Final answer:
A non-stationary time series, in which statistical properties change over time, may be influenced by exogenous or endogenous factors affecting trends. Analyzing such data often involves different analytical techniques designed to capture these underlying trends and mechanisms.
Step-by-step explanation:
If at least one of the conditions of weak stationarity are violated, the time series is then called non-stationary. The properties of a non-stationary time series can be quite different from those of a stationary one. In a non-stationary series, statistical properties like the mean, variance, and autocorrelation structure are not constant over time.
Non-stationary time series may exhibit trends, either due to exogenous influences, such as changes in the environment that affect population change but are not influenced by population numbers, or endogenous influences, which involve dynamical feedbacks affecting population numbers and could involve time lags. Determining the interplay between these drivers is crucial for understanding the underlying patterns in the time series data.
Non-stationary data often require different analytical techniques to identify underlying trends and feedback mechanisms, such as power analyses to detect trends, application of nonlinear models, or the simultaneous application of multiple time series models for analysing relative abundance counts. It is imperative when querying abundance time series to engage in the multiple working hypotheses approach to gather a more comprehensive understanding of the forces at play.