Evidence Linking Delayed Mortality of Snake River Salmon to Their Earlier Hydrosystem Experience

North American Journal of Fisheries Management 22:35±51, 2002 ᭧ Copyright by the American Fisheries Society 2002

PHAEDRA BUDY* AND GARY P. T HIEDE Utah Cooperative Fish and Wildlife Research Unit, U.S. Geological Survey, Department of Fisheries and Wildlife, Utah State University, Logan, Utah 84322-5290, USA

NICK BOUWES Eco Logical Research, 456 South 100 West, Providence, Utah 84332, USA

C. E. PETROSKY Idaho Department of Fish and Game, 600 South Walnut Street, Post Of®ce Box 25, Boise, Idaho 83707, USA

HOWARD SCHALLER Columbia River Program Of®ce, U.S. Fish and Wildlife Service, 9317 Northeast Highway 99, Suite I, Vancouver, Washington 98665, USA

Abstract.ÐThe numbers of Snake River salmon and steelhead Oncorhynchus spp. have substantially declined since the completion of the Columbia River hydrosystem. We used analytical approaches to identify management options for halting the decline of these stocks, such as removal of Snake River dams and improvements to the existing hydrosystem. The bene®ts these actions are predicted to have in terms of salmon recovery hinge on whether the mortality that takes place in the estuary and early in their ocean residence is related to earlier hydrosystem experience during downstream migration. Evidence from the literature demonstrates numerous mechanisms that would explain this delayed mortality in relation to a ®sh's experience passing through the hydrosystem. Spatial and temporal comparisons of stock performance provide indirect evidence of delayed mortality and evidence that delayed mortality is linked to hydrosystem experience. Recent mark± recapture data also provide evidence of differences in delayed mortality by route of passage through the hydrosystem. The different types of evidence discussed here suggest that the delayed mortality of Snake River ®sh is related to the hydrosystem.

Over the last several decades, Paci®c salmon tric development, an emphasis on hatchery sup- Oncorhynchus spp. and steelhead O. mykiss pop- plementation, and climatic and oceanic conditions. ulations have declined to extremely low levels. The Snake River stocks of spring and summer These declines have been especially pronounced chinook salmon O. tshawytscha and steelhead once in the Snake River and upper Columbia River re- accounted for as much as 50% of the salmon pro- gions, where the National Marine Fisheries Ser- duced in the interior Columbia River Basin (Has- vice (NMFS) has estimated a 55±100% probability semer et al. 1997). Because of their earlier decline of extinction of particular stock groups over the relative to other Columbia River stocks, the ex- next 100 years (NMFS 2000). These low popu- tinction or near extinction of some stocks, and their lation levels, and the high risk of extinction, led long migration past as many as eight large hydro- to the recent Endangered Species Act listing of 13 electric dams (Figure 1), these Snake River stocks evolutionarily signi®cant units (ESUs) in the Co- have been a topic of extensive research, modeling, and management efforts since the development of lumbia Basin (Myers et al. 1998). The decline in the hydroelectric system in the late 1960s and early Columbia River Basin salmon and steelhead is 1970s. Most recently, the fate of the Snake River probably the result of a combination of factors, stocks has received national attention, and the including harvest, habitat degradation, hydroelec- question of whether or not to breach the lower four Snake River hydroelectric dams is being discussed among scientists, managers, and politicians * Corresponding author: [email protected] (American Fisheries Society 1999a; ISG 1999; Received August 17, 2000; accepted July 19, 2001 Karieva et al. 2000).

35 36 BUDY ET AL.

FIGURE 1.ÐMap of the Columbia River Basin, showing the eight hydroelectric dams (and dates of their completion) that spring and summer chinook salmon migrate past on the Snake and lower Columbia rivers.

Until recently, management efforts for Snake ing of the four lower dams on the Snake River has River salmon and steelhead stocks have focused been proposed (USACE 1999). This management largely on technological ®xes aimed at improving option was evaluated through a multiagency mod- the survival of juvenile and adult ®sh through or eling process (the Plan for Analyzing and Testing around the hydroelectric dams during their migra- Hypotheses; PATH). The PATH forum completed tions to and from the Paci®c Ocean (ISG 1999; a detailed biological decision analysis to evaluate Nemeth and Kiefer 1999). These technological ®x- the different management options recommended es have included attempts to transport juvenile ®sh for the recovery of Snake River ESUs (Marmorek around the dams by truck or barge, allowing by- and Peters 1998; American Fisheries Society pass ®sh to migrate in-river around each dam, 1999a). In addition, NMFS, the agency responsible spilling the ®sh over the tops of dams, and de- for protecting anadromous ®shes under the En- veloping turbines having lower mortality rates for dangered Species Act, has addressed the question the ®sh that pass through them. Although these of dam breaching, in the context of improvements ®xes have gradually improved the direct survival in other areas and life stages, using a series of of ®sh around or through the dams (Raymond extinction and life history models through their 1988; Williams and Mathews 1995), the Snake Cumulative Risk Initiative (NMFS 2000). The River stocks have continued to decline. Because PATH decision analysis demonstrated that dam of this continued decline, along with the results breaching would likely lead to recovery over a from research and modeling, many ®shery scien- broad range of hypotheses and was thus the most tists now believe there is no feasible technological risk-averse management option. The NMFS Cu- ®x, short of dam breaching, that will recover Snake mulative Risk Initiative similarly concluded that River salmon and steelhead (Marmorek and Peters to minimize the risk of extinction of Snake River 1998; American Fisheries Society 1999b; ISG stocks, dam breaching may be required in com- 1999; Nemeth and Kiefer 1999; Karieva et al. bination with improvements in habitat, reductions 2000). in harvest, and changes in hatchery practices As part of the efforts to mitigate the impacts of (NMFS 2000). the hydrosystem on endangered stocks, the breach- The key point of uncertainty relates to the source EVIDENCE FOR HYDROSYSTEM-RELATED MORTALITY 37 of mortality that affects Snake River ®sh while delayed hydrosystem mortality from recent pas- they are in the estuary and ocean. Sources of es- sive-integrated-transponder (PIT) tag data, which tuary and early ocean mortality include not only documents the hydrosystem passage and survival elements of the natural ocean environment but also histories of individual ®sh. The combination of delayed effects of earlier life stage experiences. information from different life stages, however, Here we focus on the hypothesis that although this provides evidence and estimates of delayed hy- mortality takes place in the estuary and ocean, it drosystem mortality and links this mortality to fac- may be related to a ®sh's earlier experiences within tors of the earlier hydrosystem experience. In this the hydrosystem. Snake River salmon, for exam- paper, we synthesize these pieces of evidence for ple, may die in the estuary or ocean after exiting delayed hydrosystem mortality. the hydrosystem as a result of the cumulative stresses of negotiating as many as eight hydro- Study Area electric dams. Direct mortality, in contrast, is mor- A Natural River tality that can be measured directly and takes place within the hydrosystem. Delayed hydrosystem As the Independent Scienti®c Group noted (ISG mortality could also affect the upstream survival 1999), the predevelopment Columbia River sal- of returning spawners (e.g., increased straying of monid ecosystem was a network of complex in- transported ®sh as a result of poor imprinting), but terconnected habitats that had been created, al- that component is not addressed here. tered, and maintained by natural physical pro- Identifying the incidence and magnitude of de- cesses. Flow regimes produced by local and re- layed hydrosystem mortality of Snake River stocks gional climates (unencumbered by hydrosystem is crucial to the predicted outcome of a dam- regulation) provided natural spawning, rearing, breaching scenario. If delayed mortality takes and migration conditions. Before hydrosystem de- place and is related to the hydrosystem, then dam velopment, passage to and from the spawning and breaching is likely to greatly increase the survival rearing sites was unimpeded. The historical hy- of Snake River stocks and ultimately may provide drograph of the Columbia River Basin was char- a high probability of recovery. In contrast, if de- acterized by high spring runoff (April through layed mortality does not take place or is unrelated mid-June), decreasing summer and fall ¯ows (mid- to hydrosystem experience, dam breaching will June through December), and low winter ¯ows have less of a bene®t and may not be suf®cient to (January through March). Anadromous salmon recover these stocks. Whether research can reduce adapted to this ¯ow pattern by migrating down- this uncertainty is unclear (Brodeur et al. 2000); river during the spring freshet. These riverine hab- under current conditions, however, extinction is itats, now altered by human activity, were and are highly probable and decision makers will be faced critical for salmonid spawning, rearing, migration, with making a decision about dam breaching soon. feeding, and predator avoidance (ISG 1999). In this paper, we discuss evidence that Snake River Snake River salmon and steelhead have complex stocks are affected by delayed mortality and that life histories, encompassing adaptations to a va- this outcome is related to their experience in the riety of freshwater and saltwater habitats over a hydrosystem. This paper provides a comprehen- broad geographic range. These salmon spawn, sive synthesis of the evidentiary points for hydro- emerge, and rear in tributary streams 800±1,540 system-based delayed mortality. We attempt to km inland from the Paci®c Ocean at elevations document the different types of evidence currently ranging from 300 to 2,000 m. Because yearling available that may help ®shery scientists and de- smolts passively migrate to the estuary during the cision makers fairly evaluate the likely effects of spring freshet, water velocity determines their ar- a dam breaching. rival time at the estuary. Numerous morphological, Evidence for delayed hydrosystem mortality physiological, and behavioral changes (e.g., comes in several forms. In this paper, we ®rst sum- change in coloration, increased saltwater toler- marize the literature that considers mechanisms for ance, and becoming semipelagic) are associated delayed mortality and links to a ®sh's hydrosystem with smolti®cation such that, as the smolt migra- experience. In addition to our literature review, we tion progresses, the smolts are no longer adapted also evaluate spatial and temporal patterns of stock to remain in freshwater (Ebel 1977; Thorpe et al. performance to provide indirect evidence for the 1981). The salmon spend several years in the existence and the extent of delayed hydrosystem ocean, covering thousands of kilometers in their mortality. Finally, we evaluate direct evidence of ocean migration, before returning to freshwater 38 BUDY ET AL.

FIGURE 2.ÐCross-sectional view of possible routes for juvenile ®sh past the dams: (1) over the spillway; (2) through the juvenile ®sh bypass system; and (3) through the turbine. Some ®sh go into the juvenile bypass system and collection channel and are either returned to the river below the dam or transported past the dams by barge and truck. and migrating to their natal tributaries to spawn. McNary Dam on the Columbia River. Transported Each of these life history stages poses natural chal- ®sh are then released below Bonneville Dam. lenges for survival. Anthropogenic conditions may Radio-tagging studies indicate that migration increase stress and perhaps contribute to decreased rates decrease signi®cantly as smolts approach a survival at each life history stage. Our focus is on dam and a considerable proportion of ®sh are de- the smolt migration and the effects of altered con- layed in the forebay for a week or more (Venditti ditions and stresses imposed by the hydrosystem. et al. 2000). Venditti et al. (2000) also believed that increased forebay crossings and upstream ex- The Impounded River cursions of ®sh (delay) were associated with re- In the Columbia River Basin, main-stem dam duced water velocity. Fish that pass through tur- construction began in 1938 with the completion of bines are exposed to extreme pressure changes and Bonneville Dam near the mouth of the Columbia mechanical injuries while going through the tur- River. All seven dams on the lower Columbia and bines (Long et al. 1968; Mathur et al. 1996; Na- Snake rivers had been completed by the time Low- varro et al. 1996; Bickford and Skalski 2000). er Granite Dam, the eighth and uppermost dam on Some smolts are guided into a collection and by- the lower Snake River, was completed in 1975 pass system by intake screens. From the intake (Figure 1). Juvenile ®sh attempting to migrate to screens, the guided smolts are returned in 3 s back the ocean and adult ®sh migrating back to their to the surface (about 20 m) into a turbulent ga- natal Snake River freshwater tributaries to spawn tewell (Figure 2). Smolts go from 1 to 3 and back must now pass through eight large hydroelectric to 1 atmospheric pressures in about 10 s. At Lower dams and 522 km of slack water (Figure 1). The Granite Dam (S. Pettit, Idaho Department of Fish juveniles have four different possible routes past and Game, personal communication), they are then the dams: (1) over the spillway; (2) through the piped from the ®sh collection area nearly 0.4 km juvenile bypass system, with routing back to the to below the dam at high velocity (9 m/s) and river; (3) into the juvenile bypass system and being pressured through two 90Њ turns, experiencing high collected for transport in barges or trucks; or (4) turbulence and rapid deceleration at the end. through the turbines (Figure 2). Snake River ®sh Smolts are passed through a device to separate ®sh may be collected for transportation at three of the by size, dewatered, and then passed through a tube four dams on the Snake River (Lower Granite, that leads to a PIT-tag detector. Remember, this Little Goose, and Lower Monumental) and at account describes the passage through only one of EVIDENCE FOR HYDROSYSTEM-RELATED MORTALITY 39 eight dams. Typically, for transportation effec- increase in the residence time and temperature of tiveness evaluations, smolts are then held in race- the water, further stressing ®sh (Raymond 1979). ways for 24 h, transferred to a sample room, anes- In addition, reservoir environments are better suit- thetized, marked, returned to the raceways for re- ed for certain predators, which can have a direct covery, and then barged, trucked, or ¯ushed and indirect effect on smolts (Ruggerone 1986; through a 20-cm-diameter pipe for release to the Beamsderfer and Rieman 1991; Poe et al. 1991; river. In recent bypass and collection operations at Rieman et al. 1991; Vigg et al. 1991; Schreck and the dams (since 1994), bypassed smolts may not Stahl 1998; Collis et al. 1999). As much as 20% be held in the raceway but still experience the same of the smolt population can be lost during the mi- stressors up to the point of diversion. In summary, gration through a reservoir (NMFS 2000). Simi- several problems are associated with passage larly, hydrosystem operation and its effects on the through the dams by way of collection, bypass, estuary and near-shore environment may affect and turbines: (1) delay of smolts in the forebay; salmonid survival by altering the annual ¯ow cycle (2) concentration of smolts in the forebay and in and associated ecological (e.g., increased smolt the tailrace at the bypass out¯ow; (3) high pre- residence time) and environmental (e.g., reduced dation rates of smolts near the dams, where pred- productivity) changes (Sherwood et al. 1990; Pear- ators tend to congregate; 4) large pressure changes; cy 1992). This migration experience through an and (5) mechanical injury. impounded river presents the potential for an ac- The ®sh that pass dams through the spillways cumulation of stress events, which may lead to face different conditions: (1) They appear not to delayed mortality. be delayed as long as ®sh that pass by the other routes, (2) they avoid the mechanical injury and Evidence from the Literature: A Focus on dewatering of bypass and the severe pressure Impacts of Stress changes of the other routes, and (3) predators are Impacts of Stress more dispersed in the tailwater of the spillway (Faler et al. 1988). Therefore, passing a dam by Stress and its impacts on ®sh and ®sh popula- the spillway has been hypothesized to be the least tions have been widely studied and documented. stressful route and cause less mortality than pass- For further discussions of the general impacts of ing through the turbines (although gas supersatu- stress on ®sh, see major reviews by Wedemeyer et ration may present problems during uncontrolled al. (1990) and Fagerlund et al. (1995). Much of spills in high-¯ow years). In low-¯ow years, most the focus on stress in Paci®c salmon in the avail- or all of the water is diverted for power production, able literature describes acute factors or events forcing the majority of ®sh into the collection sys- such as toxicology (Meyers and Hendricks 1982; tem (for transport or bypass) or through the tur- Cairns et al. 1984; Larsson et al. 1985) or obstacles bines. In high-¯ow years, a larger proportion of to migration (Bentley and Raymond 1976; Fager- water will go over the spillway, such that a greater lund et al. 1995). However, many of these studies proportion of ®sh will pass the dam by a possibly provide a narrow perspective on stress and toler- less stressful route. Passage past a dam claims the ance in the context of delayed or latent mortality; lives of as much as 10% of the smolt population they do not consider sublethal and chronic stress. (NMFS 2000), and the remaining population has In this section, we focus on evidence and potential probably been stressed by the experience. pathways for the impacts of the hydrosystem on Reservoirs behind the dams may also create stress, the accumulation of stress, and delayed stressful conditions for smolts. Water velocity is mortality associated with stress. greatly reduced by accumulating behind the dams, Fish migrating through a reservoir and past a and thus the time and energy the ®sh expend to dam have three possible experiences: (1) They mi- get through the reservoirs are greater than that grate successfully past the dam and experience no experienced in the free-¯owing conditions for negative impacts from passage, (2) they die im- which these ®sh evolved (Williams and Mathews mediately from mechanisms discussed above, or 1995). The increased freshwater residence time of (3) they are physiologically or behaviorally in-river migrants may result in untimely physio- stressed (Figure 3). This set of outcomes is pos- logical changes for survival in saltwater that are sible for each of the eight hydroelectric dams that less suitable for the freshwater environment (Conte a Snake River ®sh must negotiate. Acute stress is and Wagner 1965; Wagner 1974; Woo et al. 1978). generally a high-intensity discrete event (e.g., by- The decrease in water velocity also results in an pass system at the dam), whereas chronic stress is 40 BUDY ET AL.

Hetherman et al. 1998; FPC 2000; Mesa et al. 2000). Other studies, evaluating nonvisual signs such as swimming performance, growth, and blood chemistry, also suggest that smolts are stressed by these events (Dawley and Ebel 1975; Dawley et al. 1975; Wright and McLean 1985; Krise et al. 1990). However, these latter signs of stress are not routinely measured and are probably underestimated in terms of the total effect of these passage experiences. Although several experiences are theoretically stressful, determining whether these events are ac- tually stressful can be dif®cult. Sometimes stress effects are visually observable, but physiological and biochemical tests generally are needed to de-

FIGURE 3.ÐFlow diagram of mortality and survival tect stress in ®sh. Despite the great strides in mea- options as a juvenile ®sh migrates past a dam for as suring stress from these tests, however, interpre- many as eight times. As a ®sh passes a dam, it either tation of their signi®cance remains dif®cult (Kru- dies, experiences no effect, or encounters stress. If zynski and Birtwell 1994). This is due, in part, to stressed, the ®sh may fully recover or the stress may the problem of obtaining background indicator accumulate or have chronic health effects on the ®sh, values: Either reference ®sh have already experi- leading to decreased physical abilities. Acute stress can lead directly to delayed mortality, or diminished abilities enced a stressor or the capture event itself was may lead to death through mechanisms of delayed mor- stressful (Congelton et al. 2000). For example, tality. Harman et al. (1980) demonstrated that stress brought on by the capture process was, if not immediately fatal, a major contribution to delayed either an accumulation of acute events or a longer mortality in freshwater drum Aplodinotus grun- term, low-intensity experience (e.g., repeated ex- niens. Documenting stress may also be dif®cult posure to high temperatures). A ®sh may recover because single indicators (e.g., cortisol levels) or from acute or chronic stressors, eventually die challenge tests used to measure stress may not from the stressor, or experience diminished phys- demonstrate the full effect of stress (Wood et al. ical abilities, which may increase vulnerability to 1983; Wedemeyer et al. 1990). The use of multiple sources of mortality. Here we describe the com- indicators may ameliorate this problem but may bination of the effects of stress and its associated also result in equivocal results, given that multiple mortality as delayed mortality. Although delayed indicators are often con¯icting (McLeay and mortality can take place within the hydrosystem, Brown 1979). Finally, some responses may be ini- for our discussion this portion of mortality is in- tially bene®cial and so interpreted (Maule et al. cluded in direct hydrosystem mortality estimates 1989); if a stress becomes chronic, however, the and not included in our de®nition of delayed hy- responses may be maladaptive and their impact drosystem mortality. Studies linking stress and may be expressed outside the time frame of the mortality that happens within the hydrosystem, evaluation (Wedemeyer et al. 1990). Because however, provide evidence that delayed mortality many of these problems produce only partial ev- can happen outside the hydrosystem. idence of stress, the prevalence of stress in ®sh populations may be underestimated. Measuring Stress Measuring the magnitude of stress experienced Stressful events related to the hydrosystem can by a ®sh is dif®cult enough; relating the level of be categorized as follows: (1) dam passage, (2) mortality brought on by stress may be even more migration conditions, and (3) collection and trans- problematic. Even if stress is observed in a ®sh, portation around the hydrosystem. Visible external this stress may not be ecologically signi®cant signs of injury, trauma, and disease in ®sh are (Sprague 1971). Indeed, stressed individuals can observed and recorded regularly at each dam col- often recover fully (Mesa et al. 1994). Conversely, lection facility (e.g., descaling, hemorrhaging on individuals that no longer show signs of stress ®ns, curved spine, fungal infections, and visible through speci®c indicators may still die from the gas bubble trauma; Williams and Mathews 1995; cumulative effects of stress (Wood et al. 1983; EVIDENCE FOR HYDROSYSTEM-RELATED MORTALITY 41

Wedemeyer et al. 1990). Moreover, death from et al. 1985). Also, physical disturbances (acute these cumulative effects may not take place for stressors) have been demonstrated to increase the some time after the stress events. Jensen et al. metabolic rate of ®sh (Barton and Schreck 1987). (1980) suggested that for half of the population of Cumulative increases in energy demand in re- juvenile salmonids to die from the stress of ex- sponse to repeated disturbances may result in de- posure to increased total gas pressures may take pleted energy reserves (Mazeaud et al. 1977; Bar- as long as 50 d. Such time frames for stress-in- ton et al. 1986). Decreased energy reserves may duced mortality would often put smolts past the lead to decreased swimming performance, disease hydrosystem, thus making it dif®cult to observe resistance, foraging ability, growth, reproductive and link hydrosystem-related stress to mortality. success, predator avoidance, and ultimately sur- Physiological stress experiments are often per- vival (Wedemeyer et al. 1990). formed in some type of holding facility (e.g., challenge tests) and evaluate an indicator of stress Stress and Predation (e.g., cortisol levels) from a single stressor (e.g., The concept of increased vulnerability to pred- high temperatures) or the interaction between the ators as a result of acute or chronic stress is ubiq- stress event and another source of mortality (e.g., uitous in ecology (Errington 1946; Peterson 1977; vulnerability to predation under high temperature; Rice et al. 1987). Many examples can be found Wedemeyer et al. 1990). However, these experi- for ®sh (Coutant 1973; Sigismondi and Weber ments generally do not capture additional stresses 1988; Kruzynski and Birtwell 1994; Mesa et al. that an organism may encounter in the wild, nor 1994; Mesa and Warren 1997), speci®cally in- do they evaluate the vulnerability of stressed increased vulnerability to predation because of hy- dividuals to multiple sources of mortality (Barton drosystem-induced stress (Rieman et al. 1991; et al. 1986). Although these experiments do pro- Mesa 1994; Schreck and Stahl 1998). Mesa (1994) vide insight into the effects of stress, they probably found that the more stress events experienced by do not capture the total effects of stress on delayed a smolt, the greater its vulnerability to predation mortality of individuals (Vaughan et al. 1984; Ad- by northern pikeminnow Ptychocheilus oregonen- ams et al. 1985; Barton et al. 1986). In addition, sis, the most abundant piscivore in the lower Co- the laboratory conditions and experiments them- lumbia River (Rieman et al. 1991). In addition, selves may cause stress. These potential limita- Schreck and Stahl (1998) followed radio-tagged tions of stress experiments done in holding facil- smolts released from barges below Bonneville ities may need to be considered, given that NMFS Dam and found that stressed smolts avoided salt- has proposed to estimate the degree of delayed water entry by remaining on the ¯oating fresh- mortality associated with different routes of dam water lens at the saltwater±freshwater interface. passage by holding ®sh collected at Bonneville That forced these smolts to the surface, where they Dam in tanks (USACE 2000). The salmon in these were subject to predation by avian predators. Giv- experiments would not be subject to the additive en the greater densities of piscine and avian pred- or multiplicative effects of all mortality agents ex- ators below Bonneville Dam (Ward et al. 1995; perienced in the wild (e.g., piscine and avian pred- Roby et al. 1998), the interaction between pre- ators, diseases, parasites, foraging, and salt-water dation and stress is probably a mechanism in- transitions). Although stress has been demonstrat- volved in hydrosystem-related delayed mortality. ed to lead to mortality eventually, the total effect of stress has probably been underestimated because of the complex interaction between one or Stress and Disease more stressors and multiple interactions with other Various types of environmental stress can also sources of mortality. affect ®sh health and condition by lowering resistance to disease (Wedemeyer 1970). Stress may Stress and Energetic Condition decrease mucus production (Mazeaud et al. 1977) Stress can cause metabolic disturbances and de- and impair immune function (Morgan et al. 1996), creased energetic condition (Leach and Taylor which may increase vulnerability to disease. Ray- 1980). For example, if an extended migration mond (1988) found that during downriver migra- through reservoirs increases metabolic require- tion, transportation and seawater transition can ments by more than the energy intake, a smolt's cause stress and activate bacterial kidney disease. energy reserves would be depleted, which might For example, ®sh exposed to dissolved gas levels eventually lead to lower survival rates (Rondorf of 100% or more have shown increased suscep- 42 BUDY ET AL. tibility to disease and fungal infections (Weitkamp et al. 1985; Wedemeyer et al. 1990; Korman et al. and Katz 1980; White et al. 1991). 1994). The effects of cumulative stress have been ob- Stress and Physiology served in smolts as they migrate through the Snake and Columbia River hydrosystem. Mathews et al. Many investigators have concluded that sal- (1986) documented an increase in stress levels as monids undergo physiological changes, such as chinook salmon smolts moved through the collec- changes in osmoregulatory capacity, to success- tion and transport system at Lower Granite Dam. fully enter saltwater (Zaugg and Wagner 1973; Similarly, evidence exists for cumulative effects Wagner 1974; Zaugg 1981; Zaugg et al. 1985; of increased temperature and dissolved gas con- Rodgers et al. 1987). The proper timing of this centrations in the hydrosystem on smolt condition physiological change has been referred to as the and ultimately on survival. For example, Coutant ``biological window'' of successful saltwater entry (1999) suggested that the cause of a massive ®sh (Walters et al. 1978; Bilton et al. 1982; Boeuf and kill at McNary Dam was the result of cumulative Harache 1982; Holtby et al. 1989). The hydrosys- exposure to high-temperature water over several tem has prolonged the migration time for Snake days. Ebel et al. (1975) reviewed several labora- River salmon populations by decreasing water ve- tory and in situ studies and concluded that a sub- locity and delaying ®sh at dams (Bentley and Ray- stantial number of smolts die as a result of chronic mond 1976). Extended freshwater residency is as- versus acute exposure to high concentrations of sociated with reversion to parrlike characteristics dissolved gas. Given that the accumulation of due to a lower salinity tolerance in some species stress and often the resulting mortality have been of salmon (Conte and Wagner 1965; Wagner 1974; observed in the hydrosystem where these studies Woo et al. 1978). In contrast, transport of ®sh in are feasible, one can reasonably expect that this trucks and barges greatly shortens the time it takes stress may not result in mortality until after smolts ®sh to travel through the hydrosystem relative to leave the hydrosystem and thus would be consid- historical free-¯owing conditions and may result ered delayed hydrosystem mortality. in premature saltwater entry. The effects of premature saltwater entry (e.g., incomplete smolti®- Indirect Evidence for Delayed Hydrosystem cation) can result in high mortality and, in many Mortality: Retrospective Analyses and of the survivors, a reduction in or cessation of Downstream Stock Comparisons growth (Fagerlund et al. 1995). Retrospective comparisons of stock performance among regions and time periods provide Accumulation of Stress indirect evidence of delayed mortality linked to Although stresses are generally measured as sin- hydrosystem experience (Deriso et al. 1996; gle-point±speci®c events, stress may accumulate Schaller et al. 1999). The comparisons are based in ®sh and these additive effects may cause mor- on spawner and recruit (offspring of spawners) tality at a later stage. Kruzynski and Birtwell data describing the overall survival (or mortality) (1994) have referred to mortality resulting from over the life cycle of spring and summer chinook exposure to a legacy of physiologically sublethal salmon for brood years 1957±1990 (Figure 4). For stressors as an ``ecological death.'' Barton et al. the comparisons, life cycle mortality was separated (1986) observed an accumulation of stress when into two components. The ®rst component, direct juvenile chinook salmon subjected to repeated mortality (25±73%) takes place while juvenile and handling exhibited more stress indicators (e.g., in- adult salmon pass by dams and travel through res- creases in plasma lactate, sodium, potassium, an- ervoirs that constitute the hydrosystem and are aerobic metabolism, or ionoregulatory perfor- transported around the hydrosystem by barge or mance) than after a single disturbance. In fact, the truck (discussed in more detail below; Bouwes et accumulation of stress resulting in ``sudden death al. 1999). The second component of mortality is syndrome,'' induced by multiple handling, has what occurs in the Columbia River downstream been documented for rainbow trout (Lanno and from Bonneville Dam, in the estuary, and in the Dixon 1996). The cumulative effects of multiple ocean. After accounting for direct hydrosystem stress events on ®sh populations has been sug- mortality, natural spawner and recruit processes gested to cause the decline or collapse of ®sh pop- (e.g., density-dependent mortality), and environ- ulations, even though the effects of a single stress mental or ocean effects, the additional mortality may be insigni®cant (Vaughan et al. 1984; Adams that Snake River ®sh experience may be 37±68% EVIDENCE FOR HYDROSYSTEM-RELATED MORTALITY 43

FIGURE 4.ÐSchematic life cycle diagram for Snake River salmon and steelhead showing passage past the dams and the different indices of survival across different life stages. Smolts/spawner provide an estimate of freshwater and rearing spawning survival. The survival rate of smolts to returning adults (smolt-to-adult), or SAR, is measured at Lower Granite Dam; R/S is recruits per spawner, measured by back-calculating the number of recruits (progeny) from the number of spawners (parents) based on redd counts, age structure, and harvest. Direct and overall survival of ®sh that migrated through the hydrosystem and in barges is based on PIT tag mark±recapture experiments.

(Deriso et al. 1996). Two types of comparisons tem-related delayed mortality (Schaller et al. yield information on the degree or source of this 1999). These declines continued despite reduction additional mortality for Snake River stocks of or elimination of freshwater harvest and decreases spring and summer chinook salmon. We discuss in direct mortality from transportation. In addition, comparisons of overall survival rates for Snake spring and summer chinook salmon stocks expe- River stocks before and after the development of rience little ocean harvest (Berkson 1991). Most the hydrosystem and of stocks originating from degradation of Snake River spawning and rearing different regions with different migration experi- habitats occurred before the start of stock declines, ences but similar life histories. and populations in degraded and good habitat re- Comparison of historical and current empirical sponded similarly (Schaller et al. 1999; Petrosky life stage survival estimates provide evidence that et al. 2001). Furthermore, no climatic index was changes in delayed mortality have taken place in found that could explain why Snake River stock Snake River spring and summer chinook salmon declines were greater than their downstream coun- coincident with the completion of the hydrosystem terparts (Marmorek et al. 1996). Evidence sug- (Schaller et al. 1999). Retrospective comparisons gests, however, that each stock group experienced of stocks originating from different regions dem- common-year effects, presumably from climatic onstrate that completion of the hydrosystem in the conditions experienced by both (Deriso et al. late 1960s through the mid-1970s was immediately 1996). Moreover, although there was some cor- followed by considerably sharper declines in over- respondence between overall hatchery production all survival rates of Snake River stocks than of levels and declines in Snake River spring chinook downstream stocks over the same period (Schaller salmon, overall hatchery production showed little et al. 1999). The sharper decline in overall survival to no correspondence to declines in the down- observed for the upstream stocks probably re¯ect- stream counterparts. Similarly, little to no corre- ed the increasing amounts of direct and hydrosys- spondence was seen between hatchery production 44 BUDY ET AL. and population decreases in individual subbasins tween hatchery ®sh and wild ®sh, and increases (Wilson et al. 1996; Budy et al. 1998; Williams et in exotic predators (Schaller et al. 2000; Zabel and al. 1998). Therefore, the lack of support for other Williams 2000). The evidence discussed above hypotheses indicates that the hydrosystem is prob- suggests that delayed effects of the hydrosystem ably partially responsible for the greater mortality are likely to be partially responsible for overall that was observed in the Snake River stocks, rel- declines in survival. ative to their downstream counterparts, after the completion of the hydrosystem. Direct Evidence: Relationship to Hydrosystem The potential for stressful migration conditions Experience is high under low-¯ow conditions because a ma- Comparisons of overall survival of different jority of ®sh will experience increased delays at groups of ®sh that experience different combina- dams (given that a majority of ®sh pass dams tions of passage routes (Figure 2) and transpor- through turbines and bypass systems) and longer tation around the dams provide direct evidence of travel times through reservoirs. In years with high delayed hydrosystem mortality in relation to hy- ¯ows, ®sh experience fewer delays at dams (more drosystem experience. In 1994±1996, NMFS ex- ®sh pass over spillways) and shorter travel times panded their studies on transportation and in-river through reservoirs, which would reduce stressful survival of Snake River spring and summer chi- conditions. Columbia River salmon stocks gen- nook salmon by using PIT-tagged ®sh, thus esti- erally respond favorably to good water years with mating rates of survival by route of passage around high runoff during smolt migration, but the re- each dam (e.g., bypass or transport; Figure 4; sponse is relatively greater for upriver stocks than NMFS 1999; Sandford and Smith, in press). The for downriver stocks (Deriso et al. 1996). This tagged smolts are not detected if they pass through supports the notion that poor migration conditions turbines or over spillways. When the adults return, caused by the additional mainstem dams are the the detectors record their successful passage past primary cause of delayed mortality of upriver the uppermost dam on the Snake River, Lower stocks relative to their downriver counterparts. In Granite Dam (Figures 1 and 4). This detection re- fact, in recent years, the returns of wild Snake cord allows estimation of the overall SAR of var- River spring and summer chinook salmon have ious groups of ®sh that experience different pas- increased coincident with the relatively high ¯ow sage routes throughout the Snake and Columbia and spill (the route of passage thought to be the rivers (Bouwes et al. 1999; NMFS 1999; Sandford least stressful) during their downstream migration. and Smith, in press). The overall survival estimated from spawner For ®sh that are not transported and migrate in- and recruits described above can be further divided river, evidence for delayed hydrosystem mortality into an egg-to-smolt survival rate (smolts per in relation to hydrosystem experience can be eval- spawner) and a smolt-to-adult survival rate (SAR; uated by comparing the overall SARs of ®sh that Figure 4). Empirical estimates for these two in- were not collected and bypassed around the dams dices of survival from the period before three of with those ®sh that were collected and bypassed the four Snake River dams were in place (1975) one or more times. Spill over the top of a dam is indicate that the number of smolts per spawner has the route of passage most similar to a natural river, not changed in a pattern consistent with the steep whereas collection and bypass have a much greater decline in SAR for the Snake River stocks (Pe- potential of stressing the ®sh (discussed above). trosky et al. 2001). The decline in survival overall However, direct mortality is generally estimated cannot be attributed to changes in freshwater to be similar between the spill and bypass routes, spawning and rearing survival (e.g., habitat deg- both of which have lower rates than the turbine radation) and instead may be explained in part by route (Marmorek and Peters 1998). We would ex- the combination of direct and delayed mortality pect direct survival rates for collected and by- during downstream migration. This result is not passed ®sh to be generally higher because a portion surprising, given that several of these populations of the ®sh that are not collected go through the spawn and rear in wilderness areas that have re- turbines. Therefore, the more times ®sh go through mained relatively pristine since before the com- the bypass system, the lower their direct mortality pletion of the Snake River hydrosystem (Figure attributable to the hydrosystem and thus a higher 1). Alternative explanations for increased mortal- expected smolt-to-adult return rate (in the absence ity have emphasized over®shing, degradation of of delayed mortality). freshwater and marine habitats, interaction be- The apparent direct survival bene®ts of the by- EVIDENCE FOR HYDROSYSTEM-RELATED MORTALITY 45

press). This same general pattern is also observed for steelhead (Sandford and Smith, in press). These data further indicate that although direct mortality may be lowest for ®sh that are bypassed, some delayed mortality related to the hydrosystem may explain the patterns of overall survival. These results explain only the difference in delayed mortality between different routes of passage rather than the total effect of dam passage on delayed mortality. Furthermore, when ®sh are transported from lower dams (and avoid the bypass systems in the Snake River), the overall SAR appears to decrease as the number of navigated hydrosystem dams increases (to the point of transport). The SARs were greater for smolts transported from the upper two dams (Lower Granite and Little Goose) than for those transported from the lower dams (Lower Monumental and McNary) in 1994, the only year with an adequate sample size (Figure 5; Bouwes et al. 1999). This pattern, in part, provides evidence of delayed hydrosystem mortality for non- FIGURE 5.ÐThe top panel shows estimated SAR of wild spring and summer chinook salmon by in-river pas- bypassed smolts in that all groups were collected sage history: nondetected (0ϫ), and bypassed one, two, only once (for transportation). If we assume that and three or more times (1ϫ,2ϫ, 3 or more ϫ, respec- survival in the barge is always high, this pattern tively), 1994±1996. The bottom panel shows estimated is probably the result of the increasing stress of SAR of wild spring and summer chinook salmon for ®sh multiple project passage before transportation. that were transported from Lower Granite (LGR), Little In addition to comparing the overall survival of Goose (LGO), Lower Monumental (LMO), and McNary ®sh that migrate by different routes in-river, the (MCN) dams, respectively. Data are shown for 1994, the only year with suf®cient sample sizes for analysis. SAR of transported ®sh, relative to in-river ®sh, provides evidence of hydrosystem-related delayed mortality. The direct mortality of ®sh that are pass route of passage, however, do not translate transported through the hydrosystem appears to be well into SARs of wild spring and summer chinook very low (about 2%), whereas the direct mortality salmon. The SARs of uncollected ®sh were sig- of the ®sh that migrate in-river is generally greater ni®cantly higher than those of the bypassed group than 50%. Therefore, if no additional delayed mor- in 1995±1996 and may have been higher, or no tality resulted from being collected and transport- different, in 1994 (R. Kiefer, Idaho Department of ed, we would expect the transported ®sh to return Fish and Game, personal communication). Fish in as adults at a rate nearly twice that of the in-river the uncollected group used a combination of spill ®sh (given almost all transported ®sh survived and turbine routes and have a lower expected direct through the hydrosystem and half of the in-river survival than those in the bypassed group (by- migrants died while migrating). A posthydrosys- passed from one to four times). In addition, the tem SAR can be estimated after accounting for this uncollected group returned at rates as good as or estimated direct mortality. The ratio of transport better than those for the transported group in 1995 to in-river posthydrosystem SARs is termed D. If and 1996, but in 1994 the transported smolts may the posthydrosystem SARs of transported and in- have had the higher return rates. Similarly, esti- river ®sh were the same, D would be equal to one. mates of SARs for in-river migrating smolts were A ratio of less than one means that the transported examined each year for groups that were collected ®sh die at a higher rate in the estuary and ocean and bypassed 0±3 times or more through the four than do the in-river migrants. collector dams (Bouwes et al. 1999). These PIT- The degree of the delayed mortality of trans- tag data indicate that SARs decreased when the ported ®sh relative to ®sh that migrate in-river (D) number of times ®sh were bypassed increased is controversial. For Snake River spring and sum- (Figure 5; NMFS 1999; Sandford and Smith, in mer chinook salmon, estimates of D from two 46 BUDY ET AL.

PATH-developed models were 0.48 and 0.66 (Mar- an empirical basis as well as biological rationale. morek et al. 1998). When estimates were made Recent PIT-tag data are direct evidence that de- directly from the PIT-tag data, estimates of D for layed mortality of both in-river and transported spring and summer chinook salmon ranged from smolts is related to the hydropower system. More 0.52 (1994±1998; Budy 2001) to 0.63±0.73 speci®cally, the evidence suggests that, at least for (1994±1997; NMFS 2000). NMFS used similar collected and bypassed smolts, there is a difference methods and assumptions to estimate D for Snake between the patterns of direct passage survival River steelhead and obtained values ranging from rates and SARs. These ®ndings provide direct ev- 0.52 to 0.58 (NMFS 2000). PIT-tag estimates for idence that the delayed mortality of Snake River fall chinook salmon are not as reliable, but the spring and summer chinook salmon is related to recent estimate of D is approximately 0.24, with the juvenile migration hydrosystem experience. very wide con®dence intervals (Marmorek and Pe- ters 1998; NMFS 2000). Despite some disagree- Conclusion ment about the true D values, all methods indicate Substantial evidence supports the existence of that D is less than 1.0, which is evidence of delayed delayed mortality for Snake River chinook salmon mortality of transported ®sh. Although D describes and steelhead and links delayed mortality to hy- the difference in delayed mortality between trans- drosystem experience. This evidence comes in the ported ®sh and ®sh that migrate in-river, the ab- form of published literature on causes of stress and solute amount of delayed mortality for both groups any resulting delayed mortality, indirect evidence may plausibly be substantially greater. from life cycle modeling based on historical data The differences in hydrosystem experience may and known mechanisms, and direct evidence from help explain why delayed mortality is greater for ®sh-tagging experiments. The literature supplies transported ®sh than for nontransported ®sh. The numerous mechanisms that explain how the ob- stressful experience of being delayed in the fore- served stressors of the hydrosystem could be cu- bay, collected, and bypassed is most relevant to mulative and eventually lead to mortality at a later transported ®sh at collection dams because most life stage. The literature also demonstrates that the of the ®sh entering the collection and bypass sys- effects of stress can be hard to detect and dif®cult tem at these dams are subsequently transported. In to link empirically to delayed mortality in ®eld contrast, in-river smolts may be bypassed, go experiments. Nevertheless, there is little debate through turbines, or over the spillway. In addition, that ®sh condition and stress affect overall survival transported smolts are subjected to the stress of to adulthood or that the hydrosystem causes direct crowding and injury during collection, holding, mortality and imposes stress on those ®sh that sur- and transport (e.g., each liter of water would hold vive. Similarly, the retrospective life cycle anal- up to approximately the equivalent of one wild ysis provides indirect evidence of increases in de- chinook salmon and one wild steelhead; Congelton layed mortality in Snake River spring and summer et al. 2000). These high loading densities, in com- chinook salmon coincident with completion of the bination with the stressed state of the ®sh, provide Snake River hydrosystem. The decreases in sur- an effective environment for disease transmission vival rates of Snake River ®sh were considerably or manifestation (IDFG 1998). Also, transported sharper than those of downriver stocks over the smolts may be delivered to the estuary without same period. Further, most decreases in survival regard for their physiological readiness to enter rate were in the smolt-to-adult life stage, rather saltwater (Fagerlund et al. 1995). Because fall chi- than the spawner-to-smolt life stage. Data from nook salmon are still rearing while migrating, the PIT tags provide direct evidence that delayed mor- difference in the time required to be transported tality of both in-river±migrating and transported (approximately 1±2 d) versus migrating in-river smolts was related to the hydropower system. The (up to 6 weeks) is far greater than for spring and patterns of SAR could not be explained on the summer chinook salmon. Thus premature saltwater basis of estimates of direct survival within the hy- entry for fall chinook may explain why D is much drosystem but instead must result from delayed lower for fall chinook salmon than for spring and mortality related to speci®c hydrosystem experi- summer chinook salmon. All these factors may be ence. responsible for the greater delayed mortality ex- This evaluation has implications for the analyses perienced by transported ®sh. of management options for the recovery of Snake As discussed above, the hypothesis of delayed River salmon and steelhead. In those analyses, mortality resulting from hydrosystem passage has dam breaching appears to increase the survival of EVIDENCE FOR HYDROSYSTEM-RELATED MORTALITY 47

Snake River chinook salmon and may perhaps lead Columbia River Inter-Tribal Fisheries Commission, to eventual recovery of these stocks if delayed Portland, Oregon. mortality is related to hydrosystem experience. Bickford, S. A., and J. R. Skalski. 2000. Reanalysis, and interpretation of 25 years of Snake-Columbia River Dam breaching is predicted to have less of an ef- juvenile salmonid survival studies. North American fect and be insuf®cient for recovery if delayed Journal of Fisheries Management 20:53±68. mortality is unrelated to hydrosystem experience. Bilton, H. T., D. F. Alderdice, and J. T. Schnute. 1982. Regardless of the action chosen to recover these In¯uence of time and size at release of juvenile coho stocks, the amount of hydrosystem mortality to salmon (Oncorhynchus kitsutch) on returns at ma- compensate for must include both direct and de- turity. Canadian Journal of Fisheries and Aquatic Sciences 39:426±447. layed components. Given the evidence discussed Boeuf, G., and Y. Harache. 1982. Criteria for adapta- here, we ®nd signi®cant support for the hypothesis tions of salmonids to high salinity seawater in that Snake River ®sh are subject to delayed mor- France. Aquaculture 28:163±176. tality and that this effect is related to the hydro- Bouwes, N., H. Schaller, P. Budy, C. Petrosky, R. Kiefer, system. P. Wilson, O. Langness, E. Weber, and E. Tinus. 1999. An analysis of differential delayed mortality experienced by stream-type chinook salmon of the Acknowledgments Snake River. Oregon Department of Fisheries and We thank Earl Weber for comments on past Wildlife, Technical Report, Portland. drafts of this manuscript and Paul Wilson for in- Brodeur, R. D., G. W. Boehlert, E. Casillas, M. B. Eld- formation used in the historical comparisons. We ridge, J. H. Helle, W. T. Peterson, W. R. Heard, S. T. Lindley, and M. H. Schiewe. 2000. A coordinated also thank Paul Bunn and Donna Allard for graph- research plan for estuaries, and ocean research on ics assistance and the Fish Passage Center in Port- Paci®c salmon. Fisheries 25(6):7±16. land, Oregon, for providing information about the Budy, P. 2001. Analytical approaches to assessing re- hydrosystem. covery options for Snake River chinook salmon. Utah Cooperative Fish and Wildlife Research Unit References (UCFWRU) 2001(1):1±86. Budy, P., C. Petrosky, P. Wilson, O. P. Langness, and N. Adams, S. M., J. E. Breck, and R. B. McLean. 1985. Bouwes. 1998. Hatchery ``extra mortality'' hy- Cumulative stress-induced mortality of gizzard shad pothesis: response to submissions 1 and 2 of the in southeastern U.S. reservoir. Environmental Bi- WOE. Pages 19.1±19.4 in D. R. Marmorek and C. ology of Fishes 13:103±112. N. Peters, editors. PATH submissions to weight of American Fisheries Society. 1999a. Breaching dams is evidence report. ESSA Technologies, Ltd., Vancou- best alternative to recover salmon, says report. Fish- ver. eries 24(2):37. Cairns, V. W., P. V. Hodson, and J. D. Nriangu, editors. American Fisheries Society. 1999b. Western division: 1984. Contaminant effects on ®sheries. Wiley, New dam removal resolution passed. Fisheries 24(10):28. York. Barton. B. A., and C. A. Schreck. 1987. Metabolic cost Collis, K., S. Adamany, D. D. Roby, D. P. Craig, and of acute physical stress in juvenile steelhead. Trans- D. E. Lyons. 1999. Avian predation on juvenile actions of the American Fisheries Society 116:257± salmonids in the lower Columbia River. 1998 An- 263. nual Report to Bonneville Power Administration, Barton. B. A., C. A. Schreck, and L. A. Sigismondi. Project 9702400, Portland, Oregon. 1986. Multiple acute disturbances evoke cumula- Congelton, J. L., W. J. La Voie, C. B. Schreck, and L. tive physiological stress responses in juvenile chi- Davis. 2000. Stress indices in migrating juvenile nook salmon. Transactions of the American Fish- chinook salmon and steelhead of wild and hatchery eries Society 115:245±251. origin before and after barge transportation. Trans- Beamsderfer, R. C., and B. E. Rieman. 1991. Abundance actions of the American Fisheries Society 129:946± and distribution of northern squaw®sh, walleyes, 961. and smallmouth bass in John Day Reservoir, Co- Conte, F. P., and H. H. Wagner. 1965. Development of lumbia River. Transactions of the American Fish- osmotic and ionic regulation in juvenile steelhead eries Society 120:439±447. trout (Salmo gairdneri). Comparative Biochemistry Bentley, W. W., and H. L. Raymond. 1976. Delayed and Physiology 14:603±620. migrations of yearling chinook salmon since the Coutant, C. C. 1973. Effect of thermal shock on vul- completion of Lower Monumental and Little Goose nerability of juvenile salmonids to predation. Jour- dams on the Snake River. Transactions of the Amer- nal of the Fisheries Research Board of Canada 30: ican Fisheries Society 105:422±424. 965±973. Berkson, J. 1991. Coded wire tag analysis on Snake Coutant, C. C. 1999. Perspectives on temperature in the River spring, summer, and fall chinook. Letter dated Paci®c Northwest's fresh waters. Oak Ridge Na- February 19, 1991 to the U.S. National Marine Fish- tional Laboratory, Report ORNL/TM-1999/44, Oak eries Service for Endangered Species Act record, Ridge, Tennessee. 48 BUDY ET AL.

Dawley, E., T. Blahm, G. Snyder, and W. Ebel. 1975. the emigration timing of coho salmon (Oncorhyn- Studies on effects of supersaturation of dissolved chus kitsutch) smolts and fry and chum salmon (O. atmospheric gas on juvenile chinook salmon and keta) fry from Carnation Creek, British Columbia. steelhead trout. U.S. National Marine Fisheries Ser- Canadian Journal of Fisheries and Aquatic Sciences vice, Northwest Fisheries Center Final report to the 46:1396±1405. Bonneville Power Administration, U.S. Bureau of IDFG (Idaho Department of Fish and Game). 1998. Is- Reclamation, and U.S. Army Corp of Engineers, sue paper: ®sh health and stress related to trans- Seattle. portation. Appendix 3.3 in Idaho's anadromous ®sh Dawley, E. M., and W. J. Ebel. 1975. Effects of various stocks: their status and recovery options. IDFG, concentrations of dissolved atmospheric gas on ju- IDFG 98-13, Boise. venile chinook salmon and steelhead trout. U.S. Na- ISG (Independent Scienti®c Group). 1999. Return to the tional Marine Fisheries Service Fishery Bulletin 73: river: scienti®c issues in the restoration of salmonid 787±796. ®shes in the Columbia River. Fisheries 24(3):10±19. Deriso, R., D. Marmorek, and I. Parnell. 1996. Retro- Jensen, J. O. T., J. Schnute, and D. F. Alderice. 1980. spective analysis of passage mortality of spring chi- Assessing juvenile salmonid response to gas su- nook of the Columbia River. In D.R. Marmorek and persaturation using a general multivariate dose±re- 21 coauthors, compilers, and editors. Plan for an- sponse model. Canadian Journal of Fisheries and alyzing and testing hypotheses (PATH): ®nal report Aquatic Sciences 43:1694±1709. of retrospective analysis for ®scal year 1996. ESSA Karieva, P., M. Marvier, and M. McClure. 2000. Re- Technologies, Ltd., Vancouver. covery and management options for spring/summer Ebel, W. J. 1977. Panel 2: ®sh passage problems and chinook in the Columbia River basin. Science 290: solutionsÐmajor passage problems. Pages 33±39 in 977±979. E. Schweibert, editor. Columbia River salmon and Korman, J., D. R. Marmorek, G. L. Lacroix, P.G. Amiro, steelhead. American Fisheries Society, Special Pub- J. A. Ritter, W. D. Watt, R. E. Cutting, and D.C. E. lication 10, Bethesda, Maryland. Robinson. 1994. Development and evaluation of a Ebel, W. J., H. L. Raymond, G. E. Monan, W. E. Farr, biological model to assess regional-scale effects of and G. K Tanoka. 1975. Effects of atmospheric gas acidi®cation on Atlantic salmon (Salmo salar). Ca- saturation caused by dams on salmon and steelhead nadian Journal of Fisheries and Aquatic Sciences trout of the Snake and Columbia rivers. U.S. Na- 51:662±680. tional Marine Fisheries Service, Seattle. Krise, W. F., J. W. Meade, and R. A. Smith. 1990. Effect Errington, P. L. 1946. Predation and vertebrate popu- of feeding rate and gas supersaturation on survival lations. Quarterly Review of Biology 21:144±177. and growth of lake trout. Progressive Fish-Culturist Fagerlund, U. H. M., J. R. McBride, and I. V. Williams. 52:45±50. 1995. Stress and tolerance. Pages 461±503 in C. Kruzynski, G. M., and I. K. Birtwell. 1994. A predation Groot, L. Margolis, and W. C. Clarke, editors. Phys- bioassay to quantify the ecological signi®cance of iological ecology of Paci®c salmon. UBC Press, sublethal responses of juvenile chinook salmon Vancouver. (Oncorhynchus tshawytscha) to the antisapstain fun- Faler, M. P., L. M. Miller, and K. I. Welke. 1988. Effects gicide TCMTB. Canadian Journal of Fisheries and in variation in ¯ow on distribution of northern Aquatic Sciences 51:1780±1790. squaw®sh in the Columbia River below McNary Lanno, R. P., and D. G. Dixon. 1996. The induction of Dam. North American Journal of Fisheries Man- lethality in rainbow trout exposed to thiocyanite by agement 8:30±35. application of a stressor: acute or chronic response? FPC (Fish Passage Center). 2000. Fish passage center Canadian Journal of Fisheries and Aquatic Sciences smolt monitoring and gas bubble trauma monitoring 1774:241±242. program protocols. FPC, Portland, Oregon. Larsson, A., C. Haux, and M. L. Sjobeck. 1985. Fish Harman, B. J., D. L. Johnson, and L. Greenwald. 1980. physiology and metal pollution: results and expe- Physiological responses of Lake Erie freshwater riences from laboratory and ®eld studies. Ecotoxi- drum to capture by commercial shore seine. Trans- cology and Environmental Safety 9:250±281. actions of the American Fisheries Society 109:544± Leach, G. J., and M. H. Taylor. 1980. The role of cortisol 551. in stress-induced metabolic changes in Fundulus Hassemer, P. F., S. W. Kiefer, and C. E. Petrosky. 1997. heteroclitus. General and Comparative Endocrinol- Idaho's salmon: can we count every last one? Pages ogy 42:219±227. 113±125 in D. G. Stouder, P. A. Bisson, and R. J. Long, C. W., R. F. Krcma, and F. J. Ossiander. 1968. Naiman, editors. Paci®c salmon and their ecosys- Research on ®ngerling mortality in Kaplan turbines. tems: status and future options. Chapman and Hall, U.S. National Marine Fisheries Service, Northwest London. and Alaska Fisheries Center, Seattle. Hetherman, L., and 18 coauthors. 1998. Juvenile ®sh Marmorek, D. R., and 21 coauthors. 1996. Plan for transportation program. 1997 annual report to the analyzing and testing hypotheses (PATH): ®nal re- U.S. Army Corps of Engineers, Walla Walla, Wash- port on retrospective analyses for ®scal year 1996. ington. ESSA Technologies, Ltd., Vancouver. Holtby, L. B., T. E. McMahon, and J. C. Scrivener. 1989. Marmorek, D. R., and C. N. Peters, editors. 1998. Plan Stream temperatures and inter-annual variability in for analyzing and testing hypotheses (PATH): ®nal EVIDENCE FOR HYDROSYSTEM-RELATED MORTALITY 49

report for ®scal year 1998. ESSA Technologies, 1998. Status review of chinook salmon from Wash- Ltd., Vancouver. ington, Idaho, Oregon, and California. NOAA Tech- Mathews, G. M., D. L. Park, S. Achord, and T.E. Ruehle. nical Memorandum NMFS-NWFSC-35, Seattle. 1986. Static seawater challenge test to measure rel- Navarro, J. E., D. J. McCauley, and A. R. Blystra. 1996. ative stress levels in spring chinook salmon smolts. Turbine passage at four low-head hydroelectric fa- Transactions of the American Fisheries Society 115: cilities in northeast Michigan. North American 236±244. Journal of Fisheries Management 16:182±191. Mathur, D., P. G. Heisey, E. T. Euston, J. R. Skalski, Nemeth, D. J., and R. B. Kiefer. 1999. Snake River and S. Hays. 1996. Turbine passage survival esti- spring and summer chinook salmonÐthe choice for mation for chinook salmon smolts (Oncorhynchus recovery. Fisheries 24(10):16±23. tshawytscha) at a large dam on the Columbia River. NMFS (U.S. National Marine Fisheries Service). 1999. Canadian Journal of Fisheries and Aquatic Sciences An assessment of lower Snake River hydrosystem 53:542±549. alternatives on survival and recovery of Snake Riv- Maule, A. G., R. A. Tripp, S. L. Kaattari, and C. B. er salmonids. Anadromous ®sh appendix in Draft Schreck. 1989. Stress alters immune function and lower Snake River juvenile salmon migration fea- disease resistance in chinook salmon (Oncorhyn- sibility study/environmental impact statement. U.S. chus tshawytscha). Journal of Endocrinology 120: Army Corps of Engineers, Walla Walla, Washing- 135±142. ton. Mazeaud, M. M., F. Mazeaud, and E. D. Donaldson. NMFS (U.S. National Marine Fisheries Service). 2000. 1977. Primary and secondary effects of stress in Biological opinion of the Federal Columbia River ®sh: some new data with a general review. Trans- power system including the juvenile ®sh transpor- actions of the American Fisheries Society 106:201± tation program and the U.S. Bureau of Reclama- 212. tion's 31 dams including the entire Columbia Basin McLeay, D. J., and D. A. Brown. 1979. Stress and project. NMFS, Seattle, Washington. chronic effects of untreated and treated bleached Pearcy, W. G. 1992. Ocean ecology of North Paci®c kraft pulpmill ef¯uent on the biochemistry and sta- salmonids. University of Washingon Press, Seattle. mina of juvenile coho salmon (Oncorhynchus kis- Peterson, R. O. 1977. Wolf ecology and prey relation- utch). Journal of the Fisheries Research Board of ships on Isle Royale. U.S. National Park Service Canada 36:1049±1059. Scienti®c Monograph Series 11. Mesa, M. G. 1994. Effects of multiple acute stressors Petrosky, C. E., H. A. Schaller, and P. Budy. 2001. Pro- on the predator avoidance ability and physiology of ductivity and survival rate trends in the freshwater juvenile chinook salmon. Transactions of the Amer- spawning and rearing stage of Snake River chinook ican Fisheries Society 123:786±793. salmon (Oncorhynchus tshawytscha). Canadian Mesa, M. G., T. P. Poe, D. M. Gadomski, and J. H. Journal of Fisheries and Aquatic Sciences 58:1196± Petersen. 1994. Are all prey created equal? a review 1207. and synthesis of differential predation on prey in Poe, T. P., H. C. Hansel, S. Vigg, D. E. Palmer, and L. substandard condition. Journal of Fish Biology 45: A. Prendergast. 1991. Feeding of predaceous ®shes 81±96. on out-migrating juvenile salmonids in John Day Mesa, M. G., and J. J. Warren. 1997. Predator avoidance Reservoir, Columbia River. Transactions of the ability of juvenile chinook salmon (Oncorhynchus American Fisheries Society 120:405±420. tshawytscha) subjected to sublethal exposures of Raymond, H. L. 1979. Effects of dams and impound- gas-supersaturated water. Canadian Journal of Fish- ments on migrations of juvenile chinook salmon and eries and Aquatic Sciences 54:757±764. steelhead from the Snake River, 1966 to 1975. Mesa, M. G., L. K. Weiland, and A. G. Maule. 2000. Transactions of the American Fisheries Society 108: Progression and severity of gas bubble trauma in 505±529. juvenile salmonids. Transactions of the American Raymond, H. L. 1988. Effects of hydroelectric devel- Fisheries Society 129:174±185. opment and ®sheries enhancement on spring and Meyers, T. R., and J. D. Hendricks. 1982. A summary summer chinook salmon and steelhead in the Co- of tissue lesions in aquatic animals induced by con- lumbia River basin. North American Journal of trolled exposures to environmental contaminants, Fisheries Management 8:1±24. chemotherapeutic agents, and potential carcino- Rice, J. A., L. B. Crowder, and F. P. Binkowski. 1987. gens. U.S. National Marine Fisheries Service Ma- Evaluating potential sources of mortality for larval rine Fisheries Review 44(1):1±17. bloater (Coregonus hoyi): starvation and the vul- Morgan, J. D., S. K. Balfry, M. M. Vijayan, and G. K. nerability to predation. Canadian Journal of Aquatic Iwama. 1996. Physiological responses to hyposa- Sciences 44:467±472. line exposure and handling and con®nement stress Rieman, B. E., R. C. Beamsderfer, S. Vigg, and T. P. in juvenile dolphin (mahimahi: Coryphaena hip- Poe. 1991. Estimated loss of juvenile salmonids to purus). Canadian Journal of Fisheries and Aquatic predation by northern squaw®sh, walleyes, and Sciences 53:1736±1740. smallmouth bass in John Day Reservoir, Columbia Myers, J. M., R. G. Kope, G. J. Bryant, D. Teel, L. J. River. Transactions of the American Fisheries So- Lierheimer, T. C. Wainwright, W. S. Grand, F. W. ciety 120:448±458. Waknitz, K. Neely, S. T. Lindley, and R. S. Waples. Roby, D. D., D. P. Craig, K. Collis, S. L. Adamany. 50 BUDY ET AL.

1998. Avian predation on juvenile salmonids in the USACE (U.S. Army Corps of Engineers). 2000. Eval- lower Columbia River. 1997 Annual Report to Bon- uate comparative delayed mortality of juvenile neville Power Administration, Contract 97BI33475, salmon utilizing different routes of passage at the and U.S. Army Corps of Engineers, Contract dams on the lower Columbia and Snake rivers. E96970049, Walla Walla, Washington. USACE, Northwest Division, Walla Walla, Wash- Rodgers, J. D., R. D. Ewing, and J. D. Hall. 1987. Phys- ington. iological changes during seaward migration of wild Vaughan, D. S., R. M. Yoshiyama, J. E. Breck, and D. juvenile coho salmon (Oncorhynchus kisutch). Ca- L. DeAngelis. 1984. Modeling approaches for as- nadian Journal of Fisheries and Aquatic Sciences sessing the effects of stress on ®sh populations. Pag- 44:452±457. es 259±278 in V. W. Cairns, P. V. Hodson, and J. Rondorf, D. W., M. S. Dutchuk, A. S. Kolok, and M. E. Nriagu, editors. Containment effects on ®sheries. L. Gross. 1985. Bioenergetics of juvenile salmon Wiley, Toronto. during spring out-migration. 1983 Annual Report Venditti, D. A., D. W. Rondorf, and J. M. Kraut. 2000. to Bonneville Power Administration, Project 82-11 Migratory behavior, and forebay delay of radio- (DE-AI70-82BP35346), Walla Walla, Washington. tagged juvenile fall chinook salmon in a lower Ruggerone, G. T. 1986. Consumption of migrating ju- Snake River impoundment. North American Journal venile salmonids by gulls foraging below a Colum- of Fisheries Management 20:41±52. bia River dam. Transactions of the American Fish- Vigg, S., T. P. Poe, L. A. Pendergast, and H. C. Hansel. eries Society 115:736±742. 1991. Rates of consumption of juvenile salmonids Sandford, B. P., and S. G. Smith. In press. Estimation and alternative prey ®sh by northern squaw®sh, of smolt-to-adult return percentages for Snake River walleyes, smallmouth bass, and channel cat®sh in Basin anadromous salmonids, 1990±1997. Journal John Day Reservoir, Columbia River. Transactions of Agricultural, Biological, and Environmental Sta- of the American Fisheries Society 120:421±438. tistics. Wagner, H. H. 1974. Photoperiod and temperature reg- Schaller, H. A., C. E. Petrosky, and O. P.Langness. 1999. ulation of smolting in steelhead trout (Salmo gaird- Contrasting patterns of productivity and survival neri). Canadian Journal of Zoology 52:219±234. rates for stream-type chinook salmon (Oncorhyn- Walters, C. J., R. Hilborn, R. M. Peterman, and M. J. chus tshawytscha) populations of the Snake and Co- Staley. 1978. Model for examining early ocean lim- lumbia rivers. Canadian Journal of Fisheries and itation of Paci®c salmon production. Journal of the Aquatic Sciences 56:1031±1045. Fisheries Research Board of Canada 35:1303±1315. Schaller, H. A., C. E. Petrosky, and O. P.Langness. 2000. Ward, D. L., J. H. Peterson, and J. J. Loch. 1995. Index Reply to Zabel and Williams' comments on ``Con- of predation on juvenile salmonids by northern trasting patterns of productivity and survival rates squaw®sh in the lower and middle Columbia River for stream-type chinook salmon (Oncorhynchus and in the lower Snake River. Transactions of the tshawytscha) populations of the Snake and Colum- American Fisheries Society 124:321±334. bia rivers'' by Schaller et al. (1999). Canadian Jour- Wedemeyer, G. 1970. Stress of anesthesia with MS-222 nal of Fisheries and Aquatic Sciences 57:1742± and benzocaine in rainbow trout (Salmo gairdneri). 1746. Journal of the Fisheries Research Board of Canada Schreck, C. B., and T. P. Stahl. 1998. Evaluation of 27:909±914. migration and survival of juvenile salmonids fol- Wedemeyer, G. A., B. A. Barton, and D. J. McLeay. lowing transportation. Annual Report to the U.S. 1990. Stress and acclimation. Pages 451±489 in C. Army Corps of Engineers, Project MPE-W-97-4, B. Schreck and P. B. Moyle, editors. Methods for Walla Walla, Washington. ®sh biology. American Fisheries Society, Bethesda, Sherwood, C. P., D. A. Jay, R. B. Harvey, P. Hamilton, Maryland. and C. A. Simestad. 1990. Historical changes in the Weitkamp, D. E., and M. Katz. 1980. A review of dis- Columbia River estuary. Progressive Oceanography solved gas supersaturation literature. Transactions 25:299±352. of the American Fisheries Society 109:659±702. Sigismondi, L. A., and L. J. Weber. 1988. Changes in White, R. G., G. Phillips, G. Liknes, G. Brammer, W. avoidance response time of juvenile chinook salmon Conner, L. Fidler, T. Williams, and W. Dwyer. 1991. exposed to multiple acute handling stresses. Trans- Effects of supersaturation of dissolved gases on the actions of the American Fisheries Society 117:196± ®shery of the Bighorn River downstream of the Yel- 201. lowtail Afterbay Dam. Final report to the U.S. Bu- Sprague, J. B. 1971. Measurement of pollutant toxicity reau of Reclamation, Bozeman, Montana. to ®sh III. Sub-lethal effects and ``safe'' concentra- Williams, J., G. Mathews, J. Myers, S. G. Smith, T. tions. Water Research 5:245±266. Cooney, and C. Toole. 1998. Hatchery extra mor- Thorpe, J. E., L. G. Ross, G. Struthers, and W. Watts. tality hypothesis. Pages 1±16 in D. R. Marmorek 1981. Tracking Atlantic salmon Salmo salar L. mi- and C. N. Peters, editors. PATH submissions to gration. Journal of Fish Biology 19:519±537. weight of evidence report. ESSA Technologies, USACE (U.S. Army Corps of Engineers). 1999. Draft Ltd., Vancouver. lower Snake River juvenile salmon migration feasi- Williams, J. G., and G. M. Mathews. 1995. A review bility report/environmental impact statement. USACE, of ¯ow and survival relationships for spring and Walla Walla, Washington. summer chinook salmon, Oncorhynchus tshawyts- EVIDENCE FOR HYDROSYSTEM-RELATED MORTALITY 51

cha, from the Snake River basin. U.S. National Ma- Zabel, R. W., and J. G. Williams. 2000. Comments on rine Fisheries Service Fishery Bulletin 93:732±740. ``Contrasting patterns of productivity and survival Wilson, P. With support form the hatchery evaluation rates for stream-type chinook salmon (Oncorhyn- group. 1996. Chapter 11. PATH: hatchery impacts. chus tshawytscha) populations of the Snake and Co- In D.R. Marmorek and 21 coauthors, compilers, and lumbia rivers'' by Schaller et al. (1999). Canadian editors. Plan for analyzing and testing hypotheses Journal of Fisheries and Aquatic Sciences 57:1739± (PATH): ®nal report of retrospective analysis for 1741. ®scal year 1996. ESSA Technologies, Vancouver. Zaugg, W. S. 1981. Advanced photoperiod and water Woo, N. Y. S., H. A. Bern, and R. S. Nishioka. 1978. temperature effects on gill Naϩ-Kϩ adenosine tri- Changes in body composition associated smolti®- phasphatase activity and migration of juvenile steel- cation and premature transfer to seawater in coho head (Salmo gairdneri). Canadian Journal of Fish- salmon (Oncorhynchus kitsutch) and king salmon eries and Aquatic Sciences 38:758±764. (O. tshawytscha). Journal of Fish Biology 13:421± Zaugg, W. S., E. F. Prentice, and F. W. Waknitz. 1985. 428. Importance of river migration to the development Wood, C. M., J. D. Turner, and M. S. Graham. 1983. Why do ®sh die after exercise? Journal of Fish Bi- of seawater tolerance in Columbia River anadro- ology 22:189±201. mous salmonids. Aquaculture 51:33±47. Wright P. B., and W. E. McLean. 1985. The effects of Zaugg, W. S., and H. H. Wagner. 1973. Gill ATPase aeration on the rearing of summer chinook fry (On- activity related to par±smolt transformation and mi- corhynchus tshawytscha) at the Puntledge hatchery. gration in steelhead trout (Salmo gairdneri) in¯u- Canadian Technical Report of Fisheries and Aquatic ence of photoperiod and temperature. Comparative Sciences 1390. Biochemistry and Physiology 45:955±965.