Identifying freshwater systems that are at risk from anthropogenic stressors is a pressing management problem. In particular, the detection of metal pollution is often constrained by data availability and resources. To address this challenge and develop a tool to identify susceptible systems, we tested whether land cover could be predictive of stream sensitivity to metal pollution, as determined by the biotic ligand model (BLM). We used water chemistry data from the conterminous United States to estimate metal sensitivity in streams using two BLMs (i.e., HydoQual, Bio-Met). Subsequently, we combined the sensitivity estimates with land cover and physiochemical data from the GAGES-II database to build predictive models of sensitivity to metals in streams. When combined, our predictor variables (e.g., land cover, mean annual temperature, mean annual precipitation) generally explained about half of the variation in our dataset. In each model, the percent of wetlands in a watershed was strongly correlated with reduced sensitivity to metals, likely due to increased concentrations of dissolved organic carbon associated with wetlands. To validate the utility of the models, we used them to predict metal sensitivity in sites where metal concentrations had been collected, but where the full suite of BLM parameters were unknown. We were able to classify several hundred sites which are likely at risk to metal pollution. Our work highlights the value in considering metal toxicity at the landscape-scale and describes a new approach to estimate metal sensitivity when site-specific chemical parameters are unknown.
Keywords: Biotic ligand model; Copper; Land cover; Land use; Landscape; Landscape ecotoxicology; Nickel; Watershed; Zinc.
Copyright © 2018. Published by Elsevier Ltd.