Toward river health assessment using species distributional modeling

The reference condition of a river ecosystem is often used as a benchmark against which restoration activities are framed. However, the reference condition approach is not suitable for all river health assessments given undisturbed reference sites similar to a test site may not exist. We tested the ability of applying individual Species Distribution Models (SDMs) to predict the “natural” fish assemblages in a regulated coastal Australian river. Environmental data and fish collections from training sites were used to generate predictive models for 21 species. Model accuracy was high with the area under the receiver operating curve greater than 0.80 for the majority of species. The SDMs were then used to determine the natural fish assemblages and the condition of the sites in the test river was then assessed as the observed to expected species ratio (O/E). We expected that the O/E ratio would decline upstream through a series of weirs that created barriers to fish passage. Results confirmed that an inverse relationship existed between the number of weirs present on the test river and the O/E ratio, largely due to the decline in diadromous species. Our study clearly demonstrates the usefulness of the SDM approach to riverine health assessment in situations where undisturbed reference sites do not exist.