A comparison of the sensitivity of stream benthic community indices to effects associated with mines, pulp and paper mills, and urbanization

This study examined the relative sensitivities of seven commonly used indices of stream benthic community composition and three multivariate indices to effects associated with mines, pulp and paper mills, and urbanization. The indices included total abundance, number of taxa, diversity (H'), evenness, Hilsenhoff's biotic index (HBI), the BioMAP water quality index (WQI), the percent model affinity (PMA), and the first three ordination axes from a correspondence analysis. The second objective of the study was to determine the degree of redundancy among these indices. Six data sets (two from each of the three types of development) were used to address the objectives. In each data set, replicate benthic samples were collected from reference areas as well as one or more downstream areas exposed to a point-source or non-point-source discharge. The PMA approach and the ordination axes indicated significant differences between the reference and downstream communities for all six data sets (p < 0.05). With the exception of H', each of the other metrics revealed significant effects associated with one or two, but not all three, types of development. For example, the HBI and WQI indicated significant effects associated with pulp and paper mills and urbanization, but not mining. In all studies, effect sizes (i.e., the standardized difference between means for the reference and exposed areas expressed in units of standard deviations [SDs]) exceeded 2 SDs. However, effect sizes for the PMA and the first or second ordination axis scores were generally larger than effect sizes for the other metries, indicating that these indices were often the most sensitive indicators of development. In addition, a high degree of redundancy was found among the various metrics, perhaps because the effects were large (i.e., >2 SDs). The greater sensitivity of the ordination axes and the PMA approach emphasizes their value as indices of benthic community composition. As a result, we recommend that any suite of indices used for assessing benthic communities should include these types of multivariate metrics.