1. Introduction

The degradation of aquatic habitats from human activities has been occurring for centuries, but became particularly pronounced in the late nineteenth and twentieth century in Europe, the United States of America, Australia, and other developed countries (Brookes, 1992; Welcomme, 1994; Arthington and Pusey, 2003). In most developing countries modification, degradation, and pollution of aquatic habitats began in the late twentieth century and continues today (Parish, 2004; Welcomme and Petr, 2004a). Estimates from throughout the globe suggest that 75 to 95 percent of riverine habitats are degraded (Benke, 1990; Dynesius and Nilsson, 1994; Muhar et al., 2000). Dynesius and Nilsson (1994) estimated that 77 percent of the 139 largest rivers (discharge > 350m³/s) in the Northern Hemisphere have modified hydrologic regimes. Inventories for specific countries are even higher. For example, 80 percent of large rivers in Austria are moderately to heavily impacted by human development (Muhar et al., 2000) and it is estimated that only 2 percent of the river kilometres in the United States of America are pristine (Benke, 1990).

Human activities such as forestry, agriculture, channelization, power production, industrialization, water use and many others can have a variety of negative consequences for fisheries and aquatic resources. Forest management practices have negatively impacted many streams by increasing fine and coarse sediment, altering stream hydrology, disrupting delivery of woody and organic debris, and simplifying habitat (Meehan, 1991; Murphy, 1995; Erskine and Webb, 2003). Improving the navigation of rivers and estuaries through dredging and snagging (removal of wood) has greatly simplified many rivers and is still widely practiced today (Sedell and Froggatt, 1984; Buijse et al., 2002; Collins et al., 2003). Agricultural activities have had detrimental effects on estuaries, floodplains, wetlands, and low-gradient tributaries through dredging, draining, filling, pollution, channelization of waterways, and diversion of water for irrigation (NRC, 1992; Cowx and Welcomme, 1998; Welcomme and Petr, 2004a). Irrigation and the over appropriation of stream flows particularly in arid regions has led to low stream flows, higher water temperatures, reduced total wetted habitat, reduced ability of the stream to transport sediment, and other deleterious effects (Orth, 1987; Hill et al., 1991; World Commission on Dams, 2000; Parish, 2004). Mining and other extraction industries have had many negative effects on streams from direct alteration and removal of substrates to pollution and release of toxic substances (Nelson et al., 1991). Residential development, industrialization, and urbanization have lead to a suite of problems for aquatic habitats including filling and channelization, changes in hydrology from increased impervious surface area, pollutants from point and nonpoint sources, elimination of riparian zones, and simplification of habitat (Booth, 1990; Booth et al., 2002; Riley, 1998; Konrad, 2003). All these factors have contributed to the degradation and simplification of inland aquatic habitats across entire ecosystems as well as the loss of biodiversity and are the basis for the development of numerous rehabilitation techniques.

Sustainable inland fisheries depend on well functioning ecosystems and adequate habitat. The importance of fishery habitat is acknowledged in the FAO Code of Conduct for Responsible Fisheries which was adopted by the FAO Conference "to ensure sustainable exploitation of aquatic living resources in harmony with the environment" (FAO 1995). Articles 2g) and 6.1of the Code call on States to, "promote protection of living aquatic resources and their environments..." and "... conserve aquatic ecosystems", respectively. Article 6.8 of the Code specifically addresses habitat rehabilitation: "All critical fisheries habitats... should be protected and rehabilitated as far as possible and where necessary". A review of the effectiveness of habitat rehabilitation efforts is in Article 7.6.7, "In the evaluation of alternative^[1] conservation and management measures, their cost-effectiveness... should be considered", and Article 7.6.8, "The efficacy of conservation and management measures should be kept under continuous review." FAO recognizes that conservation and management strategies must be viewed realistically taking into account numerous alterations and perturbations to inland water ecosystems and recommended that strategies should avoid "vain attempts to restore a substantially altered ecological balance" (FAO, 1997).

In response to the degradation of aquatic habitats from a variety of human activities, rehabilitation of aquatic habitats has become commonplace throughout developed countries and is increasing in developing countries (NRC, 1992; Cowx and Welcomme, 1998). These efforts are often undertaken to restore or improve natural resources that are of economic, cultural, or spiritual importance. Rehabilitation efforts typically occur throughout a watershed including both riparian and upland activities as well as activities in the lowlands such as reconnection of floodplains and addition of habitat structures (e.g. logs, boulders, weirs) in streams. The vast majority of these efforts have been undertaken to restore fisheries resources and in some cases large sums of money are spent on a single species or group of species. For example, hundreds of millions of dollars are spent annually in western North America in an effort to increase Pacific salmon (Oncorhynchus spp.) runs that once sustained large commercial and sport fisheries but are now threatened with extinction. Other ecosystem restoration programmes have been initiated in the Florida Everglades, the Missouri, Mississippi, and Sacramento rivers, the Louisiana Delta, Chesapeake Bay, the Great Lakes and other major basins (Northeast Midwest Institute, unpublished data; http://www.nemw.org/restoration_summary.htm). Similar efforts are underway in Europe to rehabilitate and reconnect habitats throughout large river basins such as the Rhine and Danube as well as many other rivers and water bodies (Buijse et al., 2002). Many are funded through the European Union’s environmental programs (i.e. LIFE or The Financial Instrument for the Environment). Interest in inland fisheries rehabilitation is increasing in many developing countries not only a result of declines in fisheries resources, but also because of desertification in arid regions owing to over appropriation of stream flows or increased flooding in temperate and tropical areas due to poor land use practices (Parish et al., 2004). For example, large efforts are underway to reforest areas and restore floodplain wetlands in the Yangtze River basin in China to reduce flooding, while efforts are underway in the arid Timar River basin to restore steamflows and wetlands and halt desertification and loss of biodiversity (Parish, 2004). Similarly, reflooding large portions of the Mesopotamian Marshes in Iraq, which almost disappeared following extensive draining in the 1990s, is currently underway (Richardson et al., 2005).

Restoration ecology is a relatively young, interdisciplinary field and the literature on aquatic rehabilitation is extensive yet somewhat fragmented (Buijse et al., 2002). Several existing publications that discuss different techniques for rehabilitation in North America (Hunter, 1991; NRC, 1992; Hunt, 1993; Slaney and Zoldakas, 1997; FISRWG, 1998), Europe (Brookes and Shields, 1996; Petts and Calow, 1996; RSPB et al., 1994; Cowx and Welcomme, 1998; Vivash, 1999), Australia (Rutherfurd et al., 2000), and east Asia (Parish et al., 2004) are applicable to many regions throughout the world. Other texts discuss the ecological basis for restoration (e.g. Naiman and Bilby, 1996; Calow and Petts, 1994; Perrow and Davy, 2002) and some regional papers or grey literature discuss effectiveness of different techniques (e.g. Binns, 1999; Roni et al., 2002; Avery, 2004). Unfortunately, no comprehensive review of the effectiveness of various rehabilitation techniques has been completed. In particular, the effectiveness of these efforts at restoring natural watershed processes, improving fish habitat, and increasing fish and biota abundance, and the cost and benefits of various techniques needs to be examined (Roni et al., 2002).

The goals of this document are to synthesize the available information on the effectiveness of various habitat rehabilitation techniques and provide guidance for restoration of aquatic ecosystems. By outlining the shortcomings of techniques as well as our understanding of aquatic systems, we strive to improve our understanding of how to plan and prioritize projects as well as to monitor and evaluate their success. To provide some common background, we begin with a general discussion of rehabilitation and restoration terminology and provide an overview of ecological processes that create aquatic habitats and their importance in planning rehabilitation projects. Next we synthesize what is known about the effectiveness of different categories of techniques at restoring natural processes, improving physical habitat, and increasing fish and biotic production. We then discuss the costs and cost-effectiveness of various activities, and provide recommendations for prioritizing restoration as well as monitoring and evaluation. We focus on freshwater habitats and habitat modifications and while other factors are important for successful watershed rehabilitation, such as remediation of pollutants, toxicology, water quality and quantity, we do not discuss them here. Rehabilitation of estuarine and marine habitats, an equally large field, will be covered in another volume.

1.1 Habitat restoration vs. rehabilitation

There are many different definitions of restoration and rehabilitation and practitioners and researchers are in disagreement as to what constitutes restoration (Gore, 1985; Cairns, 1988; NRC, 1992; Kauffman et al., 1997). The term restoration, which in the most formal sense is returning an ecosystem to its original predisturbance state, has commonly been used to refer to all types of habitat manipulations including enhancement, improvement, mitigation, habitat creation, and other situations (Table 1). These activities are more accurately termed rehabilitation, as most do not truly restore a system and in many areas were the land use is predominantly agricultural, residential, urban or industrial, true restoration is not feasible in the foreseeable future (Stanford et al., 1996). Therefore, we use the term habitat rehabilitation throughout this document to refer to the various activities and where appropriate use more specific terminology (see Table 1).

TABLE 1
Commonly used terminology and general definitions*

* The authors use the term rehabilitation because it includes both full restoration (which is difficult to achieve in many populated areas) as well as the other activities described below. Modified from Roni (2005).

1.2 Ecological basis for rehabilitation

River rehabilitation has often been pursued with little knowledge of the natural structure and function of rivers, and projects have commonly attempted to create habitats considered suitable for a single species of interest (Frissell and Nawa, 1992; Muhar et al., 1995; Ward et al., 2001; Roni et al., 2002; Ormerrod, 2004). However, many scientists have pointed out that worldwide declines of fishes and other aquatic species in fresh waters are partly a result of trying to manage individual species and habitat characteristics rather than managing whole ecosystems (e.g. Doppelt et al., 1993; Muhar et al., 1995; Muhar, 1996; Frissell et al., 1997). Scientists and managers alike have also recognized that rehabilitation actions are more likely to be successful at restoring individual or multiple species and preventing the demise of others if they are considered in the context of the surrounding watershed or ecosystem (Doppelt et al., 1993, Muhar et al., 1995, Reeves et al., 1995, Beechie and Bolton, 1999; Habersack, 2000). The watershed and ecosystem contexts are also critical to understanding the effectiveness of various practices, as well as why they succeed or fail (Harper et al., 1998, Jungwirth et al., 2002). International organizations have also adopted an ecosystem approach to fisheries development and conservation. The Convention on Biological Diversity (CBD) (UNEP, 1998) and FAO (FAO, 1995; Sinclair and Valdimarsson, 2003) have recognized the necessity of, inter alia, understanding ecological processes, the bio-physical-chemical qualities of aquatic habitat, nutrient cycling, the importance of nontarget species, predator-prey relationships and even the role of humans in ecosystems in order to ensure long-term survival of aquatic species. Therefore, before discussing the effectiveness of habitat rehabilitation techniques, some background on watershed processes is necessary to assist in understanding the variability in effectiveness among techniques as well as among studies on a given rehabilitation method.

Effective planning, implementation, and evaluation of rehabilitation actions requires assessment of disrupted ecosystem functions that reduce the productivity of river systems and are responsible for declines in aquatic ecosystem integrity (Beechie and Bolton, 1999; Buijse et al., 2002). The goal of such assessments is to identify alterations of key processes that affect stream habitats and specify management actions required to restore or rehabilitate those processes that sustain aquatic habitats and support biological integrity (e.g. Muhar et al., 1995; Tockner et al., 1999; Buijse et al., 2002; Beechie et al., 2003a). In this approach, restoring specific fish populations (or populations of any other single organism) is subordinate to the goal of restoring the ecosystem that supports multiple species. As long as all rehabilitation actions are consistent with the overriding goal of restoring ecosystem processes and functions, habitats will be restored for multiple species.

The scientific basis for this approach can be summarized in two important characteristics of biota and their habitats (adapted from Beechie and Bolton, 1999):

1. Biota are adapted to local environmental conditions.
2. Spatial and temporal variations in landscape processes create a dynamic mosaic of habitat conditions in a river network.

These statements imply that biota are adapted to spatially and temporally variable habitats, and suggest that such environmental variability is important to the long-term survival of fish stocks or races. Thus it does not make sense to manage for the same conditions in all locations, or to expect conditions to remain constant in any single location. This has been recognized in scientific critiques of many management issues in the past decade, including "one-size-fits-all" habitat standards (Muhar et al., 1995; Bisson et al., 1997), not managing for spatial or temporal variation in habitats (Reeves et al., 1995; Bisson et al., 1997; Habersack, 2000), and addressing symptoms of a disrupted ecosystem rather than the causes (Frissell and Nawa, 1992; Muhar et al., 1995; Spence et al., 1996; Ormerrod 2004). Such approaches generally do not consider that local populations are adapted to the natural potential habitat conditions within their range, and that those conditions vary in space and time. By contrast, identifying the root causes of degradation (i.e. impaired ecosystem processes and functions) focuses rehabilitation actions on those processes that form and sustain habitats. This allows each part of the river network to express its natural potential habitat, and helps conserve and restore the natural spatial and temporal variation of habitats to which aquatic communities are adapted (Muhar, 1996; Beechie and Bolton, 1999).

We stress that identifying the root causes of ecosystem degradation is important for two main reasons. First, our understanding of most of the linkages between landscapes, habitat, and aquatic biota is fraught with uncertainty, and we cannot predict exactly how land uses alter aquatic habitat conditions or how those habitat changes alter biota. In fact, it can be argued that we are not yet even aware of all the aspects of aquatic ecosystems that significantly affect fish populations. This lack of knowledge has in the past led to significant habitat degradation. For example, the role of wood debris in habitat formation was poorly understood until the 1970s, and removal of wood and channelization of rivers over the past 200 years resulted in dramatic alteration of river habitats (Sedell and Luchessa, 1982; Muhar et al., 1995; Collins and Montgomery, 2002). As recently as the 1980s biologists recommended widespread wood removal for habitat improvement, not recognizing its importance to the structure and function of aquatic ecosystems. Today wood removal is far less common (but still occurs), however, the example serves to illustrate that we could have avoided significant habitat loss by choosing management actions that preserved riparian forest processes and natural wood functions in channels, even without understanding the value of wood in aquatic ecosystems.

Second, traditional rehabilitation actions such as boulder and log structures or spawning gravel placement attempt to build habitats that do not move in space or time, whereas natural habitats are typically created by movement of river channels, wood debris, and sediment. Therefore, many rehabilitation actions fail to restore habitats because they do not recognize the integrated nature of physical and ecological processes in watersheds (Frissell and Nawa, 1992; Ward et al., 2001; Ormerrod, 2004). As we shall see in subsequent sections of this document, this relative lack of knowledge leads to two main types of failure: 1) site-prescribed engineering solutions can be overwhelmed by altered watershed processes that are far removed from degraded habitats (e.g. increased sediment supply from upslope sources can bury instream structures and pools), or 2) such measures can prevent habitat formation that would otherwise naturally occur (e.g. bank protection prevents formation of new floodplain habitats). Avoiding these types of project failure requires that we focus on restoring ecosystem processes and functions that form and sustain aquatic habitats, rather than on the habitats themselves. Understanding these factors will also help the reader to understand project successes and failures discussed in subsequent sections of this report.

Beechie et al. (2003b) proposed a simple conceptual framework detailing relationships among ecosystem processes, rehabilitation actions, habitat conditions, and biota (Figure 1). Using this framework we discuss actions that restore either ecosystem processes (e.g. sediment reduction actions, riparian planting, floodplain rehabilitation) or connectivity of habitats (e.g. restoring fish passage) and contrast those with rehabilitation actions that directly manipulate or modify stream habitats (e.g. instream structures, bank hardening, construction of floodplain ponds). We recognize that most evaluations of habitat rehabilitation have focused on instream habitats, with a lesser degree of emphasis on connectivity. There have been relatively few evaluations of the effectiveness of actions to restore sediment supply or hydrologic regime, in part because these processes are naturally highly variable and relatively small changes are hard to detect at the watershed scale. Moreover, such processes often require decades for results to manifest in stream channels, making linkage of such actions to changes in habitat characteristics and biota even more difficult (Beechie et al., 2005).

River ecosystems throughout the world are driven by the same fundamental sets of processes, such as supply of water, sediment, and organic matter (Leopold et al., 1964; Welcomme, 1985; Knighton, 1998). However, the specific mechanisms driving each of the inputs vary depending on geologic, topographic, and climatic setting (Table 2). Therefore, restoration or rehabilitation actions will address the same sets of processes, but techniques will vary depending on the specific mechanisms that must be addressed. For example, sediment supply to the stream network should be considered in any watershed, but certain processes of sediment supply may be emphasized depending on location. Sediment supply is dominated by landslides in humid mountainous regions (Sidle et al., 1985), but is more commonly a function of surface erosion and gullying in gentler terrain or semiarid regions (Dunne and Leopold, 1978; Darby and Simon, 1999). Land uses that alter these erosion processes also vary by region, and techniques needed to address changes in sediment supply must be adapted to specific erosion processes and land use impacts.

Conducting watershed assessments and understanding important processes driving habitat development and degradation are critical elements in planning rehabilitation actions. Identifying these processes can help determine the success of the various stream and lake rehabilitation techniques we will discuss.

FIGURE 1
Diagram of linkages between landscape controls, habitat-forming processes, habitat conditions and biological responses. (Source: Beechie et al., 2003b)

TABLE 2
Examples of variation in dominant ecosystem processes and human impacts across ecological settings. Assessments should target those processes that are locally important within each region or watershed.