Quantitative risk model for polycyclic aromatic hydrocarbon photoinduced toxicity in Pacific herring following the Exxon Valdez oil spill.

Phototoxicity occurs when exposure to ultraviolet radiation increases the toxicity of certain contaminants, including polycyclic aromatic hydrocarbons (PAHs). This study aimed to (1) develop a quantitative model to predict the risk of PAH phototoxicity to fish, (2) assess the predictive value of the model, and (3) estimate the risk of PAH phototoxicity to larval and young of year Pacific herring (Clupea pallasi) following the Exxon Valdez oil spill (EVOS) in Prince William Sound, Alaska. The model, in which median lethal times (LT50 values) are estimated from whole-body phototoxic PAH concentrations and ultraviolet A (UVA) exposure, was constructed from previously reported PAH phototoxicity data. The predictive value of this model was confirmed by the overlap of model-predicted and experimentally derived LT50 values. The model, along with UVA characterization data, was used to generate estimates for depths of de minimiz risk for PAH phototoxicity in young herring in 2003/2004 and immediately following the 1989 EVOS, assuming average and worst case conditions. Depths of de minimiz risk were estimated to be between 0 and 2 m deep when worst case UVA and PAH conditions were considered. A post hoc assessment determined that <1% of the young herring population would have been present at depths associated with significant risk of PAH phototoxicity in 2003/2004 and 1989.

Incorporating covariates into fisheries stock assessment models with application to Pacific herring.

We present a framework for evaluating the cause of fishery declines by integrating covariates into a fisheries stock assessment model. This allows the evaluation of fisheries' effects vs. natural and other human impacts. The analyses presented are based on integrating ecological science and statistics and form the basis for environmental decision-making advice. Hypothesis tests are described to rank hypotheses and determine the size of a multiple covariate model. We extend recent developments in integrated analysis and use novel methods to produce effect size estimates that are relevant to policy makers and include estimates of uncertainty. Results can be directly applied to evaluate trade-offs among alternative management decisions. The methods and results are also broadly applicable outside fisheries stock assessment. We show that multiple factors influence populations and that analysis of factors in isolation can be misleading. We illustrate the framework by applying it to Pacific herring of Prince William Sound, Alaska (USA). The Pacific herring stock that spawns in Prince William Sound is a stock that has collapsed, but there are several competing or alternative hypotheses to account for the initial collapse and subsequent lack of recovery. Factors failing the initial screening tests for statistical significance included indicators of the 1989 Exxon Valdez oil spill, coho salmon predation, sea lion predation, Pacific Decadal Oscillation, Northern Oscillation Index, and effects of containment in the herring egg-on-kelp pound fishery. The overall results indicate that the most statistically significant factors related to the lack of recovery of the herring stock involve competition or predation by juvenile hatchery pink salmon on herring juveniles. Secondary factors identified in the analysis were poor nutrition in the winter, ocean (Gulf of Alaska) temperature in the winter, the viral hemorrhagic septicemia virus, and the pathogen Ichthyophonus hoferi. The implication of this result to fisheries management in Prince William Sound is that it may well be difficult to simultaneously increase the production of pink salmon and maintain a viable Pacific herring fishery. The impact can be extended to other commercially important fisheries, and a whole ecosystem approach may be needed to evaluate the costs and benefits of salmon hatcheries.