An Ailuravine Rodent from the Lower Eocene Cambay Formation at Vastan, Western India, and Its Palaeobiogeographic Implications

A new ailuravine rodent, Meldimys musak sp. nov. (Mammalia: Rodentia, Ischyromyidae), is recorded from the lower Eocene lignites of western India. It is the oldest record of Rodentia from India. M. musak is more derived than the earliest Eocene ailuravine Euromys cardosoi from Portugal and more generalized than late early Eocene E. inexpectatus and Ailuravus michauxi from France. Its dental morphology closely corresponds to the middle early Eocene species M. louisi, which lived about 52 Ma in Western Europe. Meldimys was previously known only from Europe, and ailuravines were previously reported only from Europe and North America. Its occurrence in India allows the first direct correlation between the early Eocene land mammal horizons of Europe and India, and raises the possibility of a terrestrial faunal exchange between India and Eurasia close to the Palaeocene—Eocene transition.


Introduction
Fossil rodent systematics and biochronology of South Asia have provided a fairly continuous record extending from the early Eocene to Neogene and Quaternary sediments. Among the best studied Palaeogene assemblages are the ones from the Eocene Subathu (India) and Kuldana (Pakistan) forma− tions and coeval sediments (Hussain et al. 1978;de Bruijn et al. 1982;Hartenberger 1982a;Kumar et al. 1997a, b). The previously known Indian Palaeogene rodent assemblages were dominated by ctenodactyloids including the Chapatti− myinae and the Baluchimyinae.
Here we describe a new early Eocene ailuravine rodent from the Vastan lignite mine (Rana et al. 2004(Rana et al. , 2005Sahni et al. 2006), which has interesting palaeobio− geographic implications since most plate tectonic recon− structions (Sahni and Kumar 1974;Klootwijk et al. 1991Klootwijk et al. , 1992Mattauer et al. 1999) place the Indian landmass as an island subcontinent in pre− or penecontemporaneous col− liding phase at that time, and it has been argued that the India−Asia collision may have triggered some of the late Palaeocene-early Eocene global climatic and biotic events (Beck et al. 1998). The new rodent is based on isolated cheek teeth and incisor fragments and appears to have clear affini− ties with the early Eocene ailuravine (Ischyromyidae) rodents from Europe (Michaux 1968;Hartenberger 1975;Wood 1976;Escarguel 1999).
Prior to this report the ailuravine rodents were known only from Europe and North America. In Europe, they have been recorded from at least seven biochronological reference levels within the Eocene (Appendix 1). In North America, they are mainly known from the middle Eocene Uintan Land Mammal Age (ca. 46-41 Ma), with species ranging from the early Eocene (Wasatchian) to the late Eocene (Chadronian) (Wood 1962(Wood , 1980Korth 1984Korth , 1988Korth , 1994. The Vastan rodent represents a new species of the other− wise European ailuravine genus Meldimys. The common presence of the genus Meldimys in the Vastan as well as Eu− ropean sections is significant because (1) it allows for a biostratigraphic correlation between south Asia and Europe, and (2) it opens up exciting new possibilities for examining migration routes across the Tethys Sea which was still in ex− istence during the early Eocene, enveloping most of the northern and northeastern margin of the Indian Plate (Sahni and Kumar 1974). Although such a connection with Asia had been proposed by several workers, including an early (near the Cretaceous/Tertiary Boundary) collision of India and Asia (Sahni et al. 1983;Jaeger et al. 1989;Klootwijk et al. 1991Klootwijk et al. , 1992Beck et al. 1995;, connections with Europe were considered to be more tenuous except of course for the presence of a few taxa, e.g., the dichobunid artiodactyl Diacodexis, and the ray fish Igdabatis, which were common to the two widely separated regions (Thewissen et al. 1983;Kumar and Jolly 1986;Soler−Gijon and López−Martínez 1998).

Geological setting
The new rodent material described herein was recovered from the subsurface beds of the Cambay Formation (= Cambay Shale) dug out in an open cast lignite mine at Vastan, about 40 km (60 km road distance) NE of Surat (N 21°12', E 72°55') in Gujarat, western India (Fig. 1). In the Vastan mine, the Cambay Shale is 20-45 m thick but at the rodent site (N 21°25' 59", E 73°06' 59") only about 31 m thickness has been exposed so far (Fig. 1). It is composed of an alternating se− quence of lignite, dark grey to black and greenish grey shales with vertebrate−rich sandy lenses, and clay−marls. There are two main lignite seams in the mine section: Lignite 1 at the top and Lignite 2 at the mine floor. The rodent−bearing level lies about a meter above the Lignite 2 (Fig. 1) to as the Amravati Formation (sensu Sudhakar and Basu 1973) followed by Recent alluvium. Surface exposures of the Cam− bay Formation occur as thin strips to the east of the Gulf of Cambay (= Khambhat) and along the Saurashtra coast (Bhan− dari et al. 1991).
The Cambay Formation in the Vastan mine section is quite rich in animal as well as plant remains. Molluscs (mostly ma− rine) are highly concentrated in the lower part of the section forming conspicuous shell bands, which are also rich in di− verse vertebrates including fish (mainly sharks and rays), snakes, lizards, frogs, chiropterans, artiodactyls, anthracobu− nids, perissodactyls, rodents, insectivores, and primates (e.g., Rana et al. 2004Rana et al. , 2005Bajpai et al. 2005;Rose et al. 2006;Sahni et al. 2006;Smith et al. 2007). Other major recorded groups are plants, ostracodes, and foraminifers. The foramini− fera−rich interval lies about 10 m above the rodent−bearing ho− rizon. The only index species in the Cambay Formation in general and the Vastan mine section in particular is the fora− minifer Nummulites burdigalensis burdigalensis, which oc− curs~18-20 m above the rodent−bearing horizon . It assigns the section to the early Cuisian Shallow Ben− thic Zone SBZ 10 (= Planktonic Foraminiferal Zone P6b of Berggren et al. 1995 and tropical planktonic foraminiferal zone E4 of Berggren and Pearson 2006) corresponding to about early Ypresian (Serra−Kiel et al. 1998;Berggren and Pearson 2006).

Materials and methods
The rodent fossils from the Vastan lignite mine consist of 16 isolated cheek teeth as well as incisor fragments, all referable to a single genus and species. All were recovered from the same level and locality by screen−washing of matrix using only water. The quality of preservation of isolated teeth is variable but generally good. A few teeth have suffered dam− age which is ascribed to post−depositional breakage. The enamel surface in at least one tooth is deeply pitted probably from chemical erosion.
In this paper, the upper premolars and molars are indicated by P or M, respectively, lower molars by m, and deciduous teeth are preceded by d. The dental terminology adopted here follows Escarguel (1999 Michaux, 1968Genus Meldimys Michaux, 1968 Type species: Meldimys louisi (Michaux, 1964), early Eocene (middle Ypresian) of Avenay, Paris Basin, France.
Revised diagnosis.-Smaller than Ailuravus and North− American ailuravines, and about the same size as the most primitive ailuravine Euromys. Differs from other ailuravines in having lower−crowned cheek teeth with more rounded cusps, noticeably less developed hypocone and less pro− nounced lingual sinus in P4-M3, more globular P4 with less developed anteroloph and posteroloph but better developed mesostyle, and p4 with metaconid markedly less developed than protoconid. Further differs from Ailuravus and Euromys in having upper molars with typical tetrahedral paracone and occlusal surface markedly more curved along the antero−pos− terior median line, and lower molars with hypoconulid nearly merged into posterolophid. Upper cheek teeth with short crest descending from protocone into trigon basin (this crest absent in Ailuravus). Lower cheek teeth with mesoconid con− nected to hypoconid, in contrast to Euromys, which has mesoconid completely separated from both hypoconid and metaconid. Diagnosis.-Teeth similar in size and occlusal morphology to M. louisi. Differs from M. louisi in having P4 with strong metaloph; upper molars with a small median conule between protoloph and metaloph and less buccally projecting meso− style; M1 with less prominent parastyle positioned almost in line with paracone rather than being buccally extended; M2 with less rounded occlusal outline and a protocone usually more anteriorly situated compared to the more central proto− cone of Meldimys louisi; posterior width of M3 less short than in M. louisi; M3 with sub−complete metaloph and reduced lin− gual sinus (between hypocone and protocone); m1-2 with antero−posteriorly reduced trigonid and weaker anterolophid and "waist" (= slight constriction between trigonid and talo− nid), and no paralophid (= metalophulid I) and metalophulid II; talonid cusps considerably lower than trigonid cusps.

Meldimys musak
Description.-The dP4 and M1-M3 are triple−rooted; P4 is double−rooted, its postero−labial root fused with the lingual root. The m1-2 are double−rooted. The p4 and m3 are not represented in our collection. The dP4 is triangular, trans− versely elongated (L/W = 0.86), quite narrow lingually and somewhat heart−shaped with a faint notch (ectoflexus) in the labial margin. Its paracone is large, slightly labially placed in comparison to the metacone and also posteriorly shifted, leaving a wide shelf anterior to the protoloph. The hypocone represented by an elongated ridge is much lower than the protocone. The protoconule is anteriorly placed rather than aligned with the paracone and protocone. P4 is somewhat globular with a convex anterior contour and a centrally slightly concave posterior contour. The pro− toloph is nearly transverse, the metaloph strong and oblique and the paracone posteriorly shifted as in dP4. It differs from dP4 in being more ovoid, and in having a thick and labially extended anteroloph, a short posteroloph restricted to the lin− gual half, and a more noticeable hypocone. Its lingual aspect is broad and rounded rather than narrow as in dP4.
Upper molars have occlusal surfaces somewhat concave in lateral profile, forming a shallow "groove" along their antero− posterior median line (Figs. 3, 4C-G, and 5A-D). M1 is nar− row lingually, its labial half has a rounded contour. The anteroloph, posteroloph, protoloph, and metaloph are distinct. The protocone is the largest cusp; the paracone is massive and tetrahedral but slightly transversely elongated, much larger than the metacone, and medially shifted relative to the labial margin of the tooth. The protocone and paracone are equally high from the crown base. The parastyle is strong and extends more labially than the paracone. The mesostyle is also quite prominent and labially projecting relative to paracone and metacone. The hypocone is moderately developed and notice− ably lower than the protocone. The metaconule is better devel− oped than the protoconule. A long sloping shelf occurs ante− rior to the protoloph. In unworn teeth a short crest originates from the protocone and descends into the trigon; a distinct conule is present in the middle of trigon. A weak sinus is lin− gually situated between protocone and hypocone.
M2 is slightly larger than M1 with a somewhat rounded contour and better developed hypocone than in other upper teeth. The paracone is labio−lingually elongated, the para− style and metastyle distinct and the mesostyle labially pro− truded. The metaconule is much larger than the protoconule. A crest arises from the protocone and descends into the trigon basin extending beyond the protoconule and meta− conule and directed towards the mesostyle. A small conule in the trigon basin between the protoconule and metaconule is more distinct than in M1. It is usually isolated but lies close to the buccal end of the crest that descends into the basin. The anteroloph and posteroloph are prominent. The lingual sinus between the hypocone and protocone is more pronounced than in M1.
M3 is triangular with subequal length and width. Its para− cone is large and labio−lingually elongated as in the more an− terior molars. The hypocone is small but distinct in most teeth and slightly labially positioned relative to the protocone (  The lingual sinus between the hypocone and protocone is weak. Our collection has only three lower cheek teeth, all of which are quite worn and or partly damaged. In m1, the trigonid is antero−posteriorly reduced, the anterolophid is weak, and talonid cusps are considerably lower than trigonid cusps. They lack a paralophid (= metalophulid I), while the metalophulid II is either absent or not discernible due to strong wear.
The m1 has a distinct "waist" (constriction) between the trigonid and the wider talonid (Fig. 5A). The metaconid is the highest cusp, and entoconid the smallest and lowest. The protoconid is larger than the metaconid and somewhat posteri− orly placed in comparison to the latter. The hypoconid is mas− sive, and hypoconulid tiny, almost merged into the postero− lophid. An ectolophid is short but well developed, particularly posteriorly; it bears a distinct mesoconid which is connected to the hypoconid but separated from the protoconid. An antero− lophid is poorly developed. A faint hypolophid can be seen in one of the two m1's (GU/RSR/VAS 604; Fig. 5A). A deep talonid notch is present between the entoconid and metaconid along the lingual margin of tooth.
The m2 differs from the m1 in having more widely sepa− rated protoconid and metaconid, a hypoconid smaller than the protoconid, better developed ectolophid and anterolo− phid, a much shallower talonid notch between entoconid and metaconid and a clearer yet weak hypolophid, and in lacking a "waist" between the trigonid and talonid. It has a faint metastylid.
Comparisons.-The new Indian rodent is referred to the subfamily Ailuravinae (Family Ischyromyidae) based on dis− tinctive occlusal morphology of its cheek teeth-upper mo− lars with relatively less developed hypocone (primitive char− acter), tetrahedral paracone, strongly developed parastyle, distinct and externally extended mesostyle and a short crest ascending from the protocone into the trigon basin, and lower molars with a large mesoconid connected to the hypo− conid by a strong ectolophid, and an isolated entoconid con− nected to a smaller hypoconulid. Most of these characters are not present in chapattimyids and yuomyids (Ctenodactylo− idea), which include all middle Eocene rodents reported thus far from India and Pakistan (Hussain et al. 1978;Kumar et al. 1997a, b). Similarly they are not seen in other contemporary ctenodactyloids such as ctenodactylids known from Central Asia (e.g., Dawson et al. 1984 Escarguel et al. 1997) in the Paris Basin (France), and is most similar to the latter. Features including rounded occlusal outlines, lower crown height, a hypocone noticeably lower than the protocone, a markedly concave occlusal surface creating a "groove" along the antero−posterior median line, and protocone and tetrahe− dral paracone reaching nearly the same level above the crown base, are clearly indicative of its closer affinity to Meldimys than to Euromys. In contrast to Euromys, the Indian ailuravine has lower molars with a better developed trigonid and a hypoconulid more strongly connected to the hypoconid. The upper molars of the Indian ailuravine differ from those of Euromys mainly in having poorly developed hypocone and lingual sinus and in the relative size of the metaconule and protoconule; in the new taxon the metaconule is considerably larger than the protoconule whereas in Euromys the two conules appear subequal. Among the four known species of Euromys, E. cardosoi (Estravís, 2000) (for generic assignment see Escarguel 1999) is similar in tooth size to M. musak sp. nov., but M. musak appears morphologically more similar to E. thaleri (Michaux, 1964) (Escarguel 1999).
Prior to this report, Meldimys was represented by a single species, M. louisi (considering that the species E. cardosoi from the earliest Eocene of Silveirinha, Portugal, cannot be re− ferred to this genus; Escarguel 1999, contra Estravís 2000, which has cheek teeth remarkably similar to those of the new Indian species M. musak. M. musak is presently represented by dP4, P4-M3, and m1 and m2 but m1-m2 are too worn or dam− aged to allow detailed comparisons with those of M. louisi. The dP4 of M. musak, though also quite worn, is almost indis− tinguishable from that of M. louisi (Fig. 6A) illustrated and de− scribed by Escarguel (1999: pl. 3: d)   M. louisi is generally heart−shaped ( Fig. 6A; Escarguel 1999: pl. 3: d; pl. 4: f) due to its narrow lingual aspect and a promi− nent central notch in the broader labial aspect, but this charac− ter (heart shape) is variable-dP4 of Meldimys is not always heart-shaped (Escarguel 1999: pl. 2: o). In fact a more or less heart−shaped dP4 is "archetypal" for the Ailuravinae and is probably plesiomorphic within rodents. This trait, amongst others (all plesiomorphic), led Hartenberger (1995) to propose inclusion of the Ailuravinae, together with Alagomys and Tribosphenomys, within the Alagomyidae. This proposition was based on a very limited cladistic analysis of selected taxa and characters and is now considered untenable (Dawson 2003;Marivaux et al. 2004).
The only known P4 of M. musak sp. nov., is more rounded than dP4 with a width/length ratio of 1.3. Compared to this, the P4 of M. louisi is generally ovoid with rounded margins and maximum length and width along the median lines ( Fig. 6B; Escarguel 1999: pl. 2: p; pl. 3: e), but some are transversely elongated with high (1.4) width/length ratio and others rectan− gular with low (1.2) width/length ratio (Escarguel 1999: pl. 4: a, g). The anteroloph and posteroloph in P4 of the new species are marginally higher, the metaloph is stronger and the hypo− cone is clearly more lingually placed relative to the protocone than in M. louisi. In addition, in the single known P4 of M. musak the postero−labial root is fused with the lingual root, whereas in M. louisi the lingual root is separate and the two la− bial roots are fused together. However, based on a single spec− imen it is difficult to assess if this rather unusual configuration is really apomorphic in M. musak or just a teratological, unrep− resentative state for this isolated tooth.
Among the anterior upper molars of M. musak sp. nov., M1 differs from that of M. louisi mainly in having a less prominent parastyle positioned almost in line with the para− cone rather than being anterolabially extended (Fig. 3C). The M1 parastyle in M. louisi is so strong that the anterolabial as− pect of tooth appears distinctly protruded or extended (Fig.  6C). M2 of M. musak differs from that of M. louisi in having a normally positioned protocone rather than a more centrally positioned one as in most specimens of M. louisi (Michaux 1968;Escarguel 1999). Moreover, in M. louisi these teeth have a more rounded occlusal outline than in M. musak. M3 differs in having posterior width less short than anterior width (Fig. 4) rather than considerably shorter posterior width in two out of three M3s of M. louisi (Fig. 6E;Escarguel 1999: pl. 3: c, i). The metaloph on M3 of the new species is more or less continuous and better developed than in M. louisi, in which the metaconule appears as an isolated cone. Further, in M. musak the metaloph and posteroloph converge on the metacone and hypocone to form a small enclosure (Fig. 4B), which is not seen in M. louisi. Most upper molars of M. musak possess a small but distinct conule in the middle of the trigon basin between the metaconule and paraconule (Figs. 3A 1 , 4F 1 , G 4 , and 5A-D). This character is lacking in M. louisi. The m1-2 of M. musak differ from corresponding teeth of M. louisi mainly in having an antero−posteriorly re− duced trigonid and in lacking a paralophid (= metalophulid I) and metalophulid II.
than those of the late early Eocene species E. inexpectatus Escarguel, 1999 and Ailuravus michauxi Hartenberger, 1975 from Prémontré in the Paris Basin. Occlusal surface charac− ters of upper cheek teeth, such as trigon shape, relative size of protoconule and metaconule, transversely elongated para− cone, and labially protruded mesostyle, are highly suggestive of the same evolutionary stage for Meldimys musak sp. nov., as seen in M. louisi and E. thaleri at ca. 52 Ma (middle early Eocene) in western Europe (Paris Basin localities). These ailuravine rodent assemblages are clearly different from those known at the same time in North America, which in− clude the species ?Mytonomys coloradensis (Wood, 1962) from the lower to upper Wasatchian of Colorado and Wyo− ming, and M. wortmani (Wood, 1962) from the upper Wasat− chian ("Lostcabinian") of Wyoming (Wood 1962;Korth 1984Korth , 1988Korth , 1994. ?M. coloradensis and M. mytonensis (Wood, 1962)  Eocene beds of the northern parts of the Indian subconti− nent have produced a well known rodent fauna, most assem− blages (e.g., Kalakot and Subathu in India and Ganda Kas in Pakistan) being of an early middle Eocene age and some lev− els (e.g., Barbara Banda in Pakistan) believed to be early Eocene in age. The middle Eocene assemblages comprise chapattimyids and yuomyids (ctenodactyloids), but the lower Eocene locality (Barbara Banda) has yielded only a few isolated teeth of indeterminate ctenodactyloids and a p4 of an ischyromyoid (de Bruijn et al. 1982: fig. 6). In the light of the new early Eocene ailuravine from India, the affinity of the ischyromyoid Rp4 from Barbara Banda must be reexam− ined here. A comparison of this tooth with p4's of known ischyromyoids from Europe and North America suggests that it is neither a paramyine nor a reithroparamyine (inclu− sive of Microparamys). The overall occlusal shape and ar− rangement of cusps, most particularly its anterior part, simple trigonid with well separated proto− and metaconid, a weak connection between the hypoconulid and the entoconid, a hypoconulid completely merged into the posterolophid, and presence of a mesostylid, which is variable but rather fre− quently observed tiny cusp in Meldimys louisi as well as M. musak (e.g., Escarguel 1999: pl. 3: j, and Fig. 5A, C herein) support an ailuravine affinity for this tooth. However, its smaller size [closely corresponding to p4 of Pseudoparamys teilhardi (Wood, 1962), but ca. 30% smaller than that of M. louisi and ca. 10% smaller than that of Euromys cardosoi, the smallest ailuravine species described to date], small length/ width ratio, very strong anterior part of the ectolophid (poste− rior arm of the protoconid), a seemingly weak mesoconid totally fused with the ectolophid, and rectilinear (not rounded) posterior side favour a pseudoparamyine affinity (see Pseudoparamys teilhardi, Escarguel 1999: pl. 12: l). Al− http://app.pan.pl/acta53/app53−001.pdf RANA ET AL.-EOCENE AILURAVINE RODENT FROM INDIA 9 Fig. 7. Distribution of ailuravine rodents and possible routes of their dispersal during the early Eocene (based on Smith et al. 1994, with additional data mainly from Gaetani and Garzanti 1991;Gingerich et al. 1997). Black dots mark the position of Vastan in India and of ailuravine−yielding sites elsewhere; dashed lines indicate the possible/discussed dispersal routes, which include: Europe to Asia via the Turgai Strait and then to India via its leading edge (the favoured route); Europe to India via Africa and Madagascar; Europe to India directly through Turkey.
though the possibility that this p4 belongs to a small ailura− vine species cannot be eliminated, at present it seems more likely that it belongs to a species close to Pseudoparamys teilhardi than to an ailuravine. It is pertinent to mention here that both ailuravines and pseudoparamyines are likely to have shared the same biogeographical history. They are well diversified European groups known very early in the Eocene with a subsequent diversification in North America (Myto− nomys and Eohaplomys for Ailuravinae and Franimys for Pseudoparamyinae). Now that there is an Indian ailuravine, also having a pseudoparamyine in the subcontinent makes a good working hypothesis rather than appearing unusual. The occurrence of only chapattimyid and yuomyid (cteno− dactyloids) rodents from the middle Eocene localities of the subcontinent and of some indeterminate cocomyids and para− myines from a single lower Eocene locality led several work− ers (Hussain et al. 1978;de Bruijn et al. 1982;Hartenberger 1982b;Kumar et al. 1997a, b) to believe that these mainly northern middle Eocene assemblages reflected the first immi− grants from Central Asia following the Tethyan withdrawal, because similar forms had earlier been reported from the erst− while USSR (Shevyreva 1976). However, the Vastan assem− blage, which may be diachronous in age to the more northerly localities (e.g., Barbara Banda in Pakistan), offers evidence that ailuravine rodents were widespread across the Holarctic region. The presence of primitive ailuravines in lower Eocene strata of India is tantalizing and unexpected. It opens the possi− bility that the subfamily could have originated in southern Asia rather than Europe, as suggested by previous evidence. However, it does not provide further insight on rodent origins and dispersal because the Vastan material comes from a single horizon and there are no data for the basal−most Eocene or Palaeocene on the Indian subcontinent.The affinities of the pre−Palaeocene mammals known from India are also obscure.
Whether or not the Indian ailuravines were replaced by the chapattimyids when the latter appeared in India in the early middle Eocene (~48-47 Ma) is not known. From dental morphology it is apparent that ailuravines and chapattimyids (Ctenodactyloidea) did not share ancestor−descendant rela− tionship, implying that at least two groups of Eocene rodents lived in the subcontinent, but whether they lived together at any point of time is not known. Judging by the European ex− ample the ailuravines lasted in Europe at least until the latest middle Eocene (MP 16 = late Bartonian), whereas they are not found in known early middle Eocene faunas of the Indian subcontinent. In North America, ailuravines are known throughout the Eocene, from Wasatchian to Chadronian Land Mammal Ages, showing the highest diversity during Uintan times (Wood 1962(Wood , 1980Korth 1984Korth , 1988Korth , 1994.
The occurrence of a rodent genus common to two areas as widely separated as Europe and India during the early Eocene presents a palaeobiogeographic problem. We note that Mel− dimys is neither the first nor the only European genus that has been recognized in the Eocene of India. Another well known European (and North American) genus Diacodexis, a dicho− bunid artiodactyl, has been known from the lower middle Eocene beds of India and Pakistan for about a quarter of a cen− tury (Thewissen et al. 1983;Kumar and Jolly 1986). Recently diacodexeine material has also been recorded (Bajpai et al. 2005) from the same horizon which produced the new ailuravine described here. Though Bajpai et al. (2005) referred it to a new genus and species, Gujaratia indica, neither this species nor G. pakistanensis (Thewissen, Russell, Gingerich, and Hussain, 1983) (originally allocated to Diacodexis) is suf− ficiently distinct from Diacodexis to justify generic separation (see also Kumar 2006). Apart from Meldimys and Diacodexis, three genera (from three families) of chiropterans (Hassi− anycteris, Archaeonycteris, and Icaronycteris) are also com− mon to India and Europe (Smith et al. 2006b(Smith et al. , 2007.
The faunal similarity between Vastan in India and the Paris Basin in France at the middle-late early Eocene is somewhat comparable to the resemblance between Dormaal in Belgium and the Wasatchian−0 (Wa−0) interval in North America at the earliest Eocene (Smith et al. 2006a). In the context of this apparent similarity between Indian and Euro− pean early Eocene mammal taxa, it is interesting to note that Godinot and Lapparent (2003) recently presented arguments favouring mammalian and reptilian dispersal from Asia to Europe during the Palaeocene/Eocene boundary interval. Earlier, Escarguel (1999: fig. 24) proposed a four−phase palaeobiogeographical hypothesis, based primarily on ro− dents, of which the first phase is of particular relevance here. It envisaged (1) migration of alagomyid and ischyromyid ro− dents (no ailuravines) at ca. 56.5 Ma, (Ti6/Cf1 limit, latest Palaeocene) from Asia to North America, and (2) arrival around the same time in Europe (exclusively in peri−Tethyan domain) from Asia of ailuravines, the incertae sedis Corbarimys, as well as artiodactyls, perissodactyls, and hyaenodontids. The latter hypothesis, consistent with the new evidence, offers an alternative to Beard's (1998Beard's ( , 2002 "East of Eden" hypothesis on origin and dispersal of mam− mals at the Palaeocene/ Eocene boundary. Previously a direct faunal dispersal between Asia and Eu− rope was considered untenable because of the existence of the West Siberian Sea or Obik Sea and Turgai Strait acting as for− midable physical barriers. However, recent studies of late Palaeocene-early Eocene dinoflagellate cysts and the se− quence stratigraphy of the Turgai region have revived the pos− sibility of such a dispersal owing mainly to drop in sea level as well as climatic factors (Iakovleva et al. 2001). Similarly, Hooker and Dashzeveg (2003), based on cladistic analysis and palaeogeographic reconstructions, recognized at least three cases of multiple land mammal dispersal across the Turgai Straits from Asia to Europe in the late Palaeocene and early Eocene. More recently Smith et al. (2006a) concluded that the earliest Eocene primate Teilhardina, known from all three northern continents, originated in Asia and quickly dispersed westwards to Europe and from there to North America.
In recent years several studies have favoured an "Out of In− dia"/"Out of Asia" hypothesis for the origin and dispersal of various mammals, reptiles and amphibians as well as some plants (e.g., Krause and Maas 1990;Bossuyt and Milinkovitch 2001;Bowen et al. 2002;Conti et al. 2002;Gower et al. 2002;Karanth 2006) based partly on other evidence suggesting Asia to Europe dispersal around the Palaeocene/Eocene boundary. Whether this dispersal was bidirectional or just one way is not known because other than the evidence of ailuravines and Diacodexis, there is little to suggest Europe to Asia dispersal around the same time. The problem with the presently avail− able evidence from Vastan is that it comes from a single level and site. The assumption of ailuravines, diacodexines and other mammals reaching India from Europe raises the ques− tion of what route was taken-Europe to Asia via the Turgai Strait and then to India via its leading edge, Europe to India via Africa and Madagascar, or Europe to India directly through Turkey? So far Meldimys (and more generally ischyromyoids) and Diacodexis have neither been reported from central or western Asia nor from Africa−Madagascar to support either of the two possibilities (the validity of the record of diacodexeine from Asia by Gabunia 1973, is unknown). The middle Eocene mammalian faunas of the Indian subcontinent are known to be more endemic than cosmopolitan, though their migratory ele− ments clearly show North American and Eurasian and particu− larly the Central Asian influx, and little or no African influ− ence (Kumar 2001). Similarly the early Eocene mammalian fauna reported thus far from Vastan, does not show any notable relationship with African faunas. This may seem to weaken or negate the possibility of incoming migrants having taken the African route during the Eocene, but we know so lit− tle of the African Eocene record that negative evidence means a little. The Asia−North America connection, which is impor− tant for several other mammals at the beginning of the Eocene, is less pertinent for the Indian ailuravine, which is clearly more closely related to European than to the North American forms. Proponents of Asia to Europe dispersal around the Palaeo− cene/Eocene (PE) boundary generally cite global warming (the Palaeocene/Eocene Thermal Maximum or PETM) as the driving force for northward migration of faunas. This would support the origin of Meldimys and Diacodexis in India in the earliest Eocene and northward migration to Europe later, and at the same time would explain the lack of substantial Europe to Asia dispersal. At present there is no way to test any of these hypotheses, and waiting for new discoveries particularly in southeast and western Asia seems to be the only option. The distribution of Eocene ailuravine rodents and possible routes of their dispersal are depicted in Fig. 7.
While debating mammalian palaeobiogeography, the tim− ing of India−Asia collision must also be considered. Presently there are two hypotheses, one favouring collision around the Palaeocene/Eocene boundary and the other around the Creta− ceous/Tertiary (K/T) boundary. So far there is not enough palaeontological evidence to accept or negate definitively ei− ther of these hypotheses. Yet if faunal interchange between In− dia and the northern continents and especially between India and Europe can be demonstrated in the early Eocene, it will be an argument favouring a land connection between India and Asia around the Palaeocene/Eocene boundary if not earlier.

Conclusions
Ailuravine rodents are reported for the first time from out− side Europe and North America, indicating their wide− spread distribution in the Eocene. The Indian ailuravine, a new species of European genus Meldimys, is morphologi− cally closer to European lineages than to North American lineages. Meldimys musak sp. nov., is more derived than the earliest Eocene ailuravines Euromys cardosoi and Euromys sp., which occur close to the Palaeocene/Eocene boundary in Portugal and Belgium respectively, and more primitive than the late early Eocene species E. inexpectatus and Ailuravus michauxi occurring during the late early Eocene in the Paris Basin. In grade of evolution it closely corre− sponds to E. thaleri and M. louisi, which occur at~52 Ma in the Paris Basin (Western Europe).
Contrary to the earlier belief based on data from sub−Hi− malayan localities in India and Pakistan (Hussain et al. 1978;de Bruijn et al. 1982;Hartenberger 1982a;Kumar et al. 1997a, b), the oldest fossil rodents from India belong to the cosmopolitan Ischyromyidae rather than to the endemic Chapattimyidae, and so far there is no evidence to know whether these two groups lived together at any point of time.
The Indian ailuravine−yielding section at Vastan mine is chronologically closer to Paris Basin (France) sections than to Dormaal (Belgium) and Silveirinha (Portugal) as indicated by closely similar dental morphology of European M. louisi and Indian M. musak sp. nov., and the common presence of Diaco− dexis and some bat genera in the two sections. The faunal sim− ilarity between Vastan in India and the Paris Basin in France at the middle-late early Eocene is comparable to the resem− blance between Dormaal in Belgium and the Wasatchian−0 (Wa−0) interval in North America at the earliest Eocene.
The presence of a rodent genus common to India and Eu− rope allows the first direct correlation between Indian and European mammal−yielding sections. Further, it shows that generic level similarities in terrestrial mammalian palaeo− faunas across continents are not uncommon. The Eocene mammalian fauna known so far from India seems to have more in common with Europe than with Asia, but this could simply reflect that contemporaneous faunas are otherwise unknown (or poorly known) from southern Asia. II, Montpellier, France) for sharing their expertise on rodent taxon− omy and providing the first assessment of the Vastan rodent based on images. We also thank Jean−Louis Hartenberger (Montpellier, France) and Mary Dawson for constructive comments on the original manuscript. The help provided by the General Manager, and Staff of the Vastan lignite mine (Gujarat Power Corporation Limited) is thankfully acknowledged. Contribution UMR5125−08.002.