Results

Samples were collected from 89 sites in upper, mid and lower reaches of 30 randomly selected estuaries in Victoria and southern NSW (to Batemans Bay). At each site two methods were used to collect samples. An Ekman Grab was used to target infauna and a dredge for epibenthos. It was intended to use these as separate data sets to investigate which was the most appropriate sampling method. However, neither method collected sufficient numbers of species per site to develop an AusRivAS type model. To overcome this we amalgamated the two data sets. This was possible because AusRivAS type models require only that a standard sampling method be employed and uses presence/absence data rather than abundances of the different taxa.

Almost a fifth of the estuaries visited were 'layered'. These did not have tidal mixing because the mouth was closed and had only limited opportunity for wind mixing because they were narrow, i.e. they did not flow into coastal lakes. Layered estuaries usually had fresh water on the surface and a submerged body of denser brackish water. Probably because of the low availability of oxygen, such estuaries had very sparse macrofauna, with numbers in samples being too low to give an accurate representation of the community. Data from some other sites were excluded from model construction due to incomplete biological or environmental data. The pilot model was based on 58 sites. This compares with several hundred reference sites in most AusRivAS models. Nevertheless, with the exclusion of the layered estuaries we found that our model was relatively accurate in predicting O/E ratios for our test sites.

The second part of our project was to assess if an AusRivAS type model would be of value for monitoring estuarine health. Sensitivity in detecting environmental impacts depends on the number of expected taxa in the O/E ratio. If there are 20 expected taxa, a 25% species loss results in the observed taxa being 15. A loss of five taxa would be seen as significant. However, if there are only four expected taxa, a 25% loss is the disappearance of only one taxon. This may be due to chance. IN this case even a 50% loss of expected taxa is a decrease of only two taxa which would probably not be considered significant. With an expected number of four taxa, a loss of 75% would "set alarm bells ringing". But at this stage the environmental degradation would probably be obvious and not require the collection and sorting of samples. The numbers of expected taxa for our six test sites ranged from 4 to 13 with the median being 6.6. These are too low for the model to be of practical value. Without any additional work the numbers of expected taxa could probably be increased slightly by using a different sampling design from that in the pilot study. However, to increase the numbers of expected taxa to values similar to those in the river models would require a massive sorting effort.