Potential biases in site and species selection for ecological monitoring

A number of the contributions to this workshop (particularly in the session on biodiversity and population monitoring) have stressed the importance of choosing appropriate species and locations for ecological monitoring. This is of pragmatic necessity: it is impossible to monitor all species at all locations. The purpose of this note is to caution against some possible pitfalls in the selection of species and sites. These pitfalls are a result of the statistical concept of regression.