RESPONSE OF EPILITHIC DIATOM ASSEMBLAGES TO WATER-POLLUTION IN RIVERS IN THE TOKYO METROPOLITAN-AREA, JAPAN

1. During the spring of 1992, fifty-two quantitative diatom samples were collected from twenty-eight rivers located in the Tokyo Metropolitan area, Japan, to study the response of the diatom assemblages to water pollution (assessed using physical and chemical data determined monthly from April 1987 to March 1992). 2. Species composition was analysed by means of biotic indices (Pantle and Buck's saprobic index) and multivariate analyses [two-way indicator species analysis (TWINSPAN) for classification and canonical correspondence analysis (CCA) for ordination]. Species-abundance relationships were analysed using diversity indices (species richness, Shannon's diversity index and Pielou's evenness index) and rank-abundance patterns (rank-abundance curves). 3. CCA revealed two major gradients. The first corresponded to organic pollution and eutrophication. The second corresponded to variables related to geographical location. Four main station groups were determined by TWINSPAN. The location of the indicator species of groups 1-3 along the CCA axis 1 is consistent with their known pollution tolerance characteristics. Indicator species for group 4 had larger scores on CCA axis 2, and are representative of brackish water environments. 4. Species richness tended to be higher in the intermediate range of water pollution. Pielou's evenness index and Shannon's diversity index followed the same tendency but only weakly. 5. The rank-abundance patterns of diatom assemblages were more or less constant in all stations. The curves were very similar in shape, differing only in length and gradient (directly related to species richness and evenness, respectively). 6. The results of this study indicate that the response of diatom assemblages to environmental change can be observed in species compositional variation. Multivariate analyses and pollution indices revealed this response and are to be preferred to species diversity measures.