A new species of snail-eating snakes of the genus Pareas Wagler, 1830 (Reptilia: Serpentes) from eastern Himalayas, India

1,3 Bombay Natural History Society, Mumbai, Maharashtra 400001, India. 2 A/2, Ajinkyanagari, Karvenagar, Pune, Maharashtra 411052, India. 4 Annasaheb Kulkarni Department of Biodiversity, Abasaheb Garware College, Pune, Maharashtra 411004, India. 4 Department of Biotechnology, Fergusson College, Pune, Maharashtra 411004, India. 5 Department of Biosciences, Veer Narmad South Gujarat University, Surat, Gujarat 395007, India. 6 National Centre for Biological Sciences, TIFR, Bangalore, Karnataka 560065, India.


Introduction
Members of the snake subfamily Pareinae Romer, 1956 are strictly nocturnal, arboreal and are one of the most fascinating groups of snakes that feed on rather unusual prey (Whitaker & Captain 2004;Hoso et al. 2007;Vitt & Caldwell 2014). Members of the subfamily that specialize in feeding on snails present a remarkable case of dietary specialization where they have asymmetry in their dentition on the maxilla to devour prey like snails and slugs (Hoso et al. 2007). Despite their evolutionary signifi cance, these snakes remain poorly studied concerning their habits, habitat and their diversity. The subfamily currently comprises three genera, namely, Pareas Wagler, 1830, Aplopeltura Duméril, 1853 and Asthenodipsas Peters, 1864 (Deepak et al. 2018). The genus Pareas is the most species-rich among members of the subfamily, with 19 nominate species recognized today (Wallach et al. 2014;You et al. 2015;Vogel et al. 2020), with three species currently recorded from India, i.e., Pareas andersonii Boulenger, 1888, P. monticola Cantor, 1839 and P. modestus Theobald, 1868 (Whitaker & Captain 2004;Lalbiakzuala & Lalremsanga 2019;Vogel et al. 2020).
In the course of a herpetological expedition, we surveyed several localities across the northeastern Indian state of Arunachal Pradesh. During the surveys, we collected specimens of Pareas which looked visually similar to each other that were later identifi ed to belong to two sympatric species based on the structure of the hemipenis. One population was attributed to the species P. monticola based on ventral scale number and lack of keels on dorsal scales and shape of hemipenis. The other, however, could not be attributed to any of the species reported from India, owing to the differences in scalation and hemipenis morphology. To identify the second population that was collected from near Kamlang Wildlife Sanctuary in eastern Arunachal Pradesh, we generated molecular data for the collected specimens and compared them with available museum material. Results from the molecular analysis show that the specimens are genetically distinct form congeners and comparison of morphology of known species further attests the fi ndings. We herein describe the population of Pareas from eastern Arunachal Pradesh as a new species following an integrative taxonomic approach incorporating morphological, osteological and molecular data.

Morphology
The study was conducted under permit nos. CWL/Gen/173/2018-19/Pt.V11/2434-43 and CWL/ Gen/173/2018-19/Pt.V11/2421-33 issued by the Forest Department of Arunachal Pradesh. Four specimens of the new species were collected in the fi eld by hand, photographed and later euthanized with halothane within 24 hours of capture following ethical guidelines for animal euthanasia (Leary et al. 2013). The specimens were fi xed in 8% formaldehyde buffer and later stored in 70% ethanol. Liver tissue was collected for molecular work and stored in 95% molecular grade ethanol prior to fi xation. The specimens have been deposited in the collection of the Bombay Natural History Society (BNHS), Mumbai, and research collection at the National Centre for Biological Sciences, Bangalore. Measurements were taken with the help of a digital calliper to the nearest 0.1 mm, except SVL and TaL, which were taken with a string and then measured using a scale.
Morphological data for the new species were compared with the type specimens available for examination and other voucher specimens of the congeners and with data from Vogel (2015) and You et al. (2015). Dentition asymmetry index was calculated following Hoso et al. (2007).
Abbreviations CL = cephalic length measured from tip of snout to the constriction of the neck DSR = dorsal scale rows, counted at approximately one head length behind the head, midbody, and one head length before vent. ED = eye diameter, widest diameter of the eye El = eye to labial height measured from the lowest border of the eye to the lower border of the labial EN = eye to nares distance ES = eye to snout length HL = head length measured from snout tip to the angle of the jaw HW = head width measured at the widest part of the head NW = neck width measured at the constriction of the neck SVL = snout to vent length TaL = tail length TL = total length (SVL + TaL) V = ventral scales, counted as directed by Dowling (1951)  Micro-CT scans were generated for the paratype male NCBS BH655 using a Bruker ® Skyscan 1272 (Bruker BioSpin Corporation, Billerica, Massachusetts, USA). The head of the specimen was scanned for 210 minutes at a resolution of 5.4 μm and recording data for every 0.4° rotation for 360° with (AL) 1 mm fi lter. The source voltage for the scan was 65 kV and source current was 153 uA. Volume rendering was performed with CTVox (Bruker BioSpin Corporation) and images were edited in Adobe Photoshop CS6. Osteological description is based on volume renders retrieved from CTVox following the terminology of the skull described by Heatwole (2009). Images from CT scans of the new species were compared with existing literature on dentition of Pareas spp. (Wang et al. 2020) and with CT scan images of museum specimens (Anonymous 2015). Comparative material examined is listed in Appendix 5.

Molecular analysis
Genomic DNA was isolated from the preserved tissues of the holotype and the female paratype and a specimen of P. monticola from Kamlang WLS and from near Pakke Tiger Reserve using QIAGEN DNeasy kits following protocols directed by the manufacturer. Molecular methods largely follow Mirza et al. (2016) and Mirza & Patel (2018). A fragment of the mitochondrial cytochrome b (cyt b) and 16S rRNA were amplifi ed using primers used by Pyron et al. (2013) and Mirza et al. (2016). A 22.4 μl reaction was set, containing 10 μl of Thermo Scientifi c DreamTaq PCR Master Mix, 10 μl of molecular grade water, 0.2 μl of each 10 μM primer and 2 μl of template DNA, carried out with an Applied Biosystems ProFlex PCR System. The thermo-cycle profi le used for amplifi cation was as follows: 95°C for 3 minutes, (denaturation temperature 95°C for 30 seconds, annealing temperature 48°C for 45 seconds for cyt b as well as 16S rRNA, elongation temperature 72°C for 1 minutes) × 36 cycles, 72°C for 10 minutes, hold at 4°C. PCR product was cleaned using a QIAquick PCR Purifi cation Kit and sequenced with an Applied Biosystems 3730 DNA Analyzer. Sequences of related taxa available from GenBank ® were downloaded for molecular phylogenetic reconstruction following taxa sampling as in Wang et al. (2020) (Appendix 1), and the sequences were aligned in MegaX (Kumar et al. 2018) using ClustalW (Thompson et al. 2002) with default settings. The aligned dataset was subjected to Maximum Likelihood (ML) and Bayesian Inference (BI) on the online portal of W-IQ-TREE at http://iqtree.cibiv.univie.ac.at/ (Trifi nopoulos et al. 2016). The model selection for the analysis was set to auto and the analysis was run with an ultrafast bootstrap analysis for 1000 iterations. Non-parametric bootstrap pseudo-replicates were used to assess support of the clades. For the optimal partitioning strategy and evolutionary substitution model, aligned data was analyzed using PartitionFinder ver. 1.1.1 (Lanfear et al. 2012), implementing a greedy search algorithm under the Akaike Information Criterion (AIC). Bay esian Inference (BI) was implemented in MrBayes ver. 3.2.2 (Ronquist & Huelsenbeck 2003) and was run for 10 million generations and sampled every 1000 generations. The BI run included fi ve parallel chains, three hot and two cold chains. The standard deviation of split frequencies of the analysis reached were below 0.01, after which the analysis was not continued further. Twenty-fi ve percent of the trees generated were discarded as burn-in. Data were subjected to phylogenetic reconstructions with a generalised time-reversible (GTR) + gamma (G) model as the sequence substitution model, based on the optimal partitioning scheme suggested by PartitionFinder for BI. Un-corrected pairwise p-distance (% sequence divergence) was calculated in MegaX (Kumar et al. 2018) with pairwise deletions of missing data and gaps.

Results
Molecular phylogenetics based on 1117 bp of the mitochondrial cytochrome b gene revealed that the specimens from eastern Arunachal Pradesh were embedded in a clade containing P. hamptoni (Boulenger, 1905), P. mengziensis Wang et al., 2020 andP. formosensis (Van Denburgh, 1909), and was recovered basal to the entire clade with high support (ML bootstrap 89, BI posterior probability 0.99). Genetic divergence with congeners is 12-24%. Keeled dorsal scales and hemipenis morphology further support the distinctness of the species from other members of the clade and congeners (see below). Molecular phylogeny based on 16S rRNA recovered similar relationships as with cyt b.  (4) loreal not touching orbit, (5) ventrals 160-183, (6) subcaudals 65-70 in males, 52 in one female specimen, (7) hemipenis short, unilobed, (8) 6-7 maxillary teeth, (9) dorsum brown with thin black transverse bands, the head with a large black blotch from which two longitudinal black stripes (3-4 scales wide) run on each side of the neck leaving a pale central portion.
The new species shares several characters with members of its clade and is here compared to each species in greater detail based on differing and non-overlapping characters. The new species differs from P. formosensis in bearing keeled dorsal scales (vs smooth in P. formosensis), dentition asymmetry index 4.55 in males (vs 16.13 in P. formosensis).
The new species differs from P. mengziensis Wang, Che, Liu, Li, Jin, Jiang, Shi & Guo, 2020 in bearing 6-7 maxillary teeth (vs 3-5 in P. mengziensis) and in having the dorsum with thin black bands (vs connected black reticulations throughout the body in P. mengziensis).
The new species is most similar to P. hamptoni in sharing the plesiomorphic state, where the loreal shield does not touch the orbit and is separated by the preocular. However, the new species differs from the species as follows: ventrals 160-183 (vs 197-202); two anterior temporals (vs a single temporal scale in P. hamptoni); subcaudals 65-70 in males, 52 in female (vs 96 in P. hamptoni); bearing 8 keeled dorsal scales (vs only a single row keeled in P. hamptoni); hemipenis unilobed and not forked (vs deeply forked in P. hamptoni).

Etymology
The specifi c epithet is a patronym honoring wildlife photographer Sandesh Kadur for his contribution to biodiversity documentation in the Himalayas, in particular Arunachal Pradesh, as well as for his constant support to the authors during the expedition.

Description
Holotype ♂ (BNHS 3574) (Figs 1-4) The specimen is in good condition, preserved in a coil with its head resting outside the coil (Fig. 2). The specimen bears incisions. The hemipenis is partly everted.
Head short, 15.45 mm comprising 2.22% of total length; high, 6.47 mm, with steeply domed snout in lateral view; upper jaw visible from ventral side. Head distinctly broader (9.3 mm) than neck (4.4 mm).
Snout abruptly tapers, rounded tip in dorsal view (Fig. 3). Rostral subpentagonal, reaching top of the snout; as wide as high with a distinct furrow towards its ventral edge. Upper jaw distinctly longer than lower jaw. Nostrils large, 2.04 long and 1.32 high in the centre, elliptical-shaped, positioned in the centre and posterior half of nasal scale. Paired internasals, wider (2.07 mm) than long (1.26 mm); smaller than prefrontals. Prefrontals slightly wider (2.83 mm) than long (2.59 mm). Frontal hexagonal, 3.87 mm at the widest portion, median length 5.06 mm. Parietals 6.62 mm long, 3.87 mm at its widest anteriorly. Temporals 2+3+3 on both sides, subequal in size, posterior one inserts deeply between supralabial sixth, seventh and eighth. Five nuchal scales, slightly larger than adjacent dorsal scales, bordering parietals. Supraocular larger than preocular; preocular small, subequal. Loreal slightly longer (1.68 mm) than high (1.54 mm). Two postoculars, subequal in size. Eye large, circular, 3.13 mm (eye diameter/head height 0.48) diameter with a spherical pupil. Seven supralabials, seventh longest. First to third supralabials smallest, fi rst supralabial only contacts second supralabial, rostral and nasal. Second supralabial in contact with nasal, preocular, loreal and fi rst and third supralabials. Third supralabial in contact with preocular, second and forth supralabials and making contact remotely with loreal. Supralabials separated from the orbit by a crescent-shaped subocular.
Mental short, broad, triangular. Infralabials 7, anterior fi ve infralabials short and narrow, fi fth onwards larger. First infralabials of both sides in broad contact, separate the mental from the genials. Sixth infralabial broadest. First six infralabials in contact with the genials. Anterior genials almost twice as long as wide; anterior genials in broad contact, posterior genials only in remote contact.  (2020) 60 Body laterally compressed, ventral surface a little fl attened. Dorsal scales in 15:15:15 rows. Dorsal scales imbricate, regularly arranged, vertebral and adjoined scale rows enlarged and larger than the outermost dorsal scales. Eight scale rows on the mid-dorsum (including vertebral rows) keeled; other dorsal scale rows scales smooth and glossy, lacking apical pits. Ventral scales 160 in number + 2 preventrals. Anal shield undivided, slightly larger than last ventral scale, its posterior margin overlaps nine small, irregular scales on each side, in addition to pair of larger subcaudals medially. Subcaudals paired, 70 in number. Tail terminates in a sharp tapering apical spine. Total length 694 mm, tail length 144 mm, tail/total length ratio 0.207.

Hemipenial morphology (paratype NCBS BH655)
The organ is fully everted and expanded (Fig. 4). Hemipenis short, unilobed, stout and unicaliculate; lobe extends for about 60% of the hemipenis; capitulum restricted to sulcate and dorsal surfaces of the organ, covering nearly half of the lobe's length at the level of the sulcus spermaticus; capitulum smooth, except for two rows of calyces spanning almost the entire width of the organ; on asulcate surface, lobes ornamented with three to four parallel broken rows of mediolaterally enlarged and papillate body calyces; sulcus spermaticus simple; the sides of the sulcus spermaticus are smooth; truncus and hemipenial base is wrinkled and completely nude. (Fig. 2) Overall, in a shade of brown with 28 paired black transverse bands from nape to the vent. Some of these bands are distinct, whereas some are merely black spots that connect rudimentarily to form bands. The head bears a large black blotch from which two black longitudinal stripes (3-4 scales wide) run on each side of the neck leaving a pale central portion. The ventral scales are white or cream colored with sparse black mottling.

Variation shown by paratypes and referred specimens
The paratypes match the holotype in all respect except for the details noted herein: dorsal scales of female paratype BNHS 3575 are smooth, lacking keels, likely a character that is sexually dimorphic, in addition to fewer subcaudal scales. The color of individuals varies greatly in being light brown to dark blackish brown to reddish orange. Other differing characters are listed in Table 1.

Natural history
The type specimens were captured on low bushes along roads on the outskirts of Kamlang Wildlife Sanctuary at night. All the specimens were observed actively foraging after dusk. The habitat at the type locality is contiguous with the adjoining Namdapha Tiger Reserve and Mehao Wildlife Sanctuary, which lie in the Mishmi Hill range. Mishmi Hills lie between the Himalayas and the Indo-Burma biodiversity hotspot. The adjoining areas of Myanmar also share similar biotope and it is likely that the new species will be distributed in Myanmar in addition to India. The species was found in sympatry with Pareas monticola, Boiga siamensis Nutaphand, 1971, Bungarus niger Wall, 1908, Ahetulla sp. and Trimeresurus popeiorum Smith, 1937. The new species is common throughout the sampled area ranging from an elevation of 300 m to 1200 m, whereas P. monticola appears to be rare and only a single specimen was found at lower elevation (< 300 m).

Distribution
Currently, the new species is known only form the type locality, a tropical wet evergreen forest.

Discussion
The phylogenetic relationships recovered in the present work based on 1117 bp of mitochondrial cytochrome b gene (Fig. 6) are congruent with those of other studies (You et al. 2015;Zaher et al. 2019;Vogel et al. 2020;Wang et al. 2020), in which the new species, P. kaduri sp. nov., is sister to the clade containing P. hamptoni, P. formosensis and P. mengziensis with high support (ML bootstrap 89, BI posterior probability 0.99). The new species differs from P. hamptoni and P. mengziensis in exhibiting an uncorrected pairwise sequence divergence of 12% (Appendix 2). From other congeners,  the new species shows an un-corrected p-distance of 13-24% (Appendix 2). The molecular phylogeny based on 16S rRNA recovered similar relationships (Appendix 4). The genetic data hint at the presence of undocumented diversity within Pareas, especially in the broadly distributed species of the genus (see Vogel et al. 2020). Further work on the widespread members of Pareas must be undertaken to ascertain the taxonomic status of genetically distinct lineages. Discovery of the new species of Pareas of the P. hamptoni clade is the fi rst record of a member of this clade in India and the basal relationship of the species in it hints at a Himalayan origin of the clade. Our work brings the total number of species recognized within the genus Pareas to 20.
The discovery of yet another snake species from the same expedition to Arunachal Pradesh, after Trachischium apteii Bhosale, Gowande & Mirza, 2019(Bhosale et al. 2019 and Trimeresurus salazar Mirza, Bhosale, Phansalkar, Sawant, Gowande & Patel, 2020(Mirza et al. 2020, is not surprising as this region has received less attention in terms of documentation of diversity of reptiles. These discoveries advocate the need for extensive exploration across northeast India as a whole, to document the diversity of reptiles and perhaps other poorly studied taxa. Large expanses of forested habitats outside protected areas are under severe anthropogenic pressure and efforts must be made to safeguard these habitats especially in Arunachal Pradesh.