monitoring. Experience has shown that potential impacts must be considered at different spatial and temporal scales and different levels of
biological organization. Not all scales and levels need to be analyzed in
detail, depending on the severity of estimated impacts. Results of analyses
at each of these scales need to be considered in a weight-of-evidence
approach.
2. We recommend that estimated values for any given measure include
consideration of the normal variability in that measure (Ambrose et al.
1996; Coutant 2000). As in any risk assessment, scientific certitude is an
illusion; thus, a precautionary approach is appropriate (Hilborn et al. 2001;
Schnute and Richards 2001). This precautionary approach allows establishment of more conservative levels of acceptable loss when faced with higher
uncertainty.
3. We recommend that 316(b) decisions require continued monitoring
of selected measures with reassessments at each repermitting interval
(Ambrose et al. 1996; Coutant 2000). This approach of adaptive resource
management is particularly appropriate when uncertainty is high, fishery
resources at risk are highly valued, and other negative or positive changes
(e.g., water quality or regional temperature regime) are occurring in the
system. The frequency of this monitoring and reassessment might be
relaxed in the future if additional data continued to support the conclusions
of the original scientific assessment of no AEI.
4. We recommend that analyses and model applications focus on
relative risk (USEPA 1998) by comparing estimates of short-term impacts
(e.g., 5 to 10 years, as opposed to 50 to 100 years) of alternative management actions and decisions and not on absolute or long-term impacts
(Barnthouse et al. 1984; Van Winkle 2000). Emphasizing model predictions
of risk of percent reduction or quasi-extinction over the lifetime of a power
plant alone is not likely to be accepted. Nor are results of meta-analyses of
time series of data on spawning stock and subsequent recruitment alone
likely to be accepted (e.g., spawner-recruit curves and associated indices).
Both scientists and nonscientists have valid reasons and past experience to
be skeptical about such model predictions and analyses claiming to provide
an adequate basis for making site-specific decisions concerning long-term,
population consequences (Boreman 1997; Hilborn et al. 2001; Hutchings
2001; Rose 2000; Rose and Cowan 2000; Schnute and Richards 2001; Van
Winkle 2000). Scientists need to be aware when regulators or stakeholders
are expecting (or interpreting) more certain conclusions than science can
provide.
5. We recommend that, if a population declines over a period of years
during which losses from entrainment and impingement cannot be judged
as trivial, the responsible scientific conclusion is that to some unknowable
extent these losses may have contributed to the decline. Because of confounding changes in physical, biotic, and anthropogenic variables during the
same period, it will never be possible to prove that losses from entrainment
64
Webb Van Winkle and John Kadvany
biological organization. Not all scales and levels need to be analyzed in
detail, depending on the severity of estimated impacts. Results of analyses
at each of these scales need to be considered in a weight-of-evidence
approach.
2. We recommend that estimated values for any given measure include
consideration of the normal variability in that measure (Ambrose et al.
1996; Coutant 2000). As in any risk assessment, scientific certitude is an
illusion; thus, a precautionary approach is appropriate (Hilborn et al. 2001;
Schnute and Richards 2001). This precautionary approach allows establishment of more conservative levels of acceptable loss when faced with higher
uncertainty.
3. We recommend that 316(b) decisions require continued monitoring
of selected measures with reassessments at each repermitting interval
(Ambrose et al. 1996; Coutant 2000). This approach of adaptive resource
management is particularly appropriate when uncertainty is high, fishery
resources at risk are highly valued, and other negative or positive changes
(e.g., water quality or regional temperature regime) are occurring in the
system. The frequency of this monitoring and reassessment might be
relaxed in the future if additional data continued to support the conclusions
of the original scientific assessment of no AEI.
4. We recommend that analyses and model applications focus on
relative risk (USEPA 1998) by comparing estimates of short-term impacts
(e.g., 5 to 10 years, as opposed to 50 to 100 years) of alternative management actions and decisions and not on absolute or long-term impacts
(Barnthouse et al. 1984; Van Winkle 2000). Emphasizing model predictions
of risk of percent reduction or quasi-extinction over the lifetime of a power
plant alone is not likely to be accepted. Nor are results of meta-analyses of
time series of data on spawning stock and subsequent recruitment alone
likely to be accepted (e.g., spawner-recruit curves and associated indices).
Both scientists and nonscientists have valid reasons and past experience to
be skeptical about such model predictions and analyses claiming to provide
an adequate basis for making site-specific decisions concerning long-term,
population consequences (Boreman 1997; Hilborn et al. 2001; Hutchings
2001; Rose 2000; Rose and Cowan 2000; Schnute and Richards 2001; Van
Winkle 2000). Scientists need to be aware when regulators or stakeholders
are expecting (or interpreting) more certain conclusions than science can
provide.
5. We recommend that, if a population declines over a period of years
during which losses from entrainment and impingement cannot be judged
as trivial, the responsible scientific conclusion is that to some unknowable
extent these losses may have contributed to the decline. Because of confounding changes in physical, biotic, and anthropogenic variables during the
same period, it will never be possible to prove that losses from entrainment
64
Webb Van Winkle and John Kadvany