conducted scavenger-removal experiments to correct estimates for potential biases.
The most important element in designing a study is deciding on the study objective. Once the study objective is determined, other essential issues include the following:
The area of interest,
Time period of interest,
Species of interest,
Potentially confounding variables,
The time and budget available for the required studies, and
The magnitude of the impact being evaluated.
The following is a general discussion of methods, metrics, and study design for achieving objectives commonly addressed in the study of wildlife impacts from wind-energy development. For a more detailed discussion of this topic, readers are referred to Green (1979), Underwood (1994), Anderson et al. (1999), Manly (2001), and Morrison et al. (2001). There is no fundamental difference between monitoring and research, but a commonly used criterion for distinguishing them is the duration of study. Monitoring schemes are essentially repeated surveys (Manly 2001) and are usually designed to detect changes and trends in the variable of interest. Because considerations in study design are essentially the same for both monitoring and observational studies, no effort will be made to further discriminate between the two.
Reliable study designs available for environmental impact assessments are limited. The before-after/control impact (BACI) design is commonly used in observational studies (e.g., Stewart-Oaten 1986) and has been considered the optimal impact-study design by Green (1979). As the name implies, this type of study involves the collection of data in the assessment area and a similar (control) area both before and after an impact occurs (Morrison et al. 2001). An effect typically is measured as a change in the difference between estimates of a variable for the control and an assessment area following an impact. Confidence intervals can increase the reliability of an impact estimate when data from more than one control area are available (Underwood 1994). Ideally, control areas should be randomly selected from a population of similar sites (Manly 2001). Study areas within the assessment and control area may be matched to reduce the natural variation common in impact studies (Skalski and Robson 1992), although