Coordinator, Narmada Dam Project, World Wildlife Fund,
Professor of Ecotoxicology, Cornell University, Ithaca, NY
Pages: 21 Published: Jan 1993
Prediction of PCB-residue distributions in biota near hazardous-waste sites and other polluted ecosystems is presently limited and may ignore food-web relationships, relies on theoretical considerations not readily extended to real systems, or fails to represent temporal/spatial variation in residues or both. A model has been developed to predict PCB exposure of different parts of an aquatic/terrestrial food web of the Erie/Ontario Lake Plain biogeographic region and its output compared to data from the GM-Akwesasne Superfund site on the St. Lawrence River.
A one-compartment pharmacokinetic model of individual species, taxa, or guilds was linked to a mass-based food web to estimate the residues over time for each element therein. Ranges of pharmacokinetic parameters (or suitable surrogates) and feeding rates from the literature or calculated from published empirical and theoretical relationships for over 300 species of vertebrates and invertebrates of the biogeographic region were used to characterize the uncertainty of estimates.
As a primary simplification, plants and invertebrates were assigned to 14 food item categories linked to abiotic PCB concentrations or to 10 guilds feeding upon these items. Vertebrates (269 species) were clustered into 110 guilds using a subroutine based on dietary overlap, habitat, feeding rate and seasonal presence. Separate summer and winter food webs were thus established for four sets of linked guilds with uncertainty defined by ranges of parameters of species included therein. Monte Carlo simulation of monthly distributions of residues in a food item were used as input to consumers. Model outputs were means and standard deviations of lognormal distributions of simulated PCB residues within guilds over a 2-year period.
The model was validated against site-specific data by using lognormally distributed abiotic PCB residues as initial values. The 95% confidence intervals of the outputs covered 89% and 100%, respectively, of measured values in 40 and 36 of 45 spp. at the site. Exceptions were largely migratory or long-lived species or samples represented by a single item.
Simulations for human beings (for example, Mohawk Indians) subsisting off this food web inferred that they are expected to develop excessive residues. Simulations of mink (largely extirpated from this biogeographic region) indicate that if 50 to 100% of their food supply comes from the system, >20% would carry lethal residues and the majority would have residues associated with reproductive failure.
foodweb model, PCBs, landscape
Paper ID: STP19242S