COMPARISONS

Prasad and Manhas (1999) briefly compared the morphology of the present upper premolariform tooth with docodonts and shuotheriids and placed it in neither of the two groups. As its coronal morphology, particularly asymmetrical occlusal outline, two labial cusps, a wide talon, and pinching of the crown lingual to the labial cusps, is strongly reminiscent of docodont upper teeth, Prasad and Manhas (2001) assigned the tooth to the Family Docodontidae. A detailed comparison of the premolar with the casts of various docodont taxa and other early mammalian groups was carried out at MNHN, Paris at a later stage. The Family Docodontidae is known by Haldanodon (Kühne and Krusat 1972), Docodon (Marsh 1881), Simpsonodon (Kermack et al. 1987), Borealestes (Waldman and Savage 1972), Peraiocynodon (Simpson 1928, Averianov 2004), Krusatodon (Sigogneau-Russell 2003), Dsungarodon (Pfretzschner et al. 2005), Cyrtlatherium (Freeman 1979, Sigogneau-Russell 2001), Tegotherium (Tatarinov 1994), Tashkumyrodon (Martin and Averianov 2004), Sibirotherium (Maschenko et al. 2003), indeterminate docodontids (Averianov et al. 2005), Itadodon (Lopatin and Averianov 2005), Castorocauda (Ji et al. 2006), and possibly Delsatia (Sigogneau-Russell and Godefroit 1997). Among these, the first seven taxa are known from both upper and lower dentitions, whereas the rest are known from the lower dentition only.

The coronal morphology of VPL/JU/KM/12 is more reminiscent of the premolar morphology of docodonts than their molars. Among all known docodonts, upper premolars are known only in Haldanodon exspectatus Kühne and Krusat 1972 and Docodon Marsh 1881. Indeed, a comparison of VPL/JU/KM/12 with P3 of Docodon superus Simpson 1929 (USNM 2715), shows several similarities: asymmetrical triangular crown outline, development of cingular crests anteriorly and posteriorly from X, and X higher than C. However, in marked contrast to VPL/JU/KM/12, on the P3 of D. superus, A is a trenchant cusp occupying more than 75% of the crown and extending almost to the lingual margin of the crown, C is only incipient, and B and E are rudimentary. Moreover, P3 of D. superus has a tiny accessory lingual cusp (Y) split from the posterior part of X. Finally and mostly there is no talon, and X is completely posterior in position in the P3 of D. superus.

In Haldanodon, the transverse width of molars is greater than their labial length, but in the premolars the reverse is the case, as on VPL/JU/KM/12. In fact, P3 of Haldanodon (Krusat 1980, VJ 1008-155, figure 20 D) is morphologically closer to VPL/JU/KM/12 than to its molars in having an asymmetrically triangular crown, a distinct cusp C, a narrow labial cingulum, a posteriorly protruding lingual part, and in the absence of cusp D. The lingual extension and its posterior protrusion are less extensive than in the Indian specimen, and the tooth has already developed two lingual cusps (X and Y). A deciduous premolar (dP3) of Haldanodon (Krusat 1980, VJ1030-155, pl.VIII, figures F-H) resembles VPL/JU/KM/12 in its more posteriorly placed lingual lobe, transversely narrow and small X, absence of Y, C cusp leaning away from A, and a more or less similar anterolabial corner. Only minor differences, such as in size, degree of posterior rotation of the lingual lobe and tilt of C, distinguish the two. But VJ 1030-155 is more like an upper molar than a premolar in the almost equal indentation of the anterior and posterior margins and the near symmetrical outline of the crown.

The upper premolar of Delsatia rhupotopi (Sigogneau-Russell and Godefroit 1997, figures c-e), considered as the most primitive member of docodonts (see Butler (1997) for a contrary view), is distinguished from VPL/JU/KM/12 in possessing a transversely narrow crown, a distinct medially interrupted labial cingulum, and no lingual extension of the crown. The only similarities between these two taxa are in the presence of cusp B and a detached E cuspule.

In Haldanodon, the contour of upper molar crowns is an asymmetrical triangle, and the molars are indented on the anterior and posterior borders lingually to the labial cusps. However, these borders of Haldanodon upper molars are nearly at right angles to the labial face, whereas on VPL/JU/KM/12, only the posterior face is at right angles to the labial face, whereas the anterior face is obliquely oriented to the latter. On the upper molars of Haldanodon, the labial cingulum is narrow, as seems to have been the case on VPL/JU/KM/12. Also in both taxa, the labial cusps are flat labially and convex lingually; A is the largest cusp; C is nearly half the height of A; X is lower than A, higher than C; talon basin is deep; and the cusp A is connected anterolabially to B by a crest. However, in Haldanodon, cusps A and C are separated close to their tips by a V-shaped valley, and the long axes of the two cusps are almost parallel to each other contrary to their diverging nature on VPL/JU/KM12 (the angle between them being close to 90°). Finally, again in both taxa, a cingular crest extends from the tip of X to E (anterior to the base of B). A second crest descending from the posterior tip of Y reaches the posterolabial base of C in Haldanodon; on VPL/JU/KM/12, as there is no cusp Y, a posterior crest from cusp X merges with C at its lingual base. Absence of a well-developed crest from the lingual base of A to X, less steep labial face of X, absence of cingular cuspule D, and shallow sulcus between B and E on VPL/JU/KM/12 further differentiate it from Haldanodon upper molars.

Upper molars with triangular occlusal outline and unequal labial cusps, as in Haldanodon and VPL/JU/KM/12, are also known from the Upper Bathonian Kirtlington mammal bed. VPL/JU/KM/12, however, differs from these teeth referred to Borealestes serendipitus Waldman and Savage 1972 in its asymmetrical crown and the absence of cusp Y. The upper molars attributed to B. mussetti Sigogneau-Russell 2003 are also distinguished from VPL/JU/KM/12, in the presence of a supplementary cuspule on the crest from A to X and hollowing of the lingual face of A.

Though the upper molar of Dsungarodon zuoi Pfretzschner et al. 2005 has a triangular occlusal outline, the crown is not asymmetrical as in VPL/JU/KM/12. Further, the upper molar of Dsungarodon differs from VPL/JU/KM/12 in having a crest originating at the lingual base of cusp A instead of its apex, strongly labially bent hook-like cusp X, a crest from the lingual base of A to X forming the anterior border of the talon basin, and a wide labial shelf with a deep ectoflexus.

Likewise, the upper molars attributed to Krusatodon kirtlintonensis Sigogneau-Russell 2003 are triangular in outline as in VPL/JU/KM/12. But in Krusatodon the labial cusps form an arch in occlusal view; the crest from A to X is absent; the lingual part of the crown is inflated both anteriorly and posteriorly labial to X. The upper molars of Simpsonodon oxfordensis Kermack et al. 1987 and those attributed to Peraiocynodon major Sigogneau-Russell 2003 are also quite distinct from the present specimen in their trapezoidal or squarish crowns, in the development of labial cusps, the presence of extensive basins both anteriorly and posteriorly to the central crest from A to X, and the nearly symmetrical anterior and posterior borders with deep indentations. No upper premolars are known for these taxa.

VPL/JU/KM/12 is thus morphologically closer to P3 of Haldanodon than to that of Docodon and as such identified as an ultimate upper premolar. The only characters which would negate its referral to the family Docodontidae are the absence of a distinct crest connecting A with the lingual cusp X and the absence of an additional lingual cusp Y (M5 of Haldanodon also lacks cusp Y, Krusat 1980, figure 20A; but, in contrast to VPL/JU/KM/12, it also lacks cusp C and with respect to the lingual part its labial part is small and compact). Absence of these two features might, however, be attributed to the chipping of the crown in the position of Y and to the rolled state of VPL/JU/KM/12. The weak crest from A-X on VPL/JU/KM/12 might be an artefact of the poor preservation of the tooth. It may also be a feature of the premolar. Similarly, absence of cusp Y is a premolar condition and therefore, does not preclude its referral to docodonts.