Introduction

As the number of decadent dams in the United States and the awareness of the ecological cost of damming rivers increases (Poff and Hart 2002), so does the need for research on large-scale dam removal projects (Poff et al. 2003). To date, no studies of high-head (> 30 m) dam removal have been conducted in the United States (Gregory et al. 2002). Extant studies of small dam removal projects describe the potentially much less habitat altering process generated by unleashing systems with lower discharge and sediment accumulation (Doyle et al. 2005). With some exceptions (e.g., Toth 1995), available studies are short-term, qualitative, examine few system components with insufficient spatial and temporal replication (Hart et al. 2002) and describe relatively minor geomorphic change. Additionally, identifying causal pathways resulting from any dam removal process is complicated by confounding responses to simultaneously changing abiotic factors (Hart et al. 2002). Consequently, the question remains as to whether the impacts of dams are indeed reversible and by what pathways change occurs. There is evidence that ecologic changes following dam removal will be much more complex than a simple reversal of changes caused by dam construction (Auble et al. 2007). These changes will remain as an ecological legacy (e.g., Harding et al. 1998) such as that seen for Yellowstone National Park following the fires of 1988 (Gresswell 1999, Turner et al. 2003) and the eruption of Mt. Saint Helens in 1980 (Dale et al. 2005).

While several high-head dams are under review for removal (Gregory et al. 2002), removal of the Elwha and Glines Canyon dams from the Olympic Peninsula's Elwha River in Washington State will be the largest dam removal project to date in the United States. This provides an opportunity for research on removal of high-head dams and to address weaknesses and information gaps of previous dam-removal studies. With over 80% of the watershed protected from other land-use changes by National Park Service management policies, the Elwha River ecosystem is ideal for these studies. The Olympic National Park (ONP) boundary encompasses the upper portions of the Elwha River, the Lake Mills reservoir, and part of the middle river section between the dams. At this time, the Park also owns the Lake Aldwell reservoir with eventual dispensation of this area to be determined (for map, see Duda et al. 2008). The lower section of river downstream of the dams contains a mixture of private ownership and lands managed by the Lower Elwha Klallam Tribe (LEKT) and Washington Department of Fish and Wildlife. Portions of the estuary and nearshore are managed by Clallam County, the City of Port Angeles, and the LEKT.

Planning for research and monitoring is most effective as a collaborative effort among scientists representing many disciplines, as well as managers and other stakeholders (Poff et al. 2003). Here, we identify specific research and monitoring needed to address management concerns described during several workshops sponsored variously by ONP, U.S. Geological Survey, University of Washington and the National Park Foundation. Workshop participants represented federal agencies (i.e., ONP, U.S. Geological Survey, U.S. Fish and Wildlife Service, U.S. Bureau of Reclamation, U.S. Army Corps of Engineers, USDA Forest Service, National Oceanic and Atmospheric Administration, National Marine Fisheries Service, U.S. Environmental Protection Agency, Pacific Northwest National Laboratory), state agencies (i.e., Washington Department of Fish and Wildlife, Washington Department of Ecology, Washington Department of Natural Resources, Puget Sound Action Team), tribal entities (i.e., Lower Elwha Klallam Tribe, Jamestown S'Klallam Tribe, Point No Point Treaty Council, Northwest Indian Fisheries Commission), local governments (i.e., Clallam County, City of Port Angeles) and academics from numerous universities, private foundations and consultants (CCMRC 2004, Schreiner and Winter 2005, Stolnack and Naiman 2005, Stolnack et al. 2005). These workshops affirmed that recovery of anadromous fish populations and trophic interactions, and restoration of the reservoirs within ONP are high management priorities. Additionally, conditions in the lower river and nearshore affect fish recovery upstream, so there is also interest in sediment effects on fish and fish habitat in these areas (Shaffer et al. 2008). Finally, it is hoped that restoration of sediment input from the Elwha River will help mitigate erosion from Ediz Hook, a spit protecting the mouth of Port Angeles harbor.

The workshops provided initial steps toward achieving collaboration among parties with interest in the Elwha River. Subsequent integration among agencies and scientists has continued, albeit at a lower level. Monitoring and research projects in the Elwha, at least at this stage, are primarily ad hoc and researcher-driven instead of part of a comprehensive plan (McHenry and Pess 2008, Winter and Crain 2008). Under such conditions, understanding ecosystem response is potentially compromised, and consistent funding will be hard to obtain. Having identified integration as an important need, our purpose here is to present conceptual models as a framework upon which to reinvigorate efforts at collaboration among all parties with interest in Elwha restoration. Derived from the 2001–2005 workshops, these models describe impacts to the Elwha River and how the ecosystem will respond to dam removal, together with associated priority management needs for research and monitoring.

Conceptual Models for Research and Monitoring of Elwha River Restoration

Conceptual models of ecosystems are graphic representations of interactions among key ecosystems components, processes and drivers. The processes of building and evaluating conceptual models aim to explain current understanding of ecosystem function, bring common understanding among interested parties, and clarify underlying assumptions and hypotheses. Conceptual models can include any degree of detail depending on the scope of the immediate issue. We present a general framework of models that provide context for studies reported in this issue of Northwest Science relative to management concerns.

As an overview, we first consider general consequences of the dams and planned restoration activities within each section of the Elwha River. The two dams effectively partition the river into four distinct areas experiencing different impacts, adding to the natural variability in biologic and physical processes present due to effects of gradient changes, cumulative area drained, patterns of bedrock geology and other factors (Table 1). The upper river section above Glines Canyon Dam and tributaries of the middle and upper river lack anadromous fish but are otherwise in pre-dam condition. The middle river section between the two dams lacks anadromous fish and inputs of sediments, large woody debris (LWD) and other forms of organic matter from upstream. The result is a deeply incised river channel in some areas, coarse substrates, and a less dynamic floodplain (Kloehn et al. 2008). The lower section below both dams is similarly incised, lacks sediment and organic matter inputs from upstream, but also supports hatchery-raised and wild anadromous fish. The reservoirs are extremely altered from pre-dam conditions, as the dams created a lentic ecosystem from a lotic one. Dam removal is expected to restore pre-dam processes to the river, with the different river segments experiencing some distinct and some similar phases (Figure 1). To facilitate restoration, ONP and the LEKT will actively re-establish vegetation in the former reservoirs and in partnership with other agencies will out-plant fish in many reaches (Table 1; see also Ward et al. in press, McHenry and Pess 2008, and Pess et al. 2008).

Figure 1.

Paths by which different sections of the Elwha River will be restored through time. The middle/lower sections and upper section follow the same path to restoration but begin at different points. The reservoirs begin on a different path that eventually converges with the path of the river sections. LWD = large woody debris; OM = organic matter.

TABLE 1.

Summary of the major direct and indirect effects of dams on sections of the Elwha River and adjacent nearshore. MDN = marine derived nutrients; NA = not applicable.

In addition to impacts within the river channel, the dams have also reduced sediment input to the Elwha estuary and adjacent nearshore areas (Table 1). These areas support eelgrass and kelp bed habitats (Warrick et al. 2008) that are important for the rearing, migration, feeding and smoltification of anadromous fish, particularly Chinook salmon and chum salmon (Healy 1982, Simenstad et al. 1982, Shaffer 2004). The estuary and nearshore habitats also are inhabited by forage fish, shellfish and other economically and ecologically important marine creatures (e.g., rockfish, halibut, herring, sea otters; Miller et al. 1980, Shaffer 2004, Simenstad et al 1979). Removal of the two dams in the Elwha River is expected to partially restore nearshore substrates and beach shape to allow eelgrass and kelp beds to re-establish. However, the Elwha is the source of only 30% of the sediments that naturally replenish the nearshore, with the remainder coming from a bluff that has been armored to protect housing and a landfill (US Army Corps of Engineers 1971). With bluff armoring in place, merely re-establishing sediment dynamics from the Elwha will likely be insufficient to fully restore estuarine and nearshore habitat (Shaffer et al. 2008).

Specific components, processes, and interactions that resource managers consider most important vary by river and intertidal segment (Figure 2) and reflect management goals. System dynamics are described in more detail in other papers in this special issue. Specific hypotheses illustrated in Figure 2 are articulated in Table 2.

Figure 2.

Hypothesized results of dam removal on three sections of the Elwha River: above both dams and tributaries throughout (Upper Elwha and Tributaries), below one or both dams (Middle Elwha and Lower Elwha), the reservoirs currently behind each dam (Reservoirs), and the Elwha estuary and associated nearshore area (Estuary/Nearshore). Arrows in boxes indicate increase (↑), decrease (↓), or unknown (±) change in component; components with no arrow are not expected to change in abundance, at least in the short term; arrows between boxes indicate causal links. Thick grey arrows indicate potential for transport of exotic plants and animals, diseases/pathogens, sediment, large woody debris, wildlife, plant propagules, fish, and nutrients. Grey boxes indicate components with a lagged response time.

TABLE 2.

Summary of management goals, hypotheses, specific research and monitoring needs, and monitoring indicators identified by staff of Olympic National Park and attendees of research and monitoring workshops. Under monitoring indicators, “research” indicates that the associated need requires research rather than monitoring. LWD = large woody debris; MDN = marine derived nutrients.

TABLE 2 Continued..

TABLE 2 Continued..

Designing research and monitoring around these components (Figure 2) requires careful consideration of spatial and temporal extents as well as the type of monitoring that is needed (Figure 3). Monitoring types are defined by the questions addressed and endpoints measured. Implementation monitoring answers how well management plans were followed (i.e., administrative endpoints such as acres of invasive species treated); effectiveness monitoring answers whether the management plan achieved the desired outcome (i.e., environmental endpoints such as restoration of a fish run); validation monitoring tests assumptions (i.e., hypotheses, models) about cause and effect (Busch and Trexler 2003, Karr and Yoder 2004). Done carefully, monitoring of restoration can provide a vital link between science and resource management, including adaptive management (Naiman et al. 1992, Stanford and Poole 1996).

Figure 3..

Spatial and temporal extents of management goals. Indicators of spatial extent on the x-axis are categorical variables, therefore difference in spatial scale within category is not intended to be represented. Management goals listed in the lower sub-box within “reservoirs” are expected to be achieved before those in the upper box. Reservoirs are expected to begin changing before the other sections, with the upper section changing most slowly.

Management goals vary according to spatial scale and time for achievement. At the smallest temporal and spatial scale, ONP will take direct management actions at particular sites, and monitoring projects will be used to inform adaptive management. These include monitoring the implementation and effectiveness of planting vegetation, exotic plant control, and sediment control measures during dam removal to protect a municipal water supply. These projects will occur primarily at the sites of management actions, and involve monitoring to evaluate achievement of goals such as proportion of seedlings established or suspended sediment levels (Figure 3). Goals should be established using conceptual and quantitative models that describe cause and effect relationships between initial management actions and eventual restoration. If monitoring shows that the actions fail to achieve the desired outcome, this information can support adaptive management. Management goals and target conditions have been determined for fish restoration (Ward et al. in press) and management of suspended sediment in municipal water supply (Randle et al. 2006), but not for other aspects of the Elwha River project.

Management goals at a slightly larger temporal and spatial scales (i.e., river segment/nearshore scale; Figures 2 and 3), reflect that the reservoirs are expected to undergo dramatic geomorphic change, as lake beds become river channels. Initial management goals are to stabilize sediments and restore vegetation on terraces and within the riparian corridor. Over the longer term, the goal is to restore fish and their habitat, wildlife, vegetation, and dynamics of sediment and organic matter. In the middle and lower reaches, the initial focus will center on effects of eroding reservoir sediments on downstream biota and water quality. Over time, the focus will shift to changes in channel morphology due to accumulation of sediments and large woody debris, and resultant effects on creation of fish habitat, and fish restoration. In the estuary and nearshore, the goal is for improved rearing and migratory habitat for anadromous fish, which may result in greater success for adult fish. Eventually, greater adult success should impact the river because greater numbers of fish may return to spawn. These changes will not necessarily be strictly sequential or concurrent throughout the Elwha ecosystem and will require effectiveness and validation monitoring.

The upper section above the dams is the least altered; best available information indicates that it includes all of its natural components except anadromous fish (Figures 2 and 3). Because this area is encompassed by a national park, there are few other stressors on the system, thus providing almost ideal conditions for research investigating the effect of anadromous fish on the trophic structure and nutrient cycling in aquatic and terrestrial forested ecosystems. Over the 30 years predicted to achieve the management goal of salmon restoration (DOI 1995), the use of validation monitoring to examine response dynamics of the upper river ecosystem to the return of anadromous fish will assist in identifying thresholds that may have bearing on effective escapement goals. Additionally, returning fish will drive interactions among other components of the system (Figure 2) making it possible to better understand ecosystem dynamics.

Considerations for Research and Monitoring of Restoration

Because so little is known about ecosystem recovery following removal of large dams, most research questions will be answered using effectiveness or validation monitoring. That is, most research questions will be answered by following trends in environmental endpoints, and comparing actual outcomes with hypotheses and models. Consequently we consider research and monitoring to be nearly one and the same for Elwha restoration.

The foundation of any monitoring program is the indicators that are monitored. Indicators were identified in the various planning workshops (Table 2), but they have not as yet been put into a conceptual framework. There are many published criteria and requirements for good indicators (e.g., NRC 2000, Dale and Beyeler 2001, Young and Sanzone 2002, Niemi and McDonald 2004) and all lists include the need for a scientific basis linking the indicator to the ecosystem. It is recommended that indicators be supported by a conceptual model, which will provide the rationale for choosing it and the interpretation of observed changes (Landres 1992, NRC 2000, Young and Sanzone 2002). While we have provided overview models, much more detail is required to support indicator selection.

Many restoration projects are hampered by lack of baseline and reference data that provide a measure of site potential and establish pre-treatment conditions (Frissell and Ralph 1998, Roni et al. 2005). Monitoring of restoration projects can address these common weaknesses by adopting a before-after control-impact (BACI) study design. According to this method, data are collected before and after the onset of restoration activities in both the treated area and an untreated, pristine reference area. The greater the number of pre-treatment data collected, the better the estimate of natural variation in treated and control sites (Frissell and Ralph 1998). The biggest impediment to this approach from a design perspective is the difficulty in finding reference sites (Block et al. 2001, McHenry and Pess 2008). There are no perfect controls in field studies, and pristine sites of any kind are challenging to locate. Nevertheless, it is important to find the best reference sites possible (Minns et al. 1996). Even if no data are collected before the treatment, comparing an area after treatment with an untreated reference area is a reasonable option. If no reference sites can be found, a before-after (BA) design will describe the outcome of a treatment better rather than a poorly controlled BACI design (Roni et al. 2005). The lack of reference sites can sometimes be addressed by using mechanistic models (Shugart 1989).

Possibilities for location of reference sites assessing various responses to Elwha dam removal can be found elsewhere within the Elwha basin and on the Olympic Peninsula (Table 3). None of the potential reference sites are ideal because environments and ecosystems of the Olympic Peninsula vary depending on aspect, geology, climate, elevation, and land management objectives (McHenry and Pess 2008). Consequently, available controls will depend on the objectives and variables of each research or monitoring project. For example, monitoring of large woody debris dynamics in the middle reach of the Elwha should probably use areas dominated by old-growth Douglas-fir (Pseudotsuga menziesii) of reasonably similar climate. This might include the upper Sol Duc River but not the Hoh River where floodplains are predominantly Sitka spruce (Picea sitchensis) and precipitation is approximately double that of the Elwha. On the other hand, for studies of bear-use of riparian corridors in the upper reach of the Elwha, control areas may have to include the Hoh and Quinault Rivers because anadromous fish are currently present, even though the difference in vegetation may have an effect on bear distribution. The best control area for nearshore monitoring is offshore from the Dungeness River mouth, approximately 16 km to the east near Sequim.

TABLE 3..

A subset of research and monitoring topics and associated potential reference sites for comparison to the Elwha River (See also McHenry and Pess 2008).

The spatial sample frame of restoration research and monitoring should recognize that ecosystems are hierarchically nested structures in time and space (Frissell and Ralph 1998). In aquatic systems, habitat units are nested within reaches; reaches are nested within valley segments; and valley segments are nested within watersheds. The rate of controlling dynamics slows as spatial extent increases. This structure calls for a likewise nested, hierarchical approach to monitoring, with functional controls at several spatial scales (Frissell and Ralph 1998, Minns et al. 1996, Poole et al. 1997). The hierarchical approach to restoration monitoring has rarely been applied and perhaps never in the Pacific Northwest (Imhoff et al. 1996).

In addition to hierarchical nesting, co-location or coordination of sample frames among studies of different resources will likely provide more information for the same cost. For example, co-located information about riparian vegetation, wildlife behavior and fish carcass abundance would provide greater understanding of system dynamics than the same studies located independently.

Full implications of restoration management actions in the Elwha River will likely require a long time to evolve because the time scale of ecosystem response to perturbation is proportional to the size of the ecosystem, habitat diversity (Trexler and Busch 2003) and size and intensity of the disturbance. Short-term changes at this scale may be misleading because it may be hard to differentiate effects of management actions from natural variation. For example, macroinvertebrate density and richness following a dam removal in Oregon did not immediately respond to dam removal, but over time responded to reservoir erosion (Stewart 2006). However, changes at smaller spatial scales such as the outcomes of local management actions may be easier to detect in shorter time. Bryant (1995) suggested that pulsed monitoring is an effective temporal design for restoration monitoring. He advocates that monitoring should include extensive long-term surveys repeated at intervals of 10–15 years, interspersed with intensive short-term 3–5 year studies focused on specific questions.

Needs for Information and Coordination

Resource management staff members of ONP have five priority information needs related to removal of the dams on the Elwha River, all of which can be addressed by monitoring (see also McHenry and Pess 2008):

Additionally, monitoring is required to support adaptive management (Table 2). Some of these needs are being met, including baseline data collections for some topics. Other needs are for integrated, long-term research and monitoring following dam removal. Our conceptual models and list of potential indicators is a first step towards developing the plan. In the near future, resource managers must develop refined models for indicators and prioritize the indicators as a basis for selecting a feasible number. They must also identify restoration targets for ecosystem components other than fish and sediment, and develop a sample frame for all indicators. These activities will be less effective if undertaken independently by each interested party.

Development of a common framework to evaluate, prioritize and coordinate the many goals associated with this large restoration project is a potential role of the Elwha Research Consortium. An intellectual framework could be built by coordinating more detailed conceptual and quantitative model development to integrate hypotheses supporting research and monitoring needs for agency management, and those of other research projects. An important aspect would be to address spatial and temporal hierarchies inherent in the results of large-extent restoration projects. An organizational framework would include integrating research through coordination of sample designs, and co-location of data collection among research and monitoring projects. Additionally, there is a need for data management, project-wide data standards, information management, protocols, and coordinated reporting (Conquest and Ralph 1998).

Sustaining support for long-term monitoring will be a daunting challenge due to the low priority generally given to monitoring of restoration (Frissell and Ralph 1998, Roni et al. 2005) and evidenced by the fact that no money for monitoring was included in the appropriation supporting the Elwha River dam removal project (Winter and Crain 2008). Nevertheless, this challenge must be addressed because this project is an unprecedented research opportunity, many research questions cannot be answered with a short-term effort, and the information can be used to improve the success of other anticipated dam removal projects. Long-term monitoring is the most practical and effective way to assess restoration and to document recovery. An intellectual and organizational framework including data management and reporting will form a strong platform from which to justify funding for necessary monitoring.