Distribution Patterns of Mercury in Lakes and Rivers of Northeastern North America
- Cite this article as:
- Dennis, I.F., Clair, T.A., Driscoll, C.T. et al. Ecotoxicology (2005) 14: 113. doi:10.1007/s10646-004-6263-0
- 167 Views
We assembled 831 data points for total mercury (Hgt) and 277 overlapping points for methyl mercury (CH3Hg+) in surface waters from Massachussetts, USA to the Island of Newfoundland, Canada from State, Provincial, and Federal government databases. These geographically indexed values were used to determine: (a) if large-scale spatial distribution patterns existed and (b) whether there were significant relationships between the two main forms of aquatic Hg as well as with total organic carbon (TOC), a well know complexer of metals. We analyzed the catchments where samples were collected using a Geographical Information System (GIS) approach, calculating catchment sizes, mean slope, and mean wetness index. Our results show two main spatial distribution patterns. We detected loci of high Hgt values near urbanized regions of Boston MA and Portland ME. However, except for one unexplained exception, the highest Hgt and CH3Hg+ concentrations were located in regions far from obvious point sources. These correlated to topographically flat (and thus wet) areas that we relate to wetland abundances. We show that aquatic Hgt and CH3Hg+ concentrations are generally well correlated with TOC and with each other. Over the region, CH3Hg+ concentrations are typically approximately 15% of Hgt. There is an exception in the Boston region where CH3Hg+ is low compared to the high Hgt values. This is probably due to the proximity of point sources of inorganic Hg and a lack of wetlands. We also attempted to predict Hg concentrations in water with statistical models using catchment features as variables. We were only able to produce statistically significant predictive models in some parts of regions due to the lack of suitable digital information, and because data ranges in some regions were too narrow for meaningful regression analyses.