DIPLOCRATERION VERSUS SKOLITHOS: MODELING AND SPATIAL ANALYSIS OF TRACE FOSSILS USING R-STUDIO

In the wide-spread, near-shore, sandy marine environments of the early Cambrian of the southern Appalachians, vertical tubular trace fossils first occur in abundance in the upper Chilhowee Group. Historically, these traces have been identified as Skolithos (single vertical tubes). In practice however, it can be difficult to differentiate between Skolithos and Diplocraterion (U-shaped tube resulting in paired vertical tubes). The ability to recognize these ichnogenera has implications for their use as up-direction indicators and for paleoecological interpretations.

The goal of this project was to model the distribution of Diplocriterion versus Skolithos openings on bedding plane surfaces using the statistical program R. Models were created that generate random spacing of points on an X-Y coordinate plane to simulate Skolithos, and used pairs of points separated by 1.4 units to simulate Diplocriterion. Multiple simulations were run, controlling for variation in (1) population density and (2) sample area size. Both are needed in order to establish minimum rock/slab size and maximum tube density in practical application. Variables were systematically altered in models in an effort to determine a critical point, at which burrow density (overprint) would be too high to allow for recognition of random single tubes versus paired tubes. We used the Ripley’s K-hat test, using R-code modified from SPATSTAT v 1.43, which tests for nearest neighbor values within a certain area of each point, and plotted the values on a line graph with areas of expected deviation.

For population-density model, we observed a random value for Skolithos and a nonrandom value for Diplocriterion, until a critical saturation point causing all points to appear random (simulating time averaged overprinting). For the area size-model Skolithos was random for all sizes, whereas Diplocriterion was nonrandom until the area was decreased to a critical size at the distribution was random. In both cases, the oversaturation point for Diplocriterion was much lower than expected, which restricts this application to relatively low density, large area size samples.