Detecting Regime Shifts in Small Ecological Datasets: Antarctica as a Case Study

This thesis explores the detection of ecological change in regions hampered by difficulty in collecting data. We investigae four statistical methods for detecting change points within time series data, looking at the Pettitt test, Lanzante method, Rodionov test, and Wild Binary Segmentation. Our analysis reveals that the Pettitt method consistently performs well in accurately identifying the number and location of change points. We then use the Pettitt test to look for patterns of change in penguin colonies in Antarctica, an ecology experiencing large change and consisting of few time series data points, and find significant changes in nesting counts.