Analyses of structural changes in ecological time series (ASCETS)

Assessing status of natural resources and ecosystem components is pivotal for management, where indicators or indices often are used as proxies of ecological state. Many indicators, however, lack reference points and are associated with sampling errors and environmental noise, limiting their usefulness in management. Here we present a method for assessing state changes in ecological indicator from time-series: Analyses of Structural Changes in Ecological Time Series (ASCETS). ASCETS enables both quantitative boundary levels for changes in indicator states (e.g. for management targets), and the confidence for a change in state during an assessment period. Thereby it can be used in risk assessments and is suitable for aggregation or integration of different indicator states across sites, or for an ecosystem based approach to management. With extended information about ecological state during a reference period, ASCETS can support reference levels for defining ecological status of an indicator. ASCETS first identifies structural changes in time-series to determine reference periods with coherent indicator dynamics. Next, from the observed indicator values during the reference period, a distribution of resampled median values is used to set boundary levels as a tolerable range of indicator variation reflecting the same state as during the reference period. Finally, a confidence of a change in indicator state is evaluated during an assessment period as the proportion of resampled median values of the assessment period overlapping the boundary levels of the reference period. Simulations indicate ASCETS correctly detects changes in indicator state when changes in indicator values are at least twice as large as the coefficient of variation, with a false rate of changes around 5%. We apply ASCETS to indicators for bird and fish communities used within the Marine Strategy Framework Directive to illustrate how indicator boundary levels can be set where reference levels may be ecologically and analytically troublesome. An R-script is provided for further use and modification. We propose ASCETS as a flexible and generic method for assessing changes in ecological states from time-series to support identification of management targets.