extinction of atlantic salmon

7/31/2019 Extinction of Atlantic Salmon

1/9

Fitness Reduction and Potential Extinction of Wild Populations of Atlantic Salmon, Salmosalar, as a Result of Interactions with Escaped Farm SalmonAuthor(s): Philip McGinnity, Paulo Prodhl, Andy Ferguson, Rosaleen Hynes, Niall .Maoilidigh, Natalie Baker, Deirdre Cotter, Brendan O'Hea, Declan Cooke, Ger Rogan, JohnTaggart and Tom CrossReviewed work(s):Source: Proceedings: Biological Sciences, Vol. 270, No. 1532 (Dec. 7, 2003), pp. 2443-2450Published by: The Royal SocietyStable URL: http://www.jstor.org/stable/3592212 .

Accessed: 08/05/2012 03:47

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

The Royal Society is collaborating with JSTOR to digitize, preserve and extend access to Proceedings:

Biological Sciences.

http://www.jstor.org
http://www.jstor.org/action/showPublisher?publisherCode=rslhttp://www.jstor.org/stable/3592212?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3592212?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=rsl


2/9

THE ROYALs SOCIETYReceived 15 May 2003

Accepted 9 July 2003Published online 15 October 2003

Fitness reduction and potential extinction of wildpopulations of Atlantic salmon, Salmo salar, as aresult of interactions with escaped farm salmonPhilip McGinnity', Paulo Prodohl2, Andy Ferguson2*, Rosaleen Hynes2,Niall 0 Maoileidigh', Natalie Baker2, Deirdre Cotter', Brendan O'Hea',Declan Cooke', Ger Roganl, John Taggart3 and Tom Cross4'Aquaculture and Catchment Management Services, Marine Institute, Newport, Co Mayo, Ireland2School of Biology and Biochemistry, Queen's University, Belfast BT7 INN, Northern Ireland3Department of Biological and Molecular Sciences, University of Stirling, Stirling FK9 4LA, UK4Department of Zoology and Animal Ecology, National University of Ireland, Cork, Ireland

The high level of escapes from Atlantic salmon farms, up to two million fishes per year in the NorthAtlantic, has raised concern about the potential impact on wild populations. We report on a two-generation experiment examining the estimated lifetime successes, relative to wild natives, of farm, F1and F2 hybrids and BC1 backcrosses to wild and farm salmon. Offspring of farm and 'hybrids' (i.e. allF1, F2 and BC1 groups) showed reduced survival compared with wild salmon but grew faster as juvenilesand displaced wild parr, which as a group were significantly smaller. Where suitable habitat for theseemigrant parr is absent, this competition would result in reduced wild smolt production. In the experi-mental conditions, where emigrants survived downstream, the relative estimated lifetime success rangedfrom 2% (farm) to 89% (BC1 wild) of that of wild salmon, indicating additive genetic variation for survival.Wild salmon primarily returned to fresh water after one sea winter (1SW) but farm and 'hybrids' producedproportionately more 2SW salmon. However, lower overall survival means that this would result inreduced recruitment despite increased 2SW fecundity. We thus demonstrate that interaction of farm withwild salmon results in lowered fitness, with repeated escapes causing cumulative fitness depression andpotentially an extinction vortex in vulnerable populations.Keywords: escaped farm salmon; common garden experiment; DNA profiling; outbreeding depression;

lifetime success; extinction vortex

1. INTRODUCTIONThe increase of Atlantic salmon culture in the NorthAtlantic to a current level of ca. 700 000 tonnes, togetherwith the vulnerability to damage of marine net cages, hasmeant that large-scale escapes are now frequent occur-rences. Escapes occur both during routine handling andas a result of large-scale accidents, with, for example,600 000 salmon escaping in a single storm incident in theFaroes in 2002 (Atlantic Salmon Federation 2002). It isestimated that some two million salmon escape each yearin the North Atlantic region, which is ca. 50% of the totalprefishery abundance of wild salmon in the area (basedon data in Atlantic Salmon Federation 2002). In Norway,on average, about one-third of adult salmon entering riv-ers are escaped fishes, rising to over 80% in some rivers(Fiske & Lund 1999). On the east coast of North America,escaped farm salmon outnumber wild fishes by as muchas 10 to one in some rivers (Atlantic Salmon Federation2002). Fears have been expressed (Hansen et al. 1991;Hindar et al. 1991; Youngson et al. 1998; McDowell2002) about the potential detrimental genetic and otherchanges that may occur in wild populations as a result ofescaped farm salmon entering rivers and interacting with

*Authorfor correspondence([email protected]).Proc.R. Soc. Lond. B (2003) 270, 2443-2450DOI 10.1098/rspb.2003.2520

wild populations. Because natural populations of Atlanticsalmon are a major resource for angling, tourism and com-mercial exploitation, as well as being an important compo-nent of biodiversity with cultural and aestheticsignificance, these intrusions are of increasing concern,especially as the species is now extinct, or in critical con-dition, in over 27% of rivers, and endangered or vulner-able in a further 30% (WWF 2001).Farm Atlantic salmon are genetically different from wildpopulations as a result of geographical origin (e.g.Norwegian farm strains are widely used in all salmon-farming countries) and founder effects, as well as direc-tional selection, inadvertent selection and genetic driftduring domestication (Skaala et al. 1990; Gjedrem et al.1991; Fleming & Einum 1997; Gjoen & Bentsen 1997;Johnsson et al. 2001; Fleming et al. 2002). Although theirbreeding performance has been shown to be inferior tothat of wild salmon (Fleming et al. 1996, 2000), escapedfarm salmon do breed successfully and hybridize with wildfishes (Lura & Saegrov 1991; Crozier 1993, 2000; Cliffordet al. 1998), thereby potentially changing the geneticmake-up, fitness (i.e. juvenile recruitment in subsequentgenerations) and life-history characteristics (e.g. age andtiming of life-history events) of wild populations. Thus far,discussions on the topic have been largely theoretical orinferential (Hutchings 1991; Youngson et al. 1998). Thefew empirical studies undertaken (McGinnity et al. 1997;

2443 ? 2003 The Royal Society


3/9

2444 P. McGinnity and others Fitness and extinctionof wild Atlantic salmon

Fleming et al. 2000) have involved only F1 hybridsbetween wild and farm salmon. However, it is necessaryto undertake studies for at least two generations, becauseF1 hybrids often show intermediate or even enhanced per-formance compared with their parents (hybrid vigour),but F2 hybrids can show reduced performance(outbreeding depression) (Templeton 1986). BecauseAtlantic salmon have both freshwater juvenile and marinelife-history phases, it is necessary to study both thesephases as well as migrations.The only current direct method of examining quantitat-ive genetic differences among wild, farm and 'hybrid' sal-mon is to carry out 'common garden' experiments, wherefishes are reared from egg to adult in a communal environ-ment. As environmental variability is eliminated, any dif-ferences found in performance will reflect geneticdifferences (with the exception of maternal physiologicaleffects). The development of DNA profiling has enabledaccurate parentage identification and allows direct com-parison of groups from the egg stage onwards under natu-ral conditions (Ferguson et al. 1995). McGinnity et al.(1997) reported the freshwater performance of F1 hybridsbetween wild and farm salmon (in 1993 and 1994cohorts). Here, we extend this previous study by examin-ing the freshwater performance of second-generation F2hybrids and BC1 backcrosses to wild and farm salmon (a1998 cohort), as well as adult return from the sea for allcohorts. The results from all cohorts are combined toallow estimation of two-generation lifetime success.

2. MATERIAL AND METHODSThe experimentwas undertakenin the Burrishoolesystem inwestern Ireland (McGinnity et al. 1997). This system consistsof a freshwater ake (Lough Feeagh), connected to Lough Fur-

nace, a tidal brackish ough, by two outlet channelswith perma-nent smolt and adult trapping facilities ('sea entry traps'), anda number of afferent rivers (figure 1). One of these rivers(Srahrevagh: a. 7250 m2 of juvenilesalmonid habitat)was usedfor the freshwaterstages of the experiment and was equippedwith a furthertrap capable of capturingall downstreamjuvenilemigrants and upstream adults (hereafter referred to as the'experimentriver'and 'experiment trap'). Experimentaldetailsfor the 1998 cohort freshwaterstage were as for the 1993 and1994 cohorts (McGinnity et al. 1997), with the followingchanges and additions for the 1998 cohort groups. Native wildBurrishoole salmon of one sea winter maturity(1SW) and 2SWfarmsalmon (NorwegianMowi origin)were used, whereas3SWand 4SW farm fisheswere used forthe earliercohorts.ReturningF, hybridAtlantic salmon (2SW), which had been ranchedfromthe 1994 cohort, were captured at the Burrishoole traps fromAugust to November 1997 and used to produce the F2 hybridsand BC1backcrosses(table 1). Family tie or full reciprocalmat-ing designs were used (Winkelman & Peterson 1994), givingsome 500 to 800 eggs per family, with all families being estab-lished on the same day. A muscle-tissue specimen from eachparent was retained for DNA profiling. Fertilized eggs wereincubated in the hatchery on the Burrishoole system until thedevelopmental stage when eyes were visible ('eyed eggs'), withcumulative mortality being recorded daily. At this stage, liveeggs were counted accurately(table 1) and families were mixedand planted out in the experiment river in artificialredds con-structed according to Donaghy & Verspoor (2000). Ten eggs

Srahrc\agh Ri\crseper'imentr11'et1'r

LoughlFeeag-h

sea entry ralbps

(brackish

r> (sea)u %- -'

\I

Figure 1. Sketch map of the Burrishoole river systemshowing the location of the experiment river and traps.from each family were retained in the hatcheryin a communaltank, with additional eggs (ca. 300 per family) (but not F2hybrids)being reared in separate group tanks. The communallyrearedhatchery parrwere sampled (n = 523) in March 1999, i.e.at ca. 11 months of age. Juvenileswere sampled (n = 295) fromthe experimentriverby electrofishingin August 1998, and theexperimenttrap was inspected daily from 24 April 1998 to 30June 2001 (parr,n = 169; smolts, n = 177). All 1998 cohort parrand smolts from the experimenttrap were sacrificed.Because insufficient adult returns would have been obtainedfrom the smolts produced in the experiment river, the marinephase of the life cycle of all cohorts (with the exception of F2hybrids) was examined by ranching, i.e. smolts were rearedinthe hatcheryand released to the sea to complete their life cycle.Prior to release, smolts were tagged with coded wire microtags(Wilkinset al. 2001). Communallyrearedsmolts from the 1993and 1994 cohorts were each given a single code, whereas eachgroupof the 1998 cohort was reared n a separate ank and couldbe assigneda unique group code. Smolts (table 1) of the 1993,1994 and 1998 cohorts were released to Lough Furnace on 3May 1994, 3 May 1995 and 29 April 1999, respectively.Microtags and tissue specimens were recoveredfrom returning

Proc. R. Soc. Lond. B (2003)


4/9

Fitness and extinctionof wild Atlantic salmon P. McGinnity and others 2445

Table 1. Experimental groups of Atlantic salmon in the 1993, 1994 and 1998 cohorts.

cohort code number of number of number of eyed eggs tofemales males families riveraBurrishoole wildfarmF1 hybrid: wild females x farm malesF, hybrid: farm females x wild malesBurrishoole wildfarmF1 hybrid: wild females x farm malesF1 hybrid: farm females x wild malesBurrishoole wildfarmF2 hybrid: F1 hybridx F1 hybridBC1 wild backcross: F1 hybridx wildBC1 farm backcross: F1 hybridx farm

199319931993199319941994199419941998199819981998

Wild93Farm93F1HyW93F1HyF93Wild94Farm94F1HyW94F1HyF94Wild98Farm98

F2HyBCiW1998 BC1Fa The number of eggs planted out in the experiment river.b The number of microtagged smolts released to sea.

adult salmon (1993 cohort: n=49; 1994 cohort: n=67; 1998cohort: n = 535) caught by anglingin Lough Furnace and at thesea entry upstream traps on the Burrishoole system. For the1998 cohort, additional returning fishes (n = 357) were alsoobtained from the commercial net fisheries around the Irishcoast through the National Microtag Recovery Programme(Wilkins et al. 2001). Fecundity (F) was estimated from theweight (W) (in grams) of returningfemales using the formulagiven by Mangel (1996): F=cWk, where c=4.832 andk = 0.8697.

Sampled individuals, except returning adults of the 1998cohort (identifiable from microtags), were identified bymicrosatelliteprofiling involving six microsatellite loci as threesets of two multiplexed primers:Ssa 407 (Cairneyet al. 2000)and Ssa 202 (O'Reilly et al. 1996); SSSP2201 and SSSP2210;and SSSP1605 and SSSP2215 (Verspoor et al. 2002). PCRamplification protocols varied for each of the three sets of pri-mers. Common components in all three assays (12 gl finalvolume) were: 100 ng template DNA, 20 mM Tris-HCl (pH of8.4), 50 mM KCl, 1.5 mM MgC12,100 ILM NTP and 0.5 unitsTaqDNA polymerase (Invitrogen).The forwardprimerof eachprimer set was fluorescently labelled (IRD 700 Li-Cor for Ssa202, SSSP2201, SSSP2210 and SSSP1605; IRD 800 Li-Cor forSsa 407 and SSSP2215). The varyingamounts of each primerused in the multiplex PCR reactions were as follows: Ssa 407(4 pmol) and Ssa 202 (1.5 pmol); SSSP2201 (3.5 pmol) andSSSP2210 (0.5 pmol); and SSSP1605 (0.3 pmol) andSSSP2215 (3 pmol). Following an initial cycle at 94 ?C for4 min, PCR cycling parameterswere specificto each of the threeprimersets: Ssa 407 and Ssa 202, 24 cycles of 94 ?C for 50 s,57 ?C for 50 s and 72 ?C for 50 s; SSSP2201 and SSSP2210,29 cycles of 94 ?C for 40 s, 55 ?C for 40 s and 72 ?C for 40 s;SSSP1605 and SSSP2215, 28 cycles of 94 ?C for 1 min, 62 ?Cfor 1 min and 72 ?C for 1 min. Following PCR, 4 gl of stop sol-ution (95% formamide, 10mM NaOH, 10mM EDTA and0.01% pararosaniline)was added to each 12 gl reaction. Theresulting amplified products were denatured at 80 ?C for 3-4 min, and 1 gtlof the reaction was loaded onto 96-well, 25 cmlong, 6%polyacrylamidegels (1 x TBE buffercontaining 5.6 Murea) mounted in an automated Li-Cor double-dye system. Acommercially available size ladder for the Li-Cor system(MicroStep-20a from Microzone, UK) was run every 15 speci-

mens to size allelic fragments.Gels were run at a constant powerof 40 W and at a temperatureof ca. 50 ?C for 1-2 h. The GENE-PROFILER genotyper software (Scanalytics) was used to scoregenotypic data from the Li-Cor, with all databeing subsequentlychecked manually. Progeny were identified to family and groupparentage using the FAPprogram (J. B. Taggart, unpublisheddata) as previously described (McGinnity et al. 1997). Overall96.7% of individuals were unambiguously assigned to a singlegroup.As relative sizes of groups in some samples is determined byboth survivaland migration, this is referred o as representation.Differences in survival and representation,relative to the wildgroup, were tested using G-tests incorporatingWilliams'correc-tion (Sokal & Rohlf 1995; McGinnity et al. 1997), and wereexpressed relative to a wild value of 1.0. Length data did notmeet the requirements for parametricanalyses and were ana-lysed using the Kruskal-Wallis non-parametric one-wayANOVA and, if this showed significant overall heterogeneity,unplanned pairwisecomparisonswere carriedout using Dunn'smultiple comparison test.

3. RESULTS(a) Fertilization to eyed-egg stage and hatcherycontrols

The highest egg mortality occurred in the F2 hybridgroup (median 68%); this was significantly higher thanin all other groups (e.g. wild 3%; figure 2). Because thebackcrosses, which used aliquots of the same eggs as F2hybrids, showed significantly lower mortality (8%), thishigh F2 hybrid mortality is not caused by maternal or egg-quality effects. There was also no significant difference inmortality between the two paternal groups of families,indicating that it is not a paternal effect. Thus, this high F2mortality most probably reflects outbreeding depression.No significant differences in survival between groupswere found in the hatchery control sampled at 11 months.However, given that total mortality was less than 10%under 'protected' hatchery conditions, there was littleopportunity to detect differential survival. The lack of dif-ferences in the hatchery controls serves to demonstrate

Proc.R. Soc. Lond.B (2003)

groupsmolts to

seab

61568111111118615

1515

615681111111159255

615681111111112 (24 sea)33264545

527314997588686591053710537105371053787879832833795499928

184217221962191495411381211102825449131056617297


5/9


2.01.8

? 1.6-o, 1.4-

a O ~~~~~~~~~~~~1.2 ~~~~~n.s..s.1.0- n.s.n.s. n.s. n.s.a%~~ n.s.(Dr~~~~~~1~~ ~n.s.n.s.0.8-0.6-0.4-0.2 -

0 3 - m Fm m [ mm m m m 3 m v m msmolts smoltsfertilization:eyed egg 0 + parr parrmigrants actualexperiment estimatedseatrap entry

Figure 2. Survival/representation n samples of farm and 'hybrid' groups relative to a wild value of 1.0. Data for thefertilization to eyed-egg stage are shown for all cohorts and numbers indicate total number of families. Parr and smolt data areshown for the 1998 cohort only (earlierdata in McGinnity et al. 1997) and numbers indicate individuals in sample.Significance of pairwise differences from the wild group (x2) are indicated above the bars: n.s., non-significant; - indicatessignificantly less; + indicates significantly more; -/+, 0.05 < p < 0.01; --/++, 0.01 < p < 0.001; ---/+++, p < 0.001.Two measures of smolt output are given: smolts actual experiment trap is the actual count of smolts leaving the experimentriver;smolts estimated sea entry is the estimated output of smolts to the sea based on the assumption that displaced parr havethe same survival as parr of the same group remaining in the experiment river.

that all groups were potentially equally viable and that thedifferential survival apparent in the wild was the result ofgenetic and/or maternal differences.(b) Juveniles in fresh waterAs found for the 1993 and 1994 cohorts (McGinnity etal. 1997), in the samples of 0+ parr (i.e. first-yearjuveniles) of the 1998 cohort from the experiment river atthe end of the first summer, farm salmon showed signifi-cantly lower representation than wild salmon. The BC,backcross to farm salmon also showed significantly lowerrepresentation but the F2 hybrid and BC1 backcross-to-wild groups were not significantly different from wild sal-mon (figure 2). As previously (McGinnity et al. 1997),the relative numbers of parr in the river underestimate thedifferences in survival, as downstream movement occurredprior to the sample being taken. During the period fromMay 0+ to September 1+ (i.e. the second year), the high-est proportion of emigrant parr, caught in the experimenttrap, was from the wild group and the lowest from thefarm group, with 'hybrids' intermediate in representation.There was significant heterogeneity in length (Kruskal-Wallis X2, p < 0.001) among the groups in the 0+ parrwith the decreasing order of size being: farm; BClbackcross to farm; F2 hybrid; BC1 backcross to wild; andwild. The inverse relationship between emigration and sizeindicates competitive displacement of the wild salmon by

the larger farm and 'hybrid' fishes, in agreement withearlier findings for the 1993 and 1994 cohorts (McGinnityet al. 1997).Final migration from the river occurred in two phases:an 'autumn' migration of presmolts in the period fromNovember to January, and a typical 'spring' smoltmigration from March to May (McGinnity et al. 1997).In the autumn presmolts of the 1998 cohort, 81% weremature male parr, with wild parr showing greatest rep-resentation, farm parr the least and hybrids intermediate.As there was no significant difference (figure 2) inpresmolt + smolt output between wild and farm salmon,this difference in parr maturity cannot be explained bydifferential survival or previous parr emigration. Mostmigration (90%) occurred in the second autumn and thefollowing spring, i.e. 2+ smolts, with 6% as 1+ smolts and4% as 3+ smolts. There were no significant differencesbetween groups in the age of smolting. As previously,smolt output was assessed in two ways (McGinnity et al.1997; figure 2). The first approach was to consider onlythe actual number of migrants (presmolts + smolts)caught in the experiment trap, which assumes that earlieremigrant parr do not survive. As a result of displacementof the wild parr, the F2 hybrid and BC1 backcross-to-farmgroups produced significantly more smolts than the wildgroup, but farm numbers were not significantly different,in spite of this group showing the lowest parr emigration.



6/9


n.s.1.(c)

141 12

10o10

c642c; - -

1993

ao

-cl0au

c,'-o++ .>

_ Q3U CZ19981998

Figure 3. Return of adults relative to a wild value of 1.0: (a) 1SW adults; (b) 2SW adults; and (c) 1SW + 2SW adults withfemale 2SW fish numbers being multiplied by 2.3 to correct for increased fecundity; 1993, 1994 and 1998 refer to the threecohorts. Numbers within the bars for (a) and (b) indicate the numbers of individuals, and for (c) the corrected number ofindividuals. Sampling of the 1993 cohort 1SW was incomplete and the two age groups cannot be combined. No wild fishesreturned as 2SW for the 1994 cohort and relative return cannot be calculated. Significances of pairwise differences from thewild group are indicated as for figure 2.

As suitable habitat is present in the river downstreamof the experiment trap and in freshwater Lough Feeagh,emigrant parr were potentially able to survive and producesmolts, as was demonstrated for the 1993 cohort(McGinnity et al. 1997). The second measure of smoltoutput assumed that emigrant parr would have had thesame survival downstream as parr of the equivalent groupremaining in the experiment river, and combines the esti-mated number of smolts produced from these with theactual experiment-trap migrants to give an estimate of sea-entry smolts. The farm and BC1 backcross-to-farm groupshad significantly lower estimated sea-entry smolt pro-duction relative to wild salmon, with farm salmon produc-ing 55% smolts relative to wild salmon (figure 2).(c) Marine survival and maturity: all cohortsAdult salmon returned from the sea after one or twowinters (1SW and 2SW). In the 1SW returns, all groups,except the BC, backcross to wild, showed a significantlylower return relative to wild salmon (figure 3). In the 2SWreturns, all groups showed a proportionately greater returnthan wild salmon. However, the Burrishoole wild popu-lation is primarily a 1SW stock, and the 2SW return wasonly 2.5% of the total return of 203 fishes. Farm salmonhave been bred for late maturity, a trait with high herita-bility under such conditions (J6nasson et al. 1997). As eggdeposition, rather than sperm, is likely to be a limitingfactor in salmon recruitment, it is necessary to takeaccount of the greater egg production of 2SW femalescompared with 1SW fishes. Based on the mean weights ofreturning females in the two groups, it was found that the2SW fishes potentially produced 2.3 times more eggs.However, even taking account of this differential egg pro-duction of 1SW and 2SW females, the total potential eggdeposition was significantly lower than for wild salmon inall groups except BC1 backcross to wild (figure 3). If thiscorrection for increased fecundity is not applied, the rela-tive performances of farm and 'hybrid' groups (except BCIbackcross to wild) were even lower. Thus, based on num-bers of fishes alone, farm salmon had a return relative to

wild salmon of 4% but, when corrected for the greaterreturn of 2SW salmon in the farm group, the value is 7%of potential egg deposition. Whether or not a correctionfactor is applied, and the exact value used, does notchange the rank order of adult returns and has only aminor effect on the magnitude of the estimated overalllifetime success.(d) Overall lifetime successThe product of the survivals at the different life-historystages can be used as a quantitative estimate of overalllifetime success, which, by taking account of differentialegg production between 1SW and 2SW females, can beequated to potential fitness (table 2). It should be notedthat the same fishes were not used for the freshwater andmarine stages, and thus combined effects of selection inboth stages are excluded. In the experimental conditionswhere parr emigrants survived, farm salmon had an esti-mated lifetime success of 2% relative to wild fishes. The'hybrids' showed intermediate success and decreased insurvival in the order: BC1 backcross to wild; F1 hybrid(wild mother); F2 hybrid (but marine stage not measuredfor this group); BC1 backcross to farm; and F, hybrid(farm mother). Under the scenario where only the actualsmolts at the experiment trap are considered, all groupsincreased their relative survival, as a result of displacementof wild parr, but the rank order is similar.

4. DISCUSSIONAND CONCLUSIONSPrevious studies (Fleming et al. 1996, 2000) haveshown that farm salmon differ in their breeding behaviourfrom wild fishes and have lower breeding success. Our

experiment, starting with fertilized eggs, was designed toeliminate behavioural differences between spawning adultsand to examine the effect of genetic differences on survivaland performance. The concordance of the results in thethree cohorts, where the same groups of fishes were com-pared, i.e. Burrishoole wild salmon and farm fishes, con-siderably increases confidence in the findings. Thus farm


(a)

o

s

CA

(b)1.2-1.0'0.8-0.6-0.4-0.2.

0-

1994 1998

m+

---~~- ~n -lC4 m-c :4-:4 ~


7/9


Table 2. Lifetime successes of the wild, farm and 'hybrid' groups.(Results averaged over several cohorts where available (this study and McGinnity et al. (1997)). Survival of the wild group istaken as 1.0. Where another group is not significantly different from the wild group, at a particularstage, it is also given a valueof 1.0. Where a group is significantly different from the wild group, then the actual survival relative to the wild group is used.Note that data for marine survival of the F2 hybrid group are not available and a value of 1.0 is used, hence lifetime successvalues are maximum estimates.)fertilization-eyedegg

1.01.01.00.870.341.00.79

eyed egg-smolta eyed egg-smoltb smolt-adult lifetime successa lifetime successb1.00.890.730.501.00.790.41

1.01.01.00.631.841.590.76

1.01.00.580.61n.a.0.390.07

1.00.890.420.27(0.34)0.310.02

1.01.00.580.33(0.63)0.620.04

a This assumes that displaced parrhave the same survival as parr of the same group remaining in the experiment river, i.e. thatthe river is not at its parr carryingcapacity and spare habitat is available for displaced parr.b This assumes that displaced parr emigrating from the experimental river do not survive, i.e. that the river is at its parr carry-ing capacity.salmon consistently show the lowest freshwater and mar-ine survival in all cohorts. There is no evidence for hybridvigour, with F1 and BC1 hybrids being intermediatebetween wild and farm salmon in survival, growth andparr maturity. There is clear evidence of outbreedingdepression in the F2 hybrids, as might be expected froma breakdown of coadapted sets of alleles following recom-bination of parental chromosomes. However, this out-breeding depression appears to be limited to the earlydevelopmental stage, with allelic combinations that sur-vive this stage causing no further reduction in survival.The order of lifetime success among the experimentalgroups is indicative of additive genetic variation for sur-vival, with genetic changes having occurred in the farmstrain that reduce its survival under natural conditions.Intentional and unintentional selection and genetic driftduring domestication have resulted in many geneticallybased behavioural and physiological changes that couldreduce survival under natural conditions. At least some ofthese differences appear to be the result of selection forfaster growth in farm strains (Gjoen & Bentsen 1997).Recent evidence (Fleming et al. 2002) suggests that thisselection has indirectly targeted individuals with highergrowth-hormone production and that increased aggressionand decreased response to predation risk (Fleming &Einum 1997; Johnsson et al. 2001) are consequences ofthis increased endocrine activity (Jonsson et al. 1996).The difference in overall survival between the wild-motherand farm-mother F1 hybrids indicates that there is also amaternal component to survival. As expected, this compo-nent primarily influences early survival in both the fertiliz-ation-to-eyed-egg and freshwater juvenile stages.Our results for marine survival and maturity differ fromthose of Fleming et al. (2000) in that they did not findany difference in survival between farm and wild salmon.They also found that F1 hybrids had a lower mean age atmaturity as a result of earlier smolting. In part, at least, thedifferences are probably caused by the fact that Fleming etal. (2000) used farm salmon from the largest Norwegianbreeding programme whereas we used farm fishes derivedfrom the Norwegian Mowi strain. The life-history charac-

teristics of the wild populations used, and the differentnatural conditions under which the experiments wereundertaken in Norway and Ireland, are also likely to havecontributed to the differences observed.In many cases, where escaped farm salmon enter a river,

production of F1 hybrids rather than pure farm offspringis the outcome (Clifford et al. 1998; Fleming et al. 2000).Thus part of the potential wild juvenile recruitment is con-verted to hybrids in the first generation and to backcrossesin the second and later generations. Inevitably the lowerlifetime success of such 'hybrids' will reduce the fitness ofthe wild population. However, these 'hybrids' can alsoresult in an increase in 2SW salmon in rivers that are nat-urally primarily 1SW producers. While this may be desir-able from an angling perspective because it increases thenumber of larger fishes, given their reduced lifetime suc-cess, these larger 'hybrids' would not compensate for theloss of wild recruitment, thereby resulting in a decrease infitness in the population. Also, in this study, only thepotential egg deposition was estimated, and these largerfishes may, in addition, have reduced survival in freshwater (Fleming 1996) and possibly reduced mating suc-cess, further decreasing fitness. Thus our results providelong-awaited empirical support for many of Hutchings'(1991) predictions on fitness reduction and extinction inwild salmon populations, which were based on modellingthe influence of spawning intrusions of farm fishes andassuming fitness reduction.

Backcrossing of F1 hybrids to wild fishes in the secondand subsequent generations will result in introgression ofalleles from farm salmon to wild salmon. As only a fewfarm strains are in general use, this gene flow will resultin a reduction in the genetic heterogeneity that is presentamong Atlantic salmon populations (Youngson et al.2003). Given the evidence (Taylor 1991; Youngson &Verspoor 1998) for local adaptation of Atlantic salmonpopulations, loss of genetic heterogeneity will reduce theadaptive potential of the species. Because, typically, farmstrains show reduced variability (Mjolnerod et al. 1997;Clifford et al. 1998), introgression will produce areduction in genetic variability in wild populations


groupwildBC1WFiHyWFiHyFF2HyBC1Ffarm


8/9


(Tufto & Hindar 2003) potentially resulting in inbreedingdepression (Wang et al. 2001, 2002).In addition to direct genetic effects, offspring of escapedfarm fishes may also reduce the size of the wild populationdue to competition. Although overall survival of farm and'hybrid' fish was lower in the experiment, due to theirlarger size, surviving fish resulted in competitive displace-ment of wild parr. Given the selection for increasedgrowth in the farm strain, this larger size is not surprising.Fleming et al. (2002) have shown that farm salmon showincreased aggression, which may further favour such fishesin competitive encounters. The impact of this displace-ment on wild smolt production will depend on whether ornot these displaced parr can survive downstream. In theexperimental conditions of the present study, these parrcould survive as there was unoccupied juvenile habitatdownstream and in Lough Feeagh. Such survival wouldnot occur if suitable unoccupied habitat is absent, forexample, when a river is at its parr carrying capacity orwhere the spawning area debouches directly to the sea, asmay be typical for escaped farm salmon spawning in somerivers (Webb et al. 1991) but not others (Webb et al. 1993;Fleming et al. 2002). Fleming et al. (2000) noted a morethan 30% reduction of smolt output in their simulatedfarm-escape experiment. As our experiment indicates,because farm salmon especially and 'hybrids' show lowermarine survival than wild salmon, they do not make upfor this loss of wild smolt production. Thus the competi-tive effect on its own serves to reduce the fitness of thewild population irrespective of any genetic changes to thepopulation. Reduction of the effective population size ofa wild population will result in lowered genetic variability.Mature male parr have been shown to fertilize a substan-tial proportion of eggs (Thomaz et al. 1997; Martinez etal. 2000; Taggart et al. 2001) and thus parr maturity maybe an extremely important factor in increasing effectivepopulation size. The reduced level of parr maturity, asfound in our experiment for farm and hybrid parr, mayfurther reduce effective population size.The overall extent of reduction in fitness in the wildpopulation, as a result of both interbreeding and compe-tition, will depend on many factors, including availabilityof unoccupied juvenile habitat, relative numbers of wild,farm and hybrid salmon, and mating success. However,the quantitative data available for the first time, to ourknowledge, from this study of both F1 and F2 generations,and from the study of an F1 generation by Fleming et al.(2000), will enable much more accurate modelling of theimpact of farm escapes under a range of scenarios. Irres-pective of the exact extent of fitness reduction, the factthat farm escapes are repetitive, often resulting in annualintrusions in some rivers, means that such reductions infitness are cumulative, which could potentially lead to anextinction vortex in endangered populations, i.e. whererecruitment is close to the replacement level. Thus only asmall reduction in fitness is required to change popu-lations with a marginally positive growth into populationswith negative growth.There have been recent calls in the angling press forsurplus (often 2+) smolts from the salmon-farming indus-try to be used for deliberate stocking in an attempt toenhance population levels in depressed wild populations.Our results indicate that such a practice would be highly

detrimental and would have the reverse effect to thatintended. Indeed the effects would be even greater thanthose produced by accidental escapes owing to the largenumber of farm fishes introduced, relative to wild fishes,and to the regular nature of such deliberate stocking. Simi-lar effects would also be expected from deliberate stockingwith other domesticated strains of salmon and trout, theimpact depending on the extent of domestication. Indeedthis study provides a general model for the likely impli-cations of the deliberate or accidental introduction of farmfishes and other aquatic animals into areas where naturalpopulations exist. A reduction in fitness does not requirethat the wild populations are locally adapted, but only thatdomestication results in genetic changes that reduce thefitness of farm and hybrid offspring in the wild. As dom-estication inevitably occurs in captivity, this study mayalso have implications for the release of zoo-bred animalsinto areas where wild conspecifics still exist (Rodriguez-Clark 1999).This project was funded by the Marine Institute Ireland, theEuropean Commission and the UK Natural EnvironmentResearch Council. We thank E. Verspoor for providingmicrosatellite primers and for assistance with experimentaldesign, and K. Whelan for logistical support. The commentsof two anonymous referees greatly improved the manuscript.

REFERENCESAtlantic Salmon Federation 2002 Atlantic salmon aquaculture:a primer. See http://www.asf.ca/backgrounder/asfaquacbackgrounder.pdf.Cairney, M., Taggart, J. B. & Hoyheim, B. 2000 Characteris-ation of microsatellite and minisatellite loci in Atlantic sal-mon (Salmo salar) and cross-species amplification in othersalmonids. Mol. Ecol. 9, 2175-2178.Clifford, S. L., McGinnity, P. & Ferguson, A. 1998 Geneticchanges in Atlantic salmon (Salmo salar) populations ofnorthwest Irish rivers resulting from escapes of adult farmsalmon. Can. J. Fish. Aquat. Sci. 55, 358-363.Crozier, W. W. 1993 Evidence of genetic interaction betweenescaped farmed salmon and wild Atlantic salmon (Salmosalar L.) in a Northern Irish River. Aquaculture113, 19-29.Crozier, W. W. 2000 Escaped farmed salmon, Salmo salarL.,in the Glenarm River,Northern Ireland:genetic status of thewild population 7 years on. Fish. Mngmt Ecol. 7, 437-446.Donaghy, M. J. & Verspoor, E. 2000 A new design of instream

incubator for planting out and monitoring Atlantic salmoneggs. N. Am. J. Fish. Mngmt 20, 521-527.Ferguson, A., Taggart, J. B., Prodohl, P. A., McMeel, O.,Thompson, C., Stone, C., McGinnity, P. & Hynes, R. A.1995 The application of molecular markers to the study andconservation of fish populations, with special reference toSalmo. J. Fish Biol. 47(Suppl. A), 103-112.Fiske, P. & Lund, R. A. 1999 Escapes of reared salmon incoastal and riverine fisheries in the period 1989-1998. NINAOppdragsmelding03, 1-23.Fleming, I. A. 1996 Reproductivestrategiesof Atlantic salmon:ecology and evolution. Rev. Fish Biol. Fish. 6, 379-416.Fleming, I. A. & Einum, S. 1997 Experimental tests of geneticdivergence of farmed from wild Atlantic salmon due to dom-estication. ICES J. Mar. Sci. 54, 1051-1063.Fleming, I. A., Jonsson, B., Gross, M. R. & Lamberg, A. 1996An experimental study of the reproductive behaviour andsuccess of farmed and wild Atlantic salmon (Salmosalar).J.Appl. Ecol. 33, 893-905.

Proc. R. Soc. Lond.B (2003)


9/9


Fleming, I. A., Hindar, K., Mjolnerod, I. B., Jonsson, B., Dal-stad, T. & Lamberg, A. 2000 Lifetime success and interac-tions of farm salmon invading a native population. Proc. R.Soc. Lond. B 267, 1517-1523. (DOI 10.1098/rspb.2000.1173.)Fleming, I. A., Agustsson, T., Finstad, B., Johnsson, J. I. &Bj6rnsson, B. Th. 2002 Effects of domestication on growthphysiology and endocrinology of Atlantic salmon (Salmosalar). Can. J. Fish. Aquat. Sci. 59, 1323-1330.Gjedrem, T., Gjoen, H. M. & Gjerde, B. 1991 Genetic originof Norwegian farmed salmon. Aquaculture98, 41-50.Gjoen, H. M. & Bentsen, H. B. 1997 Past, present, and futureof genetic improvement in salmon aquaculture. ICES J.Mar. Sci. 54, 1009-1014.Hansen, L. P., Hastein, T., Navdal, G., Saunders, R. L. &Thorpe, J. E. (eds) 1991 Interactions between cultured andwild Atlantic salmon. Aquaculture98, 1-324.Hindar, K., Ryman, N. & Utter, F. 1991 Genetic effects ofcultured fish on natural fish populations. Can.J. Fish.Aquat.Sci. 48, 945-957.

Hutchings, J. A. 1991 The threat of extinction to native popu-lations experiencing spawning intrusions by cultured Atlan-tic salmon. Aquaculture98, 119-132.Johnsson, J. I., H6jesj6, J. & Fleming, I. A. 2001 Behaviouraland heart rate response to predation risk in wild and dom-esticated Atlantic salmon. Can. J. Fish. Aquat. Sci. 58,788-794.

J6nasson, J., Gjerde, B. & Gjedrem, T. 1997 Genetic para-meters for return rate and body weight in sea-ranchedAtlan-tic salmon. Aquaculture154, 219-231.Jonsson, E., Johnsson, J. I. & Bj6rnsson, B. Th. 1996 Growthhormone increases predation exposure of rainbow trout.Proc. R. Soc. Lond. B 263, 647-651.Lura, H. & Sagrov, H. 1991 Documentation of successfulspawning of escaped farmed female Atlantic salmon, Salmosalar, in Norwegian rivers.Aquaculture98, 151-159.McDowell, N. 2002 Stream of escaped farm fish raises fearsfor wild salmon. Nature 416, 571.

McGinnity, P., Stone, C., Taggart, J. B., Cooke, D., Cotter,D., Hynes, R., McCamley, C., Cross, T. & Ferguson, A.1997 Genetic impact of escaped farmed Atlantic salmon(SalmosalarL.) on native populations: use of DNA profilingto assess freshwaterperformanceof wild, farmed, and hybridprogeny in a natural river environment. ICES J. Mar. Sci.54, 998-1008.Mangel, M. 1996 Computing expected reproductive successof female Atlantic salmon as a function of smolt size. J. FishBiol. 49, 877-892.Martinez, J. L., Moran, P., Perez, J., De Gaudemar, B., Beall,E. & Garcia-Vazquez, E. 2000 Multiple paternity increaseseffective size of southern Atlantic salmon populations. Mol.Ecol. 9, 293-298.Mjolnerod, I. B., Refseth, U. H., Karlsen, E., Balstad, T.,Jakobsen, K. S. & Hindar, K. 1997 Genetic differencesbetween two wild and one farmedpopulation of Atlantic sal-mon (Salmosalar) revealed by three classes of genetic mark-ers. Hereditas127, 239-248.O'Reilly, P. T., Hamilton, L. C., McConnell, S. K. & Wright,J. M. 1996 Rapid analyses of genetic variation in Atlanticsalmon (Salmo salar) by PCR multiplexing of di-nucleotideand tetra-nucleotide microsatellites. Can. J. Fish. Aquat. Sci.53, 2292-2298.Rodriguez-Clark, K. M. 1999 Genetic theory and evidencesupporting current practices in captive breeding for conser-vation. In Genetics and the extinctionof species (ed. L. F.

Landweber & A. P. Dobson), pp. 47-73. Princeton Univer-sity Press.Skaala, 0., Dahle, G., Jorstad, K. E. & Naevdal, G. 1990Interactions between natural and farmed fish populations:information from genetic markers.J. FishBiol. 36, 449-460.Sokal, R.R. & Rohlf, F. J. 1995 Biometry, 3rd edn. NewYork: Freeman.Taggart, J. B., McLaren, I. S., Hay, D. W., Webb, J. H. &Youngson, A. F. 2001 Spawning success in Atlantic salmon(Salmo salar L.): a long-term DNA profiling-based studyconducted in a natural stream. Mol. Ecol. 10, 1047-1060.Taylor, E. B. 1991 A review of local adaptationin Salmonidae,with particular reference to Pacific and Atlantic salmon.Aquaculture98, 185-207.Templeton, A. R. 1986 Coadaptation and outbreedingdepression. In Conservationbiology: he scienceof scarcityanddiversity (ed. M. E. Soule), pp. 105-116. Sunderland,MA: Sinauer.Thomaz, D., Beall, E. & Burke, T. 1997 Alternative repro-ductive tactics in Atlantic salmon: factors affecting matureparr success. Proc. R. Soc. Lond. B 264, 219-226. (DOI

10.1098/rspb. 1997.0031.)Tufto, J. & Hindar, K. 2003 Effective size in management andconservation of subdivided populations. J. Theor.Biol. 222,273-281.Verspoor, E., Piertney, S., Patterson, S. & Knox, D. 2002 Anovel set of Atlantic salmon Salmo salar microsatellite locifor population studies and pedigree analysis. GenBankAY081807, AY081808, AY081810, AY081812.Wang, S., Hard, J. J. & Utter, F. 2001 Salmonid inbreeding:a review. Rev. Fish Biol. Fish. 11, 301-319.Wang, S., Hard, J. J. & Utter, F. 2002 Genetic variation andfitness in salmonids. Conserv.Genet. 3, 321-333.Webb, J. H., Hay, D. W., Cunningham, P. D. & Youngson,A. F. 1991 The spawning behaviour of escaped farmed andwild adult Atlantic salmon (Salmo salar L.) in a northernScottish river. Aquaculture98, 97-110.Webb, J. H., McLaren, I. S., Donaghy, M. J. & Youngson,A. F. 1993 Spawning of farmed Atlantic salmon, Salmo salarL., in the second year after their escape. AquacultureFish.Mngmt24, 557-561.Wilkins, N. P., Cotter, D. & 6 Maoileidigh, N. 2001 Oceanmigration and recapturesof tagged, triploid, mixed-sex andall-female Atlantic salmon (SalmosalarL.) released from riv-ers in Ireland. Genetica111, 197-212.Winkelman, A.M. & Peterson, R. G. 1994 Heritabilities,dominance variation, common environmental effects andgenotype by environment interactions for weight and lengthin chinook salmon. Aquaculture125, 17-30.WWF 2001 The status of wild Atlantic salmon: a riverby river

assessment. See http://www.wwf.org.uk/filelibrary/pdf/atlanticsalmon.pdf.Youngson, A. F. & Verspoor, E. 1998 Interactions betweenwild and introduced Atlantic salmon (Salmo salar). Can. J.Fish. Aquat. Sci. 55(Suppl. 1), 153-160.Youngson, A. F., Hansen, L. P. & Windsor, M. L. 1998 Inter-actions between salmon culture and wild stocks of Atlanticsalmon: hescientificandmanagementssues.Trondheim, Nor-way: NINA.Youngson, A. F., Jordan, W. C., Verspoor, E., Cross, T.,McGinnity, P. & Ferguson, A. 2003 Management ofsalmonid fisheries in the British Isles: towards a practicalapproach based on population genetics. Fisheries Res. 62,193-209.As thispaperexceeds the maximum ength normallypermitted, heauthorshaveagreed o contribute o production osts.

Proc. R. Soc. Lond.B (2003)

extinction of atlantic salmon

Documents