The biostratigraphic surveys provide new information about the taphonomy and paleoecology of the Siwalik faunas and suggest many avenues for further investigation. Although additional multivariate analysis and statistical treatments of our datasets are beyond the scope of this paper, there are a number of issues raised by these preliminary analyses, which we enumerate briefly below as points of departure for future research.
The survey data have possible sampling biases that could affect our results, including inconsistent or incorrect identification, variability in surveyor ability to see small versus large bones, and the effects of sunny versus overcast days and outcrop conditions on fossil visibility. Also, samples size varies considerably in the different survey blocks and intervals, affecting the consistency of our secular patterns. It will be possible in future research to examine the above variables by analyzing the records of particular surveyors, light and slope conditions, area sampled, and the size of fragments recorded (greater than or smaller than 5 cm). We can remove skeletal elements that may be problematic (e.g., humeri and femora, which may not have been correctly attributed to equid versus bovid) and re-analyze the dataset to calibrate the impact of inconsistent identification. We can also analyze blocks within the same survey level and calculate error bars based on sample size for the trends through time. Additional survey data from the Chinji Fm. and parts of the Potwar Plateau will also strengthen our analysis, once it can be put in digital format.
On the plus side, there should be no systematic biases in the types of vertebrate remains recorded through the survey levels. We did not have any particular expectation of what we would find, or search images for specific taxa, that could influence the types of trends portrayed in Figure 6, Figure 9, Figure 11, Figure 12, and Figure 13. The surveys were not done in stratigraphic order; we jumped around to different parts of the sequence, thus the sampling was “blind” in the sense that individuals did not have any prior knowledge of what the trends would be in the compiled data.
In spite of possible sampling biases discussed above, the results of this study show that significant taphonomic and paleoecologic information is preserved in the biostratigraphic survey data. This phenomenon is most apparent in the trends through time in skeletal parts (Figure 6), major vertebrate groups (Figure 10), and equid versus bovid patterns (Figure 11, Figure 12). The overall characterization of the mammalian families in Figure 13 also contains ecological information, in spite of acknowledged problems with preservation biases of larger versus smaller taxa. The challenge is to figure out ways of distinguishing taphonomic versus ecological signals when we have no direct information on original population sizes of the component taxa, and only a limited basis for assessing the impact of differential skeletal part preservation/visibility on taxonomic representation (e.g., the example of equid versus bovid based on all skeletal parts versus teeth-only). Nevertheless, it should be possible to marshal multiple lines of evidence to test the relationships of these biostratigraphic patterns to sedimentological, geochemical, and other paleontological trends. If we can eliminate or control for taphonomic trends using this approach, we will be able to better justify any patterns that remain as paleoecological in origin. The taphonomic patterns have their own value as well, and it would be interesting if the increasing proportion of teeth upward in the Dhok Pathan could indeed be linked with the tectonic evolution of the sub-Himalayan foreland basin.
In general, standardized surveys in the vertebrate fossil record are relatively uncommon, perhaps because so much of the field effort in vertebrate paleontology has been to locate and collect “good specimens,” which usually are rare. Yet such surveys can provide large samples and valuable information in addition to collectible specimens, as demonstrated by a number of previous and ongoing studies (e.g., Behrensmeyer 1975; Badgley 1986b; Smith 1980, 1993; Eberth 1990; Morgan 1994; Bobe and Eck 2001; Bobe et al. 2002; Blumenschine et al. 2003; Behrensmeyer et al. 2004) that have gathered data to address questions about associations of particular types of taphonomic assemblages with different lithofacies or paleoenvironments. Such research then is used as a foundation for exploring various aspects of the paleoecology of the preserved faunas. Many of these studies focus on developing multi-locality datasets of fossils for specific facies where bones are concentrated. Others are tapping into the scatter of bones between the patches, examining distributions and trends in relation to fluvial architecture (Smith 1980, 1993; Bobe et al. 2002; Campisano et al. 2004), tracing single productive levels in the approach labeled “landscape paleontology” (Potts et al. 1999; Blumenschine et al. 2003) and documenting fossil occurrences across large areas of exposures using GIS-based technology (Sagebiel et al. 2004; Straight 2004).
There have been few attempts, however, to apply standardized methods to specific biostratigraphic problems such as illustrated in this paper. One such approach tested for dinosaur diversity and hypothesized decline prior to the bolide impact at the Cretaceous – Tertiary boundary (Sheehan et al. 1991). A large group of field workers was organized to systematically census a sequence of fossiliferous strata in the uppermost Cretaceous Hell Creek Fm. of Montana and North Dakota. The censuses logged 15,000 search hours and recorded only in situ fossils, documenting map location, skeletal part, taxon, stratigraphic level, and lithofacies. This resulted in 556 specimens (MNI) from 8 different dinosaur families. Based on this sample, there was no decline in ecological diversity through three successive stratigraphic intervals for three different fossiliferous facies in two collecting areas, providing support for abrupt rather than gradual extinction of the dinosaurs. That study differed from the Siwalik biostratrigraphic surveys because it was limited to in situ specimens, but it is similar in the use of standardized search and recording procedures to investigate biostratigraphic patterns through time.
The Siwalik sequence of the Potwar Plateau was an ideal context in which to develop and test the methods outlined in this paper. The 10-15o tilt of the strata, laterally continuous strike-valley exposures between sandstone ridges, thick, continuous fluvial sequence, and the availability of willing surveyors all contributed to the success of this approach. However, the biostratigraphic survey methodology can be adapted to other, perhaps less ideal geological and paleontological circumstances to address similar or other types of questions. Potentially serious problems, especially in horizontal or faulted fossil-bearing deposits, include: 1) mixing on outcrop surfaces of fossils from many different stratigraphic sources; and 2) difficulty in identifying and following specific strata or intervals that are producing the lags of surface fossils. Careful use of topography and marker beds, as well as documentation of in situ fossils, can help to control the problem of mixed source levels. It is often possible to select a particularly favorable combination of topography and lithology, such as a plateau-forming sandstone or a laterally continuous gravel that forms a marked break in slope, and restrict surveying to these situations. The condition of the fossils themselves (e.g., fresh versus highly abraded or fragmented) also can be a useful indicator of post-exposure history.
Of course, documenting only in situ fossils is the most conservative and accurate approach to establishing biostratigraphic trends through time (Sheehan et al. 1991), but usually this requires a large amount of effort and results in limited samples. Paleontologists learn to gauge whether surface fossils are derived from a particular source bed or stratigraphic interval, thus enabling them to exploit rich accumulations of naturally excavated specimens. The biostratigraphic survey approach seeks to harness and standardize this expertise in order to recover more information from the fossil record relating to ecological and taphonomic changes across space or time. The nature of the fossil-bearing deposits and the question(s) being addressed must ultimately determine survey design. Whatever this design, it is very important to record the details of the field approach so that the strengths and limitations of the samples are clear to other researchers.