Australian River Assessment System: Interpretation of the Outputs from AusRivAS (Milestone Report)
Monitoring River Health Initiative Technical Report Number 24
Dr L.A. Barmuta, Dr B.C. Chessman and Professor B.T. Hart - Land and Water Resources Research and Development Corporation
Environment Australia, 2002
- Australian River Assessment System: Interpretation of the Outputs from AusRivAS (Milestone Report) (PDF - 407 KB)
About the report
AusRivAS is a series of procedures and associated software for the rapid assessment of river conditions or "health" using macroinvertebrate communities. AusRivAS includes a set of predictive computer models relevant to particular Australian States and Territories, seasons (spring, autumn or spring and autumn combined) and habitats for river macroinvertebrates (e.g. riffles and pool edges). The models attempt to predict the macroinvertebrate families in a specified subsample from a standard kick or sweep net sample, assuming natural or near-natural conditions at the sampling site. The predictions are based on physical, chemical and vegetative features of the site. The predicted fauna can then be compared with the fauna actually observed to infer departures of the macroinvertebrate community from its natural state.
In this project we developed methods for interpretation of the raw output from the AusRivAS models, which consists of the probabilities of occurrence of the various families of macroinvertebrates. We used a series of structured workshops with river scientists, managers and interested community representatives, and a telephone conference, to facilitate the development of outputs tailored to various end users and uses. The latter include environmental summary reporting (e.g. State of the Environment), agency regulatory activities, catchment planning and management, community-based assessment, and education and training.
The interpretation is based on two ratio indices. One incorporates the number of families expected to occur in a standard subsample and the number of those actually observed (O/E FAMILIES). The other is based on the modelled and observed values of the SIGNAL biotic index (O/E SIGNAL); this ratio incorporates the tolerances of different macroinvertebrate families to common types of water pollution. Detailed guidelines are provided for combining values of these indices in each habitat, to generate bands reflecting varying degrees of departure of macroinvertebrate communities from model predictions. Band A represents conditions similar to those at most sites that are minimally disturbed by human activities (reference sites). Bands B, C and D comprise sites with progressive reductions in numbers of families, or in the average pollution sensitivity of the families. Band X includes sites with more families than predicted, which may be unexpectedly biodiverse either because of natural factors or mild nutrient enrichment as a consequence of human activities.
The present models, indices and banding scheme are seen as in need of considerable further testing and improvement. Specific areas for research and development include the following:
- More standardised rules for the selection of reference sites on which the models are based, including clear ecological and management-based criteria for the degree of human disturbance that is acceptable at a reference site (see Section 3.1).
- Quantification and, where possible, reduction of the sources of variability and statistical error in the derivation of both the predicted and the observed macroinvertebrate community composition at a site (see Section 3.4).
- A more rigorous approach to the selection and use of predictor variables in the models, especially those variables whose values at test sites can be affected by human activities (see Section 3.2).
- Further investigation of the merits of alternative statistical and numerical methods of prediction, in particular detailed comparisons of DDRAM and e-ball with the present method (see Section 3.5).
- Testing of the merit of combining predictions from different habitats to obtain an overall prediction for a site using weighted probabilities (see Section 3.3).
- Further development of SIGNAL or other approaches using tolerance information to provide diagnosis of causes of impairment of macroinvertebrate communities at test sites (see Section 3.7).
- The inclusion and reporting of non-predictor environmental variables for diagnostic purposes (see Section 3.8).
- Investigation of alternative banding schemes including schemes based on defined effect sizes (degrees of family loss) that jeopardise attainment of policy objectives or can be shown to be detrimental to ecosystem functioning (see Section 3.4).
- A close, formal liaison between agencies responsible for environmental audits to ensure integration of AusRivAS into State of the Environment reporting procedures in a scientifically valid manner (see Section 3.9).