Environmental Engineering Reference
In-Depth Information
precipitation patterns (e.g., timing, intensity, and amount) may not persist as they
are replaced by species better adapted to the changing precipitation patterns. From a
wetland perspective, this is, as yet, an understudied potential consequence of global
climate change. Recognition of a human-defined temporal scale relative to distur-
bance makes it difficult to fully understand the role of natural disturbance in
wetlands beyond a few decades.
Because BACI studies are considered pseudoreplicated due to lack of true
replication, the inferential scope of these studies is limited to those wetlands studied
and not to a larger target population. Such designs are very common in wetland
studies to assess effects of proposed anthropogenic activities, especially when
mitigation is involved. The basic approach to a BACI-type study design is the
collection of a sample prior to the disturbance and another taken after the distur-
bance at both the disturbed site and representative “undisturbed” control site.
A measurable effect due to the disturbance would be represented as a statistical
difference in the average value of the dependent variable between the control and
disturbed sites prior to and after the disturbance. A wetland that is proposed to be
impacted by some activity (e.g., dredging, filling, change in hydrology) has not
been chosen randomly, but the impacts of the activity on wetland function must be
known to mitigate any negative effects of the action. A BACI study can be used to
quantify these impacts. Following identification of the impact site, a particular
wetland that is a geographic neighbor and similar enough to the impact site such
that both wetlands would be subjected to the same nuisance variables is subjectively
identified as the control site (i.e., reference site not experiencing the impact of
interest). None of the experimental units were randomly chosen nor were treatments
randomly assigned; the goal of this study design is to make inference for only the
impacted wetland by measuring the effect size of the impact. Data are collected
simultaneously in both wetlands in the same manner prior to the impact and after
the impact. Statistical analyses and resulting inference are on the comparison of the
magnitude of the differences in recorded data between impact and control sites
prior to and after the impact (Stewart-Oaten et al.
1986
). Underwood (
1991
,
1994
)
details a variety of approaches for statistical analyses of impact studies.
Assessment of impacts not defined prior to impact occurring can be studied using
an impact assessment approach, but the lack of pre-impact data results in weaker
inference. Unpredicted environmental events such as hurricanes, wildfire, floods,
drought, and wind blowdowns are frequently studied for impacts to wetlands
without any pre-impact impact data. Skalski and Robson (
1992
) described an
approach to these types of studies defined as an accident assessment study. They
suggested using a control site and creating a time series of measurements for the
control and impacted site. Inference regarding impact would be based on
comparisons of trajectories of the data as the impacted site recovers. In situations
where a reference area is not available, an impact gradient study may be appropriate
(Skalski and Robson
1992
). This approach assumes that impacts related to a
disturbance are greatest at the core of the disturbance and declines as one moves
away from the core. Therefore, sampling is conducted on the spatial linear distance
