TAXONOMIC STUDIES OF INDIAN BANDICOOT RATS ( RODENTIA : MURIDAE : MURINAE ) WITH DESCRIPTION OF A NEW SPECIES

Gray (1842) separated the bandicoot rats from the house rats under the genus Nesokia. Thomas (1907), however, divided them into three genera, Nesokia, Gunomys and Bandicota. Later, Wroughton (1908, 1919) maintained 6 species (gigantea, malabarica, elliotana, indica, nemorivaga and savilei) under the genus Bandieota, 7 species (bengalensis, gracilis, wardi, varius, lordi, sindicus and kok) under Gunomys and 4 species (indica, huttoni, griffithi and beaba) under Nesokia from the Indian subregion. Subsequently, Ellerman (1947, 1961) retained the genus Nesokia for the highly specialised bandicoot rats from Palaearctic and North-west India, Bandicota for the more generalised Indo-Malayan forms, and synonymised Gunomys with Bandieota. Further, based on the body colour and tnorphological characters, Ellerman (loc. cit.) maintained a single species indica under the genus N esokia and two species, namely, indica and bengalensis under Bandicota. While doing so, he merged all the large-sized bandicoot rats with indica except nemorivaga and savilei which were given subspecific ranks under it. Tiwari et al. (1971), however, stressed the need of retaining malabariea from the Western Ghats as a separate subspecies of B. indica. Later, Pradhan et. al. (1989), with the help of biochemical analysis found polymorphic populations in the species Bandicota indica which created confusion as to the status of different species synonymised with it. Hence, it was decided to undertake the study of large bandicoot rats afresh, covering all pos~ible aspects like osteo-morphological, biochemical and hair sculpture studies. The present work is based on the data collected for the following research projects :1. Ecological and taxonomic studies of the rats (subfamily Murinae) from Pune and adjacent areas. 2. Chaemotaxonomic studies of the commensal rodents and shrews from Bombay-Pune region.


INTRODUCTION
separated the bandicoot rats from the house rats under the genus Nesokia.Thomas (1907), however, divided them into three genera, Nesokia, Gunomys and Bandicota.Later, Wroughton (1908Wroughton ( , 1919) maintained 6 species (gigantea, malabarica, elliotana, indica, nemorivaga and savilei) under the genus Bandieota, 7 species (bengalensis, gracilis, wardi, varius, lordi, sindicus and kok) under Gunomys and 4 species (indica, huttoni, griffithi and beaba) under Nesokia from the Indian subregion.Subsequently, Ellerman (1947Ellerman ( , 1961) ) retained the genus Nesokia for the highly specialised bandicoot rats from Palaearctic and North-west India, Bandicota for the more generalised Indo-Malayan forms, and synonymised Gunomys with Bandieota.Further, based on the body colour and tnorphological characters, Ellerman (loc.cit.) maintained a single species indica under the genus N esokia and two species, namely, indica and bengalensis under Bandicota.While doing so, he merged all the large-sized bandicoot rats with indica except nemorivaga and savilei which were given subspecific ranks under it.Tiwari et al. (1971), however, stressed the need of retaining malabariea from the Western Ghats as a separate subspecies of B. indica.Later, Pradhan et.al. (1989), with the help of biochemical analysis found polymorphic populations in the species Bandicota indica which created confusion as to the status of different species synonymised with it.Hence, it was decided to undertake the study of large bandicoot rats afresh, covering all pos~ible aspects like osteo-morphological, biochemical and hair sculpture studies.
The present work is based on the data collected for the following research projects :-1.Ecological and taxonomic studies of the rats (subfamily Murinae) from Pune and adjacent areas.Nlore than 100 bandicoot specimens along with their skulls were studied in detail for the present work.The material, in addition to the freshly collected specimens, included the specimens present at the Bombay Natural History Society, Bombay, Zoological Survey of India, Calcutta, and the Western Regional Station of the Zoological Survey of India, Pune.The freshly collected specimens have been depOsited at the Western Regional Station, Pune.
For osteomorphological studies, all measurements were taken after Roonwal and Agrawal (1966).The freshly collected material as well as the already identified specimens (vide Ellerman 1961) were reidentified with the help of keys provided with by W 7 roughton (1919) and then compared.For comparison of data only adult specimens were taken into consideration.The significance or student ,t' test was applied fo every character showing statistically significant differences (P = 0.05) in the average measurements.
For biochemical studies, haemoglobins were separated according to method described by \Xlright (1974).The samples of haemoglobin and plasma protein were resolved in individual patterns using polyacrylamide gel electrophoresis (PAGE) after Davis and Ornstein (1961), Whitaker (1967), andGordan (1980).PAGE separation was carried out under carefully controlled factors like gel concentration (7.5%), pH of stacking (8.3) and running (9.5) gels, buffer system (Tris-glycine, pH 8.3), voltage current (4 watt per tube) temperature (4°C ± 1°C), the time of run, etc.The dye, bromophenol blue, mixed with sample before loading on the gel columns in the neutral glass tubes, served as a marker.To identify various specific proteins, gels were stained after Gordon (1980) for plasma proteins, Brewer and Sing (1970) for Lactate dehydrogenase (LDH) and non-specific esterases, and Ornstein (1967) for haemoglobin (Hb) fractions.The eye lenses were extracted according to Smith (1971), with certain modifications (Pradhan and Bhagwat 1990).After PAGE separation, the eye lens proteins were stained by the method of Gordon (1980).Consolidated protein profiles were prepared by analysing each sample in several replicates and averaging the electrophoretic mobilities with reference to marker (Rm values) for individual specimen.The final Rm values, obtained for individuals, were clubbed together to obtain characteristic profiles for the three populations of Bandicota under investigation.
For the analysis of hair structure, hair samples were collected from the region posterior to the neck on the dorsal surface.Five specitnens each of Bandicota spp.under study were selected for the present work.For light microscopic study, the hair smaples were first washed in warm water and then transferred to detergent solution (Teepol, 1 % v/v).After this treatment, hair were repeatedly washed with warm distilled water and transferred to 1 : 1 mixture of ether and alcohol as suggested by Dreyer (1966).After shaking this mixture, hair were once again washed with distilled water and dried in clean watch glasses.For routine study impressions of of hair sculptures were obtained on gelatin or polyvinyl acetate (Brunner and Coman, 1974) and photographed using Olympus microphotography attachment at x 200.
OBSERVATION &.DISCUSSION Dsteo-morphological Study The external and skull-measurement of the large-sized bandicoot rats (Tables 1   & 2) show that out of about 100 specimens examined from the distributional range of indica, malabarica and nemorivaga, 15 have the occipitonasal length more than the condylobasallength.The reverse is true in the rest.Not only that, these specimens are, on average, larger in size.When the average measurements (with standard deviations) of these specimens were plotted on a graph against the average measurements of indica, malabarica and nemorivaga (Figs. 2, 3, 4 &. 5 ) for comparative study, distinct differences were noticed in the lengths of occipitonasal, condylobasal, palate and diastema and width of zygomatic arches.All these differences were found to be statistically significant (P = 0•05).Although the measurements of these specimens come quite close to those of B. indica malasbarica, yet these differ in the length of occipitonasal, condylobasal and palate, and width of zygoma.Moreover, the longer occipitonasal, wider zygomatic arches and inflated occiput {Fig.6 & 7) give a somewhat triangular shape to the head of these large-sized rats (Fig. 8).
From the above study it is clear that Bandicota i. malabarica not only differs from the other two populations, viz., B. indica indica and B. indica nemorivaga in the length of nasals, diastema and palate, but is also allopatric in distribulion.Hence,.
it is treated as a separate subspecies of Bandicota indica (Bechstein).Our view finds support from the earlier work of Tiwari et al. (1971) who maintained malabarica as a separate subspecies of Bandicota indica.
As mentioned above, the large-sized bandicoot rat, Bandicota sp.differs from Bandicota indica (all three subspecies) in the occipitonasal length being more than condy1obasallength, and in the width of zygomatic arches and length of mandibles (Tables 1-3).Although these bandicoot rats (Bandicota sp.) come very close to Bandicota indica malabarica, yet cannot be placed as a subspecies of Bandicota indica due to its India-wide distribution.Hence, it deserves a specific rank.

BIOCHEMICAL STUDY
The consolidated population profiles for five specific proteins of three bandicoot populations in question are represented in Figs. 9 &.10.The data on protein, separation was used to calculate Genetic Identity (I) at the specific locii (Nei, 1972).:r 1) ", c: Table 4 represents the I values for individual proteins as well as mean I (1) for all . ..• .,. ..
.'.:1_ the proteins separated during the present study.On the basis of the data on I values .lDdalso applying the UPGMA method of cluster analysis, dendrograms shoWing relationships of the Bandicota species were also constructed (Fig. 11).

FIG. 6
Occiput region of  observed that for the two enzyme fractions, B. bengalensis showed, a greater gene identity with the proposed Bandicota sp.than with Bandicota indica which had the least gene identity.The genes representing low molecular plasma proteins showed greater identities in the populations of B. bengalensis and B. indica (0.85).Here ag~in, the genetic identity between B. indica and proposed Bandicota sp. was the least (0.64).

---
Dendrograms constructed from the above data (Fig. 11) clearly establish the patterns of branching in phylogeny of the three species.The dendrogram representing average genetical identities at locii controlling the five specific proteins suggests that all the three species of genus Bandicota were separated from each other more or less at• the same time.However, it was B. indica that got separated early from the common ancestral stock, whereas B. bengalensis and proposed Bandicota sp.sep-arated at a latter stage in the phylogeny.Selander and Yang (1969) have suggested that subspecies should not have less than 90% identity at genomic level.They further Rtate that sibling species show an identity close to 50% ; and when identity is about 30% the population should be treated as a distinct species.Our results average about 50% identity for five protein expressions studied.'Therefore, if one looks at the entire genome level and with a larger number of species' specific proteins, this identity might come down to the level of distinct species.To conclude, therefore, it may be stated that on the basis of the analysis of five protein expressions the populations of B. indica and proposed Bandicota sp.cannot be treated as a single species.All the three species (bengalensis, indica and B. sp.) appear to be genetically distinct and hence, should be given the status of independant species..ri1C"'1 -en _( 2) . ( 5) _ ( 5) _( 11) _ ( 7) -") ,-(~) -( 1) _ ( 6) _ ( 5) -5 -( q ) _ ( 4) . ( 4) -( 1) -( 1 _( 2) _( 6) -( 6) .251_(2) (8) -( 15) _( 8) -( 12) _( 4) -( 5) Frequency of occurance of each band in populations oran e51S Dendrograms constructed using UPGMA method of cluster analysis 8& data in table Haemoglobi n s ESTERASES , S.b.
Occiput flattened, ridges prominent.The pattern of hair was analysed on the basis of nomenclature given by Brunner and Coman (1974) and Keogh (1983Keogh ( , 1985)).Recently, Ingale (1986) studied hair sculpture pattern of some rodents using SEM, and used the patterns to establish phylogenetic relationships amongst them.The hair sculpture and scale patterns of Bandicota indica and Bandicota sp. are represented in PIs.I and II.Under low magnification, the hair of B. indica shows an irregular waved mosaic pattern at near base and half-way mark.There are not more than two scales across the width of the hair.The scales are of fairly uniform depth.The margins of the scales are slightly rippled and crenate.A very shallow groove on the hair is also visible at lower magnification.The scale margins appear to be distant.The scale characteristics more or less remain constant even in the near apical region of the hair, however, due to reduction in diameter, the number of scales across the width of the hair is further reduced to one scale.
The hair of Bandicota sp.(Plates I & II) exhibits a distinct chevron pattern which in the near middle region appeJfS double chevron.Thus, there is only one scale across the width of the hair in the basal and the near middle regions.The scales are wider then deep and their ends overlap the base in front.The margins of the scale are almost smooth at lower magnification.Though the general scale pattern remains more or less identical in the middle and apical regions, the margin patterns become sharper in the near middle region.
The scale pattern in the hair of Bandicota bengalensis (vide Ingale, 1986) is petaloid, with several scales across the width of the hair.Scales are of uniforln size and have crenate margins.Thus the pattern of scales in Bandicota bengalensis does not match with those exhibited by Bandicota indica and Bandicota sp.
The differences on the scale pattern observed in the present study on B. indica and Bandicota sp. are very distinct, waved mosiac (Plate lA) and chevron (Plate IB) respectively.

CONCLUSION
From the above study it is clear that four populations of the large-sized bandicoot rats occur in India, which differ from each other in one or more characters (Tables 1-4, figs.2-5).The three, namely, indica, nemorivaga and malabarica are allopatric in distribution, hence, treated here as three subspecies of Bandicota indica; malabarica occurring in Western Ghats, nemorivaga in West Bengal and northeastern India and indica in the rest of India.
The fourth population of the large-sized bandicoot rats is India wide in distribution and differs from the other three (indica, nemorivaga and malabarica) together in the structure of skull, biochemical characters and hair-sculpture.Hence, the same is described below as a new species.

SYSTEMATIC ACCOUNT
Pradhan et a!.(1989) described this population of large-sized bandicoot rats as Bandicota gigantea non Hardwicke.But since the skull of the type of B. gigantea present in the British Museum is broken, it is not possible to confirm (the main key character), whether the ONL was more than CBL in that specinlen or not.Hence, it is descr~bed here as a new species.All the collecti.onsare deposited at the Western Regional Station, Zoological Survey of India, Pune.All measurements are in millimetre (Table 5).
Description: A very large-sized rat (Fig. 8), with triangular head, rounded snout, and tail shorter than head and body length.Body covered with smooth coarse

FIG.12
OCCiput region of §. maxima sp.n.(M/S8) Skull (Figs. 12-14 and Table 5) more or less similar to that-of Bandicota indiclJ except the swollen occiput (Fig. 7).,Occipito nasal length equal to or more thaa Distribution: The species was recorded from Gujarat, Maharashtra, Rajasthan, Karnataka, Kerala, Andhra Pradesh and West Bengal; also Nepal and Bangladesh.Hence, its distribution appears to be throughout India.
Habitat: Bandicota maxima normally occurs near human habitation and lead epizotic life.It is nocturnal and fossorial.It makes burrows in open yards, gardens, under the foundations of residential premises, granaries, store houses, etc.Its preferred food is grains and vegetables but can switch over to other diet.
To accommodate the new species Bandicota maxima, the genus Bandicota may be redefined as large rats having proodont I orthodont incisors, condylobasal length may or may not exceed occipitonasallength, anterior palatal foramina more than 6.5 mm or over 15 % of ONL, and the postero-internal cusp present in first and second upper molars. 1.

Key to species of genus Bandicota
Occipitonasal length, in Indian species, less than 45 mm; Zygomatic width nlore than 57%, bulla more than 20%, and nasals less than one-third of occipito-nasal length.
Occipitonasallength less than condylobasal length; sculpture pattern of dorsal hair mozaic (at lower magnification).
Occipitonasal length equal to or more than condylobasallength ; sculpture pattern of dorsal hair chevron (at lower magnification).
Nasals and diastema less than 40 0 / 0 and 33% of ONL respectively.(1961) reduced the number and merged all species into two species viz.B. bengalensis and B. indica.His studies were based on the British Museum material.However, it has been found out by the present workers that it is rather difficult to allot any taxonomic status to the freshly collected bandicoot material based an Ellerman's keys (1961).It was, then, decided to undertake a detailed comparative osteo-morphological, biochemical and hair impression analysis studies of such a bandicoot population which is not fitting in Ellerman's keys.The studies show that large-sized bandicoot rat populations belong to a separate species.This species has been named as B. maxima.Keys to the identification and description of the new species have also been given in the present communication.
•=====••±,==~~~\ areas have been shown in Figure 1.The rodent collection in the sampling areas was made during the extensive surveys carried out over a veriod of four years from.1982 to 1986, MATERIAL AND METHODS

Fig
Fig. No.3 ie.... .,.,~ Ii Fig.No.5 ~. indica s\(tull (M / l.15) Pradhan et al. (1989) have discussed at Ienght the status of the large-sized bandicoot population and have doubted its inclusion in B. indica.Along with several eateomorphological characters they had used two protein fractions, Hb and eye lens -proteins, to examine the differences.During the present study, additional proteins namely LDH, non-specific esterases and plasma low molecular proteins in the albumin lOne representing a total of about 52 gene locii in the populations were used to •IHamine homologies at functional (enzyme) levels.From the tests (Table 4) it is PBADHAN et.al: Taxo, nomic studies of Indian B(), ndico()t rats A PLATB-I B A• Photograph showiog cudcular impressio.D pattern of B. indica (M/41S) hair between basal and middle region.Kindly igDore air bubbles.(Magnification: Pbotographed at X200).B: Photogr, aph showing cuticular impre.sslonpattero of Bandic, ota sp.(MI98) hair betwe, en basal and middle region. .Kindly ignore air bubbles.(Magnification: Pho• togr, apbed at X200).

Table No
Genetic Identities (above diagonal) and Genetic Distances (below diagonal) for the locH representing five specific proteins of the three Bandicota species in question.

Table No
bristles present in hind quarter.Tail thinly covered with short hairs but has a leathery texture due to presence of broken scales along its entire length.Dorsal colour varies from dark slaty to light brown, ventral greyish white; specimens from Calcutta lighter in overall coloration than those from Bombay-Pune region.Tail dark and unicoloured.Thumb rudimentary but with a blunt claw.Legs having 5 toes, studded with prominent claws.Soles dark, bearing six plantar pads.Mammae 3 + 3 Records of the Zoological Survey of IndiaZygolnatic width less than 55% of ONL ; hindfoot more than one-fifth of head and body length.. .•Zygomaticwidth more than 55% of ONL ; hindfoot less than one-fifth of head and body length.Rodent genus Bandicota was split earlier, into a number of species.Ellerman SUMMARY B. i. indica B. i. nemorivaga