Risks of toxic contaminants to exploited fish populations: Influence of life history, data uncertainty and exploitation intensity

Abstract

We investigated three aspects of the use of toxicity test data for population-level risk assessment: (a) the influence of life history characteristics on vulnerability to contaminant-induced stress, (b) the importance of test data availability and (c) the influence of exploitation intensity. We quantified population-level effects of chronic contaminant exposure by coupling standard toxicity test data to matrix-type population models derived from long-term field studies of the Gulf of Mexico menhaden (Brevoortiu patronus) and Chesapeake Bay striped bass (Morone suxutilis) populations. We used statistical regressions to quantify the uncertainty inherent in using test data ranging from life cycle tests to quantitative structure-activity relationships (QSARs) to estimate effects of contaminants on the survival and reproduction parameters of the population models.We found that because of differences in life history, menhaden and striped bass differ in terms of their capacity to sustain the same level of contaminant-induced mortality. Changes in exploitation intensity affect the responses of both populations to the same level of additional contaminantinduced mortality. However, the quantitative effects of both factors are negligible compared to the : uncertainty introduced by estimating long-term effects from short-term tests or QSARs. Our results suggest that consideration of life history may be important primarily for site-specific assessments. For screening-level assessments, the substantial differences in uncertainty associated with different types of test data are of much greater concern.

Military Ecological Risk Assessment Framework (MERAF) for Assessment of Risks of Military Training and Testing to Natural Resources

“…Demographic population models developed for this purpose have been adapted to estimate risks from contaminants (Barnthouse et al 1987;Barnthouse et al 1990). …”

“…For example, if the mechanisms of loss of vegetation due to vehicle passage are modeled (e.g., plant crushing, soil compaction, and soil churning), then measures of plant loss at one site can be used to estimate plant loss at other sites. In ecological risk assessment for chemicals, mechanistic models have been used to extrapolate toxic effects from individual fish to fish populations, and from various aquatic species to lake ecosystems (Barnthouse et al 1990;Bartell et al 1992 ).…”

“…If other modes of action, such as reduced fecundity or increased emigration, were potentially significant, the resulting mortality would be used with those modes in a demographic model to estimate population-level effects. For an example, refer to the combination of toxic effects on survival and fecundity of fish with the effects of harvesting on survival (Barnthouse et al 1990). …”

“…However, interactions among the exposure and response processes generally require a deeper level of mechanistic description. For example, combined effects of contaminants and harvesting on fish populations have been modeled using relatively simple demographic models by assuming that the contaminants only reduce survivorship and fecundity, and harvesting only effects survivorship (Barnthouse et al 1990). However, if one wishes to estimate the effects of compensatory processes, such as reduced interference among survivors and increased food supply, one must incorporate mechanisms that affect individual fish (Rose et al 1993).…”

Guidance for treatment of variability and uncertainty in ecological risk assessments of contaminated sites

“…The results can be generalized by aggregating pairs of congeneric species, species in different genera within a family, species in different families within an order, and species in different orders within a class (Suter et al 1983;Barnthouse et al 1990;Suter 1993). The prediction interval (PI) is the appropriate measure of the variance in these relationships (Sloof et al 1986;Barnthouse et al 1990;Suter 1993). Calabrese and Baldwin (1 994) adopted these results and generated more conservative factors by calculating the 95% tolerance interval on the 95% prediction interval.…”