Main content area

Evaluating the spatial variation of total mercury in young-of-year yellow perch (Perca flavescens), surface water and upland soil for watershed-lake systems within the southern Boreal Shield

Gabriel, Mark C., Kolka, Randy, Wickman, Trent, Nater, Ed, Woodruff, Laurel
Science of the total environment 2009 v.407 no.13 pp. 4117
mercury, chemical concentration, Perca flavescens, perch, surface water, hydrochemistry, upland soils, soil chemistry, spatial variation, watersheds, lakes, peatlands, regression analysis, statistical models, Superior National Forest
The primary objective of this research is to investigate relationships between mercury in upland soil, lake water and fish tissue and explore the cause for the observed spatial variation of THg in age one yellow perch (Perca flavescens) for ten lakes within the Superior National Forest. Spatial relationships between yellow perch THg tissue concentration and a total of 45 watershed and water chemistry parameters were evaluated for two separate years: 2005 and 2006. Results show agreement with other studies where watershed area, lake water pH, nutrient levels (specifically dissolved NO₃ ⁻-N) and dissolved iron are important factors controlling and/or predicting fish THg level. Exceeding all was the strong dependence of yellow perch THg level on soil A-horizon THg and, in particular, soil O-horizon THg concentrations (Spearman ρ =0.81). Soil B-horizon THg concentration was significantly correlated (Pearson r =0.75) with lake water THg concentration. Lakes surrounded by a greater percentage of shrub wetlands (peatlands) had higher fish tissue THg levels, thus it is highly possible that these wetlands are main locations for mercury methylation. Stepwise regression was used to develop empirical models for the purpose of predicting the spatial variation in yellow perch THg over the studied region. The 2005 regression model demonstrates it is possible to obtain good prediction (up to 60% variance description) of resident yellow perch THg level using upland soil O-horizon THg as the only independent variable. The 2006 model shows even greater prediction (r ² =0.73, with an overall 10 ng/g [tissue, wet weight] margin of error), using lake water dissolved iron and watershed area as the only model independent variables. The developed regression models in this study can help with interpreting THg concentrations in low trophic level fish species for untested lakes of the greater Superior National Forest and surrounding Boreal ecosystem.