Since Clements' ‘phytometer’, bioindicators have been used in various fields of ecological research. Bioindicators include both indicator species and biological indices (e.g., diversity index) that indicate a particular physical environment, community or partial ecosystem. In this paper, I examine some logical problems with the concept of indicator species and evaluate its utility. Supporters of the indicator species concept emphasize that it can detect the average, cumulative or extreme effect of the physical factor in question often by a simple short-term study, and that the results are more suggestive of the impacts on living organisms, including humans. Critics point out that indicator species are less decisive than physicochemical parameters and are sometimes invalid in different geographic sites. The problems of logic associated with the indicator species concept are summarized as follows: Its definition is inductive, its application amounts to affirming the consequent, and the logic is circular. Concerning the criticism of induction, perpetual in-situ tests of the indicator species are needed. The problem of affirming the consequent could be solved by repeated confirmation of the coincidence of the indicator species and the factors concerned in various situations, for which statistical significance tests are necessary. Circularity of logic is avoided by excluding the indicator species itself from its definition and basing the definition only on physical factors. When an indicator species is used for falsification of some previous assertion, or when plentiful background information is already available, such as in the case of temperature indicator species based on geographic distribution (e.g., tropical species), its interim employment is plausible. But in general, the use of indicator species is inappropriate without the process of standardization, because of the latent logical problems of the concept.
