Molecular and morphological revision of small Myotinae from the Himalayas shed new light on the poorly known genus Submyotodon (Chiroptera: Vespertilionidae)

The systematics status of the constituent species of the M. mystacinus morphogroup in the Himalayan region has long been marred by uncertainty. Lack of integrative studies combining morphological and genetic data from specimens recently collected in this region has hampered our understanding of cryptic variations in this complex taxonomic group. To address this issue, new material from the Himalayan region of India and Nepal was obtained and vouchered specimens in the holdings of various museums were also re-examined. As comparative material, a large series of relevant specimens from South and Southeast Asia were also included in this revision. Using a combination of multivariate analysis of craniodental characters and molecular reconstructions, we critically evaluated the systematic position of the small Myotinae in the Himalayas. We establish that M. nipalensis forms a very distinct lineage (which also includes the recently described M. annatessae) and refute previous taxonomic suggestions that it is related to M. davidii. Our study also conclusively proved the common occurrence of the poorly known genus Submyotodon in the Himalayan region (Afghanistan, Pakistan, India, Nepal and China) and evidenced species-level divergences within that genus. Submyotodon species share nyctalodont or semi-nyctalodont lower molar configuration with few other small and unrelated Myotinae from Asia suggesting that these unusual dental characters are homoplasious in this subfamily. We also noticed a very confused taxonomic situation associated with many DNA sequences of Asian Myotis deposited in public repositories and call for possibilities of better data curation.


Introduction
The Himalayan foothills are a megadiverse ecoregion supporting a high number of mammalian taxa (Olson et al. 2001;Orme et al. 2005). Its bat fauna has been studied in some details during the last centuries, especially during the Handling editor: Allan McDevitt.
British Raj, when naturalists like Blyth, Hodgson, and Hutton collected many specimens from these hilly regions, some of them representing new species (e.g. Hodgson 1835;Dobson 1878). During the twentieth century, global revisions of the speciose genus Myotis based on extensive museum material of Eurasian origin (Tate 1941;Hill 1983) suggested that many of those named forms from the Himalayas were only colour or minor size variants of more widespread taxa, which resulted in a considerable reduction in the numbers of accepted species (Ellerman and Morrison-Scott 1951;Corbet and Hill 1992). However, owing to the limited resolving power of classical morphological taxonomy in groups such as Myotis characterized by remarkably conservative and convergent phenotypic characters (Ruedi and Mayer 2001;Stadelmann et al. 2007; Morales et al. 2019), part of these taxonomic inferences based solely on morphology likely do not reflect natural groupings. One of these debated species complexes consists of a group of small-sized Myotis collectively known as the mystacinus-morphogroup and thought to be widely distributed over the Palearctic and Indomalayan regions (Thomas 1926;Benda and Tsytsulina 2000;Benda et al. 2016). These small "whiskered" Myotis (forearm 30-35 mm) are all characterized by small feet, rather short ears and wing membranes inserted to the base of toes and indeed resemble to the European M. mystacinus (Kuhl 1817) (type locality: Germany).
A new revision of members of the whole mystacinus morphogroup concentrating on morphometric, bacular and dental characteristics of Central Asian and Western Palearctic specimens (Benda and Tsytsulina 2000), however, again challenged the prevailing taxonomic arrangement of Asian whiskered bats and demonstrated that M. mystacinus sensu stricto (s. str.) is a species confined to the Western Palearctic region and the Caucasus with northern Iran representing its easternmost limit of distribution (Benda et al. 2012). In that region, it occurs in sympatry with another distinct morphotype initially called M. aurascens (Kuzjakin 1935) (type loc. Caucasus). Aided by more comprehensive series of samples from Central and Eastern Asia, Tsytsulina et al. (2012) again surveyed the variation of these bats with both genetic and morphometric analyses and corroborated most of Benda and Tsytsulina (2000) hypotheses, including the distinctness of European M. mystacinus s. str. and the species status and very widespread occurrence of M. aurascens across Eurasia. However, because they lacked any material from south of the Himalayas, they could not definitively evaluate the status of the Himalayan taxon M. nipalensis but only assumed that the steppe form przewalskii could be a subspecies of that species. The same taxonomic conclusions (including doubts about the systematic position of nipalensis sensu Benda and Tsytsulina) emerged from an independent survey of genetic variation of East Palearctic bats (Kruskop et al. 2012).
In a last twist given to the confusing taxonomic arrangement of Asian members of the mystacinus-morphogroup, Benda and colleagues (Benda and Gaisler 2015;Benda 2016) (Benda and Gaisler 2015:327). Nevertheless, no genetic material from the Himalayas was evaluated to substantiate this new hypothesis of synonymy.
Although M. muricola was widely accepted to be a common bat occurring throughout most of the Indomalayan region, with numerous taxa listed under its synonymy, genetic evidence from both mitochondrial and nuclear genes (Lack et al. 2010;Ruedi et al. 2013) showed that at least one of the synonymized taxon from Taiwan, "Myotis" muricola latirostris (Kishida 1932) was phylogenetically not only distant from other forms of M. muricola, but also very divergent from all assayed species of Myotis. These authors concluded that this taxon should therefore be classified in a distinct genus. Owing to its peculiar skull and molar shapes, Ruedi et al. (2015) placed it in the morphologically similar fossil genus Submyotodon described by Ziegler (2003) from Miocene deposits of Europe. These authors further suggested that some small forms formerly considered synonyms of M. muricola (i.e. caliginosus, blanfordi and moupinensis) also corresponded to Submyotodon, a view substantiated in a morphological revision by Benda and Gaisler (2015). Species in this genus all share a nyctalodont (see Fig. S3 in Ruedi et al. 2013) or semi-nyctalodont configuration of the first two pairs of lower molars (Ziegler 2003;Ruedi et al. 2015), unlike in most other Myotis that have a myotodont configuration (Menu and Sigé 1971). The only exceptions among the Myotinae in this respect are taxa included in the M. siligorensis species complex (Borisenko et al. 2008;Tiunov et al. 2011).
Despite many decades of efforts, the paucity of museum specimens and lack of reference genetic data from the Himalayas has hampered progress in our understanding of species diversity and taxonomy in this difficult group of bats. We thus setup a series of field surveys in various parts of Northern India and Nepal with the aim to get primary access to new specimens from this area that includes several type localities for these taxa. We present here results of these surveys of Himalayan Myotinae and, with help of a combined morphological and molecular approach, we shed new light on the taxonomy of members of the mystacinus morphogroup and also on the little-known genus Submyotodon.

Materials and methods
The current revision is based on a series of field studies conducted in the Himalayan region of Nepal (between 1994 and 2016) and India (between 2009 and2019) and also reexamination of museum specimens lodged in various institutions of Asia and Europe. In

Morphological methods
Bats were captured with mist nets, hand nets, and a harp trap set across bodies of water or across presumed flyways in forests. Animals were preliminarily identified in the field with keys in Bates and Harrison (1997), measured, photographed and biopsied (Worthington-Wilmer and Barratt 1996) before being released within an hour of capture. A small number of voucher specimens were taken for more thorough identification involving cranial or dental characters. Animals were handled according to the recommendations of the American Society of Mammalogists (Sikes and Animal Care and Use Committee of the American Society of Mammalogists 2016). Except for hindfoot length (HF) which included claws, standard external measurements were taken as defined by Bates and Harrison (1997) and taken to the nearest 0.1 mm with a dial calliper. These measurements were forearm length (FA), tail length (TL), head and body length (HB), tibia length (TIB), ear length (EAR), tragus length (TRA) and body mass (W, expressed in grams). For vouchered specimens, the following craniodental measurements were also taken with a digital calliper (to the nearest 0.01 mm) and also corresponded to those proposed by Bates and Harrison (1997): greatest length of skull including incisors (GTLi); condylo-basal length (CBL); condylo-canine length (CCL); zygomatic breadth (ZB); breadth of braincase (BB); interorbital breadth (IOB); width across the upper canines (CCW); width across the upper molars (M3M3W); maxillary toothrow length (CM3L); mandible length, including incisors (MLi); mandibular toothrow length (cm3L); in addition, we also took the height of the coronoid process measured from its tip to the closest point of the base of the ramus (coh) (Yu et al. 2020). The prepared specimens were compared to original descriptions of the types, or directly with the type material available in museums. To visualise shape and size variation, measurements of intact skulls were also subjected to multivariate analyses via principal components extracted from the correlation matrix in SPSS-X (Norusis 1993); few missing data from slightly damaged skulls were replaced by mean values of the species.

Material for comparisons
Besides the newly captured bats, we also examined and reidentified small Myotinae specimens from Northern India and Nepal and some extra-limital material held in various museum collections, the acronyms of which are BMNH: the Natural History Museum (formerly British Museum (Natural

DNA analyses
Wing biopsies or tissue samples preserved in alcohol were extracted for genomic DNA using the Qiagen DNeasy blood and tissue kit (Bangalore, India) according to manufacturer's manual and eluted in a final volume of 200 µl TE buffer. Total genomic DNA from extra-limital samples was obtained with the same methods. Two mitochondrial genes commonly used in bat systematics were amplified: the cytochrome b (abbreviated hereafter CYTB) and the traditional barcode gene, part of the cytochrome C oxidase subunit 1 (COI). They were amplified with Molcit-F/Cytb-H (Ibáñez et al. 2006;Weyeneth et al. 2008) and UTyr/C1L705 (Hassanin et al. 2012) or VF1d/VR1d (Ivanova et al. 2007), respectively. PCR were carried out in 25 µl reactions containing 50-500 ng of genomic DNA. Each PCR was run with negative controls to monitor possible contaminants. PCR products were cleaned using ExoSAP Kit (Thermo Fisher Scientific, India) and sequenced in both directions with the PCR primers. Chromatographs were checked, assembled, and edited using Sequencher 4.1 (Gene Codes Corp., USA). In addition, we downloaded a set of published and unpublished sequences of various representatives of Eurasian Myotis taken from both the GenBank and BOLD (see Online Resource 1). This was a complicated task as several hundred CYTB and COI sequences of potential interest were available, but many are associated with problematic species labels. The spotted problematic species identifications included obvious labelling errors, old names that were not updated following recent taxonomic changes, and inaccurate or wrong specimen identification. To illustrate this problem, we cite here the whole mitogenome from a M. muricola specimen from Vietnam (Platt et al. 2018), which is wrongly labelled in the Genbank as "Myotis atacamensis", an endangered species endemic to Chile. In the Online Resource 1, we list the downloaded sequences used in the phylogenetic reconstruction and give the original species name as it appears in the databank along with our revised species assignation.

Phylogenetic analyses
Because most publicly available sequences came from disparate specimens for both markers or lacked data for either gene, aligned sequences were analysed separately for the COI and CYTB genes. Phylogenetic relationships were estimated with maximum likelihood (ML) and Bayesian inferences (BI). All analyses were done with a fully partitioned model, the best-fitting model for these partitions being evaluated with Modelfinder (Kalyaanamoorthy et al. 2017) and the Akaike Information Criterion. We inferred the ML tree using the edge-linked partition model in iQ-TREEv 2.0 (Nguyen et al. 2015;Chernomor et al. 2016) and obtained branch supports with 1000 non-parametric bootstrap resampling. For the BI, we run twice independently four parallel MCMC chains for 10 million generations obtained in MrBayes v3.2.0 (Ronquist and Huelsenbeck 2003). The chains were checked for convergence and appropriate effective sample size (ESS > 200) with TRACER v.1.5 (Rambaut and Drummond 2009). The initial 10% of sampled trees were discarded as burn-in. Posterior probabilities (PP) were subsequently computed from the consensus of the remaining trees. For all molecular reconstructions we used three species of Indian Miniopterus (Table 1) as a composite outgroup to Myotinae.
Level of genetic divergence between sequences were evaluated with the Kimura two parameter model implemented in MEGA X (Kumar et al. 2018) for comparison with values given for bats (Bradley and Baker 2001;Kruskop et al. 2012;Benda et al. 2016).

Results
Based on re-examination of the chiropteran holdings at ZSI Solan, we revised taxonomic identification of a few Myotis species from Himachal Pradesh (Saikia et al. 2011). The first revision pertains to a female specimen (registration number M-50) which was collected near Chamba, at 2480 m a.s.l. and was identified as "M. muricola" by Saikia et al. (2011). We captured six new individuals of the same taxon in various places of Himachal Pradesh, including in the forested areas of Mt. Karol, Sangla, and Narkanda. Externally, this small whiskered Myotis had a dense, shaggy dark brown fur. Individual hairs on the dorsum had light brown tips and the roots were blackish. The lighter tips gave a glossy sheen to the dorsal pelage. The ventral hairs also had dark roots and greyish brown tips but lacked the overall glossy appearance. The animals had a dark, hairy face with prominent bristle-like hairs and small feet with a wing insertion to the base of toe (Fig. 1a). However, contrary to the usual notched rear edge of ears found for instance in M. mystacinus or M. muricola (Fig. 1b), the deep emargination in the ears of this unusual taxon was even more pronounced and projected forwards as an arched lobe near the base of the ear (Fig. 1a). This peculiar emargination of the ear was noticed by Tomes as a typical attribute of V. caliginosus (Tomes 1859) and resembles the antitragus of Rhinolophus, as also mentioned by Dobson in the original description of V. blanfordi (Dobson 1871: 214). This ear lobe is also clearly visible in Submyotodon latirostris from Taiwan (Lin et al. 2004;Cheng et al. 2015) and on specimens of V. moupinensis from China (see Online Resource 4). The shape of the tragus also differs in being curved forwards and only tapering near the tip (Fig. 1a), reminiscent of many pipistrelles but unlike that of most Myotis species, the tragus of which being usually straight and tapering regularly towards the tip. In lateral view, the braincase of our specimens appears particularly flattened when compared to other species in the region (Fig. 2a); the skull also has a raised occipital part and pointed rostrum. This typical skull shape closely resembles that of the type of V. moupinensis (Fig. 1e) or S. latirostris (e.g. Figure 4a in Ruedi et al. 2015). The above three taxa also share a nyctalodont or semi-nyctalodont configuration of lower molars characterising the genus Submyotodon (Ziegler 2003), and all differ considerably from the skull of few other small nyctalodont Myotinae such as M. alticraniatus or M. siligorensis (Fig. 2d) (Kruskop and Borisenko 2013). The distinctive specimens from Himachal Pradesh, thus, correspond in all respects to Vespertilio (= Submyotodon) blanfordi described from the same area (Tomes 1859;  (Table 1) could be easily distinguished externally by their larger dimensions, by a combination of relatively uniform dark-brown fur (Fig. 1b), small ears and feet, and strong canines (significantly exceeding the height of third upper premolars, Fig. 2b), as noted by Benda (2010). The lower molars had a myotodont configuration, as observed in the examined type specimen.
Another misidentified specimen included in the ZSI Solan collections is a female individual (No. M-38) collected in the city of Solan (at 1500 m a.s.l.) and reported as "M. siligorensis" by Saikia et al. (2011). This specimen had a whitish ventral pelage (albeit barely discernible due to prolonged exposure to excessive UV-light), a globose braincase without sharply raising frontal parts and a relatively strong dentition, which are all characters excluding genuine siligorensis (Bates and Harrison 1997) but instead are typical for another poorly known Himalayan species, M. nipalensis. We captured two adults of this same taxon with mist nets set over the Narag River, 12 km SE of Solan. One was a pregnant female, which was quickly released after basic measurements and the other, a male, kept as a voucher for detailed morphological examination (Fig. 1c). The pelage of these animals was characterized by light brown dorsum and extensive whitish ventrum (usually only slightly lighter in other Himalayan Myotinae), strong but relatively small feet with wing membranes attaching to the base of the outer toe, and unnotched ears (Fig. 1c). Thumbs were thin and relatively long. The skull had a globose braincase, a slight concavity in the frontal parts when viewed in profile and a much-elevated coronoid process compared to the condyloid process (Fig. 2c). The dentition was relatively weak, however, not to the extent as seen in M. siligorensis where the canine hardly exceeds the length of the third upper premolar (Fig. 2d). The lower molars were all myotodont, like in most other Myotis. These morphological features concur very well with the original descriptions of M. nipalensis (e.g. Dobson 1871Dobson , 1878 and are in contradiction to the key characters (Bates and Harrison 1997) given for M. siligorensis.
Indeed, specimens of genuine M. siligorensis captured further east in North-eastern India (c. 400 km away from type-locality of that species) and housed in the ZSI Shillong collection (Table 1) were different. Although they superficially resembled M. muricola or M. nipalensis, they were darker and had only a slightly lighter venter without the prominent white fur typical of the latter (Fig. 1d) and smaller body dimensions than the former. They differed from the similar-sized S. caliginosus by their relatively larger feet and thumbs (compare Fig. 1a and d) and by the ears lacking deep emargination to the outer edge. The wing insertion, as in the other "whiskered" Myotis, was to the base of the outer toe. Craniodental characteristics of M. siligorensis were very distinctive, with a weak dentition (particularly the tiny, conical lower canines, see Benda 2010), sharply raised frontals and a globose braincase (Fig. 2d). The first two lower molars (m1 and m2) were nyctalodont but the last one (m3) was myotodont, in conformity with the type of M. siligorensis examined.
Finally, we also re-examined the series of specimens housed in the ZSI Solan collection identified initially as "M. mystacinus" by Saikia et al. (2011). They were all captured around Solan and apparently were relatively common. The relatively long feet and wing membrane attaching to the ankle immediately indicated that this taxon does not belong to a species from the mystacinus-morphogroup, but to a typical trawling Myotis in the longipes-laniger group (Topál 1997). Compared to the series of Kashmiri M. longipes used by Topál in his revision of the group (Topál 1997), these Himachal Pradesh specimens proved to be morphologically identical (albeit somewhat darker). As these bats could be easily distinguished by external characters from true members of "whiskered" Myotis, they were not included in the following morphometric analyses.

Multivariate morphological comparisons
Global PCA analyses of craniodental variation of Indomalayan whiskered bats suggest that size was correlated with the main component, which accounted for 70% of the total variance (Fig. 3). Indeed, all variables were highly and positively correlated with this first component. Specimens were thus ordered accordingly along this component, the skulls of the smaller species (alticraniatus) appearing on the negative side, and the larger ones (muricola) on the positive side of the axis (Fig. 3). The second component accounted for 11.3% of the total variance and separated essentially the Myotis from the Submyotodon species with little overlap between both taxa in this morphospace. The skull of nyctalodont M. alticraniatus were thus strongly differentiated from all Submyotodon individuals possessing similar cusp structure on lower molars.
To focus on the more difficult groups, we repeated this PCA on the Myotis species only. For this second analysis, the first axis (83.3% of variance) was also essentially a size component, which ordered the measured skulls according to overall size (Fig. 4). The second axis (6.8% of variance) was more strongly correlated with two measurements expressing the broadness of skull (BB and POC). According to this analysis, all M. muricola specimens were set in a distinct group, while those identified as M. davidii and M. nipalensis stood out owing to their relatively broader skulls. This morphospace separated most M. alticraniatus from the skulls of other small species (M. siligorensis and M. annatessae), while the single M. phanluongi was even more distinct. It is interesting to note that the five intact skulls of M. siligorensis from the Himalayas and the three skulls of M. annatessae analysed here were very similar in this morphospace, although both possess distinct lower molar configurations (nyctalodont versus myotodont, respectively) (Kruskop and Borisenko 2013).

Molecular analyses
We successfully generated 26 new sequences of CYTB and 40 of COI from a total of 48 specimens (Table 1). Except for the older specimens from the HNHM collections for which we could only amplify a shorter segment (about Fig. 1  1 3 400 bp) of the CYTB gene, these sequences were 1140 bp and 652 bp long for the CYTB and COI, respectively. All new sequences were deposited in the GenBank under numbers MW054869-MW054932 (Table 1).
For both markers, mean intraspecific comparisons within the Himalaya were within 1% of sequence divergence (K2P model; Online Resources 2 and 3). Species sampled over wider geographic areas were more distinct, with mean K2P genetic distances ranging up to 5% within M. muricola sensu lato (s.l.). Comparisons involving distinct species were usually higher than 9%. Exceptions for the CYTB dataset included the two partial sequences of paratypes of M. csorbai from Nepal that were almost identical to the western Himalayan M. longipes (K2P 0.2-0.5%), topotypes of M. badius and M. alticraniatus from China (< 1%), or M. annamiticus and some M. laniger (down to 2.2%). These pairwise comparisons were much below what is expected for inter-specific comparisons (Baker and Bradley 2006). Within the genus Submyotodon inter-specific distances ranged from 9 to 14% for both mitochondrial markers, whereas they were higher when compared to sequences of Myotis (16 to 23% K2P distance).
Phylogenetic relationships of the Myotinae inferred with the different methods (BI and ML) and for each gene were mostly congruent (Figs. 5 and 6). Discordances were essentially due to the poorly resolved, deeper nodes (with less than 75% support), indicating either that the datasets were  insufficient to resolve these ancient nodes, or that basal relationships in these Myotinae reflect unresolved polytomies (Whitfield and Lockhart 2007). One consistent relationship was the basal position of the strongly monophyletic Submyotodon clade, which branched off early in the Myotinae radiation. Crown groups were much better resolved and concordant between both genes. The focal taxa from the Himalayas were all placed with strong support in the following clades. The diverse clade comprising the samples of M. muricola s.l. was monophyletic and formed a sister group relationship with another group of closely related taxa including M. nipalensis, M. annatessae (Fig. 5) and M. ancricola (Fig. 6). Indian samples of M. siligorensis were firmly placed as a sister species of M. longipes with strong support in both genes. The latter species included the two partial CYTB sequences of M. csorbai as well, supporting their close relationship (Topál 1997 and 6). This M. davidii clade included notably the two partial sequences of CYTB (AY699844) and COI (JX008074) considered as "M. nipalensis" by Tsytsulina et al. (2012) and Kruskop et al. (2012), respectively.

Discussion
Concordant with previous molecular reconstructions based on multiple mitochondrial and nuclear markers (Lack et al. 2010;Ruedi et al. 2015;Morales et al. 2019), nyctalodont or semi-nyctalodont species of Myotinae do not form a monophyletic unit as the three Submyotodon species and the few Asian Myotis taxa showing this dental character (siligorensis, alticraniatus, phanluongi, badius) (Tiunov et al. 2011) were phylogenetically unrelated (Figs. 5 and 6). Clearly, the special molar configuration in these small Myotinae evolved several times independently and may not serve as a good indicator of phylogenetic relationship in this subfamily. Reexamination of the external morphology of the three known species of recent Submyotodon also indicates that they can easily be distinguished from true Myotis by their unique shape of the ears: they all possess a prominent, arched lobe to the rear edge projecting forwards at a right angle (Fig. 1a and Online Resource 4). This arched lobe was not observed in any Myotis and thus provides an external diagnostic character to identify members of Submyotodon in the field. Genetically, all three species of Submyotodon examined here differ conspicuously from one another (9.4-11.9%; Online Resources 2 and 3), which supports their specific distinction.
A further sample referred here as S. cf. moupinensis sampled north of the Qinling mountains in central China (Online Resource 1) differs genetically from these three species, suggesting additional taxonomic diversity in this poorly known genus.
Submyotodon caliginosus appeared to be relatively common in parts of Himachal Pradesh (Dobson 1871;present study), Uttarakhand (Chakravarty et al. 2020) and Sikkim (India), and also in Langtang Himal (Nepal; material in HNHM). This species also occurs in Punjab and Khyber Pakhtunkhwa, Pakistan (material in BMNH and HZM, respectively) and according to Benda and Gaisler (2015) occurs as far west as the Hindu Kush mountains in eastern Afghanistan, suggesting that the species is relatively widespread in the southern slopes of the Himalayas. On the other hand, S. moupinensis is probably much rarer and known from a few specimens recorded in southern China (Szechwan, Yunnan and Fujian) (Milne-Edwards 1872; Allen 1938), whereas S. latirostris is quite common but endemic to the mountains of Taiwan (Cheng et al. 2015;Ruedi et al. 2015). Museum collections holding potential representatives of these poorly known species need to be re-examined, to provide realistic distribution ranges for all of them.
Besides the distinctive Submyotodon species, our phylogenetic reconstructions showed that a number of small Asian Myotis characterized by a raised frontal part of the skull, doomed braincase, very small canines (Fig. 2)  This again suggests that similar craniodental traits shared by M. siligorensis s. str. and the latter species result from convergent morphological evolution.
Taxonomically, these new morphological and phylogenetic results imply that M. siligorensis s. str. (including the junior synonym darjelingensis Horsfield, 1855) must be considered as a species on its own. Its distribution range is thus probably restricted to Central and Eastern Himalaya, including parts of India, Nepal and Myanmar (Bates et al. 2005). The other taxa found further east into China and Indochina should be referred to as M. alticraniatus or to the allied taxa M. thaianus, M. phanluongi and M. badius. Sequences from topotypes of M. badius are intermingled among various geographic lineages of M. alticraniatus (Figs. 5 and 6) suggesting that they represent variants of a single species. Notice that all Chinese sequences available in the GenBank and labelled as "M. davidii", including a complete mitochondrial genome (KM233172; Wang et al. 2016), certainly represent M. alticraniatus as well (see Online Resource 2) as they are all within 0-5% of sequence divergence compared to our reference data in this group. Genuine M. davidii, as redefined by Benda and Gaisler (2015) and as understood here, are part of another, much more divergent (13-18%) clade related to M. mystacinus (Morales et al. 2019;Coraman et al. 2020). The main phylogenetic relationships presented here (Figs. 5 and 6) are largely corroborated by more global molecular analyses of over 100 Myotis species and based on a much denser sampling of up to 1610 nuclear loci (Morales et al. 2019), and hence unlikely to reflect only particular relationships of the mitochondrial genome.
The widespread and polytypic species M. davidii itself has also been the source of much taxonomic confusion. This Chinese taxon, originally described from Beijing, is quite  (Dobson 1878;Allen 1938;Benda and Gaisler 2015;Benda et al. 2016), but fell into oblivion when it was included in the long list of purported synonyms of "M. mystacinus" (e.g. Findley 1972) or ignored in global revision of Asian Myotis (e.g. Corbet and Hill 1992). The status of this taxon was revised by Benda and colleagues with a combination of morphologic and genetic characters, first under "M. nipalensis" (Benda and Tsytsulina 2000), then under "M. aurascens" (Tsytsulina et al. 2012) and finally as M. davidii (Benda et al. 2016). Although we concur with these revisions that M. davidii is a welldefined species that is morphologically and genetically distinct from any other species in the mystacinus morphogroup (Zhigalin 2019), we demonstrate here that the overlooked  (Dobson 1878). Likewise, records of "M. mystacinus" from the Indian Subcontinent (Bates and Harrison 1997) and Myanmar (Bates et al. 2005) also fit to the characters of genuine M. nipalensis and should all be reassigned to that species. According to this new taxonomic view, M. nipalensis is a typical species of the southern slopes of the Himalayan region but its precise geographic range needs to be evaluated with more museum material. The novel systematics arrangement of M. nipalensis proposed here also calls into question the possible occurrence of M. "annatessae" s. l. into the Himalayas and the specific distinction of the closely related M. ancricola (Kruskop and Borisenko 2013;Kruskop et al. 2018). Indeed, none of these new taxa were adequately compared with genuine M. nipalensis as the later was wrongly assumed to be related to M. davidii. Our phylogenetic results rather suggest that M. nipalensis, M. annatessae, M. cf. annatessae and M. ancricola are genetically remarkably similar (Online Resources 2 and 3), despite considerable geographic distances separating the samples. Compared to genetic distances observed in other Myotis bats, these taxa should be considered conspecific. Morphologically they also appeared similar in our PCA analyses (Fig. 5), but specific comparisons involving the type material should be done to support this taxonomic interpretation.
Given the accumulation of confused or wrong species identifications associated with sequences deposited in the GenBank (Bidartondo et al. 2008;Leray et al. 2019), a more general consequence of the taxonomic changes proposed here for several Myotinae of the Himalayas is that it becomes urgent to curate such useful genetic databases with more taxonomic attention. It is indeed more and more popular to identify species from difficult groups such as Myotis bats with barcode approaches (Francis et al. 2010;Clare et al. 2011), including in epidemiological studies of zoonotic diseases (Nadin-Davis et al. 2012;Mohammadi et al. 2019), but such approaches depend entirely on the reliability of taxonomic identification of reference sequences. Unfortunately, with an exponentially increasing number of submitted sequences with questionable species tagging, it becomes even harder now to evaluate the accuracy of these reference databases. The BOLD system partly solved this problem by building a reference system based on barcode index numbers-BIN (Ratnasingham and Hebert 2013) rather than on binomial names, which group similar sequences together in a MOTU, regardless of their actual species identification. Still, when we submitted to the BOLD identification machine (as of 18/07/2020) our COI sequence of Myotis badius topotypes, no less than four different species (daubentonii, mystacinus, siligorensis and davidii) received a 100% probability of correct placement, none of which corresponding to the correct identification of that sequence. This system is furthermore largely restricted to sequences of the COI gene, impeding critical comparisons with other genes. GenBank has a far greater scope of genetic markers (Porter and Hajibabaei 2018), but because there is currently no possibility to correct or even flag problematic species assignation of sequences, it may actually become increasingly misleading to rely on it for automatized barcoding approaches (Shen et al. 2013). For instance, in a recent review evaluating species-level genetic diversity for mammalian species based on CYTB, Schrago and Mello (2020) provided practical guidelines to improve the performance of unsupervised algorithms for species delimitation. To highlight species deserving attention for possible cryptic lineages, they give the example of the endangered M. atacamensis showing "unusually high genetic differences" in that Neotropical species, overlooking the fact that one of the two reference sequences taken from the Genbank (MF143493; Platt et al. 2018) is actually a mislabelled M. muricola (see Online Resource 1), hence inflating their "intra-specific" statistics.
As shown in this study, a key to the success in resolving long-standing taxonomic riddles among Indomalayan Myotinae is the inclusion of new genetic material from near the type localities of these taxa. We think this approach should be applied to solve other taxonomic problems in the Indomalayan fauna or flora. Given the rapidly accelerating habitat degradation (Betts et al. 2017) and the possible loss of these unique elements of biodiversity (Ceballos and Ehrlich 2009), this should be considered as a priority area of research.