CHAPTER 6 • Coastal Environmental Management in Southeast Australia: The Roles of Ecology
113
Thom 1996). Such areas are called reference areas. Successful restoration predicts that the
variables used to measure structure or function will initially differ between the damaged
habitats and the reference areas, will change after the start of restoration, will converge on
average conditions in the reference areas and will thereafter show similar time-courses
(Fig. 6.4a). Therefore, to evaluate the success of restoration, the environmental variables of interest (diversity, productivity, etc.) must be compared between the restored
habitat and a number of so-called reference areas (Chapman and Underwood 1997a).
Although such an experimental design is appropriate for measuring environmental impacts and other disturbances (e.g. Underwood 1992), it is, however, not adequate
for measuring the success of restoration. Although a change in the environmental variable coincident with the start of restoration of the habitat is essential to successful restoration, it is not in itself sufficient to demonstrate that the managed restoration is the
cause. J;Iabitats change naturally without intervention. To conclude without doubt that
the changes were a response to restoration, it is necessary to compare the outcome with
similarly damaged areas which are not managed (i.e. control locations). It is therefore
predicted that changes in those control pieces of habitat will not mirror those in the
restored habitats (Fig. 6.4b)
This methodology has led to concerns about the quality of data used to assess success of restoration. Typically, inadequate or shoddy sampling (or measurement) will
give very imprecise estimates, with large standard errors and confidence intervals, associated with mean values. The prediction that restored habitats change to become
similar to reference habitats will be supported if standard errors around the mean
values in the different habitats are large, even when the mean values are very different
and restoration has not been achieved (McDonald and Erickson 1994; Fig. 6.5a).
Under the same conditions, precise sampling (with associated small standard errors) would lead to the conclusion that restoration has not occurred (Fig. 6.5b). Therefore, imprecise data which are least likely to be able to measure restoration are also
more likely to cause one to conclude that restoration has been successful! This has led
to the concept that one must, a priori, state the desired measurement that would signify successful restoration. For example, one might decide that restoration would be
considered successful if the confidence limits in the restored habitat fall within ±100/0
of the confidence intervals around average conditions the reference locations (Fig. 6.5c).
Small errors and confidence limits in the restored location (and, hence, more precise
sampling) will be more likely to indicate successful restoration. Unfortunately, such
quantitative predictions about the end-points and goals of management are not yet
commonly made in restoration projects.
Finally, one needs to consider problems associated with structure and function ofhabitats. Many managerial strategies involve changes to the structure of the restored habitat (planting trees, removing exotic species, changing flow of water). Restoring structure is relatively easy, as is - within the limitations described above - measuring changes
to structure. Functional aspects of damaged habitats are, however, more important to
restore if restoration is to be sustainable. Function includes productivity, capacity for
self-generation, maintenance of biodiversity, networks of interactions, food-webs, etc.
Even when appropriate structure and function are each stated to be goals of restoration, structure is often the only means by which restoration is assessed (reviewed by
Kentula et al. 1992). It has been assumed that if an area looks right, it will function
adequately. A few recent studies have shown that constructed and natural wetlands are
113
Thom 1996). Such areas are called reference areas. Successful restoration predicts that the
variables used to measure structure or function will initially differ between the damaged
habitats and the reference areas, will change after the start of restoration, will converge on
average conditions in the reference areas and will thereafter show similar time-courses
(Fig. 6.4a). Therefore, to evaluate the success of restoration, the environmental variables of interest (diversity, productivity, etc.) must be compared between the restored
habitat and a number of so-called reference areas (Chapman and Underwood 1997a).
Although such an experimental design is appropriate for measuring environmental impacts and other disturbances (e.g. Underwood 1992), it is, however, not adequate
for measuring the success of restoration. Although a change in the environmental variable coincident with the start of restoration of the habitat is essential to successful restoration, it is not in itself sufficient to demonstrate that the managed restoration is the
cause. J;Iabitats change naturally without intervention. To conclude without doubt that
the changes were a response to restoration, it is necessary to compare the outcome with
similarly damaged areas which are not managed (i.e. control locations). It is therefore
predicted that changes in those control pieces of habitat will not mirror those in the
restored habitats (Fig. 6.4b)
This methodology has led to concerns about the quality of data used to assess success of restoration. Typically, inadequate or shoddy sampling (or measurement) will
give very imprecise estimates, with large standard errors and confidence intervals, associated with mean values. The prediction that restored habitats change to become
similar to reference habitats will be supported if standard errors around the mean
values in the different habitats are large, even when the mean values are very different
and restoration has not been achieved (McDonald and Erickson 1994; Fig. 6.5a).
Under the same conditions, precise sampling (with associated small standard errors) would lead to the conclusion that restoration has not occurred (Fig. 6.5b). Therefore, imprecise data which are least likely to be able to measure restoration are also
more likely to cause one to conclude that restoration has been successful! This has led
to the concept that one must, a priori, state the desired measurement that would signify successful restoration. For example, one might decide that restoration would be
considered successful if the confidence limits in the restored habitat fall within ±100/0
of the confidence intervals around average conditions the reference locations (Fig. 6.5c).
Small errors and confidence limits in the restored location (and, hence, more precise
sampling) will be more likely to indicate successful restoration. Unfortunately, such
quantitative predictions about the end-points and goals of management are not yet
commonly made in restoration projects.
Finally, one needs to consider problems associated with structure and function ofhabitats. Many managerial strategies involve changes to the structure of the restored habitat (planting trees, removing exotic species, changing flow of water). Restoring structure is relatively easy, as is - within the limitations described above - measuring changes
to structure. Functional aspects of damaged habitats are, however, more important to
restore if restoration is to be sustainable. Function includes productivity, capacity for
self-generation, maintenance of biodiversity, networks of interactions, food-webs, etc.
Even when appropriate structure and function are each stated to be goals of restoration, structure is often the only means by which restoration is assessed (reviewed by
Kentula et al. 1992). It has been assumed that if an area looks right, it will function
adequately. A few recent studies have shown that constructed and natural wetlands are