MODELING NONMARINE FOSSIL OCCURRENCES IN A SEQUENCE STRATIGRAPHIC FRAMEWORK USING AN EMBEDDED MARKOV CHAIN METHOD

Simulating the nonmarine fossil record within a sequence stratigraphic framework furthers our understanding of the degree to which stratigraphic architecture controls the distribution of fossils. The empirical model presented here simulates a stratigraphic column with occurrences of plant, invertebrate, and vertebrate fossils. The model uses an embedded Markov chain to simulate a column, based upon observed probabilities of transition among facies and the observed distribution of thicknesses of each facies. Fossil occurrences in the simulated columns are based on observed probabilities of collection of each taxon within each facies. We tested the behavior of the model using data from the fluvio-deltaic Pennsylvanian Breathitt Group of eastern Kentucky (Aitkin and Flint, 1995). Based on previous studies of nonmarine sequence stratigraphy, we expect the strata to contain low-accommodation systems tracts (LAST) with multistory, multilateral channel deposits and thin floodplain deposits, and high-accommodation systems tracts (HAST) with single-story channel deposits encased in thicker floodplain deposits. Analysis of the stratigraphic columns from Aitkin and Flint (1995) generally supports this expected architecture. Observed and simulated floodplain facies are proportionally thicker and more common in the HAST than in the LAST. Channel facies are proportionally thicker and more common in the LAST. With few exceptions, transitions between the facies also follow expectations. Transitions between floodplain facies are common in the HAST, creating thick floodplain intervals, and transitions from channel sands to channel bases, creating multistory channels, are more common in the LAST. The model is able to simulate strata realistically from the data, although simulated columns do not always match the idealized versions of HAST and LAST, often showing intermediate architectures. Model results indicate that there should be no difference in the probabilities of collection of fossils in the HAST versus the LAST, beyond differences in the relative abundance of facies. However, the mean probability of collection of plants is more variable than for invertebrates and vertebrates. The model will be further tested using stratigraphic and fossil data from the Judith River Formation of north-central Montana.