implemented in cost-effective and statistically sound
ways to achieve the supposed goals.
Thus, if the potential disturbance to some area of an
estuary is the impact due to hydrocarbon leaks from
marinas, the types and amounts of contaminating hydrocarbons can be specified, as can their locations and probable quantities. Designing sampling to detect such releases
and any ecological consequences is then relatively
straightforward.
If, in contrast, sampling is supposed to detect changes
in some poorly defined concept (e.g., “ecosystem health”),
no relevant variables are defined and no amount of anticipated deterioration in ecological variables can be identified. It is then impossible to determine what, where, and
when to sample, let alone how to do so effectively. Nevertheless, such “monitoring” is commonly done.
Types of changes
Different types of study are required for different types of
disturbances. For example, if the problems are due to
chemical contamination, it is quite straightforward to design
sampling to detect the background levels of chemicals.
Provided the amounts of change to be detected are defined
in advance, it is not difficult in principle to define sampling
that is appropriate to identify such changes.
It is, however, less clear that measuring chemicals is
a good way to identify impacts. Contamination is the presence of unnatural chemicals in the environment. Pollution
is some biological or ecological response to the contamination. It is common for the concentrations of chemicals to
increase without any particular response by organisms. It
is equally possible for there to be no change in concentrations of chemicals, but for there to be a large-scale impact.
Consider the concentrations of nitrogen in estuarine
water. Increased input of nitrogen can lead to rapid and
excessive growth of algae in the plankton or sediments.
Algae take up the extra nitrogen as they grow, resulting
in no change in the concentration of nitrogen in the water
column. Thus, monitoring for impacts on algae by measuring nitrogen is useless; the measure will not change regardless of the amount of impact of increased nitrogen. In this
case, it would, instead, be necessary to monitor the algae
themselves. Using indicators of required measures is often
not sensibly planned.
Impacts occur in many forms. The most common are
due to press as opposed to pulse disturbances (Bender
et al., 1984). Press disturbances are those which last
a long time. For example, building a marina will cause
long-term changes to water flow, shading, etc. A pulse disturbance is a relatively short-term problem that then ends.
For example, short-term dredging can cause increased
sedimentation in surrounding areas, but this will stop
when the dredging ceases.
An impact in response to a press disturbance can be
a press impact, where the monitored variables change to a
new average value and then show any natural temporal
fluctuation around that value. Alternatively, a press
disturbance can cause a pulse impact, where the monitored
variables fluctuate in response to the disturbance. Pulse
disturbances can also cause pulse or press impacts.
A press impact would occur when the system being affected
continues to be affected even though the original
disturbance has ceased, for example, because a short-term
disturbance causes long-term changes to a habitat. Identifying and measuring press as opposed to pulse responses
to disturbances require different types of sampling
(Underwood, 1994).
Spatial and temporal issues in monitoring
Spatial replication
Spatial replication is necessary in any sampling because of
large variability in all ecological variables. Suppose
amphipods are sampled in one area where dredging has
disturbed the sediments and in one similar area where sediments have not been disturbed. Any difference found may
be due to the disturbance or to any ecological processes
affecting the numbers of amphipods differently in the
two areas (Hurlbert, 1984). Any conclusion about an
impact is confounded because the difference may also be
due to natural variation.
It is necessary to sample several dredged and several
undisturbed areas so that natural variability that is not
due to dredging can be measured under disturbed and control conditions. An overall difference between the dredged
and undisturbed areas that is larger than natural variation
among the areas of each type would then be an unambiguous evidence for an impact.
Spatial scales of sampling
Sampling at only one spatial scale can, however, cause
problems. Suppose that several places in an estuary have
been dredged, but the dredged sediments are contaminated
by metals which may affect benthic infauna in
areas around the dredged sites. Sediments are thought to
carry metals up to 100 m away. Sampling is done
in several dredged and several controls. In each area, the
infauna are counted in replicate cores of sediment. Natural
variation from replicate to replicate and area to area
can then be estimated, and analysis could reveal any
systematic difference between dredged and control areas.
Suppose, however, that contaminated sediments actually get dispersed over the entire area of the estuary, not
just to sites within 100 m of dredged sites. Any impacts
will now cause changes in benthic infauna in all of the
sites sampled. There can be no apparent impact, because
the controls are affected in the same way as are sites close
to dredged areas. The interpretation would now be that
some estuary-wide change had occurred to the infauna,
which was not due to the dredging.
Other estuaries where dredging was not being done
should also be sampled as controls. Uncertainty about the
scale of potential impacts requires sampling at several
scales. Analysis of data in such situations has been
described in detail by Green (1979) and Underwood (1994).
224
ECOLOGICAL MONITORING
ways to achieve the supposed goals.
Thus, if the potential disturbance to some area of an
estuary is the impact due to hydrocarbon leaks from
marinas, the types and amounts of contaminating hydrocarbons can be specified, as can their locations and probable quantities. Designing sampling to detect such releases
and any ecological consequences is then relatively
straightforward.
If, in contrast, sampling is supposed to detect changes
in some poorly defined concept (e.g., “ecosystem health”),
no relevant variables are defined and no amount of anticipated deterioration in ecological variables can be identified. It is then impossible to determine what, where, and
when to sample, let alone how to do so effectively. Nevertheless, such “monitoring” is commonly done.
Types of changes
Different types of study are required for different types of
disturbances. For example, if the problems are due to
chemical contamination, it is quite straightforward to design
sampling to detect the background levels of chemicals.
Provided the amounts of change to be detected are defined
in advance, it is not difficult in principle to define sampling
that is appropriate to identify such changes.
It is, however, less clear that measuring chemicals is
a good way to identify impacts. Contamination is the presence of unnatural chemicals in the environment. Pollution
is some biological or ecological response to the contamination. It is common for the concentrations of chemicals to
increase without any particular response by organisms. It
is equally possible for there to be no change in concentrations of chemicals, but for there to be a large-scale impact.
Consider the concentrations of nitrogen in estuarine
water. Increased input of nitrogen can lead to rapid and
excessive growth of algae in the plankton or sediments.
Algae take up the extra nitrogen as they grow, resulting
in no change in the concentration of nitrogen in the water
column. Thus, monitoring for impacts on algae by measuring nitrogen is useless; the measure will not change regardless of the amount of impact of increased nitrogen. In this
case, it would, instead, be necessary to monitor the algae
themselves. Using indicators of required measures is often
not sensibly planned.
Impacts occur in many forms. The most common are
due to press as opposed to pulse disturbances (Bender
et al., 1984). Press disturbances are those which last
a long time. For example, building a marina will cause
long-term changes to water flow, shading, etc. A pulse disturbance is a relatively short-term problem that then ends.
For example, short-term dredging can cause increased
sedimentation in surrounding areas, but this will stop
when the dredging ceases.
An impact in response to a press disturbance can be
a press impact, where the monitored variables change to a
new average value and then show any natural temporal
fluctuation around that value. Alternatively, a press
disturbance can cause a pulse impact, where the monitored
variables fluctuate in response to the disturbance. Pulse
disturbances can also cause pulse or press impacts.
A press impact would occur when the system being affected
continues to be affected even though the original
disturbance has ceased, for example, because a short-term
disturbance causes long-term changes to a habitat. Identifying and measuring press as opposed to pulse responses
to disturbances require different types of sampling
(Underwood, 1994).
Spatial and temporal issues in monitoring
Spatial replication
Spatial replication is necessary in any sampling because of
large variability in all ecological variables. Suppose
amphipods are sampled in one area where dredging has
disturbed the sediments and in one similar area where sediments have not been disturbed. Any difference found may
be due to the disturbance or to any ecological processes
affecting the numbers of amphipods differently in the
two areas (Hurlbert, 1984). Any conclusion about an
impact is confounded because the difference may also be
due to natural variation.
It is necessary to sample several dredged and several
undisturbed areas so that natural variability that is not
due to dredging can be measured under disturbed and control conditions. An overall difference between the dredged
and undisturbed areas that is larger than natural variation
among the areas of each type would then be an unambiguous evidence for an impact.
Spatial scales of sampling
Sampling at only one spatial scale can, however, cause
problems. Suppose that several places in an estuary have
been dredged, but the dredged sediments are contaminated
by metals which may affect benthic infauna in
areas around the dredged sites. Sediments are thought to
carry metals up to 100 m away. Sampling is done
in several dredged and several controls. In each area, the
infauna are counted in replicate cores of sediment. Natural
variation from replicate to replicate and area to area
can then be estimated, and analysis could reveal any
systematic difference between dredged and control areas.
Suppose, however, that contaminated sediments actually get dispersed over the entire area of the estuary, not
just to sites within 100 m of dredged sites. Any impacts
will now cause changes in benthic infauna in all of the
sites sampled. There can be no apparent impact, because
the controls are affected in the same way as are sites close
to dredged areas. The interpretation would now be that
some estuary-wide change had occurred to the infauna,
which was not due to the dredging.
Other estuaries where dredging was not being done
should also be sampled as controls. Uncertainty about the
scale of potential impacts requires sampling at several
scales. Analysis of data in such situations has been
described in detail by Green (1979) and Underwood (1994).
224
ECOLOGICAL MONITORING