Environmental Change and Population History of North America from the Late Pleistocene to the Anthropocene

Abstract

The assumption that prehistoric Native American land use practices had little impact on the North American landscape persists in the literature. However, recent research suggests the effects of prehistoric burning, deforestation and agriculture may potentially have been greater than previously considered. To resolve this discrepancy, quantitative estimates of changes in human population size and forest structure and composition over the course of the Holocene are needed. This thesis addresses this need by providing radiocarbon-based paleodemographic reconstructions and pollen-inferred estimates of vegetation change, as well as analyses of associations between the two at both continental and regional scales, from the late Pleistocene to the Anthropocene. One way to estimate paleodemographic change is to use the number of radiocarbon (14C) dates from a given area to study patterns of human occupation through time. A review of the literature and compilation of existing databases relevant to this method showed there is now sufficient data to study the paleodemographic history of many regions around the world. An analysis of 14C datasets from North America and Australia compared well with model-based reconstructions of past demographic growth, and provided higher frequency fluctuations in population densities that will be important for future research. Using a kernel density estimation approach, the first estimates of prehistoric population density for North America were obtained and synthesized into a series of continental-scale maps showing the distribution and frequency of 14C dates in the Canadian Archaeological Radiocarbon Database (CARD). The maps illustrated the space-time evolution of population and migration patterns, which were corroborated by independent sources of evidence. A methodology based on the statistical evaluation of cross-correlations between population and plant abundance was then developed to analyze the associations between these population estimates and plant communities derived from pollen databases. Periods of high spatial cross-correlation (positive and negative) between population and plant abundance were irregular and did not improve over time, suggesting that ancient human impacts are not discernable at a continental scale, either due to low populations or varying human land use practices. To further examine the relationship between pollen data and human land use at a regional scale, estimates of plant density and landscape openness are needed. The REVEALS (Regional Estimates of VEgetation Abundance from Large Sites) model corrects for the non-linear relationship between pollen production and plant abundance and can therefore be used to map histories of land use and land cover change. The model was applied to pollen records from lake sediments in the deciduous forest of southeastern Quebec. A preliminary analysis comparing these results to population density revealed low population during times of high Populus abundance and high population following the appearance of the mixed temperate forest suggesting a discernable human-environment association at regional scales. Overall, the results of thesis support the growing body of literature that suggests prehistoric Native Americans impacted their environments and that these impacts can be detected and quantified by integrating archaeological and paleoecological information. However, the timing, location, and intensity of human land use has changed in both space and time, suggesting regional- to local-scale analyses of human-environment interactions are most appropriate for continental North America. The methodology presented here can be used to study additional North American regions for the purpose of developing a continental history of human-environment interaction.