Abstract
Contaminants of emerging concern (CECs) include a variety of pharmaceuticals, personal care products, and hormones commonly detected in surface waters. Human activities, such as wastewater treatment and discharge, contribute to the distribution of CECs in water, but other sources and pathways are less frequently examined. This study aimed to identify anthropogenic activities and environmental characteristics associated with the presence of CECs, previously determined to be of high priority for further research and mitigation, in rural inland lakes in northeastern Minnesota, United States. The setting for this study consisted of 21 lakes located within both the Grand Portage Indian Reservation and the 1854 Ceded Territory, where subsistence hunting and fishing are important to the cultural heritage of the indigenous community. We used data pertaining to numbers of buildings, healthcare facilities, wastewater treatment plants, impervious surfaces, and wetlands within defined areas surrounding the lakes as potential predictors of the detection of high priority CECs in water, sediment, and fish. Separate models were run for each contaminant detected in each sample media. We used least absolute shrinkage and selection operator (LASSO) models to account for both predictor selection and parameter estimation for CEC detection. Across contaminants and sample media, the percentage of impervious surface was consistently positively associated with CEC detection. Number of buildings in the surrounding area was often negatively associated with CEC detection, though nonsignificant. Surrounding population, presence of wastewater treatment facilities, and percentage of wetlands in surrounding areas were positively, but inconsistently, associated with CECs, while catchment area and healthcare centers were generally not associated. The results of this study highlight human activities and environmental characteristics associated with CEC presence in a rural area, informing future work regarding specific sources and transport pathways. We also demonstrate the utility of LASSO modeling in the identification of these important relationships.