7. Specificity and Sensitivity of Indicators

Long lists of indicators can present challenges for drawing inference about overall ecosystem status. A useful way to interpret lists of indicators in aggregate focuses on one of the primary considerations in the set of evaluation criteria introduced above, “the indicator responds predictably and is sufficiently sensitive to a specific ecosystem attribute.” Two of the terms in this criterion, “specific” and “sensitive,” can be used to organize indicators according to the type of information they provide about attributes. Rapport et al. (1985) proposed that an indicator’s specificity can be distinguished based on whether it reliably tracks few or many attributes. An indicator that provides information about many attributes (even attributes of multiple PSP goals) is non-specific but perhaps broadly informative of ecosystem status. An indicator that serves well as a proxy for fewer attributes can be thought of as diagnostic of changes in specific ecosystem characteristics. For example, in Figure 8 harbor seals are a non-specific indicator for Species and Food Webs attributes whereas jellyfish are a diagnostic one.

Another informative axis on which to interpret an indicator is in terms of its sensitivity. An indicator that provides information about impending changes in attributes before they occur is an early warning or “leading” indicator. For instance, due to fast turnover rates, phytoplankton are likely to be an early warning indicator for Species and Food Web attributes in Puget Sound (Figure 8). In contrast, an indicator that reflects changes in attributes only after they have occurred is a retrospective or “lagging” indicator. Retrospective indicators, such as killer whales (Figure 8), are likely to be characterized by slow turnover rates, but can nonetheless be useful for interpreting cumulative impacts and ecosystem-wide shifts in attribute values.