The metropolitan city of Perth has its beginning with the establishment of the Swan River Colony in 1829, around the Swan-Canning River estuary. Within the last 183 years the city has expanded with almost 1.4 million people inhabiting the metropolitan city impacting heavily on the ecology of the River system. Several natural streams discharging into the system have been drastically modified by intense urbanisation. The River system and its tributaries, streams and drains have been the recipients of nutrients, effluents, and waste water from the urban and rural catchments resulting in severe symptoms of eutrophication over the past several decades. While the rural catchments have been implicated in the eutrophication and pollution of the estuary, the contribution of urban catchments to the degradation of the system is relatively less known. Through urban catchment groups, and the Integrated Catchment Management Policy, the importance of urban streams and drains is just beginning to be recognised.
The current project was aimed at developing a predictive model for biomonitoring the urban streams using diatoms as tools. The investigation lasting from 1996 to 1999 focussed on classification of the urban streams using water quality parameters and 'stream conditions' and subsequent development of a predictive model using diatoms as biomonitors. Close to 180 sites were sampled in summer 1996, spring-summer 1997 and autumn-winter 1997 recording as many as 30 environmental variables. These sites represented most of the streams and drains (including 'concrete') in the urban and semi-rural areas of Perth. The sampling sites were established after analyzing the historical data on water quality and catchment conditions, as well as from preliminary observations.
A large number of 'reference sites' or 'relatively pristine sites' were required for the model. The semi-rural sites provided the best examples of reference sites, as they were least impacted by urbanisation. Diatom samples from both reference and impacted sites were collected using an artificial sampling device the JJ Periphytometer. All sites were classified on the basis of seven environmental variables, with the highest correlation coefficient with the sites. Agglomerative hierarchical fusion method with flexible unweighted pair group mean averages (UPGMA) within the multivariate pattern analysis program PATN was used. However, currently, PRIMER SOFTWARE (Plymouth Routine In Mutivariate Ecological Research) is considered to be a more efficient multivariate program. The dendrogram with the sites clustered according to their similarity was further analysed and the status of some of the sites revised, recognising a third group 'the Intermediates'.