No abstract

1. The paper describes diatom communities from a series of linked urban lakes in relation to water chemistry and uses mullivariate statistical techniques to show how indicator groups can be defined.2. Diatoms are classified into ecological groups using two-way species indicator analysis (TWINSPAN). Each ecological group is assiciated with a specific range of water-quality conditions. The headwater stream environments are differentiated from the lake habitats at level 1. Al level 2 of TWINSPAN, the sampling sites are grouped into five ecological groups. according to their water chemistry and irrespective of the time of year.3. Detrended correspondence analysis (DCA) and principal components analysis (PCA) provide two statistically independent methods of assessing the environmental gradients along which the ecological groups are distributed. The most important physico-chemical parameters are total hardness, specific conductance and pH, followed by phosphates and nitrates.4. Ecological groups I-V correspond to an environmental gradient ranging from the forested headwaters, which are acidic, of low specific conductance, total hardness and nutrient content, through the urban lakes, which are alkaline and of relatively high total hardness, specific conductance and nutrient content.5. Twelve site groups are identified at level 3 of TWINSPAN. each of which corresponds to a narrower range of water-quality conditions within the ecological groupings. A specific diatom assemblage is associated with each site group.6. Benthic diatoms form an intregral part of the diatom assemblages found in the water column and this is identified as a topic for further study.

SUMMARY. 1. Five diatom assemblages and their corresponding site groups were associated with the urban lakes. When treated as a single dataset, DC A and PCA grouped these diatom assemblages along an environmental gradient from circum natural. dilute waters to strongly alkaline, concentrated, nutrient‐rich waters. 2. When treated as individual datasets, 75% of the known variance was explained by the first axis of DCA in all five diatom assemblages; this indicated the existence of a strong primary environmental gradient. The PCA results showed that the identity of this primary environmental gradient can be different for each assemblage. 3. The statistical analyses showed that the physicochemical gradients were continuous and not discrete. Nevertheless, fine‐scale analysis can identify smaller, more precise species and site groups. 4. Two types of species response could be identified within most sub‐assemblages in the urban lakes: the first was a quasi‐seasonal response and was characterized by a gradual rise to a peak relative abundance of one dominant or several co‐dominant species which could be sustained for a period of weeks before a gradual decline. A quasi‐seasonal response was sometimes interrupted by a storm response. The storm response was characterized by a sudden increase in the relative abundance of one or more diatom species following a storm, after which the species declined to pre‐storm abundances within 7 days. 5. The effects of urban runoff on the water quality of urban lakes can override the effects of true seasonal factors such as water temperature and day length on the relative abundance of species.

No abstract

No abstract