2.3 Floodplain rehabilitation^[2]

Floodplain wetlands, lakes, and other aquatic habitat support large fisheries and aquaculture operations in many parts of the world (Welcomme, 1985; Thompson and Hossain, 1998; Hoggarth et al., 1999; Ghosh and Ponniah, 2001; Welcomme and Petr, 2004a,b). The channelization of rivers for industrial, urban, agricultural, and other human uses has lead to a decline in fish production in many floodplain areas (Bayley, 1995; Modde et al., 2001; Pretty et al., 2003; Miranda and Lucas, 2004; Welcomme and Petr, 2004a,b). Habitats such as wetlands, sloughs, and oxbow lakes were once common in large river floodplains, but are now disconnected or rare in many river systems because of channelization and other human activities (Sowl, 1990; Galat et al., 1998).

In the last decade, the rehabilitation of large rivers, and reconnection of isolated floodplains and their associated habitats in particular, has become a critical component of river ecosystem rehabilitation in developed countries (Holmes and Nielsen, 1998; Sear et al., 1998; Buijse et al., 2002; Florsheim and Mount, 2002) and more recently in developing countries (Gopal, 2004). Techniques used to reconnect river-floodplain systems and rehabilitate larger streams (> 20m bankfull width) are at the early stage of development (Cowx and Welcomme, 1998), though they have recently been the focus of study on the Danube and Rhine rivers (Buijse et al., 2002) and other rivers in Europe (Iversen et al., 1993; Lusk et al., 2003). Many of the river systems where such rehabilitation efforts have taken or will take place are heavily engineered systems that rely on human intervention to be maintained in a specified state (Bayley et al., 2000; Buijse et al., 2002). Thus many projects and subsequent techniques have been and will be a compromise between allowing natural processes to function and engineering solutions (Cowx and Welcomme, 1998).

2.3.1 Overview of floodplain techniques

The objective of most floodplain or large river rehabilitation efforts is to restore or improve lateral (connection with floodplain), longitudinal (upstream-downstream), and occasionally vertical connections (hyporheic connections). Some of the most common rehabilitation techniques include connection of isolated floodplain habitats (e.g. side channels, ponds, wetlands, excavated ponds), levee removal or setback, and construction of a new channel or realignment or meandering of existing channel or creation of new floodplain ponds (Table 7; Figure 7 and 8). Other techniques include dam removal, restoration of flood flows, and rehabilitation of riparian areas, which are covered in other sections of this document. Floodplain habitats include numerous types of channels, ponds, lakes, and depressions. These areas provide valuable overwintering habitat for fishes that prefer lentic habitats (Peterson and Reid, 1984; Swales and Levings, 1989; Cowx and Welcomme 1998; Buijse et al., 2002). Moreover, in many cases floodplain habitat rehabilitation is recommended over instream habitat improvement in larger channels (>12m bankfull width) and low gradient channels (<2 percent) (Dominquez and Cederholm, 2000; Roni et al., 2002; Pess et al., 2005). In the following subsections we describe various techniques and what is known about their effectiveness. We distinguish between wetland rehabilitation and floodplain rehabilitation by focusing on habitat rehabilitation activities for fish while wetland rehabilitation can include restoration of plants, birds, mammals, and other vertebrates and invertebrates and often focuses on habitat not directly connected with a stream. We view all of these as important goals of habitat rehabilitation, but focus our discussion on fish habitat, rather than purely wetland science (see Kusler and Kentula, 1989, Wheeler et al., 1995, and others for more information on wetlands).

2.3.1.1 Connecting isolated habitats

The level of connectivity of floodplain habitats has a direct effect on fish species utilization and composition (Berrebi-dit-Thomas et al., 2001; Miranda and Lucas, 2004). Reconnecting isolated habitats through weir removal, levee breaching, or replacement of a culvert or stream crossing to reconnect a relic channel or larger floodplain water bodies, such as a pond or lake to the main stem, is one of the more common techniques. Notched dikes, culverts, submersible dikes, and controllable flow gates can be used to reconnect existing relic channels in areas where flow needs to be regulated (Lister and Finnegan, 1997; Cowx and Welcomme, 1998). Two common problems that occur with such projects are: 1) elevation differences between the mainstem and relic channels due to mainstem channel incision from the time of floodplain channel disconnection, and 2) water quality degradation in the floodplain. For example, rehabilitation opportunities identified in the Stillaguamish River in Washington State (USA) include the reconnection of two former meander bends that are now floodplain sloughs (Pess et al., 1999). However, the river channel incised 1 to 2 m between the period when the bends where cut off in 1929 and 1991 making reconnection of the former meanders difficult (Collins et al., 2003). This problem is common to many stream channels in Europe, North America, and throughout the world.

TABLE 7
Floodplain rehabilitation techniques, modified from Pess et al., (2005). Barrier and dam removal are discussed in sections 2.1 and 2.4, respectively.

Submersible check dams are technique used to raise the bed elevation of rivers with incised mainstem channels and reconnect them with their former channels and floodplain. For example, in the Danube River in Slovakia, check dams in lateral artificial channels have been used to aggrade the river to the point where older channels are now reconnected, improving the retention time of water in the reach (Cowx and Welcomme, 1998). On the Kissimmee River, Florida (USA), channel filling has been used to reconnect old meanders and floodplains (Toth et al., 1993; Figure 8). The main channel has been filled from levee material at points in the river where the meander crosses the main channel. This method only works if the floodplain or some of its key habitats remains intact (Cowx and Welcomme, 1998).

FIGURE 7. Photographs of some common floodplain rehabilitation techniques including bank protection and removal of bank protection (Photo A) straightened reach, Photo B downstream restored reach, River Aggar, Germany), active gravel mining operation (Photo C) and reclaimed gravel pit on the Rhine River, Germany (Photo D), reconnection of former oxbow or side-channel (Photo E, Rhine River, fishway in foreground), and excavation of new off-channel habitat (Photo F, River Drau, Austria).

FIGURE 8. Example of Kissimmee River before channelization (top right), after channelization (top left), and following restoration of channel meanders (bottom right).

2.3.1.2 Levee breeching and setbacks

A levee breeching or setback project, whether full or partial, allows a river to migrate and subsequently create and maintain different floodplain channel types, thereby increasing the habitat diversity of a floodplain. The relative extent of such rehabilitation actions is often limited by land constraints and thus many such projects are a combination of full and partial levee removal. One technique is a "beaded approach" to floodplains, where small sectors of the full floodplain width (e.g. 6 to 8 km long) are allowed to function, alternating with leveed sections of the river. In these full floodplain width sections, several habitat features are included such as floodplain channels and wetlands. This "family" of habitats will allow the river to function properly and possibly create sediment, flow, and nutrient pulses (Sparks, 1995). This technique allows portions of the floodplain to be inundated and encourages scour, erosion, and deposition in those areas. Such a technique can be used in areas where natural channel migration is normally high such as the confluence with tributary junctions.

2.3.1.3 Channel reconstruction or meandering

Depending on the stream power and sediment load and composition, a channel may revert back to a meandering channel if left undisturbed for many decades (Brookes, 1992). However, most floodplain channels tend to be low energy and require direct intervention to be restored to a meandering channel connected with its floodplain. Therefore, one of the more common techniques for rehabilitating rivers and reconnecting them to their floodplains is to construct a new meandering channel (Cowx and Welcomme, 1998; Pess et al., 2005). This technique is particularly common in Denmark and northern Europe where most rivers are low gradient, highly channelized, and unlikely to return to their historical sinuous channel type without intervention (Brookes, 1992; Hansen, 1996). Restoring the sinuosity of a stream is done by either 1) pulling back levees and constructing a new meandering channel (rather than allowing the stream to naturally form its own channel as with a levee set back or removal project), 2) constructing a meandering channel adjacent to the straightened channel and then diverting the river into the new channel, or 3) if remnants of the original meandering channel exists adjacent to the straightened channel, diverting the straightened channel into its old channel, or 4) some variation of these three approaches. Many smaller channels have been straightened and piped and "daylighting" or exposing these concealed watercourses and then restoring natural remeandering pattern is an increasing common technique in urban and agricultural landscapes (Nielsen, 1996; Riley, 1998). Sometimes in highly incised channels it may not be feasible to aggrade the channel and reconnect it with the historical floodplain; a new floodplain is created by excavating the banks and widening the channels (Cowx and Welcomme, 1998). This is often coupled with a two-stage channel design, which incorporates a low flow channel and a new high-flow channel or new floodplain above the low flow channel but below the historic floodplain (Iversen et al., 1993). All these methods seek to reconnect the river with its floodplain, increase the diversity of habitats (mixture of slow and fast water habitats), and lengthen the stream channel (reduce the gradient).

2.3.1.4 Construction of new floodplain habitats

The creation of new floodplain habitats is a form of habitat enhancement that involves active construction of new floodplain channel, ponds, or backwaters. Creation of new habitats can be an important form of enhancement, replacing natural floodplain channels that have been lost owing to flood control activities and other floodplain-isolating actions (Beechie et al., 2001). There are two major habitat types that are typically created in river floodplains: side channels (surface and groundwater-fed) and backwaters or off-channel ponds (e.g. gravel pits, mill ponds, mine dredge ponds, and alcove ponds).

Creation of surface or groundwater-fed side channels are constructed to provide spawning habitat for adult fishes and rearing habitat for juveniles (Lister and Finnegan, 1997). They are typically connected at both the upstream and downstream ends to the main river channel, with an intake structure on the upstream end. Due to the surface water source, there is flexibility in choosing a project location. However, along with this water source comes sedimentation problems. Because of siltation concerns, this type of channel may be unsuitable for systems with high suspended sediment loads. Also, the location of the channel intake is critical for controlling sediment and organic introduction into the channel. Typically, a surface-fed side channel requires a dike or control structure and bank armouring at the upstream end to protect the channel from river flooding and bank erosion (Lister and Finnigan, 1997).

If the water table is low, groundwater-fed side channels can also be excavated. Groundwater-fed channels offer stable year-round water flows with little sediment, clean substrates and stable water temperatures (Sheng et al., 1990; Bonnell, 1991). Groundwater channels are excavated parallel to the main river channel, often along an existing intermittent stream or relic channel. Many of these side channels were originally constructed as channels with nearly uniform depth and substrate to serve as spawning habitat for salmonids (Sheng et al., 1990 from Giannico and Hinch; Giannico and Hinch, 2003).

Off-channel ponds can be created through several excavation or construction methods including excavation, blasting, damming or flow control structure on an existing perennial or intermittent stream, pond, or wetland (Cederholm et al., 1988; Slaney and Zoldakas, 1997; Cowx and Welcomme, 1998). The construction of alcoves or backwaters directly connected and adjacent to the river is another technique that is popular in both North America and Europe (Slaney and Zoldokas, 1997; Cowx and Welcomme, 1998). Alcoves have been constructed through excavation of a pond directly adjacent to the river channel. Other forms of constructed habitats include the construction of new side channels (surface or groundwater fed), excavation of backwaters or alcoves adjacent to the main channel, and excavation of ponds and other wetlands. Regardless of how the ponds or channels are constructed, it is important to ensure that these constructed habitats are connected to the river as least seasonally, but preferably consistently throughout the year. However, floodplain ponds in developing countries such as Nigeria are sometimes purposefully excavated so that water and fish will be isolated when floodwaters recede-thus providing fishing opportunities (Neiland and Ladu, 1998).

Floodplain mining and other floodplain excavation activities have resulted in the creation of gravel pits, mill ponds, and mine-dredge ponds along river channels (Norman, 1998). The restructuring and connection of these ponds with the river channel provides opportunities for additional floodplain habitats. These areas can provide important rearing habitat for a variety of fishes that prefer slow water habitats, though in some areas they are thought to provide refuge from predators and exotic species (Bryant, 1988; Swales and Levings, 1989; Richards et al., 1992; Norman, 1998; Hall et al., 2000). The successful conversion of gravel pits, mill ponds, and other constructed floodplain lakes depends on several factors including depth, morphology, adequate prey resources, cover, water quality, and adequate water quality (Norman, 1998). Shallow ponds with complex shapes provide the best opportunities for successful use by juvenile fishes, as large and deep ponds will provide less prey, as well as substantially higher risk of avulsion (Peterson, 1982; Norman, 1998).

2.3.2 Effectiveness of floodplain rehabilitation techniques

2.3.2.1 Isolated habitats

Similar to other rehabilitation techniques, the effectiveness of floodplain rehabilitation can be measured through recovery of processes (sediment transport, hydrology, connectivity), and physical and biological response. The rehabilitation of connectivity and physical and biological response have been frequently examined, though the affects on watershed processes have infrequently been measured. In terms of restoring connectivity and improvements in physical habitat, the benefits are obvious and can be measured via the amount of physical habitat connected. For example, in the Danube River the reconnection of a former side channel and canal in resulted in an additional 50 km of habitat. Reconnected floodplain ponds has proven to be effective at providing habitat for juvenile salmonids such as coho (Oncorhynchus kisutch) and Chinook (O. tshawytscha) salmon (Richards et al., 1992; Norman, 1998; Roni et al., 2002) in western North America. They are also known to provide critical rearing habitat for a number of other fishes including cyprinids (minnows), catastomids (suckers), and many other warm and coolwater fishes. Schmutz et al., (1994) reported the colonization of a reconnected section of the Danube River and found over 40 species of fish after only one year. In the lower Rhine River, reconnection of floodplain lakes and channels lead to an increase in rheophilic cyprinids (Grift et al., 2001; Simons et al., 2001). The level of flow and connection of habitats to the main Rhine was positively correlated with rheophilic cyprinid presence and abundance (Grift et al., 2001). Similarly, the reconnection of a side channel/canal in the Danube allowed rapid colonization by many species, though pike-perch (Stizostedion lucioperca; an important sport and commercial fish) did not respond as favourably as other species due to their aversion to bypass channels around weirs (Schmutz et al., 1998). Rapid colonization is not altogether surprising given that many freshwater fish species are entirely or partially dependent upon floodplain habitats for all or portions of their life history (Mann, 1996; Welcomme, 1985; Cowx and Welcomme, 1998; Buijse et al., 2002).

In Bangladesh and India, reconnection of secondary channels and floodplain lakes has been shown to be an effective practice to increase fish catch (Thompson and Hossain, 1998; Rahman et al., 1999). For example, the reconnection of Singharagi Beel lead to an increase in annual fisheries yield from 1 863 kg/ha to 11 384 kg/ha (partly because of increased effort) and the percentage of catch composed of large migratory catfish and carps increased from 2 to 24 percent (Payne and Cowan, 1998). While not often evaluated, reconnection of isolated habitats is believed to be one of the more effective techniques from both a physical and biological standpoint, partly because it relies on existing habitat rather than enhancing habitat or creating new habitat (Roni et al., 2002; Pess et al., 2005). Studies from various developed and developing countries support this hypothesis.

2.3.2.2 Levee breaching

Most of the information on project effectiveness for levee breaching and setbacks has focused on physical aspects and the lateral hydrologic connectivity of habitats. The removal of levees and bank armouring allow the channel to migrate naturally and recover its former sinuosity and is becoming increasingly common. Early results suggest these projects lead to successful improvements in floodplain processes such as nutrient transport and lateral erosion. Change in nutrient transport levels have been reported for levee removal projects that allow reconnection with wetlands (Childers et al., 1999a,b). Florsheim and Mount (2002) demonstrated that levee breaching on the Cosumnes River, California, allowed for the successful restoration of floodplain features such as sand-splays complexes. Evidence from ongoing studies in Austria indicates that following levee removal river channels begin to move laterally and recover some sinuosity and habitat complexity fairly quickly (Jungwirth et al., 2002; Muhar et al., 2004). Levee set backs and modifications on the Danube River have also been shown to benefit not only rheophilic fishes, but also amphibians and dragonflies (Chovanec et al., 2002). Moreover, Hein et al., (1999) demonstrated increased plankton biomass in reconnected habitats and that plankton production declined as connectivity of habitat decreased. These and other Austrian studies have demonstrated not only improvements in physical habitat and restoration of natural erosional and channel migration processes, but also improvements in both fish and riparian diversity and age structure (Jungwirth et al., 2002).

2.3.2.3 Remeandering

The effectiveness of remeandering can be measured in part by increase in the total stream length as the amount of total stream length can increase dramatically by reinstating meanders. In large rivers, the amount of length lost due to straightening can be dramatic. For example, it is estimated that channelization of the Mississippi River in the United States reduced stream length by 229 km (Gore and Shields, 1995). Hvidsten and Johnsen (1992) reported that channelization of a sinuous reach of a Norwegian stream reduced stream length from 7.5 to 2.5 km. The increase in stream length by remeandering can be equally impressive. For example, in a review of Danish stream rehabilitation, Iversen et al. (1993), reported increases in stream length ranging from 17 percent to more than 60 percent for five river meander reinstatement projects. Clearly restoring stream and floodplain meanders dramatically increases total available habitat and habitat diversity.

The remeandering on the Brede, Cole, and Skerne rivers in Denmark and England is one of the more thorough ongoing evaluations of stream remeandering. Evaluation of these projects indicated an obvious improvement in habitat complexity and channel morphology, flood frequency, and amount of water passing onto the floodplain (floodplain connectivity), as well as an increase in sediment deposition and sediment-associated phosphorous (Kronvang et al., 1998; Sear et al., 1998). Studies of macroinvertebrates, fish fauna, and aquatic vegetation in the River Gelså and other remeandered Danish streams have shown some positive results (Iversen et al., 1993; Hansen, 1996; Friberg et al., 1998). Improvements in both physical habitat and fish species diversity have also been reported from Austrian streams (Jungwirth et al., 1995).

The positive benefits of restoring or recreating meanders on stream morphology and hydrology and sediment transport have been documented. Similarly, recovery of macroinvertebrates and other biota from construction impacts appears to be relatively rapid. However, long-term studies conclusively documenting recovery or increase in fish production have not yet been completed. As with many structural manipulations of physical habitat, if potential water quality problems are not addressed, they can impair biotic responses to floodplain habitat rehabilitation.

2.3.2.4 Constructed habitats

The connectivity of the habitats to the main river channel plays a large role in their physical effectiveness. If the habitats remain connected through a variety of flows they are much more likely to be used by a variety of fish life stages. Constructed floodplain habitats have been shown to be particularly effective at both providing habitat for juvenile salmonids and increasing their survival (Lister and Bengeyfield, 1998; Solazzi et al., 2000; Giannico and Hinch, 2003). For instance Solazzi et al. (2000) concluded that creation of slow-water floodplain channel habitats increased overwinter survival for coho salmon, cutthroat trout (Oncorhynchus clarki), and steelhead (O. mykiss). Chovanec et al. (2002) reported rapid colonization by rheophilic fish species (e.g. Chondrostoma nasus) of constructed backwaters and side channels created on the Danube River in Austria. Similarly, Langler and Smith (2001) found age 0 coarse fish numbers and diversity higher in created backwater habitats than unmanipulated habitats in the Huntspill River, England. Habersack and Nachtnebel (1995) found that a constructed side channel of the River Drau, Austria, had higher diversity of habitats, substrates, macroinvertebrates and higher densities of fish in the side channel than in other river reaches. Excavation of groundwater channels is a particularly popular technique for creating spawning habitat for salmonid fishes (Bonnell, 1991; Cowan, 1991; Hall et al., 2000). Similar to surface-fed side channels, groundwater-fed channels also provide rearing habitat for juvenile fishes, particularly coho salmon (Bryant, 1988; Bonnell, 1991; Richards et al., 1992). These studies demonstrate that properly constructed floodplain habitats can provide important spawning and rearing areas for fishes.

2.3.3 Conclusions - floodplain habitats

Floodplain rehabilitation is a relatively new science and long-term studies documenting biological effectiveness are not currently available. In addition, the goals are typically broad ecological and cultural objectives. Thus trying to evaluate a purely fisheries response to a project is difficult. It is clear that most techniques described can lead to improvements in physical and hydrologic and other natural processes, provide additional slow water habitats, and additional habitat for fishes. Below we summarize some of the key findings and concerns.

• Reconnection of isolated floodplain habitats is particularly effective at improving habitat diversity, providing habitats for various fishes, and increasing species diversity.

• Levee removal, channel remeandering, and construction of floodplain habitats have all shown promising results both physically and for biota, but long term data on their success are not yet available

• Many factors influence the physical and biological effectiveness of projects including: habitat complexity, depth, wood volume, connectivity with the main channel, channel incision, flow volume and source, exotic species, project size, and upstream flow regulation and water quality.

2.4 Dam removal and flood restoration

2.4.1 Techniques for dam removal and flood restoration

Dam removal, weir removal, creation of fish passage or bypass channels, and restoration of peak flows in regulated rivers are employed primarily to restore longitudinal connectivity of stream and floodplain habitats, improve fish access, allow for natural transport of water, sediment, organic material and nutrients, and maintain or restore natural riverine processes that create and maintain fish habitat (Pess et al., 2005). While construction of dams and weirs is still occurring at a rapid pace in developing countries, many dams and diversion weirs in developed countries have reached the end of their useful life, and are being considered for removal or restructuring because of ecological concerns, as well as the safety concerns and costs of continued operation or repair of older structures (Stanley and Doyle, 2003). For example, in the last 20 years over 500 dams have been removed in the United States, mostly smaller dams less than 20 meters high with a storage area of less than 100 acre-ft (123 000 m³) (Poff and Hart, 2002; Stanley and Doyle 2003). Similarly, removal of weirs and low-head dams or installation of bypass channels and fish passage structures is one of the most common techniques for improving fish passage in Europe (Iversen et al., 1993; Cowx and Welcomme, 1998; DVWK, 2002) and other parts of the world (Larinier, 2001; Larinier and Marmulla, 2004).

Dams negatively impact aquatic resources by blocking migration of fishes, dramatically impacting watershed processes by changing sediment, nutrient, hydrologic and temperature regime downstream of the dam and by converting areas above the dam from a lotic to a lentic environment (Pejchar and Warner, 2001; Jackson and Marmulla, 2001; Pizzuto, 2002; Poff and Hart, 2002; Arthington et al., 2004). Most studies of dams in other regions have focused on the impacts of dams rather than removal of dams. Similar to dam removal studies, these studies demonstrate changes in physical conditions or biota (e.g. Grzybkowska et al., 1990; World Commission on Dams, 2000). The consideration of these and other negative ecological consequences of dams on aquatic resources, particularly migratory fish, have made the removal of both small and large dams an increasingly common method of rehabilitation (Chisholm, 1999).

Similar to dam construction, dam removal affects three areas-upstream of a dam reservoir, a reservoir/impoundment area, and below a dam. Biotic exchange between a river system below and above a dam increases after dam removal, as does the role of aquatic migratory species (e.g. salmonids) that may have been blocked because of a dam (Hart and Poff, 2002). The former reservoir or impoundment will experience dramatic changes in aquatic and riparian biota, flow characteristics, sediment erosion and deposition, and the channel is likely to be unstable in the first few years after dam removal. Dam removal typically increases downstream sediment supplies and changes flow patterns and result in significant changes in instream and riparian biota (Bushaw-Newton et al., 2002; Nislow et al., 2002; Shafroth et al., 2002). This suggests that the riparian environment will change dramatically as floods remove or change vegetation composition and create habitat suitable for more flood-tolerant species. Increases in sediment can generally be expected to raise bed elevation and increase lateral movement where there is room within the floodplain (Pizzuto, 2002; Shafroth et al., 2002). Such changes should result in more depositional bars and other newly created landforms that would favour early successional riparian species as well as create a diversity of aquatic habitats.

Another approach to restoring a natural hydrologic regime when dam removal is not feasible is the restoration of natural flood or flow regime. The regulation of river flows has lead to changes in sediment, habitat connectivity, biotic diversity, colonization of riparian plants, nutrient cycling and reduction in primary productivity (Osmundson et al., 2002). Many riparian species such as cottonwood become established under specific flow conditions, such as floods (Fenner et al., 1985; Auble and Scott, 1998). When floods are eliminated, the natural establishment of such species becomes problematic. Additionally, species that would not become established because of the scouring of floods appear in places they are not regularly seen. Flooding is also important for colonization of constructed riverine habitats by fishes. For example, in a constructed riverine wetland in the upper Mississippi River, fish species diversity increased dramatically following flooding (Theiling et al., 1999). Thus the timed release of waters from dams to mimic natural flows at certain times is becoming recognized as a valuable rehabilitation technique (Rood and Mahoney, 1993).

2.4.2 Effectiveness of dam removal and flood restoration

2.4.2.1 Effectiveness dam removal

Because dam removal is relatively new technique, there is not extensive published literature on its effectiveness, though many of the benefits are inherently obvious (e.g. fish access and passage). Below we summarize the results of studies that have examined the effects of dam and weir removal on connectivity, processes, habitat and biota below the dam, and processes, physical habitat, and biota immediately above the dam.

The restoration of upstream-downstream connectivity and fish access is one of the most clearly demonstrated effects of dam removal. Hart et al. (2002) summarized the results of several dam removal projects in the United States and reported more than 10 cases of dam removal that resulted in rapid colonization of former impoundment sites and upstream areas by both migratory and resident fishes in both warm and coolwater rivers. For example, dam removal on the Clearwater River, Idaho (USA) in 1963 reconnected the main stem, increasing both habitat quality and Chinook salmon runs (Shuman, 1995). Similarly, removal of 150-year-old Edwards Dam on the Kennebec River in Maine (USA) resulted in large numbers of American eel (Anguilla rostrata), alewife (Alosa pseudharengus), and Atlantic and shortnose sturgeon (Acipenser spp.) moving upstream within the first year, as well as juvenile downstream migrants in subsequent years (Hart et al., 2002). Smith et al., (2000) reported improved fish passage following removal of a 3 m high dam on a stream in Oregon (USA), but continued water withdrawal and other factors upstream of the dam prevented full recovery of both physical and biological conditions. Kanehl et al., (1997) also examined the effects of removal of a low head dam in the Wisconsin (USA) and found improvement in habitat quality, biotic integrity, and smallmouth bass (Micropterus dolomieu) abundance and biomass five years after dam removal. While not discussed extensively in this report, the installation of bypass channels to allow fish access above diversion weirs has also proven to be highly successful rehabilitation technique for fishes in European streams (DVWK, 2002; Iversen et al., 1993). Clearly dam removal has a number of benefits for migratory and lotic fishes.

Below a dam several major changes occur, the most obvious being a change in sediment and channel form due to a change in sediment flux. Several studies have demonstrated changes in sediment transport and fine sediment, but the changes in sediment depend upon the composition and levels of fine sediment trapped behind the former dam. For example, Doyle et al. (2003) examined low-head dam (<3 m high) removal in two Wisconsin rivers and erosion of fine sediment deposited in the former reservoir and increased deposition of fine sediment downstream. Hart et al. (2002) reviewed 20 dam removals in the US, 14 of which documented increased sediment transport, but few studies were long enough to document changes in the channel downstream of the dam. Other changes include a return to a more natural temperature regime, plant colonization, and a greater exchange of nutrients and organic matter with upstream portions of a watershed (Hart et al., 2002). Downstream effects from dam removal on ecological attributes ultimately depend on how reservoir-derived deposits move into and through downstream reaches (Stanley and Doyle, 2003). Changes in downstream water temperatures also typically occur following dam removal as do shifts in the macroinvertebrate community (Hart et al., 2002). Whether temperature increases or decreases following dam removal depends on several factors including the previous operations and structure of the dam and reservoir.

Chisholm (1999) reported anecdotal evidence on the positive effects of 25 dam removals in the United States. However, he also provided information on one dam removal (Fort Edward Dam on the Hudson River in New York State) that is considered a failure because it released heavy metals and pollutants trapped in reservoir sediments, which continue to have negative consequences on aquatic resources downstream from the former dam site. In addition, increase in turbidity and sediment immediately following and shortly after dam removal are not uncommon.

Former impoundments are affected by dam removal because they are returned to river, riparian, and floodplain habitats (Stanley and Doyle, 2003). This physical change reduces the residence time of water in a former reservoir reach, and subsequently reduces the amount of sediment and other materials stored within a reach. This in turn shifts the biota from a lentic to a lotic system (Hart and Poff, 2002). For example, fish and macroinvertebrates adapted to a high sediment supply reservoir environment gave way to riverine fish and macroinvertebrates within a year of two separate dam removal projects in Wisconsin (Stanley et al., 2002; Stanley and Doyle, 2003). In a related study on the Baraboo River, Wisconsin, Stanley et al. (2002) found that within 1 year macroinvertebrate assemblages above in formerly impounded stream reaches (reservoir) were similar to those in upstream and downstream reaches. Clearly dam removal will also result in changes in algae and other aquatic vegetation as well as riparian vegetation in the former reservoir sites (Hart et al., 2002; Shafroth et al., 2002). Shafroth et al. (2002) found that plants that colonized newly exposed soil along former reservoir bottoms were often composed of hydrophilic, monotypic species such as stinging nettle (Urtica dioica) and red canary grass (Phalaris arundinacea), over time sites became drier and dominated by xeric species. These studies demonstrate the dramatic changes in physical habitat and riparian and aquatic flora and fauna that occur following dam removal. They also suggest that there are some negative consequences of dam removal such as short-term channel instability or colonization of newly exposed riparian areas by invasive riparian species.

2.4.2.2 Effectiveness of flood restoration

Because restoration of flood flows is a relatively new technique, there is somewhat limited information on its effectiveness, though the results have been very positive. The most notable test case has been the attempts to restore flood flows in the Grand Canyon in the United States. Along the Colorado River in the Glen Marble and Grand Canyon reaches, forests and marshlands have become established in areas that were formerly scoured by floods (Stromberg, 2001). Results of high flow tests have show some promising results in this large river system that formerly had a very dynamic flow regime prior to regulation. Recent restoration activities that simulate floods have resulted in changes to the colonization of riparian areas that more closely mimic natural colonization patterns and reduced nearshore woody vegetation that artificially established as a result of flood control (Mahoney and Rood, 1990; Ellis et al., 2001; Stevens et al., 2001). Stevens et al., (2001) monitored the effects of flood simulations downstream of Glen Canyon Dam on the Colorado River and found that floods restored sandbar habitat and inundated patches of woody vegetation with as much as 1 m of sand. However, the woody vegetation was not eliminated and backwater marshlands that had established after dam construction also remained. Similarly, Hill and Platts (1998) monitored the effect of altered flows in the form of flood simulations and increased base flows in the Owens River, California. They found that riparian vegetation rapidly recovered following the changed flow regime and that instream habitat and fish abundance substantially improved concurrent with the changes in vegetation. Speierl et al. (2002) reported an increased in both species diversity and abundance following restoration of natural flow dynamics in the Main and Rodach rivers in Germany. Other studies have indicated that survival and growth of seedlings of some riparian plant species is higher under natural versus regulated flow conditions (Johansson and Nilsson, 2002). Similarly, restoration of instream flows can lead to recovery of riparian forests, birds, and fish populations (Rood et al., 2003). Studies on natural floods suggest the importance of these events for riverine ecosystems and support the use of flood restoration in regulated rivers. Large floods in the 1980s and 1990s on the Missouri River reconnected many wetlands and floodplain lakes isolated through decades of river regulation and channelization (Galat et al., 1998). Post flood studies indicated that reconnected areas had a more diverse fish fauna than areas that remained isolated (Galat et al., 1998). Similarly, high water years lead to reconnection of floodplain wetlands in the Green River, Utah (USA), and provided rearing habitat for endangered larval razorback sucker (Xyrauchen texanus) not available in the main river channel (Modde et al., 2001). These studies on natural floods and initial studies from restoration of flood flows suggest positive benefits for sediment transport, riparian vegetation, and aquatic biota.

Similar to increases in high flows, adequate instream flows or base flows, are needed to maintain aquatic and riparian habitat and production of aquatic ecosystems and biota (Petts and Maddock, 1996; Stanford et al., 1996; Annear et al., 2002; Arthington and Pusey, 2003). Increasing instream flows on rivers with water diversions or rewatering stream reaches are methods known to reduce water temperatures, improve water quality, and generally benefit biota (Weisberg and Burton, 1993; Petts and Maddock, 1996). There is an extensive body of literature on instream flows and it would be difficult to treat them adequately in a review of physical habitat rehabilitation techniques. However, along with restoring floods it is important that minimum instream flows are established to protect biota and assure the success of other habitat rehabilitation actions.

2.4.3 Conclusions - dam removal and flood restoration

Initial studies on dam removal and restoration of flood flows and a natural hydrology regime show promising early results at restoring access, natural processes, habitats, and recovering lotic biota. Based on the review of existing studies we provide the following recommendations:

• Dam or weir removal appears to be highly effective at restoring processes, though long-term monitoring of recovery is lacking.

• Fish and other migratory biota rapidly colonize areas upstream of dams.

• Some negative impacts such as fine sediment or water quality may be apparent depending on the level and contents of sediments stored in the reservoir and whether attempts are made to remove or stabilize them.

• In the absence of dam removal, restoration of flood flows has shown promising results in terms of restoration processes, reconnecting habitats, and restoring flood-dependent biota.