Selecting and applying indicators of ecosystem collapse for risk assessments.

ABSTRACT

Ongoing ecosystem degradation and transformation are major threats to biodiversity. Measuring ecosystem change toward collapse relies on monitoring indicators that quantify key ecological processes. Yet little guidance is available on selection and use of indicators for ecosystem risk assessment. We reviewed indicator use in ecological studies of ecosystem collapse in marine pelagic and temperate forest ecosystems. We examined indicator-selection methods, indicator types (geographic distribution, abiotic, biotic), methods of assessing multiple indicators, and temporal quality of time series. We compared how these factors were applied in the ecological studies with how they were applied in risk assessments by using the International Union for Conservation of Nature's Red List of Ecosystems (RLE), for which indicators are used to estimate risk of ecosystem collapse. Ecological studies and RLE assessments rarely reported how indicators were selected, particularly in terrestrial ecosystems. Few ecological studies and RLE assessments quantified ecosystem change based on all 3 indicator types, and indicators types used differed between marine and terrestrial ecosystems. Several studies used indices or multivariate analyses to assess multiple indicators simultaneously, but RLE assessments did not because as RLE guidelines advise against them. Most studies and RLE assessments used time-series data that spanned at least 30 years, which increases the probability of reliably detecting change. Limited use of indicator-selection protocols and infrequent use of all 3 indicator types may hamper accurate detection of change. To improve the value of risk assessments for informing policy and management, we recommend using explicit protocols, including conceptual models, to identify and select indicators; a range of indicators spanning distributional, abiotic, and biotic features; indices and multivariate analyses with extreme care until guidelines are developed; time series with sufficient data to increase ability to accurately diagnose directional change; data from multiple sources to support assessments; and explicitly reporting steps in the assessment process.