Evaluation of two methods for prediction of bioaccumulation factors

Two methods for deriving bioaccumulation factors (BAFs) used by the U.S. Environmental Protection Agency in development of water quality criteria were evaluated using polychlorinated biphenyl (PCB) data from the Hudson River and Green Bay ecosystems. One method predicts BAF(L)fd values (BAFs based upon concentrations of freely dissolved chemical in ambient water and in the lipid fraction of tissue) using field-measured biota-sediment accumulation factors (BSAFs): greater than 90% of the predicted BAF(L)fd values were within a factor of 5 of their measured values for both ecosystems. The second method predicts BAF(L)fd values as the chemical's 1-octanol/water partition coefficient (Kow) times a food chain multiplier: greater than 90% of the predicted BAF(L)fd values were within a factor of 5 of their measured values for the Green Bay ecosystem and for three of the six sampling locations on the Hudson River. Poorer predictive ability with the Kow method for the other three sampling locations was believed to be due to the existence of environmental conditions not representative of the longer term ecosystem conditions assumed for the method. BAF(L)fd and BAF(T)t values (BAFs based upon concentrations of total chemical in ambient water and in wet tissue) were compared. The within ecosystem and across ecosystems comparisons demonstrated a 2-5-fold decrease in variability (expressed as ratios of coefficients of variation, percentile ranges, and confidence ranges) for predicted BAF(L)fd values in comparison to BAF(T)t values.