Effects of Physical Habitat and Ontogeny on Lentic Habitat ...

Transactions of the American Fisheries Society 135:1191–1204, 2006 American Fisheries Society 2006 DOI: 10.1577/T05-281.1

[Article]

Effects of Physical Habitat and Ontogeny on Lentic Habitat Preferences of Juvenile Chinook Salmon CHRISTOPHER J. SERGEANT1

AND

DAVID A. BEAUCHAMP

U.S. Geological Survey, Washington Cooperative Fish and Wildlife Research Unit,2 School of Aquatic and Fishery Sciences, Box 355020, University of Washington, Seattle, Washington, 98195-5020, USA Abstract.—We experimentally tested the habitat preferences of juvenile Chinook salmon Oncorhynchus tshawytscha to evaluate whether habitat availability was limited for a lake-rearing population in Lake Washington (Seattle, Washington). A segment of this population resides in shallow (,1 m deep), nearshore areas of the lake during January–May and migrates to saltwater at age 0. During the shallow, nearshore phase, fish are exposed to varying degrees of bottom slope, substrate, and cover (e.g., overhead docks and woody debris) formed by the natural and modified shorelines of this highly urbanized system. The effects of these three habitat variables on patch use or electivity were tested in combination with ontogeny and the presence or absence of piscivores. Fry and presmolts avoided steeper bottom slopes, but presmolt responses were strongest. Responses to substrate and cover options were weak, although fry exhibited some coherent preference for finer substrates. The habitat preferences displayed by both life stages corroborated the observations from nearshore field surveys in Lake Washington. No direct effects on habitat preference from diel period or piscivore presence were evident. These results, combined with field observations, suggest that juvenile Chinook salmon may risk exposure to predation in order to utilize preferred habitat and to forage at a high rate. Therefore, nearshore habitat restoration projects aimed at increasing preferred juvenile Chinook salmon habitat should consider this risk-prone behavior. Future experiments in larger experimental arenas could clarify the importance of heterogeneous nearshore habitat and further examine predation effects on the productivity of lake-rearing juvenile Chinook salmon.

Habitat and predation risk strongly affect the behavior and distribution of species in aquatic environments (Werner et al. 1977). This is apparent in nearshore freshwater habitats, where fish density is typically high among complex littoral habitat features (Hosn and Downing 1994). As structural complexity increases in nearshore habitats, predation efficiency rates generally diminish (Crowder and Cooper 1982). However, selection for habitat edges often involves a tradeoff between predator avoidance behavior and prey availability (Tabor and Wurtsbaugh 1991; Koehler 2002). Salmonid juveniles often inhabit shallow, nearshore areas in lotic (Lister and Genoe 1970; Sempeski and Gaudin 1995) and lentic (Tabor and Wurtsbaugh 1991; Graynoth 1999; Tabor and Piaskowski 2001) habitats. Anthropogenic habitat degradation in these areas has been listed as an important cause of salmonid * Corresponding author: [email protected] 1 Present address: Shared Strategy for Puget Sound, 1411 4th Avenue, Suite 1015, Seattle, Washington 98101, USA. 2 The Unit is jointly sponsored by the U.S. Geological Survey, University of Washington, Washington Department of Ecology, Washington Department of Fish and Wildlife, Washington Department of Natural Resources, and Wildlife Management Institute. Received November 8, 2005; accepted March 9, 2006 Published online August 31, 2006

population declines (Nehlsen et al. 1991). In estuarine habitats, higher juvenile survival in nearshore areas may significantly enhance survival to adulthood for Chinook salmon Oncorhynchus tshawytscha. However, survival mechanisms are poorly understood in nearshore areas, and it is unclear whether mortality can actually be decreased through habitat restoration (Greene and Beechie 2004). Before recommending specific restorations in nearshore freshwater habitats, it is important to understand the mechanisms that affect salmonid survival in these areas and how various types of nearshore habitat control these mechanisms. Wild, ocean-type Chinook salmon commonly rear in Lake Washington (Seattle, Washington) for 2–5 months before emigrating to sea (Tabor and Piaskowski 2001). Because wild Chinook salmon in the Lake Washington basin are part of the Puget Sound Chinook salmon evolutionarily significant unit, which is listed as threatened under the Endangered Species Act, knowledge of stage-specific habitat requirements during the extended lake-rearing phase is essential for informed management. In the Lake Washington basin, most naturally reproducing Chinook salmon spawn in the Cedar River, the largest tributary to the lake, and their progeny can be found in the lake and river from January to July (Seiler et al. 2003). Juvenile Chinook salmon of wild and hatchery origin also enter Lake

1191

1192

SERGEANT AND BEAUCHAMP

Washington from the Sammamish River and other smaller tributaries. The wild Chinook salmon from the Cedar River exhibit two different life history strategies: lake rearing and stream rearing. Soon after emergence, the lake-rearing fry migrate to the lake (January–April); peak immigration to the lake occurs from late February to early March (Seiler et al. 2003). In contrast, streamrearing fry feed and grow in streams for several months before migrating to the lake in May and June. Lakerearing fry grow more rapidly than stream-rearing fish and feed mainly on chironomid pupae (Koehler 2002; Koehler et al., in press). Because the temperature regimes of Lake Washington nearshore areas and the Cedar River follow similar trends, high growth rates are probably attributable to the greater food supply available in the lake. This food supply may also explain why lake-rearing fry attain sizes that are comparable to those of juvenile Chinook salmon in Puget Sound (Duffy et al. 2005). From field observations, Tabor and Piaskowski (2001) described habitat usage by Chinook salmon rearing in Lake Washington, which to date is the most extensive work on juvenile Chinook salmon in lentic systems. During January through mid-May, lake-rearing fry primarily occupied shallow littoral (,1 m) regions with a gentle slope and sand–silt substrate. Overhead cover (e.g., docks) was generally avoided during all diel periods, except for occasional daytime use during February and March. Depth in the water column varied with diel period. During the day, most Chinook salmon aggregated near the surface and displayed feeding behavior, whereas aggregations dissipated during the night and fish stayed inactive near the lake bottom (Tabor and Piaskowski 2001). Most lake-rearing fry leave nearshore areas in May and June and migrate through the lake, becoming smolts that enter Puget Sound during June and July. Once in Puget Sound, higher numbers of nearshore juvenile Chinook salmon are found in sand–silt substrates than in cobble (Footen 2000). Human development of Lake Washington’s shoreline has modified the types of habitat available to juvenile Chinook salmon. As of 2000, there were 2,737 docks on Lake Washington (Toft 2001). The shade and composition of docks may alter the natural movements of Chinook salmon by causing the fish to swim around these structures into deeper water instead of through shallow, nearshore areas. Dock areas were often associated with armored shorelines (riprap or bulkheads), which made up 71% of the total lake shoreline. Armored shorelines limit recruitment of overhanging vegetation and woody debris to the littoral zone as cover for juvenile salmonids. These revetments can also have localized effects on bottom slope and

substrate type by altering wave energy and silt deposition. Before assessing availability of beneficial habitat, life-stage-specific habitat preferences should be identified and integrated into any evaluation of potential habitat conservation or restoration actions. The field observations of Tabor and Piaskowski (2001) on juvenile Chinook salmon in Lake Washington described important habitat associations and set the stage for controlled experiments. Because naturally produced fish were never simultaneously presented with a full array of littoral habitats before entering the lake and did not have perfect knowledge of the types of habitat available at large spatial scales, it is difficult to evaluate whether their observed habitat use is actually preferred or merely a function of habitat availability. Therefore, to aid interpretation of habitat use observed in the field, we conducted habitat preference experiments under controlled conditions. The objectives of this study were to experimentally examine lake-rearing Chinook salmon habitat selection patterns (1) across a range of bottom slopes, (2) for a factorial combination of substrate and cover types, (3) in the presence and absence of predators, (4) among diel periods, and (5) ontogenetically (i.e., between fry and presmolt stages). Based on the field observations of Tabor and Piaskowski (2001), we expected that Chinook salmon fry would select patches with gradual bottom slopes and fine substrate and would generally avoid all cover types. Any use of cover would be greatest during the day. Habitat preference patterns would weaken in presmolts because larger juveniles move offshore and deeper into the lake. We expected juvenile Chinook salmon under predation pressure to use deeper water and to stay near the bottom of the arena (Gregory 1993a). Bottom usage would also be common during the night and would lessen during the day. Study Area Lake Washington is a large, glacially formed, nowurbanized lake. It is 32.2 km long, averages 2.5 km in width, and has a maximum depth of 66 m. Two major tributaries enter Lake Washington: the Cedar River in the south and the Sammamish River in the north. The lake flows into Puget Sound via the Lake Washington Ship Canal and the Hiram M. Chittenden Locks, where lake surface elevation is controlled. The completion of the locks in 1916 lowered lake elevation by 2.4 m and altered the natural drainage of the Cedar River into the Duwamish River and Puget Sound by channeling it into the lake (Edmondson 1991). This modified the emigration corridor of Chinook salmon smolts and

CHINOOK SALMON LENTIC HABITAT PREFERENCES

1193

FIGURE 1.—Overhead schematic views of experiments in artificial habitats where Chinook salmon fry and presmolts were exposed to four bottom slopes (5, 10, 15, and 20%) to enable examination of slope preferences. Panel (A) shows the configuration in 2004, whereby the arena was a raceway configured with one slope unit per bottom slope (habitat patch); the open areas between units were covered in light brown burlap. Panel (B) shows the configuration in 2003, when pilot slope experiments employed a circular concrete pond configured with eight habitat patches, or two patches per bottom slope; the areas between each patch were covered in burlap, and each patch had clear side walls to limit habitat choice to the neutral area or an individual patch.

forced them to swim through the lake to reach saltwater. Methods Three independent sets of experiments were used to examine the nearshore lentic habitat preferences of juvenile Chinook salmon exposed to a range of bottom slopes, substrate types, and substrate–cover combinations. The habitat options were selected from a representative range of habitat types found in the littoral zone of southern Lake Washington (Toft 2001). In areas where Tabor and Piaskowski (2001) observed lake-rearing wild Chinook salmon, slopes of 0–20% made up 77% of the littoral zone within 15 m from shore and 74% of the area within 30 m of shore. These measurements describe the general bathymetry of south Lake Washington’s nearshore areas but do not describe smaller-scale slope areas formed at the terrestrial– aquatic margins of modified shorelines (e.g., riprap). Lake surface elevations averaged 6.10 m above mean sea level in January 2004, the month when bottomslope measurements were taken. From March to May 2004, lake surface elevations averaged 6.25–6.71 m (U.S. Army Corps of Engineers, http://www.nwd-wc. usace.army.mil/nws/hh/basins/).

Pilot slope experiments.—Pilot experiments were undertaken in 2003 to examine the feasibility of testing bottom-slope preferences of juvenile Chinook salmon in a large arena. The experimental arena was built within a large, circular concrete hatchery pond 12.2 m in diameter at Seward Park in Seattle, Washington (Figure 1). Eight slope patches were arranged around a circular middle section called the ‘‘neutral area,’’ where water depth was maintained at 0.5 m. Clear walls on each slope patch aided the observation of experimental fishes by confining them to one patch at a time and minimizing edge effect. Light brown burlap fabric was stretched between the clear walls to further mimic a smooth transition between units and minimize edge effect. Cinder blocks were propped under each slope patch to provide four unique slope choices: 5, 10, 15, and 20%. Two replicates of each slope were randomized within the experimental arena. The slope preference experiments were conducted in the absence (21 May 2003) and presence (29 May 2003) of predators. We used 24 Chinook salmon presmolts for both the predator-present and predator-absent experiments (mean fork length [FL] ¼ 73 mm, SD ¼ 7.4). Experimental Chinook salmon were transported from

1194


the Washington Department of Fish and Wildlife (WDFW) Issaquah Creek Hatchery on 29 April 2003 and were held at Seward Park. After being randomly selected from the holding tanks, experimental fish were released into a plastic mesh cage in the neutral area. After acclimating for 30 min in the neutral-area cage, fish were released and allowed to roam the experimental arena. Habitat observations were not recorded until 24 h after release from the neutral-area cage. During the second experiment, one cutthroat trout O. clarkii (FL ¼ 160 mm) and one smallmouth bass Micropterus dolomieu (FL ¼ 168 mm) were added outside of the neutral-area cage at the same time as the experimental Chinook salmon were added. All other pilot experiment protocols followed the descriptions given for the main suite of habitat experiments below. Main slope experiments.—During 2004, a second set of bottom-slope experiments were conducted for fry and presmolts in a large outdoor raceway (30 m long 3 3 m wide 3 1.8 m tall) at the University of Washington (UW) hatchery. The length of the raceway was necessary to accommodate a realistic range of depths (0.5 m; Tabor and Piaskowski 2001) and bottom slopes (5–20% slope) occupied by juvenile Chinook salmon in Lake Washington. For example, a habitat unit with a 5% slope required a horizontal distance of 10 m to accommodate depths up to 0.5 m. Four habitat units simultaneously presented unique slope choices of 5, 10, 15, and 20%. Two sets of two habitat units faced each other and were separated by a neutral area (3.0 m long 3 2.7 m wide 3 0.5 m deep) in the middle. The neutral area contained the deepest water and was the release point for experimental fish (Figure 1). The slopes of the habitat units were reconfigured twice during experimental trials to avoid tank-effect biases. Sand was used because it was the most heavily utilized substrate for juvenile Chinook salmon in Lake Washington (Tabor and Piaskowski 2001). We used 80 naı¨ve (previously unexposed to the experimental arena) Chinook salmon for each slope trial. Temperature ranged from 88C to 188C during the experiments (mean ¼ 12.98C, SD ¼ 2.88C). Substrate experiments.—The four most common substrates found in Lake Washington (Toft 2001) were used for substrate preference experiments: sand, gravel, sand with embedded cobble, and cobble. In our experiments, cobble was 64–175 mm in diameter and gravel was 4–20 mm in diameter. These experiments were conducted in outdoor experimental troughs under a canopy at the U.S. Geological Survey (USGS) Western Fisheries Research Center in Seattle, Washington. Water for the troughs was supplied from Lake Washington at ambient temperatures and filtered for particulate matter. Six fiberglass experimental troughs

(12.2 m long 3 1.5 m wide 3 1.5 m deep) were used. Each experimental block was composed of an area (1.5 m long 3 1.0 m wide 3 0.4 m deep) divided into four rectangular patches (0.75 m long 3 0.5 m wide). Each patch contained one of the four substrate types, which were randomly assigned. Each of three troughs contained two experimental blocks, allowing for six concurrent randomized blocks. Twenty naı¨ve Chinook salmon juveniles were used in each block, resulting in a density of 13 fish/m2. Temperature ranged from 78C to 168C during the experiments (mean ¼ 12.08C, SD ¼ 2.88C). Substrate–cover experiments.—The four substrates (sand, gravel, sand with embedded cobble, and cobble) were tested in a factorial combination with three cover types (no cover, overhead cover emulating a dock, and small woody debris) in each arena. The same troughs used in the substrate experiments were reconfigured to test habitat preferences for all factorial combinations of substrate and cover. However, to accommodate the additional three cover types, the experimental block size was expanded to 2.8 m long by 1.5 m wide. Two experimental blocks containing all substrate–cover combinations were constructed in each of two troughs, allowing for four concurrent experimental blocks. The size of substrate patches in these arenas was equal to the patch sizes in the substrate-only experiments. The 12 substrate–cover combinations were randomly assigned within each experimental block. Sixty naı¨ve Chinook salmon juveniles were used in each block, resulting in a density of 14 fish/m2. Temperature ranged from 148C to 168C during the experiments (mean ¼ 15.08C, SD ¼ 0.48C). Design for all experiments.—For all experiments, trials were repeated for both fry and presmolt stages of Chinook salmon; this allowed us to examine the potential ontogenetic shifts in habitat preference (Table 1). The fish used in these experiments came from the WDFW Issaquah Creek Hatchery. Fish from the hatchery are released into Issaquah Creek, where they migrate through the Sammamish River and Lake Washington to Puget Sound. Fry were acquired from the hatchery on 23 February 2004, and presmolts were obtained on 10 May 2004. Fish were held at the UW and USGS facilities and were fed daily rations equal to 1–2% of body weight to maintain a consistent size during the trials (Table 1). Cutthroat trout and prickly sculpin Cottus asper are both predators of juvenile Chinook salmon in Lake Washington (Tabor et al. 1998; Nowak et al. 2004) and were used as piscivores during our experiments (Table 1). We collected 12 cutthroat trout on two sampling dates during winter 2004 via beach seining in the littoral regions of Lake Washington. Twenty-three

1195


TABLE 1.—Fork length (FL), dates, and number of replicates used for juvenile Chinook salmon habitat preference experiments performed in 2004 in artificial habitats (N is the number of fish used in each trial). Piscivore lengths and experiment dates are also presented. Total replicates Life stage and experiment type (N) Fry Slope (80) Substrate (20) Substrate-cover (60) Presmolt Slope (80) Substrate (20) Substrate-cover (60) a b

FL (SD)

No piscivore

Piscivore

Predator length (SD)

Dates

45.4 (4.3) 50.0 (4.3) 50.0 (4.3)

4 12 8

4 12 8

184.7 (6.7)a 113.6 (13.4)b 113.6 (13.4)b

24 Mar–10 May 7 Apr–23 Apr 5 May–20 May

78.9 (4.0) 79.4 (4.8) 79.4 (4.8)

2 12 4

2 2 2

184.7 (6.7)a 113.6 (13.4)b 113.6 (13.4)b

23 May–30 May 20 May–9 Jun 28 May–9 Jun

Cutthroat trouth FL. Prickly sculpin total length.

prickly sculpin were collected on 1 April 2004 by dipnetting near the mouth of the Cedar River in south Lake Washington. Piscivores were fed a combination of live nightcrawlers and Chinook salmon fry from the Issaquah Creek Hatchery twice per week. Before the start of each experiment, naı¨ve Chinook salmon were chosen at random from holding tanks. Fish in slope experiments were transported to a wiremesh cage in the neutral area; fish in substrate and substrate–cover experiments were transported to buckets floating in the experimental arenas. After acclimating for 30 min, fish were released and allowed to roam the experimental arena. Upon release into the arena, Chinook salmon juveniles generally explored the entire arena in loose aggregations before dispersing as individuals and settling into different habitat patches. Agonistic behavior among individuals was rarely observed among all arena types. During pilot experiments performed in 2003 and 2004, we determined that 12 h was sufficient for fish to return to nonexploratory behavior. Once fish were acclimated to the arena for at least 12 h, observations were recorded over a 24-h period divided into five distinct diel periods: (1) dawn, from 1 h before to 1 h after sunrise; (2) full-light AM, from 1 h after sunrise until 1200 hours; (3) full-light PM, from 1200 hours to 1 h before sunset; (4) dusk, from 1 h before to 1 h after sunset; and (5) full darkness, from 1 h after sunset to 1 h before sunrise. No artificial light sources were used to manipulate diel periods. Pilot experiments at Seward Park and substrate–cover experiments at USGS were situated in areas where light levels were dark enough at 1 h before sunset to be considered crepuscular. The UW hatchery facilities were not as protected as the other two facilities, so fish in slope experiments may have experienced slightly higher light levels during the dusk period.

For the slope experiments, we recorded the number of fish per habitat unit, their horizontal location and behavior within each unit (cruiser [individuals that were moving within a habitat patch], center, nearshore, or offshore), and depth in the water column (top, middle, or bottom third of the water column) once during each of the five diel periods. In the substrate and substrate–cover experiments, only the number of fish per habitat patch was recorded. Each habitat patch was observed for approximately 60 s. Locations of individuals were recorded based on where each fish spent the majority of its time during the 60-s observation period. Daytime counts of fish in each experimental unit were aided by use of polarized sunglasses. For the substrate and substrate–cover experiments, mirrors were suspended above each arena to aid observations during daylight. Side observation windows were covered with black cardboard material to minimize outside disturbances. For nighttime observations, we used short illumination pulses (about 1–2 s) with a white spotlight into the water column. Fish were typically motionless when illuminated but would move if exposed to prolonged periods of white light; therefore, care was taken by the observer to use short pulses and to memorize the locations of observed fish. Piscivorous fish were added to the experimental blocks after the first 24-h observation period. The acclimation and observation protocol described above was repeated for another 24-h observation period. Piscivore location and activity level within each block were noted during the observation periods. One to three cutthroat trout per trial were used for slope preference experiments in the large raceway at the UW hatchery. These predators were too large for the other experiments, so instead we used one to two prickly sculpin per trial for the substrate experiment and six prickly sculpin per trial for the substrate–cover experiment.

1196


of fish using each habitat patch were arcsine transformed for the analyses (Zar 1996). Post hoc comparisons between multiple treatments used the Tukey honestly significant difference (HSD) test (equal error variance) or Dunnett’s T3 test (unequal error variance; Zar 1996). The significance level was 0.05 for all main effects and 0.10 for all interaction effects. Data analyzed by ANOVA were grouped over diel period and piscivore presence or absence (hereafter, predator treatment). Although the diel observations were not truly independent observations, the nonsignificance (P . 0.50 in all cases) of diel period and predator treatment in the following results lends credence to this approach. Chesson’s (1978) alpha electivity index was used to classify habitat preference for all slope experiments. This was important because steeper slopes offered less surface area than shallower slopes over the 0.0–0.5-m depth range of the arena. Chesson’s alpha values (range ¼ 0.0–1.0) were used to detect whether each of N habitat types were randomly chosen (Chesson’s a ¼ 1/ N), preferred (a . 1/N), or avoided (a , 1/N). Chesson’s alpha is computed as ri =pi a¼ X ; ri =pi

ð1Þ

i

FIGURE 2.—Slope preference (Chesson’s alpha) of Chinook salmon presmolts in three diel periods (day, crepuscular, night) and in the presence or absence of predators (cutthroat trout or smallmouth bass) during 2003 pilot experiments in artificial habitats. Only one experiment was conducted for each piscivore treatment. The dashed line at the alpha value of 0.2 represents random preference; points above the line represent preference, and points below represent avoidance.

Because each experimental arena contained vertical surfaces (fiberglass walls on the sides, PVC-mesh dividers on the ends), the potential for edge attraction or repulsion effects was a concern. These concerns were lessened by the pilot slope experiments conducted in 2003, which indicated that Chinook salmon juveniles generally aggregated over the interior of the habitat patches rather than along the edges, and test fish repeated these patterns in 2004. Data analysis.—The effects of slope, substrate, cover, predators, and diel period on the differences in proportional use and electivity (Chesson 1978) of fish utilizing each experimental habitat patch were analyzed using generalized linear models (analysis of variance [ANOVA]; Neter et al. 1996) in the Statistical Package for the Social Sciences version 11.5.0. The proportions

where ri is the relative use of habitat type i and pi is the relative availability of habitat type i. Results Slope Preference Experiments Pilot slope experiments in 2003.—Daylight electivity values for all habitat patches were nearly random (Figure 2). Chinook salmon presmolt electivity was higher for the deep neutral area than for all other slope choices during crepuscular and night periods. During these two diel periods, neutral-area preference was higher when predatory cutthroat trout and smallmouth bass were absent. Main slope experiments in 2004.—For Chinook salmon fry, habitat electivity varied significantly among slopes (ANOVA: P , 0.001; Figure 3). Cutthroat trout treatment (P . 0.50) and diel period (P . 0.50) did not affect overall slope use, but a significant interaction existed between slope and predator treatment (P ¼ 0.033). In the absence of cutthroat trout, electivity by fry was higher for the deep neutral area than for all other slope choices (Tukey HSD: P , 0.010 for all), and 10% (P ¼ 0.020) and 15% (P ¼ 0.037) slopes were preferred over the 20% slope (Figure 3). When cutthroat trout were present, the deep neutral area was still preferred over all other slopes (P , 0.010 for all), and 5% (P ¼ 0.050) and


FIGURE 3.—Slope preference (Chesson’s alpha 6 2 SEs) of Chinook salmon fry and presmolts in three diel periods (day, crepuscular, night) and in the presence or absence of predators (cutthroat trout) during 2004 experiments in artificial habitats (NA ¼ neutral area). The dashed line at the alpha value of 0.2 represents random preference; points above the line represent preference, and points below represent avoidance.

15% (P ¼ 0.001) slopes were preferred over the 20% slope (Figure 3). For presmolts, habitat electivity varied significantly among slopes (ANOVA: P , 0.001; Figure 3). Cutthroat trout treatment (P . 0.50) and diel period (P . 0.50) did not affect overall slope use patterns by presmolts, and interaction terms were not significant. Presmolts preferred the deep neutral area over all other slope choices (Tukey HSD: P , 0.001 for all), and 5% and 10% slopes were preferred over 15% and 20% slopes (P , 0.003 for all; Figure 3). Slope patch location and water column usage patterns.—Patterns in the horizontal and vertical locations of fry within slope patches were evident. Fry location within each slope patch (nearshore, center, offshore, or cruiser) varied significantly (ANOVA: P , 0.001; Figure 4). Cutthroat trout treatment (P . 0.50) and diel period (P . 0.50) did not affect fry horizontal location, but a significant interaction existed between diel period and horizontal location (P ¼

1197

FIGURE 4.—Proportional use þ 2 SEs of horizontal locations (N ¼ nearshore; C ¼ center; O ¼ offshore; cruise ¼ individuals moving within a habitat patch) by Chinook salmon fry and presmolts within slope patches in three diel periods (day, crepuscular, night) and in the presence or absence of predators (cutthroat trout) during 2004 experiments in artificial habitats.

0.014). During all diel periods, significantly more fry were either actively cruising within the habitat unit or occupying the deeper offshore regions than were occupying the shallower nearshore or central portions of the slope unit (Tukey HSD: P , 0.015 for all; Figure 4). The vertical position of fry (top, middle, and bottom of the water column) varied significantly within all slope patches (ANOVA: P , 0.001; Figure 5). The cutthroat trout treatment (P . 0.50) and diel period (P . 0.50) main effects on vertical position of fry were not significant, but the interaction of diel period and cutthroat trout treatment was significant (P ¼ 0.001). Most fry occupied the middle of the water column during daylight but were more strongly oriented toward the bottom during crepuscular and night periods, especially when cutthroat trout were present (Table 2; Figure 5). For presmolts, the horizontal and vertical location patterns shifted somewhat from those observed for fry. Presmolt proportions varied significantly among horizontal locations within the slope patches (ANOVA: P , 0.001; Figure 4). Higher proportions of presmolts

1198


(P . 0.50) and diel period (P . 0.50) on the vertical position of presmolts were not significant, the interaction of diel period and vertical position was significant (P , 0.001). During crepuscular periods, more presmolts used the middle than the upper portion of the water column (Dunnett’s T3: P , 0.001). Presmolts concentrated in the middle of the water column during daylight but dispersed into the middle and bottom layers during crepuscular periods; they became more bottom oriented at night (Tukey HSD: P , 0.001 for all). During slope experiments, cutthroat trout generally occupied the offshore portions of the 5% slope unit or remained in the deep neutral area. Several attacks on fry and presmolts were observed during most trials, but no actual predation was observed. Substrate Preference Experiments

FIGURE 5.—Proportional use þ 2 SEs of vertical positions (bottom, middle, and top of water column) by Chinook salmon fry and presmolts in three diel periods (day, crepuscular, night) and in the presence or absence of predators (cutthroat trout) during 2004 slope experiments in artificial habitats.

cruised or occupied the offshore than center or nearshore portions of the slope patches during day and crepuscular periods (Tukey HSD: P , 0.001 for all). These horizontal patterns were weaker but persisted at night. Cutthroat trout treatment (P . 0.50) and diel period (P . 0.50) did not affect presmolt location. Presmolt vertical position in the water column varied significantly (ANOVA: P , 0.001; Figure 5). Although the main effects of cutthroat trout treatment

Both fry and presmolts responded weakly to the different substrate choices (Figure 6). Across all diel periods and predator treatments, the average proportional use for any of the four substrates varied from 13% to 38%, whereas a 25% usage rate was expected under uniform usage of substrates. Proportional patch use by fry varied significantly among substrates (ANOVA: P , 0.001; Figure 6). Prickly sculpin treatment (P ¼ 0.87) and diel period (P ¼ 0.99) did not affect overall substrate use patterns by fry, but substrate and diel period interacted significantly (P ¼ 0.045). During the day, sand (Tukey HSD: P ¼ 0.015) and sand–cobble (P ¼ 0.026) were used more than cobble. During crepuscular periods, sand was used significantly more than cobble (P ¼ 0.007). At night, sand was used significantly more than gravel (P ¼ 0.028), sand–cobble (P ¼ 0.003), and cobble (P , 0.001). Gravel was also used significantly more than cobble (P ¼ 0.007) at night. Proportional patch use by presmolts, which was generally more variable than use by fry throughout all treatments, exhibited no significant pattern among substrates (ANOVA: P ¼ 0.068; Figure 6), predator

TABLE 2.—Multiple comparison results for vertical position patterns of Chinook salmon fry in the water column during slope experiments in artificial habitats. Each column represents an experiment, and comparisons were made within the three vertical positions. Experiments were conducted by diel period and cutthroat trout (CTT) presence or absence. Significant P-values are in parentheses, and Xs denote nonsignificant comparisons. Crepuscular

Day

Night

Vertical position

CTT

No CTT

CTT

No CTT

CTT

No CTT

Top Middle

X Top (0.008)

X X

X Top (0.010)

X X

X X

Bottom

Middle (0.038) Top (,0.001)

X

X

X Top (,0.001) Bottom (,0.001) X

Middle (0.014) Top (0.014)

X


FIGURE 6.—Proportional use 6 2 SEs of substrates (S ¼ sand; G ¼ gravel; SC ¼ sand with embedded cobble; C ¼ cobble) by Chinook salmon fry and presmolts in three diel periods (day, crepuscular, night) and in the presence or absence of predators (prickly sculpin) during 2004 experiments in artificial habitats.

treatments (P ¼ 0.85), or diel periods (P ¼ 0.97). The exception was at night, when cobble was used significantly less by presmolts in the presence of prickly sculpin than in the absence of these predators (two-sample t-test: P ¼ 0.048). The average proportional use appeared to decline with increasing substrate complexity in predator treatments during day and night periods, but because of the high variability within some of the substrate types, these patterns were not significant. Despite using all substrate patches, prickly sculpin were most often observed in cobble habitat patches and least often in sand. Several prickly sculpin attacks on fry and presmolts were observed during some trials, but no actual predation was observed. Substrate–Cover Preference Experiments Neither fry nor presmolts exhibited strong selection for any habitat combination. All patches were utilized, and the average proportional use of unique substrate– cover combinations only varied from 5% to 16%, whereas 8.25% was expected under uniform usage

1199

FIGURE 7.—Proportional use 6 2 SEs of cover types within substrate types (S/C ¼ sand and cobble) by Chinook salmon fry and presmolts during 2004 experiments in artificial habitats. Proportions were averaged over replicate substrate– cover and piscivore (prickly sculpin) treatments for fry (N ¼ 8 for both piscivore treatments) and presmolts (N ¼ 2 for piscivore present; N ¼ 4 for piscivore absent).

(Figure 7). Fry use patterns differed significantly among substrates (ANOVA: P ¼ 0.002) and cover types (P , 0.001; Figure 7). Prickly sculpin treatment (P ¼ 0.85) and diel period (P ¼ 0.78) did not affect substrate–cover patch use patterns, but significant interactions existed between substrate and cover (P ¼ 0.014) and between cover and predator treatment (P ¼ 0.007). When prickly sculpin were present, the use of different cover types differed significantly within and among substrate patches: patches without cover were used more often than patches with overhead cover in gravel (P ¼ 0.024), sand–cobble (P ¼ 0.009), and cobble (P , 0.001). In cobble, coverless patches were also used more often than patches with woody debris (P , 0.001). In general, when prickly sculpin were present, fry use of coverless patches was greater than or equal to use of patches with woody debris or overhead cover, and the use of woody debris patches was greater than or equal to use of overhead cover patches. When prickly sculpin were absent, fry used complex substrates significantly more in the coverless patches. Cobble was used more than sand (Tukey HSD: P , 0.001), gravel (P ¼ 0.004), and sand–cobble (P ¼ 0.022). Sand was used less than gravel (P ¼ 0.032) and sand–cobble (P ¼ 0.006). Use did not differ among substrates within overhead cover or woody debris

1200


patches, but use within cobble substrates was significantly higher for coverless patches than for woody debris or overhead cover patches (Dunnett’s T3: P , 0.001 for both). Presmolt use patterns differed significantly among substrates (ANOVA: P ¼ 0.022) but not among cover types (P ¼ 0.516; Figure 7). Prickly sculpin treatment (P ¼ 0.80) and diel period (P ¼ 0.76) did not affect substrate–cover patch use patterns; however, significant interactions existed between cover and diel period (P ¼ 0.041) and between cover and predator treatment (P ¼ 0.079). During night periods in the absence of prickly sculpin, overhead cover was used more than coverless (P ¼ 0.018) or woody debris (P ¼ 0.006) patches. Prickly sculpin used every available combination of substrate and cover during the experiments but mostly occupied overhead cover patches of all substrate types. Attacks and predation on fry and presmolts were observed during most trials. Discussion In these experiments, juvenile Chinook salmon avoided steeply sloped littoral habitat but exhibited relatively weak responses to a wide range of substrate and cover types among predator treatments and among different diel periods. Presmolts showed stronger responses to slope effects than did fry, but fry exhibited more-coherent responses to substrate and cover. The fry generally used all portions of slope habitat patches, from nearshore to offshore, whereas presmolts strongly selected for offshore regions. No direct effects from predator treatment or diel period were found, but both factors interacted significantly with substrate and cover. Although the effects of different types of predators (cutthroat trout or prickly sculpin) on habitat use were similar, the results of the habitat use responses in the presence of different predators are not directly comparable because habitat function for juvenile Chinook salmon could differ between ambush and cruising predators. These results support the habitat use observations of nearshore Chinook salmon juveniles in Lake Washington (Tabor and Piaskowski 2001). The deep and flat neutral area was the most preferred habitat patch during slope experiments for fry and presmolts. It may be that fish preferred a flat habitat or that they were searching out the deepest available water. Tabor and Piaskowski (2001) reported that the modal depth for fry was centered at 0.4 m in Lake Washington; in our experiments, a depth of 0.5 m was provided in the slope experiments and a depth of 0.4 m was used in the substrate and substrate–cover arenas. Perpendicular transects snorkeled in 2004 in south Lake Washington during March, April, and May (R.

Tabor, U.S. Fish and Wildlife Service, personal communication) indicated that Chinook salmon gradually moved to deeper water as the year progressed and as they became larger. The maximum depth of the experimental slope arena (0.5 m) accommodated the field depths observed in March, but Chinook salmon were observed in areas as deep as 0.68 m in April and 0.94 m in May. These data suggest that use of a deeper arena (1.0 m) may help clarify the importance of the flat neutral area and provide a more realistic picture of bottom-slope preference. Substrate preferences shifted to more-complex substrate when factorial combinations of substrate and cover were offered simultaneously. In the substrateonly experiments with predators during day and crepuscular periods, such preferences may have been lacking because prickly sculpin did not have overhead cover to enhance the visual predation advantage offered by shade (Helfman 1981). Indeed, the only actual prickly sculpin predation events were observed during substrate–cover interaction experiments for both fry and presmolts. In south Lake Washington, prickly sculpin are mainly found in areas with larger substrates like cobble and riprap (Tabor et al. 1998). Diel period had little effect on the habitat preferences of Chinook salmon in our experiments. Diel shifts in habitat preference have been reported in many studies on salmonid habitat usage. Sockeye salmon O. nerka often exhibit diel vertical migration in lakes and are found in deeper waters during the day than during periods of lower light levels (Clark and Levy 1988; Burgner 1991). At night, Chinook salmon rearing in streams often move inshore to slower waters and finer substrates, and then return to deeper portions of the channel with higher flows and larger substrates during daylight (Healey 1991). Our experimental fish showed a significant preference for finer substrates at night, but for all other experiments the diel effect was muted. This may be attributed in part to the urban setting of Lake Washington and the experimental arenas. The light levels near the UW and USGS arenas were similar to ambient levels found on the lake. Artificial lighting on the Cedar River has been shown to enhance the nighttime predation of sockeye salmon fry by cottids (Tabor et al. 2004a). The most notable interaction of diel period and predation effects concerned the vertical distribution of Chinook salmon fry during slope experiments. Fry spent more time on the bottom when cutthroat trout were present but spent more time suspended in the water column during the day. This behavior may be a result of lowered piscivore capture success during high light levels (Mazur and Beauchamp 2003), when prey fish are able to detect predators before the


predators detect them (Howick and O’Brien 1983), and thus, prey are more likely to maintain a safe distance from predators while continuing to feed actively in the water column. At night, when detection capabilities of prey are lower, high activity levels might increase predation vulnerability. Prey responses could change dramatically if prey are confronted with a combination of predators that employ different sensory mechanisms and foraging tactics. Except for changes in the vertical distribution of Chinook salmon fry at night, we found that predation risk had little effect on experimental habitat preferences of fry or presmolts, despite numerous field and laboratory studies that have reported significant effects of predators on prey fish behavior (Cooper and Crowder 1979; Gregory 1993a, 1993b; Walters and Juanes 1993). The lack of response may have been due to several factors, including (1) the use of hatchery Chinook salmon, (2) the riskier behavior of Chinook salmon juveniles exposed to predators than other salmonids, and (3) the low threat posed by the experimental predators during experiments. The effect of hatchery fish use on our overall results is unknown, but it is important to note that over 30% of the adult Chinook salmon carcasses recovered from the naturally spawning population in the Cedar River during 2004 originated from Issaquah Creek Hatchery (Burton et al. 2005), which was the source of our experimental fish. Although the spawning success of these hatchery adults is unknown, we believe that it is reasonable to assume that hatchery fish contribute significantly to the gene pool of lake-rearing juveniles in this basin. Nonetheless, even if the hatchery fish exhibited the appropriate innate responses during these experiments, we cannot definitively exclude the possibility that their behavior might have been modified by hatchery rearing practices that occurred before experimentation. Several studies have suggested that the behavior of hatchery fish in natural environments does not differ significantly from wild fish behavior when the appropriate genetic stocks are used (Brannon et al. 2004). Olla et al. (1998) suggested that predation avoidance in hatchery Chinook salmon, sockeye salmon, and coho salmon O. kisutch can be improved by exposing them to predatory stimuli before release into the wild. However, it is unclear whether survival is improved through habitat selection that reduces predator encounters, predator success, or both. Aquaria experiments on hatchery Chinook salmon from two different watersheds and of similar sizes used in our experiments demonstrated that predator-naı¨ve fish exhibited fright responses to predatory stimuli from piscivores that were endemic and nonendemic to their watersheds (Berejikian et al. 2003). These responses

1201

included a reduction in feeding rate, increased time spent motionless, and increased time spent near substrates. In our experiments, fry exposed to cutthroat trout in slope arenas spent significantly more time in the bottom of the water column during night and crepuscular periods. The nonresponse to predation pressure by our experimental fish may have also been typical for juvenile Chinook salmon, which are more willing than other salmonids to risk exposure to predators (Abrahams and Healey 1993). In Lake Washington, lakerearing Chinook salmon foraged at a high rate and grew faster than stream-rearing Chinook salmon and lake-rearing sockeye salmon in the basin (Koehler et al., in press). Because ocean-type Chinook salmon typically migrate to salt water as age-0 fish, it may be important for them to grow quickly and move to salt water during an optimal time of year. They may therefore accept a higher predation risk as a tradeoff for more aggressive feeding and faster growth that would improve survival in subsequent life stages (Schindler 1999; Biro et al. 2005). Both cutthroat trout and prickly sculpin were considered effective predators during these experiments. Based on observed predator–prey size relationships in Lake Washington (Nowak et al. 2004), cutthroat trout in our trials were capable of eating Chinook salmon fry and were marginally capable of eating presmolts; we also observed their predatory attacks during the experiments. Prickly sculpin used during substrate and substrate–cover experiments averaged 114 mm, and the lengths of our experimental Chinook salmon ranged from 40% to 67% of the total lengths of the prickly sculpin predators. Although the length of juvenile Chinook salmon approached the size threshold of these gape-limited predators, actual predation by prickly sculpin on larger presmolts was observed during substrate–cover experiments. Tabor and Chan (1996) reported that cottids in Lake Washington consumed prey whose lengths were up to 60% of predator length. The single habitat-type experiments (i.e., slope only or substrate only), considered in conjunction with field research (Tabor and Piaskowski 2001), suggest that lake-rearing Chinook salmon may prefer homogenous shorelines devoid of cover if they are characterized by low slopes and finer substrates. Shallow water and a lack of cover may limit the number of predators that are present (Power 1984; McIvor and Odum 1988; DeVries 1990). Steep slopes could allow pelagic piscivores more access to nearshore zones and could limit salmonid escape responses by only allowing vertical instead of horizontal movement. Still, it is important to clarify the difference between

1202


the natural bathymetry of Lake Washington and the altered nearshore bottom-slope habitat, such as riprap and bulkheads. These retained shorelines make up 71% of Lake Washington’s shoreline, but field studies have indicated that rearing Chinook salmon are rarely seen near these structures (Tabor and Piaskowski 2001). During our slope experiments, Chinook salmon generally avoided steep slopes, used the interior portions of the habitat arenas, and were not in close proximity to the concrete edges of the raceway. Thus, natural and unretained shorelines should still be considered preferred habitat. This study, coupled with ongoing field research, was an initial examination of important habitat attributes for Chinook salmon rearing in lakes. Predation may be a mechanism controlling productivity of juvenile Chinook salmon in Lake Washington. Our experiments indicated that these fish did not change habitat preferences in the presence of predators; therefore, when numbers of Chinook salmon rearing in Lake Washington are low, their risk-prone behavior could seriously reduce their overall productivity. Our results do not conclusively demonstrate the value of shoreline restoration to wild Chinook salmon rearing in Lake Washington. Field and experimental research suggests that juvenile Chinook salmon in the lake may benefit from shorelines with a low slope and fine substrate; however, without knowing how docks and other anthropogenic habitat features affect juvenile distribution, behavior, feeding, growth, and survival, we cannot discern the benefit of a series of noncontiguous natural shorelines. Improving connectivity between preferred habitat types has been identified as an important aspect of watershed restoration (Roni et al. 2002). Future research on Chinook salmon should consider these questions: (1) does a predator density threshold influence shifts in juvenile habitat preference; (2) if there is a predator density threshold, does it change among species of predators; (3) what are the roles of inwater and overhead cover on juvenile habitat preference and interactions with predators; and (4) does the magnitude of connectivity between natural and developed shorelines influence juvenile habitat selection and productivity? Larger-scale, in situ experiments should be conducted to examine the broader issues presented in questions (1) and (2) above. The scale of our substrate– cover interaction experiments was not large enough to allow use of cutthroat trout as predators or to expand overhead cover (docks) to a realistic scale. Interpretation of interactions between combinations of cover, substrate, and predation pressure is therefore difficult. Lake Washington field observations have shown that

juvenile Chinook salmon use shorelines with overhead vegetation and small woody debris during certain periods of lake rearing (Tabor et al. 2004b). A larger experimental scale would enable more-realistic combinations and densities of predatory species and an easier interpretation of how fish relate to cover. This level of experimentation would allow biologists to begin answering the question that encompasses the essence of habitat preference experiments for juvenile Chinook salmon in lentic environments (Rosenfeld 2003): are preferred nearshore habitats essential to the fitness of the individual? Acknowledgments Seattle Public Utilities and the Washington Cooperative Fish and Wildlife Research Unit funded this study. Three anonymous reviewers provided thoughtful review and greatly improved this manuscript. R. Tabor and T. Quinn provided study design assistance and extensive reviews of earlier documents. L. Conquest offered valuable statistical advice. A. Cross, S. Damm, L. Duffy, D. Garrett, J. Hall, K. Kurko, A. Lind, H. Limont, J. Mattila, M. Mazur, S. McCarthy, J. McIntyre, N. Overman, E. Schoen, and S. Wang provided field assistance and thoughtful criticism. J. Duda and R. Reisenbichler coordinated the research facilities at USGS. J. Wittouck and D. Rose coordinated the research facilities at UW. The WDFW Issaquah Creek Hatchery provided experimental fish in 2003 and 2004. References Abrahams, M. V., and M. C. Healey. 1993. A comparison of the willingness of four species of Pacific salmon to risk exposure to a predator. Oikos 66:439–446. Berejikian, B. A., E. P. Tezak, and A. L. LaRae. 2003. Innate and enhanced predator recognition in hatchery-reared Chinook salmon. Environmental Biology of Fishes 67:241–251. Biro, P. A., J. R. Post, and M. V. Abrahams. 2005. Ontogeny of energy allocation reveals selective pressure promoting risk-taking behaviour in young fish cohorts. Proceedings of the Royal Society of London B 272:1443–1448. Brannon, E. L., D. F. Amend, M. A. Cronin, J. E. Lannan, S. LaPatra, W. J. McNeil, R. E. Noble, C. E. Smith, A. J. Talbot, G. A. Wedemeyer, and H. Westers. 2004. The controversy about salmon hatcheries. Fisheries 29(9):12– 31. Burgner, R. L. 1991. Life history of sockeye salmon Oncorhynchus nerka. Pages 3–117 in C. Groot and L. Margolis, editors. Pacific salmon life histories. University of British Columbia Press, Vancouver. Burton, K., L. Lowe, and H. Berge. 2005. Cedar River Chinook salmon (Oncorhynchus tshawytscha) redd and carcass surveys: annual report 2004. Seattle Public Utilities, Annual Report, Seattle, Washington.


Chesson, J. 1978. Measuring preference in selective predation. Ecology 59:211–215. Clark, C. W., and D. A. Levy. 1988. Diel vertical migrations by juvenile sockeye salmon and the antipredation window. American Naturalist 131:271–290. Cooper, W. E., and L. B. Crowder. 1979. Patterns of predation in simple and complex environments. Pages 257–267 in H. Clepper, editor. Predator-prey systems in fisheries management. Sport Fishing Institute, Washington, D.C. Crowder, L. B., and W. E. Cooper. 1982. Habitat structural complexity and the interaction between bluegills and their prey. Ecology 63:1802–1813. DeVries, D. R. 1990. Habitat use by bluegill in laboratory pools: where is the refuge when macrophytes are sparse and alternative prey are present? Environmental Biology of Fishes 29:27–34. Duffy, E. J., D. A. Beauchamp, and R. M. Buckley. 2005. Early marine life history of juvenile Pacific salmon in two regions of Puget Sound. Estuarine Coastal and Shelf Science 64:94–107. Edmondson, W. T. 1991. The uses of ecology: Lake Washington and beyond, 1st edition. University of Washington Press, Seattle. Footen, B. 2000. Preliminary results of an investigation into the impacts of piscivorous predation on juvenile Chinook (Oncorhynchus tshawytscha) and other salmonids in Salmon and Shilshole Bays, King County, Washington. Muckleshoot Indian Tribe: Fisheries Department. Available: http://dnr.metrokc.gov/WTD/fish/docs/ footen-b-juvenile-pred.pdf (February 2006). Graynoth, E. 1999. Recruitment and distribution of juvenile salmonids in Lake Coleridge, New Zealand. New Zealand Journal of Marine and Freshwater Research 33:205–219. Gregory, R. S. 1993a. Effect of turbidity on the predator avoidance behavior of juvenile Chinook salmon Oncorhynchus tshawytscha. Canadian Journal of Fisheries and Aquatic Sciences 50:241–246. Gregory, R. S. 1993b. The influence of ontogeny, perceived risk of predation, and visual ability on the foraging behavior of juvenile Chinook salmon. Pages 271–284 in D. J. Stouder, K. L. Fresh, and R. J. Feller, editors. Theory and application in fish feeding ecology. Belle Baruch Library in Marine Science 18. University of South Carolina Press. Columbia. Greene, C. M., and T. J. Beechie. 2004. Consequences of potential density-dependent mechanisms on recovery of ocean-type Chinook salmon Oncorhynchus tshawytscha. Canadian Journal of Fisheries and Aquatic Sciences 61:590–602. Healey, M. C. 1991. Life history of Chinook salmon Oncorhynchus tshawytscha. Pages 331–340 in C. Groot and L. Margolis, editors. Pacific salmon life histories. University of British Columbia Press, Vancouver. Helfman, G. S. 1981. The advantage to fishes of hovering in shade. Copeia 1981:392–400. Hosn, W. A., and J. A. Downing. 1994. Influence of cover on the spatial-distribution of littoral-zone fishes. Canadian Journal of Fisheries and Aquatic Sciences 51:1832–1838. Howick, G. L., and W. J. O’Brien. 1983. Piscivorous feeding behavior of largemouth bass: experimental analysis.

1203

Transactions of the American Fisheries Society 112:508–516. Koehler, M. E. 2002. Diet and prey resources of juvenile Chinook salmon Oncorhynchus tshawytscha rearing in the littoral zone of an urban lake. Master’s thesis. University of Washington, Seattle. Koehler, M., K. Fresh, D. Beauchamp, S. Simenstad, J. Cordell, and D. Seiler. In press. Diet and bioenergetics of lake-rearing juvenile Chinook salmon in Lake Washington. Transactions of the American Fisheries Society. Lister, D. B., and H. S. Genoe. 1970. Stream habitat utilization by cohabiting underyearlings of Chinook Oncorhynchus tshawytscha and coho O. kisutch salmon in the Big Qualicum River, British Columbia. Journal of the Fisheries Research Board of Canada 27:1215–1224. Mazur, M. M., and D. A. Beauchamp. 2003. A comparison of visual prey detection among species of piscivorous salmonids: effects of light and low turbidities. Environmental Biology of Fishes 67:397–405. McIvor, C. C., and W. E. Odum. 1988. Food, predation risk, and microhabitat selection in a marsh fish assemblage. Ecology 69:1341–1351. Nehlsen, W., J. E. Williams, and J. A. Lichatowich. 1991. Pacific salmon at the crossroads: stocks at risk from California, Oregon, Idaho, and Washington. Fisheries 16(2):4–21. Neter, J., M. H. Kutner, C. J. Nachtsheim, and W. Wasserman, editors. 1996. Applied linear regression models, 3rd edition. Irwin, Chicago. Nowak, G. M., R. A. Tabor, E. J. Warner, K. L. Fresh, and T. P. Quinn. 2004. Ontogenetic shifts in habitat and diet of cutthroat trout in Lake Washington, Washington. North American Journal of Fisheries Management 24:624–635. Olla, B. L., M. W. Davis, and C. H. Ryer. 1998. Understanding how the hatchery environment represses or promotes the development of behavioral survival skills. Bulletin of Marine Science 62:531–550. Power, M. E. 1984. Depth distributions of armored catfish: predator-induced resource avoidance? Ecology 65(2):523–528. Roni, P., T. J. Beechie, R. E. Bilby, F. E. Leonetti, M. M. Pollock, and G. R. Pess. 2002. A review of stream restoration techniques and a hierarchical strategy for prioritizing restoration in Pacific Northwest watersheds. North American Journal of Fisheries Management 22:1– 20. Rosenfeld, J. 2003. Assessing the habitat requirements of stream fishes: an overview and evaluation of different approaches. Transactions of the American Fisheries Society 132:953–968. Schindler, D. E. 1999. Migration strategies of young fishes under temporal constraints: the effect of size-dependent overwinter mortality. Canadian Journal and Fisheries and Aquatic Sciences 56(Supplement 1):61–70. Seiler, D., G. Volkhardt, and L. Kishimoto. 2003. Evaluation of downstream migrant salmon production in 1999 and 2000 from three Lake Washington tributaries: Cedar River, Bear Creek, and Issaquah Creek. Washington Department of Fish and Wildlife, Report FPA 02–07, Olympia. Sempeski, P., and P. Gaudin. 1995. Size-related changes in diel distribution of young grayling Thymallus thymallus.

1204


Canadian Journal of Fisheries and Aquatic Sciences 52:1842–1848. Tabor, R. A., and W. A. Wurtsbaugh. 1991. Predation risk and the importance of cover for juvenile rainbow trout in lentic systems. Transactions of the American Fisheries Society 120:728–738. Tabor, R. A., and J. Chan. 1996. Predation on sockeye salmon fry by cottids and other predatory fishes in the lower Cedar River, 1996. U.S. Fish and Wildlife Service, Annual Report, Lacey, Washington. Tabor, R. A., J. Chan, and S. Hager. 1998. Predation of sockeye salmon fry by cottids and other predatory fishes in the Cedar River and southern Lake Washington, 1997. U.S. Fish and Wildlife Service, Annual Report, Lacey, Washington. Tabor, R. A., and R. M. Piaskowski. 2001. Nearshore habitat use by juvenile Chinook salmon in lentic systems of the Lake Washington basin, annual report, 2001. U.S. Fish and Wildlife Service, Annual Report, Lacey, Washington. Tabor, R. A., G. S. Brown, and V. T. Luiting. 2004a. The effect of lighting intensity on sockeye salmon fry

migratory behavior and predation by cottids in the Cedar River, Washington. North American Journal of Fisheries Management 24:128–145. Tabor, R. A., J. A. Scheurer, H. A. Gearns, and E. P Bixler. 2004b. Nearshore habitat use by juvenile Chinook salmon in lentic systems of the Lake Washington basin, annual report, 2002. U.S. Fish and Wildlife Service, Annual Report, Lacey, Washington. Toft, J. D. 2001. Shoreline and dock modifications in Lake Washington. University of Washington, Report SAFSUW-0106, Seattle. Walters, C. J., and F. Juanes. 1993. Recruitment limitation as a consequence of natural-selection for use of restricted feeding habits and predation risk-taking by juvenile fishes. Canadian Journal of Fisheries and Aquatic Sciences 50:2058–2070. Werner, E. E., D. J. Hall, D. R. Laughlin, D. J. Wagner, L. A. Wilsmann, and F. C. Funk. 1977. Habitat partitioning in a freshwater fish community. Journal of the Fisheries Research Board of Canada 34:360–370. Zar, J. H. 1996. Biostatistical analysis, 4th edition. Prentice Hall, Englewood Cliffs, New Jersey.