Transactions of the American Fisheries Society 132:1076±1088, 2003 ᭧ Copyright by the American Fisheries Society 2003

Winter Ecology of Kokanee: Implications for Management

GEOFFREY B. STEINHART*1 AND WAYNE A. WURTSBAUGH Aquatic, Watershed, and Earth Resources Department and the Ecology Center, State University, Logan, Utah, 84322-5210, USA

Abstract.ÐWe sampled various limnological parameters and measured growth and diet of age-0 kokanee Oncorhynchus nerka (lacustrine ) during two winters in a high-mountain lake of the Sawtooth Valley, . Although winter has been recognized as an important period for many warmwater ®shes and for stream-dwelling salmonids, winter limitations have only re- cently been studied for coolwater and coldwater species. Ice and snow cover in winter limited light penetration. As a result, chlorophyll-a and zooplankton density were lower in ice-covered periods than during ice-free periods. The weight of stomach contents was often below a maintenance ration, yet the incidence of empty stomachs was extremely low (1 of 63) and the weight of stomach contents increased as energy reserves declined, indicating that kokanee were actively foraging during winter. Kokanee length and weight remained constant during the winter of 1993±1994 but increased from November through May in 1994±1995. Condition factors, however, declined sig- ni®cantly over the winter in both years, and lipid content approached levels associated with mortality. Differences in growth patterns may have been caused by a combination of changes in zooplankton density and kokanee abundance and in kokanee behavior to defend energy reserves or avoid predation. Results demonstrated the ambiguity of some growth measurements and the importance of choosing the correct metric for measuring growth in ®shes. Because juvenile kokanee and sockeye salmon are ecologically similar, management efforts to restore the endangered Snake River sockeye salmon to the Sawtooth Valley lakes should recognize that winter conditions might be a bottleneck for this species.

Winter can be a critical period for many ®shes class strength (Toneys and Coble 1979; Henderson because cold temperatures and reduced productiv- et al. 1988; Miranda and Hubbard 1994). Even for ity can limit survival (Toneys and Coble 1979; coldwater species such as salmonids, winter can Miranda and Hubbard 1994). Identifying mecha- be an extremely harsh period (Cunjak 1988; Met- nisms underlying winter mortality is essential to calfe and Thorpe 1992). understanding recruitment and population dynam- Winter growth rate and lipid loss affect not only ics of ®shes. For young ®sh, winter mortality is overwinter survival but also the behavior, future often the result of starvation, predation, or os- survival, and choice of life history tactics in salm- moregulatory failure (Toneys and Coble 1979; Mi- on. Most young ®sh rely on energy stores to meet randa and Hubbard 1994). Signi®cant lipid losses, metabolic demands in winter, but when lipid con- a consequence of starvation, during winter have tent declines, salmon show increased appetite (i.e., been documented in sand smelt Atherina boyeri increased consumption) and foraging activity (Henderson et al. 1988), smallmouth bass Microp- (Metcalfe and Thorpe 1992; Kadri et al. 1996; Ma- terus dolomieu (Oliver and Holeton 1979), large- clean and Metcalfe 2001). Increased foraging ac- mouth bass M. salmoides (Ludsin and DeVries tivity can reduce dependence on limited energy 1997) pikeminnow Ptychocheilus lucius reserves, but it increases predation risk (Danne- (Thompson et al. 1991), and many other species. witz and Petersson 2001; Reinhardt 2002). If win- Lipid loss in age-0 ®sh, owing to their high met- ter conditions or energy reserves do not allow abolic demands, may be so severe as to cause mor- salmon to reach adequate size or lipid content by tality (Shuter and Post 1990). Size-selective mor- spring, postwinter mortality may increase owing tality (i.e., small individuals dying) has been doc- to metabolic demands of smolti®cation and mi- umented for many ®shes and may control year- gration (Hoar 1976; Rondorf et al. 1985; Jonsson and Jonsson 1998). Smolt-to-adult survival im- * Corresponding author: [email protected]. proves with increased size and condition of smolts 1 Present address: Aquatic Ecology Laboratory, The (Holtby et al. 1990; Henderson and Cass 1991). Ohio State University, 1314 Kinnear Road, Columbus, In addition, the timing and expression of smolti- Ohio 43212-1156, USA. ®cation and maturation are at least partially con- Received September 20, 2002; accepted March 25, 2003 trolled by size and energy stores (Rowe et al. 1991;

1076 WINTER ECOLOGY OF KOKANEE 1077

Berglund 1995; Duston and Saunders 1999; Ri- little time for them to build lipid reserves before kardsen and Elliot 2000). Poor condition shortened smolti®cation (Bjornn et al. 1968). Many ®shes the spawning migration some individuals in a pop- deplete lipids during migration; thus, increased en- ulation of Atlantic herring Clupea harengus (Slotte ergetic costs of migration could reduce survival or 1999), whereas low lipids halted the migration of even prevent migration (Rondorf et al. 1985; Sve- silver-stage Euorpean eels Anguilla anguilla (Sve- daÈng and WickstroÈm 1997; Leonard and McCor- daÈng and WickstroÈm 1997). mick 1999). Winter conditions are especially important for Efforts are currently underway to augment the salmon in lakes at high latitudes or in high moun- remnant populations of sockeye salmon in three tains, such as the lakes of the Sawtooth Valley of Sawtooth Valley lakes, but all ®ve lakes contain central Idaho. Historically, ®ve lakes in the valley self-sustaining populations of kokanee. Although supported Snake River populations of sockeye they usually have different life histories, kokanee salmon Oncorhynchus nerka, kokanee (lacustrine can migrate to the ocean and have established sockeye salmon) or both (Bjornn et al. 1968). Life anadromous populations in lakes previously de- history tactics distinguish sockeye salmon and ko- void of sockeye salmon (Taylor et al. 1996; Quinn kanee: kokanee spend their entire life in fresh- et al. 1998). In fact, kokanee may have been re- water, but sockeye salmon migrate to the ocean at sponsible for the return of sockeye salmon to the age 1 or 2 and return 1±3 years later to their natal Sawtooth Valley lakes after a dam entirely block- stream to spawn as adults (Burgner 1992). Unlike ing access to the ocean was breached in 1934 many warmwater species, O. nerka (we will use (Bjornn et al. 1968). Sockeye salmon, however, O. nerka when referring collectively to sockeye sometimes remain in freshwater for their entire life salmon and kokanee) are active at temperatures (Ricker 1938), and it is now believed that these below 4ЊC, so winter is not a period of dormancy ``residual'' sockeye salmon (i.e., another fresh- (Burgner 1992). In fact, at 4ЊC, sockeye salmon water resident that differs from kokanee) in the must consume about 0.1% of their wet weight per Sawtooth Valley lakes gave rise to the current day to meet basic metabolic requirements (Brett anadromous population. et al. 1969). Due to low zooplankton abundance, We sought to understand what in¯uenced the however, fewer than 2% of sockeye salmon in a winter growth and condition of kokanee in Stanley coastal Alaskan lake consumed food during winter Lake, in the Sawtooth Valley, and how that might (Ruggerone 1992). Even in a eutrophic coastal sys- affect the recovery of the Snake River sockeye tem such as Lake , declining energy salmon. Because Snake River sockeye salmon reserves caused juvenile sockeye salmon to be- were not present during our study, we used ko- come more active in late winter than in early winter kanee as a surrogate. Sockeye salmon and kokanee (Eggers 1978). Unlike coastal lakes, many O. ner- eat similar food (Wood et al. 1999), undergo com- ka systems, including the Sawtooth Valley lakes, parable smolti®cation processes (Foote et al. 1992, are ice covered for more than half the year. Ice 1994; Rice et al. 1994), and are more closely re- cover reduces light penetration and photosynthe- lated to each other within a lake than between lakes sis, which, in turn, limit winter productivity and (Foote et al. 1989; Taylor et al. 1997). Speci®cally, O. nerka foraging ability (Steinhart and Wurts- we quanti®ed kokanee length, weight, and ener- baugh 1999). getic content from November through May and Understanding winter conditions in the Saw- examined how those variables related to zooplank- tooth Valley lakes may provide insight into the ton abundance and kokanee behavior. recruitment success of Snake River sockeye salm- on. Snake River sockeye salmon declined to such Methods low numbers that they were listed as an endan- Study site.ÐStanley Lake, located in the Snake gered species in 1991. Sockeye salmon smolts River drainage in the Sawtooth Valley of central from the Snake River travel over 1,400 km and Idaho (44ЊN, 115ЊW) at an altitude of 1,985 m, is pass eight hydroelectric dams en route to the typically ice covered from late November until ocean. Each dam may delay migration by a week early May. It covers 81 ha and has mean and max- or more and increase a smolt's exposure to pi- imum depths of 13 and 26 m. During our study, scivorous ®sh (Venditti et al. 2000; Budy et al. the ®sh community of Stanley Lake was predom- 2002). Not only is this one of the longest seaward inated by kokanee and Salvelinus na- migrations among sockeye salmon populations, maycush but also included O. clarki, but migration begins before ice-out, so there is rainbow trout O. mykiss, brook trout S. fontinalis, 1078 STEINHART AND WURTSBAUGH mountain white®sh Prosopium williamsoni, north- and during periods of ice cover, we used sinking ern pikeminnow P. oregonensis, suckers Catosto- horizontal gill nets (each with three panels of mus spp., redside shiner Richardsonius balteatus, square mesh: 22, 27, and 35 mm), ¯oating hori- dace Rhinichthys spp., and sculpin Cottus spp. zontal gill nets (each with four panels of square (Taki and Mikkelsen 1997). We sampled limno- mesh: 25, 32, 38, and 44 mm), and vertical gill logical characteristics and the kokanee population nets 25 m deep by 3 m wide (each with one panel monthly during October±May and every 2 weeks of square mesh: 19, 25, or 38 mm) set in holes cut during June±September 1993±1995. by chainsaw (Hubert 1996). Initial sampling re- Limnological sampling.ÐOn each sampling vealed that few kokanee were captured during day- date, we sampled three stations near the center of light, so gill nets were set for 24-h and checked Stanley Lake, where depth exceeded 23 m. During between 0800 and 1000 hours. After capture, we ice cover, we cut holes with an ice auger or a measured maximum total length (TL; nearest mm) chainsaw. We measured temperature and dissolved and wet weight (WW; nearest 0.01 g) and saved oxygen (1-m intervals to 10 m, and every 5 m individuals for gut and otolith analyses. below 10 m) using either a YSI model 58 dissolved Kokanee less than 150 mm TL were considered oxygen meter (November through May) or a Hy- to be age 0, graduating to age 1 on 1 January. This drolab Surveyor III meter (June through October). size threshold is appropriate because September Light intensity (photosynthetically active radiation trawling in Stanley Lake revealed age-0 kokanee 400±700 nm) was measured with a submersible to be 75 mm mean TL (range, 35±97 mm) and age Li-Cor Model LI-188B radiometer. The probe was 1 to be 161 mm (range, 111±179 mm; Paul Kline, lowered at 1-m intervals to the bottom of the lake Idaho Department of Fish and Game, unpublished or until light was undetectable (Ͻ0.02 W/m2). A data). Using a threshold of 150 mm, the mean TL surface light sensor corrected for changes in sur- of age-0 kokanee in Stanley Lake in November face intensity due to cloud cover. During ice cover, was 113 mm (range, 71±135 mm), well below the the light sensor was lowered through a 0.75-m2 age-1 September mean total length. hole. To avoid light contamination from the hole, Kokanee condition.ÐWe calculated relative we extended the sensor 3 m laterally from the hole condition (Kn; Anderson and Neumann 1996) us- and ®lled the hole with snow. We collected water ing population-speci®c coef®cients from the samples (0±6 m and 23 m) for chlorophyll-a anal- length±weight regression for all captured kokanee ysis; two replicate 50-mL aliquots per sample were (i.e., we included kokanee Ͼ150 mm TL): ®ltered through 0.45-␮m cellulose acetate ®lters Ϫ63.11 and frozen. Filters were placed in 6 mL of 100% wet weight ϭ 4 ϫ 10 ´ (TL ) buffered methanol, extracted for 24 h and analyzed (r2 ϭ 0.99; N ϭ 329). (1) with a Turner model 111 ¯uorometer to within Ϯ 0.1 ␮g cholorophyll a/L (Holm-Hansen and Rie- To determine dry weight (DW), ®sh were thawed, mann 1978). We collected zooplankton at each sta- cut into pieces, and dried for 24 h at 60ЊC (i.e., a tion with an 80-␮m mesh closing conical net (35 temperature that will not oxidize fats; Kerr et al. cm diameter and 150 cm long), equipped with a 1982). Lipids were quanti®ed by Soxhlet lipid ex- General Oceanics ¯owmeter. Samples were im- traction, using petroleum-ether as a solvent to pre- mediately preserved in 10% formalin-sucrose so- serve structural fats (Bligh and Dyer 1959). After lution and subsequently identi®ed, counted, and drying, lipids were extracted from individual ®sh measured. Zooplankton densities were corrected for 24 h. Because lipid content is related to ®sh for the volume of water samples, and dry biomass size, we used residual lipid weights to avoid con- was calculated using length±weight regressions founding effects of ®sh size (Sutton et al. 2000). (McCauley 1984; Koenings et al. 1987). Residual lipid weight (RLW) was calculated as the Fish sampling.ÐKokanee were collected difference between observed lipid weight (LW) monthly from November to May with a combi- and predicted LW from the regression: nation of trawls and gill nets. In November and May, when the lake was ice-free, we collected ko- log10 (LW) ϭ 3.301 ´ log 10 (TL) Ϫ 6.931 kanee at or near the dark phase of the moon using (r2 ϭ 0.49; N ϭ 191). (2) a midwater beam trawl (3 m wide, 7 m deep). Oblique trawl hauls, 30 min in duration at a speed We compared the between-year variation of ko- of about 1 m/s, were made at various depths from kanee pre- and postwinter with respect to TL, WW, the surface to 20 m. In May of 1994 and 1995, Kn, and RLW with t-tests (␣ϭ0.05) and examined WINTER ECOLOGY OF KOKANEE 1079

TABLE 1.ÐAverage limnological conditions during ice-free (mid-May through early November) and ice-covered periods (December±April) in Stanley Lake, Idaho. Surface temperature was measured at 0 m, surface chlorophyll a was the mean from 0±6 m, hypolimnetic chlorophyll a was at 1 m above the bottom (typically 25 m), and zooplankton were collected from the surface to the bottom.

Surface Chlorophyll a (mg/L) temperature Depth of Zooplankton Year(s) Season (ЊC) 1% light (m) Surface Hypolimnetic (␮g/L) 1993 Ice free 11.1 14.40 1.26 0.91 18.3 1993±1994 Ice covered 3.6 3.85 0.92 0.12 2.35 1994 Ice free 14.1 14.95 0.52 1.1 22.71 1994±1995 Ice covered 3.7 5.30 0.55 0.08 11.32 changes among months within a year and between Johnson 1992), we predicted prey consumption years with analysis of covariance (ANCOVA). that matched observed changes in wet weight and Kokanee diet.ÐGut contents were collected estimated changes in caloric density during No- from just 63 of the 217 kokanee captured because vember through May. Model parameters included we randomly selected individuals for other exper- bioenergetic equations for kokanee (Beauchamp et iments and we did not sample diets from ®sh that al. 1989) and observed water temperatures (as- clearly had been in the gill net for an extended suming O. nerka were at 10 m; Steinhart and period (e.g., discolored or stiff). Kokanee used for Wurtsbaugh 1999). Copepods were the primary diet analyses were placed on wet ice, and lavage food (61% by weight) of kokanee in both years without dissection was conducted with a water- (Steinhart 1997), and we estimated of copepod en- ®lled syringe within 2 h. Stomach contents were ergy density to be 2.5 kJ/g (Hewett and Johnson preserved in 95% ethanol and subsequently iden- 1992). Monthly kokanee energy density was es- ti®ed, counted, and measured. We calculated the timated using energy densities for protein (24 kJ/ total undigested-weight of ingested prey based on g), lipid (40 kJ/g), and carbohydrates (17 kJ/g; zooplankton length±weight regressions (Mc- Brett et al. 1969). Lipid weight was measured di- Cauley 1984; Koenings et al. 1987). At winter lake rectly, carbohydrate weight was estimated as 2% temperatures near 4ЊC, O. nerka take over 24 h to of kokanee wet weight, and protein weight was the digest 75% of their stomach contents (Brett et al. difference (Brett et al. 1969). 1969). Although back-calculated weights may un- derestimate total ration somewhat, such a slow di- Results gestion rate should still provide insight into chang- Limnology es in daily ration. Using bioenergetic simulations (Hewett and Snow and ice (0.4±0.8 m thick) covered Stanley Lake from late November until early May in both years, and blocked up to 99% of incident light from reaching the water. Reduced light penetration and low temperatures during ice cover produced hy- polimnetic chlorophyll-a concentrations 9±12 times lower than at ice-free periods (Table 1). Sur- face (0±6 m) chlorophyll a did not vary seasonally; photoinhibition limited production in surface wa- ters during ice-free periods (Gross et al. 1998). Dissolved oxygen levels were suf®cient to support salmon (Ͼ5 mg/L) through most of the water col- umn; however, on occasion, oxygen levels fell be- low 5 mg/L at 15 to 25 m. Zooplankton biomass during ice-free months exceeded that during ice cover (Table 1; Figure 1) by sixfold in 1993±1994 and by twofold in 1994± FIGURE 1.ÐSeasonal patterns of crustacean zooplank- ton biomass in Stanley Lake, Idaho, from 1993 to 1995. 1995. During ice-free periods, Daphnia spp. were Data do not include copepod nauplii, which were not the most abundant zooplankton, but cyclopoid co- found in kokanee diets during this study. pepods and Bosmina spp. predominated during ice 1080 STEINHART AND WURTSBAUGH

there was no difference between the trawl and combined gill-net distributions (KS test: D ϭ 0.12, P ϭ 0.66). Therefore, we combined all catches because our comparisons between months were not heavily biased by collection method, especially considering that trawls were used in both Novem- ber and May and that residual lipid weight analysis corrected for possible TL biases. Kokanee did not grow from November to May in 1993±1994 but did grow over the winter of 1994±1995 (Table 2; Figure 3). Kokanee were sim- ilar in length (t ϭ 0.81, df ϭ 43, P ϭ 0.42) and weight (t ϭ 1.04, df ϭ 45, P ϭ 0.30) in November of both years but grew more during the winter of FIGURE 2.ÐStacked-bar histogram of the total-length 1994±1995 than in 1993±1994. Although kokanee frequencies of three different collection techniques lengths in May were not signi®cantly different be- (midwater trawl, horizontal gill net, vertical gill net) tween years (117 mm versus 126 mm; t ϭ 1.65, used to capture kokanee in Stanley Lake, Idaho, during November±May of 1993±1995. All kokanee less than df ϭ 16, P ϭ 0.06), mean kokanee wet weights 150 mm total length were assumed to be age 0 during were 3 g heavier after the winter of 1994±1995 November±December and age 1 during January±May. (t ϭ 1.7, df ϭ 17, P ϭ 0.05) and ANCOVA results show that kokanee TL and wet weight increased cover. Mean zooplankton biomass from May to during 1994±1995 (Table 2). Wet weight did not November was 25% higher in 1994 than 1993, and increase proportionately with TL, as evidenced by during ice cover, mean zooplankton biomass was signi®cant declines in condition factor in both ®ve times higher in 1994±1995 than in 1993±1994 years (Table 2; Figure 3). Kokanee condition fac- (Table 1). tors in November were signi®cantly higher in 1994 than in 1993 (t ϭ 2.64, df ϭ 34, P ϭ 0.01), but Kokanee Growth condition factors converged by May (t ϭ 0.77, df Kokanee (Ͻ150 mm TL) captured in trawls, hor- ϭ 16, P ϭ 0.45). izontal gill nets and vertical gill nets were similar Although kokanee grew in both length and in length, although the trawl caught more small weight in 1994±1995, their RLW declined during kokanee than did the gill nets (Figure 2). Total November through May in both years (Table 2; length of kokanee caught in horizontal and vertical Figure 3). Kokanee began with 50% lower RLW gill nets did not differ signi®cantly (Kolmogorov± in November 1993 than in November 1994 (t ϭ Smirnov [KS] test: D ϭ 0.15, P ϭ 0.37; Sokal and 2.54, df ϭ 9, P ϭ 0.032), but in each year, RLW Rohlf 1981), so catches from these two net types had converged to similarly low values by May (t ϭ were combined. The midwater trawl caught dif- 1.10, df ϭ 10, P ϭ 0.30). Month had a signi®cant ferent kokanee sizes than gill nets (KS test: D ϭ effect on RLW, and there was a signi®cant differ- 0.22, P ϭ 0.04), but the difference was due to a ence between years (Table 2). Owing to changes few small kokanee captured in the trawls (Figure in wet weight and RLW, calculated mean energy 2). If the four smallest kokanee were removed, content of kokanee declined from 89 to 55 kJ over

TABLE 2.ÐAnalysis of covariance results of kokanee growth parameters from Stanley Lake, Idaho. Year columns contain regression coef®cients and statistics for individual ice-covered seasons; sample month was the independent variable. Year comparison presents month±year interactions, signi®cant values (in bold) indicating that measured growth rates differed between years. Abbreviations: TL ϭ total length, WW ϭ wet weight, Kn ϭ residual condition factor, and RLW ϭ residual lipid weight. Sample sizes are presented in Figure 3.

1993±1994 1994±1995 Year comparison Growth parameter Slope tPSlope tPFP TL 1.26 1.56 0.12 2.70 5.59 Ͻ0.001 2.46 0.12 WW 0.18 0.77 0.44 0.78 5.35 Ͻ0.001 4.93 0.027 Kn Ϫ0.02 Ϫ3.69 Ͻ0.001 Ϫ0.01 Ϫ2.96 0.004 0.56 0.46 RLW Ϫ0.05 Ϫ4.95 Ͻ0.001 Ϫ0.08 Ϫ20.30 Ͻ0.001 13.27 Ͻ0.001 WINTER ECOLOGY OF KOKANEE 1081

FIGURE 3.ÐChanges in mean total length, wet weight, condition factor, and residual lipid weight for kokanee from Stanley Lake, Idaho, over the winters of 1993±1994 and 1994±1995. Whiskers represent the 10th and 90th percentiles, the boundaries of the boxes indicate the 25th and 75th percentiles, and the bars indicate the medians. Sample sizes for total length, wet weight, and condition factor are presented in the top panels, and sample sizes for residual lipid weight are presented in the bottom panels. the winter of 1993±1994 but only declined from we compared November TL quantiles to May TL 89 to 85 kJ in 1994±1995. quantiles (5, 10, 25, 50, 75, 90, and 95) and looked Observed growth patterns may have been con- for size-selective mortality or growth (Figure 4; founded by size-selective overwinter mortality, so for detailed methods and interpretation, see Post 1082 STEINHART AND WURTSBAUGH

FIGURE 5.ÐKokanee lipid content (percent of wet weight) as a function of percent water. Data are for ko- FIGURE 4.ÐQuantile±quantile (QQ) and growth in- kanee captured in Stanley and Alturas lakes, Idaho, as crement plots for kokanee over the winters of 1993± well as for kokanee from a Utah hatchery. The critical 1994 and 1994±1995 in Stanley Lake, Idaho. If kokanee level is the mean percent lipid of kokanee that died due grew from November to May, points on the QQ plots to starvation (points indicated by plus signs were not (panels A and C) fall above the 1:1 line. Growth incre- used in the multiple regression). Although the graph only ment plots (panels B and D) show the magnitude of shows the relationship between water content and lipids, growth for each November's total length (TL) quantile. total length (TL) also had a signi®cant (P Ͻ 0.0001) The slopes of the QQ plots indicated that no size-selec- in¯uence on lipids. tive processes occurred (H0: slope ϭ 1; Post and Evans 1989). Likewise, no size-selective processes are indi- cated in the growth increment plots (H0: slope ϭ 0; t Both percent water and total length were signi®- and P statistics are identical to QQ plots). cant factors (P Ͻ 0.0001) in the regression equa- tion. Sources of these data used for this regression and Evans 1989). Growth increments were greater analyses were (1) Stanley Lake (this study), (2) during 1994±1995 than 1993±1994, which cor- kokanee collected from a second Sawtooth Valley roborated our ANCOVA results that kokanee grew lake (Alturas Lake) by midwater trawl and gill nets only during the 1994±1995 winter. Size-selective in 1993±1995 (Steinhart 1997), and (3) kokanee processes did not affect Stanley Lake kokanee in obtained from a Utah hatchery (original stock 1934±1994 winter. Slopes of TL quantile±quantile source was Lake Whatcom, Washington, but these (QQ) and growth increment plots, along with no offspring were progeny of parents stocked into change in size, demonstrate that neither size-se- , Utah). Six of the Utah lective mortality nor size-selective growth oc- kokanee were subjected to a food-deprivation ex- curred (case 1 in Post and Evans 1989). The QQ periment and died with a mean lipid content of plot and growth increment slopes suggest that ei- 2.3% (Steinhart 1997). Field lipid values measured ther no size-selective processes operated in 1994± in kokanee from Stanley and Alturas lakes in May 1995 (case 4) or that a combination of size-selec- 1994 and 1995 approached this critical lipid con- tive growth (small Ͻ large) and size-selective mor- tent (Figure 5). tality (for small or large individuals; cases 8b and 9b in Post and Evans 1989) may have been pro- Kokanee Diet and Bioenergetics duced the apparent growth. The mean weight of kokanee stomach contents There was a strong relationship between total was extremely low between early and mid winter, length, percent water, and percent lipid for kokan- but it increased as lipid content decreased in late ee (Figure 5). We derived the following multiple winter (Figure 6). Light limited foraging ability regression, which may be used for to estimate per- when Stanley Lake was ice covered (Steinhart and cent lipid in kokanee from populations other than Wurtsbaugh 1999), but during the ice-free months Stanley Lake: of November and May, kokanee consumed less percent lipid ϭ 73.8 Ϫ 0.85´ (percent water) prey when their energy reserves were high (No- vember) than when lipids were nearly exhausted Ϫ 0.03´ TL (May). Regurgitation may have occurred in some (r2 ϭ 0.93; N ϭ 341). (3) kokanee taken by gill nets, but that did not appear WINTER ECOLOGY OF KOKANEE 1083

winters. Length and weight changes, however, dif- fered among winters; in 1993±94, kokanee did not grow in length or weight but they did in 1994± 1995. May lipid levels converged in both years at levels approaching the critical limit where mor- tality may occur. The lipid levels in May were similar to values when Colorado pikeminnow die (Thompson et al. 1991) and, when the only re- maining lipids are structural, in Chinook salmon O. tshawytscha (Rondorf et al. 1985). Although we could not rule out the possibility of size-selective growth or mortality in 1994±1995, we found no size-dependent changes in 1993±1994. Given the small size of age-0 kokanee compared with their FIGURE 6.ÐRelationship between mean weight of gut contents and residual lipid weight of kokanee captured lake trout predators, it is reasonable that size-de- during November±May, 1993±1995, in Stanley Lake, pendent predation was not an important factor. Idaho (r2 ϭ 0.74; P Ͻ 0.006). Note that gut weight Therefore, we believe that kokanee growth pat- generally increased from November to May in each year, terns during winter may be explained by different whereas residual lipid weight decreased. prey abundances and behavioral changes to con- serve limited energy reserves. Limited prey abun- to overly bias our results because the lowest gut dance during winter may be accentuated by re- weights were found in kokanee caught during No- duced foraging ability resulting from low light vember trawl sampling. In addition, May samples penetration through ice and snow covering the lake included ®sh taken by both gill nets and trawls, (Steinhart and Wurtsbaugh 1999). and were the only samples in which the weight of Growth of sockeye salmon and kokanee is often the stomach contents exceeded the estimated main- density dependent and related to zooplankton den- tenance ration for O. nerka at 4ЊC (0.1% wet sity (Hyatt and Stockner 1985; Rieman and Meyers weight/d; Brett et al. 1969). Sixty-two of 63 stom- 1992). In 1994, kokanee condition factor and lipid achs examined between November 1993 and May content in November were higher than in 1993, 1995 contained prey, suggesting that kokanee were which may have been the combined result of 25% actively feeding during our sampling. Observed more zooplankton and 75% fewer kokanee during kokanee consumption (Figure 6) was similar in ice-free periods in 1994 (Taki and Mikkelsen 1993±1994 and 1994±1995 during both ice cov- 1997). In addition, kokanee grew more (in TL and ered (t ϭ 0.08, df ϭ 29, P ϭ 0.94) and the ice- WW) and their predicted November±May con- free periods in November and May (t ϭ 0.25, df sumption was greater for 1994±1995, when mean ϭ 33, P ϭ 0.80). zooplankton biomass was ®ve times greater, than From the bioenergetics simulations, the predicted it was in 1993±1994. These results suggest that kokanee consumption rate was 53% higher in 1994± zooplankton abundance can limit the growth of 1995 than in 1993±1994. From November 1993 to Stanley Lake kokanee in winter and that compe- May 1994, kokanee consumed an average of 0.017 tition with abundant O. nerka may further limit g of zooplankton/d, or 0.14% of their wet weight/ growth (Taki and Mikkelsen 1997). d (near their maintenance ration of 0.1% WW/d; The weight of kokanee gut contents increased Brett et al. 1969). In 1994±1995, average kokanee as lipid content decreased. It is presumed that the consumption was 0.026 g of zooplankton/d (0.21% weight of gut contents can be used as a proxy for WW/d). Consumption differences were re¯ected in appetite (Metcalfe and Thorpe 1992) or food avail- energy loss rates. From November 1993 to May ability. Increased appetite with decreasing energy 1994, kokanee energy content declined at an esti- stores occurs in Atlantic salmon Salmo salar (Met- mated 175 J/d, but during 1994±1995, when zoo- calfe and Thorpe 1992; Silverstein et al. 1999), plankton densities were higher, energy content de- brown trout Salmo trutta (Dannewitz and Petersson clined at 20 J/d. 2001), and Arctic char Salvelinus alpinus (Jobling Discussion and Miglavs 1993). Furthermore, the timing of life Kokanee Growth during Winter history events affects appetite, growth rate, and In Stanley Lake kokanee, percent lipid and es- behavior (Thorpe 1987; Metcalfe 1998). For ex- timated energetic content declined during both ample, when Atlantic salmon fry were split into 1084 STEINHART AND WURTSBAUGH two groups beginning in their ®rst summer; the er sockeye salmon recovery efforts. Smolts leaving larger fry fed throughout the winter in preparation the Sawtooth Valley are some of the largest of any for smolting in the spring, whereas the smaller fry sockeye salmon system (Bjornn et al. 1968; Taki fed less, were less active, and smolted 1 year later. and Mikkelsen 1997), and they must undergo a The behavior of the small fry reduced predation 1,400-km migration. This arduous migration may risk over the winter, although both groups of ®sh explain why smolts traditionally left the Sawtooth still showed increased appetites and activity when Valley lakes at such a large size. Changes along energy reserves declined (Metcalfe and Thorpe this migration route, speci®cally the addition of 1992; Kadri et al. 1996). eight dams, have made this journey more perilous Although kokanee do not typically migrate to (Budy et al. 2002). Therefore, to successfully re- the ocean, they do undergo many of the physical store the Snake River sockeye salmon, it is im- changes involved with smolti®cation and matu- portant to determine if the rearing lakes produce ration (Foote et al. 1992, 1994; Rice et al. 1994). smolts capable of surviving this migration. Until Both processes require energy and can be stressful recently, researchers had ignored the effects of (Hoar 1976); therefore, kokanee must be physi- winter in many Paci®c salmon systems, but we cally prepared before undertaking these changes. believe winter conditions could have a profound In 1993±1994, kokanee entered the winter with affect on salmon growth and survival. Alternate less energy than in 1994±1995, and their behavior life histories (e.g., anadromy versus freshwater could be compared with the smaller Atlantic salm- residency) are found in many salmonids; the fast- on fry in the above example. Kokanee in 1993± est and slowest growing ®sh may remain in fresh- 1994 did not grow over winter, possibly because water, and ®sh growing at average rates may mi- of limited zooplankton availability. A combination grate to sea (Nordeng 1983; Metcalfe 1998). of reduced light penetration and low zooplankton Therefore, winter conditions also could affect life density reduced kokanee foraging ability during history strategy for some species. Under poor winter, and as a result, kokanee may have foregone growth conditions, some individuals may delay growth to adopt behaviors (e.g., diel vertical mi- migration until suf®cient size and energy reserves gration) that reduced predation risk from pisci- are accumulated, and others may mature in fresh- vores (Steinhart and Wurtsbaugh 1999). Converse- water (e.g., residual sockeye salmon). ly, in 1994±1995, kokanee were in better condition Like kokanee, sockeye salmon growth is often and may have foraged more actively (as indicated density dependent and limited by food resources by bioenergetics consumption estimates) but still (Hyatt and Stockner 1985; Rieman and Meyers declined in condition because they utilized their 1992). Monitoring programs in the Sawtooth Val- energy reserves to continue growth and, perhaps, ley lakes have documented an inverse relationship to hasten smolting and maturation, or avoid pre- between O. nerka abundance and zooplankton bio- dation. Morgan et al. (2000) suggested that adding mass (Taki and Mikkelsen 1997). Food limitation mass is easier than increasing skeletal growth, so may be exacerbated during winter, when low un- if rapidly attaining a large size confers advantages der-ice temperatures and light intensities limit in this population, then diverting lipid stores for phytoplankton and zooplankton production (Stein- growth in length during winter could be bene®cial, hart 1997; Steinhart and Wurtsbaugh 1999). The at least until lipids were nearly exhausted. In sum- combination of low zooplankton density and low mer, when light and zooplankton are not as limiting light intensities during periods of ice cover may as during ice-covered periods, the larger kokanee have restricted O. nerka foraging ability to such then would be well prepared to add mass or lipid an extent that they were unable to meet the daily stores. maintenance ration on some days (Steinhart and Wurtsbaugh 1999). During ice-covered periods in Implications for Snake River Stanley Lake, kokanee grew more and predicted Sockeye Management consumption was higher when zooplankton were In both study years, kokanee had dif®culty meet- most abundant. ing their daily maintenance ration due to low zoo- Therefore, winter conditions should be consid- plankton density and reduced light intensity during ered in recovery strategies, especially because the periods with ice cover (Steinhart and Wurtsbaugh long seaward migration of Snake River sockeye 1999). As a result, kokanee reached lipid levels salmon necessitates that they leave the nursery approaching lethal limits. This result could have lakes physiologically prepared for migration to the important implications for endangered Snake Riv- ocean. For example, Chinook salmon smolts in the WINTER ECOLOGY OF KOKANEE 1085

Columbia River depleted their available lipid re- on length and weight can be misleading even if serves after traveling only 370 km (Rondorf et al. population-speci®c coef®cients are used in the cal- 1985). In other salmon populations, survival dur- culations (Cone 1989). Unless a researcher is ex- ing migration and after arrival at sea has been re- plicitly interested in changes in weight or length lated to the size or lipid content of the smolts (Bil- (e.g., for the purpose of calculating swimming ve- ton et al. 1982; Holtby et al. 1990; Henderson and locity or the time a ®sh may outgrow predators), Cass 1991). Rondorf et al. (1985) found that a 1% then the goal is usually to reveal changes in ®sh increase in starting lipid stores in Chinook salmon energetic content. Length and weight measure- smolts resulted in nearly an 8% increase in sur- ments alone will not consistently reveal whether vival. ®sh are gaining or losing calories. We suggest that Many sockeye salmon nursery lakes are con- the most appropriate way to measure ®sh growth stantly changing due to environmental and an- is measuring energetic content directly by proxi- thropogenic effects. Because seaward migrations mate or calorimetric analysis (Busacker et al. can be stressful and energetically expensive, 1990) with corrections for size biases (Sutton et changes in winter conditions that in¯uence growth al. 2000). Because these processes are often time may result in smolts experiencing different sur- consuming and expensive, regressions of percent vival rates after migration. For example, Red®sh lipid or caloric density to percent water, such as Lake in the Sawtooth Valley was stocked with those presented here and in Thompson et al. sockeye salmon fry from a broodstock program in (1991), can provide an inexpensive and rapid way 1994. Assuming that stocking increased O. nerka to estimate lipid content or caloric density. Alter- densities, competition for zooplankton may have natively, some researchers have successfully es- increased, especially during winter when foraging timated lipid content with total body electrical is already limited. If increased competition exac- conductivity estimates (Gillooly and Baylis 1999), erbates energetic losses during winter, sockeye nonlethal tissue samples (Hendry and Berg 1999), salmon may be smaller and have fewer lipids, thus reducing migration survival. Alternatively, some RNA±DNA ratios (Bulow et al. 1981), or multiple individuals may delay migration for a year or fore- morphometric parameters (Adams et al. 1995). go migration and residualize. To improve rearing These latter methods are especially useful, as they conditions in Red®sh Lake, the lake was fertilized are nondestructive methods that can be repeatedly to increasing zooplankton abundance and biomass used on the same ®sh through time. (Taki and Mikkelsen 1997; Wurtsbaugh et al. 1997). Our results suggest that monitoring of smolt Acknowledgments size, lipid content, and additional research on the We would like to thank all those people who metabolic demands of the seaward migration helped collect the data necessary for this project. would provide additional insights into the resto- Odette Brandt, Michelle Gregg, Chad Mellison, ration efforts in the Sawtooth Valley, and that it Fredrick Norrsell, Rick Orme, John Ossowski, may be prudent to consider winter conditions in Mike Slater, and Darek Staab were all generous other salmonid systems. enough to help on many cold days and nights. We Growth Metrics would also like to thank Dave Beauchamp, Phae- Different measures of kokanee growth in Stan- dra Budy, Howard Gross, and Chris Luecke for ley Lake highlight the ambiguity of growth mea- sharing their data and discussing issues facing the surements, reinforcing the need for proper choice recovery of the Snake River sockeye salmon. Spe- of a growth metric for ®shes. Changes in kokanee cial thanks to Doug Taki, David Teuscher, and Paul lengths and weights did not re¯ect changes in con- Kline for help in the ®eld, sharing of data, and dition factor, lipid content, or estimated energetic discussions of salmon ecology. Roy Stein, Grace content. Wet weight may provide misleading Kilbane, HaÊkkan Turesson, Melo Maiolie, Patrick growth estimates because ®sh can replace tissue Martinez, and Bruce Shepard all provided thought- with water and experience no loss of weight, even ful comments that improved the quality of this though energy content decreases (Brett et al. 1969; manuscript. Funding for this project came from Busacker et al. 1990). In lieu of measuring changes Utah State University, the Shoshonne-Bannock In- in length or weight, many researchers have turned dian Tribes (through Bonneville Power Adminis- to one of several measures for condition (Anderson tration project 91±71), and the Ohio State Uni- and Neumann 1996). But condition factors based versity. 1086 STEINHART AND WURTSBAUGH

References tion? Canadian Journal of Fisheries and Aquatic Sciences 45:443±452. Adams, C. E., F. A. Huntingford, and M. Jobling. 1995. Dannewitz, J., and E. Petersson. 2001. Association be- A non-destructive morphometric technique for the tween growth, body condition and anti-predator be- estimation of body and mesenteric lipid in Artic haviour in maturing and immature brown trout parr. charr: a case study of its application. Journal of Fish Journal of Fish Biology 59:1081±1091. Biology 47:82±90. Duston, J., and R. L. Saunders. 1999. Effect of winter Anderson, R. O., and R. M. Neumann. 1996. Length, food deprivation on growth and sexual maturity of weight, and associated structural indices. Pages Atlantic salmon (Salmo salar) in seawater. Canadian 447±482 in B. R. Murphy and D. W. Willis, editors. Journal of Fisheries and Aquatic Sciences 56:201± Fisheries techniques. American Fisheries Society, 207. Bethesda, Maryland. Eggers, D. M. 1978. Limnetic feeding behavior of ju- Beauchamp, D. A., D. J. Stewart, and G. L. Thomas. venile sockeye salmon in Lake Washington and 1989. Corroboration of a bioenergetics model for predator avoidance. Limnology and Oceanography sockeye salmon. Transactions of the American Fish- 23:1114±1125. eries Society 118:597±607. Foote, C. J., I. Mayer, C. C. Wood, W. C. Clarke, and Berglund, I. 1995. Effects of size and spring growth on J. Blackburn. 1994. On the developmental pathway sexual maturation in 1ϩ Atlantic salmon (Salmo to nonanadromy in sockeye salmon, Oncorhynchus salar) male parr: interactions with smolti®cation. nerka. Canadian Journal of Zoology 72:397±405. Canadian Journal of Fisheries and Aquatic Sciences Foote, C. J., C. C. Wood, W. C. Clarke, and J. Blackburn. 52:2682±2694. 1992. Circannual cycle of seawater adaptability in Bilton, H. T., D. F. Alderdice, and J. T. Schnute. 1982. Oncorhynchus nerka: genetic differences between In¯uence of time and size at release of juvenile coho sympatric sockeye salmon and kokanee. Canadian salmon (Oncorhynchus kisutch) on returns at ma- Journal of Fisheries and Aquatic Sciences 49:99± turity. Canadian Journal of Fisheries and Aquatic 109. Sciences 39:426±447. Foote, C. J., C. C. Wood, and R. E. Withler. 1989. Bio- Bjornn, T. C., D. R. Craddock, and D. R. Corley. 1968. chemical genetic comparison of sockeye salmon and Migration and survival of Red®sh Lake, Idaho, kokanee, the anadromous and nonanadromous sockeye salmon, Oncorhynchus nerka. Transactions forms of Oncorhynchus nerka. Canadian Journal of of the American Fisheries Society 97:360±373. Fisheries and Aquatic Sciences 46:149±158. Bligh, E. G., and W. J. Dyer. 1959. A rapid method of Gillooly, J. F., and J. R. Baylis. 1999. Reproductive total lipid extraction and puri®cation. Journal of success and the energetic cost of parental care in Biochemical Physiology 37:911±917. male smallmouth bass. Journal of Fish Biology 54: Brett, J. R., J. E. Shelbourn, and C. T. Shoop. 1969. 573±584. Growth rate and body composition of ®ngerling Gross, H. P., W. A. Wurtsbaugh, and C. Luecke. 1998. sockeye salmon, Oncorhynchus nerka, in relation to The role of anadromous sockeye salmon in the nu- temperature and ration size. Journal of the Fisheries trient loading and productivity of Red®sh Lake, Ida- Research Board of 26:2363±2394. ho. Transactions of the American Fisheries Society Budy, P., G. P. Thiede, N. Bouwes, C. E. Petrosky, and 127:1±18. H. Schaller. 2002. Evidence linking delayed mor- Henderson, M. A., and A. J. Cass. 1991. Effect of smolt tality of Snake River sockeye salmon to their earlier size on smolt-to-adult survival for Chilko Lake hydrosystem experience. North American Journal sockeye salmon (Oncorhynchus nerka). Canadian of Fisheries Management 22:35±51. Journal of Fisheries and Aquatic Sciences 48:988± Bulow, F. J., M. E. Zeman, R. J. Winningham, and W. 994. F. Hudson. 1981. Seasonal variations in RNA-DNA Henderson, P. A., R. H. A. Holmes, and R. N. Bamber. ratios and in indicators of feeding, reproduction, 1988. Size-selective overwintering mortality in the energy storage, and condition factor in a population sand smelt, Atherina boyeri Risso, and its role in of bluegill, Lepomis macrochirus Ra®nesque. Jour- population regulation. Journal of Fish Biology 33: nal of Fish Biology 18:237±244. 221±233. Burgner, R. L. 1992. Life history of sockeye salmon Hendry, A. P., and O. K. Berg. 1999. Secondary sexual (Oncorhynchus nerka). Pages 1±117 in C. Groot and characters, energy use, senescence, and the cost of L. Margolis, editors. Paci®c salmon life histories. reproduction in sockeye salmon. Canadian Journal University of Press, Vancouver. of Zoology 77:1663±1675. Busacker, G. P., I. R. Adelman, and E. M. Goolish. 1990. Hewett, S. W., and B. L. Johnson. 1992. A generalized Growth. Pages 363±387 in C. B. Shreck and P. B. bioenergetics model of ®sh growth for microcom- Moyle, editors. Methods for ®sh biology. American puters. University of Wisconsin Sea Grant Institute, Fisheries Society, Bethesda, Maryland. UW Sea Grant Technical Report WIS-SG-92-250, Cone, R. S. 1989. The need to reconsider the use of Madison. conidition indices in ®sheries science. Transactions Hoar, W. S. 1976. Smolt transformation: evolution, be- of the American Fisheries Society 118:510±514. havior, and physiology. Journal of the Fisheries Re- Cunjak, R. A. 1988. Physiological consequences of search Board of Canada 33:1234±1252. overwintering in streams: the cost of acclimatiza- Holm-Hansen, O., and B. Riemann. 1978. Chlorophyll WINTER ECOLOGY OF KOKANEE 1087

a determination: improvements in methodology. Oi- of Fisheries and Aquatic Sciences 55(Supplement kos 30:438±447. 1):93±103. Holtby, L. B., B. C. Andersen, and R. K. Kadowaki. Metcalfe, N. B., and J. E. Thorpe. 1992. Anorexia and 1990. Importance of smolt size and early ocean defended energy levels in over-wintering juvenile growth to interannual variability in marine survival salmon. Journal of Ecology 61:175±181. of coho salmon (Oncorhynchus kisutch). Canadian Miranda, L. E., and W. D. Hubbard. 1994. Length-de- Journal of Fisheries and Aquatic Sciences 47:2128± pendent winter survival and lipid composition of 2194. age-0 largemouth bass in Bay Springs Reservoir, Hubert, W. A. 1996. Passive capture techniques. Pages Mississippi. Transactions of the American Fisheries 157±192 in B. R. Murphy and D. W. Willis, editors. Society 123:80±87. Fisheries techniques. American Fisheries Society, Morgan, I. J., I. D. McCarthy, and N. B. Metcalfe. 2000. Bethesda, Maryland. Life-history strategies and protein metabolism in Hyatt, K. D., and J. G. Stockner. 1985. Responses of overwintering juvenile Atlantic salmon: growth is sockeye salmon (Oncorhynchus nerka) to fertiliza- enhanced in early migrants through lower protein tion of British Columbia coastal lakes. Canadian turnover. Journal of Fish Biology 637±647. Journal of Fisheries and Aquatic Sciences 42:320± Nordeng, H. 1983. Solution to the ``char problem'' 331. based on Arctic char (Salvelinus alpinus) in Norway. Jobling, M., and I. Miglavs. 1993. The size of lipid Canadian Journal of Fisheries and Aquatic Sciences depotsÐa factor contributing to the control of food 40:1372±1387. intake in Arctic charr, Salvelinus alpinus? Journal Oliver, J. D., and G. F. Holeton. 1979. Overwinter mor- of Fish Biology 43:487±489. tality of ®ngerling smallmouth bass in relation to Jonsson, N., and B. Jonsson. 1998. Body composition size, relative energy stores, and environmental tem- and energy allocation in life-history stages of brown perature. Transactions of the American Fisheries trout. Journal of Fish Biology 53:1306±1316. Society 108:130±136. Kadri, S., D. F. Mitchell, N. B. Metcalfe, F. A. Hunt- Post, J. R., and D. O. Evans. 1989. Size-dependent over- ingford, and J. E. Thorpe. 1996. Differential pat- winter mortality of young-of-the-year yellow perch terns of feeding and resource accumulation in ma- (Perca ¯avescens): laboratory, in situ enclosure, and turing and immature Atlantic salmon, Salmo salar. ®eld experiments. Canadian Journal of Fisheries Aquaculture 142:245±257. and Aquatic Sciences 46:1958±1968. Kerr, D. C., C. D. Ankney, and J. S. Millar. 1982. The Quinn, T. P., E. Graynoth, C. C. Wood, and C. J. Foote. 1998. Genotypic and phenotypic divergence of effect of drying temperature on extraction of petro- sockeye salmon in New Zealand from their ancestral leum ether soluble fats of small birds and mammals. British Columbia populations. Transactions of the Canadian Journal of Zoology 60:470±472. American Fisheries Society 127:517±534. Koenings, J. P., J. E. Edmundson, G. B. Kyle, and J. M. Reinhardt, U. G. 2002. Asset protection in juvenile Edmundson. 1987. Limnological ®eld and labora- salmon: how adding biological realism changes a tory manual: methods for assessing aquatic pro- dynamic foraging model. Behavioral Ecology 13: duction. Department of Fish and Game, 94±100. Fisheries Rehabilitation, Enhancement, and Devel- Rice, S. D., R. E. Thomas, and A. Moles. 1994. Phys- opment Division Report 71, Juneau. iological and growth differences in three stocks of Leonard, J. B. K., and S. D. McCormick. 1999. Effects underyearling sockeye salmon (Oncorhynchus ner- of migration distance on whole-body and tissue- ka) on early entry to seawater. Canadian Journal of speci®c energy use in American shad (Alosa sapi- Fisheries and Aquatic Sciences 51:974±980. dissima). Canadian Journal of Fisheries and Aquatic Ricker, W. E. 1938. ``Residual'' and kokanee salmon in Sciences 56:1159±1171. Cultus Lake. Journal of the Fisheries Research Ludsin, S. A., and D. R. DeVries. 1997. First-year re- Board of Canada 4:192±218. cruitment of largemouth bass: the interdependency Rieman, B. E., and D. L. Meyers. 1992. In¯uence of of early life history stages. Ecological Applications ®sh density and relative productivity on growth of 7:1024±1038. kokanee in ten oligotrophic lakes and reservoirs in Maclean, A., and N. B. Metcalfe. 2001. Social status, Idaho. Transactions of the American Fisheries So- access to food, and compensatory growth in juve- ciety 121:178±191. nile Atlantic salmon. Journal of Fish Biology 58: Rikardsen, A. H., and J. M. Elliot. 2000. Variations in 1331±1346. juvenile growth, energy allocation and life-history McCauley, E. 1984. The estimation of the abundance strategies of two populations of Arctic charr in and biomass of zooplankton in samples, Pages 228± North Norway. Journal of Fish Biology 56:328± 265 in J. A. Downing and F. Rigler, editors. A man- 346. ual on methods of secondary productivity in fresh- Rondorf, D. W., M. S. Dutchuk, A. S. Kolok, and M. waters. Blackwell Scienti®c Publications, Oxford, L. Gross. 1985. Bioenergetics of juvenile salmon UK. during the spring outmigration. Bonneville Power Metcalfe, N. B. 1998. The interaction between behavior Administration, Annual Report 1983, Project 82-11, and physiology in determining life history patterns Portland, . in Atlantic salmon (Salmo salar). Canadian Journal Rowe, D. K., J. E. Thorpe, and A. M. Shanks. 1991. 1088 STEINHART AND WURTSBAUGH

Role of fat stores in the maturation of male Atlantic ville Power Administration, Annual Report 1996, salmon (Salmo salar) parr. Canadian Journal of Project 91±71, Portland, Oregon. Fisheries and Aquatic Sciences 48:405±413. Taylor, E. B., C. J. Foote, and C. C. Wood. 1996. Mo- Ruggerone, G. T. 1992. Winter ecology of sockeye salm- lecular genetic evidence for parallel life-history on in the Chignik Lakes. Fisheries Research Insti- evolution within a Paci®c salmon (sockeye salmon tute, University of Washington, Annual Report, and kokanee, Oncorhynchus nerka). Evolution 50: Seattle. 401±416. Shuter, B. J., and J. R. Post. 1990. Climate, population Taylor, E. B., S. Harvey, S. Pollard, and J. Volpe. 1997. viability, and the zoogeography of temperate ®shes. Postglacial genetic differentiation of reproductive Transactions of the American Fisheries Society 119: ecotypes of kokanee (Oncorhynchus nerka) in Okan- 314±336. agan Lake, British Columbia. Molecular Ecology 6: Silverstein, J. T., K. D. Shearer, W. W. Dickhoff, and E. 503±517. M. Plisetskaya. 1999. Regulation of nutrient intake Thompson, J. M., E. P. Bergersen, C. A. Carlson, and and energy balance in salmon. Aquaculture 177:1 L. R. Kaeding. 1991. Role of size, condition, and 61±169. lipid content in the overwinter survival of age-0 Slotte, A. 1999. Effects of ®sh length and condition on Colorado squaw®sh. Transactions of the American spawning migration in Norwegian spring spawning Fisheries Society 120:346±353. herring (Clupea harengus L.). Sarsia 84:111±127. Thorpe, J. E. 1987. Environmental regulation of growth Sokal, R. R., and F. J. Rohlf. 1981. Biometry. Freeman, patterns in juvenile Atlantic salmon. Pages 463±474 . in R. C. Summerfelt and G. E. Hall, editors. Age Steinhart, G. B. 1997. Winter limnology and the ecology and growth of ®sh. Iowa State University Press, Ames. of Oncorhynchus nerka in high-mountain lakes of Toneys, M. L., and D. W. Coble. 1979. Size-related, ®rst Idaho. Master's thesis. Utah State University, Lo- winter mortality of freshwater ®shes. Transactions gan. of the American Fisheries Society 108:415±419. Steinhart, G. B., and W. A. Wurtsbaugh. 1999. Under- Venditti, D. A., D. W. Rondorf, and J. M. Kraut. 2000. ice diel vertical migrations of Oncorhynchus nerka Migratory behavior and forebay delay of radio- and their zooplankton prey. Canadian Journal of tagged juvenile fall chinook salmon in a lower Fisheries and Aquatic Sciences 56(Supplement 1): Snake River impoundment. North American Journal 152±161. of Fisheries Management 20:41±52. Sutton, S. G., T. P. Bult, and R. L. Haedrich. 2000. Re- Wurtsbaugh, W. A., H. P. Gross, C. Luecke, and P.Budy. lationships among fat weight, body weight, water 1997. Nutrient limitation of oligotrophic sockeye weight, and condition factors in wild Atlantic salm- salmon lakes of Idaho (USA). Internationale Ver- on. Transactions of the American Fisheries Society einigung fuÈr theoretische und angewandte Limnol- 129:527±538. ogie Verhandlungen 26:413±419. SvedaÈng, H., and H. WickstroÈm. 1997. Low fat contents Wood, C. C., C. J. Foote, and D. T. Rutherford. 1999. in female silver eels: indications of insuf®cient en- Ecological interactions between juveniles of repro- ergetic stores for migration and gonadal develop- ductively isolated anadromous and non-anadromous ment. Journal of Fish Biology 50:475±486. morphs of sockeye salmon, Oncorhynchus nerka, Taki, D., and A. Mikkelsen. 1997. Snake River sockeye sharing the same nursery lake. Environmental Bi- salmon habitat and limnological research. Bonne- ology of Fishes 54:161±173.