Archaeological and Palaeoenvironmental Time-series Analysis

The effects of modern climate change will be felt for centuries to come. Planning for that future right now is very difficult, however. We do not know how human societies respond to climate change over the long term. Modern and historically recent cases cannot provide us with a solid basis for making predications about the future because modern climate change has not been going on long enough to see its full effects. Instead, we need to look to the archaeological record for examples of long-term human responses to climate change. Despite more than a century of effort, though, archaeologists have made limited progress in understanding past human-environment dynamics. Archaeological and palaeoenvironmental datasets have improved markedly, but attempts to link those records have so far been unconvincing. The primary reason for this is a lack of appropriate quantitative tools. Archaeological and palaeoenvironmental data contain idiosyncrasies—namely temporal autocorrelation and chronological uncertainty—that undermine statistical methods. Given the seriousness of modern climate change, we need to rectify this situation. In this dissertation, I lay the groundwork for developing a quantitative toolkit for analyzing long-term human-environment dynamics. The dissertation is comprised of four studies involving time-series methods. The first two look at the impact of climate changes on the Classic Maya using two types of time-series analysis, and the last two use simulations to probe the limits of these methods. Together, the four studies demonstrate that the idiosyncrasies of archaeological and palaeoenvironmental data create challenges for quantitative analyses. Reviewing the studies, I identify the main methodological challenges and sketch out some potential solutions, illuminating a path for future methodological development.