Scaling up the value of bioindicators.

S pecies whose presence or abundance readily reflect some measure of the character of the habitat within which they are found are often identified as bioindicators 1 , most frequently to monitor changes within a particular habitat. In spite of the intuitive appeal of bioindication, largely as a consequence of its cost-effectiveness in the face of urgent conservation issues, indicator studies frequently fall short of providing spatially explicit and objectively determined indicator species or species subsets. This is particularly true of studies involving invertebrates. A recent paper by Dufrêne and Legendre 2 outlines and applies a straightforward and efficient method for addressing this shortcoming. Although previously-developed and commonly-used techniques seek to identify bioindicators within groups of taxa (e.g. I-tests, correspondence analysis, two-way indicator species analysis), many of these suffer from methodological problems 3,4. For example, one is frequently constrained in the choice of site clusters for the identification of indicator sets by the site classification procedure used. Also, many of these methods use relative species abundance measures that are easily biased by sampling methods with differential sampling efficiency for different taxa. This use of compositional data can also severely bias the patterns that are generated by classification procedures 5. Furthermore, many techniques have a tendency to select rare species as being distinctive of particular habitats or sample units 3. The risks associated with such choices include the possibility of selecting vagrant species, or individuals from non-viable or sink populations 6. The value of bioindicators would thus be highest if species whose indicator values were calculated independently of other species in the assemblage were truly representative of a group of sites, by being unique to that site group (high specificity), as well as by being abundant and widespread within it (high fidelity). Such species would not only have a high information content, but also a high probability of being sampled during monitoring and assessment. If, at the same time, one had the freedom to combine sites in as many meaningful ways as required, the optimum group of sites necessary for the conservation of the indicator species could be established. This approach is of particular value when dealing with taxa, such as invertebrates, where there is a paucity of information on the distribution and biology of the majority of species sampled. This is the essence of the approach and method advocated by Dufrêne and Legendre 2. The indicator value (IndVal) of a …