INTRODUCTION

Will Downs was one of the grand old men of the Siwaliks. His knowledge of Siwalik fossils, localities, and stratigraphy was extensive and, due mainly to his tenacity and drive, the biostratigraphy of Siwalik microfauna was built as arguably the finest long sequence of superposed microfaunas in the world. One hundred screenwash localities span the 18 to 6 Ma Potwar Plateau sequence in northern Pakistan. Will also added other sites, notably those from the older Zinda Pir Dome in western Pakistan. Ten thousand small mammal specimens document this composite sequence. Some portions of the record are represented by about 10 fossil sites per million years; of these, some sites are separated by less than 100,000 years. Elsewhere, the record is poorly documented with gaps approaching a million years. Ten years ago, our primary small mammal collecting agenda was to reduce the number of substantial gaps in the fossil record by targeting specific intervals. One of us (MEM) had the good fortune to find a rich microsite in the lower part of the Nagri Formation, an interval otherwise poorly known by only a few sparse large mammal localities. Will enthusiastically returned to Y797 the following year to recover more material. Among the abundant small mammal teeth in this site are two specimens of an odd, large rodent. Our purpose is to describe this rodent and its provenance, and discuss both with respect to their Siwalik context.

Biostratigraphic Sequence

The following discussion of the fossiliferous sequence of the Potwar Plateau considers major collection localities, both those yielding large mammals and/or small mammals. The Chinji Formation of the Potwar Plateau produces many classic Siwalik fossils from well-known horizons (Colbert 1935). In our experience (Barry et al. 1995), the Chinji Formation, comprised predominantly of reddish mudstones, is densely fossiliferous in its middle part, and becomes less fossiliferous upwards. The superposed Nagri Formation, dominated by thick sandstones, has a low density of localities, although near the top the famous Sethi Nagri locality complex (our site Y311) is highly productive. Above this are the many localities of the Dhok Pathan Formation.

Whereas the density of small mammal fossil horizons is on the order of one per 100,000 years in the middle Chinji Formation (Flynn et al. 1998), that frequency declines upward. There was a gap of about a million years between upper Chinji and younger Nagri sites when Barry et al. (1990) analyzed the overall rodent record. In that analysis, the gap centered around 10 Ma; current paleomagnetic time scales make it older.

The youngest well-sampled small mammal localities of the Chinji Formation in the Gabhir Section east of Chinji and Bhilomar are Y504 and Y76, 11.5 and 11.3 Ma, respectively, using the Cande and Kent (1995) time scale. Y311 is much younger, at 10.0 Ma. In the Kaulial Kas section, approximately 50 km to the northwest from the Gabhir Section, locality Y259 is a productive small mammal site lower in the Nagri Formation at 10.4 Ma. We attempted explicitly to fill this gap of nearly one million years by intensive field surveys in 1989. We screened previously identified sites and found several new localities in the lower part of the Nagri Formation.

Locality Y799 produces one of the oldest local occurrences of a hipparionine horse. Its age is 10.7 Ma (Pilbeam et al. 1996). Lower still are two localities, Y791 at 11.2 Ma and Y797 at 11.1 Ma. The former has few fossils, but does yield interesting tree shrew material. The latter proved to be one of our richest small mammal localities, currently about 330 catalogued specimens. Although not squarely in the middle of the long "small mammal gap," Y797 provides a large enough sample (Flynn et al. 1998) to offer an informative view of the Siwalik small mammal fauna in the early late Miocene.

Geological Setting

Locality Y797 is located in the lower portion of the Nagri Formation, a formation dominated by large sandstones. In this region the beds dip at approximately 15 degrees to the northwest. The sandstone locally referred to as GB2 is superposed over the locality and is a complex, multi-storied sand more than 15 m thick that can be traced laterally for several kilometers (Willis 1992). Beneath Y797 is the sandstone GB1.5, locally several meters thick. The lithology of Y797 consists of lenses of red silt and gray sand with pockets of a sandy clay and calcium carbonate nodule conglomerate that are part of a small-scale floodplain channel sandwiched between GB1.5 and GB2. The eastern edge of the channel was observed clearly, but a convincing western edge was not identified in the field. We presume that the western edge was cut out by an adjacent channel to the west. The locality and capping and underlying sandstones are illustrated in Figure 1.

This floodplain channel, measuring approximately 75 m wide and 3 m thick, is comprised primarily of coarse to mixed sand and silt with preserved bedding. The upper portion (1 to 1.5 m) is mostly red silt and is not very fossiliferous. The underlying 2 m of section are comprised of red silt and gray sand lenses, with a sandy clay and calcium carbonate nodule conglomerate, which could represent a small-scale splay deposit. Small bone is densest in the conglomerate, which appears to thicken and become coarser and less sandy to the west. The small mammals were sampled from two places within this lower layer (see Figure 2). One sampling location is visible in the photo. The second place is stratigraphically slightly higher and is on the south side of the hill. Surface collected large mammal bone is generally quite weathered and white in color, in contrast to better fossil preservation at most localities of similar lithology in the Gabhir Kas section.