Can the detection of salinity and habitat simplification gradients using rapid bioassessment of benthic invertebrates be improved through finer taxonomic resolution or alternative indices?

Benthic macroinvertebrates are commonly used in rapid bioassessment programs of rivers and streams with a few standard measures used to assess condition. These include AUSRIVAS predictive models, indices such as SIGNAL and EPT and a simple measure of taxa richness, number of families. Data in these programs are usually identified to family level. Until recently, there has been no assessment of the performance of these standard measures to specific environmental gradients. In addition, the level of identification remains contentious, with family level considered adequate in broadscale assessments whereas finer levels of resolution may be preferable in other situations. We tested whether finer levels of taxonomic resolution (of all taxa and just EPTs) improved detection of two environmental gradients - salinity and habitat simplification (through increased sandiness) - and also whether alternative indices enhanced detection of these gradients. Multivariate techniques demonstrated that different invertebrate communities were readily apparent for the salinity gradient (measured as electrical conductivity at the time of sampling) but less so for the habitat simplification gradient (measured as %sand in the riffle). By contrast, the standard measures (listed above) had low r (2) values with EC (< 0.34) for the salinity dataset but higher r(2) values with %sand (up to 0.72) in the habitat simplification dataset. Species level data and species EPT data did not account for much more variation in the data than their family level equivalents. Reformulating the EPT data from numbers of taxa to % of the community improved these relationships in some cases. Alternative indices based on Crustacea and Mollusca showed strong relationships to the salinity gradient but not to habitat simplification. These findings are discussed regarding the application of the RBA data to specific disturbances and the need to develop more diagnostic indices.