On the maturing of restoration: Linking ecological research and restoration


  • P. S. Lake

  • P. S. (Sam) Lake holds a Personal Chair in Ecology in the Department of Biological Sciences and the Cooperative Research Centre for Freshwater Ecology, Monash University (PO Box 18, Clayton, Victoria 3800.

    Email: sam.lake@sci.monash.edu.au).

In Australasia and throughout the world, there is now a rapidly growing drive to restore terrestrial and freshwater environments. Restoration is the process of inducing and assisting abiotic and biotic components of an environment to recover to the state that they existed in the unimpaired or original state (Bradshaw 1997). The original state may mean the state prior to human-induced damage, but in many cases knowledge of such a state is simply not available and return to that state is impossible. The restoration effort may range from restoring populations of a particular species to restoring an entire ecosystem. The goal or target of the restoration effort may be set by the presence of undamaged reference areas, or by reliable historical data, or by the compilation from many fragmentary pieces of evidence of an idealized state or scenario.

Restoration differs from rehabilitation in that the latter seeks to improve the condition of a selected area, but not necessarily in the direction of the pre-existing undamaged state (Bradshaw 1997). Both activities may be carried out either passively (where the degrading forces are abated so that natural recovery processes then drive the restoration) or actively (where not only are the degrading forces abated or stopped but the course of restoration is, to a greater or lesser extent, driven by interventions such as reinstatement of dynamic processes, removal of exotics or reintroduction of species).

In most cases restoration and rehabilitation projects have ecological goals, whether clearly enunciated or not. Restoration, and especially rehabilitation, may have non-ecological goals, such as aesthetic or recreational improvements.

Restoration ecology

Ecological restoration — although a field of management practice existing in its own right — can also be seen as a test of ecological concepts and theory (Jordan et al. 1987, Palmer et al. 1997). Restoring an ecological community, for example, can be viewed as a test of concepts in community and ecosystem ecology (e.g. Bradshaw 1987; Cairns 1987; Gilpin 1987; Hobbs & Hopkins 1990; Palmer et al. 1997; Parker & Pickett 1997; Lockwood & Pimm 1999). Restoration ecology involves not only the testing of ecological ideas in the execution of restoration projects, but also involves conceiving and testing new ideas and concepts specifically to assist restoration and rehabilitation. Thus, the links between ecological restoration and restoration ecology represent an effective mutualism (Clewell 1993; Palmer et al. 1997).

Restoration ecology is a new and immature branch of ecology. Like all areas of ecology, its development depends on the steady accumulation of empirical observations and data as well as the testing of hypotheses derived from models based on the observations. As many restoration projects are long-term projects, the gathering of useful data and the testing of some of the hypotheses in these projects will be, by necessity, a slow and long-term process. Currently there are large-scale restoration projects both in Australasia and internationally that have yet to report results in-progress or final results.

Five obstacles to the development of restoration ecology

Excluding the problem of gaining results from long-term projects, it is still obvious that the development of restoration ecology is being impeded by five obstacles. These include the reluctance of resource managers to undertake significant restoration projects; the poor design of many restoration projects; the lack of satisfactory monitoring of projects; the astounding lack of reporting on the progress and outcomes of projects; and resolution of problems of spatial and temporal scale in restoration projects. The first four of these are correctable by the restoration team itself, while the last (the recognition of the overriding importance of spatial/temporal scale of the location and its surrounds) is yet to be accepted by the broader community and, therefore, more readily accommodated in projects.

1. Need for participation (by resource management agencies) in significant projects

Successful restoration projects are invariably centred on degraded areas where considerable economic and social values exist alongside ecological values. Such areas are managed to varying degrees of competence and thoroughness by one or multiple resource management agencies. While ecologists can help in the design, monitoring and evaluation of a restoration, the active ‘works’ and ongoing maintenance of the project requires resource and logistic inputs from the appropriate resource management agency. Such a partnership can be effectively achieved by adaptive management involving the implementation of the Adaptive Environmental Assessment and Management (AEAM) process (Walters 1986). Rather than carrying out many small-scale restoration projects, such as under the recent NHT Program in Australia, it would be better (both for the resource management agency and for ecologists) to undertake a limited number of well-designed, large-scale projects. To date, at least in Australia, resource management agencies have shown a great reluctance to do this, even though we have a plethora of cases of large-scale ecosystem degradation. This reluctance is a major impediment to the development of restoration ecology.

2. Necessity for adequate design of restoration projects

Restoration projects may be undertaken without adequate consideration of such design necessities as the gathering of before-data, the need for replication (if possible), the provision of control and/or reference sites, the setting of feasible goals, and the capacity to test even the most elementary form of hypothesis. In recent times, in Australia, the practice of undertaking poorly designed restoration projects has been widespread; driven by political pressures to initiate community-based projects without interference from monitoring or research. ‘On-ground works’ have been supported and funded without any provision for linked research (Toyne & Farley 2000). In terms of ecology, very little of value has been learnt from such projects, and in terms of restoration, design inadequacy has meant that lessons learned on the sites cannot be reliably evaluated to improve restoration on these sites or elsewhere.

3. Monitoring of restoration projects

Monitoring is essential if progress in a project is to be evaluated. This necessity is linked with project design. In restoration projects, monitoring may be necessary of (i) the state of the inputs, (ii) the restoration manipulation, and (iii) the ecological responses. Inputs in projects that may require monitoring are such things as the volumes of environmental flows, the positioning of added logs in stream channels, exotic biomass removed, and the initial establishment of plantings or seedlings. Ecological responses in projects that may require monitoring include such things as colonization by plants and animals, accumulation of soils and soil biota, and successional development.

In both cases, monitoring depends on the selection of appropriate indicators. Desirable properties of indicators include ease of sampling and processing, relative low cost, lack of ambiguity such as taxonomic uncertainty, high sensitivity to the restoration measures (pragmatic in terms of the measurable characteristics of the system), and of course, direct incorporation into the hypotheses being tested (Keddy 1999).

4. Need for reporting of restoration projects

The lack of reporting on the progress and outcomes of restoration projects is a major problem. Ideally, the outcomes of a project, along with its hypotheses and its rationale, would be reported in the refereed scientific literature. There is a rapid proliferation of suitable journals. Even reporting in the ‘grey’ literature is a step in the right direction. In gaining knowledge on restoration, it is important to recognize that reasoned reporting of a failure may be as valuable, if not more so, than the reporting of a success. Otherwise faulty procedures may continue unchecked and money may continue to be wasted.

5. Consideration and resolution of scale in restoration projects

Temporal and spatial scale must be considered in restoration projects. It is important to realize that there is usually a great mismatch between the rate at which humans may damage ecosystems and the rate at which a damaged system can be restored, even with active intervention. This discrepancy in rates, that may be termed the hysteresis of repair, is well illustrated by activities such as land clearance. Hectares may be cleared in a morning’s work with a bulldozer; while to restore such cleared land would probably take decades, if not longer.

The rates at which natural processes operate, be they abiotic or biotic, are usually positively correlated with the spatial scale (Wiens 1989). Thus, ecological processes, such as nutrient cycling, occur at a faster rate at the small scale (plots) than they do when considered at larger scales, such as entire landscapes. This spatial–temporal scale correlation may govern the rate of restoration. It may be possible to restore an individual wetland in a few years but the restoration of a floodplain complex may take decades. In Florida, there is the ambitious project on the Kissimmee River to restore 70 km of river channel and 11000 hectares of floodplain (Arrington 1995). This is a large-scale project with a 20-year plan and costing US$8 billion (Pelley 2000). The time scale and the funding reflect the reality of large-scale projects. However, in many large-scale restoration projects, while the planning reflects the spatial–temporal scale or scope of the project, funding is usually only short-term. Unfortunately, the short-term nature of funds for effective long-term restoration projects is a function of the short-term programmes of funding bodies and of the short-term mind set of many managers and politicians.

Different biota have different rates of responding to restoration measures. This is obvious; annuals grow faster than trees. These differences largely govern the rate of succession in restoration. In the Kissimmee River project in Florida, it is projected that aquatic plants will recover in 3–8 years, invertebrates in 10–12 years and fish in 12–20 years (Trexler 1995). Such differences in recovery rates of biota also mean that these differences need to be considered in selecting indicators. Different rates of responses by different ecological indicators may mean that a graded sequence of hypotheses, from the short term to the long term, may be nested and tested in any one large-scale restoration project.

Initial steps in a restoration project

Assessment of condition and ‘restorability’

In any restoration project, as stressed by Hobbs and Norton (1996), a considerable amount of reconnaissance and assessment needs to be carried out prior to carrying out the restoration measures. In any landscape/waterscape selected for restoration, initial steps are to identify the degrading forces, to assess their current status and strength, to rank them in a priority order of severity, and to determine the likelihood of whether restoration will allow abatement of degradation and consequent recovery. Such an exercise may be difficult and, invariably, is multidisciplinary. In a stream restoration exercise such as the Granite Creeks Project (of the Cooperative Research Centres for Freshwater Ecology and of Catchment Hydrology) near Euroa, Victoria, this assessment involved hydrologists, geomorphologists, biologists, resource managers and Landcare staff. The outcomes were compiled into a report (Davis & Finlayson 2001) detailing the current geomorphological condition of the creeks and their catchments as well as the history of stream degradation. This study was then followed by an assessment of the ecological condition of the creeks, as judged by invertebrate and fish fauna; and from this assessment, indicators were selected. The whole 3-year exercise stresses the very desirable necessity of obtaining reliable before-data in any restoration project.

Assessment of linkages

Even if the main focus is the restoration locality itself, it is also very important to determine the form and strength of linkages between the locality and its surrounds. In cases of stream restoration, for example, it is critical that linkages with catchments are recognized. Catchment land use is usually a dominant driver of stream condition. Harding et al. (1998) found that ‘whole watershed use in the 1950s was the best predictor of present-day diversity’ in the streams they studied.

In both terrestrial and freshwater restoration projects, connections between the restored locality and sources of colonists and resources are important. Disruption of such linkages can disrupt restoration. In the case of streams, the maintenance of hydrological connectivity (Ward 1989; Townsend 1996) upstream–downstream and between streams, is critical for the movement of biota. Dams and barriers can break the connectivity. Levees on floodplains can break the connectivity between a river and its floodplain.

In assessing linkages, it is important to realize that there may be critical buffer zones or transition zones that mediate linkages between major types of environment. Riparian zones between catchments and their streams are an example. In constrained streams, riparian zones are usually narrow vegetated strips and are particularly susceptible to damage. Most of the damage comes from the terrestrial side and includes such forces as fire and grazing. Such damage may break the continuity of riparian zones, impeding the movement of biota and disrupting the catchment-stream linkages.

The concerns about linkages and investigating the surrounds of a selected locality raise the need to determine whether there are influences in the surrounds that may impede the restoration effort. There may be a conflict between aquatic and terrestrial management practices; an example of which may be the potential conflict between restoring a stream in an area susceptible to contamination by chemical weed control in adjacent terrestrial sites.

A further critical step in planning a restoration project is to ascertain the whereabouts and state of undamaged areas harbouring intact communities and viable populations; and if possible, to make sure that these areas are protected. Such areas may serve two key functions: they may be sources of colonists, and they may act as reference areas for both the setting of restoration goals and for statistically evaluating the progress of the restoration effort.

Developing restoration ecology by hypothesis testing

While there has been an ever-increasing interest in restoration, there has not been an accompanying growth in knowledge and the development of principles of restoration ecology (Hobbs & Norton 1996; Palmer et al. 1997, Ehrenfeld 2000). As mentioned before, there are at least five major impediments to the development of restoration ecology. Four are rectifiable by the restoration team within relatively confined time scales, while the problems due to scale require cooperation from a much broader base, often over long time periods. Linked with spatial–temporal scale in impeding progress in restoration ecology is the fact that, in many projects (especially large-scale ones) the gathering of useful data may be a long drawn-out process.

Lamenting the lack of progress in restoration ecology, Hobbs and Norton (1996) stated that it has ‘largely progressed on an ad hoc, site- and situation-specific basis, with little development of general theory or principles that would allow the transfer of methodologies from one situation to another’. Thus, seeking to place restoration ecology on a more solid footing and to allow it to develop general principles, Hobbs and Norton (1996) proposed a set of seven sequential steps or processes that need to be considered in restoration projects. However, while their key processes are valuable, if not crucial, to the success of a restoration project, they did not propose the need to incorporate hypothesis testing into restoration ecology. In many restoration projects, monitoring of varying degrees of rigour has been carried out to assess and evaluate progress, but explicit hypothesis testing is rare. As stressed by Michener (1997) and Chapman and Underwood (2000), it is imperative for scientific progress in restoration ecology that explicit hypothesis testing be carried out. Understanding the causal links and the critical steps in a restoration project does not come from generating models and simply carrying out monitoring. There is the clear need to generate hypotheses based on observations, available information and the project’s goals and to test the hypotheses in the context of the project. Within the large spatial–temporal scope of large restoration projects, elucidation of some of the critical causal links in the restoration may be gained from well-designed hypothesis-testing experiments at smaller scales (Havens & Aumen 2000).

The setting of goals and selection of indicators in restoration

In any restoration project there is the clear need to set goals. Ecological goals or targets may be set at various ecological levels: species populations, communities, ecological processes and ecosystem services. Targets may be set by regulation, requiring certain levels of compliance to be met. This particularly applies to environmental quality attributes such as water quality. Targets may be quantified in terms of their biotic and statistical properties, may be informed by specific attributes of reference sites (Hobbs & Norton 1996), or may be drawn from reference information found in historical and contemporary sources (White & Walker 1997). Reference sites need not be in perfect pristine condition (in fact, such sites are becoming increasingly impossible to find) but may simply be in good condition and typical of undamaged sites in the region.

To monitor changes in the project it is necessary to select indicators. In the case of single species restoration, the targets and the indicators may be identical. For communities and ecosystems, however, separate indicators have to be selected. This selection must cover the parameters postulated in any hypothesis. Desirable properties of indicators include their being relatively easy and inexpensive to measure (very important in long-term projects); they must have no taxonomic difficulties or measuring uncertainties; they need to be sensitive to the restoration measures; they need to respond at different rates over different time spans; and preferably they need to be linked with each other in their ecological functioning. In setting indicators, progress in a project may be either detected by increases in desirable biota or properties or by decreases in undesirable biota, such as weeds or exotics. In large-scale, long-term projects, it is a good tactic to select abiotic and biotic parameters that respond at different rates, short term to long term; hence, for example, the use in the Kissimmee River project of a range of indicators (Dahm et al. 1995; Trexler 1995).

In setting goals or targets and in subsequent monitoring, it should be borne in mind that at least three major types of difficulties may arise. First, the hypothesized response in the target biota may show a marked lag response (Huxel & Hastings 1999). For example, colonists may take time to reach and establish themselves; some key resource may be in short supply, or unanticipated disturbance may set back progress. Second, instead of progressing to the set goal, the system may proceed to an unforeseen and stable alternative state (Westoby et al. 1989; Hobbs & Norton 1996), and recovery of a system to its natural state ‘may require massive management inputs’ (Hobbs & Norton 1996). It may be that the system is now being exposed to a disturbance regime that did not exist in the original state. Indeed, setting goals that are resistant and/or resilient to the prevailing disturbance regime is an important consideration (Westman 1991). Third, the system undergoing restoration may not have any clear trajectory and stability. This may be because the system is being exposed to unforeseen disturbance and/or the system is but a fragment of the original state (which, with a high perimeter to area ratio, means it is being influenced by surrounding factors). In this case, restoration may require continual intervention, such as controlling weed in small patches of native vegetation.

Design of monitoring programs in restoration projects

Along with the needs for developing hypotheses from observations, establishing available information and modelling, selecting indicators, and setting goals, a further crucial need exists. This is the need for designing the project and the monitoring regime such that information can be gained to evaluate progress and to test hypotheses. Both inputs and outcomes may be monitored. In a stream restoration project, for example, it may be the condition and persistence of added coarse woody structures (inputs) and the diversity and population sizes of fish in the treated stream reaches (outcomes) that are monitored. While both inputs and outcomes require monitoring, it is imperative that selected indicators of projected outcomes are rigorously monitored.

In situations where reference sites are available, it should be possible to design a monitoring programme that both allows the testing of hypotheses and the evaluation of restoration progress. This would involve a monitoring design where before-intervention (B) and after-intervention (A) data are collected from the sites to be restored where intervention occurs (I) as well as the reference (R) sites (a multiple BARI design). In some instances, there may be only one site to be restored but a number of reference sites. Replication is very desirable as it increases the certainty of statistical analyses and increases the level of generalization.

In many cases, reference sites are simply not available. This particularly applies in urban and densely settled areas where sites are either less available or all of the sites are earmarked for treatment within a relatively short period. In other scenarios, however, there may be many degraded control (C) sites so that hypothesis testing and evaluation of progress may be carried out by determining departure in the restored site (I) from the state of the control sites (M BACI design). In the most powerful design, both reference and control sites are available and the project design (M BARCI design) incorporates both into the monitoring programme (Chapman & Underwood 2000). Departure from the control sites and convergence with the reference site conditions may be carried out with the bioequivalence statistical procedure detailed by McDonald and Erickson (1994).

As the spatial scope of the restoration project increases (such as from a small upland stream to a large lowland river), not only is it inevitable that the temporal scope will increase, but it is also inevitable that the presence of replicated control and/or reference locations will disappear. In this case it may be only possible to compare the ‘to-be restored’ site (I) with a degraded control (C) and/or a fairly intact reference location (R): BARCI, BARI or BACI designs.

Finally, and this is quite common, one may attempt to restore a location for which controls and reference locations are missing. If goals cannot be readily set for either departure from the degraded condition or convergence to the intact state, then it may be possible to set directions of restoration from historical data or from information from analogous but not identical locations. Evaluation of progress may be by a ‘levels of evidence’ approach outlined in Downes et al. (2001). Alternatively or together, a Bayesian analysis (Wade 2000) may be appropriate.

To carry out large-scale restoration projects as large-scale hypothesis-testing experiments, it is absolutely essential that solid and durable partnerships are formed between resource management agencies, concerned stakeholders and scientists. Until now such partnerships are rare. Thus, building partnerships that are centred on large restoration projects that involve the implementation of strong restoration measures is a critical advance. Even when such partnerships are formed and projects initiated, it is important to be aware that such partnerships may fracture and break causing the project to fail (Walters 1998).


I am grateful to Dr Ralph MacNally of the Department of Biological Sciences, Monash University for urging me to present this paper in draft form at the 2000 Annual Conference of the Ecological Society of Australia and to two referees for valuable criticism and comments.