• Danielson, Thomas John
  University of Maine Graduate School
  

The purpose of this study was to test and develop algal methods of evaluating the condition of Maine streams and rivers. The primary objective was to develop a statistical model to predict attainment of Maine's aquatic life criteria for water quality classes A, B, and C. I collected 298 samples of algae on rocks from 193 locations across the state. The major pattem in species composition related to conversion of forests to urban, residential, and agricultural land uses. I calculated preferred environmental conditions of 236 algal taxa for 1) concentrations of nitrogen, phosphorus, and dissolved ions in the water, 2) percent of watershed land cover that is not forested, 3) and percent of watershed land cover that is impervious, such as pavement. I then tested and identified algal community metrics that responded to increasing watershed development. Metrics derived from Maine data performed better than metrics developed in other parts of the world. Five biologists with Maine's Department of Environmental Protection (MDEP) grouped samples based on attainment of aquatic life criteria (i.e., A, B, C, and non- attainment) by interpreting algal species abundances and community metrics. I developed a statistical model to replicate biologist assignments, which currently classified  95% and 91% of samples used to build and test the model. The second objective was to develop models based on algal community composition to estimate concentrations of nitrogen and phosphorus in stream water. A multiple linear regression model and a variation of weighted averaging that weights estimates using localized subsets of data performed the best. The final objective was to use nutrient diffusing substrates to determine if growth of benthic algae in the Sheepscot River was limited by phosphorus or nitrogen. It was co-limited by nitrogen and phosphorus. Although my statistical models have limited transferability to adjacent regions with similar ecological conditions, methods used to build the models have wide transferability. MDEP could use the first model to determine if streams and rivers attain water quality classes A, B, and C. MDEP could then use nutrient inference models and diffusing substrates to better diagnose and manage enrichment of phosphorus and nitrogen