The discovery of two new species in the Cyrtodactylus irregularis group highlights that hidden diversity remains in the largest clade of the mega-diverse genus Cyrtodactylus

. The Cyrtodactylus irregularis group, originally considered to consist of only one taxon, has been split into 26 species. We herein present the distribution of all species within the group in Cambodia, Laos and Vietnam and describe two new species based on integrative analyses. Cyrtodactylus chumuensis sp. nov. is discovered from Dak Lak Province and distinguished from the remaining taxa by more than 11.86% genetic divergence and by the following distinct morphological characters: size medium (SVL 67.5 mm); enlarged femoral scales on each thigh 4–5, femoral pores 0–2 in males; precloacal pores 6–7 in males; ventral scale rows 43–45; lamellae under toe IV 17–21. Cyrtodactylus arndti sp. nov. is described from Binh Dinh Province and genetically differentiated from its congeners by a minimum of 11.42% and by the following characters: adult size medium (SVL 73.4–80.8 mm); enlarged femoral scales on each thigh 5–11; femoral pores 0–2 in males; 6 precloacal pores in males, females with 6 pitted precloacal pores; ventral scale rows 26–38; lamellae under toe IV 17–22; subcaudal scales transversely enlarged. Additionally, we highlight the potential cryptic diversity with the taxon currently regarded as C. pseudoquadrivirgatus and understudied areas in Vietnam where new species will likely be discovered.

We herein present data on the distribution of known taxa in the C. irregularis group and discuss potential cryptic diversity based on our updated molecular analyses. In addition, we also describe two new species collected during our recent fi eldwork in Dak Lak and Binh Dinh provinces in south-central Vietnam based on integrative taxonomy, viz. combination of morphological and genetic evidence.

Sampling
Field surveys were conducted in June 2014 in M'Drak District, Dak Lak Province and in August 2016 in Van Canh District, Binh Dinh Province, Vietnam. Specimens were collected between 19:00 and 22:00 h (Fig. 1). After being photographed in life, specimens were euthanized in a closed vessel with a piece of cotton wool containing ethyl acetate (Simmons 2002), fi xed in 80% ethanol for fi ve hours, and later transferred to 70% ethanol for permanent storage. Tissue samples were preserved separately in 70% ethanol prior to fi xation. Voucher specimens referred to in this paper were deposited in the collections of the Institute of Ecology and Biological Resources (IEBR), Hanoi, Vietnam and the Zoologisches Forschungsmuseum Alexander Koenig (ZFMK), Bonn, Germany.
Afterwards, sequences were aligned by ClustalX ver. 2.1 (Thompson et al. 1997) with default settings. Data were analyzed using Bayesian inference (BI) as implemented in MrBayes ver. 3.2.7 (Ronquist et al. 2012), maximum likelihood as implemented in IQ-TREE ver. 1.6.8 , and Maximum Parsimony (MP) implemented in PAUP*4.0b10 (Swofford 2001). For MP analysis, a heuristic analysis was conducted with 100 random taxon addition replicates using tree-bisection and reconnection (TBR) branch swapping algorithm, with no upper limit set for the maximum number of trees saved. Bootstrap support was calculated using 1000 pseudo-replicates (BP) and 100 random taxon addition replicates. All characters were equally weighted and unordered. For the ML analysis, we employed a single model and 10 000 ultrafast bootstrap replications (UFB). The optimal model for nucleotide evolution was determined using jmodeltest ver. 1.2.4 (Darriba et al. 2012).
For the Bayesian analyses, we used the optimal model determined by jmodeltest with parameters estimated by MrBayes ver. 3.2.7. Two independent analyses with four Markov chains (one cold and three heated) were run simultaneously for 10 7 generations with a random starting tree and sampled every 1000 generations. Loglikelihood scores of sample points were plotted against generation time  to detect stationarity of the Markov chains. Trees generated prior to stationarity were removed from the fi nal analyses using the burn-in function. The posterior probability values (PP) for all clades in the fi nal majority rule consensus tree were provided. Nodal support was evaluated using BP as estimated in PAUP, UFB in IQ-TREE ver. 1.6.7.1, and PP in MrBayes ver. 3.2. UFB and PP ≥ 95% and BP ≥ 70% are regarded as strong support for a clade (Hillis & Bull 1993;Ronquist et al. 2012;Nguyen et al. 2015). The optimal model for nucleotide evolution was set to GTR+I+G for ML and Bayesian analyses. The cut-off point for the burn-in function was set to 47 in the Bayesian analysis, as -lnL scores reached stationarity after 47 000 generations in both runs. Uncorrected pairwise divergences were calculated in PAUP*4.0b10.

Morphological characters
Measurements followed Ziegler et al. (2002) and Luu et al. (2015) with slight exceptions and were taken with a slide-caliper to the nearest 0.1 mm. Measurements were taken on the right side of the specimens unless otherwise indicated. Scale counts were taken using a stereo microscope (Euromex NexiusZoom). Bilateral scale counts were given as left/right or as one value (on each side), unless otherwise indicated.

AG
= axilla-groin length, from insertion of posterior margin of front limbs to insertion of anterior margin of hindlimbs BW = maximum width of body CrusL = crus length, from heel to fl exed knee ED = greatest diameter of ear opening EyeEar = distance between posterior margin of orbit and anterior margin of ear opening FemurL = femur length, from limb insertion to knee ForeaL = forearm length, from elbow fl exed to base of palm HH = maximum height of head posterior to orbits HL = head length, from posterior margin of retroarticular process of jaw to tip of snout HW = maximum width of head posterior to orbits IND = internarial distance ML = maximum length of mental MW = maximum width of mental OD = orbital diameter, greatest diameter or bony orbit RW = maximum width of rostral RH = maximum height of rostral SE = distance between tip of snout and anterior margin of orbit SVL = snout-vent length, from tip of snout to anterior margin of cloaca TaL = tail length, from cloaca to end of tail Scale counts DTR = dorsal tubercle rows counted transversely across the midbody between ventrolateral folds FP = femoral pores in males EFS = enlarged femoral scales EPS = distinctly enlarged precloacal scales, at the least twice as large, as the surrounding scales GST = granular scales surrounding dorsal tubercles IL = infralabials were counted from the fi rst labial scale to posterior corner of mouth (except for granular scales) IN = internasals LD1 = number of subdigital lamellae on fi rst fi nger LD4 = number of subdigital lamellae on fourth fi nger LT1 = number of subdigital lamellae on fi rst toe LT4 = number of subdigital lamellae on fourth toe N = nasal scales, surrounding naris, from rostral to labial, except rostral and labial PAT = postcloacal tubercles PM = postmentals PP = precloacal pores in males or pitted precloacal pores in females SL = supralabials were counted from the fi rst labial scale to corner of the mouth V = ventral scales at midbody, counted from one ventrolateral fold to the other

Phylogenetic analyses
The fi nal matrix consisted of 61 terminals, consisting of fi ve from this study and 56 from previous works. The ingroup taxa contained described species (except C. buchardi and true C. pseudoquadrivirgatus) and two undescribed populations of the Cyrtodactylus irregularis group. Both ML and BI produced very similar topologies based on a total of 652 aligned characters with no internal gaps and using a single model of molecular evolution (Fig. 2). In the MP analysis, 254 characters were parsimony informative. A single most parsimonious tree with 1649 steps was recovered (Consistency index = 0.25; Retention index = 0.59). The results show that Cyrtodactylus condorensis and Cyrtodactylus grismeri are nested within the Cyrtodactylus irregularis group in the MP analysis but placed in a separate clade in ML and BI (Fig. 2).
In our analyses, Cyrtodactylus badenensis is considered a member of the Cyrtodactylus irregularis group in both ML and BI analyses. Similar to the results reported in Ngo et al. (2022), Cyrtodactylus cf. pseudoquadrivirgatus was recovered in three distinct places of the tree, two closely related to C. taynguyenensis with 8.2 % and 10.6% genetic divergence from the latter and one as a sister taxon to C. cryptus with 9.3 % divergence between the two.  (Smith, 1921) and C. grismeri Ngo, 2008 Table 2 Diagnosis The new species can be distinguished from remaining congeners of the irregularis species group by a combination of the following characters: maximum SVL 67.5 mm; dorsal pattern with 6 irregularly shaped and short longitudinal stripes on the neck; nuchal band thin, interrupted, reaching the posterior margin of the orbits; the absence of transversely enlarged median subcaudal scales; 4 or 5 enlarged femoral scales on each thigh, 17-19 distinctly enlarged precloacal scales; males with 0 or 1 femoral pore on each thigh, 6 or 7 precloacal pores in a continuous series, ˄-shaped; ventral scales 43-45; dorsal tubercles in 20 irregular longitudinal rows; precloacal groove absent; internasal scales 2; supralabials 8-14; infralabials 9-11; number of subdigital lamellae on fourth fi nger 16-19 and on fourth toe 19-21.

Etymology
The new species is named after its type locality, Chu Mu Mountain in Dak Lak Province. We propose the following common names: Chu Mu Bent-toed Gecko (English), Thạch sùng ngón chư mư (Vietnamese).

Coloration in preservative
Dorsal surface of head, body and limbs light-brown with some dark-brown bands, pattern without light bordering; occiput marbled with small, irregular dark-brown blotches; rostral, mental and infralabials creamy white, supralabials dark-beige with short greyish brown vertical stripes; neck bands dark-brown,  (2023) 80 extending in two thin stripes along lateral sides of the snout to the orbits, broader on the neck, interrupted on the left side; two dark-brown longitudinal stripes, disconnected from the neck band extending to shoulders, one dark-brown blotch next to each stripe; dorsal pattern consisting of 6 irregular bands, each formed by two triangularly shaped blotches, shifted medially along the body axis; dorsolateral region covered with small, irregular dark-brown blotches arranged in a longitudinal row from neck to groin; a blurry dark-brown transverse band on dorsal surface of original part of the tail, regenerated part greyish beige and speckled with very small light-greyish brown blotches; dorsal surface of limbs with 3 or 4 dark-brown, blurry bands; phalanges brown with creamy white knuckles; tubercles white or dark-brown depending on position on pattern or background; venter creamy white; ventral tail greyish beige without bands. For coloration of the paratype in life that closely resembles the holotype in life see Fig. 3.

Variation
The paratype is a subadult and therefore differs greatly in size. Its original tail showed some dark-brown irregular bands, although broken at the base. The number of precloacal pores is 7 and it lacks femoral pores. For more morphological characters see Table 2.

Comparisons
The new species can be distinguished from all other member of Cyrtodactylus irregularis group from Vietnam by morphological characteristics (see Table 2).
Cyrtodactylus chumuensis sp. nov. differs from C. badenensis by having more ventral scale rows (43-45 vs 25-29 in C. badenensis), the presence of enlarged femoral scales (4-5 vs absent in C. badenensis), the presence of precloacal pores in males (6-7 vs absent in C. badenensis), and the absence of transversely enlarged subcaudals (vs present in C. badenensis); differs from C. bidoupimontis by having a smaller size (SVL 67.5 mm vs 74.0-86.3 mm in C. bidoupimontis), fewer enlarged femoral scales (4 or 5 vs 8-10 in C. bidoupimontis), a different dorsal color pattern (irregularly banded with longitudinal stripes on the neck vs transversal bands with light borders in C. bidoupimontis), and a thin discontinuous nuchal band (vs well developed, widened posteriorly in C. bidoupimontis); differs from C. bugiamapensis by having fewer enlarged femoral scales (4 or 5 vs 6-10 in C. bugiamapensis) and the different dorsal color pattern (irregularly banded with longitudinal stripes on the neck vs unclear transversal bands formed by irregular round to oblong, dark-brown spots in C. bugiamapensis); differs from C. buchardi by having more ventral scale rows (30 vs 43-45 in C. buchardi), the presence of enlarged femoral scales (4-5 vs absent in C. buchardi), more subdigital lamellae under the fourth fi nger (16-19 vs 14 in C. buchardi), more subdigital lamellae under the fourth toe (17-21 vs 12 in C. buchardi); differs from C. cattienensis by having more ventral scale rows (43-45 vs 28-42 in C. cattienensis), more subdigital lamellae under the fourth fi nger (16-19 vs 12-16 in C. cattienensis), and different dorsal color pattern (irregularly banded with longitudinal stripes on the neck vs irregular dark-brown banded, fi rst band on the shoulder x-shaped C. cattienensis); differs from C. caovansungi by having a smaller size (SVL 67.5 mm vs 90.4-94 mm in C. caovansungi), fewer enlarged femoral scales (4 or 5 vs 8 in C. caovansungi), fewer femoral pores on each thigh in males (0-1 vs 6 in C. caovansungi), fewer precloacal pores in males (6 or 7 vs 9 in C. caovansungi), fewer lamellae under the fourth fi nger (16-19 vs 22 in C. caovansungi), fewer lamellae under the fourth toe (17-21 vs 23-25 in C. caovansungi), and the absence of transversely enlarged subcaudal plates (vs present in C. caovansungi); differs from C. chungi by having more ventral scale rows (43-45 vs 30 or 31 in C. chungi), more dorsal tubercle rows (20 vs 18 in C. chungi), different dorsal color pattern (irregularly banded with longitudinal stripes on the neck vs irregular transversal bands with a closed nuchal band), and a thin, discontinuous nuchal band (vs continuous nuchal band in C. chungi); differs from C. cryptus by having fewer ventral scale rows (43-45 vs 47-50 in C. cryptus), the presence of enlarged femoral scales (vs absent in C. cryptus), fewer precloacal pores in males (6 or 7 vs 9-11 in C. cryptus), a thin, discontinuous nuchal band (vs well developed, widened posteriorly in C. cryptus), and different dorsal color pattern (irregularly banded with short, longitudinal stripes on the neck vs irregular transverse bands in C. cryptus); differs from C. cucdongensis by having more dorsal tubercle rows (20 vs 16-19 in C. cucdongensis), fewer enlarged femoral scales (4 or 5 vs 5-9 in C. cucdongensis), more enlarged precloacal scales (20-21 vs 6-13), and a different dorsal colour pattern (irregularly banded with short, longitudinal stripes on the neck vs irregular dark brown transverse bands); differs from C. culaochamensis by having a smaller size (SVL 67.5 mm vs 69.8-79.8 mm in C. culaochamensis), the presence of enlarged femoral scales (vs absent in C. culaochamensis), fewer lamellae under the fi rst fi nger (11 vs 13 or 14 in C. culaochamensis), and fewer lamellae under the fi rst toe (10 or 11 vs 13-15 in C. culaochamensis); differs from C. dati by having fewer femoral pores in males (0-2 vs 3 or 4 each side in C. dati), fewer lamellae under the fi rst toe (10 or 11 vs 12 or 13 in C. dati), the presence of blotches on head (vs absent in C. dati), and different dorsal color pattern (irregularly banded with longitudinal stripes on the neck vs irregular dark blotches); differs from C. gialaiensis by the presence of enlarged femoral scales (vs absent in C. gialaiensis), fewer precloacal pores in males (6 or 7 vs 9 or 10 in C. gialaiensis), and more subdigital lamellae under the fourth fi nger (16-19 vs 14 or 15 in C. gialaiensis) as well as under the fourth toe (17-21 vs 15-17 in C. gialaiensis); differs from C. huynhi by having more dorsal tubercle rows in males (20 vs 16-18 in C. huynhi), fewer lamellae under fi rst fi nger (11 vs 12-15 in C. huynhi), fewer lamellae under fi rst toe (10 or 11 vs 13-17 in C. huynhi), and a thin discontinuous nuchal band (vs well developed, widened posteriorly in C. huynhi); differs from C. irregularis by having a smaller size (SVL 67.5 mm vs 72.0-86.0 mm in C. irregularis), more ventral scale rows (43-45 vs 37-42 in C. irregularis), fewer enlarged femoral scales (4 or 5 vs 7 or 8 in C. irregularis), and different dorsal color pattern (irregularly banded with short longitudinal stripes on the neck vs blotched in C. irregularis); differs from C. kingsadai by having a smaller size (SVL 67.5 mm vs 83.0-94.0 in C. kingsadai), fewer enlarged femoral scales (4 or 5 vs 9-12 in C. kingsadai), the absence of transversely enlarged subcaudal plates (vs present in C. kingsadai), and more internasals (2 vs 1 in C. kingsadai); differs from C. orlovi by having more ventral scale rows (43-45 vs 36-39 in C. orlovi); a thin, discontinuous nuchal band (vs continuous nuchal band in C. orlovi), and different banded pattern ranges (6 vs 3-5 in C. orlovi); differs from C. phnomchiensis by having a smaller size (SVL 67.5 vs 76.1-80.7 mm in C. phnomchiensis), more precloacal pores in males (6 or 7 vs 5 in C. phnomchiensis), and different dorsal color pattern (irregularly banded vs banded in C. phnomchiensis); differs from C. phuocbinhensis by having a larger size (SVL 67.5 mm vs 46.0-60.4 mm in C. phuocbinhensis), different dorsal color pattern (irregularly banded vs stripes or blotches in C. phuocbinhensis), and dark-brown transverse banded of the tail than light-brown interspaces (vs dark transverse banded wider than light interspaces in C. phuocbinhensis); differs from C. phumyensis by having more ventral scale rows ( the presence of enlarged femoral scales on each thigh (vs absent in C. taynguyenensis); irregularly banded of the tail (vs banded in C. taynguyenensis), and different dorsal color pattern (irregularly banded vs blotched in C. taynguyensis); differs from C. yangbayensis by having a smaller size (SVL 67.5 vs 78.5-92.3 mm in C. yangbayensis), more subdigital lamellae under the fourth toe (17-21 vs 15-17 in C. yangbayensis), fewer subdigital lamellae under the fi rst toe (10-11 vs 18-20 in C. yangbayensis), and the absence of transversely enlarged subcaudal plates (vs present in C. yangbayensis); differs from C. ziegleri by having a smaller size (SVL 67.5 vs 84.6-93.0 mm in C. ziegleri), more ventral scale rows (43-45 vs 33-39 in C. ziegleri), and fewer enlarged femoral scales (4 or 5 vs 8-10 C. ziegleri).

Natural history
Specimens were found at night between 19:00 and 22:00, on granite rock, along a rocky stream, approximately 0.5-1.0 m above the ground, at elevations between 400 and 500 m a.s.l. The surrounding habitat was evergreen forest of medium and small hardwoods mixed with shrubs and vines (Fig. 5). The humidity was approximately 50-71% and the air temperature ranged from 27.5 to 32.1°C. Other reptile species found at the sites included Gekko gecko (Linnaeus, 1758), Hemidactylus platyurus (Schneider, 1792), Ahaetulla prasina (Boie, 1827), Lycodon sp., and Oligodon sp.

Etymology
We name this species in honor of our colleague, Prof. Dr. Hartmut Arndt, Institute of Zoology, University of Cologne, Germany, in recognition of his support for biodiversity research in Vietnam. As common names, we suggest Arndt's Bent-toed Gecko (English) and Thằn lằn ngón arndt (Vietnamese).

Coloration in preservative
Dorsal surface of head, body and limbs light-brown with some dark-brown pattern, without light bordering; occiput marbled with small, irregular dark-brown banded; rostral, mental, fi rst three supralabials and fi rst infralabials greyish brown, remaining infralabials light beige, some with greyish brown speckles or frames; nuchal band discontinuous, consisting of two stripes extending from the orbits to the neck, ending by a dark blotch on each side and a third blotch medially; dorsum with 7 irregular bands, the fi rst two interrupted; dorsolateral region covered with round or elongate dark-brown blotches; tail with 6 dark-brown bands, fade ventrally, some small dark-brown spots arranged in a line along the lateral side of tail, tail tip dark-brown; dorsal surface of limbs with 6 or 7 irregular dark-brown bands; phalanges brown with beige knuckles; dorsal tubercles white or dark-brown depending on position; tubercles on dorsal surface of limbs and tail light-brown; venter greyish brown.

Sexual dimorphism and variation
The female (IEBR R.4933) differs from the males by the absence of hemipenal swellings. All male specimens have 6 precloacal pores but the females has 6 pitted scales only. Three males (IEBR R.4930, ZFMK 103910, ZFMK 103911) and the female lack femoral pores (IEBR R.5077, IEBR R.4933). For further morphological characters see Table 3.

Comparisons
The new species can be distinguished from all other member of Cyrtodactylus irregularis group from Vietnam by morphological characteristics (see Table 4).

Discussion
Cyrtodactylus irregularis was described by Smith in 1921 based on specimens collected from the Cam Ly River Valley, Langbian Plateau (now known as Lam Dong Province, Vietnam) (Smith 1921;Grismer et al. 2021). After its description, the name Cyrtodactylus irregularis was applied to all specimens of Cyrtodactylus collected in north-central and south-central Vietnam for more than 80 years. However, as C. irregularis is still being investigated, we referred to the specimens of Cyrtodactylus from Lac Duong, Lam Dong Province, Vietnam as C. cf. irregularis in this study. At the moment, the Cyrtodactylus irregularis group is split into 28 species (including the two species in the present study) and it is the largest group within the genus Cyrtodactylus Ngo et al. 2022). Most members of the Cyrtodactylus irregularis group are distributed in north-central and south-central Vietnam which is also known as the Truong Son Range (Fig. 10) Ngo et al. 2022). Only a few species inhabit eastern Cambodia and southeast Laos, i.e., C. buchardi, C. cryptus, and C. phnomchiensis (David et al. 2004;Luu et al. 2016b;Neang et al. 2020;Grismer et al. 2021).
Species in this group occur in different habitat types, including granitic montane and limestone evergreen forests, and coffee farms. The adaptation ability allows its members to successfully diversify by occupying different ecological niches (Grismer et al. 2020;Ngo et al. 2022). Many areas in Cambodia, Laos, Vietnam where members of the C. irregularis group might occur are still poorly studied, e.g., the Central Highlands in Vietnam (Fig. 10). It is also noted that a number of species complexes and several potentially new species have been reported within the range of these broadly distributed taxa Ngo et al. 2022). For example, there exist three C. pseudoquadrivirgatus clades with high genetic divergence from each other as shown by this and previous studies. Cyrtodactylus pseudoquadrivirgatus was described by Rösler et al. (2008) from A Luoi, Thua Thien Hue Province; Huong Hoa, Quang Tri Province; Ba Na Nature Reserve, Da Nang Province; Kon Plong, Kon Tum Province. Thus, further studies are required to determine the actual distribution of the C. irregularis group by sequencing and examining the type specimens (i.e., C. pseudoquadrivirgatus, C. irregularis) to show whether it is in fact a species complex, viz. containing multiple taxa.
Similar to those relationships reported by Grismer et al. (2021) using ND2, the phylogeny supported by BI and ML in our study showed that C. grismeri belongs to the C. condorensis group although the former was placed in both the C. irregularis and C. condorensis groups in the barcoding study by Brennan et al. (2017) using a combination of COI and ND2. Moreover, the phylogenetic position of C. badenensis is still unclear. Grismer et al. (2021) considered C. badenensis the sister species to the C. condorensis group in their ML analysis but this placement was not supported in the BEAST analysis, using the mitochondrial gene ND2 and tRNAs. However, it was supported as a member of or sister to the C. irregularis group in this study and in several previous studies using COI gene or combining mito-nuclear data in both ML and BI analyses (i.e., Brennan et al. 2017;Nguyen et al. 2017;Grismer et al. 2021;Ostrowski et al. 2021). In addition, the species was not recognized as representative of the C. irregularis group based on morphological characters. It has a different dorsal pattern compared to that of remaining species within the C. irregularis group. Instead of an irregular brown dorsal pattern on a beige background and the occiput marbled in the same color, C. badenensis shows regular white transverse bands on a black background and a yellow occiput (Nguyen et al. 2006). Cyrtodactylus badenensis is tentatively placed in the C. irregularis group in the present study pending more comprehensive molecular study.
for resolving taxonomic issues but also for providing a better understanding of the population status, threats, and distribution of species within the largest group within the mega-diverse Cyrtodactylus.