Detecting Changes in Ecological Time Series

Abstract
Some practical techniques are discussed for analyzing time series whose statistical properties are changing with time. We first consider how principal component analysis can reduce the multidimensional nature of certain series and, in particular, apply this technique to the analysis of changing seasonal patterns. Discussions of trend, changes in oscillatory behavior, and unusual events follow. The problem of making inferences regarding causation is briefly considered. We conclude with a call for flexibility in approach.

This publication has 12 references indexed in Scilit: