A recent monograph by Hurlbert raised several problems concerning the appropriate design of sampling programs to assess the impact upon the abundance of biological populations of, for example, the discharge of effluents into an aquatic ecosystem at a single point. Key to the resolution of these issues is the correct identification of the statistical parameter of interest, which is the mean of the underlying probabilistic "process" that produces the abundance, rather than the actual abundance itself. We describe an appropriate sampling scheme designed to detect the effect of the discharge upon this underlying mean. Although not guaranteed to be universally applicable, the design should meet Hurlbert's objections in many cases. Detection of the effect of the discharge is achieved by testing whether the difference between abundances at a control site and an impact site changes once the discharge begins. This requires taking samples, replicated in time, Before the discharge begins and After it has begun, at both the Control and Impact sites (hence this is called a BACI design). Care needs to be taken in choosing a control site so that it is sufficiently far from the discharge to be largely beyond its influence, yet close enough that it is influenced by the same range of natural phenomena (e.g., weather) that result in long-term changes in the biological populations. The design is not appropriate where local events cause populations at Control and Impact sites to have different long-term trends in abundance; however, these situations can be detected statistically. We discuss the assumptions of BACI, particularly additivity (and transformations to achieve it) and independence.

Because studies of environmental accidents must be initiated after the fact and because the accidents cannot (or should not) be replicated, sampling cannot be entirely randomized, and investigations involve some degree of confounding and pseudoreplication. The study designs that can be used carry methodological limitations and ecological assumptions, which must be considered in evaluating results. The methodological issues relate to consistency in sampling methods and to the adequacy of sampling of levels of environmental disturbance or contamination, while the ecological assumptions derive from spatial and temporal variation in biological resources and the factors that affect them. We assess how these methodological issues and ecological assumptions affect study designs based on before–after comparisons and on single‐time or multiple‐time sampling after an accident. Designs that rest on the assumption of a steady‐state equilibrium in resource‐environment relationships (baseline and time‐series designs) must be interpreted with particular care, and baseline designs are sensitive to the effects of pseudoreplication and inconsistencies in methods. Other designs (pre/post paired samples, impact level‐by‐time, impact trend‐by‐time) assume only that environmental variations are equivalent among areas and/or contamination levels (a dynamic equilibrium) and are less affected by pseudoreplication. Single‐time designs (comparisons between impact and reference sites, between matched paired sites, or over a contamination gradient) have fewer methodological limitations, but assume that other natural factors that may influence the response of a resource are equal among all samples. If measurements of other factors are included in the design, covariance analysis may help to reduce this problem. In evaluating the effects of unplanned environmental impacts, post facto study designs that document both initial effects and subsequent recovery (impact level‐by‐time, impact trend‐by‐time) or that treat effects as continuous rather than categorical variables (gradient or trend designs) may be more useful than before–after comparisons.

Gauging whether or when a population, species, or community recovers from an environmental accident or disturbance, such as an oil spill or forest fire, is complicated by environmental variation in time and space, and therefore depends on the assumptions one makes about equilibrium. These ecological assumptions about equilibrium affect how one designs and interprets studies to assess recovery from environmental accidents or disturbances. We use examples from studies conducted following the Exxon Valdez oil spill to illustrate several approaches to assessing recovery and their sensitivity to the form of equilibrium one assumes. Baseline study designs, which compare levels of a resource after the disturbance to pre‐disturbance levels for impact data only, are generally inadequate because they rest on the unrealistic assumption of steady‐state equilibrium. Since data for the impact area only are used, recovery and temporal variation are confounded. Unlike baseline designs, before–after control–impact (BACI) designs use impact and reference data, and relax this sensitivity by incorporating both temporal and spatial variation. Studies that compare impacted with reference areas in a single year following the disturbance assume spatial equilibrium and therefore may confound recovery with systematic spatial differences between the areas. Sampling and analytical strategies such as stratified random sampling or the use of environmental measures as covariates may lessen the sensitivity to this assumption. Multiyear studies that include comparisons between impacted and reference areas or that sample areas along a gradient of disturbance rest on the more realistic assumption of dynamic equilibrium. Understanding the underlying assumptions and how they relate to the approach one uses must be part of assessing the recovery of biological resources from an environmental accident. Because the dynamics of different populations, species, and communities and the environments they occupy vary and exhibit different dependencies on the scale of disturbance (and the scale of analysis), there is no single “best” approach to assessing recovery. Discussions about recovery should include an explicit and honest consideration of the underlying ecological assumptions, the likelihood that they hold in the system being studied, and the consequences if the assumptions are violated.

No abstract

Previous research in the Milne Point oilfield in northern Alaska showed that the density of caribou (Rangifer tarandus) in the calving period within 1 km of a road was significantly lower after construction than before construction of the road. This was interpreted as displacement from the road and a functional loss of habitat and has been extensively cited as a documented effect of oilfield development on caribou. We continued this study with additional aerial surveys flown during 1991–2001 and compared caribou numbers and density in 6 1‐km intervals from the road over 3 time periods: pre‐road construction (1978–1981), early post‐road construction (1982–1987), and recent post‐road construction (1991–2001). During the recent post‐road period, the densities of calves and of all caribou were not significantly lower within 1 km of the road than the densities in the pre‐road period. In addition, calf density was higher in the interval within 1 km of the road than in the intervals 1–6 km from the road during the recent post‐road construction period. The total number of calving caribou observed in the study area has declined since pre‐road construction and early post‐road periods, but we found no evidence that caribou using the area during calving avoided areas near the road during the recent post‐road period. Numbers of caribou in the study area during the post‐calving period (after 20 June) during 1991–2001 were highly variable, but generally were higher than during calving. Analyses of relationships between calving and post‐calving caribou densities and distance intervals from Milne Point Road suggested that distributions of calves and adult caribou were not strongly influenced by presence of the road.