It can be argued that the very act of selecting indicators of any kind is biased by the observer’s perspective[34] but separation of goals from descriptions has been advocated as a step toward transparency: “A separation of descriptive and normative indicators is essential from the perspective of the philosophy of science … Goals and values cannot be deduced directly from descriptions … a fact that is emphasized repeatedly in the literature of environmental ethics … Hence, we advise always specifying the definition of indicators and propose clearly distinguishing ecological indicators in science from policy indicators used for decision-making processes.”[31]
And integration of multiple, possibly conflicting, normative indicators into a single measure of “ecosystem health” is problematic. Using 56 indicators, “determining environmental status and assessing marine ecosystems health in an integrative way is still one of the grand challenges in marine ecosystems ecology, research and management”[35]
Another issue with indicators is validity. Good indicators must have an independently validated high predictive value, that is high sensitivity (high probability of indicating a significant change in the indicandum) and high specificity (low probability of wrongly indicating a change). The reliability of various health metrics has been questioned[36] and “what combination of measurements should be used to evaluate ecosystems is a matter of current scientific debate.”[3] Most attempts to identify ecological indicators have been correlative rather than derived from prospective testing of their predictive value[37] and the selection process for many indicators has been based upon weak evidence or has been lacking in evidence.[38]