The Effects of Substrate Type on Diatom-Based Multivariate Water Quality Assessment in a Tropical River (Monjolinho), São Carlos, SP, Brazil

Diatoms are good indicators of water quality in lotic systems. Unlike in the temperate region, the effect of substrate on diatom-based water quality assessment in tropical streams is not fully understood. The purpose of this study was to assess the effect of substrate on diatom-based multivariate water quality assessment. Epilithic, epiphytic, epipsammic and epipelic diatom community and water quality sampling was done four times at 10 sites during the dry season (2008 and 2009). Artificial substrates (brick and glasses) were also placed at the sampling sites during this period and sampled after 1 month. Cluster analysis was performed to show the main differences and similarities in community composition amongst substrates sampled and amongst sampling sites. The IndVal method was used to identify indicator species characterising different substrates. Canonical correspondence analyses (CCAs) were performed to relate the structure of diatom communities from different substrates to predictor variables. A gradient of increasing metal and organic pollution, eutrophication and ionic strength was observed from the agricultural/forested area to the urban area. Diatom community structure closely reflected this gradient, with communities from polluted sites (8, 9 and 10) being different from other communities. Polluted sites were associated with such species as Nitzschia palea, Plantago lanceolata, Achnanthes exigua, Caloneis hyaline, Cyclotella meneghiniana, Gomphonema parvulum, Fallacia monoculata, Luticola goeppertiana, Pinnularia microstauron, Pinnularia subcapitata and Sellaphora pupula. Indicator species analysis showed that common diatom species were not restricted to a single substrate, though preference was generally high for natural (especially macrophytes) compared to artificial substrates. Six CCAs corresponding to six substrates performed to relate diatom community structure to simultaneous effects of predictor variables explained ∼50% of the diatom species variance in all cases and roughly separated highly polluted sites from the rest of the sites. This indicates that the results of diatom-based multivariate water quality assessment based on different substrates may be interchangeable. Only one substrate has to be collected at each site for water quality assessment surveys, thus avoiding unnecessary expensive and time-consuming oversampling. Given the limitations of artificial substrates, sampling of natural substrates is highly recommended.