PCoA, bearing an F ratio of 1.12 (p-adjusted = 1.00) between the Bornean and Sumatran lineages and an F ratio of 53.62 (p-adjusted = 0.0001) between the east and west peninsular lineages.
The ANOVA analyses and subsequent Tukey-HSD post hoc tests of the meristic and morphometric data sets recovered various combinations of statistically different mean values between various combinations of lineage pairs at varying levels of significance (Fig. 5;Tables 6,7,8,9). The analyses recovered the west peninsular lineage and the Bornean lineage as being the most significantly divergent species pair in both the meristic and morphometric data set. The east and west peninsular lineages were also significantly Table 4. Summary statistics and principal component analysis scores for the Gekko smithii complex. Abbreviations are listed in the Materials and methods. well-separated in both data sets. Nonetheless, despite many of these characters having statistically different mean values between the eastern and western lineages, their ranges overlap considerably, rendering them uninformative as discretely diagnostic characters. Numeric and morphometric trends in these characters among all the lineages are illustrated in Figure 6.
Geckos from the west peninsular lineage have a considerable amount of dark dorsal blotching in the interspaces between the transversely aranged white spots which often tend to form bands (Fig. 7). Dorsal blotching is faint to absent in geckos from the east peninsular lineage or appears as fine dark speckling. Geckos from the west peninsular lineage also have a thin dark nuchal loop running from the back of one eye to the back of the other eye. The nuchal loop is usually absent in geckos from the east peninsular lineage, and if present, it is very faint and usually does not contact the eyes. The white dorsal spots on the body of geckos from the west peninsular lineage tend to be larger and surround the dorsal tubercles or border them posteriorly, whereas in geckos from the east peninsular lineage, the spots are usually very small and restricted to the tubercles and/or border them anteriorly.
Given that the multivariate and univariate diagnoses corroborate the geographically structured phylogenetic delimitation of distinct allopatric mitochondrial lineages within Gekko smithii across prominent biogeographic barriers (i.e. mountain ranges and seaways), we hypothesize these are non-reticulating lineages that should be designated as separate species. Because Penang Island is the type locality for Gekko smithii (Gray 1842), individuals in the western peninsular lineage retain that name and a new name is proposed for the eastern peninsular lineage. Based on the descriptions of Platydactylus albomaculatus Giebel (Giebel, 1861) from Bangka Island, Sumatra, the specific epithet albomaculatus is available for the Sumatran populations (see discussion below). Based on the description of Gecko albo-fasciolatus (Günther, 1867) from Banjermassin (= Banjarmasin), Kalimantan, Indonesian Borneo (see Rösler et al. 2011), the specific epithet albofasciolatus is available for the Bornean populations (see discussion below).        (Figs 1, 7). Stoliczka (1870) reports "Gecko Smithii" from Java based on a juvenile specimen (SVL 86.3 mm) specimen sent to him but not collected by him. However, his description of the color pattern is well within the range of variation of both G. smithii and G. gecko. Stoliczka (1870) stated his specimen had 12 longitudinal rows of "small flattened sub-equal granules [=longitudinal rows of dorsal tubercles], slightly varying in size on the posterior part of the body and especially at the sides". The dorsal tubercles of species in the smithii complex are distinctly raised and sub-conical in adults but less so in juveniles. Furthermore, of the 93 specimens of the smithii complex examined here, only one specimen from Borneo had 12 rows dorsal tubercles. The others ranged from 8-11. Awal Riyanto (pers com. in lit. 2021) says he nor any of his colleagues working in Java have ever seen G. smithii. Photographic material and specimens we have examined from Java cataloged as G. smithii in the Zoological Museum Amsterdam, now officially part of Naturalis Biodiversity Center in Leiden, are G. gecko. We believe reports of G. smithii from Java stem from the possible misidentification of this species by Stoliczka (1870) as no new naturally occurring populations have been reported or observed to our knowledge. The less likely scenario exists, however, that Stoliczka described a juvenile G. albomaculatus from Bangka island, Indonesia as he notes "I have also added a complete description of the rare Gecko smithii, Gray, a specimen of which was sent to me from Java, and that of what appears to be a full grown specimen of Tetragonosoma [Lycodon] effrene, CANT., from the island Banca." The ambiguity in this sentence could mean these two species were sent to him from Java but were collected on Bangka Island. Stoliczka (1870) reported Gekko smithii (as Gecko stentor) from the coastal cities of Chittagong, Bangladesh and Akyab (= Sittwe = Sittway, Rakhine [Arakan] State), Myanmar which was followed by Theobald (1876), Annandale (1906), Boulenger (1912), and Taylor (1963). Smith (1935) stated "I do not know of any specimens to prove that this Gecko inhabits Burma, as has been stated."  examined a specimen (MCZ 3120) putatively from "Burma: Rangoon" (=Myanmar, Yangon). MCZ has no other data on this specimen and the only species of Gekko we have seen in Yangon and throughout all of Myanmar over the last five years of field work has been Gekko gecko. Furthermore, the herpetological surveys in Myanmar by the California Academy of Sciences (www. calacademy.org/research/herpetology/myanmar) which resulted in the collection of over 14,000 specimens, found no Gekko smithii. We do not consider this species as part of the Burmese herpetofauna. However, the type material of Gekko gecko azhari Mertens, 1955 from Barkal, Chittagong Hill Tracts, Bangladesh does bear some resemblance to G. smithii in head shape, gular scales, and dorsal tubercle shape (Rösler 2001). Mahony et al. (2009) reported on 23 additional specimens from throughout the Chittagong Division that they considered G. g. azhari.

Gekko smithii Gray, 1842
Variation in color pattern. The dorsal body tubercles of Gekko smithii from northernmost Peninsular Malaysia and Thailand north of the Kangar-Pattani Line (generally running from Gunung Jerai, Kedah, Peninsular Malaysia to Songkhla, Songkhla, Thailand) are surrounded by a large white ocellus whereas in southern populations, the ocelli are smaller and may only border the tubercles posteriorly (Figs 7, 8). Although the latter creates a general resemblance to G. hulk sp. nov., the small white spots in   (Fig. 8F). Although rare, other geckos from Selangor state may resemble east peninsular lineage geckos in lacking a dark nuchal loop and having a darkly speckled dorsum (Fig. 8C). Iris color can also range from turquoise to lime green in the same population (Figs 8A, 8D) and along with ground color (which can vary from yellow to dark gray (Figs 8B, 8E)), is not a diagnostic character (contra Grossmann 2006). See Tables 9 and 10 for variation in morphology.

Natural history (adapted from Grismer 2011a,b).
Gekko smithii is an arboreal nocturnal species that is well-established in all types of primary and secondary forests. It is particularly well-suited for inhabiting occupied and abandoned human dwellings along forest edges where it is most easily observed. In forested habitats, geckos occur on the trunks of large trees and boulders from where males are commonly heard calling during the day and early in the evening.  Wermuth 1965;Kluge 1991Kluge , 1993Bauer 1994;Rösler 2000). De Rooij (1915) obviously also followed Boulenger (1885), because she listed the distribution of G. albofasciolatus based on Günther (1872) under the species G. smithii. Recently, Kluge (2001) revalidated G. albo-fasciolatus at the specific rank, but Malkmus et al. (2002) and Das (2004) did not consider G. albofasciolatus as part of the Bornean herpetofauna." Günther (1867) noted that Gekko albofasciolatus had a reddish-olive dorsum marbled with grayish and a uniformly whitish venter. Specimens with a reddish-olive dorsum with gray marbling also occur in Sabah (Fig. 9E) and preserved material examined from western and eastern Borneo all have uniform whitish venters, although in life, some have faint-yellow marbling which presumably fades after preservation. Günther (1867) also noted his specimen had "a narrow horseshoe-shaped band across the neck, the convexity being directed backwards". This is clearly in reference to the nuchal band which was not present in specimens we examined or in photographs we acquired (Fig. 9). Contra to Das (2004Das ( , 2007, all Bornean specimens we examined had the characteristic dark Y-shaped marking on the head common to G. smithii and geckos of the eastern peninsular lineage (Figs 7, 8, 9). Grossmann (2006) illustrated a specimen from eastern Kalimantan (i.e. eastern Borneo) with a brown vertebral stripe and small white dorsal flecks. All adult Bornean specimens examined herein, have small white flecks and variable degrees of dorsal striping and ground color (Fig. 9). Therefore, dorsal color pattern is not a reliable character separating populations from Borneo west of the Iran Mountains and populations east of the Iran Mountains. Rösler et al. (2011) noted that northern Bornean populations might be different on the basis of Günther's G. albofasciolatus (eastern  Borneo) having 26 rows of ventral scales versus 29-39 in G. smithii s.l. However, specimens examined from northern Borneo (N = 7) have 23-33 scale rows and a specimen examined from eastern Kalimantan at Kelay (Fig. 9C) has 27 scale rows. Therefore, at this point, other than the spotted nuchal band, there are no morphological or color pattern differences between G. albofasciolatus populations east of the Iran Mountains and populations west of the Iran Mountains (i.e. from the Malaysian states of Sarawak and Sabah and the Sultanate of Brunei). The molecular data clearly indicate that the Bornean samples from Sarawak are not conspecific with G. smithii s.s. or any other lineage (Fig. 2). Therefore, the name albofasciolatus, is available for Bornean populations. However, owing to an absence of molecular data from eastern populations, all populations west of the Iran Mountains are referred to as G. cf. albofasciolatus pending the outcome of a molecular analysis. We believe this approach is justified based on the fact that the Banjaran Titiwangsa of Peninsular Malaysia separate G. smithii and the east peninsular lineage and the Iran Mountains may be doing the same with populations from Borneo. Investigations on Bornean populations are currently underway (Grismer et al. unpubl.) Resurrection of Gekko albomaculatus (Giebel, 1861)  N made no mention of P. albomaculatus, and thus that name remained valid and his taxonomy was followed by several prominent authors of the time (e.g. Boulenger 1889Boulenger , 1890Boulenger , 1912Müller 1895 first report from Sulawesi;Flower 1896Flower , 1899de Rooij 1915 andSmith 1930). Boettger (1886) synonymized Gecko Stentor Cantor from "Insel Nias" with (Gecko) fasciolatus (sic) Günther (1867). Additionally, Stoliczka's (1870) description of a specimen from Java as Gecko Smithii was most likely based on a specimen of Gekko gecko. It wasn't until Smith (1935) stated "Gray's description of Gekko smithi, brief though it is, cannot well apply to any other Gecko coming from Penang; I therefore reinstate his name, which has priority over stentor." that Platydactylus Stentor formally became recognized as a junior synonym of G. smithi. Smith (1935) stated that G. smithi ranged throughout the Malay Peninsula as far north as Pattani [Thailand] and the Malay Archipelago. Grossmann and Ulber (1990) reverted to the original spelling of the specific epithet "smithii" following Mertens (1946).

Gekko albofasciolatus (Günther 1867)
Comparing the short description of the two syntypes of Platydactylus albomaculatus by Giebel (1861) and the slightly more detailed redescription of a lectotype (the larger of the two syntypes) by Müller (1941), we cannot differentiate them from the specimens examined from Sumatra, islands off the west coast of Borneo, or G. smithii. They can be putatively separated from the eastern peninsular lineage by the male having 15 precloacal pores as opposed to 6-13 (n=16) and having dark blotching on the body as opposed to its absence or reduction to fine speckling (Fig. 10G). Müller (1941) noted that the lectotype had a dark-brown horseshoe-shaped band on the back of the head running from eye to eye that was bordered posteriorly by a row of moderately large white blotches. Only some Sumatran specimens have a dark nuchal band and the white blotches are often fused, forming a thin white band that is unique to specimens from Sumatra and its adjacent islands (Fig. 10). Bangka Island is Indonesia's ninth largest island and lies less than 15 km off the southeastern coast of Sumatra from where G. "smithii" have been reported (e.g. de Rooij 1915; Fig. 10). The seaway between Bangka Island and mainland Sumatra is less than 25 m deep and these landmasses had broad intermittent subaerial connections with one another, Peninsular Malaysia, and Borneo over the last 10 million years (Hall 2013). Nonetheless, lacking sequence data from topotypic material and mainland Sumatra, the possibility exists, however unlikely, that the Bangka Island population is not conspecific with other Sumatran populations. The molecular data clearly indicate that the Nias and Banyak Islands populations are not G. smithii or any other lineage. Therefore, based on current geography and geographic history, we refer to these insular populations and mainland Sumatran populations as G. cf. albomaculatus. Investigations on these populations are currently underway (Grismer et al. unpubl.). Table 10. Diagnostic character states (shaded cells) separating Gekko hulk sp. nov. from the other species of G. (Gekko). Data on G. gecko, G. nutaphandi, G. reevesi, G. siamensis, G. stoliczkai, and G. verreauxi come from Grossmann and Ulber (1990), Ota and Nabhitabhata (1991), Bauer et al. (2008), Rosler et al. (2011), and Chandramouil et al. (2021.  Diagnosis. Gekko hulk sp. nov. can be separated from all other species of Gekko in the G. smithii species complex by having the combination of a maximum SVL of 161.3 mm, 10-15 supralabials, 9-13 infralabilas, 3-5 internarial scales, 13-24 frontal scales, 4-8 chin scales, 84-110 midbody scales, 16-21 paravertebral tubercles, 9-11 longitudinal rows of tubercles, 22-28 ventral scales, 14-19 1st toe subdigital lamellae, 18-24 4th toe lamellae; 6-13 precolacal pores in males (absent in females); subcaudals enlarged; thin, white nuchal band at base of occiput composed of closely spaced spots; thin dark nuchal band absent or faded and never contacting eyes; small white ocelli confined to dorsal tubercles or their anterior margin in six or seven transverse rows; and no thick dark reticulum on body (Tables 9, 10).
Description of holotype (Fig. 12). Adult male SVL 147.5 mm; head moderate in length (HL/SVL 0.28), width (HW/HL 0.65), somewhat flattened, distinct from neck, triangular in dorsal profile; lores concave slightly anteriorly, weakly inflated posteriorly; prefrontal region concave; canthus rostralis rounded; snout elongate (SE/ HL 0.41), rounded in dorsal profile; eye large (OD/HL 0.26), pupil vertical, margins crenulated; ear opening elliptical, obliquely oriented, moderate in size; eye to ear distance slightly less than diameter of eye; rostral rectangular, bordered posteriorly by large left and right supranasals and smaller azygous postrostral, bordered laterally by first supralabials; external nares seated anteriorly in nasal scale bordered anteriorly by rostral, dorsally by large anterior and smaller posterior supranasals, posteriorly by three small postnasals, ventrally by first and second supralabials; five internasals; 12 (R,L) rectangular supralabials, first supralabial slightly larger than second; 12(R) 10(L) infralabials tapering smoothly to angle of jaw; scales of rostrum and lores flat, larger than flat scales on top of head and occiput; scales of occiput intermixed with distinct, small tubercles; superciliaries elongate, largest dorsally; mental not enlarged, subtriangular, bordered laterally by first infralabials and posteriorly by left and right trapezoidal postmentals contacting medially for 60% of their length posterior to mental; one row of slightly enlarged, elongate chin scales extending posteriorly to fifth (R) and sixth (L) infralabial; gular and throat scales small, flat, juxtaposed, grading posteriorly into larger, smooth, imbricate, pectoral scales which grade into larger ventral scales. Body slightly flattened, relatively long (AG/SVL 0.51) with well-defined ventrolateral folds; dorsal scales small, flat, juxtaposed, interspersed with larger, smooth subconical, regularly arranged tubercles; 110 longitudinal rows of scales at midbody; 17 paravertebral tubercles, those at midbody surrounded by 9-12 smaller scales; 10 longitudinal rows of tubercles at midbody; body tubercles extend from occiput onto base of tail forming transverse rows, terminating near end of original tail; smaller tubercles in temporal and postocular regions; 25 longitudinal rows of flat, imbricate, ventral scales much larger than dorsal scales between body folds; and eight large, pore-bearing, precloacal scales.
Forelimbs moderately robust, relatively short (FL/SVL 0.13); large, imbricate scales of upper arm and anterior surface of forearm larger than those on posterior surface which are interspersed with large tubercles; palmar scales flat, subimbricate; digits well-developed, inflected at penultimate interphalangeal joints, arising from digital pad; subdigital lamellae wide, transversely expanded forming digital pad; claws well-developed, claw base sheathed by a dorsal and lateral scale; digit I clawless; hind limbs more robust than forelimbs, moderate in length (CL/SVL 0.16), covered anteriorly by large, flat, imbricate scales, dorsally and posteriorly by much smaller flat, juxtaposed scales interspersed with large, subconical tubercles; ventral scales of hind limbs large, flat, imbricate, abruptly contacting small postfemoral scales; femoral pores absent; plantar scales flat, subimbricate; digits well-developed, inflected at penultimate interphalangeal joints, arising from digital pad; distal subdigital lamellae wide, transversely expanded forming digital pad; 15 transverse lamellae beneath digit I, 19 transverse lamellae beneath digit IV; claws well-developed, claw base sheathed by a dorsal and lateral scale; digit I clawless; small amount of webbing between digits I-IV.
Tail original, 139.0 mm in length, tapering to a point; dorsal scales flat, square, bearing transverse rows of six large, subconical tubercles; tubercle rows separated by six or seven transverse rows of dorsal scales; large, paired, median, transversely expanded subcaudal scales; and base of tail bearing hemipenal swellings, each with two conical postcloacal tubercles.
Color pattern (Figs 11, 12). Ground color of all dorsal surfaces yellowish-brown bearing slightly darker faint mottling; top of head bearing small white spots and dark-colored, diffuse Y-marking; thin white nuchal band composed of closely spaced spots extends from one ear opening to other, edged anteriorly by faint, dark-colored nuchal band running between postocular regions; incomplete series of obliquely aligned, small white spots immediately anterior to forelimb insertions parallel the white nuchal spots; series of five rows of transversely arranged, widely separated, small white spots between limb insertions, another between hind limb insertions; white spots on body generally confined to tubercles or border them anteriorly; limbs bearing incomplete, thin white bands; white spots on base of each digit and one or two more on each digit; venter beige, mottled with faint dark-colored markings, weakest in gular region, most dense in subcaudal region; and center of iris gold, transitioning distally to green then turquoise.
Variation (Figs 7, 8). Color pattern variation in Gekko hulk sp. nov. is not as extensive as that in G. smithii. The hue and intensity of the ground color changes from day to night and is generally lighter and less bold during evening hours. The description here is of the daytime coloration. The dorsal ground color in life can be dark brownish green, light-green, tan or yellowish brown. There are no thick, dark-colored reticulations on the dorsum although some specimens have faint, brown markings. The white dorsal spots are small and typically confined to the tubercles although in some specimens they may border the tubercles anteriorly or both. The light-colored caudal bands can vary in both width and boldness. In one specimen from Pulau Tioman (LSUHC 5152; Fig. 8K), the white nuchal spots form a solid band as in G. albomaculatus. The color pattern of hatchlings and juveniles is more boldly marked. Variation in morphology is presented in Tables 11 and 12.
Distribution. Gekko hulk sp. nov. ranges from at least the southeastern corner of southernmost Thailand in the Hala-Bala Wildlife Sanctuary, Narathiwat Province, southward east of the Banjaran Titiwangsa through Peninsular Malaysia to Singapore. It approaches the west coast of Peninsular Malaysia south the Banjaran Titiwangsa at Gunung Ledang, Johor. It is known from the east coast islands of Perhentian Besar and Redang, Terengganu in the north and from the islands of Tulai and Jahat, Johor and Tioman, Pahang in the south (Figs 1, 7).  Contact zone with Gekko smithii (Fig. 7). As noted above, the DFA placed THNHM 01841 from the Hala-Bala Wildlife Sanctuary, Waeng District, Narathiwat, Province, Thailand in Gekko cf. albomaculatus with a 79.5% PsP and in G. smithii with a 20.2% PsP. The DAPC placed it in G. smithii with a 100% PsP. THNHM 01841 bears the color pattern of the northern populations of G. smithii having white ocelli surrounding the tubercles (e.g. Figs 7I, 8G) although the ocellei are somewhat reduced in width (Fig. 13). However, photographs of two other specimens from the Hala-Bala Wildlife Sanctuary that were not examined, have the typical G. hulk sp. nov. color pattern with small white spots generally restricted to the tubercles (LSUDPC 12880 and 12953; Figs 7J and 8J, respectively). Examination of a speci-men of G. smithii from Pa Phru Sirinthorn Research Station, Su-ngai Kolok District, Narathiwat (THNHM 01844) and a photograph of another specimen of G. smithii from Waeng, Narathiwat (LSUDPC 12877, Fig.  7I) approximately 32 km and 13 km to the north of the Hala-Bala Wildlife Sanctuary, respectively, also have the typical northern G. smithii color pattern. Both the DFA and DAPC placed THNHM 01844 in G. smithii with a 100% PsP. These data indicate the G. smithii and G. hulk sp. nov. are very likely to be sympatric in the Hala-Bala Wildlife Sanctuary and that the color pattern of THN-MH 01841 may be the result of hybridization. A genomic data set is currently being developed to further investigate this contact zone. Etymology. The specific epithet "hulk" is a noun in apposition in reference to ''The Incredible Hulk'', who is a fictional character and superhero created by Stan Lee and artist Jack Kirby in 1962 and appears in the Marvel Comics publications. When angry, The Incredible Hulk becomes a large, green-skinned, muscular beast possessing great physical strength and a very aggressive temperament-all characteristics of Gekko hulk sp. nov.
Comparisons. Wood et al. (2020a) erected the subgenus Gekko (Gekko) based on the phylogenetic relationships of the holotype of G. hulk sp. nov. (LSUHC 6284) and its inferred close relationship to the generotype species G. gecko  in order to contain the species G. albofasciolatus; G. gecko (Linnaeus, 1758); G. nutaphandi Bauer, Sumontha, and Pauwels, 2008; reevesii (Gray, 1831); G. siamensis Grossmann and Ulber, 1990; G. smithii s.s.; and G. verreauxi Tytler, 1864. To this we add G. albomaculatus, G. hulk sp. nov., and G. stoliczkai. Although several authors have used dorsal ground color as a diagnostic character (e.g. Rösler et al. 2011;Otal et. al. 1991), it is too variable in all species to be of diagnostic significance (see above). Chandramouli et al. (2021) noted several statistically significant differences in meristic and morphometric characters between G. smithii s.l. and G. stoliczkai. No data or formal analysis has ever been forwarded to argue that Gekko taylori Ota and Nabhitabhata, 1991 is a junior synonym of G. siamensis. Thus, the former remains in G. (Gekko) pending the outcome of current investigations (Grismer unpubl). Gekko hulk sp. nov. can be separated from other species of the subgenus G. (Gekko) by a number of discrete characters (Table 10). Gekko hulk sp. nov. differs from G. gecko and G. smithii by having a much smaller maximum SVL (161.3 versus 185.0 and 191.0, respectively) and from G. nutaphandi (SVL 117.0) and G. stoliczkai (SVL 128.4) by having a larger maximum SVL. It differs from G. verreauxi by having the rostral scale in contact with the external nares as opposed to them being separated by a small scale. Gekko hulk sp. nov. differs from G. gecko and G. reevesi by having occipital scales and those on the top of the head smaller than the scales on the rostrum as opposed to them being the same size and having enlarged versus small subcaudal scales. It differs from G. reevesi by having 9-11 versus 12-18 longitudinal rows of dorsal tubercles. Male G. hulk sp. nov. differ from G. albofasciolatus, G. nutaphandi, G. reevesi, G. smithii, and G. stoliczkai in having fewer precloacal pores (6-13) as opposed to 13-15, 17-22, 13-20, 13-15, and 13-15, respectively. Gekko hulk sp. nov. has a turquoise/green iris which separates it from G. gecko and G. reevesi that have a gold/copper to olive/brown iris, from G. nutaphandi which has a bright-red iris, and from G. verreauxi which has a golden iris. It differs from G. albomaculatus by not having a thin, white, solid nuchal band (except for one specimen from Pulau Tioman). Gekko hulk sp. nov. differs from G. albofasciolatus by having as opposed to lacking a white nuchal band. Gekko hulk sp. nov. differs from G. smithii in that the small white dorsal spots, if not confined to the tubercles, tend to border them anteriorly, whereas in G. smithii, the dorsal tubercles are surrounded by a large white ocellus in northern populations and the smaller ocelli in the southern populations only tend to border the tubercles posteriorly. Gekko hulk sp. nov. has a number of significantly different mean meristic and morphometric values as well as significantly different morphospatial placement that separate it from G. albofaciolatus, G. albomaculatus, and G. smithii (Figs 3-6; Tables 6-9).
Natural history. Much like other species of Gekko (Gekko), G. hulk sp. nov. is an arboreal nocturnal species that is well-established in all types of primary and secondary forests as well as buildings on forest edges. We observed lizards 3-4 m above the ground on the trunks of large trees on Gunung Tebu (Fig. 14) and heard males calling in the afternoon. On Pulau Tioman, we observed individuals 4-5 m above the ground on tree trunks, in tree cavities, and on cement light poles. On Pulau Tulai, lizards were seen on trees during the day but collected off the metal stair railing at night. On Perhentian Besar, geckos occur is high densities and are more common on granite boulders where they take refuge between boulders and in rock cracks. Lizards are far less common on tree trunks.  reports finding eggs attached to the undersides of boulders on Pulau Perhentian Besar during September and a gravid female was collected during July on Pulau Tioman.

Discussion
Several integrative taxonomic analyses of Peninsular Malaysian genera and species across a number of taxonomic groups have consistently found the Banjaran Titiwangsa to be an effective geographic barrier separating sister clades and sister species (Grismer et al. , 2014a(Grismer et al. ,b, 2015(Grismer et al. , 2018a(Grismer et al. ,b, 2019Chan et al. 2017Chan et al. , 2018Wood et al. 2020b). Therefore, it is not surprising that a lineage such as the Gekko smithii complex that had not been examined in depth since its description in 1842, shows the same biogeographic pattern. Also, the relationships of several Peninsular Malaysian taxa such as G. hulk sp. nov. being more closely related to Sumatran lineages as opposed to other Peninsular Malaysian or Bornean lineages is also not uncommon (Loredo et al. 2013;Matusi et al. 2014Matusi et al. , 2019Chan et al. 2016Chan et al. , 2017Chan et al. , 2018Chan et al. , 2020Harvey et al. 2016;Grismer and Davis 2018;O'Connell et al. 2018O'Connell et al. , 2019. The shared episodic environmental fluctuations of these major Sundaic landmasses since the Miocene accounts for a plethora of overlapping and distinct com- binations of concordant phylogeographic patterns across a broad range of taxa (de Bruyn et al. 2013, 2014 andreferences therein). When the results from integrative taxonomic analyses of more groups become availableespecially those that include the vast number of unstudied lineages from Sumatra (Eric Smith pers. com. in lit. 2017) and Indonesian Borneo-these patterns will likely become even more commonplace.