Periphyton, water quality, and land use at multiple spatial scales in Alberta rivers

The ability of land use to replace water quality variables in predictive models of periphyton chlorophyll a was tested with a 21-year data set for Alberta rivers. Nutrients (total dissolved P and NO2 + NO3) explained 23%-24% of the variability in seasonal chlorophyll a, whereas land use (human population density) explained 25%-28% of the variability. The best models included the combination of total dissolved P and population density, explaining 32%-34% of periphyton chlorophyll a variability. However, analysis of variance of chlorophyll a by ecoregions and ecozones explained about as much variability (28%-30%), and the inclusion of an ecoregion term into the regression models showed a diminished importance of land use as a predictor of chlorophyll a, with best models based on the combination of nutrients and ecoregion and explaining up to 43%-44% of periphyton chlorophyll a variability. Within ecoregions, land use was sometimes a good surrogate for nutrient data in predicting chlorophyll a concentrations. Overall, land use is a suitable surrogate for nutrients in regression models for chlorophyll a, but its inclusion in general models may reflect regional differences in nutrient-chlorophyll relationships rather than true land use effects on chlorophyll a.