Published online by Cambridge University Press: 05 December 2018
Vertebrate microfossil assemblages in a stratigraphic sequence often yield similar assortments of taxa but at different relative abundances, potentially indicative of marginal paleocommunity changes driven by paleoenvironmental change over time. For example, stratigraphically younger assemblages in the Dinosaur Park Formation (DPF) yield proportionally more aquatic taxa, consistent with marine transgression. However, individual deposits may have received specimens from multiple source paleocommunities over time, limiting our ability to confidently identify ecologically significant, paleocommunity differences through direct assemblage comparisons. We adapted a three-source, two-tracer Bayesian mixing model to quantify proportional contributions from different source habitats to DPF microfossil assemblages. Prior information about the compositions of separate, relatively unmixed terrestrial, freshwater, and marine assemblages from the Belly River Group allowed us to define expected taxon percent abundances for the end-member habitats likely contributing specimens to the mixed deposits. We compared the mixed assemblage and end-member distributions using 21 different combinations of vertebrate taxa. Chondrichthyan, dinosaur, and amphibian occurrence patterns ultimately allowed us to parse the contributions from the potential sources to 14 of the 15 mixed assemblages. The results confirmed a significant decline in terrestrial contributions at younger DPF sites, driven primarily by increased freshwater specimen inputs—not incursions from the adjacent marine paleocommunity. A rising base level likely increased lateral channel migration and the prevalence of freshwater habitats on the landscape, factors that contributed to increased paleocommunity mixing at younger channel deposit sites. Bayesian methods can account for source-mixing bias, which may be common in assemblages associated with major paleoenvironmental changes.