Species distribution models (SDMs) are used to predict biodiversity patterns given expected changes in climate, however, they assume species exist in isolation even though research shows that a species response to climate is often mediated by their interactions and associations with other species. Recently developed methods, called community level models (CLMs), incorporate species associations in order to predict species distributions and communities properties. Although encouraging, the ability of these models to predict biodiversity to novel climates is unknown. Our research uses the fossil pollen record of eastern North America and paleoclimate simulations for the past 21,000 years to compare the ability of 5 SDMs and 5 CLMs to predict species distributions and community composition across periods of climatic change. We are also working on extending our results to climate scenarios for the future.
Furthermore, when mega-herbivores went extinct in the late Pleistocene, plant community composition changed, suggesting a cross-trophic cascading effect. We are investigating this phenomenon by incorporating mammal and plant species co-occurrences into CLMs. This work involves updating and verifying mammal fossil assemblages in eastern North America from the Neotoma and Faunmap databases. In the process we are also developing quality control methods to standardize the use of radiocarbon dates for mammals in paleoecological analyses.
This NSF-funded work has been led by postdoc Kaitlin Maguire and is collaborative with Matt Fitzpatrick and his postdoc Diego Nieto-Lugilde, and Jack Williams. Collaborator Simon Ferrier at CSIRO and climate modeler Dave Lorenz, as well as many others, contribute to various parts of the project.