Relationships between lakewater pH and sedimentary diatoms in the Matamek Watershed,Northeastern Quebec,Canada

Twenty lakes in the Matamek and nearby watersheds were sampled in the summer of 1983 for water chemistry and surficial sediments. Thirty-two physical and chemical variables, including pH, alkalinity, cations and metal concentrations were measured on samples from the epilimnion and hypolimnion of each lake. In three lakes, two to four replicate cores were collected to estimate spatial variability of the sediment flora.All lakes were acidic (pH 4.59 to 5.80), highly colored and poorly buffered. Aluminum and magnesium concentrations reached 494 and 70 µg l^-1, respectively. The pH of the lakes appears to be declining, as indicated by a comparison of our results with those from a survey done in 1970. An empirical chemical model based on the alkalinity/sulfate ratio and the regression of pH on calcium also indicated that these lakes may be undergoing acidification.Analyses of the diatom flora of the surficial sediments showed strong dominance of 6 species that cooccurred with a large number of rare species (in all, 229 taxa were found). Variability among samples within a lake was as high as among lakes. Calculation of Nygaard's alpha index for each sample and the regression of its log-transformed value on surface pH yielded a relationship that was significantly from other published models. The regression model was applied to a down-core analysis of the diatom flora of the sediments of two lakes. Although the large confidence intervals on pH values predicted by the model obscured any evidence of pH change with sediment depth, there was a significant increase of acidophilous and a decrease of circumneutral species over time, suggesting that a change in the flora, possibly correlated with pH, is taking place.In making comparisons among lakes, the surficial-sediment flora did not provide clear evidence of a relationship with the pH of surface water. Indices computed from the surficial-sediment flora are apparently insensitive to differences in pH over a narrow range, particularly when the lakes being compared are similar chemically. Nygaard's alpha index is shown to be unduly sensitive to outliers. The currently accepted assumption of a progressive linear change in communities may be inadequate for the quantification of acidification processes. Other models, derived from catastrophe theory, may prove more fruitful.