weak empirical relationship between spill volume and bird mortality points out the need to better understand the other sources of uncertainty (such as spill timing).
Assessing recovery after a pollution event is perhaps even more challenging than assessing initial damage. Recovery is further removed in time from the acute phase of the damage, and thus may be occurring in a different environmental framework than that which existed at the time of the accident. If there is variation in time, but the long-term mean remains stable, recovery might be judged by some to have been complete when the environmental variable of concern returns to within the normal range of variation (see Fig 5-1A, Wiens, 1995). In contrast, if the long-term environmental mean is changing, then recovery would occur when the variable of concern returns to within a range of variation around a short-term mean that will be quite different from that when the perturbation occurred (Fig. 5-1B). To assess recovery quantitatively requires either a well designed BACI approach, or one that compares measurements of the environmental variable of interest along a gradient of perturbation (Wiens, 1995). This gradient can be in space or time. One must be certain that, when numbers of organisms are being compared for assessment of recovery, attributes such as age or reproductive potential be taken into account. For example in marine birds, young, inexperienced animals do not have the same value to the population as experienced breeding adults. The natural variability inherent in estimates of populations introduces considerable uncertainty in assessing impact and recovery from pollution events. Confidence limits in excess of 20 percent of the mean size are usual in wildlife censuses. Such variability in the estimated mean makes it certain that population changes will be difficult to detect without a high degree of replication spatially and temporally before and after an event. More importantly, under some circumstances estimates of recovery based on the population returning to a “window” of natural fluctuation could minimize the time to true recovery. Other important considerations in evaluating oil pollution effects are the roles that laboratory studies, mesocosms and impact modeling play in complementing, or, in some cases, replacing the field observations discussed above.
Laboratory studies avoid the aforementioned problem of lack of control, but their improved precision disallows the wide range of possible interactions and indirect effects that can occur in complex ecosystems. Such indirect effects might be substantial. For example, in the Exxon Valdez and Torrey Canyon oil spills, destruction of the algal cover had indirect impacts on limpets and other invertebrates (Southward and Southward, 1978; Peterson, 2001). Such successional, reverberating or cascading indirect effects in a complex ecosystem may be very important, but are not captured by laboratory studies. The bulk of laboratory studies have examined oil impacts on organism mortality and health using dissolved oil or seawater suspensions. Most experiments are conducted for short durations (Capuzzo, 1987), which does not take into account long-term effects.
Field observations and laboratory experiments, as ways of knowing effects, represent two ends of a spectrum. Field observations allow little or no control of interactions between the full complement of ecosystem variables; laboratory experiments allow control of the interaction of single components that have been removed from the ecosystem. Taken together they still may not tell the whole story of oil impact. As a result, efforts have been made to bridge the gap between these two ends of the experimental control-field complexity continuum. Intermediate approaches include: laboratory experiments with multiple species, or communities that include environmental components (micro-and mesocosms); and field experiments, for example that put oiled sediments into the environment to be colonized by natural populations of animals and plants.
The modeling of the impacts of oil spills and their potential effects provides another route for predicting the potential effects of spilled oil. Oil spill impact modeling, which was originally applied to predicting the fate of oil in the environment, has recently been extended to prediction of effects (McCay, 2001).
In this chapter, we provide a brief review of progress in addressing the research recommendations of the 1985 Oil in the Sea report (NRC, 1985). We then examine the acute and chronic effects of oil at the organism, population and community/ecosystem levels. In the review, we single out marine birds and mammals for special attention because of their high visibility in spills and the great public concern for their welfare. It has been our intent to focus on the significant advances in knowledge and perceptions of the effects of oil in the sea, rather than to provide a detailed examination of the many research papers that have been published since the completion of the NRC (1985) or the Boesch and Rabalais (1987) reviews.
Since the major review of oil in the sea conducted by the National Research Council and published in 1985, there have been thousands of individual studies contributing to our overall understanding of the acute and chronic toxicity of oil in the marine environment and the restoration and recovery of oiled habitats. The major recommendations of the 1985 report were:
To expand studies of effects of low concentrations of petroleum hydrocarbons on marine organisms, especially larval and juvenile stages;
To examine the apparent coincidence of petroleum hydrocarbon exposure with increased prevalence of pollution-related disease in marine organisms;
To examine the impacts of petroleum hydrocarbons in polar and tropical habitats;
To better integrate laboratory studies with field investigations;