CHAPTER 6 • Coastal Environmental Management in Southeast Australia: The Roles of Ecology 103
6.2.2
Temporal and Spatial Variation
The inevitable variation in space and time (points (ii) and (iii» has important consequences for the sorts of information that managers can get and the sorts of studies
that ecologists and other environmental scientists must do to provide it. To take a relatively simple example, consider the need to detect changes in abundances of fauna on
sandy beaches following a managerial decision to prevent access by recreational vehicles in one popular tourist area. It had been a source of concern that vehicles running
over beaches were causing loss of polychaete worms, various crabs and burrowing
bivalves, because the vehicles compact the sediments and prevent the animals feeding.
Numbers of worms (and the other species) do, however, vary greatly from place to place
and time to time. This is normal.
The prediction made by the managers is part of an entirely scientifically logical
sequence (summarized in Underwood 1990, 1995). It has been observed (or, at least, it
is assumed) that the numbers of worms are smaller where recreational vehicles are
numerous. It has been proposed that the numbers are smaller because of the vehicles.
This model for what is going on led to the decision to remove vehicles, with the prediction or hypothesis that numbers will rise (because the cause of decreased abundances has been removed). Implementation of the new management is, in fact, an experimental test of the hypothesis. The issue remains, however, that someone needs to
verify that management has been effective. This requires the hypothesis to be tested
by appropriate quantitative sampling.
This would be relatively simple if the numbers of worms were relatively constant
from place to place and time to time (i.e. if there were a balance of nature). All that
would be needed is some measurement of numbers before management is implemented
and another measure after. The anticipated increase would either be evident or not,
depending on whether or not the proposed model was correct. Such "before-after" sampIing is often done iIi the belief that there are no natural changes, so that any change
in numbers must be due to change of management.
But numbers of worms will always change naturally, so an increase in the managed
area is not evidence that removal of vehicles has solved the problem. Implementing it
elsewhere will not necessarily lead to increases in fauna. So, first, there must be sufficient temporal sampling to be sure that changes from before to after implementation
of management are more than would normally be encountered under natural processes of change (e.g. Bernstein and Zalinski 1983; Stewart-Oaten et al. 1986). Second,
control areas must also be sampled to demonstrate that the differences in the managed area are not a general change affecting their locations, but that happens to be a
larger increase than generally occurs (e.g. Underwood 1992, 1993a). The appropriate
hypothesis is therefore the one pointed out by Green (1979) in his pioneering book. If
management is effective, there should be a greater change from before to after in the
managed area than occurs during the same period in unmanaged areas (illustrated in
Fig. 6.2). The appropriate statistical interaction should be detected (Green 1979;
Underwood 1993a, 1994a). In this particular case, there is an inevitable asymmetry -
there is only one managed location, but there can be appropriate replication of control, unmanaged locations (see Underwood 1992, 1993a, 1994a for all details of design
and analysis of such sampling).
6.2.2
Temporal and Spatial Variation
The inevitable variation in space and time (points (ii) and (iii» has important consequences for the sorts of information that managers can get and the sorts of studies
that ecologists and other environmental scientists must do to provide it. To take a relatively simple example, consider the need to detect changes in abundances of fauna on
sandy beaches following a managerial decision to prevent access by recreational vehicles in one popular tourist area. It had been a source of concern that vehicles running
over beaches were causing loss of polychaete worms, various crabs and burrowing
bivalves, because the vehicles compact the sediments and prevent the animals feeding.
Numbers of worms (and the other species) do, however, vary greatly from place to place
and time to time. This is normal.
The prediction made by the managers is part of an entirely scientifically logical
sequence (summarized in Underwood 1990, 1995). It has been observed (or, at least, it
is assumed) that the numbers of worms are smaller where recreational vehicles are
numerous. It has been proposed that the numbers are smaller because of the vehicles.
This model for what is going on led to the decision to remove vehicles, with the prediction or hypothesis that numbers will rise (because the cause of decreased abundances has been removed). Implementation of the new management is, in fact, an experimental test of the hypothesis. The issue remains, however, that someone needs to
verify that management has been effective. This requires the hypothesis to be tested
by appropriate quantitative sampling.
This would be relatively simple if the numbers of worms were relatively constant
from place to place and time to time (i.e. if there were a balance of nature). All that
would be needed is some measurement of numbers before management is implemented
and another measure after. The anticipated increase would either be evident or not,
depending on whether or not the proposed model was correct. Such "before-after" sampIing is often done iIi the belief that there are no natural changes, so that any change
in numbers must be due to change of management.
But numbers of worms will always change naturally, so an increase in the managed
area is not evidence that removal of vehicles has solved the problem. Implementing it
elsewhere will not necessarily lead to increases in fauna. So, first, there must be sufficient temporal sampling to be sure that changes from before to after implementation
of management are more than would normally be encountered under natural processes of change (e.g. Bernstein and Zalinski 1983; Stewart-Oaten et al. 1986). Second,
control areas must also be sampled to demonstrate that the differences in the managed area are not a general change affecting their locations, but that happens to be a
larger increase than generally occurs (e.g. Underwood 1992, 1993a). The appropriate
hypothesis is therefore the one pointed out by Green (1979) in his pioneering book. If
management is effective, there should be a greater change from before to after in the
managed area than occurs during the same period in unmanaged areas (illustrated in
Fig. 6.2). The appropriate statistical interaction should be detected (Green 1979;
Underwood 1993a, 1994a). In this particular case, there is an inevitable asymmetry -
there is only one managed location, but there can be appropriate replication of control, unmanaged locations (see Underwood 1992, 1993a, 1994a for all details of design
and analysis of such sampling).