A probabilistic model for silver bioaccumulation in aquatic systems and assessment of human health risks

Abstract

Silver (Ag) is discharged in wastewater effluents and is also a component in a proposed secondary water disinfectant. A steady-state model was developed to simulate bioaccumulation in aquatic biota and assess ecological and human health risks. Trophic levels included phytoplankton, invertebrates, brown trout, and common carp. Uptake routes included water, food, or sediment. Based on an extensive review of the literature, distributions were derived for most inputs for use in Monte Carlo simulations. Three scenarios represented ranges of dilution and turbidity. Compared with the limited field data available, median estimates of Ag in carp (0.07-2.1 micrograms/g dry weight) were 0.5 to 9 times measured values, and all measurements were within the predicted interquartile range. Median Ag concentrations in biota were ranked invertebrates > phytoplankton > trout > carp. Biotic concentrations were highest for conditions of low dilution and low turbidity. Critical variables included Ag assimilation efficiency, specific feeding rate, and the phytoplankton bioconcentration factor. Bioaccumulation of Ag seems unlikely to result in toxicity to aquatic biota and humans consuming fish. Although the highest predicted Ag concentrations in water (> 200 ng/L) may pose chronic risks to early survival and development of salmonids and risks of argyria to subsistence fishers, these results occur under highly conservative conditions.

Site‐Specific Applications of Probabilistic Health Risk Assessment: Review of the Literature Since 2000

“…Bosgra et al (2005) used probabilistic risk characterization to plot the confidence in concluding that contaminant intake by humans is below regulatory limits. Warila et al (2001) used a probabilistic bioaccumulation model and a distribution for exposure parameters to calculate a hazard quotient for human exposure to silver in surface water. Hamed (2000) studied the effects of probability distribution shape (for both exposure and toxicity factors) on the human cancer risk resulting from uptake of benzo(a)pyrene‐contaminated soil.…”

Time and space issues in ecotoxicology: Population models, landscape pattern analysis, and long-range environmental chemistry

“…They defined the concept of elasticity, which in modeling parlance is equivalent to the sensitivity of population growth rates to different life‐stage parameters (juvenile mortality, adult reproduction rate, etc.). In a case where probabilistic analysis was applied to the simulation of silver bioaccumulation in aquatic ecosystems [53], it was shown that the statistical distribution of many toxicologically relevant parameters is nonnormal. Accordingly, realistic ecotoxicological modeling should include adequate estimates of the probability functions of the various parameters characteristic of population dynamics.…”

A bayesian approach to parameter estimation for a crayfish (Procambarus spp): Bioaccumulation model

“…The recent publication of Volume 3 of the Risk Assessment Guidance for Superfund [2] suggests several probabilistic modeling approaches with varying degrees of sophistication to yield a probability distribution that characterizes uncertainty and/or variability in risk within a population. In the past, one-dimensional Monte Carlo (MC) simulation [3][4][5][6][7][8] and two-dimensional Monte Carlo simulation [9][10][11] were applied to bioaccumulation models to address the importance of uncertainty and/or variability of model parameters. In these approaches, model parameters are randomly drawn from distributions determined from the available reference literature and/or site-specific data.…”