2. Effectiveness of common techniques

Dozens of methods and techniques have been developed to rehabilitate freshwater habitats, ranging from methods that try to restore natural processes (e.g. riparian replanting, sediment reduction) to those that seek to create immediate changes in physical habitat (e.g. placement of instream habitat structures) in hopes of creating rapid increases in target species. Typically many types of rehabilitation are undertaken at the same time at a given site or within a catchment. We categorize these rehabilitation activities based on the general area where they occur or the processes they seek to restore including: 1) roads and uplands, 2) riparian, 3) floodplain, 4) dam removal, 5) instream structures, 6) structures in lakes and ponds, 7) nutrient enrichment, and 8) miscellaneous activities (e.g., habitat protection, bank protection, beaver reintroduction or removal) (Table 3). We briefly describe each type of rehabilitation and then describe what is known about its effectiveness at improving natural processes (delivery of wood, water, sediment, nutrients, etc.), physical habitat, and biota. For biotic responses we focus primarily on fishes and to a lesser extent on macroinvertebrates, but report on plants and other biota when available and appropriate. Detailed information on specific techniques and their designs can be found in RSBP et al. (1994) Cowx and Welcomme (1998), Slaney and Zoldakas (1997), FISRWG (1998), and Vivash (1999).

To assess the effectiveness of rehabilitation we conducted an extensive review of the existing literature focusing primarily on peer reviewed literature, supplementing that with information from grey literature when available. We utilized databases such as Aquatic Fisheries and Science Abstracts and Web of Science to locate relevant papers, books, and technical reports on restoration theory (general background information) and, more importantly, evaluations of project effectiveness. We also searched library catalogues of the United Nations Food and Agriculture Organization, the US National Oceanic and Atmospheric Administration’s Northwest Fisheries Science Center, and the University of Washington, as well as conducted Internet searches with relevant keywords (e.g. aquatic restoration, rehabilitation, habitat improvement). We obtained more than 700 papers on freshwater habitat rehabilitation and restoration including 334 that reported results of scientific evaluations of the effectiveness of one or more habitat rehabilitation techniques (Figure 2a). There were also many other books and reference material examined. The vast majority of the studies on both rehabilitation and effectiveness were from the United States, Canada and Western Europe, with a relatively small number from other countries (Figure 2b). This likely reflects the size of the economy, funding for research programmes, and money spent on habitat rehabilitation in these developed countries versus elsewhere in the world. Most of the published literature on effectiveness focuses on instream rehabilitation or nutrient enrichment, while most of the general or theoretical literature focuses on floodplain restoration or more general references on various types of restoration. Thus the reader needs to be conscious that the following discussion on the effectiveness of various techniques is biased as it is limited by the amount of published information available on effectiveness for a given techniques.

TABLE 3
Common categories of habitat rehabilitation discussed in this review and examples of common actions. More detailed descriptions of techniques are provided in the section on effectiveness.

2.1 Road rehabilitation

The construction of both paved and unpaved roads can have a number of negative impacts on aquatic ecosystems (Trombulak and Frissell, 2000; Gucinski et al., 2001). Roads alter hydrologic regimes (Harr et al., 1975; King and Tennyson, 1984; LaMarche and Lettenmaier, 2001) and sediment supply to streams (e.g. Sidle et al., 1985), which influence channel and habitat characteristics (e.g. Cederholm et al., 1982; Hicks et al., 1991; Nyssen et al., 2002) and ultimately impact aquatic biota (Waters, 1995; Gucinski et al., 2001; Gibson et al., 2005). Forest road drainage connections to stream channels (e.g. ditches draining to streams) can alter the amount and timing of water delivery to streams, as well as the delivery of sediment eroded from hill slopes or road surfaces (Croke and Mockler, 2001; Madej, 2001). Roads also alter sediment supply through increased frequency of landslides (Dyrness, 1967; Megahan and Kidd, 1972; Sidle et al., 1985; Best et al., 1995) and increased surface erosion (Reid and Dunne, 1984; Bilby et al., 1989). Ultimately, the changes in sediment and hydrology have a negative impact on many fishes and other aquatic biota.

Much of the research on the impact of roads and techniques for rehabilitation has focused on unpaved or gravel roads in forested, mountainous, or rural areas. However, roads and other impervious surfaces have a dramatic effect on watershed processes in urban areas as well. For example, in urban areas roads and other impervious surface areas lead to increases in frequency and magnitude of peak flows, channel incision, and simplification of instream habitat (Scott et al., 1986; Booth, 1990; Wong and Chen, 1993; Moscrip and Montgomery, 1997). When roads and impervious surfaces exceed 10 percent of the total watershed area, changes in hydrology and negative impacts on aquatic habitat and biota will occur (Klein, 1979; Booth and Jackson, 1997).

Road crossings such as culverts can also prevent or inhibit the upstream or downstream migration of many fishes and other aquatic organisms (Larinier, 2002c; Roni et al., 2002). The amount of stream habitat made inaccessible because of human infrastructure is daunting. There are an estimated 1.4 million stream-road crossings in the United States and an estimated 2.5 million artificial barriers prevent fish passage in the United States (US Fish and Wildlife, National Fish Passage Programme, Arlington, Virginia, unpublished data). While massive efforts have focused on replacing impassible culverts in streams in the United States and Canada, fish passage through culverts and other road crossings is a problem throughout North America (e.g. USGAO, 2001; Langill and Zamora, 2002; Warren and Pardew, 1998), Europe (e.g. Yanes et al., 1995; Glen, 2002; Larinier, 2002c), and elsewhere in the world.

2.1.1 Techniques for reducing road impacts

A number of rehabilitation actions have been attempted to mitigate for the negative impacts that roads have on aquatic ecosystems. These can be divided into three categories: sediment reduction, restoration of hydrology, and connectivity. Connectivity is extensively discussed in the floodplain rehabilitation section, but here we will discuss it in terms of road crossings of small streams.

A variety of methods are used to reduce sediment delivery and landslides induced by roads or road construction. These include resurfacing roads, reducing traffic, increasing the number of stream crossings, stabilizing cut and fill slopes, and replacing stream crossings to improve the natural transport of sediment and biota (Table 4; Figure 3). In addition, complete road removal or abandonment of roads, which may use many of the previously mentioned techniques, is also a common method to reduce road impacts.

Actions to reduce hydrologic effects of roads in rural or forested areas include some of the same activities as sediment reduction as well as increasing the number of cross drain structures, water bars to distribute water onto the forest floor or into existing stream channels, and other techniques to prevent the road or road ditches as serving as channel networks. Reducing the amount of road surface draining directly to streams can also reduce fine sediment delivery.

MARY ANN MADEJ, UNITED STATES GEOLOGICAL SURVEY.

MARY ANN MADEJ, UNITED STATES GEOLOGICAL SURVEY.

MARY ANN MADEJ, UNITED STATES GEOLOGICAL SURVEY.

FIGURE 3. Examples of forest road removal before, immediately after, and a few years after removal of a stream crossing and road fill in Redwood National Park, California.

In urban areas several new techniques have been applied in recent years to control increased runoff and increase storm-water infiltration. Those include but are not limited to porous paving, removing hard surfaces where possible and planting with vegetation, a variety of detention, retention, or seepage basins or ponds, overflow wetlands, addition of rainwater cisterns, high-flow bypass channels, alternative drainage systems, and low-impact development for new construction (Sieker and Klein, 1998; Riley, 1998) (Figure 4). Much of the work on urban systems focuses on protecting waterways in suburban or rapidly urbanizing areas by limiting impervious surface areas and protecting riparian zones and uplands. In contrast, work in existing urban centres has focused on preventing further degradation and structural improvements to channels and riparian areas when possible (Sailer, 1994; Horner and May, 1999).

Culverts, bridges, and fish migration ladders are used to restore connectivity among stream reaches and channels at road crossings. Most are designed to provide adequate adult fish passage at road crossings, but not all provide passage for juvenile fishes or maintain natural processes (e.g. sediment and wood transport) and many affect channel morphology (Table 5). Bridges may be costly, but do not constrain the channel as much as culverts and allow the passage of other materials and formation of a natural stream channel. Open bottom culverts or embedded (e.g. countersunk) pipe-arch culverts allow a natural substrate to form within the channel and are effective at passing both juvenile and adult salmonids (Furniss et al., 1991; Clay, 1995; Larinier, 2002c). However, such culverts may constrain the stream channel if the culvert size does not account for large flow events or the volume of sediment and wood transported by the stream (Robison et al, 1999). Other design options include backwatering culverts at the outlet or inlet and placing baffles within the culvert to reduce flow velocity. Clay (1995) and Larinier (2002c) provide concise reviews of culvert designs and methods for retrofitting impassible culverts. While most culverts are designed to pass adult fish, additional research is needed to confirm which types effectively pass juvenile fish at a variety of flows. An intensive review of fish passage structures and their effectiveness for different species is beyond the scope of this document, but can be found in Clay (1995), DVWK (2002) and Larinier (2002a,b).

TABLE 4
Overview of rehabilitation techniques commonly used to reduce sediment inputs and hydrologic effects from roads (Modified from Beechie et al., 2005). D = road decommissioning, I = road improvement, B = commonly used in both decommissioning and improvement projects. Techniques are described in detail in Reid and Dunne (1984) Bilby et al. (1989), Burroughs and King (1989), Chatwin et al. (1994), Furniss et al. (2000), and Madej (2001).

2.1.2 Effectiveness of techniques for reducing road impacts

With the exception of stream crossings such as bridges and culverts, which can inhibit the migration of fishes and other organisms, all of the research on the effectiveness of road rehabilitation efforts has focused on their physical effects on the stream channel and occasionally on processes such as erosion rates and landslides in mountainous regions. While many of the techniques discussed above are widely utilized in mountainous areas, few published evaluations of their effectiveness exist (Switalski et al., 2004). Evaluations of road-surface erosion reduction techniques have generally been limited to comparisons of fine sediment concentrations in road runoff at different traffic levels and with different surfacing materials. For example, Bilby et al. (1989) found a positive relationship between traffic levels and fine sediment delivery to stream channels. Reducing traffic levels in the Clearwater River watershed in Pacific Northwest US reduced surface erosion by a factor of 10 (Reid and Dunne, 1984).

Other studies have demonstrated that reducing traffic levels and tire pressure of trucks can reduce sediment delivery from forest roads (Foltz, 1998; Foltz and Elliot, 1998). Reid and Dunne (1984) demonstrated that increasing the thickness of surfacing material to 15.2 cm reduces surface erosion by about 80 percent. Similarly, Kochenderfer and Helvey (1987) demonstrated that surfacing an Appalachian Mountain forest road with 7.6 cm (3 inches) of gravel reduced soil loss from 47 to 6 tons per acre. Burroughs and King (1989) reviewed unpublished studies evaluating different methods for reducing sediment erosion from roads and indicated that surfacing the road with 7.5 to 15 cm of crushed rock greatly reduced erosion. Moreover, reducing truck traffic on unpaved roads or paving of the road surface greatly reduced erosion (Burroughs and King, 1989). These studies, while limited to the western United States, suggest that these methods when implemented properly can reduce surface erosion.

CITY OF SEATTLE PUBLIC UTILITIES

CITY OF SEATTLE PUBLIC UTILITIES

FIGURE 4. Example of recontouring and reduction of impervious surfaces to reduce storm flows in an urban watershed. Reduction of impervious surface, addition of vegetation, and swales to capture rain and storm water. Preliminary results indicate that project reduced storm flow and runoff from street to stream by more than 95 percent.

TABLE 5
Summary of various stream crossing structures and whether they allow for fish passage (juvenile and adult salmonids) and the transport of sediment and large woody debris (LWD) or impact stream morphology by constraining the channel. Modified from Roni et al. (2002).

^a Fish passage depends on culvert slope and length
^b Depends on size of culvert or bridge relative to channel and floodplain width

Road removal or abandonment has also been demonstrated to greatly reduce sediment delivery (Madej, 2001; Hickenbottom, 2000; Switalski et al., 2004). For example, Hickenbottom (2000) demonstrated in a Montana watershed that 12 months after road removal and recontouring runoff and erosion approached those of natural slopes. The type of treatment following road removal or closure (e.g. recontouring of slope, ripping of road surface, removal of stream crossing, placement of mulch, seeding, planting) can influence the sediment production and infiltration capacity of the former road bed (Cotts et al., 1991; Maynard and Hill, 1992; McNabb, 1994; Luce, 1997; Elseroad et al., 2003). In general, these studies have found that recontouring slope and site preparation (ripping and mulching of former road bed) are the most effective at inducing plant growth and reducing fine sediment production, though position of the road on the slope and time since treatment also play a role in the effectiveness of these methods.

Very few evaluations of rehabilitation techniques for landslide hazard reduction have been conducted. Harr and Nichols (1993) provided anecdotal evidence that road removal resulted in reduced landslide rates in mountainous areas. Cloyd and Musser (1997) examined a subset of over 1 200 kilometres of treated and untreated forest roads stabilized or obliterated in Oregon and found higher problem rating (erosion and mass failure severity) on untreated roads associated with stream crossings. Changes in turbidity associated with stream crossing and road rehabilitation have occasionally been examined; however, results generally show short-term, construction-related increases in turbidity (Brown, 2002). Finally, it is important to note that the geology plays an important role in the level of natural sediment delivery, the sediment delivery from roads, and the overall success of sediment reduction effort through road resurfacing or complete road removal (Bloom, 1998).

Much of the literature on urban stormwater management and hydrology has focused on modeling of potential changes (e.g. Sieker and Klein, 1998; Johnson and Caldwell 1995) and few published evaluations exist. Booth et al. (2002) indicated that stormwater retention ponds have been generally inadequate to alleviate channel erosion or restore hydrology in highly urbanized areas because of the small size of most the ponds. Newer methods for stormwater reduction and reducing hydrologic and water quality impacts through natural drainage systems (e.g. constructed swales, wetlands, vegetated areas) have shown promise, but are in early stages of development and evaluation.

Physical studies on channel response following culvert removal are rarely conducted or published. In a rare study on channel adjustment following culvert removal, Klein (1987) reported channel adjustments following removal of culverts in a California watershed were minimal when large woody debris or other channel roughness elements were present. Most published studies have, however, focused on biotic response or recolonization -usually the major objective of culvert replacement projects. Fish often colonize new habitats relatively quickly and several studies have demonstrated the effectiveness of replacing stream culverts at allowing fish migration (Iversen et al., 1993; Bryant et al., 1999; Glen, 2002; see also section on floodplain rehabilitation). However, if fish numbers are extremely low or if a culvert is only passable at some water levels or seasons, it may take several years for fish to colonize new habitats and monitoring to evaluate success may need to be long term. Studies comparing different types of habitat rehabilitation techniques for salmon have shown removing barriers that block fish migrations leads to some of the largest increases in fish production (Scully et al., 1990).

In addition to benefiting fishes, culvert replacement or other barrier removal projects may benefit invertebrates and wildlife (Yanes et al., 1995; Vaughan, 2002). For example, Yanes et al., (1995) found that culverts both provided and inhibited migration of many mammals and reptiles, and Vaughan (2002) suggested that culverts inhibit upstream movement of many aquatic macroinvertebrates. Finally, the ability of various types of stream crossing to allow for fish and other aquatic organism passage involves a complex relationship between the physical characteristics of the stream crossing (e.g. depth, velocity, roughness) and the behaviour and swimming performance of the fish or other biota (Clay, 1995; Larinier, 2002b,c). While the published literature is limited, it is clear that replacing culverts and other road crossing structures that prevent or inhibit fish migration is a highly successful technique to increase fish habitat and production.

2.1.3 Conclusions - techniques for reducing road impacts

Road removal, abandonment, replacement of impassible culverts and other techniques provide some obvious benefits in terms of sediment reduction, restoring natural hydrologic regimes and habitat connectivity in aquatic systems. There are, however, relatively few published studies on recovery of watershed processes such as hydrology and sediment delivery following road improvements or removal because of the high cost of long term monitoring needed to detect these changes. Little long term monitoring is typically conducted or published on removal of culverts that block fish access because they provide almost immediate and relatively obvious results. Based on our review of the literature on rehabilitation methods to reduce impacts of roads on aquatic systems, we provide the following recommendations with the understanding that new studies are needed to provide more specific recommendations.

• Several techniques (e.g. traffic reduction, resurfacing, removal) have been developed that appear successful at reducing erosion and landslides and improving hydrology associated with roads in forested areas.

• Many techniques exist for reducing impacts of urban stormwater, though their effectiveness is unclear.

• Replacing culverts or other stream crossings that prevent migration of fish is a highly effective technique for restoring connectivity, though stream crossing types vary in their ability to transport sediment and restore other watershed processes.

2.2 Riparian rehabilitation

Riparian zones are the interface between terrestrial and aquatic environments that provide important chemical, physical, and biologic functions within a watershed, including processing nutrients, delivering woody debris and organic matter to a stream, providing shade, stabilizing soils, regulating microclimate, and many other important functions (Naiman et al., 1993; Pollock et al., 2005). They are also important habitat for both terrestrial and aquatic biota and are zones of particularly high productivity. Unfortunately, because of their proximity to water, transportation and the productivity of the forests and soils, they have been and continue to be highly impacted by anthropogenic activities (Pollock et al., 2005). Common activities in riparian areas include agriculture, grazing, logging, recreation, building of roads and transportation, as well as various residential, urban, and industrial developments (see also section on floodplains). In an effort to improve habitat for fish and wildlife and for a variety of cultural and ecological reasons, large efforts have been initiated to restore riparian areas and improve stream habitat quality.

2.2.1 Techniques for riparian rehabilitation

Several strategies have been developed for either restoring or improving riparian condition typically with the ultimate goal of improving aquatic habitats. These fall into two major categories: 1) silviculture treatments and 2) fencing and grazing reduction. We discuss the two separately as the approaches to rehabilitation of riparian areas are quite different. Silviculture treatments seek to restore riparian areas through replanting of trees and other vegetation, typically with protection from further harvest or vegetation removal. Grazing management or reduction efforts typically seek to remove or limit pressure on riparian areas (grazing) and allow vegetation to recover naturally (sometimes called passive restoration).

2.2.1.1 Riparian silviculture

A variety of silviculture treatments have been used to improve or restore riparian conditions and riparian forests. These include seeding, planting and removal of trees or removal of competing understory vegetation, as well as the removal of a disturbance suppressing vegetation or the alteration of physical conditions (e.g. flood regimes or sediment supply) such that desired vegetation becomes established or undesirable vegetation dies (Smith, 1986; Verry et al., 1999; Stromberg, 2001; Pollock et al., 2005) (Figure 5). Riparian rehabilitation methods can be categorized based on the whether they remove or kill vegetation (thinning/harvest, girdling, competitive release, pruning) or plant vegetation (seeding, planting, coppicing, staking and layering) (Table 6). Most of these techniques were adapted from techniques in upland forests and much of the silvicultural literature focuses on techniques to grow tall, straight, densely packed, commercially important trees, often as monocultures, rather than focusing on creating diverse and functional ecosystems (Smith, 1986; Burkhart et al., 1993; Curtis et al., 1998). There are also a number of nonsilviculture techniques that have been used to improve riparian zones including modifying flow regimes, bank stabilization, and managing invasive exotic plant species.

Because of their natural high rates of disturbance, which create unoccupied patches, riparian areas are relatively susceptible to colonization by exotic species, and in some river systems, the number of exotics is quite high (Nilsson, 1986; deWaal et al., 1995). As such, the problem of controlling exotic species in riparian areas is global in nature (Dudgeon, 1992; Rowlinson et al., 1999; Tickner et al., 2001). However, most of these species do not significantly alter ecosystem functions or out compete native species and thus are not considered invasive or noxious. In contrast, invasive species are aggressive and often form monocultures or become the dominant species if left unchecked. In some instances, riparian areas are dominated by nonnative species and restoration involves removing such species. Tamarisk (Tamarix ramosissima) and salt cedar (T. chinoensis), invasive species in the American Southwest, for example, have light seeds, enabling them to disperse easily and readily occupy recently disturbed riparian areas, and have caused considerable degradation of riparian habitat throughout the American Southwest (Sudbrock, 1993; Zavaleta, 2000). Other problematic riparian invasive species common in North American and Europe include but are not limited to Japanese knotweed (Fallopia japonica), reed canary grass (Phalaris arundinacea), giant reed (Arundo donax), and leafy spurge (Euphorbia esula). There are undoubtedly countless others invasive species in other parts of the world.

FIGURE 5. Examples of riparian replanting including live trees (top) and planting of willow stakes (bottom) (River Drau, Austria).

Other methods of riparian rehabilitation include more passive efforts such as letting vegetation recover natural or restoration of flood waters. The restoration of flooding or groundwater through dam removal or restoration of normative flows, covered in a later section, is another method for restoring riparian vegetation and river channels

2.2.1.2 Fencing and grazing reduction

Heavy livestock grazing has lead to degradation of riparian and stream habitats throughout the world (e.g. Platts, 1991; NRC, 1992; Robertson and Rowling, 2000; Jansen and Robertson 2001). Riparian areas tend to be more heavily impacted from grazing than upland areas because they are more heavily utilized by livestock and native ungulates owing to their proximity to water and high quality forage (Armour et al., 1991). Livestock grazing can negatively impact riparian and stream environment by changing and reducing vegetation, destabilizing banks, increasing sediment, channel widening, channel aggradation, changing hydrology, and lowering the water table (Armour et al., 1991; Platts, 1991; Belsky et al., 1999). Effects on fish habitat typically include reduction in shade, cover, and terrestrial food supply, increase in water temperature and quality, alteration of stream morphology, and reduction in pools and depth (Armour et al., 1991; Platts, 1991; Belsky et al., 1999).

TABLE 6
Common techniques used in riparian rehabilitation. Pruning and harvest are also occasionally used to improve or restore riparian functions (Modified from Pollock et al., 2005). Dam removal and restoration of floods, which also can improve riparian conditions, are discussed in a separate section.

Riparian and stream channel habitats affected by grazing are restored primarily by completely removing grazing, excluding livestock from part of the riparian zone with fences or other structures, or by implementing a grazing reduction or management system that enables riparian vegetation to recover (Elmore, 1992) (Figure 6). Various grazing management systems have been implemented throughout world; almost all involve control of duration of grazing, ungulate numbers, or both. One particularly common one is rest-rotation grazing management, which includes alternating periods of grazing and nongrazing often rotated on multiple pastures. Other techniques to keep livestock out of riparian areas include providing water sources away from the stream (i.e. nose pumps) or providing salt away from the stream (Platts and Nelson, 1985).

2.2.2 Effectiveness of riparian rehabilitation

2.2.2.1 Effectiveness of riparian silviculture

The effectiveness of riparian techniques has been primarily evaluated through short term (<10 years) examination of vegetation survival and growth and few long term evaluations exist (Jorgensen et al., 2000; Pollock et al., 2005). A number of factors effect the growth and survival of riparian plantings including understory or overstory control and grazing by herbivores. Sweeney et al. (2002) examined the success of different oak (Querus spp.) species over four years under various riparian treatments and silviculture techniques. Protection from herbivores was the most important factor in determining oak seedling survivorship. Emmingham et al. (2000) examined over 30 riparian projects designed to convert deciduous riparian areas to coniferous riparian forests in coastal Oregon and suggested that initially riparian silviculture treatments show promise at establishing conifers in hardwood-dominated riparian zones. However, lack of understory and overstory control, protection from browsing by deer (Odocoileus spp.), elk (Cervus elaphus), beaver (Castor canadensis), and mountain beaver (Aplodontia rufa) affected project success. Any rehabilitation activity that involves planting and monitoring vegetation should recognize that herbivores can undermine rehabilitation efforts (Emmingham et al., 2000; Opperman and Merenlender, 2000). Techniques for reducing animal damage to planted vegetation include fencing, plastic mesh cages, netting, and chemical sprays (Black, 1992).

FIGURE 6. Cartron Stream, County Sligoi, Ireland before (top left) and three years after fencing (bottom left) and Vernon Creek, Wisconsin before (top right) and fifteen years after removal of grazing (bottom right).

No thorough research on fish response to riparian planting exists and only a few studies have examined other instream biota. Penczak (1995), however, found that the fish species diversity increased in the Warta River, Poland, from 11 to 16 species as riparian vegetation regenerated after removal. A recent examination of riparian fencing and replanting in New Zealand showed that replanted riparian buffers showed improvements in water quality and channel stability, but nutrients and fecal contaminants responses were variable and no improvement in macroinvertebrate communities towards clean water varieties was observed (Parkyn et al., 2003). It suggested that larger or longer buffers were needed to affect changes in water temperature and water quality.

Evaluation efforts for invasive species typically focus on success of removal efforts and success of recolonization by native species (often cottonwoods and willows). Taylor and McDaniel (1998), Roelle and Gladwin (1999), Sprenger et al. (2002) and others have monitored and examined the success of various chemical, burning, mechanical, and hydrological treatments in removing tamarisk and the subsequent recolonization of native species. Application of a combination of these techniques in addition to cottonwood and willow plantings and timed irrigations have produced diverse riparian habitat (Taylor and McDaniel, 1998). Roelle and Gladwin (1999) were able to prevent the establishment of tamarisk seedlings at a rehabilitation site by regular, controlled fall flooding. For other invasive species such as Japanese knotweed (Beerling, 1991), which is known to disrupt riparian habitat throughout Europe and North America, there has been little in the way of monitoring to determine at what rate riparian habitat is being lost to the species.

Evidence from the few published studies on passive riparian restoration indicates that natural recovery rates are sometimes sufficient such that hands-on rehabilitation is not necessary and potentially could be counterproductive. Briggs (1996) describes the recovery of Arivaipa Creek, Arizona, after a 500-year flood destroyed most of the riparian vegetation. A massive recovery effort was implemented to accelerate recovery that included planting of thousands of cottonwood stakes. However, just 8 years after the flood, the riparian areas had recovered so robustly with natural vegetation that the artificial plantings could not even be found. Natural recovery of forests is also a common practice to restore seasonally inundated forests that are critical fish spawning and rearing habitat. For example, in Cambodia reforestation is used as an important fisheries rehabilitation technique to create fish spawning, rearing, and feeding areas in seasonally flooded forests adjacent to Tonle Sap (Great Lake) and the Mekong River (Thuok, 1998). Little long term monitoring of natural recovery exists, but it appears to be an effective tool under the right circumstances.

2.2.2.2 Effectiveness of fencing and grazing reduction

The various studies on the effectiveness of grazing reduction methods at restoring riparian areas have examined a broad array of different grazing systems, which makes drawing firm conclusions about the relative effectiveness of a management system other than removal of grazing difficult. Below we summarize what is known about grazing removal or reduction at improving riparian condition, physical habitat, and aquatic biota. This sequence also represents the chronological order in which systems recover following reduction in grazing. When possible we provide examples of effectiveness of different grazing reduction methods.

Riparian conditions - The relatively rapid improvement (5-10 years) of riparian vegetation and riparian functions such as shade, sediment storage, and hydrologic effects (i.e. water storage and aquifer recharge) following livestock exclusion or dramatic reductions in grazing intensity have been documented in several studies (Elmore and Beschta, 1987; Myers and Swanson, 1995; Clary et al., 1996; Kauffman et al., 1997; Clary, 1999; O’Grady et al., 2002). Bank stability, channel geometry, habitat complexity, and other channel characteristics also recover quickly unless the channel is deeply incised (Elmore and Beschta, 1987; Myers and Swanson, 1995).

The most successful strategy for vegetation appears to be complete livestock removal. For example, in a comparison of currently grazed sites and sites where grazing had been removed from 2 to 50 years prior, Robertson and Rowling (2000) found understory vegetation and tree abundance were 1 and 3 orders of magnitude higher, respectively on ungrazed sites. The percentage of bare soil was lower, and the levels of fine coarse and fine particulate organic matter were higher at sites that were still being grazed. Similarly, Platts (1991) reviewed grazing strategies in the US, identified 17 riparian grazing systems, and indicated that light use and complete livestock exclusion provided adequate protection to riparian and fisheries resources. However, the success of rest-rotation grazing systems at allowing vegetation recovery is influenced by many factors including number of days of grazing, season, seasonal livestock dispersal behaviour, and level of compliance (Myers, 1989). Myers (1989) reviewed grazing systems in Montana (USA) and reported that rest-rotation grazing systems that allowed for successful recovery of vegetation averaged 28 days in duration, while unsuccessful averaged 59 days. Platts and Nelson (1985) examined rest-rotation grazing and found compliance during scheduled rest periods was difficult to achieve and cattle grazed less in the riparian areas during the early part of the growing season when high quality upland vegetation was abundant. Historically, grazing systems have not differentiated between riparian and upland range areas (Clary and Webster, 1989), but grazing systems that control the intensity and timing of use of riparian and upland areas are also thought to be beneficial (Elmore, 1992). Spring grazing, for example, can result in equal distribution of stock between riparian and upland areas and maintain herbaceous stubble heights that adequately protect erodable stream banks (Clary and Webster, 1989). Rest-rotation and other seasonal grazing strategies have shown promise at protecting riparian and aquatic habitat when coupled with intensive monitoring (Myers and Swanson, 1995), but these strategies need additional evaluation for effectiveness (Clary and Webster, 1989).

While grazing by native ungulates may have historically degraded some riparian and stream areas, there is evidence that ungulates can help maintain diversity of riparian vegetation (Medina et al., 2005). In a recent study in Scotland, Humphrey and Patterson (2000) found that reintroduction of cattle grazing increased plant community diversity. Conversely, the exclusion of excessive grazing by native herbivores can also assist in the recovery of riparian vegetation. For example, Oppermann and Merenlender (2000) found significantly higher levels of woody plants (Salix spp.) along reaches of a California stream where blacktail deer (Odocoileus hemionus columbianus) had been excluded with fencing. Moreover, the exclusion of livestock may be followed by an increase in native ungulates, which can also negatively impact efforts for recovery of riparian areas and the stream channel (Medina et al., 2005). These studies suggest that in some case limited grazing by livestock can have some positive effects on riparian vegetation, while grazing of wild ungulates can have some negative consequences.

Physical habitat and channel conditions-Instream habitat conditions recover slower than many riparian conditions (e.g. shade, plant growth); however, several studies in arid environments have reported positive effects of complete grazing removal or fencing on bank stability and in channel features (Platts, 1991; Medina et al., 2005). Myers and Swanson (1995) found that both complete livestock exclusion and rest-rotation grazing increased bank stability, tree cover, and pool habitat, though other management activities (e.g. road crossing, removal of coarse woody debris) negatively impacted instream conditions. Clary et al., (1996) and Clary (1999) reported improved width/depth ratios and substrate embeddedness following complete removal and reduction grazing in studies in both Oregon and Idaho. Connin (1991) reviewed several riparian rehabilitation projects that involved both bank protection and fencing or removing of grazing in the western United States and found in general an increase in bank stability, riparian vegetation growth, and improvements in channel conditions. If livestock are excluded from only a portion of the riparian zone, the width of the exclusion or buffer has been demonstrated to be positively correlated with the level of fine sediment retention (Hook, 2003). These studies demonstrate that grazing systems may also lead to recovery of stream banks, particularly if historic grazing was extremely heavy. Complete grazing removal appears to allow for better recovery of other in-channel factors such as width to depth, channel entrenchment, bank angle, and fine sediment.

Biotic responses-Relatively few published studies exist that examine the effects of grazing reduction on fishes and other aquatic biota (Rinne, 1999). Our understanding of the effects of grazing and grazing reduction on fish populations has been developed primarily through examining the effects of grazing on stream habitat characteristics (Platts, 1991; Clary, 1999). Much of the information is based on studies comparing sites with different levels of grazing and habitat quality and fish numbers, which generally show higher stream temperatures and lower fish numbers in grazed than ungrazed sites or watersheds (e.g. Li et al., 1994; Myers and Swanson, 1991; Platts, 1991; Wu et al., 2000).

Most of the studies on grazing impacts on biota have occurred in the western United States where large tracks of public land are managed for livestock grazing. Platts (1991) and Rinne (1999) conducted thorough reviews on grazing in the western United States and indicated that while some studies demonstrated increases in fish numbers, fish and aquatic biota were rarely the focus of the studies. Rinne (1999) found that most studies were of inadequate experimental design or duration to effectively determine significant changes in fish numbers due to removal or reduction of grazing. In a study in Oregon streams, Kauffman et al. (2002) reported improvements in vegetation, stream morphology, and densities of young-of-year rainbow trout (Oncorhynchus mykiss) but not adult or juvenile trout following livestock exclusion. They suggested that the lack of response of juvenile and adult fishes was because of the short reaches sampled in their study. Medina et al. (2005) reported on three long-term case studies in the American Southwest and found inconclusive results in all three cases due to limitations in study design, species interactions, upstream or watershed-scale effects, limitations of study design, introduction of exotic species, and fisheries management (stocking and changes in fishing regulations).

Few published studies have examined reduction in grazing on fishes or biota in other parts of the world. O’Grady et al. (2002), for example, reported an increase in Atlantic salmon (Salmo salar), brown trout (Salmo trutta), and minnow (Phohincus phohincus) numbers following stabilization of banks and fencing in an Irish basin impacted by severe overgrazing. Parkyn et al. (2003) found that fencing and planting of riparian buffers produced rapid improvements in water quality and channel stability, but had no detectable effect on macroinvertebrate fauna in New Zealand streams. Similar to North American studies, they indicated that upstream, watershed-scale factors and short buffer reaches likely influenced results for biota. The results of grazing studies from throughout the world emphasize the need for long term, well designed, watershed-scale monitoring that considers many external factors when evaluating grazing reduction.

Other factors-Reduction in grazing can also influence water quality, fish diet, and other biota. In a rare watershed-scale study, Meals and Hopkins (2002) reported that riparian fencing coupled with other bank protection actions reduced P levels up to 20 percent. Similarly, Sovell et al. (2000) found lower water quality at continuously versus rest-rotation grazed sites. Removal of grazing can lead to improvements in avifauna most likely through change in riparian community and associated with change in water table (Diaz et al., 1996; Dobkin et al., 1998). Grazing can also affect fish by changing the plant community, which affects fish prey, and in turn fish diet, as demonstrated by Laffaille et al. (2000) in a study in a French marsh.

2.2.3 Conclusions - riparian silviculture and fencing and grazing reduction

Different grazing management strategies and silviculture treatments to restore riparian areas and improve fish habitat are common and effective techniques with obvious benefits to riparian vegetation and less obvious benefits to instream factors. Few published evaluations have examined instream factors and many factors can influence the success of different techniques at improving riparian and instream conditions. We provide the following recommendations based primarily on the few published evaluations discussed above and experience.

• The success of planting depends upon a number of factors including site conditions, competition, grazing and browsing, and exotic species.

• Passive restoration of riparian areas may be an effective technique if the disturbance can be removed and invasive species do not compete with native vegetation.

• Complete exclusion and fencing of livestock have shown the most promising results in terms of vegetation, bank, and instream characteristics.

• Rest-rotation or other grazing systems have shown promise under proper management and the right physical, morphological, and climatic conditions.

• Response of stream channel and biota to grazing and riparian silviculture treatments may lag behind vegetation recovery.

• Response of aquatic biota and in-channel conditions, while rarely examined, are influenced by upstream or watershed-scale features and thus may not respond to changes in riparian conditions.

• Several factors may influence success of grazing and riparian rehabilitation techniques including geology, channel type, climate, exotic species, native ungulates, effective control of grazing intensity and duration, size of exclusion or buffer, and upstream processes or impacts.