area and that treatment-induced effect could be distinguished as a deviation from
the historical pattern.
Monitoring a comparable treatment and control site simultaneously is the most
effective design to detect impact. Monitoring a control site provides the data to
separate the impact of treatment from the variability shared by both the treatment
and control. One option is to monitor similar stream stations in paired watersheds-
-one in which there is a management action and the other, without. Likewise, a
survey of treated and reference lakes may show treatment effects. Implementation
can be at the same time or staggered through time to track and account for factors
(e.g., climate) that affect all lakes at once.
Using one or more references can account for system variability (e.g., biological
response, life cycle, population fluctuations, and hydrologic changes), therefore
reducing the time needed to detect improvement, and providingstrongerstatistical
evidence of cause-and-effect. Disadvantages include the difficulty of finding a
suitable reference site, the need for coordinated monitoring in both systems, and
To determine causality, a system of a control site and a treatment site is needed.
Monitoring a control site is necessary to distinguish changes in a variable due to
natural variability from those due to treatment. Mosteller and Tukey (1977)
identify four conditions to show causality or cause-and-effect: association,
consistency, responsiveness, and a mechanism.
Association is shown by demonstrating a relationship between two variables (e.g.,
a correlation between the intensity of management and the apparent reduction in
Consistency can be confirmed by observation only and implies the relationship
does not change in different populations (e.g., management action was imple-
mented in several areas and pollutant loading was reduced, depending upon the
effect of treatment, in each case).
Responsiveness is shown in an experiment when a treatment
there is a corresponding change in a variable.
a plausible step-by-step explanation of how the management
cause the observed change. For example, conservation
reduced the edge-of-field losses of sediment, thereby removing a known fraction
of pollutant from runoff to a stream. The result was decreased suspended sediment
concentration in the water column.
Formulating and testing a hypothesis are central to a meaningful monitoring
program for detecting trends and impacts or showing causality. The hypothesis is
not needed for the objectives of evaluating current conditions or documenting the
water quality problem. The remaining discussion will focus on experimental
design objectives. The experimental design is part of an important framework for
hypothesis testing and the analysis of results.
The hypothesis is based on the monitoring objective and it provides structure to
the design. The null hypothesis (Ho:) is a statement reflecting that no change or