IDAHO COOPERATIVE FISH AND WILDLIFE RESEARCH UNIT

Transcription

1 Technical Report DRAFT IDAHO COOPERATIVE FISH AND WILDLIFE RESEARCH UNIT MIGRATION BEHAVIOR AND SPAWNING SUCCESS OF SPRING CHINOOK SALMON IN FALL CREEK AND THE NORTH FORK MIDDLE FORK WILLAMETTE RIVER: RELATIONSHIPS AMONG FATE, FISH CONDITION, AND ENVIRONMENTAL FACTORS By R. D. Mann, C. C. Caudill, M. L. Keefer, C.A. Peery 1 U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit University of Idaho, Moscow, ID C.B. Schreck and M. L. Kent Oregon Cooperative Fish and Wildlife Research Unit Departments of Fisheries and Wildlife and Microbiology Oregon State University, Corvallis, Oregon For U.S. Army Corps of Engineers Portland District, Portland OR Current affiliation: US Fish & Wildlife Service, 4147 Ahsahka Road, Ahsahka, ID 83520

2 Acknowledgements Many people assisted with work on this report and its successful completion was made possible through their efforts. This project was funded by the U.S. Army Corps of Engineers (USACE). We would like to thank: Dan Calvert for assisting with field work and data collection; Greg Taylor, Doug Garletts, Greg Gauthier, and Todd Pierce from the USACE Lookout Point office for field help and project coordination; David Griffith from USACE Portland District; Oregon Department of Fish and Wildlife staff, including Tom Friesen, Ken Kenaston, Fred Monzyk, Dan Peck, Tim Wright, Marika Dobos, and Kris Clemons; Oregon State University personnel for disease screening, especially Kristin Berkenkamp, Virginia Watral, and Trace Peterson, D.V.M.; and others who contributed their time to this project. Karen Johnson, Rose Poulin, Dan Joosten, and Mike Jepson from the U. of Idaho provided also provided support. ii

3 Table of Contents Acknowledgements... ii Table of Contents... iii Abstract... iv Introduction... 1 Methods... 3 Study Sites and Facilities... 3 Tagging and Assessment of Condition... 6 Proximate Analysis... 8 Temperature Monitoring... 9 Spawning Ground Surveys and Spawning Success Disease Screening Results Tagging Fall Creek Middle Fork Willamette Proximate Analysis River Conditions Migration Spawning Ground Surveys and Spawning Success Disease Screening External Findings Gross Findings Histological Findings Discussion Summary Energetic Condition and Spawning Success Environmental Effects Pathology Results Management Implications References Appendix iii

4 Abstract In recent years, high percentages (80-90%) of adult Chinook salmon transported above dams in some Willamette River tributaries have died prior to spawning. In 2008 and 2009 we surveyed the energetic status and survival rates of two populations of Willamette River spring Chinook salmon, monitored river environmental conditions, and investigated the relationships among prespawn mortality and a suite of potential causative factors including assessment of disease. A total of 395 Chinook salmon were sampled at Fall Creek, including 195 in 2008 and 200 in Fish were collected, assessed for energetic condition, PIT-tagged and/or radiotagged, and then transported above the dam and allowed to spawn naturally. In 2008, a total of 32 PIT-tagged salmon were recovered from spawning ground surveys on Fall Creek, a recapture rate of 16.4%. Three (9.4%) of the recovered fish were prespawn mortalities, and there were no clear relationships between mortality, initial energetic state, or fish morphometry in this small prespawn mortality sample. Notably, water temperatures in Fall Creek were uncharacteristically low and discharge rates were high in 2008, which may explain the relatively high survival rates. In 2009, prespawn mortality estimates were 84.8% for PIT-tagged fish and 89.5% for radiotagged fish. River temperatures in 2009 were very high (>25 C) at the outplant release site in late July. Prespawn mortality in 2009 was associated with rising temperatures and was significantly related to overall physical condition of adult salmon at the time of tagging. Due to small run size, tagged salmon were not outplanted into the North Fork Middle Fork (NFMF) Willamette River in 2008, but 239 fish were collected and tagged in To assess the effects of holding fish on prespawn mortality, the 2009 sample was separated into two groups: fish immediately outplanted and fish held at Willamette Hatchery prior to outplanting in late summer. Overall prespawn mortality of NFMF outplants was estimated to be 27% for radiotagged fish and 40% for PIT-tagged fish in Carcass recovery rates from the NFMF were low in 2009, precluding statistical analysis of holding effects. However, observational data suggested that holding in the hatchery may reduce prespawn mortality and we recommend further evaluation of this potential management strategy. Differences in prespawn mortality between Fall Creek and the NFMF release groups in 2009 were attributed to differences in thermal regimes and outplant release dates. Results from disease screening of adults collected during spawning surveys revealed that all of the fish examined had one or more of the following conditions that were, at least in part, responsible for death: swelling or hemorrhaging of internal organs, external evidence of trauma, and histological evidence of severe pathogenic infection. Severe infections of five pathogens (Ceratomyxa shasta, Nanophyetus salminocola, Apophallus sp., Echinochasmus milvi, and Parvicapsula minbicornis) were more frequent in adults that died prior to spawning compared to adults that successfully spawned. Nearly two-thirds of the fish that died prior to spawning were severely infected with one or more of these parasites. All adults were infected. Results of this study, in combination with previous Willamette River Chinook salmon studies, suggest prespawn mortality is caused by an interaction of environmental factors (particularly water temperature), fish condition and disease load, and energetic status. Multi- iv

5 year sampling of adult energetic status and other condition metrics will: 1) provide insights into the factors causing prespawn morality; 2) determine how mean salmon condition varies from year to year in response to environmental factors such as main stem conditions and ocean conditions; and 3) will assist in the development of effective management strategies to reduce prespawn mortality in Willamette River spawning tributaries. v

6 Introduction The numbers of adult spring-run Chinook salmon (Oncorhynchus tshawytscha) returning to the Willamette River, including tributaries managed as part of the USACE Willamette Valley Project (WVP), have fluctuated widely and have been near historic low levels in recent years. Development of the WVP began in 1941 and currently includes 13 dams and reservoirs on the Long Tom, Santiam, McKenzie, Middle Fork Willamette, and Coastal Fork Willamette subbasins. The WVP is managed for flood control, recreation, irrigation, fish and wildlife management, and power generation. Upper Willamette Chinook salmon populations in the WVP have declined for a variety of reasons, including habitat degradation, habitat loss associated with dams, land use practices, overharvest, pollution, changes in hydrologic and thermal regimes, and direct and indirect effects of artificial propagation (NMFS 2008). Due in part to these concerns, the upper Willamette River spring Chinook salmon run was listed as threatened under the U.S. Endangered Species Act in 1999 (NMFS 1999). Returning adults in many WVP populations can no longer reach much of their historic spawning grounds due to impassable dams on tributaries. Therefore an adult transportation program was initiated in the 1990 s to make use of surplus hatchery broodstock with the objectives of restoring a source of marine-derived nutrients and supplementing the prey base of native resident fish and wildlife, including other threatened species (i.e., bull trout, Salvelinus confluentus) (Beidler and Knapp 2005; Schroeder et al. 2007). Secondary benefits of outplanting include facilitating natural spawning of these populations above the dams and reconnecting habitats, and these secondary objectives have been elevated in recent years. There has been high prespawn mortality observed in adults some years since the start of the adult transportation program. Rates have been widely variable among years and among sub-basin populations (Schroeder et al. 2007; Kenaston et al. 2009; Keefer et al. 2010; Keefer and Caudill in review) and underlying mechanisms are not fully understood. Factors most likely to contribute to adult prespawn mortality are environmental stressors (especially water temperature), infectious disease, and poor energetic condition. The migration corridors of many rivers in the WVP have been altered by habitat degradation, hydroelectric installations, and climate change. In addition to the direct effects of passage barriers and lost access to spawning habitat, the operation of dam and reservoir systems for power production, recreation, and flood control can indirectly affect salmon and their migrations. Some important indirect effects are the alteration of river flow and temperature regimes. In many river systems, operating dams for flood control has resulted in more consistent flow regimes during migration. Depending on dam operation, water stored in reservoirs can either warm or cool downstream reaches when it is released. In the Columbia and Snake rivers, peak main stem water temperatures occur earlier in the year and warm temperatures persist later in the fall, compared to historic patterns (Quinn and Adams 1996; Quinn et al. 1997). In the Willamette system, tributary dams tend to cool downstream reaches in the spring and early summer and tend to increase water temperatures in the late summer and fall compared to the undammed system (e.g., Rounds 2007). The physiological effects of increased water temperatures during Chinook salmon migration, both below dams, and in tributaries during holding and spawning, may have negative effects on energy use and gonad development, potentially resulting in lower reproductive fitness for these populations. 1

7 Stress from trapping and transport efforts, in combination with disease, may also contribute to prespawn mortality (Schreck et al. 2001). The role of pathogens in prespawn mortality may be overlooked and underestimated because all salmon and most steelhead die shortly after they spawn and there are few attempts to document the proportion that die prematurely. Spawning fish are severely immune compromised, and thus even those that survive past spawning often are infected with a variety of pathogens. Therefore, infections and lesions in adult salmon in freshwater are considered normal, and commonly post-spawned fish exhibited a variety of infections and lesions. However, if infections become too severe, fish may succumb days or weeks before spawning, and thus may significantly impact the abundance of fry in the subsequent generation. It is well recognized that mature salmon are often infected with opportunists, such as saprolegnia fungus (Neitzel et al. 2004). Sockeye salmon in northern British Columbia succumbed to gill infections by Ichthyophthirius multifiliis (Traxler et al. 1998). This was later discovered to be a mixed infection of Ich and the microsporidium Loma salmonae, as both caused macroscopic white spots in gills (Higgins and Kent 1997; Shaw et al. 2000). Kent was the first to describe Parvicapsula minibicornis, a myxosporean that causes severe kidney disease and mortality in adult salmon once they return to freshwater (Kent et al. 1997; Jones et al. 2003). Kocan et al. (2004) concluded that chronic infections by protist, Ichthyophonus hoferi, has caused significant prespawning mortality in Chinook salmon for multiple years in the Yukon River, and a non-lethal PCR test was developed to track the prevalence of the infection in Chinook salmon from this river (Whipps et al. 2005). Other pathogens that have been documented to cause prespawn mortality include Dermocystidium, IHN virus, Ceratomyxa shasta (Bartholomew et al. 1989), and the bacterium Aeronomas salmonicida (Jones et al. 2004). Severe infections leading to significant prespawn mortality have consistently been associated with high water temperatures or other poor water quality conditions (Cooke et al. 2004). Many of these fish pathogens, and others, have been detected in Willamette River salmon (see Table 13). Migrating adult Chinook salmon do not feed during their upstream freshwater migration but rely on finite energy reserves accumulated while feeding in the ocean. Adult salmon die within days to weeks of spawning, indicating that energy stores are likely fine-tuned by past selection to maximize reproductive output (spawning and gametes) while also providing adequate energy to fuel upstream migration. The energetic costs of migration and spawning activities may have changed as a result of altered hydro- and thermographs, degradation of main stem and tributary habitat, and the effects of climate change. Thus it is possible that energy stores in Chinook salmon returning to the Willamette Valley may currently be mismatched to present conditions and possibly insufficient to allow successful spawning. Energy is primarily stored as lipids and energy content tends to be higher in populations traveling greater distances or that return to higher elevations (e.g., Crossin et al. 2004a). Within populations, there is evidence that energetic condition depends on growth conditions experienced in the ocean prior to return migration. For example, adult sockeye salmon return with lower reserves in years following relatively poor ocean feeding conditions (Crossin et al. 2004b). More generally, poor energetic condition at river entry (Crossin et al. 2004b; Rand et al. 2006) and 2

8 temperature regime during migration and on spawning grounds (Mann 2007; Crossin et al. 2008; Keefer et al. 2008) has been associated with higher probability of prespawn mortality. The primary goal of this study was to evaluate factors potentially associated with prespawn mortality in adult Chinook salmon, including environmental stressors, maturation status, disease, parasites, and initial energetic condition. Adults were collected at dams, assessed and tagged, and released above dams into spawning tributaries. In 2009, a subsample of adults collected at one site (Dexter Dam) were held at Willamette Fish Hatchery prior to outplanting to test the hypothesis that holding would decrease prespawn mortality rate. Carcasses were collected from spawning grounds and evaluated for spawning success and potential mortality sources. This included data collected from fish morphometrics, lipid content, and gross signs of disease and injury. Otoliths and scales were also collected from unmarked adults. Methods Specific objectives were: 1) Use non-lethal methods to estimate adult condition, size, and energy content upon tributary entry; 2) Analyze lipid and protein contents of the skin, muscle, viscera and gonad tissue to determine mean population levels before and after spawning; and 3) Determine fate of individual fish to investigate the inter-relationships among initial fish condition, river environment variables, fish transportation, and spawning success. A second component of the study examined disease to: 1) Determine pathogen infection rates in prespawned mortalities; 2) Determine trauma and lesions in prespawned mortalities; 3) Conduct preliminary comparison of pathogen infection rates and lesions in fish that died before spawning with those that survived to spawn in the wild and at Willamette hatchery; and 4) Determine if maturational state could affect rate of prespawning mortality. Study Sites and Facilities Tagging for this study took place at two sites in the upper Willamette Valley, west of Eugene, OR (Figure 1). The first site was at Fall Creek Dam on Fall Creek, a tributary of the Middle Fork of the Willamette River. The second was at Dexter Dam on the Middle Fork of the Willamette River. Dexter Dam regulates the outflow from Lookout Point Dam. The Fall Creek trap included a small ladder that led to a finger weir in front of a large collection area. A mechanical sweep was used to crowd trapped fish and raise them into a chute that dropped the fish into an anesthetic tank. The tank was lifted using a fixed crane and placed on the ground to facilitate fish tagging. 3

9 The Dexter trap was operated by Oregon Department of Fish and Wildlife (ODFW) and work at the site was coordinated with this agency. ODFW uses the Dexter facility to collect broodstock for the Willamette Fish Hatchery in Oakridge, OR. A fish ladder led to a slot weir at the entrance to a holding raceway. At the time of sorting, fish were mechanically crowded into an elevator which lifted them to an anesthetic tank. After fish were sedated, they were sorted and loaded into a hauling truck for transportation to either Willamette Fish Hatchery (WFH) or to release above the dams into the North Fork of the Middle Fork (NFMF) of the Willamette River. Tagging and assessment of energy condition occurred at both the Dexter Trap and the Willamette Hatchery. Only fish above the hatchery s quota for broodstock were transported above the Dexter and Lookout Point Dams for natural spawning (2009 only). To study the effects of holding on prespawn mortality in 2009, salmon tagged at Dexter Dam were included in one of two experimental groups. One group was transported to Willamette Fish Hatchery following tagging, and was held in the hatchery s adult holding pond until a release date just prior to spawning in the NFMF Willamette. The second group was released immediately following tagging into the NFMF Willamette. Figure 2 outlines the 2009 study design. 4

10 Figure 1. Map of the Middle Fork Willamette River basin showing Chinook salmon collection and outplant sites. Dams are numbered: 1 = Dexter Dam, 2 = Fall Creek Dam, 3 = Lookout Point Dam, and 4 = Hills Creek Dam. 5

11 Figure 2. Study design for All fish tagged at Fall Creek trap were immediately outplanted into Fall Creek. Fish collected and tagged at Dexter Dam were either immediately outplanted or were transported to Willamettee Fish Hatchery and held and outplanted later in the summer in the NFMFF Willamettee River. Tagging and Assessment of Condition At both trap sites fish were fully anesthetized prior to handling. Adults were anesthetized in approximately 60 ppm eugenol at Fall Creek trap. Sampling at Dexter trap and the Willamette Hatchery used MS-222 according to ODFW protocols. Following tagging, fish were loaded d into a truck filled with fresh river water and transported to an upstream release site or to the Willamette Hatchery. Oxygen was monitored during transportation with a targett concentration of 10 ppm. Tagging temperaturee was recorded and was generally less than 16 C because bottom-draw reservoir water was used for the anesthetic tank and hauling truck at both sites. While fish were anesthetized, fish were sexed and inspected for clips or markings. A composite condition score was recorded based on injuries, marine mammal marks, headburn, parasites, and descaling. A score of three indicated no obvious damage or minimal healed 6

12 scrapes, two indicated minor or healed injuries with potential scarring, and one indicated a fish had open/severe wounds or a large number of minor injuries. Fish were PIT tagged anterior to the pelvic fins in 2008, with the tag inserted into the muscle tissue of the pelvic girdle. In 2009, fish were PIT tagged in the dorsal sinus, near the back of the dorsal fin in an effort to increase tag retention on scavenged carcasses. Fork lengths to the nearest half centimeter were taken as well as four morphological measures previously used to estimate energetic status (Mann et al. 2009). These measurements were used to evaluate the relationships between fish body size and shape and estimates of energy content (i.e., Fatmeter estimates) as well as possible associations with prespawn mortality. The metrics were: mid-eye to hypural length, hump height, depth at anus, and breadth at anus (Figure 3). Mid-eye to hypural length was defined as the distance along the lateral line from the middle of the eye to the end of the scales on the hypural plate on the caudal peduncle. Hump height was the distance from the anterior origin of the dorsal fin to the lateral line, perpendicular to the lateral line. Depth at anus was the total depth of the fish perpendicular to the lateral line at the anal opening. Breadth at anus was the width of the fish at the intersection of the lateral line and a theoretical line perpendicular to the lateral line at the anus (not collected in 2008). Morphometric measurements were taken using calipers and recorded to the nearest mm. Fish weights (to the hundredth of a kilogram) were collected using a flat table scale (Ohaus washdown bench scale, Ohaus Corp., Pine Brook, NJ). The percentage of lipids in the muscle tissue was used as the primary estimate of energy condition because lipids are the primary energy reserve fish use during migration and spawning (Brett 1995). Lipid levels were estimated using a Distell Fatmeter (Distell Industries Ltd., West Lothian, Scotland). The Fatmeter was developed in the commercial fish industry to estimate the percent of lipids in a trimmed fillet. The meter uses a low energy microwave sensor to estimate water content in the muscle tissue. Based on the inverse relationship between water and lipid levels in fish tissue (Craig et al. 1978; Higgs et al. 1979), the meter estimates the percent lipid in Chinook salmon muscle tissue using a proprietary algorithm. We used proximate analysis of tissues in each study year (see below) to validate estimates and correct for any error in the Fatmeter algorithm. Four readings were taken just above the lateral line, progressing toward the posterior of the fish and the average recorded for each fish. A sub-sample of fish was also radio-tagged in both years (Fall Creek: 2008 n =7, 2009 n =25; NFMF: 2009 n =12). A 7-volt transmitter (Lotek Wireless Inc., New Market, Ontario; MCFT-7A, 80 mm 16 mm diameter, 29 g in air) or 3-volt transmitter (MCFT-3A, 43 mm 14 mm diameter, 11 g in air) was inserted gastrically through the mouth. A latex band was placed on each transmitter to reduce regurgitation (Keefer et al. 2004). Fish >78 cm were randomly selected for radio tagging. The purpose of radio tagging a small portion of fish was to verify if fish were moving upstream after release and to determine if fish migrated back into the reservoir. In past years, the latter behavior has been linked with prespawn mortality (Keefer et al. 2010). Additionally, the use of radio tags aided in the collection of carcasses for prespawn mortality assessments. In 2009, all radio-tagged fish were sampled for blood from the sub-vertebral caudal vessel posterior to the anal fin. The blood sample was centrifuged for a minimum of four minutes until the red blood cells separated from the plasma. A hematocrit was recorded for each fish. The plasma was transferred to a vial using a pipette, and immediately stored on ice. 7

13 Samples were frozen as soon as possible. Plasma samples were frozen for use in determining sexual maturation based on hormone levels. HH MeH Da Ba Figure 3. Diagram of morphometrics collected. MeH = Mid-eye to hypural length, HH = Hump height, Da = Depth at anus, Ba = Breadth at anus. Proximate Analysis A small sub-sample of fish was lethally sampled to estimate mean lipid, protein, water, and ash amounts in tissues and to validate the accuracy of the Fatmeter estimates of energy condition. Processing fish entailed partitioning the fish carcass into 4 tissues types; muscle, skin, viscera and gonads (e.g., Mann et al. 2009). Each of the tissues was removed as entirely as possible from a carcass, and weighed to the nearest gram to establish the total weight of each tissue type. Then each tissue was homogenized independently in a Cuisinart food processor and a 50 gram subsample of the homogenate was taken. The samples were frozen and later transported to Washington State University where they underwent proximate analysis. Proximate analyses were performed using established methods. Lipid amounts were calculated by passing volatized ether through the 50 g tissue samples which removed all ethersoluble products including lipids. Lipids were then extracted from the ether, dried and weighed (AOAC 1965). Ash content was calculated by combusting weighed samples at ºC for 12 hours and reweighing (AOAC 1965; Craig et al. 1978). The percent moisture in the samples was obtained by placing a weighed sample in a freeze drier at -40º C for 24 to 36 hours and reweighing. Protein content was determined by subtraction (% protein =100 - % water - % fat - % ash), as in other studies on salmon energetics (e.g., Berg et al. 1998; Hendry and Berg 1999; Hendry et al. 2000). Carbohydrate content was assumed to be negligible. After lipid weights were calculated for each 50 gram subsample, we calculated total lipid per tissue and total body lipid levels. Energy density or gross somatic energy was calculated as kj of energy per kg of fish mass, assuming energy equivalents for fat and protein of 36.4 kj g -1 and 20.1 kj g -1, respectively (Brett 1995). Total energy included gonadal tissues. Gross somatic energy density (kj/kg) was used as a second measure of energy condition and was calculated for the lethally sampled fish in 2008 and Gross somatic energy density 8

14 represents the energy density contained within somatic tissues of the fish and is a measure of energy contained not only in the muscle tissue, but also the viscera and skin (Crossin and Hinch 2005). Because it is standardized by mass, it can be directly compared among individuals. Gross somatic energy density was regressed on lipid percentage (natural log [log e ] transformed) estimated by the Fatmeter (non-standardized values, see below) to examine the relationship between Fatmeter estimates and gross somatic energy density (e.g., Colt and Shearer 2001; Crossin and Hinch 2005). We used linear regression to estimate the relationship between muscle lipid content and Fatmeter readings. The relationship was then used to estimate muscle lipid content for each outplanted fish by inverse prediction (Sokal and Rohlf 1995) using Fatmeter measurements taken at the time of tagging. Henceforth, we refer to the corrected lipid estimates for outplanted fish as standardized lipid percentage. We used regressions without an intercept for standardization to force lines through the origin because neither proximate analysis results nor Fatmeter percentages can have negative values. We also estimated relationships using standard regression (with an intercept) to evaluate the statistical significance and strength of the relationships because statistics from regression without an intercept (e.g., r 2 ) are difficult to interpret (Sokal and Rohlf 1995). Morphometric measures listed above were also taken on all lethally sampled fish. A suite of multiple linear regressions were performed between the morphometrics, the standardized Fatmeter results, and gross somatic energy calculated from proximate analysis. We evaluated the association between energy parameters and several ratios estimating aspects of fish shape (e.g., breadth at anus: fork length, a measure of fat vs. skinny ). Multi-model selection (AIC) analysis was used to evaluate which models might be suitable to estimate energy content of nonlethally sampled fish. Temperature Monitoring Temperature recorders were installed in 2009 at ten locations in Fall Creek and the NFMF Willamette River. In Fall Creek, loggers were located at the release site (rkm 23.0), the bridge near Johnny Creek (rkm 30.6), near the mouth of Portland Creek (rkm 34.1), the bridge at forest road 1828 (rkm 37.0), and at the unnamed falls that act as a fish barrier (rkm 47.2). In the NFMF Willamette River, loggers were placed at the USACE screw trap site (rkm 2.3), the release site (rkm 30.0), below the bridge near Kiahanie campground (rkm 37.2), at the forest road 1944 bridge (rkm 44.6), and above Skookum Creek (rkm 58.0). Temperatures were logged at 1 h intervals from early June to mid-october. Temperatures were available from 2008 at the 1828 Bridge on Fall Creek, and were used for comparison between years. We used ibcod submersible temperature loggers (Alpha Mach, Inc., Mont-St-Hilaire, Quebec; 35x25x12 mm, 12g in air) to record internal temperatures on a small subsample of radio-tagged fish. The tags were slipped on the wire of the radio tags, and inserted gastrically. The temperature recorders were recovered during carcass surveys and were downloaded. Spawning Ground Surveys and Spawning Success 9

15 After translocation to spawning areas above the dams, salmon were allowed to spawn naturally and spawning areas were monitored to collect carcasses and assess spawning success. In 2008, most carcass surveys were conducted from August to the end of October, although limited monitoring did occur in July. In 2009, carcass surveys were conducted by both UI and ODFW on a regular basis from the beginning of releases through the spawning period (June through early October). Fish encountered during spawning ground surveys were inspected by UI and/or ODFW personnel for FLOY and PIT tags. When the carcass of an individual from this study was located, it was inspected to determine spawning status, its general condition was noted (how recently it died, obvious wounds, fungus levels, or other apparent visual cues that caused mortality) and measured again using the Distell Fatmeter (2008 only). If a fish had recently died (gills were pink), tissue samples were collected as described above for proximate analyses. Tissue samples were collected for histology as described below. Surveys included the reach below Dexter Dam and results were compared to the transported group. Surveys were not conducted below the Fall Creek collection site because few adults typically spawn in that reach (G. Taylor, pers. comm.). Spawning success was assessed by inspecting the gonads of females and estimating the proportion of gametes remaining to the nearest 25%. A successfully spawned fish was defined as having less than 25% of gametes remaining in the body cavity (Pinson 2005). Survival to the first day of spawning activity was also used in analyses as a metric of reproductive success because the proportion of remaining gametes could not be reliably estimated in most males or in some carcasses that had been scavenged. Multi-model selection, including univariate and multiple logistic regression models with adult fate (spawned, prespawn mortality) as the dependent variable were used to evaluate potential correlates with prespawn mortality. Disease Screening We evaluated the health status of fish using histology as our primary diagnostic tool, coupled with macroscopic changes recorded at necropsy in the field. We sampled from moribund or very recently dead fish. We conducted a complete necropsy on available fish. Necropsies consisted of close visual examination of each individual for signs of injury and lesions, both on the exterior and interior. All fish and abnormalities were photographed. Tissues were then collected and fixed in buffer formalin for later histological analysis. Heart, brain, gills, liver, spleen, kidney, gonads, pyloric caeca, and lower intestine were collected and fixed in 10% neutral buffered formalin. Additional kidney and spleen samples from each fish were archived at -80 C for potential PCR identification of bacterial and viral pathogens, as well as potential confirmatory identification of parasites. Fixed tissues were processed using routine microtechniques involving H and E staining in the laboratory; specialty stains were then employed as appropriate. Resultant microscope slides were read by a trained histopathologist to determine presence of pathogenic organisms and an assessment of infection rate was made. The medical interpretation was then confirmed by a second trained pathologist. The data allowed assessment of the potential for morbidity associated with these pathogens. Moribund fish were identified as follows: still alive, but listless, lethargic and showing marked ataxia (inability to maintain proper body position in 10

16 water). Fish not showing ataxia, but that were lethargic to the extent that they do not respond to physical stimuli (e.g., poking) by swimming away were considered moribund if needed to meet minimum sample sizes. Maturation status of radio-tagged fish will be determined from blood samples taken at the time of tagging and the samples currently await analysis at the time of this writing. Plasma was separated by centrifugation and stored at -80 C. Sex steroids will be quantified by radioimmunoassay as described by Fitzpatrick et al. (1986). These assays with EPA-level QA/QC criteria are routinely run in Schreck s laboratory. The OSU laboratory worked out the ability to determine reproductive development of Coho salmon based on sex hormone levels in the blood (Fitzpatrick et al. 1986). We realize that Chinook salmon have a different maturational chronology than Coho salmon, but some of the same principles likely apply. Results Tagging Fall Creek In 2008, PIT tagging occurred from 15 May to 14 July. A total of 195 fish (87 females, 108 males) were tagged at Fall Creek (Figure 4). Seven of the 195 fish at Fall Creek were also radio-tagged, and all were transported above Fall Creek Dam. Three fish (1.5%, 2 females, 1 male) had adipose fin clips. In general, fish were in good overall condition (mean condition score = 2.34, Table 1), with low incidence of headburn (4.1%) and few marine mammal scars. The mean fork length was 73.1 cm, mean weight was 5.0 kg, and mean standardized lipid percentage was 7.8% (Table 1). Lipid percentages averaged close to 10% at the beginning of the run and were mostly 4-5% at the end of the run. A steady decline in lipid percentages was seen through the migration season (Figure 5), indicating a loss of energy reserves as the summer progressed. In 2009, tagging occurred from 26 May to 24 August. A total of 200 fish (85 females, 115 males) were PIT tagged, and 25 were also radio-tagged (Figure 6). Twenty-eight of the 200 fish had adipose fin clips, although this was not an accurate representation of the proportion of hatchery fish in the run because no adipose-clipped fish were transported above the dam after 15 June. The mean condition score in 2009 was 2.17, mean fork length was 74.0 cm, mean weight was 4.7 kg, and mean lipid percentage was 4.7% (Table 2). As in 2008, lipid percentages decreased through the season in 2009 as energy stores were depleted (Figure 7). The mean lipid percentage in 2009 was lower than recorded in 2008, but the difference was related at least in part to later collection dates in Additionally, there were differences in equipment readings between years that may account for a portion of the betweenyear difference (see the results on proximate analysis for more details). 11

17 70 60 Fall Creek n = Tagged Salmon Dexter n = 105 5/5/08 5/19/08 6/2/08 6/16/08 6/30/08 7/14/08 7/28/08 Date Figure 4. Numbers of adult Chinook salmon tagged in The Fall Creek data were representative of the overall run timing, as nearly every fish that returned to the trap from 15 May to 12 July was tagged. All the fish tagged at Dexter trap were transported to Willamette Hatchery and used for broodstock; none were outplanted. 12

18 Table 1. Adult Chinook salmon size, lipid content, and condition metrics for fish sampled at Fall Creek trap in MeH = Mid-eye to hypural length, Da = Depth at anus, HH = Hump height, % Lipid = Standardized % lipid in muscle tissue, wet weight. Fall Creek (n = 195) Fork Length (cm) Weight (kg) MeH (cm) Da (cm) HH (cm) % Lipid Condition Score Mean St. Deviation Max Min Table 2. Adult Chinook salmon size, lipid content, and condition metrics for fish sampled at Fall Creek trap in MeH = Mid-eye to hypural length, Da = Depth at anus, Ba = Breadth at anus, HH = Hump height, % Lipid = Standardized % lipid in muscle tissue, wet weight. Fall Creek (n = 200) Fork Length (cm) Weight (kg) MeH (cm) Da (cm) Ba (cm) HH (cm) % Lipid Condition Score Mean St. Deviation Max Min

19 20 18 Standardized Fat Meter % n=10 n=12 n=14 n=26 n=75 n=11 n=25 n=14 n=8 Before May 25 May 26 June 2 June 9 June 16 June 23 June 30 July 7 July 14 Week Figure 5. Weekly distributions of Fatmeter results for Chinook salmon tagged at Fall Creek trap in Box plots represent median (solid line), mean (dashed line), 25 th and 75 th percentiles (ends of boxes), 10 th and 90 th percentiles (whiskers), and individual outliers (solid circles). 14

20 50 40 Fall Creek n = Tagged Salmon Dexter Held n = 105 Outplant n = 135 Col /4/09 5/18/09 6/1/09 6/15/09 6/29/09 7/13/09 7/27/09 8/10/09 8/24/09 Date Figure 6. Number of adult Chinook salmon tagged in White bars represent fish that were sent to Willamette Hatchery and held before being outplanted on August 24 th. Black bars represent fish that were immediately outplanted to the NFMF Willamette River. 15

21 Standardized Fat Meter % n=30 n=15 n=29 n=56 n=8 n=21 n=11 n=10 n=20 May 25 June 1 June 8 June 15 June 22 June 29 July 6 Late July August Week Figure 7. Weekly distributions of Fatmeter results for Chinook salmon tagged at Fall Creek trap in Box plots represent median (solid line), mean (dashed line), 25 th and 75 th percentiles (ends of boxes), 10 th and 90 th percentiles (whiskers), and individual outliers (solid circles). Middle Fork Willamette In 2008, fish were not sampled at Dexter trap until the trap was opened to collect fish for broodstock at Willamette Hatchery on 18 June. A total of 204 fish (120 males, 84 females) were then sampled at the hatchery at approximately one week intervals coinciding with the operation of the Dexter trap by ODFW. Adults collected at Dexter were slightly larger than the Fall Creek sample, on average. Mean length was 77.3 cm, mean weight was 5.4 kg, mean condition score was 2.42, and mean standardized percentage of lipids was 5.4% (Table 3). The latter may reflect the one-month later start of collection at Dexter trap. In 2009, fish that were held at Willamette Hatchery before outplanting were tagged from 24 June to 9 July. This group had 104 fish (23 males, 81 females). The mean condition score 16

22 was 2.30, mean fork length was 77.6 cm, mean weight was 5.4 kg, and mean standardized lipid percentage was 5.1% (Table 4). Tagging of the immediate outplant group began on 14 July and continued until 17 August. This group included 135 fish, with mean length of 74.9 cm, mean weight of 4.5 kg, mean condition score of 2.25, and mean standardized lipid percentage of 3.5% (Table 5). On August 24 th, 2009, fish originally tagged at Dexter Dam trap and held at Willamette hatchery were outplanted into the NFMF Willamette River. A second Fatmeter reading was taken on all fish before outplanting; however, no individuals showed lipid percentages greater than 3.0% on reassessment (Figure 8). Table 3. Adult Chinook salmon size, lipid content, and condition metrics for fish collected at Dexter trap and sampled at Willamette Hatchery in All fish were used for broodstock. MeH = Mid-eye to hypural length, Da = Depth at anus, HH = Hump height, % Lipid = Standardized % lipid in muscle tissue, wet weight. Dexter (n = 204) Fork Length (cm) Weight (kg) MeH (cm) Da (cm) HH (cm) % Lipid Condition Score Mean St. Deviation Max Min Table 4. Adult Chinook salmon size, lipid content, and condition metrics for fish collected and sampled at Dexter trap and held at Willamette Hatchery in MeH = Mid-eye to hypural length, Da = Depth at anus, Ba = Breadth at anus, HH = Hump height, % Lipid = Standardized % lipid in muscle tissue, wet weight. Dexter (n = 104) Fork Length (cm) Weight (kg) MeH (cm) Da (cm) Ba (cm) HH (cm) % Lipid Condition Score Mean St. Deviation Max Min Table 5. Adult Chinook salmon size, lipid content, and condition metrics for fish collected and sampled at Dexter trap and then immediately outplanted in MeH = Mid-eye to hypural length, Da = Depth at anus, Ba = Breadth at anus, HH = Hump height, % Lipid = Standardized % lipid in muscle tissue, wet weight. Dexter (n = 135) Fork Length (cm) Weight (kg) MeH (cm) 17 Da (cm) Ba (cm) HH (cm) % Lipid Condition Score Mean St. Deviation Max Min

23 16 Standardized Fatmeter Percentage Tagging Outplant Tagging Outplant Tagging Outplant Week 1 Week 2 Week 3 Figure 8. Distributions of Fatmeter results for Chinook salmon held at Willamette Hatchery and then outplanted in Tagging data were collected on the dates fish were tagged, and outplant data were for when fish were reassessed and outplanted on 24 August. Box plots represent median (solid line), mean (dashed line), 25 th and 75 th percentiles (ends of boxes), 10 th and 90 th percentiles (whiskers), and individual outliers (solid circles). 18

24 Proximate Analysis In 2008, proximate analysis was performed on 11 salmon, including five males from Fall Creek, and two females and four males from Dexter (Table 6). Lethal takes for proximate analysis were conducted from 16 June to 11 July. Females had gonadal lipid compositions of 7-9%, while males were much lower at approximately 1%, as expected because eggs have high lipid content. The average muscle lipid level was 6% and ranged from %. Males need less energy for the gamete production, but potentially store more energy in muscles than females for use in spawning activity. In 2009, proximate analysis was performed 15 fish (seven females, eight males) at Fall Creek Dam, and 14 fish (seven females, seven males) at Dexter Dam (Table 7). As in 2008, females had higher percentage of lipids in gonads compared to muscles and males had higher lipid content in muscles than gonads. Table 6. Mean tissue composition of Chinook salmon used in proximate analysis in Tissue % Moisture % Crude Lipid % Total Ash % Protein Combined (n = 2 females, 9 males) Gonads Muscle Skin Viscera Fall Creek (n = 5 males) Gonads Muscle Skin Viscera Dexter (n = 2 females, 4 males) Gonads Muscle Skin Viscera

25 Table 7. Tissue composition of Chinook salmon used in proximate analysis in Tissue % Moisture % Crude Fat % Total Ash % Protein Combined (n = 14 females, 15 males) Gonads Muscle Skin Viscera Fall Creek (n = 7 females, 8 males) Gonads Muscle Skin Viscera Dexter (n = 7 females, 7 males) Gonads Muscle Skin Viscera Fatmeter readings were taken on proximate analysis fish at the time of trapping to simultaneously assess the accuracy of the fat meter readings and provide regression equations to calculate standardized values. The relationship appeared to vary between years (see Figure 9) so we used separate regressions from each year to standardized lipid estimates within each year and to compare 2008 and 2009 results. We used linear regression, formulated without an intercept term to force lines through the origin because neither proximate analysis results nor Fatmeter percentages can be negative values. The results from the Fatmeter were compared to the results for crude lipid percentage in the muscle tissue (by wet weight; Figure 9). Modest positive correlations were found in both years (r 2 = 0.18 in 2008 and r 2 = 0.38 in 2009; r 2 based on regressions with constants; Table 8). The relationship for males was stronger than for females in The 2008 result was not as strong a relationship as we have seen in previous studies (e.g., Anderson 2007). The weak relationship was most likely a result of performing proximate analysis across a small range of lipid percentage values during a narrow window of the migration season. Performing proximate analysis on post-spawning or earlier-run fish in 2008 would likely have resulted in a stronger overall relationship because a wider range of lipid levels would have been present in the sample. The standardized lipid percentage in adults collected at Fall Creek trap declined seasonally in both years (Figure 10). After standardizing the Fatmeter results, lipid levels in 2009 were slightly higher than those recorded in 2008 for each week and this difference could reflect difference in initial condition or seasonal timing (or both). 20

26 Fat Meter Percentage Males 2008 Females 2009 Males 2009 Females 2008 Regression 2009 Male Regression 2009 Female Regression Proximate Analysis (% Lipid in Muscle, Wet Weight) Figure 9. Fatmeter percentages versus percent muscle tissue lipid calculated in the proximate analysis. Open symbols are for 2008 and black symbols are for Circles are for males and triangles are for females. Linear regressions forced through the origin are shown for reference. Sample sizes did not permit separate regressions by sex in Equations are given in Table 8. We chose to estimate lipid in muscle for tagged fish, because the Fatmeter was designed to estimate this parameter and because a large proportion of energy is stored in muscle tissue. In addition, we also tested whether the estimates from the Fatmeter provided good estimates of total energy in all body compartments combined (muscle, skin, and viscera). Specifically, we estimated whole-body somatic energy density (KJ/kg), which also standardizes energy content for differences in fish size. Although the relationship in 2008 was non-significant, presumably due to small sample size and a relatively small range of variation measured, we found a strong positive relationship between Fatmeter readings and energy density in 2009 (Figure 11). This result suggests that the Fatmeter of muscle lipid content also provided accurate non-lethal estimates of whole-body fish energy condition. 21

27 Table 8. Linear regression results that show the relationships between Fatmeter percentages (FM) and percent lipid in wet weight muscle tissue calculated in proximate analysis (PA). These equations were used to obtain standardized Fatmeter estimates for individual adults. *Statistical results shown are based on regressions with constants (see Methods). Year n Equation F* P* r 2* FM = 0.946*PA (3.228) (0.106) (0.182) 2009 Combined 29 FM = 0.601*PA (17.814) (<0.0001) (0.375) Male 15 FM = 0.510*PA (36.822) (<0.0001) (0.719) Female 14 FM = 0.786*PA (10.195) (0.008) (0.414) Our evaluation of the relationships between fish size and morphometry metrics and both standardized Fatmeter % and gross somatic energy showed that it may be possible to use measurements of salmon size and shape to estimate lipid content. Because the morphometry and size variables were intercorrelated in our sample (29 fish with proximate analysis data in 2009; Appendix Table 1) we compared all possible combinations of these variables and used AIC to identify the best-fit model(s) (Appendix Table 2). The basic conclusion was that about 45% of the variability in the standardized Fatmeter % and about 60% in the variability in gross somatic energy for the full sample (males and females combined) could be explained by the fish metrics. We also found that the best models differed for males and females and for the two energy content measures. In general, the ratios of breadth at anus to fork length (Ba:FL) or breadth at anus to depth at anus (Ba:Da) were included in most models. Each of these terms is an indicator for the overall shape of the fish (i.e., skinny versus fat). Models that included fish collection date explained more of the variability than models that included only fish measurements. The date term reflected the loss of lipid reserves through the summer (for fish of a given size), as clearly shown in the correlations between date and several of the size metrics (Appendix Table 1). 22

28 Standardized Fatmeter % Week Figure 10. Weekly standardized Fatmeter lipid percentages for Chinook salmon tagged at Fall Creek trap in 2008 and Standardized Fatmeter percent estimated the lipid percentage in muscle tissue, wet weight. 23

29 Gross Somatic Energy (kj/kg) : y = x R 2 = P < : P = Ln Fatmeter Percentage Figure 11. Relationship of Chinook salmon energy density (kj/kg) to log e transformed raw Fatmeter percentages. Energy density data were calculated from proximate analysis. The regression line is for 2009 data only; the relationship for 2008 was non-significant. River Conditions The spring and summer of 2008 were marked by high flows and cool water temperatures in the Willamette migration corridor resulting from a large snow pack in the Cascade Range. During the peak of the summer freshet, discharge reached 25,000 cfs at Albany, Oregon, approximately the half-way point for Middle Fork migrants in the Willamette River (Figure 12). River discharge did not reach regular maintenance flows (~5000 cfs) until mid-july. In 2009, river discharge conditions were closer to average, with a large early-spring freshet, followed by discharge returning to maintenance levels in late June. Daily mean temperatures in the mid-sections of the Willamette River did not exceed 15 C until mid-june, much later than has been observed in recent years, and maximum temperatures were slightly above 20 C (Figure 13) temperatures were favorable for Chinook salmon migration, as they were within the thermal preferendum of this species and were 3-4 C lower than incipient lethal temperatures (Orsi 1971; Coutant 1977; Richter and Kolmes 2005). River water temperatures in the lower reaches of Willamette River tributaries were cooler than average in 2008 (Figure 14). 24

30 Figure 12. Willamette River discharge data collected at the USGSS monitoring site near Albany, Oregon. The site represents the approximate half-way point for migration from the mouth of the Willamette River to Fall Creek. Data from are plotted with the daily mean from

31 24 22 Temperature ( C) Date Figure 13. Willamette River temperature data collected at the USGS monitoring site near Albany, Oregon. The site represents the approximate half-way point for migration from the mouth of the Willamette River to Fall Creek. Data from are plotted. 26

32 Figure 14. South Fork McKenzie River water temperature data collected at the USGS monitoring site. These data weree used as a proxy for the Middle Fork Willamette because of its proximity to the field sites, and the lack of dam-regulated flow. There are no USGS sites within the Fall Creek or Middle Fork Willamette basins that are unaffected by reservoirs and dam outflow. There were substantial differences in thermal regimes between years and between Fall Creek and the NFMFF Willamettee River. Temperatures were available at a single site from Fall Creek in both 2008 and 2009, and water temperatures at this site were higher in 2009 than in 2008 (Figure 15). In 2009, temperature loggers were deployed to record ambient temperatures at multiple sites in Fall Creek and the NFMF Willamette River. Water temperatures in Fall Creek in July 2009 were extremely high, due to high summer air temperatures and lack of precipitation. Maximumm water temperatures exceeded 25 C in the lower sections of the river in the last four days of July (Figure 16), temperatures that exceeded the incipient lethal temperature for Chinook salmon. Minimum exposures to these high temperatures can cause rapid death or other non- lethal complications. Daily mean temperatures exceededd 23 C at the Fall Creek release site (Figure 16). Temperatures were also monitored in the NFMF at five locations from June to October in 2009 (Figure 17). Daily mean temperatures never exceeded 16 C at the release site. These temperatures were near the thermal preferendum for Chinook salmon. Temperatures were higher downstream at the screw trap site, reaching 20 C. 27

33 Temperature ( C) Temps 2009 Temps Date Figure 15. Comparison of mean daily water temperatures collected in Fall Creek in 2008 and 2009 near the Forest Road 1828 bridge (rkm 37.0). The 2008 data were supplied by the USFS. 28

34 Temperature ( C) Release Site Johnny Cr. Portland Cr Bridge Fish Barrier 8 6 Date Figure 16. Daily mean water temperatures in 2009 at five sites in Fall Creek. The loggers represent a progression upstream from the release site (rkm 23.0) to the fish barrier (rkm 47.2). Although the release sites at Fall Creek and the NFMF Willamette were located 16.5 miles from each other, there were large differences in temperature in 2009 due to differences in elevation and watershed characteristics (Figure 18). In particular, daily mean river temperatures in the NFMF averaged 5.7 C lower than in Fall Creek at the release sites during the July and August holding period. This difference likely had a large impact on fish outplanted into the two river systems. 29

35 Temperature ( C) Screw Trap Site Release Site Kiahanie 1944 Bridge Skookum Cr. Date Figure 17. Daily mean water temperatures in 2009 at five sites in the North Fork Middle Fork Willamette River. The loggers represent a progression upstream from the screw trap site (rkm 23.0) to Skookum Creek (rkm 47.2). 30

36 Figure 18. Comparison of mean daily water temperatures collected in Fall Creek and NFMF Willamette River in Sites in Fall Creek are in gray, and sites in the NFMFF Willamettee River are in black. Temperatures at sites where Chinook salmon were outplanted are shown with dotted lines. Migration In both years, all fish tagged at Fall Creek were transported above the dam to facilitate natural spawning. The 2008 radiotelemetry data indicated that fish released into Fall Creek moved upstream 2 km from the release site within one day of release. No fish were recorded returning to the reservoir after release in In 2009, at least two fish returned to the reservoir. This behavior has been associated with prespawn mortality in previous years (Keefer et al. 2010). Extensive radio tracking was conducted at least once a week through the migration season in The majority of fish were observed and most fish movements occurred within a week of release (Figure 19). After this initial movement, fish spent as many as 80 days within the same river reach or in sites with deep holding water (i.e., pools). Fish then resumed movement two to three weeks before spawning, presumably to find suitable spawning habitat. Of the two fish that used the reservoir in the summer, one moved back into the river prior to 31

37 spawning successfully and the other was not recorded again and apparently died in the reservoir. Both fish were released on 9 July, 2009 as water temperatures were increasing (Figure 16). Chinook salmon migration behaviors differed in the NFMF Willamette River compared to Fall Creek. Unlike in Fall Creek, we saw very little movement immediately after release in the NFMF (Figure 20). Several factors may have contributed to the difference in behavior. The release site for the NFMF was in a reach with ideal spawning habitat, and fish were released much later in the spawning season than they were in Falll Creek due to the broodstock collection protocols at Dexter Dam/Willam mette Hatchery. Additionally, there were two large holding pools immediately upstream and downstream of the NFMF release site. The downstream pool, in particular, held ~60 or more fish until spawning started. Finally, temperatures were generally cooler in the NFMF. A sub-sample of Fall Creek radio-tagged fish had a temperature tag to record internal temperatures during migration above the dams and to examine possible thermoregulatory behavior; eight temperature tags were recovered, and seven had retrievable data. Figure 19. Movement summaries for 25 radio-taggedd Chinook salmon released in Fall Creek in Each fish is represented by a line that starts at the release site (rkm 23.0) and follows their migration as recorded during mobile radio tracking. The line ends when the transmitterr and/or fish was recovered. The two fish thatt survived > 80 d were the two successful spawners. River kilometers of the temperature logger sites and top of reservoir are shown for reference. 32

38 Figure 20. Movement summaries for 11 radio-taggedd Chinook salmon released in the NFMF Willamette River in Each fish is represented by a line that starts at the release site (rkm 30.0) and follows their migration as recordedd during mobile radio tracking. The line ends when the transmitter and/or fish was recovered. River kilometers of the three temperature logger sites are shown for reference. Temperatures recorded on internal temperature loggers matched in-stream temperature data when matched with mobile tracking locations (see Figures 21 and 22 for two examples) ). These results suggest that there was little opportunity for behavioral thermoregulation within individual river reaches (i.e., temperatures were fairly homogenous throughout the stream and we did not observe adults with body temperatures lower than the nearest downstream logger temperature). Movements into upper river reaches appeared to be the only way fish reducedd body temperature. While some fish did move rapidly upstream to cooler sections of Fall Creek, fish weree observed in deep pools throughout the entire available river (Figure 22). It is unknown whether fish that stayed in warmer lower reaches lacked energy reserves to seek cooler temperatures upstream, if other behavioral factors were at play (e.g., territorial or innate anti- predator behaviors), or if there was some energetic or developmentall benefit to holding in warmer water. 33

39 Temperature ( C) Thermograph Average Release Johnny Cr Portland Cr 1828 Bridge Fish Barrier Mobile Tracking 47.0 Fish Barrier Bridge 35.0 Portland Cr Johnny Cr Rkm 8 6/14/2009 6/19/2009 6/24/2009 6/29/2009 7/4/2009 7/9/2009 7/14/2009 Date 23.0 Release Site Figure 21. Internal temperature recorder data collected after outplanting in Fall Creek of the Chinook salmon with radio transmitter The light line background shows the daily fluctuations in body temperature, while the dashed black line is the daily mean body temperature. Daily means of the in-stream temperature logger sites are shown with corresponding locations on the second y-axis. Dotted line indicates fish position as obtained by mobile tracking. At release, body temperature was similar to the release site and quickly declined as the adult moved upstream; thereafter body temperature was intermediate between temperatures recorded at the nearest up- and downstream location (fish barrier and 1828 bridge respectively). 34

40 Figure 22. Internal temperature recorder data collected after outplanting in Fall Creek of the Chinook salmon with radio transmitter The light line background shows the daily fluctuations in body temperature, while the dashed black line is the daily mean body temperature. Daily means of the in-stream temperature logger sites are shown with corresponding locations on the second y-axis. Dotted line indicates fish position as obtained by mobile tracking. At release, body temperature was similar to the release site and quickly declinedd as the adult moved upstream; thereafter body temperature was intermediate between temperatures recorded at the nearest up- and downstream location (Portland Creek and Johnny Creek, respectively). This fish was one of two that spawned. Spawning Ground Surveys and Spawning Success Spawning success differed between years within Fall Creek and between the Fall Creek and the NFMF study sites. Fall Creek spawning ground surveys were conductedd from 21 July through 2 October in Fifty salmon were recovered in carcass surveys, and energetic conditionn was assessed for 32. The recovery rate was 16.4% of the PIT-tagged fish sampled at the Fall Creek trap. The mean lipid level of carcasses was 0.55% (range %). These estimatess were consistent with what we have seen in other studies on post-spawning fish (e.g., Pinson 2005). Threee of the 32 sampled carcasses were unsuccessful spawners; the 2008 estimated prespawn mortality rate was 9.4% (Table 9). There was no clear association between spawning success and initial energy condition, though the sample of prespawn mortalities was very low (Figure 23). 35

41 Table 9. Final estimated fates of Chinook salmon that were PIT-tagged or double-tagged (PIT- and radio-tagged) in Fall Creek in 2008 and Year Group Spawned Prespawn mortality Unknown/Poached Unrecovered 2008 PIT-tagged % of released 15% 2% - 84% % of recovered 91% 9% 2009 PIT-tagged % of released 3% 14% 1% 83% % of recovered 15% 85% 2009 Double-tagged % of released 8% 76% 8% 8% % of recovered 11% 89% Standardized Fatmeter % Before 25-May May 2-June 9-June 16-June 23-June 30-June Week Figure 23. Weekly mean standardized lipid levels of Fall Creek Chinook salmon as calculated from Fatmeter readings collected at the time of tagging in Circles represent the lipid percentage for successfully spawned fish, crosses represent pre-spawn mortalities. The points were included in the week that they were tagged. Sample sizes shown for reference. 7-July 14-July 36

42 In 2009, Fall Creek spawning ground surveys were conducted from mid-june until the second week in October. Ninety-nine carcasses were recovered including 33 PIT-tagged fish. The PIT-tag recovery rate was 16.5%. A number of carcasses had been scavenged prior to recovery, and it is possible that PIT tags were lost from fish before they were scanned. This hypothesis was supported by the observation that the proportion of released adults with PIT tags (59% of those outplanted) was higher than the proportion of carcasses with PIT-tags (33%). Carcasses were primarily scavenged by turkey vultures during the day (based on observations) and likely by raccoons at night (based on scat). Prespawn mortality was estimated using several parameters including using PIT-tagged fish only, double-tagged fish only, and all tagged and untagged carcasses. Some carcasses or double-tagged fish were excluded from the analyses because spawning success could not be determined or because the fish presumably died from unnatural means (i.e., poaching). Prespawn mortality estimates were similar across groups: estimated to be 84.8% using PITtagged fish, 89.5% using double-tagged fish, and 77.3 % for untagged adults. Although prespawn mortality was high, redd: fish ratios were comparable to other years with lower prespawn mortality. We note that the latest arriving fish were untagged (tagging limits had been reached) and these adults were transported above the dam just before spawning. Qualitative observations suggest this late-arriving group had a higher probability of spawning than earlier arriving adults (including those in the tagged sample) and this probably contributed to the lower PSM estimate for untagged adults. The final distribution of PIT-tagged fish indicated the majority of spawning occurred near the mouth of Hehe Creek (rkm 39.9) and a smaller concentration was upstream from Big Pool campground (rkm 38) (Figure 24). At least two fish were apparently harvested illegally by anglers, based on the recovery of radio tags that were hidden in rock jams and root wads out of the river. On several occasions, we observed activities that suggested harassment or poaching of Fall Creek Chinook salmon in Figure 24. Distribution of PIT-tagged Chinook salmon carcasses that were recovered in Fall Creek spawning ground surveys in 2009, including their spawning status. Two unsuccessful spawners at rkm 42.6 were harvested illegally. 37

43 Prespawn mortality was associated with rising temperatures in late July (Figure 25). We began to recover carcasses as daily mean temperatures exceeded 20 C. The recovery of PITtagged carcasses began to slow as water temperatures decreased in Fall Creek before spawning (Figure 26). Only one fish tagged and released between 1 July and 1 August was known to survive until spawning. Radiotelemetry records indicated that this fish spent the majority of the summer in Fall Creek Reservoir. This behavior may have allowed the fish to avoid lethally high temperatures in lower Fall Creek. We also tested for associations between fate and a suite of factors potentially related to prespawn mortality using logistic regression and multi-model selection techniques (Table 10). In univariate models, there was no clear association between standardized lipid percentages at tagging and spawning success (Figure 27). Among the univariate models tested, only condition score was significantly associated with prespawn mortality at P < Survival rates for condition scores of 1, 2, and 3 were 0%, 18%, and 40%, respectively. In the multi-model logistic regression comparison, the most parsimonious model included condition score, standardized fatmeter %, and the four morphometry metrics (Table 10). In this model, condition (χ 2 = 4.43, P = 0.035), standardized fatmeter (χ 2 = 2.48, P = ), and MeH (χ 2 = 4.24, P = 0.040) were significant; Ba (χ 2 = 3.80, P = 0.051), Da (χ 2 = 3.46, P = 0.063), and HH (χ 2 = 1.23, P = 0.267) were not significant at the P = 0.05 level. Three other models had strong statistical support, with ΔAIC < 2.0. These models were: 1) Condition + Sex, 2) Condition +morphometry, and 3) Condition + Sex + standardized Fatmeter. Notably, models that included only morphometric variables (ΔAIC = 3.67), only timing-related variables that were a surrogate for seasonal changes in environmental conditions (ΔAIC = 8.83), and only salmon size variables (ΔAIC = 9.22) had substantially less statistical support. Overall, the model comparison results indicated that condition score and some combination of fish sex, lipid content, and fish shape (which were all correlated to some degree) were important predictors of prespawn mortality. 38

44 Figure 25. Release, holding period and date of last observation for Chinook salmon released in Fall Creek in 2009 by fate. Each horizontal bar represents one PIT-tagged fish from the tag date to the recovery date, ranked by release date. Fish that survived to the first spawning day are in gray, while prespawn mortalities are in black. Releasee site temperatures are plotted for reference. The fish indicated by the arrow was a radio-tagged fish that spent the summer in the reservoir. Fate was not clearly associated with release date. 39

45 Figure 26. Release, holding period and date of last observation for Chinook salmon released in Fall Creek in 2009 by fate as in Figure 20, ranked by recovery date. The majority of adults that died did so during or just after a period of warming water temperature in mid- to late July. 40

46 Table 10. Selection statistics for logistic regression models that used prespawn mortality in Fall Creek (2009) as the dependent variable and a variety of predictor variables. AIC = Akaike information criteria, ΔAIC = AIC current -AIC best. Models in shaded grey had strong statistical support (ΔAIC < 2), and the model in bold text was most parsimonious. Variable definitions: Condition = overall physical condition; Finclip = presence/absence of adipose clip; Tagdate = release date; Tagtemp = release date water temperature; StFatmeter = standardized Fatmeter; GSE = gross somatic energy; MeH = mideye to hypural; HH = hump height; Da = depth at anus; Ba = breadth at anus; FL = fork length; Weight = weight; and K = Condition Factor (10 5 *W/L 3 ). Model type Variable(s) AIC ΔAIC Univariate Condition Sex Finclip Tagdate Tagtemp StFatmeter GSE MeH HH Da Ba FL Weight K Timing Tagdate Tagtemp Morphometry MeH HH Da Ba Size FL Weight Condition + Condition Sex Condition Finclip Condition StFatmeter Condition GSE Condition Timing Condition Morphometry Condition Size Condition Sex Finclip Condition Sex StFatmeter Condition Sex Timing Condition Sex Morphometry Condition Sex Size Condition Clip StFatmeter Condition Clip Timing Condition Clip Morphometry Condition Clip Size Condition StFatmeter Timing Condition StFatmeter Morphometry Condition StFatmeter Size

47 18 16 Standardized Fat Meter % May 1-June 8-June 15-June 22-June 29-June 6-July 13-July 20-July 27-July 3-August 10-August 17-August Week Figure 27. Weekly mean standardized lipid levels of Fall Creek Chinook salmon as calculated from Fatmeter readings collected at the time of tagging in Circles represent the lipid percentage for successfully spawned fish, crosses represent pre-spawn mortalities. The points were included in the week that they were tagged. Sample sizes shown for reference. 42

48 A total of 209 fish were released into the NFMF Willamette River in 2009 including the immediate outplants (n = 122) and fish held at Willamette hatchery and later outplanted (n = 87). Eleven PIT-tagged carcasses were recovered during 2009 including one fish that was tagged as a juvenile as part of a separate study. The recovery rate was substantially lower than at Fall Creek (4.8%), probably because of the larger size of the NFMF and other river characteristics such as large woody debris traps that made it difficult to find NFMF carcasses. Though sample sizes for double-tagged fish were low (Table 11), comparison of recovery rate to the PIT-tag group reveal much higher recovery rate and suggest a larger double tag sample would improve recovery rates (as planned for 2010). Prespawn mortality in the NFMF was estimated to be 40.0% based on PIT-tagged carcasses, and 27.3% based on radio transmitter recoveries (Table 11). Because sample sizes were low, prespawn mortality was also estimated using results from all carcasses recovered (41.3%). All three estimates indicate that prespawn mortality in the NFMF was considerably lower than in Fall Creek in In the NFMF, spawning activity was concentrated in three main river reaches (Figure 27). These included a reach near the release site, in the large log jam section at rkm 35.3, and near Kiahanie campground. A total of 200 redds were counted in 2009, and all but seven were upstream from the release site. Unfortunately, the low recovery rate precluded statistical comparison of prespawn mortality rates between the immediate outplant and the group held at Willamette hatchery prior to outplanting on 24 August. Qualitative comparison of the few recoveries hint at a potential benefit of holding because the prespawn mortality rate among PIT-tagged adults in the hatchery held group was 5.8% (6 of 104) compared to ~30% mortality in adults outplanted immediately. Table 11. Final fates of PIT tagged and radio tagged subset in the NFMF Willamette. Double-tagged fish were only included in the PIT tagged numbers if the whole carcass was recovered, and not just the radio tag. *This value only includes outplanted fish, and excludes adults unrecovered from the hatchery holding pond (n = 9) or a small group of fish tagged and outplanted by ODFW to the Middle Fork Willamette (n = 13). Year Group Spawned Prespawn mortality Unknown/Poached Unrecovered 2009 PIT-tagged * % of released 3% 2% 0% 95% % of recovered 60% 40% 2009 Double-tagged % of released 67% 25% 0% 8% % of recovered 72% 27% 43

49 Figure 27. Distribution of PIT-tagged (including radio tagged) Chinook salmon carcasses that were recovered in the NFMF Willamette River spawning ground surveys in 2009, including their spawning status. Disease Screening Thirty-four Chinook salmon carcasses from the study areas that had recently died (fresh mortalities, many still with red gills) prior to spawning were obtained, transported to the laboratory at Oregon State University, and necropsied within a few hours of collection. These individuals spanned the time Chinook salmon were in the study systems, from July into September, In addition, two fresh mortalities that died immediately after spawning in the wild and 14 other individuals that survived to be spawned at Willamette hatchery were sampled. External Findings External trauma was observed in many of the fish. The location and severity of the trauma varied with each fish and ranged from dermal erosion (Figure 28) to tissue maceration of the head (Figure 29). Other forms of trauma observed included dermal ulceration (Figure 30) hematoma formation (Figure 31) and post-traumatic scoliosis (Figure 31). Because of the freshness of the carcasses (e.g., pink gills and red blood apparent in wounds), we suspect that most, if not all, of these injuries occurred ante mortem and not after death. 44

50 Figure 28, Head tissue maceration, dermal erosion, and ulceration in a prespawn mortality suspected to have occurred ante-mortem. Figure 29. Caudal peduncle, skin: Multifocal dermal ulceration and hematoma formation, prespawn mortality. 45

51 Figure 30. Vertebrae: Post traumatic scoliosis, prespawn mortality. 46

52 Macroscopic Internal Changes Many fish were observed to have swelling or hemorrhaging of the kidney and/or the spleen. A few fish had lesions of the GI tract; the specific portion of the GI tract was not noted. Histological Findings All of the fish that died prespawning had histological signs of infection, injury and internal signs of disease (Table 12). Indeed, all but one of these fish exhibited moderate to severe histological changes. The most common infections associated with significant pathological changes were either due to digenean metacercariae (Fig 31) or myxozoan parasites (Fig 32). Metacercariae of Nanophyetus salminicola were found in the heart, kidney, and a mix of Echinochasmus miivi, Apophallus sp., and Nanophyetus were observed in the gills. The former two were associated with cartilage metaplasia in the gill filaments, and in heavy infections a significant amount of the gills was affected. Nanophyetus was associated with mild to moderate inflammatory changes regardless of the organ that it infected. Regarding the myxozoans (Figure 32), severe, systemic Ceratomyxa shasta infections were observed in various organs, including the intestine, liver and heart, and were consistently associated with severe necrotic changes. Parvicapsula minibicornis (Myxozoa) was observed in the renal glomeruli and tubules, and in severe infections essentially all glomeruli were affected. Other, less common, histopathological changes observed included inflammatory changes suggestive of bacterial kidney disease, severe fungal infection of the gills, and idiopathic hepatitis. In addition, all fish exhibited diffuse, chronic inflammation of the lamina propria of the intestine and pyloric caeca (Fig 33). In one fish, this lesion appeared neoplasitic, with the appearance of a diffuse lymphosarcoma. 47

53 Table 12. Lesions and pathogens in prespawn and post-spawn mortalities from various locations in the upper Willamette River basin. Date is collection date. 0 = absent, 1 = light, 2 = moderate, 3 = prominent or severe infections. Prespawn Morts Fall Creek Fish Gill Nanophyetus Nanophyetus C. shasta Parvicapsula Macroscopic & Other Sex Date Number Metacercariae Heart Kidney Systemic Kidney lesions 41 M 17/ F 2%, enlarged spleen 171** F 23/ F 28/ F!% 166 M 29/ HL, F 15% 37** F 31/ F 8/ HL, F 20% 98** M 13/ F 18/ F 20/ Severe gill mycosis 251 M 26/ F1% 131? 28/ F 1%, grey kidney 103 F 9/ F 1% 160 F 9/ M 11/ External abrasions 118? 11/ Prespawn Morts Middle Fork Willamette Fish Gill Nanophyetus Nanophyetus C. shasta Parvicapsula Macroscopic & Other Sex Date Number Metacercariae Heart Kidney Systemic Kidney lesions 225** M 24/ F 24/ Multiple liver and GI whitish lesions 214 M 24/ Enlarged liver with white spots 256** F 23/ M 11/

54 Prespawn Morts North Fork of the Middle fork of the Willamette Fish Gill Nanophyetus Nanophyetus Sex Date Number Metacercariae Heart Kidney C. shasta Systemic Parvicapsula Kidney Macroscopic & Other lesions 60 F 5/ F 5/ F 5/ ** M 5/ F 5/ M 5/ Multiple granulomas in gut 11 F 7/ F 7/ * M 7/ ** M 7/ M 13/ M 11/ F 5% 215 F 12/ M 12/ Severe hepatitis, unknown cause 49

55 Post Spawn Morts North Fork Middle Fork Willamette River Fish Gill Nanophyetus Nanophyetus C. shasta Parvicapsula Macroscopic & Other Sex Date Number Metacercariae Heart Kidney Systemic Kidney lesions 100 F 14/ Slightly swollen kidney 113** M 14/ Post Spawn Morts Willamette Hatchery Fish Gill Nanophyetus Nanophyetus C. shasta Parvicapsula Macroscopic & Other Sex Date Number Metacercariae Heart Kidney Systemic Kidney lesions A M 8/ One Ich parasite observed B M 8/ C F 8/ D F 8/ Salminicola copepod on gills E F 8/ F M 8/ G F 8/ H M 8/ I M 22/ J F 22/ ` 0 Trichodinids and copepods on gills. Granulomas suggestive of BKD in kidney K M 22/ L F 22/ Granulomas suggestive of BKD in kidney M M 22/ Numerous trichodina on gills N F 22/

56 Figure 31. Metacerariae infections. A. Kidney with heavy Nanophyetus infection (white area). B. Heart with heavy Nanophyetus infection. Gill with severe Echinochasmus/Apophallus sp. (A) infection in lamellae. N = Nanophyetus in gill arch. D) Insert shows cartilage metaplasia associated with encysted Apophallus 51

57 Figuree 32. Myxozzoan infectioons. A) Masssive liver damage in fissh with cerattomyxosis. B). B Histologiical section of o liver of saame fish shoowing severee necrosis. Bar B = 200 µm m. C) High magnification of liveer showing sppores of C. shasta s (arrow ws) within neecrotic lesioon. D). Parvicapsulosis, with h damage to glomeruli. Arrows A = developmenta d al stages of Parvicapsula P a minibicornis in blood d vessels. 52

58 Figure 33. Inflammatory changes in (a) the pyloric caeca and lower intestine seen in essentially all prespawn mortalities and survivors. Image (b) respresents extreme cases suggestive of lymphosarcoma. 53