Date of Award

Fall 12-15-2017

Level of Access

Open-Access Thesis

Degree Name

Master of Science (MS)

Department

Earth Sciences

Advisor

Stephen Norton

Second Committee Member

Aria Amirbahman

Third Committee Member

Sean Smith

Abstract

Understanding lake vulnerability with respect to eutrophication and loss of water quality is important for sustainability of aquatic ecosystems. This project aims at identifying and quantifying the effects of relevant physiochemical, climate, and watershed characteristics on lake vulnerability in order to develop management decision tools for the Maine Department of Environmental Protection (MEDEP). In a changing chemical and physical environment, using independent variables from each of these categories and then relating them to the summer lake epilimnetic phosphorus (P) concentrations allows for development of models to inform stakeholders of lake vulnerability to eutrophication problems.

We studied 24 lakes covering a range of trophic states (oligotrophic to mesotrophic) in Maine, USA. The lakes are classified as either dimictic or polymictic and may develop anoxic hypolimnia during stratification. Lake water samples were collected twice in June and August 2015, and analyzed for a variety of elements, with a primary focus on P. August epilimnetic P ranged from 1.9 to 21.0 µg/L (henceforth ppb). Sediment samples from the deepest point were collected in June 2015, and were sequentially extracted and analyzed for P, aluminum (Al), and iron (Fe). The results show that lakes with sediment having a NaOH-extractable Al to dithionite-reducible Fe ratio (AlNaOH:FeBD) > 3 and a NaOH-extractable Al to dithionite-reducible P ratio (AlNaOH:PBD) > 25 are less susceptible to internal P release, and have lower epilimnetic P concentrations. Ratios can thereby be used as sediment indicators for hypolimnetic P release under anoxic conditions.

Three types of regression models (regression tree analysis, multiple linear regression (MLR), and quantile regression (QR)) were developed in order to broaden understanding of different aspects impacting lake eutrophication using data from the 24 study lakes that represented relevant lake physiochemical, climate, and watershed characteristics. A larger database of lakes from the Maine Department of Environmental Protection (96), and the Lake Environment Association (23) were then used to validate the models by analyzing the goodness of fit. The regression tree analysis was performed to detect dominant drivers in relation to the August epilimnetic P concentrations, revealing that to best predict the lake epilimnetic P, parameters representing physiochemical, climate and watershed characteristics are necessary independent variables. Of the approaches tested for MLR, the best fits to the observed data were obtained by one or more physiochemical variables and one watershed variable (R2 > 0.78). Regression quantiles were used to estimate changes in epilimnetic P as a function of the agriculture area: watershed area (Ag:WA) ratio ranked by sediment AlNaOH:PBD and area depth (Zavg), all parameters that were shown to be important predictors in the MLR models. The structure of QR is robust for developing nutrient reduction targets for lake management. Using this approach, we determined that the reduction in Ag:WA to meet a specific epilimnetic P target (15 ppb) should be the first priority to mitigate eutrophication in Maine lakes. Using multiple regression models to identify and quantify factors that influence lake eutrophication allows us to classify susceptible lakes and inform stakeholders about appropriate practices for lake stewardship.

Files over 10MB may be slow to open. For best results, right-click and select "save as..."

Share