Geomorphology and Ecology

Key Elements

• McKenzie River channel morphology was evaluated using reach characterization, historical channel mapping, and specific gage analysis methods.
• The upper McKenzie River basin along Reach 1 is narrow, steep, and confined within a relatively narrow floodplain. The channel and floodplain width increase substantially along Reach 2. Exemplar species include spring Chinook salmon and bull trout.
• The middle McKenzie River basin (Reaches 3–8) is moderately steep and has a sometimes confined valley floor with narrow floodplains. Exemplar aquatic and terrestrial species include spring Chinook salmon, bull trout, Oregon chub, Pacific and western brook lamprey, western pond turtle, red-legged frog, cottonwood, and alder.
• The lower McKenzie River basin (Reaches 9–12) occupies a wide alluvial corridor that increases in width as it approaches the Willamette River confluence. Exemplar species includes those listed above.
• Bed-material transport has presumably decreased appreciably since the construction of Cougar and Blue River Dams, which are located on historically sediment-rich tributaries. Decreased sediment supply resulting from the dams likely contributed to a decrease in gravel bars between 1939 and 2005 in downstream reaches, particularly along the lower McKenzie River basin (Reaches 9–12). Previous studies have shown that the upper McKenzie River basin has very low sediment yield rates; therefore, Trail Bridge Reservoir probably has less of an effect on sediment transport than the Cougar and Blue River Reservoirs.
• Between 1939 and 2005, secondary channel features and sinuosity also decreased, particularly along the lower reaches.
• The historical channel changes observed for 1939–2005 are likely a function of streamflow reduction and bed-material trapping by the dams, but other influences, including bank stabilization, land use, channel clearing, and natural variability in streamflow regime also are important factors.

This section provides a background assessment of geomorphic, ecological, and biological characteristics within the McKenzie River basin and also discusses the effect of dams, bank hardening, and land-use changes. Instream physical habitat and riparian ecosystems are linked through a series of complex interactions. Within the McKenzie River riparian corridor, key areas of interest include pools, offchannel aquatic habitat, depositional bed features, and floodplains. The biological assessment draws primarily from Gregory and others (2007a), who identified key species and ecological processes in the Coast and Middle Fork Willamette basins that are applicable to the McKenzie River basin. For this study, nine exemplar aquatic and terrestrial species selected for the McKenzie River basin, including five native fish species (spring Chinook salmon, bull trout, Pacific and western brook lamprey, Oregon chub), red-legged frog, western pond turtle, black cottonwood, and white alder. These nine species were selected based on information in the literature and discussions with regional and local biologists.

Endangered Species Act Background

The Willamette Project Biological Opinion (National Marine Fisheries Service, 2008b) identified key threats and limiting factors in the McKenzie River basin to fish survival. These included (1) the effects of reduced peak streamflows on channel complexity, gravel and large wood recruitment, maintenance of riparian vegetation, and formation of pools, (2) the lack of gravel recruitment because of capture by upstream reservoirs, and (3) elevated water temperatures downstream of dams, resulting in premature hatching and emergence of salmonid fry.

The USACE Supplemental Biological Assessment report (U.S. Army Corps of Engineers, 2007) summarizes major issues regarding critical habitat for endangered species, particularly spring Chinook salmon. They highlight many of the same issues described above, including water temperatures downstream of the dams, lack of spawning gravel and coarsening of substrate downstream of dams, low recruitment of large wood, and the loss of side channels, islands, and habitat channel complexity. They also discussed other concerns, including (1) the loss of natural floodplain function (which has been most prominent along the lower parts of the mainstem McKenzie River and South Fork McKenzie River), (2) alteration of important seasonal streamflow events, (including flows during the fall spawning period, winter and spring floods, and flushing streamflows), and (3) the occurrence of unnatural streamflow fluctuations (caused by the high ramping rates of dam streamflow releases). Alteration of discharge and temperature, which are major cues for salmonids during different life stages, can affect the timing, growth, and survival of riverine salmonid life stages. An earlier USACE report (U.S. Army Corps of Engineers, 2000) also recommended that habitat improvement projects, streamflow targets, and ramping rates be established, researched, and monitored for spring Chinook salmon.

Methods

To better understand habitat requirements and ecological processes in the McKenzie River basin, the analysis draws largely on the Coast and Middle Fork Sustainable Rivers Project recommendations (Gregory and others, 2007b), the USACE Willamette Supplemental Biological Assessment (U.S. Army Corps of Engineers, 2007), and the Willamette Project Biological Opinion (National Marine Fisheries Service, 2008b). Together, these reports provide a historical summary and environmental baseline of the Willamette River basin as well as the current status of salmonid stocks and factors limiting their survival/recovery. Gregory and others (2007a) provide an overview of streamflow needs for riparian and floodplain vegetation, birds and mammals, aquatic invertebrates, amphibians, reptiles, and eight exemplar species of fish in the Willamette River basin. Discussions on critical issues and important species related to environmental flows from agency and university biologists who work within the McKenzie River basin also were incorporated.

A series of reconnaissance level geomorphic analyses were conducted to evaluate the role of intrinsic controls (those imposed by geology and physiography) versus external factors (including discharge, sediment supply, and large wood) on channel morphology. The geomorphology-related activities used in this study included reach characterization, historical channel mapping, and specific gage analysis.

Reach Characterization

Channel morphology along each of the 12 reaches was characterized by mapping the active channel from aerial photographs taken in 2005, where the active channel was considered the area inundated during annual high water and included the primary (wetted) channel as well as sparsely vegetated gravel bars and secondary channel features (side channels, alcoves). Mapping of the active channel was performed at a scale of 1:5,000 using publicly available digital orthophotographs taken in 2005 and produced by the U.S. Department of Agriculture National Agriculture Imagery Program (Oregon State University, 2009). The geologic floodplain, representing areas that were the floodplain during the Holocene, was digitized at a scale of 1:10,000. The geologic floodplain was compiled from previous mapping (O’Connor and others, 2001), from areas corresponding with the 100- and 500-year flood inundation zone (University of Oregon, 2009), and from the 10-m Digital Elevation Model (DEM). The valley floor, into which the floodplain is incised, is defined by Pleistocene sediments. This unit was mapped as Quaternary alluvium by Sherrod and Smith (2000).

Simple metrics describing sinuosity, bed slope, channel width, floodplain width, valley bottom width, and the degree of confinement imposed by valley walls were calculated from the channel and floodplain maps and from a longitudinal profile developed from the 10-m DEM (fig. 2). Descriptions of channel-level classification (for example, pool-riffle, plane bed), were taken from previous studies and compared against aerial photographs and field observations.

Historical Channel Mapping

Digital maps of the channel were produced from aerial photographs for two periods, 1939 and 2005, to document net changes in channel conditions during the mid to late 20th century. Channel morphology was mapped from 1939 aerial photographs acquired from the U.S. Army Corps of Engineers, Portland District, and then scanned and rectified to the 2005 orthophotographs. The 1939 photographs encompassed only Reaches 2–12 (from the McKenzie Bridge to the Willamette River confluence). The 1939 and 2005 aerial photographs were taken during summer at relatively low discharges, facilitating the mapping of low-lying bars, alcoves, and other features. Daily mean discharge at the USGS streamflow-gaging station at Vida (14162500) ranged from 2,100 to 2,300 ft³/s during the July to August dates of the 2005 photographs, and was slightly higher (3,100 ft³/s) on May 31, 1939, when the 1939 photographs were taken.

The photographs from 1939 and 2005 were used to map active gravel bars (defined as bars greater than 700 yd² in area, with less than 25 percent vegetative cover) and channel features. Channel features were characterized by digitizing centerlines along the primary channel, as well as secondary channel features, which included side channels and alcoves that were at least partially wetted and located within the active channel. Although other “dry” channels were evident in the aerial photographs, only channels that were at least partially wetted were mapped in order to definitively characterize offchannel habitat that would be inundated during high streamflows. This mapping protocol undoubtedly underestimates the actual length of offchannel habitat available during high streamflows, but it nonetheless provides a repeatable, screening-level indication of offchannel habitat availability in each of the study reaches. All mapping was conducted at a scale of 1:5,000 with a goal of characterizing broad-scale differences in the availability of gravel bars and offchannel habitat.

A major limitation to this reconnaissance-level analysis is our reliance upon two sets of channel maps. Ideally, aerial photographs from multiple, relatively narrow time intervals would be used so as to better link channel adjustment with specific drivers of channel change. For example, the role of floods could be evaluated by assessing pre- and post-flood photographs. Here, however, the net channel change is evaluated across a 65-year period in which the channel experienced numerous, overlapping, natural and anthropogenic disturbances. Major anthropogenic effects to the river basin between 1939 and 2005 include dam construction, bank stabilization, timber harvest, conversion of riparian forests to agriculture and development, and improvement of Highway 126 along the McKenzie River. Many floods of varying magnitude also occurred between 1939 and 2005, with several large floods occurring in the 1940s and 1950s; the largest recorded flood occurred on December 22, 1964, and a more recent, but much smaller flood event on February 7, 1996 (fig. 7).

The 1939 channel maps document a fluvial system that had already been subject to anthropogenic influences for nearly 100 years, although the magnitude and extent of these early effects (which included construction of the Leaburg and Walterville diversions as well as channel clearing, bank stabilization, timber harvest, and land conversion) probably had less of an effect on channel conditions than the larger-scale effects experienced in the mid to late 20th century. Furthermore, the 1939 channel characteristics were influenced by the numerous small floods that occurred in the 1920s and 1930s (fig. 7), and may also be a result of the exceptionally large regional floods that occurred in 1881 and 1890, and possibly even the largest known historical flood, which occurred in 1861.

Specific Gage Analysis

To assess the vertical stability of the channel bed at various sites in the study area, specific gage analyses were conducted at eight streamflow-gaging stations. The specific gage analyses followed the methodology of Klingeman (1973), using rating curves from USGS streamflow-gaging stations to plot the temporal variation in stage (water elevation) for a particular discharge. For each of the eight stations, five discharges were selected representing relatively low streamflows (75- and 95-percent exceedance), mid-range streamflow (50-percent exceedance) and high streamflows (5-percent exceedance and 2-year flood discharge). A measured stage for a particular discharge that is relatively constant over time indicates little net change in the cross-section geometry at a station, whereas large changes in the stage–discharge relation can indicate changes in cross-sectional geometry. When evaluated within the context of multiple discharges spanning a range of streamflow conditions and when the site characteristics are well documented, the specific gage curves can be used to infer potential deposition or incision and to indicate the overall susceptibility of the site to changes in channel geometry. The locations of streamflow-gaging stations are not necessarily representative of reach-scale conditions because these sites are selected for their relative stability. Hence, there is potential for a “false” observation of vertical stability, when an observation of instability should be “true.”

Factors Governing the Distribution of Key Habitat Types

The selection of key habitat types (pools, off channels, bars, and floodplains) in the analysis was based on the life history stages of the nine exemplar species that were used. A similar approach was used by Gregory and others (2007a) for the Coast and Middle Fork Willamette Sustainable Rivers Project study.

Pools

Pools are formed either freely by the interaction of channel form and streamflow that drive secondary currents and sediment transport or forced by obstructions such as bedrock, boulders, and large wood (singularly or as debris jams) (Montgomery and others, 1995). Accordingly, the primary factors controlling pool dimensions and frequency include streamflow magnitude, sediment load, large wood and other streamflow obstructions, channel geometry, and bed material size (Buffington and others, 2002). Pools generally increase in size with increasing discharge along the length of a river. Steeper, coarser grained reaches often have smaller pool volumes than lower gradient, finer grained reaches (Wohl and others, 1993; Buffington and others, 2002). Because pools typically are alluvial features scoured from bed material, they can be highly sensitive to sediment supply and bed-material characteristics. If sediment supply exceeds transport capacity, then aggradation may result in diminished pool depth (Lisle, 1982). However, a sharp reduction in sediment supply can reduce pool frequency and depth because pool geometry will be limited by the thickness of deposited alluvium (Montgomery and others, 1996). Furthermore, reductions in sediment supply and large wood can substantially reduce pool frequency because channel-spanning blockages of large wood and sediment are a dominant pool-forming mechanism. For example, Montgomery and others (1995) report that in forested streams up to 100 ft wide, 73 percent of the pools were forced by large wood, 18 percent were self-forming, and the remaining 9 percent were formed by boulders and bedrock outcrops.

Deep pools provide habitat; depths greater than about 2 ft provide cover for fish, amphibians, and mammals, and additional cover can be provided by large boulders, overhanging vegetation, undercut banks, or large wood. Large, deep pools with abundant cover are important habitats for many aquatic species and can significantly increase the carrying capacity of rivers and streams (Murphy and Meehan, 1991; Montgomery and others, 1995). Bjornn and Reiser (1991) cite literature showing that when large wood was removed from a stream, Coho salmon production declined coincident with a decrease in the number and size of pools, whereas the average water velocity increased.

Multiple studies have documented historical declines in the frequency and size of pools that can be related to reductions in pool-forming structures, including loss of large wood (Sullivan and others, 1987; Meehan, 1991). Channel clearing and snag removal for navigational purposes is well documented for the mainstem Willamette River (Sedell and Froggatt, 1984), but channel clearing by private landowners also occurred along the McKenzie River and its tributaries (Minear, 1994). As large wood was systematically removed from the river channel, riparian forests were repeatedly harvested, and downed and standing wood was selectively removed from proximal roads, which reduced wood recruitment. Inputs of large wood to the mainstem river were further reduced through the construction of dams along the Blue River and South Fork McKenzie River, and in the upper McKenzie River basin. The USACE flood control dams substantially diminish peak streamflows, which reduces opportunity for wood recruitment by way of bank erosion. Furthermore, the EWEB dams in the upper McKenzie River basin and USACE facilities trap large wood, which along with other factors, lessens the volume of wood entering downstream reaches (Stillwater Sciences, 2006b; U.S. Army Corps of Engineers, 2007; National Marine Fisheries Service, 2008b).

Pool depth can be diminished by sedimentation, which can be exacerbated by land management activities that include forest harvest and road building (Megahan, 1982). Numerous studies document channel aggradation in northern California streams following extensive hillslope erosion triggered by the December 1964 flood (Hickey, 1969; Kelsey, 1980; Lisle 1981). Such erosion and re-deposition can result in diminished pool–riffle morphology (for example, Lisle, 1982). However, the McKenzie River generally supports high sediment transport capacities relative to sediment supply (Lignon, 1991; Stillwater Sciences, 2006b); therefore, the potential role of aggradation on reducing pool complexity must be evaluated on a reach-by-reach basis by considering the local balance of sediment availability versus transport capacity, because in some locations, the regulated, post-dam streamflows might be insufficient to scour and maintain deep pools.

Aquatic Offchannel Habitat

Offchannel habitat, which includes side channels, alcoves, sloughs, and backwaters, are important habitat components for a large variety of species and life stages and is often critical for juvenile and rearing fish. These habitats provide refuge from high velocity waters, particularly during high streamflows in winter, and refuge from predators. They also provide important food resources because of their habitat complexity and their natural abundance of overhanging and riparian vegetation. Offchannel habitats are used extensively by Oregon chub, juvenile salmonids, amphibians, and turtles.

The suite of riparian landforms that comprise offchannel habitats are most abundant along unconfined, alluvial river corridors, where the channel is more likely to experience meander migration and avulsions, as well as creation and maintenance of side channels. Through frequent shifting, new channels are created. Older channels, progressively filled with fine sediment, become inundated only during high streamflows, and eventually they are abandoned. The degree of channel dynamism and the overall vegetation characteristics of the floodplain and riparian corridor dictate the availability and diversity of offchannel habitats. Like other gravel-bedded rivers in the Pacific Northwest, the primary drivers of channel adjustment on the McKenzie River include floods as well as upstream inputs of sediment and large wood, which form obstacles to streamflow and hence trigger bank erosion (O’Connor and others, 2003). Meander migration into forested floodplains introduces additional large wood and sediment into the channel, which further enhances channel shifting (Fetherston and others, 1995; Abbe and Montgomery, 2003; O’Connor and others, 2003).

Reaches that are most likely to support these processes, and hence also support a diverse array of offchannel habitat features, typically have unconfined active channels situated within a relatively wide forested floodplain, which maximizes opportunities for meander migration and side-channel formation. A wide alluvial corridor also helps to ensure that the channel is flowing through erodible bank materials that support high rates of meander migration and avulsions (Wallick and others, 2006). Artificial bank hardening through the construction of revetments can substantially diminish channel shifting and habitat complexity, particularly when large tracts of the alluvial corridor have been revetted (as documented on the Willamette River by Gregory and others, 2002a, and Wallick and others, 2006). Naturally occurring “hard” features such as bedrock outcrops can also exert a stabilizing influence when present for longer reaches (the scale of one or more bends), but when isolated bedrock outcrops are located within an otherwise alluvial reach, they can potentially contribute to overall habitat complexity by establishing forced pools (Montgomery and Buffington, 1997).

Healthy native riparian vegetation is an important component of offchannel habitat because vegetation has beneficial effects on water quality, provides cover for wildlife and fish, traps sediment, increases stream shading (thereby reducing water temperatures), and provides food, nutrients, and organic matter to the stream (Murphy and Meehan, 1991; Johnson and Buffler, 2008). Riparian forests also can supply large wood to the channel, which creates jams that help force channel migration and provides important physical cover for various species, especially salmonids.

Floods are another key factor contributing to channel dynamism and the creation of offchannel habitat. Along the upper Willamette River, small to moderate-sized floods, in which streamflow is mainly confined within the active channel, have historically triggered rapid rates of meander migration, whereas exceptionally large floods have supported high overbank velocities carving meander-cutoffs and triggering numerous avulsions (Wallick and others, 2007). Hydraulic modeling and historical channel mapping also have shown that large historical floods were capable of creating and scouring secondary channels along the floodplain and high surfaces within the active channel, whereas smaller, controlled, post-dam floods (such as the February 1996 flood) inundated floodplain swales, but were only able to scour side channels along low-lying, relatively erodible surfaces within the active channel (Wallick and others, 2007).

Alluvial Bed Features

A wide variety of bedforms composed of gravel to cobble size sediments are found along the McKenzie River and other gravel bed streams. These bedforms include a diverse array of bar features, such as point bars and midchannel bars, as well as riffles and pools. Gravel bars are important ecologically because they provide newly created areas for vegetation colonization, enhance channel complexity, and promote hyporheic exchange (Dykaar and Wigington, 2000; Fernald and others, 2001). Submerged bedforms, such as riffles, can provide habitat for benthic macroinvertabrates and serve as spawning beds for salmon and trout (Kondolf and Wolman, 1993). In western rivers and streams, the highest diversity and productivity in benthic macroinvertebrates usually is in fast-flowing riffles with gravel to large cobble-sized substrate habitats (Minshall, 1984; Allan, 1995), such as those in McKenzie River basin.

The frequency of gravel bars and other alluvial bed forms is largely dictated by the interaction between the supply of coarse bed material (sediment supply) and the ability of the channel to transport and redeposit this sediment (transport capacity). Previous studies have shown a decreasing number and size of gravel bars downstream of dams, which is attributable to trapping of coarse sediment behind the dams and stabilization of formerly active bars because of reduction in peak streamflows and vegetation encroachment (Nadler and Schumm, 1981; Williams and Wohlman, 1984). In addition, depositional features require a unique set of hydraulic circumstances; for example, point bars form along the inside of bends, whereas midchannel bars typically form at channel expansions, and frequently occur upstream of obstructions to flow posed by boulders, large wood, or semi-stable islands. Activities that result in channel simplification (such as removal of large wood and bank stabilization) may ultimately promote a straighter planform, which can enhance transport capacity and diminish opportunities for bed material deposition. Furthermore, by limiting bank erosion, channel stabilization reduces local recruitment of sediments that could be remobilized into gravel bars and spawning gravels (Lignon and others, 1995).

Bed substrate frequently coarsens downstream of dams, as gravels and fines are winnowed from the bed, leaving the bed armored with coarse deposits of larger gravels (Dietrich and others, 1989). Bed coarsening decreases habitat availability for salmonids as they require certain size ranges of gravel to create spawning beds (redds): the median diameter of spawning gravels for smaller trout is approximately 15 mm, whereas large salmon can utilize gravels up to 100 mm in diameter (Kondolf and Wolman, 1993). Another key factor determining the availability of suitable spawning habitat is the “looseness” of the cobbles and gravels composing redds. Fine sediment within the interstitial spaces can also negatively affect the diversity and production of benthic macroinvertebrates (Roy and others, 2003), decrease the supply of oxygenated water for eggs (Everest and others, 1987), and block fry from emerging from bed sediments (Tappel and Bjornn, 1983).

Floodplain Habitat and Riparian Vegetation

Floodplain surfaces are distinguished from the active channel by their age and relative elevation to the water surface. Whereas the active channel (and its various aquatic habitats) are inundated on an annual basis, floodplains represent a gradient of infrequently inundated, channel-adjacent surfaces constructed of sediment transported and deposited by the streamflow regime (Nanson and Croke, 1992). Along the lower McKenzie River basin, floodplain formation follows a similar trajectory as that of the upper Willamette River, whereby within-channel bar forms gradually aggrade and coalesce with older floodplain surfaces (Dykaar and Wigington, 2000). Along free-flowing alluvial rivers, a suite of geomorphic processes create a mosaic of different-aged surfaces at varying elevations, with each surface blanketed by distinct sediments according to its position relative to the channel (Ward and Stanford, 1983). The streamflow regime is superimposed upon this spectrum of surfaces, which can range from coarse gravel bars to rarely inundated floodplain swales, to establish a dynamic physical template for vegetation colonization and succession (Hupp and Osterkamp, 1985).

Channel shifting carves newly created surfaces out of older, established riparian plant communities, and hence increases patch heterogeneity. Multithread channels, such as in the historical lower McKenzie River basin, experience moderately frequent disturbances (in the form of meander migration and avulsions), and are more likely to maintain a mix of young and old surfaces, which in turn support higher levels of biological diversity than upslope habitats (Gregory and others, 1991; Beechie and others, 2006).

The habitats afforded by low-lying floodplain surfaces include wetlands, swales, and relict channels, as well as terrestrial communities composed of varying-age stands of riparian species (such as alder, cottonwood, and willow), whereas upland communities may be on the highest floodplain surfaces (Gregory and others, 1991). There has been extensive research on the numerous physical and ecological functions of riparian zones to their parent streams. For example, rarely inundated floodplain channels provide refuge for aquatic species during large floods, whereas low-lying, seasonal wetlands preclude the establishment of predatory species and provide important habitat for red-legged frogs (Kiesecker and Blaustein, 1998).

Factors that hinder floodplain formation and contribute to overall simplification of the riparian corridor include processes or direct actions that limit the connectivity between the active channel and floodplain, or minimize floodplain creation. For example, reduction of peak streamflows not only minimizes overbank deposition, but also reduces the processes that create and recycle floodplains (such as bar formation and bank erosion). Similarly, channel stabilization limits bank erosion, and contributes to planform simplification, which in turn slows the creation of bar forms that would eventually evolve into incipient floodplain. In addition to anthropogenic factors, there are inherent physiographic and geological controls on floodplain formation. Steep, highly constrained reaches support narrow floodplains, characterized by thin riparian communities which resemble upslope forests, whereas wide, unconstrained alluvial reaches are more likely to support broad, diverse floodplains with greater patch diversity (Gregory and others, 1991).

The life histories of riparian species depend upon the characteristic streamflow regime, that is., the timing and magnitude of streamflow, and hence can be negatively affected by streamflow regulation and diversions. Gregory and others (2007a) summarize the effects of altered streamflow on terrestrial vegetation in the Willamette River basin as follows:

• The magnitude of winter floods is reduced, which reduces floodplain inundation, sedimentation, and available bare soil for germination.
• The magnitude of summer low streamflow is increased, which can decrease regeneration and seedling survival by inundating low bars.
• The timing of the transition between winter high streamflow and summer low streamflows (recession) is shifted, which decreases survival of black cottonwood and other early successional-stage species.
• Alteration of streamflow magnitude and timing increases the potential for invasion of nonnative species, such as giant knotweed and reed canary grass.

Additional factors that diminish the abundance and diversity of riparian vegetation include land-use and channel stabilization. Vegetation maps produced by the 1851 General Land Office survey show that the McKenzie River floodplain historically supported an extensive deciduous forest composed of ash, red alder, big leaf maple, black cottonwood, white oak, and dogwood (Gregory and others, 2002b; dataset available from the Pacific Northwest Ecosystem Research Consortium (PNWERC) at http://www.fsl.orst.edu/pnwerc/wrb/access.html, last accessed July 30, 2009). However, the area of riparian forests along major tributaries of the Willamette River basin declined by two-thirds between 1850 and 1990. The dominant floodplain landcover change has been the direct conversion of riparian forests and wetlands to agricultural lands (Gregory and others, 2002b). Even along floodplain areas of the mainstem Willamette River that were not converted to agriculture or developed areas, riparian vegetation communities have become increasingly homogenous and have decreased patch diversity (Dykaar and Wigington, 2000; Gutowsky, 2000; Fierke and Kauffman, 2005, 2006).

Description of Channel Morphology and Habitat Limitation for Key Species

The dominant factors controlling channel morphology and physical habitat vary for each reach of the McKenzie River. As a result, some reaches may be more prone to channel adjustment and reaches may be more sensitive to potential imbalances between sediment supply and discharge. Moreover, channel morphology and factors governing physical habitat availability will dictate whether environmental flows alone can achieve specific habitat goals for a reach, or if other factors (for example, sediment supply, bank materials, or large wood) must also be considered.

Upper McKenzie River Basin—Reaches 1 and 2

Channel Morphology and Controls on Physical Habitat Availability

The McKenzie River between McKenzie Bridge and Trail Bridge Dam (Reach 1) is a highly stable, predominantly single-thread stream with very few gravel bars. The channel morphology is largely a function of the geological history of the upper McKenzie River basin. The young High Cascades lava flows support a groundwater-driven hydrologic system, which leads to muted flood flows and low sediment yield rates (Stillwater Sciences, 2006b). The broad, glacially carved valley bottom is approximately 3–4 times wider (about 1,500–2,000 ft) than the incised floodplain (about 500 ft), and is composed of coarse (bouldery) alluvium. Along Reach 1, the channel is narrow (reach average width is 141 ft), steep (rise/run slope is 0.0092), and confined within a relatively narrow floodplain that is only about three times wider than the active channel. The channel occupies a relatively straight and predominantly single-thread planform, with no visible gravel bars in the 2005 aerial photographs (table 20). Minear (1994) and Stillwater Sciences (2006b) determined Reach 1 to have plane bed morphology, characterized by featureless channel beds that lack bedforms (such as pool-riffle sequences or bars) and occurring at moderate to high gradients (Montgomery and Buffington, 1997). Channel and floodplain width increase substantially along Reach 2, particularly below the confluence of Horse Creek, a large tributary draining the Western and High Cascades terrains. The channel adopts pool-riffle morphology along Reach 2, where reach average width increases to 316 ft, and slope diminishes to 0.0067 (table 20).

Channel planform is stable along Reach 1, but becomes slightly more dynamic along the lower parts of Reach 2 (fig. 24). Although 1939 aerial photographs of Reach 1 were not available, Minear (1994) mapped historical channel change above McKenzie Bridge for the period 1949–1986 and detected little change in sinuosity and channel position, but did detect a 7,300 yd² increase in gravel bar area. A specific gage analysis shows that the channel elevation at the USGS streamflow gage near Trail Bridge Dam (14168850) has been very stable for the period 1959–2008, indicating that the channel near the gage is not prone to substantial incision or aggradation (fig. 25). The major changes in this time period are associated with the December 1964 flood, when 0.5 ft of deepening occurred, followed by about 0.5 ft of aggradation between 1968 and 1971. From 1971 to 2008, gradual lowering was less than 0.5 ft, apparent at low to moderate streamflows (580–1,700 ft³/s), with little variation in stage for high streamflows (4,600 ft³/s).

Along Reach 2 the channel has largely remained in the same overall position between 1939 and 2005 with the exception of three avulsions. The two largest of these occurred at historically dynamic areas downstream of the Horse Creek confluence where the floodplain widens and the channel adopts a more sinuous path. An avulsion at Dearborn Island (RM 63) occurred during the December 1964 flood, whereas the avulsion at Delta Campground (RM 61) is associated with the 1986 high streamflow event (Minear, 1994). Although the avulsions decreased the centerline length, they increased the length of secondary channel features as both sites continue to experience meander migration and channel shifting. More side channels are in the 2005 aerial photographs than in the 1939 photographs (fig. 26A).

Deep pools in the upper McKenzie River basin typically form where the channel either impinges upon valley walls (which is the dominant pool-forming mechanism along Reach 1) or as a result of scour related to midchannel accumulations of sediment and wood (as observed at Delta Campground in the 2005 aerial photographs). Minear (1994) compared the frequency of large pools (greater than 6.5 ft deep and 430 ft² in area) between historical habitat surveys in 1937–1938 and field observations in 1991. Along Reach 1, there was a 40-percent decrease in large pools, whereas a 60-percent decrease was observed along Reach 2. Decreases in pool frequency along the upper McKenzie River basin could be a result of multiple causes, including reductions in large wood that are possibly associated with land management, reservoir storage, and channel clearing (Minear, 1994).

Sediment transport in Reach 1, and the distribution of alluvial bedforms, including gravel bars, is limited primarily by the inherent geology and hydrology of High Cascades terrain, but also is influenced by sediment trapping within the Carmen-Smith–Trail Bridge dam complex (Stillwater Sciences, 2006b). Field investigations have shown that the Pleistocene sediment composing the valley floor is much coarser than modern bedload, and remains in the channel as a coarse armor layer that can be transported only during extreme flood events (Stillwater Sciences, 2006b).

Although Western Cascades terrain comprises only 15 percent of the drainage basin above McKenzie Bridge, tributaries draining this terrain have high sediment yield rates and constitute the dominant sediment source to the McKenzie River (U.S. Forest Service. 1995; Stillwater Sciences, 2006b). The high sediment yield rates stem from a variety of mass-wasting processes, including deep-seated earth flows, debris flows, and debris slides. In contrast, the much more abundant High Cascades terrain is marked by low drainage densities and low sediment production, which supports groundwater-discharge dominated channels with little sediment transport. Although Trail Bridge and Smith Reservoirs trap 100 percent of coarse sediment entering from upstream, the overall rate of sediment capture is relatively low (approximately 478,400 ft³ of sediment per year), which is roughly equivalent to one or two medium-sized gravel bars in the lower McKenzie River basin. Based on sediment budget calculations, tracer rock studies, and mapping of inchannel sediment storage Stillwater Sciences (2006b) concluded that the McKenzie River along Reach 1 was supply limited. The upper parts of the reach (downstream of Trail Bridge Dam) display strong supply limitation, while downstream areas with greater sediment contributions from Western Cascade tributaries were closer to equilibrium. Along Reaches 1 and 2, there were no large (greater than 700 yd²) patches of bare gravel mappable from the 1939 or 2005 aerial photographs, which lends further evidence that any transportable sediment delivered to the channel is transferred to lower reaches during annual high water.

Offchannel habitat formation and maintenance along Reaches 1 and 2 are largely constrained by inherent geological controls. The steep gradient, narrow channel and relatively narrow floodplain along Reach 1 supports high sediment transport capacities that exceed the supply of sediment entering the reach. Although the terraces flanking the modern floodplain are largely composed of alluvial materials deposited during Pleistocene glaciations, these materials were deposited in a much wetter, higher energy glacio-fluvial setting. Hence, large floods such as those in December 1964 or February 1996 probably are required to remobilize local deposits of stored coarse sediments, but even these infrequent events are unlikely to trigger fundamental reach-scale changes in channel morphology. Lateral migration and side-channel formation are further restricted by revetment placed alongside roads.

It is unlikely that environmental flow releases from Smith and Trail Bridge Reservoirs would instigate reach-scale changes in side-channel formation, gravel bar frequency, or pool complexity because the overall channel morphology in Reach 1 is highly stable and subject to minimal alteration from large floods (figs. 24 and 25). However, gravel bar frequency in the lower gradient, downstream part of Reach 1 may possibly be increased if environmental flow releases coincide with natural or anthropogenic events that increase the supply of bed material (such as tributary inputs from large floods or gravel augmentation)and large wood. Environmental flow releases along Reach 2 are more likely than Reach 1 to create and maintain side-channels particularly along the lower gradient wider alluvial areas near Dearborn Island (RM 63). Changes at these sites are likely to be influenced by flood discharges and delivery of sediment and large wood from Horse Creek and smaller Western Cascade tributaries.

Habitat Limitation

The upper McKenzie River basin is home to the largest naturally spawning population of bull trout west of the Cascades, although they migrate seasonally downstream to the lower basin. The upper McKenzie River basin historically probably was not a predominant area for many of the other aquatic exemplar species, such as red-legged frogs, western pond turtle, Oregon chub, and brook lamprey because (among other reasons) these species rely upon secondary channel features which are sparse in the upper basin. Spring Chinook salmon are excluded from a part of their historical range in the upper McKenize River basin by Trail Bridge and Smith dams (National Marine Fisheries Service 2008b). Key habitat issues in Reaches 1 and 2 include an insufficient level of large wood (important for habitat formation such as pools and cover for fish) and spawning gravel, loss of rearing and spawning habitat (as a result of reservoirs and blockage posed by dams), and the reduction of large pools (National Marine Fisheries Service, 2008b).

Hydrologic alteration is not a primary issue for the aquatic/riparian ecosystem in the upper McKenzie River basin. The key habitat limitations in Reaches 1 and 2 include insufficient large wood and spawning gravel, loss of rearing and spawning habitat (as a result of blockage posed by dams), and the reduction of large pools (National Marine Fisheries Service, 2008b).

The 40–60 percent reduction in large pool habitat in the upper McKenzie River basin has potentially affected native fish production, particularly spring Chinook and bull trout, which depend on deep pools with cover. Rearing fish use pools, and fish abundance is closely tied to availability of cover provided by turbulence, large wood, overhanging vegetation, and depth. Adult fish frequently use pools as resting areas, to spend extended periods waiting for migratory or spawning environmental cues.

Middle McKenzie River Basin—Reaches 3–8

Channel Morphology and Controls on Physical Habitat Availability

The middle McKenzie River basin extends from the confluence of South Fork McKenzie River near the town of Blue River to the Leaburg Canal diversion dam (Reaches 4, 6, 7, and 8). The overall morphology and historical patterns of channel change along the lower parts of the South Fork McKenzie River (Reach 3) and Blue River (Reach 5) are similar to the mainstem McKenzie River.

The middle McKenzie River basin channel is moderately steep, with channel gradients varying from 0.0033 in Reach 4 to 0.0023 in Reach 8 (fig. 2, table 20). The town of Blue River (near the downstream end of Reach 4) marks the farthest advance of valley floor glaciations; hence, the valley bottom is wide above this point but narrows by more than 50 percent below Reach 4 (along Reaches 6–8). A relatively narrow floodplain (about 500–800 ft) is incised within this valley bottom, where the floodplain width is typically two to three times the width of the active channel (about 300 ft), although the ratio of floodplain to active channel width is highly variable depending on the width of the valley floor and the position of the channel relative to valley walls. In some areas, the channel is relatively confined, with floodplain width only slightly greater than active channel width (for example, Reach 7 near the town of Vida), although there are limited areas (such as Reach 4 above Finn Rock) where the floodplain is relatively wide, with several sinuous side channels in 1939 and 2005.

The availability of offchannel habitat and secondary channel features is largely controlled by floodplain width and local valley floor constraints; in areas where the channel is highly confined, few side channels are in the 1939 or 2005 aerial photographs, but in wider, alluvial stretches, side-channel frequency increases. The middle McKenzie River basin is highly stable with little net change in its planform and secondary channel features from 1939 to 2005 (figs. 24 and 26B). A specific gage analysis conducted at the USGS streamflow gage near Vida (14162500) shows that cross-sectional geometry at the gage has been stable for the period 1925–2008, with lower discharges having virtually no variation in stage over time, and moderate to high discharges having a maximum variation in stage of only ±0.3 ft (fig. 25).

The number of large pools along the middle McKenzie River basin has decreased by approximately 35 percent between 1939 and 1991 (from data provided by Sedell and others, 1991, and Minear, 1994). Pools within the middle McKenzie River basin are primarily forced pools, which occur where the channel impinges on bedrock, making an abrupt turn. Although forced pools are still along the river, and the overall position of pool-riffle sequences is similar in the 1939 and 2005 aerial photographs, Minear (1994) report that the overall size and complexity of pools has diminished and attributes these changes to reductions in large wood (which historically provided cover) and reduction of peak streamflows (which historically scoured pools).

Along Reaches 6–8, there were large decreases 
(80–100 percent) of bare gravel bars mapped between 1939 and 2005 (fig. 24). Reach 4 shows an increase in gravel bar area mainly a result of the growth of a bar near the confluence with the South Fork McKenzie River. However, the magnitudes of the 1939 and 2005 Reach 3 gravel bar area are much smaller than those of Reaches 6–8. Bar losses along Reaches 6–8 were typically associated with vegetation colonization, which reduced the area of bare gravel. The bars in the 1939 photographs typically appear in the same locations as their 2005 counterparts, (and generally along wide areas of the floodplain), but the overall area of bare gravel in 2005 is diminished, and the area of vegetation has increased (fig. 26B). Minear (1994) report similar results when mapping gravel bars above Leaburg between 1945 and 1986. Decreases in bar area are probably in part a result of reduction of flood peaks (which has allowed vegetation to colonize and stabilize formerly active bar surfaces that were located along lower-gradient areas). Although not quantified, sediment supply presumably has diminished appreciably following construction of the USACE flood control dams which block the transport of coarse bed material from the historically sediment-rich tributaries of South Fork McKenzie and Blue River. Reduced sediment supply would limit bar building unless offset by inputs from other tributaries or bank erosion. Creation of new gravel bars is also limited by the channel and valley morphology as there is a paucity of low-velocity areas where gravels and cobbles can be deposited. Hydraulic modeling by Lignon (1991) showed that despite flow regulation, the steep, confined reaches of the middle McKenzie River basin continue to support high transport capacities that may exceed the current sediment supply. Supply limitation is also indicated by visual assessments of bed substrate from 1939 and 1991, which suggests that the bed material has coarsened in response to dam construction (Minear, 1994). Bed coarsening can have a direct effect on the availability of spawning habitat so that in many areas the particle size is too coarse for salmon spawning. The resulting reduction in supply of spawning gravel has led to redd superimposition, or redds being built on top of one another (Lignon and others, 1995).

Habitat Limitation

Historically, the middle McKenzie River basin was probably an important region for all nine exemplar species discussed in this report, although less for bull trout compared to the upper basin. In this region there are large tributaries such as Blue and South Fork McKenzie Rivers that would also have provided abundant habitat for spring Chinook salmon, lamprey, and potentially bull trout in the upper parts of these subbasins. The middle McKenzie River basin mainstem is an important area for spring Chinook salmon and lamprey. The floodplain and offchannel habitats in this region are also important for Oregon chub, red-legged frogs, western pond turtles, and key riparian vegetation species (cottonwoods, alder, willows, and other terrestrial species).

Key habitat issues and limiting factors include reduced recruitment of large wood and spawning gravel (as a result of blockages imposed by dams), loss of rearing and spawning habitat, alteration of natural streamflow and temperature regimes, and loss of instream, riparian, and floodplain habitats and/or complexity. The reduction in spawning habitat can be specifically related to a reduction in cobble/gravel recruitment caused by dams and channel and bank stabilization. Reduced high streamflow events (high streamflow pulses and small to large floods) affect channel and offchannel formation and maintenance. Aside from decreased flooding in the winter and spring, dam streamflow releases are typically elevated in the summer. In the late summer elevated streamflows can encourage spawning along the wetted channel margins. Later these locations can dry up if streamflow levels are not sustained throughout the incubation period. Dam streamflow releases in the summer are also unnaturally cooler, which can affect the timing of juvenile salmonid growth and outmigration.

The 35-percent reduction in large pools in the middle McKenzie River basin can negatively affect many species, particularly spring Chinook salmon, bull trout, and lamprey by decreasing available cover and habitat necessary for rearing and resting areas. Salmon typically stay under available cover and are less inclined to migrate during daylight hours. Daytime streamflow downramping at a dam can then severely affect fish if they become stranded in small disconnected pools or dry channel beds.

Lower McKenzie River Basin—Reaches 9–12

Channel Morphology and Controls on Physical Habitat Availability

The McKenzie River floodplain below the town of Leaburg widens, occupying an alluvial corridor greater than 3,000–4,000 ft in width from this point to its confluence with the Willamette River (fig. 27). The lower reaches of the McKenzie River have displayed the greatest degree of dynamism of all the study reaches; although there is little net change in historical channel position along much of the middle and upper McKenzie River basin, there historically have been more avulsions and a greater degree of lateral migration along Reaches 10–12.

Channel planform in the lower McKenzie River basin is generally that of a “wandering gravel bed river” (Church, 1983). Wherein the channel is predominantly single threaded, but can also have multichanneled reaches. Historical surveys by the General Land Office (GLO) in the 1850s depict the channel along Reaches 9–12 as narrow, sinuous, and flowing through a dense riparian forest that extended throughout the Holocene floodplain (digital map produced by Pacific Northwest Ecosystem Research Consortium, 2002; and The Nature Conservancy and available at http://www.fsl.orst.edu/pnwerc/wrb/access.html, last accessed July 30, 2009). Between 1850 and 1939, there were at least seven avulsions along Reaches 10, 11, and 12 that resulted in the abandonment of large-amplitude bends. Avulsions were often coupled with rapid migration along adjacent bends, as multiple locations show the enlargement and downstream migration of bends. Along the nearby Willamette River, numerous avulsions also occurred during the large regional floods of the late 19th century (particularly the 1861 flood of record), whereas rapid meander migration was dominant from the 1890s to 1930s, (a period in which there were frequent small floods) (Wallick and others, 2007). The lower McKenzie River basin probably followed a similar trajectory of change until the onset of flood control and channel stabilization in the mid-20th century. Many of the areas along the McKenzie River that displayed meander migration, side-channel shifting, and avulsions historically were bordered by more erodible Holocene alluvium that has been subsequently stabilized with revetment. In contrast, bends that impinge upon resistant valley walls (composed of Tertiary volcanics) have generally remained stable between 1850 and 2005, with little net change in position (fig. 27).

Between 1939 and 2005, there have been large reductions in the length of secondary channel features and abundance of active gravel bars (fig. 24). Several of the large, semi-stable islands in the 1939 photographs appear increasingly stabilized by vegetation in the 2005 aerial photographs, and the historically active side channels that once separated these islands from the floodplain appear less distinct (fig. 26C). Despite these reductions, many relict secondary channel features are still present within the active channel and floodplain. Large decreases in the area of active gravel bars have occurred: Along Reach 9 in 1939, seven bare gravel bars, as well as other, much larger unmapped bars had greater than 25-percent vegetation cover; by 2005, these bare gravel bars had either disappeared entirely, or were covered in mature tree canopy, resulting in a 100-percent decline in the mapped area of bare gravel bars along Reach 9. Reach 10 had a 25 percent decrease in gravel bar area, and Reaches 11 and 12 had a greater than 70-percent loss of bare gravel (fig. 24).

The abundance of gravel bars in the 1939 aerial photographs, and particularly the appearance of numerous “bare” gravel features along point bars and side channels, and at the heads of islands, indicates that these alluvial features were created and remobilized on a frequent basis. The large decrease in the area of bare gravel bars between 1930 and 2005 is presumably a result of a combination of factors, including (1) vegetation encroachment and stabilization because of reduced peak streamflows, (2) decreased sediment supply resulting from trapping of coarse sediment by dams, (3) decreased recruitment of local sediment sources because of dampened bank erosion rates resulting from flood control and bank stabilization, and (4) reduction in the number of low-velocity, depositional areas as a result of channel narrowing and simplification following vegetation encroachment and bank stabilization.

In addition to losses in secondary channel features and gravel bars, the lower McKenzie River basin between reaches 9 and 12 had a 63-percent decrease in large pools between 1939 and 1991 (Sedell and others, 1991). Historically, the main pool-forming agents along these dynamic, alluvial reaches would presumably have been inchannel obstructions (for example, midchannel bars or islands) that are enlarged and stabilized by large accumulations of wood, triggering scour and pool formation (for example, Montgomery and others, 1995; O’Connor and others, 2003). Pools also would be expected where the channel impinges upon Tertiary valley walls and possibly where the channel flows against older, more indurated (difficult to erode) bank materials at the base of Pleistocene terraces. Reductions in pool frequency are most likely a result of an overall decrease in pool-forming agents, namely gravel bars and large wood, which are not only dependent upon an ample supply of bed material and wood, but also require sufficiently high streamflows to form the channel-spanning obstructions and scour the resultant plunge pools.

Determining the relative importance of sediment supply versus transport capacity would require quantitative analyses that include development of a sediment budget and modeling of transport capacity for a variety of streamflow scenarios, which were beyond the scope of this project. Previous work by Lignon (1991), however, indicates that transport capacities in the lower McKenzie River basin are sufficiently high to transport available sediment. If this is the case, then trapping of coarse gravels by dams may be a factor in the decrease of total gravel bar area in addition to other factors, such as reductions in local recruitment and the stabilization of formerly active bars.

A specific gage analysis at USGS streamflow gages near Coburg (14165500) (Reach 12) shows substantial lowering of stage for all streamflows, indicating a net lowering of the streambed elevation by approximately 8 ft (fig. 25). Much of this lowering occurred in the period after the December 1964 flood and prior to 1972 when the gage was discontinued. Measurements by the USGS in October 2007 indicate that incision slowed during 1972–2007, but that the channel continued to degrade by a total of about 2 ft. Klingeman (1973) attributed bed degradation at Coburg to several possible causes, including commercial sand and gravel removal downstream of the gage and the presence of a historic irrigation dam that may have artificially stabilized the streambed until the pilings in the dam were destroyed. Another possible contributing factor to the channel lowering is a series of avulsions that occurred downstream of the gage at the site currently occupied by a sand and gravel mining operation. The avulsions occurred sometime between 1939 and 2005, and likely are attributable to the December 1964 flood. By locally increasing channel slope, these avulsions may have triggered an upstream migrating knickpoint that traveled upstream to the Coburg gage (as described on other rivers by Kondolf, 1997). A specific gage analysis conducted 20 mi upstream from Coburg at the USGS streamflow-gaging station near Walterville (14163900) (in Reach 10) shows that the channel geometry has been stable during the 11 years that the gage has been operational (1989–2009) (fig. 25). The Walterville gage is located at a stable location where the channel flows against the valley wall and has experienced very little shifting between 1939 and 2005.

Bed degradation can potentially be problematic from a floodplain restoration perspective because, as the channel bed incises, the discharge required to overtop the banks increases. Thus, the 2-year recurrence interval, or “bankfull flow” may not inundate the floodplain, and could lead to further bed scouring. Additional work is needed to determine the extent of incision in the vicinity of the Coburg gage, and to ascertain whether the bed lowering at the gage is indicative of systematic reach-scale incision in the lower McKenzie River basin. Field observations by biologists in the McKenzie River basin indicate that local incision at various points throughout the lower McKenzie River basin is potentially associated with avulsions through gravel mining pits and meander cutoffs triggered by floods (Greg Taylor, U.S. Army Corps of Engineers, and Andrew Talabare, Eugene Water and Electric Board, oral commun., 2009). Lignon (1991) also documents local incision on the order of 1–5 ft occurring downstream of Leaburg Dam. However, widespread incision may be limited by bedrock outcrops and other “hard points” that exert vertical control on channel elevations (Greg Taylor, U.S. Army Corps of Engineers, and Andrew Talabare, Eugene Water and Electric Board, oral commun., 2009).

Habitat Limitation

In addition to serving as a migration corridor for salmonids and lamprey, the lower reaches historically were probably most important for key floodplain and offchannel riparian species such as red-legged frogs, turtles, and Oregon chub.

Like the middle reaches, key habitat issues for the lower McKenzie River basin include reduced recruitment of large wood and spawning gravel, loss of rearing and spawning habitat, alteration of natural streamflow and temperature regimes, and loss of instream, riparian, and floodplain habitats and/or complexity. Reduced high magnitude streamflow (high streamflow pulses, small and large floods) affect channel formation and maintenance, instream and offchannel habitats, riparian structure, and habitat complexity and diversity. Bank stabilization with the use of revetments is also a factor in the reduction of channel complexity in the middle and lower reaches.

As with the upper and middle reaches, large-pool habitat in the lower McKenzie River basin has decreased. There also has been a significant loss of secondary channels (36–70 percent) and bare gravel bar area (25–100 percent), which are important for a number of native species, including juvenile salmonids (particularly during overwintering periods), Oregon chub, western pond turtle, and red-legged frog. Bare gravel bars that are inundated during fall or spring flows are potentially important spawning areas for Chinook salmon or other native species.

Aerial photographs of the lower McKenzie basin show that by 1939, much of the historical floodplain forest depicted in the 1850s GLO vegetation maps already had been converted to agriculture (fig. 28). Between 1939 and 2005, increases in riparian forest canopy occurred on some areas of the floodplain (including formerly logged areas as well as historically active, nonvegetated bar surfaces), and other areas experienced declines in native vegetation (typically where floodplain forest had been converted to gravel pits or agriculture). In general, however, large (greater than 50 acres) patches of riparian forest are still present, particularly along more dynamic areas of the channel that are unsuitable for agriculture or development. Nonetheless, black cottonwoods and mature alders are currently dying in large numbers along the lower reaches of the river, which may be related to channel incision and subsequent declines in the water table elevation (Greg Taylor, U.S. Army Corps of Engineers, and Andrew Talabare, Eugene Water and Electric Board, oral commun., 2009). Recent cottonwood and alder recruitment does not appear to be sufficient to replace the dying trees, which is similar to findings for the mainstem Willamette River, where young black cottonwoods are unable to replace mature stands (Dykaar and Wigington, 2000). Another explanation for the lack of new recruitment of riparian trees may be the significant reduction in bare ground within the active channel and floodplain, as a result of the reduction in bare gravel bars.

First posted February 8, 2010