A QUANTITATIVE APPROACH TO MAMMALIAN PALEOECOLOGY BASED ON MODERN ECOSYSTEMS

Approaches to reconstructing paleoecology should have a foundation in the modern biota. A biological extension of the geological axiom ‘the present is the key to the past' implies that rigorous examination of modern environments can estimate paleoenvironmental conditions. Failure of a proposed paleoecological model within this modern, relatively well-known, slice of evolutionary time should preclude its application to past environmental reconstruction. In reality, paleoecological models are rarely tested rigorously in modern ecosystems before they are applied to fossil ecosystems.

Recent publications have shown the correlation of species-level diversity in some groups of rodents (arvicolines and murines) with climatic variables. I have revised those correlations and documented new ones with numerous arrangements of phylogenetic clades and ecological grades based on the mammalian fauna from each ecoregion of the United States and Canada. With these correlations, quantitative estimates can be made based on many different groupings of mammals. Faunal lists were compiled for specific localities within each ecoregion using data from hundreds of published studies. Preference was given to localities having better-studied mammalian faunas and more complete climate data. Localities were as far apart from ecoregion boundaries as possible, while keeping the two localities within each ecoregion as far apart as possible from each other in order to encompass the greatest possible climatic variation. Climate data were taken from US National Climatic Data Center and the Canadian Department of the Environment.

Application of the predictive equations (based on modern biota) to fossil deposits gives mixed results. Long, relatively-continuous sequences of Pliocene and Pleistocene deposits were examined using the predictive equations. Temperature patterns based on fossil mammals matched independent data; however, precipitation estimates differed greatly. The ability of these predictive equations to produce reliable estimates appears to be more dependent on the diversity of the particular group of mammals being examined, than on the correlation statistics of the equations generated.