Striped Ichthyophis Fitzinger, 1826 (Amphibia: Gymnophiona: Ichthyophiidae) from the Western Ghats of peninsular India in the collection of London's Natural History Museum and their implications for taxonomy and distribution

Abstract. Eleven specimens of striped species of the Asian caecilian amphibian genus Ichthyophis from peninsular India are present in the collections of London's Natural History Museum. The most recently collected specimen of these was recently reported upon. We reassessed the other specimens that were accessioned in the period 1874-1955. We identified four species, I. beddomei, I. kodaguensis, I. longicephalus and I. tricolor, with five specimens reidentified from previous catalogue entries. Morphological data and photographs are presented. These specimens provide insights into the taxonomic history of Ichthyophis in India, especially G.A. Boulenger's synonymy of I. beddomei with I. glutinosus, and E.H. Taylor's resurrection of that species and elevation of I. tricolor from a ‘variety’ to full species. The material also includes, as far as is known, the earliest collected specimens of I. kodaguensis (accessioned 1882) and I. longicephalus (accessioned 1874), species described in 2007 and 1986, respectively. One specimen of I. tricolor from the Nilgiri hills proves this species is not restricted to the southernmost Western Ghats, south of the Palakkad (Palghat) Gap biogeographic barrier.


Introduction
The Asian ichthyophiid caecilian amphibian genus Ichthyophis Fitzinger, 1826 is the most speciose and widely distributed caecilian genus (e.g.Frost 2023) and needs further taxonomic investigation, with many species being poorly circumscribed and littleknown (e.g.Kupfer & Müller 2004, Kotharambath et al. 2012a, Wilkinson et al. 2014).Some of the limits in current knowledge are likely caused by a general lack of dedicated fieldwork, small available samples for many species, insufficient effort applied to the larger samples available for some species, relatively few distinguishing characters in operation, inadequate understanding of intra-and interspecific variation, and some key material (including types) having only vague locality information.In addition, many museum specimens are larvae, the basis for morphology-based identification of which is currently lacking.Taxonomic uncertainty precludes an accurate understanding of distribution, one of the key determinants of caecilian conservation assessments (e.g.Gower & Wilkinson 2005, Doherty-Bone et al. 2011, Kotharambath et al. 2012a).
In India, Ichthyophis occur primarily in the two main centres of amphibian diversity, in the Northeast, and in the Western Ghats region of peninsular India, each with their own endemic species and with no species known from both regions (Gower et al. 2017).For identification purposes, Ichthyophis can be divided into two non-monophyletic groups of species, those with a lateral yellow stripe on each side of the body (and usually at least part of the head) and those unstriped as adults (e.g.Taylor 1968).In the Western Ghats, there are five currently recognised endemic species of striped Ichthyophis: I. beddomei Peters, 1880, I. tricolor Annandale, 1909, I. longicephalus Pillai, 1986, I. kodaguensis Wilkinson, Gower, Venu & Venkatachalaiah, 2007, and I. davidi Bhatta, Dinesh, Prashanth, Kulkarni & Radhakrishnan, 2011, with no species names currently in junior synonymy.
In the 1800s and early 1900s, many natural history collections were sent from South Asia to Europe, especially the UK.As a consequence, London's Natural History Museum (NHM), formerly the British Museum (Natural History) (BMNH), holds important Indian biodiversity material.Although there are no type specimens of striped Indian Ichthyophis species in the NHM collection, the non-type NHM material nonetheless has played an important part in the taxonomic history of Western Ghats' Ichthyophis.For example, Boulenger (1882) considered I. beddomei Peters, 1880 as a junior synonym of I. glutinosus (Linnaeus) based on variation in numbers of annuli and position of the tentacle in NHM specimens, but Taylor (1960) considered the former to be a distinct, valid species because 'an examination of the material that he (Boulenger) studied now remaining in the British Museum shows that he has confused two or more forms'.Taylor (1968) used one of the London specimens to describe the morphology of I. tricolor, which he argued was a valid species rather than a colour variety of I. glutinosus.
There have been substantial advances in Indian caecilian taxonomy in recent years, with a great increase in the number of recognised species from 17 in 1998 to 40 now (plus an additional family and genus: Kamei et al. 2012), despite five species being relegated to junior synonymy in the same period (e.g.Gower et al. 2007, 2013, Kamei & Biju 2016).Additionally, some aspects of morphological variation have been documented in detail (e.g.Kotharambath et al. 2012aKotharambath et al. , b, 2015)), some new identification keys have been published (e.g.Giri et al. 2003, 2004, Bhatta et al. 2011b), and phylogenetic relationships for several species have been inferred from DNA sequence data (e.g.Gower et al. 2002, Wilkinson et al. 2002, Kotharambath et al. 2012a).However, much of this progress has come from the collection and examination of new specimens and comparison with types and less from reassessments of non-type historical material, including that held in London.
Here, we present a reassessment of the remaining ten striped Ichthyophis specimens from peninsular India in the NHM collection (Table 1), all collected before 1960.We identified three species (with five specimens Table 1.Summary of the ten pre-1956 specimens of striped, peninsular Indian Ichthyophis in the NHM collection (specimen numbers with BMNH prefix).Note that although precise collection dates are unknown, the catalogue numbers correspond to accession dates into the London collection.For example, specimen 1874.4.29.50 was the 50 th specimen accessioned into the NHM collection on April 29, 1874.Locality information presented here is as recorded in the catalogue -see main text for comments.* Although these specimens are listed as 'I.glutinosus' in the main BMNH catalogue, a note inside the relevant jar written by E.H.Taylor shows that the correct identifications had been made previously (probably before 1960 -the date of Taylor's Ichthyophis taxonomy paper that included assessments of this taxon).Taylor (1968).The specimen numbers are based on dates of accession into the NHM collection.For example, 1874.4.29.50 was the 50 th specimen to be accessioned on the 29 th of April, 1874.For an explanation of abbreviations of characters in the left-hand column, see Methods.In addition, AGs post-vent -number of AGs posterior to the vent; AGs vent -number of AGs interrupted by the vent; stripe width -approximately mean width of lateral yellow stripe measured at midbody; snout width -width of head halfway between nares and TAs (see Kotharambath et al. 2012a).An em dash (-) indicates the minimum distance between two features; for example, TA-N is the minimum distance between the tentacular aperture and naris.Vertebral counts are approximate (+/-1) because the precise number of vanishingly small terminal vertebrae is uncertain.

Material and Methods
In addition to the material reported here, we have examined the type specimens of all Western Ghats ichthyophiid caecilians.Measurements were taken to the nearest 0.1 mm using a digital calliper (Mitutoyo, Japan) except for total length, which was measured to 1 mm by stretching specimens along a ruler.Vertebral counts were made from radiographs using a Matchlett Solus Schall, beryllium window, copper target tube.Sex was determined by direct examination of gonads and other sexually dimorphic features of the urogenital system (Wilkinson 1990).
We identified the species based on our understanding derived from examining the types, and we used more than twenty taxonomic characters, including the morphometric and meristic features listed in Table 2.We followed previous workers (see Wilkinson et al. 2014) in using the following abbreviations for morphological features and ratios of measurements: AG -annular groove; E -eye; IM -inner mandibular tooth; N -naris; NG1 -first nuchal groove; NG2second nuchal groove; NG3 -third nuchal groove; OM -outer mandibular tooth; PM -premaxillarymaxillary tooth; ST -snout tip; TA -tentacular aperture; TL -total length; TN/TE -distance between TA and naris divided by distance between TA and eye; and VP -vomeropalatine tooth.Place names and physical geographical features referred to in the text are shown in the map in Fig. 1.

Results
The studied material (Table 1) comprises ten specimens of striped Ichthyophis from localities with mostly regional rather than precise collection data.Five specimens are from Nilgherries (Nilgiri hills), three from Nilgiri, Waynaad District (=Wayanad, a district of Kerala, north of the Nilgiris and a highland region that historically was sometimes considered as part of the Nilgiri hill complex (e.g.Burton 1851,  Western Ghats striped Ichthyophis in NHM London collections Francis 1908)), one from Wynad (Wayanad), and one from Piermerd (Peerumedu, also spelt Peermade, Peermed), Travancore.The latter region (Travancore) at the time of collection in the late 1800s was a kingdom (and native state of British India), approximately equivalent to the southernmost seven districts of present-day Kerala state.The collection dates of these ten specimens are not known, but the dates of accession must predate them in the collections of the Natural History Museum, London (Table 1).with qualitative characters including head shape and colour pattern and comparisons with type material, were used to identify specimens.We identified four species in the sample: one specimen each of I. kodaguensis and I. longicephalus, two of I. tricolor, and six of I. beddomei (Table 1).The I. tricolor specimens (Figs. 7,8) were identified primarily by their distinctive colour pattern with a whitish midventer, but also by number of annular grooves, tentacle position, and head shape (see e.g.Wilkinson et al. 2007, Kotharambath et al. 2012a).The I. beddomei specimens (Figs.4-6, 9-11) have a characteristically small and pointed head, tentacle only a little closer to the eye than to the naris, broad stripe, and a low number of annular grooves (see Peters 1880, Taylor 1968).The single male I. beddomei in the sample (BMNH 82.12.12.3) is not fully mature.It has a welldeveloped musculus retractor cloaca, but it has small (ca.1.5 × 0.7 mm), relatively smooth surfaced testes lobes and lacks copulatory loops (see Wilkinson 1990).Based on superficial examination (i.e.without further dissection), it does not appear to have a heavily ornamented phallodeum as expected of a mature male Ichthyophis (Gower & Wilkinson 2002).The I. longicephalus specimen (Fig. 2) has many more annular grooves than the other three species, a tentacle almost twice as far from the naris as from the eye, and a more U-shaped and rounded head (in dorsal view) than I. beddomei and I. tricolor (see Pillai 1986, Kotharambath et al. 2012a).The I. kodaguensis specimen (Fig. 3) has a similar number of annular grooves to I. tricolor but lacks a whitish midventer, and it has substantially fewer inner than outer mandibular teeth (see Wilkinson et al. 2007).

Discussion
Boulenger (1882) considered I. beddomei a junior synonym of I. glutinosus because examination of Beddome's material in London revealed a lack of clear correspondence between the numbers of annular grooves and the relative position of the tentacle.From our identifications, it appears that Boulenger was confused because Beddome's sample included not only I. beddomei but also specimens of I. kodaguensis and I. longicephalus (two species that had yet to be described) as well as I. tricolor (the latter seemingly considered by Boulenger as a colour variety of I. glutinosus, following Annandale 1909).Thus, Boulenger likely struggled to reconcile the variation he observed with Peters' (1880) description of I. beddomei based on a single specimen with 240 annular grooves and a tentacular aperture barely closer to the eye than the naris.Thus, we agree with Taylor (1960) that the NHM material available to Boulenger (1882) is consistent with I. beddomei being distinct from I. glutinosus, and we agree with Taylor (1968) that I. tricolor is also distinct at the species level.Mitochondrial DNA sequence data support the understanding (based on morphology) that I. beddomei and I. tricolor are distinct species, because specimens referable to these species differ substantially, and these species might not even be each other's closest relative (Gower et al. 2002, Kotharambath et al. 2012a).
Ichthyophis longicephalus was described in 1986 (Pillai 1986), more than 100 years after I. beddomei and more than 75 years after I. tricolor.However, our re-identification of NHM material demonstrates that at least one specimen of this species has been in a scientific collection since 1874.Similarly, I. kodaguensis was described as recently as 2007, despite a specimen occurring in the NHM collection since at  Western Ghats striped Ichthyophis in NHM London collections least 1882.Both the I. kodaguensis and I. longicephalus specimens were probably examined by scholars, including at least Boulenger and Taylor.These were two of the most enthusiastic and prolific describers of new species in the history of herpetology, yet in this case, both overlooked two undescribed species, underlining how challenging Ichthyophis taxonomy can be, especially when confronted with small samples and lack of precise collection data.
The distributions of striped Ichthyophis species in peninsular India are not well established.In general (e.g.Pillai & Ravichandran 1999, Gower et al. 2002, Kotharambath et al. 2017), I. tricolor is considered to be restricted to the southernmost part of the Western Ghats, south of the Palakkad Gap, a 25-30 km wide pass in the Western Ghats mountains (Fig. 1), and a noted biogeographic feature (e.g.Subramanyam & Nayar 1974, Gower et al. 2007, Vijayakumar et al. 2016).However, BMNH 1882.12.12.5 is reported to be from the Nilgiris, north of the Palakkad Gap.This specimen (Fig. 7) has the typical ventral colouration of the species, and its morphology also agrees with I. tricolor specimens (e.g.Fig. 8) from the south of Palakkad.Assuming that Beddome's locality information is correct, the range of I. tricolor extends further north than previously understood.
Verified records of I. beddomei are mostly from the Nilgiris and Wayanad and further north into southern Karnataka (e.g.Gower et al. 2004).Historical reports of I. beddomei from much further north into Goa and Maharashtra are likely of I. davidi, described in 2011 from northernmost Karnataka (Bhatta et al. 2011b), or of one or more, as-yet undescribed species that might remain to be discovered.Of less certain status are reports of I. beddomei from far south of the Palakkad Gap, in southern Kerala and Tamil Nadu states (e.g.Inger et al. 1984, Pillai & Ravichandran 1999, Subramanian et al. 2013).Addressing the precise range of I. beddomei is beyond the scope of this work, but the historical NHM I. beddomei specimens (Table 1) are all recorded as being collected from the Nilgiris (the type locality of the species) or the adjacent Wayanad plateau.
Ichthyophis longicephalus was originally described from Silent Valley (Pillai, 1986), a short distance north of the  Western Ghats striped Ichthyophis in NHM London collections Palakkad Gap, at the far southern end of the Nilgiris.This striped Ichthyophis species has subsequently been documented from further north, along the western escarpment of the Western Ghats of northern Kerala, including Wayanad (Kotharambath et al. 2012a) and into nearby southern Karnataka and far north-western Tamil Nadu (Venu et al. 2020).Ichthyophis kodaguensis has been reported only from southern Karnataka, from around Kodagu and up to 125 km to the north (Bhatta et al. 2011a, Subramanian et al. 2013, Venu et al. 2021) -it has not been recorded from Kerala or Tamil Nadu.Although the I. kodaguensis specimen BMNH 1882.12.12.1 is from 'Nilgherries', that term historically could have included Wayanad (e.g.Burton 1851, Francis 1908).Whether from the Nilgiris (as understood today) or Wayanad, this specimen suggests that the geographic range of I. kodaguensis extends some distance to the south of Kodagu.
Among the five currently recognised species of striped Ichthyophis occurring in the Western Ghats, only I. davidi is seemingly not represented among the NHM material.That species is known from around the tri-border between Karnataka, Goa and Maharashtra states (Bhatta et al. 2011b, Subramanian et al. 2013), so its known range lies far (ca.400 km) to the north of the localities possibly represented by the ten historical NHM specimens discussed here.
Reassessment of the NHM London material of striped Indian Ichthyophis sheds light on the taxonomic history of species belonging to this genus in the Western Ghats.It provides additional data on morphological variation in four of these poorly known species and, in one case (I.tricolor), challenges the current understanding of geographic distribution.Although new fieldwork is essential, reexamination of understudied museum specimens can still contribute to advances in caecilian systematics.

Fig. 1 .
Fig. 1.Map of southwestern peninsular India showing states (uppercase text), towns and villages (marked with circles), and geographical features mentioned in the text.
Western Ghats striped Ichthyophis in NHM London collections requiring re-identification), and our interpretations provide insight into the causes of previous taxonomic confusion.The available material also challenges the understanding that I. tricolor is restricted to the southernmost Western Ghats, south of the Palakkad (also spelt Palghat) Gap.

Fig. 2 .
Fig. 2. Ichthyophis longicephalus specimen BMNH 1874.4.29.50.The whole animal is in two views, and the head and collars (upper row) and posterior end of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 3 .
Fig.3.Ichthyophis kodaguensis specimen BMNH 1882.12.12.1.The whole animal is in two views, and the head and collars (upper row) and posterior end of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 4 .
Fig.4.Ichthyophis beddomei specimen BMNH 1882.12.12.2.The whole animal is in two views, and the heads and collars (upper row) and posterior ends of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 5 .
Fig. 5. Ichthyophis beddomei specimen BMNH 1882.12.12.3.The whole animal is in two views, and the heads and collars (upper row) and posterior ends of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 6 .
Fig.6.Ichthyophis beddomei specimen BMNH 1882.12.12.4.The whole animal is in two views, and the heads and collars (upper row) and posterior ends of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 7 .
Fig. 7.Ichthyophis tricolor specimen BMNH 1882.12.12.5.The whole animal is in two views, and the head and collars (upper row) and posterior end of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 8 .
Fig. 8. Ichthyophis tricolor specimen BMNH 1893.4.18.26.The whole animal is in two views, and the head and collars (upper row) and posterior end of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 9 .
Fig. 9. Ichthyophis beddomei specimen BMNH 1955.1.2.54.The whole animal is in two views, and the heads and collars (upper row) and posterior ends of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 10 .
Fig. 10.Ichthyophis beddomei specimen BMNH 1955.1.2.55.The whole animal is in two views, and the heads and collars (upper row) and posterior ends of the body (lower row) are in four views.Scale bar increments in mm.

Fig. 11 .
Fig. 11.Ichthyophis beddomei specimen BMNH 1955.1.2.56.The whole animal is in two views, and the heads and collars (upper row) and posterior ends of the body (lower row) are in four views.Scale bar increments in mm.