Development of a sampling framework for the Australian Alps stream health monitoring project

Nerida Davies and Richard Norris, Cooperative Research Centre for Freshwater Ecology | April 1999


Executive Summary

1. Predictive models for the assessment of river health have been successfully implemented in Britain (RIVPACS), the United States (the BEAST), Australia and Indonesia (AUSRIVAS). These models match test sites to a set of reference sites using habitat and environmental features of the site. Taxa likely to occur at test sites are predicted using the reference data and compared to those collected, the differences providing an assessment of damage. This provides an independent method of comparison, avoiding problems of upstream/downstream controls and comparisons to predetermined thresholds that may be unrealistic to expect for many sites.

2. Reference sites are used to characterise the region based on their physical, chemical and biotic attributes. The condition at a test sites will be described in the reference data so that accurate site comparisons can be made.

3. Selection of reference sites involved choosing more than 100 sites from relatively unmodified regions of the Alps. These sites were chosen to describe the range of altitudes (> 1000m), vegetation types, slope and conditions experienced within the national parks of the Australian Alps.

4. Test sites were chosen to determine if the AUSRIVAS model for the Alps will be able to detect degradation in biological condition.

5. Physical and chemical water quality variables will be measured at all sites to relate to the biota and assist in identifying if any biological impairment at a site results from water quality issues or land use or habitat degradation.

6. Habitat variables will be measured at all sites and will be related to biological data in model development. Ultimately the habitat and environmental data will be used to determine the macroinvertebrate community that should occur at a site in the absence of the effects of human activities.

7. The predictive technique involves classifying a large number of reference sites into groups using the biological attributes. A discriminant function analysis is used to determine how well the groups describe the structure in the environmental habitat data. The habitat features at a site are then used to determine into which reference group a test site should belong and ultimately the taxa that should occur (EXPECTED) at that site in the absence of environmental stress. These EXPECTED taxa are then compared to what was found at a site (OBSERVED) to provide an assessment of the sites (OBSERVED to EXPECTED ratio -O/E).

8. Biological bands are used to allocate a level of biological impairment to the O/E ratio for the purposes of management, for example Band A the biological condition is equivalent to reference through to Band D that indicates that the biological condition of that sites is impoverished.

9. Ongoing use of the model only requires sampling of test sites to assess the condition of streams and rivers across the national parks of the Australian Alps.