Evidence for competition at sea between Norton Sound chumsalmon and Asian hatchery chum salmon

Gregory T. Ruggerone & Beverly A. Agler &

Jennifer L. Nielsen

Received: 30 July 2010 /Accepted: 12 May 2011# Springer Science+Business Media B.V. 2011

Abstract Increasing production of hatchery salmonover the past four decades has led to concerns aboutpossible density-dependent effects on wild Pacificsalmon populations in the North Pacific Ocean. Theconcern arises because salmon from distant regionsoverlap in the ocean, and wild salmon populationshaving low productivity may compete for food withabundant hatchery populations. We tested the hypoth-esis that adult length-at-age, age-at-maturation, pro-ductivity, and abundance of a Norton Sound, Alaska,chum salmon population were influenced by Asianhatchery chum salmon, which have become excep-tionally abundant and surpassed the abundance ofwild chum salmon in the North Pacific beginning inthe early 1980s. We found that smaller adult length-at-age, delayed age-at-maturation, and reducedproductivity and abundance of the Norton Sound

salmon population were associated with greater pro-duction of Asian hatchery chum salmon since 1965.Modeling of the density-dependent relationship, whilecontrolling for other influential variables, indicated thatan increase in adult hatchery chum salmon abundancefrom 10 million to 80 million adult fish led to a 72%reduction in the abundance of the wild chum salmonpopulation. These findings indicate that competitionwith hatchery chum salmon contributed to the lowproductivity and abundance of Norton Sound chumsalmon, which includes several stocks that are classifiedas Stocks of Concern by the State of Alaska. This studyprovides new evidence indicating that large-scalehatchery production may influence body size, age-at-maturation, productivity and abundance of a distant wildsalmon population.

Keywords Arctic-Yukon-Kuskokwim . Alaska .

Chum salmon . Hatchery versus wild salmon .

Competition . Density-dependence . Tragedy of thecommons

Introduction

Competition among salmon for food in the ocean canlead to reduced growth and survival (Zaporozhets andZaporozhets 2004; Ruggerone and Nielsen 2009).Salmon may compete with local salmon populationsduring early marine life in coastal areas (Peterman1984a; Levin et al. 2001) or during late marine life

Environ Biol FishDOI 10.1007/s10641-011-9856-5

G. T. Ruggerone (*)Natural Resources Consultants, Inc.,4039 21st Avenue West, Suite 404,Seattle, WA 98199, USAe-mail: GRuggerone@nrccorp.com

B. A. AglerAlaska Department of Fish and Game, Division ofCommercial Fisheries, Mark, Tag, and Age Lab,10107 Bentwood Place,Juneau, AK 99801, USA

J. L. NielsenThe Evergreen State College,2700 Evergreen Parkway NW,Olympia, WA 98505, USA

when maturing salmon concentrate along the migratorypathway to their natal river (Rogers and Ruggerone1993). Salmon may compete in the ocean with salmonof the same species (Rogers 1980; Peterman 1984a, b;Kaeriyama 1998; Pyper and Peterman 1999; Helleet al. 2007) and with salmon of different species(Peterman 1982; Ruggerone et al. 2003, 2005;Ruggerone and Nielsen 2004). Competition in offshorewaters often involves populations originating fromdistant regions and even different continents becausesalmon migrate long distances and are broadly distrib-uted at sea (McKinnell 1995; Ruggerone and Nielsen2004, 2009; Myers et al. 2007).

Scientists have raised concerns about increasingabundances of hatchery salmon and their possibledensity-dependent effects on wild salmon (Peterman1991; Beamish et al. 1997; Cooney and Brodeur1998; Hilborn and Eggers 2000; Kaeriyama andEdpalina 2004; Zaporozhets and Zaporozhets 2004).Concerns arise because hatcheries release numerousjuvenile salmon into the ocean each year even thoughocean conditions vary and may not provide sufficientprey to fully support both hatchery and wild salmon.For example, production of adult hatchery chumsalmon from Asia increased rapidly beginning in1970, and hatchery chum salmon began to exceedtotal production of wild adult chum salmon from Asiaand North America in the early 1980s (Kaeriyama etal. 2009; Ruggerone et al. 2010). Unlike sockeye andpink salmon, whose abundance doubled after theocean regime shift in the mid-1970s, the abundance ofwild chum salmon in the North Pacific remainedrelatively stable (Ruggerone et al. 2010). Since 1980,approximately 2.2 billion hatchery chum salmon peryear were released from Asian hatcheries into theNorth Pacific Ocean and adjacent seas compared withapproximately 0.7 billion chum salmon from NorthAmerican hatcheries (Ruggerone et al. 2010). Hatcherychum salmon are much more numerous than otherspecies of hatchery Pacific salmon. The large produc-tion of hatchery chum salmon in Asia was associatedwith a significant reduction in growth of Asian chumsalmon (hatchery and wild) and delayed age-at-maturation (Ishida et al. 1993; Kaeriyama 1998;Kaeriyama et al. 2007; Zavolokin et al. 2009).However, while some Russian scientists (Klovatch2000; Zaporozhets and Zaporozhets 2004) claim thatwild chum salmon in Russia have declined in responseto increasing production of hatchery chum salmon in

Asia, Morita et al. (2006a) noted that there is littleempirical evidence to support this claim.

Asian hatchery chum salmon, most originating fromJapanese hatcheries, are broadly distributed in theBering Sea and North Pacific Ocean and their distribu-tion at sea overlaps with that of wild chum salmonoriginating from the Arctic-Yukon-Kuskokwim (AYK)region of western Alaska (Myers et al. 2007, 2009;Beacham et al. 2009; Urawa et al. 2009). The greatabundance of Asian hatchery chum salmon and theirdistribution overlap with western Alaska chum salmonled Myers et al. (2004) to hypothesize that Asianhatchery chum salmon compete with AYK chumsalmon for prey. Hatchery and wild chum salmon fromNorth America may also compete with AYK chumsalmon, but their overlap at sea and abundance is lesscompared with that of Asian hatchery chum salmon(Myers et al. 2007, 2009; Beacham et al. 2009;Urawa et al. 2009).

Potential competition between Asian hatcherysalmon and AYK chum salmon is a concern tocommunities in this large region because abundancesof AYK chum salmon have been low since themid-1990s or earlier (Krueger and Zimmerman2009). Abundance of some AYK stocks, such asNorton Sound chum salmon in northwestern Alaska,have declined since the late 1980s, leading torestrictions on commercial fisheries (e.g., an 80%decline in commercial harvests after 1988), reducedharvests of salmon for subsistence, and significanthardship for people in the region (AYK SSI 2006;Banducci et al. 2007; Menard et al. 2009; Wolfe andSpaeder 2009). Three chum salmon stock aggregatesin Norton Sound are currently classified as “Stocks ofConcern” by the State of Alaska because harvestshave been consistently low compared with previousharvests (AYK SSI 2006; Menard and Bergstrom2009). Subsistence fishing in the Nome subdistrict ofNorton Sound has been restricted since the mid-1980s. Additionally, chum salmon in the Yukon River(summer and fall runs) and Kuskokwim River wereclassified as “Stocks of Concern” until recently(Brannian et al. 2006). Factors causing the declineof AYK chum salmon are largely unknown and amajor initiative was undertaken in the region toidentify potential factors (AYK SSI 2006; Kruegerand Zimmerman 2009). Stock-recruitment analysesindicated that the declining productivity of AYKchum salmon was synchronous and indicative of a

Environ Biol Fish

region-wide factor of decline that has yet to beidentified (Hilborn et al. 2007).

We examined the hypothesis that large-scale produc-tion of hatchery chum salmon from Asia has influencedthe growth, age-at-maturation, productivity, and abun-dance of chum salmon originating from Norton Sound,Alaska. We also tested for potential effects of density-dependent interactions with abundant Eastern Kam-chatka pink salmon (Radchenko et al. 2007) that mayaffect chum salmon (Tadokoro et al. 1996; Morita et al.2006b; Khrustaleva and Leman 2007) and whetherseasonal sea surface temperature and ocean regimeshifts influenced abundance. Chum salmon in NortonSound are not highly productive (adult returns perspawner is low), and their distribution in the BeringSea and North Pacific Ocean overlaps that of hatcherychum salmon originating from Asia (Myers et al.2009). This investigation addresses the question ofwhether large-scale hatchery production limits theproductivity of a distant wild salmon population.

Methods

Our approach for evaluating the potential effects ofhatchery chum salmon on wild Norton Sound chumsalmon involved regression analysis and three responsevariables: length-at-age, age-at-maturation, and produc-tivity of Norton Sound chum salmon. The primaryexplanatory variable considered in the analyses wasabundance of Asian hatchery salmon. Potential effectsof total chum salmon abundance (hatchery and wild),pink salmon abundance, parent chum spawners, sea-sonal sea surface temperature, air temperature at Nome,ice cover in the Bering Sea, and ocean regime shiftswere also evaluated as a means to explain variability inthe response variables.

Norton Sound chum salmon

The wild chum population that served as a responsevariable in this investigation spawns in the KwiniukRiver, a tributary to Norton Sound in northwesternAlaska (Fig. 1). The Kwiniuk River drains into thenorth side of Norton Sound just east of Moses Point,approximately 160 km east of Nome, Alaska. Kwiniukchum salmon were the major contributor to thecommercial fishery that began in 1962 near MosesPoint. However, significant commercial harvests of

Kwiniuk chum salmon have not occurred since 1988 inspite of achieving sufficient parent spawners (Kent2007; Volk et al. 2009), indicating that the productivityof the population had declined. Subsistence fishingoccurs in the Kwiniuk River and in nearshore marinewaters without significant restrictions to limit harvestsneeded for food (Menard et al. 2009; Wolfe andSpaeder 2009). Tagging studies indicated few Kwiniukchum salmon were captured in adjacent harvest areasin Norton Sound (Gaudet and Schaefer 1982).

The Kwiniuk chum salmon population was selectedfor this investigation because it has the most compre-hensive dataset in Norton Sound and the trends inKwiniuk chum salmon abundance are representative ofother stocks in the region (Hilborn et al. 2007; Menardet al. 2009). Therefore, in this investigation, we usedthe Kwiniuk chum salmon population as a proxy forchum salmon in Norton Sound. The Alaska Depart-ment of Fish and Game (ADF&G) has estimatedage-at-maturation, spawner abundance, and total abun-dance of Kwiniuk chum salmon since 1965. NortonSound chum salmon migrate to sea immediately afteremergence from gravel (age-0.X), typically spend three(age-0.3) or four (age-0.4) winters at sea, and return tospawn as four or five-year old fish, respectively(Ruggerone and Agler 2008). Age-specific adultreturns and spawning abundances were used tocalculate the number of adult chum salmon returningfrom parent spawners (R/S), brood years 1965–2001(T. Hamazaki, S. Kent, ADF&G, pers. comm.). TheR/S data incorporated adult returns from 1965–2007.

Length-at-age of Kwiniuk chum salmon wasobtained from the ADF&G. Approximately 350 age-0.3 and age-0.4 chum salmon were measured eachyear, 1974–2005, except 1984 and 1992 which hadless than 20 measured fish and were excluded fromthe analysis. Prior to 1974, length data were collectedinfrequently. The index of chum salmon growth wasbased on the mean of male and female salmonmaturing at age-0.3 and age-0.4 in each year ofreturn. This mean of mean approach accounted fordifferences in length associated with gender and agewhile providing a single robust index of chum salmongrowth given that large numbers of fish in eachcategory were not measured each year.

Average age of maturing Kwiniuk chum salmonproduced by each brood year was calculated fromage-specific returns to the river. Kwiniuk chumsalmon mature at age-0.2, age-0.3, age-0.4 and

Environ Biol Fish

age-0.5, corresponding to three- to six-year old fish,respectively.

Salmon and environmental data

Estimates of annual abundances of hatchery- andwild-origin salmon in Asia and western Alaska wereobtained from Ruggerone et al. (2010). The largemajority of Asian hatchery chum salmon originatedfrom Japan, but some were from Russia. Negligiblehatchery chum production (<1% of total) occurs inwestern Alaska (e.g., a few years of production inKotzebue). Hatchery and wild stocks from central andSoutheast Alaska were excluded from this analysisbecause the degree of overlap and influence ongrowth was much less than that of Asian and westernAlaska chum salmon (Myers et al. 2007, 2009;Beacham et al. 2009; Urawa et al. 2009). Potentialdensity-dependent interactions at sea were examinedby comparing length-at-age, age-at-maturation, andproductivity (see below) of Kwiniuk chum salmonwith abundances of 1) Asian hatchery chum salmon,and 2) total chum salmon (hatchery and wild)returning to Asian and western Alaska (watershedsdraining to the Bering Sea) since the mid-1960s. Eachbrood year of Kwiniuk chum salmon overlapped withother chum salmon stocks two, three and four yearsafter the parent spawning year of Kwiniuk salmon.

Therefore, mean abundances of adult chum salmonreturning to Asia and western Alaska were calculatedfor these three years for comparison with productivityof Kwiniuk chum salmon (see below).

Eastern Kamchatka pink salmon are highly abun-dant in the Bering Sea, especially during odd-numbered years (Radchenko et al. 2007). Adult pinksalmon returning two years after the chum parentspawning year were compared with productivity ofKwiniuk chum salmon. These pink salmon wouldoverlap with Kwiniuk chum salmon during their firstwinter and second spring in the ocean.

Environmental data were tested as potential varia-bles to explain characteristics of Kwiniuk chumsalmon. Seasonal sea surface temperature (SST) dataand ice cover indices were obtained from http://www.beringclimate.noaa.gov. Additional sea surface tem-perature (SST) data were derived from COADS dataprovided by the US National Center for AtmosphericResearch and the US National Oceanic and Atmo-spheric Administration (Woodruff et al. 1998; http://dss.ucar.edu/datasets/ds540.1/data/msga.form.html).Monthly air temperature at Nome was obtained fromhttp://climate.gi.alaska.edu. Large scale shifts inocean productivity of the Bering Sea and NorthPacific Ocean occurred in 1976/1977, 1989, and1997 (Kruse 1998; Hare and Mantua 2000). Theeffect of these shifts on productivity of Kwiniuk chum

Fig. 1 Map of NortonSound, Alaska, showing thelocation of the KwiniukRiver

Environ Biol Fish

salmon was tested using dummy variables (0, 1) inthe statistical model (see Eq. 1) whereby the yearswithin the shift period were coded as “1” and otheryears were coded as “0”. For example, since chumsalmon fry enter the ocean during the springfollowing parent spawning, the 1976/1977 oceanregime shift in relation to other regime shift periodswas examined by coding the dummy variable as“1” during brood years 1976–1987, 1976–1995, or1976–2001.

Data analysis

We extended the linear form of a Ricker recruitmentcurve (Hilborn and Walters 1992; Peterman et al.1998) to determine whether abundance of hatcherychum salmon (Hi) and other factors explained thevariability in Kwiniuk chum salmon productivity(Loge R/S) during brood years 1965–2001 afteraccounting for density-dependent effects associatedwith parent spawners (Si):

Loge Ri=Si ¼ !� " Sið Þ �<ðHiÞ þ % Eið Þ þ (i; ð1Þ

where Ri is the adult return of progeny produced byparent spawners (Si) during brood year i, Hi ishatchery or total chum salmon abundance, Ei is anenvironmental variable, and εi is the unexplainedresidual or deviation from expected recruitment.Stepwise and multiple regression, estimates of auto-correlation among model residuals, and collinearitybetween independent variables (Variance InflationFactor [VIF]) were used to evaluate whether theindependent variables explained variability in theproductivity of Kwiniuk chum salmon (Kutner et al.2005). Statistical significance of a variable in themodel was determined when both the partial p-value(P) was <0.05 and the Akaike’s Information Criterion(AIC) was reduced by at least three points (Burnhamand Anderson 1998). A maximum VIF of 10 or morewas used to indicate unsatisfactory collinearity amongindependent variables (Kutner et al. 2005). Partialresidual analysis was used to examine the effect ofeach independent variable in a multiple regressionwhile accounting for the effect of other variables inthe model (Larsen and McCleary 1972).

Preliminary analyses indicated that productivity ofKwiniuk chum salmon was related to abundance ofhatchery chum salmon and other factors, as shown in

Eq. 1. Therefore, the effect of hatchery chum salmonon the abundance of Kwiniuk chum salmon wasestimated by solving Eq. 1 for adult returns (R):

R ¼ Se!�" Sð Þ�< Hð Þþ% Eð Þe( ð2Þ

Kwiniuk spawner abundance (S) and other environ-mental variables in this model were set at their meanvalue during the study period, whereas abundance ofhatchery chum salmon (H) was allowed to to vary bythe approximate range in abundance since 1965 (10million to 80 million fish).

Linear regression analysis was used to test whetheradult length-at-age of Kwiniuk chum salmon wascorrelated with chum salmon abundance and environ-mental variables. Generalized least squares (GLS)regression with restricted maximum likelihood esti-mation (R Development Core Team 2010) was usedto evaluate the relationship between average age ofKwiniuk chum salmon and abundance of chumsalmon because preliminary analysis indicated signif-icant autocorrelation among the residuals. The GLSregression model has the same form as the linearmodel (e.g., Eq. 1) except the error values (εi) areassumed to be correlated and are accounted for in themodel.

Results

Kwiniuk chum salmon length

Length-at-age (mean of age-0.3 and age-0.4 male andfemale salmon) of adult Kwiniuk chum salmon wasnegatively correlated with both Loge Asian hatcherychum salmon and Loge total abundance of Asian andwestern Alaska chum salmon that returned to theirnatal stream during the same year, 1974–2005(Fig. 2). Chum salmon abundance (total or hatchery)explained approximately 36% of the variability inlength-at-age of Kwiniuk chum salmon. Autocorrela-tion at lags 1–6 years was non-significant (P>0.05).Adult chum length was negatively correlated withSST during the winter prior to adult return (r=−0.47,P<0.05), an unexpected pattern. However, environ-mental variables, including seasonal SST, Nome airtemperature, and Bering Sea ice index, did notimprove the model that included chum salmonabundance (P>0.05).

Environ Biol Fish

Kwiniuk chum salmon age at maturation

The dominant age of Kwiniuk chum salmon returningfrom brood years 1965–2001 was age-0.3 (59% oftotal) followed by age-0.4 (36%), age-0.2 (3%), andage-0.5 (2%). Average age of chum salmon in thebrood return to Kwiniuk River ranged from 3.9 to 4.8years, and age increased with greater abundance ofAsian hatchery chum salmon (P=0.004). The rela-tionship between average age of chum salmon andabundance of Asian hatchery chum salmon during the37-year period was explained by the following GLSmodel (Fig. 3a):

Chum salmon age yearsð Þ¼ 3:76þ 0:166 Loge Hatchery chum abundanceð Þ:

ð3Þ

This GLS regression incorporated second orderautoregressive (AR2) terms (ϕ1=0.39, ϕ2=−0.39)as a means to account for residual memory whenestimating model parameters. The AIC of this modelwas at least 3.6 points lower than the AR1 and AR3models and the model that assumed no autocorrela-tion. Average age of Kwiniuk chum salmon alsoincreased with greater total abundance of Asian andwestern Alaska chum salmon, based on the same GLSregression approach (Fig. 3b).

Kwiniuk chum salmon productivity

Adult runs of chum salmon to the Kwiniuk Riveraveraged 35 671±22 034 (SD) fish per year, 1965–2007. Adult returns per spawner (R/S) averaged1.8±1.6 fish during brood years 1965–2001. Residualsfrom the Ricker recruitment curve were relatively high

Fig. 2 Mean length-at-ageof Kwiniuk River (NortonSound) chum salmon inrelation to a) Loge abun-dance of Asian hatcherychum salmon, and b) Logetotal abundance of Asianand western Alaska chumsalmon during the sameyear of return, 1974–2005

Environ Biol Fish

during 1965–1980, low from 1981 to 1994, and verylow from 1995 to 2001 (Fig. 4). Since 1965, 37% of thebroods failed to produce sufficient adult returns (catchand escapement) to replace the parent spawningescapement (R/S<1). Since 1980, 57% of the broodsfailed to replace themselves.

Approximately 48% of the variability in theproductivity (Loge R/S) of Norton Sound chumsalmon during brood years 1965–2001 was explainedby the following model (Table 1):

LogeR=S ¼ 2:69

� 0:016 Total chum abundanceð Þ� 0:009 Kamchatka pink salmonð Þ� 0:039 Spawnersð Þ: ð4Þ

This model indicates that productivity decreased withgreater total adult abundance of Asian and westernAlaska chum salmon two to four years after theKwiniuk chum salmon brood year (millions), de-creased with greater adult abundance of EasternKamchatka pink salmon two years after the chumsalmon brood year (millions), and decreased withgreater abundance of parent spawners (1000s). AICvalues decreased by at least 4 points with theinclusion of each new variable into the previousbest-fit model, indicating that all explanatory varia-bles in Eq. 4 were important (Table 1). Autocorrela-tion among residuals at lags of one to six years wasnon-significant (P>0.05). Collinearity among theindependent variables was negligible, as indicated byVariance Inflation Factor (VIF) values of 1.09-1.10.Environmental variables (e.g., seasonal SST, ice cover,

3.6

3.8

4.0

4.2

4.4

4.6

4.8

5.0

2.2 2.5 2.8 3.1 3.4 3.7 4.0 4.3 4.6

AAgg

eeoo

ffKK

wwiinn

iiuukk

cchhuu

mmssaa

llmmoo

nn((yy

eeaarrss

))Loge Number of Asian hatchery chum salmon (millions)

Age = 3.76 + 0.166 (Loge Hatchery chum abundance)

AIC = -7.38, P < 0.001

3.6

3.8

4.0

4.2

4.4

4.6

4.8

5.0

3.5 3.7 3.9 4.1 4.3 4.5 4.7 4.9

AAgg

eeoo

ffKK

wwiinn

iiuukk

cchhuu

mmssaa

llmmoo

nn((yy

eeaarrss

))

Loge Number of total chum salmon (millions)

Age = 3.26 + 0.262 (Loge Total chum abundance)

AIC = -7.92, P < 0.001

a

b

Fig. 3 Average age of adultchum salmon returning tothe Kwiniuk River (NortonSound) from each broodyear (1965–2001) in relationto a) Loge average Asianhatchery chum salmonabundance, and b) Logeaverage total chum salmonabundance two to four yearsafter the Kwiniuk salmonbrood year. The regressionequations reflect a fit to thedata using generalized leastsquares regression withautoregressive (AR2) corre-lation structure (AIC valuesdecreased by at least 3.6over the AR1 model and by5.0 over the simple linearmodel that assumed no au-tocorrelation)

Environ Biol Fish

winter and spring air temperature in Nome) did notimprove the fit of the model.

Asian hatchery chum salmon averaged approxi-mately 61% of total chum salmon production in Asiaand western Alaska during the study period, but thecontribution of hatchery salmon increased to 68%, onaverage, after 1980. Productivity of Kwiniuk chumsalmon was negatively correlated with abundance ofhatchery chum salmon, as shown in the followingmodel (Fig. 5):

LogeR=S ¼ 2:37

� 0:018 Hatchery chum abundanceð Þ� 0:009 Kamchatka pink salmonð Þ� 0:039 Spawnersð Þ:

ð5ÞThis model explained 50% of the variability inproductivity of Kwiniuk chum salmon. Productivity

of Kwiniuk chum salmon declined with greaterabundance of hatchery chum salmon two to fouryears after the Kwiniuk chum salmon brood year,greater abundance of Eastern Kamchatka pink salmontwo years after the chum salmon brood year, andgreater abundance of parent chum salmon spawners inKwiniuk River (Table 1). Autocorrelation amongresiduals at lags of one to six years was non-significant (P>0.05). The AIC values decreased byat least 4 points with the inclusion of each newvariable into the previous model (Table 1). Collinear-ity among the independent variables was negligible,as indicated by Variance Inflation Factor (VIF) valuesof 1.09–1.1. Standardized regression coefficientsindicated that Kwiniuk spawner abundance was themost influential independent variable followed byhatchery chum salmon abundance (Table 1).

Abundance of wild Asian and western Alaskachum salmon did not improve the model (P>0.05),indicating that abundance of hatchery chum salmon

Fig. 4 Time series of a)adult chum salmon recruit-ment to Kwiniuk River,Norton Sound, b) produc-tivity of Kwiniuk Riverchum salmon, c) abundanceof adult Asian hatcherychum salmon four yearsafter the brood year, d)abundance of adult Asianand western Alaska chumsalmon (hatchery & wild)four years after the broodyear, and e) abundance ofpink salmon returning toEastern Kamchatka twoyears after the brood year

Environ Biol Fish

was more important than wild chum salmonabundance in explaining variability in productivityof Kwiniuk chum salmon. Environmental variables didnot improve the model (P>0.05). Most ocean regimeshift variables did not improve the model (P>0.05),but the regime shift period incorporating broodyears 1976–1995 was statistically significant (partial

P=0.018, AIC change: −4.0), suggesting that produc-tivity of the 1976–1995 broods may have beensomewhat more productive after accounting for otherfactors in the model. However, the regime shift variablewas the least influential variable in the model and it wasmoderately collinear with other variables, therefore thiscomplex model was not considered further.

Table 1 Standardized model coefficients, AIC, and partial P-values of multivariate models used to explain the variability inadult chum salmon returns per spawner in the Kwiniuk River,

Norton Sound, Alaska, brood years 1965–2001 (see Eq. 1).Numerical superscript values identify the independent variables(V) in the model

Model variables P-value adj R2 AIC Std model coefficients Partial P-values

V1 V2 V3 V1 V2 V3

Spawners1 0.004 0.19 −18.5 −0.46 0.004

Total chum salmon

Spawners + Total chum2 <0.001 0.40 −28.7 −0.58 −0.48 <0.001 <0.001

Spawners + Total chum + Pink salmon3 <0.001 0.48 −33.0 −0.65 −0.43 −0.31 <0.001 0.002 0.02

Hatchery chum salmon

Spawners + Hatchery chum2 <0.001 0.42 −29.9 −0.56 −0.50 <0.001 <0.001

Spawners + Hatchery chum + Pink salmon3 <0.001 0.50 −34.0 −0.64 −0.45 −0.30 <0.001 0.002 0.02

Fig. 5 Multivariaterelationship showing theeffect on Kwiniuk River(Norton Sound) chumsalmon return per spawner(Loge) of a) average Asianhatchery chum salmonabundance two to four yearsafter the Kwiniuk salmonbrood year, b) abundance ofEastern Kamchatka adultpink salmon abundance twoyears after the Kwiniukchum salmon brood year,and c) spawning escapementof parent chum salmon in theKwiniuk River, 1965–2001.Plots are based on partialresidual analysis (Larsen andMcCleary 1972)

Environ Biol Fish

Kwiniuk chum salmon abundance

The effect of hatchery chum salmon on the abundanceof Kwiniuk chum salmon was examined by solvingfor returns (R) of adult Kwiniuk chum salmon inEq. 5. For this analysis, the mean number of bothparent spawners in Kwiniuk River (24 800 fish) andpink salmon returning to Eastern Kamchatka (37.6million fish) were held constant in the equation, butthe number of Asian hatchery chum salmon wasallowed to vary from 10 million to 80 million adultsalmon, a range that spanned the observed hatcherysalmon production during the study period. Thisanalysis indicated that increasing hatchery chumsalmon from 10 million to 80 million fish wouldcause abundance of Kwiniuk chum salmon to declinefrom 60 900 fish to 17 100 fish, representing a 72%reduction (Fig. 6).

Discussion

Kwiniuk chum salmon, a key population in NortonSound, Alaska, have experienced reduced adultlength-at-age, greater age-at-maturity, lower produc-tivity (Ricker residual), and lower abundance sincethe early 1980s, corresponding with the period ofincreased production of hatchery chum salmon inAsia. Age-at-maturation of Kwiniuk chum salmontended to be delayed in relation to increasingabundance of hatchery chum salmon, potentiallycontributing to the observed lower productivity ofthe wild chum salmon because older salmon have ahigher risk of mortality. Productivity and adult lengthof Kwiniuk chum salmon were inversely correlatedwith abundance of hatchery chum salmon (avg. 59million salmon), which represented approximately68% of total adult chum abundance in Asia and

Fig. 6 The modeled effectof Asian hatchery chumsalmon on abundance ofKwiniuk River chumsalmon, based on Eq. 5 (seetext) and mean values forKwiniuk spawner abun-dance (24 800 fish) andEastern Kamchatka pinksalmon abundance (37.6million fish). The responseof Kwiniuk chum salmon isshown in a) numbers offish, and b) percentagedecline relative to thebaseline of 10 millionhatchery chum salmon.Confidence limits (95%)bounding the meanabundance prediction areshown

Environ Biol Fish

western Alaska since 1980 (Ruggerone et al. 2010).Inclusion of wild chum salmon abundance in Asiaand western Alaska did not improve the statisticalmodel, suggesting that Asian hatchery salmon was theprimary stock correlated with the decline of chumsalmon productivity and abundance in Norton Sound.The relationships involving age-at-maturation, length-at-age and productivity of Kwiniuk chum salmonsince 1965 are consistent with the hypothesis thathighly abundant hatchery salmon compete withdistant wild salmon stocks in the ocean, leading toreduced growth and productivity of the distant wildsalmon stock. This analysis provides evidence that thepreviously documented decline in AYK chum salmon(Hilborn et al. 2007) was associated with hatcheryproduction.

The analyses presented here were based on correla-tions between variables that were not controlled withinan experimental framework. The analyses suggestadverse interactions between hatchery and wild salmonat sea, based on the known overlapping distribution anddiet of the chum salmon stocks at sea, but thecorrelations do not necessarily prove the hypothesis.Nevertheless, there were multiple lines of evidencesuggesting that hatchery salmon influenced key charac-teristics of wild chum salmon. In addition to therelationships involving adult length, age-at-maturationand productivity, we found that length-at-age wasnegatively correlated with SST, rather than positivelycorrelated as expected based on studies involvingsalmon in northern latitudes (Mueter et al. 2002a, b;Ruggerone et al. 2007). This unexpected findingsuggests that density-dependent effects involvingabundance of hatchery chum salmon may have over-whelmed favorable growth conditions associated withwarmer SST. Our results were consistent with otherstudies showing reduced length-at-age and delayedmaturation of Japanese and Russian chum salmonduring the past several decades in response toincreasing abundance of hatchery chum salmon (Ishidaet al. 1993; Kaeriyama 1998; Zavolokin et al. 2009). Incontrast to total abundance of wild sockeye and pinksalmon in the North Pacific Ocean, abundance of wildchum salmon did not increase after the ocean regimeshift in the mid-1970s, possibly because the increasingabundance of hatchery chum salmon in the oceanled to reduced productivity of wild chum salmon(Kaeriyama et al. 2009; Ruggerone et al. 2010).Together, these studies provide consistent support for

the hypothesis that large-scale hatchery production canaffect the growth, age, and productivity of wild salmonin the ocean.

The statistical model (Eq. 5) indicated that anincrease from 10 million to 80 million hatcherychum salmon would lead to a 72% decline in theabundance of Kwiniuk chum salmon, assuming allother factors were held constant. The statisticalmodel explained only 50% of the variability inKwiniuk chum salmon productivity during the37-year period, so other factors were also importantand contributed to variability. Nevertheless, abun-dance of Kwiniuk chum salmon declined 60%, onaverage, from brood years 1965–1979 to 1990–2001,a period in which hatchery chum salmon productionincreased 190% (from 23 million to 67 million fish, onaverage).

The decline in productivity of Kwiniuk chumsalmon is a special concern because the Kwiniukpopulation had relatively low productivity beforelarge scale hatchery releases began in the early1970s, e.g., R/S=1.8. Six wild chum salmon pop-ulations in the AYK region of western Alaska, whichtypically inhabit relatively pristine habitats, have beendepressed during the past 15 or more years, leading to“Stock of Concern” designations by the State ofAlaska (Brannian et al. 2006). Our findings provideevidence that increasing production of Asian hatcherychum salmon may have contributed to the decline ofthese chum salmon stocks, whose distribution at seaoverlaps that of Kwiniuk chum salmon (Myers et al.2007, 2009; Urawa et al. 2009).

Our findings indicated that productivity ofKwiniuk chum salmon also declined in responseto the abundance of Eastern Kamchatka pinksalmon, which were exceptionally abundant in theBering Sea during odd-numbered years. These pinksalmon were wild fish, but production of hatcherypink salmon was also increasing in Asia and NorthAmerica (Ruggerone et al. 2010). The influence ofpink salmon on Kwiniuk chum salmon was less thanthat of hatchery chum salmon, as expected. Thisfinding is consistent with other studies that havereported negative relationships between highly abun-dant pink salmon versus chum salmon (Tadokoroet al. 1996; Morita et al. 2006b; Khrustaleva andLeman 2007), other species of salmon (Bugaev et al.2001; Ruggerone and Nielsen 2004), and marinebirds in the Bering Sea (Toge et al. 2011).

Environ Biol Fish

In recent decades, scientists have raised con-cern about the increasing abundance of hatcherysalmon and density-dependent effects on wildsalmon populations (Peterman 1991; Cooney andBrodeur 1998; Myers et al. 2004; Mantua et al.2009). The concern becomes more critical when theunderlying productivity of the wild population islow, as in populations in the AYK region of Alaskaand in the Pacific Northwest (Good et al. 2005;AYK SSI 2006; Krueger and Zimmerman 2009).Salmon stocks originating from distant regions andadjacent continents overlap in the North PacificOcean and Bering Sea, and they share a commonfood resource (Myers et al. 2004, 2009). Forexample, genetic data show numerous Japaneseand Russian origin chum salmon overlap withwestern Alaska chum salmon in both the BeringSea and in the Gulf of Alaska (Beacham et al. 2009;Urawa et al. 2009). The growing evidence forcompetition for food among conspecific salmonand between species of salmon has led somescientists to suggest the need for internationaldialog among organizations that produce numeroushatchery salmon so that the productivity of wildsalmon can be preserved (deReynier 1998; Holtet al. 2008). Our findings represent another exampleof the “tragedy of the commons” (Hardin 1968) inthat production of hatchery salmon has unintendedeffects on salmon and people in distant regions.

Conclusions

Smaller adult length-at-age, delayed age-at-maturation,and reduced productivity and abundance of Kwiniukchum salmon in Norton Sound, Alaska, were associatedwith greater production of Asian hatchery chum salmon,which have been exceptionally abundant since the early1980s. These findings, together with other observationsof density-dependence involving Asian hatchery andwild chum salmon, provide multiple lines of evidencesupporting the hypothesis that large-scale hatcheryproduction may adversely affect growth and productiv-ity of distant wild salmon populations.

Acknowledgements The manuscript was prepared in partunder award NA16FP2993 from the National Oceanic andAtmospheric Administration, U.S. Department of Commerce,administered by the Alaska Department of Fish and Game. This

investigation was funded by the Arctic-Yukon-KuskokwimSustainable Salmon Initiative. We thank ADF&G biologists,especially H. Hamazaki, S. Kent, J. Menard and G. Todd, whoprovided access to reports and data. We also appreciate effortsto gather, measure, and catalog scales by D. Folletti, M.Lovejoy, W. Rosky, and W. Whelan. Constructive comments onthe manuscript were provided by E. Volk, M. Webster, and twoanonymous reviewers. Use of any trade names or products isfor descriptive purposes only and does not imply endorsementof the U.S. Government. The statements, findings, conclusions,and recommendations are those of the author(s) and do notnecessarily reflect the views of the National Oceanic andAtmospheric Administration, the U.S. Department of Commerce,or the Alaska Department of Fish and Game.

References

Arctic-Yukon-Kuskokwim Sustainable Salmon Initiative (AYKSSI) (2006) Arctic-Yukon-Kuskokwim Salmon Researchand Restoration Plan. Bering Sea Fishermen’s Association,110 West 15th, Anchorage, AK. Available: www.aykssi.org/Home.htm (June 2010).

Banducci A, Kohler T, Soong J, Menard J (2007) 2005 annualmanagement report Norton Sound, Port Clarence, andKotzebue. Alaska Department of Fish and Game, FisheryManagement Report No. 07–32, Anchorage, AK.

Beacham TD, Candy JR, Sato S, Urawa S, Le KD, Wetklo M(2009) Stock origins of chum salmon (Oncorhynchus keta)in the Gulf of Alaska during winter as estimated withmicrosatellites. North Pacific Anadromous Fish Commis-sion Bulletin 5:15–23. Available: www.npafc.org/new/pub_bulletin.html (June 2010).

Beamish RJ, Mahnken C, Neville CM (1997) Hatchery andwild production of Pacific salmon in relation to large-scale, natural shifts in the productivity of the marineenvironment. ICES J Mar Sci 54:1200–1215

Brannian LK, Evenson MJ, Hilsinger JR (2006) Escapement goalrecommendations for select Arctic-Yukon-Kuskokwimregion salmon stocks, 2007. Alaska Department of Fish andGame, Fishery Manuscript No. 06–07, Anchorage, AK (with2-15-2007 revision). Available: www.sf.adfg.state.ak.us/FedAidPDFs/fm06-07.pdf (June 2010).

Bugaev VF, Welch DW, Selifonov MM, Grachev LE, EvesonJP (2001) Influence of the marine abundance of pink(Oncorhynchus gorbuscha) and sockeye salmon (O.nerka) on growth of Ozernaya River sockeye. FishOceanogr 10:26–32

Burnham KP, Anderson DR (1998) Model selection andinference: a practical information–theoretic approach.Springer, New York, 353

Cooney RT, Brodeur RD (1998) Carrying capacity and NorthPacific salmon production: stock enhancement implica-tions. Bull Mar Sci 62:443–464

deReynier YL (1998) Evolving principles of internationalfisheries law and the North Pacific Anadromous FishCommission. Ocean Dev Int Law 29:147–178

Gaudet DM, Schaefer G (1982) Migrations of salmon in NortonSound, Alaska determined by tagging in 1978–1979.

Environ Biol Fish

Alaska Department of Fish and Game InformationalLeaflet no. 198. Anchorage, AK.

Good TP, Waples RS, Adams P (eds.) (2005) Updated status offederally listed ESUs of West Coast salmon and steelhead.U.S. Dept. Commerce NOAA Technical Memorandum.NMFS-NWFSC-66, 598 p. Available: www.nwr.noaa.gov/Publications/Biological-Status-Reviews/Salmon.cfm(June 2010).

Hardin G (1968) The tragedy of the commons. Science162:1243–1248

Hare SR, Mantua NJ (2000) Empirical evidence for NorthPacific regime shifts in 1977 and 1989. Prog Oceanogr47:103–146

Helle JH, Martinson EC, Eggers DM, Gritsenko O (2007)Influence of salmon abundance and ocean conditions onbody size of Pacific salmon. North Pacific AnadromousFish Commission Bulletin 4:289–298. Available: www.npafc.org/new/pub_bulletin.html (September 2009).

Hilborn R, Eggers D (2000) A review of the hatchery programsfor pink salmon in Prince William Sound and KodiakIsland, Alaska. Trans Am Fish Soc 129:333–350

Hilborn R, Walters CJ (1992) Quantitative fisheries stockassessment: choice, dynamics and uncertainty. Chapmanand Hall, New York, p 570

Hilborn R, Lessard R, Goodman D, Bue B, Adkison M,Stanford J, Whited D, Knudsen E (2007) Alternativemethods for setting escapement goals in AYK. Preparedfor the Arctic Yukon Kuskokwim Sustainable SalmonInitiative, Anchorage, AK. Available: www.aykssi.org/docs/Project_Docs/Final_Reports/120.pdf (January 2011).

Holt CA, Rutherford MB, Peterman RM (2008) Internationalcooperation among nation-states of the North PacificOcean on the problem of competition among salmon fora common pool of prey resources. Mar Pol 32:607–617

Ishida Y, Ito S, Kaeriyama M, McKinnell S, Nagasawa K(1993) Resent changes in age and size of chum salmon(Oncorhynchus keta) in the North Pacific Ocean andpossible causes. Can J Fish Aquat Sci 50:290–295

Kaeriyama M (1998) Dynamics of chum salmon, Oncorhyn-chus keta, populations released from Hokkaido, Japan.North Pacific Anadromous Fish Commission Bulletin1:90–102. Available: www.npafc.org/new/pub_bulletin.html (September 2009).

Kaeriyama M, Edpalina RR (2004) Evaluation of the biologicalinteraction between wild and hatchery population forsustainable fisheries management of Pacific Salmon. In:Leber KM, Kitada S, Blankenship HL, Svasand T (eds)Stock enhancement and sea ranching: developments,pitfalls and opportunities. Blackwell Publishing Ltd.,Oxford, pp 247–259

Kaeriyama M, Yatsu A, Noto M, Saitoh S (2007) Spatial andtemporal changes in the growth patterns and survival ofHokkaido chum salmon populations in 1970–2001. NorthPacific Anadromous Fish Commission Bulletin 4:251–256. Available: www.npafc.org/new/pub_bulletin.html.(September 2009).

Kaeriyama M, Seo H, Kudo H (2009) Trends in the run sizeand carrying capacity of Pacific salmon in the NorthPacific Ocean. North Pacific Anadromous Fish Commis-sion Bulletin 5:293–302. Available: www.npafc.org/new/pub_bulletin.html (September 2009).

Kent S (2007) Salmonid escapements at Kwiniuk, Niuklukand Nome Rivers, 2006. Alaska Department of Fishand Game, Fishery Data Series No. 07–09, Anchorage,AK.

Khrustaleva AM, Leman VN (2007) Interannual variation oflinear-weight indices, age structure, and rate of growth ofchum salmon Oncorhynchus keta and factors affecting it(the Bol’shaya River basin, western Kamchatka). JIchthyol 47:385–393

Klovatch N (2000) Tissue degeneration in chum salmon andcarrying capacity of the North Pacific Ocean. NorthPacific Anadromous Fish Commission Bulletin 2:83–88

Krueger CC, Zimmerman CE (2009) Pacific salmon: ecologyand management of western Alaska’s populations.American Fisheries Society, Symposium 70, Bethesda,Maryland.

Kruse GH (1998) Salmon run failures in 1997–1998: a link toanomalous ocean conditions? Alaska Fishery ResearchBulletin 5:55–63

Kutner MH, Nachtsheim CJ, Neter J, Li W (2005) AppliedLinear Statistical Models, 5th edn. McGraw-Hill Irwin,New York, p 1396

Larsen WA, McCleary SJ (1972) The use of partial residualplots in regression analysis. Technometrics 14:781–790

Levin PS, Zabel RW, Williams JG (2001) The road toextinction is paved with good intentions: negative associ-ation of fish hatcheries with threatened salmon. Proc RSoc Lond B 268:1153–1159

Mantua NJ, Taylor NG, Ruggerone GT, Myers KW, PreikshotD, Augerot X, Davis ND, Dorner B, Hilborn R, PetermanRM, Rand P, Schindler D, Stanford J, Walker RV, WaltersCJ (2009) The salmon MALBEC project: a North Pacific-scale study to support salmon conservation planning.North Pacific Anadromous Fish Commission Bulletin5:333–354. Available: www.npafc.org/new/pub_bulletin.html (September 2009).

McKinnell S (1995) Age-specific effects of sockeye abundanceon adult body size of selected British Columbia sockeyestocks. Can J Fish Aquat Sci 52:1050–1063

Menard J, Bergstrom DJ (2009) Norton Sound subdistrict 2(Golovin) and subdistrict 3 (Moses Point) chum salmonstock status and action plan, 2010; a report to the AlaskaBoard of Fisheries. Alaska Department of Fish and Game,Special Publication No. 09–19, Anchorage, AK. Avail-able: www.cf.adfg.state.ak.us/region3/nomehome.php(June 2010).

Menard J, Krueger CC, Hilsinger JR (2009) Norton Soundsalmon fisheries: history, stock abundance, and manage-ment. Am Fish Soc Symp 70:621–673

Morita K, Saito T, Miyakoshi Y, Fukuwaka M, Nagasawa T,Kaeriyama M (2006a) A review of Pacific salmonhatchery programmes on Hokkaido Island, Japan. ICES JMar Sci 63:1353–1363

Morita K, Morita SH, Fukuwaka M, Nagasawa T (2006b)Cyclic population dynamics of pink salmon in the BeringSea. 7th International Congress on the Biology of Fish. StJohns, Newfoundland, Canada.

Mueter FJ, Ware DM, Peterman RM (2002a) Spatial correlationpatterns in coastal environmental variables and survivalrates of salmon in the north-east Pacific Ocean. FishOceanogr 11:205–218

Environ Biol Fish

Mueter FJ, Peterman RM, Pyper BJ (2002b) Opposite effects ofocean temperature on survival rates of 120 stocks ofPacific salmon (Oncorhynchus spp.) in northern andsouthern areas. Can J Fish Aquat Sci 59:456–463 plusthe erratum printed in Canadian Journal of Fisheries andAquatic Sciences 60:757.

Myers KW, Walker RV, Davis ND, Armstrong JL (2004)Diet Overlap and competition between Yukon Riverchum salmon and hatchery salmon in the Gulf ofAlaska in summer. Final Report to the Yukon RiverDrainage Fisheries Association. SAFS-UW-0407.School of Aquatic and Fisheries Sciences, Universityof Washington, Seattle.

Myers KW, Klovach NV, Gritsenko OF, Urawa S, RoyerTC (2007) Stock-specific distributions of Asian andNorth American salmon in the open ocean, interan-nual changes, and oceanographic conditions. NorthPacific Anadromous Fish Commission Bulletin 4:159–177. Available: www.npafc.org/new/pub_bulletin.html(September 2009).

Myers KW, Walker RV, Davis ND, Armstrong JL, KaeriyamaM (2009) High seas distribution, biology, and ecology ofArctic-Yukon-Kuskokwim salmon: direct informationfrom high seas tagging experiments, 1954–2006. Am FishSoc Symp 70:201–239

Peterman RM (1982) Nonlinear relation between smolts andadults in Babine Lake sockeye salmon (Oncorhynchusnerka) and implications for other salmon populations. CanJ Fish Aquat Sci 39:904–913

Peterman RM (1984a) Density-dependent growth in earlyocean life of sockeye salmon (Oncorhynchus nerka). CanJ Fish Aquat Sci 41:1825–1829

Peterman RM (1984b) Effects of Gulf of Alaska sockeyesalmon (Oncorhynchus nerka) abundance on survival,body size, growth rate, and age at maturity of BritishColumbia and Bristol Bay, Alaska sockeye populations.Fisheries and Oceans Canada, Canadian Technical Reportof Fisheries and Aquatic Sciences 1302, Ottawa.

Peterman RM (1991) Density-dependent marine processes inNorth Pacific salmonids: Lessons for experimental designof large-scale manipulations of fish stocks. ICES Mar SciSymp 192:69–77

Peterman RM, Pyper BJ, Lapointe MF, Adkison MD, WaltersCJ (1998) Patterns of covariation in survival rates ofBritish Columbian and Alaskan sockeye salmon (Onco-rhynchus nerka) stocks. Can J Fish Aquat Sci 55:2503–2517

Pyper BJ, Peterman RM (1999) Relationship among adult bodylength, abundance, and ocean temperature for BritishColumbia and Alaska sockeye salmon, 1967–1997. CanJ Fish Aquat Sci 56:1716–1720

Radchenko VI, Temnykh OS, Lapko VV (2007) Trends inabundance and biological characteristics of pink salmon(Oncorhynchus gorbuscha) in the North Pacific Ocean.North Pacific Anadromous Fish Commission Bulletin 4:7–21. Available: www.npafc.org/new/pub_bulletin.html(June 2010).

R Development Core Team (2010) R: A language andenvironment for statistical computing. R Foundation forstatistical computing. Vienna, Austria. Available: www.R-project.org (January 2011).

Rogers DE (1980) Density-dependent growth of Bristol Baysockeye salmon. In: McNeil WJ, Himsworth DC (eds)Salmonid ecosystems of the North Pacific. Oregon StateUniversity Press, Corvallis, pp 267–283

Rogers DE, Ruggerone GT (1993) Factors affecting the marinegrowth of Bristol Bay sockeye salmon. Fish Res 18:89–103

Ruggerone GT, Agler BA (2008) Retrospective analysis ofAYK chum and coho salmon. Prepared for the ArcticYukon Kuskokwim Sustainable Salmon Initiative, An-chorage, AK. Available: http://www.aykssi.org/Research/project_profile.cfm?project_id=124 (January 2011).

Ruggerone GT, Nielsen JL (2004) Evidence for competitivedominance of pink salmon (Oncorhynchus gorbuscha)over other salmonids in the North Pacific Ocean. Rev FishBiol Fish 14:371–390

Ruggerone GT, Nielsen JL (2009) A review of growth andsurvival of salmon at sea in response to competition andclimate change. Am Fish Soc Symp 70:241–266

Ruggerone GT, Zimmermann M, Myers KW, Nielsen JL,Rogers DE (2003) Competition between Asian pinksalmon and Alaskan sockeye salmon in the North PacificOcean. Fish Oceanogr 3:209–219

Ruggerone GT, Farley E, Nielsen J, Hagen P (2005) Seasonalmarine growth of Bristol Bay sockeye salmon (Oncorhyn-chus nerka) in relation to competition with Asian pinksalmon (O. gorbuscha) and the 1977 ocean regime shift.Fish Bull 103:355–370

Ruggerone GT, Nielsen JL, Bumgarner J (2007) Linkagesbetween Alaskan sockeye salmon abundance, growth atsea, and climate, 1955–2002. Deep-Sea Res II 54:2776–2793

Ruggerone GT, Peterman RM, Dorner B, Myers KW (2010)Magnitude and trends in abundance of hatchery and wildpink, chum, and sockeye salmon in the North PacificOcean. Marine and Coastal Fisheries: Dynamics, Manage-ment, and Ecosystem Science. In press.

Tadokoro K, Ishida Y, Davis ND, Ueyanagi S, Sugimoto T(1996) Change in chum salmon (Oncorhynchus keta)stomach contents associated with fluctuations of pinksalmon (O. gorbuscha) abundance in the central subarcticPacific and Bering Sea. Fish Oceanogr 5:89–99

Toge K, Yamashita R, Kazama K, Fukuwaka M, Yamamura O,Watanuki Y (2011) The relationship between pink salmonbiomass and the body condition of shorttailed shearwatersin the Bering Sea: can fish compete with seabirds? ProcRoy Soc. doi:10.1098/rspb.2010.2345

Urawa S, Sato S, Crane PA, Agler B, Josephson R, Azumaya T(2009) Stock-specific ocean distribution and migration ofchum salmon in the Bering Sea and North Pacific Ocean.North Pacific Anadromous Fish Commission Bulletin5:131–146. Available: www.npafc.org/new/pub_bulletin.html (June 2010).

Volk EC, Evenson MJ, Clark RH (2009) Escapement goalrecommendations for select Arctic-Yukon-Kuskokwimregion salmon stocks, 2010. Alaska Department of Fishand Game, Fishery Manuscript No. 09–07, Anchorage,AK. Available: www.sf.adfg.state.ak.us/FedAidpdfs/FMS09-07.pdf (June 2010).

Wolfe RJ, Spaeder J (2009) People and salmon of the Yukonand Kuskokwim drainages and Norton Sound in Alaska:fishery harvests, culture change, and local knowledgesystems. Am Fish Soc Symp 70:349–379

Environ Biol Fish

Woodruff SD, Diaz HF, Elms JD, Worley SJ (1998) COADSrelease 2 data and metadata enhancements for improvementsof marine surface flux fields. Phys Chem Earth 23:517–527

Zaporozhets OM, Zaporozhets GV (2004) Interaction betweenhatchery and wild Pacific salmon in the Far East of Russia:a review. Rev Fish Biol Fish 14:305–319

Zavolokin AV, Zavolokina EA, Khokhlov YN (2009)Changes in size and growth of Anadyr chum salmon(Oncorhynchus keta) from 1962–2007. North PacificAnadromous Fish Commission Bulletin 5:157–163.Available: www.npafc.org/new/pub_bulletin.html (June2010).

Environ Biol Fish