Expansion of cultured Pacific salmon into marine ecosystems

:It/ui~cultutu~. 98 ( I99 I ) 173- 183 tlscrrrr

Science Puhlishrrs

173

B.V.. A:llltcrdam

Expansion of cultured Pacific Amon into marine eccsystems

ABXRAClMcNeil,

I”-!

W.J..

1991. Expans%n of cultured

Px~fic

salmon Into mz.rinc csosywms.

.ly~ur~t/r~tr~. 98:

83.

Global harvc:r of salmon has iqcrrascd cle:!di:y sinr-: rhe mid-1970 b. Record harvest Ic\cls arc currcnlly being csiablished largclq r’rom production c1’1:rlccd ana rant led salmon. Farmed salmon :?hnOn arc r&aSCU as Juwnlk% 11110 IIalWdi arc itcld .xpllve throughout their li\.rs. whPrcas r;xhra cccjystems where Ihey mix with wild salmon. Obk-rvation ofa nonlmc?. tiZ~i~,~. ;molr relationship III ranched salmon car! hc mtc:Fr,. l ed 2s cv;$znce that marlnc 1 ,atcrs h;l.# z bi2-r. overs!ockcd. provided the ralio ti,f adults IO smalls declines as Ihc r.Jmbcr of smol!s incrcasrs. Thrcr% rjrl:.,ing program5 Oregon cohc. ;;pan chum. and Alaska pink salmon - arc cxamlnrd hcrc for ~~IICVIICCthat ocean mortalil) from natural causes 1s directI> dcpcndcnt on ~bundanccofsmol~s. lntcrpr ,latlon of thedata dots not s*~ppofl the hypothesis that marine Natcrc have been ovcrstockrd with ha’chey s.~olls In a lcvcl Hhcre survival of salmon has been rrduccd. Howrrer. the poss:bilily ~hdl rndustrial prodl;rti.)n r,f rbnchcd salmon sma!rs ;s reducing. or #nay soon rcducc. suv~val oi‘fishvp rcsourccs should not hc ignorrd.

SALMON

PRC)DUCTION

TREYDS

--:tfit cf salmon (-5t.2 species of OncorhyntAu.~ plus Sta:i:tic: CXIglobal hUi S&no sallr) indicate ;i ;hronology of increase, deciine, and recovery ( Fig. I )- P.n early pericd cf sustained growth, which ended in the late 1930’s. was Followrd by a pl ccipitous decline. Relatively stable production at reduced levels began in the 1950’s and continued into thti 197i)‘s. Rapid recovery began in :he mid-1950’s and continues today. Production lev& currently exceed thosr oT ;hc 1930’s. Peak catches it1 the lath !93C’h WCLCditai:-4 !argely through exploitation of wil;l stocks of Pdc;tic salmon. Wild stocks (Jf Atlantic salm\ -I (S&no .CUIUI- 1 hdJ long since been depleted (Netooy. i 963 i. Recent rapid growth 4:i global ‘larvest has been achieved largely through ciliture - farmine, and ranching. Farming is u.~heres;ilmon are cultured in c>!Jti\ Ity to market size.

I74

tXl?4NSION

(IF C‘lJL >1fRED PAC’IFIC SALMW

INTt’ SIAR!NE E<‘OSYSTEMS

175

is difficult becausefish from both sourcesare captured largely in mixed stock fisheries, and catch statistics typtcally fail to differentiate ranched from wild salmon 1 calculated my estimate of the biomass of ranched salmon harvested from the north Pacific from estimated numbers of juveniles released from culture facilities. assumed fraction harvested, hnd dssumcd avcragc wcig!? (Table 1). Numbers of ranchedjuveniles have doubled per decadein the last folr dec-‘ ades. More than 99% of ranchedjuveniles are releasedinto the ncrth Pacific. Productiort in other oceanic basins presently has little impact on contribution to global harvest, but this couid change in the future. The major producing countries are Japan ( =2.0X JO9juveniles), USA (= 1.5X lo9 juveniles), USSR ( z 1.0~ 10“juveniles) and Canada ( ~0.5~ 1O’jcveniles). Numbers of jllveniles releasedby speciesand reported in Tabie 1 were obtaiqed largely from personal contacts with fishery agencyreprescntytives, unpublished data and recent statistics reported by fishery agencies.These estimctcs jn numbers of Juveniles are preliminaij, but they are believed to be conscr ‘ative. Assumptrrlns on fraction of ranched juveniles harvested and their a~era,; -*nsbetween culture:! and wild stocks. This interest has frcl rsedia qyi; x :z3!9?P!l! sa!r??cn since tky 3re re!esseJ from culture farilttl;fa into natural ecosystemswhere thsy a:‘~ free tc mix with wild tisn. Ivrortalltv in marine waters has received considcrablc attention b: analysts who hav: examined aduit-tosmoh rclatitinships in z,n effort to ottein evidence abc ut thz nature of mortality p:ocesscs.

Three general shapesc!‘curvcj relating the number $ sllrviving adults to number: of smrAts tniciing marine ecosystem‘are s??wy?in Fig. 2. Wire

I ‘h

W.J. MCNEIL

----IiUMEER

* OF JUVENILES

mertrllltv ;s nui&nsit\ dependent. the fraction surviving natural nlul;&tl is ildcp&icnt of the number of smolts, i.e., the adult-to-smolt relationship is lirwr. \Vhcre mortality is directly densi!y dependent, the fraction sur:ivi!!g natural morirliity decreases as the nu;i:ber of smolts increases. Where ;nut I;liitl is irl\~crr:;ely density dependent. the fraction surviving natural mortti!ity increases L the number of smelts increases. ~ontiI;~~cd groivth of salmon ranching increases the likelihood that direct densit!-dcpcndent mortalitv will rec;uce survival of both wild and ranched ktcek< which o\~erlap ill their marine distrihutlons. 1 hree examples are exam;rwd hc:e: coho salmon in Oregon. chum salmon in Japan, and pink salmon in A!~skd.

I hc ctistcrn Pacific salmon fishing area bcrdered by California, Oregon, 2nd Southern V;izshingtqq is referred to as thy “Oregon Produrtion Index*’ 1OPl ) area Estimated “timber5 of cuho salmon caught by sport and commercial fishers and returning to hatchci,ics and sp~;:*niz~ rtrcsm.5 arc added tc cllmir m:i~tics TJ OPI (*at& pl11rescapement. Edirrated numbers of wild plus rrln*:hcd jLvcnilcs incrcascd nearly five-fold from 15 x 10”to 72 x IOh in the ;) :ril d 1C)hO-YZ( Nickelson. 1986 ). Reialionships between number of ju I L,cniltc and rLsLlting adults have been w4uated (Clark end McCarl. 1983: McCarl_ and Rettig. 1983: Peterman and Routledge, 1983: Walker et al., 1983; ,.. . I,,. I a. \.l>,.... . I\, ..&...-‘I.- , \,.;*a..., . . . -... 5, ?83: l’eLerm,m. 1989). Inctinclusivc and

equivocal results have left the cen:ral question about density-dependent mortality large!) Lqresotved Overall marine stirviva of OPI coho typically ranges from 2.5 to 10.0% (Fig. 3 ). OPI coho come from three sources: wild stocks, public hatcheries, and private ha0 <;ies. Smolt-to-adult survival averaged about 8.0% for wild stocks, 5.0% for public hatcheries, and 2.0% for private hatcheries over the period 1960-85. Numbers of smolts produced by public and private hatcheries have increased about eight-fold in this period, whereas; numbers of wild smo:rs have declined fractionally. Combining survival statistics 05:~the three sources d OPT coho. eal;h ot which exhibited a different capacity for survival dur ;ng the pericrj ;~:IQ~~‘LoIIsideration, introduces bias fgvnring the hypothesi i that mortality is directly density dependent. The correlation coefficient for .he combined groups is ~0.05. This weak correlation lends credence to the hypothesis that the relationship between adults and smelts is nonlinear. However, separation of the three sources of OPI coho yields m*lch stronger individual correlations (r=0.53 for wild, r=0.30 for public hatchery, and r=0.97 for private hatchcry fish). Fig. 4 shows camp arisons of adults and smelts for the three groups oi OPI coho. Three linear curves shown in Fig. 4 represent average marine survivsl fog wild ($%). public hatchery (5%). and private natchery (29’0) smelts. More recent years are not included in the above assessment because pub-

10

20 NJMBER

37 OF JIIVENILES

40

5n

63

70

(M1L~lOrrlS)

b’ig. 3. Rclarionship be1wcn numlwr 4’ rcturnlng a111 . ~4: cntcring the ocean from comhrxd WIN. and public ant: priks !‘r(IduCtio-. lndcs ;lrcti. i ihta from Michelson. 19)oh. 1

.,;,,wn to numhcr rll’~u\ cntlcs : hatcher) stocks in the c)rcgon

w 1.MCNEII

178

,“,W ild 4

I

. Publrc It Private

Hatchery Hatchery

Nu,.lLtri

OF AJVtNiCEb

(MILLIONS)

Fig. 4. Rclatlonship betwxn numhcr OI :vturling adult coho salmon IO number of juvcnilcs t’rom WI!& public hatchz-y and prtbatc hatchcry slacks shown srparartl~ in rhc Oxgon Productlm ;I&*\ arcs. ( lhla frcm Nickcfson. 1986. )

lisheddata comparing the three sourcesof coho smolts are limited to the 196082 period. However. unpublished statistics on catch and escapementof private hatchery fish subsequent to 19812show improved ocean survival ap proaching or exceedingthat of public hatchery fish. Analysts who have evamined Japanesehatchery chum salmon for evidence of density-dependentmortality have arrived at different conclusions. Results of an early study (‘faguchi and Abe, 1966) covering the 1932 through 1959 brood years at Hokkaido hatcheries indicated a dome shaped adult-to-smelt relationship. Smelt numbers ranged from 95x lo6 to 417~ 106. Peterman ( 1978, 19: I ) and McCarl and Rettig ( 1983) examined Hokkaido adult-tosmelt relationships for the brood years 1950 through 1969 (smolt numbers ranged from 140x 10’to 550 x I 06) and rejected the hypothesis that mortality was density dependent.However. Lin and Williams ( 1988) using the same data for the 1950 through 1969 brood yearsacceptedthe hypothesisthat mortality was directly dependent on density. More recent unpublished data for the 1965through 1984 orood yearsprovided by the Honshu Salmon Farming Development Association, and including hatchery production from both Honshu and Ho!..kaido Is!ands, supports a third hypothesis that mortality 1s inversely density dependent. Numbers of adults returning from given numbers of smelts releasedare plotted in Fig. 5. Marine survival averagedlessthan 1% for each island when

179

. rbkkaldo H;rtcher!es n Honshu HaIc%~r~er

J. i

l # J

0

l

0

J 0 -0

// /

:.

J JJ

l

NUMBER OF JUVENtLES (MILLIONS)

Fig. 5. Relationship of adult chum salmon to number of hatchery juveniles in Japan for *hc period I96S- 1985. ( UnpubL data from bnshu Salmnr! Devclopmcnt ASSGYation. 1986. )

juvenile production was less than 450~ IC’. Survival increasedseveral fold as number?;of smelts increased to 841)XIfI6 on Honshu and I 100x 1V on Hokkaido. The numbers of juveniles ariii their stir,-ical >o:h i;:Lil;.tJ, wi:h time. The apparent inverse density-dependentsurvival ( Fig. 5 ) might be explained in two ways: ( 1) an artifact related to improved technologv over time; or (2) satiation of predatnrs at higher levels of juvenile production. Pink snlmon in Ahska

Pink salmon ranching was initiated in Prince William Sound, Alaska, in the 1970’sto rehabilitate a depressedfishery economy,Production of ranched juveniles increasedfrom 1Ox I O6in I978 to 532 x IO6in 1988. Total harvest of wild plus ranched pink salmon has increasedseven-fold from 3 x lo6 ( 8year averagebefore hatchery supplementation) to 20~ lo6 or more annually (Fig. 6). Production in excessof 500x IO6hatchery juveniles is e>petted to continue, and future results should be examined for evidenceof density-dependent mortality. The 20-year averageharvest in Prince William Sound pre-

I80

H’l MrNLIL

h--l----T 130

f---l

200

NUMBER OF HATCHECIY

300 JUVENILES

400 RELEASED

500

600

hlILLION9)

t-lg. h. Kt~li~Iitm’.lllp twwccn numtwr ol’w~ld plus ranched aduh pink salmon harvcstcd m Prinrc Wtlllanr Sound. ,AM.J. xnd the numhrr of ranched juvcnilcs rclcascd for the 1977 through I ‘US l~uod yiar\. ( I’npuhl. daln from Masku Dcpartmcnt of Fish and Game, 1990. )

cecding production of hatchery fish was 3.1 million per year. If the recent incrcnscto about 20 million is due to added hatchery fish, more than 80% of the current catch comes from ranching and less than 20% from natural product ion. DIS(‘I’SSIOK

Salmon ranching is WC!!established throughout the north Pacific where nearly 5 x 10”crrlturcd juvcni!.:s, mostly pink and chum salmon, are presently relcascd.Glob;t! production of ranchedjuveniles is expectedto place continucd emphasison r;rnched pink and chu.n salmon in the north PacitTcdue to highly favorable ratios betweenbiomassof adults harvestedand juveniles released( McNeil, I39 I ). Developmentof industrial-scaleranching in the north Atlantic wil! depend largely upon demonstration of economic feasibility with Atlantic salmon where ratios between biomass of adults harvested and juveniles releasedare likely to be far lessfavorablethsn with pink and chinmsalmon in t!le north Pacitic. &cause salmon are not native to the south Pacific, devclopment of ranching dependson successfulacclimation of salmon and subsequent demonstration of economic feasibility. Chinook salmon have been acclimated to New Zealand and introduced to Chile. Chinook ranching is in a developmentalstagein both countries. Many questions rcletzd to successfulexpansion of cultured salmon into

marine ecosystemsremain unresolved, Our ultimate goal is to determine the number (or biomass) of ranLhed juveniles that ran be releasedfrom culture facilities withuut dcpisting wild stocks and/or reducing efficiency of production of ranched and wild salmon. Eventu; lly. we need to develop an in-depth understanding of mortality processes which determine adult-to-smelt relationships. The three ranching examples examined in this report (Oregon coho. Japanesechum, and Alaska pink) collectively contribute nearly,two-thirds of the total global biomass nf ranched salmon harvested at prestnt. Nor.? of these pxamplespresentsclea1’tivtdence of decreasedrate of adult return to fisheries and hatcheries with increased number of ranchedjuveniles. However, even if the true adult-to-smolt relationship departs negatively from linearity, high variability in adult return is likely to obscure t.herelationship. Coho salmon 19rhc OPI area have already been the subject of considerable study. Even though evidence about linearity oi aduit-to-smolt relationships is inconciusive, there is solue information on effects of predators and food availability on survival of coho in the ‘I!Pl acea. Sea birds have been observed to aggregatein the vicinity ofcoastal salmon releasesites and feed on ranched smelts in ltiwer rcdchnsof Oregon estuaries (Bayer. 1986; McNeil et al., 199I ). The common murr: ( L’rio adguc) was thi: principal avian predator observed, but c,ther sueciesof seabirds exhibited similar behavior. An observed high incidence of salmon smolts in murre stomach contents indicated a functional responseof the predator since murres do not t>pica?i target on salmon as a preferred food ( Mathews. I983 ). There was a r!multaneous numerical :esponseas evidenced by increased density of murres when smelts were temporarily concentrated in lower estuaries. There observed changes in murre distribution and feeding behavior are consisttnt with the hypothesis that mortaiity is directly de&y deppndent. Other workers failed to implicate a scarcity of food as a factor contributing to mortality. Fi&r and Pearcy ( 1988! observed similar growth rates of OPI area juvenile coho in years of low and high productivity of nearshorewaters. Food availabihty alsir was not thought to be a factor limiting ranched Prince Niliiam Sound pink salmon based on estimates of annual secondary productivity agd nearshore distribution of pink salmon iuvenile? (Mooney et al.. 1978). Production of ranchedjuvenile salmon has approximately doubied per decade since the 1950’s in the north Pacific. The over41 abundance of ranched plus wild salmon probably exceedshisto..rip high levels, Continued geometric growth of ranchi ng raises a prospect of competition for spaceand food. Will competition be consned largely among salmon stocks or wiil it affect largely nonsaimonid species’?Will the biomass of salmon be self-reguiatmg through density-dependent processesor Gli it increase linearly with increased number of ranched juveniles at the expenseof other fishery resources?Is salmon

H.J McNtlL

I82

ranching a major first step toward managing an oceanic ecosystem for production of selccied crops as has occurred in terrestrial ecosystems? Science has provided a foundation for a technolqy to ranch salmon on an industrial scale. Science now faces the added challenge of determining the implications of this txhnolog) on the future of food production from temperate oceanic basins. \CKNOW’LED
Funding fcr preparation of this report was provided by the Oregon Sea Grant Program which receives financial support from the National Oceanic and Atmoiphcric Administration (Grant No. NA83AA-D-SG108).

KEFEKENCES Ba!c~. K.D., 1986. Scah~rds near an Orrgor, rstuarinc salmon ha!ctlcry in 1981 and during thlr I YK3 ti XI~II). I‘iih.

Buil.. 8-I: 2 IY-?!XJ.

C‘larh J. and McC‘arl. B.. 1983. An invcsti&!ion of the rctationship bclwct’n Oregon coht> salmon ( thrt orl~rrr&rrr AI.\I~~& ) halchcp rclcascs and adult produc;ion utilizing law of minimum rcgrcssiou. C‘an. J. Fish. .?quat. Sci.. 40: 516-523. (‘oonc’!. K.T.. I’rquhan. I’.. Ncvt; t?.. Hilsingcr. J.. c.lasby. R. and Barnard. D., 1978. Some aspects of the carrymg capacity of Prince Wlttiam %und. Alaska. for hatchery relcareti pink. and chum salmon fry. Univ. Alasha Sea Grant Rep.. K 78-3.98 pp. Eldrnr. B.M.. 1989. The Norwegian fisheries induslr): from capture to cultl\afion. World AquaCUlt”rc. 20( 3 ): (IO-68. Fisher. 1.P. and Pcarcy. W.G.. IY8K C’*owrh of juvcnilc coho salmon ( Om,nrl7!,nclrrrs kisrrrzlr ) olTOrcgor: and Washington, USA, In lears ofditlicring coastal upwelling. Can. J. Fish. .\quat. Sci.. 35: 1036-1044. Lln. B.-H. and Willi;rms. N.A.. I U88. Spctiislng a funcrional form for the influcncc of hatcher) snnlt rclcasc on adult salmon production. Fish. Bult.. 86: 655-662. htathcws. t?.K.. 1983. Frcding rcology of the common murre, l’rro xlgc. off the Oregon coast. M.S. Thrsls, Itnlbcr\ity ofOrcgon, Eugene, OR, 108 pp. XIKarl. B. 4. and Rcttig. R.B.. 1983. tnfluencc of hatchen ;molt releases on adult salmon produclion and its variability. (‘an. J. Fish. Aquat. ,cI.. 40: 1X80-I 886. McNeil. V?.J.. I99 I. Fulun of salmon. Fisheries iioengineering Symposium. Ponlallcl. OR. October 1988. .Am. Fish. Sot, Sgmp. IO. Mcf:cil. W-J., Gowan. R. a Id Scverson. R.. 1991, Offshore rctcasc of salmon smelts. Fisher& Clocnginccrlng Symposium, Ponland. OR. October 1988. Am. Fish. Sot. Symp. IO. : ‘lsaka. Y.. 1088. Satmonrd programs and public policy in Japan. In: W.J. McNcll (Editor). Salmon Production. Management. and Allocation. Oregon Slate Univ. Press. Corvattis. OR, p;’ 25-31. Netboy. A.. 1968. The Atlantic Salmon. A Vanishing Species? Houghton Mifftin. Boston. MA. 157 pp. Nlckctson. T.E.. 1984. lqftuencc of upwelting. ocean tcmpcraturc. and smott abundance on surcival of coho salmon (Or?,.l,rlr!.r~c.l~~rrXisrrr:4) m the Oregon Production Area. Can. J. Fish. Aqua1 Sci 43: 527-535.

Pclcrman. H.M.. 1978. Testing for dcnsq dcpcndcnr marlnc sutv~\.al in I’,lcllic salmonld\. J Fish. Res. 8oard (‘an.. 35: 14341450. Pclcrmdn. R.M.. 1981. Form of random varialiun in salmon smelt-to-ad& rclation5 and its iniluwcr: cn production rstimatcs. Can. J. Fish. Aqw. Sri.. 313: 1 i I:I I I 9, Pcwrman. R.M.. 1989. Application of slatistital pows’r analysis IO the 0rcgon tAo xall:wn (Omw~ync~lrw Risrtrt4) probIem. Can. J. Fish. Aqua!. Sci.. 46: 1 18% I 187. Petcrman. R.M. and Routledgc. R.D.. 1983. Exlwimcntal managemsnl ol’Orcgt>ll cob ) salmon (li,tc,c)r~t~,n~,ltlrr ~rslrrr4): dcsigmng for ),icld of informalion. Car,. J. Fish. -lqua!. %-I.. JO: l’l’ ---l”3. “Taguchi. K. and .Abe. S.. 1966. Contribc!ions tcr arllliclal propagation. !n: t’ropagaliur ot’thr i-hum Mlmon II-I Japan. Japan tish. KCSOUrCe cons. .+ssoc. lokyo, pp, 32-55. Walker. K.D., RN&, R.B. and Hilborn. R., 1983. .Ana!!r!s of nwll~plc obJcctlvcs in ( brcgon coho salmon policy. Can. J. Fish. Aqual. Sci.. 4C: 580-587.