Further assessment of the Genus Neodon and the description of a new species from Nepal

Recent molecular systematic studies of arvicoline voles of the genera Neodon, Lasiopodomys, Phaiomys, and Microtus from Central Asia suggest the inclusion of Phaiomys leucurus, Microtus clarkei, and Lasiopodomys fuscus into Neodon and moving Neodon juldaschi into Microtus (Blanfordimys). In addition, three new species of Neodon (N. linzhiensis, N. medogensis, and N. nyalamensis) have recently been described from Tibet. Analyses of concatenated mitochondrial (Cytb, COI) and nuclear (Ghr, Rbp3) genes recovered Neodon as a well-supported monophyletic clade including all the recently described and relocated species. Kimura-2-parameter distance between Neodon from western Nepal compared to N. sikimensis (K2P = 13.1) and N. irene (K2P = 13.4) was equivalent to genetic distances observed between recognized species of this genus. The specimens sampled from western Nepal were recovered sister to N. sikimensis in the concatenated analysis. However, analyses conducted exclusively with mitochondrial loci did not support this relationship. The occlusal patterns of the first lower (m1) and third upper (M3) molars were simpler in specimens from western Nepal in comparison to N. sikimensis from eastern Nepal and India. Twelve craniodental characters and four external field measurements were examined from specimens of N. sikimensis from eastern Nepal and India, N. irene, and Neodon from western Nepal. Neodon from western Nepal were significantly different from N. sikimensis from eastern Nepal and India in ten out of 16 characters measured and from N. irene for all characters except ear height. Specimens from western Nepal were smaller in size than N. sikimensis from Eastern Nepal and India and larger than N. irene. Together the results of the molecular and morphological analyses indicate that Neodon from western Nepal are distinct under the phylogenetic, genetic and morpho species concepts.


Introduction
The genus Neodon (Tribe Arvicolini, Subfamily Arvicolinae) was described by Horsfield [1] and was retained as a genus in early revisions [2][3][4]. Allen [5] recognized Neodon as a subgenus of Microtus whereas Ellerman and Morrison-Scott [6] included it in the genus Pitymys, a large and rather ill-defined genus with taxa occurring in the Palearctic and Nearctic. Gromov and Polyakov [7] considered Neodon a subgenus of Microtus whereas Corbet [8] followed the suggestion of Ellerman and Morrison-Scott [6] including Neodon in the genus Pitymys. Revisions in the 1990's either recognized Neodon as a distinct genus [9,10] or as a subgenus of Microtus [11][12][13] whereas more recent revisions [14,15] have recognized as a distinct genus. Central Asia voles including the genera Neodon, Blanfordimys, and Phaiomys have been hypothesized to be Pleistocene relicts due to their primitive molar occlusal patterns resembling the fossil genus Allophaiomys [16]. Additionally, N. juldaschi possesses a karyotype (2n = 54 or 56) thought to be ancestral to the Arvicolini [10,17].
Neodon sikimensis occurs on both the western and eastern sides of the Himalayas (Fig 1) with specimens reported from southern Tibet on the eastern side and in Nepal, India, and Bhutan on the western side [14]. In Nepal, the reported distribution of N. sikimensis extends well into western Nepal [26]. Several specimens from Nepal identified as N. irene reside at the American Museum of Natural History (AMNH) and the Natural History Museum in London (NHMUK). However, Pearch [26] reclassified specimens at the American Museum of Natural History identified as N. irene collected from Shey Gompa in western Nepal (AMNH 238055-AMNH 238058, AMNH 238060-AMNH 238063) as N. sikimensis citing the restricted distribution of N. irene to China [14,27]. Specimens at the Natural History Museum in London identified as N. irene (cataloged as Pitymys irene forresti: NHMUK 1954.10. 52-1954.10.77) collected from Saipal, Urai Lagna, Garanphu, and Seti in western Nepal could also be excluded from representing N. irene due to their distribution. Although N. irene has been included as a synonym of N. sikimensis by some authors [8,28,29], these two species have been distinguished based on differences in dentition and size [30,31], leading others, including most recent reviews, to treat N. irene as a distinct species [6,7,14,15]. Nadachowski and Zagorodnyuk [16] observed variation in dental characters among specimens from Nepal great enough to suggest that N. sikimensis may constitute a group of sibling species. Because of a lack of agreement of the number of species of Neodon that occur in Nepal a closer examination is needed.
We report on samples of Neodon collected from western Nepal and compare mitochondrial and nuclear DNA sequences obtained with existing sequences from Neodon and related species. The objectives of this study are: (1) to compare Cytb and other mitochondrial and nuclear sequence data from specimens of Neodon to determine if the samples from western Nepal represent a distinct species under the genetic and phylogenetic species concepts; (2) to determine if Neodon from western Nepal represent a distinct species under a morphospecies concept by comparison of craniodental measurements and external field measurements with N. sikimensis from eastern Nepal, India and Tibet, and with N. irene; and (3) to sample both mitochondrial and nuclear loci to provide better resolution of the phylogenetic relationships between Neodon and other genera and subgenera of the Arvicolini.   (dotted), and N. irene (blue). Collecting localities of museum specimens of N. sikimensis (circles), N. irene (triangle), and N. species novum (squares) examined in the study, with the type locality of N. sikimensis indicated with a star. Localities of sequenced specimens of N. species novum (black squares), approximate location of specimens of N. sikimensis sequenced by Liu et al. [20,22] (black circle), and N. irene misidentified as N. sikimensis in the FMNH (black triangle). Range shapefiles of N. sikimensis and N. irene acquired from Terrestrial 2013 and 2015 respectively (Table 1). Permits were secured from the Department of National Parks and Wildlife Conservation in Nepal to collect samples from the Nepalese field mouse (Apodemus gurkha), at the time considered an endangered species, hence specimens were to be released following the collection of ear punches under the permit conditions. Voucher specimens were prepared for all individuals that perished in the traps overnight (Table 1). Specimens were collected following methods outlined in the American Society of Mammalogists Guidelines [32]. Mitochondrial (Cytb and COI) and nuclear (Rbp3 and Ghr) sequences for select Arvicolinae and outgroup taxa (Mesocricetus auratus, Subfamily Cricetinae) were obtained from GenBank (S1 and S2 Appendices).

Specimens and taxonomic sampling
A total of 170 specimens cataloged as N. sikimensis from the Field Museum of Natural History (FMNH; n = 168, including a juvenile) and the National Museum of Natural History (NMNH; n = 2), were measured and examined for molar occlusal patterns (S3 Appendix). In addition, 6 specimens cataloged as N. sikimensis from Sichuan and Yunnan [outside the reported range of N. sikimensis] provinces in China (east of Tibet) in the FMNH (n = 4) and the NMNH (n = 2) were examined. Measurements were also taken from 8 specimens of N. irene in the FMNH (n = 2) and the NMNH (n = 6) and from 3 new specimens of Neodon collected from western Nepal (S3 Appendix).

DNA extraction, sequencing, and alignment
DNA extraction and sequencing for Neodon sampled in Nepal was carried out at the Center for Molecular Dynamics, Nepal (CMDN). Approximately 25 mg of ear tissue stored in 95% ethanol was used for DNA extractions that were conducted using DNeasy Blood and Tissue Kit (Qiagen, Germany). Tissue samples were air dried for 30 min prior to extraction. Each tissue sample was cut into small pieces (approx. 20) and incubated at 56˚C in 180 μl of ATL (Tissue Lysis Buffer) and Proteinase K in a shaking incubator until the tissue samples were completely lysed. DNA was finally eluted in 50 μl of Elution buffer and the quantity and quality of the extracted DNA was assessed on a NanoVue Plus (GE Healthcare Life Sciences) spectrophotometer.
Cytochrome b (Cytb), cytochrome oxidase I (COI), interphotoreceptor retinoid binding protein (Rbp3), and growth hormone receptor (Ghr) were amplified with the primer pairs: L14274/H15915 [33], COIF/COIR [34], Rbp3217/Rbp31531 [35], and Ghr5/Ghr4 [36] respectively. Amplification of Cytb was performed with initial denaturation at 94˚C for 5 min followed by 31 cycles of denaturation at 95˚C for 1 min, annealing at 50˚C for 1 min and extension at 68˚C for 1 min and final extension at 68˚C for 10 min. Attempts to amplify Cytb from "crusties" from six museum specimens included in the morphological analyses were unsuccessful. Amplification of Rbp3 was carried out with initial denaturation at 94˚C for 5 min followed by 35 cycles of denaturation at 94˚C for 1 min, annealing at 50˚C for 1 min and ExoSAP (Exonuclease and Shrimp Alkaline Phosphatase) (ThermoFisher Scientific) was used to purify the PCR products following manufacturer's protocol prior to sequencing reactions. Sequencing reactions (10 µl) were conducted with 1 μl of ExoSAP product, 3 µl primer (5 µM/ µl), 2 µl nuclease free water and 4 µl of Ready reaction mix containing fluorescent tagged terminator (BigDye v3.1; Applied Biosystems Foster City, CA). The sequencing reaction was further purified using Big Dye X-Terminator Purification Kit (ThermoFisher Scientific) (45 µl SAM solution + 10 µl Big Dye X-terminator solution along with all sequencing PCR product (10 µl) in a 0.5 ml tube. The reactions were optimally diluted in ABI 310 Sequencing Strip tubes for running the final sequencing reaction on ABI 310 avoiding the formation of any air bubbles in the strip tubes. Chromatograms outputted from the sequencers were visualized and edited in Chromas 2.6.2 (Technelysium: http://technelysium.com.au/wp/chromas/). Sequences from two mitochondrial genes, Cytb (1140 bp) and COI (1542 bp) and two nuclear genes Rbp3 (1227 bp) and Ghr (869 bp) were aligned using ClustalW [37] in Mesquite 3.50 [38]. The codon positions for each locus were assigned and the alignments of the genes were concatenated in Mesquite. Concatenations for species of Neodon were compiled based on sequences from the same specimen or specimens from the same locality when possible (S1 Appendix).  genetic distances between and within genera were determined from Cytb sequences of Neodon (S2 Appendix) to compare levels of differentiation observed with values reported for other mammals that have been used to characterize species differentiation under the genetic species concept [40,41]. These measures of genetic differentiation were estimated with MEGA6.0 [42] based on the Cytb alignment of the individuals in Table 1 and Neodon sequences from GenBank (S2 Appendix).

Genetic distance and phylogenetic analyses
The concatenated data set was divided into 12 individual partitions a priori with each gene (Cytb, COI, Rbp3, and Ghr) divided by codon positions. Partition finder 2.1.1 [43] was used to determine the best partitioning scheme and best model for each partition based on the AIC criterion (Table 2) under a likelihood framework using PhyML [44] and the greedy algorithm [45].
Maximum Likelihood (ML) analysis was conducted with RAxML [46] for the concatenated sequences with 1000 bootstrap replicates using the GTR+I+G model on all partitions shown in Table 2. The majority rule consensus (MRC) tree of the bootstrapped analysis was constructed in Mesquite. A partitioned Bayesian analysis was conducted on the CIPRES portal [47] using the 11 partitions and models in Table 2 for the concatenated dataset in MrBayes 3.2.3 [48]. Two simultaneous runs of 10,000,000 generations with sampling every 1000 generations was carried out. The MrBayes log files for both runs were examined in Tracer 1.6 [49] and a burnin of 2,500,000 generations was set for each run. The runs were combined after discarding the burnin and the MRC tree with posterior probability values was constructed in Mesquite. In addition, partitioned Bayesian analyses of mitochondrial genes (Cytb and COI), nuclear genes (Rbp3 and Ghr) and each gene were also carried out separately using the models from Table 2 and same parameters as mentioned above.

Morphometric analyses
From examination of photos (S1 Fig) of the dentition of the specimens used by Nadachowski and Zagorodnyuk (see Fig 2 [16]) to illustrate the variation observed in the occlusal pattern of the first lower molar (m1) and third upper (M3) molar among specimens of N. sikimensis, diagrams of the discrete patterns of the m1 and M3 were constructed (Fig 2). The numerical designations of the original illustration (see Fig 2 [16]) were used as code for the patterns observed. The occlusal patterns of the first lower molar varied from a tooth with an anterior loop followed by one pair of confluent triangles followed by three alternating closed triangles and a posterior loop resulting in four lingual and three labial folds (Fig 2, pattern 7) to a tooth with an anterior loop followed by two pairs of confluent triangles followed by three alternating closed triangles and a posterior loop resulting in five lingual and four labial folds (Fig 2, pattern  11). Patterns 8-10 (Fig 2) vary in the degree of development of the second pair of confluent triangles. The upper third molar varied from a tooth with an anterior loop followed by three alternating triangles, the first two which were closed and a third which was either opened or closed, and a posterior loop without a medially inflected end resulting in two lingual folds ( Fig  2, pattern 12) to a tooth with and anterior loop followed by three alternating closed triangles and a posterior loop with a clearly defined medially inflected end resulting in three lingual folds (Fig 2, pattern 14). Pattern 13 (Fig 2) has three closed triangles and a weakly defined medially inflected end of the posterior loop.
Discrete numerical codes were assigned to the m1 and M3 molar patterns of 177 specimens of Neodon examined (S3 Appendix). Variation in the distribution of occlusal patterns was examined from a geographic prospective.
Eight cranial characters were measured (±0.01 mm) with dial calipers including: skull greatest length (SGL), skull basal length (SBL), condylobasal length (CBL), zygomatic breadth (ZB), mastoidal breadth (MB), least interorbital width (IOW), skull height (SH), and auditory bullae length (ABL). In addition, four dental characters were also measured including: length of maxillary tooth-row (LMxT), greatest width across upper molars (MM), length of the mandibular tooth-row (LMbT), and breadth across the two upper incisors (TUIB). External measurements (in mm) including total body length (TBL), tail length (TL), length of hind foot (HF), and ear length (EL) were obtained from skin tags on the specimens. Measurements for the three voucher specimens housed in Nepal are provided in S1 Table. Estimation of descriptive statistics (mean, range, and standard error) of all measurements for each species (S3 Appendix) and one-way ANOVA with pairwise (t-test) comparisons Table 2. Partitioning scheme and best molecular evolution models selected by PartitionFinder [43].

Data Partitions Best Model
Cytb codon 1 GTR+I+G  separation among the species. Variable loadings used in the CVA were based on the product of correlated coefficients of principal components of the log transformed measurements. A scatter plot displaying the first two canonical axes was plotted along with convex hulls, where the axes represent linear combinations of the original variables in a Principal Component Analysis. Both the MANOVA and CVA were performed in the statistical package PAST [50].

Nomenclatural acts
The electronic edition of this article conforms with the requirements of the amended Code of Zoological Nomenclature, and hence the new name contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is urn:lsid:zoobank.org:pub: E296Ø27B-E533-4DF7-9720-BFFDBC5FB48D. The electronic edition of this work was published in a journal with an ISSN and has been archived and is available from the following digital repositories: PubMed and LOCKSS.  genetic distances between specimens of Neodon sequenced from wester Nepal and N. sikimensis (mean = 13.1) or N. irene (mean = 13.4) were comparable to the K2P distances between recognized species of Neodon (mean = 11.1-14.4) ( Table 3). Large genetic distances between the western Nepal Neodon sampled from Dolpo and Myagdi Districts, Nepal were also observed (mean = 7.8; range = 7.6-8.0) ( Table 4).

Phylogenetic analyses
Neodon formed a strongly supported clade in both the Maximum Likelihood (ML = 99) and Bayesian analyses (PP = 1.00) using the concatenated dataset of four genes (Fig 3).    3). The species novum/sikimensis clade recovered in the analysis of the concatenated dataset was sister to a clade containing the other 7 species of Neodon examined in this study (Fig 3).
The taxon, previously recognized as N. juldaschi, was recovered in a clade with taxa of

Molar occlusal patterns
The new specimens of Neodon collected from western Nepal had occlusal patterns of their m1 (S8 Fig) that were best characterized as discrete pattern 9 (Fig 2) and of their M3 that were best characterized as patterns 13 (n = 1) or 14 (n = 2). These patterns were simpler than the patterns observed in a paratype of N. sikimensis and specimens from N. sikimensis reported from Tibet [20] or examined from Sikkim (Table 5). Occlusal patterns were more variable among specimens cataloged as N. sikimensis from Nepal with both the m1 and M3 varying across the range of patterns identified by Nadachowski and Zagorodnyuk [16]. However, more complex  13 & 14) were observed in specimens in eastern Nepal whereas simpler patterns (m1 -patterns 7-9 and M3 -patterns 12 &13) were characteristic of specimens from western Nepal (Table 5). Based on the observed geographic variation of occlusal patterns in specimens from Nepal, we hypothesize that the specimens from western Nepal, including the newly collected material that are genetically distinct from N. sikimensis, represent a distinct species based on morphological species concept. Specimens from Sichuan and Yunnan catalogued as N. sikimensis had m1 occlusal patterns assigned to pattern 7 (Fig 2) and M3 occlusal patterns assigned to pattern 12, identical to some of the patterns reported for N. irene from Sichuan [16]. A few populations of Neodon examined from northeastern Nepal, Khumjung, Num and Balutar (S9 Fig), were observed to have molar with rather simple occlusal patterns (m1 -patterns 8-10; M3 -pattern 12).

Morphometric analyses
Pairwise t-tests showed that specimens in the FMNH from Sichuan and Yunnan catalogued as N. sikimensis were only significantly differenty from N. irene for two characters (LMxT and LMbT), whereas they differed from N. sikimensis by all characters except MB. The specimens catalogued as N. sikimensis from Sichuan and Yunnan were pooled with N. irene for all subsequent statistical analyses.
One-way ANOVA comparing Neodon species novum, N. sikimensis, and N. irene resulted in statistically significant (p<0.0001) differences between the species for all measurements ( Table 6). Neodon species novum from western Nepal were significantly smaller than N. sikimensis and larger than N. irene (Fig 4 and Table 6). ANOVA indicated significant differences among the three species (N. species novum, N. sikimensis, and N. irene) for all observed characters (p<0.0001: Table 6). Post hoc pairwise comparisons using a t-test between the species revealed that N. species novum differed from N. sikimensis and N. irene with 10 and 15 characters out of 16 respectively (Table 6).

Neodon of Western Nepal
Presently two species of Neodon are recognized as occurring in Nepal, N. sikimensis and N. leucurus [26], though the molar patterns of two specimens (NHML 541076 & NHML 662583) referred to as N. irene were illustrated by Nadachowski and Zagorodnyuk [16]. In addition, specimens cataloged as N. irene collected from locations in Nepal are housed in the American Museum of Natural History and the Natural History Museum of London. Pearch [26] included some of these specimens reported as N. irene as N. sikimensis based on the range of N. irene being restricted to China [14,27] rather than examination of these specimens. The molar occlusal patterns of the specimens N. irene from Nepal illustrated by Nadachowski and Zagorodnyuk (see Fig 3 [ 16]) are very similar to the m1 pattern 9 and the M3 patterns 12 and 13 presented for N. sikimensis (see Fig 2 [16]). Sequence data obtained from specimens on Neodon from western Nepal demonstrated considerable genetic differentiation (Cytb K2P) from N. sikimensis, N. irene, N. leucurus, and other species of Neodon. Mean genetic distance (K2P) of Cytb between N. sikimensis and the N. species novum (13.1) exceeds the K2P values calculated between N. sikimensis and all other species of Neodon (range = 11.1-12.5) except for N. fuscus (14.4) and is comparable to genetic distances between recognized species of Neodon (range = 11.1-14.9) (Table 3). Moreover, the mean genetic distance derived between N. sikimensis and N. species novum (K2P = 13.1) is larger than the K2P distances reported within Rodentia for sister species (mean K2P = 7.3, range = 1.3-13.0) and comparable to intrageneric distances (mean K2P = 10.9, range = 4.9-16.9) for species of rodents [41]. In a comparable "genus", Microtus, the reported mean K2P  New species of Neodon from Nepal

Systematics of Neodon
Hinton [2] provided a description of the essential features of the genus Neodon and commented on the presence of three closed triangles on the m1 as in Pitymys, instead of the five, characteristic of Microtus. Neodon is not as highly modified for subterranean habits as Pitymys, with moderately long ears clearly visible above the fur [2]. Our molecular analyses support the conclusion of other recent studies [20][21][22] and revisions [14,15] in the recognition of Neodon as a distinct genus of the Arvicolini. Unlike previous molecular studies where the Neodon clade was recovered as a monophyletic clade within the Microtus clade [20][21][22], it was recovered in our phylogenetic analysis as the sister group to a Microtus-Alexandromys clade though with rather low support. Furthermore, our molecular phylogeny supports the conclusion regarding the taxonomic treatment of M. (Blanfordimys) juldashi [15,19,25] rather than retaining it in the genus Neodon [14].
Several misplaced sequences were initially detected in our phylogeny, including sequences identified as M. clarkei (AY641526) from Luo et al. [51] and N. leucurus (AM392371 and AM392394) from Galewski et al. [36] which appear to represent misidentified specimens of N. irene and N. fuscus respectively (represented as N. irene 12 and N. fuscus 9 in Fig 3). The Cytb sequence AY641526 was originally placed in GenBank by Luo et al. [51] as from M. clarkei but identified as N. irene (though included in their appendix as M. clarkei) by Liu et al. [20,22]. This sequence (AY641526) has a 95% identity with Cytb sequences from N. irene and only an 88% identity with sequences from M. clarkei, thus we conclude this sequence is representative of N. irene. Similarly, Cytb sequence (AM392371 and Ghr sequence AM392394 were submitted to GenBank by Galewski et al. [36] as sequences from N. leucurus but changed to Lemmus leucurus in Chen et al. [23] and to N. fuscus in Liu et al. [22]. The Cytb sequence AM392371 has a 99% identity with Cytb sequences from N. fuscus and an 89% identity with N. leucurus sequences. In addition, the Rbp3 sequence AY163593 submitted by Weksler [52] as from a N. sikimensis was derived from specimen USNM449126 which is cataloged as N. irene from Qinghai, China. This sequence (AY163593) was concatenated with the Ghr sequence AY294924 from a specimen (USNM449173) of N. irene collected from the same location in our analysis (N. irene 6). The Rbp3 sequence AM919400 was excluded from our concatenated analyses due to uncertainty as whether it was derived from N. fuscus or N. irene but was included in the Rbp3 gene tree (S6 Fig). Arvicolines are often difficult to distinguish morphologically, subsequently leading to misidentification of the taxon that was the source of molecular data. Unfortunately, no voucher specimens are associated with these sequences from misidentified specimens other than AY163593. This further demonstrates the importance of voucher specimens and the identification of those vouchers in association with sequences in GenBank and in the publications reporting new sequence data. In addition, once these misidentifications are recognized the database (GenBank) needs to be corrected. Without correction of the taxon misidentification in the database the use of these misidentified sequences will continue to cause confusion in the literature and contribute further to the confusion in resolving the phylogeny of an already problematic group.
All species of Neodon, including N. species novum, form strong supported clades (ML = 100; PP = 1.00) that are reciprocally monophyletic with other Neodon clades. This well support reciprocal monophyly with other clades of Neodon, supports the recognition of N. species novum as a distinct species under a phylogenetic species concept [53,54]. The phylogeny (Fig 2) supports the currently recognized content of the genus [15] with the addition of N. species novum and presents greater resolution of the relationships among Neodon species (excluding N. forresti for which no data are available). However, there are discrepancies among relationships recovered within Neodon between mitochondrial and nuclear loci. All phylogenies show a distant relationship between N. irene and either N. species novum or N. sikimensis. Phylogenies recovered from different concatenation of genes (mitochondrial+nuclear or mitochondrial and nuclear separately) differ in the placement of Neodon specimens from western Nepal. Use of concatenated mitochondrial+nuclear loci recovers N. species novum sister to N. sikimensis, whereas use of just mitochondrial loci recovers N. species novum in a weakly supported clade containing N. leucurus, and N. fuscus and the use of nuclear loci exclusively does not resolve the N. species novum/N. sikimensis relationship. The use of mitochondrial loci to construct phylogenies for arvicolines has been shown to yield low support values and polytomies among the deeper nodes [36,55]. Most Neodon species (N. clarkei, N. leucurus, N. linzhiensis, N. medogensis, and N. nyalamensis) have only sequence from mitochondrial loci available. However, well supported sister taxa relationships were recovered between N. clarkei  and N. medogensis, N. leucurus and N. fuscus, and N. irene and N. linzhiensis. Our analyses support the need for both mitochondrial and nuclear loci to resolve species relationships among arvicolines, as addition of just a few nuclear loci markedly improved the support values.

Molar occlusal patterns
Among the samples of Neodon from western Nepal for which sequence data were obtained, three specimens were vouchered (Table 1). These three N. species novum had molar occlusal pattern (S8 Fig) more complex than observed in N. irene and less complex than observed in specimens of N. sikimensis from Sikkim, reported from Tibet [20] and observed in most specimens of Neodon from eastern Nepal. The molar patterns were most similar with those observed in other specimens of Neodon from western Nepal. We combined all of the Neodon from western Nepal into N. species novum based on their simpler m1 and M3 occlusal patterns (S9 Fig). Population sampled from western Nepal, N. species novum, and N. sikimensis from eastern Nepal and Sikkim have marked differences in the predominant occlusal patterns of their m1 and M3 molars (Fig 6).

Morphometrics
Neodon species novum was significantly smaller than N. sikimensis in four cranial, three dental and three external measurements, but was significantly larger in breadth across the upper incisors. Compared with N. irene, N. species novum was significantly larger in all measurements examined other than ear height. Although N. species novum was significantly larger than N. irene in all measurements, the MANOVA failed to differentiate N. nepalensis and N. irene based on the non-significant Bonferroni-corrected p-value in the pairwise comparisons when all three species were included but yielded a highly significant p-value differentiating the two species when N. sikimensis was excluded. The pairwise comparisons conducted in PAST for MANOVA use within-group covariance matrix pooled over all groups, which could lead to discrepancies in p-values if groups are included/excluded.
In the two-dimensional plot in multivariate space N. species novum was more similar to N. sikimensis, though all three species were clearly differentiated. Measurements reported for N. sikimensis from Tibet included only mean, maximum, and minimum values [20] and without standard error or deviation being reported, limited statistical analyses could be conducted. A CVA including the mean, maximum, and minimum values of Tibet specimens [20] cluster with the N. sikimensis specimens.
All three species were clearly separated by canonical variance analysis and the discriminant analysis correctly identified over 97% of the specimens with only two specimens of N. sikimensis being identified as N. species novum. Specimens from western Nepal, N. species novum, are morphologically distinct from both N. sikimensis and N. irene.

Neodon sikimensis Horsfield, 1851 -Sikkim Vole
One of the problems in trying to address the variation observed among Neodon occurring in Nepal is the lack of a good description of N. sikimensis. Not only is this taxon poorly described [7] but there is some confusion about the author and date of the limited description that is available. Kaneko and Smeenk [56] clarified that the author and date of the publication of the Sikkim vole N. sikimensis is Horsfield (1851) [1] rather than Hodgson (1849). The paper of 1849 was a letter published in the Annals and Magazine of Natural History written by Horsfield [57] commenting on a specimen that Hodgson considered to be a new species and pointing out that the molars had some unique feature that would be described in detail later by Hodgson. Hodgson's expected description never appeared and Horsfield (1851:145-146) [1] formally described the genus Neodon and N. sikimensis based on characteristics of a single specimen from Sikkim. Kaneko and Smeenk [56] concluded that specimen BM 79.11.21.395 in the British Museum is the holotype and reported that there is no skull for this specimen. Horsfield [1] described N. sikimensis as having very soft and silky uniform pelage that was deep brownish black, with a slight rusty shade dorsally, with the chin, breast, and abdomen being a deep bluish gray, with a slightly ferruginous shade. The ears were described as moderate in size and hairy externally. The tail was described as short (1½ inches) and a head and body of 5 inches. Jerdon [58] provided a similar description of the external morphology and reported measurements of the head and body as 4¾ inches, tail length 1½ inches, and hind foot ¾ of an inch of an individual collected in Sikkim, near Darjeeling.
Three specimens on N. sikimensis all collected by B. H. Hodgson from Sikkim were donated to the British Museum of Natural History in 1879 by the Museum of the East-India Company (P. Jenkins, pers. com.). In addition to the holotype (1879.11.21.395) specimen 1879.11.21.397 (skin only) and 1879.11.21.396 (skin with a damaged skull) were collected from the type locality. Specimen 1879.11.21.396 displays complex occlusal patterns of the m1 and M3 (S10 Fig) characteristic of the descriptions provided by Hinton [3] and Agrawal [59] with an anterior loop followed by two pairs of confluent triangles, three closed triangles and a posterior loop in the m1 (pattern 11, Fig 2) while the M3 possesses a long posterior loop with a well-developed third inner angle (pattern 14, Fig 2). Variation in the occlusal patterns of the m1 and M3 has been reported [16] and variation within and between populations was observed in our examination of additional specimens (S9 Fig). In general, specimens from eastern Nepal, had more complex m1 and M3 occlusal patterns characteristic of N. sikimensis from Sikkim and Tibet.
The Sikkim vole has been characterized by its large size (hind foot 20 mm and condylebasal length of skull 28 mm) [3] and more complex cheek teeth [3,59]. Eight cranial characters were examined from a single specimen collected in 1915 from Lachen, Sikkim [3] and standard external and six cranial characters were examined from 12 and 16 specimens from India [59], respectively. In addition, standard external and 11cranial characters were examined in 13-16 specimens from Tibet [20]. Unfortunately, there is limited overlap in the cranial measurements reported among these three studies [3,20,59] but there would appear to be differences in several characters between populations sampled from Tibet and India. For example, the head and body to tail ratio is 2.14 for specimens from eastern Nepal and India compared to 2.55 for specimens from Tibet.
The holotype of N. sikimensis (BM 79.11.21.395) identifies the type locality only as Sikkim, India [1,56]. Jerdon [58] stated "it [N. sikimensis] had only been procured in Sikkim, near Darjeeling, at heights varying from 7,000 to 15,000 feet." Thus, we propose to restrict the type locality to Sikkim, near Darjeeling. The distribution of the Sikkim vole extends from eastern Nepal into northern India and southern Tibet [1,16,20,30,58] and has not been verified to occur in either Sichuan or Yunnan [27].
Together, the results from the DNA sequence data, the and molecular phylogenetic analyses, the geographic distribution of the variation in molar occlusal patterns, and the morphometric analyses supports the hypothesis that the individuals (previously referred to as N. sikimensis) from western Nepal represent an undescribed species of Neodon. Below, these populations of Neodon from western Nepal are formally described as a new species. Diagnosis. A species of Neodon with the following characteristics: size medium for the genus; tail about one-third the length of the head and body; hind foot medium; ear small; dorsal coloration dark (Buffy Brown at tips, Deep Mouse Gray at base, with guard hairs projecting 3-5 mm above pelage with lighter Cream Color tips; color nomenclature following Ridgeway [60]; sides Clay Color; venter Clay Color tips, Mouse Gray base, and lighter Cream Color middle band; tail bicolored, Buffy Brown above and Clay Color below; feet Cream Color with claws about 2.5 mm in length; ears Mummy brown; vibrissae black, 10-15 mm in length, some with white tips; cranium square, 1.7 times as long as wide; rostrum short, NL 85.6% of RL and RL 32. Measurements. External measurements of the holotype as taken in the field (in mm) are: total length, 130.5; tail length, 31; hindfoot, 19.9; and ear, 12.7. Cranial measurements were obtained using dial calipers (in mm) and are as follows: SGL, 25 Measurements of the three specimens in the private collection in Nepal are presented in S1 Table and the mean measurements and standard errors for these and additional specimens are presented in Table 6.
Comparisons. A species of Neodon resembling N. sikimensis and N. irene but is smaller than N. sikimensis and larger than N. irene. Tail shorter but with a higher head and body to tail ratio (2.99) than N. sikimensis (2.14-2.55) and longer than in N. irene but with lower head and body to tail ratio (3.13). Much smaller but with a longer tail than N. leucurus which has a head and body to tail ratio of 3.33 or greater. The ventral pelage of N. nepalensis is brown (Clay colored tips) compared with the lighter venters of N. sikimensis (Cream Buff tips) and N. irene (Slaty washed with gray white). Neodon nepalensis is significantly smaller than N. sikimensis in total body length, tail length, ear height, and 6 cranial measurements including greatest length of the skull, condylobasal length, mastoid breadth, skull height, length of the mandibular tooth row, width across molars but is significantly larger in breath of incisors. This newly described taxon is significantly larger than N. irene in all measurements other than ear height.
The molar patterns of the m1 and M3 observed in N. nepalensis are simpler than those observed in N. sikimensis and are more similar to those observed in N. irene. A m1 with an anterior loop followed by a single pair of confluent triangles, three alternating closed triangles, and a posterior loop resulting in four lingual and three labial folds is characteristic of N. nepalensis (Fig 2, pattern 8) though some individuals showed an unpaired addition confluent triangle following the anterior loop (Fig 2, pattern 9). The m1 of N. sikimensis contained two pairs of confluent triangles following the anterior loop (Fig 2, pattern11), though specimens with only one triangle of the second pair were observed (Fig 2, pattern 10). Specimens of N. irene examined also displayed a m1 with a single pair of confluent triangles (Fig 2, pattern 7) as has been previously reported [16,19,30]. The m1 of N. sikimensis contained five lingual and four labial folds whereas four lingual and three labial folds were observed of the m1 of N. nepalensis and N. irene.
A third upper molar with an anterior loop followed by two alternating closed triangles and an open triangle that was confluent with a posterior loop without a medially inflected end was characteristic of N. nepalensis (Fig 2, pattern 12), though some individuals had either a weakly defined medially inflected end of the posterior loop or an additoional medial triangle that was not closed (Fig 2, pattern 13). The more complex M3 of N. sikimensis had a clearly defined medially inflected end of the posterior loop (Fig 2, pattern 14, though this was weakly defined in some specimens (Fig 2, pattern 13). No medial inflection was observed of the posterior loop of the M3 of N. irene and this tooth had only 2 lingual folds (Fig 2, pattern 12). The M3 of N. nepalensis contained either two lingual folds or two with a hint of a third fold whereas N. sikimensis usually contained a M3 with three lingual folds and rarely two with a hint of a third.
Neodon nepalensis demonstrates differentiation from other species of Neodon, including Neodon sikimensis from Tibet in both of the mitochondrial genes examined (Cytb and COI). Although sequence data are not available from N. sikimensis from eastern Nepal or India, N. nepalensis is morphologically distinct from samples examined from those regions.
Habitat. Inhabits edges of rhododendron (Rhododendron sp.) forest, coniferous (Abies sp., Cedrus sp., Picea sp., Pinus sp.) forests, scrub habitat with peashrub (Caragana sp,) and honeysuckle (Lonicera sp.) and ecotones between scrub and forest at elevations between 2400 and 4200 m. Also reported from areas with clumps of ephedra (Ephedra sp.) and stunted junipers (Juniperus sp.) [61]. Restricted to the western Himalayan broadleaf and coniferous forest, whereas N. sikimensis occurs in the eastern Himalayan broadleaf and coniferous forest in Nepal.
Remarks. Results from the phylogenetic, genetic distance and morphological analyses support the recognition of Neodon nepalensis as a distinct species under phylogenetic, genetic and morpho species concepts.
Etymology. This species is named to recognize its endemic distribution to Nepal. The common name Nepalese vole is recommended.
Specimens examined. Additional material, including specimens examined, are identified in S3 Appendix.

Other potentially unidentified cryptic species of Neodon in Nepal
Genetic distances between N. nepalensis sampled from Myagdi District (GPN16 and GPN25) and Dolpo District (RCH4, RGM4, RGM6, and RGM11) were between 7.6 to 8.0, which are comparable to the reported mean between sister species of Microtus [41]. The topography between these regions where these specimens were collected in western Nepal are separated by deep river valleys created by the Kali Gandaki River and numerous high mountains which may act as barriers to gene flow. However, the current sampling represents two populations without any intermediate sampling of localities and the divergence observed may be an effect of isolation by distance. The current sampling does not provide data to exclude either of these alternative hypotheses. Hence, more intensive sampling of N. nepalensis is need, especially in the areas between the localities from which sequence data were obtained, to determine if additional cryptic species of Neodon occur in western Nepal.
Specimens (BMNH 1922 [62] and the type locality for this taxon has been reported as northern Nepal [7,63]. The original description of P. everesti [62] indicated that it resembled P. waltoni but was considerable smaller and P. waltoni [64] was described as smaller than P. blythi. Since its description [62] this small taxon has been considered as a subspecies of Microtus (Neodon) leucurus [7], a synonym of M. leucurus waltoni [65], a synonym of M. leucurus [11], or a synonym of P. leucurus [7]. Phaiomys leucurus now recognized as N. leucurus [20] has also been reported to occur in the Mustang District of northcentral Nepal [26,61,66]. A few specimens examined, currently recognized as N. sikimensis from northeastern Nepal (Solukhumbu and Sankhuwasabha Districts) collected by the Arun Valley Wildlife Expedition (1972)(1973)  Joshi and Priya Joshi for providing high quality sequences. We also thank Dr. Bruce Patterson, Dr. Adam Ferguson, and John Phelps at the Field Museum and Darrin Lunde and Nicole Edmison at the Smithsonian National Museum of Natural History, for providing access to the specimens in their research collections. Dr. Jan Decher of the Zoological Research Museum Alexander Koenig kindly provided photos of skulls previously examined by Nadachowski and Zagorodnyuk [16] and Dr. Paula Jenkins of the Natural History Museum in London kindly provided clarification of the status and source of the type series of N. sikimensis. We greatly appreciate the constructive comments and suggestions provided by two anonymous reviewers.