Hard-bottom invertebrates were visually sampled in a man-modified bay during a harbour enlargement scheme. Several environmental variables were repeatedly measured. Faunal assemblages were analysed using classification and ordination (MDS) techniques at different levels of resolution (species, families and indicators). Indicator taxa were selected from the full species data set by performing preliminary correlations between faunal and environmental variables. Similar results were observed at the three levels investigated. This suggests the existence of redundant information in the species level data set when applied to this case study. The implications of our findings for environmental monitoring are discussed.