Evidence necessary for taxa to be reliable indicators of environmental conditions or impacts

Using taxa as indicators of environmental impacts is widespread. Indicators are chosen because they are considered to be easy to measure, sensitive to stresses and respond to stresses in predictable ways. Here, we review these criteria by addressing the nature of the relationships between some characteristic of taxa and the environmental variables they are supposed to indicate. It is crucial that variables measured as indicators be strongly and consistently correlated (through space and time) with levels of the environmental variables. Appropriate experiments must be done to establish that an observed correlation is causal, or the correlation cannot be considered sufficient to identify a useful indicator. Finally, it is necessary to establish that the taxa directly respond to changes in the environmental variables they are supposed to indicate. Appropriate methodologies to establish these criteria are considered and we evaluate studies in which these criteria have or have not been met.