Identity of the holotype and type locality of Rhabdophis leonardi (Wall, 1923) (Colubridae: Natricinae), with notes on the morphology and natural history of the species in southwestern China

Abstract The original description of Natrix leonardi (currently Rhabdophis leonardi) by Frank Wall in 1923, based on a specimen from the “Upper Burma Hills,” lacked important morphological details that have complicated the assignment of recently collected material. Furthermore, although the holotype was never lost, its location has been misreported in one important taxonomic reference, leading to further confusion. We report the correct repository of the holotype (Natural History Museum, London), together with its current catalog number. We also describe key features of that specimen that were omitted from the original description, and provide new details on the morphology of the species, including sexual dichromatism unusual for the genus, based upon specimens from southern Sichuan, China. Rhabdophis leonardi is distinguished from its congeners by the following characters: 15 or 17 DSR at midbody and 6 supralabials; distinct annulus around the neck, broad and red in males, and narrow and orange with a black border in females; dorsal ground color light green or olive; some lateral and dorsal scales possessing black edges, the frequency of black edges gradually increasing from anterior to posterior, forming irregular and ill‐defined transverse black bands; eye with prominent green iris; black ventral spots with a red edge, most numerous at midbody but extending halfway down the length of the tail. In southwestern China, this species is frequently found at 1730–2230 m elevation. It has been documented to prey upon anuran amphibians, including toads. A recently published phylogenetic analysis showed this species to be deeply nested with the genus Rhabdophis, as a member of the R. nuchalis Group. That analysis also revealed the existence of two closely related but geographically distinct subclades in the molecular analysis, one of which may represent an unnamed taxon.


| INTRODUC TI ON
The natricine genus Rhabdophis Fitzinger, 1843 is widely distributed across southern and eastern Asia, from northeastern India and Sri Lanka through China to Japan, and south to the islands of Malaysia and Indonesia. Note that the Natricinae, considered a subfamily here and by many authors, (e.g., Zheng & Wiens, 2016), is regarded as a family by other recent authors (e.g., Burbrink et al., 2020;Zaher et al., 2019). At either rank, the content and relationships of the clade are equivalent. Malnate (1960) assigned 15 species to Rhabdophis when he partitioned it from the expansive nominal genus Natrix. He characterized Rhabdophis as having a terrestrial habitus, with enlarged posterior maxillary teeth, usually following a diastema. He noted that many of the species reportedly possessed nuchal or nucho-dorsal glands. Those integumentary glands had first been described in Rhabdophis tigrinus by Nakamura (1935) and soon thereafter were reported in other congeners by Smith (1938).
Initially of unknown function, those glands were only known from Rhabdophis, some species of Macropisthodon, and the monotypic genus Balanophis Boulenger, 1893, erected by Smith to accommodate Natrix ceylonicus (Smith, 1938).
In the years since Malnate's resurrection of Rhabdophis, the genus has expanded to include 32 currently recognized species, several of them identified or included on the basis of molecular analyses. Formerly widespread species, such as the nominal R. subminiatus (the type species), R. tigrinus, and R. nuchalis, have been found to harbor cryptic diversity at the specific level (David & Vogel, 2021;Takeuchi et al., 2012;Zhu et al., 2022). Furthermore, a comprehensive analysis of the Asian natricines that possess nuchal or nucho-dorsal glands, together with their nominal congeners, was conducted by Takeuchi et al. (2018), who determined that the genus Macropisthodon, as constituted at that time, was paraphyletic. Two of the species included in that study (M. flaviceps, the type species, and M. plumbicolor) were found to be nested among the species assigned to Rhabdophis, as was Balanophis ceylonensis, whereas M. rudis lay far outside the so-called "nuchal gland clade" of Natricinae. Meanwhile, Figueroa et al. (2016) had found that M. rhodomelas also lies within Rhabdophis. Therefore, Takeuchi et al. (2018) formally synonymized Balanophis and Macropisthodon with Rhabdophis, uniting all species with nuchal or nucho-dorsal glands within that single genus, together with a few populations that appear to have lost the integumentary glands secondarily. Rhabdophis spilogaster was recently sequenced, and it was moved to the genus Tropidonophis (Deepak et al., 2022), whereas the phylogenetic positions of Rhabdophis auriculatus, R. chrysargus, and R. conspicillatus are unclear at this time and remain to be resolved (Deepak et al., 2022). Of the 32 currently recognized species of Rhabdophis, 14 occur in China, including Taiwan: R. adleri Zhao, 1997;R. chiwen Chen, Ding, Chen and Piao, 2019;R. confusus David & Vogel, 2021;R. formosanus (Maki, 1931); R. guangdongensis Zhu et al., 2014;R. helleri (Schmidt, 1925); R. himalayanus (Günther, 1864); R. lateralis (Berthold, 1859); R. leonardi (Wall, 1923); R. nigrocinctus (Blyth, 1855); R. nuchalis (Boulenger, 1891); R. pentasupralabialis Jiang & Zhao, 1983;R. siamensis (Mell 1931); and R.
swinhonis (Günther, 1868). A recent study by Zhu et al. (2022) revealed additional cryptic diversity within the derived worm-eating clade of southwestern China, designated the R. nuchalis Group, to which R. leonardi belongs.
Meanwhile, the morphology and chemistry of the nuchal and nucho-dorsal glands have been extensively studied and substantially clarified. The glands are now known to serve a defensive function by releasing noxious steroidal compounds known as bufadienolides that the snakes sequester from toxic prey (Hutchinson et al., 2007;Mori et al., 2012). Superimposing both the anatomical distribution of the integumentary glands (i.e., whether they occur along with the entire length of the body, as nucho-dorsal glands, or are limited to the neck, as nuchal glands) and the dietary source of the toxins (whether from amphibians or insects) on the phylogeny, it is clear that Rhabdophis likely ancestrally possessed nucho-dorsal glands that contain bufadienolides derived from a diet of anuran amphibians (Takeuchi et al., 2018;Yoshida et al., 2020). Importantly, the phylogeny also reveals that some members of a deeply nested clade that occurs in southwestern China and adjacent regions underwent a shift in their primary diet from frogs to earthworms. Accompanying that change in diet was a shift in the source of the sequestered defensive toxins from toads (Bufonidae) to lampyrid firefly larvae, both of which contain bufadienolide steroids (Yoshida et al., 2020).
In the course of identifying specimens belonging to the wormeating clade of Rhabdophis from Sichuan Province, China, we encountered inconsistencies in our comparison of the new specimens with the original description of R. leonardi (Wall, 1923). Furthermore, our attempts to compare our specimens to the holotype of that species were confounded by erroneous information on the repository of the holotype, a problem exacerbated at the time by travel restrictions and museum closings associated with the COVID-19 global pandemic. Wall's (1923) original description of the holotype was insufficient to distinguish the species from among the greater diversity of Rhabdophis recognized today, and some critical attributes of the type specimen had been omitted, notably its sex. Ironically, this species appears to have the most extreme sexual dichromatism of any member of the genus. Finally, after the holotype had been located and our specimens had been determined to conform morphologically with R. leonardi, our efforts were further complicated by the presence of two well-differentiated clades within that nominal species, as documented in the molecular assessment of the R. nuchalis Group by Zhu et al. (2022).
Here we correctly identify the repository of the holotype of Natrix leonardi Wall, 1923, describe the historical context behind the discovery of that specimen and the significance of the type locality, and provide important additional details on the morphology of the holotype, including its sex. We also address the phylogenetic relationships of species based on the recent work of Zhu et al. (2022) and a new molecular phylogenetic analysis. We describe in detail the morphology of a male specimen (for comparison with the female holotype; for the determination of sex of the holotype, see below), briefly describe several additional female specimens based on recent material, and describe a hatchling from the Sichuan population, including the coloration of these specimens in life and in preservative. We report on two additional localities in neighboring Yunnan Province, based upon field observations and photographs, and provide information on the natural history of the species in China.
Finally, we discuss the likely association of the name Rhabdophis leonardi with one of the two molecular clades that currently bear that name and suggest fruitful directions for future studies.

| Phylogenetic analysis
A phylogenetic tree was inferred from analyses of concatenated sequences of one mitochondrial (cyt b) and one nuclear (c-mos) gene.
Sequences were aligned by Mega 5.0 (Tamura et al., 2011). A total of 1599 base pairs for 50 samples were analyzed in this study, and Natriciteres olivacea (Peters, 1854) was selected as an outgroup. The sequences used for constructing our phylogenetic tree are listed in Table 1. The phylogenetic analysis employed maximum likelihood (ML) and Bayesian inference (BI) methods. The best-fitting model of sequence evolution for the BI analysis was determined using PartitionFinder 2.1.1 (Lanfear et al., 2017), separating all genes by codon position and identifying the best-fitting partitions scheme, employing the Akaike information criterion (Akaike, 1974), and the phylogenetic analysis was performed using MrBayes 3.1.2 (Huelsenbeck & Ronquist, 2001

| Morphological data
The holotype of Natrix leonardi Wall, 1923, once located, was closely examined at its repository, and the data were compared to the original description to verify its identity. Photographs of the holotype were taken by Kevin Webb of the Photo Unit of the Natural History Museum, London.

Rhabdophis leonardi
The topology of the maximum likelihood (ML) tree was consistent with that of the Bayesian inference (BI) tree ( Figure 1). With respect to the position of Rhabdophis leonardi and its substructure, our phylogenetic results largely conform to those of Zhu et al. (2022), who focused on relationships among the R. nuchalis Group, to which R. leonardi belongs.

| Type locality of Natrix leonardi
In 1923  has several spellings and appears in two physical locations on older maps, whereas it does not appear at all in searches of recent online maps.
In the 1920s P. M. R. Leonard, a member of the Burma Frontier Service, traveled repeatedly to the region near the Chinese border.
By the time a survey of the aquatic mollusks had been conducted in the region, to assess the presence Schistosoma, Leonard was cited in the acknowledgments as Assistant Superintendent of the Northern Shan States, serving at Kutkai (Rao, 1928). Today, Kutkai is approximately 240 km by road (via Highway 3) southeast of Bhamo, the closest major city to the type locality (Google Maps, 2022) and, even in colonial times, an important port along with the Irrawaddy (or Ayeyarwady) River.
Wall (1923)   The type locality, Sinlum Kaba, lies in a triangular mountainous region above and to the east of Bhamo, in southeastern Kachin State.
Recently recorded family history, published in a series of articles in the Katchinland News, sheds some light on the still confusing history of Sinlum Kaba (Lahpai, 2017(Lahpai, , 2020Pangmu, 2011), from which the following account is synthesized. From those published clan histories, it appears that Sinlum Kaba was established by the Gauri Lahpai tribal leader Zau Bawm in the early to mid-1800s (Lahpai, 2017).

By the late 1800s Baptist missionaries William Henry Roberts and
Ola Hansen had established a church and school in Bhamo, where they baptized Zau Tu, a member of the Gauri Lahpai chieftain clan, and Hka Jan (Lahpai, 2017;Pangmu, 2011 (Pangmu, 2011).
The British outpost grew in importance, with a fortified military station, a center for handling legal matters involving the Kachin people, and plantings of fruit trees imported from England (Lahpai, 2020).

Meanwhile, a high school that had been established in Pangmu by
Zau Tu and Hka Jan in 1903 was moved to Sinlum Kaba in 1928.
Lahpai (2020) notes that Hka Jan was recognized for her contributions as an educator at a Durbar (a gathering of regional dignitaries) in Sinlum Kaba in 1923. Such an event may well have involved a visit by colonial authorities such as Leonard, affording him an opportunity to acquire the specimens reported that year by Wall.
A search for "Sinlum Kaba" or "Sinlumkaba" in current online maps (Google Maps or Google Earth) fails to return any records. clearings, which are located just over 2 km apart. Both clearings appear to contain substantial settlements, including several large structures, especially at the southern site, which also appears to include a helipad. These two clearings presumably represent the two

| Location of the holotype
Recent information on the location and identity of the holotype of Natrix leonardi presented similar ambiguities. Both of Wall's reports on collections of snakes from Sinlum Kaba (Wall, 1921(Wall, , 1923, each describing one new species of snake as a patronym for P. M. R. Leonard, imply that the specimens were donated to the Bombay Natural History Society (BNHS), whose museum collection was earlier recognized by the acronym BNHM (Leviton et al., 1985).
Since 2020 the standard acronym of the BNHS museum collection is BNHS, and the acronym BNHM is now obsolete (Sabaj, 2020).
However, only one of the holotypes of those two species has been specifically described in the literature as being deposited at the BNHS (Das & Chaturvedi, 1998), and even that designation may be in error. Importantly, the Reptile Database cites the holotype of Natrix leonardi as "BNHS = BNHM 466" (Uetz et al., 2022;accessed 25 March 2023). One of us (RK) examined and photographed that specimen and found it to be a colubrine snake, not a natricine. Indeed, that specimen is listed in the BNHM (now BNHS) type catalog (Das & Chaturvedi, 1998) as the holotype of Coluber leonardi (Wall, 1921; Figure 3), described in the Journal of the Bombay Natural History Society, and now considered a synonym of Archelaphe bella (Schulz et al., 2011). We note, however, that BNHM 466 appears in the photograph to be considerably larger than the holotype of Coluber leonardi (279 mm SVL; Wall, 1921). We suggest that BNHM 466 may, in fact, be the third and much larger specimen of Coluber leonardi (685 mm SVL), which was reported by Wall (1923) in the same paper in which he described Natrix leonardi. Thus, the identity and repository of the holotype of Coluber leonardi remains to be clarified, but resolution of that question is beyond the scope of this report.
Because the holotype of Natrix leonardi was not found in the BNHM and was not listed in the type catalog for that collection, we next examined the holdings of the Natural History Museum, London (formerly British Museum (Natural History)), whose catalog number prefix for historical material is BMNH, and identified a specimen of Rhabdophis leonardi, BMNH 1946.1.12.86 (formerly BMNH 1923, from the Upper Burma Hills (Figure 4). Upon examination by two of us (DG and VD) and comparison with Wall's (1923) description, that specimen was determined to be the holotype of N.
leonardi and, indeed, it was listed as a "type" in the BMNH catalog. As with other type specimens in the London collection, that specimen Green squares represent the locality of referred material, including two localities described in Wall (1921Wall ( , 1923 (Smith 1921) (now Archelaphe bella) and of Natrix leonardi (Smith 1923) (now Rhabdophis leonardi). The specimen is definitively not the holotype of Natrix leonardi and may not be the holotype of Coluber leonardi. See text for a full discussion of this specimen.
was re-cataloged and assigned a new number after the end of World War II when specimens were returned from safekeeping.
Although designated as a type specimen and bearing the more accurate coordinates for the type locality than those published by Wall (1921Wall ( , 1923 to the confusion over the identity and repositories of the two type specimens is the fact that the recommended acronym for the herpetological collection of the Bombay Natural History Society, BNHM, is an anagram of that for the Natural History Museum, London, BMNH (Leviton et al., 1985). Furthermore, the catalog number of the presumed holotype of Coluber leonardi in Mumbai is BNHM 466 (Das & Chaturvedi, 1998), which coincidentally is the same number as the page on which the description of Natrix leonardi appeared 2 years later, in the same journal (Wall, 1923).

| Partial redescription of the holotype of Natrix leonardi
The original description of Natrix leonardi (Smith, 1923), based on a single specimen, provides sufficient detail to confirm with confi- We note that Parker (1925), in a study of dorsal scale row reduction in specimens he identified as Natrix nuchalis, placed the reduction from 17 to 15 dorsal rows at approximately 45% of "Body Length" in the "Type" of Natrix leonardi (presumably then cataloged as BMNH 1923.10.13.39) and at over 50% of "Body Length" in a second specimen recognized, doubtfully, as the same species. Importantly, Parker did not define "body length," but he distinguished the body from a region he identified as the "neck" (but also did not define).
Parker considered that all the species he examined were conspecific with what is now Rhabdophis nuchalis, but that conclusion has not been supported by recent analysis of that complex species group .

| Diagnosis
Rhabdophis leonardi is characterized by the following combination    (Figure 7).

| Description of females
Morphological data for the other female specimens are listed in Table 2 and Figures 5,6. The female specimens differ most importantly from the male in the size and color of the nuchal collar.
In adult females the collar is narrower (extending for about three to four scales) and is orange with narrow, irregular anterior and posterior black borders ( Figure 6). Although our sample is small, we believe this represents sexual dimorphism in the color of the nuchal collar, a feature that should be confirmed by future studies. The female hatchling generally resembles the adult females in nuchal coloration, although the collar is yellow rather than orange, contrasting more strongly with its prominent black borders ( Figure 5c).

| Comparisons
The number of dorsal scale rows in R. leonardi (17-17/15-15) distinguishes it from all congeners except R. auriculatus (Table 3), in which the scale rows reportedly also are reduced from 17 to 15 at about the midbody (Leviton, 1970). However, R. auriculatus, which is endemic to the Philippines, has white lines on the sides of the body and behind the eyes, as well as white spots on the dorsum.
Within China, males of Rhabdophis leonardi most closely resemble R. helleri and R. siamensis in coloration. R. helleri was recently elevated from a subspecies of R. subminiatus to a full species by David and Vogel (2021), an action supported on molecular evidence by Liu et al. (2021), and the species has a wide range across southern China. R. siamensis, also resurrected from the synonymy of R. subminatus by David and Vogel (2021), oc-  (David & Vogel, 2021;Zhao, 2006).  Table 3.

| Habitat, behavior, and distribution
Residents of Xiaoshanbao Village (Figure 8) report that they occasionally encounter this species. Therefore, we suggest that this species occurs in southern Sichuan and adjacent Yunnan at approximately 1730-2230 m elevation, although it occurs at higher elevations at more northern localities.
A blurry photograph of a toad being consumed by this species (Figure 5d) also was obtained from a resident of Huangcao Village.
R. leonardi, therefore, is known to feed on earthworms and slugs (Yoshida et al., 2020;Zhao, 2006;Zhao et al., 1998), as well as anuran amphibians. We also note a catalog entry at the California Academy of Science indicating that CAS 215027, identified in the catalog as R.
leonardi, contained snails in its stomach.
Like other Rhabdophis, R. leonardi is oviparous. We also note here the presence of eight specimens at the

| DISCUSS ION
Species of the genus Rhabdophis are widely distributed across southern and eastern Asia, with an elevational range from sea level to more than 3000 m (Zhao, 2006). The existence of cryptic species of Rhabdophis has been suspected (Liu et al., 2021;Takeuchi et al., 2018), and several of those have recently been described (David & Vogel, 2021;Zhu et al., 2022). R. leonardi, as currently recognized, has a wide distribution in southwestern China and adjacent Myanmar. However, our data and those of Zhu et al. (2022) suggest that the two well-defined molecular lineages of nominal R.
leonardi, referred to here as Clades B and C of Zhu et al. (2022) Rhabdophis leonardi appears to prey in part upon anuran amphibians, including bufonids, in contrast to at least some of the other, and generally smaller, members of the earthworm-eating R. nuchalis Group to which R. leonardi belongs (Piao et al., 2020;Yoshida et al., 2020). For example, R. pentasupralabialis has a total length of approximately 483 mm and R. chiwen has a total length of about 536 mm (Piao et al., 2020), versus about 600-700 mm in R. leonardi.
In all these aspects of diet and size, R. leonardi appears intermediate between other members of the R. nuchalis Group and more general- Finally, our molecular results confirm previous findings that earlier concepts of Rhabdophis were not monophyletic. Specifically, R.

ACK N OWLED G M ENTS
This study was supported by a grant from the National Natural History. We are grateful to Yangyang Liu for assistance with specimen collection and examination. We also thank Gernot Vogel for assistance with revisions and for professional advice. We are grateful to Kevin Webb for the photographs of the holotype, and we thank Zhijie Jian for help with some additional photos. VD's contribution was supported in part by Humboldt fellowship hosted by Uwe Fritz at the Senckenberg Dresden.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.